You are on page 1of 12

Bacterial Transposon Tn5: Evolutionary Inference8

Douglas E. Berg,* Claire M. Berg,? and Chihiro Sasakawa*$

of Microbiology and Immunology,
Washington University School of Medicine

Sciences Group,


and of Genetics,

of Connecticut

elements induce spontaneous
promote genome rearrangements, regulate gene expression, and participate in the horizontal spread of
genes encoding traits such as antibiotic resistance among bacterial genera too
distantly related to undergo homologous recombination.
Here we review the bacterial
Tn5 and focus on those aspects of its functional organization
which provide insights into how it and other elements may have
arisen, proliferated, and evolved.


Transposable elements are discrete DNA segments that exhibit the specialized
ability to move from site to site in a genome independent of extensive DNA sequence
homology. Their existence was first documented by McClintock in her prescient
genetic analyses of unstable alleles in maize. More recently the existence and importance
of transposable elements has been established in many groups of organisms. As
McClintock first showed, transposable elements often cause spontaneous mutations,
regulate the expression of genes near their insertion sites, and induce cycles of chromosome breakage and rearrangement. Transposition-like phenomena are also implicated in the DNA rearrangements that occur during the normal development of
the immune system and in the unscheduled chromosomal translocations associated
with many types of cancer (for reviews see Shapiro [ 19831).
Transposable elements play a special role in bacterial evolution because of their
ability to move between the chromosome and the various plasmid and temperate
phage DNAs resident in a bacterial cell and, when piggybacked on these molecules,
to move between unrelated bacteria in a population. Virtually any gene can become
associated with a transposable element, and complex elements called transposons
containing genes whose functions are unrelated to movement are now commonplace.
Those that encode resistance to clinically useful antibiotics have been studied most
intensively because their resistance determinants have provided selectable genetic
markers that are valuable in basic studies of transposition mechanisms and in the
use of transposons as mutagens in many bacterial species, and because the epidemic
spread of antibiotic resistance among bacterial populations has severely compromised
strategies for the treatment and prevention of infectious disease. Consideration of the
clonal nature of most bacterial growth, the participation of transposable elements in
1. Key words: insertion sequences, antibiotic resistance, interspecific gene exchange, mechanism of
transposition, DNA sequence recognition.
$ Current address: Institute of Medical Science, University of Tokyo, Minato-ku, Tokyo 108, Japan.
Address for correspondence and reprints: Douglas E. Berg, Department of Microbiology and Immunology, Box 8093, Washington University School of Medicine, 4566 Scott Avenue, St. Louis, Missouri
Mol. Biol. Evol. 1(5):41 l-422. 1984.
0 1984 by The University of Chicago. All rights reserved.


4 12

Berg, Berg, and Sasakawa

a network of exchange that transcends classical barriers between species and genera,
and the rapidity of the spread of resistance following the onset of routine usage of
antibiotics in medicine and agriculture suggest that it is especially during periods of
drastic environmental change that transposable elements make their greatest contributions to the adaptability and evolution of bacterial populations.
How transposable elements have evolved is of great interest, and insights into
their evolution are emerging from analyses in bacteria of their functional organization,
and the mechanism and regulation of their movement. Among the best known of
the procaryotic elements is the kanamycin resistance transposon Tn5, and it is on
this element that the following discussion focuses.


Functional Organization
Tn5 is a 5,700-base pair (bp) composite element in which a pair of simpler
mobile elements, the 1,534-bp insertion sequences ISSOR and ISSOL, are present in
inverted orientation and bracket a central region that contains genes encoding resistance
to kanamycin (kan) and to streptomycin (str; fig. 1; for review, Berg and Berg [ 19831).
Tn5 was discovered as a component of a bacterial R factor plasmid. It exhibits no
significant similarity to the half-dozen transposable elements indigenous to the genome
of Escherichia coli K-12 or to sequences in the vast majority (>95%) of several
hundred enteric isolates that have been screened. Tn5 transposes with high frequency
and inserts into many sites, including into a small number of hot spots. When inserted
into an operon, Tn5 blocks the normal transcription of distal genes. At certain sites,
however, it causes low-level constitutive expression of distal genes. Like several of
the other bacterial elements, Tn5 generates a direct 9-bp duplication of the target
sequence at its site of insertion, probably reflecting staggered cuts made in the target
DNA during transposition and limited repair synthesis to fill in the resulting gap.






