DNA damage alters binding conformations of E. coli single-stranded DNA-binding protein

Abstract

Single-stranded DNA-binding proteins (SSBs) are essential cellular components, binding to transiently exposed regions of single-stranded DNA (ssDNA) with high affinity and sequence non-specificity to coordinate DNA repair and replication. Escherichia coli SSB (EcSSB) is a homotetramer that wraps variable lengths of ssDNA in multiple conformations (typically occupying either 65 or 35 nt), which is well studied across experimental conditions of substrate length, salt, pH, temperature, etc. In this work, we use atomic force microscopy to investigate the binding of SSB to individual ssDNA molecules. We introduce non-canonical DNA bases that mimic naturally occurring DNA damage, synthetic abasic sites, as well as a non-DNA linker into our experimental constructs at sites predicted to interact with EcSSB. By measuring the fraction of DNA molecules with EcSSB bound as well as the volume of protein bound per DNA molecule, we determine the protein binding affinity, cooperativity, and conformation. We find that, with only one damaged nucleotide, the binding of EcSSB is unchanged relative to its binding to undamaged DNA. In the presence of either two tandem abasic sites or a non-DNA spacer, however, the binding affinity associated with a single EcSSB tetramer occupying the full substrate in the 65-nt mode is greatly reduced. In contrast, the binding of two EcSSB tetramers, each in the 35-nt mode, is preserved. Changes in the binding and cooperative behaviors of EcSSB across these constructs can inform how genomic repair and replication processes may change as environmental damage accumulates in DNA.

Significance

Single-stranded binding proteins (SSBs) bind transiently exposed single-stranded DNA (ssDNA) during DNA replication, recombination, and repair. SSBs both protect ssDNA from degradation and recruit additional proteins to aid in essential cellular processes. Escherichia coli SSB (EcSSB), a well-studied model system, binds ssDNA in multiple conformations, occluding variable lengths of substrate. We examine EcSSB binding to ssDNA substrates at a single-molecule level and find that modifying DNA to imitate naturally occurring DNA damage alters the preferred binding conformation of EcSSB without reducing its high binding affinity. Our results suggest that EcSSB can bind damaged ssDNA in a site-directed manner that could help facilitate specific remediation of individual bases.

Introduction

Single-stranded binding proteins (SSBs) are a class of proteins that bind preferentially to single-stranded DNA (ssDNA) with high affinity. This binding specificity allows SSBs to quickly and stably bind regions of ssDNA that are transiently exposed during essential cellular processes such as DNA replication, recombination, and repair (1,2,3,4). The presence of SSB prevents the formation of secondary structure that can inhibit polymerization and degradation by nucleases. SSBs can also recruit other proteins to perform genome maintenance functions (5,6,7,8,9).

The SSB of Escherichia coli (EcSSB), perhaps the most well-studied SSB, is a stable homotetramer, with each 177-amino-acid, 18.9-kDa subunit containing an oligonucleotide/oligosaccharide-binding (OB) fold and a disordered C-terminal tail (10,11). The OB folds are structured, both individually and when forming a tetramer (12,13), and each can independently bind ssDNA substrates. Thus, depending on substrate length and solution conditions, a single ssDNA can wrap around the OB fold tetramer in different conformations. In its largest binding size conformation, an EcSSB tetramer can accommodate a single 65-nucleotide (nt) ssDNA that binds to each OB fold as it wraps around the tetramer surface. This conformation is most stable in vitro at high salt concentrations and low ratios of protein to DNA (14). At lower salt concentrations and in the presence of excess protein, however, EcSSB can bind to a 35-nt length of ssDNA (15), such that more tetramers can be accommodated on a substrate of defined length. In the 35-nt state, not all OB folds directly interact with the ssDNA, and a structure has been resolved in which two 35-nt ssDNAs bind to one EcSSB tetramer (12). Besides the main 65 and 35 states, other less stable binding states have been proposed or observed under different experimental conditions (16,17). In contrast, the C-terminal tail, which consists of an acidic tip attached to the OB fold by a long, disordered linker, does not function primarily through direct interaction with ssDNA substrates. Instead, the C-terminal tail primarily interacts with other proteins, including other EcSSB tetramers (18,19,20,21).

