Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis

Article

Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis Graphical Abstract

Authors Boon Chong Goh, Juan R. Perilla, Matthew R. England, Katrina J. Heyrana, Rebecca C. Craven, Klaus Schulten

Correspondence [email protected]

In Brief Goh et al. present an atomic model of the Gag lattice, using RSV as the model virus. Construction of the model is based on cryo-EM densities of M-PMV, together with high-resolution structures of RSV. Validity of the model is evaluated by MD simulations, biochemical experiments, and structural comparison with the immature capsid of HIV.

Highlights d

Arrangement of RSV NTDs is similar in M-PMV, yet different in HIV-1 particles

d

Gag monomers are crosslinked at the NTD by the p10 peptide

d

The SP-NC subdomain oligomerizes into a six-helix-bundle, stabilized by salt bridges

d

Removal of the upstream and downstream regions of Gag results in lattice disassembly

Goh et al., 2015, Structure 23, 1–12 August 4, 2015 ª2015 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.str.2015.05.017

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Structure

Article Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis Boon Chong Goh,1,3 Juan R. Perilla,1,3 Matthew R. England,2 Katrina J. Heyrana,2 Rebecca C. Craven,2 and Klaus Schulten1,* 1Department

of Physics and Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA of Microbiology and Immunology, Penn State University College of Medicine, Hershey, PA 17033, USA 3Co-first author *Correspondence: [email protected] http://dx.doi.org/10.1016/j.str.2015.05.017 2Department

SUMMARY

Defining the molecular interaction between Gag proteins in an assembled hexagonal lattice of immature retrovirus particles is crucial for elucidating the mechanisms of virus assembly and maturation. Recent advances in cryo-electron microscopy have yielded subnanometer structural information on the morphology of immature Gag lattices, making computational modeling and simulations feasible for investigating the Gag-Gag interactions at the atomic level. We have examined the structure of Rous sarcoma virus (RSV) using all-atom molecular dynamics simulations and in vitro assembly, to create the first all-atom model of an immature retroviral lattice. Microseconds-long replica exchange molecular dynamics simulation of the spacer peptide (SP)-nucleocapsid (NC) subdomains results in a six-helix bundle with amphipathic properties. The resulting model of the RSV Gag lattice shows features and dynamics of the capsid protein with implications for the maturation process, and confirms the stabilizing role of the upstream and downstream regions of Gag, namely p10 and SP-NC.

INTRODUCTION Retroviruses encode Gag polyproteins that contain subdomains corresponding to the matrix (MA), capsid (CA), and nucleocapsid proteins (NC), as well as other smaller peptides. In the infected cell, thousands of Gag proteins oligomerize into an immature retroviral lattice at the plasma membrane, driving the assembly and budding of the virus particle (Briggs et al., 2004; Carlson et al., 2008). Once budded from the cell surface, the immature virus undergoes maturation triggered by proteolytic cleavage of Gag, and marked by structural rearrangement of the internal constituents of the virion, formation of a mature capsid structure composed of CA, and activation of virion infectivity (Bell and Lever, 2013; Sundquist and Kra¨usslich, 2012; Adamson and Freed, 2007). The Gag lattices from viruses of the Orthoretrovirinae subfamily share similar morphologies in vivo, that is, a nearly spherical

but incomplete shell (Briggs et al., 2009; de Marco et al., 2010; Schur et al., 2015). The MA proteins are arranged at the outermost layer of the Gag lattice with the N termini bound to the viral membrane; they are followed by the CA proteins, and subsequently by the NC proteins that are located at the innermost layer, closest to the viral genome (Figure 1). The CA proteins from the orthoretrovirus genera all share a common domain organization with two primarily a-helical parts (the NTD or N-terminal domain, and CTD or C-terminal domain) (Briggs et al., 2006; Campos-Olivas et al., 2000; Gamble et al., 1997; Kingston et al., 2000; Bailey et al., 2009, 2012). These two domains provide the majority of the inter-molecular interactions that determine the short- and long-range packing geometry of Gag in the immature virion and of CA in the mature capsid (Gamble et al., 1997; Datta et al., 2007). Whereas the two CA domains can make all interactions needed to build a mature capsid and CA protein alone is sufficient to build capsid-like structures in vitro (Zhao et al., 2013; Pornillos et al., 2011; Cardone et al., 2009; Benjamin et al., 2005; Bailey et al., 2009), the assembly of a Gag lattice requires the contributions of additional components of Gag, both upstream and downstream of CA (Keller et al., 2008; de Marco et al., 2010; O’Carroll et al., 2012; Bush et al., 2014; Wright et al., 2007). The Gag proteins of both the Rous sarcoma virus (RSV) and HIV type 1 (HIV-1) contain, immediately following the CA CTD, a short spacer peptide subdomain called SP (12 amino acid residues in RSV) or SP1 (14 residues in HIV-1), separating CA from NC. While the SP subdomain is absent in the MasonPfizer monkey virus (M-PMV), this subdomain is a necessary component for Gag assembly of both RSV and HIV-1 (Craven et al., 1993; Kra¨usslich et al., 1995; Wright et al., 2007; Datta et al., 2011) as it provides crucial lateral stabilization of hexameric Gag complexes (von Schwedler et al., 2003; Li et al., 2007; Larson et al., 2003). Mutations or deletions of SP/SP1 residues interfere with Gag-Gag interactions in vitro (Keller et al., 2008; Bush et al., 2014; England et al., 2014; Datta et al., 2011; Yu et al., 2001), and cause in vivo phenotypes ranging from mild to severe disturbances of virion morphology and even to the total blockage of virus assembly (Kra¨usslich et al., 1995; Liang et al., 2003; Morikawa et al., 2000). Recent cryo-electron tomography studies of RSV, HIV-1, and M-PMV, members of the alpha-, lenti-, and beta-retrovirus genera, respectively, have provided comparative analyses of the organization of the Gag lattice at resolutions that now exceed subnanometer detail for the two latter viruses (de Marco et al.,

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 1

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

1

155 176 211

MA

p2

p10

239

389

476 488 496

CA-NTD CA-CTD SP

NC

577

701

PR

Figure 1. Schematic Diagram of the Full-Length Gag Protein of RSV The shaded rectangles illustrate the Gag construct used in the MD simulations (Table 1). The simulated construct includes the entire CA-SP region plus flanking residues belonging to p10 and NC.

