DNA Interactions Probed by Hydrogen-Deuterium Exchange (HDX) Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Confirm External Binding Sites on the Minichromosomal Maintenance (MCM) Helicase*

The archaeal minichromosomal maintenance (MCM) helicase from Sulfolobus solfataricus (SsoMCM) is a model for understanding structural and mechanistic aspects of DNA unwinding. Although interactions of the encircled DNA strand within the central channel provide an accepted mode for translocation, interactions with the excluded strand on the exterior surface have mostly been ignored with regard to DNA unwinding. We have previously proposed an extension of the traditional steric exclusion model of unwinding to also include significant contributions with the excluded strand during unwinding, termed steric exclusion and wrapping (SEW). The SEW model hypothesizes that the displaced single strand tracks along paths on the exterior surface of hexameric helicases to protect single-stranded DNA (ssDNA) and stabilize the complex in a forward unwinding mode. Using hydrogen/deuterium exchange monitored by Fourier transform ion cyclotron resonance MS, we have probed the binding sites for ssDNA, using multiple substrates targeting both the encircled and excluded strand interactions. In each experiment, we have obtained >98.7% sequence coverage of SsoMCM from >650 peptides (5–30 residues in length) and are able to identify interacting residues on both the interior and exterior of SsoMCM. Based on identified contacts, positively charged residues within the external waist region were mutated and shown to generally lower DNA unwinding without negatively affecting the ATP hydrolysis. The combined data globally identify binding sites for ssDNA during SsoMCM unwinding as well as validating the importance of the SEW model for hexameric helicase unwinding.

The archaeal minichromosomal maintenance (MCM) helicase from Sulfolobus solfataricus (SsoMCM) is a model for understanding structural and mechanistic aspects of DNA unwinding. Although interactions of the encircled DNA strand within the central channel provide an accepted mode for translocation, interactions with the excluded strand on the exterior surface have mostly been ignored with regard to DNA unwinding. We have previously proposed an extension of the traditional steric exclusion model of unwinding to also include significant contributions with the excluded strand during unwinding, termed steric exclusion and wrapping (SEW). The SEW model hypothesizes that the displaced single strand tracks along paths on the exterior surface of hexameric helicases to protect singlestranded DNA (ssDNA) and stabilize the complex in a forward unwinding mode. Using hydrogen/deuterium exchange monitored by Fourier transform ion cyclotron resonance MS, we have probed the binding sites for ssDNA, using multiple substrates targeting both the encircled and excluded strand interactions. In each experiment, we have obtained >98.7% sequence coverage of SsoMCM from >650 peptides (5-30 residues in length) and are able to identify interacting residues on both the interior and exterior of SsoMCM. Based on identified contacts, positively charged residues within the external waist region were mutated and shown to generally lower DNA unwinding without negatively affecting the ATP hydrolysis. The combined data globally identify binding sites for ssDNA during SsoMCM unwinding as well as validating the importance of the SEW model for hexameric helicase unwinding.
DNA unwinding by hexameric replicative helicases is required for DNA replication elongation, providing the singlestranded DNA (ssDNA) 5 templates for leading and lagging strand synthesis. These hexameric helicase complexes form ring-like structures that preferentially encircle one of the DNA strands, providing stability in DNA binding and enhancing processivity for unwinding long stretches of DNA. In the steric exclusion model for unwinding, the motor domains within the central channel translocate along the encircled ssDNA while physically excluding the complementary ssDNA on the exterior. Although DNA unwinding by helicases can be effectively measured in gel-based and fluorescence assays, the specific molecular interactions with DNA are poorly described. Previously, we have proposed that the excluded strand is not a passive component of unwinding; rather, it will interact along specific exterior paths on the hexameric helicase surface to both stabilize the complex and promote forward unwinding (1). Although structures of hexameric helicases that include small stretches of ssDNA have been solved recently (2,3), the interactions are constrained to central channel residues, and no molecular information is available at the duplex region or with respect to the excluded strand.
Hexameric DNA replication helicases are known to exist across the three domains of life, including viruses. Although their overall structural toroidal features are conserved, their amino acid sequences are not, and they are generally classified into three families by their unwinding polarity and specific protein folds (4,5). Bacteria and many phages are members of the superfamily 4 helicases, which include RecA motor domains and 5Ј-3Ј unwinding polarity. In contrast, eukaryotic viruses from superfamily 3 (i.e. E1 and SV40 Large T) and eukaryotic and archaeal superfamily 6 minichromosomal maintenance (MCM) helicases are members of the AAA ϩ (ATPase associated with various cellular activities) enzymes and translocate 3Ј-5Ј.
