Prior to discussing E. coli Y-family polymerases DinB (Pol IV) and UmuD'₂C (Pol V), the functions of the E. coli umuD gene products will be briefly described, as the umuD gene products play critical roles in regulating the activities of both DinB and UmuC. UmuD is the product of one of the genes whose expression is coordinately upregulated along with dinB and umuC [13,28,29]. The umuDC genes are organized in an operon; however, the levels of UmuD appear to correlate more closely with those of DinB than UmuC [30]. Notably, UmuD is not found in all species that have UmuC present, indicating that UmuD is not generally required to regulate UmuC or DinB, although other proteins may fulfill this function in organisms that lack UmuD [8]. UmuD₂ is the predominant form of the protein for the first 20 to 40 minutes after induction by the SOS response [31]. UmuD, in conjunction with UmuC, acts in a DNA damage checkpoint [32]. When E. coli cells are grown at 30 °C and UmuD and UmuC are present at elevated levels, they inhibit DNA replication in a role distinct from their function in TLS [31,33]. They also inhibit the replication process after the cell has been exposed to UV light [31]. When UmuD₂ interacts with the RecA/ssDNA nucleoprotein filament, the filament facilitates UmuD autocatalytic cleavage, thereby removing the 24 N-terminal amino acids of UmuD to form UmuD' [34,35,36]. UmuD cleavage is similar to the autocatalytic cleavage of LexA, also facilitated by the RecA/ssDNA nucleoprotein filament [37,38,39]. However, the catalytic efficiency of cleavage is much greater for LexA than it is for UmuD₂[34]. UmuD₂ cleavage typically occurs about 20 to 40 minutes after the initiation of the SOS response [11,31]. The cleaved form of UmuD₂, UmuD'₂, then interacts with UmuC to form UmuD'₂C (Pol V), which is capable of performing TLS [11,40,41,42]. UmuD' and UmuC prevent RecA-dependent homologous recombination as a result of the interaction between UmuD'₂C and the RecA/ssDNA nucleoprotein filament [43,44,45]. Full-length UmuD₂ is involved in prevention of mutagenesis by UmuC or DinB, whereas UmuD'₂ is involved in facilitation of mutagenesis via Pol V; thus, cleavage of UmuD represents a switch from a non‑mutagenic state to a mutagenic state of a cell [46].

DinB, initially identified as the product of the dinP gene [47], was discovered in 1980 as one of the DNA damage-inducible genes and was named dinB [48]; both names are used in the literature. The dinB gene encodes one of the two E. coli Y-family DNA polymerases (DinB, or Pol IV) capable of bypassing lesions in DNA via translesion synthesis [49]. DinB is the only Y-family DNA polymerase that is conserved throughout all domains of life, although S. cerevisiae apparently lacks a DinB ortholog [8,50]. In non-SOS conditions, DinB is expressed at approximately 250 molecules per cell; however, the number of DinB molecules increases by approximately 10-fold under SOS-induced conditions [51]. During the SOS response, DinB is expressed at the highest level of all five DNA polymerases in E. coli [51]. Like other Y-family polymerases, DinB is a DNA-dependent DNA polymerase that is capable of copying damaged primer-template structures and lacks 3'-5' exonuclease proofreading abilities [49]. The fidelity of the dinB gene product is lower than that of the replicative polymerase in E. coli, namely the Pol III holoenzyme [52]. The presence of DinB during TLS can increase mutagenesis as a result of its relatively low fidelity and the lack of 3'-5' exonuclease activity. [49,51,53,54,55,56].

In addition to being upregulated by the SOS response, DinB is involved in a process known as adaptive mutagenesis [57,58]. In an experiment using Lac reporter strains of E. coli, there was a seven‑fold decrease in mutations in strains lacking the dinB gene compared to in the wild-type strain, suggesting that DinB can induce mutations [57]. These mutations, which result from the adaptive mutagenesis process, can cause cells to have a selective advantage during stressful conditions [59]. The precise mechanism for adaptive mutagenesis and how it leads to high levels of DinB expression is not fully understood [60,61,62], although one mechanism may be through DinB involvement in error‑prone double-strand break repair [63]. Overall, DinB expression can be considered a general stress response mechanism that can lead to high rates of mutagenesis and could ultimately result in antibiotic resistance, as described below [57,64,65,66]. Moreover, DinB contributes to cellular fitness and long‑term survival in stationary phase [67]. DinB also has a non-catalytic function, in that elevated levels of DinB slow the replication fork in a checkpoint-like phenomenon [68,69,70].

