Vol. 169, No. 6JOURNAL OF BACTERIOLOGY, June 1987, p. 2643-26490021-9193/87/062643-07$02.00/0Copyright C 1987, American Society for Microbiology

Isolation of Haemophilus influenzae Genes That SuppressEscherichia coli polA Mutations

GENE L. WILLIAMS, BRENT SEATON,t AND DAVID McCARTHY*Department ofBotany and Microbiology, University of Oklahoma, Norman, Oklahoma 73019

Received 20 January 1987/Accepted 18 March 1987

HaemophUus influenzae was found to produce a DNA polymerase that was similar to polymerase I ofEscherichia coli. E. coli polA mutants were used as backgrounds for the selection of H. influenzae polAsuppressor genes. Six different H. influenzae fragments were isolated that could suppress E. coli polAmutations. None of the suppressors appeared to encode the H. influenzae equivalent of the E. coli polA gene.Oiie type of clone, represented by pGW41, caused a polymerase I activity to appear in a suppressed poUlImutant. Plasmids from the pGW41 class contained two genes (pol-2 and pol-3) that were both required for polAsuppression. Mutated nonsuppressing derivatives of the pGW41 class were used to create H. influenzae mutantsthat were deficient in polymerase I.

We screened a library of Haemophilus influenzae chromo-somal fragments for the gene that encodes polymerase I(polI) as a preliminary step in the construction of polymer-ase-defective mutants. We used Escherichia coli polA mu-tants as backgrounds for our attemnpts to isolate the putativeH. influenzae analog of the polA gene, defined here as thepol-I gene. Six regions of the H. influenzae chromosome thatsuppressed polA mutations of E. coli were isolated. Therecombinant plasmids were characterized on the basis ofrestriction enzyme analysis, in vitro gene expression, andthe production of polymerase activities. None of the sixclasses of suppressors encoded an analog of E. coli polymer-ase I. One of the classes of suppressor plasmids (representedby pGW41 and pDM19) affected the level of polI accumula-tion in an E. coli polA mutant. Mutated derivatives ofpDM19 and pGW41 with defects in either of two suppressorgenes (called pol-2 and pol-3) failed to suppress polA muta-tions. Mutations in the pol-2 gene blocked the expression ofthe pol-J gene in H. influenzae.

MATERIALS AND METHODSBacterial strains and plasmidg. The E. coli strains used in

this work are listed in Table 1. Wild-type H. influenzaeBC200 was used as a recipient for mutagenized DNA. ThePstI library of H. influenzae chromosomal fragments was inthe high-copy-number shuttle vector pDM2. pDM2 wasconstructed by inserting the chloramphenicol resistancegene from plasmid p2265 into plasmid pRSF0885 (Apr) (19).pDM2 can replicate as a high-copy-number plasmid in bothH. influenzae and E. coli. A version of the H. influenzaechromosomal library that was established in E. coli WA946(polA+ hsdM+ hsdR) was used for the selection of polAsuppressors (19). pCJl is a pNG16 derivative that containsthe polA gene of E. coli (11).

Transformation. E. coli cells were grown in LB mediutn.CaCl2-competent cells were transformed as described previ-ously (18). H. influenzae cells were grown in brain heartinfusion broth supplemented with 10 ,ug of hemin per ml and2 ,ug of NAD per ml. BC200 was transformed by the MIVmethod (8).

* Corresponding author.t Present address: Department of Biology, University of Califor-

nia at San Diego, La Jolla, CA 92093.

Isolation of nonsuppressing pDMl9::TnS and pDM-19::mini-TnlO kan. E. coli K37(pDM19) was mutagenized byinfection with either X b221 rex: :TnS c1857 Oam8 Pam29 [6])or X1105 (mini-TnlO kan [26]). Mutagenized cells wereselected on LB plates containing kanamycin (20 ,ug/ml) andchloramphenicol (30 jig/ml). Kanr colonies were pooled tomake a plasmid preparation (18). RS5064 was transformedwith the pooled plasmid DNA, and the Kanr Camr transform-ants were screened to define their temperature-sensitive (Ts)phenotypes. Of 369 plasmid transformants that were tested,69 had retained their Ts- phenotype, indicating that the polAsuppressor gene was inactivated.

Selection for H. influenzae polI-deficient mutants. H.- influ-enzae TnS mutants were constructed by transforming wild-type cells with TnS-mutagenized chromosomal inserts.BC200 cells were made competent by incubation in MIVmedium. The cells were transformed by a collection ofplasmids that were assumed to contain TnS in the suppressorgenes based on their inability to suppress the E. colipolA480mutation. The transposon was expected to integrate into thechromosome by homologous recombination involvingflanking sequences that were homologous to the recipientchromosome (additive transformation) (25). polI-deficientmutants were tentatively identified based on the assumptionthat 'they would be unable to maintain pDM2. The loss ofpDM2 from H. influenzae colonies was detected by a starch-iodine colorimetric plating assay for P-lactamase activity (4).Brain heart infusion-starch plates were used instead ofLB-starch plates. The polI-deficient designation of individ-ual Kanr strains was confirmed by examining extracts in apolymerase activity gel.