Although EcSSB must be able to bind ssDNA in a sequence-independent manner, such as when it cycles over the full genome during replication, there is possibility for substrate/sequence-specific effects. In addition, EcSSB plays a vital role in DNA repair (22) and localizes in response to DNA damage (23). DNA damage can result from chemical reactions, radiation exposure, and enzymatic activity (24). In particular, depurination of DNA bases results in apurinic/apyrimidinic (AP) sites, also known as abasic sites. Unrepaired AP sites can stall DNA polymerization at the replication fork and result in mutation (25,26,27,28). EcSSB has been shown to play a role in the SOS response to DNA damage, interacting with repair proteins such as RecA (29), RadD (30), and Exonuclease I (31), and cells deficient in SSB display increased mutagenesis (32). It is less understood, however, whether EcSSB itself binds to sites of DNA damage differently than to undamaged, canonical DNA bases.

To test the hypothesis that sites of DNA damage modulate EcSSB binding, we directly observe EcSSB binding in vitro at a single-molecule level to ssDNA substrates with modified bases that mimic DNA damage. Due to the prevalence of abasic sites in DNA, occurring spontaneously approximately once per generation in E. coli and more frequently under stress conditions (33,34), we chose to utilize a stable abasic site mimic in these studies. We utilized a 67-nt-long sequence that can accommodate one tetramer in the 65 state or two in the 35 state (without excess unbound ssDNA) and chose bases for modification expected to interact closely with bound proteins based on structural models (12). Although EcSSB is able to bind these damaged substrates with nanomolar affinity, similar to its binding to undamaged DNA, we find the exact binding conformation is modulated, favoring the simultaneous binding of two proteins flanking the damage site, even under conditions for which the binding of a single protein is favored for the undamaged ssDNA substrate. These results suggest that the exact spatial binding pattern of EcSSB along a longer length of ssDNA could be influenced by specific locations of DNA damage, which in turn could provide a mechanism to direct DNA repair machinery.

Materials and methods

Protein purification

Wild-type (WT) EcSSB (UniProt P0AGE0) was purified as previously described (16,35). The plasmid encoding WT EcSSB pEAW134 was a gift from Dr. Mark Sutton of the University at Buffalo. Briefly, EcSSB was expressed in E. coli BL21 Tuner cells, precipitated with Polymin P followed by ammonium sulfate, and then further purified on an ssDNA-cellulose column. Protein concentration was determined spectroscopically using an extinction coefficient of ε₂₈₀ = 1.13 × 10⁵ M⁻¹cm⁻¹ for the EcSSB tetramer (35).

DNA-binding substrates

Hybrid ds/ssDNA substrates were prepared as previously described (16). A 268-bp dsDNA was produced by PCR amplification using pUC19 plasmid template and Taq DNA polymerase (New England Biolabs [NEB]). The primers are listed in Table 1 and 30 PCR cycles of denaturing (95°C, 30 s), annealing (56°C, 30 s), and extension (68°C, 60 s) were performed. The product was digested by BamHI (37°C, 4 h), resulting in a 248-bp dsDNA with a 4-nt 5′ overhang. The cut dsDNA was then incubated with a 10× molar excess of ssDNA substrates and linker oligos (Table 1). The DNA mixture was heated to 50°C for 5 min then gradually cooled to 16°C, allowing the linker oligo to anneal to both the dsDNA overhang and the ssDNA substrate. The constructs were ligated overnight (4°C, 16 h) with T4 DNA ligase. The sample was gel purified to remove excess ssDNA, ensuring all ssDNA in the final product was ligated to a dsDNA marker. The final ligated product was 260 bp of dsDNA with a 67-nt ssDNA overhang. In addition to the ds/s DNA hybrid, ligation of the dsDNA to itself produced a 500-bp dsDNA and this additional gel band was also excised for use as a molecular calibration ruler. All DNA oligos and enzymes were purchased from Integrated DNA Technologies and New England Biolabs, respectively.

