J. Mol. Biol. (1990) 211, 351-358
Mutagenesis by Proximity
to the recA Gene of Escherichia cdi
Shi-Kau Liu and Irwin Tessmant Department of Biological Sciences Purdue University West Lafayette IN 47907, U.S.A. (Received
7 June 1989, and in revised form
Escherichia coli recA(Prt”) strains, which produce protease constitutive RecA proteins in the absence of DNA-damaging treatments, display an increased frequency of spontaneous mutations. These mutations occurred preferentially in the neighborhood of the recA gene. This &s-like mutagenic effect was observed in the recA, rexAB, phoE and bio genes. The localized mutagenesis can be explained by the ease with which RecA(Prt”) proteins are activated to the protease state, which implies that there should be a relatively high concentration of activated RecA protein near the recA gene, where the protein is synthesized. The unusually high frequency of mutation in the recA gene is a novel example of an overactive gene preferentially turning itself down by mutation.
1. Introduction Exposure of Escherichia coli to DNA-damaging agents results in the induction of a diverse set of physiological effects collectively termed the SOS response, which is under the control of the recA and ZexA gene products (Walker, 1984; Peterson et al., 1988). The RecA protein is activated to a so-called protease state by signals generated in cells having damaged DNA, and the activated RecA promotes the proteolytic cleavage of the LexA protein (Little et al., 1980; Horii et al., 1981), the repressor of SOS genes (Mount, 1977; Brent & Ptashne, 1981). It is central to this paper that the RecA protein can also be activated to its proteolytic state by designated reeA(Prt”), which confer mutations, constitutive protease activity on the protein even in the absence of DNA-damaging treatments (Tessman & Peterson, 1985a,b). In vitro studies of two highly RecA proteins, active protease-constitutive RecA1202 and RecA1211, show that in contrast to the normal RecA+ protein these mutant proteins can be activated to cleave the LexA repressor by an enlarged number of cofactors; in particular, they can use natural nucleoside triphosphates in addition to ATP and dATP and can use RNA in addition to single-stranded DNA (Wang et al., 1988a,b). Greatly enhanced mutagenesis is one of the dramatic characteristics of the SOS response (Walker, 1984; Peterson et aZ., 1988). The products of the recA and umuDC genes are required for most of the spontaneous mutagenesis in recA(Prt”) strains
as well as for the SOS mutagenesis induced by u.v.$ light and a variety of chemical agents that damage DNA (Bagg et al., 1981; Witkin & Kogoma, 1984; Ennis et aZ., 1985; Tessman & Peterson, 1985a). Intact recA and umu genes are also essential for Weigle reactivation of u.v.-irradiated phage (IBagg et al., 1981; Defais, 1983). Because ZexA(Def) recA+ cells fail to show increased mutagenesis or Weigle reactivation in the absence of DNA-damaging treatments (Ennis et aZ., 1985), a second role for the RecA protein in these processes is indicated, and this role appears to be the proteolytic processing of UmuD protein to provide a C-terminal fragment that is necessary and sufficient for the role of UmuD in SOS mutagenesis et al., 11988; (Shinagawa et al., 1988; Burckhardt Nohmi et al., 1988). However, the UmuD cleavage product fails to restore mutagenesis to ZexA(Def) recA(Def) strains, implying that there is yet a third role of RecA in SOS mutagenesis (Nohmi e%al., 1988; Dutreix et al., 1989), quite possibly a direct one. Although the biochemical basis for this putative direct role is still not clear, both in wivo and in vitro studies on the interaction between RecA and the E subunit of DNA polymerase III suggest that RecA may enable the polymerase to read through1 the DNA lesions by reducing or inhibiting the proofreading function of the E subunit, and, as a 5 Abbreviations used: u.v., ultraviolet; kb, lo3 bases or base-pairs; XGal, 5-bromo-4-chloro%indolyl-b-ngalactoside; XP, 5-bromo-4-chloro-3-indolyl-phosphatep-toluidine.
t Author to whom all correspondence should be addressed.