Currently, no crystal structure of DinB has been solved; however, homology models [71,72,73] have been constructed using the crystal structures of Dpo4 from Sulfolobus solfataricus [23] and Dbh from Sulfolobus acidocaldaricus [16] as templates. These models allow for the prediction of specific residues involved in DinB function. For example, the steric gate residue, which is the amino acid residue of a DNA polymerase that prevents ribonucleotides from entering the active site [74,75], of DinB is F13 [73]. Changing the steric gate residue of DinB increases the frequency of ribonucleotide incorporation from less than 10⁻⁵ to 10⁻³, as well as increases the ability of DinB to replicate undamaged DNA [73].

DinB is known to bypass certain dG adducts (Table 1). For example, DinB bypasses N²-furfuryl dG [73], N²-benzo[a]pyrene-dG [72,76], N²-(-1-carboxyethyl)-dG [77], N²-N²-dG interstrand crosslinks [78], and γ-hydroxypropano-dG [79]. DinB is effective in bypass of N²-dG adducts formed from benzo[a]pyrene (B[a]P), a bulky polycyclic carcinogen which is metabolically activated to form 7R,8S-dihydroxy-9S,10R-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene [72,76]. In the presence of DinB, N²-B[a]P-dG adducts are bypassed with relatively high fidelity and efficiency with a misincorporation frequency of 10⁻² to 10⁻⁴ [76]. In addition, DinB has been shown to accurately and efficiently bypass N²-(1-carboxyethyl)-2'-deoxyguanosine (N²-CEdG) adducts [77]. N²-CEdG minor groove adducts are formed endogenously from methylglyoxal, which is a byproduct of glycolysis [80,81], and are detected at one lesion per 10⁷ nucleotides in human melanoma cells [80].

An E. coli strain containing a deletion of the dinB gene is sensitive to both nitrofurazone (NFZ) and 4-nitroquinoline-1-oxide (4NQO) [73]. Both of these DNA-damaging agents are thought to form N²‑dG adducts in vivo [82,83]. DinB shows greater accuracy and 15-fold increased proficiency of dCTP insertion opposite N²-furfuryl-dG, an N²-dG adduct likely formed from NFZ, than opposite undamaged DNA [73]. DinB has also been shown to accurately bypass N²-N²-dG interstrand crosslinks (ICLs), which can disrupt DNA replication [78]. Interstrand crosslinks are typically repaired by cooperation between homologous recombination and nucleotide excision repair [84], but recent work has shown that TLS by DinB may also play a role in repair of ICLs [78]. The γ-hydroxypropano-dG adduct, as well as other adducts formed from α,β-unsaturated aldehydes, can form DNA-peptide crosslinks [79]. DinB has been shown to bypass these acrolein-mediated adducts as well as the interstrand crosslinks and the peptide conjugates that form from these adducts [79,85].

DinB confers resistance to the alkylating agent methyl methanesulfonate (MMS) [86]. A cluster of DinB residues referred to as the ‘aromatic triad’, F12, F13, and Y79, is important for survival of E. coli cells in the presence of MMS [87]. E. coli strains that contain single-point mutations in the ‘aromatic triad’ residues within DinB show fewer MMS-induced mutations than nitrofurazone-induced mutations, which suggests that these residues not only play a significant role in TLS, but also are involved in modulating the accuracy of DinB in bypassing specific lesions [87].

DinB is also involved in bypass of lesions that result from various reactive oxygen species leading to A:T → G:C transitions [88]. In studies with defined lesions, DinB was shown to preferentially insert dATP opposite 5-formyluracil (5-fodU) and 5-hydroxymethyluracil (5-hmdU); both dCTP and dATP opposite 7,8-dihydro-8-oxoguanine (8-oxo-dG) with low efficiency; and both dCTP and dTTP opposite 1,2-dihydro-2-oxoadenine (2-oxo-dA) with dCTP inserted more efficiently [88]. In addition, DinB was found to incorporate 8-hydroxy-dGTP opposite both adenine and cytosine and 2-hydroxy-dATP opposite both guanine and thymine in vitro [89,90]. Oxidation of dGTP to 8-oxo-dGTP is the cause of cell death that results from treatment with antibiotics and from elevated levels of DinB, due to increased incorporation of 8-oxo-dGTP by DinB [89]. The evidence supports a model in which cytotoxicity results from double strand breaks caused by incomplete repair of 8-oxo-dG lesions that are closely spaced [89].

