Proc. Nati. Acad. Sci. USAVol. 87, pp. 6772-6776, September 1990Microbiology

Transformation of Halobacterium halobium: Development of vectorsand investigation of gas vesicle synthesis

(Haloferax voklani/restriction endonuclease/plasmnd stability/gas vesicle protein gene)

ULRIKE BLASEIO AND FELICITAS PFEIFER*Max-Planck-Institut fuir Biochemie, D-8033 Martinsried, Federal Republic of Germany

Communicated by Herbert W. Boyer, June 18, 1990 (received for review March 20, 1990)

ABSTRACT We developed vector plasmids for the trans-formation of Halobacterium halobium, using the replicon re-gion from the halobacterial phage OH or from' the plasmidpHHl together with a DNA fragment conferring resistance tomevinolin. H. halobium P03, a strain lacking pHHI as well asthe restriction endonuclease activity found in wild-type H.halobium, was used as the recipient strain. AlR H. halobiumfragments tested for autonomous replication as well as theHaloferax volcanii vector pWL102 enabled stable plasmidmaintenance in this strain. A frequent loss of all vectors(including pWL102) was observed inHf. vokanfi, where >90%of the mevinolin-resistant colonies obtained after transforma-tion had lost the vector, presumably because of restrictionendonuclease activity and concomitant recombination of themevinolin resistance marker with the chromosome. Theexpression of gas vesicle-encoding genes (vac) was analyzed byusing a 4.5-kilobase-pair (kbp) fragment containing the plas-mid-encoded p-vac gene from H. haobium or an 11-kbpfragment containing the mc-vac chromosomal gene from Halo-ferax mediterranei for transformation experiments with H.halobium and. Hf. vokanii. These experiments indicated thatthe mc-vac fragment contains all information necessary tosynthesize gas vesicles, whereas in the case of the smaller p-vacfragment, complementation by other genes was required for aVac+ phenotype.

A characteristic feature of the archaebacterium (archaeobac-terium) Halobacterium halobium is the synthesis of protein-aceous gas vesicles. H. halobium contains two vac genes[p-vac present on the 150-kilobase pair (kbp) plasmid pHH1and c-vac on the chromosome] that encode the major struc-tural protein of the respective gas vesicle (1-3). Whereas thep-vac gene is expressed in wild-type H. halobium, the c-vacgene is only active in mutants completely lacking the p-vacregion (3). A homologous chromosomal gene (mc-vac) en-coding a major gas vesicle protein was isolated from the moredistantly related Haloferax mediterranei (4).-The analysis of halobacterial gene regulation by in vitro

mutations and reintroduction into halobacteria requires atransformation system. Transfection of H. halobium hasbeen demonstrated with phage OH DNA (5), and transfor-mation of the 1.7-kbp multicopy plasmid pHSB1 of Halo-bacterium sp. SB3 has been shown by colony hybridization(6). For Haloferax volcanii, a species that neither producespurple membrane nor gas vesicles, a transformation vector(pWL102) is available (7) that contains the replicon region ofthe 6.4-kbp Hf. volcanii plasmid pHV2 combined with achromosomal DNA fragment conferring resistance to mevi-nolin, an inhibitor of the eucaryotic and archaebacterial3-hydroxy-3-methylglutaryl CoA-reductase. Transformation

of H. halobium with pWL102 has not been demonstrated sofar.To develop transformation vectors for H. halobium, we

tested putative H. halobium replicons for autonomous rep-lication. One of these replicons is derived from the 59-kbpDNA of phage OH and is present on the 12-kbp circulardefective prophage p4HL (8), and the second replicon isderived from the endogenous 150-kbp plasmid pHH1 of H.halobium. A common region of4.3-kbp DNA was determinedfor all pHH1-type plasmids during the analysis of variousdeletion derivatives (9). One of the two smallest derivatives,pHH9 (5.7 kbp), is found as a major plasmid, whereas theother one, pHH8 (6.3 kbp), is always present at a 1:1 ratiotogether with the parental plasmid pHH4 (10).

In this paper, we describe the transformation of various H.halobium strains as well as Hf. volcanii with plasmid con-structs containing the Hf. volcanii mevinolin-resistance genecombined with these putative H. halobium replicons. As thefirst application, the smallest H. halobium vector, pUBP2,and the Hf. volcanii vector pWL102 were used to study gasvesicle synthesis.