0 1990 Bcademic Press Limited
Liu and lb. Newman
result, mutations would be generated when DNA replication bypasses the lesions with reduced et al., 1988; fidelity (Lu et al., 1986; Jonezyk Foster & Sullivan, 1988). The recA(Prt’) mutants display increased spontaneous mutagenesis (Tessman & Peterson, 1985a). We show here a novel localized mutagenic phenomenon in which the spontaneous mutations in the recA1202 strain occur preferentially in the recA gene itself and regions proximal to it
(a) Bacteria, bacteriophage and plasmids All of the strains used in this study are isogenic derivatives of E. coli K12 and are listed in Table 1. The Ts’ starting strain, EST1515, is stable against Mu lysis and Mu d(Zac) transposition (Tessman & Peterson, 1985a). Phages TC45 and U3 were provided by C. A. Schnaitman. A lysogen containing i recA1202 integrated into the proBA site was identified by its Pro- phenotype and the prophage location was confirmed by interrupted mating with a Pro’ Hfr donor (CGSCZSS) that has a point of origin between 97 and 98 min on the E. coli genetic map (Bachmann, 1972). This lysogen was made isogenic with the strain containing 1, recA1202 integrated at the att,I site: the deleted bio-gal region was restored by mating with an Hfr B8 strain (CGSC5015; Broda, 1967) and selecting the recombinants on an MS-minimal-Amp plate lacking added biotin and containing galactose as the sole
carbon source. A iysogen contraining 3,Teed - integrated at t,he proBA site was prepared in the same way. Because expression of the phoE gene is repressed when cells are grown in rich media containing abundant phosphate (Tommassen & Lugtenberg, 1980; Overbeeke &, Lugtenberg, 1980), the expression was made constitutive mut,ation (Wanner & by introduction of a phoG McSharry, 1982), which is located in the PST-phoU gene cluster near 83 min (Wanner. 1987). The plasmid pSKL2, a AumuC derivative of pSEll7. contains a deletion of 1.3 kb. It was constructed by digesting pSEl17 DNA with SmaI and religating with T4 DNA ligase. The plasmid containing the deletion was selected by the ability of a transformed ZexAjl(Def) recit T strain to form colonies at 35°C; by contrast, the pSElI7 transformant is cold-sensitive, being unable to grow at or below 35°C (Marsh & Walker, 1985). (b) &Yedia and
MS-CAA medium was M9 salts (Miller, 1972) supplemented per 1 with 2 g glucose, 5 g vitamin-free Casamino acids (Difco), 2 mg thiamine, 10 PM-FeCl,; 1 mM-MgSO,. and 0.1 mM-CaCl,. MS-CAA plates cont’ained in addition 15 g of agar/l. M9-CAA-XGal agar contained in addition 80 pg of XGal/ml. XP was added to agar plates by spreading 75 ~1 of a 20 mg/ml solution (in dimethyl formamide). LB agar contained (per 1) 10 g tryptone, 5 g yeast extract, 5 g NaCl, and 15 g agar. For the scoring of Biomutants; avidin (Sigma) was added to MM9minimal plates at 1 pg/ml to eliminate trace amounts of biotin. M9 minimal agar was similar to M9-C$A agar except that the Casamino acids were replaced with just 7 amino acids at
Table I Bacterial Strain EST1515 EST1813 IT2247 IT2228 BW6504 IT1022 IT2274 IT2373 EST1 122 EST1262 EST1301 IT 1366 EST2050 IT2276 IT2279 IT2312 ~IGSC5015 IT2378 IT2391 IT2499 LTZ506 IT2526 IT2530 IT2532 C1GSC259 pAE117 pSE137 pSKL2
genot,ype and phenotype
sulAl1 Alac-169 thr-1 leuB6 hi&4 IhiargE3 ilv(Ts) supE44 dinDl::Mu d(Ap lac) AweA srl::TnlO As EST1515, but I recA1202 CT ind 4s EST1813 but TetS As IT2247.‘& X13” Alac-169 phoU35 pho-510 thi his str sup gal-3 ilo Y::TnlO As BW6504. but iZvY::TnlO As IT2247, but phoU35 ilvY::TnlO As ESTl515, but Tet” As EST1 122, but /I recA + cl ind As EST1122, but i reeA1202 c1 ind Agal-attkbio nadA::TnlO -4s EST1301, but Agal-attkbio nadA::TnlO As EST2050, but proBA.9. reeAl202 e1 ind As IT2276; but Tet” As IT2279, but phoU35 ilvY::TnlO Hfr Broda 8 (point of origin: PO1 18) As IT2312, but bio+ attAfYgal+ nadA+ As IT2279. but bio+ &t/2+ ual+ nadA+ Ss EST20j0, but proBA:&cA+ e1 ind As IT2499, but Tet” As IT2506, but bio+ attA+ gal- na.dA+ As Il’2526, but phoU35 i2vY::TnlO As EST1262, but, pholT35 ilvY::TnlO Hfr H (Point of origin: POl) umuD+C+ Kan’ Amp’ (high copy number) umuD+C+ Spc’ (low copy number) As pSEll7, but AumuC
t Procedure of Maioy & Dunn (1981).