The eukaryotic heterohexameric MCM2-7 helicase interacts with and unwinds DNA as a part of a larger GINS-Cdc45-MCM2-7 (CMG) complex that also includes Cdc45 and GINS components (6,7). Interactions with ssDNA within this complex have been noted on the exterior, specifically with the Cdc45 subunit (8). Recently, a high resolution electron microscopy structure of the yeast MCM2-7 complex shows twisted and tilted double hexamers on DNA, but the conformations of the individual hexamers are symmetrical, and contacts with DNA are constrained to the central channel (9). The archaeal replicative MCM helicase generally exists as a homohexamer that shares homology with each of the six sequence-distinctive subunits in eukarya (MCM2 to MCM7) and is essential for replication initiation and elongation (10). Archaeal MCM helicases from Sulfolobus solfataricus (SsoMCM) and Methanothermobacter thermoautotrophicus serve as simplified models for understanding mechanisms of ATP hydrolysis, DNA binding, and unwinding for more complex eukaryotic hexameric DNA replication helicases (11). Crystal structures of archaeal MCMs with the N-terminal domain (NTD) alone (12,13) as well as nearly full-length structures with both the NTD and AAA ϩ domain (14,15) have been solved and show similar symmetry and subunit arrangements. It is interesting to note that an EM structure of M. thermoautotrophicus MCMs has identified a binding site for DNA on the external surface of the helicase, possibly implicated as a prebound engaged state prior to loading and encircling onto DNA at replication origins (16).
We previously elucidated the unwinding mechanisms for the SsoMCM helicase and found specific and required interactions with the excluded ssDNA (1,17). Specific electrostatic mutations on the exterior surface disrupted the excluded strand interactions and reduced unwinding rates. These results advanced the traditional steric exclusion unwinding model to include external surface interactions with the excluded strand and designated a new helicase unwinding model termed steric exclusion and wrapping (SEW). Although we identified specific sites of interaction with the excluded strand through mutagenesis, a more comprehensive and global description of the external binding path is needed to validate and better describe the SEW model.
Using hydrogen/deuterium exchange mass spectrometry (HDX-MS), we are able to identify specific contacts with both the encircled and excluded strands and the SsoMCM homohexamer in solution. The benefit of using HDX-MS to probe protein/DNA interactions is that it can measure H/D exchange correlated with binding interfaces in the native solution state without any perturbations (18,19). HDX-MS detects protein backbone amide proton exchanges that are accessible to solvent and can identify regions that are protected upon ligand binding or through conformational changes (20,21) as validated by comparison with HDX exchange observed by proton NMR (22). Hydrogen atoms on the side chains or on the N and C termini of the peptide also undergo H/D exchange. However, at these rapidly exchanging sites, incorporated deuteriums backexchange by H 2 O in the mobile phase during HPLC separation. Hence, the mass increment is solely due to the deuterium exchange of the backbone amide hydrogens (23).
Ultrahigh resolution Fourier transform ion cyclotron resonance MS (24), improved instrument automation (25)(26)(27), and efficient protein digestion (28) allow for detection of specific ssDNA interactions with single subunits within a redundant MCM homohexamer. Using HDX-MS, we are able to probe interfacial residues within the SsoMCM hexamer from the monomeric x-ray structure (14) and define a more global interaction path of the excluded strand with the exterior surface, thus validating the SEW model as a bona fide unwinding mechanism.

Experimental Procedures
Materials-ATP was obtained from Invitrogen. Optikinase was purchased from U. S. Biochemical Corp. All other materials were from commercial sources and were analytical grade or better. Oligonucleotides were purchased from IDT Corp. (Coralville, IA) and gel-purified as described previously (1). [␥-32 P]ATP was purchased from PerkinElmer Life Sciences. Helicase buffer is used in all assays consisting of 125 mM potassium acetate, 25 mM Tris acetate (pH 7.5), and 10 mM magnesium acetate.
SsoMCM Cloning and Purification-SsoMCM R33A, K111A, K111A/R113A, K129A, and R226A were created using a standard QuikChange protocol (Agilent, Santa Clara, CA) with KAPA HiFi DNA polymerase (KAPA Biosystems, Woburn, MA). Primer sequences included silent mutations for unique restriction site screening. Mutations were confirmed using the DNA Sequencing Facility at the University of Texas (Austin, TX). WT, the point mutants, and previously cloned K323A/ R440A were purified as described (1). Briefly, autoinduced SsoMCM was heat-treated at 70°C for 20 min, and the supernatant was applied to MonoQ, heparin, and S-200 gel filtration columns (GE Healthcare) to isolate the purified hexameric species.