( str )*




Pvu IL












FIG. 1.-Top, Functional organization of transposon Tn5 (Rothstein and Reznikoff 1981; Berg et al.
1982). The gene encoding resistance to streptomycin is indicated in parentheses because its transcript is
not translated in Escherichia coli (Putnoky et al. 1983; Selvaraj and Iyer 1984). Bottom, The structure of
ISSOR, showing the DNA sequence of its essential transposase recognition sites. ISSOLis identical except
for a single base pair change 112 bp from the inside end (Auerswald et al. 1980; Isberg et al. 1982; Johnson
et al. 1982; Sasakawa et al. 1983). 0 = outside; I = inside.



4 13

Of the two IS elements in Tn5, ISSOR encodes a cis-acting protein, transposasc,

that is necessary for IS50 and Tn5 movement (Isberg and Syvanen 198 1; Rothstein
and Reznikoff 198 1). Transposase is thought to act by binding distinctive 19-bp
nucleotide sequences near the ends of its cognate (IS50) elements (Johnson and
Reznikoff 1983; Sasakawa et al. 1983). ISSOL differs from ISSOR in that it contains
the promoter used for expression of the kan gene in TnSs central region, and also
an ochre allele of the transposase (tnp) gene. Both the kan promoter and the mutant
tnp allele arise from a substitution of one nucleotide pair 112 bp from ISSos inside
end (Rothstein and Reznikoff 198 1).
Derivatives of Tn5, with direct rather than inverted terminal repeats of IS50,
have been generated and found to transpose about as well as Tn5 wild type. They
are somewhat unstable in recombination proficient host cells, however, because homologous recombination between the IS elements results in the loss of one IS50
element plus TnSs central region. In addition, the central segment of Tn5 can be
replaced by other DNA segments without impairing transposition. Thus we view Tn5
as a composite element, selected because of its antibiotic resistance and its transposability. Its mobility arises simply from the ability of pairs of IS elements to move
in unison, carrying with them interstitial segments; the inverted orientation of IS
elements in many transposons could be fortuitous or a reflection of selection for
stability in the face of homologous recombination.
The Ends of IS50
Typically, the base sequence at one end of an IS element is a short, imperfect,
inverted repeat of the sequence at the other end, for example, 15/ 16 bp in IS5,
18/23 bp in ISI, and 36/37 bp in y6 (Iida et al. 1983). Because specific sequences of
such lengths are unlikely to occur in the bacterial genome by chance alone, these
repeats are usually equated with transposase recognition sites, and the sequence mismatches in them suggest that transposition does not involve direct base pairing between
IS element ends. (These short terminal inverted repeats within IS elements should
not be confused with the reverse duplication of entire IS elements in transposons
such as Tn5). The ends of IS50 are intriguing because they are only well matched
over eight of the first nine base pairs (fig. 1, bottom). These 9 bp seemed too short
to constitute an entire transposase recognition site, and recent analyses of IS50 mutants
have shown that the recognition sites actually extend 19 bp in from each end, with
only 12 of the 19 positions matched (Johnson and Reznikoff 1983; Sasakawa et al.
1983). Because the outside 19 bp of IS50 seem to be necessary only for transposase
recognition, whereas the 19 bp at the inside end also contain the last few codons of
the transposase gene, the dissimilarity of these two ends may reflect an evolutionary
compromise between the needs for an effective transposase and for an efficient
transposase binding site.
Tests such as those diagrammed in figure 2 had shown that ISSos outside and
inside ends differ in activity. Transposition mediated by a pair of outside ends is lOOto 1,OOO-foldmore frequent than inverse transposition mediated by a pair of inside
ends. In addition, transposition mediated by one inside end plus one outside end
occurred with an efficiency similar to that mediated by a pair of outside ends. These
results suggested a model in which the c&acting transposase is activated by binding
to the outside end and then tracks along the DNA molecule until it binds a second
outside or inside end with less discrimination (Isberg and Syvanen 198 1; Sasakawa
and Berg 1982).

