MICROBIOLOGICAL REVIEWS, Mar. 1983, p. 46-83 Vol. 47, No.

1
0146-0749/82/010046-38$02.00/0
Copyright 0 1983, American Society for Microbiology

Bactericidal and Bacteriolytic Activity of Serum Against

Gram-Negative Bacteria
PETER W. TAYLOR
Bayer AG, Institut fdr Chemotherapie, Pharma-Forschungszentrum, 5600 Wuppertal-], Federal Republic of
Germany
INTRODUCTION ............................................................. 46
MEASUREMENT OF BACTERICIDAL AND BACTERIOLYTIC ACTIVITY .................. 47
Deectio of Serum Bactericidal Activity ............................................................. 47
Deteto of Serum Bacterlolytic Activty ............................................................. 51
MECHANISM OF SERUM BACTERICIDAL AND BACTERIOLYTIC ACTION ........ ..... 51
Requirenents for Serum-Mediated Kilng: Complement ............................................ 51
Role of Antibody ............................................................. 54
Role of Lysozyme ............................................................. 57
Additional Fact ............................................................. 58
Sequence of Events After Exposure to Serum ........................................................... 59
RESISTANCE TO SERUM BACTERICIDAL ACTIVITY .................... ........................ 62
Defiition of Serum Restance. ............................................................. 62
Location of Serum Resbtance Determinats at the Outer Membrane ........... ................. 62
Role of Lipopolysaharldes .................. ........................................... 64
Role of Acidic Polysaccharides ............................................................. 66
Role of Plasmid-Determined Factors ............................................................. 68
Role of Other Outer Membrane Proteins .............................................................. 70
SERUM BACTERICIDAL ACTIVITY AS AN IMMUNE DEFENSE MECHANISM .......... 70
Clnical Observations on the Relevance of Serum Resistance as a Bacterial
Pathoenkity Determinant ............................ ................................. 71
Evidenc from Experimental Infecions of Laboratory Animals .................................... 74
CONCLUDING REMARKS ............................................................. 74
LITERATURE CITED ............................................................. 74

INTRODUCTION acetylmuramoylhydrolase; EC 3.2.1.17); this

Exposure of many strains of gram-negative
bacteria to suitable concentrations of human or
animal serum results in loss of viability, and
sometimes dissolution, of the bacterial cells.
Since the recognition, towards the end of the last
century, that the bactericidal and bacteriolytic
properties of serum are destroyed by heating at
56°C, an extensive literature has accumulated
indicating that the killing process is effected by
deposition on or insertion into the bacterial
envelope of thbe assembled terminal proteins of
the complement cascade, the membrane attack
complex (MAC). It is now clear that activation
of complement by gram-negative bacteria can
occur via the classical or the alternative path-
way; the former usually requires for its activa-
tion recognition of bacterial surface antigens by
certain antibody classes, whereas activation of
the latter can be initiated and amplified, in the
absence of antigen-antibody interactions, by
poorly understood structural or conformational
characteristics of the cell surface. Killing by
serum is often, but not invariably, accompanied
by bacteriolysis, an event dependent on ade-
quate amounts of lysozyme (mucopeptide N-
46
basic enzyme degrades peptidoglycan to form
monomers or multiples of the disaccharide-tetra-
peptide unit (265).
The majority of investigators of bactericidal
and bacteriolytic phenomena have utilized se-
rum from humans (26, 222, 328) or from domes-
tic (194, 2%) and laboratory animals (7, 47, 144)
as sources of complement; the capacity to kill
gram-negative bacteria is associated, however,
with serum from a wide variety of warm-blooded
and cold-blooded animals, including fish (218,
322), and probably reflects the distribution of
immunological responsiveness and the comple-
ment system in the animal kingdom. Susceptibil-
ity to the serum bactericidal system is a wide-
spread characteristic of gram-negative bacteria;
in addition to the many well-documented in-
stances of enterobacterial susceptibility to com-
plement, serum is known to possess bactericidal
and bacteriolytic activity against susceptible
representatives of practically every gram-nega-
tive genus so far examined. In fact, any procary-
ote that presents a lipid bilayer membrane to the
external environment would appear to be poten-
tially susceptible to complement killing. Al-
though outside the scope of this review, comple-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
ment in conjunction with specific antibody is
also capable of lysing lipid-enveloped viruses
(67).
In addition to susceptible strains, there are
gram-negative bacteria that appear refractory to
the serum bactericidal and bacteriolytic sys-
tems; these resistant strains are frequently iso-
lated as causative agents of infections involving
tissue damage. It has therefore been suggested
that serum resistance is an important determi-
nant of virulence in at least some infections due
to gram-negative bacteria; although much of the
literature does tend to show a relationship be-
tween insensitivity to serum and the ability to
cause infection, the precise role and relative
importance of serum resistance at any stage of
the infection process is at present not clear.
It has often been noted that smooth strains of
gram-negative bacteria, synthesizing a lipopoly-
saccharide with a high degree of substitution of
core units by 0-specific side chain moieties, are
more resistant to serum than rough isolates or
mutants that have lost the ability to either syn-
thesize or attach the 0 antigen component of
lipopolysaccharide (214, 222, 260, 300, 329).
Similarly, resistant strains can be sensitized to
serum by agents known to disrupt the integrity
of the outer membrane (246). Serum-resistant
Escherichia coli strains are less likely to be
agglutinated by anti-O antigen antibodies than
susceptible strains (201); 0 inagglutinability is
often correlated with the presence on the bacte-
rial surface of acidic polysaccharide K antigens
(227). These and many other observations make
it likely that resistance to serum is determined
by the presence of structures at or near the
bacterial surface capable of interfering with the
formation, attachment, or subsequent activity of
the MAC. The recent realization that some
antibiotic resistance plasmids carry a determi-
nant which reduces the serum susceptibility of
E. coli strains (244) has led to a substantial
increase in interest in the mechanism of serum
resistance, partly due to the fact that plasmid-
determined serum resistance is an easily quanti-
fiable parameter that lends itself to investigation
by recombinant DNA techniques (165) and also
because it is now important to determine wheth-
er or not administration of antibiotics may select
bacteria with undesirable invasive properties
(33).
The large number of studies devoted to serum
bactericidal and bacteriolytic reactions have not
been the subject of a full literature survey since
the excellent review by Inoue (127), which
appeared in 1972. Since then, a number of im-
portant advances in our understanding of the
activation of complement pathways by viable
gram-negative bacteria and of the nature of the
bacterial surface in relation to host-parasite in-
47
teractions justify a reexamination of these reac-
tions. Bactericidal and bacteriolytic systems will
also be discussed in relation to the defense of the
host against infection.
MEASUREMENT OF BACTERICIDAL AND
BACTERIOLYTIC ACTIVITY
Detection of Serum Bactericidal Activity
In its most simple form, determination of
serum bactericidal activity involves exposure of
a suspension of viable organisms to a suita-
ble concentration of antibody and complement,
incubation at the optimum temperature for com-
plement activity, and determination, after suit-
able periods of time, of the absolute concentra-
tion of surviving organisms by some form of
viable counting. Unfortunately, no single tech-
nique has emerged as a readily acceptable and
widely applicable standard assay for estimation
of serum activity, and the reader is confronted
by a bewildering array of assays that frequently
make meaningful comparisons between studies
difficult. As a number of parameters have been
defined that significantly influence the outcome
of the serum bactericidal reaction, it is some-
what surprising that no definitive study of opti-
mun conditions for bactericidal activity has so
far been undertaken. It is, for example, well
established that the serum susceptibility of a
gram-negative strain is influenced by the growth
conditions employed for preparation of the bac-
terial inoculum; this may be partly due to envi-
ronmental influence on the biosynthesis of cell
surface structures that affect the degree of serum
susceptibility (302) and partly to its effect on the
metabolic state of the target cell (106). In batch
culture, both the nature of the medium and the
phase of growth during which the cells are
harvested may influence the response to serum.
It has been repeatedly observed that E. coli and
other gram-negative bacteria are more readily
killed by serum when in the early logarithmic
phase than in either the lag or stationary phases
(54, 55, 263). In fact, susceptible strains may
appear to be completely refractory to the serum
bactericidal system if the bacteria are harvested
before the onset of rapid cell division (54).
Melching and Vas (180) demonstrated that
logarithmic-phase E. coli 0111:B4 cells were
more susceptible when grown in a simple salts
medium than in a nutritionally complex medium.
Furthermore, cells offered glycerol or acetate as
a carbon and energy source were less suscepti-
ble than glucose- or succinate-grown cells; the
authors postulated that these differences might
be related to the ability of the cell to repair
complement-mediated membrane damage. Simi-
lar observations were made by Maaloe (163,
164), who noted that Salmonella typhimurium
48 TAYLOR
grown in diluted broth was more susceptible to
human serum than when grown in undiluted
broth; the experimental conditions were ar-
ranged in such a way that the growth rates in
these media were identical. Addition of various
monosaccharides and organic acids to the sys-
tem led to a phenotypic increase in serum resist-
ance, whereas addition of ammonium salts had
the opposite effect. The serum susceptibility of
chemostat-grown E. coli urinary isolates was
markedly dependent on the nature of the limiting
nutrient and, to a lesser extent, on the dilution
rate (302); in addition, the reproducibility of
serum bactericidal kinetics was higher with che-
mostat cultures than when batch-cultured cells
were used. These observations make it essential
that particular attention be given to the proce-
dure for inoculum preparation for serum bacteri-
cidal assays; when batch culture is utilized,
bacteria should be harvested during the early
logarithmic phase of growth from, if possible, a
medium that can be formulated with a minimal
amouiit of batch-to-batch variation. Bacteria
should be grown at a temperature at or near the
optimum, as cultivation at suboptimal or elevat-
ed temperatures frequently results in alteration
of serum killing rates (144, 229). In a comparison
of three established serum bactericidal assays,
DeMatteo and colleagues (55) found that the
system demonstrating the highest degree of kill-
ing activity was the only one using cells in early
logarithmic-phase culture. It is therefore unfor-
tunate that many procedures use bacteria in an
undefined and variable metabolic state, includ-
ing those washed from overnight agar slope
cultures and suspended in diluent (138, 183,
296).
The procedure adopted for harvesting and
washing of cells before exposure to serum may
influence the outcome of the bactericidal reac-
tion; the frequently adopted procedure of cen-
trifugation at 0 to 4°C is likely to temperature
shock the cells and should be avoided (342).
Cells that have been extensively washed with
buffer may show increased susceptibility to se-
rum, and a more accurate estimation of true
serum susceptibility may be obtained by resus-
pension in buffer without washing (78).
Complement activation in the bactericidal re-
action may depend on an initial antigen-antibody
interaction at or near the surface of the bacterial
cell. Classical pathway function normally re-
quires the formation of a complex involving
immunoglobulin M (IgM) or IgG molecules (28,
253, 261), and amplification of the alternative
pathway may also require the participation of an
immune reaction (81, 151). The presence in the
reaction mixture of antibody molecules that are
able to participate in complement activation is
necessary, therefore, to properly assess the out-
MICROBIOL. REV.
come of a bactericidal reaction, particularly as
both pathways may function simultaneously to
kill susceptible bacteria (242, 282, 306). Early
colonization of the intestinal tract by commensal
bacteria ensures that small quantities of antibod-
ies directed against the surface antigens of many
types of gram-negative bacteria are present in
the blood and tissue fluids of humans (150, 168)
and a variety of animals (150, 289). Thus, IgM
and IgG antibodies directed against surface anti-
gens of E. coli (45) and other enterobacteria
(187, 189) as well as against Neisseria species
(45, 100) and Haemophilus influenzae (219) can
be detected in the serum of adults in the absence
of any history of immunization or specific infec-
tion. In those cases in which the presence of
antibodies capable of participating in comple-
ment activation can be demonstrated, the serum
can be utilized in the bactericidal assay as a
source of both antibody and complement. Anti-
bodies against some overtly pathogenic gram-
negative bacteria, such as Vibrio cholerae and
Salmonella typhi, are generally absent from non-
immune sera, and it has been common practice
to supplement the bactericidal system with an
antibody source in the form of heated (56°C)
immune serum when investigating such organ-
isms (261). This procedure is, however, compli-
cated by the fact that natural and immune anti-
bodies differ in physicochemical properties that
affect their behavior in bactericidal tests (189).
Also, immune sera are rich in IgG, which func-
tions poorly in complement activation in com-
parison to IgM; at high concentrations, IgG
molecules may inhibit the complement-mediated
killing mechanism (213, 220). These factors must
be taken into account when considering the
design of serum bactericidal assays.
Since the realization (254) that a very high
proportion of cases of neonatal meningitis are
caused by E. coli strains carrying the Kl poly-
saccharide antigen, much effort has been spent
on examining the possibility of a correlation
between Kl carriage and serum susceptibility of
clinical isolates; in these cases, it would seem
particularly relevant to assess the antibody con-
tent of sera, as the Kl antigen is the major
surface polymer in these strains (255), is a poor
immunogen (255), and may prevent effective
activation of the alternative pathway (290).
Possible sources of error may materialize
when bactericidal systems containing heteroge-
neous sources of antibody and complement are
used. Interaction between serum proteins of one
animal species and antibodies of others occur
and may lead to fixation and deviation of com-
plement from the bactericidal system (261).
Large differences have also been reported when
sera from different animal species have been
compared. For example, Ogata and Levine (221)
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
compared the ability of rabbit, human, and guin-
ea pig sera to kill E. coli K-12 strains harboring
plasmid R100; they found that human serum
killed E. coli K-12 J6-2 more efficiently than
either rabbit or guinea pig serum and that the
nature of the serum also affected the ability of
the plasmid to modify the serum response.
Schwab and Reeves (271) compared the bacteri-
cidal activity of sera from eight species of verte-
brates and found considerable variation in the
efficiency with which the sera could kill a num-
ber of enterobacterial strains, although it was
not possible to put the various sera in an unam-
biguous order with respect to this property.
Within a given animal species, however, there
appears to be little variation in the bactericidal
activity of serum from different healthy individ-
uals, reflecting the ubiquity of the complement
system and a shared spectrum of antibody speci-
ficities. This has been particularly well estab-
lished for the human subject (222, 242, 328), and
Olling and associates (223) have demonstrated
that the reactivities of sera from sailors based in
various European, Asian, and South American
countries are identical.
The bactericidal action of serum is demonstra-
ble over a wide range of serum concentrations,
although the rate of killing is known to generally
increase with increasing serum concentration
(203, 259). For the study of serum bactericidal
activity in relation to infection, the majority of
investigators have employed systems containing
relatively high concentrations of serum to en-
sure that killing is not limited by availability of
essential components of either the classical or
alternative complement pathway; thus, concen-
trations in the 20 to 100% range have been
successfully utilized in a number of studies (26,
105, 162, 222, 236, 310). However, assembly of
the alternative pathway from the 11 pathway
proteins purified from human serum has re-
vealed that its bactericidal and bacteriolytic ac-
tivity is insignificant at protein concentrations
equivalent to a 1:16 dilution of serum (268). Clas
and Loos (43) also detected almost no alterna-
tive pathway activity when guinea pig serum
was used at dilutions of greater than 1:10. Re-
cently, E. coli K-12 strains have been used by a
number of investigators to elucidate the nature
of plasmid-determined serum resistance, and of
necessity, very low concentrations, often in the
region of 1 to 4%, have been utilized to achieve a
demonstrable difference between isogenic pairs
(25, 79, 196). It is likely, therefore, that only
classical pathway activity is being observed in
these studies.
Antibody is needed in such small amounts for
complement activation that it is rarely a limiting
factor in bactericidal systems (95). Attention has
been drawn to the possibility that agglutination
49
of bacteria during the bactericidal reaction may
lead to an underestimation of survival rates,
particularly when immune sera are used as a
component of the test system (179). Such an
effect is detectable by using a control containing
serum in which complement activity has been
abrogated by heating at 56°C. The use of low
numbers of bacterial cells will favor the occur-
rence of gross antibody excess and may lead to
inhibition of serum bactericidal action (339).
Essentially identical rates of killing of suscepti-
ble bacteria are seen over a wide range of
inoculum sizes, and inocula up to 107 cells per
ml may be used in bactericidal assays (310).
Olling (222) found no variation in the serum
response of 10 E. coli strains over a concentra-
tion range of 104 to 1010 cells per ml.
A wide range of buffers and other diluents
have been incorporated into serum bactericidal
systems; that these solutions can have signifi-
cant and unpredictable effects on the efficiency
of complement killing was recently demonstrat-
ed by Clas and Loos (43). They found that the
killing of Salmonella minnesota by guinea pig
serum, which proceeded efficiently in the pres-
ence of 0.02 M Tris, pH 7.4, or 0.004 M thiogly-
colate buffer, pH 7.1, was totally abolished in
0.1 M phosphate-buffered saline, pH 7.6, an
effect which might be due to a nonspecific
inhibition of complement action by high cation
concentration (328; P. W. Taylor, unpublished
data) or to a specific inhibition by phosphate
ions (7, 172). Wardlaw (328) found that bacteri-
olysis by human serum of E. coli Lilly was
markedly dependent on the pH and ionic
strength of the system. Somewhat surprisingly,
the pH optimum for bacteriolysis was found to
be 8.4, in comparison to an optimum for hemoly-
sis of 7.15 to 7.35 (170). Similarly, an optimum of
0.06 for ionic strength was found, with a sharp
decrease in activity either side of this peak; this
contrasts to the erythrocyte system, in which
near-maximal rates of lysis occur at an ionic
strength of 0.147 and significant hemolysis oc-
curs at pR = 0.177. This suggests that the require-
ments for lysis of E. coli Lilly may be rather
unusual, as the ionic strength of human serum is
about 0.183 (95). The Tris buffer system at pH
8.4 used by Wardlaw has been widely adopted
for studies of serum bactericidal activity (95,
241, 310), often without due consideration that
optimum conditions required for bacteriolysis
and bactericidal action may be different. A high
pH is, in fact, known to promote the action of
lysozyme on gram-negative bacteria (216, 217).
If relatively high concentrations of serum are
used, the strong buffering capacity of serum may
limit the pH range of the serum-buffer mixtures
and keep any effect due to pH at a minimum
(310). Although the use of Tris buffer has been
50 TAYLOR
criticized on the basis that divalent cations nec-
essary for complement activity may interact
disadvantageously with this buffer (244), it ap-
pears to be comparable to many buffers used in
bactericidal systems (43, 310). In the absence of
definitive data on optimum conditions for bacte-
ricidal activity that apply to the majority of
strains of bacteria commonly investigated, it
would seem advisable to use one of the buffer
systems that have been developed for studies of
immune hemolysis (170); gelatin-veronal-buff-
ered saline plus Mg2> and Ca2, pH 7.35 (89),
provides the essential divalent cations necessary
for complement activity at optimum concentra-
tions, does not affect deleteriously the viability
of gram-negative bacteria, and provides an envi-
ronment in which very high rates of serum
killing of susceptible bacteria can be achieved
(268, 306, 342).
Addition of complex nutrient solutions, such
as Hanks balanced salts solution (222, 344) or
nutrient medium (54, 194), to serum bactericidal
systems may have adverse effects on the ability
of the system to kill normally susceptible strains
(55); nutrient broth is known, for example, to be
anticomplementary at high concentrations (208).
The modification of bactericidal assays by po-
tentially metabolizable sustrates has been stud-
ied in detail by Michael and Braun (184). They
found that high concentrations of various com-
plex nutrient media protected Shigella dysenter-
iae from the bactericidal action of human serum,
whereas concentrations below 0.1% enhanced
its susceptibility. Glucose significantly en-
hanced bacterial susceptibility, whereas other
carbohydrates not metabolized by S. dysenter-
iae had no significant effect; the authors suggest-
ed a relationship between bacterial metabolism
and susceptibility to serum which has subse-
quently been studied in detail (106, 306). A
complex situation was apparent when individual
amino acids were added to the system. Those
amino acids known to stimulate metabolic activ-
ity also enhanced serum killing, others had a
protective effect, and some were without effect
on the rate of killing. Two analogs, 5-methyl-
tryptophan and fluorophenylalanine, interfered
with bacterial killing in the presence of usually
bactericidal concentrations of serum. Nutrient
factors necessary for high killing rates are pre-
sent in adequate concentrations in the serum
used for the assay (306). Therefore, addition to
the system of complex nutrient solutions serves
no purpose and modifies, in an incompletely
understood way, the serum killing kinetics.
The majority of studies of serum bactericidal
activity have used any of a number of viable
counting procedures to determine bacterial sur-
vival rates after exposure to serum-buffer mix-
tures; variations inherent in viable counting are
MICROBIOL. REV.
'1
E
0D la
4-
C
10
0
0
U
0
a.
0 1 2 3
hours
FIG. 1. Response of Citrobacter sp. strain C14 (0),
E. coli CS (0), and E. coli C42 (A) to incubation at
37°C with 75% human serum. Data of Taylor and
Hughes (304).
well known (240) and will not be reiterated here.
However, it has been common to select only one
point in time, usually in the 1- to 2-h range, at
which to evaluate survival; the validity of this
practice must be questioned, as it has been
frequently observed that killing of some strains
does not proceed at a constant rate. For exam-
ple, many smooth, serum-susceptible strains
show a characteristic response (Fig. 1) in which
there is little change in the viable count during
the first hour of exposure, but the number of
survivors after 3 h is generally less than 1%
(299). The rate of killing after alternate pathway
activation is lower than that associated with the
classical pathway (256), and strains of Serratia
marcescens also show a delayed serum-suscep-
tible response as a result of selective activation
of the alternative pathway (320). The true nature
of the interaction between bacterial cells and
activated complement components can only be
evaluated, therefore, when the number of bacte-
rial survivors is estimated at frequent time inter-
vals during the reaction.
Although measurement of the serum response
in the presence of a large excess of the serum
components essential for complement killing
would appear to be the most direct way of
determining the degree of serum killing, some
workers have preferred to estimate the survival
of bacteria over a range of serum dilutions, the
extent of bactericidal action being expressed as
the reciprocal of the highest serum dilution
showing killing activity (194, 263, 271, 296, 318).
It is possible that differences in bactericidal titer
merely reflect different quantitative require-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA 51
ments for some complement components at high 277), involving spotting serum dilutions onto
serum dilutions, and differences established un- seeded agar plates, are difficult to control, and
der conditions of limiting complement concen- some batches of agar possess high anticomple-
trations may not be apparent when bactericidal mentary activity. However, a rapid method de-
kinetics are determined in the presence of ex- veloped by Provonchee and Zinner (241) and
cess amounts of all required serum factors. For using the Steers-Foltz replicator (288) has been
example, the small increases in survival of E. successfully used for the screening of phenotyp-
coli K-12 J6-2 resulting from acquisition of plas- ic modifications to serum susceptibility induced
mid R100 are demonstrable only over a very by growth in sublethal antibiotic concentrations
limited range (0.5 to 3%) of human serum con- (303). A rapid colorimetric assay (195) has made
centrations (25, 221); at these concentrations, possible the genetic analysis of plasmid-borne
key components of the classical pathway may be serum resistance genes by gene cloning tech-
present in limiting amounts and alternative path- niques (1%).
way components will have been diluted to such Techniques that do not rely on determination
an extent as to be nonfunctional. of cell viability have also been used to quantify
The inherent disadvantages of viable counting serum bactericidal activity. Kato and Bito (144)
procedures for estimation of survival after se- took irreversible loss of protein (P3-galacto-
rum treatment led Muschel and Treffers (208) to sidase) synthesis as an index of bacterial killing,
develop an alternative assay based on the ability because serum-treated cells may be active in
of surviving bacteria to remultiply when re- both respiration and macromolecular synthesis
moved from the action of antibody and comple- but inactive in cell division. Fierer and col-
ment. This widely used assay involves exposure leagues (78) measured the release of 5 Cr-la-
of bacteria to suitable serum concentrations for beled lipopolysaccharide from serum-suscepti-
a reaction period of 1 h followed by the addition ble and serum-resistant bacteria and found a
of a standardized volume of growth medium, correlation between serum susceptibility and
such as brain heart infusion broth; the anticom- release of label. However, even heat-inactivated
plementary activity of the medium terminates serum released approximately 20% of 51Cr from
the bactericidal action of the serum, and the serum-susceptible E. coli cells.
density of the cultures after a second incubation
period will be proportional to the number of Detection of Serum Bacteriolytic Activity
organisms surviving the initial exposure to se- Relatively few bacterial strains are lysed by
rum. The optical densities obtained are convert- serum, even in the presence of adequate
ed to percentages of the control values and amounts of lysozyme; this contrasts with the
plotted on a probit scale against the log of the many strains that are killed without obvious
serum volume. The resulting linear relationship lysis (328). Lysis can be followed by measuring
enables the 50% lethal dose of serum to be the optical density of
determined. For obvious reasons, the assay is regular intervals (95,serum-exposed
328).
bacteria at
Alternatively, the
only valid when antibody concentration is varied release of radiolabeled intracellular markers (29,
against a background of constant complement 286) or cytoplasmically located enzymes
concentration, or vice versa. It also depends on can be monitored. Amano and co-workers (7), (130)
the assumption that surviving bacteria that have their extensive studies of immune in
been exposed to potentially damaging amounts bacteriolysis,
of complement will multiply at an identical rate developed a technique in which sodium deoxy-
cholate is added to suspensions of serum-ex-
to bacteria that have not been so exposed after
transfer to nutrient media. However, one could posed bacteria. Only spheroplasts formed by
immune bacteriolysis are solubilized by this
envisage that the cellular activity of serum-
exposed bacteria might be directed towards re- procedure;
after centrifugation and removal of
protein by precipitation, released nucleic acids
pair of complement-damaged sites, resulting in are measured spectrophotometrically.
diminution or temporary suspension of cell divi-
sion. Michael and Braun (185) compared the MECHANISM OF SERUM BACTERICIDAL
photometric assay with a conventional direct AND BACTERIOLYTIC ACTION
plating method and consistently found a greater
number of survivors with the latter procedure, Requirements for Serum-Mediated Killing:
suggesting a continuation of complement-medi- Complement
ated bactericidal or bacteriostatic activity after Evidence that serum-mediated killing of gram-
addition of broth. negative bacteria via the classical pathway nor-
The necessity of screening large numbers of mally requires all nine complement components
bacterial cultures or clones has led to the devel- of that pathway was provided by Inoue et al.
opment of a number of rapid assays for serum (134). They reacted E. coli with guinea pig serum
bactericidal activity. Plaque assays (77, 212, that had been depleted of the classical third
52 TAYLOR
complement component, thus forming a com-
plex of bacterial cells, antibody to an unidenti-
fied bacterial surface antigen, and the first three
components of the classical pathway (BAC142
cells).
Functionally pure C3, C5, C6, C7, C8, and C9
preparations were then added in various combi-
nations to the sensitized bacteria; only when all
components were present in the reaction mix-
ture could lysozyme-mediated bacteriolysis be
demonstrated. Goldman and co-workers (98)
studied the antibody and complement require-
ments for the killing of an E. coli K-12 strain. An
antigen-antibody interaction involving IgM mol-
ecules was necessary for classical pathway-
mediated killing of this rough strain. A require-
ment for Cl, C4, C2, and C9 was directly
demonstrated; addition of Cl-deficient human
serum to BAC1 cells, generated with functional-
ly pure human Cl, resulted in killing of the
sensitized cells. BAC142 cells were killed after
addition of a human C3 through C9 source, and
BAC1 cells were only killed when a human C4
through C8 source and functionally pure human
C9 were present; omitting C9 resulted in surviv-
al of all bacterial cells. These workers also found
that treatment of antibody-sensitized bacteria
with high concentrations of functionally pure
guinea pig or human Cl rendered the organisms
resistant to killing by whole sera; this effect was
partially dependent on the amount of antibody
used to sensitize bacteria. Such unpredictable
effects may account for subsequent failures to
generate bactericidal systems completely from
functionally pure classical pathway compo-
nents.
That gram-negative bacteria can also be killed
by the alternative pathway of complement acti-
vation was suggested by the observation that
C4-deficient guinea pig serum possessed bacteri-
cidal activity against E. coli B and C strains
(256); C4 is essential for classical but not for
alternative pathway function (Fig. 2). These
strains were, however, killed significantly more
slowly by C4-deficient than by normal guinea pig
serum, and killing by C4-deficient serum was
preceded by a lag phase, during which there was
no change in the number of viable organisms in
the incubation mixture. Using proteins purified
to homogeneity (269) and at concentrations cor-
responding to those found in serum, Schreiber et
al. (268) have shown that antibody-independent
bactericidal activity can be generated from the
11 components (Fig. 2) of the alternative path-
way. Comparable kinetics of activation by E.
coli K-12 W1485 in the isolated component
mixture and in C4-depleted serum were ob-
served; lysis but not killing was dependent on
the presence of lysozyme. Properdin (factor P)
was found to be a nonessential component for
MICROBIOL. REV.
bactericidal and bacteriolytic activity, but its
presence in physiological amounts resulted in a
threefold increase in activity. Killing was depen-
dent on the presence of C9.
These studies strongly suggest that deposition
of the assembled proteins of the MAC onto the
surfaces of susceptible bacteria is responsible
for serum-mediated killing and is a necessary
prerequisite for lysozyme-mediated bacterioly-
sis. Depending on the bacterial strain selected
for investigation, generation of the MAC may
result from either classical or alternative path-
way activation; indeed, both pathways may be
activated simultaneously (242, 282, 306). Only
the five terminal components, C5 through C9,
are directly involved in the assembly of the
macromolecular protein complex on the target
membrane (152, 315) after proteolytic activation
of C5 to C5b. Assembly of the MAC involves the
transformation of hydrophilic molecules into a
macromolecular amphiphilic complex capable of
integration into lipid-containing bilayers (21, 23).
The complex is a short, hollow cylinder 150 to
160 A in length with an internal diameter of 100
A and is rimmed by an annulus of an external
diameter of 200 to 220 A at one end (Fig. 3). The
other terminus bears an apolar surface 40 A in
length that inserts into the hydrophobic mem-
brane interior; after insertion, the MAC behaves
as an integral membrane protein complex, as it
resists elution from the membrane by ionic ma-
nipulations, binds nonionic detergents, and can
be reincorporated into artificial lipid vesicles
(21). The major part of the vertically oriented
cylinder, including the annulus, is located exteri-
or to the lipid bilayer, and it has been proposed
that the MAC penetrates through the membrane,
forming a channel (319) that destroys the integri-
ty of the attached membrane; channel formation
would seem likely, as the terminal complement
components produce large increases in ion per-
meability across planar lipid bilayers (171). The
MAC appears, therefore, to be identical to the
classical complement-induced lesions found on
sheep erythrocyte (126) and E. coli (95) mem-
branes. Available evidence suggests that the
MAC is a stable complex (molecular weight,
2,050,000) of the terminal complement proteins
and has the formula (C5b, C6, C7, C8, 6C9)2
(C5b-9)2 (237). Very recently, however,
Tschopp and Podack (238, 323) found that, un-
der certain conditions, C9 undergoes spontane-
ous polymerization to form a dodecamer with a
structure very similar or identical to the MAC.
This raises the possibility that the transmem-
brane channel may be poly(C9) and that C5b-8
functions as a poly(C9) generator complex by
reducing the activation energy of C9 polymeriza-
tion. If this is the case, C5b-8 must remain
attached to poly(C9) after channel formation, as
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA 53
C1Ag-Ab C1