Hydrogen/Deuterium Exchange-HDX was performed as described previously (29). Stock solutions of 9 M SsoMCM were prepared in helicase buffer either alone or in the presence of 3Ј-tail DNA or with 3Ј-tail/5Ј-tail DNA in an equivalent molar ratio (1:1). Similar buffer conditions were applied for the preparation of the corresponding D 2 O buffers. HDX experiments were optimized and automated with an HTC Pal autosampler (Eksigent Technologies, Dublin, CA). 5 l of SsoMCM was mixed with 45 l of corresponding buffer in D 2 O to initiate each H/D exchange period. For the blank control, the initial dilution was made in H 2 O buffer. Reactions were performed in triplicate at 1-2°C to reduce back-exchange for HDX incubation periods of 0.5, 1, 2, 4, 8, 15, 30, 60, 120, and 240 min, each followed by acid quench and proteolysis for 3 min. Each sample was quenched by rapid mixing with 25 l of 200 mM tris(2-carboxyethyl)phosphine, 8 M urea in 1.0% formic acid and 25 l of a 40% saturated protease type XIII (Sigma-Aldrich) solution in 1.0% formic acid (final pH ϳ2.3). On-line LC electrospray ionization 14.5-tesla Fourier transform ion cyclotron resonance MS with an LTQ Velos (Thermo Fisher Scientific, San Jose, CA) front end yielded high mass resolving power (m/⌬m 50% Ͼ 200,000 at m/z 400) (30). There were 772 peptides for apo-SsoMCM, 722 peptides for SsoMCM 3Ј-tail, and 676 peptides for SsoMCM 3Ј-tail/5Ј-tail. Data were analyzed by an in-house Predator software package (26). The measured deuterium uptake and associated S.D. values were recorded for each incubation period and fitted by a maximum entropy method (MEM) (31). The average relative deuterium uptake difference (ARDD) between free and complexed protein was calculated by the following equation, in which A is the deuterium uptake for sample A (apo-SsoMCM) at a specified time (t i ), and B (SsoMCM/DNA) is the deuterium uptake for sample B at a specified time (t i ) (32).
A focused analysis of each peptide was also performed to more accurately reveal highly sequence localized changes in solvent accessibility that can be obfuscated by the averaging of multiple sequence overlapped peptides of varying length as is done in the global analysis. This direct comparison of identical peptides between experimental conditions can often uncover localized changes that may otherwise be overlooked. Potentially poorer quality data with S.D. values Ͼ0.5 were excluded for this analysis. This procedure reduced the data to 574 peptides for each of the three experimental data sets. We then assigned values to each peptide by calculating the midpoint of the peptide (first and last amino acid number) to align the peptides semisequentially. If the assigned value was the same for two different peptides, the peptide with the lower first amino acid number was listed first. For those peptides with two or more charges, we averaged the values for each incubation period, parsing the data to 497 peptides. The three data sets were then compared at 10 incubation periods. Only those peptides that yielded values for each incubation period were retained, reducing the number of peptides to 238. The data were then plotted as relative deuterium uptake versus assigned peptide number for the butterfly plots. Differences between free and complexed protein were determined based on significant differences in absolute H/D exchange as well as exponential rate constants.
DNA Unwinding Assays-Unwinding assays were performed as described previously (1). Briefly, Optikinase was used to 32 Plabel the 5Ј-end of the duplex. SsoMCM variants (2.4 M monomer, 400 nM hexamer) were incubated with 15 nM 5Ј-radiolabeled DNA for 5 min at 60°C and initiated with 2 mM ATP. Reactions were quenched with an equal volume of glycerol quench (0.5% (w/v) SDS, 50% (v/v) glycerol, 0.1% (w/v) bromphenol blue, 100 mM EDTA, pH 8.0, and 300 nM trap ssDNA). Quenched reactions were stored on ice until the addition of 4 l of Proteinase K 20 mg/ml (Thermo Fisher Scientific) and incubation at 37°C for 1 h for digestion. DNA was resolved on the 16% acrylamide gels with 0.1% SDS to remove any remaining bound protein. Gels were then phosphorimaged with a Storm 820 phosphor imager (GE Healthcare) and quantified for the unwinding rate.
ATPase Assays-SsoMCM variants (2 M) were incubated in the absence or presence of unlabeled forked DNA as described previously (33). Briefly, 30-l reactions were incubated at 60°C for 5 min, and ATP was added to initiate the reaction. Samples were quenched at 5, 10, and 15 min after initiation in equal volumes of 0.7 M formic acid. A total of 0.8 l of quenched reaction was spotted on Millipore TLC PEI Cellulose F, allowed to dry, resolved in 0.6 M potassium phosphate (pH 3.5), phosphorimaged, and quantified for the ATPase rate (pmol/min).

