Functional Dynamics in Replication Protein A DNA Binding and Protein Recruitment Domains

Functional Dynamics in Replication Protein A DNA Binding and Protein Recruitment Domains

Article Functional Dynamics in Replication Protein A DNA Binding and Protein Recruitment Domains Graphical Abstract Authors Chris A. Brosey, Sarah E...

3MB Sizes 0 Downloads 42 Views

Article

Functional Dynamics in Replication Protein A DNA Binding and Protein Recruitment Domains Graphical Abstract

Authors Chris A. Brosey, Sarah E. Soss, ..., Kavita Dorai, Walter J. Chazin

Correspondence [email protected]

In Brief Replication Protein A scaffolds DNA processing machinery through dynamic interactions with ssDNA and partner proteins. Brosey et al. demonstrate motional autonomy of ssDNA binding and protein interaction modules and propose a critical role for inter-domain linker length in controlling RPA functional dynamics.

Highlights d

DNA binding dramatically reorients and couples the interdomain motion of RPA70AB

d

RPA70N protein interaction domain has structural and dynamic autonomy from RPA70AB

d

RPA70N remains autonomous from RPA70AB when ssDNA is engaged

d

Linkers between the globular domains are proposed to control RPA functional dynamics

Brosey et al., 2015, Structure 23, 1028–1038 June 2, 2015 ª2015 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.str.2015.04.008

Structure

Article Functional Dynamics in Replication Protein A DNA Binding and Protein Recruitment Domains Chris A. Brosey,1,6 Sarah E. Soss,1 Sonja Brooks,2 Chunli Yan,3 Ivaylo Ivanov,3 Kavita Dorai,5 and Walter J. Chazin1,4,* 1Departments

of Biochemistry and Chemistry, Center for Structural Biology, Vanderbilt University, Nashville, TN 37232-8725, USA of Biological Sciences, Vanderbilt University, Nashville, TN 37232, USA 3Department of Chemistry, Georgia State University, Atlanta, GA 30302-4098, USA 4Department of Chemistry, Vanderbilt University, Nashville, TN 37232, USA 5Department of Physics, Indian Institute of Science Education and Research (IISER), Mohali, Sector 81 Manauli PO, SAS Nagar, Punjab 140306, India 6Present address: Department of Biochemistry and Molecular Biophysics, Washington University, St. Louis, MO 63110, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.str.2015.04.008 2Department

SUMMARY

Replication Protein A (RPA) is an essential scaffold for many DNA processing machines; its function relies on its modular architecture. Here, we report 15 N-nuclear magnetic resonance heteronuclear relaxation analysis to characterize the movements of single-stranded (ss) DNA binding and protein interaction modules in the RPA70 subunit. Our results provide direct evidence for coordination of the motion of the tandem RPA70AB ssDNA binding domains. Moreover, binding of ssDNA substrate is found to cause dramatic reorientation and full coupling of inter-domain motion. In contrast, the RPA70N protein interaction domain remains structurally and dynamically independent of RPA70AB regardless of binding of ssDNA. This autonomy of motion between the 70N and 70AB modules supports a model in which the two binding functions of RPA are mediated fully independently, but remain differentially coordinated depending on the length of their flexible tethers. A critical role for linkers between the globular domains in determining the functional dynamics of RPA is proposed.

INTRODUCTION Replication protein A (RPA), the primary eukaryotic singlestranded (ss) DNA binding protein, is an essential factor required for maintenance and propagation of the genome. RPA functions as a scaffold interacting with the substrate DNA and other proteins to facilitate the assembly and disassembly of complex DNA processing machines (Fanning et al., 2006; Sugitani and Chazin, 2015; Wold, 1997). Its ability to bind and integrate assemblies in constant flux arises from RPA’s own dynamic, modular architecture. RPA is a heterotrimer (RPA70, RPA32, RPA14) with seven globular and one disordered domain, which are organized into five distinct structural modules connected by flexible linkers (Figure 1). The core of the trimer comprises

one domain from each subunit (RPA70C/32D/14). Three of the remaining modules are attached to the core by flexible linkers (RPA32N, RPA32C, RPA70AB), and the fourth, RPA70N, is flexibly linked to 70AB (Figure 1). Except for the disordered RPA32N, structures of these RPA modules have been determined at atomic resolution, and their respective biochemical contributions to ssDNA binding and protein interaction during DNA processing have been characterized (Bochkarev et al., 1997, 1999; Bochkareva et al., 2001, 2002, 2005; Deng et al., 2007; Fan and Pavletich, 2012; Jacobs et al., 1999; Mer et al., 2000). To translate structural information on the full-length protein into understanding functional outcomes, however, it is essential to define the time-dependent disposition of each module (architecture) within the full-length protein, the relative movements of the domains, and the alterations in these movements associated with different functional states. The two modules RPA70AB and RPA70C/32D/14 form the DNA-binding core of RPA (RPA-DBC), whereas the 70N and 32C modules are dedicated to interactions with other DNA processing protein partners. The ssDNA is bound with a 50 / 30 polarity from domains A to D as a byproduct of the higher ssDNA affinity of the tandem RPA70AB domains relative to the RPA70C and RPA32D domains (Fanning et al., 2006). X-ray crystal structures, and NMR and SAXS studies, of RPA70AB generated initial insights into how this module binds ssDNA (Arunkumar et al., 2003; Bochkarev et al., 1997; Bochkareva et al., 2001; Pretto et al., 2010). Recent studies of RPA-DBC using scattering and computational approaches have provided a more complete picture of DNA binding (Brosey et al., 2013). RPA has two DNAbinding modes. The first mode involves the tandem high affinity DNA-binding domains RPA70AB engaging 8–10 nucleotides of substrate. In the second mode, the lower affinity DNA-binding domains of the trimer core (70C, 32D) bind, extending the occluded site size to 24–30 nucleotides. RPA70AB also participates in protein-protein interactions, which are understood to be important in modulating its interaction with ssDNA and facilitating interconversion between different DNA-binding modes (Arunkumar et al., 2005; Jiang et al., 2006). Previous analyses of full-length RPA and tandem domain fragments by nuclear magnetic resonance (NMR) spectroscopy and small-angle X-ray scattering (SAXS) have revealed that the five modules of RPA are structurally independent and occupy a

1028 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved

Figure 1. Domain Organization of RPA RPA is a heterotrimer of subunits RPA70, RPA32, and RPA14, which associate through interactions between domains 70C, 32D, and 14.