Table 1.

DNA Oligos Used for Construction of DNA Substrates

Oligo	Sequence (5′–3′)
PCR primer 1	CAGGTCGACTCTAGAGGATCCC
PCR primer 2	ACTGAGAGTGCACCATATGCG
Linker	GATCGGGAAGGG
ssDNA substrate	CGTTACTCAGATCAGGCCTGCGA AGAXYTGGGCGTCCGGCTGCAGCT GTACTATCATATGCCTATATCCCTTCCC

PCR primers generate a 268-bp product from pUC19. The linker oligo is complementary to both the 5′ overhang generated by BamHI digestion and the 3′ end of the ssDNA substrates (underlined). For ssDNA substrates, the 27^th (X) and 28th (Y) bases (italic) are both C (undamaged), one is replaced with an apurinic/apyrimidinic (AP) site (AP27 and AP28), both are replaced by an AP site (AP27AP28), or both are replaced with a triethylene glycol spacer (SP2728).

Atomic force microscopy imaging and analysis

Varying concentrations of EcSSB were added to DNA substrates diluted to a concentration of 1 nM in a buffer containing 145 mM NaCl, 5 mM NaOH, 100 μM spermidine, and 10 mM HEPES (pH 7.5). Samples were incubated for 5 min at 37°C, then 5 μL of solution was deposited on a freshly cleaved mica surface. After 1 min of deposition time, the mica was washed with an excess of deionized (DI) water and then dried with argon gas. The sample was imaged with a MultiMode 8 atomic force microscope and Nanoscope V controller (Bruker) using peak force tapping mode and analyzed using Gwyddion software (version 2.55). For each condition measured (specific ssDNA substrate and EcSSB concentration), three or more biological replicates of EcSSB-ssDNA incubation were performed. For each incubation, the surface was imaged in multiple locations to observe a large number of ssDNA substrates. Although the exact number of substrates in each image frame naturally varies, an average of 408 substrates (ranging from 219 to 558) were imaged per condition. SE of the mean (error bars in plots) was calculated based on deviations in average binding fraction and protein volume per distinct incubation.

Results

Measuring EcSSB-ssDNA binding

The binding of unlabeled protein to unlabeled, short ssDNA substrates is difficult to detect at a single-molecule level by many popular methods. For atomic force microscopy (AFM) imaging specifically, the large size of the EcSSB tetramer (75.5 kDa) obscures the presence of any tightly bound ssDNA substrate small enough to be fully bound by the protein (∼65 nt or ∼20 kDa). Thus, free proteins and proteins bound by unlabeled oligos are nearly indistinguishable, preventing accurate measurement of binding affinities. In this study, we utilize a ds/ssDNA hybrid (36) where the ssDNA substrate of interest is ligated to the end of a dsDNA marker (Fig. 1 A). The rigidity of dsDNA both prevents its binding to EcSSB directly and increases its visibility in AFM imaging, where the 260-bp marker appears as an 85-nm-long line. Additionally, the dsDNA region controls the hybrid construct’s migration through a gel, ensuring unligated ssDNA, which migrates further through the gel, is completely removed from the sample during purification. Thus, EcSSB tetramers bound to the ssDNA substrate will only colocalize with one end of the dsDNA marker (Fig. 1 B), enabling accurate numeration of ssDNA substrates with and without protein bound. This method allows for the ligation of any ssDNA substrate to the dsDNA marker and for this project we utilize a 67-nt sequence from M13 bacteriophage, which our lab has previously used in polymerase assays (37). The sequence has 54% GC content and contains limited secondary structure (no large stable hairpins) and can be considered representative of the mixed base composition, naturally occurring sequences with which EcSSB interacts. The length was chosen such that either one EcSSB tetramer can bind ∼67 nt or two tetramers can bind ∼33.5 nt each, as the exact binding site sizes of the two conformations have been determined to be 35 ± 2 nt and 65 ± 3 nt (12,38,39). We intentionally chose a substrate on the smaller end of this range to limit protein shifting on the substrate (40,41), which could affect which nucleotides interact with specific protein residues.