Tessman & Pet,eraon (1985n) Tessman & Peterson (1985a) Tet” from EST1813$ Selection of St-l’ S. L. Wanner Our collection Pl (TT1022) x BW6504 Pl (IT2274) x IT2247 Tet” from ESTl515t Our collection Our collection Our collection PI (IT1365) x EST1301 This work Tet’ from IT2276SPI (IT2274) x IT2279 CGSC IT9319 da x CGSCSOlS / IT2279 x CGSC6015 This work Tet” from IT3499S IT2506 x GGSC5015 P1 (IT2274) x IT2526 I’1 (IT2274) x EST1262 CGSC Marsh & Walker (1985) Elledge h Walker (1983) This work
cis-like action of the recA gene 40 pg/ml each: threonine, leucine, arginine, isoleucine, valine, histidine and proline. Antibiotics were used at the following concentrations: kanamycin, 30 pg/ml; rifampicin, 25 pg/ml; spectinomycin, 50 pg/ml; ampicillin, 50 pg/ml; tetracycline, 25 pg/ml. (c) Scoring mutants When recA1202(Prt”) cells were transformed with the high copy number umuDC plasmid pSE117, there occurred an exceptionally high frequency of mutation in the recA1202 gene, which led to the experiments reported here. The frequency of Ret-, Bio-, Rex-, and auxotrophic mutants were compared. This was done by growing the cells for only 3 or 4 generations following transformation, then plating for colonies on LB agar containing kanamycin to select transformants; and then testing the colonies for the various phenotypes. The transformants were deliberately grown for only a few generations in order to limit the fraction of cells that would lose their mutability due to mutation in the recA1202 gene. In the case of cells transformed with the low copy number umuDC plasmid pSE137 (Elledge & Walker, 1983), the mutation frequency was low enough that it was possible to purify the transformants and to initiate growth of a culture from a small inoculum. Bio- mutants were scored by their ability to grow on M9 minimal agar when supplemented with 1 pg biotin/ml, but not when supplemented with I pg avidin/ml. Rexmutants were scored by spotting the lambda lysogens on top agar containing the phage T4 rII deletion mutant NB668 (Tessman, 1962) at 3 x 106/ml; Rex- mutants do not form a visible spot. For recA, Aux, Bio- and Rex- mutant frequencies, the error values were determined from the number of observed mutants. For Rif, PhoE; LPS and PhoB+PhoU mutant frequencies, the error values were determined experimentally from the standard deviation of repeated observations.
353 gpf PhoE
Figure 1. E. coli chromosome map of 2 distant lambda attachment sites, adapted from Bachmann (1983).
plasmid pSE117. Fifteen of these mutant transformants chosen at random were found to display at least four distinctly different RecA-related phenotypes based on variations in colony color, u.v.sensitivity, and Weigle reactivation of phages lambda and S13 (data not shown). An examination of five independent pale blue transformants, including a representative of each phenotypic class, showed by Pl transduction mapping that the pale blue colony color was in each case linked to at& each mutation was approximately 6 o/o cotransducible with a nadA::TnlO marker (Fig. 1). It appeared, therefore, that the changes in the RecA phenotype were due to a high frequency of a variety of mutations in the recA gene, which is located near attl. An explanation for the high proportion of mutant transformants is that the combinati’on of the recA(Prtc) allele and high copy number pSE117 plasmid produces a powerful mutagenic effesct on the recA gene itself.
(a) Spontaneous induction of a variety of RecA mutant phenotypes in a recAlZOB(Prt”) strain containing the high copy number umuDC plasmid pSEll7
(b) Preferential mutagenesis in the recA gene of the recA1202 strain
In cells with a lac fusion to an SOS promoter, the expression of /?-galactosidase can be used as an indicator of the protease strength of the RecA protein (Kenyon & Walker, 1980; Tessman & Peterson, 1985a). For example, a 2 recA1202(PrtC) lysogen with the dinD::Mu d(Ap Zac) fusion (EST1813; Tessman & Peterson, 1985a) has very high protease activity and consequently forms a dark blue colony on XGal plates. It was striking, however, that more than 10% of the EST1813 cells produced pale blue colonies when plated only three to four generations after being transformed with the high copy number umuDC
It seemed unlikely that the unusually high spontaneous mutation frequency observed in the recA gene of recA1202/pSE117 cells could be tolerated throughout the genome. As a test of the mutation frequency for other loci on the chromosomle, we measured the mutant frequencies for rifampicin resistance (RIP) and auxotrophy, and compared these with the frequency of mutations in the recA gene that produced a change of colony color (dark blue-+pale blue) on MS-CAA-XGal plates (Table 2). For J. recA1202 lysogens without a umu plasmild, the preference for mutations in the recA gene was evident by a 22-fold higher frequency than for Rif’;
The growth rate of bacteria in MS-CAA at 37°C was determined from the A,,,, of log-phase cells at 2 time points 50 to 60 min apart.
