Responses to Ligand Binding in the Bacterial DNA Sliding Clamp “β‐Clamp” Manifest in Dynamic Allosteric Effects

ABSTRACT

The homo‐dimeric, ring‐shaped bacterial DNA sliding clamp, β‐clamp, is a central hub in DNA replication and repair. It interacts with a plethora of proteins via their short linear motifs, binding to the same hydrophobic binding pocket on β‐clamp. Although the structure, functions, and interactions of β‐clamp have been amply studied, less focus has been on understanding its dynamics and how this is influenced by ligand binding. In this work, we have made a backbone nuclear magnetic resonance (NMR) assignment of the 83 kDa dimeric β‐clamp and used NMR in combination with hydrogen‐deuterium exchange mass spectrometry to scrutinize the dynamics of β‐clamp and how ligand binding affects this. We found that the binding of a short peptide from the polymerase III α subunit affects the dynamics and stability of β‐clamp. The effect not only appears locally around the binding pocket but also globally through dynamic allosteric connections to distant regions of the protein, including the dimer interface. The dissipated dynamic effect from ligand binding is likely a consequence of a unique binding pocket architecture that connects distant parts of the structure and may reflect a mechanism of structural plasticity in protein hubs, where different ligands impose differential responses in the structure and dynamics of β‐clamp, resulting in diverse functional responses.

Abbreviations

BADCOP

beta/alpha decoupling pulse

CBM

clamp binding motif

CSP

Chemical shift perturbation

GOODCOP

gradient optimized CO decoupling pulse

HDX‐MS

hydrogen‐deuterium exchange mass spectrometry

NMR

nuclear magnetic resonance

pol

polymerase

SCS

secondary chemical shift

SLiM

short linear motif

1. Introduction

In the bacterium E. coli , DNA replication is carried out by a multiprotein complex consisting of a polymerase III complex (pol III core complex), a helicase/primase complex, and a clamp loader complex in addition to a DNA sliding clamp, called β‐clamp [1, 2]. β‐clamp is a homo‐dimeric, toroidal protein that encircles DNA [3] tethering polymerases to the primer/template junction during DNA synthesis, ensuring processive DNA replication [1, 2]. β‐clamp is loaded onto DNA in an ATP‐dependent crab‐claw‐like motion by the clamp loader complex [4, 5] and constitutes a central hub interacting with numerous proteins important in translesion synthesis (TLS) [6, 7, 8], DNA mismatch repair [9, 10] or otherwise involved in DNA homeostasis [10, 11, 12].

The structure of each β‐clamp subunit consists of three topologically similar domains (I, II and III) connected by interdomain connecting loops (IDCLs) forming a head‐to‐tail ring‐shaped homodimer (Figure 1A). The two subunits create an inner channel lined by 12 α‐helices and an outer surface composed of 24 β‐strands forming antiparallel sheets [14, 15]. The many binding partners interact with β‐clamp through a short linear motif (SLiM) [16, 17] called the clamp binding motif (CBM), which spans five to six residues and is defined as Qφx[L/M][F/L] and Qφx[L/M]x[F/L] (φ = aliphatic, x = any residue) [12] (Figure 1B). All known natural β‐clamp ligands bind to the same two identical highly conserved binding pockets, one in each subunit located between domains II and III (Figure 1A). Here, the Gln and the following aliphatic amino acid of the CBM bind to subsite II of the binding pocket, and the two C‐terminal hydrophobic residues bind to an adjacent subsite I [12]. Subsite I is composed of residues Arg152, Leu155, Thr172, Gly174, Arg176, Leu177, Pro242, Arg246, Val247, and Val360, and subsite II of Asn320, Tyr323, Val344, Pro363, Met364, and Arg365. In addition, His175 and Met362 bridge the two subsites [12]. Remarkably, these residues are distributed and separated by almost 100 and 200 residues, respectively, resulting in a unique binding pocket architecture. The binding pocket is overall hydrophobic, surrounded by positively charged residues (Figure 1B). Although the isolated CBMs bind to β‐clamp with micromolar affinity, residues outside the CBM also contribute to affinity [18, 19], binding interface [7, 11, 20] and to increased lifetime of protein:DNA complexes [21].

FIGURE 1.

β‐clamp structure and binding pocket. (A) Structure of E. coli β‐clamp in complex with a CBM containing peptide from pol III α (PDB ID: 3D1F [13]). β‐clamp domains I, II and III are shown in white, light gray and dark gray, respectively, and IDCLs in blue and the pol III peptide in red. (B) The β‐clamp binding pocket with the electrostatic potential computed in ChimeraX. The pol III peptide is shown in gray with key SLiM‐interacting residues highlighted as sticks. Two methionine residues of β‐clamp, Met362 and Met364 laying within the binding pocket are colored green.

β‐clamp is a stable dimer in solution with an estimated K _D in the low picomolar range in the absence of DNA [22, 23] and a half‐life for spontaneous DNA dissociation of ~1–2 h [22, 24, 25]. Still, studies have found β‐clamp to be dynamic in solution. While hydrogen‐deuterium exchange mass spectrometry (HDX‐MS) studies concluded that domain I is highly dynamic [26, 27], a high‐resolution crystal structure of β‐clamp showed increased B‐factors in domains II and III [15]. It was also suggested that the β‐clamp dimer can open spontaneously, existing in an open‐closed equilibrium [20, 26], while several other studies have found that the major state of β‐clamp is a closed dimer [20, 28, 29, 30]. Thus, there is an apparent inconsistency regarding how and in what ways β‐clamp is dynamic, possibly reflecting different conditions or methodology. Still, very little is known about the dynamic properties of β‐clamp and how this is influenced by ligand binding.

Here we combine NMR spectroscopy with HDX‐MS to address the conformational dynamics of β‐clamp, and we scrutinize how the binding of CBM‐carrying ligands affects this. With the dynamics mapped on several timescales, we found that peptide binding influences both fast and slow‐timescale dynamics in β‐clamp, its conformational stability, and a tendency to self‐associate at high concentrations. Not surprisingly, regions in domains II and III that entail the binding pockets become stabilized upon ligand binding. Remarkably, however, ligand binding also allosterically affects the dimer interface far from the binding pockets, likely facilitated by the unique architecture of the binding pocket. This structural and dynamic adaptability of β‐clamp could have implications for its function as a hub and allow β‐clamp to interact with a wide variety of protein binding partners through the same shared site. With variations in the CBMs, different ligands may affect the dynamics of β‐clamp differentially and lead to diverse functional outcomes.

2. Results

2.1. Strategy to Circumvent Unfavorable Dynamics for Backbone Assignment of β‐Clamp

To investigate the dynamics and interactions of the bacterial DNA sliding clamp, we made an NMR backbone assignment of recombinantly produced β‐clamp. Due to its very large size (83 kDa) β‐clamp was expressed in ~97% deuterated minimal media supplemented with ²H, ¹³C₆‐labeled glucose, and ¹⁵N‐labeled NH₄Cl. Expression and purification of ²H, ¹³C, ¹⁵N β‐clamp yielded ~20–30 mg pure protein per 0.5 L expression media. Although the TROSY‐HSQC of ²H, ¹³C, ¹⁵N labeled β‐clamp showed nice dispersion and fair intensities of the peaks (Figure S1), the resonance assignment proved challenging. HNCA and HNCO experiments yielded decent spectra; however, the spectral quality of HNCACB spectra was inadequate for assignment. Instead, modified HNCA experiments named gradient optimized CO decoupling pulse (GOODCOP) and beta/alpha decoupling pulse (BADCOP) were implemented [31]. In the different GOODCOP/BADCOP spectra, C^α peaks display different splitting patterns depending on the C^β chemical shifts of the residue. Thus, these spectra provide indirect information on the C^β chemical shift via selective decoupling and help resolve ambiguities of sequentially matching peaks in the HNCA spectra during assignment (Figure S2). Furthermore, 3D‐¹⁵N‐TROSY‐NOESY [32] spectra were obtained that, together with crystal structures of β‐clamp, aided assignment of spatially close residues in the β‐clamp structure. Despite these useful NMR experiments, assignments of the β‐clamp stalled after approximately 50%–60% of the protein backbone resonances. The main reason was the low spectral quality of the 3D spectra, resulting in many missing peaks in the HNCA spectra.