L C4a
04 00
Mg + C4b,2a
a C4b,2a,C3b

classicalpathway -C
2b A3b C6
C98
C 5a
CC55b XC 5b-9
alternative pathway
A
C3 - C3b

VC 3 C3b C3b,CBb
C3b Bb C 3b,P,Bb,C 3b

C 3 (20) ,D C 3 (H20) Bb

FIG. 2. Schematic representation of the classical and alternative pathways of complement activation. The
classical pathway is usually initiated by an interaction between antigen-antibody complexes involving IgM or IgG
and Cl; this complement component is composed of three distinct subunits, Clq, Clr, and Cls, and circulates as
aCa2+-dependent complex of one Clq, two Clr, and two Cls molecules. The Clq subunit combines with the Fc
portion of IgM or IgG and is known to have six binding sites for immunoglobulin. Critical conformational
changes in Clq lead to conversion of the proenzymes Clr and C1s to Clr and C1s, molecules with serine protease
activity (243). The cleavage of Cls to Cls is catalyzed by Clf and results in formation of two peptide chains
(molecular weights, 56,000 and 30,000) held together by a disulphide bond. Cls then cleaves C4 (molecular
weight, 204,000) and C2 (molecular weight, 98,000) to generate the C3-cleaving enzyme C4b,2a. Fluid-phase C4
is cleaved into two fragments, C4a and C4b; the larger fragment, C4b, may bind covalently to the activating
surface, but C4a is released into the fluid phase and takes no further part in the activation of subsequent
components. C2 is cleaved by Clsinto two fragments, C2a and C2b; C2 cleavage is inefficient unless native C2 is
bound to C4b in the presence of Mg2+. The larger fragment, C2a (molecular weight, 72,000), remains bound to
C4b, and this complex constitutes the "C3 convertase." Cls can generate a large number of C4b,2a complexes.
C4b,2a cleaves C3 (molecular weight, 190,000) to form two fragments, C3a and C3b (molecular weight, 181,000);
the enzymatic site for C3 cleavage resides on the C2a moiety ofC3 convertase. C4b binds C3 in a way that makes
it susceptible to enzymatic attack. C3a is an anaphylatoxin and is released, whereas C3b binds to the activating
surface through a short-lived reactive carboxyl group; when this happens in the vicinity of a C4b,2a complex, a
new complex, C4b,2a,3b, is generated and constitutes the "C5 convertase." The binding ofC3b to surfaces is
inefficient, and many C3b molecules are released into the fluid phase. C4b,2a,3b then cleaves C5 (molecular
weight, 206,000) into the small fragment CSa, a chemotactic factor, and CSb. C5 cleavage initiates a self-
assembling process, the membrane attack pathway, which results in the formation of a stable transmembrane
complex involving CSb, C6 (molecular weight, 128,000), C7 (121,000), C8 (153,000), and C9 (79,000). The
alternative pathway is activated when factor D (molecular weight, 24,000), a serine protease present in active
form in blood, cleaves molecules of factor B (molecular weight, 100,000) that are transiently associated with C3
in the presence of Mg2+. In this way, small amounts of a C3-cleaving enzyme, C3,Bb, are formed, which
producesC3b in normal serum or blood at a low rate.C3b can associate transiently with B, which when cleaved
by D yields C3b,Bb, also aC3-cleaving enzyme. These two enzyme complexes are unstable and the subunits
spontaneously dissociate; C3b,Bb may complex with factor P (molecular weight, 184,000; previously known as
properdin) in the presence of Mg2+ to form a stable C3 convertase which may then bind an additional C3b
fragment to be converted to C3b,P,Bb,C3b, the C5 convertase of the alternative pathway. C3 and C5 cleavage is
catalyzed by the Bb fragment, an unusual serine protease. Under normal physiological conditions, the pathway is

P1H P1H
subject to strong negative control by (molecular weight, 150,000) and the enzyme C3bINA (C3b inactivator;
molecular weight, 88,000). binds stoichiometrically toC3b and prevents its interaction with other proteins,
whereas C3bINA splits C3b in the presence of,1H to prevent formation of C3b,Bb. However, on certain
bacterial surfaces, C3b,Bb,P can be formed in substantial amounts because the conformation of certain surface
structures protects the complex against P1H- andC3bINA-mediated negative control.

the terminal (C5b-9) components are complexed
to each other after their isolation from mem-
branes by Triton X-100 solubilization (22).
Evidence that polymerized CSb-8 may also be
capable of mediating bacterial killing has come
from studies of C9-deficient serum. Although, in
some cases, serum bactericidal activity may
require the presence of physiological amounts of
C9 (98, 268), this is clearly not always so. Serum
from a patient with a selective, total deficiency
of C9 possessed bactericidal activity against a
clinical isolate of E. coli, although killing oc-
curred at a rate of about 3% of that observed in
normal serum (160). Normal serum depleted
54 TAYLOR MICROBIOL. REV.

FIG. 3. Side (s) and axial (a) projections of C5b-9 MACs incorporated into artificial lipid vesicles. Vesicles
not carrying complexes (asterisks) do not contain stain (sodium silicotungstate). Scale bars represent 100 nm.
Electron micrographs courtesy of S. Bhakdi, Giessen, Federal Republic of Germany, and J. Tranum-Jensen,
Copenhagen, Denmark.

immunochemically of C9 and serum from C9- mented by an unidentified component of the