Preparation of cell extracts. Cell extracts were made asdescribed by Spanos et al. (24). The protein concentrationsof the extracts were determined by the Bradford method (3).DNA polymerase activity gels. DNA polymerase activity

gels were prepared as described previously (24). Polyacryl-amide (10%) gels contained 150 ,ug of nicked calf thymusDNA per ml (24), 50 ,ug of fibrinogen (type I-S; SigmaChemical Co., St. Louis, Mo.) per ml, and 0.1% (wt/vol)ultrapure sodium dodecyl sulfate (SDS) (Bio-Rad Laborato-ries, Richmond, Calif.). Cell extracts (100 ,ug) were mixedwith 5 ,ul of loading buffer (4% [wt/vol] SDS, 2% [vol/vol]2-mercaptoethanol, 36% [vol/vol] glycerol) and then wereloaded onto the gels after a 5-min incubation at 37°C. The

2643

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

2644 WILLIAMS ET AL.

TABLE 1. E. cOli strains

Strain Relevant genotype Source or reference

K37 polA+ H. I. MillerMM383 polAl2 21P3478 polAl 7RS5064 polA480 15

samples were fractionated at 275 V. The gels were subjectedto a series of washes after the electrophoresis step to removethe SDS and to promote renaturation of the fractionatedproteins (2). The gels were washed in 10 mM Tris hydro-chloride (pH 7.6-5 mM 2-mercaptoethanol-20% isopropa-nol, followed by an incubation at 4°C for 24 h in 500 ml of 50mM Tris hydrochloride (pH 7.6)-5 mM 2-mercaptoethanol-1mM EDTA. The in situ polymerase reaction was at 25°C for24 h in 100 ml of 70 mM Tris hydrochloride (pH 7.6)-7 mMMgCI--10 mM dithiothreitol-12 ,uM each of dGTP, dCTP,dATP-0.5 to 1 ,uCi of [a-32P]dTTP per ml (3 kCi/mmol) (2).The reaction was terminated by washing the gels in 5%(wt/vol) trichloroacetic acid-1% (wt/vol) sodium PPi for 48 h.The gels were then dried on Whatman 3MM paper andexposed at -70°C to Kodak XAR-2 film plus a CronexLightning-Plus intensifying screen.

In vitro transcription-translation. Plasmid-encoded pro-teins were made in vitro with a procaryotic DNA-directedtranscription-translation kit by the procedure supplied by themanufacturer (Amersham Corp., Arlington Heights, Ill.).[35S]methionine (New England Nuclear Corp., Boston,Mass.) was included at a concentration of 1.1 FM (1,171Ci/mmol). The products of in vitro gene expression weredetected in 15% SDS-polyacrylamide gel by autoradiog-raphy.

Southern blots. DNA fragments from 0.8% agarose gelswere electrophoretically transferred to nitrocellulose filtersas described by Smith et al. (23). The filters were blockedwith Blotto (9). The prehybridization solution was 6x SSC(lx SSC is 0.15 M NaCl plus 0.015 M sodium citrate)0.1%SDS-0.05% Blotto45% deionized formamide (9). The hy-bridization was done at 25°C for 48 h after the addition of 100ng of 32P-labeled pCJ1 (1.05 x 108 cpm/pg). Theheteroduplex bands were visualized by autoradiography.

RESULTS

H. influenzae and E. coli extracts appeared to produceidentical patterns of 32p incorporation in a polymeraseactivity gel as shown in the autoradiograph in Fig. 1. Ahigh-molecular-weight band correlated with purified poll ofE. coli, and a satellite band of incorporation at 68 kilodaltons(kDa) correlated with purified large fragment of poll (5, 14).The presence of the 103- and 68-kDa fragments in the H.influenzae extracts implied that a polA analog exists in H.influenzae that has the same structure as its E. coli counter-part. We refer to the gene that encodes H. influenzae poll asthe pol-I gene, in keeping with the nomenclature establishedfor other H. influenzae DNA metabolism genes.

Isolation of H. irfluenzae chromosomal fragments thatsuppress E. coli pokA mutations. Three E. coli polA mutantswere used to select for H. influenzae polA suppressor genes.MM383 encodes the, temperature-sensitive mutation poUA12that inactivates the polymerase activity of poll at 42°C butdoes not affect the growth of the cells at restrictive temper-atures (21). Our H. influenzae-E. coli plasmid shuttle vector,pDM2, required a functional poll for maintenance in E. coli.