4 14

Berg, Berg, and Sasakawa

,.._. _





4I a

L_____ _________________________+_.

I _ - - ___ _ __-.




I 3/710

c-- t
34 /60
0 1 -_______ _29t3,1_2.9________ __ _____ ___--_--++___d x :



to A

______ 0








26 /60

? , O/60

D (-


FIG. 2.-Functional
test of the transposition proficiency of ISWs outside (0) and inside (I) ends.
Shown at left is the 12-kb dimeric pBR333::Tn5 plasmid marked with resistances to ampicillin (a) and
kanamycin (k). IS50 elements are indicated by jagged lines, and the h target sequences are indicated by
dashed lines. The relative positions of the 2.9- and 3.1-kb fragments generated by digestion with XhoI were
used to deduce the structures of the Ran Amp trans&xition products (Sasakawa and Berg 1982).

Mechanism of Tn5 Transposition

Transposition mediated by IS50 always involves a fragment of the donor DNA
molecule. This suggested that ISSO-mediated transposition is conservative (fig. 3). On
this model, double-strand cleavages separate the mobile DNA segment from its vector,
the element is ligated to its target DNA, and the remainder of the donor DNA
molecule is released as a linear DNA fragment and lost, probably through exonucleolytic degradation. Precocious replication of one of the donors siblings fills the
niche created by destruction of the donor molecule, so that in future generations the
mobile element is found at the original site as well as the new location, just as if it
had replicated while (instead of after) transposing (Berg 1977, 1983).
Transposition of other unrelated elements (e.g., Tn3, ISI, and temperate phage
Mu) often produces co-integrate molecules in which the donor and target DNAs are
joined together by direct repeats of the mobile element (See Heffron 1983; Iida et al.
1983; Toussaint and Resibois 1983). The co-integrate structure is generally interpreted
to indicate replicative transposition, and one popular model for how such transposition
might occur is drawn in figure 4.
Until recently there had been widespread acceptance of the notion that all elements
were like Tn3, replicating during transposition. But the findings that Tn5 probably
transposes conservatively, and that phage Mu and ISI can transpose without replicating
(Liebhart, Ghelardini, and Paolozzi 1982; Akroyd and Symonds 1983; Biel and Berg
1984), indicate that, consistent with the known diversity of transposable elements,
there may be a variety of transposition mechanisms, some replicative and some nonreplicative.
Regulation of Transposition
The increase in mobile element copy number associated with conservative as
well as replicative transposition to new sites contributes to the fixation of these elements



4 15






FIG. 3.-A model of conservative transposition (Berg 1977, 1983). The junctions between transposable
element and vector sequences undergo a blunt-end double-strand cleavage and the target DNA molecule
undergoes a staggered double-strand cleavage, the element being inserted between the broken ends of target
DNA, and ligated in place. The ends of the vector DNA are not religated because, unlike the element itself,
they do not contain sites which would be held together by transposase. Hence, a viable (circular) product
of Tn5 excision is not formed during transposition. The resultant casting off of the remains of the donor
molecule creates an empty niche which is refilled by precocious replication of a sibling molecule present
in the same cell (the new DNA synthesis is indicated by the interrupted lines at bottom).