Because β‐clamp tended to precipitate over time at high concentrations at 37°C, we aimed to stabilize the protein by adding a ligand. Remarkably, the addition of a small eight‐residue peptide harboring the C‐terminal CBM of pol III α (Ac‐EQVELEFD, hereafter denoted as pol III) clearly improved the quality of the TROSY‐¹H, ¹⁵N‐HSQC spectrum, but was accompanied by substantial movement of many peaks (Figure S1). Importantly, in the presence of pol III, the quality of the 3D‐spectra markedly improved; HNCACB spectra yielded distinct peaks of both C^α and C^β nuclei, and the minor peaks missing in the HCNA/GOODCOP/BADCOP spectra of unbound β‐clamp became intense (Figure S3). The resonances of pol III‐bound β‐clamp were therefore assigned using HNCA, HNCACB, GOODCOP, BADCOP, and TROSY‐NOESY experiments. A TROSY‐¹H, ¹⁵N‐HSQC titration enabled us to transfer many assignments to the unbound β‐clamp (Figure 2A). To ensure a correct match between the bound and unbound peaks, especially in parts of the spectrum with substantial spectral overlap in the TROSY‐HSQC, C^α chemical shifts and NOESY connections were compared. This ligand‐bound state strategy resulted in the assignment of 306/346 ¹⁵N and ¹H (88%), 328/366 ¹³C^α (90%), 269/341 ¹³C^β (79%) and 213/366 ¹³C’ (58%) of the pol III bound β‐clamp and 294/346 ¹⁵N and ¹H (85%), 317/366 ¹³C^α (87%) and 212/366 ¹³C’ (58%) of the free β‐clamp (excluding the His₆‐tag). Unassigned residues mainly locate to flexible loops or small turns between connecting β‐strands (Figure S4). Their peaks were likely missing due to structural heterogeneity in the intermediate exchange regime or due to fast proton exchange with the solvent at the high temperature. A few residues were only assigned in the β‐clamp + pol III complex. Met364, which interacts closely with pol III [13], was only assigned in free β‐clamp. Although Ala147, Tyr153, and Gly157 were not assigned in the free β‐clamp, they were visible in the titration spectra with pol III.

FIGURE 2.

Peptide binding impacts the dynamics of the protein binding pocket on β‐clamp. (A) ¹H, ¹⁵N TROSY‐HSQC titration of a ²H, ¹³C, ¹⁵N β‐clamp with pol III peptide. (B) Chemical shift perturbations between unbound β‐clamp and β‐clamp:Pol III (1:2) for all assigned residues. Red stars indicate residues that form the binding pocket as defined in the introduction. (C) Chemical shift perturbations upon pol III binding mapped onto the structure of β‐clamp in complex with a pol III peptide (PDB ID: 3D1F [13]). The peptide is in green and highlighted with a dashed circle. Unassigned residues are colored gray.

2.2. Ligand Binding Induces Global Changes in β‐Clamp Dynamics

The large and widespread spectral changes induced by binding of the eight‐residue peptide indicated a global change in the structure and/or dynamics of β‐clamp. We therefore mapped the chemical shift perturbations (CSPs) for each residue (Figure 2B). The CSPs mainly located to domain III, with domain II less affected, and only minor perturbations in domain I. Mapping the CSPs onto the structure of β‐clamp clarified that the largest CSPs were from residues located in or around the binding pocket (Figure 2C). Although we expect residues interacting with the peptide to display large CSPs, residues further away from the binding site were also strongly affected, suggesting that ligand binding causes allosteric structural changes or a redistribution of the conformational ensemble of β‐clamp [33]. Secondary chemical shifts (SCSs), which report on secondary structures, were generally consistent with the secondary structure content of the β‐clamp crystal structures (Figure S5). Yet, the SCSs of β‐clamp with and without the addition of pol III were overall very similar, except for the extreme C‐terminal residues directly involved in the interaction with pol III (Figure S5). This suggests that changes upon ligand binding do not occur on the secondary structure level. Furthermore, β‐clamp has been crystallized in the absence [14, 15] and presence of a similar nine‐residue peptide of the pol III α C‐terminal CBM [13]. These crystal structures align with an RMSD of C^α atoms of only 0.95 Å (Figure S6). Thus, differences in the 3D structures cannot explain the large spectral changes we observe upon ligand binding, suggesting them to originate from changed dynamics, which is likely not visible in the crystal structures due to a common crystal lattice.

To address whether and how ligand binding affects the dynamics of β‐clamp, peak intensity ratios from the unbound and pol III‐bound β‐clamp were mapped for each residue (Figure S7). An overall increase in intensity of the peptide‐bound β‐clamp was in line with the enhanced quality of the NMR spectra (Figure 3A,B). Notably, peak intensities of three residues located in the loop formed by residues 145–157 increased markedly, indicating that ligand binding stabilizes this loop. This is also the loop where most residues are either not assigned or assigned only in the pol III‐bound state. Some residues also show decreased peak intensities in the ligand‐bound state, but these are mostly located at the binding interface (Figure 3A,B) and likely experience faster relaxation due to dipolar coupling to protons of the pol III peptide.

FIGURE 3.

Peptide binding stabilizes β‐clamp dynamics and reduces self‐association. (A) Peak intensity ratios between pol III bound and unbound β‐clamp. (B) Intensity ratios mapped onto the β‐clamp structure in complex with a pol III peptide (green) (PDB ID: 3D1F [13]). Residues for which peak intensities decrease upon peptide biding are colored blue. Residues with the largest increase in peak intensities are highlighted by a dashed square. (C) R ₂ values of β‐clamp in the absence (black) and presence (blue) of two‐fold molar excess of pol III. Top and bottom plots are R ₂ values of β‐clamp at concentrations of 0.5 and 0.3 mM, respectively.

Because the peak intensities are related to the transverse relaxation, we recorded ¹⁵N R ₂ relaxation rates on β‐clamp in the presence and absence of pol III (Figure 3C). R ₂ values were generally high and uniform, characteristic of a large, folded protein. However, at a concentration of 0.5 mM β‐clamp, we noticed that the unbound β‐clamp had overall larger R ₂ values compared to the pol III‐bound protein (Figure 3C). We speculated whether β‐clamp tends to self‐associate at high concentrations. Whereas the pol III‐bound β‐clamp R ₂ values were essentially unaffected by concentration changes, R ₂ values of free β‐clamp decreased with decreasing concentration, approaching those of the ligand‐bound state. These results suggest that β‐clamp tends to self‐associate at high concentrations (~300 μM and higher) and that ligand binding abolishes this, explaining in part the improvement in spectral quality. The ¹⁵N, ¹H HSQCs in Figure 2A, from which the CSPs in Figure 2B and the peak intensities in Figure 3A were derived, were obtained at 200 μM β‐clamp where the self‐association should be minimal. We also did not see large changes in the ¹H, ¹⁵N chemical shifts at the different β‐clamp concentrations used for the different experiments (assignment spectra in Figure S1, ¹⁵N R ₂ relaxation spectra and the spectra in Figure 2A). Hence, we attribute the large CSPs we see upon addition of pol III to allosteric changes in β‐clamp dynamics upon ligand binding and not resulting from self‐association.