HDX-MS Analysis of SsoMCM Homohexamer-We used
HDX-MS to study the SsoMCM hexamer solution conformation and the ssDNA binding locations. Digestion with protease XIII yielded Ͼ574 overlapping peptides covering 98.7% of the sequence for SsoMCM (Fig. 1, A and B). Deuterium uptake (deuteration level, percentage) was measured for three experimental conditions: 1) apo-SsoMCM, 2) SsoMCM with 3Ј-tail DNA, and 3) SsoMCM with 3Ј-tail/5Ј-tail fork DNA ( Fig. 1, C-F). Examination of the apo-SsoMCM data reveals significant correlation with solvent-exposed regions within a previously developed hexameric model (6-fold symmetry) of the monomeric SsoMCM crystal structure (Protein Data Bank code 3F9V) ( Fig. 2) (14). There is one other nearly full-length crystal structure of a hexameric archaeal MCM that consists of a hybrid of S. solfataricus (residues 7-269) and Pyrococcus furiosus (residues 257-361 and 729 -965) (Protein Data bank code 4POG) (15). Although the two structures have similar folds and interactions between subunits, the A-subdomain in the hybrid structure is displaced outward compared with 3F9V, resulting in even more solvent-exposed surface between the A-subdomain and the C-subdomain as well as between subunits, which is less consistent with our HDX data. Therefore, we shall primarily correlate our HDX-MS data with the hexameric model, 3F9V.
Analysis of the relative deuterium uptake for the apo-structure alone and binning the values at each residue show regions (i, ii, and iii) that are solvent-exposed in the hexamer model, with low deuterium uptake (Fig. 2, B-D). We compared the HDX data with both hydrophobic and electrostatic surfaces for regions i-iii (Fig. 3). Region i is consistent with a hydrophobic patch on the N-terminal domain. This apparent inconsistency can be explained as the segment that interacts with the first 1-6 residues that are absent from the x-ray structure.
Region ii exists in a combined hydrophobic and electrostatic pocket within the interface between subunits (residues 108 -115 with 219 -226). Our HDX data seem to indicate a tighter compaction between subunits in this area than represented in the model. Given that the hexameric structure for SsoMCM is modeled based on 6-fold symmetry from a monomeric crystal structure (14), precise intersubunit contacts containing specific hydrophobic patches may be poorly represented by the model. Analysis of a helical filamentous SsoMCM structure (34) shows a tighter compaction in this area (region ii) between subunits (data not shown), but it is arranged in a much larger oligomeric structure and shifted slightly between subunits, which is not consistent with our hexameric solution state. It is possible that the solution state of hexameric SsoMCM adopts an intermediary structure between 6-fold symmetry and helical filament that protects region ii from significant deuterium exchange.
The third apparent surface-exposed area with minimal deuterium uptake (region iii) is in a hydrophobic patch in the C-terminal domain and consists of residues 328 -334, 472-473, and 562-568. Region iii is in a hydrophobic pocket at the base of the  (14). Because this is the only other region that shows significant deviation between the hexameric model and our HDX data, we propose that the difference is due to the flexible C-terminal helix-turn-helix (HTH) domain interacting at this site. The HTH domain is absent from all available x-ray structural data and is typically removed for crystallization trials because of its flexibility; however, a recent NMR structure of the HTH domain alone has been determined and docked onto the nearly full-length x-ray structure at the ␣6 -␣8 helices (35). Unlike other data supporting this hypothesis, our HDX experiments include all residues of SsoMCM and can therefore directly identify protected regions unobserved in other structurally modified experiments. Our results most likely indicate that the HTH domain is indeed binding directly to the ␣6 -␣8 helix region or otherwise causing some conformational rearrangement at that site, transmitting DNA binding information through to the ATPase site.
Deuterium Uptake Kinetic Differences: SsoMCM Bound to DNA Substrates-To detect possible DNA binding surfaces, we performed HDX-MS experiments for SsoMCM bound to 3Ј-tail DNA or fork DNA with a 3Ј-tail/5Ј-tail. The difference in deuteration level for each DNA binding experiment relative to apo-SsoMCM is averaged over time for every residue (Equation 1). Even if the ssDNA binds the SsoMCM surface stably, the experimental HDX data are difficult to interpret because each subunit is identical. Thus, very small differences contributed from DNA-bound subunits may be lost in averaging due to contributions from the unbound and unperturbed subunits. In fact, using ARDD, we were able to detect significant perturbations only for the 3Ј-tail/5Ј-tail substrate relative to the apo-SsoMCM experiment. Instead, for the 3Ј-tail substrate alone, we analyzed relative fractional exchange for each peptide along with the corresponding S.E. values (36). Regions that do not absolutely mirror one another have differences in relative deuterium uptake or the H/D exchange rate constant indicative of either structural changes or DNA binding. "Relative fractional exchange" describes the deuterium uptake for each peptide segment. For example, if 5-deuterium incorporation is observed for a peptide with 10 exchangeable amide hydrogens, the "relative fractional exchange" for that peptide is 0.5. ARDD describes the difference in deuterium incorporation under two distinctive conditions. For example, if 5-deuterium uptake is observed for a peptide from the apoprotein, and that same peptide incorporates 4 deuteriums after binding of DNA, ARDD  for that peptide is calculated to be Ϫ20%. The relative fractional exchange butterfly plots are able to identify HDX differences within identical peptides between experimental conditions for a more focused analysis.
