A single hydrophobic cleft in the Escherichia coli processivity clamp is sufficient to support cell viability and DNA damage-induced mutagenesis in vivo

Background The ubiquitous family of DnaN sliding processivity clamp proteins plays essential roles in DNA replication, DNA repair, and cell cycle progression, in part by managing the actions of the different proteins involved in these processes. Interactions of the homodimeric Escherichia coli β clamp with its known partners involves multiple surfaces, including a hydrophobic cleft located near the C-terminus of each clamp protomer. Results A mutant E. coli β clamp protein lacking a functional hydrophobic cleft (βC) complemented the temperature sensitive growth phenotype of a strain bearing the dnaN159 allele, which encodes a thermolabile mutant clamp protein (β159). Complementation was conferred by a βC/β159 heterodimer, and was observed only in the absence of the dinB gene, which encodes DNA polymerase IV (Pol IV). Furthermore, the complemented strain was proficient for umuDC (Pol V) -dependent ultraviolet light (UV) -induced mutagenesis. Conclusions Our results suggest that a single cleft in the homodimeric E. coli β sliding clamp protein is sufficient to support both cell viability, as well as Pol III, Pol IV, and Pol V function in vivo. These findings provide further support for a model in which different Pols switch places with each other on DNA using a single cleft in the clamp.


Background
Viability of all organisms depends upon a capacity to both accurately repair damaged DNA, as well as tolerate DNA lesions that for whatever reason evade repair [1]. In contrast to repair, which acts to either directly reverse the damage, or to excise modified bases so that the affected sequence may be re-synthesized, DNA damage tolerance mechanisms act to enable replication past the damaged site, without catalyzing repair of the lesion(s). Generally speaking, DNA damage tolerance mechanisms fall into one of two broad classes: (i) daughter strand switching, which refers to a collection of recombinational mechanisms that act to physically restructure the DNA at the replication fork to enable the complementary daughter strand to act as template to support replication beyond the damaged site(s) [1,2]; and (ii) translesion DNA synthesis (TLS), which refers to the process by which one or more specialized DNA polymerases (Pols) are recruited to catalyze replication past damaged sites in the DNA [3]. Since most Pols capable of catalyzing TLS display remarkably low fidelity on undamaged DNA, their actions must be very tightly controlled in vivo to guard against unwanted mutations [4,5]. Although multiple mechanisms likely contribute to the coordinate regulation of replicative and TLS Pols, considerable effort over the past decade has been devoted to understanding the roles played in this process by the ubiquitous family of DnaN sliding clamp proteins [5,6].
Bacterial sliding clamps, termed β or DnaN, are encoded by the dnaN gene, and function as homodimers. Like their eukaryotic counterparts, these clamps are loaded onto DNA in an ATP-dependent manner by a multi-subunit clamp loader complex [7]. Once loaded, they recruit the replicative Pol (Pol III), as well as other partner proteins involved in various aspects of DNA replication, repair, and damage tolerance [5]. E. coli β clamp, like other DnaN family members, contains a hydrophobic cleft positioned near the C-terminus of each protomer that interacts competitively with a conserved clamp-binding motif (CBM) sequence present in most, if not all partners (see Figure 1 &2; [8]). Since the β clamp functions as a homodimer, it contains two such clefts, suggesting it may simultaneously manage the actions of two different partner proteins on DNA by acting as a molecular 'toolbelt.' In this model, each partner is bound to a different cleft in the clamp (see Figure  1A; [9][10][11]). Consistent with this model, a single cleft in the clamp is sufficient to support assembly of the clamp onto DNA, as well as processive replication by the replicative E. coli Pol (Pol III) using an in vitro system reconstituted with purified components, suggesting the other cleft is available for physical interaction with a second partner protein [12,13].
Although the CBM-clamp cleft interaction is essential for biological function of all known clamp partners examined to date, it is becoming increasingly evident that partners make functionally important contacts with non-cleft surfaces of the clamp as well (reviewed in [5]). For example, residues E93 and L98 of the clamp, which are located on the rim, interact with TLS Pols IV and V (see Figure 2C; [13][14][15][16]). Using a heterodimeric clamp protein (β C /β + ) bearing a mutant protomer lacking a functional cleft (β C ) in complex with a

B.