in bacterial populations. Balanced against this, however, are detrimental effects: each
transposition event is potentially mutagenic, copies of any DNA segment scattered
in the genome facilitate rearrangements via homologous recombination, and excessive
expression of transposable element genes may be inherently deleterious. Tn5 (IS50)
and several other elements are now known to control the frequency of their own
transposition and thereby their proliferation within a genome. The gene in ISSOR,
which encodes transposase, also encodes a second protein, inhibitor, that diminishes
the transposition of IS50 and Tn5.
The inhibitor, unlike transposase, acts in trans, probably interferes directly with
transposase action, and appears to reduce transposition in proportion to the copy
number of ISSOR (Biek and Roth 1980; Isberg et al. 1982; Johnson et al. 1982). By
decreasing the chance that an invading IS50 (or Tn5) element will establish itself,
the inhibitor is advantageous for the resident IS50 Furthermore, although at least


Berg, Berg, and Sasakawa


FIG. 4.-A
model of replicative transposition
(adapted from Arthur and Sherratt [ 19791; Shapiro
[ 19791). At the onset of transposition
the target DNA undergoes a double-strand
cleavage, one strand at
each end of the mobile element is cleaved, and the free ends of transposable
element and target DNAs are
joined together forming replication forks. DNA synthesis proceeds inward from both ends until the element
is fully replicated. The resultant product is a co-integrate
in which donor and target DNAs are joined
together by direct repeats of the element. Co-integrates can be slowly resolved by homologous recombination
between the repeats in recA+ bacterial cells. However, many members of the Tn3 family encode a recombination function, resolvase, which mediates resolution by a site-specific crossover event independent
the host ret function. The Tn.%encoded resolvase operates so efficiently that T&-based
co-integrates have
only been isolated using mutants altered in resolvases structural gene or at its site of action. However,
resolution is not necessarily associated with the preceding steps, and hence this step is diagrammed
an interrupted
rather than a solid arrow.

Transposon Tn5


one copy of IS50 is beneficial (Hart1 et al. 1983), the inhibitor reduces the rate of
accumulation of additional copies of the resident element to deleterious levels.
Tn3 and ISIO, in contrast, limit transposition by inhibiting transposase synthesis:
Tn3 encodes a protein repressor of the transcription of its transposase gene (Heffron
1983). IS10 encodes a small mRNA molecule that interferes with the translation of
its transposase protein and whose synthesis may interfere with the synthesis of the
transposase message (Symons and Kleckner 1983). In addition, DNA molecules containing one copy of Tn3 generally cannot be used again as targets for Tn3 insertion
(Heffron 1983). Thus, there are numerous ways of controlling transposable element
movement, each of which diminishes the probability that a cell lineage containing
one of these elements will be recolonized by a closely related and potentially competing


We will consider how transposable elements may have arisen from preexisting
immobile genes, then proliferated and diverged.
IS50 Element Formation
Evolution is opportunistic, since genes that initially carry out one role are often
recruited for other functions. We imagine that ISSos transposase may be the end
product of a phylogeny begun by chance fusions between genes encoding proteins
that bind to, track along, nick, and reseal DNA molecules, proteins with properties
much like present-day repressors, helicases, and topoisomerases. Some of the
transposase recognition sites currently found at the ends of transposable elements
may have evolved from operators to which ancestral repressor proteins had bound.
Therefore, neither the ancestral genes nor the primordial recognition sites need to
have been part of a preexisting mobile element.
The recognition sites at both ends of an IS element could have arisen by a reverse
duplication of the first binding site. However, the available data also suggest models
in which the two ends do not share a common ancestry (fig. 5). For example, an
incipient IS element may have been created by the linking of a transposase (tvlp) gene
and a single recognition site, 0. The primitive transposase activated by binding
this 0 site would migrate on the DNA molecule, occasionally encounter another
sequence for which it possessed fortuitous affinity, and cause transposition of the
segment bounded by 0 and this second sequence. Natural selection would favor
mutant elements which transpose efficiently, and mutations would accumulate in the
chance binding sites, in the tnp gene, or, following changes in transposase specificity,
in the 0 sequence itself. Because chance binding sites would be rare, the earliest
IS elements may have been quite long, but shorter derivatives could then have arisen
by spontaneous internal deletions drawing the distant end closer to the tnp gene.
IS elements could also become truncated by the occasional binding of transposase
to other sequences closer to the tnp gene. Once a shortened IS element had moved
from its earlier location, the recapture of its previous recognition site would not occur.
Rather, selection for more efficient transposition would speed the evolution of the
new recognition site, of the tnp gene, and of the original 0 recognition site as well.
The current form of IS50, the dissimilarity between its two ends, and the overlap
between essential coding and recognition sequences may reflect a pressure for compactness inherent in the transposition process itself. Some elements (e.g., y6 and
ISI 721; see Heffron [ 19831) are larger and encode, in addition to transposase, a second
enzyme called resolvase which catalyzes site-specific reciprocal crossovers and the