Figure 1.

AFM imaging of EcSSB binding to ssDNA. (A) A ds/ss DNA hybrid is constructed by ligating a restriction enzyme digested dsDNA to a target ssDNA sequence using a linker oligo complementary to both. The end product contains 260 bp of dsDNA and a 67-nt 5′ ssDNA overhang. The constructs are incubated with varying concentrations of EcSSB, enabling binding specifically to the ssDNA end. (B) DNA/protein solutions are imaged using AFM. Colocalization of the EcSSB tetramer (white spots) with the end of the dsDNA marker (red lines) indicates bound ssDNA substrates. (C) Increased EcSSB concentration results in a greater fraction of substrates bound. (D) The fraction of ssDNA substrates bound as a function of EcSSB concentration is well fitted (reduced χ² ≈ 1) by a simple binding isotherm (Eq. 1, dotted line). Error bars are mean ± SE for N ≥ 3 biological replicates for all data shown. To see this figure in color, go online.

To measure the binding affinity of EcSSB to our ssDNA substrates, we incubated varying concentrations of EcSSB with a fixed 1 nM concentration of DNA in a 150 mM Na buffer, as moderate salt conditions allow EcSSB to bind ssDNA in both the 35- and 65-nt mode. Samples were incubated at 37°C for 5 min to ensure equilibrium binding, in accordance with previous kinetic measurements showing equilibration of ssDNA with 100 pM EcSSB occurring on a 100-s timescale (16). The number of ssDNA substrates bound or unbound by EcSSB are counted at each protein concentration (Fig. 1 D). As expected, the fraction of ssDNA that is bound by protein increases with EcSSB concentration (c), with a trend that can be well fitted as:

$$f\left(c\right)=\frac{c}{c+\left(K_{\mathrm{D}}\right)}$$ (1)

Here, K_D is an effective dissociation constant, the protein concentration at which half the substrates are bound. Note, the concentration used in this analysis is the concentration of free protein after the system equilibrates, not the concentration initially added to the incubation. The free protein concentration is calculated by subtracting the product of total ssDNA concentration (1 nM), the fraction of ssDNA bound (f), and the average bound state from the total protein concentration:

$$c_{free}=c_{total}-\left(\left[ssDNA\right]·f·⟨BoundState⟩\right)$$ (2)

The bound state is either one or two proteins per bound ssDNA, the average value of which is measured for each condition (discussed later). This simple binding isotherm fits the experimental data within experimental error (reduced χ² ≈ 1) with the ssDNA substrates transitioning from mostly unbound to mostly bound around a free protein concentration of 1 nM.

EcSSB binding to damaged ssDNA

The binding affinity experiment was repeated using different ssDNA substrates (Fig. 2 A), each the same length, but with the 27^th and/or the 28^th base (from the 5′ end) modified. When a single base was replaced with an AP site, we observed nearly identical EcSSB binding for these substrates (Fig. 2 B). The fraction of ssDNA bound is still well fitted by a simple binding isotherm with an effective K_D similar to that for binding to the undamaged ssDNA.

Figure 2.

EcSSB binding to damaged ssDNA. (A) Binding experiments are repeated for 67-nt ssDNA substrates with the 27^th and/or 28^th (magenta) bases modified. These C bases (undamaged) are replaced with either an abasic site (AP) or non-DNA spacer (SP). (B) When a single base is replaced by a stable abasic site (blue 27, green 28) both substrates are well fitted by Eq. 1 (dotted lines) and exhibit the same binding behavior as the undamaged DNA (black dotted line fitted from Fig. 1D). (C) Replacing both the 27^th and 28^th bases with two abasic sites (red) or one triethylene glycol spacer (yellow) results in a sharper transition between mostly unbound and mostly bound substrates. Comparing all five DNA substrates, the damage sites do not alter the apparent binding affinity K_D to a significant degree (D), but the substrates with two sites modified are no longer fitted by Eq. 1, as measured by reduced χ² (E). To see this figure in color, go online.