A. The 'toolbelt' model posits that both a replicative and a TLS Pol associate with a single clamp, with each Pol binding to a separate cleft. In this model, the clamp acts to regulate sequential access of the two Pols to the DNA [9,10]. Displacement of Pol III from the clamp by Pol II, Pol IV, or Pol V is suggested to arrest replication in response to replication blocking DNA damage via a primitive DNA damage checkpoint control [18,20,38]. (B) In an alternative model, surfaces in addition to the cleft are postulated to play an important role in controlling access of the TLS Pol to the DNA (Residues E93 and L98 of the clamp control access of Pol IV [13,14], and possibly Pol V [16], for TLS), such that a single cleft in the clamp is sufficient to coordinate the switch (reviewed in [5]). TLS Pols may subsequently displace Pol III from the clamp to enable the checkpoint. Alternatively, binding of two or more TLS Pols may be required to displace Pol III and enable the checkpoint. See text for further details regarding these and other models. The green ring represents the β clamp, the small black circle in each clamp protomer represents the cleft, DNA is depicted in stick form, the blue oval represents Pol III, the red oval represents a TLS Pol, and the blue and red curved lines protruding from the respective Pols represents their CBMs, which must contact the clamp cleft to enable access of each respective Pol to the DNA. wild-type protomer (β + ), we recently determined that a single cleft, together with the adjacent rim contact were both required for Pol IV to switch with a stalled Pol III in vitro (see Figure 1B; [13]). These results suggest a single clamp may utilize a combination of cleft and non-cleft surfaces to simultaneously manage the actions of more than two partners on DNA [5,13].

Checkpoint
Besides their roles in TLS, Pols IV and V additionally act in a primitive DNA damage checkpoint control [17][18][19]. Pol II is also suggested to act in a checkpoint [19]. Interactions of these Pols with components of the replication machinery, particularly the β clamp, are suggested to slow or even arrest DNA replication following replication blocking damage to allow time for accurate repair (see Figure 1; [18][19][20]). However, the mechanistic relationship between the proposed checkpoint and Pol switching is presently unknown.
In addition to its protein partners, the clamp also interacts with the DNA template that it encircles [21,22]. Clamp-DNA interactions involve the cleft, as well as two additional clamp surfaces, one of which (residues H148-R152) also interacts with both Pol II and Pol IV [21]. Taken together, these results suggest that the clamp may play a direct role in physically sensing damaged DNA, and upon doing so, alter the way in which it interacts with the DNA template to enable recruitment of one or more DNA repair proteins, and/ or specialized Pols to enable a series of concerted switches designed to coordinate replication with DNA repair and TLS [21].
Our current appreciation of clamp functionality is based largely on results of in vitro biochemical assays (reviewed in [5,7,23]). For example, recent in vitro experiments exploiting heterodimeric clamps comprised of two distinct β protomers revealed that a single cleft of the clamp was both necessary and sufficient for supporting a switch between Pol IV and a stalled Pol III [5,12,13]. However, many questions concerning the biological significance of this as well as other current models remains largely untested, due to the lack of a method for analyzing defined heterodimeric clamp proteins in vivo. As part of an effort to address this deficiency, we engineered a synthetic dnaN gene expressing tandem clamp protomers fused head-to-tail by a short amino acid linker bearing a His 6 tag. Based on Western blot analysis using both antibeta clamp and anti-Penta·His (Qiagen) antibodies, the linker sequence was susceptible to a significant level of proteolysis in vivo (J.M. Duzen and M.D. Sutton, unpublished results). We therefore pursued a separate strategy that exploited an inactive, mutant form of the β clamp lacking a functional cleft (β C ) that we determined to be capable of complementing the temperature sensitive growth phenotype of the dnaN159 strain to provide a defined population of β C /β159 heterodimeric clamp protein for genetic analysis. Using this system, we tested several critical predictions of the models discussed above and summarized in Figure 1. Our findings, discussed below, suggest that a single cleft in the E. coli β clamp protein is sufficient to support cell viability, as well as manage the actions of Pols III, IV, and V during DNA replication and TLS in vivo.