range of inter-domain orientations in solution (Arunkumar et al., 2003; Brosey et al., 2009, 2013; Pretto et al., 2010). Binding of ssDNA couples the two modules of the DNA-binding core (RPA70AB and RPA70C/32D/14) (Brosey et al., 2013; Fan and Pavletich, 2012) and restricts their inter-domain orientations (Brosey et al., 2009, 2013; Pretto et al., 2010) but does not appear to influence the modules dedicated to interactions with DNA processing proteins (70N and 32C) (Brosey et al., 2009; Pretto et al., 2010). This autonomy between the ssDNA binding and protein interaction modules in both DNA-free and DNAbound states has largely been assumed, indirectly supported by NMR chemical shift data from the full-length protein (Brosey et al., 2009) and by modeling of SAXS data collected on tandem domain fragments RPA70AB and RPA70NAB (Pretto et al., 2010). Whether DNA binding imposes a subtler alteration in the dynamic sampling of architecture by the protein interaction modules of RPA, which could influence the coordination of protein and DNA substrates of RPA, is unknown. Such knowledge is critical to obtaining a fundamental understanding of scaffolding function of RPA. To address this gap in knowledge, we report here characterization of the functional domain dynamics of the RPA70AB and RPA70NAB tandem domain fragments of RPA. NMR 15N relaxation experiments were performed in the absence and presence of the ssDNA substrate to determine directly how the binding of ssDNA affects the movement of these domains. To extend the analysis, the domain diffusion parameters extracted from these experiments were integrated with hydrodynamic modeling from molecular dynamics (MD) simulations and compared with previously reported SAXS analysis of RPA. Our results support a model in which the dynamic modular architecture of RPA enables independent but coordinated binding of the ssDNA substrate and partner proteins, the degree of which is determined by the length of the inter-domain linker. We propose that these differences in the coordination of domain motions are critical to the ongoing remodeling required to scaffold dynamic DNA processing machinery. RESULTS The Inter-Domain Motion of 70A and 70B Is Partially Coupled Previous investigations of the principal DNA-binding domains of RPA, 70A and 70B, have indirectly detected the presence of inter-domain dynamics (Arunkumar et al., 2003; Brosey et al.,

Figure 2.

15

N-T1/T2 and NOE Parameters Obtained for 15N-RPA70AB

Parameters in the absence (A) and presence (B) of dT10 oligonucleotide at 800 MHz point to independent rotational motion for domains 70A and 70B in the absence of substrate and coupled motion upon binding ssDNA. Residues in the inter-domain linker are colored black. Dark horizontal lines overlayed upon 15N-T1/T2 plots represent trimmed average T1/T2 values for each domain (see Table 1). Plots are shown without error bars here for clarity; see Figures S1–S3.

2009; Pretto et al., 2010). To quantitatively characterize the motional properties of each domain, we measured 15N-T1, T2, and nuclear Overhauser effect (NOE) relaxation parameters for 15 N-enriched RPA70AB at 800 MHz (Figures 2A and S1; Table S1). Plots of T1 values versus the RPA70AB sequence reveal a slightly higher average for 70B relative to 70A, whereas T2 values exhibit a slightly lower average (Figure S1; Table S1). The differences in relaxation parameters of the two domains are most evident in the ratio between T1 and T2 (T1/T2), a compound parameter that scales with the global rate of rotational diffusion (Figure 2A, top panel). The variation in T1/T2 values between domains indicates that the two domains tumble in solution with

Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved 1029

Figure 3. RPA70AB Rotational Diffusion Tensors Support Independent Rotational Motion for Domains 70A and 70B in the Absence of Substrate and Coupled Motion upon Binding ssDNA Rotational diffusion is displayed with ellipsoids for RPA70AB (A), RPA70AB/ dT10 (B), and HYDRONMR simulations of domains 70A and 70B and 70AB (PDB: 1FGU, chain A) (C). Ellipsoid size is scaled to the rate of diffusion, and ellipsoid alignment is referenced to the inertial frame of each domain. The primary axis of diffusion, Dpar, is indicated by a thick line highlighted by crosses along the major axis of the ellipsoid.

distinct rates of diffusion. NOEs are sensitive to local flexibility within and between the two domains, and reduced NOE values are observed in the two DNA-binding loops within each domain (70A L12 residues 212–219 and L45 residues 268–274; 70B L12

residues 335–341 and L45 residues 387–390), the connecting inter-domain linker (residues 290–300), and the disordered 70B C terminus (residues 416–422) (Figure 2A, lower panel). This pattern of rapid, local motions is consistent with our previous 15 N relaxation study of domain 70A (Bhattacharya et al., 2002) and aligns well with regions of disordered density observed in published 70AB crystal structures (Bochkareva et al., 2001). Notably, the flexibility observed for residues of the 70AB linker is consistent with differences in domain diffusion reflected in the 70A and 70B relaxation parameters. To extend the analysis in a more quantitative manner we calculated rotational diffusion tensors from the relaxation data for each domain individually and together for the pair (Figure 3; Table 1). A subset of residues was selected for this analysis to ensure that only those with dynamics dominated by the global rotational motion were used. Residues with dynamics dominated by fast, local motions (NOE values less than 0.65), exchange between conformational substates, or exceptionally large deviations from the T1/T2 average (T1/T2 more than 2 SDs from the average) were excluded from this calculation. An initial comparison of the resulting diffusion tensor calculations reveals that the data are better fit when each domain is considered individually, rather than as a single, tandem unit (cf. reduced c2 values in Table 1), suggesting autonomous domain diffusion of the two domains within RPA70AB. For both domains, the diffusion is best modeled as axially symmetric rotation. Notably, the average rates of diffusion, Diso, for each domain are distinct (1.15 3 107 s1 for 70A; 1.07 3 107 s1 for 70B), and correspond to a 1-ns difference in correlation times (70A, 14.5 ns; 70B, 15.5 ns) (Table 1). This difference arises at least in part because the orientation of the AB linker with respect to the molecular frame is different for the two domains (see Figure 8A) and because domain B has two flexible extensions from the globular core (the AB linker and C-terminal tail), whereas domain A has only one (the AB linker). The orientations of domain diffusion with respect to the inertial frame (determined from the RPA70AB crystal structure [Bochkareva et al., 2001]) are also non-identical, indicating that each domain tumbles with respect to its own internal frame of reference (Figure 3A; Table 1). The diffusion tensor analysis supports the presence of independent domain motion within RPA70AB. Evaluating the degree of inter-domain independence is difficult, however, without external references for domain motion. To clarify whether interdomain diffusion is more akin to the diffusion of single, isolated domains or to the diffusion of a single structural module with fixed inter-domain orientation, we used HYDRONMR (Bernado et al., 2002) to simulate 15N relaxation and rotational diffusion parameters for RPA70A, RPA70B, and RPA70AB using coordinates from the published RPA70AB crystal structures (Bochkareva et al., 2001). The T1/T2 values simulated for individual 70A and 70B domains fall below the observed experimental averages, but reproduce the slightly higher average T1/T2 ratio of 70B relative to 70A (Figure S4; Table S2). In contrast, The T1/T2 values simulated for the tethered domains are much higher than the experimental data, suggesting that RPA70AB domain diffusion more closely resembles that of the isolated domains (Figure S4). This trend is not the same as for the rotational diffusion parameters, as experimental correlation times (tm) (14.5 ns and 15.5 ns for A and B, respectively) fall approximately halfway

1030 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved

Table 1. Rotational Diffusion Parameters for 15N-RPA70AB and 15N-RPA70NAB in the Absence and Presence of dT10 Oligonucleotide Substrate at 800 MHz No. Residues Model RPA70AB