MATERIALS AND METHODSMaterials. Restriction endonucleases, T4 DNA polymer-

ase, and T4 DNA ligase were obtained either from Boeh-ringer Mannheim or from Pharmacia. [a-32P]dATP (3,000Ci/mmol; 1 Ci = 37 GBq) was from Amersham. Nylonmembranes used for Southern transfer were purchased fromPall. Chemicals for the isolation of DNA fragments fromagarose gels by the "Gene clean" method were obtainedfrom Bio 101. The Hf. volcani vector pWL102 (7) andpIBI-mev containing the mevinolin-resistance gene in theEscherichia coli plasmid pIBI31 were provided by W. L.Lam and W. F. Doolittle (Halifax, Canada). Plasmid pIBI31confers ampicillin resistance to E. coli. Mevinolin was pro-vided by D. Oesterhelt.

Halobacterial Strains and Media. H. halobium strains usedare listed in Table 1. Hf. volcanji WFD11 (obtained fromW. F. Doolittle) lacks the endogenous plasmid pHV2 (11).Rich media, minimal media, and solutions used for transfor-mation were prepared as described (5, 12). Solid minimalmedium used to grow Hf. volcanji transformants contained15% sucrose (added as powder after sterilization) and 15 ,uMmevinolin. H. halobium transformants were selected on solidmedium (5) supplemented with 25 juM mevinolin. Liquidcultures were grown at 370C with shaking at 200 rpm, andplates were incubated at 420C.

Preparation of DNA. Total DNA of halobacteria was iso-lated from 0.5 ml of a late logarithmic culture as described(13) with the exception that Hf. volcanii cells were directlyresuspended and lysed in the water/deoxycholate solution.

*To whom reprint requests should be addressed.

6772

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Apr

il 23

, 202

1

Proc. Natl. Acad. Sci. USA 87 (1990) 6773

Table 1. Strains and species

EndogenousStrains and species plasmid Genotype PhenotypeH. halobium

Wild type pHH1 p-vac', res+ Vac'MCH1 pHH1 ISH27::p-vac, res+ Vac-PHH4 pHH4 Ap-vac, res- Vac+P03 Ap-vac, res- Vac+

Hf. volcaniiWFD1I no pHV2 vac-, res+ Vac-

Plasmids were purified by CsCI/ethidium bromide densitygradient centrifugation (9).

Southern Hybridizations. Stringent hybridizations weredone at 42°C in a solution containing 45% formamide, 0.75 MNaCl, 0.075 M sodium citrate (pH 7), 20 mM sodium phos-phate (pH 6.5), 100 ,g of salmon sperm DNA per ml, and0.1% sodium dodecyl sulfate. Restriction fragments werelabeled as described by Feinberg and Vogelstein (14). Afterhybridization, filters were washed as described (2). Forlow-stringency hybridization of total Hf. volcanii DNA withthe p-vac, c-vac, and mc-vac genes, the formamide concen-tration was lowered to 30%, and the filters were washed twicefor 15 min at room temperature. Filters were exposed byusing Trimax-X100 films and intensifying screens.Copy Number Determination of Halobacterial Plasmids.

Cells contained in 1.5 ml of a culture in late logarithmicgrowth were lysed in the same volume of 10 mM Tris/1 mMEDTA, pH 7.5, and treated with 10 jig ofRNase per ml. Afterelectrophoresis through a 1% agarose gel in 1 mM EDTA/20mM Tris acetate, pH 7.5, the intensity of the ethidiumbromide-stained plasmid and chromosomal DNA bands wasdetermined with a microdensitometer (Joyce-Loebl).

Preparation of Cell-Free Extracts for Analysis of RestrictionEndonuclease Activity. Two hundred milliliters of a culture inlate logarithmic growth was used for the preparation of thecell-free extract (15). Plasmid pBR322 containing the 10.8-kbp Cla I fragment of phage 4H DNA was used as substratein the analysis of restriction endonuclease activity.

Transformation of Halobacteria. Transformations weredone as described by Cline et al. (5, 12) with the followingmodifications. After transformation cells were kept at 42°Cfor 12-24 hr; then 1/10th of the transformation assay wasspread on solid medium and incubated up to 2 weeks. At themevinolin concentration used, the spontaneous resistancefrequency of a stationary H. halobium culture was 1 x 10-6.

ISH1.8

Colonies grew more slowly on mevinolin plates than withoutselective pressure. In comparison, Hf. volcanii cells wereless fragile during the transformation procedure, whereas H.halobium cells had a strong tendency to lyse.

