Probing E. coli SSB protein-DNA topology by reversing DNA backbone polarity

Abstract

Escherichia coli single-strand (ss) DNA binding protein (SSB) is an essential protein that binds ssDNA intermediates formed during genome maintenance. SSB homotetramers bind ssDNA in two major modes, differing in occluded site size and cooperativity. The (SSB)₃₅ mode in which ssDNA wraps, on average, around two subunits is favored at low [NaCl] and high SSB/DNA ratios and displays high unlimited, nearest-neighbor cooperativity forming long protein clusters. The (SSB)₆₅ mode, in which ssDNA wraps completely around four subunits of the tetramer, is favored at higher [NaCl] (>200 mM) and displays limited low cooperativity. Crystal structures of E. coli SSB and Plasmodium falciparum SSB show ssDNA bound to the SSB subunits (OB folds) with opposite polarities of the sugar phosphate backbones. To investigate whether SSB subunits show a polarity preference for binding ssDNA, we examined EcSSB and PfSSB binding to a series of (dT)₇₀ constructs in which the backbone polarity was switched in the middle of the DNA by incorporating a reverse-polarity (RP) phosphodiester linkage, either 3′-3′ or 5′-5′. We find only minor effects on the DNA binding properties for these RP constructs, although (dT)₇₀ with a 3′-3′ polarity switch shows decreased affinity for EcSSB in the (SSB)₆₅ mode and lower cooperativity in the (SSB)₃₅ mode. However, (dT)₇₀ in which every phosphodiester linkage is reversed does not form a completely wrapped (SSB)₆₅ mode but, rather, binds EcSSB in the (SSB)₃₅ mode with little cooperativity. In contrast, PfSSB, which binds ssDNA only in an (SSB)₆₅ mode and with opposite backbone polarity and different topology, shows little effect of backbone polarity on its DNA binding properties. We present structural models suggesting that strict backbone polarity can be maintained for ssDNA binding to the individual OB folds if there is a change in ssDNA wrapping topology of the RP ssDNA.

Significance

Single-stranded DNA (ssDNA) binding (SSB) proteins are essential for genome maintenance. Usually homotetrameric, bacterial SSBs bind ssDNA in multiple modes, one of which involves wrapping 65 nucleotides of ssDNA around all four subunits. Crystal structures of Escherichia coli and Plasmodium falciparum SSB-ssDNA complexes show ssDNA bound with different backbone polarity orientations, raising the question of whether these SSBs maintain strict backbone polarity in binding ssDNA. We show that both E. coli and P. falciparum SSBs can still form high-affinity, fully wrapped complexes with non-natural DNA containing internal reversals of the backbone polarity. These results suggest that both proteins maintain a strict backbone polarity preference but adopt an alternate ssDNA wrapping topology.

Introduction

Single-stranded (ss) DNA binding proteins (SSBs) are essential in all kingdoms of life and function by selectively binding to the ssDNA intermediates that are formed during genome maintenance, protecting them from degradation and inhibiting DNA secondary structures (1, 2, 3, 4). Escherichia coli SSB (EcSSB) also serves as a central hub for interaction with numerous metabolic proteins (SSB-interacting proteins (SIPs)) involved in replication, recombination, and repair (5).

EcSSB functions as a homotetramer (Fig. 1 B; (3,6)), with each subunit (177 amino acids (aa)) composed of two domains (Fig. 1 A): a structured N-terminal DNA binding domain (residues 1–112) and a C-terminal domain (residues 113–177) composed of a flexible, intrinsically disordered linker (IDL) (56 aa) (Fig. 1 A) and a nine residue “acidic tip.” This acidic tip is conserved among many bacterial SSBs and is the primary site of interaction with the SIPs (5,7, 8, 9, 10, 11, 12, 13). EcSSB binds ssDNA in two major modes, referred to as (SSB)₃₅ and (SSB)₆₅, in which the subscripts denote the average number of ssDNA nucleotides occluded (14,15). The relative stabilities of these modes depend primarily on salt concentration and type and protein/DNA ratio (binding density) (14,16, 17, 18, 19, 20, 21, 22, 23, 24), as well as applied force (25, 26, 27, 28).

Figure 1.

E. coli SSB-ssDNA structure and binding modes and its comparison with other SSB-DNA structures. (A) EcSSB (i) subunit (177 aa) and PfSSB (ii) subunit (210 aa) are composed of an N-terminal DNA binding domain (OB fold) (112 aa and 119 aa, respectively), intrinsically disordered linkers (IDL), 56 aa and 82 aa long, respectively, and 9 aa acidic tips, with different aa sequences. (iii) Reverse-polarity (RP) (dT)₇₀ constructs used in the study in which the polarity of phosphodiester backbone was reversed in the middle of the DNA (abbreviated as 5′-(dT)₇₀-5′ (green) and 3′-(dT)₇₀-3′ (dark orange)) are shown. (B) The cartoon shows EcSSB-ssDNA complex in its (SSB)₆₅ binding mode (6), with 65 nts of DNA (orange ribbon) wrapped around the tetramer, showing the polarity of DNA wrapping. The identical subunits are marked as AA′BB′. Note that the IDL and acidic tip are not visible in this structure. (C) The cartoon shows a proposed model for EcSSB-ssDNA binding to (dT)₇₀ (orange tube) in the (SSB)₃₅ binding mode using an average of only two subunits per tetramer. (D) The cartoon shows a PfSSB in complex with two molecules of (dT)₃₅ (orange tubes) showing a polarity for ssDNA binding that is opposite to EcSSB (41). (E) The cartoon shows HpSSB-ssDNA complex in the (SSB)₆₅ binding mode (42). (F) The cartoon shows PaSSB in complex with two molecules of (dT)₂₅ (44). To see this figure in color, go online.

In the (SSB)₆₅ mode, favored at [NaCl] > 0.20 M or [Mg²⁺] > 10 mM at 25°C, the ssDNA wraps around all four subunits of the tetramer (6) (see Fig. 1 B), with an ∼65 nucleotide-occluded site size. The topology of ssDNA wrapping in the (SSB)₆₅ binding mode follows the seams on a baseball, such that ssDNA enters and exits the tetramer in close proximity. On long ssDNA, the (SSB)₆₅ mode displays “limited” cooperativity between adjacent tetramers (16,24,29), as depicted in Fig. 1 B. In this mode, SSB can diffuse along ssDNA, transiently destabilizing DNA hairpins and promoting RecA filament formation (25,30). The ssDNA translocases are able to actively push SSB tetramers bound in the (SSB)₆₅ mode along ssDNA, providing a potential mechanism for reorganization and clearance of tightly bound SSBs from ssDNA (31). In the (SSB)₆₅ mode, non-nearest-neighbor SSB tetramers can interact cooperatively, possibly through IDLs (23,32). These interactions result in condensation of nucleoprotein complexes and require presence of IDL and are promoted by glutamate and acetate salts (23,28,32).