2.3. Global Changes in Chemical Shifts Are a Universal Effect of Peptide Binding to β‐Clamp

Next, we explored if the ligand‐induced global effect on the chemical shifts occurred upon binding of other peptide ligands and recorded ¹H, ¹⁵N TROSY‐HSQCs of β‐clamp in complex with peptides from polymerase IV (pol IV) and from two proteins involved in mismatch repair, MutS and MutL (Table 1, Figure S8). For quantification, we designed the peptides as 15–16‐mers including an N‐terminal Trp. For better comparison, a similar, longer version of the pol III peptide containing the Trp was included. These longer peptide variants are denoted with an asterisk. To first assess if the flanking region of the long pol III* affects the interaction with β‐clamp, we compared the binding of the 16‐mer pol III* to that of the eight‐mer pol III. Isothermal titration calorimetry (ITC) yielded affinities of K _D = 4–7 μM (Table 1), comparable to earlier studies [13, 18, 36]. ITC and ¹H, ¹⁵N TROSY‐HSQCs showed that the flanking region contributes to the interaction by forming additional contacts to β‐clamp, without largely affecting the binding affinity (Figure S9).

TABLE 1.

Peptide sequences and properties.

Peptide name	Sequence a	Net charge	K D (μM)
pol III	Ac‐EQVELEFD‐COO—	−5	4 ± 1
pol III*	NH3 +‐DWRGLIGSEQVELEFD‐COO—	−5	7 ± 2
pol IV*	NH3 +‐GWLDPQMERQLVLGL‐COO—	−2	2–8 [34, 35]
MutS*	NH3 +‐GWATQVDGTQMSLLSV‐NH2	0	ND (likely weak)
MutL*	NH3 +‐GWGEAPVCAQPLLIPL‐NH2	0	ND (likely weak)

Abbreviation: ND, not determined.

^a Key CBM interacting residues are highlighted in bold.

Comparing the NMR spectra of the complexes revealed that the global changes were also induced by binding of the pol IV* peptide (Figure 4). Titrations with the pol III* and pol IV* peptides showed large effects on β‐clamp residues of domain II and III, which surround the binding pocket. Notably, despite similar affinities and the same saturation level (> 98%) (Table 1), there were differences; while the CSPs of domain I were very similar for the two peptides, pol IV* induced slightly larger CSPs in domain II and especially in the interface between domains II and III. Conversely, pol III* generally induced larger CSPs in domain III (Figure 4). The differences could be caused merely by the different chemical nature of the residues, where for example, pol III* is more negatively charged compared to pol IV*, or because the flanking regions outside the binding motifs interact differently with β‐clamp. However, the widespread effect points towards a differential effect in the interaction of the CBM binding residues with the binding pocket, causing different impact on the structural dynamics of the β‐clamp.

FIGURE 4.

pol IV* binding causes different global changes on the chemical shifts compared to pol III*. (A) CSPs derived from ¹H, ¹⁵N TROSY‐HSQCs of ²H, ¹³C, ¹⁵N labeled β‐clamp in complex with pol III* (top) and pol IV* (bottom). (B) CSPs mapped onto the structure of β‐clamp in complex with pol III peptide (PDB ID: 3D1F [13]). Unassigned residues are colored gray. Peaks that were untraceable due to intermediate exchange are colored blue.

The CBM of MutS also induced CSPs for many of the same residues as pol III* and pol IV*, however, their amplitudes were smaller reflecting the weaker affinity (Figure S8). To our knowledge, there are no reported affinities on MutS and the β‐clamp may not be completely saturated with MutS, explaining the smaller effect. There are also no reported affinities on MutL, but the affinity is likely low [9, 37]. Although we would expect the peaks to move in fast exchange at such low affinity, the binding of the CBM of MutL appeared to cause substantial line broadening (Figure S8). Thus, MutL* may bind β‐clamp differently and do differ from the canonical CBM as it contains a Pro at the second position and a Leu/Ile combination at positions 4 and 5 [12]. Crystal structures of the complex between MutL and β‐clamp also show that the CBM binds differently in the β‐clamp binding pocket, likely caused by the conformational restraints imposed by the proline [9, 12]. As the proline succeeds an isoleucine this gives rise to a 12% cis‐population of the peptide bond [38], and line broadening may thus likely originate from structural heterogeneity within the binding pocket.

2.4. β‐Clamp Experiences Fast and Slow Hydrogen‐Exchange

To assess the impact from ligand binding on the dynamic properties of β‐clamp on slower timescales, we performed additional NMR experiments including CPMG and CEST, but the signal to noise was far too low to obtain reliable data. We therefore aimed to study slow‐timescale dynamics by measuring the hydrogen‐deuterium exchange of β‐clamp in the presence and absence of ligand. At first, we performed HDX by NMR, but due to the long measuring time of the HSQCs (> 1 h), many signals were lost already after the first time point and the time resolution of other signals were not good enough to extract meaningful rate constants. Instead, we turned to HDX‐MS on free β‐clamp and β‐clamp in complex with pol III, where measuring times can be much shorter. At first, the global deuterium uptake of intact β‐clamp was measured. A non‐deuterated control of β‐clamp yielded an experimental mass of 41 466 Da (Figure 5A), corresponding well with the theoretical mass of 41 466.07 Da of one subunit of the His₆‐tagged β‐clamp with the N‐terminal Met removed. We noticed two additional peaks in the MS spectrum, each with a mass increase of 16 Da, suggesting β‐clamp to be partially oxidized carrying up to two oxidations per subunit (Figure 5A).

FIGURE 5.

Global HDX‐MS analysis reveal a weak protection against exchange in β‐clamp upon pol III binding. (A) Deconvoluted ESI‐MS spectrum obtained from β‐clamp in H₂O‐buffer. (B) Deuterium uptake of free β‐clamp and β‐clamp in complex with pol III peptide after incubation for 10 s, 1.5 min, 13.5 min, and 121.5 min in deuterated PBS buffer at 37°C, pD 7.4. The deuterium uptake of the maximally labeled β‐clamp sample is indicated by a gray dashed line. The data points are an average of triplicate values, error bars show ±SD, but most of them are too small to be visible in this plot.

During HDX‐MS analysis, an inevitable loss of deuterium will occur due to back‐exchange with the protonated solvents. To measure the experimental maximum deuterium uptake of β‐clamp, a maximally labeled control was made. Here, β‐clamp was diluted 10‐fold in 6 M guanidinium deuterochloride (GdnDCl) and incubated for 2 h, resulting in a deuterium uptake of 257 ± 3 Da. Since the theoretical maximum uptake of β‐clamp in 90% D₂O is 317 Da, the back‐exchange of β‐clamp in the global HDX‐MS analyses is 19%.