It has been further shown that DinB is capable of adding dGTP opposite the modified pyrimidine 1,3-diaza-2-oxophenothiazine (tC) specifically but is incapable of continuing TLS beyond the modified base [91]. This is intriguing since DinB binds slightly more strongly to DNA primer/template constructs that contain the tC analog than it binds to an undamaged DNA primer/template, which may indicate a specific inability to bypass major-groove modified bases in DNA [91]. However, it was also found that DinB inserts tC opposite G in the DNA template and is capable of extending from the newly-incorporated tC, suggesting that DinB shows asymmetric discrimination against the modified DNA template and the incoming nucleotide [91].

The error frequency of DinB on undamaged DNA is approximately 2.1 × 10⁻⁴ for generating frameshift mutations and about 5.1 × 10⁻⁵ for generating base substitution mutations [52]. DinB is also known to bypass abasic sites, causing −1 frameshift mutations [52,92]. One model for this is a ‘dNTP‑stabilized’ misalignment mechanism in which dNTP is placed correctly opposite a template base downstream rather than placing an incorrect nucleotide opposite the next available template base [52]. More recently however, evidence for another mechanism involving template slippage has been observed, which provides another possible explanation for the generation of −1 frameshift mutations [93]. The template slippage mechanism causes single base deletions on DNA containing homopolymeric runs [93]. When UmuD₂ is bound to DinB, a non-slipped conformation is preferred which prevents the generation of frameshift mutations [93].

It is known that the following proteins interact directly with DinB and affect its replication efficiency: UmuD₂, RecA, NusA, Rep helicase, single-stranded DNA binding protein (SSB), and the β-processivity clamp subunit [30,94,95,96,97,98]. The presence of UmuD and RecA improves the catalytic efficiency of DinB and also reduces the number of −1 frameshift mutations generated by DinB in vitro [30]. Deletion of the umuD gene leads to an increase in the frequency of −1 frameshift mutations; however, the resistance of such cells to nitrofurazone was not affected, suggesting that the ability of DinB to perform TLS and the mechanism for generating −1 frameshift mutations are separable functions [30].

The C-terminus of DinB (residues 347–351) binds to the β-clamp subunit of the Pol III holoenzyme; this interaction contributes substantially to both enzyme processivity in TLS and dNTP-binding affinity [94,99,100,101]. A second binding site has also been observed between residues 303–305 of the DinB little finger domain and residues located near the dimer interface of the β-clamp subunit [100]. Binding of DinB to the β-clamp subunit increases the processivity of the polymerase, helps to position DinB correctly at the replication fork, and coordinates polymerase usage [100,102,103,104,105,106,107].

NusA, which has roles in elongation, termination, and anti-termination of transcription [108,109,110], has also been shown to interact with DinB. NusA recruits DinB to gaps in the DNA template strand during transcription-coupled TLS when RNA polymerase is stalled by a lesion in DNA [95,96]. NusA has been shown to be necessary for stress-induced mutagenesis by DinB [111]. The exact binding site of DinB on NusA is unknown, but it is believed to be located in the C-terminal domain of NusA [95,112]. The NusA-DinB interaction is predicted to bridge the gap between replication-coupled TLS and transcription-coupled TLS [96]; therefore, this work demonstrates a crucial connection between replication and transcription, especially in the presence of DNA damage.

E. coli Pol V is the second of the two Y-family polymerases found in E. coli. Pol V consists of two different protein subunits, one UmuD'₂ dimer and UmuC, which interact to form UmuD'₂C [42,113,114,115]. The umuC gene was discovered in the late 1970s to be required for E. coli cells exposed to UV light to mutate [114]. The ability of UmuC to bypass UV-induced DNA adducts via TLS was not discovered until well after UmuC was characterized as being involved in SOS mutagenesis. Several models were proposed over the course of the next few decades to explain the role of UmuC in mutagenesis [116,117,118,119]; however in the late 1990s, it was determined that UmuC was in fact a DNA polymerase [40,41,42,120]. This was confirmed when it was shown that UmuC exhibited low fidelity lesion bypass on its own, but its fidelity and efficiency increased in the presence of RecA, SSB, and UmuD' [40,120,121]. In a key finding, UmuC maintained its ability to function even in the absence of the Pol III holoenzyme [40,120].