Liu and I. Tesswhan
Table 2 Preferential
Strain EST1813 EST1813/pSE137
( & SEM) Rif’
2.0( iO.6) x lo-“ 47( kO.6) x 10m3
S,O(& 1.3) x 1o-6 35( &0‘5) x lo-.*
Ratio ( + plasmid/- plasmid):
umuD+CT umw,D+C+/pumuD+C+ Ratio (+plasmid/plasmid):
1.q kO.4) x 1OP 35( kO.3) x lo-’ 19(15)
3.1(+0.(i) x 1O-6 3-O(f0.7) x 1om5 9.7( F2.9)
ST 22( ,T) 5.7( +1.4) x 10m3 134( i26) ST NT
x19-CAA cultures were grown from small inocula (1000 to 2000 cells for reeAl202 strains; 10 to 20 cells for reeA1202/pSE137 strains; to approx. 2 x lOs/ml at 37°C; spectinomycin (50 pg/ml) was added for pSE137 transformed cells. For determination of mutant, frequency, the cells were then diluted and plated on the following media: MS-CAA f XGal (80 @g/ml) for recA mutant, frequency: LR. LB +rifampicin (25 pg/ml) for Rif’ mutant. and LB +spectinomycin (50 pg/ml) for auxotrophic mutant. All plates were incubated at 37 “C! for 24 h. t Each value was determined from at least 10 cultures. Aux, auxotrophic mutant. ST. not tested. SEM. standard error of mean.
this increased to 134-fold in the presenee of the low copy number umuDC plasmid pSE137. Since changes in growt’h rate could account in part for differences in mutant frequency, we measured the growth rate in MS-CAA of the recAIZOZ/pSE137 mutants that made pale blue colonies. No significant difference was found in generation time between six parent cells from four independent cultures and 12 mutant cells from eight independent cultures (parent/mutant = 1.05 + @02). Therefore, the increase in mutant frequency was apparently not due to a growth advantage but rather due to a higher mutation frequency. We estimated from a list of known E. co& Kl2 genes (Bachmann, 1983) that under our plating conditions auxot,rophic mutations in EST1515 could occur in roughly 90 or more different genes on the E. coli chromosome. (We counted genes needed for biosynthesis of the 13 amino acids, 7 vitamins, and nucleic acid precursors that were not in the medium.) Therefore, the average mutant frequency for each of these genes is at most 6.3 x IO-‘, which is 75-fold less than the frequency observed in the recA gene (Table 2). The preferential mutagenesis in the recA gene was also found in strains carrying 1 recA1202 located at the proBA secondary attachment site (Table 2). The similarity of the mutation frequencies at the two locations indicates that the location had no significant effect on the level of expression of the recA1202 gene. (c) Mutagenesis of the phoE gene located proximal and distal to the recA1202 gene
A critical test of preferential mutagenesis was obtained by comparing the effect of a i, recA1202 prophage locat’ed close to a target gene with the effect of the prophage located at a distance from the same target gene. We deleted the primary lambda attachment site and isolated a lysogen containing the /z recA1202 prophage at the proBA secondary site. This secondary site was chosen because (1) the lysogenization frequency is higher than that at most other secondary sites (Shimada et al., 1973); and (2) a mutated phoE gene, which is located close to
proBA (Fig. I), has a selectable phenotype, namely resistance to phage TC45, which arises because the PhoE protein is the receptor for TC45 (Chai $ Foulds, 1978; Pugsley et al., 1980). Our two isogenic 1” recA1202 lysogensi one containing the seeA allele proximal, the other distal, to phoE (Fig. I), were used for a comparative determination of the phoE mutation frequencies in the two strains (Table 3). The TC45’ phenotype in phoi! strains identifies three classes of mutants: PhoE-, LPS-, and PhoBor PhoV (Tommassen & Lugtenberg, 1981). U3’ characterizes LPS mutants (Beher & Schnaitman, 1981); pale blue colony color on LB-XP plates characterizes PhoBand PhoU’ mutants (Tommassen & Lugtenberg, 1981). To confirm that mutants having the phenotype TC45’. C3” and dark blue colony color were indeed in the phoE gene, we mapped two such mutants by interrupted mating with Hfr H (CGSC259) tso six minutes region on the chromosome: which is the phoE (Tommassen & Lugtenberg, 1981). The results (Table 3) indicate that phoE is subject to preferential mutagenesis. An &B-fold preference was observed when 1 recA1202 was in the proximal Bocat’ion. As expected, there was no significant difference between the two strains in their mutant frequencies for LPS and PhoB + PhoU+ When the low copy number umuDC plasmid pSE137 was the ratio of the phoE mutation introduced, frequency in the proximal to t’hat in the distal strain was increased to 14-fold. and again, no significant, difference in the mutant frequencies for LPS and PhoB + PhoU+ was observed (Table 3). The stimulating mutagenic effect of the zcrn&(: plasmid implies that the enhanced preferential mutagenesis in the recA and phoE genes (Tables 2 and 3) is produced by an SOS mutagenic process, which would mean that’ the proximal effect’ is dependent on the presence of the recAI202(PrLC) allele and is not just -the consequence of the insertion of a lambda chromosome. More direct evidence for this is provided by the control experiment in which the ;1 recA+ gene was tested in the proximal and distal locations; the phoE gene was not preferentially mutagenized to a significant extent in the
