A novel rho promoter::Tn10 mutation suppresses and ftsQ1(Ts) missense mutation in an essential Escherichia coli cell division gene by a mechanism not involving polarity suppression

An extragenic suppressor of the Escherichia coli cell division gene ftsQ1(Ts) was isolated. The suppressor is a Tn10 insertion into the -35 promoter consensus sequence of the rho gene, designated rho promoter::Tn10. The ftsQ1(Ts) mutation was also suppressed by the rho-4 mutant allele. The rho promoter::Tn10 strain does not exhibit rho mutant polarity suppressor phenotypes. In addition, overexpression of the ftsQ1(Ts) mutation does not reverse temperature sensitivity. Furthermore, DNA sequence analysis of the ftsQ1(Ts) allele revealed that the salt-remediable, temperature-sensitive phenotype arose from a single missense mutation. The most striking phenotype of the rho promoter::Tn10 mutant strain is an increase in the level of negative supercoiling. On the basis of these observations, we conclude that the ftsQ1(Ts) mutation may be suppressed by a change in supercoiling.

A large number of genes have been implicated in Escherichia coli cell division. Most of the characterized genes are clustered around the 2-and 76-min regions of the E. coli chromosome (3; for reviews, see references 8 and 9). The 2-min region has been extensively studied and includes the ftsQ, ftsA, ftsZ (sulB, sfiB), and ftsI (sep, pbpB) genes. This region has been sequenced (33,35,36,47,48), and the individual protein products have been characterized (8,9,39,48). With the exception of the ftsI gene product (8,9), little is known about the function of the individual components. Ward and Lutkenhaus (44) demonstrated that overexpression offtsZ results in minicell formation. This observation and the demonstration that FtsZ is the target of the cell division inhibitor SulA (8,9) have led to the conclusion that FtsZ may be a rate-limiting component in cell division.
TheftsQ gene product is an essential component of E. coli cell division. Conditional lethal mutations have been isolated that filament and die at restrictive temperatures on medium without salt (4). We recently demonstrated that theftsQ gene encodes a stable, inner membrane-associated protein of 31,400 Da that is expressed at low levels in the cell (39). However, overproduction of the FtsQ protein had no apparent affect upon cellular morphology in wild-type cells such as strain AMC290 (Table 1).
To further investigate the role of FtsQ in cell division and to examine potential interactions with other cell division proteins, we isolated an extragenic suppressor offtsQl(Ts).
In this report, we characterize the ftsQJ(Ts) mutation and demonstrate that mutations in either the rho gene or its promoter are capable of suppressing theftsQJ(Ts) mutation.

MATERIALS AND METHODS
Media, bacterial strains, and bacteriophages. The bacterial strains and phages used in this study are listed in Table 1. P1 transductions were performed as described previously (30). Bacteria were routinely grown in complex medium (LB [29]) or in M9 minimal medium (30) without CaCl2 supplemented with amino acids (50 ,ug/ml) as necessary. LB medium lacking NaCl (TEY) was used to determine the temperature sensitivity of the ftsQJ(Ts) mutant strains. Kanamycin and carbenicillin were each used at a concentration of 50 ,ug/ml for agar media and 25 ,g/ml for liquid media; tetracycline hydrochloride was used at a concentration of 10 ,uglml. MacConkey agar-galactose plates (Difco) contained 1% galactose. Rifampin sensitivity was determined on LB plates containing 5 or 10 ,ug of rifampin per ml (13). Anthranilate sensitivity (31,32) was determined on M9 minimal plates containing leucine (50 pug/ml) and anthranilate (100 ,ug/ml). Streptomycin sulfate was used at a concentration of 200 ,ug/ml. Mapping techniques. TnJO transposition into the chromosome from X1098 was performed as recommended by Way et al. (45). This phage contains a defective TnWO that is only capable of a single transposition event. For initial mapping of TnJO in strain AMC501, a total of 17 Hfr strains (including strain PK3Hfr) were transduced to tetracycline resistance with P1 grown on AMC501. The tetracycline-resistant Hfr donor strains were then conjugated with AMC436 as recommended by Miller (30). After 10 and 90 min, tetracyclineresistant recipients were selected on LB containing tetracycline hydrochloride and streptomycin. A time-of-entry experiment was performed by using PK3 Hfr::TnJO as the donor and AMC436 as the recipient. Aliquots of the mixed culture were disrupted at 1-min intervals and plated as described above. More precise mapping was performed using P1-mediated generalized transduction.