deficient patients killed strains of Neisseria gon-
orrhoeae and Neisseria meningitidis but at a
much slower rate than normal serum (115).
Membranes of killed bacteria lacked the typical
complement lesions, thus providing limited evi-
dence in favor of poly(C9) channel formation.
A number of workers have presented data
suggesting that the MAC may be formed as a
result of activation of mechanisms distinct from
either the classical or alternative pathways.
Steel and associates (287) reported a bactericidal
mechanism against V. cholerae that was activat-
ed by F(ab')2 fragments and was Ca2' depen-
dent and C4 independent; Ca2+ is required for
the functional assembly of the Clq, Clr, and Cls
subunits to give enzymatically active Cl esterase
(Fig. 2). Rough E. coli K-12 derivative 200P was
killed at very low efficiency by a mechanism
analogous to reactive lysis of erythrocytes; ex-
posure of unsensitized bacteria to the trimolecu-
lar complex C5b-7 in the presence of C8 and C9
produced a bactericidal effect that was enhanced
by pretreatment of cells with EDTA and aug-
pseudoglobulin fraction of human serum (97). It
is unlikely, however, that these pathways will be
of relevance to natural mechanisms of serum
killing.
Role of Antibody
The question as to the role of antibodies in the
initiation of the serum bactericidal reaction has
been the subject of long controversy (see, for
example, the review by Inoue [127]) that has
only recently been resolved owing to the realiza-
tion that both complement pathways can cause
deposition of MAC proteins onto the surface of
gram-negative bacteria and that the require-
ments for activation of these pathways are dif-
ferent. For example, Wardlaw (328) was unable
to show a requirement for antibody in the killing
of a rough strain, E. coli Lilly, by human serum.
sterzl et al. (289) found that serum obtained
from newborn piglets possessed bactericidal ac-
tivity against rough, but not smooth, strains of
Salmonella and E. coli; piglets do not receive
gamma globulins from the mother during the
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA 55
intrauterine period but only postnatally, via co- injection of Salmonella enteritidis polysaccha-
lostrum. These observations complemented ear- ride, more than 99% of the complement-activat-
lier work by Turk (324), showing that precolos- ing activity was associated with the IgM fraction
tral calf serum possessed high bactericidal of rabbit serum. Robbins et al. (253) purified to
activity against rough enterobacteria, and he homogeneity IgM and IgG anti-S. typhimurium
implicated the alternative pathway, then known antibodies from the sera of immunized rabbits;
as the properdin pathway, as the mediator of this they found IgM antibodies to be 18-fold more
activity. Other workers obtained apparently active in bactericidal assays than IgG antibod-
conflicting data. Goldman et al. (98) found an ies. Very small amounts (6 to 7 ng) of purified
absolute requirement for IgM in the killing of E. IgM from rabbits immunized with Pseudomonas
coli 200P, and Rowley (260) observed that a aeruginosa were found to be effective in bacteri-
series of rough mutants of S. minnesota required cidal assays with the homologous strain, where-
the presence of antibody directed against sur- as little or no bactericidal activity was observed
face lipopolysaccharide determinants for killing with as much as 40 ,ug of IgG (28). The molar
to take place. Sheep serum could not kill E. coli ratio of the bactericidal activities of IgM to IgG
Lilly after absorption with the homologous was greater than 1:250,000. When compared on
strain (194). It is now clear that these differences a molar basis, IgM purified from rabbits immu-
can be explained on the basis of differences in nized with S. typhimurium 0 antigen was ap-
methodology, particularly those relating to se- proximately 250 times as active as IgG in a
rum concentration and to the efficiency of the complement-dependent bactericidal assay (270).
bactericidal system employed, as rough strains Although these and other (116, 117) studies have
are known to be susceptible to both antibody- emphasized the superiority of IgM antibodies
dependent classical pathway activity and anti- over IgG as activators of the classical pathway
body-independent alternative pathway activity in the bactericidal reaction, there are large dif-
(26, 242, 306). ferences in the relative activities of the two
In the serum bactericidal reaction, initiation of antibody classes. It is known, however, that
the classical pathway generally results from for- during the development of the immune response
mation of an antigen-antibody complex at or there is a progressive increase in the association
near sites on the bacterial surface that can constant of the antigen-antibody bond (64, 112)
interact with activated molecules of the comple- and hence a concomitant increase in the biologi-
ment cascade. Extremely small amounts of anti- cal activities of antibodies produced during this
body are required for complement activation; period; differences in avidity of antibodies in
for example, Michael et al. (188) estimated that these studies are likely, therefore, to be a reflec-
10 ng of antibody per ml can be demonstrated by tion of the adoption of differing immunization
a system possessing bactericidal activity via the schedules and animal species for production of
classical pathway. Sterzl and co-workers (289) antibodies. Thus, Schulkind and co-workers
found the bactericidal reaction to be 2,500 to (270) demonstrated a 5-fold increase in the spe-
10,000 times more sensitive as a means of anti- cific bactericidal activity of IgM antibodies and a
body determination than the agglutination reac- 12-fold increase for IgG antibodies during devel-
tion. The bactericidal reaction has, therefore, opment of the immune response over a 70- to 80-
been used as a procedure for the diagnosis of day period. Antibodies of a single immunoglob-
diseases such as typhoid fever (108) and syphilis ulin class were further fractionated into
(170). populations of molecules with different activities
There is wide variation in the efficiency with in bactericidal assays, indicating that at any one
which different antibody classes can function in time during the course of immunization the
complement-mediated killing. Although natural serum represents a pool of antibodies specific
antibodies in serum may belong to the IgM, IgA, for a given antigen. Similar variations in affini-
or IgG classes (45), the predominant natural ties of antibodies during the immune response
antibodies to representative enterobacterial spe- were found by Pike and Chandler (235). These
cies are contained in the IgM fraction of human authors determined the efficiency of IgM and
plasma (187); the IgM fraction possessed 140- IgG produced in rabbits inoculated with live V.
fold more activity against S. typhi and 560-fold cholerae at intervals over a period of almost 1
more activity against E. coli than the IgG frac- year. The activity of IgM in serum bactericidal
tion on a weight basis (187). The bactericidal assays remained relatively constant over this
activity of cord serum from newborn infants was period; this contrasted with IgG activity, which
found to be low in comparison to the activity of increased markedly during the immunization pe-
the serum of the mothers (91); neonatal serum is riod. These differences could be due to varia-
markedly deficient in IgM, as only IgG molecules tions in the relative amounts of IgG subclasses in
are transferred across the placental barrier. Sim- the IgG pools, as some subclasses, such as
ilarly, Landy et al. (153) found that, after a single hunman IgG3 and IgGl, are known to activate
56 TAYLOR
complement efficiently, whereas others are vir-
tually inactive in this respect (151).
Early studies with purified specific excretory
IgA suggested that this antibody could activate
bacteriolytic mechanisms in the presence of
complement and lysozyme (3). Subsequent in-
vestigations have failed to confirm this finding
(63, 116, 117), and it has been suggested that the
IgA fractions examined by Adinolfi et al. (3) may
have contained small amounts of contaminating
active immunoglobulins of other classes (63,
261). More recent reports have, however, pro-
vided evidence that IgA may activate the alter-
native pathway in bactericidal reactions against
E. coli (281). There is clearly a need for clarifica-
tion of the role of secretory antibodies in serum
bactericidal reactions in the light of recent ad-
vances in the molecular biology of complement.
In classical pathway activation, IgM and cer-
tain IgG antibodies interact with the Clq compo-
nent of the Cl complex; Clq is a 400,000-
molecular-weight glycoprotein formed from 18
peptide chains. Electron microscopy studies
(273, 295) have revealed that Clq has an unusual
ultrastructure: six peripheral globular moieties
are attached by fibril-like strands to a central
core. Clq has six binding sites for the Fc portion
of immunoglobulins, and these are situated on
the six globular moieties (151). At least two of
the six valencies must bind to immunoglobulins
before the next stage in the complement se-
quence can take place. Studies of antibody-
mediated erythrocyte lysis have indicated that
one IgM molecule is sufficient to combine with
at least two Clq binding sites, whereas the
reaction with IgG requires that two IgG mole-
cules be present in close proximity (30). As
recently stressed by Atkinson and Frank (13),
the attachment of one molecule of IgG to an
antigenic surface does not influence attachment
of other molecules, so a very large number of
attached IgG molecules may be necessary for
the required proximity to be achieved. It thus
becomes easy to understand the relative effi-
ciencies of IgM and IgG as mediators of bacteri-
cidal phenomena.
Antibodies mediating bactericidal activity
combine with antigenic determinants on or very
near the bacterial surface so that complement
may interact with lipid membrane components.
For this reason, antibodies directed against fla-
gellar (H) determinants are ineffective (10).
There are a large number of studies demonstrat-
ing directly and indirectly that both normal and
immune sera contain complement-activating
antibodies specific for antigenic determinants
carried by lipopolysaccharide. Schwab and
Reeves (271) examined sera from a number of
species, including humans, pigs, kangaroos, liz-
ards, fowls, and toads, and found natural anti-
MICROBIOL. REV.
bodies directed against heat-stable cell surface
antigens of E. coli and a number of Salmonella
sp. Absorption of human serum containing anti-
S. typhi antibodies with a heat-killed suspension
of S. typhi resulted in loss of bactericidal activity
against S. typhi and other strains with cross-
reacting 0 antigens, but not against salmonellae
with unrelated 0 antigens (203), suggesting that
antibodies were specific for the 0 side-chain
moiety of lipopolysaccharide. Glynn and Ward
(96) showed that normal and immune sera lose
bactericidal activity against gonococci after ab-
sorption with erythrocytes coated with lipopoly-
saccharide, an observation confirmed by Tra-
mont and co-workers, using hyperimmune
rabbit antisera (318). Immunization with lipo-
polysaccharide gives rise to complement-acti-
vating antibody against homologous Salmonella
sp. (50, 270, 333) and E. coli (282, 333). Lipo-
polysaccharide from a gram-negative strain in-
hibits the bactericidal activity of both normal
and immune sera against the homologous strain
(10, 50, 189, 282), but such experiments should
be treated with caution, as lipopolysaccharides
may activate complement pathways and deplete
serum of complement in the absence of antibody
(84, 88, 199). In addition, RNA, a frequent
contaminant of lipopolysaccharide and endotox-
in preparations (336), is able to act as a nonspe-
cific inhibitor of serum bactericidal reactions
(10). Antibody for classical pathway activation
may be directed against antigenic determinants
on the core moiety of lipopolysaccharide (39,
260) and against enterobacterial common anti-
gen (59); antibody against outer membrane pro-
teins of N. gonorrhoeae and N. meningitidis
also functions in serum bactericidal systems
(318). Similarly, antibodies raised against the E.
coli receptor protein for phage A participate in
the killing of lamB+ E. coli K-12 populations
(83).
Comparatively little work has been done on
the role of anticapsular antibodies in the serum
bactericidal reaction. Antibodies to the exopoly-
saccharide of mucoid strains of P. aeruginosa
(28) and to the group-specific capsular polysac-
charides of N. meningitidis (100) are known to
activate complement in the bactericidal reac-
tion.
There is, therefore, a large body of evidence
indicating that, during the bactericidal reaction,
the classical pathway is dependent on Cl activa-
tion by antigen-antibody complexes. The lipid A
moiety of lipopolysaccharide is known to acti-
vate Cl in the absence of antibodies (199), but
this region occupies a position deep in the outer
membrane of gram-negative bacteria (338) and
would not normally be exposed on the surface of
cells; it is unlikely, therefore, to function in situ
as a Cl activator. Loos and colleagues (161)
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
demonstrated that serum-susceptible Klebsiella
pneumoniae and E. coli could bind the activated
form of Cl in the absence of detectable IgM (<2
,ug/ml) and IgG (<1 ,ug/ml). Binding of native Cl
from absorbed normal human serum was also
demonstrated, but the possibility exists that
small amountsofIgM might remain after absorp-
tion. The question of whether or not antibody-
independent Cl or CT binding leads to demon-
strable bactericidal activity has been recently
investigated by Betz and Isleker (19, 20). They
exposed E. coli J5, which is deficient in galac-
tose-4-epimerase, to radiolabeled Cl; Cl was
bound to cells in a dose-dependent manner and
essentially all bound Cl was subsequently acti-
vated. The complex consumed purified C4 and
C2 to an extent compatible with classical path-
way activity. Addition of C3-9 resulted in a 90%
decrease in bacterial viability, demonstrating
that antibody-independent activation of the clas-
sical pathway may contribute to the killing of
rough gram-negative bacteria. However, the
smooth parent strain, E. coli 0111:B4, from
which the mutant J5 was derived (177), was not
killed by this mechanism, even though it is
known to be susceptible to serum (206). This
suggests that antibody-independent classical
pathway activity may be more important in the
killing of rough rather than smooth strains.
Preincubation of cells with hyperimmune anti-E.
coli J5 IgG or IgM resulted in increased Cl
binding, although consumption of later compo-
nents was comparable to that observed with
antibody-independent activation. These studies
demonstrate unequivocally that the classical
pathway may be activated in the complete ab-
sence of antibody and that this system is rele-
vant to the killing of rough gram-negative bacte-
ria. Its relative importance in comparison with
other more established mechanisms of activa-
tion awaits further investigation.
The alternative pathway may be activated by
insoluble immune complexes (81, 281), but this
mode of initiation is likely to be of only minor
importance in serum bactericidal reactions
(151). Early evidence for an antibody-indepen-
dent bactericidal mechanism distinct from clas-
sical pathway activation was provided by Ward-
law and Pillemer (331, 332), whose discovery of
a bactericidal system requiring properdin pro-
vided the impetus that led to the elucidation of
the alternative pathway. It is now clear that
subsequent demonstrations (289, 296, 324, 328)
of bactericidal activity in the absence of anti-
body are likely to be manifestations of alterna-
tive pathway activity.
Role of Lysozyme
Although susceptible gram-negative bacteria
are killed by antibody-complement systems,
57
they lyse to a significant extent only in the
presence of lysozyme (132). Lysozyme is pre-
sent in blood and serum at a concentration in the
range of 1.54 to 9.74 ,ug/ml (mean, 5.64 pg/ml;
80). Bacteriolytic activity is negligible when the
enzyme is removed from serum by adsorption
onto bentonite (132, 328, 342) or by neutraliza-
tion with anti-human lysozyme antiserum (95,
167, 306). Bacteria treated with lysozyme-free
serum retain their rod shape but are converted to
spheroplasts after subsequent addition of lyso-
zyme in physiological amounts; these sphero-
plasts will eventually lose their integrity and lyse
unless protected from osmotic forces by hyper-
tonic solutions (48, 70, 86, 342). The rate and
extent of bacteriolysis is directly related to the
lysozyme concentration (95, 132). Pretreatment
of bacteria with lysozyme has no effect on the
rate of lysis of serum-exposed cells (9), suggest-
ing that lysozyme gains access to its peptidogly-
can substrate only after formation of a functional
C5b-9 complex has disrupted the integrity of the
outer membrane (92, 134).
The role of lysozyme in the bactericidal reac-
tion has, in contrast, been the subject of some
controversy. Serum killing occurs in the com-
plete absence of lysozyme (134, 268), and it has
therefore been asserted that the enzyme has no
effect on the rate of serum killing by complement
(86, 167, 185). Other workers have, however,
produced data that tend to contradict these
findings. Inoue and co-workers (132) removed
all detectable lysozyme from rabbit antiserum
and guinea pig complement and found that a
mixture of these reagents possessed much re-
duced bacteriolytic but still powerful bactericid-
al activity against E. coli B. The addition of 5 ,ug
of egg white lysozyme per ml to this system both
enhanced and accelerated the bactericidal effect.
Glynn and Milne (95) found that neutralization
of lysozyme activity with antilysozyme antiser-
um or removal by bentonite adsorption resulted
in reduced rates of serum killing of both a rough
and a smooth isolate of E. coli. The addition of
egg white lysozyme restored killing of the rough
strain by both sera and of the smooth strain by
antiserum-treated serum. Lysozyme addition
did not, however, completely restore the activi-
ty of bentonite-adsorbed serum to normal levels,
and Glynn and Milne suggested that bentonite
adsorption, in addition to removing lysozyme
from serum, had also resulted in the loss of a
factor, distinct from antibody and complement,
that was necessary for maximal rates of killing of
the smooth strain. Removal of lysozyme from
human serum by bentonite adsorption resulted
in reduced killing which could be completely
restored by the addition of egg white lysozyme
(70, 98). Taylor and Kroll (306) recently found
that removal of all detectable lysozyme activity
58 TAYLOR
from human serum by bentonite adsorption or
by neutralization with purified anti-human lyso-
zyme IgG resulted in a relatively small but
reproducible reduction in the rate of killing of a
rough, encapsulated E. coli strain. Optimal ac-
tivity could be restored by the addition of phys-
iological amounts of egg white lysozyme, and
the addition of larger quantities of the enzyme
further increased the rate of killing. The addition
of various quantities of lysozyme to unabsorbed
serum also resulted in increased killing kinetics.
The reasons for these conflicting findings are
unclear but may be related to methodological
differences and differing requirements for the
killing of various strains of gram-negative bacte-
ria. If the differences in the rate of killing be-
tween normal and lysozyme-free sera are small,
then great care is needed in constructing re-
sponse curves. In many of the cases in which the
influence of lysozyme on serum killing has been
dismissed as negligible or nonexistent, insuffi-
cient sampling over the defined time period has
been carried out. Also, it has been common
practice to remove lysozyme activity by benton-
ite treatment of serum, but this procedure is
known to remove gram-positive bactericidins
(211) that may possess activity against gram-
negative species (35, 61), as well as factors
essential for alternative pathway activity (306).
Addition of antilysozyme antibodies could, in
theory, result in complement deviation and loss
as a result of complement activation by antigen-
antibody complexes. One possible explanation
of the enhancing effect of lysozyme is that
depolymerization of peptidoglycan removes a
barrier to the rapid assembly of the C5b-9 com-
plex at the inner membrane, as it is likely that
the primary lesion responsible for the bactericid-
al effect occurs at this site.
Additional Factors
The alternative pathway has been entirely
reconstructed from chemically pure components
(268), and there is good evidence that classical
pathway killing requires only Cl through C9 and
some activating mechanism that usually in-
volves antibodies of the IgM or IgG class (98,
134). There are a number of reports in the
literature, however; suggesting that the presence
of additional factors in serum is necessary for
maximal rates of killing to occur. For example,
Skarnes (282) examined the bactericidal activity
of rabbit and guinea pig sera against a number of
rough and smooth enterobacterial strains and
found that complement killing of rough strains
occurred in the absence of antibody and re-
quired an additional nonspecific component that
was destroyed by freezer storage and by heating
MICROBIOL. REV.
at 52°C. Smooth organisms were killed by com-
plement in the presence of antibody, but the
presence of the labile component was also re-
quired. Killing of rough strains presumably oc-
curred after antibody-independent alternative
pathway activation, whereas killing of smooth
strains required classical pathway activation.
The author speculated that the labile component
might be an alternative pathway component or
regulator, and this raises the interesting possibil-
ity that serum killing of some strains of gram-
negative bacteria requires the participation of
both pathways, or at least one pathway and a
limited number of components of the other. That
both pathways may be activated simultaneously
is suggested by a number of studies (102, 242,
306), but the relevance of any interaction or
synergy between the two in the serum bacteri-
cidal reaction is an area that has been almost
completely ignored. Similar effects have been
noted after zymosan adsorption of serum (306).
Treatment of serum with this yeast cell wall
preparation at 17°C is known to remove proper-
din (331) and also part of the lysozyme activity
(210), but not to activate the complement cas-
cade. Zymosan-treated serum was found to pos-
sess no bactericidal activity against a rough,
encapsulated E. coli strain that was susceptible
to both classical pathway- and alternative path-
way-mediated killing, even though the serum
retained a functional classical pathway as judged
by its high lytic activity against sensitized eryth-
rocytes. Addition of an antibody source to the
system failed to restore bactericidal activity, and
the authors (306) speculated that either proper-
din or another alternative pathway component is
essential for classical pathway killing of the E.
coli strain or that zymosan removed a previously
unrecognized factor essential for classical path-
way killing. As classical pathway-mediated kill-
ing has yet to be demonstrated by assembly of
chemically pure classical components, the in-
triguing possibility remains that other compo-
nents in addition to Cl through C9 and antibody
are needed. Glynn and Milne (95) found that the
reduction in the rate of serum killing resulting
from neutralization of lysozyme by antilyso-
zyme antisera could be completely restored by
the addition of egg white lysozyme, whereas loss
of full activity after bentonite adsorption could
not. The bentonite-adsorbable factor essential
for full serum bactericidal activity has, there-
fore, properties similar to those of other en-
hancers of serum killing.
Serum from a variety of animal species con-
tains relatively low-molecular-weight cationic
proteins that have been collectively termed ,B-
lysins; the proteins possess lethal activity
against gram-positive bacterial types and appear
to be released from platelets during coagulation
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
of blood (121). Unless special care is taken,
therefore, these substances are likely to be pres-
ent in variable amounts in human and animal
sera utilized for bactericidal tests. The primary
site of action of I3-lysins is considered to be the
cytoplasmic membrane (60), and consequently,
these proteins have no direct lethal action
against gram-negative bacteria possessing an
integral envelope structure. Removal of 3-lysins
from rabbit serum, however, was shown by
Donaldson and co-workers (61) to result in de-
creased bactericidal activity against E. coli B
cells. Highest rates of killing were obtained
when 3-lysins, lysozyme, and the antibody-com-
plement system functioned in a cooperative
fashion. It would appear, therefore, that disrup-
tion of outer membrane integrity by the MAC
provides the means whereby P-lysins may reach
their target sites at the cytoplasmic membrane.
Recently, Carroll and Martinez (35, 36) isolat-
ed and characterized the major heat-stable bac-
tericide from rabbit serum. They demonstrated
that the primary bactericidal activity of normal
rabbit serum against Bacillus subtilis resides in a
single, 1,800- to 2,000-molecular-weight cationic
polypeptide which they designated PC-Ill (35,
36). They found that both E. coli ML35, a K-12
derivative, and an S. typhimurium strain were
rapidly killed after exposure to 80 to 200 ng of
PC-Ill per ml. Although gram-positive bacteria
were significantly more susceptible, PC-Ill con-
centrations in normal rabbit serum were found
to be about 1 ,ug/ml, and it is therefore likely that
antibacterial activity against gram-negative
strains would manifest itself in relatively undi-
luted serum. Any possible interrelationship be-
tween direct or indirect ,B-lysin PC-Ill activity
and the antibody-complement system awaits fur-
ther elucidation.
Sequence of Events After Exposure to Serum
Exposure of susceptible gram-negative cells
to serum is generally followed by a period during
which there is little apparent change in viability;
the length of this period varies depending on the
bacterial strain used, the test system, the serum
concentration, and the metabolic state of the
bacterial cells. For example, early-log-phase E.
coli O1l1:B4 or 0127:B8 cells grown and tested
at 37°C in the system of Davis and Wedgwood
(54) began to lose viability 6 to 12 min after
exposure to human serum, whereas Pasteurella
septica growing in undiluted horse serum sur-
vived in an identical fashion to control cells for
30 to 40 min after addition of specific antiserum
(105). Rough enterobacteria, including E. coli
Lilly, B, and K-12 strains, appear to be extreme-
ly susceptible to serum, and the lag phase may
59
be so short with these organisms as to be virtual-
ly undetectable (146, 167). Consequently, the
availability of an enormous amount of data relat-
ing to the genetics and metabolism of K-12
derivatives, as well as a wide range of well-
characterized substrains and mutants, should be
carefully weighed against the disadvantage that
the extreme susceptibility of K-12 to serum
provides only a limited opportunity to discern
early events in the bactericidal reaction that
must precede any reduction in viability.
Although direct evidence is lacking, it would
seem reasonable to assume that during this
latent period the components of the MAC are
under assembly at the site of the primary lesion.
This period may correspond to the first stage of
complement attack on E. coli, as demonstrated
by Michael and Braun (185), who found that
cells exposed to slightly diluted serum and diva-
lent cations for 7 min and then washed could be
efficiently killed by highly diluted serum in the
absence of cations. No killing could be detected
during the initial period of exposure, and highly
diluted serum had no bactericidal activity. Se-
lective removal of the late-acting complement
components from the second-stage treatment
resulted in complete loss of activity. Therefore,
once functional lesions have been assembled,
loss of viability rapidly ensues, and killing pro-
ceeds in a logarithmic fashion as reflected in
steep killing kinetics of susceptible organisms
(54, 221, 306). Rapid killing of bacteria suggests
that the bactericidal mechanism has a high de-
gree of efficiency. It has been well established
by the classic studies of Mayer (170) that lysis of
sensitized erythrocytes by complement follows
one-hit killing kinetics. The gram-negative cell
obviously presents a more complex target for
complement attack than the erythrocyte; bacte-
ria have a more complex envelope structure,
they can respond rapidly to changes in the
environment, and they have a capacity for repair
of damaged sites. Nevertheless, Inoue et al.
(128) developed an analytical method for deter-
mination of the number of lesions necessary for
killing of E. coli B that involved exposure of
increasing concentrations of antibody-sensitized
bacteria to a constant amount of complement;
