SSB as an organizer/mobilizer of genome maintenance complexes

Abstract

When duplex DNA is altered in almost any way (replicated, recombined, or repaired), single strands of DNA are usually intermediates, and single-stranded DNA binding (SSB) proteins are present. These proteins have often been described as inert, protective DNA coatings. Continuing research is demonstrating a far more complex role of SSB that includes the organization and/or mobilization of all aspects of DNA metabolism. Escherichia coli SSB is now known to interact with at least 14 other proteins that include key components of the elaborate systems involved in every aspect of DNA metabolism. Most, if not all, of these interactions are mediated by the amphipathic C-terminus of SSB. In this review, we summarize the extent of the eubacterial SSB interaction network, describe the energetics of interactions with SSB, and highlight the roles of SSB in the process of recombination. Similar themes to those highlighted in this review are evident in all biological systems.

The genomes of all cellular organisms are organized as double-stranded (ds) DNA, with the information content, in the form of nucleotide bases, sequestered in the interior of the protective double helix ^{1; 2}. To provide DNA replication, recombination, and repair machinery access to genomic information, dsDNA must be unwound to form single-stranded (ss) intermediates. Such processes are obligatory, but they are not without risks. ssDNA is prone to chemical and nucleolytic attacks that can produce breaks or lesions that are difficult to repair and can self-associate to create impediments to genome maintenance ^3-9. To help preserve ssDNA intermediates, cells have evolved a specialized class of ssDNA-binding (SSB) proteins that associate with ssDNA with high affinity and in a sequence-independent manner ^10-21. SSB binding protects ssDNA from degradation ^{10; 22-25}, and, more globally, defines the nucleoprotein substrates upon which DNA replication, recombination, repair, and replication restart processes must act. With these central roles in genome maintenance, it is no surprise that SSB proteins are conserved throughout all kingdoms of life and are indispensable for cell survival ^26-28.

Beyond their eponymous roles in DNA binding, SSB proteins have a second, less well-appreciated role in which they physically associate with a broad array of cellular genome maintenance proteins. SSB interaction with heterologous proteins targets enzymes to active genome maintenance sites and, in many cases, stimulates the biochemical activities of SSB's partner proteins. In this review, we focus on several aspects of eubacterial SSB interactions with heterologous proteins. First, we summarize the extent of SSB's interaction network by describing what is known about its many partner proteins. In this section, we focus primarily on E. coli SSB as a workhorse for understanding SSB structure and function. Second, we consider the thermodynamc mechanisms underlying the binding of heterologous proteins to SSB and the ways in which ssDNA binding by SSB influences its interaction with other proteins. These are important features for defining the specificity of binding to SSB. Finally, we summarize how heterologous proteins have adapted to carry out DNA recombination reactions in the cellular environment where mediator proteins regulate recombinase loading onto ssDNA/SSB nucleoprotein substrates. Although this review focuses on eubacterial SSBs, it is clear that eukaryotic SSB proteins have similarly evolved to interact with numerous genome maintenance enzymes as has been described in excellent reviews ^{26; 29}. Thus, throughout the kingdoms of life ssDNA/SSB complexes are not merely inert particles but are instead dynamic centers that play a key role in choreographing the processes surrounding DNA replication, recombination, and repair.

SSB PROTEIN OVERVIEW

Eubacterial SSB proteins are linked by two common structural features. The first is the use of oligonucleotide/oligosaccharide-binding (OB) domains to bind ssDNA through a combination of electrostatic and base-stacking interactions with the phosphodiester backbone and nucleotide bases, respectively ^30-37. The second is SSB oligomerization that brings together four DNA-binding OB folds in the protein's active form ^{34; 38-42}. E. coli SSB, which encodes a single OB fold in each monomer and functions as a tetramer, has served as the prototypical SSB protein for decades ^{12; 19; 34; 35; 39; 43}. Rare exceptions to the E. coli SSB-type arrangement exist, including the SSBs from the Deinococcus-thermus genera, which contain two OB folds per monomer and assemble as homodimers ^44-47 (Figure 1). SSB proteins in non-eubacterial systems have distinct quaternary structures, including the heterotrimeric eukaryotic Replication Protein A (RPA) ²⁹, which acts as a heterotrimer, and several bacteriophage and viral SSB proteins that function as monomers (T4 gp32) ⁴¹ and dimers (T7 gene 2.5) ^{41; 42}.

Figure 1.

(A) Schematic representation of SSB. The E. coli SSB OB domain (residues 1−115) is shown as a box with its structurally dynamic C-terminal tail (residues 116−177) as a line. The sequence of the E. coli SSB-Ct element is displayed with its conservation across 280 eubacterial species represented as a logo⁴²⁴ in which the height of the residue relates to its frequency at the given position. Logo residues are colored to indicate the hydrophobic (red), electronegative (blue), polar (black), or electropositive (green) nature of their side chains. (B) Ribbon diagram of the proposed structures of the E. coli (SSB)₃₅ (left) and (SSB)₆₅ (right) ssDNA binding models ³⁴. Each monomer in the tetramer is separately colored and its C-terminus is shown schematically as a dashed line. ssDNA is shown as a red tube. (C) Ribbon diagram of the crystal structure of D. radiodurans SSB ⁴⁴. OB folds are colored as for E. coli SSB, but with two OB folds in each monomer of the dimer. C-terminal tails are displayed as dots.

Eubacterial SSB proteins can bind ssDNA in a highly cooperative manner, which leads to clustering of SSB protein tracts to form protein filaments on long ssDNA ^{12; 13; 15}. However, E. coli SSB only binds ssDNA with high cooperativity in one of its binding modes (see below). The eukaryotic RPA SSB protein also does not display significant cooperativity in its binding to ssDNA ^{29; 48}. Hence the general role of this cooperativity remains unclear. Due to the presence of four ssDNA binding sites, the E. coli SSB tetramer can bind to long stretches of ssDNA in multiple binding modes differing in the number of OB-folds that interact with the ssDNA ^49-51. The primary ssDNA binding modes are denoted as the (SSB)₆₅, (SSB)₅₆ and (SSB)₃₅ modes, where the subscript reflects the average number of nucleotide residues occluded by each tetramer in the complex. In the (SSB)₆₅ mode, ∼65-nucleotides of ssDNA wrap around and interact with all four subunits of the tetramer, whereas in the (SSB)₃₅ mode, ∼35-nucleotides interact with an average of only two subunits (Figure 1). The (SSB)₆₅ binding mode is a limited cooperativity mode in which SSB shows little tendency to form protein clusters along ssDNA; the (SSB)₃₅ binding mode, on the other hand, is a high, unlimited cooperativity mode in which SSB can form long protein clusters along ssDNA ^{14; 15; 50}. The relative stabilities of the different SSB-DNA binding modes is influenced by monovalent salt concentration, Mg²⁺ concentration as well as the polyamines, spermine and spermidine ^{49; 51; 52}, with the (SSB)₆₅ mode being favored at monovalent salt concentrations above 200 mM. The (SSB)₃₅ mode is also favored at high SSB to ssDNA ratios ^{15; 50}. Whether these different SSB binding modes have specific functions in vivo is not clear, although it has been proposed that they may be used selectively in different processes in the cell. For example, under conditions where RecA protein stimulates DNA strand exchange in vitro, SSB binds primarily in the low cooperative, fully wrapped (SSB)₆₅ binding mode ⁵³. The (SSB)₃₅ mode, which binds with high nearest neighbor cooperativity, has been proposed to function in DNA replication ³⁷. The yeast RPA also undergoes a salt-dependent transition from a lower to a higher site size ssDNA binding mode ⁴⁸.

Eubacterial SSB proteins have been shown to bind more than a dozen different proteins (Figure 2). For all cases tested thus far, complex formation requires the C-terminal region of SSB (SSBCt), suggesting a conserved mechanism by which proteins can recognize and bind to SSB. Indeed, the C-terminus of eubacterial SSB proteins, which ends in an Asp-Phe-Asp-Asp-Asp-Ile-Pro-Phe sequence in E. coli SSB, is very well conserved ⁵⁴ (Figure 1). Owing to its high density of Asp residues, this region is often referred to as SSB's “acidic tail,” but the hydrophobic tripeptide that forms the extreme C-terminus is well conserved among SSBs and is critical for protein interactions (see below). Thus, the C-terminus of SSB should more accurately be considered as an amphipathic sequence element. In contrast to the well-folded OB domains, the C-termini of bacterial and bacteriophage SSB proteins appear to be structurally dynamic as they are readily removed by proteolysis ^{20; 55-57}. Further, the C-terminus is not visible in the crystal structure of full-length E. coli SSB ³⁵. Proteolysis of the SSB-Ct is stimulated by ssDNA binding, and deletion of the SSB-Ct influences the relative stabilities of the (SSB)₃₅ and (SSB)₆₅ binding modes ⁵⁸. Recent studies have shown that the C-terminal tail of the phage T7 gene 2.5 SSB protein can compete with ssDNA for binding to the OB-fold ⁵⁹.

Figure 2.