2010; Bharat et al., 2012, 2014; Schur et al., 2015), in particular showing that the packing of CA CTDs is highly conserved among the three viruses. Also, for RSV and HIV-1, an SP bundle region has been observed in cryo-electron tomograms (de Marco et al., 2010). The high-resolution density maps of HIV-1 suggest the existence of a six-helix bundle (6HB) composed of the SP1 regions belonging to six hexamerically related Gag proteins (Bharat et al., 2014; Schur et al., 2015). Similarly, the SP of RSV has also been suggested to form a 6HB by hydrophobic interactions at the SP-NC subdomains (Bush et al., 2014; de Marco et al., 2010). However, although the HIV-1 density maps are 8–9 A˚ in resolution, the density corresponding to the SP1 subdomain is not well resolved. Furthermore, solid-state nuclear magnetic resonance (NMR) measurements of the HIV CA-SP1 show that the region corresponding to SP1 is in a random coil conformation and dynamically disordered (Han et al., 2013), in contrast to solution NMR studies, which show a helical propensity of SP1 (Morellet et al., 2005; Newman et al., 2004). In addition, the SP-NC subdomains of HIV-1 have been investigated computationally by means of coarse-grained simulations (Ayton et al., 2010) and replica exchange molecular dynamics (MD) simulations (Datta et al., 2011). Although these studies provided valuable information in describing the interactions of Gag proteins in a lattice environment, neither effort could determine the atomic structure of the oligomerized SP-NC. Therefore, there is no clear model to date that elucidates the atomic details of the CA-SP1 complex. In RSV, an upstream segment of Gag contributes additional stabilizing contacts to the lattice. A 25-residue-long peptide, located upstream of the NTD of CA, p10, is essential to the assembly of Gag (Campbell and Vogt, 1997; Nandhagopal et al., 2004; Phillips et al., 2008; Krishna et al., 1998; Scheifele et al., 2007). The complete sequence of p10 contains 64 residues, but only the last 25 residues have been structurally determined and identified as an important component of the Gag lattice (Campbell and Vogt, 1997; Nandhagopal et al., 2004). Furthermore, in vitro assembly experiments (Phillips et al., 2008) suggest a key role for the shorter p10 peptide by providing inter-molecular interactions between Gag molecules in the immature lattice via p10-CA interactions as predicted via crystallography. However, the experimental data could not differentiate whether the crosslinked p10-CA molecules form a hexamer, as suggested by Phillips et al. (2008), or form a trimer of dimers. Unfortunately, the cryo-electron tomograms of RSV have yet to reach the resolution to confirm this structural feature. In this study, we present an atomic model of the Gag lattice, using RSV as the model system. RSV is an ideal candidate for such a modeling effort because of the available X-ray crystallography data, along with extensive genetic biochemical studies on

the p10, SP, and NC subdomains that are implicated in Gag assembly (Nandhagopal et al., 2004; Phillips et al., 2008; Bush et al., 2014; Keller et al., 2008; Scheifele et al., 2007). Based on the cryo-electron microscopy (cryo-EM) density map of M-PMV (Bharat et al., 2012), together with structural data from X-ray crystallography and NMR of RSV, an all-atom model of the immature RSV Gag lattice has been computationally derived. The validity of the lattice model is tested by all-atom MD simulations (Perilla et al., 2015) and biochemical experiments. We find that the Gag lattice containing only the NTD and CTD layers of CA is not stable; however, the Gag lattice that incorporates additional upstream and downstream components achieves stability. The results support the importance of p10-CA interactions in the formation of stable NTD hexamers, and identify in particular the key stabilizing role of a collar of charged residues present in the 6HB for Gag assembly and stability. RESULTS The roles played by both p10 and SP-NC in the stability of the Gag lattice were investigated by extensive MD simulations (Table 1). In addition, the interactions predicted by the model were tested by mutagenesis experiments in vitro. Modeling of the RSV Gag Lattice The building process of the Gag lattice of RSV (Figure 2) involves several docking procedures. First, to obtain the RSV CA dimer (Figure 2B), the RSV NTD (PDB: 1P7N) (Nandhagopal et al., 2004) and CTD (PDB: 1D1D) (Campos-Olivas et al., 2000) monomers were individually docked into their corresponding regions of the cryo-EM density map of the M-PMV CA dimer (EMDB: 2090) (Bharat et al., 2012) (Figure 2A). Then, to obtain a hexamer of dimers (Figure 2C), each RSV CA dimer was subsequently docked into the cryo-EM density map of the M-PMV hexameric lattice (EMDB: 2089) (Bharat et al., 2012). Additional components, namely p10 and SP-NC (Figure 2C), were independently modeled and sequentially added to the CA lattice to complete the central hexamer of the Gag lattice. The hexamer of hexamers belonging to the Gag lattice model (Figure 2D) was then obtained by docking the central hexamer, as the basis unit, to the surrounding density map. All docking steps were performed using Situs rigid body docking (Chaco´n and Wriggers, 2002; Wriggers, 2010). Role of Upstream CA Layer in Stabilizing the RSV Gag Lattice Structural and disulfide crosslinking experiments suggest that p10 interacts with the CA hexamer in an intra-hexameric manner. The p10 that is connected to a CA monomer interacts with the NTD of an adjacent CA subunit within the hexamer (Nandhagopal et al., 2004; Phillips et al., 2008). Such inter-molecular interaction between p10 and adjacent NTD was ascertained by the antiparallel p10-NTD homodimer configuration in the crystal structure (PDB: 1P7N). To incorporate p10 in the CA lattice, p10 and NTD from conjugate monomers were extracted from the crystal structure (Figure 2C). Next, the p10-NTD heterodimer from the previous step was aligned to each of the CA monomers in the lattice using the NTD as reference. Lastly, the connecting loop between p10 and its corresponding NTD, residues 240–249, was modeled

2 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Table 1. List of Simulations Performed for the Present Study Employing Molecular Dynamics and Replica Exchange Molecular Dynamics Simulations Simulation

Description

Residues

PCALres

p10-CA lattice with restraints

211–469

Time (ns)

PCAL

p10-CA lattice without restraints

211–469

100

MD

1,820,000

CAL

CA lattice without restraints

240–469

100

MD

1,800,000

BCAL

b-CA lattice without restraints

240–469

100

MD

1,710,000

PCAD

p10-CA dimer of RSV

211–469

110

MD

87,000

CAD

CA dimer of RSV

240–469

110

MD

81,000

BHBR

Bush six-helix bundle (Bush et al., 2014)

469–496

3,750

REMD

19,000

RHBR

refined six-helix bundle

469–496

3,750

REMD

18,000

RHBA

refined six-helix bundle with 4-alanine mutation

469–496

200

MD

17,000

MD

637,000

REMD

624,000

100

PCALB

p10-CA lattice with Bush six-helix bundle

211–496

100

PCALRR

p10-CA lattice with refined six-helix bundle

211–496

5,000

Method

System Size (Atoms)