In the (SSB)₃₅ mode, favored at [NaCl] < 10 mM (Fig. 1 C) or [MgCl₂] < 1 mM, and high SSB/DNA ratios (14,15,18), ssDNA wraps around only two subunits, on average, with an ∼35 nucleotide-occluded site size. In this mode, SSB binds ssDNA with unlimited nearest-neighbor cooperativity, favoring formation of long protein clusters (17,18,20,22,24,32,33), as depicted in Fig. 1 C. Structural models for the (SSB)₃₅ binding mode have been proposed, suggesting direct interactions of adjacent tetramers through the L₄₅ loops within the tetrameric core of the protein (Fig. 1 C; (6)). In this mode SSB can diffuse along ssDNA (25,26) and undergo direct or intersegment transfer between separate ssDNA molecules (34) or between distant sites on the same DNA molecule (35). The ability to undergo direct transfer appears to play a role in SSB recycling during replication (34,36). Nearest-neighbor cooperativity can also be displayed between the different SSB binding modes (24). A recent study (37) identified residues Y22 and K73 as important for cooperative interactions between adjacent tetramers. Both the linker and the conserved tip of the C-terminus are also important for cooperativity (23,24,32).

Most bacterial SSBs are homotetramers (38). Although the dimeric Deinococcus radiodurans SSB (DrSSB) is an exception (39,40), it still contains four OB folds (two OB folds per monomer). In addition to E. coli, several other homotetrameric SSB-ssDNA crystal structures have been reported, including Plasmodium falciparum (PfSSB(41)), Helicobacter pylori (HpSSB (42)), Bacillus subtilis (BsSSB-B (43) and BsSSB-A (37)), and Pseudomonas aeruginosa (PaSSB (44)) SSB. All but one of these structures shows DNA bound to the OB folds with the same phosphodiester backbone polarity orientation (following the path across the β-sheet in the direction from the 2-3 loop toward the 1-2 loop). This is exemplified by PfSSB and HpSSB in Fig. 1, D and E. The one exception is the EcSSB-DNA structure that displays the opposite 5′-3′ polarity to each OB fold (Fig. 1 B). It is interesting to note that the predicted topology of ssDNA wrapping in the (SSB)₆₅ binding mode also differs for EcSSB when compared with PfSSB and HpSSB. A crystal structure of the EcSSB tetramer bound with two (dC)₃₅ molecules (6) predicts an ssDNA binding topology that wraps around the OB folds of the upper AA′ dimer (first (dC)₃₅ molecule), then crosses the dimer-dimer interface in the direction of the B subunit α-helix, and finally wraps around the OB folds of the BB′ dimer (second (dC)₃₅ molecule), as shown in Fig. 1 B (suggested path is AA′BB′). In a PfSSB structure (41), the first and the second molecule of (dT)₃₅ are wrapped around the upper and lower dimer, similar to EcSSB (though with different polarity). However, crossing of the dimer-dimer interface occurs in the direction of the B′ subunit following a path under the α-helix of the A′ subunit, as shown in Fig. 1 D (suggested path is AA′B ′B). Essentially, the same path is predicted for the (SSB)₆₅ mode of HpSSB (Fig. 1 E; (42)) and BsSSB-B (43). Interestingly, a recent structure of PaSSB co-crystallized with (dT)₂₅ suggests that individual subunits could accommodate ssDNA with opposite polarity (e.g., (dT)₂₅, shown in yellow in Fig. 1 F, binds to subunit A with 5′-3′ polarity, whereas it binds to subunit B′ with 3′-5′ polarity).

Based on available SSB-DNA structural data, it is reasonable to expect that the polarity of ssDNA binding might affect the SSB wrapping topology and the binding properties. In this study, we examined the effects of ssDNA backbone polarity for EcSSB and PfSSB for two reasons. First, crystallographic studies suggest that they bind ssDNA with opposite polarities and slightly different topologies. Second, their ssDNA binding properties have been extensively studied in solution. In particular, it was established that PfSSB forms an (SSB)₆₅ binding mode similar to that of EcSSB, yet PfSSB does not form a stable (SSB)₃₅ binding mode (45). Yet, both proteins can diffuse on ssDNA in their fully wrapped modes and effectively melt out short hairpins (25,30,45). Here, we examine the binding of these proteins to a variety of oligodeoxythymidylates (dT₇₀) that contain a reversal of the phosphodiester backbone polarity in the middle of the DNA molecule (after residue 35), either through a 3′-3′ or 5′-5′ linkage. We also examine the binding of ap-(dT)₇₀ and ap-(dT)₃₅ DNA in which the polarity alternates after every residue.

Materials and methods

Reagents and buffers

Buffers were prepared with reagent grade chemicals, and distilled water was treated with a Milli-Q (Millipore, Bedford, MA) water purification system. Buffer T is 10 mM Tris (pH 8.1) and 0.1 mM Na₃EDTA; buffer H is 10 mM Hepes (pH 8.1) and 0.1 mM Na₃EDTA.

SSB proteins and ssDNA

EcSSB protein was expressed and purified as previously described (32). The SSB protein is a stable tetramer under all solution conditions used in this study, as determined by sedimentation velocity (32). Protein concentration was determined spectrophotometrically (14) (buffer T, 0.20 M NaCl) using ε₂₈₀ = 1.13 × 10⁵ M⁻¹ cm⁻¹ (tetramer). PfSSB was expressed and purified as previously described (41), and its concentration was determined using ε₂₈₀ = 9.58 × 10⁴ M⁻¹ cm⁻¹ (tetramer) (41,45).

Oligodeoxythymidylates possessing 70 and 35 nucleotides were used in this study ((dT)₇₀ and (dT)₃₅, respectively). Along with DNA possessing normal 5′-3′ polarity, we also examined two variations of (dT)₇₀ (Fig. 1 Aiii), in which the backbone polarity was reversed in the middle of the sequence: 5′-(dT)₃₅-3′-3′-(dT)₃₅-5′ (abbreviated as 5′-(dT)₇₀-5′) and 3′-(dT)₃₅-5′-5′-(dT)₃₅-3′ (abbreviated as 3′-(dT)₇₀-3′). Two DNA molecules, termed ap-(dT)₇₀ and ap-(dT)₃₅, were synthesized such that the backbone polarity is reversed after each residue. A similar set of (dT)₇₀ molecules were synthesized that were doubly labeled at the ends with Cy5 and Cy3 fluorescence dyes: (5′-Cy5-(dT)₇₀-Cy3-3′, 5′-Cy5-(dT)₃₅-3′-3′-(dT)₃₅-Cy3-5′, 3′-Cy5-(dT)₃₅-5′-5′-(dT)₃₅-Cy3-3′, and 5′-Cy5-ap-(dT)₇₀-Cy3-3′). All oligonucleotides were synthesized and purified to >99% homogeneity, as previously described (20,46). DNA concentrations were determined spectrophotometrically in buffer T plus 0.10 M NaCl, using ε₂₆₀ = 8.1 × 10³ M⁻¹ (nucleotide) cm⁻¹ for all unlabeled (dT)₇₀ and (dT)₃₅ constructs (47) (70 × ε₂₆₀ and 35 × ε₂₆₀, molecule). For all Cy5-(dT)₇₀-Cy3, the extinction coefficient ε₂₆₀ = 5.82 × 10⁵ M⁻¹ (molecule) was used (24,46).