their data was comparable to theoretical curves
based on one-hit killing kinetics. However,
some theoretical aspects of these estimations
have been criticized by Wright and Levine (343),
who reexamined dose-response kinetics for kill-
ing of E. coli by using human serum lacking
either C7 or C8 and then adding back, in analogy
to the hemolysis experiments of Mayer, limiting
amounts of the missing components. They found
that killing of E. coli, as well as complement-
mediated inner and outer membrane damage,
required more than one hit and that the lethal
60 TAYLOR
process closely approximated two-hit killing ki-
netics. The authors suggested that this reflected
the dimeric nature of the MAC, but it does not
preclude the possibility that damage to inner and
outer membranes are independent events.
Serum treatment of E. coli cells results in
rapid and often total release of enzymes, such as
alkaline phosphatase and 5'-nucleotidase, that
are associated with the periplasmic space (53,
71, 130, 306, 342); release is dependent on
activation of the terminal complement attack
sequence and formation of a functional MAC
(130, 342). Integration of the MAC into phospho-
lipid bilayer areas of the outer membrane there-
fore disrupts outer membrane integrity and re-
moves the barrier to hydrolysis of peptidoglycan
by lysozyme (92, 130, 167). It is unlikely, how-
ever, that damage confined to the outer mem-
brane is responsible for the lethal activity of
complement. For example, Feingold and co-
workers (70) examined the effect of lysozyme-
free human serum on E. coli K-12 cells in the
presence of 0.6 M sucrose; cells plasmolyzed in
this fashion were found to be refractory to the
serum bactericidal system. Bacteria grown for
extended periods of time in the presence of
sucrose did not undergo plasmolysis when re-
suspended in hypertonic solutions and were
susceptible to serum. The authors therefore sug-
gested that cells were killed after complement
damage to the cytoplasmic membrane and that
plasmolyzed cells escape this damage due to
retraction of the cytoplasmic membrane. Fur-
thermore, plasmolyzed cells promptly released
alkaline phosphatase (71), demonstrating the es-
sentially nonlethal nature of outer membrane
damage. These experiments have been criticized
on the grounds that the high concentrations of
sucrose employed would inhibit bactericidal ac-
tion (127, 205), yet the data presented by Fein-
gold et al. (70) demonstrate a high degree of
bactericidal activity in the presence of sucrose.
These studies (70, 71) also provided evidence
that complement damage to the cytoplasmic
membrane results in irreversible loss of viabili-
ty. The K-12 strain used, 200P, is inducible for
the cytoplasmic enzyme P-galactosidase but
lacks a transport system for P-galactosides. The
loss of the permeability barrier to o-nitrophenyl-
P-D-galactopyranoside paralleled the loss of via-
bility of plasmolyzed and nonplasmolyzed cells
after exposure to lysozyme-free serum, whereas
serum-mediated periplasmic enzyme release oc-
curred readily in the absence of lethal effects.
Loss of P-galactoside crypticity after serum
treatment has subsequently been observed by
Martinez and Carroll (167) and by Wright and
Levine (342) and found to be strongly correlated
with cell death. Serum treatment also causes
other perturbations that suggest damage to the
MICROBIOL. REV.
cytoplasmic membrane. For example, active
transport of sugars, amino acids, and 86Rb+ (a
K+ analog) are inhibited by serum treatment
(53). Exposure to lysozyme-free serum has also
been reported to cause a rapid efflux of 86Rb+
from preloaded cells (167, 342), suggesting that
serum may function by dissipating the electrical
potential of the inner membrane. We have re-
cently found, however, that significant amounts
of 86Rb+ may be released by heat-inactivated
serum and even by some buffers (306) and that
serum-resistant strains of E. coli may release as
much preloaded 86Rb+ as susceptible strains
when exposed to activated complement compo-
nents (P. W. Taylor and H. P. Kroll, unpub-
lished data). It would, therefore, seem pertinent
to treat data of serum-mediated 86Rb+ release
with caution, particularly when inadequate in-
formation regarding nonspecific marker release
is presented.
Complement-mediated damage to the inner
membrane is likely to be limited and not involve
gross disruption, as bacterial respiration remains
relatively unaffected until late in the reaction
sequence and may even be transiently stimulat-
ed (10, 272). However, it then becomes difficult
to explain how complement can cause lethal
damage to the cytoplasmic membrane after de-
position onto and insertion into the outer mem-
brane. Some of the possibilities have been dis-
cussed by Feingold et al. (71) and more recently
by Wright and Levine (342). The limited size of
the hydrophobic domain of the MAC (21) makes
it unlikely that a transmembrane channel could
be inserted across the double membrane system
of gram-negative bacteria. One possibility is that
the cytoplasmic and outer membranes are dam-
aged by two separate complement-mediated re-
actions that occur during the course of bacterial
killing. Insertion of the MAC into the outer
membrane is known to cause release of phos-
pholipid into the surrounding medium (129, 131,
133, 341), and extensive disassembly of the
outer membrane would allow subsequent MAC
insertion into the cytoplasmic membrane. Such
a mechanism would, however, appear to be
discounted by the observation of Wright and
Levine (342); they exposed E. coli cells to C8-
deficient serum, removed excess complement
components by thorough washing, and then
treated C5b-7 bacteria with purified C8 and C9.
Cytoplasmic membrane damage occurred in the
absence of any new C5b-7 deposition, thus dem-
onstrating that damage to the cytoplasmic mem-
brane is not a result of fresh complement activa-
tion on the cytoplasmic membrane after outer
membrane disruption.
Another possibility, considered by Feingold et
al. (71) and subsequently by others (167, 342), is
that effective lesions occur only when terminal
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
complexes are deposited on the bacterial surface
at the points of contact between the cytoplasmic
and outer membranes described and character-
ized by Bayer (16). Simultaneous damage would
then occur on both membranes. Evidence in
favor of this mechanism has been obtained by
Wright and Levine (342), who found that the
kinetics of release of periplasmic enzymes and
intracellular cations after exposure to lysozyme-
free serum are virtually identical, indicating that
cytoplasmic membrane damage and outer mem-
brane damage are closely coupled events. They
found, as have others (130, 306), that cytoplas-
mic marker release is selective; large intracellu-
lar molecules such as P-galactosidase (molecular
weight, 540,000) and thiogalactoside transacety-
lase (molecular weight, 63,500) are retained.
Insertion of the MAC at points of membrane
adhesion would also account for cytoplasmic
membrane damage of E. coli C5b-7 by purified
C8 and C9 in the absence of any new CSb-7
deposition (342) and for the fact that stationary-
phase cells, possessing few adhesion sites, are
relatively less susceptible to serum (55, 342). As
exponentially growing bacteria possess only 200
to 400 adhesion sites per cell (16), such a mecha-
nism suggests that many lesions formed on the
surface would not directly result in killing, mak-
ing the serum bactericidal system a relatively
inefficient phenomenon.
A resolution of the current controversy con-
cerning the molecular nature of the MAC in
favor of a poly(C9) cylindrical structure would
present another possibility for closely linked
cytoplasmic and outer membrane damage.
There are indications that complement-mediated
outer membrane damage differs from damage to
erythrocyte membranes and to bacterial cyto-
plasmic membranes. MAC lesions on erythro-
cyte membranes restrict the free diffusion of
small molecules such as sucrose and raffinose
(157, 280), whereas damage to the outer mem-
brane facilitates the rapid diffusion of macromol-
ecules through the membrane. This latter obser-
vation indicates disruption of the outer
membrane, as periplasmic enzymes are too large
to diffuse through the transmembrane MAC
channel. Extensive disruption is also suggested
by the frequently observed release of outer
membrane fragments, as lipopolysaccharide-
protein-phospholipid complexes, into the sur-
rounding environment after exposure to serum
(53, 70, 340). Thus, a detergent-like action rather
than ion-channel formation appears to be opera-
tive at the outer membrane. Erythrocytes (171)
and E. coli protoplasts (5) undergo slow C5b-8-
mediated lysis which parallels increases in ion
permeability of C5b-8-treated planar lipid bi-
layers; this process follows single-hit kinetics, in
contrast to membrane disintegration, which dis-
61
plays multihit characteristics (171). Poly(C9)
might, therefore, have evolved to effect deter-
gent-like disruption of the gram-negative outer
membrane to allow CSb-8-mediated cytoplasmic
membrane perturbation. Such a mechanism
would not necessarily require fresh deposition of
MAC proteins after outer membrane disruption,
as CSb-8 could be translocated onto active sites
at the cytoplasmic membrane subsequent to
poly(C9) generation. Analysis of the distribution
of the proteins of the membrane attack pathway
between cytoplasmic and outer membranes of
serum-treated bacteria would help resolve the
various proposed mechanisms, and such experi-
ments are presently being conducted.
Such a translocation step is suggested by the
observation that, in contrast to erythrocyte ly-
sis, the serum bactericidal reaction requires an
input of bacterially generated energy. Griffiths
(106) demonstrated that the lethal action of
antibody and complement against P. septica
could be prevented by addition of uncouplers or
inhibitors of oxidative phosphorylation 5 min
after initiation of the bactericidal reaction. Com-
plete inhibition of serum killing by an inhibitor
(cyanide) and an uncoupler (2,4-dinitrophenol)
was also found with an E. coli strain (306). E.
coli LP1092 cells were killed by human serum
after a lag period of about 10 min; if KCN was
added to the system 8 to 18 min after initiation of
the bactericidal reaction, the degree of inhibition
was much reduced. Exposure of LP1092 cells to
serum was followed by a rapid and large in-
crease in intracellular ATP levels; ATP synthe-
sis did not occur when bacteria were exposed to
dialyzed serum, which killed at a much reduced
rate. Addition of glucose or serum ultrafiltrate to
dialyzed serum restored optimal bactericidal ac-
tivity, suggesting that optimal killing of gram-
negative bacteria requires an input of bacterially
generated ATP. It is not known for which early
stage in the lethal process energy is required.
The precise nature of the lethal event after
interaction of MAC proteins with fluid bacterial
membranes (144) is unknown. A rapid efflux of
intracellular cations (167, 342) without inhibition
of respiration (10, 272) suggests that comple-
ment kills E. coli by dissipating the energized
state of the cytoplasmic membrane and that
MAC proteins form ion-permeable channels.
They would then act in an analogous fashion to
membrane-active colicins such as El, K, and Ia.
These colicins have been found to inhibit respi-
ration-linked active transport systems (74), to
lower intracellular ATP levels (74), to cause a
rapid efflux of intracellular K+ (335), and to
inhibit the energy-linked transhydrogenase reac-
tion (264) as a consequence of a rapid and drastic
reduction of the membrane potential (334). ATP
levels fall after colicin treatment because the
62 TAYLOR
bacteria attempt to maintain the concentration
gradient of K+ and possibly Mg2+ through the
expenditure of ATP. Serum-treated cells do not,
however, lose ATP, even when metabolizable
substrates have been removed from serum by
dialysis (306). The effect of MAC formation on
the electrochemical potential (A*) across the
cytoplasmic membrane has recently been evalu-
ated by measuring the uptake of the lipophilic
cation tetraphenylphosphonium and of proline
(A. F. Esser, Fed. Proc. 39:1755, 1980). Both
functions were inhibited by deposition of C5b-9,
but formation of C5b-8 with C9-deficient serum
had a similar effect. In the absence of C9 deposi-
tion, uptake and transport recovered but were
irreversibly inhibited by addition of C9. These
experiments illustrate that C5b-8 can transiently
perturb the cytoplasmic membrane but that dis-
sipation of Ai alone is insufficient to account for
the bactericidal action of serum.
Exposure to serum also results in an inhibition
of macromolecular biosynthesis, but in all cases
in which kinetics of precursor incorporation
have been followed, reduction in cell viability
was evident well before inhibition became ap-
parent. It is likely, therefore, that inhibition of
macromolecular biosynthesis occurs secondary
to the lethal event and merely reflects a running
down of cellular activity in cells that have al-
ready been rendered nonviable. For example,
loss of viability of E. coli O111:B4 cells began
almost immediately in the bactericidal system of
Melching and Vas (180), whereas RNA synthe-
sis was affected after 7 min, changing from an
increasing rate of synthesis to a constant rate of
synthesis. In contrast, the rate of DNA synthe-
sis was not affected until after 15 min. Total
RNA accumulation began to decrease after 15
min, and after 30 min much of the intracellular
RNA began to be lost. Inhibition of DNA accu-
mulation was not apparent until after 25 min,
and protein synthesis was affected even later in
the reaction sequence. Inhibition of RNA syn-
thesis followed by inhibition of protein and DNA
synthesis was also noted for antibody-comple-
ment-exposed P. septica (105) and for rough
strains of E. coli (167, 306). Martinez and Carroll
(167), however, noted significant rates of DNA
synthesis, even well after cell lysis had oc-
curred; the significance of this extended lysis
remains unclear. Until the point of inhibition,
mRNAs and their translated products remain
fully functional (167). Early suggestions (8) that
inhibition of protein synthesis is responsible for
loss of viability of serum-exposed bacteria ap-
pear, therefore, to be unfounded; indeed, a
certain amount of protein synthesis appears to
be essential for maximal rates of serum killing,
as inhibitors of protein synthesis reduce the rate
of viability loss (184, 306).
MICROBIOL. REV.
RESISTANCE TO SERUM BACTERICIDAL
ACTIVITY
Definition of Serum Resistance
Under conditions approaching the optimum,
many strains of gram-negative bacteria are rap-
idly killed by human and animal sera. Some
strains appear to be completely refractory to the
serum bactericidal system, and prolonged incu-
bation in the presence of adequate amounts of
sensitizing antibody and an excess of comple-
ment components may result in a considerable
increase in viable cell numbers (125, 302). There
are, however, many strains that fall between
these two extremes with regard to serum suscep-
tibility; significant reduction in viable count may
be apparent only after lengthy periods of incuba-
tion (125, 299, 320). Nevertheless, these strains
should be regarded as susceptible to serum, or at
least significantly different from strains that are
capable of replicating in serum, because de-
crease in viable cell numbers implies insertion of
functional MACs into sites on the surface of the
cell envelope from which they can effect killing,
albeit slowly. It is therefore unfortunate that
many studies have described as resistant strains
which have undergone significant reduction in
viability after incubation in serum (145, 221, 251,
279). In an extreme example (79), certain antibi-
otic resistance plasmids were described as con-
ferring serum resistance on E. coli K-12 strains;
in fact, numbers of plasmid-carrying strains
were reduced to 1% of the inoculum after 30 min
of incubation in 3% human serum. Although
these plasmids did reduce by a small amount the
rate of serum killing, they clearly did not confer
serum resistance. It is recommended that the
term "serum resistance" be restricted to the
description of strains that are totally refractory
to serum when tested in early logarithmic phase
in assays containing excess amounts of the com-
ponents of the bactericidal system.
Location of Serum Resistance Determinants at
the Outer Membrane
There is no evidence that resistant gram-
negative bacteria circumvent the bactericidal
action of serum by producing proteases or other
extracellular products that neutralize or destroy
the functional integrity of complement compo-
nents in the fluid phase (139). As protoplasts
derived from serum-resistant enterobacteria
(248) and even from gram-positive bacteria (204)
are rapidly lysed by complement, it appears that
the outer membrane constitutes the main barrier
to the serum killing of gram-negative bacteria.
Serum-resistant cells may be sensitized to the
bactericidal action of serum by treatment with
Tris and EDTA (246, 248), a process known to
lower the permeability barrier of the outer mem-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
brane by effecting the release of approximately
half of the complexed lipopolysaccharide (156).
Similarly, serum-resistant enterobacteria be-
come susceptible to complement after exposure
to polymyxin B (75, 293); this antibiotic is
known to adversely affect the integrity of the
outer membrane (72, 313), perhaps as a result of
its ability to complex with lipopolysaccharide
(15, 198). Further, albeit indirect, evidence for
cell envelope involvement in the determination
of serum resistance is provided by the observa-
tion that penicillin- or cephalosporin-resistant
mutants selected from serum-resistant, antibiot-
ic-susceptible populations of gram-negative bac-
teria may be serum susceptible (183, 214, 252).
It appears that serum resistance does not
result from a block in the activation of the
complement cascade. Rather, MACs formed on
the surfaces of serum-resistant strains are not
effectively inserted into the bacterial membranes
and are released without causing lethal damage
(139, 140). Reynolds et al. (247) found that
equivalent amounts of C3 were deposited on the
surfaces of serum-resistant S. typhimurium cells
and on the same bacteria that had been rendered
susceptible to serum by treatment with Tris and
EDTA. They did, however, fail to show deposi-
tion of C5 on untreated cells, an observation that
contrasts with those of other workers (139, 140,
221). Ogata and Levine (221) observed compara-
ble fixation of C4, C3, and C5 by E. coli J6-2 and
by the same strain harboring plasmid R100 after
treatment with low concentrations of either
guinea pig or human serum. However, the de-
gree of serum resistance conferred by this FII
incompatibility group plasmid is marginal and
only detectable in systems containing small
amounts of serum and is, therefore, not compa-
rable with the high levels of resistance attribut-
able to many wild strains. C3 deposition on the
surface of serum-resistant blood isolates of E.
coli has been demonstrated by Fierer and Finley
(75). Joiner and colleagues (139, 140) have re-
cently studied the interaction of C3 and some of
the terminal components of the complement
cascade with the smooth S. minnesota strain
S218 and a deep rough mutant (ReS95) of S218
that synthesizes a lipopolysaccharide containing
only lipid A and 3-deoxy-D-manno-octulosonic
acid; S218 is completely resistant to serum,
whereas Re595 is highly susceptible (139, 260).
Although both strains consumed 75 to 80% of
the C3 from human serum, twice as many mole-
cules of radiolabeled C3 were bound per cell of
S218 compared with Re595. C5, C7, and C9
depletion of human serum after incubation with
Re595 cells was found to be relatively low; in
contrast, depletion of C5 and C7 by S218 cells
was 95% after 15 min of incubation, and com-
plete inactivation of C9 occurred after 5 min.
63
Comparatively few molecules of radiolabeled
MAC components were rapidly and irreversibly
bound to the surface of Re595 cells. Therefore,
this serum-susceptible strain consumes relative-
ly small amounts of C5, C7, and C9, and these
components are efficiently and stably deposited
on the bacterial surface. Much larger amounts of
the terminal components were deposited onto
S218 cells, but binding reached a peak after
about 10 min, and a progressive loss of the
bound components was observed with pro-
longed incubation. The serum resistance of S.
minnesota S218 is not, therefore, attributable to
a defect in MAC formation on the bacterial
surface; the formed complex does not remain
associated with the surface, suggesting that the
complex fails to integrate into the outer mem-
brane. Furthermore, MACs could be eluted
from the surfaces of S218 cells, but not Re595
cells, after incubation in buffers of increasing
ionic strength (140), indicating that MACs on the
surface of the serum-resistant strain did not
behave as integral membrane proteins. Signifi-
cant loss of membrane phospholipids from the
serum-resistant strain was not observed, demon-
strating that loss of surface-bound MACs was
not due to shedding of part of the bacterial
surface. Other workers have also found that, in
contrast to serum-susceptible organisms, phos-
pholipid is not released from serum-resistant
strains (17), emphasizing that one mechanism of
serum resistance is due to a failure of amphiphi-
lic MACs to integrate into hydrophobic domains
on the bacterial envelope.
One parameter that appears to be intimately
related to the ability of MACs to integrate into
biological and artificial membranes is the degree
of fluidity of these structures. Consequently,
factors reducing the fluidity of either the outer or
cytoplasmic membranes of gram-negative bacte-
ria could be crucial in determining serum resist-
ance. The degree of disruption of phospholipid-
cholesterol liposomes varies inversely with the
concentration of membrane cholesterol (275),
and there is a correlation between susceptibility
of sheep erythrocytes to complement attack and
the ratio of membrane lecithin to sphingomyelin
(171), suggesting that increased membrane fluid-
ity enables a more efficient assembly and inte-
gration of MACs on phospholipid membranes.
Incorporation of the fluidizing and fusogenic
agent 2-(2-methoxy)-ethoxyethyl-8-(cis-2-n-oc-
tylcyclopropyl)-octanoate into sheep erythro-
cyte membranes loosens phospholipid acyl-
chain packing and thus increases membrane
lipid disorder; cells treated in this way become
extremely susceptible to complement lysis (274).
Below the phase transition temperature, mem-
brane phospholipids are in a state of gel packing
with their acyl chains in a restricted and ordered
64 TAYLOR
state; above this temperature, they are in a
liquid crystal state with the fatty acyl chains
exhibiting a high degree of molecular motion.
Kato and Bito (144) have examined the effect of
the physical state of membrane phospholipids on
the susceptibility of E. coli to immune lysis.
They found that the phase transition tempera-
ture and the temperature at which cells became
susceptible to complement were both subject to
variation depending on the conditions of bacteri-
al growth and were highly correlated. Thus,
susceptibility to complement increased rapidly
with increases in temperature, up to the point at
which the transition to liquid crystal is consid-
ered finished (27°C for E. coli B cells grown to
stationary phase in complex media at 42°C). The
authors suggested that membrane fluidity is
obligatory for the formation of functional com-
plement lesions. Parenthetically, membranes be-
come initially more rigid after insertion of the
MAC (51, 90), probably as a result of reorienta-
tion of ordered bilayered lipid effected by strong
binding of phospholipids to MAC proteins (68).
Akiyama and Inoue (4) reported that comple-
ment-resistant variants of E. coli K-12 had a
less-fluid membrane structure due to a shift of
fatty acid composition, although the basis of
these changes is unclear as resistance was lost
during maintenance of the strains.
It therefore becomes attractive to suggest that
those macromolecular components of the outer
membrane that are thought to confer serum
resistance do so by virtue of the fact that they
reduce membrane fluidity and exclude integra-
tion of MACs. Unfortunately, relatively little
information is currently available concerning the
physical state of gram-negative membranes.
Overath et al. (231) found that less than one-half
of the outer membrane phospholipid takes part
in the order-disorder transition, whereas the
bulk of cytoplasmic membrane phospholipids
become ordered, suggesting that many lipid mol-
ecules in the outer membrane cannot participate
in the lipid-lipid interactions characteristic of the
order-disorder transition owing to interaction
with protein. These lipid-protein interactions
would therefore have a fluidizing effect on the
outer membrane. In support of this contention,
Nikaido and co-workers (215) have found that
isolated outer and cytoplasmic membranes from
a S. typhimurium rough (Rc) mutant, incorporat-
ing spin-labeled fatty acid probes, produced
remarkably similar electron spin resonance
spectra, indicating minimal immobilization of
probe by the Rc lipopolysaccharide, which lacks
O side chains and sugar residues from the outer
core (338). Other studies, however, suggest that
complete, S-type lipopolysaccharide may in-
crease the microviscosity of the outer membrane
(40, 257). Outer membrane preparations contain-
MICROBIOL. REV.
ing lipopolysaccharide molecules with long 0
side chains were less fluid than preparations
containing the same number of lipopolysaccha-
ride molecules lacking 0 side chains and some
core sugars (257). Furthermore, removal of
about half of the S-type lipopolysaccharide from
membrane preparations by EDTA treatment
greatly increased the fluidity of the membranes.
Although these findings are controversial (49),
they would, if verified, suggest an attractive
mechanism for the serum resistance of smooth
gram-negative bacteria. Clearly, however, sus-
ceptibility to complement is dependent not only
upon the properties of hydrophobic membrane
domains. Modulation of hydrophilic structures
also influences the ability of complement to
interact with membranes. For example, removal
by enzymatic hydrolysis of N-acetyl-neuraminic
acid (NANA)-containing structures from the
outer surface of the membrane increases eryth-
rocyte susceptibility to complement lysis (154).
Therefore, of the structures that have been
implicated as mediators of serum resistance,
lipopolysaccharides and outer membrane pro-
teins may affect membrane viscosity, whereas
some acidic exopolysaccharides (e.g., K anti-
gens of E. coli) might function in an analogous
fashion to the NANA-containing structures on
the erythrocyte membrane.
Role of Lipopolysaccharides
The earliest attempt to systematically define
cell envelope components involved in the deter-
mination of serum resistance can be attributed to
Wardlaw (329, 330), who compared the compo-
sition of purified envelopes from a smooth,
resistant E. coli strain with those from the
rough, extremely susceptible strain Lilly. The
author originally postulated (328) that resistant
strains might contain relatively few phospholipid
complement-binding sites on the envelope in
comparison with susceptible strains. It was
found, however, that both strains contained
comparable amounts of envelope protein and
lipid but that there was a ninefold difference in
the amount of lipopolysaccharide that could be
extracted from the envelopes by the phenol-
water procedure (336). It was suggested that, in
some way, an outer membrane rich in lipopoly-
saccharide could protect celis from complement
attack. However, the phenol-water technique
removes only a proportion, sometimes very
small, of R lipopolysaccharide from rough
strains but is considerably more efficient when
applied to smooth forms (85). Efficiency of lipo-
polysaccharide extraction could, therefore, go
some way towards explaining the disparity in
yields of this macromolecule from the two
strains. Both Rowley (259) and Michael and
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
Landy (186), using only smooth enterobacteria,
showed that serum-susceptible organisms pos-
sessed less intrinsic endotoxic activity than re-
sistant ones, a function that can be directly
attributed to the lipid A moiety of lipopolysac-
charide (193). However, no significant differ-
ences could be found in the amount of lipopoly-
saccharide extractable from 28 smooth, urinary
E. coli strains of differing serum susceptibilities
(301). Similarly, mutations to serum resistance
occurred independently of quantitative changes
in E. coli lipopolysaccharide content (300), and
phenotypically induced transitions from serum
resistance to susceptibility were not accompa-
nied by differences in lipopolysaccharide yields
(308). Therefore, although strains with a low
lipopolysaccharide content may generally be
susceptible to serum, it is unlikely that the
absolute amount of polymer in the envelope is