When both the 27^th and 28^th bases were replaced with AP sites, however, a different binding behavior was observed (Fig. 2 C). To further test specificity, we also utilized an ssDNA with both the 27^th and 28^th bases replaced by a triethylene glycol spacer (SP), which gave the same results. Although comparable levels of ssDNA binding are observed for protein concentrations above K_D (approaching saturation), the fractional binding at lower protein concentrations is reduced. As a result, we observe a sharp transition in binding, where the ssDNA goes from mostly free to mostly bound over a small EcSSB concentration increase. Thus, fitting with the simple binding isotherm again returns the same approximate K_D (Fig. 2 D) but is a poor fit to the data (reduced χ² >> 1; Fig. 2 E). One possibility is that the removal of the 27^th and 28^th C bases in the damaged constructs reduces the ability of the ssDNA to form secondary structure, and stable secondary structures would be expected to inhibit EcSSB binding. However, the opposite effect is observed, with less binding observed for the damaged constructs, suggesting the removal of ssDNA secondary structure is not the primary cause of the altered binding behavior. The more likely explanation is that this model does not take into account the multiple binding modes of EcSSB and especially the cooperativity associated with the simultaneous binding of two proteins. Thus, we must examine the binding stoichiometry of the ssDNA-EcSSB complexes to better understand this behavior.

EcSSB binding stoichiometry and cooperativity

AFM imaging provides additional information through volumetric measurement of objects. By integrating over all pixels associated with an object (multiplying pixel height by area and summing over all pixels), the volume sterically occupied by an object can be measured. Although the exact value is influenced by some conditions external to the measured object (particularly the size of the AFM tip itself), it has been shown that measured volumes of multiple proteins scale linearly with their molecular weights when calibrated with a common fixed marker, such as DNA (42). Correspondingly, we independently established this linear relationship using proteins studied in our lab (16,43,44) and a 500-bp DNA as a calibration marker (Fig. 3 A). We verified this method can distinguish between one and two bound EcSSB tetramers by incubating EcSSB with the undamaged ssDNA substrate in 20 and 300 mM Na⁺ buffer, where the 35- and 65-nt binding modes, respectively, are known to predominate (45). AFM images reveal the bound protein complexes to be noticeably larger in 20 mM Na⁺ (Fig. 3 B) than in 300 mM Na⁺ buffer (Fig. 3 C). By measuring the average volume of these protein clusters under both conditions and converting the volume to an estimated molecular mass using the same 500-bp DNA marker, we find that the protein-ssDNA complexes are in fact consistent with either one or two tetramers bound to the substrate (Fig. 3 D).

Figure 3.

Salt-dependent binding modes of EcSSB. (A) The integrated volume of a DNA (red) or proteins (blue) as measured by AFM is directly proportional to the known molecular weight of the constructs if the imaging tip and solution conditions are conserved. When high concentrations of EcSSB (10 nM) are incubated with the undamaged ssDNA at low (B) or high (C) salt concentrations, the size of the bound protein cluster on the substrate varies, with larger volumes at low salt. (D) The average sizes of the protein clusters under both conditions are converted to a measured molecular weight (blue bars), using a conversion factor determined by the apparent size of the 500-bp DNA construct measured under the same conditions (black dashed line in A). Comparing these values to the molecular weight of the ssDNA substrate (20.7 kDa) with either one (red) or two (green) EcSSB tetramers bound (75.5 kDa each) confirms that one EcSSB (65-nt mode) binds the ssDNA at high salt and two EcSSB (35 mode) bind the ssDNA at low salt. To see this figure in color, go online.