pDM2 was rapidly cured from a culture of MM383 pDM2 at42°C, since selection for the plasmid marker chlorampheni-col resistance (Cm') at high temperature was lethal tovirtually every cell in the culture (data not shown). We neversuccessfully transformed pDM2 into p3478 (polAI) in numer-ous transformations experiments, as expected if the plasmidrequires poll for maintenance. Strain RS5064 encodes thetemperature-sensitive polA allele (polA480) that is lethal at420C (15, 17).We assumed that H. influenzae polA suppressor plasmids

would successfully transform the E. colipolA mutants if theycould use the products of the suppressor genes to promotetheir own maintenance. A single type of H. influenzae polAsuppressor plasmid was repeatedly isolated after transfor-mation of P3478 (polAl) with the library of cloned chromo-somal fragments. Eighty-six chloramphenicol-resistant colo-nies were recovered after transformation of the E. colipolAlmutant with the library. All but one of these plasmidsproduced the same restriction patterns when digested withPstI or EcoRI (data not shown). Both of the missense polAmutants (MM383 and RS5064) were transformed with thepDM2-based library of H. influenzae chromosomal frag-ments followed by incubation under nonpermissive condi-tions. Ten suppressor plasmids were isolated that could becategorized into six groups based on PstI digestion patterns(Fig. 2A). Several of the plasmids (represented by pBS3,pBS4, and pDM19) were isolated independently in more thanone of the mutants. pBS1 and pDM18 were identical topDMll and pDM12 which were previously isolated based ontheir abilities to suppress a DNA ligase mutation in E. coli(20). All the plasmids were interchangeable in their abilitiesto suppress the polAJ2, polA480, and polAl defects inplasmid maintenance. Furthermore, all the plasmids couldmaintain the viability of RS5064 at 42°C (data not shown).Table 2 summarizes the characteristics of representativeplasmids from each of the six groups. pDM19, one of the sixpolA suppressors that was isolated in a polA missensebackground, was identical to the majority of suppressorplasmids that were isolated in the polAl background asshown by comparing the PstI and EcoRI digests ofpDM19 tothe polAl suppressor plasmids (data not shown). ThepGW41-pDM19 plasmids were preferentially isolated fromthe library in the polAl background even though the othertypes of plasmids could transform P3478 efficiently. Noattempt was made to measure the relative efficiencies ofP3478 transformation by the various types ofplasmids. Noneof the members of the six groups of plasmids were isolated

1 2

.-KPoll

Kienowim-

FIG. 1. Polymerase activities in extracts of H. influenzae and E.coli. Crude extracts of E. coli K37 (lane 2) and H. influenzae BC200(lane 1) were examined in a polymerase activity gel.

J. BACTERIOL.'

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

H. INFLUENZAE polA SUPPRESSORS 2645

A B1 2 3 4 5 6 7 8 9 10 11_ _

_ ,w g I_w . X I ,

:: :.s | g 1 2 3 4

;.

5 6 7 8 9 10 11

FIG. 2. Restriction digests ofpolA suppressor plasmids and their corresponding Southern analysis with pCJ1 as a probe. The samples andthe restriction enzymes used to cut them in panels A and B correspond to: lanes 1, E. coli K37, HindIII; lanes 2, K37, PstI; lanes 3, H.influenzae MAP9, PstI; lanes 4, pBS2, PstI; lanes 5, pBS4, PstI; lanes 6, pBS1, PstI; lanes 7, pDM18, PstI; lanes 8, pDM19, PstI; lanes 9,pDM20, PstI; lanes 10, pGW41, PstI; lanes 11, pCJ1, HindIII. pDM20 contained a larger vector fragment than expected because a deletionremoved a PstI junction region between the vector and insert. The exposure of the X-ray film for the pBS2 sample (lane 4) was extended tocompensate for the small amount of plasmid DNA. The hybridization in lanes 4 through 10 was due to homology between the vectorsequences of the immobilized DNAs and the probe. Panels A and B are not aligned by molecular weight.

coincidently as a result of selection for suppressing hostmutations, because all the plasmids suppressed fresh polAcultures under nonpermissive conditions.Lack of nucleotide sequence homology between E. colipoU

gene and H. influernvae suppressor inserts. Figure 2B repre-sents a Southern blot of immobilized H. influenzae DNAshybridized with radioactive E. coli polA plasmid pCJ1. PstIdigestion of H. influenzae chromosomal DNA produced asingle fragment that hybridized at low stringency with pCJ1(Fig. 2B, lane 3). We assume that this fragment encodespol-l, the H. influenzae analog of the E. coli polA gene. Thepol-i gene was not included in our collection of polAsuppressor plasmids. Southern blots between PstI digests ofall the H. influenzae polA suppressor plasmids and thecloned E. coli polA gene in pCJ1 showed no apparenthomology between the chromosomal inserts and the polAgene when the hybridizations were done under conditions oflow stringency (Fig. 2B, lanes 4 to 10). In every samplehybridization was detected between the P-lactamase se-

TABLE 2. H. influenzae polA suppressor plasmids

Plasmid Insert size PolI Relative spontaneous(kb) activity mutation frequencyb

pBS1, pDM11 1.0 - 1pBS2 1.7 - 1pBS3, pBS5, pDM12, 3.4 - 189pDM18

pBS4, pGW86 3.0 - 1pDM19 (pGW41) 6.5 -(+) 1pDM20 NAC - 1

a Plasmids with the prefixes pBS, pGW, and pDM were isolated in RS5064,P3478, and MM383, respectively. pDM11 and pDM12 were isolated in theligase mutant N2668 (20).

b Frequency of Rifr MM383 with plasmid/Rifr MM383 without a plasmid.c The junction of the insert and vector was deleted resulting in the fusion of

the insert to the vector.

quence of each vector fragment and the homologous vectorsequence of pCJ1. We did not rule out the possibility thatpart of the pDM20 insert shares homology with polA,because we could not cleanly separate the entire insert fromthe vector (Fig. 2, lanes 9). However, the observed hybrid-ization between pCJ1 and pDM20 could have been duesolely to their common P-lactamase sequences.

