Mol Gen Genet (1984) 193:493-499 ~ Springer-Verlag 1984

Characterization of the New Insertion Sequence IS91 from an Alpha-Hemolysin Plasmid of Escherichia cob'* Esmeralda Diaz-Aroca, Fernando de la Cruz, Juan C. Zabala, and Jos6 M. Ortiz Departamento de B]oquimica, Facultad de Medicina, Pollgono de Cazofia s/n, Santander, Spain

Summary. IS91 is a 1.85 kb insertion sequence originally resident in the ~-hemolytic plasmid pSU233. The element was transposed sequentially from this plasmid to pACYC184, to R388, and to pBR322. Both cointegrates and simple insertions of the element were obtained. A detailed restriction enzyme map of the element is presented. This does not bear any relationship to the maps of previously described insertion sequences. Furthermore, hybridization between these sequences and 1S91 could not be demonstrated. Deletion derivatives of ISgl were constructed which are unable to transpose. However, their transposition can be complemented in trans by wild-type elements. One of these deletion derivatives has been genetically labeled with a kanamycin resistance marker from Tn5. When this new element was complemented for transposition, only about 2% of the transposition products were cointegrates. Thus, the behavior of IS91 is better explained by transposition models that allow direct transposition.

Introduction

Trhnsposable elements are specific sequences of DNA that can transpose more or less at random from one replicon to another. Transposable elements contain genes whose expression is absolutely required lbr their own transposition (transposase genes), and flanking inverted repetitions at which the transposases are presumed to act (see Kleckner 1981 for a recent review). Transposable elements provide sophisticated mechanisms for the dissemination of genes among structurally unrelated replicons, and thus can be considered to be active devices that promote evolutionary change in bacteria. We are interested in the evolutionary relationship among ~t-hemolysin-producing plasmids (Hly plasmids). It has previously been shown that Hly plasmids belonging to four incompatibility groups contain almost identical hly * Part of this work was carried out by E. Diaz-Aroca as a requirement for her degree in Sciences. The work is published (Esmeralda Diaz-Aroca. Tesina de Licenciatura, Universidad Aut6noma de Madrid, 1983) and it contains the complete details of procedures and results of the cloning experiments and the restriction maps of the plasmids shown in this work. It is available from the authors upon request Off'print requests to: F. de la Cruz

genes (de la Cruz et al. 1980), suggesting that the hly genes can transfer between plasmids. During the course of a study to detect transposable elements in these Hly plasmids, it was discovered that all of them contain multiple copies (2-6) of an IS element called IS91 (Zabala et al. 1982). This work focusses on the genetic andmolecular characterization of IS91. It shows that IS91 is a new IS element, unrelated to the previously described ones, and describes some of the essential features of its mechanism of transposition. Materials and Methods Strains and Plasmids. The strains used are described in Ta-

ble 1 and the plasmids in Table 2. Plasmid-containing strains were constructed either by conjugation on the surface of agar plates (Bradley et al. 1980) or by transformation (Brown et al. 1979). All strains were checked for the presence of plasmids by their antibiotic resistance patterns and by physical visualization of the plasmids using a rapid mini-scale preparation (Arthur and Sherratt 1979). The recA character was checked by UV sensitivity. Genetic Experiments. Bacteria were grown at 37 °C, unless otherwise indicated. The liquid medium used was Penassay broth (Difco, Detroit, USA); solid medium was usually Penassay broth solidified with 1.5% agar. Where indicated, M9 minimal medium (Miller 1972) solidified with 1.5% agar was used; this was supplemented with Casamino Acids (Difco) 0.5% plus L-tryptophan (20 ~tg/ml). The concentrations of drugs (and abbreviations) used for selection were: carbenicillin (Cb), 500~tg/ml; chloramphenicot (Cm),

Table 1. Strains used

Strain

Relevant genotype

Reference

UB5201

pro met recA56 gyrA

UB1636 UB1637

his lys trp rpsL his lys trp recA56 rpsL

SU121 SU141

UB5201(R388+pSU234) UB5201(pSU239+pBR322)

de la Cruz and Grinsted (1982) Grinsted et ai. (1982) de la Cruz and Grinsted (1982) This work This work

494 Table 2. Plasmids used

Plasmid

Derivation

Relevant phenotype

References

pACYCI84 pBR322 pSU233 pSU234 pSU235 R388 ~SU238 ~SU239 ~SU240 sSU241 sSU242 sSU243 sSU248 ~SU253