RESULTS AND DISCUSSIONConstruction of H. halobium Vector Plasmids. Copy-

number determination of the plasmids chosen for the vectorconstructions (except pHH4) revealed 15-20 copies perchromosome for p4HL and 8-10 copies in the case of pHH8and pHH9 (data not shown). The latter two plasmids occur inthe same copy number as determined for pHH1 in wild-typeH. halobium (16).

Restriction fragments from these plasmids were insertedinto the ampicillin-resistant E. coli plasmid pIBI containing a3.5-kbp Hf. volcanii DNA fragment conferring resistance tomevinolin. Plasmid pIBI-mev does not contain halobacterialreplication functions, thus providing a test system for theautonomous replication mediated by the inserted DNA frag-ments. We tested as putative halobacterial replicons (i) a10.8-kb Cla I fragment of phage OH containing most of thesequences maintained in plasmid p4HL except 800 bp of thefusion region and (it) three fragments of pHH1-type plas-mids-i.e., an 8.8-kbp EcoRI fragment ofpHH4 (as parentalwild-type form) and the entire EcoRI-linearized plasmidspHH8 and pHH9 (Fig. 1). The latter three fragments overlapby 4.3 kbp (besides ISH27 insertion sequences). The resultingconstructs are summarized in Table 2, and a restriction mapof the smallest H. halobium vector, pUBP2, is presented inFig. 2.Transformation of H. halobium. H. halobium P03 lacking

pHH1-type plasmids entirely (F. P., unpublished data) wastransformed with the vector containing the Cla I fragment ofphage OH (pUBP1), the three constructs containing thepHH1-derived fragments, and the Hf. volcanii vectorpWL102 (7). With all vectors we obtained 104-106 transfor-mants per jig of DNA (depending on the experiment). Anal-ysis of the plasmid DNA indicated that each transformantcontained the plasmid construct used for transformation(data not shown). To test for vector plasmid stability, thepUBP2 and pUBP8 transformants were grown for 3 weekswith or without selective pressure, and the plasmid contentof 50 single colonies was analyzed. Without selection pres-sure, about 30% of the colonies had lost the constructtogether with the mevinolin resistance, whereas with mevi-nolin selection the vectors were stably maintained. Mevinolin

ISH1 .8

1_L- 0.5 xH

ISH27

TI 2.0 1 4.1 10.51I ;c1 2.0 A.0 10$J 1§ K 8.8 kb pHH4ISH27

I.1 10.7 1 *1K pHH8&3 1

KHHEISH27

p

1.7 ,_ lK pHH9

FIG. 1. Restriction map of the four halobacterial fragments used for the vector construction. Numbers indicate sizes in kbp. B, BamHI; C,Cla I; E, EcoRI; and K, Kpn I. The orientation of these fragments in the vector constructs is identical to the orientation presented for pUBP2in Fig. 2. ISH1.8 and ISH27 are insertion elements.

Microbiology: Blaseio and Pfeifer

Dow

nloa

ded

by g

uest

on

Apr

il 23

, 202

1

6774 Microbiology: Blaseio and Pfeifer

Table 2. Vector constructs used for transformation

Vector

Name Size, kbp Source of replicon

pUBP1 17.4 OHpUBP2 12.3 pHH9pUBP8 12.9 pHH8pUBP4 15.4 pHH4pWL102* 11.2 pHV2

All vectors contain the mevinolin resistance gene of H. volcaniiand the E. coli plasmid pIBI conferring resistance to ampicillin in E.coli.*Constructed by W. Lam and W. F. Doolittle (7).

resistance was always connected with the presence of thevector plasmid; thus, recombination between the selectivemarker and the chromosomal wild-type copy of this genedoes not occur in H. halobium. These experiments prove thatthe 10.8-kbp Cla I fragment of phage OH and the 4.3-kbpregion common to all pHH1-type plasmid fragments containthe functions necessary for plasmid replication and mainte-nance. The minimal size of both replicons will be determinedby introducing further deletions.To test for plasmid incompatibility, the pHH1-derived