RPA70AB/ dT10

RPA70NAB

tm (ns)

p Valuea

Diso (3 107 s1) Dpar/Dper

a (degrees)b b (degrees)b Reduced c2

70A

55

axial

0.006

14.5 (0.7) 1.15 (0.06)

1.26 (0.07)

86 (14)

165 (72)

1.4

70B

64

axial

0.006

15.5 (0.7) 1.07 (0.05)

1.21 (0.07) 173 (18)

60 (17)

1.3

70AB 119

axial

0.007

15.0 (0.6) 1.11 (0.06)

1.17 (0.08) 166 (17)

16 (74)

2.1

70A

57

axial

2.06 3 106

16.7 (0.5) 1.00 (0.04)

1.24 (0.05) 172 (17)

73 (16)

1.9

70B

66

axial

1.92 3 107

16.7 (0.4) 1.00 (0.03)

1.28 (0.05) 182 (12)

100 (9)

2.4

70AB 124

axial

1.07 3 1012 16.7 (0.3) 0.99 (0.02)

1.27 (0.04) 124 (8)

176 (117)

2.2

1.5 3 104

12.0 (0.6) 1.39 (0.07)

1.36 (0.08)

66 (14)

9.5

17.6 (0.4) 0.95 (0.02)



70N

47

axial

70A

27

isotropic –

70 (16) –



6.3

70B

39

axial

0.005

17.3 (2.3) 0.96 (0.15)

1.65 (0.48)

22 (18)

145 (26)

5.2

RPA70NAB/ 70N dT10 70A

51

axial

3.9 3 104

11.3 (0.6) 1.48 (0.08)

1.29 (0.09)

74 (20)

80 (15)

11.4

20

isotropic –

20.4 (0.5) 0.82 (0.02)







3.4

70B

32

isotropic –

21.0 (0.4) 0.79 (0.01)







4.5

a

A p value cutoff of 0.01 was used to determine the significance of applying a more complex diffusion model to the data. F statistics were calculated by ROTDIF. b Orientation angles a and b describe sequential polar and azimuthal rotations around the inertial z-axis (a) and y-axis (b).

between the two calculations (70A: 7.6 ns versus 22.4; 70B: 9.5 ns versus 22.4 ns; Table 2). However, the differences in the trends are not surprising because, although they reflect similar characteristics, the relationship between the T1/T2 and tm parameters is not linear. Examination of the orientation of diffusion reveals that the calculated rotational diffusion tensor for the isolated domains is perfectly aligned to their inertial frames, while that of the linked domains is rotated nearly 90 (Figure 3C). The orientation of the experimental diffusion tensor is slightly offset from the inertial frame for both domains, but more similar to the calculations for the isolated domains than the linked domains (cf. Figures 3A and 3C). Overall, the HYDRONMR calculations suggest that RPA70AB motion is dominated by independent diffusion of the individual domains, but with some influence of their being tethered, resulting in partial coupling of their motion. Our observations of tethered motion are consistent with the systematic analysis of domain diffusion from NMR relaxation analysis of tethered protein GB1 domains (Walsh et al., 2010), as well as matrix metalloproteinases (MMPs) 1 and 12 (Bertini et al., 2008, 2009). While the crystal structures represent informative extremes for 70AB domain diffusion (fully independent or fully coupled), these single, static snapshots are limited in their ability to capture the full range of conformational space sampled by the 70A and 70B domains. To determine if we could generate a more realistic representation, we performed an ensemble-based HYDRONMR calculation using a set of 150 RPA70AB coordinates extracted from a previously published MD trajectory of the RPA-DBC (Brosey et al., 2013). Calculations were performed with and without the 13 residues at the C terminus of 70B, as the helix formed by these residues was observed to unravel and detach from the base of the 70B domain during the course of the simulation. Notably, ensemble-averaged T1/T2 values excluding the C-terminal residues corresponded well to those calculated from the RPA70AB crystal structures (Figure S4A), while inclusion of the mobile C terminus elevates the global T1/T2 average across both domains (Figure S4B). In both cases, T1/T2 values are higher than the experimentally measured values. Similarly, the

average rates of rotational diffusion for the MD ensemble are slower than that measured experimentally (Table S3). Thus, ensemble averaging of diffusion parameters across multiple, static conformations of RPA70AB appears to be essentially equivalent to simulating the diffusion of a single, static conformation, as represented by the RPA70AB crystal structure. We conclude that treating the solution ensemble as a population of conformers with fixed inter-domain orientations in the hydrodynamic calculation fails to fully capture the diffusive motion of RPA70AB, i.e. that there is additional motion that is not adequately accounted for in this model and that a more complex modeling approach is required (vide infra). DNA Binding Reorients and Couples 70A and 70B Inter-Domain Motion To assess the impact of DNA binding upon the motion of the 70A and 70B domains, analysis of 15N relaxation parameters was extended to 15N-enriched RPA70AB bound to a ten-nucleotide ssDNA substrate. In contrast to the DNA-free state, plots of 15N relaxation parameters including T1/T2 ratios for the RPA70AB/dT10 complex are uniform between 70A and 70B (Figures 2B and S2; Table S1), indicating that the two DNA-binding domains share similar rates of rotational diffusion when bound to substrate. The higher values of T1/T2 reflect a slowing of the rotational diffusion of the DNA-bound state relative to the DNA-free state (Table S1). Consistent with coupled inter-domain motion, NOE values for residues within the 70AB linker are significantly elevated relative to the DNA-free state (Figure 2B), reflecting a loss in local flexibility and in architectural sampling. This arises from the alignment of the two globular domains to bind the ssDNA, as seen in the crystal structure of RPA70AB/dC8 complex (Bochkarev et al., 1997). Similarly, NOE values for residues within the DNA-binding loops of each domain are now closer to the average, as binding the ssDNA quenches their local flexibility (Bhattacharya et al., 2002). In contrast to the DNA-free state, diffusion tensors calculated for the DNA-bound complex are fit equally well whether the 70A and 70B domains are considered individually or together

Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved 1031

Figure 4. RPA70AB Bound to ssDNA Substrate, but not the Free Protein, is Well Described by Classic Model-Free Analysis Residue plots of order parameters (S2) and internal correlation times (te) for RPA70AB (A) and RPA70AB/dT10 (B). Error bars represent standard fitting errors reported by the relax software.