To probe the effect of pol III on the global deuterium uptake, β‐clamp was diluted 10‐fold into deuterated buffer to initiate H‐to‐D exchange in the presence and absence of pol III. After various periods of deuteration, the exchange reaction was quenched by acidification and cooling followed by mass measurement. The results showed that within the first 10 s, a fast exchange of the amide protons occurred with an average deuterium uptake of 92 ± 0 Da in unbound β‐clamp and 84 ± 2 Da in pol III‐bound β‐clamp (Figure 5B). After ~2 h, the deuterium uptake reached 162 ± 2 Da for both samples, which was still far from the maximally labeled β‐clamp of 257 Da. Even after more than 4 h, the deuterium uptake only reached 176 Da for free β‐clamp and 173 Da for pol III‐bound β‐clamp (Figure S10). The hydrogen‐deuterium exchange is therefore defined by a fast exchange process, likely occurring in the unprotected turns and flexible loops in β‐clamp, followed by a slower exchange process where a substantial part of β‐clamp remains unexchanged after several hours. A weak, yet significant protection induced by binding of pol III is evident by the lower D‐uptake of pol III‐bound β‐clamp vs. free β‐clamp at the early time points. As the protection is weak, this difference in deuterium uptake disappears after prolonged deuteration. This suggested that the overall exchange process is similar for the bound and unbound β‐clamp.

2.5. Methionine Oxidation in the Binding Pocket Affects Pol III Binding

By local HDX‐MS analysis of proteolyzed β‐clamp, we identified methionine oxidations located at the extreme C‐terminus. Peptide 359–366 was identified as non‐oxidized and as oxidized (+16 Da) at either Met362 or Met364. Both methionines are located within the binding pocket (Figure 1B). Deuterium uptake plots of peptide 359–366 clearly showed that a protection induced by pol III binding at 10 s in the non‐oxidized β‐clamp was almost completely abolished when Met362 was oxidized (Figure 6A). Oxidation of Met364 had the same effect (Figure S11). This suggests that oxidation of methionine residues in the binding pocket decreases pol III affinity. We also noticed that upon NMR titrations of β‐clamp with various peptide ligands, two populations of the C‐terminal Leu366 were present (Figure 6B). Thus, a small population of what appears to be the unbound β‐clamp state persists until it disappears with excess amounts of ligand. This small population likely corresponds to the oxidized protein and further supports different ligand affinity between non‐oxidized and oxidized forms of β‐clamp.

FIGURE 6.

Methionine oxidation in the β‐clamp binding pocket impacts binding. (A) Local HDX‐MS deuterium uptake plots of peptide 359–366 in its non‐oxidized form (left) and with oxidation of Met362 (right). (B) Zoom on the C‐terminal Leu366 from the ¹H, ¹⁵N TROSY‐HSQC titrations of ²H, ¹³C, ¹⁵N β‐clamp with addition of pol III (left) and pol III* (right).

2.6. Peptide Binding Stabilizes β‐Clamp Against Global Unfolding

We discovered serendipitously by HDX‐MS that β‐clamp slowly undergoes global unfolding when diluted 10‐fold in 4 M urea, and that the unfolding kinetics are readily monitored by the emergence of bimodal peaks in the global uptake plots of intact β‐clamp, corresponding to two different populations. The low mass peak represents a population of folded β‐clamp with a deuterium uptake comparable to that observed without urea, while the high mass peak represents the maximally labeled, globally unfolded β‐clamp. We furthermore noticed that the two states were differently populated in the presence and absence of pol III, where pol III suppressed the population of the unfolded state relative to the folded state. This led to the hypothesis that pol III may impact the stability of β‐clamp. We therefore designed a urea experiment, where β‐clamp with and without pol III were diluted 10‐fold in 4 M deuterated urea and the reaction was quenched at different time points. Again, the MS spectra showed two peaks corresponding to a fully and a partially deuterated protein, respectively (Figure 7A). Quantification of the two states as a function of time revealed that the unfolding reaction is slow (Figure S12). After incubation for only 10 s, the fully unfolded population was not observed. Over time, the population of the unfolded, fully deuterated state gradually increased until the entire population was fully unfolded after 24 h (Figure 7B). The unfolded population increased by 14.1% and 11.6% per hour for unbound β‐clamp and pol III‐bound β‐clamp, respectively. At time points between 1 and 4 h, where both states are present, a significant population difference between free β‐clamp and pol III‐bound β‐clamp was observed, with the fully unfolded state less populated in the pol III‐bound sample. Thus, pol III has an overall stabilizing effect on the slow unfolding of the β‐clamp.

FIGURE 7.

Pol III peptide stabilizes β‐clamp against global unfolding. (A) ESI‐MS spectra showing the two populations present in the samples of free β‐clamp and pol III bound β‐clamp diluted 10‐fold in 4 M deuterated urea and incubated at 25°C for 4 h (two middle panels). Top panel is the non‐deuterated control (in H₂O PBS buffer, pH 7.4), and the bottom panel is the maximally labeled control (6 M GdnDCl, 37°C, 2 h). (B) Fraction (%) of the fully deuterated and unfolded state for free β‐clamp and pol III bound β‐clamp at different time points. Unpaired, two‐tailed t‐test for significance between unbound and pol III bound samples; 1 h: P = 0.0116 (*); 2 h: P = 0.0317 (*); 4 h: P = 0.0006 (***).

To test if the unfolding by urea was reversible or irreversible, we made a pulse‐labelling experiment. Here, β‐clamp with and without pol III were incubated in non‐deuterated urea. After 2 h, the samples were diluted into deuterated urea and incubated for 1 min before the exchange reaction was quenched, allowing readily exchangeable hydrogens to exchange with deuterium. Here, we saw a substantial amount of the fully deuterated state, indicating either an irreversible unfolding process or slow kinetics in the unfolding/refolding equilibrium (Figure S12).

2.7. Pol III Binding Stabilizes the Binding Pocket and Imposes Allosteric Effects at Distant Sites

To investigate the native state β‐clamp dynamics in the absence of urea, we analyzed the local deuterium uptake. First, peptic peptides of β‐clamp were identified, and their deuterium uptake measured (the peptides and uptake listed in Table S1, sequence coverage map in Figure S13, example spectra shown in Figure S14). The N‐terminal His₆‐tag experienced a high amount of back‐exchange (Table S1), as previously noted [39, 40], and was excluded from the analysis. The average amount of back‐exchange for all remaining peptides was 36%.

Representative peptides from β‐clamp and their individual HDX in the absence of pol III were mapped onto the β‐clamp structure (Figure 8A). To increase the resolution, the peptides were selected to cover as much of the β‐clamp structure with as short peptides as possible. The average deuterium uptake for these selected peptides was 45% after 10 s and 76% after 2 h. Peptides with more than 50% deuterium uptake after 10 s were all located in the IDCLs, in the very C‐terminus, or in other flexible loops or linkers, whereas peptides from structured parts of β‐clamp remained relatively protected. The same was true after 2 h, where many of the forementioned loops had completely exchanged with deuterium, while structured parts generally were more protected. The overall HDX in the three domains of β‐clamp did not vary greatly. Notably, after 2 h, peptides 7–16 and 317–325 displayed high deuterium uptake compared to the average (97% and 89%, respectively) despite not covering any loops (Figure S15), hinting that these parts of the structure are highly dynamic compared to other structured regions of β‐clamp. Interestingly, the dimer interface appeared to be one of the most stable parts of the structure, with only 13% (10 s) and 55% (2 h) uptake for peptide 98–111, and 23% (10 s) and 64% (2 h) for peptide 284–306 (Figure S15).

FIGURE 8.

Hydrogen‐deuterium exchange mapping shows stabilization of the binding pocket and allosteric effects at distant sites upon ligand binding. PDB ID: 3D1F [13]. pol III is colored green and unassigned peptides are colored gray. (A) Overall hydrogen‐deuterium exchange of selected peptides from β‐clamp in the absence of pol III mapped onto the structure of β‐clamp. (B) The difference in deuteration (%) between free β‐clamp and β‐clamp + pol III, where positive values (blue) is the result of a lower deuterium uptake in the presence of pol III and negative values (red) is the result of a higher deuterium uptake in the presence of pol III. Bottom panel shows zooms of the protein binding pocket after 10 s (left), 2 h (middle) and of the dimer interface after 2 h (right).