Similar to the SOS regulation of dinB expression, the genes encoding the protein constituents of Pol V, umuC and umuD, are both regulated by the SOS response but the umuD and umuC genes are located within the same operon [28,29,122]. Also similar to DinB, upon induction of the SOS response, the expression of the Umu proteins increases 10-fold, with UmuC increasing from approximately 15 to 200 molecules and UmuD increasing from approximately 180 to 2,400 molecules [123]. DNA repair processes such as nucleotide excision repair typically remove a lesion once it has formed in DNA [124,125]; however, in the event that nucleotide excision repair does not take place, replication will recover upon induction of Pol V [125]. It has been shown that when the umuC gene is deleted, there is moderate DNA synthesis recovery and when the recJ gene is deleted, there is poor recovery of DNA synthesis after damage [126]. When both the umuC and recJ genes are deleted, there is no recovery of DNA synthesis after damage, indicating that recJ and umuC both act to restore stalled replication forks [126].

The ability of Pol V to perform TLS is dependent on the formation of the UmuD'₂C complex and the presence of RecA [40,127]. Full-length UmuD is involved in preventing UmuC from engaging in TLS and therefore preventing mutagenesis by UmuC. Changing the active site Ser60 residue to Ala in full-length UmuD prevents autocatalytic cleavage of UmuD₂[35]. Full-length UmuD is involved in prevention of mutagenesis; thus it was found that UmuD₂ harboring the S60A mutation results in a large reduction of UV-induced mutagenesis [31,128,129]. Cells that contain UmuD₂-S60A with UmuC experience greater sensitivity to UV light relative to cells that contain wild-type UmuD and UmuC [35,128,129]. Molecular interactions between UmuC and UmuD have been difficult to determine; however, through immunoprecipitation, glycerol gradient analysis, and yeast two hybrid assays, the physical interaction of UmuD' and UmuC was confirmed [115,130]. The interaction of full-length UmuD and UmuC was also confirmed using affinity chromatography and velocity sedimentation analysis [115]. The physical interaction between UmuD' and UmuC consists of one UmuC protein bound to a dimeric UmuD' [115], with UmuD' interacting with the 25-amino acid C‑terminal end of UmuC [32,115].

To date, there is no experimentally-determined structure of UmuC. Homology models have been constructed of the polymerase and little finger domains [24,128,131], but as the C-terminal domain possesses little homology to proteins of known structure, a model of the entire protein cannot be constructed at this time. Still, the model has allowed specific predictions of the functions of particular residues to be tested. In particular, the steric gate residue (Y11) [131,132,133], hydrophobic residues (I38, A39) near the nascent base pair [134], and a cluster of residues (N32, N33, D34) near the incoming nucleotides have all been shown to contribute to UmuC function [135].

In general, Pol V is capable of bypassing lesions formed in DNA that DinB is incapable of bypassing. Pol V can bypass lesions caused by exposure to UV light such as thymine-thymine (T-T) cis-syn cyclobutane pyrimidine dimers (CPD) and T-T (6-4) photoproducts (Table 1) [101,105,106,136]. In addition to bypassing lesions from UV light, Pol V is capable of bypassing abasic sites, C⁸-dG lesions such as N-2-acetylaminofluorine (C⁸-AAF), as well as being required for replicating DNA containing 5'S-8,5'-cyclo-2'dG [92,101,105,106,136,137]. Pol V bypasses lesions caused by UV light with greater efficiency when in the presence of the β-clamp subunit, the RecA/ssDNA nucleoprotein filament, and SSB [136]. Pol V can be mutagenic when carrying out TLS. For example, Pol V inserts dGTP opposite the 3'T in T-T (6-4) photoproducts instead of dATP with a six-fold greater frequency [136]. However, Pol V is also known to bypass certain lesions with high accuracy. For instance, Pol V faithfully inserts dATP opposite both Ts of T-T CPDs [136] as well as dCTP opposite C⁸-AAF-dG [106]. Pol V also bypasses N²-benzo[a]pyrene-dG adducts, N⁶-benzo[a]pyrene-dA adducts and adducts resulting from oxidation with varying accuracy [72,76,138]. Pol V replicates undamaged DNA with error frequencies of 10⁻³ to 10⁻⁴, which is a much lower fidelity than the Pol III holoenzyme and a lower fidelity than DinB [136]. It should be noted that Y-family DNA polymerases can be accurate or error-prone depending on the cellular and DNA context in which they are acting.