Cloning vector Cloning vector Natural isolate pACYCI84::ISgl pSU233 :: IS91 ::pACYC184 Natural isolate R388 :: IS91: :pSU234 R388:: IS91 pBR322 (bla:: IS91) pACYCI84:: IS91zt pSU234 Km~derivative pSU240 Km~derivative pACYC184:: IS91A kan R388: : IS91zl kan:: pSU248

Cm'TcrMob- (R388)Rep(pl 5A) AprTc~Mob- (R388)Rep(pMB8) Hly *TraF IncFVlI CmrTcr CmrTc~Hly Tp'Suq'raW IncW Cm'TcVI'prSu' Tp'Su~ Tc' T'Cm' Cm'Km"

Chang and Cohen (1978) Bolivar et al. (1977) de la Cruz et al. (1979) Zabala et al. (1982) Fig. 1 Zabala et al. (1982) Ward and Grinsted (1982) Fig. 2 This work, Fig. 2 This work, Fig. 2 This work, Fig. 1 This work, Fig. 3 This work, Fig. 3 This work, Fig. 3 This work, Fig. 3 This work, text

Krn r

Km~CrnrTcr Km~Cmel'c,Tp~Su,

25 lag/ml; nalidixic acid (Nx), 25 lag/ml; kanamycin sulphate (Km), 50 lag/ml; trimethoprim (Tp), 25 lag/ml in M9 minimal medium; tetracycline hydrochloride (Tc), 10 lag/ml and streptomycin sulphate (Sm), 200 lag/ml. Superscripts r or s to the abbreviations denote resistance or sensivity to that drug. Transposition Experiments. Donor strains, containing the transposon donor and recipient plasmids, were incubated in stab cultures at room temperature for at least 4 days, and then mated with suitable recipients. In all experiments. either the donor or the recipient molecule was a nonmobilizable plasmid (a derivative of pBR322 6r pACYCI84) and therefore, could not be mobilized unless physically linked with the Tra + plasmid (the other component of the donor strains). Physical linkage is brought about by cointegrate formation between both molecules, by means of a transposable element, and leads to transfer of the sequences of the nonmobilizable plasmid. This special kind of mobilization has been termed conduction (Clark and Warren 1979) and has been widely used to detect transposition of unselectabte transposons (Guyer 1978; de la Cruz and Grinsted 1982). The conduction frequency is the number of transconjugants containing the marker of the nonmobilizable plasmid (Td or Cm ' or Km ~ in the experiments described here) divided by the total number of transconjugants (i.e., in this case, the number of Tp t transconjugants, since plasmid R388 was the Tra ÷ plasmid). Transcomplementation of defective IS elements was cartied out in essentially the same way except that the donor strain contained, as well as both donor and recipient plasraids described above, a compatible helper ptasmid that contained the wild-type IS element. DNA Preparation and Manipulation. The method of preparation of DNA depended on its intended use: for estimating the number of plasmid species and their sizes, lysis with sodium dodecyl sulfate followed by a 15min clearing spin in an Eppendorf 5412 centrifuge (Arthur and Sherrat 1979) was used; for restriction enzyme analysis, plasmid DNA was prepared from overnight cultures (10 ml) by the method of Holmes and Quigley (1981); and for large-scale preparations, CsCl-ethidium bromide gradients were used, the method being essentially that described by Humphreys et al. (1975). Digestion with restriction endonucleases and analysis were as described by Grinsted et al. (1978). For the gen-

eration of in vitro recombinants, DNA from CsCl-ethidium bromide gradients was used. Specific DNA restriction fragments were isolated after horizontal slab gel electrophoresis by the electroelution methof of Yang et al. (1979). Filling-in of recessed Y-ends of restriction fragments was carried out with the Klenow-fragment of DNA polymerase I (New England Biolabs, Boston, USA) as described by Maniatis et al. (1982). Ligation of specific DNA fragments (using T4 DNA ligase from Biolabs) was carried out at DNA concentrations favoring inter- or intramolecular annealing (Dugaiczyk et al. 1975), as appropriate. Results