Table 3 Preferential Strain
by proximal Mutant PhoE
IT2378 (proximal) IT2373 (distal)
proBA::lrecA1202 phoU35 umuD+C+ attk:LrecA1202 phoU35 umuD+C+ proximal/distal ratio:
IT2378/pSE137 (proximal) IT2373/pSE137 (distal)
proBA::/lrecA1202 phoU35 umuD+C+ pumuD+C+ attl::lrecAlZOZ phoU35 umuD+C+ pumuD+C+ proximal/distal ratio:
IT2530/pSE117 (proximal) IT2532/pSE117 (distal)
proBA::IrecA+ phoU35 umuD+C+ pumuD+C+ attl::IrecA+ phoU35 umuD+C+ pumuD+C+ proximal/distal
frequency( + SEM) LPS
PhoB + PhoU+
52( f 1.2) x 1o-5
35( * 15) x 1o-4
1.4( iO.8) x 1om4
8.6( * 3.9) 2.7( + 1.1) x 10m3
1,3( [email protected]
2,5( * 1.9)
3,7( kO.6) x 1O-4
1.9( kO.2) x 1o-4
5,4( 52.1) x 1O-4
14(i6) 1.2( kO.3) x lo-*
1.4( kO.1) x 1o-5
1.9( kO.9) x 1om6
9.3( f 1.4) x 10-C
1.6( kO.5) x 10-a
Bacterial cultures were grown as described in Table 2 except 100 to 200 cells were inoculated for reeA1202 strains and 1000 to 2000 cells for recA+/pSEll7 strains. The cells were then centrifuged and concentrated 10 x in MS-CAA, and 0.1 ml were then mixed with 91 ml of phage TC45 (2.5 x lO“‘/ml) at 37°C. After 10 min, cell/phage mixtures were diluted and plated on LB agar with an additional 0.1 ml of TC45 (2.5 x 10’O/ml). The TC45’ colonies that emerged after 24 h were further tested for sensitivity to phage U3 and for color on LB-XP (50 pg/ml) by cross streaking each presumtive TC45’ colony with TC45 and U3 and LB-XP plates. Mutants were defined as follows: PhoE by TC45’, U3”, and dark blue on LB-XP plates (Tommassen & Lugtenburg, 1981); LPS by TC45’ and U3’ (Beher & Schnaitman, 1981); PhoB+PhoU+ by TC45’, U3’, and plae blue or white on LB-XP (Tommassen & Lugtenburg, 1981). For recA1202 strains, each value is the average for at least 6 cultures; for weA+ strains, each value is the average for at least 9 cultures.
proximal magnify plasmid
location even though we attempted to any effect by use of the high copy number pSEll7 (Table 3).
(d) Mutagenesis of the bio genes located proximal and distal to the recA1202 gene We measured the frequency of Bio- mutations in strains containing 2 recA1202 proximal (IT2228) and distal (IT2391) to the bio genes. To increase the mutation frequency and thereby facilitate the scoring, the cells were transformed with the high copy number umuDC plasmid pSEl17 and then grown three to four generations before plating for colonies. For each strain a total of 15,154 colonies were scored from six independent cultures. In the proximal strain there were 17 Bio- mutants among 116 auxotrophs; in the distal strain there were seven Bio- mutants among 85 auxotrophs. The probability that the same frequency distribution would give values of 17 and seven is only about 4%, i.e. the proximal value can be said to be greater than the distal one at the 96% confidence level. The proximal/distal ratio of 2.4 is nevertheless quite small. (e) Preferential
of the ,I rex genes
To further prove that mutations in the R recA1202 strains preferentially occur in regions near the recA gene, we measured in strain IT22281 pSEl*17 a mutant frequency for the rexAB genes, which are native to lambda. The mutant rex lysogens are easily identified by their ability to grow T4 rII mutants (Howard, 1967). The result was a
frequency of 3.4( f 1.1) x lop3 for the rex genes, which can be compared to 7.7( f0.7) x 10m3 (1 IS/IS, 154 described above) for auxotrophs. The latter represents the total frequency for the more than 90 genes, implying that the average mutation frequency for individual genes is no greater than approximately 8.6 x 10-5, which is 20-fold smaller than the average mutant frequency for each of the two rex genes. Thus, the /l rex genes, which are located proximal to lrecA1202, are also hypermutable. In the absence of the umuDC plasmid the number of rex mutants was 0 in 900 lysogens tested, suggesting as before that an SOS mutagenic process dependent on an activated RecA protein was involved. For this reason we did not look for further confirmation with a test of the mutation frequency in a L recA+ lysogen.