In addition to facilitating the autocatalytic cleavage of the LexA repressor protein as well as facilitating the cleavage of full-length UmuD₂ to UmuD'₂, RecA/ssDNA nucleoprotein filaments also play a direct role in TLS [40,41,139,140,141]. RecA has been determined to stimulate both nucleotide insertion and extension [141]. One model suggests that Pol V and two RecA molecules form a complex that activates Pol V for TLS in the presence of ATP [127,142]. There is evidence that a distinct RecA/ssDNA nucleoprotein filament transfers RecA and an ATP from the 3' end of this trans filament to Pol V, which activates Pol V for TLS [143,144]. However, other work suggests that the RecA/ssDNA nucleoprotein filament acts in cis on DNA directly downstream of pol V rather than in trans to facilitate the activation of Pol V and TLS [107]. These differences are likely the result of differences in experimental details. The complex of activated RecA with UmuD'₂C is termed the Pol V mutasome [127].

In addition to RecA, the β-processivity clamp and the γ clamp loader increase the processivity of Pol V by allowing the enzyme to remain bound to the DNA and by providing additional stability [145]. The β clamp increases processivity approximately three-fold to five-fold in one study [146] and about 100-fold in another study [107]. Clearly, the β clamp stimulates processivity of Pol V, but the extent of the increase in processivity depends on the specifics of the experimental system [19].

As both DinB and Pol V can be mutagenic depending on the cellular context and the nature of DNA lesions encountered, it was speculated that they could contribute to antibiotic resistance [64,65,66]. Indeed, in E. coli cells grown under stress conditions, DinB and to a lesser extent Pol V are responsible for base substitution mutations in the ampD gene that result in resistance to the β-lactam antibiotic ampicillin [147]. Moreover, DNA Pols II, IV, and V all contribute to E. coli resistance to the antibiotic ciprofloxacin, as well [148]. While the contribution of each polymerase was examined in bacterial cultures, in a mouse model of infection with pathogenic E. coli, LexA cleavage was required for resistance to ciprofloxacin or rifampicin [148]. Thus, SOS-regulated DNA polymerases and possibly other genes under LexA control contribute to the evolution of antibiotic resistance in bacteria both when they are grown in the free-living state and during infection within a mammalian host.

A recent study has shown that DinB of uropathogenic E. coli (UPEC) is required for virulence of UPEC strains in bladder infections in mice [149]. Deletion of dinB in all UPEC isolates tested results in a reduced ability to colonize host bladders. No reduction in virulence of the dinB-deletion mutant is observed in mice that have a reduced Toll-Like Receptor-4 (TLR4)-dependent inflammatory response, indicating that DinB is important in helping UPEC cope with the stresses produced by host inflammation. In contrast, deletion of umuDC does not reduce the virulence of these UPEC strains. Surprisingly, cells of the dinB-deletion mutant recovered from the host have a mutation frequency similar to that of the wild-type parent. This is in contrast to the phenotype observed in culture, in which the dinB-deletion mutant of the UPEC strain UTI89 has a reduced rate of spontaneous mutation when grown in either rich medium or in human urine. This study demonstrated a clear role for DinB in UPEC pathogenesis and virulence. However, DinB does not appear to influence the acquisition of mutations by UPEC in the host environment.

