AIDS RESEARCH AND HUMAN RETROVIRUSES

Volume 19, Number 12, 2003, pp. 1105–1113

© Mary Ann Liebert, Inc.

Humans Have Antibodies Reactive with Bovine Leukemia Virus

GERTRUDE CASE BUEHRING, 1 SEAN M. PHILPOTT,1,2 and K. YEON CHOI1,3

ABSTRACT

Bovine leukemia virus (BLV) is an oncogenic retrovirus that commonly infects cattle and causes B cell leuko-
sis in 1–5% of infected cattle. BLV-infected cells are present in marketed beef and dairy products. In the
decade after the discovery of BLV in 1969, studies using agar gel immunodiffusion and complement fixation
assays failed to find antibodies to BLV in human sera. This led to the prevailing opinion that exposure of hu-
mans to BLV and/or the potential for infection are not significant and therefore the virus is not a public health
hazard. We reexamined this issue using more sensitive immunological techniques available today. Using im-
munoblotting to test the sera of 257 humans for antibodies of four isotypes (IgG1, IgM, IgA, and IgG4 ) to the
BLV capsid antigen (p24), we detected at least one antibody isotype reactive with BLV in 74% of the human
sera tested. The specificity of the reactivity was strongly suggested by competition studies and by ruling out
cross-reacting antibodies to other chronic human viruses. Our results suggest that antibodies reactive with
the BLV capsid antigen may serve as a biomarker for exposure to BLV and this exposure may be widespread.
The results do not necessarily mean that humans are actually infected with BLV; the antibodies could be a
response to heat-denatured BLV antigens consumed in food. They do, however, suggest that further studies
in this area could be important.

INTRODUCTION tein (gp51) and/or capsid protein (p24). Based on these stud-
ies, the prevailing opinion5,12,13 for over two decades has been

B OVINE LEUKEMIA VIRUS (BLV) is an exogenous retrovirus

that infects the majority of dairy and beef cattle herds in
the United States and Canada.1 Transmission of BLV through
the herd can occur through blood and secretions, and in the cat-
tle industry today is primarily through transfer of infected lym-
phocytes on blood-contaminated veterinary and agricultural in-
struments not disinfected between animals.2 BLV causes bovine
leukosis, a malignancy of B lymphocytes, but less than 5% of
infected cattle actually develop clinical leukosis and are ex-
cluded from the market. Most infected cattle remain healthy,
produce optimal amounts of milk, and therefore are not culled
from the herd. BLV-infected B lymphocytes circulate through
the blood of infected cattle and are present in beef. Infectious
BLV is also found in the milk of infected cows and can be trans-
mitted to suckling calves via the milk.3
BLV was first isolated in 19694 and for several years after
its discovery there were intense efforts to determine whether it
might infect humans through consumption of bovine foodstuffs
or occupational exposure to cattle. Eight serological surveys5–12
failed to detect human antibodies to BLV envelope glycopro-
“BLV is not transmissible to humans, and no human disease
has ever been attributed to BLV.” 13
Although these carefully performed early studies used sero-
logic techniques [complement fixation and agar gel immuno-
diffusion (AGID)] that were state of the art at that time, they
are extremely insensitive compared to more modern techniques.
During the past decade it has become increasingly apparent that
many cattle testing negative for anti-BLV antibodies by AGID
are actually positive by immunoblotting or enzyme-linked im-
munosorbent assay (ELISA). 14–16 One group determined that
ELISA was 50–100 times more sensitive than AGID for the de-
tection of anti-BLV antibodies.15 Two groups determined that
AGID failed to detect anti-BLV antibodies in 53%16 and 70%,14
respectively, of cattle seropositive by immunoblotting. In light
of these results, we were curious whether antibodies would be
detected in human sera using serological techniques more sen-
sitive than those used in the 1970s. This report describes a sur-
vey of sera from 257 humans using immunoblotting to test for
antibodies of four isotypes (IgG1, IgM, IgA, and IgG4 ) to the
BLV p24 antigen. For comparison, we also tested some of the

1School of Public Health, University of California, Berkeley, California 94720.

2Current address: Wadsworth Center, New York State Department of Health, Albany, New York 12201.
3Current address: Investigen, Inc., Alameda, California 94501.