Results & Discussion
b C complements the temperature sensitive growth phenotype of the dnaN159 strain provided that Pol IV is inactivated β C lacks a functional cleft due to deletion of the C-terminal five residues (see Figure 2), rendering it completely inactive for loading onto DNA, as well as for supporting Pol III replication [12]. In stark contrast, a heterodimeric form of the clamp bearing a recombinant β C protomer in complex with a wild-type protomer (β C / β + ) was recently determined to be indistinguishable from the wild-type clamp with respect to its ability to be loaded onto primed DNA, and stimulate processive DNA synthesis by Pol III in vitro [12,13]. The β C /β + heterodimer was also comparable to the wild-type clamp with respect to coordinating a switch between Pol IV and a stalled Pol III in vitro [13]. Based on these findings, we hypothesized that if a clamp bearing a single cleft was competent for supporting E. coli viability, then β C might complement the temperature sensitive growth phenotype of the dnaN159 strain via formation of a functional β C /β159 heterodimeric clamp protein in vivo (see Figure 2B). The dnaN159 allele expresses a mutant clamp bearing G66E and G174A substitutions (β159; see Figure 2), and strains bearing this allele are unable to grow at temperatures above 37°C [24]. Of importance to the work discussed herein, β159 does not undergo detectable proteolysis during incubation at elevated temperatures (see Figure 3; [24][25][26]), and was capable of forming heterodimers with either the wild-type or a mutant β clamp protein bearing alanine substitutions of residues 148-152 in vivo [21].
We initiated these studies by first asking whether physiological levels of the wild-type, β159, or β C clamp proteins, when expressed separately from a low-copynumber plasmid (Figure 3), were capable of complementing the temperature sensitive growth phenotype of the dnaN159 strain. Consistent with previous reports [26,27], strain MS101 bearing either the empty plasmid control (pWSK29), or the β159-expressing plasmid (pJD109), grew at 30°C, but not at 42°C (Figure 4). In contrast, this same strain containing a plasmid expressing the wild-type clamp (pJD100) grew equally well at both 30°and 42°C, indicating that β + fully complemented the temperature sensitivity of the dnaN159 strain. Despite the fact that a single cleft in the clamp was sufficient for supporting both clamp loading as well as Pol III function in vitro [12], β C was unable to complement temperature sensitivity of the dnaN159 strain ( Figure 4). Based on Western blotting, β C was expressed at physiological levels (Figure 3), implying that its inability to complement the dnaN159 strain was due to a functional defect.
We have previously described circumstances under which Pol II, Pol IV and/or Pol V are capable of interfering with viability of the dnaN159 strain [21,24,26,28,29]. In light of these findings, we hypothesized that the inability of β C to complement the dnaN159 strain may be due to the action of one or more specialized Pols. In order to Steady-state levels of the different clamp proteins were examined as described in 'METHODS.' As noted in 'METHODS,' we previously determined the number of clamp proteins expressed in E. coli MS101 bearing either pWSK29, or pWSK29-derived plasmids expressing different clamp proteins [27]. Note that both β159 and β C-G66E possess altered mobility in SDS-PAGE relative to the wildtype clamp, or the other mutants, due to the G66E substitution [24]. : Ω], and MS147 (dnaN159 ΔumuDC595::cat) bearing plasmids directing physiological levels of expression of the indicated clamp proteins were spotted onto Luria-Bertani agar plates supplemented with Amp and IPTG. Plates were photographed following overnight incubation at 30°or 42°C, as indicated. At least two independent transformants for each strain were examined. Representative results are shown. 'Control' refers to the strain bearing the empty pWSK29 plasmid. See Table 1 for details concerning the E. coli strains and plasmids used. distinguish between a model in which the inability of β C to complement temperature sensitive growth of the dnaN159 strain was attributable to a requirement of both clamp clefts for E. coli viability, and a model in which a single cleft in the clamp is sufficient to support viability, but one or more TLS Pols was acting to impair function of the β C /β159 heterodimer at 42°C, we asked whether β C could support temperature resistant growth of a dnaN159 strain lacking either Pol II [Δ(araD-polB):: Ω], Pol IV [Δ(dinB-yafN)::kan], or Pol V (ΔumuDC596:: ermGT) function. As shown in Figure 4 β C was able to fully complement temperature sensitive growth of the dnaN159 strain lacking Pol IV function [Δ(dinB-yafN):: kan]. In contrast, β C was unable to complement growth of the dnaN159 strain lacking Pol II (MS146) or Pol V (MS147). Thus, Pol IV may outcompete Pol III for interaction with the β C /β159 clamp to impair growth. Alternatively, the β C /β159 clamp may be sensitized to the checkpoint function of Pol IV. Regardless of the mechanism(s) by which Pol IV acts to impair growth of the β Cexpressing strain, the fact that β C , on its own, is nonfunctional [12], together with our finding that β159 cannot support cell viability at 42°C, even when expressed at an elevated level (Figure 3 &4), argues strongly that growth at 42°C of the dnaN159 strain relies entirely on the ability of a temperature resistant β C /β159 heterodimeric clamp to fulfill all essential clamp functions.