the major factor determining resistance.
Apparent differences in lipopolysaccharide
yields obtained from smooth and rough strains
may be due to the fact that R-type lipopolysac-
charides have a significantly lower molecular
weight than S-type molecules, because R lipo-
polysaccharides lack the side chain moiety and
sometimes a portion of the core. Therefore,
comparisons between smooth and rough strains
with regard to serum reactivity may not be valid
and in any case will provide no information
concerning the nature of the serum susceptibility
exhibited by many smooth enterobacteria. Lipo-
polysaccharides do, however, play a central role
in determining the susceptibility of enterobac-
teria to serum. For example, mutations from the
smooth to the rough colonial form, usually but
not invariably associated with loss of the ability
to synthesize lipopolysaccharide 0 side chains
(137, 227, 338), are accompanied by drastic
increases in serum susceptibility (127, 166, 214,
260, 300, 314). In a detailed study of this phe-
nomenon, Dlabac (56) examined the susceptibil-
ity to piglet serum of a series of rough mutants
derived from a smooth S. typhimurium strain;
although the smooth strain was completely resis-
tant, nearly all of the rough mutants were sus-
ceptible. There was a progressive increase in
serum susceptibility corresponding to sequential
loss of sugar residues from the lipopolysaccha-
ride core. S. typhimurium mutants deficient in
UDP-galactose-4-epimerase form incomplete li-
popolysaccharides which lack both 0 side
chains and that part of the core distal to the point
of the biosynthetic lesion, unless supplied with
exogenous galactose (230). Dlabac (57) found
that, when grown in galactose-free medium,
these mutants were susceptible to serum; when
galactose was supplied to growing cells, they
became increasingly serum resistant with time.
Similarly, the serum resistance of mutants defi-
65
cient in epimerase increased in direct proportion
to the amount of galactose added to the growth
medium.
Further evidence of the influence of the 0 side
chain moiety of lipopolysaccharide on serum
susceptibility was obtained by Roantree and co-
workers (214, 291), who derived serum-suscepti-
ble mutants from resistant, smooth Salmonella
strains by selection for resistance to cephalo-
sporin or penicillin. The majority of such mu-
tants were rough or part rough as a result of
partial or total loss of the 0 side chain. A small
number, however, possessed lipopolysaccharide
with an 0 side chain sugar-to-core sugar ratio
identical to that found in the parent strains,
indicating that other factors must be involved in
the determination of resistance of the smooth
parental strains. Feingold (69) found that serum-
resistant E. coli and Pseudomonas strains be-
came susceptible after growth in media supple-
mented with low concentrations of
diphenylamine. Analysis of lipopolysaccharides
extracted from diphenylamine-grown cells re-
vealed a striking reduction in the ratio of 0 side
chain sugars to core sugars when compared with
cells grown in the normal way. Some serum-
resistant, smooth E. coli strains are converted to
serum susceptibilty after growth in the presence
of the 6-f3-amidinopenicillanic acid mecillinam
(303); lipopolysaccharides from cells grown in
this way contain reduced amounts of 0 side
chain components (307).
In all of these studies, increased susceptibility
to serum was accompanied by a reduction in 0
side chain sugars of lipopolysaccharide. Howev-
er, these structural changes were relatively dras-
tic inasmuch as serum-susceptible cells showed
many of the characteristics of rough organisms.
Many susceptible clinical isolates are complete-
ly smooth by cultural, morphological, and sero-
logical criteria and are indistinguishable in this
respect from serum-resistant organisms. As the
0 side chain length and degree of substitution of
core stubs by 0 side chains is known to be
subject to phenotypic and perhaps genotypic
variation (42, 46, 99, 312), the possibility arises
that the degree of serum resistance expressed by
smooth strains may simply be a reflection of
these parameters.
In general, more 0 side chain material was
associated with lipopolysaccharide from serum-
resistant than from serum-susceptible urinary
strains of E. coli, but differences did not reach
levels of statistical significance (301). Similarly,
no quantitative differences in lipopolysaccharide
sugar composition could be found between se-
rum-resistant mutants and their respective E.
coli parent strains (300). It appears unlikely,
therefore, that the serum resistance of gram-
negative bacteria is determined solely by the
66 TAYLOR
length and numiber of 0-antigenic polysaccha-
ride chains associated with lipopolysaccharide,
although these studies do not discount the possi-
bility that a subtle rearrangement of lipopolysac-
charide components within the molecule might
account for the observed differences in serum
reactivity.
Many strains carrying a full complement of
lipopolysaccharide 0 side chains may be suscep-
tible to serum although, unlike rough strains,
they are not usually promptly killed but exhibit a
delayed sensitive response (299). That this delay
is related to the presence of 0 side chains is
indicated by observations made with colonially
rough variants derived from a serum-resistant
mutant (300, 305). One rough form had lost the
ability to synthesize the lipopolysaccharide 0
side chain as a result of a mutation in the his-
linked rbf gene locus that specifies sugar trans-
ferases participating in 0 repeat unit biosynthe-
sis (227) and was rapidly killed by serum,
whereas another colonially rough form retained
serological 0 specificity, produced a full com-
plement of lipopolysaccharide 0 side chains,
and was killed in a delayed fashion. Inheritance
by a promptly susceptible rough E. coli strain of
the rJb locus from a serum-resistant, K-negative,
smooth 08 Hfr donor resulted in smooth recom-
binants that displayed the delayed serum killing
response (300, 311). E. coli urinary isolate
LP729 (serotype 09) displays delayed serum
killing kinetics (300, 302); when grown in the
chemostat under conditions of carbon limitation
and magnesium limitation, the degree of killing
after 1 h of incubation in human serum was
directly related to the ratio of the lipopolysac-
charide 0 side chain sugar (mannose) to the core
sugar L-glycero-D-mannoheptose (308). These
observations make it likely that 0 side chains do
not determine serum resistance per se but that a
high degree of substitution of lipopolysaccharide
core stubs by long 0 side chains is responsible
for the delayed serum killing response character-
istic of many smooth isolates. The mechanism
by which 0 side chains may do this is unclear;
the attractive possibility that low membrane
fluidity, resulting from synthesis of a complete
lipopolysaccharide, reduces the rate of insertion
of functional MACs into the cell envelope has
not so far been investigated. It has been fre-
quently suggested that long 0 side chains cause
the antibody-mediated activation of complement
at some distance from the site of lesion forma-
tion, and intermediates therefore decay before
they can be incorporated into functional MACs
(204, 206, 261, 262). Such a concept could also
be extended to include antibody-independent
activation of the alternative pathway, as the
polysaccharide moiety of lipopolysaccharide is
known to activate complement (199).
MICROBIOL. REV.
Role of Acidic Polysaccharides
Muschel et al. (203) examined the acidic exo-
polysaccharide (Vi antigen) content of clinical
isolates of S. typhi and suggested that serum-
resistant S. typhi strains produced larger
amounts of Vi antigen than susceptible strains.
Later, a similar relationship was proposed for E.
coli strains isolated from both intestinal and
extraintestinal infections (201). Correlations
were far from complete, however, and in each of
these studies only eight strains were examined.
Other work also suggests a relationship be-
tween acidic exopolysaccharide production and
serum resistance. Glynn and Howard (94) found
that saline extracts of serum-resistant strains of
E. coli were, in general, more able to non-
specifically inhibit the agglutination of sheep
erythrocytes by antierythrocyte serum than ex-
tracts of serum-susceptible strains. The authors
interpreted the inhibitory effects as being due to
the presence of K antigens in the preparations;
the contribution of other components in the
extracts is unclear, although purified K antigen
polymers from two strains were shown to pos-
sess agglutination-inhibiting activity. It was also
shown that a purified hexuronic acid-containing
K antigen from a serum-resistant strain had
greater inhibitory activity than equal amounts of
Ki antigen, a NANA homopolymer, from a
susceptible strain. It has been subsequently es-
tablished, however, that hexuronic acid-contain-
ing E. coli exopolysaccharides are more potent
inhibitors of hemagglutination systems than
NANA-containing polymers, regardless of the
reactivity in serum of the strain from which the
polymer was extracted (301). It has been sug-
gested that acidic polysaccharides may exert an
influence on the extent of bacterial killing by
serum due to an ability to impede antibody
binding and subsequent attachment of comple-
ment components to the bacterial surface (94).
Current evidence would tend to suggest, howev-
er, that polysaccharide capsules do not repre-
sent a diffusion, permeability, or adsorption
barrier to macromolecules such as IgG (147) or
other proteins (146) and that 0 inagglutinability
frequently associated with the presence of acidic
polysaccharides superficial to the bacterial sur-
face is most likely due to surface protein compo-
nents (136) or to inhibition of lattice formation
between adjacent bacteria (38).
No obvious relationships were found when
acidic exopolysaccharides were prepared by two
methods from a large number of urinary E. coli
isolates of various serum susceptibilities (301).
Some susceptible and resistant strains produced
large amounts of K antigen with high hemagglu-
tination-inhibiting activity, but both groups also
contained strains that produced small quantities
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
of polymer with no detectable activity. A similar
lack of correlation was found for E. coli bacter-
emic strains by McCabe et al. (174). Mutations
to serum resistance did not result in increased
exopolysaccharide production (300). Inheritance
of the his-linked genes for K27 antigen produc-
tion by E. coli recipients did not result in an
altered serum response, regardless of whether
recombinants produced R lipopolysaccharides
or had simultaneously inherited 0 side chains
(311). Van Dijk and co-workers (325) also failed
to find a correlation between K antigen produc-
tion as determined by the hemagglutination inhi-
bition method and the serum resistance of blood
and fecal isolates of E. coli. Klebsiella strains
are frequently found to be susceptible to serum
(77, 103), even though most members of this
genus produce copious amounts of hexuronic
acid-containing acidic polysaccharide (294). P.
aeruginosa isolates originating from the respira-
tory tracts of patients with cystic fibrosis are
known to be generally more susceptible to se-
rum than strains of this species from other
sources (122); cystic fibrosis isolates are gener-
ally extremely mucoid (58), owing to the produc-
tion of a polyuronide of composition similar to
alginic acid (159).
There appears, therefore, to be substantial
evidence against a major role for acidic exopoly-
saccharides as mediators of serum resistance, at
least in a general sense that can be discerned by
comparative epidemiological investigation.
These polymers form a distinctly heterogeneous
group of bacterial products with regard to chem-
ical composition, antigenicity, molecular size,
biophysical properties, and the intimacy of their
association with the cell surface. These factors
are likely to distort any such complex relation-
ship as that between serum resistance and the
interaction of the bacterial surface with activat-
ed complement components unless steps are
taken to minimize these effects. It is in any case
clear that any increase in serum resistance di-
rectly attributable to the presence of acidic
polysaccharides cannot form part of a general
mechanism for resistance of gram-negative bac-
teria, as many groups, such as Salmonella spp.
(56, 209, 214), possess a high level of intrinsic
serum resistance and yet do not produce acidic
polysaccharides.
E. coli strains synthesizing the Kl antigen, an
a-2,8-linked NANA homopolymer with a degree
of polymerization of 150 to 200 sialyl residues in
situ (321), are isolated with high frequency from
cases of neonatal meningitis (254) and renal
infection in infancy (142). Thus, the Kl antigen
or the presence of closely linked traits may be a
virulence factor related to the pathogenesis of
these diseases. Several groups have considered
the possibility that the Ki-mediated pathogenici-
67
ty of these strains is related to a capacity to
resist the bactericidal activity of serum. The
relative homogeneity of Kl strains as a group
(276) makes them considerably more attractive
material for comparative epidemiological studies
than random E. coli isolates.
Two groups have reported on the cloning into
plasmids of the chromosomally located, serA-
linked kpsA locus determining synthesis of Kl
(228) and its expression in E. coli strains. Silver
et al. (277) used a cosmid cloning vector in
combination with in vitro packaging techniques
to clone the Kl genes; kpsA specified synthesis
of a NANA polymer in E. coli K-12 that was
indistinguishable chemically and immunological-
ly from that of wild-type Kl strains. In contrast
to the K-12 host, E. coli K-12 expressing the
cloned Kl antigen was reported to be serum
resistant on the basis of efficiency of plating on
agar containing equine meningococcal group B
antiserum; the non-0-acetylated form of the Kl
antigen and the group B polysaccharide are
identical (255). Timmis et al. (316), using an
identical cosmid cloning-packaging system, in-
troduced a number of BamHI or PstI-generated
fragments into an E. coli K-12 strain. In 6%
rabbit serum, all K-12 cells were killed after 3 h,
whereas all Kl+ derivatives survived; such dif-
ferences were not apparent at higher serum
concentrations. These results complement those
of other investigators who have assigned a role
for the Kl antigen on the basis of genetic studies
of serum resistance. Kl-negative mutants of E.
coli Kl clinical isolates were found to be more
susceptible to high concentrations of serum than
their parent strains (32). Rough, Ki-positive E.
coli bacteremic isolates were resistant to 10 to
20% human serum, whereas single-step isogenic
Kl-negative derivatives, with an identical outer
membrane protein profile, were found to be
extremely serum susceptible (87). These studies
therefore suggest a direct role for the Kl antigen
in the protection of E. coli against serum killing.
As Taylor and Robinson (311) had earlier failed
to observe any increased serum resistance after
inheritance of genes determining biosynthesis of
the K27 antigen, Gemski and co-workers (225)
constructed strains expressing both Kl and K27
antigenic determinants. It is possible to prepare
such hybrids because gene loci for these anti-
gens are nonallelic (227). Hybrid strains express-
ing both Kl and K27 antigens exhibited serum
resistance, but at levels lower than those found
with a Kl parental strain. An isogenic Kl-
negative derivative expressing only the K27
antigen was fully serum susceptible. Loss of
capacity to produce Kl antigen by chemostat-
grown cells is accompanied by a large increase
in serum susceptibility (308).
When, however, E. coli Kl clinical isolates
68 TAYLOR
are examined and compared, no clear picture
emerges with regard to a relationship between
Kl carriage and serum resistance. Bjorksten et
al. (26) found no consistency in the way fecal,
blood, or cerebrospinal fluid isolates of E. coli
Kl reacted in a number of test procedures
involving bacteria-serum interactions, including
serum bactericidal systems. No significant dif-
ferences could be found between the amount of
acidic polysaccharide produced by urinary and
fecal Kl strains and their reactivity in human
serum (222). E. coli Kl strains from cases of
neonatal meningitis were more susceptible to
serum than those from cases of neonatal or adult
sepsis or adult meningitis (236). In these studies,
many serum-susceptible Kl isolates were re-
ported, and as a group, Kl strains appear to be
rather serum susceptible. It is therefore difficult
to reconcile these data with the results of the
genetic studies cited above, particularly as the
latter have generally utilized E. coli K-12 strains
as hosts for KI genes; effects attributable to Kl
carriage would therefore be seen against a back-
ground of high serum susceptibility. Discrepan-
cies could be due, however, to the fact that anti-
Kl antibodies are not usually present in
reasonable amounts in normal sera or, indeed, in
the sera of patients recovering from either E.
coli Kl or meningococcal B infection (266).
Sialic acid-containing polymers, including the
Kl capsule (290), restrict activation of the alter-
native pathway, and so killing of such strains
would be expected to proceed via the antibody-
dependent classical route. If Kl is the only
major surface antigen available for interaction
with serum factors, then apparent resistance
may be due to lack of a mechanism for classical
pathway activation. The addition of an effective
antibody source to the bactericidal system might
then result in killing. The high incidence of
serum-susceptible Kl isolates observed in most
clinical studies emphasizes that the Kl antigen
cannot be universally considered as providing
protection against serum killing mechanisms.
Role of Plasmid-Determined Factors
The earliest report of an altered serum re-
sponse after the acquisition of foreign DNA
appears to be that of Muschel and associates
(202, 207), who found increased serum resist-
ance after lysogenization of E. coli K-12 with A
phage. In 1974, Williams Smith (338a) noted that
strains of E. coli invasive for humans and do-
mestic animals harbored ColV plasmids which,
when transferred to K-12 and other E. coli
strains, were responsible for increased lethality
for chickens and mice and a greater ability of the
host bacteria to survive in blood and serum.
Binns et al. (24) have cloned various fragments
of CoIV,I-K94, a plasmid specifying production
MICROBIOL. REV.
of colicins V and I; a BamHI-generated fragment
increased the resistance to fresh rabbit serum of
a Col- E. coli 078:K80 strain. The gene specify-
ing serum resistance, the iss determinant, was
mapped to a 5.3-kilobase sequence within the
fragment and found to be closely linked, though
not coincident with, genes for colicin V produc-
tion. It is likely that the iss gene product, in
addition to inhibiting insertion or activity of the
MAC on the outer membrane (25), is responsible
for the increased virulence associated with CoIV
carriage. That the iss gene product may be
important in relation to the ecology of these
organisms is suggested by the observation that a
high proportion of virulent E. coli strains isolat-
ed from infections of hospitalized patients carry
and express ColV-determined functions (52).
Reynard and Beck (244) reported that the
plasmids Rl and R100 (NR1), in addition to
encoding resistance against a number of antibi-
otics, were able to confer relatively high degrees
of resistance against rabbit serum on a number
of E. coli K-12 strains. This ability appeared to
be restricted to F-like plasmids; plasmids from 4
of 16 clinical isolates examined were able to
protect E. coli K-12 J6-2N against the bactericid-
al action of rabbit serum (245). Although it has
been repeatedly confirmed that certain F plas-
mids are able to afford host protection against
the bactericidal effects of serum, there has been
some controversy over the degree of resistance
that plasmids may confer that has its origins in
the fact that, as discussed earlier, the use of
widely differing techniques and the utilization of
serum from a variety of sources and at different
concentrations makes meaningful comparisons
between such studies difficult.
Thus, a number of other workers have been
unable to confirm that either Rl or NR1 protect
K-12 strains against serum killing. Fietta et al.
(79) could not detect increased survival after
inheritance by strain J6-2 of plasmid RI when a
system employing highly dilute human serum
was used; they did, however, report that 8 of 26
plasmids examined conferred relative serum
resistance to E. coli K-12 strains. A plasmid was
said to have conferred resistance if the amount
of serum needed to reduce the viable count to
1% after 30 min was greater for R+ progeny than
for R- parents. The amounts of serum required
to effect this degree of killing were extremely
small, usually about 3% of the total reaction
mixture, and at such concentrations classical
pathway components may be present in limiting
amounts and alternative pathway activity is al-
most certain to be nonexistent (268). Taylor and
Hughes (304) examined by two assay methods
the effect of carriage of a wide range of R and
Col plasmids, including Rl and NRl, on the
susceptibility of six E. coli K-12 strains to nor-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
mal human serum. With both methods, all K-12
strains examined were rapidly killed by serum,
regardless of plasmid carriage. Survival in rabbit
serum was also unaffected by the presence of
these plasmids. Part of this discrepancy can be
attributed to the source of serum; Reynard and
Beck (244) used commercially available lyophi-
lized rabbit serum as source of antibody and
complement, and the material was reconstituted
with physological saline, whereas in other stud-
ies, fresh rabbit or human serum was used.
Human serum appears to be an inappropriate
source, as Reynard and co-workers have them-
selves reported that plasmid Rl does not provide
K-12 strains with protection against human se-
rum (245). Much of this confused situation has
been clarified by Ogata and Levine (221), who
examined the effect of various concentrations of
rabbit, guinea pig, and human sera on E. coli
strains carrying plasmid R100. They found that
differences in the serum reactivity of pairs of K-
12 strains with and without R100 could be de-
tected provided a high bacterial cell density and
relatively low concentrations of rabbit serum
were used. At a given serum concentration,
survival of plasmid-carrying strains decreased as
the bacterial concentration decreased. Howev-
er, in 10% rabbit serum, significant killing of all
five plasmid-carrying K-12 strains was ob-
served, and in four cases, incubation for 50 to 60
min resulted in less than 10% survival. It is
therefore clear that what is being measured is
not serum resistance as defined earlier but a
reduction in the rate of killing that is only
manifest under certain experimental conditions.
One could imagine that insertion of novel plas-
mid-encoded proteins with unrelated cell func-
tions into the outer membrane might fortuitously
reduce membrane fluidity and hence the rate at
which MACs are inserted into the membrane,
especially when concentrations of serum are
limiting. E. coli J6-2(R100) survived better in
guinea pig serum than E. coli J6-2, but in human
serum, which is a much more potent source of
bactericidal activity than either rabbit or guinea
pig serum, a barely discernible effect was re-
corded over a serum concentration range of 0.5
to 2%. The authors (221) suggested that failure
of other groups (245, 304) to detect a protective
effect of R100 against human serum was due to
the use of serum concentrations greater than
10% and cell concentrations 10- to 100-fold
lower than used in their study.
As is clear from the studies of Binns et al. (24),
more readily detectable effects of plasmids on
bacterial survival in serum can be achieved if
strains with a certain intrinsic level of resistance
are used as hosts. Taylor and Hughes (304)
cured a number of enterobacteria isolated from
polluted river water of antibiotic resistance and
69
bacteriocin determinants. One strain, E. coli C8,
displayed delayed human serum killing kinetics;
curing of all markers resulted in a small but
significant increase in serum susceptibility. In-
troduction into the cured derivative of plasmids
from six of eight other river isolates and of
plasmids Rl and R100 resulted in significant
increases in survival in serum. Rl and R100
were unable to modify the serum response of a
cured strain derived from a promptly susceptible
isolate, and the authors therefore suggested that
lipopolysaccharide 0 side chains, the surface
components responsible for the delay in serum
killing, were essential for expression of plasmid
factors that modify susceptibility to serum, at
least in systems containing adequate concentra-
tions of fresh human serum. That this might be
so was indicated by the results of a study of
plasmid-determined factors expressed in recom-
binants from Hfr crosses carrying defined com-
binations of polysaccharide surface antigens
(311). Plasmid factors encoded by Rl and R100
failed to modify the promptly susceptible re-
sponse to human serum of either E. coli F470 or
F470 progeny that had inherited and expressed
his-linked genes for K27 antigen production.
Recombinants inheriting the rJb locus for 08
side chain production showed delayed serum
killing kinetics, and serum susceptibility was
additionally decreased by the presence of either
of the two plasmids. Although not completely
serum resistant, a small but significant contribu-
tion towards the partial resistance to serum of
such recombinants was made by plasmid-deter-
mined genes.
Plasmid R6-5 is, like R100, a large, low-copy-
number, conjugative, multiple antibiotic resist-
ance plasmid of the FIl incompatibility group.
Timmis and co-workers have recently shown
(195, 196, 316) that R6-5 encodes a factor that
increases the survival of the moderately serum-
susceptible fecal isolate E. coli 59 in rabbit
serum. Using gene cloning techniques, they lo-
calized these functions to a segment of the
plasmid specifying conjugal transfer functions.
Analysis of serum-susceptible deletion and in-
sertion mutants demonstrated that the gene
specifying serum resistance was coincident with
the traT locus, one of two loci involved in
surface exclusion, the poor recipient ability in
conjugation of F factor-carrying cells (1). The
traT protein is present at about 21,000 copies per
cell and is situated on the outer surface of the
outer membrane. It has, therefore, the capacity
to prevent cell-protein interactions that would
normally result in assembly and insertion of
functional MACs. It is likely to do this in a
highly specific way, as plasmids containing point
mutations in the traT gene no longer confer
protection against serum, even though bacteria
70 TAYLOR
contain more of the altered gene product in their
outer membrane than do bacteria harboring the
parental plasmid (196). The studies cited above
suggest that the presence of plasmids of incom-
patibility group FII may increase the rate of
survival of suitable host bacteria on serum,
although it is clear that the degree of resistance
conferred is not enough to explain the total
resistance to human serum of many clinical
isolates. The bactericidal assays used in much of
this work have been specifically tailored to
demonstrate the maximum possible difference in
serum reactivity between isogenic pairs. Differ-
ences demonstrable with rabbit serum but not
apparent in human serum may be of little rele-
vance to an understanding of the role of serum
resistance as a host defense mechanism. Plas-
mids encoding products that decrease the serum
susceptibility of suitable recipients have been
found in wild strains that are completely suscep-
tible to serum (304), indicating that the ecology
of these determinants is likely to be complex.
The limited amount of epidemiological data
available suggests that the carriage of R and Col
plasmids shown previously to be capable of
conferring increased levels of serum resistance
on individual E. coli strains isolated from other
sources does not play an important role in
determining the serum susceptibility of E. coli
populations, at least in urinary tract infections
(125). Indeed, the obviously multifactorial na-
ture of serum resistance makes it unlikely that
any convincing association between resistance
and expression of surface structures would
emerge from such an approach. Hopefully, a
better understanding of the molecular nature of
the bactericidal reaction will allow conclusions
to be drawn about the biophysical characteris-
tics of the cell surface that enable gram-negative
bacteria to circumvent the potentially lethal ac-
tion of complement.
Role of Other Outer Membrane Proteins
A serum-resistant mutant derived from a
smooth, delayed-susceptible urinary isolate of
E. coli produced more of an envelope protein of
molecular weight 46,000 than the parent (309);
the amount of this protein was subject to varia-