(A) Schematic representation of SSB interactions. SSB proteins (yellow) are depicted at tetramers with C-termini (Ct) interacting with ovals symbolizing proteins involved the major genome maintenance pathways of DNA replication (teal), recombination (purple), replication restart (orange), and repair (green). (B) List of proteins that are known to physically interact with SSB with their requirement for the SSB-Ct for interaction given. Citations for the interactions are given in the text. Highlighting colors indicate the major genome maintenance activities of the proteins (color coding as in (A)) and the sections in which each is described (except for Topoisomerase III, which is described in the recombination section with RecQ). (C) Binding site for the E. coli SSB C-terminus on Exonuclease I ⁵⁴. Surface representation of Exonucelase I is stained in blue, red, and white to highlight positive, negative, and hydrophobic electrostatic features, respectively. The final four residues of the SSB C-terminus are shown in ball and stick form. Features shown to be critical for SSB binding by Exonuclease I are labeled.

Mutations within the SSB C-terminus have detrimental effects on E. coli cell survival. One well-studied mutation (ssb113) changes the penultimate Pro of the E. coli SSB C-terminus to a Ser ^{60; 61}. This mutation confers temperature-sensitive lethality by producing an SSB variant that is competent to bind DNA but can no longer support DNA replication at non-permissive temperatures ⁶⁰ and is hypersensitive to DNA damage even under permissive conditions ^{23; 62-65}. A second set of mutations that alter the C-terminal-most E. coli SSB residue, Phe177, similarly impair viability ^66-68. In both cases, mutations that alter the SSB-Ct sequence produce proteins that fail to interact properly with at least some of SSB's binding partners. Deletion of 10 amino acids from the C-terminus of SSB renders the cells expressing that protein inviable ⁶⁹. These dramatic phenotypes have greatly illuminated the importance of SSB's interactions with heterologous proteins to cellular genome maintenance pathways.

I. INTERACTION OF SSB WITH DNA METABOLISM PROTEINS IN EUBACTERIA

In this section, we gauge the extent of SSB's interaction network by reviewing known eubacterial SSB-interacting proteins, with a particular emphasis on E. coli proteins. Members of the interaction network are grouped through their involvement in DNA replication, recombination, replication restart, repair, or as “other SSB-binding proteins”, but it should be recognized that in many cases these proteins function in several of the listed areas. Rather than provide an extensive background on each protein, we have limited discussion of each to focus primarily on what is known about the protein interactions with SSB and the biochemical and cellular consequences of such interactions.

DNA REPLICATION

DNA Polymerase III – χ subunit. (χ binds SSB via the SSB-Ct; disruption of this binding is lethal to cells)

The DNA polymerase III holoenzyme (Pol III HE) is the multi-subunit replicative DNA polymerase in E. coli ^70-77. Within the Pol III HE, the γ clamp loader (comprised of γ, δ, δ', χ and ψ subunits) forms a subcomplex that loads the β processivity factor onto DNA and helps tie the holoenzyme together through a network of protein-protein interactions ^{74; 78; 79}. Although χ and ψ are not required for clamp loading onto DNA ⁸⁰, they do form a complex that facilitates the assembly of the clamp loader itself ⁸¹. Moreover, χψ binds SSB directly via the χ subunit ^{67; 68; 82; 83}, which allows the Pol III HE to clear potentially inhibitory SSB proteins from lagging-strand DNA during DNA replication ^{62; 67; 82}. SSB113 and an SSB truncation variant lacking the C-terminal 26 amino acids of the protein fail to bind χ, establishing the SSB C-terminus as the site of the protein-protein interaction ^{67; 68; 83}. The χ/SSB interaction plays a crucial role in Pol III HE function by driving detachment of primase from RNA primers, which stimulates primer hand-off to the Pol III HE ⁶⁸. This activity appears to be essential for cell viability in E. coli and accounts for the conditional-lethal ssb113 phenotype ^{67; 84}.

Primase. (Primase binds SSB, possibly at the SSB-Ct)

E. coli Pol III HE cannot initiate DNA synthesis but instead extends preformed nucleic acid primers. RNA primers in bacteria are generated by a specialized RNA polymerase called primase (the product of the dnaG gene) ^{70; 77; 85-88}. Primase constitutes the lone priming protein in E. coli ^{86; 89; 90}, functioning in both leading- and lagging-strand synthesis in oriC-dependent replication and in replication restart processes ^{70; 87; 89; 91; 92}.

E. coli primase interacts with the replicative helicase, DnaB, through its C-terminal protein interaction domain ^92-101. DnaB/primase complex formation recruits primase to the replication fork, helps coordinate leading and lagging strand synthesis by Pol III by regulating Okazaki fragment synthesis, and initiates bi-directional replication at oriC ^{92; 96; 100; 102-105}.

In addition to its association with DnaB, primase also interacts with SSB. Primase/SSB interaction strengthens the association between primase and the RNA primers it synthesizes ⁶⁸ and is disrupted by the χ subunit, forming the basis of an involved handoff mechanism in which primase dissociates from the RNA-DNA duplex, allowing clamp-loader assembly to occur ⁶⁸. The domains of primase and SSB that are required for complex assembly have not been identified, but apparent competition between primase and χ for SSB suggests that primase might bind to the C-terminus of SSB.

DNA RECOMBINATION

RecQ DNA helicase. (RecQ binds SSB via the SSB-Ct; interaction stimulates RecQ helicase activity)

E. coli RecQ functions as the DNA helicase in the RecF-recombination pathway ^106-110, which helps repair gapped and UV-damaged DNA and can repair dsDNA breaks in recBC-deficient cells ^111-119. Notably, many RecF-recombination pathway proteins interact with SSB, as will be described below and in section III. E. coli RecQ also plays roles in the SOS DNA damage response ¹²⁰ and in the suppression of illegitimate recombination ¹²¹. RecQ promotes cell death in ruv recA(ts) uvrD E. coli cells, apparently by driving the accumulation of excessive recombination intermediates ¹²². RecQ-mediated recombination initiation ¹²³, plasmid DNA catenation and supercoiling reactions ^{124; 125}, and converging replication fork resolution ¹²⁶ have been reconstituted in vitro. The latter two activities required the addition of Topoisomerase III, a type-Ia topoisomerase that appears to have coordinated activities with RecQ proteins in eukaryotes and bacteria ^127-131. Interestingly each of RecQ's reconstituted reactions is stimulated by or requires SSB to proceed.

SSB has been shown to physically associate with RecQ and to stimulate RecQ DNA helicase activity ^132-134. SSB interaction with RecQ is mediated by the 9 C-terminal-most residues of the SSB-Ct ¹³². SSB increases the efficiency of RecQ-mediated unwinding of a 71-basepair duplex formed between an oligonucleotide and M13 circular ssDNA; gp32 from bacteriophage T4 also stimulates this unwinding ¹³³. SSB stimulates unwinding of a 30-basepair duplex DNA with a 70-base single-stranded 3’ overhang; in this case, gp32 and RPA inhibit unwinding, as does an E. coli SSB variant that lacks its SSB-Ct ¹³². These studies indicate that in some contexts interaction between RecQ and SSB is required for SSB stimulation. Moreover, E. coli Topoisomerase III has recently been shown to interact with SSB ^{126; 135}, which could indicate that it acts with RecQ in a complex that is nucleated by SSB. Deletion analysis has shown that the RecQ winged-helix subdomain is the site of interaction with SSB ¹³². This subdomain appears to be utilized as a platform for protein interactions in eukaryotic RecQ helicases as well ^136-138.

RecJ exonuclease. (RecJ DNA binding and exonuclease activities are stimulated by SSB in vitro)

E. coli RecJ is an exonuclease that degrades ssDNA in a 5’ to 3’ direction ¹³⁹. It functions as a member of the RecF-recombination pathway ^{10; 11; 140-142} and, in conjunction with RecQ, RecJ acts at stalled replication forks to degrade nascent lagging strand DNA prior to resumption of replication ^{111; 112; 115; 143}. Additional roles for RecJ in base excision repair ^{144; 145} and in the excision step of methyl-directed mismatch repair ^{146; 147} have been reported.

In vitro, RecJ binds to the 5’ end of ssDNA and requires a 5’ overhang for cleavage ¹⁴⁸. Unlike most nucleases, RecJ DNA binding and degradation are stimulated by SSB ¹⁴⁸. Because T4 gp32 does not provide similar enhancement and RecJ is able to supershift SSB-bound DNA, this enhancement is likely to be due to a specific physical interaction between E. coli SSB and RecJ ¹⁴⁸. Strengthening this view, RecJ has been observed in complex with SSB in affinity purification studies ^{132; 135}. The domains of RecJ and SSB that mediate their interaction have not been identified.

RecG DNA helicase. (RecG binds SSB (likely via the SSB-Ct); RecG DNA binding and ATPase activities are stimulated by SSB in vitro)

RecG is a monomeric DNA helicase ^149-152 that binds forked DNA structures ¹⁵³ and promotes regression of stalled replication forks ^154-156. RecG has been implicated in a multitude of genome maintenance activities, including ssDNA gap repair and recombinational repair of dsDNA breaks ^{10; 157}, chromosome segregation ¹⁵⁸, stabilization of stalled replication forks ¹⁵⁹, and resolution of Holliday junctions ^{11; 160-163}. RecG binds and remodels numerous nucleic acid structures, including 3-way and 4-way DNA junctions^{160; 161}, D-loops¹⁶⁴, and R-loops^{149; 165}. RecG can promote rapid, ATP-dependent regression of replication forks in vitro with low processivity ¹⁶⁶ and can inhibit RecA-mediated strand exchange under conditions that are suboptimal for RecA ^166-168.