MD

1,820,000

The full-length Gag construct is shown in Figure 1 for reference. See also Figure S1A. MD, molecular dynamics; REMD, replica exchange molecular dynamics.

using VMD and refined using Modeller (Sali and Blundell, 1993). The resulting model of the lattice shows that the corresponding p10 of one CA monomer crosslinks in a counter-clockwise manner to the NTD of the adjacent subunit within a hexamer (Figure 3), similar to the model previously proposed by Phillips et al. (2008). To evaluate the key residues that provide stabilization to the p10-CA lattice, the system was simulated for 100 ns (simulation PCALres in Table 1) with positional restraints at the boundary of the NTD lattice and at all helices of CTD. The boundary restraints were applied to replicate the spatial confinement on the lattice present in a fully assembled Gag lattice; the CTD helices were held in position to mimic the stabilizing effect provided by the missing domains downstream, namely SP-NC. In all cases, the restraints were applied such that the average displacement is within 1.0 A˚ from the initial configuration. For the MD simulations, the p10-CA lattice was solvated in a sufficiently large water box (Figure S1B). Simulation PCALres shows that the central hexamer remains largely unchanged, with a root-mean-square deviation (RMSD) of below 4.0 A˚. The p10-NTDs of the central hexamer were then analyzed to identify key stabilizing residues at the interfaces. The respective residues are summarized as part of Table 2. The p10-CA lattice model also provides novel information regarding the trimeric interface of the NTD layer. Due to the allatom nature of MD simulations, the effects of ions and water can be directly addressed. In particular, it is observed that residues K346 coordinate chloride ions at the trimeric interface (Figure S5C). Similarly, the model shows that D310 plays a complementary role at the same interface by coordinating sodium ions (Figure S5D). To simulate the commencement of maturation upon cleavage of p10 and SP-NC from Gag, the p10-CA, CA, and b-CA lattices were simulated for 100 ns without any restraints (simulations PCAL, CAL, and BCAL in Table 1). The b hairpin (residues 241– 250) in the b-CA lattice can only be formed once p10 has been cleaved. It is therefore a characteristic of the mature capsid. The b hairpin was modeled into the CA lattice based on the

PDB: 1EM9 (Kingston et al., 2000) using the same protocol as the one described for p10. The NTDs of the p10-CA lattice were significantly more stable than the NTDs of the CA and b-CA lattices as measured by rootmean-square fluctuation calculations and quantified in Figure 4. Furthermore, the NTD hexamers at the circumference of the b-CA lattice were sufficiently distorted such that one of the six surrounding hexamers was dissociated from the lattice (Figure 4), suggesting that the presence of the b hairpin upstream of CA could accelerate the distortion and disassembly of the Gag lattice. The effect of the b hairpin in the dissociation of b-CA lattice is consistent with an experimental study showing that modifying the N terminus of CA could affect Gag assembly and capsid maturation (Von Schwedler et al., 1998). Role of the SP-NC Subdomains in Stabilizing the Gag Lattice Although the NTD hexamers of the p10-CA lattice are more stable than those of the CA and b-CA lattices (Figure 4), the distortions at the CTD region were prominent in all three cases (Figures S3A and 6B). Such fragile CTD lattices led us to predict that the residues downstream of CA are essential to maintain the integrity of the immature CTD hexamers. To further test this prediction, an all-atom model of the SP-NC region was constructed. It has been previously proposed that SP-NC of RSV arranges in a six-helix bundle (6HB) (Wright et al., 2007), and an initial model of such arrangement was recently proposed by Bush et al. (2014). The Bush 6HB model was derived via homology modeling, using a de novo designed ion channel (PDB: 3R47) as the template (Zaccai et al., 2011). To assess the structural integrity of the Bush 6HB model, we ran 3.75 ms of replica exchange MD (REMD) simulations (simulation BHBR). Of the 75 tempered replicas, >50% of the Bush 6HB model collapsed radially inward due to the large amount of hydrophobic residues at the inner surface of the bundle (Figure 5B) even with the presence of positional restraints at the N terminus of the 6HB (see Experimental Procedures). To test the stability of those replicas which did not collapse, the replica with lowest RMSD value

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 3

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

A

C

D

p10 NTD

Cryo-EM density map of a M-PMV CANC tube

p10 with a neighboring NTD

Amphipathic 6HB

B

NTD

90o

CTD

Figure 2. Summary of the Gag Lattice Model Building Process (A) Cryo-EM density map of the M-PMV Gag lattice. (B) Atomic structure of RSV CA (NTD in green and CTD in red) docked to the density in (A). (C) Atomic models of p10 (blue) and SP-NC (orange) modeled into the Gag lattice. (D) Resulting model of the RSV Gag lattice in tubular form.

(2.4 A˚) from the initial 6HB model was extracted from simulation BHBR and simulated for another 60 ns at 310 K, which resulted in disrupted bundles (Figure S4A). To illustrate that the distortion of the Bush 6HB model is not due to an artifact of the positional restraints applied, the uncollapsed replica was incorporated into the p10-CA lattice and simulated for 100 ns (simulation PCALB). The 6HB collapsed within 20 ns (Figure S4B) and exhibited similar characteristics as observed in simulation BHBR. To improve the 6HB structure, the Bush model was adjusted in search of an energetically favorable conformation. Conformational search was performed by helically shifting one to three residues in all six helices of the bundle. Shifting by one or three residues results in energetically unfavorable conformations. Shifting by two residues, which results in an effective 180 clockwise rotation along its helical axis from the original model, places residues R493 and E494 within distance to potentially form salt bridges at the outer circumference. After 3.75 ms of REMD (simulation RHBR), a 6HB model was selected based on the results from the partitioning around medoids (PAM) (Kaufman and Rousseeuw, 1990) clustering scheme described in Experimental Procedures. The improved 6HB model, which shows an inter-helical hydrophobic interface (Figure 5D) and a complex salt bridge network at the C-terminal region (Figures 5E and 5F), is referred to henceforth as the refined 6HB model. To complete the model of the Gag lattice, the refined 6HB was then incorporated into the p10-CA lattice using VMD (Humphrey et al., 1996). In addition, the connecting loops between the 6HB and the CTDs of the lattice were built using Modeller (Sali and