Isolation of extragenic suppressors of ftsQl(Ts  (12). The phage DNA was packaged in vitro by using a Gigapack-Plus kit obtained from Stratagene. Packaged phage was plated on E. coli Q359 to select for recombinant phage. Plaque transfers onto nitrocellulose and hybridization conditions were as recommended by the supplier (Schleicher & Schuell, Inc.). The AMC436 library was screened with a 970-bp EcoRI-PvuII fragment, which was isolated from pDSC73 (39) and labeled with [(X-32P]dCTP by using an oligolabeling kit (Pharmacia LKB) as specified by the supplier. This 970-bp fragment contains all of theftsQ gene as well as the 5' end offtsA. One of the hybridizing phages was designated XDSC104. The AMC501 library was screened with [a-32P]dCTP-labeled pNK82 (14). This hybridization probe contains TnlO sequences from the ISIOR to the EcoRI site within the tetA gene. One of the hybridizing phages was designated XDSC100.
Plasmid constructions. The plasmids described in this report are shown in Fig. 1. Plasmids pDSC73 and pDSC78 have been described previously (39). Plasmid pDSC75 contains a 2,470-bp XhoI-HindIII fragment isolated from SUPPRESSION  XDSC104 cloned into Sall-HindIII-digested pT7T3-19U (Pharmacia LKB). This 2,470-bp fragment contains most of the ddl gene, all of ftsQl(Ts), and a large portion of ftsA. Plasmid pDSC76 was constructed by digesting pDSC75 with SmaI and PvuII and cloning the 1,640-bp fragment into HinclI-digested pT7T3-19U. Unidirectional deletions were constructed in plasmid pDSC76 by using the exonuclease III protocol described by Henikoff (15). Isolation of DNA. Bacterial strains were transformed by the CaCl2 method (22). Plasmid DNA was isolated from minipreparations of transformed DH5ax (16). Large-scale isolation of plasmid DNA from DH5aL was performed as described by Holmes and Quigley (17). Lambda phage DNA was obtained as described by Maniatis et al. (26). Chromosomal DNA was isolated as previously described (2). DNA sequencing. Double-stranded (plasmid) and singlestranded (M13) DNA sequencing was performed with modified T7 DNA polymerase (Sequenase; United States Biochemical) as recommended by the supplier. Plasmid pDSC76 and deletion derivatives of pDSC76 were sequenced with a standard M13 (-40) primer. The second strand was sequenced with a standard T7 sequencing primer or with synthetic oligonucleotides ( Table 2). The XDSC100 phage was partially digested with Sau3A, and the fragments were subcloned into M13mpl9. The M13mpl9 clones were screened with oligonucleotides complementary to the outside end of ISIOR inverted repeats bracketing the TnWO ( Table 2). Two of these hybridizing, recombinant M13 phage were isolated and sequenced by using a standard M13 (-40) primer and the synthetic oligonucleotides complementary to the outside end of ISIOR.
Measurement of plasmid supercoiling. Plasmid DNA was isolated from fresh overnight cultures of pBR322-transformed AMC434, AMC436, AMC436-64, AMC501, and DSC25 by using the protocol of Lodge et al. (24). The samples were electrophoresed in a 1% agarose gel containing 12 jxg of chloroquine per ml for 16 h at 3 V/cm. After the gel was washed for 4 h with double-distilled H20, the samples were capillary blotted onto GeneScreen (New England Nuclear Corp.) as recommended by the manufacturer. The DNA was hybridized with [ot-32P]dCTP random labeled pBR322 and washed as recommended by New England Nuclear.