Transposition of IS91 to DiIferent Plasmids Plasmid pSU234 was obtained by transposition of IS91 from the ~-hemolytic plasmid pSU233 to pACYC184 (Zabala et al. 1982). A detailed physical map of pSU234 is shown in Fig. 1. Strain SU121 (containing R388 and pSU234) was crossed with strain UB1636. and Cm f transconjugants selected. They arose at a frequency of about 10 -6, and presumably contained cointegrates of R388 and pSU234 (see Materials and Methods). One of the transconjugants (strain SU122) was chosen and linkage of Cm ~ and Tp r examined by crossing this strain with UB5201 : of the resulting transconjugants, only 3% were Tp r Cm s, and the rest were Tp r CmL Examples of the plasmids contained in both types of transconjugants were analyzed and their structures are shown in Fig. 2. Plasmid pSU238 (a plasmid which determines Tp r and Cm r) is a cointegrate in which the donor and recipient replicons are fused by directly repeated copies of ISgl. Plasmid pSU239 (which determines Tp ' but not Cm') is a simple insertion of IS91 in R388. It seems likely that pSU239 originated from pSU238 in UB1636 (which is a Rec" strain) by recombination between the copies of IS91, since the region of insertion of IS91 is the same in both plasmids. It should be noted that plasmid pSU238 was completely stable in the recA strain UB5201 (data not shown). Ptasmid pSU239 was then used as donor of IS91 to pB322: strain SU141 (containing pSU239 and pBR322) was crossed with UB1636 and Tc' transconjugants were selected. They presumably contain cointegrates of pSU239

495

Eco RI

Eco RI H~n d~ est Eli IF/ earn HI Hih CJIX, ~ ~ - - - - - - - - ~ - ~ , ~ ' " / Hih cE/Eof GI

Bst E~ . ~ _ _ ~ ~

'E

.

.

.

.

.

1V

.

8st E1T~

Barn HI •

(AvoI)Xh I

I's

-

L

.""

~

Hfn ca

"BgtI

I

i

\

Bgl I

/'~,,,~'~,,JJ Pst I

Hfn ell Avo I Fig. I. Physical and genetic maps of plasmids pSU234 (pACYC184:: IS91) and pSU240 (pBR322 (bla)::lS91, defining coordinate systems for the plasmids. Plasmid pSU234 was obtained by transposition of IS91 from the ~-hemolytic plasmid pSU233 to pACYC184 (Zabala et al. 1982). Plasmid pSU240 was obtained as described in the text. The cleavage sites for 11 restriction endonucleases are shown. The radial lines joining the two circles represent the Hae[l cleavage sites, pSU234 has no sites for enzymes BgllI Or Pstl. pSU240 has no sites for 8glll or Xbal. The thick black lines represent the position of the IS9l sequences. They were approximately located by comparison of the physical maps of pSU234 and pSU240 with those of pACYCI84 and pBR322 respectively, and identification of invariant fragments. The exact location of IS91 in pSU234 and pSU240 is described in the text. The location of the relevant genes and restriction sites in pACYC184 and pBR322 were taken from Chang and Cohen (1978) and Sutcliffe (1978). cat=chloramphenicol acetyltransferase: rep = replication re,on: tet = tetracycline resistance gene; bla =fl-lactamase gene 9.8

03

~

02

t.8

3.15

2.3 .

.

bg

I

b¢

c~

bt

PSU239....

J

s

9,a2

I~!

g

--~

}l! !r! 1 .z.5 3 ~ 0 . 8 2

S

b

~j

i.•

3.38

b

""

.

I ~.oSb~ bt

2.~7

.I-i e

~ "

bl

bf

'"

I°'I

bg

bt

!L 0.5

g

t

Fig. 2. Restriction endonuctease cleavage maps of the plasmids R388, pSU238 (a cointegrate of R388 and pSU234 in which both parental plasmids are flanked by a direct repeat of IS91) and pSU239 (R388::1S91). The restriction map of R388 is already known (Ward and Grinsted 1982) except lbr the BstEll sites, which were mapped for this work. The maps are drawn to scale, except for the dotted segments. The sizes (in kitobase pairs) of all fragments except the largest are marked. The thick lines represent the location of the IS91 sequences. The broken lines joining the various maps represent the fragments presumably inserted or deleted that produced the different plasmids (see the text for details). The symbols for the endonuclease cleavage sites are: e, EcoRl; b, BamHI; s, SalGI; hg. Bglll; bt. BstEll and x. Xhol

and pBR322 (see Materials and Methods). Plasmid D N A was isolated from individual transconjugant colonies and used to transform UB1636. Tc ~Tp ' transformantswere analyzed for their plasmid content. Most of them were, according to restriction data, cointegrates of pSU239 and pBR322 containing deletions o f various lengths which removed the Tp ~ encoding genes (data not shown). The rest contained plasmids with simple insertions of IS91 within pBR322. One of these, pSU240, which contains IS9l inserted in the bla gene of pBR322, was analyzed in detail and its structure is shown in Fig. 1. The Physical Map o f lS91 Plasmids pSU234 and pSU240 were analyzed in detail to characterize IS91 (Fig. 1). The maps of the plasmids are