4. Discussion Genes proximal to the reclZOZ(Prt”) allele, in particular, phoE, rexAB and most striking recA itself, exhibited higher mutation frequencies than distal genes; a small effect was seen for the bio genes. In the phoE and bio cases a comparison was made of and dist’al locations by integrating proximal il recA1202 at either the primary or a secondary attachment site. Much less mutagenesis and no significant proximal bias was seen when the recA+ gene was used, which showed that the effect we have observed is specifically due to the recA(Prt”) allele rather than to a non-specific response to the inserted prophage. The phoE case provided the most critical evidence for a proximal mutagenic effect of the
Liu and 1. Teswwn
recA1202 allele; all the other evidence strongly confirms the proximal effect’. The constitutive activation of the RecA1202 protein to the protease st,ate in the absence of external DNA damaging agents suggests an explanation for the localized mutagenicity. The coupling of translation to transcription allows the RecA protein to be synthesized in close proximity to the recA gene. And because it can use R1”\‘A in addition to single-stranded DNA as a cofactor in its activation (Wang et al., 19883), the RecAl202 protein could be immediately activated to the protease state, and, like RecA441, would probably bind to the nearby duplex DNA (Lu & Echols, 1987). When the replication fork moves through this region, the activated RecA protein will be in place to carry out its mutagenic function. One need not invoke a specific model for mutagenesis, but the possibility that mutagenesis arises from inhibition of the proofreading function of the E subunit of DNA polymerase III (Lu et al., 1986) is eminent’ly consistent with the general model. The model should apply to all recA(Prt”) mutants, but most especially to those with the greatest constitutive activit,y, which were designated class 1 mutants (Tessman & Peterson, 1985a). Another class 1 mutant, re~A1213~ which is known to be altered at a different site (Wang & Tessman, 1986), showed a mutation frequency in the recA gene very much like that of the recA1202 strain (unpublished data). The experimental error might have obscured a very small proximal mutagenic effect of the recA+ allele (Table 3); a better mutant selection syst’em would be needed to settle that point,. Evidence that enhanced localized mutagenesis was due to the plasmid umuDG genes is that cells containing the pSKL2 plasmid: a AzcmzcCderivative of pSEl17; did not produce pale blue colonies on XGal agar; all recA1202/pSKL2 transformants displayed a uniform dark blue colony color (unpublished data). Thus; the umuDC plasmids, whether high or low copy number, appear to enhance the mutagenic effect of recA1202. Furthermore, in a recAl202 cell an intact urn& gene is still required for Weigle reactivation as well as for spontaneous and u.v.-induced mutagenesis for they are all severely reduced in a umuC::TnS strain (unpublished data). The stimulating effect of the umuDC plasmids implicates SOS mutagenesis in the proximal effect, which confirms the specific involvement of the recA(Prt”) allele. It might be argued that the several types of recA mutations were produced by integration of the umuDC plasmid into different regions of the chromosome through accidental sequence homology. Such a mechanism is unlikely since pSKL2, which contains a deletion of only 1.3 kb from a total of 10.3 kb, does not produce the mutations. The recA1202/pSE137 cells that make dark blue colonies do not have any growth disadvantage on MS-CAA plates relative to the mutant recA derivatives that make pale blue colonies, which are indicative of weaker RecA protease activities. Therefore
the high frequency
of reed mut’ants derived from can be att,ributed to a high mutation frequency rather than to selection. On LB plates the recA1202/pSE137 cells that make dark blue colonies do grow more poorly than those making pale blue colonies; this disadvantage is not shown by recA1202/pSKL2 cells? which have the umuC deletion. Hence, t,he combinations of the recAl202 mutation and overproduced UmuDC proteins is noticeably harmful to the cell in rich medium. Because of the high copy number of the plasmid. the mutation frequency in the recA gene of recA1202ipSElI7 t,ransformants is so high t,hat almost all the seemingly dark blue colonies contain mutated cells that make pale blue colonies on replating. Because of their growth advantage on LB plates these mutated cells overgrow most of the original colonies, which eventually appear pale blue; it is difficult, therefore, to accurately measure the mutation frequency of the recA gene in recA1202/ pSElli cells. For the low copy number recA1202/ pSE137 transformants, however, the mutat!ion frequency of the ,recA gelle is much lower than that in the high copy number umuDC transforman& and there is no striking insta,bility of the reciZ gene; more