Levels of poll activity in suppressed polA mutants. Eachplasmid that was isolated from a polA missense mutant wastested for its ability to produce polI in the poll mutant. EachpolA suppressor plasmid was transformed into E. coli P3478.Extracts of the plasmid-bearing cells were examined with thepolymerase activity gel. P3478 was nearly devoid of polIactivity (Fig. 3A, lane 3). On prolonged exposure of theX-ray film, a faint band at the position of polI apparentlyrepresented occasional translational misreading of the polAlamber codon (13). With the possible exception of pDM19(Fig. 3A, lane 8), none of the six polA suppressor plasmidsthat were isolated from the polA missense mutants increasedthe level of polI activity in P3478. This result implies thatnone of the suppressor plasmids were able to replace themissing polI with the H. influenzae analog. This result alsotends to eliminate the possibility that any of the clonescaused increased expression of the polA gene by greatlyincreasing translational ambiguity or by encoding a suppres-sor tRNA. On the other hand, almost all the suppressorplasmids that were isolated in P3478 (polAI) were able toproduce higher levels of poll activity than were found inextracts of P3478 that lacked plasmids. The levels of poly-merase activity varied considerably among independent iso-lates that apparently contained the same plasmid (Fig. 3B).Of the 31 suppressors that were tested, 4 produced signifi-cantly more poll activity than the rest. pGW41 was studiedas a representative of this group. pDM19 was representativeof the pGW41 relatives that produced very low levels ofpolymerase I accumulation.

VOL. 169, 1987

,:: .!!:-i .0WI

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

2646 WILLIAMS ET AL.

A1 2 3 4 5 6 7 8 910 1112

L.F.b.POiI 1v-

...N-.1LF. .:X

a1 2 3 4 5 6 7 8 9 1 O

.j_

FIG. 3. Polymerase activities associated with cells that contain polA suppressor plasmids. (A) Polymerase activity gel of P3478 containingplasmids that were isolated from polA missense mutants. Lane 1, MM383 without a plasmid; lanes 2 and 3, P3478 without a plasmid. Lanes4 to 9 represent P3478 with the following plasmids: 4, pBS2; 5, pBS4; 6, pBS1; 7, pDM18; 8, pDM19; 9, pDM20; lane 10, strain K37 withouta plasmid; lane 11, purified poll; lane 12, purified Klenow fragment. (B) Polymerase activities of P3478 containing plasmids that were isolatedin the polAl(Am) background. Lanes: 1, K37; 2, P3478 without a plasmid; 3, pDM19; 4, pGW38; 5, pGW39; 6, pGW41; 7, pGW40; 8, pGW42;9, pGW43; 10, pGW44. The arrows indicate the positions of polI and the large fragment (LF) of poll.

pGW41 and pDM19 contained the same PstI fragmentcloned from the same strain; however, there was a grossdiscrepancy in the amounts of polI that were produced byeach plasmid. pGW41 produced high levels of poll, whilepDM19 produced barely detectable increases in polI activityin P3478 (Fig. 3, compare panel A, lane 8, with panel B, lane6). EcoRI digests of pGW41 and pDM19 showed that theinserts were in identical orientations relative to the EcoRIsites in the vector (data not shown). Therefore, the differ-ence in polA expression between pGW41 and pDM19 wasnot due to a possible difference in the orientations of thecloned fragments. Gene expression of each plasmid wasexamined in an E. coli in vitro transcription-translationsystem. 35S-labeled plasmid proteins are shown Fig. 4. BothpDM19 and pGW41 produced two apparently identicalpolypeptides of 29 and 27 kDa. The plasmids differed in theproduction of a third polypeptide. pDM19 produced a24-kDa protein, while pGW41 produced a 21-kDa proteinthat might have represented a truncated version of the24-kDa product. A radioactive 104-kDa protein could not bedetected in expression extracts from either plasmid eventhough pGW41 produced a 104-kDa activity band in vivo.