When identifying the SOS boxes of Mycobacterium tuberculosis and the induction levels of the genes proposed to be lexA regulated, Davis et al. found a gene annotated as dnaE2 (i.e., imuC) with a preceding M. tuberculosis LexA-binding SOS box [196]. On average, the Rv3370c (imuC) gene was up-regulated more than 10-fold following induction by MMC [196]. It has been shown that MMC induces imuC, recA, and lexA in strains containing a functional RecA, but that imuA'B and imuC are not induced by MMC in a recA-deletion mutant [162,197]. An imuC null mutant of M. tuberculosis has a reduced virulence relative to that of the wild type and is deficient in UV-induced mutagenesis [165]. This experiment showed that imuC-mediated mutagenesis is the sole source of UV-induced mutagenesis in M. tuberculosis, with a mutational spectrum that resembles that of a signature for translesion synthesis [165]. Strains with imuC reproducibly generated CC to TT mutations, consistent with bypass of a UV damage-induced pyrimidine dimer, whereas in strains without imuC, this mutation was not observed [165]. Overexpression of imuC in non-UV-treated cells does not increase the mutation frequency, suggesting that imuC requires additional subunits to function [165].

In M. tuberculosis, recA controls expression of imuA', imuB, and imuC [188,198]. MMC also induces the SOS response in M. smegmatis. Loss of imuA', imuB, or imuC individually or in combination results in the same level of hypersensitivity to MMC, suggesting that the products of these genes function as part of a single pathway for resistance to MMC [188]. M. tuberculosis imuC has three aspartic acids that correspond to the known active site acidic residues of C-family Pol III polymerase catalytic subunits. The M. smegmatis⁴⁴¹DID⁴⁴³ to ⁴⁴¹AIA⁴⁴³ mutation, which changes two of the three conserved active site residues, eliminates UV-induced mutagenesis and confers on M. smegmatis hypersensitivity to MMC, mimicking the imuC deletion phenotype [188]. These experiments established strong evidence that imuC is responsible for induced mutagenesis and survival under DNA-damage stress conditions via translesion synthesis [188].

An extensive study by Warner et al. elucidated many of the interactions between the imuA, imuB, and imuC gene products [188]. Whereas ImuC lacks a β-binding motif to interact with the processivity factor at the replication fork, ImuB does contain a β-clamp-binding motif. Only ImuB interacts with DnaE1 or with the β-processivity clamp. ImuA' and ImuC showed interactions with only ImuB and not with each other or with the β clamp [188]. The ImuB-β interaction can be disrupted by mutation of the β-binding motif or by truncation of the C-terminal end of ImuB including the β-binding motif. Truncations up to, but not including, the β-binding motif of ImuB did not disrupt the interactions with ImuA', ImuC, or β [188]. Truncation of the C-terminal 44 residues of ImuB disrupted the ImuA'-ImuB interaction [188]. Thus, each of the three proteins expressed from the imuABC operon interact in a pairwise fashion with ImuB, which leaves open the possibility of ternary complex formation with ImuB occupying a central position in such a ‘mutasome’.

In Deinococcus deserti, which was isolated in the Sahara desert [191,192], the genome contains a mutagenesis cassette in the form of lexA-(imuB-like protein)-imuC, as a single transcriptional unit [193]. This operon has unusual characteristics compared to other imuABC cassettes. The imuB-like gene is different enough from other imuBs that the gene in D. deserti was termed imuY, to recognize its homology to Y-family DNA polymerases, rather than imuB [193]. In addition there is a hypothetical protein of 243 base pairs between imuY and imuC [193]. D. deserti has three recA genes encoding two RecA products: recA_C, encoding chromosomal RecA_C, and recA_P1 and recA_P3, which both encode the same plasmid-derived RecA_P product [193]. The mutagenesis cassette was induced by RecA_C but not by RecA_P [193]. Interestingly, while transcriptional regulation of the cassette was dependent on RecA_C, recA_C mutants did not show a loss in UV or gamma radiation survival [193]. Deletion of imuY, imuC, or both imuY-imuC showed the same 10-fold decrease in UV-induced mutagenesis, and deletion of imuY could be complemented by a plasmid carrying the imuY gene [193]. To explain the lack of decreased survival upon imuY or imuC deletion, Dulermo et al. note that the conditions in the native environment of the Sahara desert starkly differ from the mild conditions in the laboratory; under the combined stress of dessication, starvation, and other environmental conditions, D. deserti may depend more heavily on imuY and imuC for survival [193].