Fluorescence equilibrium titrations

Fluorescence titrations were performed in buffers T and H, at 25°C and at the NaCl and NaBr concentrations indicated in the text and figure legends, using a Spectrofluorometer QM4 (Photon Technology International, HORIBA Scientific, Edison, NJ). Reverse titrations of SSB (0.3 μM) with (dT)₇₀ constructs (Figs. 2, 4, and 6) and normal titration of SSB (0.3 μM) with ap-(dT)₃₅ (Fig. S6) were performed by monitoring quenching of the intrinsic SSB tryptophan fluorescence (λ_exc = 296 nm and λ_em = 350 nm) and analyzed as previously described (24,32,48). Normal titrations of Cy5-(dT)₇₀-Cy3 constructs (0.1 μM) with SSB (Fig. 2, 3, and 6; Fig. S3) were performed by exciting the Cy3 donor (515 nm) while monitoring sensitized emission from the Cy5 acceptor at 665 nm and analyzed as previously described (21,24,48). The observed fluorescence emission from the Cy5 acceptor is due almost exclusively to fluorescence resonance energy transfer (FRET) from the Cy3 donor, with minimal environmental effects on the Cy5 fluorescence due to SSB binding (see Fig. S1). This contrasts with the substantial increase in Cy3 fluorescence (PIFE effect) that we have shown occurs upon SSB binding to DNA constructs labeled with the Cy3/Cy5 FRET pair (21,24,46).

Figure 2.

EcSSB can form (SSB)₃₅ and (SSB)₆₅ modes on reverse-polarity (dT)₇₀. (A) and (B) show results of equilibrium titrations of EcSSB (0.30 μM) with (dT)₇₀ (cyan), 3′-(dT)₇₀-3′ (orange), and 5′-(dT)₇₀-5′ (green) (ex: 296 nm; em: 350 nm) in 10 mM NaCl and 0.30 M NaCl (buffer T, 25°C) and plotted as a fraction of Trp fluorescence quenching versus the ratio of total DNA/total SSB tetramer concentrations. The cartoons depict the (SSB)₃₅ and (SSB)₆₅ binding modes that form during the course of the titrations. (C) and (D) show results of equilibrium titrations of Cy3/Cy5-labeled (dT)₇₀ (cyan), 3′-(dT)₇₀-3′ (orange), and 5′-(dT)₇₀-5′ (green) (0.1 μM each) (see Fig. 1Aiii), with EcSSB monitoring Cy5 fluorescence enhancement (exc: 515 nm; em: 665 nm) in 10 mM NaCl and 0.30 M NaCl (buffer-T, 25°C). The cartoons depict the (SSB)₃₅ and (SSB)₆₅ binding modes and where they form during the course of the titrations. To see this figure in color, go online.

Figure 4.

Reverse-polarity (dT)₇₀ binds to EcSSB with different affinities but binds with the same affinity to PfSSB. Results are shown for equilibrium titrations of EcSSB (A) and PfSSB (B) (0.30 μM each) with (dT)₇₀ (cyan), 3′-(dT)₇₀-3′ (orange), and 5′-(dT)₇₀-5′ (green) monitoring intrinsic Trp fluorescence quenching (ex: 296 nm; em: 350 nm) in 2.0 M NaBr (buffer-T, 25°C). Solid lines are simulated isotherms based on the best-fit parameters shown that were obtained from an NLLS fit to a single site binding model (Eqs. 1 and 2 in Materials and methods). To see this figure in color, go online.

Figure 6.

EcSSB binding to alternating polarity (ap)-(dT)₇₀. (A) Results are shown for equilibrium titrations of EcSSB (0.30 μM) with ap-(dT)₇₀ in buffer-T (25°C) in 10 mM NaCl (green), 30 mM NaCl (orange), 100 mM NaCl (cyan), and 300 mM NaCl (dark green) and plotted as fractional Trp fluorescence quenching versus the ratio of total DNA/SSB tetramer concentrations. The cartoons depict the (SSB)₃₅ and (SSB)₆₅ modes and expected transitions during the titrations. (B) Results are shown for equilibrium titrations of Cy3/Cy5-labeled ap-(dT)₇₀ (0.1 μM each) with EcSSB in 10 mM NaCl (green), 30 mM NaCl (orange), 100 mM NaCl (cyan), and 300 mM NaCl (dark green) in buffer-T (25°C), plotted as Cy5 fluorescence versus the ratio of concentrations of total SSB tetramer/total DNA. The cartoons depict the (SSB)₃₅ and (SSB)₆₅ modes and expected transitions during the titrations. To see this figure in color, go online.

Figure 3.

EcSSB binding to reverse polarity (dT)₇₀ at intermediate [NaCl]. Results are shown for equilibrium titrations of Cy3/Cy5-labeled (dT)₇₀ (cyan), 3′-(dT)₇₀-3′ (orange), and 5′-(dT)₇₀-5′ (green) (0.1 μM each) with EcSSB monitoring Cy5 fluorescence enhancement (exc: 515 nm; em: 665 nm) in 40 mM NaCl (A), 80 mM NaCl (B), and 100 mM NaCl (C) (buffer-T, 25°C) and plotted as Cy5 fluorescence signal versus the ratio of concentrations of total SSB tetramer/total DNA. To see this figure in color, go online.

Binding isotherms in Fig. 4 for (dT)₇₀ (X) binding to SSB (M) to determine equilibrium binding constants (K) were analyzed using a single site binding model described in Eq. 1 (48),

$$Q=Q_{max}\frac{KX}{1+KX},$$ (1)

where Q = (F₀ − F_obs)/F₀ is normalized Trp fluorescence quenching, with F₀ representing the fluorescence intensity of SSB alone, and F_obs, the Trp fluorescence, was measured at each point in the titration; Q_max = (F_o − F_max)/F_o is the normalized fluorescent quenching at saturation. The concentration of the free DNA (X) in Eq. 1 was obtained by solving mass conservation Eq. 2:

$$X_{tot}=X+X_{bound}=X+\frac{KX}{1+KX}{M}_{tot},$$ (2)

where X_tot and M_tot represent total concentrations of (dT)₇₀ and SSB, respectively. Nonlinear least squares fitting of the isotherms to Eqs. 1 and 2 was performed as previously described (48), using Scientist (Micromath, St. Louis, MO).