consistent with both containing a 1.85 kb + 0.1 kb insertion within t h e p a r e n t a l plasmid molecules, pACYC184 and pBR322, respectively. The sites of insertion could easily be located since the restriction maps of pACYC184 and pBR322 are known (Fig. 1). There is an XhoI site in pSU234 and pSU240 which is not present in the parental plasmid molecules. However, deletions that remove pSU234 sequences from the HindlII site to the XhoI site (plasmid pSU242; Fig. 3) are fully transposition proficient (Table 3) and, therefore, presumable contain an intact element. Furthermore, this XhoI site was neither duplicated after cointegrate formation (see pSU238 (Fig. 2) and pSU235 (a cointegrate of pSU233 and pACYCI84; Zabala et al. 1982, E. Diaz-Aroca, J.C. Zabala, F. de la Cruz and J.M. Ortiz, manuscript in preparation)), nor was it present at the original site of IS9l in

496

PLASMI[3 9e

pau2a4

~

ret

psu240

~J

Tel

pSU'242

C~

[~

~

Te.r

~''~CCat

Rep184

Re~32e ....

O~f~Ca(~

KaI

Kan

PSU2,8

~,,

•

ReOI84

Trans-Acting Gene Products in [$91

¢p I~

1980), ISIO (Hailing et al. 1982), IS15 (Labigne-Roussel and Courvalin 1983), IS21 (Willetts et al. 1981), 1S26 (Iida et al. 1982 and personal communication), IS50 (Auerswald et al. 1981), IS70 (Hille et al. 1983), IS102 (Ohtsubo et al. 1980) and IS903 (Oka et al. 1981). No similarity whatsoever could be found betwen IS91 and any of them. In addition, the 1.05 BstEII restriction fragment of pSU234 from within IS91 (Fig. 1) was isolated, labeled, and used as a probe to search for homologous sequences in various plasmids. In this way, homology of IS9l with IS/, IS2, IS3, IS8, IS10, IS21, IS50, IS903 and Tn3 (Heffron et al. 1979), Tn21 (de la Cruz and Grinsted 1982), and TnlO00 (Guyer 1978) was tested. There was no hybridization with any of the plasmids (data not shown). So, according to the conditions used (Grinsted et al. 1982), none of the above-mentioned elements is more than 85% homologous to the IS91 sequence used.

"~.~a,

Rep,84

Fig, 3. Physical maps of plasmids containing .IS91 and of derivatives constructed in vitro. Plasmids pSU234 (Zabala et al. 1982) and pSU240 are shown in Fig. 1. pSU241 was constructed by isolating a 4.95 kb fragment produced by BstEll digestion of pSU234 (Fig. I) and intramolecular ligation. Thus. pSU241 is a pSU234 derivative which "has suffered a deletion removing the internal 1.05 kb gstEll fragment of IS91 (the dotted line in the pSU241 map of the Figure). pSU242 was constructed by ligation of a 1.55 kb Hindlll-Xhoi fragment of Tn5 (Jorgensen et al. 1979) with a 3.75 kb Hh~dlll-Xhoi fragment of pSU234 (from coord. 1.5 kb counterclockwise up to coord. 3.75 in Fig. 1). Thus, the effect of this construction is the substitution of the tet gene of pSU234 by a kan gene of Tn5. pSU243 was constructed in a similar way but using a HindilI-SalGl fragment from Tn5 (Jorgensen et al. 1979) and a 5.6 kb HindllI-SalGi fragment from pSU240, pSU248 was constructed by cleavage of pSU241 at its single BstEll site and blunt-end ligation of this fragment to the Hindl i l-SalGI Kmr fragment of Tn5 alter filling the 3' recessed ends of both fragments. Thus. pSU248 contains the Kmr fragment inserted at the single BstEll site of pSU241. The maps of all the plasmids have been confirmed by digestions with Pstl. Acal. Xhol, SalGI, Hindlll and BstE|l. o:Xhoi, e: BstEIl, -: SatGI: Q: HiedllI