Blundell, 1993). To avoid involvement of the C terminus in the salt bridge network, the C-terminal residue D496 was left uncharged. The Gag lattice was then simulated for 5 ms using REMD (simulation PCALRR) to refine the entire model and test its stability. Because refining the CTD-6HB interface does not require simulating the whole lattice, only the central hexamer and its neighboring Gag proteins were included in the REMD simulation (Figure S1C). The REMD replicas were swapped at high frequency (every 48 fs) and high efficiency (65%) (Figure S6B) characteristic of optimal REMD simulations (Sindelar and Downing, 2010). The representative model of the Gag hexamer was selected using the Gromos algorithm (Daura et al., 1999) as described in Experimental Procedures; the PDB of the Gag hexamer model can be found in Supplemental Information. In the presence of the refined 6HB model, the CTD hexameric lattice remained stable throughout the 5-ms REMD simulation (Figures 6C and 6D); without the 6HB, the hexameric lattice was significantly distorted within 100 ns of simulation (simulation CAL) (Figures 6A and 6B). Properties of the Refined 6HB Model After a cumulative simulation time of almost 9 ms (simulations RHBR and PCALRR), the refined 6HB model exhibits both a salt bridge network as well as hydrophobic contacts between adjacent helices. Several of the hydrophobic residues at the SP, previously identified to be important in Gag assembly (Bush et al., 2014), are involved in stabilizing 6HB by forming hydrophobic contacts between helices (Figure 5D). In addition,

4 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

teins, which lack the membrane binding domain (MBD) of MA and protease (PR), assembled into double-ringed spherical particles (Figure 7A) as previously described (Campbell and Vogt, 1997; Yu et al., 2001; Ma and Vogt, 2002, 2004). Reversing the charge of residue 493 (R493E) results in impaired assembly of Gag in vitro (Figure 7B). However, assembly was restored by a secondary charge reversal of residue 494 (R493E/E494R) (Figure 7D). The DMBDDPR protein with the E494R substitution alone was fully competent for assembly (Figure 7C). Table 2 summarizes the key residues involved in inter- and intra-molecular interactions; the selection criteria of such residues are outlined in Experimental Procedures, and their location is shown in Figure S5B. The model reveals several novel interactions. For instance, D464 at the CTD and R472 at the 6HB form intra-hexameric salt bridges that stabilize the N terminus of the 6HB. Residues corresponding to these salt bridges and other untested interactions listed in Table 2 provide a roadmap to future experiments. Furthermore, another attribute of the refined 6HB model is the kink present at the middle of the SP-NC helix due to the presence of a helix-breaking proline (P485). While the refined 6HB is stable and maintains its hexameric shape, the relative orientation of the entire 6HB with respect to the CA hexamer fluctuates over time (Figure S6A). DISCUSSION

Figure 3. The p10 Peptides Crosslink Adjacent CAs in a CounterClockwise Manner to Stabilize the NTD Hexamer The p10 and NTD domains are shown as helices and tubes, respectively, with each p10-CA monomer colored differently. Only the NTDs are shown for clarity. Inset: Residues L229, V225, W222, I258, L295, and V298 form an intrahexameric hydrophobic core. p10 and the NTD of the adjacent CA monomer are represented in blue and olive, respectively. Such intra-hexameric interaction is absent if immature HIV-1 lattice is used as a modeling template. See also Figure S2.

an electrostatic network is formed by the last four charged residues of the 6HB (R493, E494, R495, D496) (Figures 5E and 5F). The importance of the electrostatic network is particularly evident when simulation RHBA shows the collapse of the refined 6HB within 200 ns of simulation after the four charged residues at the C terminus of the 6HB were mutated to alanine residues (Figure S4C). As residues R493 and E494 are facing outward from the 6HB center and could be important for Gag assembly, three chargeswap mutagenesis experiments were proposed, namely R493E, E494R, and R493E/E494R. The wild-type DMBDDPR Gag pro-

There has been an increasing interest in the structure of the Gag lattice, as it is considered an unexploited therapeutic target (Adamson et al., 2009; Waheed and Freed, 2012). With this report we present what is, to our knowledge, the first atomic model and the first all-atom MD simulation of a Gag lattice that consists of CA together with the upstream and downstream subdomains of p10, SP, and NC that are known to provide critical stabilization. Between the two available paradigms for Gag domain packing (HIV-1 and M-PMV), this work provides a strong argument for the M-PMV model to describe RSV structure as well as for the p10-NTD interaction first illustrated by crystallography. The computationally derived model clearly identifies essential residues at the CA, p10, and SP-NC subdomains of Gag, elucidating the role of the domains in stabilizing the immature RSV Gag lattice. The Gag lattice model provides the structure of immature particles of retroviruses in atomic detail; the model also provides a novel template to investigate the assembly/disassembly and maturation processes of retroviruses like HIV-1, and may assist in the development of new drugs. The comparative analysis of a variety of retroviruses has historically provided important insights into the mechanism of virus assembly, crucial for understanding the human virus, namely HIV-1. Cryo-electron tomography studies of HIV-1 and M-MPV Gag lattices have recently revealed structural information at subnanometer resolution about the arrangement of the CTDs and its downstream sequence, namely SP1 (Bharat et al., 2012, 2014; Schur et al., 2015); such lattices also exhibit the same CTD packing for both HIV-1 and M-MPV (Bharat et al., 2012; Schur et al., 2015), suggesting that CTD packing might be a common motif across most retroviruses (Bharat et al., 2014) as previously suggested by lower-resolution studies (de Marco et al., 2010). On the other hand, the quaternary structure of the NTDs of the HIV-1 Gag lattice is significantly different from that of M-PMV

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 5

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Table 2. Summary of the Key Interactions Identified in the Gag Lattice with the Relevant Experiments Reported for Reference Interaction

Residues Involved

Domain

Relevant Experiments

Intra-hexameric

W222 V225 L229 I258 L295 V298

p10-NTD

X-ray (Nandhagopal et al., 2004), mutagenesis (Phillips et al., 2008; Scheifele et al., 2007; Krishna et al., 1998)

E280–R380/R384

NTD

not tested

R464/K466–D472

CTD

L470 I475 A477 M479 A482 I483 P485 I487 M488 A489

SP

mutagenesis (Bush et al., 2014)

R493–E494

NC

mutagenesis (Yu et al., 2001; England et al., 2014)

NTD

not tested

E228–K256 T220/W222–D291

R495–D496 Trimeric

R297–D310 D310, K346 R339–D350

Inter-hexameric Intra-molecular

R271/R384–D418

NTD-CTD

not tested

W392 L419 A423

CTD

X-ray (Bailey et al., 2009)

E401–R409

CTD

X-ray (Bailey et al., 2009), mutagenesis (Cairns and Craven, 2001)

F403 F406 F432

CTD

X-ray (Bailey et al., 2009), NMR (Campos-Olivas et al., 2000; Kingston et al., 2000), mutagenesis (Cairns and Craven, 2001)

The interactions of several key residues at various interfaces are illustrated in Figure S5.