tion as a result of alterations in growth condi-
tions and was well correlated with environmen-
tally induced modification of the bacterial
response to serum (308). The authors suggested
that the protein factor was involved in the deter-
mination of serum resistance but that it was only
functional when superimposed on a full comple-
ment of lipopolysaccharide 0 side chains (309).
A protein modifying the response to human
serum of N. gonorrhoeae has been identified in
strains producing disseminated gonococcal in-
fection (120). Transformation of a serum-suscep-
MICROBIOL. REV.
tible strain to serum resistance with DNA from
disseminated gonococcal infection strains was
accompanied by the appearance of a new princi-
pal outer membrane protein with a molecular
weight (36,500) characteristic of the donor. The
principal virulence factor for Aeromonas sal-
monicida, the causative agent of systemic furun-
culosis in salmonid fishes, appears to be a
49,000-molecular-weight surface protein, the A
protein. Resistance of pathogenic strains to hu-
man, rabbit, and trout sera is dependent on the
presence of A protein (200).
SERUM BACTERICIDAL ACTIVITY AS AN
IMMUNE DEFENSE MECHANISM
Precise in vivo evaluation of the role of bacte-
ricidal mechanisms against gram-negative bacte-
ria has been difficult to achieve, most obviously
because of the difficulties in distinguishing com-
plement-mediated bactericidal and bacteriolytic
phenomena from other specific and nonspecific
host defense systems that may also modify the
interaction between host and invading parasite.
Furthermore, other host systems, such as im-
mune phagocytosis by blood leukocytes, may
also require at least some of the components of
complement for efficient processing and elimina-
tion of invading microorganisms. It is also likely
that distinct components of the host defense act
in concert after a challenge with potentially
pathogenic microorganisms and that comple-
ment-mediated humoral killing mechanisms will
be of more importance in the pathogenesis of
some infections than in others. For example,
only a limited role for the serum bactericidal
system can be envisaged in infections caused by
bacteria eliciting granulomatous tissue respons-
es in which the ability of the microorganisms to
survive in an intracellular location protected
from circulating noncellular mediators is a well-
established prerequisite of pathogenicity. Con-
versely, in infections such as gram-negative bac-
terial endocarditis, in which invading
microorganisms colonize a site that is apparently
inaccessible to phagocytic cells, resistance to
serum bactericidal factors appears to be an
important pathogenicity factor enabling strains
to resist elimination from the host (12, 62, 109).
Further complications arise because bacterial
surface determinants known to increase the se-
rum resistance of certain gram-negative strains
may also provide protection against other host
defense mechanisms (27, 123, 149). However,
acquisition by gram-negative bacteria of deter-
minants known to increase serum resistance is
sometimes associated with an increase in viru-
lence (24, 214), suggesting a direct role for serum
resistance as a bacterial pathogenicity factor in
certain situations. The importance of the com-
plement system in host defense is suggested by
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
the observation that the acute stage of many
gram-negative infections is characterized by in-
creased biosynthesis and turnover of comple-
ment proteins (13) and a reduction in the serum
level of natural complement inhibitors (337);
thus, in these situations complement behaves as
an acute-phase reactant. In some infections, the
primary mechanism of defense is provided by
immune phagocytosis, and many serum-resis-
tant strains can be efficiently phagocytosed after
opsonization (135, 279, 344). Nevertheless, an
extensive body of literature has steadily accu-
mulated implicating resistance to the bactericid-
al action of serum as an important determinant
of virulence in certain infections. Much of this
data is derived from two broad sources: a large
number of epidemiological studies based on clin-
ical observations are complemented by results
obtained with experimental invasive bacterial
infections in laboratory animals.
Clinical Observations on the Relevance of Serum
Resistance as a Bacterial Pathogenicity
Determinant
Are gram-negative bacteria isolated from in-
fected clinical material more resistant to serum
than random isolates from noninfected sites? It
would seem reasonable to assume that serum-
resistant gram-negative bacteria would possess a
significant survival advantage in the blood sys-
tem, and Roantree and Rantz (251) and Fierer et
al. (77) found that a much higher proportion of
strains isolated from the blood of bacteremic
patients than of those isolated from feces or
urine was resistant to killing by human serum.
Similarly, Vosti and Randall (326) found signifi-
cantly fewer serum-susceptible E. coli strains
isolated from blood cultures than from urine and
feces. Of 46 isolates of P. aeruginosa from blood
cultures of patients in a New York hospital, 42
(91%) were found to be resistant to human
serum (344); these authors also found, however,
that many apparently saprophytic pseudomo-
nads were also serum resistant. In a survey of
clinical isolates of S. marcescens, Simberkoff
and co-workers (279) found that no blood culture
isolates were serum susceptible, whereas 40% of
urinary isolates were susceptible to comple-
ment-mediated mechanisms. There appears,
therefore, to be a strong correlation between
serum resistance and the ability of a variety of
gram-negative bacteria to invade and survive in
the human bloodstream. Furthermore, bacter-
emia caused by serum-resistant E. coli strains is
more likely to be associated with shock and
death than bacteremia due to susceptible strains
(174). However, in a study of patients with
bacteremia due to gram-negative bacteria and
including a large number of elderly patients and
patients with complicating disease, Elgefors and
71
Olling (66) found that only 37% of isolates were
completely resistant to normal human serum and
that 26% were markedly serum susceptible.
The survival of apparently serum-susceptible
bacilli in the blood of bacteremic patients might
be related to a number of factors. Serum from
some bacteremic patients has been reported to
be ineffective in bactericidal assays utilizing the
infecting, homologous organism, even though
the strain is fully susceptible to serum from
healthy individuals (251). A similar phenomenon
has been reported for naturally ocurring bacter-
emia due to Brucella canis in an adult beagle dog
population (124). Bacteria may circumvent the
serum bactericidal system by adopting an intra-
cellular location in the blood or reticuloendothe-
lial system (250). For example, viable bacteria
may be found within leukocytes of patients with
infections of the blood system (283). The early
activation and consumption of complement be-
fore the onset of shock (82) may render bacteri-
cidal activity efficient only during the initial
stage of bacterial invasion (66), when cells may
be in a state of low metabolic activity and
therefore not susceptible to complement-mediat-
ed killing (54). Many gram-negative bacteria
capable of rapid growth in a nutrient-rich in vitro
environment may achieve only very low rates of
cell division when present in tissue or body
fluids (169, 181, 182, 239).
Bacteria dividing at submaximal rates may be
less susceptible to serum than more rapidly
dividing cells (302), and this may assist survival
of apparently susceptible isolates in vivo. Also,
the genetic basis for virulence may be expressed
completely only during growth in vivo: gono-
cocci adapted to growth in chambers implanted
subcutaneously into guinea pigs were found to
be resistant to human serum, but loss of this
phenotypically acquired trait occurred after a
few generations of in vitro growth (249). Serum-
susceptible E. coli acquired phenotypic resist-
ance to serum after multiplication in rat mono-
nuclear cells; resistance was rapidly lost during
subculture on solid medium (176). Residence in
monocytes from normal guinea pigs leads to
increased serum resistance of smooth strains of
Brucella abortus (292). Thus, serum resistance
attainable in vivo may not be apparent after in
vitro assay.
Perhaps, however, the most important factor
that should be considered pertains to the im-
mune status of the infected individual. Preexist-
ing disease is of major importance in the acquisi-
tion of bacteremia and in its outcome (173, 222,
345) and may therefore reduce the necessity for
invading bacteria to express virulence determi-
nants. For example, patients with sickle cell
anemia frequently contract systemic salmonello-
sis, and serum from such individuals displays
72 TAYLO)R
deficient bactericidal function against S. typhi-
murium (114), probably as a result of abnormal
alternative pathway activity. Neonates with hy-
perbilirubinemia as a result of blood group in-
compatibility are prone to infections with gram-
negative bacteria, and their serum possesses
significantly decreased bactericidal activity
(190) due to inhibition of complement function
by bilirubin (73). Sera from patients with cirrho-
sis of the liver have also been found to have
decreased bactericidal activity against E. coli
(76); patients with this disease are unusually
susceptible to bacteremic infections with gram-
negative bacilli (141, 317). Serum bactericidal
activity has been reported to be reduced in
patients with pyelonephritis (143) and chronic
renal failure (197). Clearly, it is important when
evaluating the role of serum bactericidal mecha-
nisms in relation to infection to consider the
possibility that the serum of patients under in-
vestigation may have a reduced capacity to kill
normally susceptible organisms. There is evi-
dence to suggest that cell-mediated host defense
functions may be impaired within the urinary
tract (41, 191). It is possible, therefore, that the
serum bactericidal system may constitute an
important defense mechanism in those parts of
the urinary tract in contact with antibody-com-
plement-containing fluids. Urinary isolates do
not appear to constitute a particularly serum-
resistant group of organisms (222, 299, 326),
although Kimball et al. (145) have presented
evidence indicating that E. coli 0 groups most
frequently isolated from patients with bacteri-
uria possess a higher degree of intrinsic serum
resistance than organisms belonging to 0 groups
infrequently isolated from infections of the uri-
nary tract. The significance of this observation is
unclear, as the distribution of 0 groups causing
urinary tract infections broadly reflects the rela-
tive frequencies of those 0 groups predominat-
ing in the gut flora (107). During initiation of
urinary infection, urinary pathogens are thought
to enter the bladder during micturition through
the short female urethra, and the bacteria con-
cerned are those nonanaerobes predominating at
that time in the gut flora. Once inside the urinary
tract, however, there appears to be some selec-
tivity with regard to the type of pathogen that
can successfully invade and colonize the upper
urinary tract. For example, it has been shown
that only strains of E. coli producing large
quantities of K antigen are able to infect renal
tissue (93). Gower and co-workers (103) exam-
ined the serum susceptibility of the infecting
gram-negative organism from patients with in-
fections that had been localized to either the
upper (ureters and kidneys) or lower (bladder)
urinary tract. A very high proportion of lower
infections was due to serum-susceptible organ-
MICROBIOL. REV.
isms. In contrast, the majority of upper urinary
tract isolates were not susceptible to serum
obtained from the infected patient. In some
cases, the infecting strains were inherently se-
rum resistant, but others were efficiently killed
by normal human serum. The serum of at least
some of these patients was found to contain high
amounts of IgG antibody (297) directed against
determinants within the lipopolysaccharide of
the infecting strain (298) that effectively inhibit-
ed the bactericidal action of both the patients'
sera and normal serum, probably by blocking
cell surface sites normally available to IgM
molecules. No evidence was found of a general-
ized decrease in serum bactericidal activity of
patients with urinary tract infections. The per-
sistence of serum-susceptible organisms in the
upper urinary tracts of patients whose sera con-
tain specific blocking factors that arise as a
result of bacterial invasion and colonization sug-
gests an active role for complement-mediated
bactericidal mechanisms in defending the kid-
neys against infection.
It has been suggested, however, that comple-
ment-mediated mechanisms are of limited rele-
vance to infections involving kidney tissue. As
part of an investigation of the basis of the
vulnerability of the kidney to infection, Beeson
and Rowley (18) reported that exposure of hu-
man or rabbit serum to homogenates of renal
tissue resulted in rapid loss of complement activ-
ity. They suggested that this effect could be
extrapolated to the situation in vivo, and they
attributed the anticomplementary action of renal
tissue to the ability of the kidney to produce
ammonia, which is known to inactivate the
fourth component of complement at pH 8.5 to 10
(101); Beeson and Rowley, however, conducted
their experiments at pH 7.4, and at this pH the
effect of ammonia on complement activity is
minimal (101). Contradictory findings were pre-
sented by Henkel (118), who could find no
inhibition of serum bactericidal activity by kid-
ney homogenates. The phenomenon has recent-
ly been reinvestigated by Ormrod and Miller
(226), using renal tissue maintained in vitro
under physiological conditions. There was a
rapid decrease in the hemolytic capacity of
serum after addition of kidney tissue, but C3
levels remained essentially unaltered. Liver tis-
sue, however, had a greater complement-inacti-
vating capacity than renal tissue. Inactivation
was not due to ammonia, as the concentration of
ammonia produced had no effect on the rate of
complement inactivation, and liver tissue pro-
duced only marginal amounts of ammonia. The
authors stressed that the conditions necessary
for C4 inactivation by ammonia (101) are incom-
patible with renal physiology and limit the po-
tential role of ammonia as a factor modifying
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
complement-mediated immune mechanisms.
Furthermore, the degree of complement inacti-
vation after even prolonged incubation of serum
with renal tissue was insufficient to affect com-
plement-mediated bacteriolysis of an E. coli
strain, suggesting that the significance of com-
plement inactivation in vitro as a factor modulat-
ing host defenses has been exaggerated. Com-
plement-mediated bactericidal activity is
inhibited by hypertonic, but not hypotonic or
isotonic, urine (2), an observation likely to have
significance regarding the often exaggerated
(191) susceptibility of the renal medulla to infec-
tion, as the mechanisms for concentrating the
urine are mainly located in this part of the
kidney. In fact, the susceptibility of the renal
medulla to infection is less during water diuresis
(11). The relative inability of serum-susceptible
strains to survive in the renal environment is
compatible with present concepts of host de-
fense in the urinary tract. The primary means of
disposing of bacteria that enter the bladder is
mechanical, and immune processes are probably
confined to the bladder wall and are likely to be
exclusively cellular in nature (44). In contrast,
both humoral and cellular defense mechanisms
are known to be operative in the kidney (191).
Selection within the urinary tract is also implied
by the observation that strains from patients
with asymptomatic bacteriuria may be more
serum susceptible than fecal isolates, whereas
strains from patients with symptoms of urinary
infection are more serum resistant than either
fecal isolates or isolates from asymptomatic
patients (223). In some patients with asympto-
matic bacteriuria, the infecting strain becomes
more susceptible to serum if the infection is left
untreated (158). Spontaneous clearance of
asymptomatic bacteriuria may be preceded by
an increase in the serum susceptibility of the
urinary pathogen (224).
Complement-mediated bactericidal mecha-
nisms may also be important in other types of
infection. The overwhelming majority of entero-
bacteria causing bovine mastitis were found to
be serum resistant (34). Isolates of Bacteroides
spp. from feces were significantly more serum
susceptible than isolates from patients with clini-
cal infections, including bacteremia (37). Strains
of Bacteroides fragilis, the predominant species
found in serious infections, formd a group char-
acterized by a high degree of serum resistance.
Waisbren and Brown (327) recorded specific
defects in the ability of patients' sera to kill the
infecting organism in a variety of clinical situa-
tions that included Pseudomonas pneumonia,
chronic pyelonephritis, endocarditis, peritonitis,
and septicemia; sera from similar patients were
subsequently shown to contain IgG directed
against bacterial surface components which
73
could block complement-mediated killing mech-
anisms (110). It is tempting to speculate that
such blocking mechanisms are responsible for
the survival of genotypically serum-susceptible
bacteria in tissues from which they would nor-
mally be swiftly eradicated by the host's defense
mechanisms.
Susceptibility to the serum bactericidal sys-
tem seems to be of primary importance in deter-
mining the pathogenesis of infections due to
Neisseria sp. Thus, gonococci causing localized
genital infections are generally susceptible to
human serum (65), whereas they normally in-
vade the blood system and produce disseminat-
ed gonococcal infection only if they are resistant
to killing by serum (65, 267, 285). Deficiency of
antibody does not account for inability to kill
disseminated gonococcal infection pathogens
(45) but in some cases is related to IgG blocking
mechanisms (175). Similarly, IgA-mediated bac-
tericidal blocking mechanisms (113) may aid
dissemination of N. meningitidis in the early
stages of systemic meningococcal disease (104).
Establishment of localized gonococcal infection
in the genital tract by serum-susceptible strains
might be promoted by inhibition of complement-
mediated bactericidal activity due to proteases
in seminal plasma (233). Similarly, absence of
adequate levels of antibody and complement in
cerebrospinal fluids may contribute to the main-
tenance of high concentrations of bacteria in
untreated cases of meningitis (278).
Perhaps the most convincing evidence in fa-
vor of a role for bactericidal activity in these
infections is afforded by observations on pa-
tients with total or near-total congenital deficien-
cies of a single complement protein. The critical
role of C3 in a variety of immune reactions
beneficial to the host is indicated by a general-
ized susceptibility to microbial infection in indi-
viduals deficient in C3 (6, 13). Patients with
genetic deficiencies of the proteins of the MAC
have a propensity to infections with bacteria of
the Neisseria group. Thus, deficiencies of C5
(111, 232), C6 (14, 31, 111, 234), C7 (31, 155,
234), and C8 (31, 234) are associated with epi-
sodes of gonococcal or meningococcal bacter-
emia and meningitis. The observations imply
that the late-acting complement components are
necessary for normal host defense against patho-
genic Neisseria spp., particularly with regard to
impeding dissemination of serum-susceptible
bacteria. As the absence of these components
does not impair the opsonic and leukocyte che-
motactic functions of the complement system, a
direct role for serum bactericidal activity is
suggested by these observations, particularly as
phagocytic cells are often inefficient in ingesting
these organisms.
Individuals deficient in Cl, C4, and C2 do not
74 TAYLOR
appear to be unduly susceptible to infections
with gram-negative organisms; it is likely that in
such individuals the alternative pathway is re-
sponsible for preventing serious recurrent infec-
tions (13).
Evidence from Experimental Infections of
Laboratory Animals
A clear relationship has emerged between
serum resistance and the ability of enterobac-
teria to produce endocarditis in experimental
animals. Durack and Beeson (62) found that only
serum-resistant E. coli strains were consistently
able to cause infective endocarditis in rabbits
prepared by prior placement of an intracardiac
catheter. In contrast, all five serum-susceptible
strains tested caused endocarditis in C6-defi-
cient rabbits. Similar experimental models were
used to verify this relationship with P. aerugi-
nosa (12) and S. marcescens (109).
Evidence from animal challenge experiments
suggests a relationship between virulence and
serum resistance. For example, the ability of E.
coli strains to multiply in the peritoneum and
produce infection in mice after intraperitoneal
challenge was correlated to their ability to sur-
vive in in vitro serum bactericidal systems (258).
Serum-susceptible enterobacteria are cleared
from the blood of rabbits more rapidly than
resistant organisms and are recoverable from the
viscera 24 h after infection at only about 10% of
the concentration of resistant strains (250). A
similar relationship was established for E. coli
after intravenous injection into normal mice
(178). In peritoneal membrane chambers in guin-
ea pigs, serum-susceptible S. enteritidis cells
were killed, whereas serum-resistant ones sur-
vived (291). Caution should be exercised, how-
ever, when extrapolating these results to natural
infections in humans. Much of this work has
been performed in mice, which have a relatively
inefficient serum bactericidal system. Further-
more, Salmonella spp., often employed in such
studies, are predominantly intracellular para-
sites and may be concentrated in locations
where they are protected from complement-
mediated bactericidal systems.
In estrogen-pretreated rats, only serum-resis-
tant strains were consistently able to induce
lesions indicative of pyelonephritis; this rela-
tionship was not maintained when a rat model of
obstructive pyelonephritis was used (119). In an
experimental rat model, serum-susceptible
strains were able to produce kidney infection
only after the animals had been depleted of
complement with cobra venom factor (192).
CONCLUDING REMARKS
The weight of published evidence indicates
that resistance to the bactericidal action of anti-
MICROBIOL. REV.
body-complement systems is likely to play a role
in the pathogenesis of at least some infections
due to gram-negative bacteria. It can therefore
be inferred that the serum bactericidal system
has evolved partly in response to the need of the
host to efficiently eliminate invading bacteria
from potential sites of infection. In suitable in
vitro systems, these complement-mediated reac-
tions often proceed at an impressive rate, al-
though one feels that our understanding of se-
rum killing would progress in a more orderly
fashion if there was broad agreement on the
basic methodology of serum bactericidal assay.
Recent work has emphasized the essentially
multifactorial nature of serum resistance, and a
number of distinct cell envelope polymers, such
as outer membrane proteins, lipopolysaccha-
rides, and acidic exopolysaccharides, have been
implicated as mediators of resistance to activat-
ed complement components. The relative impor-
tance of these various structures is less clear, as
is the precise way in which some gram-negative
bacteria successfully circumvent the lethal
mechanism. It is to be hoped that future work
will shed some light on the biophysical proper-
ties of the outer membrane of resistant organ-
isms that will permit a description of serum
resistance in purely molecular terms.
ACKNOWLEDGMENT
I thank my wife, Jan, for help in preparing the manuscript
and for endless encouragement.
LITERATURE CITED
1. Achtman, M., N. Kennedy, and R. Skurray. 1977. Cell-
cell interactions in conjugating Escherichia coli: role of
traT protein in surface exclusion. Proc. Natl. Acad. Sci.
U.S.A. 74:5104-5108.
2. Acquatella, H., P. J. Little, H. E. deWardener, and J. C.
Coleman. 1967. The effect of urine osmolality and pH on
the bactericidal activity of plasma. Clin. Sci. 33:471-480.
3. Adinolfi, M., A. A. Glynn, M. Lindsay, and C. M. Milne.
1966. Serological properties of yA antibodies to Esche-
richia coli present in human colostrum. Immunology
10:517-526.
4. Akiyama, Y., and K. Inoue. 1977. Isolation and proper-
ties of complement-resistant strains of Escherichia coli
K-12. Infect. Immun. 18:446-453.
5. Allen, R. J., and G. K. Scott. 1981. Human serum
complement requirements for bacterial killing and proto-
plast lysis of Escherichia coli ML308 225. J. Gen.
Microbiol. 123:179-181.
6. Alper, C. A., N. Abramson, R. B. Johnston, J. H. Jandl,
and F. S. Rosen. 1970. Increased susceptibility to infec-
tion associated with abnormalities of complement-medi-
ated functions and of the third component of complement
(C3). N. Engl. J. Med. 282:349-354.
7. Amano, T., K. Fujikawa, T. Morioka, A. Miyama, and S.
Ichikawa. 1958. Quantitative studies on immune bacteri-
olysis. I. A new method of quantitative estimation. Biken
J. 1:13-25.
8. Amano, T., K. Kinjo, M. Nishimoto, K. Inoue, and H.
Yachiku. 1955. Studies on the immune bacteriolysis. VI.
Cause of the death of bacteria by immune bacteriolysis.
Med. J. Osaka Univ. 6:57-66.
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
9. Amano, T., T. Morioka, Y. Seki, S. Kashiba, K. Fu-
jikawa, and S. Ichikawa. 1955. Studies on the immune
bacteriolysis. VIII. The mechanisms of the immune
bacteriolysis. Med. J. Osaka Univ. 6:709-716.
10. Amano, T., M. Nishimoto, K. Kinjo, K. Inoue, S. Inai, Y.
Seki, and S. Kashiba. 1954. Studies on the immune
bacteriolysis. III. Influence of immune bacteriolysis
upon bacterial respiration. Med. J. Osaka Univ. 5:145-
165.
11. Andriole, V. T., and F. H. Epstein. 1965. Prevention of
pyelonephritis by water diuresis: evidence for the role of
medullary hypertonicity in promoting renal infection. J.
Clin. Invest. 44:73-79.
12. Archer, G., and F. R. Fekety. 1976. Experimental endo-
carditis due to Pseudomonas aeruginosa. I. Description
of a model. J. Infect. Dis. 134:1-7.
13. Atkinson, J. P., and M. M. Frank. 1980. Complement, p.
219-271. In C. W. Parker (ed.), Clinical immunology.
The W. B. Saunders Co., Philadelphia.
14. Avril, M. F., J. Lebas, B. Autran, and J. Modai. 1981.
Meningite a meningocoques recidivante associee a un
deficit en 6e composant du complement. Nouv. Presse
Med. 10:3057-3059.
15. Badar, J., and M. Teuber. 1973. Action of polymyxin B
on bacterial membranes. I. Binding to 0-antigen lipo-
polysaccharide of Salmonella typhimurium. Z. Natur-
forsch. 28:422-430.
16. Bayer, M. E. 1975. Role of adhesion zones in bacterial
cell-surface function and biogenesis, p. 393-427. In A.
Tzagoloff (ed.), Membrane biogenesis. Plenum Publish-
ing Corp., New York.
17. Beckerdite-Quagliata, S., M. Simberkoff, and P. Elsbach.
1975. Effects of human and rabbit serum on viability,
permeability, and envelope lipids of Serratia marces-
cens. Infect. Immun. 11:758-766.
18. Beeson, P. B., and D. Rowley. 1959. The anticomplemen-
tary effect of kidney tissue. Its association with ammonia
production. J. Exp. Med. 110:685-697.
19. Betz, S. J., and H. Isleker. 1981. Antibody-independent
interactions between Escherichia coli J5 and human
complement components. J. Immunol. 127:1748-1754.
20. Betz, S. J., N. Page, C. Estrade, and H. Isleker. 1982. The
effect of specific antibody on antibody independent inter-
actions between E. coli J5 and human complement. J.
Immunol. 128:707-711.
21. Bhakdi, S. 1980. On the molecular nature of the comple-
ment lesion. Behring Inst. Mitt. 65:1-15.
22. Bhakdi, S., 0. J. Bjerrum, U. Rother, H. Knufermann,
and D. F. H. Wallach. 1975. Immunochemical analyses
of membrane-bound complement. Detection of the termi-
nal complement complex and its similarity to "intrinsic"
erythrocyte membrane proteins. Biochim. Biophys. Acta
406:21-35.
23. Bhakdi, S., and J. Tranum-Jensen. 1978. Molecular na-
ture of the complement lesion. Proc. Natl. Acad. Sci.
U.S.A. 75:5655-5659.
24. Binns, M. M., D. L. Davis, and K. G. Hardy. 1979.
Cloned fragments of the plasmid ColV,I-K94 specifying
virulence and serum resistance. Nature (London)
279:778-781.
25. Binns, M. M., J. Mayden, and R. P. Levine. 1982.
Further characterization of complement resistance con-
ferred on Escherichia coli by the plasmid genes traT of
R100 and iss of ColV,I-K94. Infect. Immun. 35:654-659.
26. Bjorkstin, B., R. Bortolussi, L. Gothefors, and P. G.
Quie. 1976. Interaction of E. coli strains with human
serum: lack of relationship to Kl antigen. J. Pediatr.
89:892-897.
27. Bjorksten, B., and B. Kaiser. 1978. Interaction of human
serum and neutrophils with Escherichia coli strains:
differences between strains isolated from urine of pa-
tients with pyelonephritis or asymptomatic bacteriuria.