pSU233 (E. Diaz-Aroca et al., manuscript in preparation). Therelbre, it can be assumed that the Xho[ site was produced by the juxtaposition of IS91 and pACYC184 sequences at the site of insertion in pSU234 or of IS91 and pBR322 sequences in pSU240 (i. e., the end of IS91 contains part of the Xhol recognition sequence). Hence, this XhoI site precisely locates one end of IS91 in plasmids pSU234 and pSU240. Relationships of IS91 to Preriousl.v Characteri:ed IS Elements ISgl is noticeably larger (1.85 kb) than most known IS elements. The detailed endonuclease cleavage map of IS91 (Fig. 1) has been compared with the reported maps of the tS elements I S / ( O h t s u b o and Ohtsubo 1978), IS2 (Ghosal et al. 1979), IS3 (Sommer et al. 1979), [$4 (Klaer et al. 1981), IS5 (Schoner and Kahn 1981), 1S8 (Depicker et al.

A number of IS elements have been shown to code for trans-acting functions that are absolutely required for transposition (IS/ (Machida et al. 1982), ISIO (Morisato et al. 1983), ISSO (Isberg et al. 1982; Johnson et at. 1982) and IS903 (Grindley and Joyce 1981)). To see if IS91 also coded for its own transposase, a number of derivative plasmids were constructed in vitro (their structures are shown in Fig. 3), and transposition of their IS elements was analyzed (Table 3). Plasmids pSU234 and pSU240, containing wild-type IS91 elements, were conducted by R388 at frequencies higher than 10 -6 Tc r transconjugants per Tp r transconjugant (strains SU121 and SU190 in Table 3). Plasmid pSU242 was engineered so that sequences up to the XhoI site of its parental plasmid pSU234 were substituted by a kan gene of Tn5 (Fig. 3). pSU242 is conducted by R388 at a frequency indicating that it contains an intact IS91 element (strain SU197). Therefore, the functionality of IS9! is not afected by this deletion and this helped us in mapping the element, as described above. PIasmid pSU241 contains a deleted IS91 element (IS91d; Fig. 3) in which the internal 1.05 kb BstEII fragment of IS91 has been removed, pSU241 is not conducted by R388 at a detectable frequency (strain SU192; Table 3) and, therefore, lacks one or more genes and/or sequences required for IS91 transposition. However, the results obtained with strain SU195 (Table 3) indicate that the defect in IS91d transposition could be compensated if a wild-type IS91 element was also present. In strain SU195, the wildo type element is provided by pSU243 (Fig. 3), a derivative of pSU240 engineered so that it contains a marker (Kin r) which is distinguishable from those of pSU241, while it retains its ability to bring about cointegrate formation (strain SU194; Table 3). When the Tc ~ transconjugant colonies from the cross SU195 x UB1637 were analyzed, it was found that they con° tained cointegrate plasmids of R388 and pSU241 fused by directly repeated copies of IS91A (data not shown). IS91A does not contain any selectable genetic marker, so that transposition could only be detected by looking for cointegrate formation (e. g., by conduction). A K m r marker was inserted at the single BstEII site within IS91A (plasmid pSU248; Fig. 3), so that transposi-

497 Table 3. Transposition frequencies of IS9l and derivatives with and without trans-complementation of the transposition functions Relative frequencies of transconjugant phenotypes b

Donor strain a

Denomination

Plasmids

SU 121 SU190 SU 192 SU197 SU194 SU195 SU203 SU204

R388(Tp r) + pSU234 (Tc~) R388(Tp ~)+ pSU240 (Tc0 R388 (Tp0 + pSU241 (Tcr) R388(Tp0+ pSU242(Km r) R388(Tp~)+ PSU243(Km0 R388(Tp')+pSU241 (Tcf)+ pSU243(Km ~) R388 (Tp') + pSU248 (KmrCm') R388(Tp0 + pSU248(KmrCmr) + pSU240

Tc'Sm*/Tp*Sm* 2 x 10- 6 5 x 10 -6 < 2 x 10- t o n.d. n.d. 5 x 10 -9c n.d. n.d.