SSB stabilizes E. coli RecG binding to negatively supercoiled DNA, the substrate upon which its ATPase activity is most highly stimulated ¹⁶⁹. Maximal ATP hydrolysis also greatly increases when SSB is included in RecG reactions ¹⁶⁹. In B. subtilis, RecG colocalizes with SSB at foci that are thought to be stalled replication forks ¹³⁴. This colocalization is ablated in cells where SSB lacks its 35 C-terminal most amino acids, suggesting that RecG binds the SSB-Ct and requires SSB to associate with the replisome.

RecO. (RecO binds SSB (likely via SSB-Ct), which stimulates RecOR RecA loading)

RecO is a mediator protein in the RecF recombination pathway ^{106; 170; 171}. Strains harboring mutations in recO exhibit numerous defects in DNA replication, recombination, and repair ^{107; 116; 172-180} and in the SOS response ^{181; 182}, but recO mutations can also suppress illegitimate recombination caused by an excess of RecET ^{183; 184}. Disruption of recO confers resistance to thymineless death ¹⁸⁵ and sensitivity to UV irradiation ^{119; 171; 186-188}. UV sensitivity in strains with mutations in recF ¹⁰⁶ and recR ¹⁸⁹ (also RecF-pathway genes) can be suppressed by over-expression of RecO and RecR ¹⁹⁰ as well as by certain RecA mutants ¹⁹¹, suggesting that these proteins function in a common pathway. RecO binds ssDNA and dsDNA and possesses a DNA-annealing activity ^{192; 193}. This annealing activity is stimulated by SSB and inhibited by RecR ¹⁹⁴. Together with RecF and RecR, RecO functions as a modulator of RecA activity ^{10; 195-199}. RecO and RecR facilitate RecA loading onto SSB-coated ssDNA ^200-202. RecO and RecR produce an apparent stabilization of RecA filaments that is likely related to the continued association of RecR with the RecA filament after it forms ^{201; 202}. The putative stabilization may reflect an actual suppression of RecA filament disassembly or an enhanced re-loading of any RecA protein that dissociates. Roles for RecO in DNA recombination are described in greater detail in section III of this review.

SSB directly binds RecO ²⁰⁰ and limits the formation of RecOR complexes on ssDNA ¹⁹⁹. The ability of RecOR to load RecA is greatly reduced when RPA or an SSB variant lacking the C-terminal eight amino acids are substituted for wild-type protein ¹⁹⁹, suggesting that direct physical interaction between SSB and RecO is necessary for maximal efficiency of the RecOR-stimulated reaction. A similar effect has been observed in Thermus thermophilus: RecO-assisted loading of RecA is achieved by means of the direct protein-protein interaction between RecO and SSB ²⁰³. In this case, RecO binds both SSB and ssDNA and in doing so displaces SSB from the DNA.

DNA REPLICATION RESTART

PriA DNA helicase. (PriA binds SSB via the SSB-Ct; assocaition with SSB stimulates PriA helicase activity)

The primosome was originally identified as a collection of E. coli proteins required for the conversion of the phage ϕX174 genome from its single-stranded form to its double-stranded (replicative) form ²⁰⁴. In total, the primosome consists of seven proteins: DnaB, DnaC, DnaG, PriA, PriB, PriC, and DnaT ^{76; 205}. Of these, DnaB, DnaC, and DnaG (primase) are necessary for initiation of replication of the E. coli genome at oriC ⁷⁰, whereas the remaining proteins drive origin-independent initiation (replication restart) at the sites of collapsed replication forks ^206-208. PriA initiates assembly of the PriA/PriB/DnaT primosome by binding DNA structures that result from replication failure and attracting PriB and DnaT ^{164; 207; 209-218}. PriA also appears to be an important anti-recombinase by binding stalled replication forks and preventing RecA binding and activity ²¹⁹. In B. subtilis cells, PriA continuously colocalizes with the replication machinery ¹³⁴. E. coli cells with deleterious priA mutations harbor numerous defects including UV sensitivity and defects in DNA repair, the SOS response, and chromosomal segregation ^{210; 217; 220-226}.

PriA is a helicase that unwinds DNA with a 3’-to-5’ polarity ^227-229. PriA can unwind DNA duplexes of up to 40 bp on its own but requires SSB to process longer duplexes ²²⁸. PriA can bind SSB-coated DNA ²³⁰ and can displace SSB from DNA ²³¹. Its helicase activity is stimulated by SSB on branched DNA substrates resembling replication fork lagging strands but is inhibited by SSB on partial duplex DNA ^{213; 232}. SSB is able to weakly stimulate PriA-mediated unwinding of forked substrates that have no exposed ssDNA, suggesting that part of the enhancement effect is due to SSB sequestering DNA that PriA has already unwound ²³².

SSB stimulation of PriA appears to be a consequence of the two proteins physically interacting via the SSB-Ct ²³². Neither archaeal nor viral SSB are capable of stimulating PriA activity, and E. coli SSB variants with the ssb113 point mutation or the C-terminal 10 amino acids truncated fail to stimulate PriA activity ^{60; 232}.

PriB. (PriB binds SSB in an undefined manner)

PriB is the second member of the PriA-primosome to assemble ^{230; 233}. It acts to stabilize PriA binding to ssDNA and assists in primosome assembly by facilitating PriA binding to DnaT ²³³. Mutations in priB do not exhibit the UV sensitivity, recombination deficiency, or constitutive activation of the SOS response seen in priA mutants, but contribute to a very slow growth phenotype when combined with priC mutations ²³⁴ and can influence plasmid copy number ²³⁵. PriB is a 12 kDa protein that exists as a dimer in solution ^{236; 237} and shares extensive sequence and structural homology with SSB ^{236; 238; 239}. Indeed, the similarities between SSB and PriB are of such a degree that it has been hypothesized that PriB arose due to a duplication of the ssb gene ²⁴⁰. Consistent with this hypothesis, PriB binds to ssDNA ^{236; 237}, although in an unexpected fashion: while SSB binding of ssDNA is reliant upon base-stacking with the side chains of aromatic residues as well as electrostatic interactions ^{34; 241}, PriB appears to utilize a mainly charge-based interaction made possible by multiple lysine residues ²¹¹, and is still able to strongly bind an oligonucleotide even when its lone surface-exposed tryptophan is mutated ²⁴².

PriB is also able to bind SSB-coated DNA, suggesting a protein-protein interaction ²³⁷. Like SSB, PriB stimulates PriA helicase activity on forked DNA substrates and this stimulation is further increased when SSB is present ²⁴³. Consistent with previous observations, SSB inhibits PriA helicase activity on partial duplex DNA even when PriB is present ^{232; 243}. PriB physically interacts with the helicase domain of PriA and bridges a ternary complex between PriA and DnaT ²¹¹. What role an interaction between PriB and SSB might play remains unclear.

DNA REPAIR

Exonuclease I. (Exonuclease I binds SSB via the SSB-Ct; interaction stimulates exonuclease activity; the structure of Exonuclease I in complex with SSB-Ct is known)

Exonuclease I (ExoI) is a DnaQ-family exonuclease that processively degrades ssDNA in a 3’-to-5’ direction ^244-249. The gene that encodes ExoI in E. coli is known both as xonA and sbcB owing to the two distinct phenotypes that stem from different ExoI variants ^249-251. sbcB (suppressors of recBC) mutations restore cellular recombination activity and reduce sensitivity to DNA damage from UV light in recBC- cells ^{249; 251; 252}. The strong reduction of ExoI activity in these cells appears to allow 3’ ssDNA ends to remain intact and available as RecF pathway recombination substrates ^{106; 108}. Ordinarily, ExoI would degrade the ssDNA ends, leaving DNA structures that cannot be efficiently recombined ^{106; 108; 245; 249}. xonA mutants also acquire UV-resistance but are deficient in recombination activity compared to recBC- sbcB E. coli cells ^{249; 250; 253}. The nucleolytic activity of ExoI is important for degradation of incorrectly base-paired DNA in mismatch repair ^{146; 147; 254}. SSB also plays a central role in mismatch repair, which indicates that ExoI has adapted to act on SSB/ssDNA complexes ²⁵⁴. Like RecJ, ExoI activity is stimulated by the presence of SSB. This distinguishes both RecJ and ExoI from several other nucleases that are inhibited by SSB ²⁵. ExoI also plays roles in the preservation of genome integrity by acting as a deoxyribophosphodiesterase at apurinic and apyrimidinic sites ^{255; 256} and by suppressing frameshift mutations ^{257; 258}. The former activity is stimulated by SSB in vitro ²⁵⁹.