To further address how pol III affects the dynamics and stability of β‐clamp, we compared the deuterium uptake of β‐clamp in the absence and presence of pol III and illustrated it on the β‐clamp structure (Figure 8B). The difference in %‐uptake for free β‐clamp and pol III‐bound β‐clamp for all peptides is listed in Table S1, and uptake plots for peptides with a difference in protection in unbound and pol III‐bound β‐clamp are shown in Figure S16. After 10 s, a substantial protection of peptides that form the binding pocket was observed. Most prominently, peptides 145–159, 231–244, 317–325, 339–348, and especially 359–366 gained protection in the presence of pol III (Figure 8). This is not unexpected, since the pol III peptide interacts with residues within these sequences [13]. While the C‐terminal peptide (359–366) is almost completely buried upon binding pol III, as evident from the crystal structure [13], other peptides, such as 145–159 where only side chains interact, likely get protected as the interactions stabilize the loop dynamics.

Ligand binding induces weak protection in some peptides that are part of the binding pocket (e.g., peptide 359–366), while other peptides that are also part of the binding pockets are more strongly protected, as evidenced by decreased deuterium incorporation in the presence of pol III even after 2 h, for example, peptides 170–179 and 317–325 (Figure 8B). Given a β‐clamp concentration of 5 μM, a pol III concentration of 50 μM and an approximate K _D of 4 μM (Figure S9), the calculated fraction of bound β‐clamp at any given time is ~93%. With an affinity in the micromolar range and a fast exchange between bound and unbound β‐clamp seen on the NMR timescale, the peptide would be expected to bind and unbind numerous times within 2 h. This likely explains why the peptide 359–366, which lies directly in the binding pocket, did not exhibit a reduced deuterium uptake after 2 h in the presence of pol III. Intriguingly, strong protection was also observed in peptides that are remote to the binding site, for example, peptides 136–144, 231–244, and 274–283, indicating that a long‐term stabilization manifests in other parts of the β‐clamp structure that are not involved in direct interaction with pol III. Thus, these data combined suggest that ligand binding to β‐clamp has allosteric effects.

Unexpectedly, the peptides from domain I and III located in the β‐strands of the dimer interface (peptides 98–111 and 284–306) as well as the α‐helix at the dimer interface of domain III (peptide 263–273) exhibited a slightly higher deuterium uptake in the presence of pol III (Figure 8B). Although a global stabilization and destabilized dimer interface are not mutually exclusive, we would expect that the dimer interface had a decreased deuterium uptake in the pol III bound state since pol III stabilizes β‐clamp against unfolding (Figure 7). One possible solution to this conundrum may be that the interaction with pol III represents a noncanonical scenario, in which part of the ligand‐binding energy causes a destabilization of the ground state that in turn increases the Boltzmann occupancy of the exchange‐competent transiently unfolded states [41]. It is, however, beyond the scope of the present work to elucidate the exact mechanism behind this intriguing allosteric effect.

3. Discussion

In this study, we have addressed the dynamics of β‐clamp and how it is influenced by ligand binding. Central to this, we obtained assignment of the backbone NMR resonances of β‐clamp using a strategy including protein deuteration, TROSY‐based NMR spectra, GOODCOP/BADCOP, and NOESY connections. As a result, we have NMR reporters distributed throughout this large 83 kDa dimeric protein to monitor structural and dynamic properties. An ILV methyl NMR assignment of β‐clamp has been published [42] in which methyl ¹³C and ¹H chemical shifts of Ile, Leu, and Val side chains were assigned. Recently, a backbone resonance assignment of a stabilized β‐clamp variant (T45R/S107R) was made by Mahdi et al. and transferred to the WT β‐clamp [43]. However, we observe discrepancies between the assignment of WT β‐clamp by Mahdi et al. and those we report here, likely stemming from issues related to the assignment transfer to WT β‐clamp and due to unresolved overlap issues, resolved here from the use of the set of GOODCOP/BADCOP spectra. A detailed comparison of the two assignments and the discrepancies is provided in the Supporting Information. Thus, the robustness of the assignment process of our work holds promise for being useful for other large, folded proteins. However, a nearly complete assignment was only possible with the addition of a small, eight‐residue peptide ligand comprising the C‐terminal CBM of pol III α, highlighting that β‐clamp in its free state undergoes NMR‐unfavorable dynamic processes.

Addition of a small ligand of only eight residues had a stabilizing effect on β‐clamp, affecting its global unfolding process in urea and affecting the native state dynamics of the amides both locally in the binding site and distant to it. This suggests that binding of a ligand imposes an allosteric effect in β‐clamp that redirects dynamics from the binding pocket and the associated loops to the dimer interface, where amide exchange rates increased. Long‐range allosteric effects on dynamics have been described for many other protein systems, including the Brr2 RNA helicase [43], the peptidyl‐prolyl isomerase Pin1 [44] and the serine protease thrombin [45], and now also for β‐clamp.

According to our data, the three domains of β‐clamp display similar dynamics with similar deuterium uptake. This contrasts with two studies by Fang et al. [26, 27] who concluded that domain I was the more dynamic. Since local HDX‐MS data is not residue specific, and because the deuterium uptake is an average over an entire peptide, interpretations depend on the peptides chosen to represent the exchange as well as their length. Therefore, as the peptide covering the β‐strand in the dimer interface of domain I is also connected to the highly dynamic IDCL between domains I and II, it is highly likely that domain I appears more exposed and dynamic in the Fang et al. studies, as their peptides were longer. Here, we recovered peptides that were shorter and covered the binding interface separately from the IDCL, revealing that the dimer interface is indeed very protected compared to the rest of the protein. Although measured under different conditions (salt and temperatures), we found other discrepancies between our data and those of the Fang et al. papers, which suggest that there is a difference in the stability of the β‐clamp used here and the one used in the previous studies by Fang et al., the origin of which we cannot pinpoint. One remarkable difference is the presence of EX1 kinetics in the Fang et al. data, suggesting global unfolding in the absence of urea, a scenario we do not observe.

Despite its relatively stable structure, we see that β‐clamp tends to precipitate at high concentrations. This self‐association was reflected in the R ₂‐values of unbound β‐clamp that increased upon increasing concentration without changing the chemical shifts (Figure 3C). The self‐association was reduced upon pol III binding. Thus, the pol III peptide either screens the hydrophobic binding pockets or prevents local unfolding that exposes hydrophobic residues that may cause β‐clamp to self‐interact. However, since self‐interaction occurs at high concentrations, the phenomenon may not have any biological implications. By HDX‐MS we do see that pol III binding has a stabilizing effect on β‐clamp by altering the dynamics of β‐clamp around the binding pocket and by imposing an overall stabilizing effect on global unfolding in the presence of denaturant. If and how these changes in dynamics and stability may affect additional functional steps of β‐clamp, such as its loading onto the DNA by the clamp‐loader, remains to be addressed.

Generally, for β‐clamp, we observe different types of dynamic processes occurring at different timescales; we observe a change in dynamics at a fast timescale, reflected by the peak intensity increase for some residues in the NMR spectra, and we see changes in dynamics and stability at the seconds‐to‐hours timescale with HDX‐MS. Residues Ala147, Tyr153, and Gly157 show substantial peak intensity increases in the pol III‐bound state. However, since their peak intensities are too low in the free state, we cannot accurately determine an R ₂ ‐value for these residues and thus cannot extract at which timescale this occurs Yet, these dynamic processes may relate to each other. For example, the protection of residues within the binding pocket caused by pol III binding results in a long‐term stabilization of residues more distal to the binding pocket. This allosteric effect even manifests in changes in the dimer interface, which is relatively distant (~18 Å) from the binding pocket. Unfortunately, because of the size of the system and the self‐association, we were unable to use NMR to extract the dynamics on the μs‐ms timescales.