1105
1106
sera using AGID. We were able to detect antibodies reactive
with BLV in 74% of the human sera by immunoblotting and to
provide evidence that they were not cross-reacting antibodies
to other viruses.
MATERIALS AND METHODS
Human and bovine subjects and serum specimens
The human study population was a convenience sample of
257 self-selected adults not intended to be representative of the
general population, but simply sufficient to answer the ques-
tion: Do any humans have antibodies to BLV? Subjects were
from the Berkeley community responding to notices requesting
volunteers. Blood (7 ml) was drawn by venipuncture into one
vacutainer serum separator tube with clot activator and without
anticoagulant (Beckton-Dickinson, Franklin Lakes, NJ). Serum
was separated from the clot by low-speed centrifugation (500
3 g) and stored at 4°C until use. Volunteers completed a ques-
tionnaire recording general descriptive characteristics (age,
race, socioeconomic status), some aspects of medical history,
and, for women, reproductive history. Commercially available
pooled human cord serum (Intergen, Milford, MA) was used as
a negative human serum control, for IgA and IgM isotypes,
which do not cross the placenta.
For 70 of the volunteers a second blood specimen was drawn
at least a year after the first. We also received from Northern
California Kaiser Permanente Regional Laboratory 200 stored
human sera that had been tested for antibodies to various
viruses. The protocol for the use of human subjects was ap-
proved by the University of California, Berkeley, Committee
for the Protection of Human Subjects.
Sera from Holstein and Jersey dairy cows at University of
California, Davis were used as controls in serological assays.
These sera had been previously tested for anti-BLV antibodies
at their institution of origin and in our laboratory.17
Agar gel immunodiffusion (AGID)
AGID was performed with the Leukassay B kit (Synbiotics,
San Diego, CA) according to the manufacturer’s protocol.
Briefly, 15 ml of 0.9% agarose agar type I (Sigma Chemical
Co., St. Louis, MO) in normal saline [8.5% (w/v) NaCl] was
poured into 100-mm petri plates. After gelling at room temper-
ature for at least 1 hr, a pattern of six wells surrounding one cen-
tral well was cut with an immunodiffusion template (Veterinary
Diagnostics Technology, Denver, CO). The diameter of each
well was 8 mm and the volume was 90 ml. Wells were 3 mm
apart. The antigen from the kit (enriched BLV envelope gp51
fraction) was placed into the center well, and each experimen-
tal or control antiserum diluted 1:8 in normal saline was placed
in a peripheral well, alternating with reference sera. Plates were
read for precipitin lines after 48–72 hr incubation in a humidi-
fied chamber at room temperature. Positive and negative con-
trol sera provided with the kit were run with each set of sera.
Immunoblots (western blots)
Immunoblots were run according to standard methods.18 The
antigen was a 262-amino acid recombinant p24 protein from
BUEHRING ET AL.
the BLV gag region (Synbiotics, San Diego, CA). Antigen was
diluted in assay wash buffer (Dulbecco’s phosphate-buffered
saline with 0.2% Tween 20) to a final concentration of 16.4
mg/ml, boiled 1 min, then mixed 1:1 with gel-loading buffer
[0.125 M Tris–HCl, pH 6.8, 4% (v/v) sodium dodecyl sulfate,
20% (v/v) glycerol, 10% (v/v) mercaptoethanol, 0.2% (w/v)
bromphenol blue]. Proteins were separated using sodium do-
decyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE)
for 1 hr (5% stacking gel; 10% resolving gel) for 80 min at 500
V. The gel was formed in a Hoeffer “Mighty Small II” verti-
cal slab gel electrophoresis unit (Amersham Biosciences, Pis-
cataway, NJ) using a blank comb (without teeth) to load the
sample. Separated gel proteins were transferred electrophoret-
ically to a 0.45-mm nitrocellulose membrane (Osmonics Labo-
ratory Products, Westborough, MA) by the semidry method
(trans-blot-SD, Bio-Rad, Richmond, CA) using Bjerrum and
Schafer-Nielsen transfer buffer [48 mM Tris, 39 mM glycine,
20% (v/v) methanol, 1.3 mM sodium dodecyl sulfate, pH 9.2].
After transfer, all membranes were stained with 0.1% Ponceau
S (w/v) in 5% (v/v) acetic acid to check for the presence of pro-
tein bands. Areas with no visible protein were discarded. Mem-
branes were cut into strips approximately 0.5 cm wide, and
blocked 1 hr with 1% dried milk (free of bovine anti-BLV an-
tibody) or 1.5% normal serum from goat (American Qualex,
San Clemente, CA), the animal in which most of the secondary
antibodies were produced. Both blocking solutions were diluted
in wash buffer.
Human samples were either whole sera diluted 1:100 in wash
buffer or fractions from a chromatographic (ion-exchange resin)
column (Quick-Sep System II, Isolab, Akron, OH): the “IgG
fraction” contained IgG1,2,3 and the “IgM fraction” contained
IgM, IgA, and IgG4 antibodies. After fractionation per the man-
ufacturer’s directions, the dilution factor of the IgG fraction was
1:91 and the dilution factor of the IgM fraction was 1:20. An-
tibody fractions were reacted with the blot at a final dilution of
1:100 in assay wash buffer. Positive control antibody was anti-
BLV p24 mouse monoclonal19 diluted 1:1000 in wash buffer.
Primary antibodies were reacted 1 hr at room temperature. Bi-
otinylated secondary antibodies, diluted in wash buffer, were
goat antihuman IgG (gamma chain-specific) (Vector Laborato-
ries, Burlingame, CA) (1:25,000), goat antihuman IgM (Vec-
tor) (1:25,000), goat antihuman IgA (Vector) (1:25,000), goat
antihuman IgG4 (Sigma) (1:10,000), and horse antimouse IgG
(1:2,000) (Vector), all reacted for 1 hr at room temperature. Bi-
otinylated protein G (1:500) (Sigma) was substituted for sec-
ondary antibody in some experiments. The biotin signals were
amplified with streptavidin-horseradish peroxidase (Amersham
Pharmacia, Arlington Heights, IL) diluted 1:1000 in wash buffer
and were detected using a chemiluminescence system with hy-
perfilm ECL (Amersham Pharmacia) with exposures at 2, 10,
and 30 sec. Between each of the steps membrane strips were
rinsed three times, 10 min each with wash buffer.
Although ECL protein molecular weight markers (Amer-
sham Biosciences) were run with many assays, a more defini-
tive confirmation of the location of the band representing BLV
p24 proteins was obtained with the anti-p24 mouse monoclo-
nal antibody run with each assay. Negative controls (buffer sub-
stituted for primary antibody) were run with each gel to rule
out direct reactivity of the secondary antibodies with BLV p24.
In addition, positive and negative bovine control sera from the
HUMAN ANTIBODIES TO BOVINE LEUKEMIA VIRUS 1107

Leukassay kit (1:100 in wash buffer) were run with many of termine correlation between two continuous variables. The Stu-
the assays. We assayed 51 of 257 (20%) of the human serum dent’s t test was used to determine the correlation between a
samples at least twice on different days and some samples as nominal value and a continuous variable with normal distribu-
many as four or five times, as a check on reproducibility. tion. Statistical significance was defined as p # 0.05.