(Bharat et al., 2012; Schur et al., 2015). During the building process of the RSV Gag lattice model of this study, both the HIV-1 and M-PMV lattices were considered as starting points. However, upon docking of p10-NTD into the density map of the HIV-1 Gag lattice (Schur et al., 2015), severe steric clashes between p10 and helix a1 of the adjacent NTD were observed (Figure S2). Furthermore, the relative orientation of p10 with respect to the adjacent NTD only allows for the formation of inter-hexameric p10-NTD interactions (Figure S2), rather than intra-hexameric interactions as predicted by biochemical experiments (Phillips et al., 2008). Therefore, the Gag lattice of M-PMV (Bharat et al., 2012) emerged as the template of choice, both for NTDs and CTDs, to construct and investigate the RSV Gag lattice (Figure 2). The resulting Gag lattice model not only is consistent with a large body of published genetic, biochemical, and structural data on RSV, as listed in Table 2, but also indicates several new interactions for future investigations. The tertiary structure of the capsid protein is highly conserved across all retroviruses (Coffin et al., 1997); in particular, the CA of RSV contains seven helices (a1–a7) at the NTD, and four helices (a8–a11) at the CTD (Figure S5A). The quaternary arrangement of the NTDs in the RSV Gag lattice is similar to the packing observed in the M-PMV Gag lattice and different from the one observed in HIV-1 particles. In particular, the resulting arrangement of the NTDs accommodates p10 in such a way that the p10 helix forms a three-helix bundle with helices a1 and a3 of the adjacent NTD (Figure 3) stabilized by contacts between hydrophobic residues. The resulting interactions between p10 and helices a1/a3 are similar to the ones proposed in a previous study (Phillips et al., 2008); the hydrophobic interactions at the p10-NTD interface (Figure 3) are consistent with several experimental studies (Nandhagopal et al., 2004; Scheifele et al., 2007; Phillips et al., 2008). Such

p10-NTD interaction is the primary intra-hexameric interaction in which p10 bridges the NTDs that are scarcely interacting within a single hexamer (Phillips et al., 2008). The NTDs instead contribute to a large dimeric interface that serves to hold together adjacent hexamers in the lattice. Specifically, helices a2 and a7 of conjugate NTDs are held together by the formation of salt bridges (Figure S5H). At the interaction site of three hexamers, namely at the trimeric interface, helices a4 and a5 come together by coordinating ions via residues K346 and D310 (Figures S5C–S5E). The quaternary packing of the CTDs in the RSV Gag lattice largely resembles the CTD packing of M-PMV and HIV-1 Gag lattices (Figure 8B) (Bharat et al., 2012, 2014; Schur et al., 2015; de Marco et al., 2010). The main contact point is between helices a9 at the inter-hexameric interface, although the CTDs scarcely interact with each other intra-hexamerically. As such, the CTD hexameric ring is stabilized by the downstream subdomains (SP and part of NC) rather than by adjacent CTDs. In our study, the minimal region that confers stability to the CTD hexamers spans the last seven residues of CA, the entire SP, and the first eight residues of NC arranged in an amphipathic 6HB, consistent with studies from the Vogt laboratory (Keller et al., 2008; Phillips et al., 2008; Yu et al., 2001; Bush et al., 2014). The structure of the 6HB was computationally derived and refined based on the Bush model, and remained stable in microsecond-long simulations. SP residues critical for assembly, as identified by mutagenesis experiments (Bush et al., 2014), served as validation of the inter-helix interface in the refined 6HB (Figures 5C and 5D). The resulting model also reveals a series of inter-helical hydrophobic contacts as well as a collar of electrostatic interactions around the circumference of the bundle. Thus, to test the effects on the integrity of the lattice

6 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Figure 4. NTD Hexamers in the Gag Lattice Stabilized by p10 The degree of fluctuation of each Gag residue is measured in the course of simulations by rootmean-square fluctuation. Residues in red map to regions with large fluctuation (7 A˚), while those colored blue map to regions with relatively small fluctuation (2 A˚). The red arrow highlights the breaking of an NTD hexamer at the circumference of the b-CA lattice. Only NTDs are shown for clarity. See also Figure S3.

of such interactions, a few were selected for mutagenesis experiments as part of this study (Figure 7). Since each helix of the SP-NC complex is connected to the C terminus of the CTD by a flexible loop, the 6HB is inherently flexible (Figure S6A). Such flexibility has been previously observed in NMR studies of monomeric RSV CA-SP (Kingston et al., 2000) and HIV-1 CTD-SP1 (Morellet et al., 2005; Newman et al., 2004; Han et al., 2013) constructs. At the C terminus of the refined 6HB model there is a network of salt bridges formed by the last four residues of the CA-SP8NC construct. The salt bridges formed between R493 and E494 are of particular interest due to their location at the outer circumference of the refined 6HB, in contrast to the Bush model, which locates such residues at the inner circumference (Bush et al., 2014). In the Gag assembly experiments, the ability of E494R to correct the assembly defect caused by R493E provides strong support for the importance of such electrostatic interactions (Figure 7). In addition, removal of the salt bridge network through alanine mutations resulted in a collapsed bundle in a 200-ns MD simulation (Figure S4C). Importantly, the refined 6HB model is consistent with the findings that removal of charge by mutations R493A and R495A disrupts Gag assembly in vitro (Yu et al., 2001) and that the addition of the first eight NC residues to purified CA-SP shifts its assembly mode from mature, or CA-like, to a more Gag-like behavior (England et al., 2014). Such a charged motif has been proposed to assist in the formation of helical bundles in the murine leukemia virus (MLV) via directed charged surfaces (Cheslock et al., 2003). Interestingly, the deletion experiments showed that the charged motif of MLV can tolerate minor charge imbalances without affecting particle production (Cheslock et al., 2003). Therefore, it might not be surprising that the E494R mutation in RSV did not affect Gag assembly. The robust assembly of the single charge reversal, namely E494R, might stem from the lack of the remaining NC residues and RNA that could prevent a full accounting of all possible electrostatic interactions in our model. Using our model of 6HB as reference and based on the distribution of polar and hydrophobic residues at the SP1 and NC subdomains of HIV-1 (Morellet et al., 2005), we suggest that the SP1-NC domain of HIV-1 could form an amphipathic 6HB similar to the one derived for RSV. Despite the smaller hydrophobic face of each helix in the bundle of HIV-1 compared with that of RSV, the 6HB of HIV could still form hydrophobic contacts at the inter-helical interface. Similarly, we speculate that the polar and charged residues of HIV-1 could form intra-hexameric hydrogen bonds to stabilize such 6HB. The presence of a motif