Since this global effect on the chemical shifts was also induced by other CBM‐containing peptides, we speculate if this change in dynamics may be a common mechanism for ligand binding to β‐clamp. As a hub, β‐clamp interacts with many different binding partners with different structures and variations in their CBMs that should ultimately lead to a different biological output. The fact that the CBM length varies between five or six residues may further require some structural plasticity in β‐clamp. We therefore reason that the flexibility and dynamics of β‐clamp are a way of accompanying and adapting to its many different binding partners, and that different binding partners may impose different allosteric effects. The differences in chemical shifts induced by the pol III* and pol IV* peptides may reflect such variable structural adaptability. Structural plasticity and ligand adaptability have also been reported in other hub proteins that function as coregulators of sets of transcription factors [46, 47]. We also found that oxidation of two methionines in the binding pocket affects the affinity of peptide ligands. One can therefore speculate that these methionine oxidations can have a biological relevance for example, in partner selection under oxidative stress conditions, or variation in the oxidation state during the cell cycle [48], but further experiments are needed to investigate if methionine oxidation has a regulatory function in β‐clamp interactions.

The β‐clamp binding pocket is constituted of many loops and flexible linkers which connect different parts of the structure originating from both domain II and III (Figure 9). Such intricate architecture equips the structure with different structural modulators that can effectuate the information of ligand binding into changes in dynamics at specific sites distant from the binding pocket. Thus, dynamic allostery [49, 50, 51] may allow for ligand‐dependent modulation of the binding pocket and controlled dissemination of information throughout the protein. This has wide implications and could suggest a relevant and important role also for the non‐conserved residues of the SLiMs—a hitherto neglected part of SLiMs. Many β‐clamp binding partners also bind to other parts of the β‐clamp surface [7, 11, 20]. How these contacts affect the dynamics and flexibility in the other domains and how these are disseminated throughout the structure of β‐clamp may be important for differential binding effects. Thus, binding in the context of the full‐length protein is important in many aspects of binding, including changes in dynamics. If and how a differential ligand‐induced allosteric modulation exists in the interaction between β‐clamp and its arsenal of binding partners is a highly intriguing question that deserves future investigations.

FIGURE 9.

Ligand‐induced allosteric effects on β‐clamp dynamics. Pol III interacting residues are shown as sticks, and different structural segments of the binding pocket are colored differently. The β‐clamp binding pocket is formed by loops of different structured parts, which allow for ligand‐induced modulation of the dynamics in neighboring parts of the structure.

DNA sliding clamps are essential proteins in all domains of life [52]. The bacterial β‐clamp and the eukaryotic DNA sliding clamp, proliferating cell nuclear antigen (PCNA) are functionally and structurally conserved. PCNA is a trimer, with each subunit comprising two structurally similar domains [53]. Although all DNA sliding clamps are structurally similar, their intrinsic stability differs [22]. The dynamic properties of PCNA even differ between organisms, where human PCNA is more dynamic and experiences higher deuterium uptake compared to yeast PCNA; both demonstrated by HDX‐MS [27] and HDX‐NMR [54]. Notably, HDX‐MS experiments demonstrated that the binding of a small protein to Thermococcus kodakarensis PCNA resulted in increased deuterium uptake in two regions, one of which is a β‐strand in the subunit interfaces of the PCNA trimer [55]. Furthermore, NMR studies on human PCNA have revealed substantial effects upon ligand binding even at sites distant to the binding pocket [56, 57, 58], comparable to the observations made here for β‐clamp interactions. These general observations of allosteric effects indicate that the structural plasticity and allosteric binding mechanism may be common to DNA sliding clamps across species.

4. Materials and Methods

4.1. Peptides

The eight‐mer pol III peptide (Ac‐EQVELEFD) was purchased from Innovagen (Sweden). The other 15–16‐mer peptides pol III* (DWRGLIGSEQVELEFD), pol IV* (GWLDPQMERQLVLGL), MutL* (GWGEAPVCAQPLLIPL‐NH₂) and MutS* (GWATQVDGTQMSLLSV‐NH₂) were purchased from TAG Copenhagen (Denmark). The 15–16‐mer peptides were designed to contain DW/GW located distant to and N‐terminally upstream of the CBM for easier concentration measurements. The 8‐mer pol III peptide was N‐acetylated to remove the positive charge at the N‐terminus, since this is not present in the context of the full‐length protein. MutL* and MutS* were C‐terminally acetylated since the CBMs of these proteins are internally positioned in the sequence, whereas the CBMs of pol III and pol IV are in the C‐terminus of the protein.

4.2. Protein Expression and Purification

The His₆‐tagged E. coli β‐clamp sequence in a pET16b vector was transformed into BL21 (DE3) pLysS cells. The full sequence of the N‐terminal His‐tag is (MGHHHHHH). Cells were grown on an LB agar plate containing 100 μg/mL ampicillin (Amp) and 35 μg/mL chloramphenicol (Cam). One colony was added to 10 mL LB media with antibiotics, and the culture grew overnight (o/n) at 37°C at 180 rpm. For the expression of unlabeled β‐clamp, the 10 mL o/n culture was added to 1 L LB media supplemented with antibiotics. For the expression of ²H, ¹³C, ¹⁵N labeled β‐clamp, a pre‐culture was made by adding 500 μL LB o/n to 20 mL deuterated M9 media containing 2 g/L ¹³C, ²H labeled glucose, and 1 g/L ¹⁵NH₄Cl in ~97% D₂O with 50 μg/mL Amp and 17.5 μg/mL Cam. The pre‐culture was grown o/n at 37°C, 180 rpm and added to 0.5 L deuterated ²H, ¹³C, ¹⁵N labeled M9 media the next day. Cells were grown at 37°C, 180 rpm, and at OD₆₀₀ ~0.6–0.8, (~3 h for cells in LB media and ~10 h for cells in deuterated M9 media), expression was induced by the addition of 0.5 mM IPTG. After 5–6 h of expression, cells were harvested by centrifugation at 5000 g for 15 min and resuspended in 30 mL Buffer A (20 mM sodium phosphate, 1 M NaCl, 50 mM imidazole, 5 mM β‐mercaptoethanol, pH 7.4) including a complete EDTA free protease inhibitor cocktail tablet (Roche Diagnostics GmbH). Cells were lysed by sonication (80% amplitude, 1 cycle for 10 × 15 s with 15 s breaks between). The lysate was cleared by centrifugation at 20.000 g for 30 min, and the lysate was transferred onto a gravity column containing 5 mL Ni²⁺‐NTA resin (Cytiva Sweden AB) equilibrated with Buffer A. The column was washed with 50 mL Buffer A, and the protein was eluted in 10 mL Buffer B (20 mM sodium phosphate, 1 M NaCl, 500 mM imidazole, 5 mM β‐mercaptoethanol, pH 7.4). The eluted protein was concentrated with a 15 mL 10 000 MWCO spin filter (Merck Millipore Ltd.). The sample was centrifuged at 20000 g for 15 min to remove precipitated protein prior to being loaded onto a HiLoad 16/600 Superdex 200 size exclusion chromatography (SEC) column pre‐equilibrated with 20 mM sodium phosphate, 100 mM NaCl, 1 mM dithiothreitol (DTT), pH 7.4, and the sample was eluted into the same buffer. Fractions with β‐clamp were assessed with SDS‐PAGE, and pure fractions were pooled and stored at 4°C. β‐clamp concentrations were calculated by measuring the absorbance at 280 nm on a NanoDrop ND‐1000 spectrophotometer (Thermo Fisher Scientific) and using an extinction coefficient of 15 930 as calculated by the ExPASy ProtParam web tool (https://web.expasy.org/protparam/).