Km*Sm'/Tp'Sm' n.d. n.d. n.d. 4 x 10 -6 3 x t0 -6 3 x 10 -6 < 1 x 10 - 9 3 x 10- 7a

Cm*Sm*/Tp~Sm' n.d. n.d. n.d. n.d. n.d. n.d. < 8 x 10 - ' 1 7 X 10 - 9

UB5201 derivative strains which contained the plasmids shown in the column were constructed, and used as donors in mating experiments in which UB1637 was the recipient strain. After conjugation for 3 h on the surface of agar plates, bacteria were resuspended in broth and suitable dilutions plated on drug-containing plates (see Materials and Methods) b Smf was used to counterselect donors, so the number of Tp'Sm ~ colonies on plates reflects the number of transconjugants. The other markers of the donor strains could only be transferred after transposition to R388 or cointegrate formation. Therefore, the frequency of transfer of these markers divided by the total number of transconjugants (the conduction frequency) gives an estimate of the frequency of transposition/cointegrate formation brought about by a given element. It should be noted that the negative control (i.e. that pACYCI84 is not itself conducted by R388) is provided by lack of conduction in strains SU192 and SU203. ""n.d. "': not determined All the transconjugants were Km~ (therefore they did not contain the helper plasmid pSU243). Restriction analysis of the plasmids contained in the transconjugants showed that they were all cointegrates of R388 and pSU241, linked by direct repeats of IS91A a The Km' transconjugants were replica-plated on medium containing Cm; 489/500 were KmrCms while 11/500 were Km r Cm'. As in footnote(cL plasmid analysis of the Km r Cm ~ transconjugants indicated that they contained R388::IS91A kan, the IS element being inserted at different sites of R388 in the different transconjugants. Plasmid analysis of the Km~ Cm' transeonjugants showed that they were all cointegrates of R388 and pSU248, linked by direct repeats of IsglA kan

tion o f this new element (IS91A kan) could be studied directly. ISglA kan is transposition deficient (strain SU203; Table 3) but transposition o f Km ~ could be detected after complementation with pSU240 (strain SU204. Table 3). Most of the Km ~ transconjugants obtained were Cm s (489/500), which suggests that they were simple transpositions of ISglA kan. This was confirmed by restriction endonuclease analysis o f the plasmids contained in the transconjugants. The remaining transconjugant colonies (11/500) were Km ~ C m ' and their restriction analysis was consistent with their being cointegrates o f R388 and pSU248, linked by direct repeats o f IS91d kan (data not shown). One o f these cointegrates (pSU253), was grown for 100 generations both in the Rec * strain UB1636 and in the isogenic recA - strain UB1637. and linkage o f Cm r and Kin' examined by crossing this strain with UB5201. As shown above for the cointegrate plasmid pSU238, plasmid pSU253 was completely stable in strain UB1637 and could only be slowly resolved in strain UB1636. Therefore, the R388:: IsgiA kan transconjugants o f the cross S U 2 0 4 x UB1637 probably arose by direct transposition o f ISglA kan, and not by resolution o f presumptive cointegrates, since they were shown to be stable in the recA-background o f UB1637.

Discussion IS91 is a transposable element o f 1.85 kb, able to insert at different sites in recipient replicons. This element is ubiquitous in Hly plasmids and is probably involved in the dissemination of the hly genes (Zabala et al. 1982 and manuscript in preparation). Furthermore, none o f t h e previously described IS elements is related to IS9l as demonstrated by comparison of their restriction endonuclease

cleavage maps or D N A / D N A hybridization. Therefore, IS91 probably represents a new IS element o f E. coil The inverted repeats o f IS9l, if existing, are very short since no stem-loop structures could be detected when plasmids containing IS9l were examined under the electron microscope (data not shown). Insertion of IS91 frequently generates an YhoI site at the point o f insertion. Precedents for the generation o f new restriction enzyme sites by insertion o f transposons have also been found for other elements: e.g., Trt3 sometimes generates an SrnaI site (Grinsted et al. 1978). The fact that the XhoI site is produced rather frequently in independent insertions (3 out of 10 examined) suggests that at least one o f the IS91 terminal sequences contains a substantial part (probably 5 bp) o f the XhoI target site (i.e., 5 ' - C T C G A G ) or, alternatively, that there is a considerable specificity o f insertion. A deletion within ISgl that removes internal sequences gives a transposition-defective element (IS9IA), so that cointegrates were not formed (Table 3). However, transposition of the defective element is restored if a wild-type element is provided in trans. This suggests that ISglA lacks at least one gene product necessary for transposition. However, upon complementation, the frequency o f cointegrate formation by IS91A is only about 2% as compared to the wild-type element. Inefficiency o f the complementation reaction has also been found for other elements (Isberg et al. 1982; Johnson et al. 1982; Morisato et al. 1983), and suggests that the presumed transposase acts preferentially in cis. IS91 was then genetically labeled by insertion o f a 1.45 kb D N A segment encoding K.m r, so that transposition of the element could be studied directly. Our results o f the tran.sposition of this element (Is91A kan) show that the