Several experiments have demonstrated a direct physical interaction between E. coli ExoI and SSB, and have shown that this interaction is mediated by the SSB-Ct ^{10; 16; 25; 54; 66; 259}. This interaction is relatively strong (K_a∼ 7.1 × 10⁶ M^{−1 54}) and, consistent with an ExoI/SSB-Ct interaction, SSB113 and C-terminal deletion variants fail to interact with ExoI ^{54; 66}. Recently, the structure of ExoI bound to a peptide composed of the nine C-terminal residues of SSB was determined ⁵⁴ (Figure 2C). In this structure, the C-terminal-most phenylalanine of the SSB peptide packs into a hydrophobic pocket that is flanked by a basic surface that is thought to contact acidic SSB-Ct residues that lie N-terminal to the phenylalanine ⁵⁴. Significantly, mutations that alter residues on the surface of ExoI that disrupt SSB binding also abolish SSB stimulation of ExoI activity. Similar results are seen when the SSB tail is altered or removed, suggesting that SSB acts to recruit ExoI to ssDNA ⁵⁴.

Uracil DNA Glycosylase. (Uracil DNA glycosylase binds SSB via the SSB-Ct; interaction impacts uracil excision activity in a DNA-dependent manner)

Uracil DNA glycosylase (UDG) catalyzes the first step in a base excision repair pathway by creating an abasic site through removal of uracil from DNA ^{145; 260-265}. The DNA harboring the abasic site is then degraded and resynthesized ^{144; 145; 262}. UDG has also been implicated as a generator of dsDNA breaks when two uracils, located on opposite strands of a DNA duplex and separated by seven or fewer bases, are recognized and targeted for repair in rapid succession ²⁶⁶.

SSB affects uracil excision activity by E. coli UDG in different ways depending on substrate structure. On a ssDNA substrate that lacks secondary structure, SSB decreases excision up to three-fold, but in a ssDNA molecule containing a tetraloop, SSB enhances UDG activity 7- to 140-fold depending on the position of the uracil ²⁶⁷. SSB proteins from other bacterial species are also able to stimulate UDG uracil excision in a species-specific manner, but activity is decreased when a UDG from any of the tested species is mixed with a heterologous SSB ^{268; 269}. Surface plasmon resonance experiments suggested that these changes in activity depend upon a physical interaction between UDG and SSB ²⁶⁸.

Handa and colleagues investigated UDG/SSB interactions in an exhaustive study ²⁷⁰. Interaction between E. coli and M. tuberculosis UDG and SSB proteins was demonstrated by yeast two-hybrid screens. In vitro, direct interaction between the proteins was demonstrated by far Western blot analysis and interaction on ssDNA was shown using electrophoretic mobility supershift assay ²⁷⁰. Surface plasmon resonance experiments yielded an association constant (K_a) for E. coli UDG and SSB of 5.9 × 10⁶ M⁻¹, which is similar to the K_a for ExoI/SSB complexes ^{54; 66; 270}.

E. coli UDG can bind a chimeric SSB consisting of the C-terminal 47 amino acids from E. coli SSB appended to the N-terminal 130 amino acids of M. tuberculosis SSB. It binds this chimera, designated MtuEcoSSB, with lower affinity than wild type, but it does not bind the reciprocal chimera at all (EcoMtuSSB, in which the C-terminus of M. tuberculosis SSB is present). MtuEcoSSB is capable of stimulating E. coli UDG activity, but EcoMtuSSB has an inhibitory effect ²⁷⁰. Therefore, the C-terminus of SSB appears to be required for UDG binding and the attendant biochemical stimulation.

DNA Polymerase II. (DNA Polymerase II binds SSB; interaction enhances processivity and replication beyond abasic sites)

DNA polymerase II (Pol II) is a DNA repair polymerase ^271-274. It is induced early in the SOS response up to 8-fold over basal levels ^272-280. Pol II participates in repair of and synthesis across various lesions ^281-286 including thymine dimers ²⁸⁷ and as such is especially important in replication and repair of UV-damaged DNA ^{179; 287; 288}. Pol II plays a role in maintaining the fidelity of replication ^{281; 283; 289; 290}, which is severely compromised when its proofreading exonuclease activity is removed ^{289; 291; 292}.

Pol II was the first SSB interacting partner to be identified; indeed, in the manuscript announcing the isolation of SSB, Sigal and colleagues noted that the “DNA unwinding protein” that they purified to homogeneity had a strong stimulatory effect on DNA synthesis by Pol II ^{13; 18}. Soon after, SSB was found to facilitate binding of Pol II to ssDNA, to stimulate the Pol II-associated nuclease activity, and to form a complex with Pol II in the absence of nucleic acid ^{19; 24}.

When functioning alone, Pol II is poorly processive, synthesizing approximately five nucleotides before dissociating from the template strand; however, Pol III HE processivity factors (the β subunit and the clamp-loader complex) increase Pol II processivity to about 1600 nucleotides in an SSB-dependent manner ^{293; 294}. The processivity factors and SSB are required for by-pass of abasic sites ²⁹³. β may play a role in determining when Pol II is activated in SOS-induced cells ²⁹⁵, as deactivation of Pol II in a strain with a mutant β restored viability to cells that would otherwise have been inviable ²⁹⁶.

DNA Polymerase V. (DNA Polymerase V binds SSB via the SSB-Ct; interaction with SSB is critical for translesion synthesis activity in vitro)

DNA polymerase V (Pol V) carries out translesion synthesis on damaged DNA ^{297; 298}. Pol V is encoded by the genes umuD and umuC and forms when two molecules of UmuD undergo RecA-mediated cleavage to their active UmuD’ form ^299-301 and assemble with one molecule of UmuC ^302-305. In vitro, Pol V translesion synthesis activity requires RecA and SSB ^{303; 304; 306-310}. Increasing concentrations of SSB increases initiation of Pol V bypass synthesis ³⁰⁸. This stimulation has recently been demonstrated to arise, in part, from a physical interaction between SSB and Pol V ³¹¹.

SSB increases Pol V access to the 3’ end of a DNA gap that is flanked by RecA filaments ³¹¹. The SSB113 protein and viral SSB proteins can substitute for E. coli SSB in this respect. However, when SSB113 is included in a translesion synthesis assay, little synthesis is observed, whereas the reaction is substantially more efficient in the presence of wild-type SSB ³¹¹. Interestingly, viral SSB proteins (gp32 from T4 phage and ICP8 from herpes simplex virus 1) allowed for attenuated activity in which DNA synthesis proceeds up to, but not beyond, the DNA lesion ³¹¹. When Pol III HE subunits β and γ (proteins that have been shown to assist with Pol V activity ^{308; 312}) were included in reactions containing the viral SSBs, translesion synthesis occurred. Because Pol V coprecipitates with SSB but not SSB113, a physical interaction between the SSB C-terminus and Pol V is likely to play a crucial role in maximizing Pol V translesion synthesis activity ³¹¹. This conclusion is strengthened in light of the physical interaction between SSB and MucB, a plasmid-encoded UmuC homolog ³¹³.

OTHER SSB-BINDING PROTEINS

Exonuclease IX. (Exonuclease IX binds SSB in an undefined manner)

E. coli Exonuclease IX (ExoIX) was initially identified as a putative exonuclease since it shares 60% identity with the DNA polymerase I 5’-to-3’ exonuclease domain ³¹⁴. Indeed, partially purified preparations of ExoIX appeared to possess exonuclease activity ³¹⁵; however it has since been shown that this activity is most likely due to an Exonuclease III contamination in ExoIX preparations, and that ExoIX itself is devoid of exonuclease activity ³¹⁶. The function of ExoIX in the cell remains unclear, as it has no apparent enzymatic activity and strains harboring ExoIX deletions (xni-) are indistinguishable from wild type ³¹⁷. However, ExoIX does interact directly with SSB as demonstrated by coprecipitation and crosslinking experiments in the absence of nucleic acid ³¹⁶.

Bacteriophage N4 virion RNA polymerase. (N4 RNA polymerase binds SSB (most likely via the SSB-Ct) and requires SSB to stabilize a promoter hairpin)

RNA polymerase from bacteriophage N4 (vRNAP) specifically requires E. coli SSB for early transcription ^318-320. N4 injects vRNAP into its host along with its genome in the initial stage of infection ^{321; 322}. Even though N4 encodes its own SSB protein ³²³, its binding activity appears to be specialized to destabilize a hairpin structure that functions as a promoter, making the phage reliant on E. coli SSB as well ³¹⁸. Interestingly, the N4 SSB functions as a transcriptional activator late in the phage's replication process, but does so through stimulation of E. coli RNA polymerase ^{324; 325}. E. coli SSB not only allows the N4 promoter hairpin structure to remain intact, but also assists in displacing nascent RNA from the ssDNA template, a task vRNAP is unable to accomplish alone ³²⁰. SSB binds both the ssDNA template and the RNA product, preventing the formation of a DNA-RNA hybrid. The result of maintaining both species in their non-duplex form is increased access to ssDNA and template recycling ³²⁰.

The stimulation of transcription by SSB is dependent upon the presence of the SSB-Ct element. Wild type SSB increases vRNAP transcription by twenty-fold, but variants that lack the C-terminal ten residues of SSB fail to stimulate, but do not inhibit, transcription ³²⁰. Although a direct protein-protein interaction between vRNAP and SSB has not been explicitly demonstrated, experimental evidence strongly suggests that one exists.