Infect. Immun. 22:308-311.
28. Bjornson, A. B., and J. G. Michael. 1970. Biological
activities of rabbit immunoglobulin M and immunoglob-
75
ulin G antibodies to Pseudomonas aeruginosa. Infect.
Immun. 2:453-461.
29. Blachman, U., W. R. Clark, and M. J. Pickett. 1973.
Radiobacteriolysis: a new technique using chromium-51
for assaying anti-Vibrio cholerae antibodies. Infect. Im-
mun. 7:53-61.
30. Borsos, T., and H. J. Rapp. 1965. Complement fixation
on cell surfaces by 19S and 7S antibodies. Science
105:505-506.
31. Brooks, G. F., I. Ingwer, and B. H. Petersen. 1978.
Importance of serum bactericidal and opsonic activity
and complement in Neisseria gonorrhoeae infections, p.
199-203. In G. F. Brooks, E. C. Gotschlich, K. K.
Holmes, W. D. Sawyer, and F. E. Young (ed.), Immun-
obiology of Neisseria gonorrhoeae. American Society
for Microbiology, Washington, D.C.
32. CabeDo, F. C. 1979. Determinants of pathogenicity of E.
coli Kl, p. 155-160. In K. N. Timmis and A. Puhler
(ed.), Plasmids of medical, environmental and commer-
cial importance. Elsevier/North-Holland Publishing Co.,
Amsterdam.
33. CabeDo, F., and K. N. Timmis. 1979. Plasmids of medical
importance, p. 55-69. In K. N. Timmis and A. Puhler
(ed.), Plasmids of medical, environmental and commer-
cial importance. Elsevier/North-Holland Publishing Co.,
Amsterdam.
34. Carroll, E. J., and D. E. Jasper. 1977. Bactericidal
activity of standard bovine serum against coliform bacte-
ria isolated from udders and the environment of dairy
cows. Am. J. Vet. Res. 38:2019-2022.
35. Carroll, S. F., and R. J. Martinez. 1981. Antibacterial
peptide from normal rabbit-serum. 1. Isolation from
whole serum, activity, and microbicidal spectrum. Bio-
chemistry 20:5973-5981.
36. Carroll, S. F., and R. J. Martinez. 1981. Antibacterial
peptide from normal rabbit serum. 2. Compositional
microanalysis. Biochemistry 20:5981-5987.
37. Casciato, D. A., J. E. Rosenblatt, R. Bluestone, L. S.
Goldberg, and S. M. Finegold. 1979. Susceptibility of
isolates of Bacteroides to the bactericidal activity of
normal human serum. J. Infect. Dis. 140:109-113.
38. Ceppellini, R., and M. Landy. 1963. Suppression of blood
group agglutinability of human erythrocytes by certain
bacterial polysaccharides. J. Exp. Med. 117:321-338.
39. Chedid, L., M. Parant, F. Parant, and F. Boyer. 1968. A
proposed mechanism for natural immunity to enterobac-
terial pathogens. J. Immunol. 100:292-301.
40. Cheng, S., J. K. Thomas, and C. F. Kulpa. 1974. Dynam-
ics of pyrene fluorescence in Escherichia coli membrane
vesicles. Biochemistry 13:1135-1139.
41. Chernew, I., and A. I. Braude. 1962. Depression of
phagocytosis by solutes in concentrations found in the
kidney and urine. J. Clin. Invest. 41:1945-1953.
42. Chester, I. R., and P. M. Meadow. 1975. Heterogeneity
of the lipopolysaccharide from Pseudomonas aerugi-
nosa. Eur. J. Biochem. 58:273-282.
43. Clas, F., and M. Loos. 1980. Killing of the S and Re forms
of Salmonella minnesota via the classical pathway of
complement activation in guinea-pig and human serum.
Immunology 40:547-556.
44. Cobbs, C. G., and D. Kaye. 1967. Antibacterial mecha-
nisms in the urinary bladder. Yale J. Biol. Med. 40:93-
108.
45. Cohen, I. R., and L. C. Norins. 1966. Natural human
antibodies to gram-negative bacteria: immunoglobulins
G, A, and M. Science 152:1257-1259.
46. Collins, F. M. 1964. The effect of the growth rate on the
composition of S. enteritidis cell walls. Aust. J. Exp.
Biol. Med. Sci. 42:255-262.
47. Collins, F. M. 1967. Serum mediated killing of three
group D salmonellas. J. Gen. Microbiol. 46:247-253.
48. Crombie, L. B., and L. H. Muschel. 1967. Quantitative
studies on spheroplast formation by the complement
system and lysozyme on gram-negative bacteria. Proc.
Soc. Exp. Biol. Med. 124:1029-1033.
76 TAYLOR
49. Cronan, J. E. 1979. Phospholipid synthesis and assem-
bly, p. 35-65. In M. Inouye (ed.), Bacterial outer mem-
branes: biogenesis and functions. Wiley-Interscience,
New York.
50. Cundiff, R. J., and H. R. Morgan. 1941. The inhibition of
the bactericidal power of human and animal sera by
antigenic substances obtained from organisms of the
typhoid-salmonella group. J. Immunol. 42:361-367.
51. Dahl, C. E., and R. P. Levine. 1978. Electron spin
resonance studies on interaction of complement proteins
with erythrocyte membranes. Proc. Natl. Acad. Sci.
U.S.A. 75:4930-4934.
52. Davies, D. L., F. R. Falkiner, and K. G. Hardy. 1981.
Colicin V production by clinical isolates of Escherichia
coli. Infect. Immun. 31:574-579.
53. Davis, S. D., E. S. Boatman, D. Gemsa, A. lannetta, and
R. J. Wedgwood. 1969. Biochemical and fine structural
changes induced in Escherichia coli by human serum.
Microbios 1:69-86.
54. Davis, S. D., and R. J. Wedgwood. 1965. Kinetics of the
bactericidal action of normal serum on gram-negative
bacteria. J. Immunol. 95:75-79.
55. DeMatteo, C. S., M. C. Hammer, A. L. Baltch, R. P.
Smith, N. T. Sutphen, and P. B. Michelsen. 1981. Sus-
ceptibility of Pseudomonas aeruginosa to serum bacteri-
cidal activity. A comparison of three methods with
clinical correlations. J. Lab. Clin. Med. 98:511-518.
56. Dlabac, V. 1968. The sensitivity of smooth and rough
mutants of Salmonella typhimurium to bactericidal and
bacteriolytic action of serum, lysozyme and to phagocy-
tosis. Folia Microbiol. (Prague) 13:439-449.
57. Dlabac, V. 1969. On the significance of the lipopolysac-
charide structure of Salmonella mutants in different
immune reactions. Colloq. Int. CNRS 174:297-305.
58. Doggett, R. G., G. M. Harrison, and E. S. Wallis. 1964.
Comparison of some properties of Pseudomonas aeru-
ginosa isolated from infections of persons with and
without cystic fibrosis. J. Bacteriol. 87:427-431.
59. Domingue, G. J., and E. Neter. 1966. Opsonizing and
bactericidal activity of antibodies against common anti-
gen of Enterobacteriaceae. J. Bacteriol. 91:129-133.
60. Donaldson, D. M. 1973. ,3-Lysin and host resistance, p.
316-322. In W. Braun and J. Ungar (ed.), Nonspecific
factors influencing host resistance. S. Karger, Basel.
61. Donaldson, D. M., R. R. Roberts, H. S. Larsen, and J. G.
Tew. 1974. Interrelationship between serum beta-lysin,
lysozyme, and the antibody-complement system in kill-
ing Escherichia coli. Infect. Immun. 10:657-666.
62. Durack, D. T., and P. B. Beeson. 1977. Protective role of
complement in experimental Escherichia coli endocardi-
tis. Infect. Immun. 16:213-217.
63. Eddie, D. S., M. L. Schulkind, and J. B. Robbins. 1971.
The isolation and biologic activities of purified secretory
IgA and IgG anti-Salmonella typhimurium "O" antibod-
ies from rabbit intestinal fluid and colostrum. J. Im-
munol. 106:181-190.
64. Eisen, H. N., and G. W. Siskind. 1964. Variations in
affinities of antibodies during the immune response.
Biochemistry 3:996-1007.
65. Eisenstein, B. I., T. J. Lee, and P. F. Sparling. 1977.
Penicillin sensitivity and serum resistance are indepen-
dent attributes of strains of Neisseria gonorrhoeae caus-
ing disseminated gonococcal infection. Infect. Immun.
15:834-841.
66. Elgefors, B., and S. Oiing. 1978. The significance of
serum-sensitive bacilli in gram-negative bacteremia.
Scand. J. Infect. Dis. 10:203-207.
67. Esser, A. F., R. M. Bartholemew, F. C. Jensen, and H. J.
Muller-Eberhard. 1979. Disassembly of viral membranes
by complement independent of channel formation. Proc.
Natl. Acad. Sci. U.S.A. 76:5843-5847.
68. Esser, A. F., W. P. Kolb, E. R. Podack, and H. J. Muller-
Eberhard. 1979. Molecular reorganization of lipid bi-
layers by complement: a possible mechanism for mem-
branolysis. Proc. Natl. Acad. Sci. U.S.A. 76:1410-1414.
MICROBIOL. REV.
69. Feingold, D. S. 1969. The serum bactericidal reaction.
IV. Phenotypic conversion of Escherichia coli from
serum-resistance to serum-sensitivity by diphenylamine.
J. Infect. Dis. 120:437-444.
70. Feingold, D. S., J. N. Goldman, and H. M. Kuritz. 1968.
Locus of the action of serum and the role of lysozyme in
the serum bactericidal reaction. J. Bacteriol. 96:2118-
2126.
71. Feingold, D. S., J. N. Goldman, and H. M. Kuritz. 1968.
Locus of the lethal event in the serum bactericidal
reaction. J. Bacteriol. 96:2127-2131.
72. Feingold, D. S., C. C. HsuChen, and I. J. Sud. 1974.
Basis for the selectivity of action of the polymyxin
antibiotics on cell membranes. Ann. N.Y. Acad. Sci.
235:480-492.
73. Ferrante, A., and Y. H. Thong. 1982. Inhibition of serum
bactericidal activity by bilirubin. Immunol. Lett. 4:103-
105.
74. Fields, K. L., and S. E. Luria. 1969. Effects of colicins El
and K on transport systems. J. Bacteriol. 97:57-63.
75. Fierer, J., and F. Finley. 1979. Lethal effect of comple-
ment and lysozyme on polymyxin-treated, serum resis-
tant gram-negative bacteria. J. Infect. Dis. 140:581-589.
76. Fierer, J., and F. Finley. 1979. Deficient serum bacteri-
cidal activity against Escherichia coli in patients with
cirrhosis of the liver. J. Clin. Invest. 63:912-921.
77. Fierer, J., F. Finley, and A. I. Braude. 1972. A plaque
assay on agar for detection of gram-negative bacilli
sensitive to complement. J. Immunol. 109:1156-1158.
78. Fierer, J., F. Finley, and A. I. Braude. 1974. Release of
"Cr-endotoxin from bacteria as an assay of serum bacte-
ricidal activity. J. Immunol. 112:2184-2192.
79. Fietta, A., E. Romero, and A. G. Siccardi. 1977. Effect of
some R factors on the sensitivity of rough Enterobacteri-
aceae to human serum. Infect. Immun. 18:278-282.
80. Fogelson, S. J., A. Ross, and 0. E. Lobstein. 1954. A
method for the quantitative determination of lysozyme
activity in blood. Am. J. Dig. Dis. 21:327-331.
81. Fujita, T. 1979. The activation mechanism of human
complement system by immune precipitate formed with
rabbit IgG antibody. Microbiol. Immunol. 23:1023-1031.
82. Fust, G., G. Petris, and E. Ujhelyi. 1976. Activation of
the complement system during infections due to gram-
negative bacteria. Clin. Immunol. Immunopathol. 5:293-
302.
83. Gabay, J. 1977. Specific killing by complement of Esche-
richia coli K12 strains carrying the receptor protein for
phage lambda. FEMS Microbiol. Lett. 2:61-63.
84. Galanos, C., and 0. Luderitz. 1976. The role of the
physical state of lipopolysaccharides in the interaction
with complement. High molecular weight as prerequisite
for the expression of anticomplementary activity. Eur. J.
Biochem. 65:403-408.
85. Galanos, C., 0. Luderitz, and 0. Westphal. 1969. A new
method for the extraction of R lipopolysaccharides. Eur.
J. Biochem. 9:245-249.
86. Gemsa, D., S. D. Davis, and R. J. Wedgwood. 1966.
Lysozyme and serum bactericidal action. Nature (Lon-
don) 210:950-951.
87. Gemski, P., A. S. Cross, and J. C. Sadoff. 1980. Kl
antigen-associated resistance to the bactericidal activity
of serum. FEMS Microbiol. Lett. 9:193-197.
88. Gewurz, H., H. S. Shin, and S. E. Mergenhagen. 1968.
Interactions of the complement system with endotoxic
lipopolysaccharide: consumption of each of the six ter-
minal complement components. J. Exp. Med. 128:1049-
1057.
89. Gewurz, H., and L. A. Suyehira. 1980. Complement, p.
163-174. In N. R. Rose and H. Friedman (ed.), Manual
of clinical immunology, 2nd ed. American Society for
Microbiology, Washington, D.C.
90. Giavedoni, E. B., R. P. Mason, and A. P. Dalmasso. 1978.
Complement-induced modifications in membrane fluid-
ity: studies with resealed and glutaraldehyde-treated
erythrocyte membrane ghosts. J. Immunol. 120:2003-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
2007.
91. Gitlin, D., F. S. Rosen, and J. G. Michael. 1963. Tran-
sient 19S gamma-globulin deficiency in the newborn
infant, and its significance. Pediatrics 31:197-208.
92. Glynn, A. A. 1969. The complement lysozyme sequence
in immune bacteriolysis. Immunology 16:463-471.
93. Glynn, A. A., W. Brumfitt, and C. J. Howard. 1971. K
antigens of Escherichia coli and renal involvement in
urinary tract infections. Lancet i:514-516.
94. Glynn, A. A., and C. J. Howard. 1970. The sensitivity to
complement of strains of Escherichia coli related to their
K antigens. Immunology 18:331-346.
95. Glynn, A. A., and C. M. Milne. 1967. A kinetic study of
the bacteriolytic and bactericidal action of human serum.
Immunology 12:639-653.
96. Glynn, A. A., and M. E. Ward. 1970. Nature and
heterogeneity of the antigens of Neisseria gonorrhoeae
involved in the serum bactericidal reactions. Infect.
Immun. 2:162-168.
97. Goldman, J. N., and K. F. Austen. 1974. Reaction
mechanisms of nascent C567 (reactive l! js). II. Killing
of a rough form of Escherichia coli by C567, C8 and C9.
J. Infect. Dis. 129:444-450.
98. Goldman, J. N., S. Ruddy, K. F. Austen, and D. S.
Feingold. 1969. The serum bactericidal reaction. III.
Antibody and complement requirements for killing a
rough Escherichia coli. J. Immunol. 102:1379-1387.
99. Goldman, R. C., and L. Leive. 1980. Heterogeneity of
antigenic-side-chain length in lipopolysaccharide from
Escherichia coli 0111 and Salmonella typhimurium LT2.
Eur. J. Biochem. 107:145-153.
100. Goldkchneider, I., E. C. Gotschlich, and M. S. Arten-
stein. 1969. Human immunity to the meningococcus. II.
Development of natural immunity. J. Exp. Med.
129:1327-1348.
101. Gordon,J., H. R. Whitehead, and A. Wormall. 1926. The
action of ammonia on complement. The fourth compo-
nent. Biochem. J. 20:1028-1035.
102. G6tze, O., and H. J. Mfiler-Eberhard. 1971. The C3-
activator system: an alternate pathway of complement
activation. J. Exp. Med. (Suppl.) 134:90-108.
103. Gower, P. E., P. W. Taylor, K. G. Koutsaimanis, and
A. P. Roberts. 1972. Serum bactericidal activity in pa-
tients with upper and lower urinary tract infections. Clin.
Sci. 43:13-22.
104. Griffis, J. M., and M. A. Bertram. 1977. Immunoepide-
miology of meningococcal disease in military recruits. II.
Blocking of serum bactericidal activity by circulating
IgA early in the course of invasive disease. J. Infect. Dis.
136:733-739.
105. Griffiths, E. 1971. Mechanism of action of specific anti-
serum on Pasteurella septica. Selective inhibition of net
macromolecular synthesis and its reversal by iron com-
pounds. Eur. J. Biochem. 23:69-76.
106. Griffiths, E. 1974. Metabolically controlled killing of
Pasteurella septica by antibody and complement. Bio-
chim. Biophys. Acta 362:598-602.
107. Grineberg, R. N., D. A. Leigh, and W. Brumfitt. 1968.
Escherichia coli serotypes in urinary tract infection:
studies in domiciliary, antenatal and hospital practice, p.
68-79. In F. W.O'Grady and W. Brumfitt (ed.), Urinary
tract infections. Oxford UniversityPress, London.
108. Gupta, A. K., and K. M. Rio. 1981. A simple bactericidal
antibody test for sero-diagnosis of typhoid fever. J.
Immunol. Methods 42:121-125.
109. Gutschik, E., R. S. Norwood, S. Moller, and S. Olling.
1980. Experimental endocarditis in rabbits. 4. Experi-
ments with Serratia marcescens: on the significance of
serum susceptibility and proteolytic capacity of the
strains and the influence of an indwelling catheter. Acta
Pathol. Microbiol. Scand. Sect. B 88:269-276.
110. Guttman, R. M., and B. A. Waisbren. 1975. Bacterial
blocking activity of specific IgG in chronic Pseudomonas
aeruginosa infection. Clin. Exp. Immunol. 19:121-130.
111. Haeney, M. R., R. A. Thompson,J. Faulkner, P. Mackin-
77
tosh, and A. P. Bali. 1980. Recurrent bacterial meningitis
in patients with genetic defects of terminal complement
components. Clin. Exp. Immunol. 40:16-24.
112. Haimovich, J., and M. Sela. 1969. Inactivation of bacte-
riophage T4, of poly-O-alanyl bacteriophage and of peni-
cilloyl bacteriophage by immunospecifically isolated
IgM and IgG antibodies. J. Immunol. 103:45-55.
113. Hall, W. H., R. E. Manion, and H. E. Zinneman. 1971.
Blocking serum lysis of Brucella abortus by hyperim-
mune rabbit immunoglobulin A. J. Immunol. 107:41-46.
114. Hand, W. L., and N. L. King. 1977. Deficiency of serum
bactericidal activity against Salmonella typhimurium in
sickle cell anaemia. Clin. Exp. Immunol. 30:262-270.
115. Harriman, G. R., A. F. Esser, E. R. Podack, A. N.
Wunderlich, A. I. Braude, T. F. Lint, and J. C. Curd.
1981. The role of C9 in complement-mediated killing of
Neisseria. J. Immunol. 127:2386-2390.
116. Heddle, R. J., J. Knop, E. J. Steele, and D. Rowley. 1975.
The effect of lysozyme on the complement-dependent
bactericidal action of different antibody classes. Immu-
nology 28:1061-1066.
117. Heddle, R. J., and D. Rowley. 1975. Dog immunoglob-
ulins. II. The antibacterial properties of dog IgA, IgM
and IgG antibodies to Vibrio cholerae. Immunology
29:197-208.
118. Henkel, W. 1971. Serum resistenz und "Nephropatho-
genitait" antigenetisch definierter Stiimme von E. coli. II.
Verhalten unterschiedlich serumresistenter Keime in vi-
tro. Z. Med. Mikrobiol. Immunol. 156:284-296.
119. Henkel, W., and R. Commichau. 1971. Serumresistenz
und "Nephropathogenitat" antigenisch definierter
Stamme von E. coli. III. Experimentelle Pyelonephritis
mit unterschiedlich serumresistenten Keimen. Z. Med.
Mikrobiol. Immunol. 156:297-304.
120. Hildebrandt, J. F., L. W. Mayer, S. P. Wang, and T. M.
Buchanan. 1978. Neisseria gonorrhoeae acquire a new
principal outer-membrane protein when transformed to
resistance to serum bactericidal activity. Infect. Immun.
20:267-273.
121. Hirsch,J. G. 1960. Comparative bactericidal activities.of
blood serum and plasma serum. J. Exp. Med. 112:15-22.
122. H%iby, N., and S. Oiling. 1977. Pseudomonas aeruginosa
infection in cystic fibrosis. Bactericidal effect of serum
from normal individuals and patients with cystic fibrosis
or P. aeruginosa strains from patients with cystic fibro-
sis or other diseases. Acta Pathol. Microbiol. Scand.
Sect. C 85:107-114.
123. Howard, C.J., and A. A. Glynn. 1971. The virulence for
mice of strains of Escherichia coli related to the effects of
K antigens on their resistance to phagocytosis and killing
by complement. Immunology 20:767-777.
124. Howe, C. W. S., R. Morriset, and W. W. Spink. 1973.
Specific and non-specific factors influencing the serum
bactericidal reaction. A family-specific defect of canine
species for Brucella canis, p. 358-370.In W. Braun and
J. Ungar (ed.), Non-specific factors influencing host
resistance. S. Karger, Basel.
125. Hughes, C., R. Phillips, and A. P. Roberts. 1982. Serum
resistance among Escherichia coli strains causing- uri-
nary tract infection in relation to 0 type and the carriage
of hemolysin, colicin, and antibiotic resistance determi-
nants. Infect. Immun. 35:270-275.
126. Humphrey,J. H., R. R. Dourmashkin, and S. N. Payne.
1967. The nature of lesions in cell membranes produced
by action of C' and antibody, p. 209-220. In P. A.
Miescher and P. Grabar (ed.), Mechanisms of inflama-
tion induced by immune reactions. Schwabe, Basel.
127. Inoue, K. 1972. Immune bacteriolytic and bactericidal
reactions, p. 177-222. In J. B. G. Kwapinski (ed.), Re-
search in immunochemistry and immunobiology, vol. 1.
University ParkPress, Baltimore.
128. Inoue, K., Y. Akiyama, T. Kinoshita, Y. Higashi, and T.
Amano. 1976. Evidence for a one-hit theory in the
immune bactericidal reaction and demonstration of a
multi-hit response for hemolysis by streptolysin 0 and
78 TAYLOR
Clostridium perfringens theta-toxin. Infect. Immun.
13:337-344.
129. Inoue, K., T. Kinoshita, M. Okada, and Y. Akiyama.
1977. Release of phospholipids from complement-medi-
ated lesions on the surface structure of Escherichia coli.
J. Immunol. 119:65-72.
130. Inoue, K., A. Takamizawa, T. Kurimura, and K. Yone-
masu. 1968. Studies on the immune bacteriolysis. XIII.
Leakage of enzymes from Escherichia coli during im-
mune bacteriolysis. Biken J. 11:193-201.
131. Inoue, K., A. Takamizawa, K.-I. Yano, and T. Amano.
1974. Chemical studies on damages of Escherichia coli
by the immune bactericidal reaction. I. Release and
degradation of phospholipids from damaged bacteria.
Biken J. 17:127-134.
132. Inoue, K., Y. Tanigawa, M. Takubo, M. Satani, and T.
Amano. 1959. Quantitative studies on immune bacteriol-
ysis. II. The role of lysozyme in immune bacteriolysis.
Biken J. 2:1-20.
133. Inoue, K., K.-I. Yano, and T. Amano. 1974. Chemical
studies on damages of Escherichia coi by the immune
bactericidal reaction. II. Release of phosphatidylethanol-
amine from a phospholipase A-deficient mutant of E. coli
during the immune bactericidal reaction. Biken J.
17:135-140.
134. Inoue, K., K. Yonemasu, A. Takamizawa, and T. Amano.
1968. Studies on the immune bacteriolysis. XIV. Re-
quirement of all nine components of complement for
immune bacteriolysis. Biken J. 11:203-206.
135. Jain, N. C., and J. Lasmanis. 1978. Phagocytosis of
serum-resistant and serum-sensitive coliform bacteria
(Klebsiella) by bovine neutrophils from blood and mas-
titic milk. Am. J. Vet. Res. 39:425-427.
136. Jann, B., K. Jann, G. Schmidt, I. 0rskov, and F. 0rskov.
1970. Immunochemical studies of polysaccharide sur-
face antigens of Escherichia coli 0100:K?(B):H2. Eur. J.
Biochem. 15:29-39.
137. Jann, K., and 0. Westphal. 1975. Microbial polysaccha-
rides, p. 1-125. In M. Sela (ed.), The antigens, vol. 3.
Academic Press, Inc., New York.
138. Johnson, A. P., M. F. Osborn, and D. Taylor-Robinson.
1978. Iron-mediated protection of Neisseria gonor-
rhoeae against killing by normal human serum. FEMS
Microbiol. Lett. 4:225-227.
139. Joiner, K. A., C. H. Hammer, E. J. Brown, R. J. Cole,
and M. M. Frank. 1982. Studies on the mechanism of
bacterial resistance to complement-mediated killing. I.
Terminal complement components are deposited and
released from Salmonella minnesota S218 without caus-
ing bacterial death. J. Exp. Med. 155:797-808.
140. Joiner, K. A., C. H. Hammer, E. J. Brown, and M. M.
Frank. 1982. Studies on the mechanism of bacterial
resistance to complement-mediated killing. II. C8 and C9
release C5b67 from the surface of Salmonella minnesota
S218 because the terminal complex does not insert into
the bacterial outer membrane. J. Exp. Med. 155:809-
819.
141. Jones, E. A., N. Crowley, and S. Sherlock. 1967. Bacter-
aemia in association with hepatocellular and hepatobi-
liary disease. Postgrad. Med. J. Suppi. 43:7-11.
142. KaUser, B., L. A. Hanson, U. Jodal, G. Lidin-Janson, and
J. B. Robbins. 1977. Frequency of E. coli K antigens in
urinary-tract infections in children. Lancet i:663-664.
143. Kalmanson, G. M., E. Herbert, and L. B. Guze. 1964.
Serum bactericidal activity in patients with pyelonephri-
tis. Am. J. Med. Sci. 248:285-289.
144. Kato, K., and Y. Bito. 1978. Relationship between bacte-
ricidal action of complement and fluidity of cellular
membranes. Infect. Immun. 19:12-17.
145. Kimball, H., M. Garcia, and R. G. Petersdorf. 1964. The
epidemiology of nonenteric Escherichia coli infections.
II. Relationship of prevalence of E. coli in urinary
infections to the bactericidal effect of human serum. J.
Lab. Clin. Med. 63:901-906.
146. King, B. F., M. L. Biel, and B. J. Wilkinson. 1980. Facile
MICROBIOL. REV.
penetration of the Staphylococcus aureus capsule by
lysostaphin. Infect. Immun. 29:892-896.
147. King, B. F., and B. J. Wilkinson. 1981. Binding of human
immunoglobulin G to protein A in encapsulated Staphy-
lococcus aureus. Infect. Immun. 33:666-672.
148. Kreutzer, D. L., M. Vandermaten, C. L. Buller, D. C.
Robertson, and A. A. Hirata. 1977. Role of bacterial
phospholipases in serum-mediated killing of Escherichia
coli. Infect. Immun. 18:183-188.
149. Krishnapillai, V. 1971. Uridinediphosphogalactose-4-
epimerase deficiency in Salmonella typhimurium and its
correction by plasmid-borne galactose genes of Esche-
richia coli K-12: effects on mouse virulence, phagocyto-
sis, and serum sensitivity. Infect. Immun. 4:177-188.
150. Kunin, C. M. 1962. Antibody distribution against nonen-
teropathogenic E. coli. Arch. Intern. Med. 110:676-686.
151. Lachmann, P. J. 1979. Complement, p. 283-335. In M.
Sela (ed.), The antigens, vol. 5. Academic Press, Inc.,
New York.
152. Lachman, P. J., and R. A. Thompson. 1970. Reactive
lysis: the complement-mediated lysis of unsensitized
cells. II. The characterization of activated reactor as C36
and the participation of C8 and C9. J. Exp. Med.
131:643-657.
153. Landy, M., R. P. Sanderson, and A. L. Jackson. 1965.
Humoral and cellular aspects of the immune response to
the somatic antigen of Salmonella enteritidis. J. Exp.
Med. 122:483-504.
154. Lauf, P. K. 1975. Immunological and physiological char-
acteristics of the rapid immune hemolysis of neuramini-
dase-treated sheep red cells produced by fresh guinea pig
serum. J. Exp. Med. 142:974-988.
155. Lee, T. J., P. D. Utsinger, R. Snyderman, W. J. Yount,
and P. F. Sparling. 1978. Familial deficiency of the
seventh component of complement associated with re-
current bacteremic infections due to Neisseria. J. Infect.
Dis. 138:359-368.
156. Leive, L. 1965. Release of lipopolysaccharide by EDTA
treatment of E. coli. Biochem. Biophys. Res. Commun.
21:290-296.
157. Li, C. K. N., and R. P. Levine. 1980. Molecular transport
via the functional complement lesion. Mol. Immunol.
17:1465-1474.
158. Lindberg, U., L. A. Hanson, U. Jodal, G. Lidin-Janson,
K. Lincoln, and S. Oiling. 1975. Asymptomatic bacteri-
uria in schoolgirls. II. Differences in Escherichia coli
causing asymptomatic and symptomatic bacteriuria.
Acta Paediatr. Scand. 64:432-436.
159. Linker, A., and R. S. Jones. 1966. A new polysaccharide
resembling alginic acid isolated from pseudomonads. J.
Biol. Chem. 241:3845-3851.
160. Lint, T. F., H. J. Zeitz, and H. Gewurz. 1980. Inherited
deficiency of the ninth component of complement in
man. J. Immunol. 125:2252-2257.
161. Loos, M., B. Wellek, R. Thesen, and W. Opferkuch.
1978. Antibody-independent interaction of the first com-
ponent of complement with gram-negative bacteria. In-
fect. Immun. 22:5-9.
162. Lyman, M. B., J. P. Steward, and R. J. Roantree. 1976.
Characterization of the virulence and antigenic structure
of Salmonella typhimurium strains with lipopolysaccha-
ride core defects. Infect. Immun. 13:1539-1542.
163. Maal$e, 0. 1948. Pathogenic-apathogenic transformation
of Salmonella typhimurium. Acta Pathol. Microbiol.
Scand. 25:414-430.
164. Maal$e, 0. 1948. Pathogenic-apathogenic transformation
of Salmonella typhimurium. II. Induced change of resist-
ance to complement. Acta Pathol. Microbiol. Scand.
25:755-766.
165. Maas, W. K. 1981. Genetics of bacterial virulence.
Symp. Soc. Gen. Microbiol. 31:341-360.
166. Madonna, G. S., and R. C. Allen. 1981. Shigella sonnei
phase I and phase II: susceptibility to direct serum lysis
and opsonic requirements necessary for stimulation of
leukocyte redox metabolism and killing. Infect. Immun.
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
32:153-159.
167. Martinez, R. J., and S. F. Carroll. 1980. Sequential
metabolic expressions of the lethal process in human
serum-treated Escherichia coli: role of lysozyme. Infect.
Immun. 28:735-745.
168. Mason, T. G., and G. Richardson. 1981. Escherichia coli
and the human gut: some ecological considerations. J.
Appl. Bacteriol. 1:1-16.
169. Maw, J., and G. G. Meynell. 1968. The true division and
death rates of Salmonella typhimurium in the mouse
spleen determined with superinfecting phage P22. Br. J.
Exp. Pathol. 49:597-613.
170. Mayer, M. M. 1961. Complement and complement fixa-
tion, p. 133-240. In E. A. Kabat and M. M. Mayer (ed.),
Experimental immunochemistry, 2nd ed. Charles C
Thomas, Publisher, Springfield, Ill.
171. Mayer, M. M. 1981. Membrane damage by complement.
Johns Hopkins Med. J. 148:243-258.
172. Mayer, M. M., A. G. Osler, 0. T. Bier, and M. Heidel-
berger. 1946. The activating effect of magnesium and
other cations on the hemolytic function of complement.
J. Exp. Med. 84:535-548.
173. McCabe, W. R., and G. G. Jackson. 1962. Gram-negative
bacteremia. I. Etiology and ecology. Arch. Intern. Med.
110:847-855.
174. McCabe, W. R., B. Kaijser, S. Olling, M. Uwaydah, and
L. A. Hanson. 1978. Escherichia coli in bacteremia: K
and 0 antigens and serum sensitivity of strains from
adults and neonates. J. Infect. Dis. 138:33-41.
175. McCutchan, J. A., D. Katzenstein, D. Norquist, G. Chi-
kami, A. Wunderlich, and A. I. Braude. 1978. Role of
blocking antibody in disseminated gonococcal infection.
J. Immunol. 121:1884-1888.
176. McElree, H., J. Pitcher, and J. Arnwine. 1966. Transient-
ly acquired serum resistance by cell-grown Escherichia
coli. J. Infect. Dis. 116:231-237.
177. Medearis, D. M., B. M. Camitta, and E. C. Heath. 1968.
Cell wall composition and virulence in Escherichia coli.
J. Exp. Med. 128:399-414.
178. Medhurst, F. A., and A. A. Glynn. 1970. In vivo bacteri-
cidal activity of mouse complement against Escherichia
coli. Br. J. Exp. Pathol. 51:498-506.
179. Melching, L., and S. I. Vas. 1970. The effects of serum
components on the agglutination of gram-negative bacte-
ria. Can. J. Microbiol. 16:121-124.
180. Melching, L., and S. I. Vas. 1971. Effects of serum
components on gram-negative bacteria during bactericid-
al reactions. Infect. Immun. 3:107-115.
181. Meynell, G. G. 1959. Use of superinfecting phage for
estimating the division rate of lysogenic bacteria in
infected animals. J. Gen. Microbiol. 21:421-437.
182. Meynell, G. G., and T. V. Subbaiah. 1963. Antibacterial
mechanisms of the mouse gut. I. Kinetics of infection by
Salmonella typhimurium in normal and streptomycin-
treated mice studied with abortive transductants. Br. J.
Exp. Pathol. 44:197-208.
183. Michael, J. G., and W. Braun. 1958. Relationships
between bacterial resistance to serum and penicillin.