Disruption of the Pol IV-clamp rim contact in b C fails to alleviate the lethal effect of Pol IV The little finger domain of Pol IV (Pol IV LF ) interacts with the clamp by bridging the dimer interface [14]. In solution, and in the absence of DNA, two Pol IV molecules simultaneously contact a single clamp, with each Pol IV contacting the rim of one protomer, and the cleft of the adjacent clamp protomer (see Figure 2C; [13][14][15]). As discussed above, the cleft contact is required for stimulation of Pol IV replication [13,30], while the rim contact is required for Pol IV to undergo a switch with a stalled Pol III [13]. We therefore asked whether mutating critical residues in the rim in β C , which is adjacent to the cleft in β159 (see Figure 2 panels D & C), alleviated the need to inactivate Pol IV in order for β C to complement temperature sensitivity of the dnaN159 strain. The combination of the E93K and L98K mutations in the clamp was previously demonstrated to severely impair interaction of Pol IV with the rim [13]. A clamp mutant bearing only E93K-L98K substitutions (β E93K-L98K ) was expressed at a level comparable to the wild-type clamp (Figure 3). Moreover, the β E93K-L98K mutant clamp fully supported growth of the dnaN159 strain at 42°C, indicating that these residues were dispensable for essential clamp function(s) in vivo (Figure 4). In striking contrast, β C-E93K-L98K was unable to complement the dnaN159 strain, unless Pol IV was inactivated (Figure 4), despite the fact that the mutant clamp was expressed at physiological levels ( Figure 3). Taken together, these results indicate that the ability of Pol IV to impair growth of the β C -expressing strain is not the result of Pol IV gaining access to the cleft in β159 by first binding to the rim of β C in a manner similar to that by which Pol IV switches with a stalled Pol III (see Figure 1B; [5,13]). It is possible that Pol IV outcompetes Pol III for binding to the cleft of the β159 protomer, independently of the rim contact, leading to cell death at 42°C. Alternatively, binding of multiple Pol IV molecules to some combination of rim and cleft regions of a single β C /β159 clamp may act to preclude access and/or function of Pol III, possibly via a DNA damage checkpoint response. The affinity of Pol IV for the rim of the clamp is on the order of~1 μM [13], which is intermediate to the SOS-repressed (~300 nM) and SOS-induced levels (~3,300 nM) of Pol IV [5,31], providing support for these models.
b C complements the temperature sensitive growth phenotype of the dnaN159 strain via a b C /b159 heterodimer Results discussed above suggest that growth of the dnaN159 strain at 42°C relies on a β C /β159 heterodimer. We pursued two parallel strategies to test this hypothesis. In our first approach, we sought to purify for subsequent biochemical analysis a recombinant form of the β C /β159 heterodimer using an established protocol [12]. Although this recombinant clamp protein was expressed in a soluble form, it became poorly soluble when purified to homogeneity, making it impossible to rigorously establish its purity as a heterodimer, or to accurately measure its ability to support processive DNA replication in vitro (data not shown).
In a parallel strategy, we employed a genetic approach to determine whether a β C /β159 heterodimer supported growth of the dnaN159 strain. Residues I272 and L273 of β map to the dimer interface, and their substitution with alanine in the wild-type clamp leads to its monomerization in vitro [32]. We hypothesized that substitution of these residues in β C (β C-I272A-L273A ) would destabilize the β C /β159 heterodimer in vivo (see Figure  2D), thereby impairing growth at 42°C. Consistent with our hypothesis, β C-I272A-L273A was unable to fully complement temperature sensitivity of the dnaN159 strain, irrespective of Pol IV function (Figure 4). Our finding that β C-I272A-L273A was expressed at near physiological levels ( Figure 3) suggests that its inability to complement temperature sensitivity results from a functional defect. Efforts to clone a β I272A-L273A -expressing plasmid to measure its ability to complement the dnaN159 strain as a negative control were unsuccessful, suggesting that a monomeric form of the clamp bearing a functional cleft exerts a dominant-negative phenotype in vivo.