Berg, Berg, and Sasakawa








- ---+y___




tnp r,__








FIG. 5.-A model to explain the evolution of an IS element from a simple immobile gene complex.
The boxes represent actual or potential transposase binding sites which differ in DNA sequence. 0, 0,
and 0 represent successive stages in the evolution of the 0 end binding site, and trip, tnp, and tnp
represent successive forms of the transposase gene, each encoding a protein of somewhat different specificity
(coadapted to its recognition sites). The right fork represents a spontaneous deletion within the IS element,
and the left fork represents a shortening of the IS element by transposition using an alternative fortuitous
recognition site near to or overlapping the end of the tnp gene.

breakdown of co-integrates formed during replicative transposition. Although cointegrate formation and resolution may have made the evolution of elements in the
y6 family more complicated than that of IS50, it is likely that their current forms
reflect the pressure for decreased size equivalent to that envisioned for IS50.
The Origin of Tn5
Since each of the 1,534-bp repeats of Tn5 is itself a mobile (IS50) element, it
is likely that Tn5 was formed by insertions and/or genome rearrangements which
placed a pair of IS50 elements on both sides of a previously immobile (chromosomal)
segment encoding resistances to kanamycin and to streptomycin. Either original insertions of IS50 or a subsequent rearrangement or deletion separated the kan and str
genes from their original promoter. The resulting weak resistance caused by readthrough
transcription from the transposase gene in ISSOL would have conferred a low level
of resistance to kanamycin and related antibiotics and hence would have been advantageous to bacteria growing in antibiotic-contaminated
environments. Transposability would have helped the resistance determinant proliferate and become fixed
in the genomes of its new bacterial hosts and thus allowed time for the selection of
a better kan promoter. The formation of this promoter probably required only a
single base pair change in ISSOL. The simultaneous inactivation of the tnp gene is
unlikely to have impaired mobility, since the mutant allele is recessive to the functional
tnp+ allele present in the linked ISSOR element.
Comparisons of the DNA sequences of Tn5 and of another transposon, Tn903,
indicate relatedness between their kanamycin resistance genes but not their insertion
sequences (Beck et al. 1982). This suggests the repeated incorporation of previously
immobile resistance genes into new transposons. The apparent inability to translate
the message of TnSs str gene in E. coli (Putnoky et al. 1983; Selvaraj and Iyer 1984)
suggests that Tn5 did not arise in enteric bacteria. The finding of significant sequence
homology between the kanamycin resistance genes of Tn5 and those from Streptococcus