4.3. NMR Spectroscopy

All NMR samples were prepared by adding 5% (v/v) D₂O and 125 μM DSS to protein solutions in 20 mM sodium phosphate, 100 mM NaCl, 5 mM DTT, pH 7.4. After centrifugation, samples were transferred to 5 mm Shigemi NMR tubes and spectra were recorded at 37°C on a Bruker Avance III 750 MHz spectrometer equipped with a cryoprobe. DSS was used to reference ¹H, ¹⁵N, and ¹³C referenced indirectly using the gyromagnetic radii. 2D TROSY‐¹H‐¹⁵N‐HSQC spectra [59, 60] were processed in Topspin (Bruker). T ₂ relaxation experiments and triple‐resonance assignment spectra were processed and phase corrected using qMDD [61] or nmrDraw, a component of NmrPipe [62]. All spectra were analyzed and manually assigned in CcpNmr Analysis [63].

Free β‐clamp was assigned from a sample containing ~1100 μM ²H, ¹³C, ¹⁵N labeled β‐clamp using deuterium decoupled TROSY variants of HSQC, HNCA, HNCO [64] as well as ¹⁵N edited 3D‐TROSY‐NOESY [32, 65, 66, 67, 68, 69] and gradient optimized CO decoupling pulse (GOODCOP) and beta/alpha decoupling pulse (BADCOP) spectra [31]. A sample of 610 μM ²H, ¹³C, ¹⁵N labeled β‐clamp + ~2‐fold excess of pol III was used for the assignment of pol III bound β‐clamp using the same experiments as above, including a TROSY version of HNCACB [70, 71]. 8%–12% non‐uniform sampling was used for the GOODCOP and BADCOP spectra. The GOODCOP/BADCOP spectra of free β‐clamp were collected with 256 complex points, which correspond to a resolution of 22 Hz, and pol III bound spectra were collected with 400 complex points in the carbon dimension corresponding to a resolution of 14 Hz. The assignments are deposited to BMRB under the deposition number ID 52494 and 52495. Spectra of β‐clamp in complex with pol III*, pol IV*, MutL*, and MutS* (Figure S8) contained 175 μM ²H, ¹³C, ¹⁵N β‐clamp and 700 μM pol III*, pol IV*, or MutL* or 350 μM MutS*. MutS* was added to a lower concentration due to low solubility of the peptide.

4.4. Chemical Shift Perturbations, Intensity Ratios, and Secondary Chemical Shifts

CSPs and peak intensity ratios from pol III binding were quantified from samples of 200 μM ²H, ¹³C, ¹⁵N β‐clamp with and without 400 μM pol III. CSPs from the addition of pol III* or pol IV* were quantified from titrations into 200 μM ²H, ¹³C, ¹⁵N β‐clamp with the addition of up to 800 μM pol III* or pol IV*. Identification of β‐clamp peaks in the pol III* and pol IV* bound state was based on a titration with increasing amounts of pol III*/pol IV*. For peaks in intermediate exchange, the assignments were based on the peak trajectory of the first titration point. For the pol III* assignment, peaks in intermediate exchange were also guided by the pol III assignment. Peak intensities were only included for residues where the peak volume could be fit to a parabola. CSPs were calculated using the following equation [72]:

$$\mathit{CSP}=∆\delta \left(\mathit{ppm}\right)=\sqrt{{\left(∆\mathit{\delta H}\right)}^2+{\left(0.154·∆\mathit{\delta N}\right)}^2}$$

Secondary chemical shifts (SCSs) were calculated using the following equation:

$$SCS={\delta }_{\mathrm{obs}}-{\delta }_{\text{random coil}}$$

where the intrinsic chemical shifts of a random coil were calculated using the POTENCI web tool (https://st‐protein02.chem.au.dk/potenci/) [73] with the physical parameters set to match the conditions of the NMR samples.

4.5. NMR Relaxation

A TROSY variant of the T ₂ experiment [74] was used to determine R ₂ relaxation rates of β‐clamp with and without pol III. The relaxation delays were 8, 17, 25, 34, 51, 68, 102, and 136 ms, and the spectra were recorded in triplicates in a randomized order. The peak heights at each relaxation delay were extracted and fitted to a one‐phase exponential decay in Prism to obtain the R ₂ values. Errors are standard error between the three replicas.

4.6. ITC

β‐clamp was buffer exchanged into ITC buffer (20 mM sodium phosphate, 100 mM NaCl, 1 mM tris (2‐carboxyethyl)phosphine (TCEP), pH 7.4) using a 15 mL 10 000 MWCO spin filter (Millipore). The peptides were dissolved in ITC buffer, and the pH was measured and adjusted if needed. The cell contained 25.3–29.8 μM β‐clamp for the titrations with Pol III* and 52.5–57.4 μM β‐clamp for the titrations with Pol III. The concentration of the Pol III* peptide was determined using the extinction coefficient of 5500 as calculated by the Expasy ProtParam web tool (https://web.expasy.org/protparam/) and the measured absorbance at 280 nm from a NanoDrop ND‐1000 spectrophotometer (Thermo Fisher Scientific). Protein and peptide samples were centrifuged at 20000 g for 15 min to degas the samples. For all experiments, β‐clamp was added to the cell, and the peptide was added to the syringe. The ITC experiments were recorded on a Malvern MicroCal PEAQ‐ITC instrument (Malvern, United Kingdom) at 25°C and a stir speed of 750 rpm. ITC experiments were repeated 3 times for each peptide.

The MicroCal PEAQ‐ITC Software was used to fit the data to a one‐site binding model. Because the 8‐mer pol III peptide does not contain any aromatic residues, determination of the concentration is more inaccurate, and the N‐value between the pol III titration and the pol III* titration therefore varied, even though the β‐clamp stock was identical. Since the measured enthalpy is proportional to the concentration of ligand in the syringe, the N‐value was fixed to 1, and the syringe concentration varied to be able to compare the enthalpies between the two peptides.

4.7. HDX‐MS

Protein stocks containing 50 μM β‐clamp in the absence or presence of 500 μM pol III 8‐mer peptide in 20 mM sodium phosphate, 100 mM NaCl, 1 mM DTT, pH 7.4 were used for all HDX‐MS experiments. Triplicate values were measured for all samples and controls. All hydrogen‐deuterium exchange reactions were quenched in formic acid and immediately snap frozen in liquid nitrogen. Samples were kept frozen at −70°C or in liquid nitrogen until loaded onto the LC–MS. For all experiments, each time point represents 50 pmol β‐clamp (with or without 500 pmol pol III peptide) that is loaded onto a Waters HDX‐Manager with a desalting flow of 0.23% aqueous formic acid solution (Solvent A) provided by an Agilent 1260 Infinity quaternary pump (Agilent Technologies, CA, USA) and a gradient flow comprised of Solvent A (0.23% aqueous FA solution) and Solvent B (0.23% FA in pure acetonitrile, ACN). The gradient was delivered by a nanoAcquity ultra‐performance liquid chromatography (UPLC) Binary Solvent Manager (Waters Corporation, MA, USA). Samples were analyzed by electrospray ionization (ESI) MS using an ESI Tri‐Wave Ion Mobility mass spectrometer (Synapt G2).