Isothermal titration calorimetry

Isothermal titration calorimetry (ITC) titrations were performed in buffer H (25°C, 20 mM NaCl) with a MicroCal VP-ITC microcalorimeter (Malvern Panalytical, Westborough, MA). In these experiments the SSB (1 μM) was titrated with (dT)₇₀ DNA with stock solutions thoroughly dialyzed before use. The data were analyzed using software provided by the manufacturer, as previously described (49).

Stopped-flow kinetics

Fluorescence stopped-flow experiments were performed as previously described (46) in buffer H at 25°C in the presence of 2 M NaBr using an Applied Photophysics SX.18MV instrument (Applied Photophysics, Leatherhead, United Kingdom). In experiments shown in Fig. S5, the SSB at concentration 50 nM (final) was mixed with increasing concentrations of (dT)₇₀ constructs, and intrinsic Trp florescence (λ_exc = 296 nm) signal change was monitored at (λ > 340 nm) using a long-pass filter (catalog no. 51258; Oriel Corp., Stratford, CT). The observed rates, r_obs (s⁻¹), obtained fitting signal change to one-exponential decay function were then plotted as a function of DNA concentration (Fig. S5). Linear fit of these data produced k_on and k_off as slope and intercept, respectively.

Analytical sedimentation

Sedimentation velocity experiments were performed, as described previously (23,24,32), with an Optima XL-A analytical ultracentrifuge and an An-50 Ti Analytical Rotor (Beckman Coulter, Fullerton, CA) at 42,000 rpm in buffer T (25°C) with the salt concentrations and type indicated in the text and in the legends for Figs. S2 and S4. Constant concentrations of (dT)₇₀ constructs (0.5 μM) were used while adding SSB to make solutions with different protein/DNA ratios (r = [SSB]_tot/[(dT)₇₀]_tot). The absorbance was monitored at 260 nm, reflecting predominantly the contribution from ssDNA. The contribution of SSB to the absorbance at 260 nm is small compared with the DNA at the protein/DNA ratios (r ≤ 2) used in this study. Data were analyzed using SEDFIT (www.analyticalultracentrifugation.com) to obtain continuous sedimentation (c(s)) distributions and estimate molecular weights for the species described by single peaks (50). The c(s) refers to the c(s) distribution that results from a SEDFIT analysis of the absorbance traces obtained in a sedimentation velocity AUC experiment using the Lamm equation (50). C(s) represents the concentration (in absorbance units in this case) of species with a given sedimentation coefficient range.

The densities and viscosities at 25°C were calculated using SEDNTERP for NaCl T buffer solutions. Partial specific volumes for protein-DNA complexes were calculated using the following formula: υ = (υ_DNAM_DNA + nυ_nM_P)/(M_DNA + nM_P), where υ_DNA and M_DNA are the partial specific volume and molecular weight of DNA, υ_P and M_P are the partial specific volume and molecular weight of the protein, and n is the number of SSB tetramers bound to DNA. In all of our experiments, the absorbance estimated from integrated c(s) profiles was the same as the absorbance of loaded solutions, indicative that no precipitation/aggregation occurred during the experimental runs.

Results

Effect of a reversal of backbone ssDNA polarity on EcSSB and PfSSB binding mode formation and cooperativity

Depending on solution conditions, one (dT)₇₀ is able to bind either one SSB tetramer in its fully wrapped (SSB)₆₅ binding mode (65-nts-occluded site size) (Fig. 1 B) or two SSB tetramers in its (SSB)₃₅ binding mode with high cooperativity (Fig 1 C; (20,24,51)). The proposed structure of an (SSB)₆₅ complex in Fig. 1 B (6) shows that ∼35 nucleotides first wrap around the upper two SSB subunits, then DNA crosses the dimer-dimer interface, continuing to wrap another 35 nts around the other two subunits following the same path as for the upper subunits. If the SSB tetramer binds ssDNA in a manner that maintains a strict 3′-5′ backbone polarity, then a reversal of the backbone polarity in the middle of a (dT)₇₀ might weaken or even preclude formation of the fully wrapped (SSB)₆₅ binding mode. The ssDNA wrapping topology proposed for the (SSB)₃₅ mode, although more speculative (Fig. 1 C; (6)), also suggests that a reversal of the backbone polarity in the middle of a (dT)₇₀ might also affect the relative orientation of two tetramers bound in the (SSB)₃₅ mode and affect any cooperative interactions. To test these concepts, we designed three non-natural (dT)₇₀ variants (see Materials and methods). In two variants, the backbone polarity was reversed in the middle of the DNA by incorporating a 3′-3′ or 5′-5′ phosphodiester linkage; these are denoted as 5′-(dT)₇₀-5′ and 3′-(dT)₇₀-3′, respectively. The third variant alternated a 3′-3′ and 5′-5′ phosphodiester linkage after each nucleotide and is denoted ap-(dT)₇₀. DNA molecules containing backbone polarity reversals have been used to probe directional movement of DNA translocases and helicases (31,52, 53, 54, 55, 56, 57). Binding of these unlabeled DNA molecules was monitored by the quenching of the intrinsic Trp fluorescence of SSB (24,32,48). Other versions of these (dT)₇₀ variants were made, in which the ends of each DNA were labeled with the fluorophores Cy3 and Cy5, which can undergo FRET. Use of these labeled (dT)₇₀ molecules enabled us to monitor formation of the different SSB binding modes because a higher FRET efficiency is observed in the (SSB)₆₅ mode because of the closer proximity of the two ends of (dT)₇₀ in this mode, as demonstrated previously (21,24,48).

The results of equilibrium titrations of SSB with (dT)₇₀, 5′-(dT)₇₀-5′, and 3′-(dT)₇₀-3′ monitoring SSB Trp fluorescence quenching are shown in Fig. 2, A and B in buffer T at 25.0°C at 10 mM and 0.30 M [NaCl]. At 10 mM NaCl, both the (SSB)₆₅ and the (SSB)₃₅ binding modes can form as determined by the SSB/DNA ratio, whereas at 0.30 M NaCl, only the (SSB)₆₅ binding mode is formed (14,18,19,24,51). At 10 mM NaCl (Fig. 2 A) for all (dT)₇₀ constructs the normalized quenching increases linearly with addition of DNA, indicative of stoichiometric binding (all added DNA binds to SSB). The break point at a DNA/protein ratio of 0.5 (r = [SSB]_tot/[(dT)₇₀]_tot = 2) reflects the transition from the (SSB)₃₅ mode to the (SSB)₆₅ mode, the latter becoming solely populated at r = 1, as shown previously for (dT)₇₀ (24,51). However, we note that both RP DNA constructs show a higher extent of Trp quenching in the (SSB)₃₅ mode (Q₃₅ ∼ 0.6 for green and orange isotherms in Fig. 2 A at [(dT)₇₀)]_tot/[SSB]_tot = 0.5) compared with (dT)₇₀ (cyan), for which Q₃₅ ∼ 0.5 is expected (24,51). This might reflect a slightly different positioning of the two SSB tetramers on the RP DNAs. However, independent of polarity, no difference in quenching (Q₆₅ = 0.9) is observed for the 1:1 (SSB)₆₅ complexes that form at [(dT)₇₀)]_tot/[SSB]_tot > 1. At the higher salt concentration of 0.30 M NaCl (Fig. 2 B) all three titrations are identical, showing a linear increase in Trp quenching up to r = 1, indicative of stoichiometric binding exclusively in the (SSB)₆₅ mode (24,51).