Proc. Soc. Exp. Biol. Med. 97:104-107.
184. Michael, J. G., and W. Braun. 1959. Modification of
bactericidal effects of human sera. Proc. Soc. Exp. Biol.
Med. 102:486-490.
185. Michael, J. G., and W. Braun. 1964. Analysis of sequen-
tial stages in serum bactericidal reactions. J. Bacteriol.
87:1067-1072.
186. Michael, J. G., and M. Landy. 1961. Endotoxic proper-
ties of gram-negative bacteria and their susceptibility to
the lethal effect of normal serum. J. Infect. Dis. 108:90-
94.
187. Michael, J. G., and F. S. Rosen. 1963. Association of
"natural" antibodies to gram-negative bacteria with the
Xl-macroglobulius. J. Exp. Med. 118:619-626.
188. Michael, J. G., J. L. Whitby, and M. Landy. 1961.
Increase in specific bactericidal antibodies after adminis-
tration of endotoxin. Nature (London) 191:296-297.
79
189. Michael, J. G., J. L. Whitby, and M. Landy. 1962.
Studies on natural antibodies to gram-negative bacteria.
J. Exp. Med. 115:131-146.
190. Miler, I., J. Vondraiek, and L. Hromadkova. 1979. The
bactericidal activity of sera of healthy neonates and of
newborns with hyperbilirubinaemia to Escherichia coli.
Folia Microbiol. (Prague) 24:143-152.
191. Miller, T. E., and J. D. K. North. 1974. Host response in
urinary tract infections (editorial). Kidney Int. 5:179-
186.
192. Miller, T. E., S. Phillips, and I. J. Simpson. 1978.
Complement-mediated immune mechanisms in renal in-
fection. II. Effect of decomplementation. Clin. Exp.
Immunol. 33:115-121.
193. Milner, K. C., J. A. Rudbach, and E. Ribi. 1971. General
characteristics, p. 1-65. In G. Weinbaum, S. Kadis, and
S. J. Ajl (ed.), Microbial toxins. IV. Bacterial endotox-
ins. Academic Press, Inc., New York.
194. Mittal, K. R., and D. G. Ingram. 1975. Factors involved
in bactericidal activity of sheep serum. Am. J. Vet. Res.
36:1183-1187.
195. Moll, A., F. Cabello, and K. N. Timmis. 1979. Rapid
assay for the determination of bacterial resistance to the
lethal activity of serum. FEMS Microbiol. Lett. 6:273-
276.
196. Moll, A., P. A. Manning, and K. N. Timmis. 1980.
Plasmid-determined resistance to serum bactericidal ac-
tivity: a major outer membrane protein, the traT gene
product, is responsible for plasmid-specified serum
resistance in Escherichia coli. Infect. Immun. 28:359-
367.
197. Montgomerie, J. Z., G. M. Kalmanson, and L. B. Guze.
1972. Leukocyte phagocytosis and serum bactericidal
activity in chronic renal failure. Am. J. Med. Sci.
264:385-393.
198. Morrison, D. C., and D. M. Jacobs. 1976. Inhibition of
lipopolysaccharide-initiated activation of serum comple-
ment by polymyxin B. Infect. Immun. 13:298-301.
199. Morrison, D. C., and L. F. Kline. 1977. Activation of the
classical and properdin-pathways of complement by bac-
terial lipopolysaccharides (LPS). J. Immunol. 118:362-
368.
200. Munn, C. B., E. E. Ishiguro, W. W. Kay, and T. J. Trust.
1982. Role of surface components in serum resistance of
virulent Aeromonas salmonicida. Infect. Immun.
36:1069-1075.
201. Muschel, L. H. 1960. Bactericidal activity of normal
serum against bacterial cultures. II. Activity against
Escherichia coli strains. Proc. Soc. Exp. Biol. Med.
103:632-636.
202. Muschel, L. H., L. A. Ahl, and L. S. Baron. 1968. Effect
of lysogeny on serum sensitivity. J. Bacteriol. 96:1912-
1914.
203. Muschel, L. H., R. H. Chamberlin, and E. Osawa. 1958.
Bactericidal activity of normal serum against bacterial
cultures. I. Activity against Salmonella typhi strains.
Proc. Soc. Exp. Biol. Med. 97:376-382.
204. Muschel, L. H., and J. E. Jackson. 1966. The reactivity of
serum against protoplasts and spheroplasts. J. Immunol.
97:46-51.
205. Muschel, L. H., and L. J. Larsen. 1970. Effect of hyper-
tonic sucrose upon the immune bactericidal reaction.
Infect. Immun. 1:51-55.
206. Muschel, L. H., and L. J. Larsen. 1970. The sensitivity of
smooth and rough gram-negative bacteria to the immune
bactericidal reaction. Proc. Soc. Exp. Biol. Med.
133:345-348.
207. Muschel, L. H., and K. Schmoker. 1966. Activity of
mitomycin C, other antibiotics, and serum against lyso-
genic bacteria. J. Bacteriol. 92:967-971.
208. Muschel, L. H., and H. P. Treffers. 1956. Quantitative
studies on the bactericidal actions of serum and comple-
ment. I. A rapid photometric growth assay for bactericid-
al activity. J. Immunol. 76:1-10.
209. Myers, D. E., B. A. D. Stocker, and R. J. Roantree. 1980.
80 TAYLOR
Mapping of genes determining penicillin-resistance and
serum-sensitivity in Salmonella enteritidis. J. Gen. Mi-
crobiol. 118:367-376.
210. Myrvik, Q. N. 1956. Serum bactericidins active against
gram-positive bacteria. Ann. N.Y. Acad. Sci. 66:391-
400.
211. Myrvik, Q. N., and R. S. Weiser. 1955. Studies on
antibacterial factors in mammalian tissues and fluids. I.
A serum bactericidin for Bacillus subtilis. J. Immunol.
74:9-16.
212. Nakano, M., Y. Kishimoto, and K. Saito. 1968. Spot
method in agar for the quantitative assay of anti-Salmo-
nella antibody. Jpn. J. Microbiol. 12:364-366.
213. Neisser, M. and F. Wechsberg. 1901. Uber die Wirkun-
gart bactericider Sera. Muench. Med. Wochenschr.
48:697-700.
214. Nelson, B. W., and R. J. Roantree. 1967. Analyses of
lipopolysaccharides extracted from penicillin-resistant,
serum-sensitive Salmonella mutants. J. Gen. Microbiol.
48:179-188.
215. Nikaido, H., Y. Takeuchi, S.-I. Ohnishi, and T. Nakae.
1977. Outer membrane of Salmonella typhimurium.
Electron spin resonance studies. Biochim. Biophys.
Acta 465:152-164.
216. Noller, E. C., and S. E. Hartsell. 1961. Bacteriolysis of
Enterobacteriaceae. I. Lysis by four lytic systems utiliz-
ing lysozyme. J. Bacteriol. 81:482-491.
217. Noiler, E. C., and S. E. Hartsell. 1961. Bacteriolysis of
Enterobacteriaceae. II. Pre- and co-lytic treatments po-
tentiating the action of lysozyme. J. Bacteriol. 81:492-
499.
218. Nonaka, M., N. Yamaguchi, S. Natsuume-Sakai, and M.
Takahashi. 1981. The complement system of rainbow
trout (Salmo gairdneri). I. Identification of the serum
lytic system homologous to mammalian complement. J.
Immunol. 126:1489-1494.
219. Norden, C. W., M. L. Callerame, and J. Baum. 1970.
Haemophilus influenzae meningitis in an adult. A zcudy
of bactericidal antibodies and immunoglobulins. N. Engl.
J. Med. 282:190-194.
220. Norman, B., 0. Stendahl, C. Tagesson, and L. Edebo.
1972. Characteristics of the inhibition by rabbit immune
serum of the bactericidal effect of cattle normal serum on
Salmonella typhimurium 395 MRO. Acta Pathol. Micro-
biol. Scand. Sect B 80:891-899.
221. Ogata, R. T., and R. P. Levine. 1980. Characterization of
complement resistance in Escherichia coli conferred by
the antibiotic resistance plasmid R100. J. Immunol.
125:1494-1498.
222. Olling, S. 1977. Sensitivity of gram-negative bacilli to the
serum bactericidal activity: a marker of the host-parasite
relationship in acute and persisting infections. Scand. J.
Infect. Dis. Suppl. 10:1-40.
223. Oling, S., L. A. Hanson, J. Holmgren, U. Jodal, K. Lin-
coin, and U. Lindberg. 1973. The bactericidal effect of
normal human serum on E. coli strains from normals and
from patients with urinary tract infections. Infection
1:24-28.
224. Olilng, S., K. Verrier Jones, R. MacKenzie, E. R. Verrier
Jones, L. A. Hanson, and A. W. Asscher. 1981. A four-
year follow-up of schoolgirls with untreated covert bac-
teriuria: bacteriological aspects. Clin. Nephrol. 16:169-
171.
225. Opal, S., A. Cross, and P. Gemski. 1982. K antigen and
serum sensitivity of rough Escherichia coli. Infect. Im-
mun. 37:956-960.
226. Ormrod, D. J., and T. E. Miller. 1978. Complement-
mediated immune mechanisms in renal infection. I. Ef-
fect of renal tissue in vitro. Clin. Exp. Immunol. 33:107-
114.
227. 0rskov, I., F. 0rskov, B. Jann, and K. Jann. 1977.
Serology, chemistry, and genetics of 0 and K antigens of
Escherichia coli. Bacteriol. Rev. 41:667-710.
228. 0rskov, I., V. Sharma, and F. 0rskov. 1976. Genetic
mapping of the Kl and K4 antigens (L) of Escherichia
MICROBIOL. REV.
coli. Non-allelism of K(L) antigens with K antigens of
08:K27 (A), 08:K8(L) and 09:K57(B). Acta Pathol.
Microbiol. Scand. Sect. B 84:125-131.
229. Osawa, E., and L. H. Muschel. 1964. Studies relating to
the serum resistance of certain gram-negative bacteria. J.
Exp. Med. 119:41-51.
230. Osborn, M. J., and L. I. Rothfield. 1971. Biosynthesis of
the core region of lipopolysaccharide, p. 331-350. In
G. Weinbaum, S. Kadis, and S. J. Ajl (ed.), Microbial
toxins. IV. Bacterial endotoxins. Academic Press, Inc.,
New York.
231. Overath, P., M. Brenner, T. Gulik-Krzywicki, E.
Shechter, and L. Letellier. 1975. Lipid phase transitions
in cytoplasmic and outer membranes of Escherichia coli.
Biochim. Biophys. Acta 389:358-369.
232. Peter, G., M. B. Weigert, A. R. Bissel, R. Gold,
D. Kreutzer, and R. H. McLean. 1981. Meningococcal
meningitis in familial deficiency of the fifth component of
complement. Pediatrics 67:882-886.
233. Petersen, B. H., C. J. Lammel, D. P. Stites, and G. F.
Brooks. 1980. Human seminal plasma inhibition of com-
plement. J. Lab. Clin. Med. 96:582-591.
234. Petersen, B. H., T. J. Lee, R. Snyderman, and G. F.
Broks. 1979. Neisseria meningitidis and Neisseria gon-
orrhoeae bacteremia associated with C6, C7, or C8
deficiency. Ann. Intern. Med. 90:917-920.
235. Pike, R. M., and C. H. Chandler. 1971. Serological
properties of -yG and yM antibodies to the somatic
antigen of Vibro cholerae during the course of immuniza-
tion of rabbits. Infect. Immun. 3:803-809.
236. Pitt, J. 1978. K-1 antigen of Escherichia coli: epidemiol-
ogy and serum sensitivity of pathogenic strains. Infect.
Immun. 22:219-224.
237. Podack, E. R., and H. J. Miiler-Eberhard. 1981. Mem-
brane attack complex of complement. Evidence for its
dimeric structure based on hybrid formation. J. Biol.
Chem. 256:3145-3148.
238. Podack, E. R., and J. Tschopp. 1982. Polymerization of
the ninth component of complement (C9): formation of
poly(C9) with a tubular ultrastructure resembling the
membrane attack complex of complement. Proc. Nati.
Acad. Sci. U.S.A. 79:574-578.
239. Polk, H. C., and A. A. Miles. 1973. The decisive period in
the primary infection of muscle by Escherichia coli. Br.
J. Exp. Pathol. 54:99-109.
240. Postgate, J. R. 1969. Viable counts and viability, p. 611-
628. In J. R. Norris and D. W. Ribbons (ed.), Methods in
microbiology, vol. 1. Academic Press, Inc., London.
241. Provonchee, R. B., and S. H. Zinner. 1974. Rapid method
for determining serum bactericidal activity. Appi. Micro-
biol. 27:185-186.
242. Reed, W. P., and E. L. Albright. 1974. Serum factors
responsible for killing of Shigella. Immunology 26:205-
215.
243. Reid, K. B. M., and R. R. Porter. 1981. The proteolytic
activation systems of complement. Annu. Rev. Bio-
chem. 50:433-464.
244. Reynard, A. M., and M.E. Beck. 1976. Plasmid-mediated
resistance to the bactericidal effects of normal rabbit
serum. Infect. Immun. 14:848-850.
245. Reynard, A. M., M. E. Beck, and R. K. Cunningham.
1978. Effects of antibiotic resistance plasmids on the
bactericidal activity of normal rabbit serum. Infect.
Immun. 19:861-866.
246. Reynolds, B. L., and H. Pruul. 1971. Sensitization of
complement-resistant smooth gram-negative bacterial
strains. Infect. Immun. 3:365-372.
247. Reynolds, B. L., U. A. Rother, and K. 0. Rother. 1975.
Interaction of complement components with a serum-
resistant strain of Salmonella typhimurium. Infect. Im-
mun. 11:944-948.
248. Reynolds, B. L., and D. Rowley. 1969. Sensitization of
complement resistant bacterial strains. Nature (London)
221:1259-1261.
249. Rittenberg, S. C., C. W. Penn, N. J. Parsons,
 VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
D. R. Veale, and H. Smith. 1977. Phenotype changes in
the resistance of Neisseria gonorrhoeae to killing by
normal human serum. J. Gen. Microbiol. 103:69-75.
250. Roantree, R. J., and N. C. Pappas. 1960. The survival of
strains of enteric bacilli in the blood stream as related to
their sensitivity to the bactericidal effect of serum. J.
Clin. Invest. 39:82-88.
251. Roantree, R. J., and L. A. Rantz. 1960. A study of the
relationship of the normal bactericidal activity of human
serum to bacterial infection. J. Clin. Invest. 39:72-81.
252. Roantree, R. J., and J. P. Steward. 1965. Mutations to
penicillin resistance in the Enterobacteriaceae that affect
sensitivity to serum and virulence for the mouse. J.
Bacteriol. 89:630-639.
253. Robbins, J. B., K. Kenny, and E. Suter. 1965. The
isolation and biological activities of rabbit yM- and -yG-
anti-Salmonella typhimurium antibodies. J. Exp. Med.
122:385-402.
254. Robbins, J. B., G. H. McCracken, E. C. Gotschlich,
F. 0rskov, I. 0rskov, and L. A. Hanson. 1974. Esche-
richia coli Kl capsular polysaccharide associated with
neonatal meningitis. N. Engl. J. Med. 290:1216-1220.
255. Robbins, J. B., R. Schneerson, W. B. Egan, W. Vann,
and D. T. Liu. 1980. Virulence properties of bacterial
capsular polysaccharides-unanswered questions, p.
115-132. In H. Smith, J. H. Skehel, and M. J. Turner
(ed.), The molecular basis of microbial pathogenicity.
Verlag Chemie GmbH, Weinheim.
256. Root, R. K., L. ElIman, and M. M. Frank. 1972. Bacteri-
cidal and opsonic properties of C4-deficient guinea pig
serum. J. Immunol. 109:477-486.
257. Rottem, S., and L. Leive. 1977. Effect of variations in
lipopolysaccharide on the fluidity of the outer membrane
of Escherichia coli. J. Biol. Chem. 252:2077-2081.
258. Rowley, D. 1954. The virulence of strains of Bacterium
coli for mice. Br. J. Exp. Pathol. 35:528-538.
259. Rowley, D. 1956. Rapidly induced changes in the level of
non-specific immunity in laboratory animals. Br. J. Exp.
Pathol. 37:223-234.
260. Rowley, D. 1968. Sensitivity of rough gram-negative
bacteria to the bactericidal action of serum. J. Bacteriol.
95:1647-1650.
261. Rowley, D. 1973. Antibacterial action of antibody and
complement. J. Infect. Dis. 128(Suppl.):170-175.
262. Rowley, D., and K. J. Turner. 1968. Passive sensitization
of Salmonella adelaide to the bactericidal action of
antibody and complement. Nature (London) 217:657-
658.
263. Rowley, D., and A. C. Wardlaw. 1958. Lysis of gram-
negative bacteria by serum. J. Gen. Microbiol. 18:529-
533.
264. Sabet, S. F. 1976. Effect of colicin K on a membrane-
associated, energy-linked function. J. Bacteriol.
126:601-608.
265. Salton, M. R. J., and J. M. Ghuysen. 1959. The structure
of di- and tetra-saccharides released from cell walls by
lysozyme and Streptomyces F, enzyme and the
P(1-*4)N-acetylhexosaminidase activity of these en-
zymes. Biochim. Biophys. Acta 36:552-554.
266. Schiffer, M. S., E. Olivera, M. P. Glode, G. H.
McCracken, L. M. Sarff, and J. B. Robbins. 1976. Rela-
tion between invasiveness and the Kl capsular polysac-
charide of Escherichia coli. Pediatr. Res. 10:82-87.
267. Schoolnik, G. K., T. M. Buchanan, and K. K. Holmes.
1976. Gonococci causing disseminated infection are re-
sistant to the bactericidal action of normal human serum.
J. Clin. Invest. 58:1163-1173.
268. Schreiber, R. D., D. C. Morrison, E. R. Podack, and
H. J. Mfiller-Eberhard. 1979. Bactericidal activity of the
alternative complement pathway generated from 11 iso-
lated plasma proteins. J. Exp. Med. 149:870-882.
269. Schreiber, R. D., and H. J. Muiler-Eberhard. 1978. As-
sembly of the cytolytic alternative pathway of comple-
ment from 11 isolated plasma proteins. J. Exp. Med.
148:1722-1727.
81
270. Schulkind, M. L., K. Kenny, M. Herzberg, and J. B.
Robbins. 1972. The specific secondary biological activi-
ties of rabbit IgM and IgG anti-Salmonella typhimurium
"O" antibodies isolated during the development of the
immune response. Immunology 23:159-170.
271. Schwab, G. E., and P. R. Reeves. 1966. Comparison of
the bactericidal activity of different vertebrate sera. J.
Bacteriol. 91:106-112.
272. Sevag, M. G., and R. E. Miller. 1948. Studies on the
effect of immune reactions on the metabolism of bacte-
ria. I. Methods and results with Eberthella typhosa. J.
Bacteriol. 55:381-392.
273. Shelton, E., K. Yonemasu, and R. M. Stroud. 1972.
Ultrastructure of the human complement component,
Clq. Proc. Natl. Acad. Sci. U.S.A. 69:65-68.
274. Shin, M. L., G. Hasnsch, and M. M. Mayer. 1981. Effects
of agents that produce membrane disorder on lysis of
erythrocytes by complement. Proc. Natl. Acad. Sci.
U.S.A. 78:2522-2525.
275. Shin, M. L., W. A. Paznekas, and M. M. Mayer. 1978. On
the mechanism of membrane damage by complement:
the effect of length and unsaturation of the acyl chains in
liposomal bilayers and the effect of cholesterol concen-
tration in sheep erythrocyte and liposomal membranes.
J. Immunol. 120:1996-2002.
276. Silver, R. P., W. Aaronson, A. Sutton, and R. Schneer-
son. 1980. Comparative analysis of plasmids and some
metabolic characteristics of Escherichia coli Kl from
diseased and healthy individuals. Infect. Immun. 29:200-
206.
277. Silver, R. P., C. W. Finn, W. F. Vann, W. Aaronson,
R. Schneerson, P. J. Kretschmer, and C. F. Garon.
1981. Molecular cloning of the Kl capsular polysaccha-
ride genes of E. coli. Nature (London) 289:696-698.
278. Simberkoff, M. S., N. H. Moldover, and J. J. Rahal.
1980. Absence of detectable bactericidal and opsonic
activities in normal and infected human cerebrospinal
fluids. A regional host defense deficiency. J. Lab. Clin.
Med. 95:362-372.
279. Simberkoff, M. S., I. Ricupero. and J. J. Rahal. 1976.
Host resistance to Serratia marcescens infection: serum
bactericidal activity and phagocytosis by normal blood
leukocytes. J. Lab. Clin. Med. 87:206-217.
280. Sims, P. J., and P. K. Lauf. 1980. Analysis of solute
diffusion across the CSb-9 membrane lesion of comple-
ment: evidence that individual C5b-9 complexes do not
function as discrete, uniform pores. J. Immunol.
125:2617-2625.
281. Sirot.4k, L., K. Inoue, M. Okada, and T. Amano. 1976.
Immune bactericidal reactions by guinea-pig yl and y2
antibodies. Immunology 30:435-441.
282. Skarnes, R. C. 1978. Humoral bactericidal systems:
nonspecific and specific mechanisms. Infect. Immun.
19:515-522.
283. Smith, H. 1966. Leucocytes containing bacteria in plain
blood films from patients with septicaemia. Australas.
Ann. Med. 15:210-221.
285. Spink, W. W., and C. S. Keefer. 1937. Studies of
gonococcal infection. I. The method of destroying the
whole defibrinated blood of patients with gonococcal
arthritis. J. Clin. Invest. 16:177-184.
286. Spitznagel, J. K., and L. A. Wilson. 1966. Normal serum
cytotoxicity for P32-labeled smooth Enterobacteriaceae.
I. Loss of label, death, and ultrastructural damage. J.
Bacteriol. 91:393-400.
287. Steel, E. J., B. L. Reynolds, U. Rother, and D. Rowley.
1977. The mechanism of complement fixation by rabbit
F(ab')2: bactericidal activity of chelated serum and C4-
deficient serum. Immunochemistry 14:681-686.
288. Steers, E., E. L. Foltz, and B. S. Graves. 1959. An inocula
replicating apparatus for routine testing of bacterial
susceptibility to antibiotics. Antibiot. Chemother. 9:307-
311.
289. Sterzl, J., J. Kostka, and A. Lanc. 1962. Development of
bactericidal properties against gram-negative organisms
82 TAYLOR
in the serum of young animals. Folia Microbiol. (Prague)
7:162-174.
290. Stevens, P., s. N.-Y. Huang, W. D. Welch, and
L. S. Young. 1978. Restricted complement activation by
Escherichia coli with the Ki capsular serotype: a possi-
ble role in pathogenicity. J. Immunol. 121:2174-2180.
291. Steward, J. P., L. R. Collis, and R. J. Roantree. 1966.
Effects of immunization of guinea pigs lacking bactericid-
al antibody against Salmonella enteritidis. J. Immunol.
97:224-230.
292. Stinebring, W. R., W. Braun, and A. Pomales-Lebr6n.
1960. Modified serum resistance of bacteria following
intracellular residence. Ann. N.Y. Acad. Sci. 88:1230-
1236.
293. Sud, I. J., and D. S. Feingold. 1975. Detection of agents
that alter the bacterial cell surface. Antimicrob. Agents
Chemother. 8:34-37.
294. Sutherland, I. W. 1979. Microbial exopolysaccharides:
control of synthesis and acylation, p. 1-34. In
R. C. W. Berkeley, G. W. Gooday, and C. C. Ellwood
(ed.), Microbial polysaccharides and polysaccharases.
Academic Press, Inc., London.
295. Svehag, S.-E., L. Manhem, and B. Bloth. 1972. Ultra-
structure of human Clq protein. Nature (London) New
Biol. 238:117-118.
296. Tachibana, D. K. 1965. Characterization of the bacteri-
cidal action of precolostral piglet and calf sera against
gram-negative enteric bacteria in rough phase. Folia
Microbiol. Prague 10:145-155.
297. Taylor, P. W. 1972. An antibactericidal factor in the
serum of two patients with infections of the upper
urinary tract. CGin. Sci. 43:23-30.
298. Taylor, P. W. 1972. Isolation and characterization of a
serum antibactericidal factor. Clin. Sci. 43:705-708.
299. Taylor, P. W. 1974. Sensitivity of some smooth strains of
Escherichia colt to the bactericidal action of normal
human serum. J. Clin. Pathol. 27:626-629.
300. Taylor, P. W. 1975. Genetical studies of serum resistance
in Escherichia coli. J. Gen. Microbiol. 89:57-66.
301. Taylor, P. W. 1976. Immunochemical investigations on
lipopolysaccharides and acidic exopolysaccharides from
serum-sensitive and serum-resistant strains of Escherich-
ia coli isolated from urinary tract infections. J. Med.
Microbiol. 9:405-421.
302. Taylor, P. W. 1978. The effect of the growth environment
on the serum sensitivity of some urinary Escherichia coli
strains. FEMS Microbiol. Lett. 3:119-122.
303. Taylor, P. W., H. Gaunt, and F. M. Unger. 1981. Effect
of subinhibitory concentrations of mecillinam on the
serum susceptibility of Escherichia colt strains. Antimi-
crob. Agents Chemother. 19:786-788.
304. Taylor, P. W., and C. Hughes. 1978. Plasmid carriage and
the serum sensitivity of enterobacteria. Infect. Immun.
22:10-17.
305. Taylor, P. W., and K. G. Koutsaimanis. 1975. Experi-
mental Escherichia coli urinary infection in the rat.
Kidney Int. 8:233-238.
306. Taylor, P. W., and H.-P. Kroll. 1983. Killing of an
encapsulated strain of Escherichia colt by human serum.
Infect. Immun. 39:122-131.
307. Taylor, P. W., H.-P. Kroll, and S. Tomlinson. 1982.
Effect of subinhibitory concentrations of mecillinam on
expression of Escherichia coli surface components asso-
ciated with serum resistance. Drugs Exp. Clin. Res.
8:625-631.
308. Taylor, P. W., P. Messner, and R. Parton. 1981. Effect of
the growth environment on cell envelope components of
Escherichia coli in relation to sensitivity to human se-
rum. J. Med. Microbiol. 14:9-19.
309. Taylor, P. W., and R. Parton. 1977. A protein factor
associated with serum resistance in Escherichia coli. J.
Med. Microbiol. 10:225-232.
310. Taylor, P. W., A. P. Roberts, and P. E. Gower. 1972.
Evaluation of a technique for the estimation of serum
bactericidal activity against gram-negative organisms.
MICROBIOL. REV.
Med. Lab. Technol. 29:272-279.
311. Taylor, P. W., and M. K. Robinson. 1980. Determinants
that increase the serum resistance of Escherichia coli.
Infect. Immun. 29:278-280.
312. Tempest, D. W., and D. C. Ellwood. 1969. The influence
of growth conditions on the composition of some cell
wall components of Aerobacter aerogenes. Biotechnol.
Bioeng. 11:775-783.
313. Teuber, M. 1969. Susceptibility to polymyxin B of peni-
cillin G-induced Proteus mirabilis L forms and sphero-
plasts. J. Bacteriol. 98:347-350.
314. Thj$tta, T., and E. Waaler. 1932. Dissociation and
sensitiveness to normal serum in dysentery bacilli of type
III. J. Bacteriol. 24:301-316.
315. Thompson, R. A., and P. J. Lachman. 1970. Reactive
lysis: the complement mediated lysis of unsensitized
cells. I. The characterization of the indicator factor and
its identification as C7*. J. Exp. Med. 131:629-641.
316. Timmis, K. N., P. A. Manning, C. Echarti, J. K. Timmis,
and A. Moll. 1981. Serum resistance in E. coli, p. 133-
144. In S. B. Levy, R. C. Clowes, and E. L. Koenig
(ed.), Molecular biology, pathogenicity and ecology of
bacterial plasmids. Plenum Publishing Corp., New York.
317. Tisdale, W. A. 1961. Spontaneous colon bacillus bacter-
emia in Laennec's cirrhosis. Gastroenterology 40:141-
148.
318. Tramont, E. C., J. C. Sadoff, and M. S. Artenstein. 1974.
Cross-reactivity of Neisseria gonorrhoeae and Neisseria
meningitidis and the nature of antigens involved in the
bactericidal reaction. J. Infect. Dis. 130:240-247.
319. Tranum-Jensen, J., S. Bhakdi, B. Bhakdi-Lehnen,
0. J. Bjerrum, and V. Speth. 1978. Complement lysis:
the ultrastructure and orientation of the C5b-9 complex
on target sheep erythrocyte membranes. Scand. J. Im-
munol. 7:45-56.
320. Traub, W. H., and I. Kleber. 1976. Selective activation
of classical and alternative pathways of human comple-
ment by "promptly serum-sensitive" and "delayed se-
rum-sensitive" strains of Serratia marcescens. Infect.
Immun. 13:1343-1346.
321. Troy, F. A. 1979. The chemistry and biosynthesis of
selected bacterial capsular polymers. Annu. Rev. Micro-
biol. 33:519-560.
322. Trust, T. J., I. D. Courtice, A. G. Khouri, J. H. Crosa,
and M. H. Schiewe. 1981. Serum resistance and hemag-
glutination ability of marine vibrios pathogenic for fish.
Infect. Immun. 34:702-707.
323. Tschopp, J., and E. R. Podack. 1981. Membranolysis by
the ninth component of human complement. Biochem.
Biophys. Res. Commun. 100:1409-1414.
324. Turk, J. L. 1959. Immunological activity of precolostral
calf serum. Nature (London) 183:838.
325. Van Dijk, W. C., H. A. Verbrugh, R. Peters, M. E. Van
der Tol, P. K. Peterson, and J. Verhoef. 1979. Escherich-
ia coli K antigen in relation to serum-induced lysis and
phagocytosis. J. Med. Microbiol. 12:123-130.
326. Vosti, K. L., and E. Randall. 1970. Sensitivity of serolog-
ically classified strains of Escherichia coli of human
origin to the serum bactericidal system. Am. J. Med. Sci.
259:114-119.
327. Waisbren, B. A., and I. Brown. 1966. A factor in the
serum of patients with persisting infections that inhibits
the bactericidal activity of normal serum against the
organism that is causing the infection. J. Immunol.
97:431-437.
328. Wardlaw, A. C. 1962. The complement-dependent bacte-
riolytic activity of normal human serum. I. The effect of
pH and ionic strength and the role of lysozyme. J. Exp.
Med. 115:1231-1248.
329. Wardlaw, A. C. 1963. The complement-dependent bacte-
riolytic activity of normal human serum. II. Cell wall
composition of sensitive and resistant strains. Can. J.
Microbiol. 9:41-52.
330. Wardlaw, A. C. 1964. Endotoxin and complement sub-
strate, p. 81-88. In M. Landy and W. Braun (ed.), Bacte-
VOL. 47, 1983 SERUM ACTIVITY AGAINST GRAM-NEGATIVE BACTERIA
rial endotoxins. Rutgers State University Press, New
Brunswick, N.J.
331. Wardlaw, A. C., and L. Pillemer. 1956. The properdin
system and immunity. V. The bactericidal activity of the
properdin system. J. Exp. Med. 103:553-575.
332. Wardlaw, A. C., and L. Pillemer. 1956. The demonstra-
tion of the bactericidal activity of the properdin system.
Ann. N.Y. Acad. Sci. 66:244-246.
333. Weidanz, W. P., A. L. Jackson, and M. Landy. 1964.
Some aspects of the antibody response of rabbits to
immunization with enterobacterial somatic antigens.
Proc. Soc. Exp. Biol. Med. 116:832-837.
334. Weiss, M. J., and S. E. Luria. 1978. Reduction of
membrane potential, an immediate effect of colicin K.
Proc. Natl. Acad. Sci. U.S.A. 75:2483-2487.
355. Wendt, L. 1970. Mechanism of colicin action: early
events. J. Bacteriol. 104:1236-1241.
336. Westphal, O., and K. Jann. 1965. Bacterial lipopolysac-
charides: extraction with phenol-water and further appli-
cation of the procedure, p. 83-91. In R. L. Whistler
(ed.), Methods in carbohydrate chemistry, vol. 5. Aca-
demic Press, Inc., London.
337. Whaley, K., P. H. Schur, W. R. McCabe, and S. Ruddy.
1980. Modulation of the alternative complement pathway
in patients with gram-negative bacteraemia. Parasite
Immunol. (Oxford) 2:29-37.
338. Wilkinson, S. G. 1977. Composition and structure of
bacterial lipopolysaccharides, p. 97-175. In I. Suther-
land (ed.), Surface carbohydrates of the procaryotic cell.
83
Academic Press, Inc., London.
338a.Williams Smith, H. 1974. A search for transmissible
pathogenic characters in invasive strains of Escherichia
coli: the discovery of a plasmid-controlled toxin and a
plasmid-controlled lethal character closely associated, or
identical, with colicine V. J. Gen. Microbiol. 83:95-111.
339. Wilson, G. S., and A. A. Miles. 1964. Topley and
Wilson's principles of bacteriology and immunity, p. 264,
vol. 1, 5th ed. Edward Arnold Ltd., London.
340. Wilson, L. A., and J. K. Spitznagel. 1968. Molecular and
structural damage to Escherichia coli produced by anti-
body, complement, and lysozyme systems. J. Bacteriol.
96:1339-1348.
341. Wilson, L. A., and J. K. Spitznagel. 1971. Characteristics
of complement-dependent release of phospholipid from
Escherichia coli. Infect. Immun. 4:23-28.
342. Wright, S. D., and R. P. Levine. 1981. How complement
kills E. coli. I. Location of the lethal lesion. J. Immunol.
127:1146-1151.
343. Wright, S. D., and R. P. Levine. 1981. How complement
kills E. coli. II. The apparent two-hit nature of the lethal
event. J. Immunol. 127:1152-1156.
344. Young, L. S., and D. Armstrong. 1972. Human immunity
to Pseudomonas aeruginosa. I. In vitro interaction of
bacteria, polymorphonuclear leukocytes, and serum fac-
tors. J. Infect. Dis. 126:257-276.
345. Zinner, S. H., and W. R. McCabe. 1976. Effects of IgM
and IgG antibodies in patients with bacteremia due to
gram-negative bacilli. J. Infect. Dis. 133:37-45.