It is not clear why plasmids with apparently identicalinserts should produce grossly different amounts of poll. Thevariation in the abilities of pDM19 and pGW41 to enableaccumulation of poll might be due to alternative selectivepressures on the plasmids as they established themselves inpolA+ and polA mutant hosts. The polA gene is lethal to E.coli when it is present in a high-copy-number vector (11). Itis feasible that the H. influenzae polA suppressor genes ofpDM19 and pGW41 might have been subjected to selectionfor low levels of expression in the wild-type WA946 back-ground, especially if efficient suppression led to high levelsof polA expression. When the library was transformed intothe polA missense mutants the high level of poll activity inthe mutant under permissive conditions would have pre-vented the isolation of plasmid mutations that could increasethe expression of the cloned polA gene, even if such muta-

tions could increase the stability of the plasmid. On the otherhand, if moderate levels of poll activity can increase plasmidmaintenance without killing the cell, than the low basal levelof polI in the E. coli polAl mutant would have allowed the

1 2 3

o 5243 -

24 2w

15 jW

FIG. 4. In vitro gene expression products ofpDM19 and pGW41."S-labeled proteins were fractionated in a 15% SDS-polyacrylamidegel. The numbers next to the gel are masses in kilodaltons. Lanes: 1,pDM2; 2, pDM19; 3, pGW41. The 30-kDa band corresponds to13-lactamase whose expression is blocked in pDM19 and pGW41 byinsertional inactivation. The low levels of the remaining vectorpolypeptides in lanes 2 and 3 imply that the inserts were moreefficiently expressed than the vector sequence. The 52-kDa bandrepresents the largest plasmid-encoded polypeptide that could bedetected in these extracts.

J. BACTERIOL.

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

H. INFLUENZAE polA SUPPRESSORS 2647

isolation of plasmids that had mutated to produce moderatelevels of H. influenzae polymerase I. The differences in thein vitro gene expression products of pDM19 and pGW41imply that one of the plasmids underwent mutation in thegene for the 24-kDa polypeptide. Whether this mutation hasa direct effect on polA suppression remains to be seen.Similar reasoning could explain our inability to isolate thepol-i gene from the high-copy-number library. Overproduc-tion of the H. influenzae polI could be lethal to E. coli (11).An alternative possibility for the variable polA expression

in P3478(pDM19, pGW41) strains might involve the selectionfor compensating host mutations that affect the level ofpolAsuppression by the plasmids. We did not examine curedrecipient cells to determine whether host mutations cancontribute to variable polA expression. It appears unlikelythat large differences in plasmid copy number can accountfor the differences in polA expression in the suppressedstrains. Although we did not formally measure copy num-bers for the various pDM19 and pGW41 plasmids, all theplasmid-bearing strains routinely produced equivalent yieldsof plasmid in our standard plasmid purification procedure.

Construction of pDMl9 derivatives with defects in poUlsuppression. pDM19 was mutagenized with TnS and mini-TnWO kan to produce derivatives that could no longer sup-press the polA480 mutation of E. coli. The pDM19 nonsup-pressing derivatives were isolated, and three of thetransposons were mapped by digestion of the mutated plas-mids with restriction endonucleases. The transposon muta-tions mapped within less than 1 kilobase pair (kbp) of eachend of the chromosomal insert in pDM19 (data not shown).Since the pDM19 insert encodes at least three polypeptidesit appears that the transposon mutations must map in twosuppressor genes, one at each end of the insert. Inactivationof either gene by the insertion of mini-TnJO kan was suffi-cient to prevent polA suppression by the mutated plasmid.The polAl suppressor genes in pDM19 and pGW41 aretermed pol-2 and pol-3.

Construction of poll-deficient H. influenzae mutants. Acompetent BC200(pDM2) culture was transformed with apooled collection of nonsuppressing pDM19::TnS deriva-tives. On the basis of previous studies of plasmid transfor-mation, we expected that the transforming pDM19::TnSplasmids would produce Kanr transformants of two types(22). In one class of transformants, the plasmid could enterthe cell and establish itself as an autonomous molecule. Thisoutcome would produce transformants that were Kanr Camr.In other transformants, the plasmid could undergo recombi-nation with the recipient chromosome shortly after it enteredthe cell, leading to the integration of the cloned segment andthe destruction of the vector sequences. In this case thetransformants would be Kanr Cams. The vast majority(99.5%) of the Kanr pDM19::TnS transformants were alsoCamr, indicating that the plasmids usually became estab-lished in the cell without necessarily integrating the mutatedalleles into the recipient chromosome. A small fraction of theKanr transformants (0.5%) were Cams, indicating that theTnS integrated into the chromosome causing the loss of boththe transforming plasmid and the resident pDM2 (Cmr Apr).The apparent loss ofpDM2 was confirmed, because the cellsalso became Amps. Since we assumed that pDM2-basedplasmids would require polI for maintenance in H. influ-enzae as they did in E. coli, we tentatively assumed thatthese transformants were defective for polL. Similar trans-formation experiments were done with BC200 (pDM2) re-cipient cells and transforming pDM19. IfpDM19 contained aspontaneous mutation that reduced pol-I expression, we

1 2 3 4 56 7 8 9

:.... ,^;~~~~~~~~~~~~~~~....: ..

*.:... :.