than 99q& of the colonies appear to be uniformly dark blue on M9CAAXGal plates. The mutat’ion frequencies t#o Rex- and auxotrophy were greatly increased by introduction of pSE117 into recAl202 cells. and that’ facilitated the t’edious scoring of mu&nts. Those frequencies are bound to be somewhat underestimated because of the high frequency of mutations in the recA gene t’hat eliminate th.e mutagenicity. Nevertheless, the effect on the induction of Rex- and auxotrophie mutations should be about the same and therefore a comparison of the two frequencies would be meaningful. But the comparisons are quite rough because we do not’ know how many nucleotide sites in any gene can be altered to produce observable phenotypic changes and we do not know how much of an effect nueleotide sequences have on mutability. Rif” mut’ants map in the rpoB gene and encompass roughly 20 nucleotides (Jin & Gross? 1988). The number of potential recA sites may not be considerably higher for the following reasons. (1) The number of mutational sites in recA1202 leading to the pale blue colony phenotype is probabl_v also limited because we do not score null mutations that eliminate protease function; but only those t$hat modify it to reduce the constitutivity; sequence analysis of recA(Prtc) mutants showed &at mutations resulting in a Prt” phenotype were elustered in short regions of the recA gene (Wang Bi Tessman, 1986). Furthermore, we would not have recognized mutations from the Prt” Ret+ to a Prt” Ret- phenotype because the color would still have been dark blue on XGal agar. (2) The average frequency of mutations that lead to auxotrophy is at most 6-3 x 1O-m5per gene, which is in the same range as that for Rif’ (Table 2). This suggests that recA1202ipSE137
action of the recA gene
the frequency of mutation to Rif’ is not far less than that producing defectiveness in other genes and therefore that the comparison of Rif’ with the recA mutant frequency is reasonable. For order of magnitude calculations, we made the crude assumptions that there are no exceptional hot spots for mutation to auxotrophy and that the mutation frequency is more or less uniform over most of the potentially responsible genes. In a sample of 18 amino acid auxotrophs, we found that 13 of them had acquired single amino acid deficiencies represented by eight of the 13 amino acids missing from the minimal medium; the multiple deficiencies in the five remaining mutants were not identified. These observations argue against any dominance by just a few mutational sites. Similarly, in a simple test of only three phenotypic characteristics, 15 recA mutants that were examined could be divided into at least four distinguishable groups, indicating that a variety of recA sites were mutated. The cumulative evidence is thus consistent with a proximal mutagenic effect of the recA(Prt”) gene. Mutation of the bio genes in pSE 117 transformant (high copy number) showed only a small though significant bias when recA1202 was located at the att1 site (proximal) compared to the proBA site (distal). The small effect might be due to the methodology specific to the pSE117 case. Because of the high mutagenicity in the recA gene, the transformed cells were grown for only three to four generations, which limited the opportunity for mutation and thus may have allowed the background level of Bio- mutants to dominate the ratio of the number of proximal to distal mutants. The localized mutagenic effect of the recA1202 allele is reminiscent of the classic example of gene A in the S13-4X174 phage family, which is preferentially, but not exclusively, &s-active (Tessman, 1965,1966). In both cases, the initial proximity of the gene product to its own gene can account for the exaggerated localized action. A similar comparison has been made with the preferential c&action of transposon-encoded functions (Grindley & Joyce, 1981; Foster et al., 1981). Preferential activity proximal to the site of synthesis may be a widely occurring phenomenon. The example here is a case of an overactive gene turning itself down by targeting itself, more than other genes, for mutation; furthermore, it shows that the RecA protein provides an essential function at the immediate site of the mutagenic event.
We are grateful to several people for their kind assistance: Barry L. Wanner for critically reading the manuscript and for several helpful discussions; Chien-Tsun Kuan and Won-B0 Wang for criticism of the manuscript; Patricia K. Peterson for helping us score the Biomutants; Jeffrey R. Lucas for advice on the statistical analysis of two-way tables; Graham C. Walker for providing the pumuDC plasmids; Carl A. Schnaitman for providing the TC45 and U3 test phages and advice on their use. We are also indebted to one of the reviewers for an especially careful reading of the manuscript and for
many helpful criticisms. This work was supported Public Health Service grant GM35850.