We next performed titrations of the Cy3/Cy5 fluorescently labeled (dT)₇₀ constructs with SSB at the same two [NaCl]. The results are consistent with the Trp fluorescence titrations in Fig. 2, A and B. At 10 mM NaCl (Fig. 2 C), the Cy5 FRET signal increases linearly up to r = 1, reflecting formation of the fully wrapped (SSB)₆₅ complex and then decreases linearly up to r = 2, reflecting a transition to a 2:1 (SSB)₃₅ complex with increasing [SSB]. Interestingly, the Cy5 FRET signal is similar for all 1:1 complexes (F_Cy5,65 ∼3.5 × 10⁵ at r = 1), whereas the 2:1 complexes (r > 2) show different Cy5 FRET signals. SSB bound to 5′-(dT)₇₀-5′ shows a higher Cy5 fluorescence compared with 3′-(dT)₇₀-3′ and (dT)₇₀. This suggests some difference in the DNA wrapping in the (SSB)₃₅ mode with 5′-(dT)₇₀-5′. However, the sharpness of the transitions between binding modes at r = 1 argues that SSB still binds to both RP DNA constructs with a high nearest-neighbor cooperativity (24). However, all three titrations at 0.30 M NaCl (Fig. 2 D) are identical, indicating that they can all form a fully wrapped 1:1 (SSB)₆₅ complex in these conditions, consistent with the results in Fig. 2 B.

To obtain an independent check on our conclusion that both 1:1 and 2:1 SSB complexes can be formed on both RP (dT)₇₀ constructs at low [NaCl], we performed sedimentation velocity experiments at total concentration ratios of r = 1 and 2 SSB/DNA. Fig. S2 A shows the results plotted as c(s) distributions (24,50). All three DNA molecules sediment identically in the absence of SSB. At each SSB/DNA ratio, only a single c(s) peak is observed (s_20,w = 5.1 S at r = 1 and s_20,w = 6.7 S at r = 2). The sedimentation coefficients of the r = 1 and r = 2 complexes also are the same for each DNA-SSB complex, indicating that the change in RP does not lead to major differences in hydrodynamic properties. At the higher 0.30 M [NaCl] (Fig. S2 B) only one peak at s_20,w = 5.3S is observed at both r = 1 and r = 2, indicating that only a 1:1 (SSB)₆₅ complex forms at both SSB ratios.

We showed previously that PfSSB, binds ssDNA only in an (SSB)₆₅ binding mode independent of [NaCl] with a DNA topology similar to EcSSB but with a DNA backbone polarity opposite to that observed in the EcSSB-DNA structure (41,45). Using equilibrium titrations (Fig. S3) and sedimentation velocity (Fig. S4), we demonstrate that the backbone polarity of (dT)₇₀ has no effect on the formation of the 1:1 fully wrapped complexes, which form stoichiometrically both at low (10 mM) and moderate (0.30 M) NaCl concentrations (s_20,w = 4.8 S and s_20,w = 5.4 S, respectively, Fig. S4). However, a somewhat lower Cy5 fluorescence (FRET) characterizes formation of the PfSSB-3′-(dT)₇₀-3′ complex compared with the 5′-(dT)₇₀-5′ complex and (dT)₇₀ (Fig. S3). This suggests a slight difference in DNA wrapping in the PfSSB-3′-(dT)₇₀-3′ (SSB)₆₅ binding mode.

So far, we investigated the binding mode formation at NaCl concentrations, in which either the (SSB)₃₅ ([NaCl] ≤ 0.01 M) or the (SSB)₆₅ ([NaCl] ≥ 0.3 M) modes are formed predominantly. However, at intermediate [NaCl], both modes are expected to be populated (14,15). To monitor the transition between binding modes, we performed titrations of DNA constructs with EcSSB at 40, 80, and 100 mM NaCl (Fig. 3) and found that, whereas the binding of the first SSB tetramer (formation of 1:1 (SSB)₆₅ complex) is stoichiometric for all DNAs, the transition to the (SSB)₃₅ mode upon binding the second SSB tetramer is less favorable for 5′-(dT)₇₀-5′ (green circles), whereas the isotherms for binding 3′-(dT)₇₀-3′ and (dT)₇₀ are very similar. This difference can be due to either a higher affinity (K₆₅) of EcSSB to 5′-(dT)₇₀-5′ in the (SSB)₆₅ binding mode or a lower nearest-neighbor cooperativity (ω₃₅) in the (SSB)₃₅ binding mode (24).

To examine this further, we performed equilibrium titrations of EcSSB with the RP (dT)₇₀ DNA constructs (Fig. 4 A) in the presence of 2.0 M NaBr, which lowers the binding affinity of EcSSB to (dT)₇₀ into a range in which it can be measured accurately (34,51,58). We find that K₆₅ for SSB binding to (dT)₇₀ ((1.8 ± 0.1) × 10⁷ M⁻¹) and 3′-(dT)₇₀-3′ ((3.3 ± 0.1) × 10⁷ M⁻¹) are similar under these conditions, whereas binding to 5′-(dT)₇₀-5′ ((2.4 ± 0.1) × 10⁶ M⁻¹) is much lower. Additional stopped-flow experiments (Fig. S5 A) established that the decrease in affinity for 5′-(dT)₇₀-5′ is due to an ∼12-fold increase in the dissociation rate constant, k_-5′5′ = 13.1 ± 1.9 s⁻¹ vs. k_-3′3′ = 1.1 ± 0.1 s⁻¹ (the latter was estimated from the ratio of association rate constant (Fig. S5 A)/equilibrium binding constant (Fig. 4 A) because the value estimated from the intercept k_-3′3′ = −0.1 ± 0.7 s⁻¹ (Fig. S5 A) is not very accurate). We note that the same bimolecular association rate constant of ∼4 × 10⁷ M⁻¹s⁻¹ is measured for both DNA constructs (Fig. S5 A). The lower value of K₆₅ for the 5′-(dT)₇₀-5′ construct in 2.0 M NaBr suggests that the preference for the (SSB)₆₅ binding mode for the 5′-(dT)₇₀-5′ construct (see Fig. 2) results from a significantly decreased cooperativity in the (SSB)₃₅ mode, (ω₃₅).