498 transposition products are mostly simple insertions o f IS91,J kan into the recipient molecule: cointegrates represented only 2% o f the transposition p r o d u c t s (Table 3). As IS91/I kan cointegrates are stable in a recA b a c k g r o u n d , transposition o f IS91A kan must be at least 50 times m o r e frequent than cointegrate formation. Therefore, the transposition behavior o f IS91zl kan is better explained by m o d els that allow direct transposition (Galas a n d C h a n d l e r 1981; Harshey a n d Bukhari 1981; Berg 1983). If this is also true for the wild-type IS9I then, the true transposition frequency for this element is 10 -'~ (since cointegrate f o r m a tion occurred at a frequency o f 10 -6) which is a rather high transposition frequency when c o m p a r e d to other IS elements (Kleckner 1981). Acknowledgements. This work was supported by a grant (n° 0366! 81 ) of the Comisi6n Asesora de Investigacifn Cientifica y T~nica. We thank Dr. J, Grinsted for critical reading and correction of the manuscript, and Mrs. Marta Garcia for her excellent assistance.

References

Arthur A. Sherratt D (1979) Dissection of the transposition process: A transposon-encoded site-specific recombination system. Mol Gen Genet 175: 267-274 Auerswald EA, Ludwig G, Schaller H (1981) Structural analysis of Tn5. Cold Spring Harbor Syrup Quant Biol 45: 107-I 13 Berg DE (1983) Structural requirement for IS50-mediated gene transposition. Proc Natl Acad Sci USA 80:792-796 Bolivar F, Rodriguez RL, Greene PJ, Betlach MC, Heyneker HL, Boyer HW (1977) Construction and characterization of new cloning vehicles II. A multipurpose cloning system. Gene 2:95-113 Bradley DE, Taylor DE, Cohen DR (1980) Specification of surface mating systems among conjugative drug resistance plasmids in b,:~cherichia coli K 12. J Bacteriol 143: 1466-1470 Brown MGM, Weston A, Saunders JR, Humphreys GO (1979) Transformation of E. coli C600 by plasmid DNA at different phases of growth. FEMS Microbioi Lett 5:219-222 Chang ACY, Cohen SN (1978) Construction and characterization of amplifiable multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid. J Bacteriol 134:1141-1156 Clark AJ, Warren GJ (1979) Conjugal transmission of plasmids Annu Rev Genet 13:99-125 De la Cruz F, Grinsted J (1982) Genetic and molecular characterization of Tn21, a multiple resistance transposon from RI00.1. J Bacteriol 151:222-228 De la Cruz F, M~tller D, Ortiz JM, Goebel W (1980) Hemolysis determinant common to Escherichia coil Hly plasmids of different incompatibility groups. J Bacterio| 143:825-833 De la Cruz F, Zabala JC, Ortiz JM (1979) Incompatibility among u-hemolytic plusmids studies after inactivation of the ~-hemolysin gene by transposition of TnS02. Plasmid 2:507-519 Depicker A, de Block M, lnze D, Van Montagu M, Schell J (1980) Insertion sequence-like element IS8 in RP4 plasmid and its involvement in cointegration. Gene 10:329-338 Dugaiczyk A, Boyer HW, Goodman HM (1975) Ligation Of EcoRI endonuclease-generated DNA fragments into linear and circular structures. J Mol Biol 96:171-184 Galas DJ, Chandler M (1981) On the molecular mechanism of transposition. Proc Natl Acad Sci USA 78:4858--4862 Ghosal D, Sommer H, Saedler H (1979) Nucleotide sequence of the transposable DNA-element IS2. Nucl Acids" Res 6:11111122 Grindley NDF, Joyce CM (1981) Analysis of the structure and function of the kanamycin-resistance transposon Tn903. Cold Spring Harbor Symp Quant Biol 45:125-133