We measured the average protein volume for all observed EcSSB concentrations and used this conversion process to determine the average binding stoichiometry. For the undamaged ssDNA, only one tetramer is present for bound substrates when the concentration of ssDNA is equal to or exceeds the concentration of EcSSB tetramers (Fig. 4 A). As more EcSSB is added to the system, however, more ssDNA substrates are bound by two tetramers. At the highest EcSSB concentration measured (10 nM), the average protein cluster size is consistent with all substrates being bound by two tetramers. A similar transition has been observed at moderate salt concentrations using labeled ssDNA in a fluorescence resonance energy transfer (FRET) assay (45). In contrast, for the damaged ssDNA substrates harboring two tandem abasic sites or the spacer, we observe mostly volumes consistent with two EcSSB tetramers bound to ssDNA even at low EcSSB concentration (Fig. 4 B). Even when the ssDNA and EcSSB tetramers are at equimolar concentration (1 nM), we observe more substrates bound by two tetramers than by one; most substrates are protein free, as shown in Fig. 2 C, which explains how there is enough EcSSB in the system to achieve this stoichiometry. Thus, in both the measurements of ssDNA-binding fraction (Fig. 2) and EcSSB stoichiometry (Fig. 4), we observe similar binding behavior at high EcSSB concentrations, where binding of two tetramers in the 35 mode predominates, regardless of DNA damage, but at low EcSSB concentrations the otherwise preferred binding of one tetramer in the 65 mode is inhibited by the tandem DNA damage sites.

Figure 4.

Bound EcSSB stoichiometry. (A) For each protein-bound ssDNA substrate imaged, the number of tetramers present is calculated using volumetric methods. The average binding stoichiometry is plotted as a function free EcSSB concentration, showing that ssDNA is typically bound by one tetramer at low concentrations and two tetramers at high concentrations. A sigmoidal function (dashed line) is plotted as a guide to the eye. (B) Compared to undamaged DNA (same dashed line as in A), the tandem damage site constructs show an increase in ssDNA bound by two tetramers even at low protein concentration. To see this figure in color, go online.

Discussion

Multimode model of EcSSB binding conformation

Due to EcSSB’s multiple binding modes and the ability of the ssDNA to accommodate up to two tetramers, each substrate can be in one of four conformations: unbound, bound by one tetramer in the 35 or 65 wrapping mode, or bound by two tetramers both in the 35 mode (Fig. 5 A). Transitions between these states occur when a tetramer binds or dissociates from a substrate or when an already bound tetramer swaps its binding conformation. A model that describes the occupancy of each state, accounting for cooperative binding between neighboring proteins in the 35 mode, has been previously developed (15). The average number of tetramers bound per substrate (ν) can be solved for in terms of effective association constants (K₃₅ and K₆₅) and cooperativity parameter (ω₃₅)

$$\nu =\frac{\left(S_1{K}_{35}+K_{65}\right)P_f+2{\omega }_{35}{\left(K_{35}{P}_f\right)}^2}{1+\left(S_1{K}_{35}+K_{65}\right)P_f+{\omega }_{35}{\left(K_{35}{P}_f\right)}^2}$$ (3)

Figure 5.

Multistate binding measurements and model. (A) Cartoon showing the four possible states of the ssDNA substrate and potential transitions. (B) The average number of EcSSB tetramers bound per ssDNA substrate is calculated by multiplying the fraction of substrates bound (Fig. 1D) by the average binding stoichiometry (Fig. 4A). The data are fitted using Eq. 3 (15), with the best-fit parameters displayed. (C) The average number of EcSSB tetramers per damaged ssDNA substrate shows a sharper transition, which is not fully captured by Eq. 3 (reduced χ² > 1) even with K₆₅ reduced to 0, particularly in the region where equimolar concentrations of ssDNA and EcSSB tetramers are present (∼1 nM). (D) Structure of EcSSB tetramer (12), with individual subunits in yellow and green, bound to two 35-mer oligos (cyan), which was used to model EcSSB wrapping modes (12). The 27^th and 28^th nucleotide (red) stack between residues Trp40 and Phe60 of two adjacent subunits when ssDNA binds in the 65-nt mode. (E) Cartoon showing the EcSSB 65-nt binding mode, in which the ssDNA substrate fully occupies all four OB folds on one tetramer (left), and the 35-nt binding mode, where the substrate fully occupies one OB fold and partially occupies two OB folds on two tetramers (right). The 27 and 28 nt of the substrate strongly intact with protein in the 65-nt conformation but are located between strong interaction sites in the 35-nt conformation. To see this figure in color, go online.