As part of this same strategy, we also substituted residue G66 of β C with glutamic acid (β C-G66E ). Temperature sensitivity of β159 requires both the G66E and G174A substitutions [24]. We hypothesized that introduction of the G66E substitution into β C (β C-G66E ) would effectively mimic its effect in β159, rendering both the β C-G66E mutant, as well as the β C-G66E /β159 heterodimer thermolabile. Consistent with this hypothesis, β C-G66E was unable to fully complement growth of the dnaN159 strain (Figure 4). Based on Western blotting (Figure 3), β C-G66E was expressed at physiological levels (note that the G66E substitution slows mobility of clamp in SDS-PAGE [24]). Thus, taken together, results discussed above suggest that a β C /β159 heterodimer supports growth of the dnaN159 strain at 42°C, suggesting that a single cleft in the clamp is capable of supporting all essential clamp functions in vivo.
b C supports DNA damage-induced mutagenesis in the dnaN159 strain Our finding that a β C /β159 heterodimer supports growth of the dnaN159 strain at 42°C provided us with a tractable system to determine whether a clamp bearing a single functional cleft was capable of coordinating the actions of Pol III and TLS Pols following DNA damage. Since growth of the β C strain at 42°C required that the gene for Pol IV be deleted, we were unable to analyze Pol IV function. However, we were able to analyze Pol V (umuDC) function. For this, we cultured the dnaN159 Δ(dinB-yafN)::kan strain expressing either the wild-type or β C clamp from a plasmid in liquid broth at 42°C. Exponential phase cultures were either irradiated with 254 nm ultraviolet light (UV), or mock irradiated, and appropriate dilutions of each culture were plated to determine the frequency of spontaneous and UVinduced Rif R . Both the β + (4.4 ± 1.7 × 10 -9 ) and β C (4.8 ± 1.7 × 10 -9 ) strains displayed spontaneous mutation frequencies comparable to those reported for similar E. coli strains [29,33]. DNA mismatch repair (MMR) function is required for correcting replication errors, ensuring a low (e.g., normal) spontaneous mutation frequency [1]. Inasmuch as interaction of both MutS and MutL with the clamp is required for MMR in vivo [34], these findings suggest that a single cleft in the clamp is capable of supporting MMR, and coordinating it with replication. In contrast to their spontaneous mutation frequencies, the frequency of UV-induced mutagenesis was~2.5-fold higher in the β C -expressing strain compared to the wild-type control ( Figure 5A), suggesting that the β C /β159 heterodimer may be impaired for proper coordination of Pol III and Pol V, resulting in more frequent access of Pol V to the replication fork following UV irradiation. Although we cannot rule out the possibility that partially functional β159 homodimers persist in the dnaN159 strain, we previously determined that β159 homodimers were impaired for Pol V-dependent UV mutagenesis at 37°C [24,27]. Taken together, these results argue that the β C /β159 heterodimer supports Pol V-dependent mutagenesis at 42°C.
In addition to correcting replication errors catalyzed by Pol III, MMR also acts to correct errors catalyzed by Pol V during TLS, including misinsertions opposite thymine-thymine dimers [35], as well misinsertions opposite undamaged bases adjacent to UV adducts which, if left uncorrected, result in 'hitchhiker' mutations [36]. Since the mechanism by which MMR catalyzes repair during TLS may differ from that during Pol III replication, we measured the frequency of UV-induced mutagenesis in the MMR-deficient ΔmutL::cat strain background (MS148). As summarized in Figure 5B, frequencies of UV-induced mutagenesis in the β + and β C strains were comparable, suggesting that β C /β159 was able to properly manage the actions of Pol III and Pol V in vivo. The spontaneous mutation frequency of the β C strain was elevated~2.5-fold relative to the β + strain (6.9 × 10 -6 compared to 2.5 × 10 -6 ), suggesting that one or more aspects of DNA replication and/or accurate repair were modestly affected by the β C /β159 clamp. Interestingly, a different dnaN159 ΔmutL::cat strain displayed a similarly elevated spontaneous mutation frequency at the permissive temperature of 30°C [33], suggesting that this phenotype was due to the β159 protomer. Taken together, results discussed above suggest that although a single cleft in the β clamp is sufficient to coordinate the actions of Pol III and Pol V in vivo (see Figure 1B), both clefts are required for proper MMR function during TLS.