In contrast, equilibrium binding of PfSSB under the same conditions (see Fig. 4 B) shows little difference in affinity for (dT)₇₀ (K₆₅ = (3.2 ± 0.2) × 10⁷ M⁻¹) and the two RP DNAs (K_{65, 5′5′} = (1.7 ± 0.1) × 10⁷ M⁻¹ vs. K_{65, 3′3′} = (4.0 ± 0.4) × 10⁷ M⁻¹). Furthermore, PfSSB (Fig. S5 B) shows only an ∼threefold difference in dissociation rate constants k_-5′5′ = 3.5 ± 0.5 s⁻¹ and k_-3′3′ = 1.1 ± 0.7 s⁻¹ and similar association rate constants (∼3 × 10⁷ M⁻¹ s⁻¹). We note that affinities obtained from these rate constants (K = k_on/k_off) agree well with the affinities determined from our equilibrium titration experiments (Fig. 4).

Despite the fact that the affinity in the (SSB)₆₅ binding mode might vary for the RP constructs, the most interesting result is that both EcSSB and PfSSB are still able to form fully wrapped complexes with the RP DNA variants under all conditions examined. Furthermore, these RP constructs are also able to form highly cooperative (SSB)₃₅ complexes. However, we note that quantification of the nearest-neighbor cooperativity parameters for the different DNA constructs is difficult because of the strong correlation of the binding parameters (K₆₅, K₃₅, and ω₃₅). We have recently performed such an analysis for wild type EcSSB binding to (dT)₇₀, but this required having reasonable estimates of either K₆₅ or K₃₅ (24). Such an analysis cannot be performed at this time because estimates of K₃₅ and K₆₅ are not available for SSB binding to 5′-(dT)₇₀-5′ and 3′-(dT)₇₀-3′.

We next performed ITC experiments to probe the thermodynamics of the EcSSB and PfSSB binding to the RP-(dT)₇₀ DNA in more detail. Fig. 5 shows the results of titrations of EcSSB (1 μM tetramer) with 5′-(dT)₇₀-5′, 3′-(dT)₇₀-3′, and (dT)₇₀ at low [NaCl] (buffer H, 20 mM NaCl). All EcSSB titrations (Fig. 5, A and B) show biphasic character, reflecting the initial formation of a 2:1 (SSB)₃₅ complex in the range of DNA/protein ratios ≤0.5, which then transforms to a 1:1 (SSB)₆₅ complex for DNA/protein ratios from 0.5 to 1.0. Because binding is stoichiometric for each DNA under these conditions, the normalized heats in the flat portions of the isotherms can be related to the enthalpies for formation of the binding modes as given in (3a), (3b) (49,59),

$${\text{Q}}_{\text{norm,0}-0.5}=\text{Δ}{\text{H}}_{2\text{x}35\text{,tot}}$$ (3a)

$${\text{Q}}_{\text{norm,}0.5-1.0}=2\text{Δ}{\text{H}}_{65}-\text{Δ}{\text{H}}_{2\text{x}35\text{,tot}},$$ (3b)

where ΔH_2x35,tot = 2ΔH₃₅ + ΔHω₃₅ represents overall enthalpy of (SSB)₃₅ binding mode formation (including an enthalpic contribution from cooperativity) and ΔH₆₅ is the enthalpy of SSB binding in (SSB)₆₅ binding mode. Inspection of Fig. 5, A and B indicates that EcSSB binding to 3′-(dT)₇₀-3′ is very similar to its binding to (dT)₇₀ (Fig. 5 B) and slightly more enthalpically favorable compared with binding to 5′-(dT)₇₀-5′ (Fig. 5 A). Solution of the system of equations ((3a), (3b)) enables us to quantify the differences in the binding enthalpies for the different binding modes formed on the different DNAs. For 5′-(dT)₇₀-5′, ΔH_2x35,tot = −142.3 ± 0.4 kcal/mol and ΔH₆₅ = −148.4 ± 1.6 kcal/mol, whereas for 3′-(dT)₇₀-3′ and (dT)₇₀, ΔH_2x35,tot = −152.3 ± 1.9 kcal/mol and ΔH₆₅ = −160.0 ± 2.4 kcal/mol. Hence, the interaction of EcSSB with 5′-(dT)₇₀-5′ is ∼10 and 12 kcal/mol less enthalpically favorable in the (SSB)₃₅ and (SSB)₆₅ binding modes, respectively. This somewhat less favorable binding to 5′-(dT)₇₀-5′ is consistent with the equilibrium titrations and stopped-flow results discussed above.

Figure 5.

Enthalpic contributions to binding of EcSSB and PfSSB to reverse-polarity (dT)₇₀. Results are shown for ITC titrations of (A) EcSSB with 5′-(dT)₇₀-5′ (green), (B) EcSSB with (dT)₇₀ and 3′-(dT)₇₀-3′ (cyan and orange), and (C) PfSSB with 3′-(dT)₇₀-3′ and 5′-(dT)₇₀-5′ (orange and green) in 20 mM NaCl (buffer H, 25°C) plotted as heats per injection normalized per amount of injected DNA (Q_norm, kcal/mol) versus the ratio of total DNA/total SSB tetramer concentration. The cartoons depict the (SSB)₃₅ and (SSB)₆₅ complexes and where they form during the course of the titrations. To see this figure in color, go online.

In contrast, the interaction of PfSSB with both RP (dT)₇₀ constructs at these low-salt conditions (Fig. 5 C) leads to formation of only the 1:1 (SSB)₆₅ complex, consistent with previous studies showing that PfSSB does not form an (SSB)₃₅ complex (45). PfSSB binds to DNA molecules with similar enthalpies of ΔH_65,5′5′ = −152.3 ± 2.3 kcal/mol and ΔH_65,3′3′ = −156.2 ± 1.4 kcal/mol.