Note that S₁ is a unitless statistical factor enumerating the number of exact sites where the tetramer can bind based on the excess length of substrate relative to the binding site size (S₁ = 67 − 35 + 1 = 33 for this system).

We apply this quantitative model to our data (Fig. 5 B). The average number of proteins bound per ssDNA substrate is calculated by multiplying the fraction of DNA substrates that are bound by protein (Figs. 1 and 2) by the average number of EcSSB tetramers for each bound substrate (Fig. 4). Note that this is also equivalent to calculating the average number of proteins per substrate using the average protein volume for all substrates where the protein volume for unbound substrates is zero. For undamaged DNA, this value increases gradually, first approaching one tetramer per ssDNA as EcSSB concentration surpasses the effective K_D and then approaching two tetramers per ssDNA at higher concentrations.

Eq. 3 fits the data, although the exact fitting parameters are not well defined (a large range of values fit within error and reduced χ² < 1). In particular, since the values of K₃₅ and ω₃₅ are multiplied together for the two squared terms, their values are directly dependent on that of the other. We set the minimum value for the cooperativity factor (ω₃₅ = 10⁵) from the original work (15), which used comparable conditions (69-nt poly(dA) substrate, 125 mM NaCl, pH 8.1, 25°C versus our conditions of 67-nt mixed-base substrate, 145 mM NaCl + 5 mM NaOH, pH 7.5, 37°C), and found best-fit parameters of K₆₅ = (1.22 ± 0.67) × 10⁸ M⁻¹ and K₃₅ = (9.96 ± 1.01) × 10⁵ M⁻¹ for ω₃₅ = 2 × 10⁵ with a reduced χ² value of 0.762. A similar trend was observed in the original work with the association constant for the 65 state multiple orders of magnitude larger than the 35 state (K₆₅ = 1.6 × 10⁸ M⁻¹ and K₃₅ = 1.6 × 10⁵ M⁻¹) (15). Thus, although our assay does not directly measure EcSSB wrapping conformation, this model predicts that singly bound tetramers predominately occupy the 65 state, and the 35 state is primarily observed when two tetramers bind the same substrate. This is consistent with published results from a FRET assay with the termini of the substrate labeled that directly detects ssDNA conformation (45).

For the ssDNA substrates with two damage sites, a different EcSSB binding response is observed (Fig. 5 C). Compared to the undamaged substrate, significantly less binding is observed in the ∼1:1 ssDNA to protein regime, where the substrate should be predominately bound by tetramers in the 65-nt state. In contrast, full binding is recovered at high EcSSB concentrations where more substrates should be bound by two tetramers, each in the 35-nt state. As a result, a sharper transition from ∼0 tetramers per substrate at low EcSSB concentrations to ∼2 tetramers per substrate at high EcSSB concentrations is observed, with a narrower concentration range of EcSSB showing binding of ∼1 tetramer per substrate. The simplest explanation for these features is that the binding affinity associated with the non-cooperative 65 state is reduced, whereas the ability of EcSSB to bind in the cooperative 35 state is unchanged or even slightly enhanced. Altering the parameters of Eq. 3 accordingly (reducing K₆₅) can improve the fit, although least-squares fitting returns a non-physical value of K₆₅ < 0. Instead, if we constrain K₆₅ to non-negative values, the best-fit values for K₃₅ are not significantly different than the undamaged substrate. However, these best fits still produce χ² > 1, suggesting there may be more complex effects occurring that cannot be described by Eq. 3.