FIG. 5. Polymerase activities in H. influenzae extracts ofpDM19and pDM19::TnS transformants. BC200 cultures were transformedwith either linear pDM19 PstI insert or suppressor-defective plas-mids pDM19::TnS-1 through pDM19::TnS-5. Transformants thatappeared to have lost the resident pDM2 were examined. Lanes: 1through 4, transformants of pDM19; 5, through 9, strains GW70through GW74, respectively. These strains are transformants ofpDM19::TnS-1 through pDM19::TnS-5, respectively. The top mostbands correspond to polI activity. The lower bands correspond tothe large fragment of poll. GW74 in lane 9 contains a low level ofpolI activity that can be seen only in the original autoradiograph,implying that the TnS did not disrupt the polA gene in this particularmutant.

might be able to isolate plasmid-free transformants that hadlow levels of pol-I expression because of the integration ofthe hypothetical pDM19 mutation. A collection of four AmpsCams transformants was isolated after transformation withpDM19.We screened all the apparent cured transformants for polI

activity in the polymerase activity gel. Three of the fivepDM19::TnS transformants (GW71, GW73, and GW74) wereunable to express the pol-i gene (Fig. 5, lanes 6, 8, and 9).GW74 contained a mutation that did not disrupt the pol-ireading frame since this plasmid produced a mutant thatshowed greatly reduced levels of active polI in the originalautoradiograph (Fig. 5, lane 9). All the Kanr transformantsthat still produced an active polI (GW70 and GW72) main-tained plasmids that had apparently undergone rearrange-ments that removed the Cmr and Apr markers of the residentplasmid, leading to their mistaken identity as cured strains(data not shown). Therefore, these isolates had not lost theability to maintain plasmids and presumably had not inte-grated the TnS into the cell chromosome. GW74 apparentlycontained an integrated TnS, but produced enough residualpolI to maintain a rearranged plasmid. All the apparentlycured pDM19 transformants produced high levels of polI(Fig. 5, lanes 1 to 4). The loss of the plasmid markers fromthese strains was probably a result of rearrangements thatremoved the Apr and Cmr genes from the resident plasmids.We did not test this inference for the pDM19 transformants.In any event, we did not isolate pDM19 transformants withreduced pol-i expression.GW71 and GW73 contained the integrated TnS within the

chromosomal segment that was homologous to the clonedpDM19 insert. Figure 6 shows a Southern blot of immobi-lized EcoRI digests from GW71 and GW73 hybridized toradioactive pDM19 DNA. Lane 1 represents chromosomalDNA from a wild-type strain that lacked Tn5. The pDM19insert hybridized with three EcoRI fragments of 0.9, 2, and 6kbp. The 6-kbp fragment was homologous to the internalEcoRI fragment of the pDM19 PstI insert. In the clonedinsert the internal EcoRI fragment was flanked by 0.96- and

VOL. 169, 1987

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

2648 WILLIAMS ETAL.J.BCEO.

A

23 4 2

FIG. 6. Integration of Tn5 into the GW71 and GW73 chromo-

somal regions that are homologous to the pDM19 insert. Chromo-

somal DNAs from MAP9 (wild type), GW71, and GW73 were cut

with EcoRI and analyzed by hybridization with radioactive pDM19.

(A) The exhaustive EcoRI digest for each strain. (B) The corre-

sponding Southemn blot. The samples in panels A and B correspond

to: lane 1, MAP9; lane 2, GW71; lane 3, GW73; lane 4, HindIII

fragmnents. The arrows indicate the fragments that were affected by

TnS insertion. The molecular weight standards in panel A are not

aligned with the autoradiograph.

0.2-kbp EccRI-PstI fragmnents. The poll-deficient mutants

each contained an insertion that could be explained by the

integration of Tn5 into one of the Pstl-EccRI fragments of

the pDM19 region. Lanes 2 and 3 show that the 2-kbp EccRI

chromosomal fragment, which is homologous to one edge of

the pDM19 insert, had increased in size by about 5.7 kbp, an

increment equivalent to the size of Tn5. The ethidium

bromide-stained digests in Fig. 6A indicate that all the

samples had undergone complete digestion. Therefore, the

7.7-kbp band in the DNA from the mutants did not represent

a partially digested fragment. We assume that the Tn5 was in

one of the pDM19 pclA suppressor genes that influences the

accumulation of poll in E. ccli. The Tn5 was located in the

pDM19 segment that was correlated with one of the suppres-

sor genes in the pGW19: :mini-TnlO kan deriviatives. We

define the defective gene in GW71 and GW73 as the pcl-2

gene. Since we did not align the Tn5 mutation of GW71 with

respect to the cloned insert, we do not know yet which of the

EcoRI-PstI fragments of the cloned insert represent pcl-2

and pol-3, respectively. Strains with transposons in the

second polA suppressor gene, called pcl-3, have not beenisolated, so we do not know what phenotypic effects areassociated with a pcl-3 defect.

Since we did not have the pcl-i gene to use as a hybrid-ization probe, we could not directly rule out the possibilitythat a Tn5 might also reside in pci-I in the polI-deficientmutants GW71 and GW73. We consider it unlikely that thepoll-deficient mutants could arise by transposition of the Tn5from the transforming DNA into the pcl-i gene. Almost allthe transforming pDM19: :Tn5 DNAs would enter the cells assingle-stranded molecules that would not be suitable sub-strates for transposition (1). Those transforming moleculesthat did not undergo synapsis with the homologous chromo-somal region would be rapidly degraded (12).