Butterfly plots showing the relative fractional exchange for apoversus 3Ј-tail DNA identify altered deuterium uptakes for specific peptides (Table 1), mapped onto the crystal structure (Fig. 4). Peptide 22 provides an example control peptide where no significant deuterium uptake differences are noted between conditions consistent with no binding perturbation at that site (Fig. 4B). The regions of reduced deuteration (shades of blue) exhibit putative locations in which the 3Ј-tail binds SsoMCM. Surprisingly, few of these locations are in the interior channel of the SsoMCM hexamer. Peptide 56 (residues 183-194) (Fig. 4C) makes some contacts on the interior, and residues within this region have been identified previously as part of an MCM single-stranded DNA binding motif (MSSB) (2). Another peptide 153 (residues 425-440) (Fig. 4D) spans residues associated with a C-terminal hairpin within the pre-sensor 1 (PS1) region that is involved in DNA binding (33). The remaining peptides map to external surface locations on the SsoMCM hexamer (Fig. 4,  F-H). It is known that SsoMCM binds and unwinds 3Ј-tail DNA with a slightly reduced affinity and reduced rate compared with fork DNA (1). The detection of deuteration changes on the exterior surface of SsoMCM could highlight an initial engaged DNA binding state for the helicase that is a precursor to loading. External regions of reduced deuteration occur at the extreme C terminus (residues 590 -599, ␣8-helix) near where the ssDNA-dsDNA junction is proposed to bind (33) and where the HTH domain is proposed to interact (35). The HTH domain itself shows no significant changes upon DNA binding. Other regions of reduced deuteration include the ␣/␤-␣ linker region primarily within the ␣-region of the C-subdomain (residues 522-534) and up through the waist, which is clearly on the exterior surface. Both of these regions have not been well characterized through mutagenesis, but it is possible that the ␣/␤-␣ linker and/or the ␣-region are altered upon DNA binding, possibly through sensing DNA from the HTH domain. Alternatively, direct binding of the 3Ј-tail to these external regions is an intriguing conformational predecessor to a fully loaded MCM hexamer on DNA.
Next, we compared apoversus 3Ј-tail/5Ј-tail (fork) DNA and again identified peptides with altered deuterium uptake (Table  1 and Fig. 5). Generally, these regions were identical to those from the 3Ј-tail only population but were generally enhanced, allowing for easier identification, probably representing a more stably bound complex. Specifically, perturbation at peptide 153 was enhanced with the 3Ј-tail/5Ј-tail DNA, which includes residue Arg 440 , which we previously identified as a critical residue for exterior 5Ј-tail binding (1), as well as most of the PS1-hp (residues 424 -439), which has also been implicated in DNA binding within the central channel (14,33). The PS1-hp is generally located within the interior channel and may become more engaged with the encircled strand once MCM is externally stabilized by a fork substrate. In addition, the regions around peptide 211 (residues 550 -562) (Fig. 5E) and peptide 222 (residues 590 -599), which encompass the ␣/␤-␣ linker and ␣-region, are enhanced. These areas seem to define an external ssDNA binding path from the DNA duplex region, through the CTD, and up to the waist of the SsoMCM hexamer. A comparison of SsoMCM bound to the 3Ј-tail or the 3Ј-tail/5-tail (Fig. 6) generally confirms the binding locations determined in the previous two plots and better highlights specific interactions on the exterior surface.
Analysis of the cumulative HDX data between apo-and the 3Ј-tail/5Ј-tail substrate that averages all peptides based on the ARDD method (Equation 1) is also generally consistent with this more selective analysis, but the degree of change between experiments is smaller due to averaging. Here, similar protected regions were identified, including residues within the ␣/␤-␣ linker and ␣-region (Fig. 7F). This method also found a region of decreased ARDD (residues 224 -250) corresponding to the N-terminal DNA-binding hairpin (NHP) (12,33). It is interesting to note that the amino acids 423-440 are associated with increased average deuteration when DNA is present over that of the apo-structure. This result is opposite to what was identified with the butterfly plots, in which peptide 153 (containing residues 425-440) shows decreased deuteration. It may be that DNA binding to one or two of the six subunits has a stabilizing effect on either the exterior surface or through the interior PS1 motif, whereas on average, the other four or five unbound subunits are more dynamic. Another possibility is that the peptides cover slightly different regions in the two analysis methods, where individual peptides are compared with averaging of multiple peptides across the same region. Nevertheless, the cumulative data suggest a longitudinal external binding path for the 5Ј-tail across the C-terminal domain and into the waist of SsoMCM.