SSB binding to oligodeoxythymidylates with alternating polarity (ap-(dT)₇₀ and ap-(dT)₃₅)

We next determined whether SSB is able to bind to ssDNA in which the backbone polarity is reversed after each nucleotide. We refer to these as ap-(dT)₇₀, ap-(dT)₃₅, and Cy5-ap-(dT)₇₀-Cy3 DNA (see Materials and methods). The results are presented in Fig. 6, A and B at [NaCl] in the range from 10 mM to 0.30 M. Titrations at 10, 30, and 100 mM NaCl show transition points at DNA/protein ratios of 0.5 and 1.0 (Fig. 6 A), indicating that both 2:1 (SSB)₃₅ and 1:1 (SSB)₆₅ modes can form, whereas, only a 1:1 (SSB)₆₅ complex is formed in 0.30 M NaCl. However, we note that the normalized Trp quenching only reaches values in the range Q_obs = 0.62–0.70 ([DNA]_tot/[SSB]_tot > 1.0), much lower than the value Q₆₅ = 0.9 observed for the (SSB)₆₅ mode on (dT)₇₀ (24,51) (see also Fig. 2, A and B). This suggests that ap-(dT)₇₀ is unable to form a fully wrapped (SSB)₆₅ binding mode. This conclusion is also supported by the results of titrations with the C3/Cy5 end-labeled ap-(dT)₇₀ shown in Fig. 6 B. The 1:1 (SSB)₆₅ complexes formed in 0.03, 0.1, and 0.3 M NaCl at r = 1 have Cy5 intensities in the range (2.5–2.8) × 10⁵, whereas the Cy5 signal intensity for normal (dT)₇₀-SSB complex is ∼3.5 × 10⁵ (see Figs. 2, C and D and 3).

The ap-(dT)₇₀ data are also consistent with the results for EcSSB binding to ap-(dT)₃₅. Two molecules of normal-polarity (dT)₃₅ can bind to an SSB tetramer, although with negative cooperativity that depends on salt concentration and type (51,60). All titrations of SSB with ap-(dT)₃₅ presented in Fig. S6 show that only one ap-(dT)₃₅ molecule binds to an SSB tetramer at all [NaCl]. Binding is stoichiometric in the 0.01–0.1 M NaCl concentration range (Fig. S6 A) and affinity is measurable (k_1,35 = (1.7 ± 0.2) × 10⁸ M⁻¹) at 0.3 M NaCl (Fig. S6 B), whereas binding of the first molecule of normal (dT)₃₅ is always stoichiometric (k_1,35 affinity is too high to measure) even at much higher [NaCl] (51,58). The results shown in Fig. S6 B also indicate that even at an ∼16-fold molar excess of ap-(dT)₃₅, only a small fraction of a 2:1 complex can form (k_2,35 = (6.8 ± 0.6) × 10⁴ M⁻¹ is very low). This inability to bind a second molecule of ap-(dT)₃₅ is consistent with inability of ap-(dT)₇₀ to form a fully wrapped complex. However, it is surprising that EcSSB still can bind the ap-(dT)₃₅ with substantial affinity.

Discussion

The two tetrameric SSB proteins from E. coli and P. falciparum have similar subunit organization: an N-terminal DNA binding core and an intrinsically disordered C-terminal tail (Fig. 1 A). Although the tetrameric DNA binding cores share 39% amino acid identity, and 66% homology (41), their intrinsically disordered C-terminal domains differ greatly. In EcSSB, the conserved acidic tip (9 aa) of the C-terminal tail, which serves as a binding site for numerous metabolic SIPs (5,8,13), is connected to the DNA binding core via a 56 aa linker containing few charged residues (Fig. 1 A). In contrast, PfSSB possesses a different acidic tip that lacks the conserved IPF residues of the EcSSB tip and is connected to the DNA binding core by a longer (82 aa) more highly charged linker. This difference in the C-terminal tails has a dramatic effect on the DNA binding properties, including binding mode formation and cooperative binding (23,24,32). In particular, the PfSSB is unable to form a low-site-size, (SSB)₃₅-equivalent binding mode (45) and shows no evidence of highly cooperative binding to ssDNA (32), despite the fact that the DNA binding cores of EcSSB and PfSSB are very similar.

Our focus, in this study, was whether EcSSB and PfSSB proteins bind ssDNA with strict polarity. This was motivated by the observation that crystallographic structures show that PfSSB and EcSSB bind ssDNA with opposite polarity and slightly different topology in their fully wrapped (SSB)₆₅ binding modes (compare Fig. 1, B and D). In fact, of the SSB-ssDNA crystal structures that have been reported, only the EcSSB OB fold binds ssDNA with a different polarity. The different wrapping topology of EcSSB in the crystal structure was verified by placing iodinated uracils in the DNA (6); hence, it does not appear to be artifactual.

To examine this, we designed two types of (dT)₇₀ variants with modified phosphodiester backbone polarities. In the first variant, a reversal of the backbone polarity, was introduced in the middle of the sequence, either via a 3′-3′ or 5′-5′ linkage. In a second variant, every phosphodiester backbone in the sequence was reversed (i.e., a 3′-3′ linkage alternated with a 5′-5′ linkage, designated ap-(dT)₇₀ and ap-(dT)₃₅). If either SSB binds ssDNA with strict polarity, we hypothesized that a fully wrapped (SSB)₆₅ binding mode would be inhibited. Surprisingly, we found that both 3′-(dT)₇₀-3′ and 5′-(dT)₇₀-5′ bind the same as normal (dT)₇₀ to PfSSB. Only 1:1 fully wrapped complexes with the same hydrodynamic characteristics are formed stoichiometrically at both high (0.30 M) and low (10 mM) [NaCl]. Moreover, no differences are observed in the thermodynamics and kinetics of complex formation.

However, the binding of EcSSB to these variants shows clear differences. Both (SSB)₃₅ and (SSB)₆₅ complexes can form stoichiometrically on the RP (dT)₇₀, similar to normal (dT)₇₀ at low (10 mM) and high (0.30 M) [NaCl]. However, the relative stabilities of the two modes are affected. These differences are due to weakening of the interactions (either K₃₅ and/or ω₃₅) of SSB with 5′-(dT)₇₀-5′ in the (SSB)₃₅ binding mode. This is corroborated by ITC measurements, demonstrating that the total enthalpies for forming both binding modes on 5′-(dT)₇₀-5′ are 10–12 kcal/mol less favorable compared with (dT)₇₀ and 3′-(dT)₇₀-3′. However, even for EcSSB, the effects of the backbone polarity reversal on DNA binding are relatively small.