PROTEOMIC INDENTIFICATION OF SSB-BINDING PROTEINS

Two large-scale studies that probe networks of interacting E. coli proteins have been published to date (Table 1). The first utilized dual affinity-tagged proteins to identify binding partners of essential proteins ¹³⁵, whereas the second used hexahistidine affinity-tagged variants of the majority of the E. coli proteome to define interaction networks ³²⁶. Surprisingly, the His-tagging study detected just two of the known binding partners of SSB (RecG and UDG), which were only found when the partner proteins are the tagged bait (that is, tagged SSB failed to co-purify with either partner) ³²⁶. Two other His-tagged proteins (DNA photolyase and YbcN, a hypothetical protein) also were found to co-purify with SSB, but neither interaction has been confirmed outside of the co-purification study. Only two proteins co-purify with His-tagged SSB in the study: Peptidase D and RhlE, a putative helicase. It is possible that the purification conditions for His-tagged proteins are too stringent for most SSB-interacting proteins to remain stably bound to SSB throughout the purification method, which led to the large number of apparently false negative results for SSB-interacting proteins in the study.

Table 1.

SSB-interacting proteins found from proteomic studies.

SSB-interacting protein	Found in dual-affinity experiment?	Found in His-tag experiment?
DNA polymerase III α	Yes, as bait	No
DNA polymerase III χ	Yes, as bait	No
PriA DNA helicase	Yes, as prey	No
RecG DNA helicase	Yes, as prey	Yes, as bait
RecQ DNA helicase	Yes, as bait/prey	No
RecJ exonuclease	Yes, as bait/prey	No
Exonuclease I	Yes, as prey	No
RNase H	Yes, as bait	No
DNA photolyase	No	Yes, as bait
Uracil DNA glycosylase	No	Yes, as bait
Topoisomerase I	Yes, as bait	No
Toposimerase III	Yes, as bait/prey	No
HU- α	Yes, as prey	No
SecA translocase	Yes, as prey	No
DnaK chaperone	Yes, as prey	No
Peptidase D	No	Yes, as prey
RhlE putatitve helicase	No	Yes, as prey
YbcN hypothetical protein	No	Yes, as bait

In contrast to the His-tagged proteins screen, the dual affinity-tag study identified 52 interactions involving SSB: 37 when SSB was C-terminally tagged, and 15 others in which a partner protein was tagged ¹³⁵. This study validated interactions in experiments in which co-purifying partner proteins were tagged and the same interaction was detected reciprocally in a second, separate purification. This reduced SSB interacting proteins to the 13 verified complexes listed in Table 1. It is worth noting that since this study used C-terminal affinity tags to identify protein complexes, many false negatives could arise since the SSB-Ct forms a critical binding site for its partner proteins.

While nearly all of the listed binding partners have a clear role in nucleic acid metabolism, there are some indications that neither study has sampled the complete SSB interaction network. First, between the dual-affinity and His-tag studies, only one common binding partner was detected (RecG). This indicates experimental conditions greatly altered the spectrum of identified interacting proteins. Given that interactions with SSB can be relatively weak and dependent upon solution conditions, this is not surprising. Second, at least one of the “validated” protein interaction partners, Topoisomerase I, is not believed to interact directly with SSB ³²⁷, consistent with these purification schemes detecting both direct and indirect binding partners. Finally, since validation requires that both protein partners be amenable to similar tagging and purification procedures, some interactions may be lost as false negatives in validation screens. In the dual affinity-tag screen, there were several candidates that one can imagine as interacting with SSB in the list of non-validated partner proteins, such as DNA gyrase and Topoisomerase IV. However, since the reciprocal interaction was not detected, they are considered non-validated. Selected non-validated proteins such as these may warrant further investigation.

II. THERMODYNAMICS OF SSB-PROTEIN INTERACTIONS

The proteins that have been shown to interact directly with E. coli SSB protein all appear to contact the unstructured C-terminal region of the SSB protein, in particular the last 9 amino acids. One question that arises is whether there is any specificity associated with these different interactions; i.e., is SSB binding to these different proteins determined solely by its unstructured C-terminus or are there other interactions that are specific to the protein partner? A second question is whether the stoichiometry of binding and/or specificity is influenced by SSB binding to DNA (either ssDNA or more complex junction structures) and furthermore whether the mode of SSB binding to ssDNA influences its interactions with these other proteins. Of course, answers to these questions require quantitative thermodynamics studies; however, to date, only a few of the proteins known to interact with SSB protein have been studied using direct quantitative methods (Table 2). Furthermore, even for those that have been studied quantitatively, the solution conditions used for those studies often differ and since solution conditions generally affect these interactions, questions of specificity are currently difficult to answer.

Table 2.

Thermodynamic binding data for SSB protein complexes^*

(A) χ/SSB interaction
[Salt]	Method	SSB-Ct peptide binding data	SSB binding data	SSB/ssDNA binding data
High
300 mM	AU 83	ND	N=4	N=4
			K=(2+/−1) × 105	K=(2+/−1) × 105
200 mM	ITC a	ND	ND	N=2.8+/−0.6
				K=(4.5+/−1.1) × 105
				ΔH=−7.7+/−1.1
150 mM	SPR 67;b	ND	N=2.9+/−0.1	N=5.1+/−0.4
			K=(3.3+/−0.3) × 105	K=(4.6+/−0.1) × 105
150 mM	SPR 83	ND	N=2.5	N=4.4
			K=(2.7+/−0.5) × 105	K=(4.2+/−0.5) × 105
100 mM	SPR 67; c	ND	K=3.7 × 105	K=(1 −3) × 108 d
Low
20 mM	ITC a	N=0.9+/−0.1	ND	N=4.2+/−0.3
		K=(1.3+/−0.6) × 106		K=(6.1+/−2.1) × 106
		ΔH=−8.6+/−1.8		ΔH=−9.0+/−0.4
			N=4
5 mM	AU 83	ND	K=(4.0+/−1.0) × 105	N=4
				K=(7.4+/−1.0) × 106
none	Gel filtration 67	ND	ND	K=1.9 × 107

(B) PriA/SSB interaction
[Salt]	Method	SSB-Ct peptide binding data	SSB binding data	SSB/ssDNA binding data
High
200 mM	ITC a	ND	N=3.7+/−1.4	N=4.6+/−0.5
			K=(1.0+/−0.8) × 106	K=(1.8+/−0.5) × 106
			ΔH=−5.1+/−0.6	ΔH=−6.8+/−0.3
150 mM	SPR 232	K=(4.2+/−0.3) × 105	ND	ND
Low
20 mM	ITC a	N=1.0+/−0.1	N=1.7+/−0.2	N=5.2+/−0.2
		K=(1.8+/−0.7) × 106	K=(2.2+/−1.1) × 107	K=(7.1+/−1.8) × 107
		ΔH=−6.9+/−0.6	ΔH=−28.0+/−1.0	ΔH=−28.7+/−1.2

(C) RecQ/SSB interaction
[Salt]	Method	SSB-Ct peptide binding data	SSB binding data	SSB/ssDNA binding data
High
150 mM	ITC 132	N=0.90+/−0.02	N=3.4+/−0.6	ND
		K=(1.5+/−0.3) × 105	K=(1.5+/−0.4) × 105
		ΔH= −9.3 +/−1.1	ΔH= −18.0 +/−2

(D) UDG/SSB interaction
[Salt]	Method	SSB-Ct peptide binding data	SSB binding data	SSB/ssDNA binding data
Intermediate
50 mM	SPR 268	ND	ND	K= 5.9 × 106

^*- binding parameters: N - stoichiometry of protein binding, K (M⁻¹) - association constant, ΔH (kcal/mol) - enthalpy change

^aKozlov and Lohman, unpublished data

^bParameters obtained fitting SPR data presented in Table I of Kelman et al (ref.⁶⁷), to the N noninteracting sites model

^cIncludes ψ protein

^dDetermined in the presence of γδδ’ or τδδ’ subunits of pol III HE

Binding of SSB to the χ subunit of the Pol III HE in the presence and absence of ssDNA was investigated using Surface Plasmon Resonance (SPR) ^{67; 82; 83}, analytical ultracentrifugation (AU) ⁸³and gel filtration ⁶⁷. The interactions of PriA helicase with the SSB C-terminal peptide²³², RecO with SSB ²⁰⁰, and E. coli UDG with SSB ²⁶⁸ have also been examined using SPR methods, and, AU has been used to study the interaction of exoI with SSB and SSB with mutations in its C terminal mutants ⁶⁶. Isothermal titration calorimetry (ITC) has been used to characterize the interactions of SSB and its C-terminal peptide with RecQ helicase ¹³², PriA helicase, and the χ subunit of the Pol III HE (Kozlov and Lohman, unpublished results).