Enzyme-linked immunosorbent assay (ELISA)

RESULTS
ELISA was run according to standard methods.20 ELISA
plates (Immulon 2, Dynatech, Chantilly, VA) were coated Demographic characteristics of the study population
overnight at 4°C with the recombinant p24 antigen described
above, 500 ng/well in 100 ml carbonate/bicarbonate coating Table 1 details the characteristics of the study population.
buffer (15 mM Na2CO3, 35 mM NaHCO3, pH 9.6). After re- Most of the 257 subjects were female, white, and between the
moving the coating buffer and rinsing the plate three times, 10 ages of 24 and 49. This reflects volunteer self-selection, and, to
min each, in ELISA wash buffer (Dulbecco’s phosphate- some degree, the community in which the study was conducted.
buffered saline with 0.05% Tween 20), we blocked the plates
30 min in 1.5% fetal horse serum (Cansera, Etobicoke, Ontario) Some humans have antibodies reactive with BLV
in wash buffer. This and subsequent steps were performed at antigens by immunoblotting
room temperature and utilized a volume of 200 ml per well. The Figure 1 directly compares the relative intensities of reac-
primary antibody (reacted for 60–90 min) was whole human tions of fractionated and unfractionated bovine vs. human sera,
serum diluted 1:100 in wash buffer. The secondary antibody to the recombinant BLV p24 using protein G in place of sec-
(reacted for 30–60 min) was horseradish peroxidase-labeled ondary antibodies. Included in the figure are samples of posi-
goat antihuman IgG (Chemicon, Temecula, CA) diluted 1:500 tive and negative bovine and human sera. Fetal bovine serum
in wash buffer. After the primary and secondary antibody steps, was negative, as expected, because no type of antibody crosses
the plates were rinsed five times, 1 min each with wash buffer. the bovine placenta. This served as the negative control bovine
The chromagen, o-phenylenediamine hydrochloride (Sigma), serum. Although the pooled human cord serum IgG reacted with
was dissolved according to the manufacturer’s instructions and BLV p24, there was no IgM reactivity, as expected, because
reacted for 30 min. The plate was read on an ELISA reader at IgM does not cross the human placenta. Thus, cord blood IgM
492 nm. The blank was a well in which buffer was substituted served as the negative control human serum. The intensities of
for both the primary and secondary antibodies. The secondary the unfractionated bovine sera were noticeably greater than
antibody control substituted buffer for the primary antibody those of the unfractionated human sera. Chromatographic frac-
only. The value for this control was subtracted from all exper- tionation increased the intensity of the reactivity of the human
imental values and did not exceed 5% of the lowest experi- sera but did not significantly alter the reactivity of the bovine
mental value. All samples were run in triplicate. The Leukas- sera. Figure 2 shows reactivities of four antibody isotypes of
say B AGID kit positive and negative bovine serum controls two representative human sera with the BLV p24 protein. The
(diluted 1:100 in wash buffer) were run with each assay. The frequencies of four antibody isotypes reacting to p24 by im-
secondary antibody used for these reactions was horseradish munoblotting are shown in Table 2. We were unable to run IgA
peroxidase-labeled goat antibovine IgG (Jackson ImmunoRe- and IgG4 assays for a few of the 257 specimens because Isolab
search Laboratories, West Grove, PA). Inc. discontinued manufacture of the Quick-Sep System II chro-
matographic columns and we did not want to substitute another,
Competition studies possibly slightly different, product. Reactivity of IgM antibod-
ies was the least frequent of the four antibody isotypes.
Sera from one male and one female subject with antibodies
reactive with BLV p24 on immunoblots were used. The IgG
Quick-Sep System II chromatography fractions (final dilution
1:91) of a goat polyclonal antibody to BLV and the preimmune TABLE 1. DEMOGRAPHIC CHARACTERISTICS
serum from the same goat (NCI, Biological Carcinogenesis OF THE STUDY POPULATION
Branch, distributed by Quality Biotech, Camden, NJ) were each Characteristic n (%)
diluted as 2-fold serial dilutions in the IgG1 chromatography frac-
tion (dilution 1:91) of each human serum tested. ELISAs were Age (years)
run as described above using the recombinant BLV p24 antigen 18–23 16 (6.2%)
(200 ng per well). A biotinylated mouse monoclonal to human 24–49 159 (61.9%)
IgG1 (CalTag, Burlingame, CA) (1:500 in wash buffer) was used $50 82 (31.9%)
as the secondary antibody, followed by streptavidin-horseradish Gender
peroxidase (1:1000 in wash buffer) (Amersham Pharmacia). Females 230 (89.4%)
Males 27 (10.6%)
Race
Data analysis White 215 (83.6%)
Data were double entered into and analyzed by the Statview Asian 20 (7.8%)
Black 10 (3.9%)
5.0 program (SAS Institute, Cary, NC). Significance of differ-
Hispanic 8 (3.1%)
ences between two sets of nominal values was determined by Other 4 (1.6%)
the chi-square test. Simple regression analysis was used to de-
1108 BUEHRING ET AL.

FIG. 1. Immunoblot performed as described in Materials and Methods comparing reactivity of human sera with bovine sera.
The electrophoresed antigen was BLV capsid antigen (p24) as described in the text. Lane 1, molecular weight markers (MW).
Primary antibodies in lanes 2–12: lane 2, monoclonal antibody to p24 (Mab)(positive control); lane 3, fetal bovine serum (FBS)
(negative control); lane 4, whole serum (W) (1:100) from cow 306 (AGID positive); lane 5, IgG fraction (1:100) of serum from
cow 306; lane 6, whole serum (W)(1:100) from cow 297 (AGID negative); lane 7, IgG fraction (1:100) of serum from cow 297;
lane 8, whole serum (W)(1:100) from human 1; lane 9, IgG fraction (1:100) of serum from human 1; lane 10, pooled whole hu-
man (h) cord serum (W) (1:100); lane 11, IgG fraction of pooled human cord serum (1:100); lane 12, IgM fraction of pooled
cord serum (1:100) (negative control). Protein G (1:500) was used in place of secondary antibody for all bovine and human sera.