In Deinococcus ficus, a lexA-imuB-imuC gene cassette and a dinB2 gene are carried on an accessory plasmid [199]. Disruption of either imuB or imuC showed equal loss of survival and loss of mutagenesis following UV exposure, suggesting the same pathway of action for both survival and induced mutagenesis [199]. Deinoccocus ficus naturally possesses keratinolytic activity to break down feathers, which could be used in agricultural and industrial applications [199]. Using the inherent mutator properties of ImuC, UV exposure was utilized as a mutagen to create improved keratinolytic activity [199]. Induced mutagenesis by UV light led to at least one mutant strain with a two-fold higher activity [199]. Zeng et al. suggest that increased keratinolytic activity after UV exposure could come from imuBC-dependent induced mutations, but note the possibility of the two dinB genes to contribute to this process [199]. In addition, D. grandis contains a putative imuB showing high similarity to the D. ficus imuB [199].

The two most thoroughly studied Deinoccocus species, D. radiodurans and D. geothermalis, do not contain mutagenic or translesion synthesis polymerases [193]. It has been hypothesized that it is advantageous for Deinoccocus species not to have error-prone translesion synthesis or mutagenic polymerases in order for them to accomplish their striking feats of DNA repair [200]. The discovery of a mutagenic cassette in D. deserti and D. ficus provide an interesting example of how different species within the same genus can develop different survival strategies [193,199].

In Caulobacter crescentus, imuABC genes are responsible for UV- and MMC-induced mutagenesis [186] and are strongly repressed by LexA, with increased expression in a lexA mutant strain by 15-fold over that of the wild type [201]. While deletion of a single gene in the mutagenic cassette results in slight sensitivity to UV exposure and abolishes induced mutagenesis, a double imuB-imuC deletion mutant has no further increase in sensitivity or reduction of mutagenesis, strengthening evidence that these genes function in the same pathway in this bacterium [186].

In wild type C. crescentus, UV-induced mutagenesis results in a mixture of G:C to A:T transitions, G:C transversions, A:T to G:C or C:G mutations, and tandem substitutions [186]. Under conditions with either imuB or imuC absent, the mutation spectrum drastically shifts to become dominated by G:C to A:T transitions, with the remainder of mutations being A:T to G:C transitions [186]. The dependence on imuB and imuC for G:C to C:G transversions, A:T to C:G transitions, and tandem substitutions presents a unique mutation spectrum compared to that of E. coli umuDC [56,202,203] or M. tuberculosis imuA'BC [165,186].

In Streptomyces coelicolor, the dinB2 and imuC genes overlap by 4 bp, an organization found in most Streptomyces and some other bacteria such as Sinorhizobium [204]. In S. coelicolor, imuC deletion strains have no defect in end patching of telomeres, conjugal transfer, UV survival, or UV‑induced mutagenesis, even though imuC is induced by UV exposure and MMC. The authors argue that imuC is a rapidly evolving gene and that it may be still developing a new/optimal function [204]. In Streptococcus uberis, a mutagenesis cassette has been reported that is induced by UV light and that induces mutagenesis after UV exposure; the genes composing this cassette seem to be present throughout Streptococcacaea [205].

P. putida contains an imuABC operon [185,206]; it has been shown that during stationary phase mutagenesis imuC reduces the frequency of base substitution mutations, whereas imuB increases base substitution mutations as well as 1-bp deletions [185]. When imuC was deleted, the number of base substitution mutations increased with no change in 1-bp deletions [185]. When imuB was deleted, 1-bp frameshifts were decreased with no change seen in base substitution mutations [185].

Pol I, an A-family DNA polymerase, can act as a translesion synthesis polymerase [155]. In a P. putida Pol I-deficient background, the spontaneous mutation frequency was similar in the presence or absence of imuC [206]. However, after UV exposure, A to T and A to G mutations decreased in an imuC⁻ strain [206]. Frequency of UV-induced mutations increased two-fold in imuC⁻ compared to wild type, but not in a Pol I-deficient background [206]. It was concluded that in an unstressed Pol I deficient background, ImuC does not meaningfully contribute to DNA synthesis, but in the presence of DNA damage ImuC becomes involved in DNA synthesis [206].

In P. aeruginosa in response to ciprofloxacin, imuC and dnaE1 are upregulated two- and six-fold, respectively [207]. Also in this species, imuC has been shown to be responsible for induced mutagenesis [159]. It must be noted that while P. aeruginosa imuC and P. putida imuC share 73% identity, they have phenotypically opposite effects [185], where imuC (also referred to as polC in P. aeruginosa) is an anti-mutator in P. putida [185,206] and a mutator in P. aeruginosa [159].