similar to such 6HB in immature virions of HIV-1 has been recently observed (Bharat et al., 2014; Schur et al., 2015). Noteworthy is that the cryo-EM density of such a bundle overlaps well with the model derived in this study, but only at the top half of 6HB (Figure 8B). A density corresponding to the lower half of the bundle may be absent for two reasons: (1) flexibility of the bundle that results in a washed-out density, or (2) an inherently shorter 6HB of HIV-1 compared with RSV (de Marco et al., 2010). The NTD layer of the p10-CA lattice remained stable while the CTD layer was distorted during the unrestrained simulation (Figures 3 and S3A). Therefore, our model suggests that the p10-NTD and CTD-SP-NC interaction layers contribute independently to the stability of Gag hexamers in the lattice, consistent with current structural models for HIV-1 and M-PMV (Bharat et al., 2012, 2014; Schur et al., 2015) and with the need for the PR domain to cleave within both regions to allow the CA protein to rearrange into a capsid. In addition, our model clearly identifies a set of inter-molecular NTD-CTD interactions present in the RSV Gag lattice. In particular, residue D418 forms inter-hexameric salt bridges with either R384 or R271 of the adjacent subunit. Conversely, the same residue (D418) is known to make an intra-molecular contact with R384 in a mature RSV capsid (Bailey et al., 2012). The different interaction modes in D418 and R384 between immature and mature lattices (Figures S5F and S5G) imply that D418 might play a key role in regulating the maturation process. The need for D418 to switch from an inter-molecular to an intra-molecular interaction could explain the ability to trigger in vitro assembly of CA protein into apparently identical capsid-like particles by either exposure to low pH which protonates a nearby residue D418 (Bailey et al., 2012, 2009) or a high-phosphate condition that can shield the cluster of basic residues at the NTD (Purdy et al., 2009), thereby encouraging D418 to break its inter-molecular salt bridge. Relative orientations of the individual domains of CA are significantly different between the mature and immature forms of the RSV lattice (Figures S6C–S6F). In particular, the arrangement of CA in the immature lattice results in an unstable structure, as shown in simulation CAL, where a significant distortion of the CA domain was evident (Figures 4 and S3A). Such distortion of the CA domain was also observed in the all-atom simulation of an isolated immature CA hexamer model of HIV-1 (Ayton et al., 2010). Furthermore, according to our simulations, CAs of an immature lattice require the presence of both the p10 and SPNC subdomains for stability, unlike oligomers of CAs in the mature capsid, which are known to remain stable in silico (Zhao et al., 2013). Furthermore, the dissociation pattern

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 7

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Figure 6. SP and the First Eight Residues of NC Stabilizing the CTD Hexameric Lattice Only the CTDs and SP-NC are shown for clarity. (A) Initial CA hexameric lattice. (B) Lattice in (A) dissociated after 100 ns of simulation. (C) Starting model of the CA lattice in presence of refined 6HB. (D) Structure in (C) remains intact after 5 ms of REMD simulation.

Figure 5. Complex Salt Bridge Network Stabilizing the Refined Six-Helix Bundle Model The sequence of the 28-residue-long CA-SP-8NC construct is shown at the top panel, with the key residues for assembly highlighted in blue (Bush et al., 2014). The same residues are highlighted in (A) to (D) in licorice representation, with hydrophobic and polar residues colored in white and green, respectively. (A) 6HB proposed by Bush et al. (colored in cyan). (B) The Bush 6HB model collapses after 3.75 ms of MD simulation, forming a hydrophobic core at the center of the bundle. (C) 6HB is remodeled (shown as orange cartoon), such that the hydrophobic residues are not pointing directly inward toward the bundle as in (A), but toward the inter-helical interface. (D) After 9 ms of REMD simulation, the inter-helical interactions in the 6HB model are further refined and the tertiary structure remains intact. (E and F) A salt bridge network stabilizes the bottom of the equilibrated 6HB model. The charged residues involved in forming this network are shown in licorice representation (carbon in cyan, oxygen in red, nitrogen in blue); salt bridges are shown as red lines. See also Figure S4.

observed for both NTDs and CTDs (Figures 4 and S3A) suggests that upon cleavage of p10 and SP-NC, the lattice breaks down into dimers belonging to conjugate Gag hexamers as early intermediates of the disassembly process. Such dimers are stable for more than 100 ns (Figures S3B and S3C), and are held together by interactions at the NTD (residue E280 of helix a2 and residues

R380, R384 of helix a7), in contrast to the dimers found in the mature capsid, where the dimerization interface is part of the CTD. Overall our findings favor the idea of a diffusion-driven disassembly-reassembly maturation pathway (Butan et al., 2008; Purdy et al., 2009; Bharat et al., 2012; Keller et al., 2013) over the non-diffusive displacive transition hypothesis for retroviruses (Meng et al., 2012; Frank et al., 2015). The atomic model of a Gag lattice presented here, containing segments both upstream and downstream of CA, identifies several important interactions that are shown to contribute to the stability of the lattice. The refined 6HB model presented in this study, derived from microsecond-long MD simulations, provides a refined atomic template with which to study the properties of the SP-NC subdomains, and enables the modeling of atomic structures of 6HB for the SP-NC subdomain of several other retroviruses, including HIV-1. Furthermore, extensive simulations of the Gag lattice reveal novel mechanistic aspects of Gag related to assembly, stability, and maturation. Overall we expect that the model derived in this study extends beyond RSV, offering an unique glimpse into the maturation process of retroviral particles, and constituting a new platform for useful drug targets in the development of novel antiretroviral therapies. EXPERIMENTAL PROCEDURES All-Atom MD Simulation Systems were prepared using VMD (Humphrey et al., 1996) and simulations were performed using NAMD 2.9 (Phillips et al., 2005) with the CHARMM36 force field (Best et al., 2012). The protein assemblies (Table 1) were simulated in a sufficiently large water box such that the minimal distance between solute

8 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

1999). Simulations BHBR and RHBR employed a total of 75 replicas with temperature ranges from 275 to 425 K with a target exchange ratio (Garcia et al., 2006) of 0.5, and exchanges between replicas attempted every 2 ps. For simulation RHBR the 6HB model that corresponds to the medoid of the largest cluster was selected using the PAM algorithm (Kaufman and Rousseeuw, 1990; Reynolds et al., 2006) and was then incorporated into the Gag lattice. The details of the PAM algorithm can be found in Supplemental Information. For simulation PCALRR, 113 replicas were simulated with temperature ranges from 305 to 375 K with a target exchange ratio of 0.25. The frequency of exchange was enhanced by attempting an exchange every 48 fs (Sindhikara, 2010). The list of temperatures used for each of the REMD simulations can be found in Supplemental Information. The program g_cluster from GROMACS (Pronk et al., 2013) was used to cluster the final snapshots from the 113 replicas corresponding to simulation PCALRR, into 12 different clusters using the Gromos algorithm (Daura et al., 1999). The structure corresponding to the centroid of the largest cluster, which contains 80 of the 113 replicas, was selected as the final lattice model.