RESULTS AND DISCUSSION
Characterization of the ftsQl(Ts) mutation. The region extending from the XhoI site near the amino terminus of ddl  to the PvuIl site within ftsA was subcloned from pDSC75 into pT7T3-19U, as described above (Fig. 1). Because this was a blunt-end cloning, we expected to obtain clones with the insert in either orientation. All 12 clones examined were oriented with the truncated ddl gene adjacent to the T3 promoter. We did not ascertain the reason all of the clones had this orientation, but we assume that transcription from the lactose promoter is deleterious when transcribed in the sense orientation. A series of Exolll-generated nested deletions ( Fig. 1) was sequenced with a combination of T7 sequencing primers, the M13 (-40) primer, and chemically synthesized 17-mer oligonucleotide primers. After sequence analysis of both strands (see Materials and Methods; Table  2), the only mutation detected was a guanine-to-adenine transition at position 397 (35). The transition mutation results in the substitution of a basic lysine residue for an acidic glutamate residue at amino acid 125. The insertion of an amino acid with a basic side chain could potentially disrupt a salt bridge. To verify that this G-to-A transition was responsible for the mutation, a trp-lac [Ptac-ftSQJ(Ts)I transcriptional fusion was constructed in plasmid pKK223-3. This construction (pDSC77) is identical to a previously described plasmid (pDSC78, ftsQ' [39]), except it contains the ftsQJ (Ts) allele. Plasmid pDSC77 does not complement a chromosomal ftsQJ(Ts) strain with or without induction of the ftsQJ(Ts) plasmid allele (we discuss this experiment in another context later in this paper). We then replaced the 620-bp KpnI fragment of plasmid pDSC77 with the same KpnI fragment from the wild-type ftsQ of plasmid pDSC78. The wild-type fragment restored the complementation activity. Furthermore, the DNA sequence of this region of the wild-type ftsQ gene, on plasmid pDSC73 (39), was determined to verify that our ftsQ' gene does not contain the mutation. Figure 2 is an autoradiograph from a sequencing gel comparing the nucleotide sequences of the regions from the nucleotides 369 to 423 (35) in the wild-type and mutant alleles.
Isolation and mapping of extragenic suppressors of ftsQl(Ts): a summary. We isolated extragenic suppressors of strain AMC436 ftsQl (Ts) (see Materials and Methods) and then attempted to map the temperature-resistant phenotype  (Tables 3 and 4). We have never precisely mapped suqA. We have attempted to minimize the confusion between suqA and TnJO of strain AMC501 by removing the data on suqA when possible. We now present details of the experiments concerning the isolation and mapping of the TnJO suppressor in strain AMC501, the major subject of this paper.
Isolation and chromosomal mapping of TnlO in strains AMC501, AMC502, and AMC503. A defective TnlO was randomly transposed into the chromosome by infecting a culture of AMC436-64 with X1098. Approximately 6,000 tetracycline-resistant clones were obtained after overnight incubation at 42°C on LB-tetracycline plates. P1 vir was grown on a suspension of the 6,000 tetracycline-resistant colonies. The resulting P1 lysate was used to transduce AMC436ftsQJ(Ts) to tetracycline resistance. These tetracycline-resistant transductants were then scored for temperature resistance by streaking on TEY at 41°C. Three temper- ature-resistant, tetracycline-resistant transductants were identified. P1 vir was grown on the three purified clones and used to transduce AMC436 ftsQJ(Ts) to tetracycline resistance. The resulting transductants were then scored for temperature resistance. One of the P1 lysates cotransduced tetracycline resistance and temperature resistance at a frequency of 100%. A temperature-resistant, tetracycline-resistant clone obtained from the transduction was designated AMC501. Our initial hypothesis was that suqA resulted from inactivation of gene function at an undefined locus and that TnlO transposed into this nonfunctional locus. An alternative hypothesis, that later proved to be correct, was that TnJO created a new suppressor locus. The other two lysates cotransduced the markers with a frequency of approximately 60%. Temperature-resistant, tetracycline-resistant transductants obtained from these two lysates were designated AMC502 and AMC503. To determine whether temperature resistance was due to reversion at the ftsQ locus, P1 grown on AMC5O1, AMC502, and AMC503 was used to transduce strain PK3 leuB6 to tetracycline resistance. The transductants were then scored for leucine prototrophy. Pl(AMC501) demonstrated no linkage between the TnJO and leuB. The Pl(AMC502) and Pl(AMC503) lysates cotransduced the TnJO and leuB alleles at frequencies of 97 and 64%, respectively. Because TnJO mapped in the 2-min region of the chromosome in these two strains, we assume a subpopulation of either the AMC436 or the AMC436-64 culture reverted at the ftsQ locus during the isolation of AMC502 and AMC503; thus, AMC502 and AMC503 were not characterized further. A problem encountered during the isolation of extragenic suppressors is the potential for acquiring suppressor tRNAs. The ftsMJ mutation was recently shown to be allelic with tRNAser (21), and div E42 was identified as a tRNAser suppressor (41). To evaluate our strains, the titers of several amber, ochre, and opal bacteriophages were determined on AMC436, AMC436-64, and AMC501. None of the phages were capable of growing on these strains (data not shown).