vectors pUBP2 and pUBP8 were used to transform the H.halobium strain containing pHH4. In both cases transfor-mants were obtained, and all ofthem initially harbored pHH4together with the vector. After continuous propagation ofindividual clones for more than 3 weeks without selectivepressure, pUBP2 or pUBP8 were lost in >90% of thetransformants (46 of 50), while pHH4 was maintained. Incontrast, with mevinolin selection both vectors were stablymaintained, and pHH4 was deleted (data not shown). Theseexperiments indicate that both- pUBP2 and pUBP8 expressincompatibility functions, although in the original strainpHH8 could not be enriched further than to a 1:1 ratiotogether with pHH4 (10). We assume that the incompatibilityfunctions of the pHH4 plasmid in the H. halobium straincontaining pHH4 and pHH8 are defective.H. halobium P03 and the H. halobium strain containing

pHH4 exhibited a 100-fold higher transformation rate (up to106 per ttg ofDNA) than H. halobium wild-type (103-104 per,ug of DNA), suggesting that both strains lack the restrictionendonuclease activity found in H. halobium wild type (15).Indeed, the analysis of cell-free extracts demonstrated in

H S P B H K

both strains lacked the H. halobium restriction endonucleaseactivity (Fig. 3).

Transformation of Hf. volcanji. All pUBP vectors and theHf. volcanii vector pWL102-each DNA isolated from E. coliDH5a-were also transformed into Hf. volcanii WDF11, andin each case we obtained resistant colonies with the sameefficiency. Hybridization experiments indicated, however,that .90% of the transformants had lost the E. coli and H.halobium part of each vector, while the cells were stillresistant to mevinolin (data not shown). In contrast, by usingpUBP2 or pWL102 isolated from Hf. volcanii for the trans-formation experiments, the frequency was 100-fold higher,and the vectors were stably maintained in pUBP2 (50 of 50)and pWL102 (48 of 50) transformants (data not shown).These experiments indicate that the H. halobium vector

plasmids replicate in Hf. volcanii; however, they also showa frequent loss of the constructs possibly because of arestriction/modification system. We assume that 90% of themevinolin-resistant transformants without the vector con-struct resulted from a fragmentation of the vector through arestriction endonuclease with a concomitant marker rescuevia recombination between the mevinolin-resistance geneand the wild-type copy of this gene in the chromosome. Highfrequencies of recombination have been described for linearDNA fragments transformed into Hf. volcanji (12). To ensurethe maintenance of the plasmids after transformation, theisolation of a restriction endonuclease and recombination-deficient Hf. volcanii recipient strain is highly desirable. Lossof the Hf. volcanii-propagated vector plasmid was observedin <5% of the transformants, indicating that recombinationbetween covalently closed circular DNA and the chromo-some is a rare event (data not shown).

Analysis of vac Gene-Containing Fragments. Until recently,analysis ofgenes for gas vesicle synthesis in H. halobium wasrestricted to spontaneously occurring mutants that arecaused by the integration of insertion elements (refs. 1-3 and17 and unpublished results). Now functional tests of definedgenomic fragments cloned in transformation vectors can beperformed. We analyzed DNA fragments containing part ofthe p-vac region of H. halobium (2) or the mc-vac region ofHf. mediterranei (4) for their ability to promote gas vesicle

1 2 3 4

E

S

E

K S

FIG. 2. Restriction map of pUBP2. Restriction sites are: B,BamHI; E, EcoRI; H, HindIII; K, Kpn I; P, Pst I; S, Sph I; and Sm,Sma I. Unique restriction sites (Sm, B, and P) useful for cloningadditional fragments are underlined. Also HindIII fragments can becloned into pUBP2. amp-r, Ampicillin-resistance gene; mev-r, mevi-nolin-resistance gene; ori, origin of replication.

FIG. 3. Test for restriction endonuclease activity in H. halobiumstrains. Plasmid pBR322 containing a 10.8-kbp DNA insert wastreated with cell-free extracts of H. halobium P03 (lane 1), H.halobium pHH4 (lane 2), and H. halobium wild type (lane 3). Lane4 shows the untreated plasmid DNA. The test plasmid and cell-freeextract were incubated for 30 min at 370C and extracted with phenol,and the DNA was electrophoretically separated through a 1%agarose gel.

Proc. Natl. Acad. Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Apr

il 23

, 202

1

Proc. Natl. Acad. Sci. USA 87 (1990) 6775

synthesis. The pWL102-pvac construct contained a 4.5-kbpKpn I p-vac fragment (Fig. 4), whereas pUBP2-mcvac har-bored an 11-kbp Pst I fragment from Hf. mediterranei con-taining the mc-vac gene and 7.2-kbp-upstream and 3.5-kbp-downstream DNA sequences (4). The transformation vectorswere chosen according to the presence of the appropriatecloning sites.