External Surface Electrostatic Mutations Alter Deuterium Uptake Kinetics for MCM Binding to Forked DNA-Previously, we have shown that electrostatic mutations (K323A/R440A) in the waist region of SsoMCM alter the dynamics and interactions with the excluded strand and decrease its unwinding abil- ity (1). Although HDX-MS experiments with fork DNA and K323A/R440A were attempted multiple times, unexplained precipitation during the experiment resulted in an incomplete time course for HDX kinetics. Instead, we compared the ARDD for SsoMCM (K323A/R440A) (apoversus 3Ј-tail DNA) (Fig. 7, A-E) with that above for WT SsoMCM. There were obvious changes in the external binding sites for DNA for SsoMCM (K323A/R440A), as indicated by the presence of additional peptides with reduced ARDD (109 -125 and 571-578) relative to WT (Fig. 7, F versus G). In addition, peptides containing residues 18 -30 and 501-509 shown above to have reduced ARDD upon WT binding to DNA are unchanged when SsoMCM (K323A/R440A) is bound to DNA. This altered HDX profile for the mutant SsoMCM (K323A/R440A) suggests that the DNA binding path on the external surface has been altered within the C-terminal domain and is absent in the N-terminal domain (Fig. 7G).

Further SsoMCM External Mutations
Alter DNA Unwinding Ability-Based on the external DNA interactions that we have identified by HDX-MS, we chose to test further electrostatic surface waist SsoMCM mutations (K111A/R113A, K129A, and R226A) on DNA unwinding and ATPase activities. According to the SsoMCM hexamer model and the HDX data, each of the mutation positions has a slightly different environment within the same general location (Fig.  8A). Residues Lys 111 and Arg 113 encompass a positively charged area within the waist, adjacent to Lys 129 and slightly more toward the N-terminal domain than Lys 323 and Arg 440 identified previously to disrupt unwinding when mutated (1). Lys 111 is not conserved throughout the MCM family, whereas Arg 113 is conserved (Fig. 8C). Lys 129 is a nonconserved positively charged residue (adjacent to Arg 226 ), which is not as solvent-exposed as the others and is used as a control mutation. Arg 226 is a conserved solvent-exposed residue identified from the ARDD analysis which is protected from deuterium exchange upon DNA binding.
The ATPase activity for all of the mutants is stimulated in the presence of DNA (ϳ1.4-fold) similar to wild type (1.8-fold) ( Fig.  8B and Table 2) and consistent with previously published data showing typical DNA-dependent ATPase stimulation of 1.2-2.0-fold (14,33,(37)(38)(39). The double mutant (K111A/R113A) has an overall reduced ATPase rate (0.5-0.6) compared with WT and the other mutants. This decrease in ATP hydrolysis seems most likely the result of mutating both Lys 111 and Arg 113 , because K111A has normal ATPase activity. However, the stimulation in the ATPase rate with DNA for K111A/R113A suggests that binding to DNA is not perturbed.
Although the DNA-dependent ATPase stimulation observed for the mutants is largely unperturbed, the unwinding ability of the mutants is perturbed. DNA unwinding rates of both mutants containing conserved residues, Arg 113 and Arg 226 , are significantly reduced relative to wild type ( Fig. 9 and Table 2). K111A/R113A had barely detectable unwinding, similar to the K323A/R440A mutant (1), and R226A had a 2-fold decrease in the unwinding rate relative to wild type. The almost complete loss in DNA unwinding for K111A/R113A can only be partially explained by the modestly decreased ATPase rate and is more likely the result of a loss in exterior DNA binding in combination with reduced ATPase activity. The results for K111A/R113A are very similar to those for K323A/R440A, shown previously to disrupt exterior interactions and eliminate unwinding (1). The control nonconserved K129A mutation has unwinding activity indistinguishable from that of wild type. Therefore, mutation of conserved positively charged surface residues identified within the waist of SsoMCM has severe effects on external surface DNA interactions that translate to reduced unwinding activity.