To map the TnJO insertion in strain AMC501, a time-ofentry experiment was performed by using PK3 Hfr::TnJO (strain PK3 Hfr transduced to tetracycline resistance with strain AMC501; see Material and Methods) as the donor and AMC426 as the recipient in a conjugation experiment. The time of entry was approximately 7 min, tentatively placing TnWO at 84 min on the E. coli chromosome.
Mapping of Tn1O of strain AMC5O1 by bacteriophage P1 transduction. A number of strains with genetic markers near the 84-min region of the E. coli chromosome were transduced to tetracycline resistance with P1 grown on strain AMC501. The tetracycline-resistant transductants were then scored to determine the percent linkage of the relevant genetic markers (Table 3). TnWO was 98% linked to the ilv C7 allele in strain JP58. Because the transduced strain had no isoleucine or valine auxotrophy and because the ilv operon is fairly large (-4 kb), we postulated that TnJO was immediately clockwise to the ilvC gene (Fig. 3). This region of the E. coli chromosome has been extensively characterized. Immediately clockwise to ilvC are the rep, trxA, and rho genes (5,37,42). The rep gene product, a helicase, is essential for the replication of filamentous phages such as M13mpl9 (reviewed in reference 40), and strain JM101 is a host for M13mpl9. Strain JM101 was transduced to tetracycline resistance with P1 grown on strain AMC501. Thereafter, the titers of an M13mpl9 lysate were determined on JM101 and on the two tetracycline-resistant isolates obtained by transducing JM101 to tetracycline resistance with Pl(AMC501). The titer of an M13mpl9 lysate was 4.5 x 1011 + 0.7 x 1011 PFU per ml on both strain JM101 and the JM101 TnJOcontaining strains, suggesting that TnJO was not inserted within the rep gene. The trxA gene encodes thioredoxin, an essential subunit for the T7 DNA polymerase (27). The trxA gene product is not essential for E. coli growth but is required for phage T7 replication (18). The ability of AMC436, AMC436-64, and AMC501 to support the growth of T7 was evaluated. The T7 titer was 4.0 x 109 + 0.6 x 109 PFU per ml of lysate for all three strains tested. Thus, we concluded TnJO had not disrupted trxA.
Mapping TnlO of strain AMC501 by DNA sequencing. Because of our inability to precisely map TnJO by using P1 transduction techniques, an alternative approach was adopted. A bacteriophage (XDSC100) was isolated that contained TnJO and adjacent chromosomal DNA sequences from AMC501. After XDSC100 DNA was subcloned into M13, clones of M13 containing AMC501 DNA that hybridized with oligonucleotides specific for ISIOR were examined by DNA sequence analysis (see Materials and Methods). By using the M13 (-40) and ISJOR-specific oligonucleotides as sequencing primers (Table 2), the TnJO insertion was localized to the -35 consensus sequence of the rho gene, with TnJO oriented such that transcription from the tetR gene is towards rho (Fig. 4) (22,33,36,42). The lowercase letters represent the sequence of the right side of TnlO (13). The underlined sequence is the region where a new -35 sequence could have been formed but apparently was not. The original -35 of rho is CTGGACG, the five capital letters to the left and two capital letters to the right of the indicated tetA-tetR (TnlO) insertion.