Evidence that reactivity of human antibodies to BLV
antigens is specific
Figure 3 illustrates competition against BLV-reactive human
IgG antibodies by goat polyclonal antibody to BLV and lack
of competition by preimmune serum from the same goat. Sim-
ilar results were obtained using serum from a second human
subject.
To rule out that serum antibodies to chronic human viruses
were cross-reacting with BLV antigens, we tested 20 sera neg-
ative for and 20 sera positive for antibodies against the fol-
lowing viruses: human T cell leukemia viruses (HTLV) I and
II, human immunodeficiency virus (HIV), hepatitis B virus,
hepatitis C virus, and cytomegalovirus. All sera were received
from a large clinical laboratory running serological tests for an-
tibodies to these viruses on a daily basis. There was no signif-
icant difference in reactivity with the BLV p24 antigen by im-
munoblotting for sera positive for each of these viruses versus
sera negative for the same virus (p . 0.38, chi-square test).
To rule out that reactions of human sera with BLV antigen
reflected simply nonspecific “sticking” of IgM, we coated
ELISA plates with a 1:10 dilution of each of 31 human serum
and reacted this with goat antihuman IgM. We selected sera
that represented a full range of anti-BLV p24 immunoblot re-
actions including the most intense. We ran the ELISA as de-
scribed in the methods section, then plotted these anti-IgM
ELISA values against the anti-BLV p24 ELISA value for each
subject. The results indicated no significant correlation between
the quantity of IgM adhering to the plate and reactivity with
BLV p24 (p , 0.12, r 5 0.288) (data not shown).

FIG. 2. Immunoblot performed as described in Materials and Methods showing patterns of antibody isotype reactions in two
representative humans. The electrophoresed antigen was BLV capsid antigen (p24) as described in the text. Primary (1) and sec-
ondary (2) antibodies were as follows. Lane 1, (1) monoclonal antibody to BLV p24; (2) horse antimouse IgG; lane 2, (1) whole
serum (1:100) from cow 511 (AGID positive); (2) goat antibovine IgG; lane 3, (1) wash buffer; (2) goat antibovine IgG; lanes
4–7, human 1: lane 4, (1) IgG1 fraction; (2) goat antihuman IgG1; lane 5, (1) IgM fraction; (2) goat antihuman IgM; lane 6, (1)
IgM fraction; (2) goat antihuman IgA; lane 7, (1) IgM fraction; (2) goat antihuman IgG4; lanes 8–11, human 11: lane 8, (1) IgG1
fraction; (2) goat antihuman IgG1 ; lane 9, (1) IgM fraction; (2) goat antihuman IgM; lane 10, (1) IgM fraction; (2) goat antihu-
man IgA; lane 11, (1) IgM fraction; (2) goat antihuman IgG4; lane 12, (1) wash buffer; (2) goat antihuman IgG1; lane 13, (1)
wash buffer; (2) goat antihuman IgM; lane 14, (1) wash buffer; (2) goat antihuman IgA; lane 15, (1) wash buffer; (2) goat anti-
human IgG4.
HUMAN ANTIBODIES TO BOVINE LEUKEMIA VIRUS 1109

TABLE 2. FREQUENCY OF REACTIVITY TO BLV A NTIGENS IN None of the human sera gave a positive reaction to
THESTUDY POPULATION AS DETERMINED BY IMMUNOBLOTTING BLV antigen using the AGID assay
Antibody isotype Number (%) positive We wanted to simulate the level of sensitivity of methods
used in the 1970s and see if we would obtain similar results us-
IgG 101/257 (39.3%) ing sera collected in the 1990s. From the 257 human sera used
IgM 80/257 (31.1%) in this study, we selected the 25 samples that gave the qualita-
IgA 96/256 (37.5%) tively most intense IgG1 reaction to BLV p24 by immunoblot-
IgG4 103/252 (40.9%) ting and would, therefore, be most likely to give a positive re-
Any of the above four isotypes 191/257 (74.3%)
action in a less sensitive assay. By AGID, the most common
None of the above four isotypes 66/257 (25.7%)
assay used for BLV serology studies in the 1970s, none of the
human sera was positive. Positive and negative bovine control
sera were run with each assay.
Anti-BLV reactivity of sera from individual humans
tends to remain stable over time Heat denaturation of BLV antigens does not alter their
antigenicity as measured by immunoblotting
We tested the stability of BLV-reactive antibody level using
paired serum specimens from 37 subjects drawn at least 1 year The p24 protein used as the antigen in the immunoblot as-
apart. Figure 4 shows the regression plot obtained by plotting say was subjected to 100°C for 15 min (equivalent to pasteur-
the ELISA optical density of IgG reactivity to BLV p24 of the ization conditions) versus no heating, to determine if heat de-
first specimen against the second specimen from the same sub- naturation would alter immunoblot reactivity. Neither sample
ject. The correlation was highly significant (p # 0.0001, r 5 was given the customary 1 min heating at 100°C before mix-
0.872), suggesting that for each individual, antibody level is ing with gel-loading buffer. Figure 5 shows the reaction of the
stable over a relatively long period of time. anti-p24 monoclonal antibody and two representative positive