Figure 7. In Vitro Assembly of Immature-like RSV Particle (A) Wild-type Gag. (B) R493E Gag mutant. (C) E494R Gag mutant. (D) R493E/E494R Gag mutant. DMBDDPR protein was assembled in the presence of a 50-mer DNA oligonucleotide. Assembly products were stained with uranyl acetate and visualized by transmission electron microscopy. Scale bars, 100 nm.

and box boundary was 20 A˚ along all three axes. The solvated systems were neutralized, the salinity of the solvent being set to 100 mM NaCl. 100 mM NaCl was also used in the assembly experiments (Bush et al., 2014). The ionized systems were then minimized for 20,000 steps and subsequently thermalized to 310 K within 4 ps. All equilibrium simulations were performed in the NPT ensemble. Simulations PCALres, CALres, and BCALres were performed for 100 ns with positional restraints applied to the backbone atoms of the lattice boundary and CTDs. The lattice boundary is defined as the residues that are beyond 60 A˚ from the central hexamer. Simulations PCAL, CAL, and BCAL were performed without restraints. For simulation PCALRR and PCALB, only the central hexamer along with its neighboring monomers were included in the simulation in order to limit the system size. Details and parameters used for all simulations can be found in Supplemental Information. REMD Simulations Temperature REMD as implemented in NAMD 2.9 (Phillips et al., 2005) was used to sample the conformational space of the 6HB (Sugita and Okamoto,

Selection Criteria for Key Interacting Residues The key interacting residues listed in Table 2 were selected from the replica that corresponds to temperature T = 310 K in simulation PCALRR and filtered according to the criteria outlined below. Each of the six monomers in the central hexamer were examined individually, and the results averaged over the entire set. Residues that form inter-molecular hydrogen bonds were identified by the Hbonds plugin in VMD (Humphrey et al., 1996), and identified as key residues if the average occupancies were above 50%. Residues in a hydrophobic core were identified as key contacts if the contact area between the hydrophobic side chains was more than 100 A˚2. Some of the intra-molecular interactions are included in Table 2 due to their relevance in several experimental studies listed in the same table. Mutagenesis Studies DMBDDPR Purification The largest well-behaved, soluble form of the Gag protein lacks the first 83 residues of MA and all of PR (DMBDDPR) (a kind gift from Volker M. Vogt) (Campbell and Vogt, 1997). This sequence was expressed from the pET3xc plasmid by auto-induction in BL21(DE3)-RIL Escherichia coli as previously described (Ma and Vogt, 2002). In Vitro Assembly and Analysis The DMBDDPR protein was diluted to 1 mg/ml with 50 mM 2-(N-morpholino) ethanesulfonic acid and 100 mM NaCl (pH 6.0). Assembly was triggered by the addition of a 50-mer DNA consisting of 25 GT repeats (GT50) at a 10% w/w ratio of nucleic acid to protein. The reaction was allowed to proceed for 30 min at room temperature (Yu et al., 2001; Ma and Vogt, 2002, 2004). Resulting particles were applied to carbon and formvar-coated grids and negatively stained with 2% uranyl acetate (Yu et al., 2001; Campbell and Vogt, 1997). The grids were viewed on a JEOL JEM 1400 transmission electron microscope,

Figure 8. The Complete Hexamer Model to Form a Stable RSV Gag Lattice (A) The hexamer with the refined 6HB model derived from REMD is shown in orange cartoon representation. The p10, NTD, and CTD are shown in cartoon representations in blue, green, and red, respectively. (B) The CTD and 6HB of RSV docked into HIV-1 Gag density map for comparison. See also Figure S6.

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 9

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

and an Orius SC 1000 bottom-mounted CCD camera was used to capture images.

Cryo-electron microscopy of tubular arrays of HIV-1 Gag resolves structures essential for immature virus assembly. Proc. Natl. Acad. Sci. USA 111, 8233–8238.

SUPPLEMENTAL INFORMATION

Briggs, J.A.G., Simon, M.N., Gross, I., Kra¨usslich, H.G., Fuller, S.D., Vogt, V.M., and Johnson, M.C. (2004). The stoichiometry of Gag protein in HIV-1. Nat. Struct. Mol. Biol. 11, 672–675.

Supplemental Information includes Supplemental Experimental Procedures, seven figures, and 3D molecular models and can be found with this article online at http://dx.doi.org/10.1016/j.str.2015.05.017. AUTHOR CONTRIBUTIONS B.C.G., J.R.P., and K.S. designed the research. B.C.G. performed the modeling and simulations with J.R.P.’s support. J.R.P. wrote the clustering scripts for data analysis. M.R.E., K.J.H., and R.C.C. performed protein expression, mutagenesis, and imaging. B.C.G., J.R.P., M.R.E., and R.C.C. analyzed the data. B.C.G., J.R.P., R.C.C., and K.S. wrote the manuscript with support from all the authors. ACKNOWLEDGMENTS This work was supported by the NIH grants 9P41GM104601 (K.S., B.C.G. and J.R.P.), R01 GM067887 (K.S. and J.R.P.), R01 CA100322 (R.C.C. and K.J.H.), and T32 CA060395 (M.R.E.) as well as by the National Science Foundation (NSF) grant NSF PHY1430124 (K.S. and J.R.P.). Molecular dynamics simulations were performed on the Blue Waters Supercomputers, financed by the National Science Foundation (OCI-0725070 and ACI-1238993) under Petascale Computational Resource (PRAC) grant ACI-1440026, and on Titan Supercomputers, supported by Department of Energy under Contract DEAC05-00OR22725. The authors also acknowledge the use of the Penn State College of Medicine imaging core facility, and important discussions with Volker Vogt and John Briggs during the course of this work. Received: April 13, 2015 Revised: May 18, 2015 Accepted: May 25, 2015 Published: June 25, 2015 REFERENCES