Why would a backbone polarity reversal in the middle of (dT)₇₀ have no effect on formation of the PfSSB (SSB)₆₅ binding mode and show only small effects for EcSSB and only with 5′-(dT)₇₀-5′? We hypothesize that this is related to some structural differences in the DNA binding domains (tetrameric cores) of these proteins, which dictate slightly altered pathways of ssDNA wrapping in the two (SSB)₆₅ binding modes. In EcSSB the ssDNA wraps around the upper AA′ dimer and then crosses the dimer-dimer interface toward the β-helix of the B subunit, finally wrapping around the B′B dimer in trans (AA′BB ′ topology, Fig. 1 B). Whereas in PfSSB, as well as in HpSSB (Fig. 1, D and E, respectively), ssDNA wraps around the upper and lower dimers similar to EcSSB (although with opposite polarity), but crosses the dimer-dimer interface under the β-helix of the A′ subunit in cis following different AA′B′B topology. The EcSSB trans topology appears to be dictated by the presence of 5-6 rigid loops (Fig. 1 B; (6,61)), whereas a different orientation of these loops in PfSSB (Fig. 1 D; (41)) and their absence in HpSSB (Fig. 1 E; (42)) promotes wrapping of ssDNA in cis. These different crossings of DNA at the dimer-dimer interface could explain the differences that we observe for EcSSB and PfSSB. As a matter of fact, the cis wrapping pathway that is displayed in the PfSSB structure (Fig. 7 Ai) could accommodate fully wrapped 3′-(dT)₇₀-3′ and 5′-(dT)₇₀-5′ constructs in the (SSB)₆₅ binding mode, with no change in the polarity of ssDNA binding to the individual OB folds, as shown in Fig. 7 Aii and iii, respectively. The situation differs for EcSSB, for which DNA follows the trans topology (Fig. 7 Bi). In this case, the DNA strand crossing the dimer-dimer interface (in blue) is further removed from the DNA strand wrapping around the opposite B′ subunit. Assuming that binding polarity to the individual OB folds is maintained, we propose the wrapping topologies for the 3′-(dT)₇₀-3′ and 5′-(dT)₇₀-5′ DNA constructs shown in Fig. 7 Bii and iii, respectively. Interestingly, in the proposed topology for the 3′-(dT)₇₀-3′ DNA (Fig. 7 Bii), the DNA crosses the dimer-dimer interface in the same region as proposed for normal-polarity (dT)₇₀. However, in the proposed topology for the 5′-(dT)₇₀-5′ DNA (Fig. 7 Biii), the dimer-dimer crossing occurs in the region of the 5-6 loops, which would appear to be sterically unfavorable. This difference might explain the lower affinity and faster dissociation of 5′-(dT)₇₀-5′ from EcSSB, whereas 3′-(dT)₇₀-3′ and (dT)₇₀ show similar binding properties.

Figure 7.

Proposed alternative DNA wrapping topologies in the (SSB)₆₅ binding modes for EcSSB and PfSSB formed with the RP (dT)₇₀. Proposed topology is shown for ssDNA binding to PfSSB (A) and EcSSB (B) in an (SSB)₆₅ binding mode for (dT)₇₀ with normal polarity (i), 3′-(dT)₇₀-3′ (ii), and 5′-(dT)₇₀-5′ (iii). In each case, the ends of the DNA entering and exiting the tetramer are designated by red arrows with the indicated polarity. The blue arrows crossing the dimer-dimer (AA′/BB′) interface in each particular case represent the middle regions of the DNA and show the different wrapping topologies needed to maintain the polarity of the DNA backbone bound to each individual subunit (OB fold) (3′-5′ for PfSSB and 5′-3′ for EcSSB). To see this figure in color, go online.

The ability of EcSSB to form a 2:1 (SSB)₃₅ complex on both (dT)₇₀ RP variants is not as surprising because, in that binding mode, only two subunits of each bound tetramer interact with the DNA. However, because of different orientation of those tetramers on the RP constructs, a decrease in the nearest-neighbor cooperativity might be expected because of the loss of some contacts required for such interactions. Yet, we still observe significant cooperativity for the 3′-(dT)₇₀-3′ DNA, comparable with (dT)₇₀, whereas cooperativity is decreased with the 5′-(dT)₇₀-5′ DNA. Based on a proposed model for the (SSB)₃₅ binding mode (Fig. 1 C), a 2:1 complex should be possible with both RP (dT)₇₀ constructs. A simple 180° rotation of the upper tetramer within the x-y plane relative to the lower tetramer creates a 5′-(dT)₇₀-5′ complex, whereas a similar rotation performed on the lower tetramer creates a 3′-(dT)₇₀-3′ complex. However, in both cases, the interactions between the4-5 loops that have been proposed to be important for nearest-neighbor cooperativity (6) would be lost. We speculate that alternative orientations of the two tetramers in three-dimensional space (e.g., around the y axes) could restore such nearest-neighbor cooperative contacts, and this might be easier for the 3′-(dT)₇₀-3′ construct. We note here that the model for the (SSB)₃₅ binding mode in Fig. 1 C is speculative and awaits experimental testing. Therefore, alternative wrapping pathways in this mode, which might explain the differences observed for the RP (dT)₇₀ constructs, cannot be excluded. We also know that nearest-neighbor cooperativity in the (SSB)₃₅ mode is also dependent on the IDL in the C-terminal tails (24), which are not observed in the crystal structures. If cooperativity requires direct interactions between the IDLs of adjacent tetramers, such interactions should be possible on (dT)₇₀ and both RP variants.

Interestingly, we find that EcSSB still binds stoichiometrically to the fully alternating ap-(dT)₇₀ in both binding modes (Fig. 6) over a broad NaCl concentration range. However, the affinity in a fully wrapped (SSB)₆₅ mode is weakened significantly because EcSSB is not able to bind more than one molecule of ap-(dT)₃₅ under any solution condition (Fig. S7), yet, it can still bind to EcSSB with substantial affinity (e.g., ap-(dT)₃₅ binds to EcSSB in 0.30 M NaCl with k_35,1 = 1.7 × 10⁸ M⁻¹, whereas a lower limit for normal (dT)₃₅ is ∼2 × 10⁹ M⁻¹ (51)). The normalized Trp fluorescence quenching is the same for binding ap-(dT)₃₅ and normal (dT)₃₅ (Q₁ ∼ 0.5), suggesting that the same number of SSB Trp residues are involved in both complexes and that the same path is followed on the protein. The ability of SSB to retain such high affinity to ap-(dT)₃₅ might seem surprising. However, SSB is able to bind to DNA of varying nucleotide composition with different affinities (62), which is the basis of nonspecific binding to ssDNA. We also note that SSB binding to polynucleotides of differing base composition also affects the binding mode preference (62,63).

The results reported here demonstrate that both tetrameric EcSSB and PfSSB can accommodate stretches of ssDNA with RP and still bind with high affinity. EcSSB also retains the ability to form its different binding modes and even retains high nearest-neighbor cooperativity. Only severe backbone polarity changes, such as for the ap-ssDNA constructs, show a large effect on affinity. We have proposed structural models that can explain this behavior while still retaining strict backbone polarity for binding to the individual OB folds. Hence, our results are consistent with the different ssDNA polarities observed in the crystal structures of ssDNA complexes with EcSSB and PfSSB. This adaptability may play important roles in the metabolic functions of SSB.

Author contributions

T.M.L. and A.G.K. designed the research. A.G.K. performed and analyzed the experiments. A.G.K. and T.M.L. wrote the manuscript.

Editor: Smita S. Patel.