Biological implications of damage site-directed binding

The binding conformations of EcSSB are denoted the 65- and 35-nt state in reference to their total binding site size (i.e., the occluded length of substrate to which other EcSSB tetramers cannot bind). Not all nucleotides along this length interact with the protein to the same degree, however, and both structures and models of ssDNA bound to EcSSB show some nucleotides are in direct contact with the EcSSB OB folds (12), whereas others span the distances between sites as the ssDNA wraps around the tetramer. As such, the degree to which EcSSB binding is altered by modified bases likely depends on their exact position. On a substrate significantly longer than its binding site size, an EcSSB tetramer could slide via reptation (40,41), changing the specific nucleotides interacting with specific amino acids. For this reason, we utilized a 67-nt substrate in this study, which restricts the binding of a protein in the single 65-nt state or two proteins in the 35-nt state with minimal sliding. A structure based on the model of the 65-nt state (Fig. 5 D) shows that the 27^th and 28^th base (from the 5′ end) of a bound ssDNA (12) are stacked between a phenylalanine (residue 60) of one EcSSB subunit and a tryptophan (residue 40) of another subunit. When both these nucleotides are replaced with abasic sites or a non-DNA linker, this interaction is lost (Fig. 5 E), which is likely responsible for the reduced binding observed under conditions in which non-cooperative binding of the 65 mode should predominate due to the near equimolar ratio of ssDNA and EcSSB tetramers.

Another recent study has also investigated whether modulating a fixed length ssDNA substrate can affect EcSSB binding modes by reversing the polarity of the ssDNA backbone (46). When a single reverse-polarity phosphodiester linkage was inserted into the middle of a 70-nt poly dT substrate, EcSSB continued to bind with high affinity. However, reversing the polarity between every nucleotide prohibited binding in the 65 mode and the cooperativity of the 35 mode was greatly reduced. It was proposed that, to accommodate this modified substrate and bind stably, the ssDNA follows a unique path around the EcSSB tetramer. It is possible that a similar process occurs due to the presence of abasic sites and non-DNA spacers in our assays. That is, in addition to or instead of these damage sites modulating the affinities of the canonical wrapping states, a modified ssDNA-EcSSB complex may be formed that is responsible for the unique binding response we observe.

EcSSB functions in vivo by binding variable-length segments of ssDNA, such as Okazaki fragments, which grow and shrink in length during DNA replication. EcSSB’s high local concentration (47) and binding affinity result in complete saturation of exposed ssDNA, with the total number of proteins equal to the substrate length divided by the average binding site size. Thus, although a single protein could bind in many positions along a long substrate, the full protein lattice ensures that all nucleotides along the length are in close proximity to at least one EcSSB. However, as our results show, the exact binding pattern could be altered by the presence of DNA damage. The ssDNA could remain saturated with protein (i.e., there is no contiguous length of protein-free ssDNA long enough to accommodate an additional tetramer), but the damaged nucleotide(s) could reside either between neighboring tetramers or between OB folds within a single tetramer, rather than being tightly bound. Further studies are needed to determine if the exact positioning of EcSSB can affect the ability to recruit repair proteins, but it is plausible that certain nucleotides may be more accessible when not tightly held by the EcSSB OB fold.

Conclusions

We have shown that the presence of modified bases that mimic naturally occurring DNA damage can alter the binding conformation of EcSSB without preventing protein saturation. Our results are consistent with EcSSB maximizing direct contact between undamaged DNA and its OB fold domains, leaving sites of DNA damage less tightly bound. Although previous studies have shown that EcSSB binding conformation can be modulated in vitro by changing conditions such as salt concentration and temperature that do not change dramatically in vivo, the accumulation of DNA damage is a plausible mechanism to alter EcSSB binding conformation under normal cellular conditions.

Author contributions

M.M., M.C.W., and P.J.B. designed the research. M.M., F.N.R., and M.K. performed experiments. M.M. and F.N.R. analyzed data. M.M. and J.M. produced experimental materials. M.M., F.N.R., and P.J.B. wrote the original manuscript. All authors reviewed and revised the manuscript.

Editor: Jason Kahn.