For global and local HDX‐MS experiments, non‐deuterated controls were made by diluting the β‐clamp stock 10‐fold in PBS buffer (10 mM phosphate buffer, 2.7 mM KCl, 137 mM NaCl, pH 7.4) and mixing it 1:12 (v/v) with quench solution (0.67% formic acid). Maximally labeled controls were made by diluting β‐clamp stock 10‐fold in 6 M deuterated GdnDCl dissolved in D₂O for 2 h at 37°C.

4.7.1. Global HDX‐MS Analysis

Protein stocks were diluted 10‐fold in deuterated PBS buffer (10 mM phosphate buffer, 2.7 mM KCl, 137 mM NaCl, pD 7.4) and the samples were incubated at 37°C. At 10s, 90s, 810 s, and 7290 s, 10 μL sample was withdrawn and mixed 1:12 (v/v) with quench solution and snap frozen in liquid nitrogen. The thawed samples were afterwards injected into the sample loop (200 μL), desalted for 2 min at 500 μL/min flow rate on a 2.1 mm × 5 mm MassPREP Micro Desalting Column at 0.2°C, and eluted with the following gradient: 5%–50% B for 2.5 min, 50%–90% B for 0.5 min, 90%–95% B for 0.1 min, 95%–90% B for 0.3 min, and 90%–5% B for 0.45 min at a flow rate of 50 μL/min. MS data was acquired for 7 min in positive resolution mode at m/z 300–2500 with a scan time of 0,5 s. Combined spectra were processed in MassLynx using MaxEnt1 with a width at half height set to 0.45 (0.3 for H₂O controls). The most populated mass of the non‐oxidized state was plotted.

4.7.2. Urea HDX‐MS Experiments

Protein stocks were diluted 10 times in deuterated 4 M urea, and the samples were incubated at 25°C to avoid covalent adducts. At 10s, 1 h, 2 h, 4 h, and 24 h, 10 μL sample was withdrawn and mixed 1:12 (v/v) with quench solution and snap frozen in liquid nitrogen and analyzed as described for the global HDX‐MS analysis.

A pulse labeling experiment was made to test if the unfolding by urea was reversible. Here, protein stocks were diluted 10 times in 4 M H₂O urea and incubated at 25°C for 2 h. After 2 h, the sample was diluted 2‐fold in 4 M deuterated urea and incubated for 1 min at 25°C before the reaction was quenched.

The data was processed with MaxEnt1 in MassLynx as described for the global HDX‐MS analysis. The population of the partially deuterated and the fully deuterated was determined by smoothing the spectra (Smooth window (channels) +/−: 8; Number of smooths: 2; Smoothing method: Mean), subtracting the baseline (Polynomial order: 15; Below curve (%): 30; Tolerance: 0.01) and creating a centred spectrum (Min peak width as half height (channels): 10; Center method: Top; Create centred spectrum; Centred spectrum: Areas). The relative intensities of the centred spectra were extracted, and the percent of the fully unfolded calculated as:

$$\frac{I_{\text{unfolded}}}{\left(I_{\text{folded}}+I_{\text{unfolded}}\right)}·100\%$$

4.7.3. Local HDX‐MS Analysis

Protein stocks were diluted 10‐fold in deuterated PBS buffer, incubated at 37°C, and the reaction quenched after 10s and 7290 s in the same way as described above. The samples were injected directly into the sample loop (200 μL) and digested online by agarose immobilized pepsin (ThermoFisher, MA, USA) self‐packed in a 2.0 × 20 mm column (IDEX Upchurch guard column, WA, USA) at 20°C. The sample was desalted for 3 min at 300 μL/min flow rate on an ACQUITY UPLC BEH C18 VanGuard Precolumn (130 Å, 1.7 μm, 2.1 × 5 mm, Waters) and separated using a 1.0 × 50 mm ACQUITY UPLC Peptide BEH C18 Column (300 Å, 1.7 μm, 1.0 × 50 mm, Waters) at 0.2°C. The proteolytic peptides were eluted with 5%–10% B for 0.1 min, 10%–50% B for 11.9 min, 50%–90% B for 0.5 min, 90%–90% B for 0.1 min, and 90%–5% B for 0.1 min at a constant 40 μL/min flow rate.

Peptide identifications by LC–MS/MS analysis were conducted using an Orbitrap Eclipse mass spectrometer operating in positive ion mode (Thermo FisherScientific, San Jose, CA, USA). The peptides were chromatographically separated using the same equipment and methods as for the local HDX‐MS analysis described above. Mass spectra were acquired with an automatic data‐dependent switch between an Orbitrap survey MS scan in the mass range of m/z 300 to 1500, followed by fragmentation of peptides using collisional energy of 30% (normalized) in a 1 s cycle. MS1 spectra were acquired at a resolution of 120 000 at m/z 200. A dynamic exclusion of previously selected ions for 5 s was applied. MS2 spectra were obtained at 60000 resolution at m/z 200 with a normalized AGC target of 300% using a m/z 1.2 isolation window. The maximum injection time was 246 ms (MS1) and 118 ms (MS2). The MS1 AGC target was 400 000 ions, and the MS2 AGC target was 150 000 ions. The resulting raw files were converted to MGF files using ProteoWizard, and the peptides were identified by a Mascot database search with the following parameters: variable modifications (oxidation, M), (Sulfo, STY); peptide mass tolerance ±0.1 Da; fragment ion mass tolerance ±7 ppm; enzyme, none. The Mascot database contained the proteins of interest as well as possible contaminating proteins.

The local HDX‐MS data was analyzed in DynamX. Relative deuterium uptakes were extracted from DynamX and plotted into Excel for calculations of uptake and back‐exchange. % deuteration was calculated as:

$$\frac{\text{Average uptake}\mathrm{at}\text{given timepoint}}{\text{Average uptake of}100\%{\mathrm{D}}_2\mathrm{O}\text{controls}}·100\%$$

Percent back‐exchange was calculated as:

$$\left(1-\frac{\text{Average uptake of}100\%{\mathrm{D}}_2\mathrm{O}\text{controls}}{\mathrm{Max}\text{uptake}·0.9}\right)·100\%$$

The fraction of bound protein was calculated using the following equation [75]:

$$\text{fraction bound}=\frac{P_t+L_t+K_D-\sqrt{{\left(P_t+L_t+K_D\right)}^2-4P_t{L}_t}}{2P_t}$$

Where $P_t$ is the total protein concentration and $L_t$ is the total ligand concentration.

Author Contributions

Signe Simonsen: conceptualization, investigation, formal analysis, writing – original draft, writing – review and editing, visualization. Andreas Prestel: conceptualization, formal analysis, writing – review and editing. Eva C. Østerlund: investigation, formal analysis, writing – review and editing. Marit Otterlei: writing – review and editing, funding acquisition. Thomas J. D. Jørgensen: conceptualization, writing – original draft, writing – review and editing, supervision, methodology, resources. Birthe B. Kragelund: conceptualization, methodology, writing – original draft, writing – review and editing, funding acquisition, supervision, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: prot70024‐sup‐0001‐supinfo.pdf.

Simonsen S., Prestel A., Østerlund E. C., Otterlei M., Jørgensen T. J. D., and Kragelund B. B., “Responses to Ligand Binding in the Bacterial DNA Sliding Clamp “β‐Clamp” Manifest in Dynamic Allosteric Effects,” Proteins: Structure, Function, and Bioinformatics 93, no. 12 (2025): 2193–2210, 10.1002/prot.70024.

Data Availability Statement

The data that support the findings of this study are openly available in biomagresbank at https://bmrb.io/, reference number 52494 and 52495.