Transposon Tn5


(Trieu-Cuot and Courvalin 1983) and from Streptomyces (Thompson and Gray,
1983) provides direct support for the intriguing hypothesis of Benveniste and Davies
( 1973) that the aminoglycoside resistance genes of gram-negative bacteria may have
come from the gram-positive organisms that produce aminoglycoside antibiotics.
Transposons unrelated to Tn5 probably arose by an analogous series of insertions,
genome rearrangements, and deletions which in some cases extended into the IS
element, rendering one (Tnl721) or both (Tn3) IS elements immobile (see Heffron
Transposable Element Maintenance

and Proliferation

A half-dozen different species of IS elements have been found in the genome of

Escherichia coli K- 12 (see Iida et al. 1983), a laboratory strain that was isolated from
nature many years ago. Several IS elements are present at a level of five to 10 copies
per chromosome. It is likely that similar numbers of elements are present in the
genome of many other bacterial clones, and we are now beginning to understand
factors that contribute to the abundance of transposable elements and their proliferation
within a genome; among the factors are (i) proliferation inherent in their mechanisms
of transposition, (ii) favorable mutations they cause, (iii) benefits host cells may derive
directly from expression of their transposition genes, and (iv) their participation in
networks of intergeneric gene exchange.
Doolittle and Sapienza (1980) and Orgel and Crick (1980) proposed that transposable elements constitute a special class of DNA termed selfish DNA. Although
transposable elements might increase by chance alone (Ohta and Kimura 198 l),
Doolittle and Sapienza (1980) and Orgel and Crick (1980) maintained that inherent
in the mechanism of transposition is a tendency for the element to increase relative
to other genomic sequences. Two aspects of this concept now require modification.
First, it had been argued that the current abundance of transposable elements is
attributable solely to replication during transposition. Although it is obvious that
replicative transposition increases transposable element copy number (fig. 4), the
converse, that any increase in copy number demonstrates a replicative mechanism,
is not true. As shown in figure 3, during conservative transposition the vector from
which the mobile element had been taken is lost, and this loss is compensated by
the overreplication of a sibling molecule to fill the empty niche. A cell lineage which
retains a copy of the element at the original site as well as a second copy at a new
site results just as surely as if replication had accompanied, rather than followed,
Second, Doolittle and Sapienza ( 1980) argued that no other explanation for
the origin or maintenance of transposable elements is necessary . . . the search for
other explanations [other than transposition] may prove, if not intellectually sterile,
ultimately futile. Despite the dramatic impact of this statement, such a position is
counterproductive. Given the tendency of evolution to tinker by mutation with preexisting genes, thereby recruiting them for new roles often quite different from those
they had originally fulfilled, it seems naive to imagine that transposable element genes
never confer a selective advantage on their hosts. Recent studies have shown that the
presence of ISSOR or Tn5 permits entire populations of E. coli cells to adapt more
rapidly to sudden environmental shifts in chemostat cultures (Biel and Hart1 1983;
Hart1 et al. 1983). The increased fitness of the ISSO-containing population is evident
immediately rather than after a long lag, is independent of the position of IS50 or


Berg, Berg, and Sasakawa

Tn5 in the genome, and is also independent of the movement of IS50 to new sites.
Thus IS50 seems to make its contribution through generally improved cell physiology
rather than an increased frequency of favorable mutations. This intriguing contribution
to fitness, probably mediated by ISSORs transposase or inhibitor function, may have
preceded the formation of IS50 as a mobile element.
A second contribution that transposable elements can make to bacterial fitness,
the occasional induction of favorable mutations, has been illustrated by experiments
with TnlO (Chao et al. 1983). The advantage that TnlO confers, although real, appears
less rapidly than that conferred by Tn5, probably because TnlO-induced favorable
mutations occur relatively rarely and long periods of time are required for cells
containing these mutations to outgrow those with the parental genotypes. Because
IS50 makes a direct physiological contribution to all cells in a population, the occasional
beneficial mutations which it must also induce have not been evident.
The ability of IS elements to mediate the transposition of auxiliary genes also
increases the numbers of bacteria harboring these elements. For example, once Tn5
had formed, the frequency of TnS-containing bacteria increased in many environments
simply through the preferential killing of Kan organisms that lacked Tn5 and repopulation of the niches by clones that contained Tn5. The increased frequency of
Tn5 would increase the sizes of the reservoirs from which solo IS50 elements could
As transposable elements increase in copy number, they are subject to selective
forces in addition to those operating on immobile genes. Transposition, whether
conservative or replicative, causes proliferation of the mobile elements and thus places
each member of a family in competition with its siblings resident in the same cell.
Under these conditions elements encoding transposases that act preferentially in cis
and inhibitors that act in trans would be favored. Such transposase and inhibitor
proteins should partially isolate each IS element from its siblings in the same cell
and thereby permit the gradual divergence of the members of the family. In cases of
transposase and inhibitor competing for the same sites at the ends of an element, the
divergence of members of a transposable element family would be selected directly.