As discussed above, E. coli SSB tetramers can bind to long ssDNA in a number of different binding modes that display distinct ssDNA binding properties, differing in the number of subunits that interact directly with the ssDNA, the inter-tetramer cooperativity, the affinity and the occluded site size ^{12; 34; 35; 66; 328}. The transitions among these different binding modes can be modulated by the monovalent salt concentration ⁵¹, divalent and mutivalent cations ^{49; 52}, as well as the SSB to ssDNA binding density ^{14; 50; 53; 58}. It is therefore conceivable that the ability of SSB to recruit other proteins through interactions with its C terminus might be influenced by the particular mode of SSB binding to ssDNA. In addition, due to the acidic nature of the C-terminus of SSB, there is likely to be an electrostatic component to its interaction with other proteins. For these reasons, changes in solution conditions and especially salt concentration and type are likely to affect the binding properties of these proteins to SSB and its complexes with DNA.

Protein Binding to SSB and its C-terminal peptides at high salt

Most studies of SSB binding to other proteins have been carried out under high salt conditions (100−300 mM NaCl)^{67; 83; 132}. These are conditions that at equilibrium in vitro favor the fully wrapped high site size, (SSB)₆₅ mode of binding to ssDNA. Under these conditions, the affinities of SSB (without DNA) for the χ subunit^{67; 83}, and for RecQ helicase ¹³² are within the range of 2×10⁵ to 4×10⁵ M⁻¹ (Table 2A and 2C). Similar values have been reported for the interaction of RecQ ¹³² and PriA ²³² with peptides containing the last 9 or 15 amino acids of the SSB C-terminus also at similar high salt concentrations (Table 2B and 2C). Importantly, C-terminal deletion mutants of SSB that are missing the last 8 ¹³² or 26 amino acids ⁸³ as well as the SSB-113 mutant (Pro 176 to Ser) do not show any detectable affinity for χ ^{67; 83} or RecQ ¹³². The affinities of C–terminal peptides carrying the Pro to Ser substitution are also reduced dramatically for χ ⁶⁷, RecQ ¹³² and PriA²³². On the basis of these data it appears that at high salt concentrations (100−300 mM NaCl) SSB displays little specificity for χ vs. RecQ and interacts weakly ((2−4)×10⁵ M⁻¹) with these proteins using primarily its unstructured C-terminus. The reported stoichiometries of binding (proteins per SSB tetramer) are 4 for RecQ ¹³², but only 2−3 for χ^{67; 83}. The lower values for χ might be explained by the fact that they were obtained using SPR⁸³, where chemical coupling of SSB to the chip surface could result in some occlusion of potential binding sites.

Protein binding to ssDNA/SSB complexes at high salt

Under high salt conditions no differences in affinities were reported for the binding of χ or PriA to a pre-formed ssDNA/SSB complex compared to SSB alone (Tables 2A and 2B). The equilibrium constants reported for χ in the presence of poly(dT) using AU ⁸³ and for SSB bound to (dT)₆₅ using SPR^{67; 83}, are within the same range as reported for SSB alone with stoichiometries of ∼ 4 χ proteins bound per SSB tetramer. This is also the case for PriA binding to SSB even in the presence of (dT)₇₀. Therefore, it appears that the presence of ssDNA at high salt conditions has little effect on χ-SSB interactions, although PriA does display a higher affinity for SSB than does χ.

Interestingly, the presence of the ψ subunit, which is also a component of the Pol III HE and interacts with the χ subunit, does not affect the affinity of χ for SSB alone (3.7×10⁵ M⁻¹)⁸², whereas on a DNA template coated with SSB the affinity increases to 3×10⁸ M⁻¹ when the γδδ' subunits (clamp loader assembly) are also present. Hence, the presence of ssDNA and auxiliary proteins appears to increase the affinity between SSB and χ considerably (∼1000 fold).

Protein binding to ssDNA/SSB complexes and SSB at lower salt

In contrast to the results obtained at high NaCl concentrations, the binding of ssDNA to SSB has a more pronounced effect on SSB-χ binding affinity at lower salt concentrations. The equilibrium constant for χ binding to SSB determined by AU at 5 mM NaCl ((4.0±1.0)×10⁵ M⁻¹) increases ∼ 20 fold (7.4×10⁶ M⁻¹) when SSB is complexed with poly(dT) ⁸³(Table 2A). PriA also shows an increase in affinity for SSB in the presence of ssDNA (Table 2B), with a binding affinity that is ten-fold higher than for χ, demonstrating some degree of SSB specificity for PriA.

With increasing salt concentration, the affinities of PriA and χ for ssDNA/SSB complexes decrease (Tables 2A and 2B). However, essentially no change in affinity is observed for the interaction of χ with SSB alone as the salt concentration increases from 5 mM to 300 mM NaCl⁸³. This is somewhat surprising, especially since a ∼3−5 fold increase in affinity is observed for just the C-terminal peptide (9 or 15 amino acids long) binding to χ at low salt (20 mM NaCl) as determined by ITC (Table 2A). A significant decrease in affinity upon increasing salt concentration was also reported for SSB binding to Exonuclease I ⁶⁶.

In summary, for χ and RecQ, the interaction with SSB at high salt concentrations (100−300 mM NaCl) is characterized by moderate affinities ((2−4)×10⁵ M⁻¹), which are unaffected by ssDNA (although this is shown only for χ) and are similar to the affinities determined for the interaction of these proteins with the C-terminal SSB peptide. Presently there is not enough data to estimate quantitatively the effect of low salt conditions on the equilibrium constants for SSB-protein binding, although strengthening of the interaction is expected ⁶⁶. On the other hand it is evident that at low salt both χ ^{67; 83} and PriA (Kozlov and Lohman, unpublished results) interact with ssDNA/SSB complexes with much higher affinities than with SSB alone. The affinities of these proteins for SSB and ssDNA/SSB complexes also decrease with increasing salt concentrations. This latter effect may simply reflect an effect on the electrostatic component of the interaction or it could suggest that SSB bound in different binding modes (e.g., (SSB)₃₅ at low salt or (SSB)₆₅ at high salt) possesses different binding properties for these proteins.

III. ROLE OF SSB IN BACTERIAL RECOMBINATION PROCESSES

Recombination is a process focused on the repair of DNA strand breaks, primarily double strand breaks and single strand gaps. Both types of DNA damage are found most commonly at the sites of stalled or collapsed replication forks ^{7; 11; 329-334}. The single strand gaps at stalled replication forks can be quite extensive ^{178; 208; 335; 336}. Inevitably, SSB protein binds to the ssDNA in these gaps. Its role is not simply protective. As already described, SSB is a facilitator of ssDNA metabolism, and its interactions with the proteins of recombinational DNA repair are critical to the course of that repair. In this section, we consider the roles played by SSB in recombination both as a facilitator and an impediment to the overall process.

The late 1970s and early 1980s witnessed a renaissance in the understanding of bacterial recombination, centered on the functional characterization of the RecA protein, both in vivo and in vitro. The primary roles of RecA protein, in recombination ^337-340, in the induction of the SOS response ^{341; 342}, and in SOS mutagenesis ^343-345, were established. In recombination, RecA protein promotes a series of DNA strand exchange reactions that lie at the heart of all recombinational processes ^346-348. In the induction of the SOS response, RecA protein acts as a coprotease – facilitating the autocatalytic degradation of the LexA repressor of SOS genes ^{349; 350}. In the mutagenesis that accompanies the SOS response, RecA protein acts as an essential activator and probable subunit of the error-prone translesion DNA polymerase V ^{309; 310}.

Just out of the limelight, it became apparent at about the same time that SSB played a significant role in just about everything RecA protein did ^{346; 351-356}. The RecA and SSB proteins were linked in their functions by three other proteins: RecF, RecO, and RecR ^{171; 195-197; 200; 357; 358}. Notably, as was described above, RecO protein interacts with the SSB ^{194; 197; 200}, and in particular the SSB C-terminus ¹⁹⁹.

Two SSB mutations played key roles in elucidating the function of SSB in recombinational DNA repair. The first is ssb113, which was described above as producing a SSB-Ct variant mutation (Pro176Ser) that diminishes heterologous protein interactions with the SSB C-terminus ⁶⁰. The ssb113 alteration results in a temperature-sensitive conditional lethality at 30 °C that is not suppressed by overexpression of the mutant protein ⁶⁰. The second is the ssb1 mutation, which codes for an SSB with a mutation in the OB fold (His55Tyr) and confers a temperature-sensitive phenotype, with much SSB function abrogated at 42 °C ³⁵⁹. The ssb1 mutation is known to destabilize the SSB tetramer ^359-361. Overproduction of the SSB1 protein suppresses the temperature-sensitive phenotype ³⁵⁹. Both mutant proteins confer a variety of defects in DNA metabolism in the strains expressing them, including sensitivity to UV irradiation, growth defects, and recombination defects ^{60; 61; 359; 362}. Studies of SSB113 led to some of the first suggestions that SSB interacted directly with multiple other proteins ⁶⁰. The ssb113 mutation produced severe defects in DNA synthesis and an increase in double strand breaks that led to chromosome degradation, while the ssb1 mutation produced more modest effects ⁶¹. Wang and Smith suggested that SSB played a key role in protecting exposed ssDNA during recombinational DNA repair ⁶¹. Lieberman and Witkin ³⁵⁴ noted that the DNA degradation and UV sensitivity seen at 42 °C in an ssb1 mutant cell was not rescued by the inactivation of recBCD (exonuclease V), or by overexpression of the wild type recA gene. This indicated that multiple nucleases were involved in the chromosomal degradation, and RecA did not function properly in recombinational DNA repair unless normal SSB was present. The ssb113 mutation blocks the induction of SOS at 30 °C (its restrictive temperature). The increase in UV sensitivity and decline of SOS-associated functions such as mutagenesis seen in the ssb113 strains was suppressed by the introduction of a recA allele (recA730 = recA E38K) that promotes constitutive SOS induction ³⁵⁴. The results implied that SSB played some direct role in the activation of RecA protein for SOS induction and SOS mutagenesis.