Conclusions
Results summarized in this report indicate that β C complements both the temperature sensitivity and Pol Vdependent UV-induced mutator defect of the dnaN159 strain, provided that Pol IV is inactivated (Figure 4 &5). Taken together, these findings suggest that a single cleft in the clamp is sufficient to support all essential clamp functions in vivo, and provide additional support for our model that a single cleft is sufficient to coordinately manage the actions of multiple clamp partners on DNA (see Figure 1B; [5,13]). It is presently unclear why Pol IV function prevents β C from complementing the dnaN159 strain. It is possible that Pol IV associates with surfaces in addition to the rim and cleft to somehow impair growth, possibly as part of a DNA damage checkpoint control (see Figure 1; [18,19]). Alternatively, Pol IV may simply out compete Pol III for access to the β C /β159 clamp on leading and/or lagging strand. Other scenarios are also possible. Regardless of the mechanism(s), the fact that Pol IV impairs growth of the β C strain suggests that a single cleft on the clamp is also sufficient to support Pol IV function(s) in vivo. An obvious limitation of the method used to generate heterodimeric clamps in vivo is the possibility that partially functional β159 homodimers persist. Rational design of novel mutant clamp proteins bearing site-specific mutations at the dimer interface that impair homodimerization while simultaneously enabling dimerization in trans (e.g., heterodimers) would circumvent this issue, and would provide a powerful approach for dissecting mechanisms by which the clamp manages events at a replication fork in vivo. Finally, although not addressed in this study, proteins in addition to clamp, as well as the DNA template itself contribute to Pol switching (reviewed in [5]). Defining the respective contributions of these non-clamp factors in this multifaceted process will be made simpler once the roles played by clamp are defined in molecular terms.

Bacteriological techniques
Isogenic E. coli strains and plasmid DNAs utilized in this work are described in Table 1. Strain DH5α was used as host for cloning plasmids. Strains MS146, MS147, and MS148 were constructed by generalized transduction using P1vir [37]. The presence of the Δ (araD-polB):: Ω allele was confirmed by diagnostic PCR, as described previously [26]. Strains were routinely cultured in Luria-Bertani medium (LB; 10 g/l tryptone, 5 g/l yeast extract, 10 g/l sodium chloride; [37]). Strains bearing plasmids were grown in medium containing ampicillin (Amp) at a final concentration of 150 μg/ml. When noted, IPTG was added at a final concentration of 50 μM to induce expression of physiological levels of the plasmid-encoded clamp protein [27]. At least two independent plasmid clones were separately transformed into each strain, and at least two independent transformants of each plasmid/strain were used for each experiment.

Site-directed mutagenesis
Site-directed mutagenesis was performed using the Quick-Change mutagenesis kit (Stratagene). Primers employed in mutagenesis were synthesized by Sigma-Genosys, and their sequences are listed in Table 1. Plasmid pJD100 (dnaN + ), or pMDS110 (dnaN C ; see Table 1), served as template for PCR reactions. PCR amplification was for 18 cycles of 95°C for 30 s, 55°C for 1 min, 68°C for 12 min. After PCR, reactions (50 μl) were treated with DpnI (10 U) for 1 h to digest the parental template DNA prior to transforming it into chemically competent DH5α. Transformants were selected by virtue of their resistance to Amp. Plasmid clones were individually purified using the Qiagen Spin Prep kit, and screened by restriction analysis prior to determining the nucleotide sequence of two independent clones for each (Roswell Park Cancer Center Biopolymer Facility).

UV-induced mutagenesis
UV-induced mutagenesis was performed as described previously [27]. Briefly, cultures were grown to exponential phase (OD 595~0 .5) at 42°C in liquid LB medium supplemented with Amp and IPTG, at which point cells were harvested by centrifugation, washed twice with 0.8% saline before being resuspended in saline, and either UV irradiated (25 J/m 2 ) using a germicidal lamp (254 nM, GE Healthcare) or mock irradiated. Following irradiation, cells were allowed to recover overnight at 42°C in liquid LB medium supplemented with Amp prior to plating appropriate dilutions onto LB agar plates containing Amp with or without 100 μg/ml Rif. Mutation frequency was calculated by dividing the number of Rif R CFU/ml by the total number of viable cells/ ml. UV-induced mutation frequency is expressed as the frequency of Rif R observed following exposure to UV minus the spontaneous mutation frequency observed for the mock irradiated control.  pJD100 Amp R ; pWSK29 containing the dnaN + (β + ) gene under the control of its native promoters [26] pJD109 Amp R ; pJD100 bearing dnaN159 (G66E and G174A substitutions; β159) [27] pMDS110 Amp R ; pJD100 bearing dnaN lacking residues 362-366 (dnaN C ; β C ) This work