FIG. 3. Competition of human serum by anti-BLV goat polyclonal serum (closed circles) but not by preimmune serum from
the same goat (open circles) using ELISA with BLV p24-coated plates. Human sera were fractionated as described in Materials
and Methods, then 2-fold dilutions of the IgG fraction of the goat sera were made in a constant concentration of human serum
fraction. The secondary antibody was a monoclonal to human IgG1. Antibody levels are indicated as optical density. Each point
represents the mean of triplicate values.
1110
FIG. 4. Regression line representing the comparison of serum ELISA levels (optical density) of BLV p24-reactive IgG anti-
bodies in paired blood specimens from individual subjects drawn at least 1 year apart. The correlation between the two values
from an individual subject was highly significant (p , 0.0001, r 5 0.872).
unfractionated human sera to the boiled (B) versus not heated
(N) antigen. There was no noticeable difference in reactivity of
each antibody.
DISCUSSION
We are not the first to assay human sera for the presence of
antibodies to BLV. Studies done during the 1970s5–12 using
state-of-the-art methods at that time (AGID and complement
fixation) showed no evidence of human antibodies to BLV and
formed the basis of the prevailing opinion for the past two
decades that BLV does not infect humans and therefore is not
a public health hazard. Although a few articles over the past 25
years have reported limited evidence of reactivity of human an-
tibodies to BLV, these reports have not altered the prevailing
opinion, perhaps because their results were not conclusive,21
because the number of positive humans was only three,22 be-
cause the virus was not specifically identified as BLV,23 or be-
cause BLV was not the main focus of the study.24,25 It is not
surprising that we were able to detect anti-BLV antibodies us-
ing immunoblot. In a recent study analyzing bovine sera with
the identical immunoblot procedure, we were able to detect anti-
BLV antibodies by immunoblot in 39% of sera that tested neg-
ative by AGID. 17 Other researchers have obtained varying re-
sults ranging from 2% to 70% of immunoblot-positive cattle
BUEHRING ET AL.
missed by AGID.17 Our failure to detect human antibodies by
AGID in sera from the human subjects reacting most strongly
by immunoblotting is a confirmation of the early investigators’
negative results and a reassurance that the detection of human
BLV-reactive antibodies in this study is probably due to more
sensitive techniques, rather than to overall changes in reactiv-
ity in the human population during the past 20 years.
The question remains of why some bovine sera had anti-BLV
antibody levels high enough to be detected by AGID, whereas
none of the human sera did. One explanation is that the AGID
test involves primarily reaction to the gp51 envelope glyco-
protein as the antigen and the human response to this antigen
may not be as strong as the bovine response. For the im-
munoblot we tested only response to the p24 antigen, because
data using sera from BLV-positive cattle suggested that the p24
antigen was far superior to gp51 in the immunoblot assay,14,17,26
perhaps because the gp51 antigen was degraded by chemicals
used in the SDS–PAGE procedure.14,26 Using representative
human sera positive by immunoblot, we compared the inten-
sity of the immunoblot with that of equal dilutions of bovine
sera, using protein G as a common secondary reagent to bind
immunoglobulins. Reactions of unfractionated human sera were
less intense than those of unfractionated bovine sera, suggest-
ing that the humoral immune response of humans to BLV may
not be as great as that of cattle. This observation cannot be ex-
plained by any major species differences in serum im-
HUMAN ANTIBODIES TO BOVINE LEUKEMIA VIRUS
FIG. 5. Comparison of the reaction of human sera to heat-
treated antigen (100°C for 15 min) versus unheated antigen.
Lanes 1 and 2 reacted with a monoclonal to p24; lanes 3 and
4 reacted with serum from human 221; lanes 5 and 6 reacted
with serum from human 33. B, boiled; N, not heated.
munoglobulin levels, as these are roughly equivalent in cattle
and humans.27,28 Some possible explanations are that the ex-
posure dose could be less in humans or the route of exposure
could be less immunogenic.
Several lines of evidence suggest that the antibodies we de-
tected are specific for BLV and not cross-reacting antibodies
to epitopes of other viruses or cellular proteins. Using the sera
of two representative donors with anti-BLV antibodies, ELISA
competition studies with goat polyclonal antibodies to BLV in-
dicated that the human antibodies were competed by the goat
polyclonal antibodies but not by preimmune serum from the
same goat (Fig. 3). Second, no apparent cross-reactivity was
found with the only three known human retroviruses, human T
cell leukemia viruses I and II and human immunodeficiency
virus, and three other viruses commonly causing chronic in-
fections in humans, hepatitis B virus, hepatitis C virus, and cy-
tomegalovirus. The most common chronic human virus, Ep-
stein–Barr virus (EBV), was not tested in this regard because
about 95% of people in the United States are positive for EBV29
and it is difficult to obtain an appropriate number of EBV-neg-
ative specimens for a statistically meaningful analysis. Fur-
thermore, serological tests for anti-EBV antibodies are not rou-
tinely run in clinical laboratories. It is unlikely that the reactions
of the human sera represent reactions with cellular proteins be-
cause the p24 antigen used was not grown in mammalian cells;
it was a purified recombinant protein.
We took two measures to ensure that the immunoblot reac-
tions were specific for each of the isotypes represented. First,
whole human sera were fractionated through chromatography
columns separating the eluates into one fraction containing
IgG1,2,3 and the other fraction containing IgM, IgA, and IgG4.
According to the manufacturer (IsoLab), approximately 80% of
total IgM is recovered by the elution and most or all of inter-
fering rheumatoid factor is eliminated. Second, the secondary
antibodies used were specific for different isotypes and, ac-
cording to the manufacturer’s (Vector) specifications, the anti-
human IgG, antihuman IgM, and antihuman IgA antibodies
each had ,1% cross-reactivity with the other two. The antihu-
man antibody IgG4 did not cross-react with IgGs1,2,3.30
1111
Although our study was not specifically designed to deter-
mine the route of exposure of humans to BLV antigens, sev-
eral routes are theoretically possible. Only 25 (9.7%) of the sub-