DISCUSSION

A surprising result of this study was the isolation of six H.influenzae chromosomal regions that could suppress polAmutations in E. ccli. All the suppressors relieved the polAdefects without encoding a new poll. It is not clear how anyof the clones suppress the polI defect in the missense andamber polA mutants. The suppressors relieved the pollrequirement for plasmid maintenance and also restoredviability to the conditional lethal polA mutant- RS5064 undernonpermissive conditions. Since we do not know how polIcontributes to the maintenance of pDM2, we consider thepossible mechanisms of suppression in terms of factors thataffect cell viability. We think that this is valid because theplasmids could restore viability to the conditional lethal polAmutant. The primary role for poll is the removal of RNAprimers from Okazaki fragments by nick translation (16).Either the 5'-3' exonuclease activity or the large fragment ofpoll is sufficient to maintain viability when E. ccli cellsgrown rapidly in rich medium (10). Other E. coli functionsapparently can substitute for either of these activities in polAmutants. All the poll functions are dispensable when thecells are grown slowly in minimal medium, presumablybecause the reduced demand for primer removal can be metby other E. ccli polymerases and nucleases (10). Therefore,a trivial mechanism of suppression might simply involve aplasmid-induced increase in the generation time. Our H.influenzae plasmids might also suppress the polA mutants byenhancing the E. coli functions that normally supplement themissing poll activities. For example, it is possible that thecollection of pclA suppressors includes a 5'-3' exonuclease,an RNase H activity, or a polymerase that cannot bedetected in the activity gel (10). It is unliely that any of thesuppressors encode products that form a complex with theE. ccli poll and stabilize its missense versions, because allthe plasmids suppress the polAl(Am) allele which is missinga large portion of the polA product. At least three of the polAsuppressors appear to affect DNA metabolism directly whenthey are present at a high copy number. pBSl and pBS3 areidentical to pMD11 and pDM12, two plasmids that suppressa DNA ligase mutation in E. ccli. pDMll also affects DNArepair when it is present in H. influenzae (20). The pBS3-pDM12 class of suppressors increased the spontaneousmutation frequency in an E. coli polA mutant (Table 2).pDM19 and pGW41 encode two genes (poi-2 and pol-3) thatmust act together to suppress polA mutations in E. ccli. pol-2positively controls the accumulation of poll in H. influenzae.The role of pol-3 in H. influenzae has not been tested,because pcl-3 mutants have not been constructed.

Either pDM19 or pGW41 might contain a spontaneousmutation that affects the expression of the polA gene. Both

J. BACTERIOL.

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

H. INFLUENZAE polA SUPPRESSORS 2649

plasmids are derived from the same PstI fragment of the H.influenzae chromosome; however, they show a difference intheir abilities to stimulate polI accumulation in the E. colipolAl mutant. Since pDM19 does not produce a polI band inP3478 it is reasonable to assume that this plasmid might havea mutation that reduces pol-I expression. The selectivepressure for the isolation of this type of mutation would bethe lethal effect of high poll levels in E. coli that would begenerated by the action of the wild-type cloned sequence(11). On the other hand, pDM19 might represent the wild-type version of the cloned region and pGW41 might encodea mutation that increases the expression of the polA in an E.coli background. In this case there might be positive andnegative factors that interact with each other to control thenet accumulation of poll. Inactivation of the negative factorwould lead to large increases in polI accumulation. Theappearance from pGW41 of an apparently truncated versionof the pDM19 24-kDa polypeptide is consistent with thishypothesis. Correlations between the products of the clonedinsert and the expression ofpolA will require a collection ofmutated pDM19 derivatives.Our results imply that H. influenzae chromosomal inserts

in high-copy-number vectors can undergo changes duringselection in E. coli. The pol-2 pol-3 plasmids showed grosslyvariable levels of polA expression. pDM20 also appeared tosuffer a deletion that removed one of the PstI junctionsbetween the chromosomal insert and the vector (Fig. 2A,lane 9). This variability is presumably related to fragment-specific selective pressures in E. coli. This might not repre-sent a serious problem for the study of many H. influenzaegenes that are subject to mutation in E. coli, becauseplasmid-borne mutated genes can be efficiently converted towild-type alleles when they are reintroduced into H. influ-enzae by transformation (22).

ACKNOWLEDGMENTS

We thank Catherine Joyce for the gift of plasmid pCJl.This work was supported by Public Health Service grant AI20367

from the National Institutes of Health to D.M.

LITERATURE CITED1. Barany, F., M. E. Kahn, and H. 0 Smith. 1983. Directional

transport and integration of donor DNA in Haemophilus influ-enzae transformation. Proc. Natl. Acad. Sci. USA 80:7274-7278.

2. Blank, A., J. R. Silber, M. P. Thelen, and C. A. Dekker. 1983.Detection of enzymatic activities in sodium dodecyl sulfate-polyacrylamide gels: DNA polymerases as model enzymes.Anal. Biochem. 135:423-430.

3. Bradford, M. M. 1976. A rapid and sensitive method forquantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72:248-254.

4. Boyko, W. L., and R. E. Ganschow. 1982. Rapid identification ofE. coli transformants by pBR322 carrying inserts at the PstI site.Anal. Biochem. 122:8548.