Discussion
Using a global experimental approach, we can now validate the SEW model of unwinding for the hexameric SsoMCM DNA SsoMCM and 3Ј-tail/5Ј-tail DNA (छ, negative values) relative fractional exchange versus peptide number. *, peptides with significant differences. B-E, deuterium incorporation during the H/D exchange period for four representative peptides, including one with no significant change (B) and three others (C-E) with significant differences between the two conditions (E, apo, blue; छ, 3Ј-tail/5Ј-tail, green). Time course deuterium incorporation levels were generated by an MEM fitting method as described under "Experimental Procedures." F-H, peptide regions with significant deuterium uptake differences are mapped on the surface of SsoMCM, highlighting monomer (F), the interior channel (cutaway of top two subunits) (G), and the exterior surface (H).
replication helicase. High resolution HDX-MS monitored the deuterium exchange kinetic differences, comparing unbound and bound SsoMCM with two different DNA substrates in solution. It is interesting to note that the HDX profiles were generally similar for a 3Ј-tail only substrate and the 3Ј-tail/5Ј-tail substrate, indicating both internal and external binding paths. The 3Ј-tail DNA was detected primarily on the exterior surface along the proposed external binding path, demonstrating a probable intermediate state of DNA engagement or a nonproductive binding mode. However, the forked substrate (3Ј-tail/ 5Ј-tail) generally reduced the deuterium exchange rate further relative to the 3Ј-tail only DNA, strengthening the identification of both the internal channel and external surface ssDNA binding paths and stabilizing the forked DNA-SsoMCM complex.
The symmetrical SsoMCM hexameric model (14) as a whole is a good representation of the oligomeric structure detected by HDX-MS. The HDX data were able to determine the probable binding surfaces of missing residues and domains from the x-ray structure. Discrepancies between our HDX data and the x-ray structure are also noted in the A-subdomains, which are known to adopt different conformations depending on the species and nucleotide bound state, and mutations within that site can modulate its position (15,16,34,40). This protruding A-subdomain may be one area where significant dynamics or conformational change occurs in solution to enforce contacts between subunits or engage DNA binding along the outside surface.
Within the interior channel, two ␤-hairpins (NHP and PS1hp) and another single-stranded binding element (MSSB) showed changes in deuteration kinetics upon DNA binding. The NHP and the MSSB motif are located in the NTD, whereas the PS1-hp is located in the CTD. All three motifs have been shown previously to be important for DNA binding and unwinding (2,33). The other peptides containing the PS1-hp traverse from the interior to the exterior surface, making it difficult to con- clude exactly which region shows the most change. Both residues Lys 430 (PS1-hp, internal DNA binding) (33) and Arg 440 (external ssDNA interaction) (1) are contained within this peptide. Lys 430 has been shown previously to be important for DNA binding and unwinding, whereas Arg 440 is implicated as an   Binding of fork DNA may more precisely position the duplex region at the CTD and prevent nonspecific binding of ssDNA internally at the NTD in the opposite orientation. Although the length of ssDNA (30 bases) is long enough to traverse the central channel, the exact path is unknown and might include spiraling or interactions that preclude binding to the NTD MSSB when in the preferred orientation. More than likely, the six identical NHPs within the central channel exchange binding of the encircled strand, making detection of specific binding difficult using this analytical method. Moreover, HDX data mapped onto the apo-hexameric structure of SsoMCM show that solvent exchange within the central channel is as rapid as on the external surface. Therefore, although we expected to find significant protection within the central channel upon DNA binding, in hindsight, the core is large enough to accommodate significant solvent as well as facile exchange and binding of the encircled strand with multiple subunits within close proximity. Another possible explanation for detecting primarily exterior binding surfaces is that increased temperature is required to induce conformational changes within the hexamer or promote DNA encircling through thermal energy for this thermophilic SsoMCM. Because our HDX experiments were performed near 0°C, thermally induced DNA binding would not be detected. DNA unwinding experiments for SsoMCM are generally performed at 60°C and with ATP, at which encircling can be induced. However, DNA binding experiments, including EMSA and anisotropy, performed at room temperature show no significant differences at higher temperatures (40 -60°C). 6 Of course, binding experiments only quantify total binding and cannot differentiate between DNA encircling and external surface binding.
Except for polarity, there is no molecular difference in 3Ј-tail DNA versus 5Ј-tail DNA. An external interaction will be isolated from near neighbor interactions with other subunits based on atomic distance. On the contrary, binding of the 3Ј-tail within the interior channel places it in close proximity to at least six identical DNA binding hairpins (33) that may easily exchange contact with the 3Ј-tail, blurring deuterium uptake detection. Binding to fork DNA would properly position the SsoMCM helicase with optimal polarity and exterior stabilizing interactions.