References Bachmann, B. J. (1972). Bacterial. Rev. 36, 525-557. Bachmann, B. J. (1983). Microbial. Rev. 47, 180-230. Bagg, A., Kenyon, C. J. & Walker, G. C. (1981). Proc. Nat. Acad. Ski., U.S.A. 78, 574%5753. Beher, M. G. & Schnaitman, C. A. (1981). J. Bacterial. 147, 972-985. Brent, R. & Ptashne, M. (1981). Proc. Nat. Acad. Sci., U.S.A. 78, 4204-4208. Broda, P. (1967). Genet. Res. 9, 35-47. Burckhardt, S. E., Woodgate, R., Scheuermann, R. H. & Echols, H. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 1811l1815. Chai, T.-J. & Foulds, J. (1978). J. Bacterial. 135, 1641-170. Defais, M. (1983). Mol. Gen. Genet. 192, 509-511. Dutreix, M., Moreau, P. L., Bailone, A., Galibert, F.; Battista, J. R., Walker, G. C. & Devoret, R. (1989). J. Bacterial. 171, 2415-2423. Elledge: S. J. & Walker, G. C. (1983). J. Mol. BioZ. 164, 175-192. Ennis, D. G., Fisher, B., Edmiston, S. & Mount, 1~. W. (1985). Proc. Nat. Acad. Sci., U.S.A. 82, 3325~-3329. Foster, P. L. & Sullivan, A. D. (1988). Mol. Gen. Genet. 214, 467-473. Foster, T. J., Davis, M. A., Roberts, D. E., Takeshita, K. & Kleckner, N. (1981). Cell, 23, 201-213. Grindley, N. D. F. & Joyce, C. M. (1981). Cold Spring Harbor Symp. Quant. BioZ. 45, 1255134. Horii, T.; Ogawa, T., Nakatani, T., Hase, T., Matsubara, H. 8: Ogawa, H. (1981). CeZZ,27, 515-522. Howard; B. D. (1967). Science, 158, 1588-1589. Jin, D. J. & Gross, C. A. (1988). J. Mol. BioZ. 202, 415-58. Jonczyk, P., Fijalkowski, I. & Ciesla, Z. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 912449127. Kenyon, C. J. & Walker, G. C. (1980). Proc. Nat. Acad. Sci., U.S.A. 77, 2819-2823. Little, J. W., Edmiston, S. H., Pacelli, L. Z. & Mount, D. W. (1980). Proc. Nat. Acad. Sci., lJ.S.AL. 77, 3225-3229. Lu, C. & Echols, H. (1987). J. Mol. BioZ. 196, 497-504. Lu, C., Scheuermann, R. H. C Echols, H. (1986). Proc. Nat. Acad. Sci., U.S.A. 83, 619-623. Maloy, S. R. & Nunn, W. D. (1981). J. Bacterial. 145, 1110~1112. Marsh, L. & Walker, G. C. (1985). J. Bacterial. 162; 155-161. Miller, J. H. (1972). Experiments in Molecular Genetics, Cold Spring Harbor Laboratory, Cold Spring H.arbor, N.Y. Mount, D. W. (1977). Proc. Nat. Acad. Sci., U.S.A. 74, 309304. Nohmi, T.: Battista, J. R., Dodson, L. A. & Walker. G. C. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 18161820. Overbeeke, N. & Lugtenberg, B. (1980). FEBS L,etters, 112, 229-232. Peterson, K. R., Ossanna, N., Thliveris, A. T., IEnnis, D. G. & Mount, D. W. (1988). J. Bacterial. 170, 14. Pugsley, A. P., Lee, D. R. & Schnaitman, C. A. (1980). Mol. Gen. Genet. 177, 681-690. Shimada, K., Weisberg, R. A. & Gottesman, M. E. (1973). J. Mol. BioZ. 80, 2977314. Shinagawa, H., Iwasaki, H., Kato, T. 85 Nakata, A. (1988). Proc. Nat. Acad. Sci., U.S.A. 85, 180661810. Tessman, E. S. (1965). Virology, 25, 303-321.
Liu and I. Teseman
Tessman, E. S. (1966). J. MOE. Biol. 17, 218-236. Tessman, E. S. & Peterson, P. (1985a). J. Bacterial. 163, 617-687. Tessman, E. 8. & Peterson, P. K. (198%). J. Bacterial. 163, 688-695. Tessman, I. (1962). J. Mol. Biol. 5, 442445. Tommassen, J. & Lugtenberg, B. (1980). J. Bacterial. 143, 151l157. Tommassen, J. & Lugtenberg, B. (1981). J. Bacterial. 147, 118-123. Walker, G. C. (1984). Microbial. Rev. 48, 60-93. Wang, W.-B. & Tessman, E. S. (1986). J. BacterioZ. 168, 901-910. Wang, W.-B., Sassanfar, M., Tessman, I., Roberts, J. W.
& Tessman, E. S. (1988a). J. Bacterial. 178. 48164822. Wang, W.-B., Tessman, E. S. & Tessman, I. (198%). J. Bacterial. 170, 4823-4827. Wanner, B. L. (1987). In Escherichia eoli und Suimoaella typhimurium: Cel2ulnr and Molecular Biology (Neidhardt, F. C., Ingraham, J. L., Low. K. B.. Magasanik, B., Sehaechter. M. & Umbarger. H. 15.. eds), pp. 1326-1331. American Society for Microbiology, Washington. Wanner, B. L. & McSharry, R. (1982). J. Mol. BioE. 158. 341-363. Witkin, E. M. & Kogoma, T. (1984). Proc. Nat. Acad. Sci.. U.S.A. 81, 7539-7543.
by J. W. Miller