We are grateful to D. Hart1 for stimulating discussions and critical readings of

this manuscript. This work was supported by U.S. Public Health Service Research
grants ROlAI18980
and ROlAI14267
to D. E. B. and ROlAI19919
C. M. B.


AKROYD, J. E., and N. SYMONDS.1983. Evidence for a conservative pattern of transposition

of bacteriophage Mu. Nature 303:84-86.
ARTHUR, A., and D. SHERRATT.1979. Dissection of the transposition process: a transposonencoded site-specific recombination system. Mol. Gen. Genet. 175267-274.
AUERSWALD,E., G. LUDWIG,and H. SCHALLER.1980. Structural analysis of Tn5. Cold Spring
Harbor Symp. Quant. Biol. 45: 107-l 13.
BECK, E., G. LUDWIG,E. A. AUERSWALD,B. REISS, and H. SCHALLER.1982. Nucleotide
sequence and exact localization of the neomycin phosphotransferase gene from transposon
Tn5. Gene 19:327-336.

Transposon Tn5 421

BENVENISTE,R., and J. DAVIES. 1973. Aminoglycoside inactivating enzymes in actinomycetes
similar to those present in clinical isolates of antibiotic resistance bacteria. Proc. Natl. Acad.
Sci. USA 70:2276-2280.
BERG, D. E. 1977. Insertion and excision of the transposable kanamycin resistance determinant
Tn5. Pp. 555-558 in A. I. BUKHARI, J. A. SHAPIRO,and S. L. ADHYA, eds. DNA insertion
elements, plasmids and episomes. Cold Spring Harbor Press, Cold Spring Harbor, N.Y.
. 1983. Structural requirement for ISSO-mediated gene transposition. Proc. Natl. Acad.
Sci. USA 79:792-796.
BERG, D. E., and C. M. BERG. 1983. The prokaryotic transposable element Tn5. Biotechnology
inverted repeats of Tn5 are transposable elements. Proc. Natl. Acad. Sci. USA 79:26322635.
BIEK, D., and J. R. ROTH. 1980. Regulation of Tn5 transposition in Salmonella fyphimurium.
Proc. Natl. Acad. Sci. USA 77:6047-6051.
BIEL, S. W., and D. E. BERG. 1984. Mechanism of IS1 transposition in E. cob: choice between
simple insertion and cointegration. Genetics (accepted).
BIEL, S. W., and D. L. HARTL. 1983. Evolution of transposons: natural selection for Tn5 in
Escherichia coli K12. Genetics 103:58 l-592.
CHAO, L., C. VARGAS,B. B. SPEARS,and E. C. Cox. 1983. Transposable elements as mutator
genes in evolution. Nature 303:633-635.
DOOLITTLE,W. F., and C. SAPIENZA.1980. Selfish genes, the phenotype paradigm and genome
evolution. Nature 2?34:60l-603.
HARTL, D. L., D. E. DYKHUIZEN,R. MILLER,L. GREEN, and J. DEFRAMOND.1983. Transposable
element IS50 improves growth rate of E. coli cells without transposition. Cell 35:503-5 10.
HEFFRON, F. 1983. Tn3 and its relatives. Pp. 223-260 in J. A. SHAPIRO, ed. Mobile genetic
elements. Academic Press, New York.
IIDA, S., J. MEYER, and W. ARBER. 1983. Prokaryotic IS elements. Pp. 159-221 in J. A.
SHAPIRO,ed. Mobile genetic elements. Academic Press, New York.