Although many proteins interact with the C-terminus of SSB, the details of the interaction are likely to vary from one protein to the next. A deletion of 10 C-terminal amino acids of SSB renders E. coli cells inviable ⁶⁹. This implies that key interactions required for basic processes such as DNA replication occur at the SSB C-terminus. However, the alteration of the proline at position 176 in ssb113 allows cell growth while rendering the cell UV sensitive ⁶⁰. This indicates that interactions required for DNA repair are disrupted by the Pro176Ser change in ssb113, but the interactions required in replication remain intact.

Effects of SSB on RecA protein function

SSB plays a complicated role in RecA reactions. RecA binding to ssDNA generally occurs in two phases, nucleation and filament extension. Nucleation must involve one or a few RecA monomers, and recent work suggests the number is about 4−5 ^{363; 364}. Under most conditions, RecA filament extension is relatively fast, allowing single filaments to coat long ssDNA molecules contiguously. Filament extension on ssDNA occurs in the 5' to 3' direction ^{201; 365}, with little addition of subunits to the 5′-proximal end detectable ³⁶⁶. RecA disassembles from the 5'-proximal end ³⁶⁷, in a reaction that requires ATP hydrolysis ³⁶⁶. Disassembly from filaments bound to ssDNA occurs at a rate of about 70 subunits min-1 at 37 °C ³⁶⁶. This is substantially slower than filament extension, so that the growing ends of RecA filaments nucleated on a circular ssDNA soon encounter the disassembling end of the same or other filaments on the same DNA. Filament extension is blocked or impeded by secondary structure in the ssDNA substrate ³⁵².

SSB has different effects on the two phases of RecA filament formation. RecA filament nucleation is inhibited, and under some conditions blocked entirely, if SSB is allowed to coat the DNA prior to RecA addition ^{200; 201; 346; 352; 367; 368}. This inhibition is relieved in some RecA mutant proteins. These include RecA E38K (RecA 730 ^{191; 369}), RecA V37M (RecA 803 ³⁷⁰), RecA T121I (RecA 2020 ^{371; 372}), RecA E38K, I298V (RecA 441 ^373-375), and a truncation of the RecA C-terminus of 17 amino acid residues (RecA ΔC17; E.A. Wood and M.M. Cox, unpublished results). The involvement of the RecA C-terminus (also highly negatively charged) in the suppression of RecA loading has some possible mechanistic implications. The RecA C-terminus (the last 25 amino acid residues) is a kind of autoregulatory flap, removal of which enhances a range of RecA functions ^{309; 310; 376-382}. A plausible scheme is that the RecA C-terminus buries a surface on RecA protein that is required for a RecA interaction with SSB or the ssDNA bound to SSB, an interaction that is in turn necessary for SSB displacement and RecA nucleation. One or more of the other amino acid residues (E38, V37, T121) whose mutation also creates a protein more adept at bypassing the SSB block to nucleation may also be involved in interactions needed to cover the SSB displacement surface on RecA. For wild type RecA protein, mediator proteins alleviate the slow nucleation imposed by pre-bound SSB, as described below. These mediators may interact with the RecA C-terminus as part of their function.

When SSB is added to the ssDNA after, rather than before, RecA protein in an experiment that also includes ATP, the subsequent RecA reactions are enhanced rather than inhibited ^{383; 384}. The early addition of RecA provides an opportunity to get past the slow nucleation step. In contrast to nucleation, the extension phase of RecA filament formation is facilitated by SSB. The wild type E. coli RecA protein is unable to disrupt secondary structure in ssDNA that is encountered during filament extension, leading to the formation of abbreviated filaments that do not uniformly coat the DNA. The SSB protein binds to and disrupts the ssDNA secondary structure. Extending RecA filaments readily displace SSB, allowing RecA to form a contiguous filament on the DNA. This general scheme, first proposed in the mid-1980s ^{352; 385; 386}, is now seamlessly consistent with more than two decades of work on the reactions promoted by RecA and SSB ^{387; 388}.

RecA protein promotes DNA strand exchange in vitro optimally under conditions that include relatively high concentrations of free Mg²⁺ ion (∼10 mM). The SSB binding mode under these conditions may play a major role in the course of the reactions. Under these conditions, SSB binds primarily in the low cooperative, fully wrapped (SSB)₆₅ binding mode ^{15; 49; 50}. The (SSB)₃₅ mode, which binds with high nearest neighbor cooperativity, has been proposed to function in DNA replication ³⁸⁹.

There is some evidence for a persistent association of SSB with RecA protein filaments after contiguous RecA filaments are formed on ssDNA ^{384; 390}. A similar interaction has been detected between the eukaryotic RecA-homologue Rad51 protein and the eukaryotic RPA protein ³⁹¹. However, the positive effects of SSB on RecA filament formation (and of RPA on Rad51 filament formation in eukaryotes) are not limited to species-cognate SSBs ^{392; 393}, and no indication of a persistent E. coli RecA-SSB complex has been evident in studies employing electron microscopy. If an association exists, it is relatively weak, does not occur between RecA and SSB when neither is bound to DNA ²⁰⁰, and does not play a role that is essential (or even stimulatory) to the RecA-mediated DNA strand exchange reactions commonly carried out in vitro. In those DNA strand exchange reactions, the RecA filaments form on ssDNA, and the bound DNA is then aligned with a homologous duplex DNA. A strand switch ensues, in which one strand of the duplex DNA is transferred to the original ssDNA to create a new duplex, and one strand of the original duplex is displaced. The SSB involved in facilitating the RecA filament formation prior to strand exchange is bound to the displaced ssDNA once strand exchange is complete ³⁹⁴. The RecA protein remains bound to the product duplex DNA, or dissociates, depending on solution conditions ^395-397. In vivo, a persistent association of SSB with the RecA filaments could help choreograph the efficient transfer of the SSB to the displaced strand, and the transferred SSB could serve as a target for the binding of multiple other proteins involved in post-recombinational processes. In fact, direct transfer of SSB between two DNA strands has been documented and is facilitated when SSB is bound in the (SSB)₃₅ binding mode such that two SSB OB-folds are unoccupied by ssDNA ³⁹⁸. A persistent association of SSB with a RecA nucleoprotein filament, one that leaves ssDNA binding surfaces on the SSB unoccupied, might serve a similar function. Such a role might not translate into a measurable advantage of a RecA filament-SSB interaction during in vitro reactions, but may merit further experimental investigation. Notably, the SSB protein of Mycobacterium smegmatis interacts directly and in a species-specific manner with M. smegmatis RecA nucleoprotein filaments ³⁹⁹. This interaction relies on the C-terminus of M. smegmatis SSB ³⁹⁹.

The E. coli mediators, RecF, RecO, and RecR proteins

The SSB barrier to RecA nucleation gives rise to a need for protein mediators – proteins that bypass the barrier and facilitate the nucleation process. The same problem exists in the loading of RecA-class recombinases in all organisms, and mediators and their critical loading functions are now recognized as common in bacteria, archaeans, and eukaryotes ^400-403. There is potential for damaging genomic rearrangements inherent in recombination. Mediators provide a critical opportunity for every cell to regulate recombinase function at a point prior to the initiation of any recombinational process. The E. coli RecF, RecO, and RecR proteins are considered the prototypes of this class of proteins. However, considerable mechanistic variation may exist in different species and classes of organisms, as well as phages and viruses encoding recombination systems.

The genes coding for the RecF (40.5 kDa ¹⁰⁶), RecO (27 kDa ¹⁸⁶), and RecR (22 kDa ^{189; 404}) proteins were discovered independently as functions that had modest effects on recombination and UV resistance in E. coli. The phenotypes of mutations in the three genes are very similar, and the effects of mutations in two or three of the genes are in many cases equivalent to the effects of any one of them, defining them as an epistatic group ^{170; 189}. Several additional lines of evidence indicate that these three proteins function together early in recombinational processes, and tie them to a role in facilitating RecA filament assembly on SSB-coated ssDNA. Mutations in all three genes are suppressed by the recA E38K ^{191; 369}, recA V37M ³⁷⁰, recA T121I ^{371; 372}, recA E38K, I298V ^373-375, and a truncation of the recA gene that removes 17 codons at the end encoding the C-terminus (E.A. Wood and M.M. Cox, unpublished results). As already noted, the mutant RecA proteins produced by these same genes generally exhibit an enhanced capacity to displace SSB and bind ssDNA. In addition, a gene in bacteriophage λ called ninB or orf can replace the functions of all three recFOR genes in lambda recombination ^{405; 406}. Overexpression of SSB in E. coli produces a recFOR-like phenotype ³⁵⁸, providing another link between RecFOR and the SSB barrier to recombinase nucleation. Mutant bacteria missing any of the recFOR functions exhibit a delayed activation of the SOS response, most easily interpreted as a block to the formation of the RecA filaments required to facilitate the autocatalytic cleavage of the LexA repressor ^{182; 370}.