consensus -35 sequence (Fig. 4, underlined region). Because rho appears to be an essential gene in E. coli (7,20), this was a surprising finding. Therefore, it is probable that readthrough transcription from the tetR gene is sufficient for Rho protein synthesis. The trxA transcript has been shown to overlap the -35 region of rho (23,34,37,43). On the basis of the ability of AMC501 to support the growth of phage T7, insertion of TnJO into the 3' noncoding region of the trxA transcript does not have a detrimental effect upon thioredoxin synthesis. Further transduction mapping. Because TnJO and suppression of ftsQJ(Ts) were 100% linked, we needed to discern whether the rho:promoter:TnJO insertion was responsible for suppression of the ftsQJ(Ts) mutation or whether it simply mapped near the original suqA locus. Strain DSC23 ftsQJ(Ts) ilvY::TnlO was transduced to ilv+ with P1(AMC 436-64), P1(AMC501), and Pl(Morse 2055 rho4). The transductants were then scored for temperature sensitivity [and tetracycline resistance for the P1(AMC501) transduction]. The experimental results are summarized in Table 4. The observed linkage between suppression and isoleucine-valine prototrophy (74 to 78%) was consistent with published values for the linkage between rho and ilvY (19,42). No linkage was observed between suqA and ilvY; therefore, we conclude that the suqA gene maps elsewhere on the E. coli chromosome. To quantify the level of suppression, plating efficiencies were determined for the rho' and rho mutant strains on LB and TEY at 40°C. As demonstrated in Table 5, the rho4 and rho promoter::TnJO alleles both efficiently suppressed the ftsQJ(Ts) mutation. However, the colonies were smaller than those observed with the wild-type strain AMC434. Strain DSC24 rho4 apparently suppressed better than AMC501, as indicated by the larger colony size after a 19-h incubation on TEY. In fact, the AMC501 colonies required an extended incubation (42 h) before they were large enough to facilitate a reliable plate count. The wild- type strain, AMC434, was transduced to tetracycline resistance with Pl(AMC501) and tested for the ability to grow on LB and TEY at 40°C. There was no significant difference in the growth rate on the two media. Therefore, the slow growth rate of AMC501 on TEY at 40°C can be attributed to weak suppression, as opposed to any detrimental effects associated with insertion of TnJO into the rho gene. The data in Table 5 also demonstrate that the ftsQJ(Ts) suqA strain AMC436-64 cannot be distinguished from the ftsQJ(Ts) strain AMC436 when individual cells are plated. Nevertheless, the temperature resistance of AMC436-64 could be recognized on streak plates of the same medium (TEY) incubated at 40°C from the same inoculating culture. It was this finding that discouraged our further studies of the suqA allele.
It is difficult to explain the suppression of the ftsQl(Ts) mutation by either rho4 or rho promoter::TnlO. rho mutations are defined by their ability to suppress polar nonsense mutations (reviewed in reference 46). This suppression is attributed to inefficient transcription termination at Rhodependent termination sites (46). The ftsQJ(Ts) mutation is not expected to fall into the category of Rho-suppressible mutations. Because ftsQJ(Ts) mutant strains display a saltremediable, temperature-sensitive phenotype, it was assumed that they contain a missense as opposed to a nonsense mutation. Our sequencing of the ftsQJ(Ts) and the wild-type alleles (Fig. 2) confirmed these expectations. Furthermore, a plasmid-encoded wild-type ftsQ gene complements theftsQ mutation in anftsQl(Ts) recA56 mutant strain regardless of whether it is induced by isopropyl-p-D-galactopyranoside (IPTG) (39). These data are consistent with the hypothesis that the only detrimental effect of the temperature-sensitive mutation is upon the ftsQ gene product, not the downstream ftsA or ftsZ genes.