Recipient strains were either Hf. volcanii or two H. halo-bium mutants with different vac genotypes: one was H.halobium P03 (lacking the entire p-vac region but still con-taining a chromosomal c-vac region expressed at a low levelin the late stationary phase ofgrowth), and the other one wasVac mutant MCH1 that incurred an ISH27 insertion elementin the p-vac promoter region with an otherwise unchangedpHH1 plasmid and no c-vac expression (M. Home and F.P.,unpublished data). As shown by low-stringency Southernhybridizations, Hf. volcanii does not contain gas vesicleprotein genes (data not presented), thus providing a cleangenetic background for the analysis of vac gene expression.

Transformation of strains completely lacking the p-vacregion (Hf. volcandi and H. halobium P03) with pWL102-pvacdid not reveal transformants synthesizing gas vesicles, sug-gesting that the 4.5-kbp Kpn I fragment of pHH1 is notsufficient for gas vesicle production, although the entirep-vac gene (250 bp) and 4 kbp of its surroundings are presenton this fragment. p-vac-specific mRNA was found in H.volcanii as well as in H. halobium P03 transformants but notin the appropriate recipient strains, indicating that the p-vacgene on the vector molecule is expressed but not sufficient forthe gas vesicle formation (data not shown).However, transformation of the H. halobium MCH1 with

pWL102-pvac led to the restoration of gas vesicle synthesis.As monitored by the color of the colonies on agar plates andby inspection of the cells under the phase-contrast micro-scope, about 95% of the transformants produced gas vesicles,20% of them at a level similar to H. halobium wild type(Vac+) and 75% at a lower level (Vac+). About 5% of thecolonies were Vac- (data not shown). Investigation of theplasmid DNA indicated that each colony carried pHH1 aswell as pWL102-pvac. Further purification of Vac+ colonieswithout mevinolin selection again revealed Vac+, Vac+, andVac- colonies. Plasmid investigation indicated that Vac+ andVac- colonies had lost the vector construct. Vac- coloniesstill contained the ISH27::p-vac gene in pHH1, whereas theVac+ colonies had lost the ISH27 element in the p-vacpromoter, presumably by recombination between pHH1 andthe homologous region on the vector (data not shown). Vac+colonies, however, carried the intact p-vac gene on the vector

E

ISH27

H HI I

K E KI I

K H H E K

kbp

I I I I I I I40 1 2 3 4 5 6 7 8

FIG. 4. Restriction map of the p-vac region in the pHH1 plasmid(Upper) and of the Kpn I fragment ofpHH1 present in pWL102-pvac(Lower). The integration site of the 1400-bp element ISH27 in thep-vac promoter of the Vac mutant MCH1 is indicated by a triangle.The 5.2-kbp EcoRI/HindIII fragment (shaded) and the p-vac gene(black) are marked, and the direction of transcription (1, 2, 17) isindicated by arrows.

in trans to the mutant gene on pHH1. Thus, the Vac+phenotype is due to complementation in trans, whereas a cisconfiguration leads to the Vac+ phenotype.One of the initial Vac- transformants contained a size

increase of 1.4 kbp in the 5.2-kbp EcoRI/HindIII fragmentlocated upstream of the p-vac gene (see Fig. 4), presumablyderiving from the integration of an additional insertion ele-ment (data not shown). Investigation of the p-vac region inthe recipient strain revealed that 5% of the cells had incurredan insertion (ISH) element upstream of the p-vac gene(outside of the Kpn I fragment used for transformation) asfound in the Vac- transformant (data not shown). Thus, the5% Vac- transformants can be traced back to the transfor-mation of this double mutant. Since transformation of thedouble mutant does not reveal gas vesicle-producing cells, weassume that the target region of the additional insertionelement is necessary for gas vesicle synthesis. The analysisof a variety of ISH-element-induced Vac mutants also indi-cates the additional genes located within the 4.5-kbp Kpn Ifragment and further upstream of the p-vac gene are neces-sary for gas vesicle synthesis (ref. 17; M. Home and F.P.,unpublished data). Transformation of larger p-vac fragmentsand complementation studies should help to define the size ofthe plasmid region necessary for gas vesicle formation.