ISBERG,R. R., A. L. LAZAAR, and M. SYVANEN.1982. Regulation of Tn5 by the right-repeat
proteins: control at the level of the transposition reaction? Cell 30:883-892.
ISBERG,R. R., and M. SYVANEN.198 1. Replicon fusions promoted by the inverted repeats of
Tn5: the right stem is an insertion sequence. J. Mol. Biol. 150:15-32.
JOHNSON,R. C., and W. S. REZNIKOFF. 1983. DNA sequences at the ends of transposon Tn5
required for transposition. Nature 304:280-282.
JOHNSON,R. C., J. C. P. YIN, and W. S. REZNIKOFF. 1982. Control of Tn5 transposition in
Escherichia coZi by protein from the right repeat. Cell 30:873-882.
LIEBHART,J., P. GHELARDINI,and L. PAOLOZZI.1982. Conservative integration of bacteriophage
Mu DNA into pBR322 plasmid. Proc. Natl. Acad. Sci. USA 79:436 l-4366.
OHTA, T., and M. KIMURA. 198 1. Some calculations on the amount of selfish DNA. Proc.
Natl. Acad. Sci. USA 78: 1129- 1132.
ORGEL, L. G., and F. H. C. CRICK. 1980. Selfish DNA: the ultimate parasite. Nature 284:604607.
PUTNOKY, P., G. P. Kiss, I. OTT, and A. KONDOROSI. 1983. Tn5 carries a streptomycin
resistance gene downstream from the kanamycin resistance gene. Mol. Gen. Genet. 191:288294.
ROTHSTEIN, S. J., and W. S. REZNIKOFF. 198 1. The functional differences in the inverted
repeats of Tn5 are caused by single base pair nonhomology. Cell 23: 19 1-199.
SASAKAWA,C., and D. E. BERG. 1982. ISSO-mediated inverse transposition: discrimination
between the two ends of an IS element. J. Mol. Biol. 159:259-27 1.
SASAKAWA,C., G. F. CARLE, and D. E. BERG. 1983. The sequences essential for transposition
at the termini of IS50. Proc. Natl. Acad. Sci. USA 80:7293-7297.


Berg, Berg, and Sasakawa

SELVARAJ,G., and V. N. IYER. 1984. Transposon Tn5 specifies streptomycin resistance in

Rhizobium. J. Bacterial. 158:580-589.
SHAPIRO,J. A. 1979. Molecular model for the transposition and replication of bacteriophage
Mu and other transposable elements. Proc. Natl. Acad. Sci. USA 76: 1933-1937.
ed. 1983. Mobile genetic elements. Academic Press, New York.
SYMON~,R. W., and N. KLECKNER. 1983. Translational control of ISlO transposition. Cell
34:683-69 1.
THOMPSON,C. J., and G. S. GRAY. 1983. Nucleotide sequence of a streptomycete aminoglycoside
phosphotransferase gene and its relationship to phosphotransferases encoded by resistance
plasmids. Proc. Natl. Acad. Sci. USA 80:5 190-5 194.
TOUSSAINT,A., and A. RESIBOIS. 1983. Phage Mu: transposition as a life-style. Pp. 105-l 58
in J. A. SHAPIRO,ed. Mobile genetic elements. Academic Press, New York.
TRIEU-CUOT, P., and P. COURVALIN.1983. Nucleotide sequence of the Streptococcus fuecaks
plasmid gene encoding the 35-aminoglycoside phosphotransferase type III. Gene 23:33 l341.
MASATOSHI NEI, reviewing


July 7, 1983; revision received January 23, 1984.