Is there a rho mutant phenotype associated with rho promoter::TnlO? In an effort to determine the effect of the TnJO insertion into the -35 region of the rho gene, the phenotype of AMC501 was evaluated. We examined the ability of our rho promoter: :TnJO mutation to suppress polar mutations both in the galactose operon (7) and in the tryptophan operon (31). Strain IT1022 galP3 ilv::TnJO his was transduced to ilv+ with P1 grown on AMC434, AMC436, AMC436-64, and AMC501. Transfer of the rho promoter: :TnJO allele in the transductants derived from AMC501 was verified by testing for tetracycline resistance. Seventy-eight ilv+ colonies from each transduction were then tested for polar suppression of the galP3 mutation on the basis of the ability of suppressed strains to form red colonies on MacConkey's galactose (1%) medium because of reduced transcription termination. None of the transductants utilized galactose as a carbon source. As a control, when the rho4 allele in Morse 2055 was transduced into strain IT1022, red galactose-positive colonies were obtained at high frequency. Suppression of the polar trpE9851 ochre mutation permits growth on medium containing anthranilate (31,32). Therefore, Pl(AMC501) was used to transduce Morse 2034 trpE9851 leu-277. Under conditions in which 160 tetracycline-resistant transductants were obtained per plate, no colonies were obtained on M9 medium supplemented with leucine and anthranilate. Thus, the rho::TnJO allele did not suppress polarity in either the gal or trp operon.
An effect of rho promoter::TnlO and rho4 on bacteriophage XN-independent growth. It has been demonstrated that some rho mutants inefficiently support the growth of XN mutants (11). Therefore, we determined the titers of several XN mutants on AMC434, AMC436, AMC436-64, AMC501, DSC23, and DSC24. The results are presented in Table 6. All of the phages grew well on the supE mutant strain NK5012, as expected. However, none of the experimental bacterial strains propagated either XN7 or XN7N53 well enough to produce plaques, including rho promoter::TnJO and rho4.
These results are not particularly surprising, since not all rho mutants are expected to propagate XN mutants (11). The Xnin5 mutation deletes approximately 5% of the X chromosome between the P and Q genes, including the terminator sequence tR2 and allows limited growth of XN mutants (11). Bacteriophage XN7N53nin5 produces plaques with an efficiency of approximately 5 x 10-4 on our rho' strains. In contrast, the same X produces plaques on both the rho promoter: :TnJO and the rho4 strains with an efficiency approaching 1 (Table 6). This is the only phenotype in which these two rho alleles behave similarly. We will present interpretations of these findings below.
Mutations in rho are frequently associated with hypersensitivity to rifampin (13). Therefore, we tested the sensitivity of AMC434, AMC436, AMC436-64, and AMC501 to rifampin. All of the strains were sensitive to as little as 5 jig of rifampicin per ml; thus, no conclusions could be drawn from the experiments.
Plasmid supercoiling. Recent evidence demonstrates that rho mutations can have a profound effect upon the level of supercoiling in cells (1,10). Because no demonstrable rho phenotype was found associated with the insertion of TnJO in the -35 region of rho, the degree of plasmid supercoiling in AMC501 was examined. Plasmid DNA isolated from AMC434, AMC436, AMC436-64, AMC501, and DSC25 was subjected to chloroquine gel electrophoresis. The rho promoter: :TnJO mutation in AMC501 and DSC25 had a dramatic effect upon the degree of plasmid supercoiling (Fig.  5). Plasmid DNA isolated from these strains was substantially more negatively supercoiled than DNA isolated from AMC434, AMC436, or AMC436-64, as reflected by an increase of 4 in the linking number. These data are in contrast to data normally found with rho mutants. It had been demonstrated that rho mutants frequently exhibit decreased negative supercoiling (1, 10). We were not able to recover plasmid pBR322 from a rho4 strain in several attempts, thus preventing our direct examination of the supercoiling of plasmid in this strain.