jects queried in this study indicated any direct contact with
bovines or their biological products, and these subjects were
not more likely to have BLV-reactive antibodies of any the iso-
types tested. Therefore, it is unlikely this is the primary route
of exposure. Another possible route is injection of BLV in a
vaccine or other pharmaceutical. In the early manufacture of
viral vaccines, some cell lines used to produce the viruses be-
came contaminated with bovine viruses from the bovine sera
used to supplement the culture media.31 There are no reports,
however, of contamination with BLV and inadvertent produc-
tion of BLV in vaccine cell lines. This fits with studies show-
ing that BLV is cell associated and extracellular BLV virions
are undetectable in serum if cells are separated out soon after
blood collection.32 Moreover, infection with extracellular virus
particles is difficult to effect in vitro. BLV requires cell–cell
contact for infection, and few cell lines, when infected, result
in productive infection.33
A third possible route of exposure is through consumption
of bovine foodstuffs. Barnes et al.34 showed that healthy hu-
mans had antibodies, especially of the IgG4 isotype, to proteins
in common foods, suggesting that an oral route of administra-
tion of protein antigens is effective in eliciting serum antibod-
ies. This was the rationale for including IgG4 in the panel of
isotypes to study. The highly significant constancy of antibody
levels in paired specimens from 39 donors drawn 1 year apart
suggests that if the immunogenic exposure to BLV is through
foodstuffs, everyday dietary fluctuations do not appear to dras-
tically alter the level of BLV-reactive antibodies.
The presence of antibodies to particular viruses in human
sera is generally interpreted as an indicator of a present or past
infection with the virus.35 However, the BLV-reactive human
antibodies we observed could also theoretically represent an im-
mune reaction to heat-denatured BLV ingested in cooked or
pasteurized bovine foodstuffs. Since we found no difference in
reactivity of human sera to heat-denatured versus unheated anti-
gens (Fig. 5), we cannot distinguish between these two possi-
bilities by standard immunoblot procedures.
If the antibodies to BLV we observed do represent an im-
mune response to heat-denatured BLV in pasteurized dairy
products and cooked beef, this could represent an inadvertent
form of oral immunization and might conceivably protect
against subsequent infection with BLV that might not have been
inactivated during food preparation. The implications of this
could be significant for the milk-borne transmission of other
retroviruses, e.g., HTLV and HIV. In HTLV-I endemic areas
the main route of virus transmission is mother to child via
milk.36 Nursing is also a significant mode of mother-to-child
transmission for HIV in Africa. Standard methods of pasteur-
ization inactivate both HIV and HTLV in human milk and have
been recommended as a means to prevent milk-borne trans-
mission.37 Would the consumption of pasteurized HIV- or
HTLV-contaminated human milk result in an oral immuniza-
tion and would this immunization stimulate neutralizing anti-
body that could conceivably protect the infant against infec-
tious virus subsequently introduced via a milk-borne or even
some other route? This question deserves exploration.
Although our data do not distinguish whether human anti-
1112
bodies to BLV signify infection or merely exposure to dena-
tured antigen, they do prove an important point that may have
public health implications. Human exposure to BLV may be
widespread and antibodies to BLV may be an effective bio-
marker for this exposure. Since an estimated 89% of dairy op-
erations in the United States contain BLV-positive cows,13 such
widespread exposure is not surprising. These results, coupled
with reports of in vitro susceptibility of human cells to BLV
infection,33,38 and breast milk transmission of the closely re-
lated human T cell leukemia virus to infants up to at least 1
year of age,39–41 suggest the potential for orally transmitted
BLV to infect humans, should virus present in bovine food-
stuffs not be thoroughly denatured by pasteurization or cook-
ing. The long-held assumption that BLV is not a public health
hazard was based on the failure of experiments in the 1970s to
detect human antibodies to BLV. That assumption is no longer
tenable in view of our demonstration of humans seropositive
for BLV-reactive antibodies. Our results emphasize the need to
reopen this field of investigation and determine whether hu-
mans are actually infected with BLV and whether such an in-
fection would constitute a health hazard.
ACKNOWLEDGMENTS
We gratefully acknowledge the technical assistance of
Shirley Juan, Craig Smollin, Thomas Lee, Katharine
Tansavadti, Matthew Keeler, Anita Chiu, Denise Poteat,
Katherine McGirr, Andrew Lee, Rebecca Liu, and Hua Min
Shen. We thank Patrick Kramme for the gift of the monoclo-
nal antibody to BLV p24 and Hiroshi Takahashi and Jen-Shai-
Ngai for human sera tested for reactivity to various human
viruses. We are grateful to the following for their critical re-
view of the manuscript: Drs. A. Good, H. Takahashi, and C.
Tempelis. This research was supported by funds from the Cal-
ifornia Breast Cancer Research Program of the University of
California, Grants 2B-0001 and 3IB-0242. S.M. Philpott was
supported in part by Public Health Service Grant
8T2GM007127 from the National Institute of General Medical
Sciences, National Institutes of Health, Department of Health
and Human Services, and by a Regent’s Fellowship and a
Levine Award from the University of California, Berkeley.
REFERENCES
1. Schwartz I and Levy D: Pathobiology of bovine leukemia virus.
Vet Res 1994;25:521–536.
2. Sprecher DJ, Pelzer KD, and Lessard P: Possible effect of altered
management practices on seroprevalence of bovine leukemia virus
in heifers of a dairy herd with history of high prevalence of infec-
tion. J Am Vet Med Assoc 1991;199:584–588.
3. Ferrer JF, Kenyon SJ, and Gupta P: Milk of dairy cows frequently
contains a leukemogenic virus. Science 1981;213:1014–1016.
4. Miller JM, Miller LD, Olson C, and Gillette KG: Virus-like parti-
cles in phytohemagglutinin-stimulated lymphocyte cultures with
reference to bovine lymphosarcoma. J Natl Cancer Inst 1969;43:
1297–1305.
5. Caldwell GGP: Bovine leukemia virus, public health, serologic
studies. Proc Bovine Leukosis Symp, USDA 1978;143–159.
6. Bartsch DC, Springher F, and Falk H: Acute lymphocytic leuke-
mia: An adult cluster. JAMA 1975;232:1333–1336.
BUEHRING ET AL.
7. Olson C: Bovine lymphosarcoma (leukemia)—a synopsis. J Am