5. Brutlag, D., M. R. Atkinson, P. Setlow, and A. Kornberg. 1969.An active fragment of DNA polymerase produced byproteolytic cleavage. Biochem. Biophys. Res. Commun. 39:982-989.

6. deBrugin, F. J., a,nd J. R. Lupski. 1984. The use of transposonTnS mutagenesis in the rapid generation of correlated physicaland genetic maps of DNA segments cloned into multicopyplasmids-a review. Gene 27:131-149.

7. Delucia, P., and J. Cairns. 1969. Isolation of an Escherichia colistrain with a mutation affecting DNA polymerase. Nature

(London) 224:1164-1166.8. Herriot, R. M., E. M. Meyer, and M. Vogt. 1970. Defined

nongrowth media for stage II development of competence inHaemophilus influenzae. J. Bacteriol. 101:517-524.

9. Johnson, D. A., J. W. Gautsch, J. R. Sportsman, and J. H.Elder. 1984. Improved technique utilizing nonfat dry milk foranalysis of proteins and nucleic acids transferred to nitrocellu-lose. Gene Anal. Tech. 1:3-8.

10. Joyce. C. M., and N. D. F. Grindley. 1984. Method for deter-mining whether a gene of Escherichia coli is essential: applica-tion to the polA gene. J. Bacteriol. 158:636-643.

11. Joyce, C. M., W. S. Kelly, and N. D. F. Grindley. 1982.Nucleotide sequence of the Escherichia coli polA gene andprimary structure of DNA polymerase I. J. Biol. Chem. 257:1958-1964.

12. Kahn, M. E., F. Barany, and H. 0. Smith. 1983.Transformasomes: specialized membraneous structures thatprotect DNA during Haemophilus influenzae transformation.Proc. Natl. Acad. Sci. USA 80:6927-6931.

13. Kelly, W. S., and C. M. Joyce. 1983. Genetic characterization ofearly amber mutations in the E. coli polA gene and purificationof amber peptides. J. Mol. Biol. 164:529-560.

14. Klenow, H., and I. Henningsen. 1970. Selective elimination ofthe exonuclease activity of the deoxyribonucleic acid polymer-ase from Escherichia coli B by a limited proteolysis. Proc. Natl.Acad. Sci. USA 65:168-175.

15. Konrad, E. B., and I. R. Lehman. 1974. A conditional lethalmutant of Escherichia coli K12 defective in the 5'-3' exonucle-ase associated with polymerase I. Proc. Natl. Acad. Sci. USA71:2048-2051.

16. Kornberg, A. 1980. DNA replication. W. H. Freeman and Co.,San Francisco.

17. Lehman, I. R., and J. R. Chen. 1973. Persistence of deoxyribo-nucleic acid polymerase I and its 5'-3' exonuclease activity inpolA mutants of Escherichia coli K-12. J. Biol. Chem. 248:7717-7723.

18. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning, a laboratory manual, p. 89-91. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

19. McCarthy, D., N.-L. Clayton, and J. K. Setlow. 1982. A plasmidcloning vehicle for Haemophilus influenzae and Escherichiacoli. J. Bacteriol. 151:1605-1607.

20. McCarthy, D., K. Griffin, and J. K. Setlow. 1984. Plasmidcontaining a DNA ligase gene from Haemophilus influenzae. J.Bacteriol. 158:730-732.

21. Monk, M., and J. Kinross. 1972. Conditional lethality of recAand recB derivatives of a strain of Escherichia coli K-12 with atemperature-sensitive deoxyribonucleic acid polymerase I. J.Bacteriol. 109:971-978.

22. Setlow, J. K., N. K. Notani, D. McCarthy, and N.-L. Clayton.1981. Transformation of Haemophilus influenzae by plasmidRSF0885 containing a cloned segment of chromosomal deoxy-ribonucleic acid. J. Bacteriol. 148:804-811.

23. Smith, M. R., C. S. Devine, S. S. Cohn, and M. W. Lieberman.1984. Quantitative electrophoretic transfer of DNA from poly-acrylamide or agarose gels to nitrocellulose. Anal. Biochem.137:120-124.

24. Spanos, A., S. G. Sedgwick, G. T. Yarranton, U. Hubscher, andG. R. Banks. 1981. Detection of the catalytic activities of DNApolymerases and their associated exonucleases following SDS-polyacrylamide gel electrophoresis. Nucleic Acids Res. 9:1825-1839.

25. Stuy, J. H., and R. B. Walter. 1981. Addition, deletion, andsubstitution of long nonhomologous DNA segments by genetictransformation of Haemophilus influenzae. J. Bacteriol. 148:565-571.

26. Way, J. C., M. A. Davis, D. Morisato, D. E. Roberts, and N.Kleckner. 1984. New TnlO derivatives for transposon mutagen-esis and for construction of lacz operon fusions by transposi-tion. Gene 32:369-379.

VOL. 169, 1987

on April 18, 2018 by guest

http://jb.asm.org/

Dow

nloaded from