(Table 1). The rates of rotational diffusion are identical for both calculations (1.00 3 107 s1), consistent with the uniformity observed in the 15N relaxation parameters. Notably, the orientation of rotational diffusion is also similar for each domain (Figure 3B), and the primary axes of motion are orthogonal relative to those observed in the DNA-free state (Figure 3A), suggesting that the two domains now rotate together through solution around an axis parallel to the bound ssDNA. The 15N relaxation parameters and diffusion tensor analysis indicate that the motion of 70A and 70B is coupled when bound to ssDNA substrate. To evaluate whether this is consistent with a complete dynamic integration of the domains, we compared 15N relaxation and rotational diffusion values simulated from the crystal structure of the RPA70AB/dC8 complex (Bochkarev et al., 1997) to the experimental data (Figure S4; Table S2). Unlike the DNA-free state, there is excellent correspondence between the rate and orientation of RPA70AB rotational diffusion for experimental and simulated parameters of the DNA-bound state, indicative of coupling of the motions of the two domains. Thus, binding of ssDNA reorients and unifies the tumbling of 70A and 70B about the ssDNA substrate. We also performed ensemble-based HYDRONMR calculations on 100 RPA70AB/dC10 conformers extracted from a previously published MD trajectory (Brosey et al., 2013). As for the free RPA70AB, calculations were performed with and without 70B C-terminal residues 416 to 422. The resulting ensembleaveraged T1/T2 values calculated in the absence of the mobile C terminus of domain B demonstrate close correspondence to the experimental data (Figure S4; Table S3), reflecting reduced motion between the domains in a manner similar to the crystal structure. Notably, T1/T2 values calculated including the C-terminal residues of 70B are elevated relative to the experimental data, suggesting that the additional hydrodynamic drag from the simulated C terminus slows the overall tumbling of the system (Figure S4B). The better alignment of the experimental data with simulated values calculated with exclusion of the 70B C terminus suggest that the unraveling of the C-terminal a-helix observed in the simulations does not appear to occur in solution when ssDNA is bound, or at least is not detected by these NMR relaxation experiments.

Model-Free Analysis Reflects the Transition in 70A and 70B InterDomain Dynamics upon Binding ssDNA In order to obtain deeper insights into the motional dynamics of RPA70AB and its response to binding ssDNA, we performed a model-free analysis of the relaxation data. The classic Lipari-Szabo model-free formalism provides a minimalist parameterization of protein motion, describing the timescale of global protein tumbling (tm) and the amplitude and timescale of local backbone fluctuations (S2 and te) independent of a pre-determined geometric model (Lipari and Szabo, 1982a, 1982b). Analysis of local backbone order parameters (S2) and internal correlation times (te) along the protein backbone reveal that the DNAbound complex is reasonably well described by this model of protein motion, with an average S2 of 0.90 and te of 75 ps (Figure 4B). Although the average S2 value is slightly high, these values overall are consistent with those observed for globular proteins (Goodman et al., 2000). In the absence of ssDNA substrate, however, values of S2 and te are unrealistically high, with an average S2 of 0.96 and half of the residues having te values in excess of 100 ps (Figure 4A), indicating that the classic model-free formalism fails to fit the increased complexity of RPA70AB motion in the DNA-free state. Since DNA-free RPA70AB is not well fit by the classic model, we turned to the more complex EMF formalism to better characterize the residual inter-domain motion not captured in the diffusion tensor analysis (Clore et al., 1990). We considered acquiring more accurate NOE values (Ferrage et al., 2010), but the effects on the NOEs would not be sufficiently large to obviate the extended model-free (EMF) model for the DNA-free protein. The limitations of the model-free formalism to describe systems with complex diffusive motion (time-varying rate or time-varying orientation of the diffusion tensor) are well established (Wong et al., 2009). EMF analysis adds a second component to the description of local N-H bond motion to include fast [S2f, tf] and slow [S2s, ts] motions, which in this case allows for separation of the local motion within the domain and the inter-domain motion, respectively. Since EMF analysis requires five independent parameters, we obtained additional 15N-T1 and NOE datasets at 600 MHz (Figure S5). The approach to this analysis followed our previous published study (Soss et al., 2013). Although a global correlation time (tm) value is derived from the diffusion tensor analysis, the interdomain motion and overall tumbling are specifically parsed in the EMF approach. Hence, the analysis begins with a series of

1032 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved

more realistic model parameters than the standard model-free analysis: the average S2 is 0.86 and ts 1.86 ns, respectively (Figures 5B and 5C). Notably, comparison of average EMF parameters for 70A and 70B reveal differences in S2 and ts values (Figure 5B), reflecting differences in motion similar to that observed in the diffusion tensor analysis. Thus, the EMF analysis specifically confirms the independent domain motion within RPA70AB in the absence of ssDNA. That the data for the ssDNA complex were well fit with the simpler, standard model-free analysis confirms that the inter-domain motion in RPA70AB is essentially quenched upon binding ssDNA.

Figure 5. EMF Analysis of Free RPA70AB (A) Total goodness of fit (c2) for EMF solutions across a range of global correlation times. (B) EMF parameter averages. (C) Residue plots of order parameters (S2) and internal correlation times (te) for tm 18 ns. Error bars represent standard fitting errors reported by the relax software.

simulations to estimate the overall tumbling (tm). This involved calculating EMF solutions across a range of tm values (12– 30 ns) and evaluating their goodness of fit to the data (Figure 5A). There is generally a broad range of correlation times that are reasonably well fit by the data, in this case (tm 15–20 ns). The global minimum (tm 18 ns) was selected for further analysis. Using this tm value, the EMF analysis produces a solution with

Structural and Motional Independence of the 70N Protein Interaction Domain The scaffold function of RPA requires ssDNA binding and protein recruitment activities. Having established the effect of DNA binding on the motional properties of RPA70AB, we set out to characterize the motional properties of a larger fragment of the RPA70 subunit that includes the protein recruitment domain RPA70N. To this end, a corresponding 15N relaxation analysis was performed on 15N-enriched RPA70NAB, a construct that includes 70N, 70AB, and the 60-residue linker that connects 70N to 70A (Figure 1). Plots of T1 and T2 parameters across the sequence and the averages for each domain reveal a substantial difference between the 70N and 70AB modules, including substantially lower T1 and substantially higher T2 values (Figure S6; Table S1). In contrast, NOEs, which are insensitive to global motions, are consistent across the entire protein. Comparison of T1/T2 values clearly demonstrates that the tumbling of 70N and 70AB are dramatically different, and indeed, the rate of rotational diffusion determined for 70N is considerably faster than rates calculated for 70A and 70B (Figure 6A; Table 1). Further support for the autonomy of 70N diffusion relative to 70AB diffusion was evident from the excellent agreement of diffusion rates obtained from a HYDRONMR simulation using the published NMR structure of RPA70N (Jacobs et al., 1999) (Figure 7A; Table S2). Comparison of the data for RPA70NAB and RPA70AB reveals that linking 70AB to 70N greatly increases the complexity of RPA70AB motion. This is due to the additional hydrodynamic drag created by attachment of the 60-residue N-A linker to the 70A domain. The increased complexity of motion is reflected in a decrease in the ability to fit the data to standard diffusion models (isotropic, axially symmetric) (cf. reduced c2, Table 1) and also appears to result in a higher degree of variation in the values of the 70AB 15N relaxation parameters within the A and B domains of RPA70NAB (Figure 6A). More detailed analysis of RPA70NAB inter-domain motions, even by the EMF formalism, is greatly hindered due to the complexity of analyzing three independent domains. To assess the impact of ssDNA binding on the motion of the 70N, 70A, and 70B domains, we analyzed 15N relaxation parameters for the complex with the ten-nucleotide ssDNA substrate. Consistent with previously published results (Brosey et al., 2009), the addition of dT10 to RPA70NAB does not cause perturbation to 70N peaks within the RPA70NAB 15N-1H heteronuclear single quantum coherence (HSQC) spectrum, indicating an absence of binding between 70N and the ssDNA substrate. Peaks associated with the 70A and 70B domains are shifted to

Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved 1033

Figure 7. The Rotational Diffusion of RPA70N Is Unaffected by DNA Binding (A) Overlay of 70N 15N-T1/T2 ratios in the absence (open circles) and presence of ssDNA (filled circles). 15N-T1/T2 ratios simulated from the 70N NMR structure (PDB: 1EWI) (open squares) correspond well with experimental values, indicating that the diffusion of 70N as part of RPA70NAB is similar to that of the untethered domain. (B) Ellipsoid displays of 70N rotational diffusion in the absence (left) and presence of ssDNA substrate (right). Ellipsoid size is scaled to the rate of diffusion, and ellipsoid alignment is referenced to the inertial frame of each domain. The primary axis of diffusion, Dpar, is indicated by a thick line along the major axis of the ellipsoid.

tional diffusion (Figures 6B and S7; Tables 1 and S1). Thus, as is the case for RPA70AB, the motion of 70A and 70B becomes coupled when bound to ssDNA substrate. Notably, the rotational motion of 70N remains unchanged in the presence of ssDNA; there is excellent correspondence in the relaxation parameters and the T1/T2 values between the DNA-free and DNA-bound states (Figure 7A; Table S1), and the rate and orientation of 70N rotational diffusion, are minimally altered (Figure 7B; Table 1). Thus, binding of ssDNA by 70A and 70B does not restrict the global tumbling of 70N, allowing the domain to continue to recruit and bind protein targets. DISCUSSION Figure 6. 15N-T1/T2 and NOE Parameters Obtained for RPA70NAB

15

N-

Parameters in the absence (A) and presence (B) of dT10 oligonucleotide at 800 MHz support autonomy of the 70N domain relative to 70AB. Residues in the AB inter-domain linker are colored black. Dark horizontal lines overlayed upon 15N-T1/T2 plots represent trimmed average T1/T2 values for each domain (see Table 1). Plots are shown without error bars here for clarity; see Figure S6.

positions consistent with those observed in NMR spectra of the RPA70AB/dT10 complex (data not shown). Plots of 15N relaxation parameters including T1/T2 ratios show there is a substantial effect on the 70A and 70B domains from the binding of the ssDNA, leading to an overall rise in T1/T2 ratios and slowing of the rota-

Our NMR relaxation study provides the first direct measurements of the domain dynamics of RPA. We establish that the diffusion of the high affinity DNA-binding domains 70A and 70B is partially coupled, while diffusion of protein interaction domain 70N is autonomous from the 70AB DNA-binding module. When 70A and 70B are bound to ssDNA, their motion is dramatically reoriented and fully coupled by the association with the substrate. The dynamic independence of 70N, however, remains unaffected by 70AB binding DNA. This supports a model whereby RPA maintains structural and dynamic autonomy between modules dedicated to discrete biochemical functions (i.e. binding of ssDNA or protein substrates) during DNA processing.

1034 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved

Table 2. Comparison of RPA70AB Radius of Gyration (Rg) Values from SAXS and Hydration Radii (rH) from NMR-Derived Rotational Correlation Times 70AB (A˚) 70AB/ssDNA (A˚) Rg rH

SAXSa

25.6

23.4

MDb

22.0

19.0

NMR

27.1

26.0

MDb

28.7

26.6

a

Guinier Rg from Pretto et al. (2010). MD values are calculated from models excluding residues 406–420.

b

Figure 8. The Relative Orientation of Linker Attachment Corresponds to Rotational Differences between 70A and 70B Domains (A) Comparison of linker attachment points (highlighted in red on left) and projected rotational trajectory relative to the measured diffusion tensor. (B) Cartoon demonstrating reorientation of linked domains upon binding ssDNA.

We have compared our results with our previous SAXS study of the effects of ssDNA binding on the architecture of RPA70AB and RPA70NAB (Pretto et al., 2010). Although there are currently no methods available to directly compare NMR and SAXS data, insight can be obtained by comparing radii of hydration (rH) calculated using isotropic tm values derived from the modelfree optimization to the radii of gyration (Rg) derived from Guinier analysis of SAXS data. Table 2 provides a comparison of rH and Rg values for RPA70AB in the absence and presence of ssDNA. Notably, RPA70AB rH values are higher than the corresponding Rg values measured by SAXS, which we attribute to the larger dimension of the volume swept out by protein tumbling relative to that of the intrinsic protein mass and hydration shell that is captured by Rg. To obtain further insights, we also derived rH and Rg values for RPA70AB from MD simulations. These results also produced rH values consistently higher than the corresponding Rg values (Table 2). The observation of decreases in rH and Rg in the presence of ssDNA indicates quenching of inter-domain motion and overall compaction of RPA70AB upon binding ssDNA, as was inferred previously and is now confirmed directly from the NMR relaxation analysis. Our NMR studies provide deeper insight into how the RPA70A and B domains engage ssDNA. Despite similarities in their shape and size, the two domains possess distinct orientations of rotational diffusion when linked to form the RPA70AB fragment. These differences in rotational orientation correlate well with differences in how they are attached to the intervening

linker. The 70A domain is connected to the linker behind the base of the DNA-binding cleft (Figure 8A). Hydrodynamic drag imposed at this point would be consistent with the twisting suggested by its rotational diffusion tensor. In contrast, the 70B domain is connected to the linker just beneath the DNA-binding cleft (Figure 8A). Tension applied at this point would align well with rotation about the DNA-binding cleft, as reflected in the orientation of the rotational diffusion tensor. The distinct orientations of rotational diffusion suggest how these two DNA-binding domains achieve high affinity and directional binding of ssDNA substrates. Placement of the linker away from the DNA-binding loops and the corresponding domain dynamics provides 70A with unimpeded access to engage the substrate. The rotational pivoting of 70B toward the front of the DNA-binding cleft on the one hand partially restricts its availability to bind a free ssDNA substrate, but on the other hand optimally positions it to encounter the substrate as soon as it is initially engaged by 70A. This back-to-front attachment of the linker also ensures that 70B binding occurs adjacent to the 30 side of 70A. We propose that this construction of the tandem domains is the primary driver for the 50 -30 polarity of RPA in engaging ssDNA substrates, which is fundamental to its scaffold function, because this establishes the orientation of DNA processing machinery. The length of the linker between RPA domains N and A (60 residues) is 6-fold larger than the linker between domains A and B (11 residues). Linker length plays a critical role in determining inter-domain motion, as clearly shown in a GB1 domain model system (Walsh et al., 2010) and MMPs (Bertini et al., 2008, 2009). Here, we demonstrate that substantial differences in the inter-domain dynamics within RPA correlate with the functional roles of each domain. As a domain responsible for recruiting and assembling partner proteins into the dynamic multi-protein machinery, the long linker of 70N allows it to retain complete autonomy from the DNA-binding apparatus that engages the substrate. In contrast, the short linker between domains A and B enforces the high degree of coordination that is essential to generating the high overall affinity required for the initial binding of ssDNA. Figure S8 shows alignments for the RPA70 N-A and A-B linkers from several organisms, demonstrating that linker lengths are broadly conserved. We believe that the tailoring of linker length to generate the relevant functional dynamics is a key property in the evolution of RPA and other multi-domain proteins and is critical to the ability of multi-protein complexes to function as machines. Residue content and sequence conservation are also likely to be important determinants of linker flexibility, and thus another

Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved 1035

source of functional optimization for flexibly linked domains in RPA and other modular proteins. Of note, the center of the A-B linker is marked by a highly conserved proline preceded by a hydrophobic residue (residues 295, 296; Figure S8). Along with the short length of the linker, these conserved residues may promote 70AB domain orientations that facilitate coordinated binding of ssDNA. In contrast, we do not detect any significant sequence conservation in the N-A linker, which presumably must retain dynamic flexibility along its peptide backbone to allow 70N to recruit and interface with the rest of the DNA processing protein machinery. The lack of structural order in isolated N-A linker fragments from multiple RPA homologues is consistent with flexibility within the corresponding N-A linkers (Daughdrill et al., 2007). The importance of sequence conservation (and length) in inter-domain linkers has yet to be explored but provides a fertile ground for future investigation. Characterizing the structural dynamics of modular proteins remains an ongoing challenge to the field of structural biology. NMR spectroscopy and small-angle scattering are becoming increasingly applied to define protein architecture in solution and can generate highly complementary data on the architectural ensembles populated by modular multi-domain proteins such as RPA. Scattering provides a snapshot of all architectures occupied in the ensemble and is sensitive to the population distribution. However, interpretation of scattering data for systems with high degrees of inter-domain flexibility is challenging, because the distribution of architectures within the ensemble is not defined. In particular, if the protein populates a continuum of architectures, there are multiple indistinguishable solutions that can give rise to the same scattering (Capp et al., 2014). Conversely, the NMR approach can provide local information on the distribution of architectures in solution. NMR chemical shift perturbations can report on the existence and location of short-range contacts between domains, while NMR relaxation data can supply a description of the global diffusion of each domain and establish if domain motions are correlated. Nonetheless, the architectural ensemble of the entire system can be completely defined only by NMR parameters for highly restricted domain arrangements. Proteins with substantial architectural freedom such as RPA remain a major challenge. Ideally, greater insight into the relative distribution of architectures in solution for flexible, multi-domain proteins can be achieved by directly integrating scattering data from the entire molecule with NMR-derived domain diffusion tensors. Currently, the spatial information extracted from scattering and the timedependent descriptions of motion obtained through NMR can be used independently as parameters in structure refinements. However, theoretical frameworks for integrating these experimental data and directly translating the information into models of dynamic architectural ensembles of RPA and other large modular proteins remain a significant obstacle. Once available, these formalisms for the analysis of flexible multi-domain proteins could then be extended to multi-protein machinery. Modular independence of functional domains is a critical characteristic that enables RPA to function as a flexible scaffold for the assembly and reorganization of DNA processing assemblies (Sugitani and Chazin, 2015). Conversely, it is equally important that elements of the multi-protein machinery are able to remodel the internal modular architecture of RPA to promote or disrupt

ssDNA binding. It is plausible that the native architecture of the DNA-free and DNA-bound states of RPA helps facilitate protein-driven transitions between these states. Our results indicate that the disposition of the protein interaction modules of RPA remains unaltered by association with ssDNA, suggesting that protein binding is the primary driver of RPA remodeling. Further investigations to examine how protein interaction partners alter architectural and dynamic states of RPA will provide valuable information to test and refine this hypothesis. EXPERIMENTAL PROCEDURES Materials Plasmids for RPA70AB (pSV281) and RPA70NAB (pBG100) have been described previously (Brosey et al., 2009; Pretto et al., 2010). Both constructs contain N-terminal 6X-histidine fusion tags that are cleavable by TEV (RPA70AB) or H3C (RPA70NAB) proteases. TEV and H3C proteases were produced in house. ssDNA substrates (dT10) were purchased from Integrated DNA Technologies (IDT) with standard desalting purification and resuspended in sterile water. Expression, Purification, and NMR Sample Preparation of Recombinant 15N-RPA70AB and 15N-RPA70NAB Uniformly enriched 15N-RPA70AB or 15N-RPA70NAB was expressed and purified as described in the Supplemental Experimental Procedures. NMR samples were prepared in a buffer containing 30 mM Na-citrate (pH 6.0), 100 mM NaCl, 5 mM b-mercapto-ethanol, and 10 mM ZnCl2, at a concentration of 300–500 mM. An equimolar amount of dT10 oligonucleotide was added directly to the protein concentrate for studies of the ssDNA bound state. Additional information is provided in the Supplemental Experimental Procedures. ssDNA Titration of RPA70AB by NMR To aid in transferring assignments to the DNA-bound states of domains 70A and 70B, 15N-1H HSQC spectra were acquired on a titration series of RPA70AB bound to dT10. Details are provided in Supplemental Experimental Procedures. NMR Experiments All NMR experiments were performed at 25 C using Bruker AVANCE 800 or 600 NMR spectrometers equipped with cryoprobes. Gradient-enhanced 15 N-1H HSQC spectra were recorded with 1,024 complex points in the 1H and 128 complex points in the 15N dimension. NMR data were processed using either Topspin 2.1 (Bruker Biospin) or NMRPipe (Delaglio et al., 1995) and analyzed with SPARKY v3.1 (Goddard and Kneller, 2008). Sequence-specific assignments for RPA70AB and RPA70NAB were transferred from published assignments for RPA70N and RPA70AB (Bhattacharya et al., 2004; Jacobs et al., 1999). Assignments for domains 70A and 70B in complex with dT10 were confirmed via the ssDNA titration series described above, as well as from individual ssDNA titrations of 70A and 70B reported previously (Bhattacharya et al., 2004). HSQC-based 15N-relaxation measurements of T1 and T2 values were acquired using standard inverse detected pulse sequences (Kay et al., 1989; Skelton et al., 1993), modified to include a gradient-enhanced water suppression scheme (Sklenar et al., 2002). T1 values were measured for delays of 50, 100, 200 (32), 300, 600, 1200, 2500, 4000 ms with an overall recovery delay of 5.0–6.0 s. T2 values were measured for delays 17.3, 34.6, 51.8, 69.1 (32), 86.4, 103.7, 138.2, 172.8, 207.4 ms with an overall recovery delay of 1.5 s. The 1 H-15N NOE experiment (Kay et al., 1989) was acquired with a 3-s saturation period and interleaved acquisition of saturated and non-saturated transients. Typical acquisition parameters were 128 (15N) and 1,024 (1H) points in the u1 and u2 dimensions, respectively with 24 (T1 and T2) or 100–120 (NOE) transients collected for each t1 increment. Analysis of 15N-Relaxation Data T1, T2, and NOE parameters were extracted by fitting to the relaxation data. Rotational diffusion tensors from these data were calculated using the