Vet Med Assoc 1974;165:630–632.
8. Gilden RV, Long CW, Hanson M, et al.: Characteristics of the ma-
jor internal protein and RNA-dependent DNA polymerase of
bovine leukemia virus. J Gen Virol 1975;29:305–314.
9. Caldwell GG, Baumgartener L, Carter C, et al.: Seroepidemiologic
testing in man for evidence of antibodies to feline leukemia virus
and bovine leukemia virus. In: Comparative Leukemia Research
(Clemmesen J and Yohn DS, eds.). Bibl Haemat, No. 43. Karger,
Basel, 1976, pp. 238–241.
10. Donham K, VanDerMaaten MJ, Miller JM, Kruse BC, and Rubino
MJ: Seroepidemiologic studies on the possible relationships of hu-
man and bovine leukemia. J Natl Cancer Inst 1977;59:851–853.
11. Olson C and Driscoll DM: Bovine leukosis: Investigation of risk
for man. J Am Vet Med Assoc 1978;173:1470–1472.
12. Burridge MJ: The zoonotic potential of bovine leukemia virus. Vet
Res Commun 1981;5:117–126.
13. NAHMS, National Animal Health Monitoring System: High preva-
lence of BLV in US dairy herds. Available online at www.aphis.
usda.gov/vs/ceah/cahm/d96blv.htm.
14. Walker PJ, Molloy JB, and Rodwell BJ: A protein immunoblot test
for detection of bovine leukemia virus p24 antibody in cattle and
experimentally infected sheep. J Virol Methods 1987;15:201–211.
15. Have P and Hoff-Jorgensen R: Demonstration of antibodies against
bovine leukemia virus (BLV) by blocking ELISA using bovine
polyclonal anti-BLV immunoglobulin. Vet Microbiol 1991;27:
221–229.
16. Grover YP and Guillemain B: An immunoblotting procedure for
detection of antibodies against bovine leukemia virus in cattle. J
Vet Med B 1992;39:48–52.
17. Choi KY, Liu RB, and Buehring GC: Relative sensitivity and speci-
ficity of agar gel immunodiffusion, enzyme immunosorbent assay,
and immunoblotting for detection of anti-bovine leukemia virus an-
tibodies in cattle. J Virol Methods 2002;104:33–39.
18. Sambrook J, Fritsch EF, and Maniatis T: Molecular Cloning, a
Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring
Harbor, NY, 1989, pp. 18.47–18.68.
19. Kramme PM, Thomas CB, and Schultz RD: Contribution of bovine
leukemia virus-infected B cells to the number of circulating B cells
in cattle. Comp Haematol Int 1994;4:96–101.
20. Voller A, Bidwell D, and Bartlett A: Enzyme-linked immunosor-
bent assay. In: Manual of Clinical Immunology (Rose NR and
Friedman H, eds.). American Society for Microbiology, Washing-
ton, DC, 1980, pp. 359–371.
21. Trainin Z, Merom R, Barnea A, and Ungar-Waron H: Common re-
activity of bovine and human sera towards bovine lymphoid tumor
cells. In: Comparative Leukemia Research (Clemmesen J and Yohn
DS, eds.). Bibl Haemat, No. 43. Karger, Basel, 1976, pp. 232–234.
22. Cereda PM, Debiaggi M, and Romero E: Antibody to bovine leu-
kemia virus in sera from leukemic patients. Microbiologica 1984;
7:107–112.
23. Nastac E, Stoian M, Athanasiu P, Hozoc M, Mittelu G, and
Manolescu N: Investigations on serum antibodies to a virus (3053
LB) isolated from the plasma of a leukemic cow, in milkers, cows
with leukosis and apparently healthy cows. Rev Roum Virol 1974;
25:53–59.
24. Clausen J, Hoff-Jorgensen R, and Rasmussen HB: Antibody reac-
tivity against animal retrovirus in multiple sclerosis. Acta Neurol
Scand 1990;81:223–228.
25. Malivanova TF and Litinov SV: Antibodies to retroviruses of types
C and D in female patients with benign and malignant mammary
tumours. Acta Virol 1990;34:19–20.
26. Bunger I, Khalaf H, and Rimpler M: Prufung von Antigenprapa-
rationen des virus der enzootischen Rinderleukoses auf Eignung
im Immunoblot. Dtsch Tierartl Wschr 1996;103:516–519.
27. Norman LM and Hohenboken WD: Genetic differences in con-
HUMAN ANTIBODIES TO BOVINE LEUKEMIA VIRUS
centration of immunoglobulins G1 and M in serum and colostrum
of cows and in serum of neonatal calves. J Anim Sci 1981;53:1465–
1472.
28. McMichael AJ: Principles of immunology. In: Oxford Textbook of 38.
Medicine (Weatheral DJ, Ledingham JGG, and Warrell DA, eds.),
Vol. 1, 3rd Ed. Oxford University Press, Oxford, 1996, p. 147.
29. Mueller N, Evans A, Harris NL, Constock GW, Jellum E, and Mag- 39.
nus K: Hodgkin’s disease and Epstein-Barr virus: Altered antibody
production before diagnosis. N Engl J Med 1989;320:689–695. 40.
30. Jefferis R, Reimer CB, Skvaril F, et al.: Evaluation of monoclonal
antibodies having specificity for human IgG sub-classes: Results of
an IUIS/WHO collaborative study. Immunol Lett 1985;10:223–252. 41.
31. Molander CW, Kniazeff AJ, Boone CW, Paley A, and Imagawa
DT: Isolation and characterization of viruses from fetal calf serum.
In Vitro 1972;7:168–173. 42.
32. Roberts DH, Lucas MH, Wibberly G, and Chasey D: Survival of
bovine leucosis virus in bovine whole blood, serum and plasma. J
Biol Stand 1981:9:469–473.
33. Graves DC and Ferrer JF: In vitro transmission and propagation of 43.
the bovine leukemia virus in monolayer cell cultures. Cancer Res
1976;36:4152–4159.
34. Barnes RM, Johnson PM, Finn R, and Harvey M: Human serum
antibodies reactive with dietary proteins. Int Arch Allergy Appl
Immunol 1988;87:184–188.
35. Evans AS: Causation and disease: Effect of technology on postu-
lates of causation. Yale J Biol Med 1991;64:513–528.
36. Hino S, Yamaguchi K, Katamine S, et al.: Mother-to-child trans-
mission of human T-cell leukemia virus type-1. Jpn J Cancer Res
1985;76:474–480.
37. Orloff SL, Wallingford JC, and McDougal JS: Inactivation of hu-
1113
man immunodeficiency virus type I in human milk: Effects of in-
trinsic factors in human milk and of pasteurization. J Hum Lact
1993;9:13–17.
Altaner C, Altanerova V, Ban J, et al.: Human cells of neural ori-
gin are permissive for bovine leukemia virus. Neoplasma 1989;
36:691–695.
Diglio CA, Ferrer JF: Induction of syncytia by the bovine C-type
virus leukemia virus. Cancer Res 1976;35:1056–1057.
Slavikova K, Zajac V, Reinerova M, et al.: Bovine leukemia pro-
virus in the DNA of different infected host cells. Neoplasma 1986;
33:671–678.
Takahashi K, Takezaki T, and Oki T: Inhibitory effect of maternal
antibody on mother-to-child transmission of human T-lymphocytic
virus type-I. Int J Cancer 1991;49:673–677.
Hino S, Katamine S, Miyamoto T, et al.: Association between ma-
ternal antibodies to the external envelope glycoprotein and verti-
cal transmission of human T-lymphotropic virus type I. J Clin In-
vest 1995;95:2920–2925.
Yoshinaga M, Yashiki S, Toshaka O, et al.: A maternal risk fac-
tor for mother-to-child HTLV-I transmission. Jpn J Cancer Res
1995;86:649–654.
Address reprint requests to:
Gertrude Case Buehring
School of Public Health
University of California
Berkeley, California 94720
E-mail: buehring@uclink4.berkeley.edu
This article has been cited by:

1. Dominic F. Qualley, Bethany L. Boleratz. 2014. Expression, purification, and characterization of full-
length bovine leukemia virus Gag protein from bacterial culture. Protein Expression and Purification 93,
32-37. [CrossRef]
2. Kyung-Mee Choi, Harrison Ndetan, Eric Johnson. 2012. A Cohort Mortality Study of Workers in a
Soup Manufacturing Plant. Archives of Environmental & Occupational Health 67:4, 199-205. [CrossRef]
3. E. S. Johnson, K.-M. Choi. 2012. Lung Cancer Risk in Workers in the Meat and Poultry Industries -
A Review. Zoonoses and Public Health 59:5, 303-313. [CrossRef]
4. M. N. Ruzina, B. V. Andrianov, G. O. Shaikhaev, G. E. Sulimova. 2012. Identification and classification
of bovine leukemia virus isolates in Russia and ukraine based on the pol viral gene polymorphism. Russian
Journal of Genetics 48:7, 723-730. [CrossRef]
5. Carol Marcus-Sekura, James C. Richardson, Rebecca K. Harston, Nandini Sane, Rebecca L. Sheets.
2011. Evaluation of the human host range of bovine and porcine viruses that may contaminate bovine
serum and porcine trypsin used in the manufacture of biological products. Biologicals . [CrossRef]
6. E.S. Johnson. 2011. Cancer mortality in workers employed in cattle, pigs, and sheep slaughtering and
processing plants. Environment International 37:5, 950-959. [CrossRef]
7. Stephanie R. Downing, Emmanouil P. Pappou, Nita AhujaBehavior and Dietary Modification in the
Prevention of Colon Cancer 47-64. [CrossRef]
8. Thomas A. Sellers, Robert A. Vierkant, Julie Djeu, Esteban Celis, Alice H. Wang, Nagi Kumar, James
R. Cerhan. 2008. Unpasteurized milk consumption and subsequent risk of cancer. Cancer Causes &
Control 19:8, 805-811. [CrossRef]
9. Tina Moore, Paul Brennan, Nikolaus Becker, Silvia de Sanjosé, Marc Maynadié, Lenka Foretova,
Pierluigi Cocco, Anthony Staines, Alexandra Nieters, Rebecca Font, Andrea 't Mannetje, Véronique
Benhaim-Luzon, Paolo Boffetta. 2007. Occupational exposure to meat and risk of lymphoma: A
multicenter case-control study from Europe. International Journal of Cancer 121:12, 2761-2766.
[CrossRef]
10. X ZHAO, G BUEHRING. 2007. Natural genetic variations in bovine leukemia virus envelope gene:
Possible effects of selection and escape☆. Virology 366:1, 150-165. [CrossRef]
11. James S Lawson, Walter H Günzburg, Noel J Whitaker. 2006. Viruses and human breast cancer. Future
Microbiology 1:1, 33-51. [CrossRef]