A remaining question which we do not answer in this paper but which is of considerable interest is the mechanism by which rho promoter: :TnJO causes increased negative supercoiling. The rho gene autoregulates its own transcription within narrow limits (6,28). The rho promoter::TnJO insertion does not create a recognizable new -35 consensus sequence at the site of insertion (Fig. 4, underlined sequence). The insertion might reduce autoregulation and lead to increased synthesis of rho mRNA if transcription started either within the tetR gene or at another upstream site.
Either the resulting protein might be more active (a Rho fusion protein) or there might be more of the wild-type Rho protein. If the Rho protein activity in the cell were higher than normal, increased negative supercoiling might result.
Models and their testing. One model to explain why either increased (rho promoter::TnJO) or presumably decreased negative supercoiling (rho4) could suppress the ftsQJ(Ts) allele is that local altered chromosomal supercoiling in the vicinity offtsQJ(Ts) might increase transcription and therefore translation of the mutant allele; a similar, more traditional model would be that perhaps there is a (second) polar nonsense mutation in the ddl gene immediately upstream of the ftsQl(Ts) mutation and that mutations in rho would suppress the temperature-sensitive phenotype of the ftsQJ(Ts) allele by causing polarity suppression (36). Again, increased transcription and translation of the FtsQl(Ts) protein would result. Both models were tested by varying the amplification of the transcription and translation of ftsQJ(Ts) as follows. (AMC434, AMC436, and AMC436-64) and rho promoter: :TnlO (AMC501 and DSC25) strains. Plasmid DNA was isolated from the strains, blotted onto GeneScreen, and hybridized with radiolabeled pBR322 DNA. VOL. 173, 1991 tional fusion was constructed in plasmid pKK223-3, designated pDSC77. By either model described above, plasmid pDSC77 should suppress the temperature-sensitive phenotype of the ftsQJ(Ts) strain under some conditions in which the expression of the plasmid-derived copy of the FtsQts protein in trans is varied. Overexpression of the ftsQl(Ts) allele from the Ptac promoter by induction with IPTG at maximal, intermediate, and zero levels did not restore growth at 40°C on complex medium without added salt (TEY medium). Growth on media with salt was not inhibited by IPTG at 40°C. Previously, we demonstrated that overexpression of the ftsQ wild-type allele had no deleterious effect when either complementing anftsQl(Ts) allele or in anftsQ wild-type strain on TEY media at any temperature with either no IPTG or optimal amounts for induction (39; unpublished results). We have directly demonstrated the regulated synthesis of the FtsQ protein when it was under control of the Ptac promoter (39). Thus, both models are no longer attractive. In this paper we have reported one previously undescribed phenotype of the rho4 and rho promoter::TnJO alleles; they allow bacteriophage XN7N53nin5 to plate with an efficiency approaching 1 (Table 6). While this finding can be interpreted as reducing polarity and causing increased transcription through remaining Rho protein-dependent terminators for rho4 at least (11,38,46), the molecular mechanism of this suppression may be dependent on the superhelical state of the X and/or chromosomal DNA.
At present, our working model to explain suppression of ftsQl(Ts) by the rho promoter: :TnJO and rho4 alleles is that the FtsQ protein specified by the ftsQJ(Ts) allele interacts more readily with DNA of either increased (rho promoter:: TnlO) or presumably decreased (rho4) negative superhelicity and is thereby stabilized in its function in the inner membrane (39). The hypothetical interaction of the FtsQ protein with DNA is supported by our finding that a weak SOS response, including reduced colony size and filamentous growth, occurs in Ion (but not wild-type) strains when the FtsQ wild-type protein is overexpressed; the weak SOS response is negated by mutations in recA, sulA, or sulB (unpublished observations). The alternative to our working model is that there is an unknown gene (suqA?) whose transcription is increased in either rho promoter::TnJO or rho4 mutant strains and that it is this gene whose product is directly responsible for suppression. Our strongest evidence against this model is that our rho promoter::TnJO allele exhibited none of the suppressor phenotypes of rho mutations despite an extensive search for these phenotypes. The weak suppression of ftsQJ(Ts) by suqA could be due to a different mechanism of suppression than that of either rho promoter:TnJO or rho4, since no change in supercoiling was detected in the suqA strain.