1036 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved

ModelFree script r2r1_diffusion (Mandel et al., 1995; Tjandra et al., 1995) and ROTDIF (Walker et al., 2004). The ellipsoid representations of each diffusion tensor were calculated using modified scripts in the program relax (d’Auvergne and Gooley, 2008a, 2008b). Model-free analysis was carried out with the program relax (d’Auvergne and Gooley, 2008a, 2008b), using diffusion tensors calculated from the experimental data or following the optimized fitting for global and inter-domain motions described in Soss et al. (2013). Details for all of these analyses are provided in Supplemental Experimental Procedures. Computational Modeling of 15N-Relaxation and Rotational Diffusion 15 N-Relaxation and rotational diffusion values were calculated for crystal structures of RPA70AB and RPA70N (above) using HYDRONMR (De La Torre et al., 2000). Coordinates for RPA70AB (150 models) and RPA70AB/dC8 (100 models) were extracted from previously published MD trajectories of RPA-DBC (Brosey et al., 2013) using Ptraj and were modified to remove 13 residues from the C-terminal 70B helix (410–422). Additional details on the generation of radius of hydration and radius of gyration values are provided in Supplemental Experimental Procedures. RPA70 Sequence Alignment The RPA70 sequence alignment was created using ClustalW (Larkin et al., 2007) and rendered with ESPript 3 (Robert and Gouet, 2014).

SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, eight figures, and three tables and can be found with this article online at http://dx.doi.org/10.1016/j.str.2015.04.008. AUTHOR CONTRIBUTIONS C.A.B. performed relaxation experiments, analyzed data, and generated HYDRONMR analyses. S.E.S. generated model-free analysis. S.B. processed and analyzed relaxation data. C.Y. and I.I. performed RPA70AB MD simulations and extracted PDB coordinates. K.D. benchmarked and helped design relaxation experiments. C.A.B. and W.J.C. wrote the manuscript. ACKNOWLEDGMENTS We thank Marie-Eve Chagot for technical assistance in protein expression and purification. This research was supported by NIH operating grants (R01 GM65484 and P01 CA092584 to W.J.C.), the National Science Foundation (NSF-CAREER MCB-1149521 to I.I.), start-up funds from Georgia State University to I.I., and center grants (P30 ES00267 to the Vanderbilt Center in Molecular Toxicology and P30 CA068485 to the Vanderbilt Ingram Cancer Center). K.D. was supported by an Indo-US Research Fellowship from the Indo-US Science & Technology Forum. Computational resources were provided in part by allocations from the NSF XSEDE program (CHE110042) and the National Energy Research Scientific Computing Center supported by the DOE Office of Science (contract DE-AC02-05CH11231). Received: January 22, 2015 Revised: March 30, 2015 Accepted: April 3, 2015 Published: May 21, 2015

Bernado, P., Garcia de la Torre, J., and Pons, M. (2002). Interpretation of 15N NMR relaxation data of globular proteins using hydrodynamic calculations with HYDRONMR. J. Biomol. NMR 23, 139–150. Bertini, I., Calderone, V., Fragai, M., Jaiswal, R., Luchinat, C., Melikian, M., Mylonas, E., and Svergun, D.I. (2008). Evidence of reciprocal reorientation of the catalytic and hemopexin-like domains of full-length MMP-12. J. Am. Chem. Soc. 130, 7011–7021. Bertini, I., Fragai, M., Luchinat, C., Melikian, M., Mylonas, E., Sarti, N., and Svergun, D.I. (2009). Interdomain flexibility in full-length matrix metalloproteinase-1 (MMP-1). J. Biol. Chem. 284, 12821–12828. Bhattacharya, S., Botuyan, M.V., Hsu, F., Shan, X., Arunkumar, A.I., Arrowsmith, C.H., Edwards, A.M., and Chazin, W.J. (2002). Characterization of binding-induced changes in dynamics suggests a model for sequencenonspecific binding of ssDNA by replication protein A. Protein Sci. 11, 2316– 2325. Bhattacharya, S., Arunkumar, A.I., Sullivan, S.L., Botuyan, M.V., Arrowsmith, C.H., and Chazin, W.J. (2004). 1H, 13C and 15N assignments of singlestranded DNA binding domains from the 70 kDa subunit of human replication protein A. J. Biomol. NMR 28, 195–196. Bochkarev, A., Pfuetzner, R.A., Edwards, A.M., and Frappier, L. (1997). Structure of the single-stranded-DNA-binding domain of replication protein A bound to DNA. Nature 385, 176–181. Bochkarev, A., Bochkareva, E., Frappier, L., and Edwards, A.M. (1999). The crystal structure of the complex of replication protein A subunits RPA32 and RPA14 reveals a mechanism for single-stranded DNA binding. EMBO J. 18, 4498–4504. Bochkareva, E., Belegu, V., Korolev, S., and Bochkarev, A. (2001). Structure of the major single-stranded DNA-binding domain of replication protein A suggests a dynamic mechanism for DNA binding. EMBO J. 20, 612–618. Bochkareva, E., Korolev, S., Lees-Miller, S.P., and Bochkarev, A. (2002). Structure of the RPA trimerization core and its role in the multistep DNA-binding mechanism of RPA. EMBO J. 21, 1855–1863. Bochkareva, E., Kaustov, L., Ayed, A., Yi, G.S., Lu, Y., Pineda-Lucena, A., Liao, J.C., Okorokov, A.L., Milner, J., Arrowsmith, C.H., et al. (2005). Singlestranded DNA mimicry in the p53 transactivation domain interaction with replication protein A. Proc. Natl. Acad. Sci. USA 102, 15412–15417. Brosey, C.A., Chagot, M.E., Ehrhardt, M., Pretto, D.I., Weiner, B.E., and Chazin, W.J. (2009). NMR analysis of the architecture and functional remodeling of a modular multidomain protein, RPA. J. Am. Chem. Soc. 131, 6346– 6347. Brosey, C.A., Yan, C., Tsutakawa, S.E., Heller, W.T., Rambo, R.P., Tainer, J.A., Ivanov, I., and Chazin, W.J. (2013). A new structural framework for integrating replication protein A into DNA processing machinery. Nucleic Acids Res. 41, 2313–2327. Capp, J.A., Hagarman, A., Richardson, D.C., and Oas, T.G. (2014). The statistical conformation of a highly flexible protein: small-angle X-ray scattering of S. aureus protein A. Structure 22, 1184–1195. Clore, G.M., Szabo, A., Bax, A., Kay, L.E., Driscoll, P.C., and Gronenborn, A.M. (1990). Deviations from the simple two-parameter model-free approach to the interpretation of nitrogen-15 nuclear magnetic relaxation of proteins. J. Am. Chem. Soc. 112, 4989–4991. Daughdrill, G.W., Narayanaswami, P., Gilmore, S.H., Belczyk, A., and Brown, C.J. (2007). Dynamic behavior of an intrinsically unstructured linker domain is conserved in the face of negligible amino acid sequence conservation. J. Mol. Evol. 65, 277–288.