In contrast to the p-vac fragment, transformation of Hf.volcandi with pUBP2-mcvac resulted in about 10% Vac+transformants. The Vac+ phenotype of these transfor-mants-similar to that of Hf. mediterranei-appeared in thestationary phase of growth, and inspection of these cellsunder the phase-contrast microscope indicated light refractilebodies typical for gas vesicles (Fig. SB). As tested by hy-bridization, each of these Vac+ colonies contained pUBP2-mcvac, whereas none of the Vac- transformants containedthe construct (data not shown). The frequent loss of pUBP2-mcvac with concomitant rescue of the mevinolin-resistancemarker in Hf. volcanii resembles the instability observed withpUBP2 (see above). Nonetheless, this system can be used todetermine the minimal size of the mc-vac region needed forgas vesicle synthesis.

Untransformed H. halobium P03 colonies indicated a low-level Vac phenotype in late stationary phase of growth. Aftertransformation by pUBP2-mcvac, gas vesicle synthesis oc-curred at the same low level as in the untransformed recipientstrain. Expression of the mc-vac gene was neither detectablein the recipient strain nor in the pUBP2-mcvac transformants(data not shown). It is possible that the expression signals ofthe Hf. mediterranei mc-vac gene are efficiently recognizedin the closely related Hf. volcanii but not in the more distantlyrelated H. halobium. The lack of mc-vac expression in H.halobium P03 may also be explained by a repression of themc-vac gene mediated by the c-vac region present in thisstrain. Further investigations are necessary to determine apossible influence of the c-vac region and to characterize theindividual functions of the genes involved in gas vesiclesynthesis.

A

I t . @*wti$ .X 4 .

B

,s..B~~~~~~~' 17 4p,

t s'

FIG. 5. Phase-contrast microscopic pictures of wild-type Hf.volcanii (A) and Vac+ transformants (B). (x 1050.)

Microbiology: Blaseio and Pfeifer

3

Dow

nloa

ded

by g

uest

on

Apr

il 23

, 202

1

6776 Microbiology: Blaseio and Pfeifer

We wish to thank W. Zillig for support; M. Home for mutantMCH1 and valuable discussions; W. L. Lam, S. Cline, and W. F.Doolittle for Hf. volcanii vectors and advice in the transformationprocedure; and C. Englert and W.-D. Reiter for critical reading ofthemanuscript. Part of this work was supported by the DeutscheForschungsgemeinschaft (SFB145/B6).

1. DasSarma, S., Damerval, T., Jones, J. G. & Tandeau deMarsac, N. (1987) Mol. Microbiol. 1, 365-370.

2. Home, M., Englert, C. & Pfeifer, F. (1988) Mol. Gen. Genet.213, 459-464.

3. Home, M. & Pfeifer, F. (1989) Mol. Gen. Genet. 218, 437-444.4. Englert, C., Home, M. & Pfeifer, F. (1990) Mol. Gen. Genet.,

in press.5. Cline, S. W. & Doolittle, W. F. (1987) J. Bacteriol. 169,

1341-1344.6. Hackett, N. & DasSarma, S. (1989) Can. J. Microbiol. 35,

86-91.

Proc. Nati. Acad. Sci. USA 87 (1990)

7. Lam, W. L. & Doolittle, W. F. (1989) Proc. Natl. Acad. Sci.USA 86, 5478-5482.

8. Schnabel, H. (1984) Proc. Natl. Acad. Sci. USA 81,1017-1020.9. Pfeifer, F., Blaseio, U. & Ghahraman, P. (1988) J. Bacteriol.

170, 3718-3724.10. Pfeifer, F. & Blaseio, U. (1989) J. Bacteriol. 171, 5135-5140.11. Charlebois, R., Lam, W., Cline, S. & Doolittle, W. F. (1987)

Proc. Nail. Acad. Sci. USA 84, 8530-8534.12. Cline, S. W., Schalkwyk, L. C. & Doolittle, W. F. (1989) J.

Bacteriol. 171, 4987-4991.13. Pfeifer, F. & Betlach, M. (1985) Mol. Gen. Genet. 198,449-455.14. Feinberg, A. P. & Vogelstein, B. (1983) Anal. Biochem. 132,6-13.15. Schinzel, R. & Burger, K. (1986) FEMS Microbiol. 37, 325-329.16. Weidinger, G., Klotz, G. & Goebel. W. (1978) Plasmid 2,

377-386.17. Jones, H. G., Hackett, N., Halladay, J. T., Scothom, D. J.,

Yang, C., Ng, W. & DasSarma, S. (1989) Nucleic Acids Res. 17,7785-7793.

Dow

nloa

ded

by g

uest

on

Apr

il 23

, 202

1