Grinsted J, Bennett PM, Higginson S, Richmond MH (1978) Regional preference of insertion of Tn50l and Tn802 into RP1 and its derivatives. Mol Gen Genet 166: 313-320 Grinsted J, de la Cruz F, Altenbuchner J, Schmitt R (1982) Complementation of transposition of mpA mutants of Tn3, Tn2I, TnSOl and Tn1721. Plasmid 8:276-286 Guyer MS (1978) The ~,~ sequence of F is an insertion sequence. J Mol Biol 126: 347-365 HaUing SM, Simons RW, Way JC, Walsh RB, Kleckner N (1982) DNA sequence organization of ISIO right of TnlO and comparison with IS/0-teft. Proc Natl Acad Sci USA 79:2608-2612 Harshey RM, Bukhari AI (1981) A mechanism of DNA transposition. Proc Natt Acad Sci USA 78:1090--1094 Heffron F, McCarthy BJ, Ohtsubo H, Ohtsubo E (1979) DNA sequence analysis of the transposon Tn3: three genes and three sites involved in transposition of Tn3. Cell 18:1153-1163 Hille J, van Kan J, Klasen I, Schilperoort R (1983) Site-directed mutagenesis in Escherichia coli of a stable R772 :: Ti cointegrate plasmid from Agrobacterium tumefaciens. J Bacteriol 154: 693-701 Holmes DS, Quigtey M (1981) A rapid boiling method for the preparation of bacterial plasmids. Anal Biochem 114:193197 Humphreys GO, WiUshaw GA, Anderson ES (1975) A simple method for the preparation of large quantities of pure plasmid DNA. Biochem Biophys Acta 383:457--467 Iida S, Mayer J, Linder P, Goto N, Nakaya R, Reif HJ, Arber W (1982) The kanamycin resistance transposon Tn2680 derived from the R plasmid Rtsl and carried by phage PIKm has flanking 0.8-kb-tong direct repeats. Plasmid 8:187-198 lsberg RR, Lazaar AL, Syvanen M (1982) Regulation of Tn5 by the right-repeat proteins: control at the level of the transposition reaction? Cell 30:883-892 Johnson RC, Yin JCP, Reznikoff WS (1982) Control of Tn5 transposition in Escherichia coli is mediated by protein from the right repeat. Cell 30:873-882 Jorgensen RA, Rothstein SJ, Reznikoff WS (1979) A restriction enzyme cleavage map of Tn5 and location of a region encoding neomycin resistance. Mol Gen Genet 177:65-72 Klaer R, Kuhn S. Tillmann E, Fritz HJ, Starlinger P (1981) The sequence of IS4. Mol Gen Genet 181 : 169-175 K leckner N (1981) Transposable elements in prokaryotes. Ann Rev Genet 15: 341--404 Labigne-Roussel A, Courvalin P (1983) 1SI5, a new insertion sequence widely spread in R plasmids of gram-negative bacteria. Mol Gen Genet 189:102-112 Machida Y, Machida C, Ohtsubo H, Ohtsubo E (1982) Factors determining frequency of plasmid cointegration mediated by insertion sequence IS/. Proc Natl Acad Sci USA 79:277281 Maniatis T, Fritsch EF, Sambrook J (1982) Molecular cloning. A laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York Miller J (1972) Experiments in bacterial genetics, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York Morisato D, Way JC, Kim H J, Kleckner N (1983) Tn10 transposase acts preferentially on nearby transposon ends in vivo. Cell 32:799-807 Ohtsubo H, Ohtsubo E (1978) Nucleotide sequence of an insertion element IS/. Proc Natl Acad Sci USA 75:615-619 Ohtsubo H, Zenilman M, Ohtsubo E (1980) Insertion of element IS102 resides in plasmid pSC101, J Bacteriol 144:131-140 Oka A, Sugisaki H, Takanami M (1981) Nucleotide Sequence of the kanamycin resistance transposon Tn903. J Mol Biol 147: 217-226 Schoner B, Kahn M (1981) The nucteotide sequence of IS5 from Escherichia coil Gene 14:165-174 Sommer H, Cullum J, Saedler H (1979) Integration of IS3 into IS2 generates a short sequence duplication. Mot Gen Genet 177: 85-89 Sutcliffe JG (1978) Complete nucleotide sequence of the

499 Escherichia coli plasmid pBR322. Cold Spring Harbor Syrup Quant Biol 43: 77-90 Ward JM, Grinsted J (1982) Physical and genetic analysis of the Inc-W group plasmids R388, Sa, and R7K. Plasmid 7:239-250 Willetts NS, Crowther C, Holloway BW (1981) The insertion sequence IS21 of R68.45 and the molecular basis for mobilization of the bacterial chromosome. Plasmid 6:30-52 Yang RCA, Lis J, Wu R (1979) Elution of DNA from agarose gels after etectrophoresis. Methods in Enzymol 68:176-182

Zabala JC. de Ia Cruz F, Ortiz JM (1982) Several copies of the same insertion sequence are present in alpha-hemolytic plasraids belonging to four different incompatibility groups. J Bacteriol 15 l : 472--476 C o m m u n i c a t e d by A. B r c k Received September 19 / October 17, 1983