REFERENCES Arunkumar, A.I., Stauffer, M.E., Bochkareva, E., Bochkarev, A., and Chazin, W.J. (2003). Independent and coordinated functions of replication protein A tandem high affinity single-stranded DNA binding domains. J. Biol. Chem. 278, 41077–41082.

d’Auvergne, E.J., and Gooley, P.R. (2008a). Optimization of NMR dynamic models II. A new methodology for the dual optimization of the model-free parameters and the Brownian rotational diffusion tensor. J. Biomol. NMR 40, 121–133.

Arunkumar, A.I., Klimovich, V., Jiang, X., Ott, R.D., Mizoue, L., Fanning, E., and Chazin, W.J. (2005). Insights into hRPA32 C-terminal domain-mediated assembly of the simian virus 40 replisome. Nat. Struct. Mol. Biol. 12, 332–339.

d’Auvergne, E.J., and Gooley, P.R. (2008b). Optimization of NMR dynamics models I. Minimization algorithms and their performance within the modelfree and Brownian rotational diffusion space. J. Biomol. NMR 40, 107–119.

Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved 1037

Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J., and Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293.

Mandel, A.M., Akke, M., and Palmer, A.G., 3rd. (1995). Backbone dynamics of Escherichia coli ribonuclease HI: correlations with structure and function in an active enzyme. J. Mol. Biol. 246, 144–163.

De La Torre, G., Huertas, M.L., and Carrasco, B. (2000). HYDRONMR prediction of NMR relaxation of globular proteins from atomic-level structures and hydrodynamic calculations. J. Magn. Reson. 147, 138–146.

Mer, G., Bochkarev, A., Gupta, R., Bochkareva, E., Frappier, L., Ingles, C.J., Edwards, A.M., and Chazin, W.J. (2000). Structural basis for the recognition of DNA repair proteins UNG2, XPA, and RAD52 by replication factor RPA. Cell 103, 449–456.

Deng, X., Habel, J.E., Kabaleeswaran, V., Snell, E.H., Wold, M.S., and Borgstahl, G.E. (2007). Structure of the full-length human RPA14/32 complex gives insights into the mechanism of DNA binding and complex formation. J. Mol. Biol. 374, 865–876. Fan, J., and Pavletich, N.P. (2012). Structure and conformational change of a replication protein A heterotrimer bound to ssDNA. Genes Dev. 26, 2337– 2347. Fanning, E., Klimovich, V., and Nager, A.R. (2006). A dynamic model for replication protein A (RPA) function in DNA processing pathways. Nucleic Acids Res. 34, 4126–4137. Ferrage, F., Reichel, A., Bhattacharya, S., Cowburn, D., and Ghose, R. (2010). On the measurement of 15N-{1H}-nuclear Overhauser effects. 2. Effects of the saturation scheme and water signal suppression. J. Magn. Reson. 207, 294–303. Goddard, T.D., and Kneller, D.G. (2008). SPARKY 3 (University of California). Goodman, J.L., Pagel, M.D., and Stone, M.J. (2000). Relationships between protein structure and dynamics from a database of NMR-derived backbone order parameters. J. Mol. Biol. 295, 963–978. Jacobs, D.M., Lipton, A.S., Isern, N.G., Daughdrill, G.W., Lowry, D.F., Gomes, X., and Wold, M.S. (1999). Human replication protein A: global fold of the N-terminal RPA-70 domain reveals a basic cleft and flexible C-terminal linker. J. Biomol. NMR 14, 321–331. Jiang, X., Klimovich, V., Arunkumar, A.I., Hysinger, E.B., Wang, Y., Ott, R.D., Guler, G.D., Weiner, B., Chazin, W.J., and Fanning, E. (2006). Structural mechanism of RPA loading on DNA during activation of a simple pre-replication complex. EMBO J. 25, 5516–5526. Kay, L.E., Torchia, D.A., and Bax, A. (1989). Backbone dynamics of proteins as studied by 15N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry 28, 8972–8979. Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam, H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., et al. (2007). Clustal W and Clustal X version 2.0. Bioinformatics 23, 2947–2948.

Pretto, D.I., Tsutakawa, S., Brosey, C.A., Castillo, A., Chagot, M.E., Smith, J.A., Tainer, J.A., and Chazin, W.J. (2010). Structural dynamics and singlestranded DNA binding activity of the three N-terminal domains of the large subunit of replication protein A from small angle X-ray scattering. Biochemistry 49, 2880–2889. Robert, X., and Gouet, P. (2014). Deciphering key features in protein structures with the new ENDscript server. Nucleic Acids Res. 42, W320–W324. Skelton, J.N., Palmer, A.G., Akke, M., Kordel, J., Rance, M., and Chazin, W.J. (1993). Practical aspects of two-dimensional proton-detected 15N spin relaxation measurements. J. Magn. Reson. Ser. B 102, 253–264. Sklenar, V., Piotto, M., Leppik, R., and Saudek, V. (2002). Gradient-tailored water suppression for 1H-15N HSQC experiments optimized to retain full sensitivity. J. Magn. Reson. 102, 241–245. Soss, S.E., Klevit, R.E., and Chazin, W.J. (2013). Activation of UbcH5cUb is the result of a shift in interdomain motions of the conjugate bound to U-box E3 ligase E4B. Biochemistry 52, 2991–2999. Sugitani, N., and Chazin, W.J. (2015). Characteristics and Concepts of Dynamic Hub Proteins in DNA Processing Machinery from Studies of RPA. Prog. Biophys. Mol. Biol. 117, 206–211. Tjandra, N., Feller, S.E., Pastor, R.W., and Bax, A. (1995). Rotational diffusion anisotropy of human ubiquitin from 15N NMR relaxation. J. Am. Chem. Soc. 117, 12562–12566. Walker, O., Varadan, R., and Fushman, D. (2004). Efficient and accurate determination of the overall rotational diffusion tensor form 15N relaxation data using computer program ROTDIF. J. Magn. Reson. 168, 336–345. Walsh, J.D., Meier, K., Ishima, R., and Gronenborn, A.M. (2010). NMR studies on domain diffusion and alignment in modular GB1 repeats. Biophys. J. 99, 2636–2646.

Lipari, G., and Szabo, A. (1982a). Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. Theory and range of validity. J. Am. Chem. Soc. 104, 4546–4559.

Wold, M.S. (1997). Replication protein A: a heterotrimeric, single-stranded DNA-binding protein required for eukaryotic DNA metabolism. Annu. Rev. Biochem. 66, 61–92.

Lipari, G., and Szabo, A. (1982b). Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 2. Analysis of experimental results. J. Am. Chem. Soc. 104, 4559–4570.

Wong, V., Case, D.A., and Szabo, A. (2009). Influence of the coupling of interdomain and overall motions on NMR relaxation. Proc. Natl. Acad. Sci. USA 106, 11016–11021.

1038 Structure 23, 1028–1038, June 2, 2015 ª2015 Elsevier Ltd All rights reserved