Some results indicate that the recF, recO, and recR genes possess some functional distinctions. The three genes are not ubiquitous in bacteria, nor are they reliably coincident. A survey of recombination functions in 117 bacterial species demonstrates that bacteria tend to have all three genes, only the recO and recR genes, recR alone, or none ⁴⁰⁷. The recF gene is absent from 29 species in the survey, while recR is absent from only 10. There are only two cases where a recF gene is not accompanied by both recO and recR, and in both cases it is recO that appears to be missing ⁴⁰⁷. This may reflect a high substitution rate that seems to exist for recO, and an accompanying difficulty in identifying some recO homologues by classical search algorithms. A recO gene was subsequently identified in one of the two species in question (T. thermophilus), and its protein product has been studied at some length ^{203; 408}. Taking the potential for discovery of a few more recO genes into account, the standard (or at least most common) complement of mediator functions in bacteria is either recFOR or recOR ⁴⁰⁷.

Even where all three genes are present, some results indicate that they do not always function together. In a strain lacking the function of PriA protein, the additional loss of recO produces different results than the loss of recF ^{175; 409}. Mutation of recR or recF suppresses the strong effects of recO mutation, suggesting that RecF and RecR are deleterious to the cell in the absence of RecO ¹⁷⁵. The RecF protein, but not RecO or RecR, is needed for the activation of DNA polymerase V and mutagenic translesion DNA synthesis (TLS) ¹⁷⁹, providing one instance in which RecF may function without the other two proteins. SOS induction, UV resistance, and viability at 42°C are all reduced if RecF protein is overexpressed in vivo ⁴¹⁰. Overexpression of the RecOR proteins suppresses many of the effects of either RecF overexpression ⁴¹¹ or a recF null mutation ¹⁹⁰. These varied results indicate that RecOR may function on its own, or as part of a larger RecFOR system, and RecF may have a few independent functions.

This evidently complex situation is mirrored in vitro. Structural information about these proteins is becoming available, and this should promote mechanistic insight as it is coupled to ongoing biochemical analysis. All of the structures made available to date are from the RecFOR homologues of D. radiodurans. The D. radiodurans RecF protein exhibits an unexpected structural similarity with the head domain of the eukaryotic Rad50 protein ⁴¹². However, it lacks the long coiled-coil domain of Rad50 ⁴¹². RecF is a member of the ATP-binding cassette (ABC) ATPase family of proteins, and possesses a weak ATPase activity ^413-415. RecF binds to DNA, with increased affinity for dsDNA ^{413; 414; 416}. ATP binding triggers RecF dimerization ⁴¹². ATP hydrolysis triggers dissociation from DNA ⁴¹⁵. RecR protein forms a complex with RecF and improves the stability of RecF-DNA complexes ^{413; 414}.

The D. radiodurans RecO protein contains an N-terminal domain that adopts an OB-fold, a novel α-helical domain, and a zinc-binding C-terminal domain ^{417; 418}. RecO catalyzes complementary DNA strand annealing and invasion of duplex DNA by a complementary ssDNA ^192-194. The RecO binds directly to ssDNA, a property established for the RecO proteins derived from E. coli, T. thermophilus, and D. radiodurans ^{192-194; 197; 203; 417; 418}. Notably, it is the RecO protein that interacts with SSB ^{194; 197; 199; 200}.

The RecR homologs in D. radiodurans and B. subtilis both bind DNA ^419-421, although the E. coli RecR protein has no known intrinsic enzymatic or DNA binding activities. The D. radiodurans RecR structure is a tetrameric ring, with each monomer featuring a helix-hairpin-helix motif, a zinc finger motif, a Toprim domain, and a Walker B motif ^{420; 421}.

In vitro, the E. coli and T. thermophilus RecR proteins bind to their cognate RecF and RecO proteins ^{197; 200; 202; 203; 408; 413; 414}. Both T. thermophilus RecF and RecO proteins interact with the C-terminal TOPRIM domain of T. thermophilus RecR ⁴⁰⁸, providing a plausible explanation for an apparent competition between RecF and RecO for RecR binding that has been observed for the E. coli proteins ²⁰². The structure of a D. radiodurans RecOR complex has also been elucidated ⁴²². The proteins form a heterohexamer, with two RecO subunits on opposite faces of the RecR tetramer ring, and the OB domains of the RecO subunits proximal to the RecR ring ⁴²². No structures of thse proteins with ssDNA or SSB are yet available.

The RecOR proteins clearly function together, and under many conditions these two proteins are necessary and sufficient to load RecA protein onto SSB-coated ssDNA ^{197; 199; 200; 202; 203}. No conditions have yet been found in which one protein or the other alone can mediate the RecA loading process. As already noted, the RecO protein interacts directly with SSB ^{199; 200; 203}. Significantly, removal of the 8 C-terminal residues of SSB eliminates most RecO function in the loading reaction ¹⁹⁹, indicating that a RecO interaction with the SSB C-terminus is critical to the loading pathway. Early models indicated that RecOR does not displace SSB, but instead binds to it to form a RecO-RecR-SSB complex that facilitates RecA nucleation ^{200; 201}. A recent examination of the loading process with the T. thermophilus proteins provided evidence for SSB displacement ²⁰³. The rate-limiting step in E. coli RecOR-mediated loading of RecA protein is the binding of RecO to ssDNA ¹⁹⁹. This is inhibited by SSB, in spite of the direct interaction of RecO with the SSB C-terminus ¹⁹⁹. The only set of conditions in which a small (8−10 min) lag in RecA loading was abolished was one in which the RecO was bound to ssDNA prior to the SSB ¹⁹⁹. A model for the RecOR-mediated RecA loading process is presented in Figure 3.

Figure 3.

Loading of RecA protein onto SSB-coated ssDNA by the RecOR proteins. The RecO protein, in a complex with RecR, first binds to the C-terminus of SSB. The RecOR complex with SSB is then rearranged to permit direct binding of RecOR to the ssDNA and displacement of an SSB tetramer. Once RecOR is loaded, RecA interacts with RecOR (perhaps in a way that alters the conformation of the RecA C-terminus so as to expose an intrinsic loading surface), and RecA nucleation occurs. This is followed by rapid and unassisted RecA filament extension. This figure is based on recent studies of the loading process ¹⁹⁹; ²⁰³.

Under most conditions, the RecF protein is either neutral or inhibitory for RecA loading on SSB-coated ssDNA when added to reactions containing RecOR ^{197; 199-202}. The RecF protein has other demonstrable functions on the RecA filament formation process. RecFR complexes bind tightly to dsDNA, and can block the extension of RecA protein filaments initiated in ssDNA gaps into adjacent duplex DNA regions ⁴¹⁴. RecF also interacts directly with the E. coli RecX protein, and antagonizes its function ⁴²³. The RecX protein blocks RecA filament extension, and the RecF function in this case may facilitate RecA protein extension in some instances. However, neither of these functions appears to fully explain the phenotypes of studied recF mutant strains.

When DNA substrates are used that incorporate short duplex regions on the ssDNA (generated by annealing short oligonucleotides to a bacteriophage ssDNA circle), the E. coli RecF protein has a positive effect on the RecA loading process in concert with RecR protein ¹¹⁷. This may reflect a special role for RecF protein in augmenting the loading process at the ends of DNA gaps. The positive effect of RecF is seen only when SSB is present at very high concentrations, corresponding to a 6−10 fold excess relative to available ssDNA binding sites (117; M. D. Hobbs and M. M. Cox, unpublished data), a requirement that is not yet explained.

With accumulating structural and biochemical data, this system seems poised for rapid advancement. Although RecF may have a special function in augmenting RecOR at the ends of gaps, there is no evidence that RecF binds specifically to those gap ends ⁴¹⁴. Thus, there is potential for the discovery of additional targeting proteins in this system. A complete understanding of the RecFOR loading mechanism will facilitate studies of this critical function in all organisms. It should also facilitate an improved understanding of the dynamic nature of the SSB interaction with many other proteins.

IV. SUMMARY AND PERSPECTIVE

By binding both ssDNA and proteins central to every aspect of genome maintenance, eubacterial SSB proteins form a prominent interface at which genome maintenance pathways converge. As we have attempted to highlight in this review, the notion that SSB proteins are inert protective factors in bacterial cell biology vastly underestimates the contributions of this central scaffolding protein to genomic information storage and fidelity. From defining the substrates upon which DNA replication, recombination and repair must operate to playing an active role in nucleating complexes of enzymes, SSB proteins are central players in genome biology. Future work is needed to assess whether and how protein complex formation with SSB is regulated in vivo to determine which of the many competing interactions will predominate in a particular situation. In addition, the SSB-Ct-dependent nature of SSB/heterologous protein complexes could offer distinguishing features against which novel antibacterial therapies might be developed.