Although it is always possible that unknown allosteric conformational changes in the hexamer account for the differential HDX profiles measured when DNA is bound, there are at least three arguments against this. First, previous EM structures show no significant conformational changes when DNA is included except for overall stabilization of the hexamer (16,41). Second, no nucleotide was used in these HDX experiments to prevent conformational changes leading to unwinding. This limitation may also account for little detected binding within the central channel. It may be that the observed protection pattern for ssDNA will change when the helicases enters a translocation competent state; however, that would be extremely difficult to interpret using HDX-MS. Finally, all significant HDX changes occur on the surface, and no significant interfacial changes between subunits are noted.
The exterior binding site of SsoMCM for ssDNA has been shown previously to include at least basic residues Lys 323 and Arg 440 (1); however, it is likely that more residues are involved in DNA binding, both along the waist and along the longitudinal exterior surface, as indicated by the HDX data. It is inter-6 M. A. Trakselis, unpublished observations. esting to note that HDX experiments performed on SsoMCM (K323A/R440A) show a change in the external DNA binding path from the CTD and up through the waist. Moreover, protection seen on the NTD with WT SsoMCM disappears when K323A/R440A is used. This change in the binding path, as well as reduced contact in the NTD, is consistent with the changes in single molecule FRET ssDNA interactions seen for this mutant (1). It also suggests that multiple contacts exist along the exterior length of SsoMCM, providing significant binding stability and contributing substantially to the mechanism of unwinding.
Additional conserved positively charged residues Arg 113 and Arg 226 within the waist region have now also been identified as not only being protected from deuterium exchange (in the case of Arg 226 ) but also shown to be essential for efficient DNA unwinding activity. Because both of these residues reside within the A-subdomain of SsoMCM, external interactions with ssDNA are now validated up through the CTD and into the NTD. Arg 113 and Arg 226 (in the NTD) combined with Lys 323 and Arg 440 (in the CTD) now provide unequivocal functional evidence for excluded strand interactions with the large waist region of the hexamer. The exact path and specific regions identified in the HDX ARDD method from the CTD up to the waist region still need to be validated biochemically. For example, further external surface mutations within the N-C linker (residues 270 -294) and the ␣/␤-␣ linker (residues 498 -515 and 522-533) show reduced HDX in the presence of DNA. Presumably, a collective external path of moderate electrostatic charge from the extreme CTD, where dsDNA is separated, up through the waist and into the NTD will be required to maintain strand separation and promote the forward action of the helicase during unwinding.
Since the SEW model was first proposed for SsoMCM (1), external excluded strand contacts with DNA have been revealed for other helicases and shown to be important for unwinding. Even monomeric superfamily 1b helicase, Dda, and superfamily 2 helicase, HCV NS3, make contacts with the excluded strand at the fork and during unwinding (42,43). Of course, other well studied recombination helicase nucleases, RecBCD and AddAB, have been known to interact and translocate on both strands with opposite polarity to separate dsDNA, digest ssDNA, and identify recombination hot spots (44 -46). In addition to archaeal MCM, there are other hexameric helicases for which contact with the excluded strand provides some stability to unwinding and replication at the fork, including T7 gp4 (47), Escherichia coli DnaB (48), and within the eukaryotic CMG complex (8,49).
It is interesting that this external mechanistic contact seems to be shared from monomeric to hexameric helicases across all forms of life but surprisingly has not been thoroughly characterized previously. Instead, a sole focus on contacts within the interior channel or with the translocating strand has limited our understanding of the complete DNA unwinding mechanism. In fact, the work presented here suggests that external surface contacts are probably very important to the mechanism, as evidenced by nearly complete loss of unwinding activity with mutation of certain external surface residues, while imparting essentially no allosteric change to the inner channel solvent accessibility or ATPase activity. Recently, a new model of unwinding has been proposed for eukaryotic CMG that includes a "pumpjack" motion that involves linear translocation along DNA and a large ATPase-induced conformational change that propels the complex forward (50). In this model, exterior interactions with the excluded lagging strand are proposed to stabilize forward unwinding steps consistent with our SEW model.
The external DNA contacts identified for the SEW unwinding model may also contribute during the helicase loading process. The mechanism for loading of hexameric helicase onto and encircling of DNA is still unknown, but it most likely requires origin-binding proteins and helicase accessory proteins to distort the origin of replication and pry open the hexameric helicase (51)(52)(53). Maintaining contact with the excluded strand while encircling the translocating strand would be more effective in sustaining an open dsDNA bubble for loading helicases. Additional work will be required to determine whether the external interactions detected here are also utilized for loading and activation of archaeal MCM and other hexameric replication helicases or whether mutations at these sites reduce loading at origins. The diversity and length of contacts on the exterior surface of hexameric helicases with the excluded strand may also be an opportunistic target for inhibition of loading and unwinding across domains.