Briggs, J.A.G., Johnson, M.C., Simon, M.N., Fuller, S.D., and Vogt, V.M. (2006). Cryo-electron microscopy reveals conserved and divergent features of Gag packing in immature particles of Rous sarcoma virus and human immunodeficiency virus. J. Mol. Biol. 355, 157–168. Briggs, J.A.G., Riches, J.D., Glass, B., Bartonova, V., Zanetti, G., and Kra¨usslich, H.G. (2009). Structure and assembly of immature HIV. Proc. Natl. Acad. Sci. USA 106, 11090–11095. Bush, D.L., Monroe, E.B., Bedwell, G.J., Prevelige, P.E., Phillips, J.M., and Vogt, V.M. (2014). Higher order structure of the Rous sarcoma virus SP assembly domain. J. Virol. 88, 5617–5629. Butan, C., Winkler, D.C., Heymann, J.B., Craven, R.C., and Steven, A.C. (2008). RSV capsid polymorphism correlates with polymerization efficiency and envelope glycoprotein content: implications that nucleation controls morphogenesis. J. Mol. Biol. 376, 1168–1181. Cairns, T.M., and Craven, R.C. (2001). Viral DNA synthesis defects in assembly-competent Rous sarcoma virus CA mutants. J. Virol. 75, 242–250. Campbell, S., and Vogt, V.M. (1997). In vitro assembly of virus-like particles with Rous sarcoma virus Gag deletion mutants: identification of the p10 domain as a morphological determinant in the formation of spherical particles. J. Virol. 71, 4425–4435. Campos-Olivas, R., Newman, J.L., and Summers, M.F. (2000). Solution structure and dynamics of the Rous sarcoma virus capsid protein and comparison with capsid proteins of other retroviruses. J. Mol. Biol. 296, 633–649. Cardone, G., Purdy, J.G., Cheng, N., Craven, R.C., and Steven, A.C. (2009). Visualization of a missing link in retrovirus capsid assembly. Nature 457, 694–698. Carlson, L.A., Briggs, J.A., Glass, B., Riches, J.D., Simon, M.N., Johnson, M.C., Mu¨ller, B., Gru¨newald, K., and Kra¨usslich, H.G. (2008). Three-dimensional analysis of budding sites and released virus suggests a revised model for HIV-1 morphogenesis. Cell Host Microbe 4, 592–599.

Adamson, C.S., and Freed, E.O. (2007). Human immunodeficiency virus type 1 assembly, release, and maturation. Adv. Pharmacol. 55, 347–387.

Chaco´n, P., and Wriggers, W. (2002). Multi-resolution contour-based fitting of macromolecular structures. J. Mol. Biol. 317, 375–384.

Adamson, C.S., Salzwedel, K., and Freed, E.O. (2009). Virus maturation as a novel HIV-1 therapeutic target. Expert Opin. Ther. Targets 13, 895–908.

Cheslock, S.R., Poon, D.T.K., Fu, W., Rhodes, T.D., Henderson, L.E., Nagashima, K., McGrath, C.F., and Hu, W.S. (2003). Charged assembly helix motif in murine leukemia virus capsid: an important region for virus assembly and particle size determination. J. Virol. 77, 7058–7066.

Ayton, G.S., Lyman, E., and Voth, G.A. (2010). Hierarchical coarse-graining strategy for protein-membrane systems to access mesoscopic scales. Faraday Discuss. 144, 347–357. Bailey, G.D., Hyun, J.K., Mitra, A.K., and Kingston, R.L. (2009). Proton-linked dimerization of a retroviral capsid protein initiates capsid assembly. Structure 17, 737–748. Bailey, G.D., Hyun, J.K., Mitra, A.K., and Kingston, R.L. (2012). A structural model for the generation of continuous curvature on the surface of a retroviral capsid. J. Mol. Biol. 417, 212–223. Bell, N.M., and Lever, A.M.L. (2013). HIV Gag polyprotein: processing and early viral particle assembly. Trends Microbiol. 21, 136–144. Benjamin, J., Ganser-Pornillos, B.K., Tivol, W.F., Sundquist, W.I., and Jensen, G.J. (2005). Three-dimensional structure of HIV-1 virus-like particles by electron cryotomography. J. Mol. Biol. 346, 577–588. Best, R.B., Zhu, X., Shim, J., Lopes, P.E.M., Mittal, J., Feig, M., and MacKerell, A.D. (2012). Optimization of the additive charmm all-atom protein force field targeting improved sampling of the backbone f, j and side-chain c1 and c2 dihedral angles. J. Chem. Theor. Comput. 8, 3257–3273.

Coffin, J.M., Hughes, S.H., and Varmus, H.E. (1997). Retroviruses (Cold Spring Harbor Laboratory Press). Craven, R.C., Leure-duPree, A.E., Erdie, C.R., Wilson, C.B., and Wills, J.W. (1993). Necessity of the spacer peptide between CA and NC in the Rous sarcoma virus Gag protein. J. Virol. 67, 6246–6252. Datta, S.A.K., Zhao, Z., Clark, P.K., Tarasov, S., Alexandratos, J.N., Campbell, S.J., Kvaratskhelia, M., Lebowitz, J., and Rein, A. (2007). Interactions between HIV-1 Gag molecules in solution: an inositol phosphate-mediated switch. J. Mol. Biol. 365, 799–811. Datta, S.A.K., Temeselew, L.G., Crist, R.M., Soheilian, F., Kamata, A., Mirro, J., Harvin, D., Nagashima, K., Cachau, R.E., and Rein, A. (2011). On the role of the SP1 domain in HIV-1 particle assembly: a molecular switch? J. Virol. 85, 4111– 4121. Daura, X., Gademann, K., Jaun, B., Seebach, D., van Gunsteren, W.F., and Mark, A.E. (1999). Peptide folding: when simulation meets experiment. Angew. Chem. Int. Ed. Engl. 38, 236–240.

Bharat, T.A.M., Davey, N.E., Ulbrich, P., Riches, J.D., de Marco, A., Rumlova, M., Sachse, C., Ruml, T., and Briggs, J.A.G. (2012). Structure of the immature retroviral capsid at 8 A˚ resolution by cryo-electron microscopy. Nature 487, 385–389.

England, M.R., Purdy, J.G., Ropson, I.J., Dalessio, P.M., and Craven, R.C. (2014). Potential role for CA-SP in nucleating retroviral capsid maturation. J. Virol. 88, 7170–7177.

Bharat, T.A.M., Castillo Menendez, L.R., Hagen, W.J.H., Lux, V., Igonet, S., Schorb, M., Schur, F.K.M., Kra¨usslich, H.G., and Briggs, J.A.G. (2014).

Frank, G.A., Narayan, K., Bess, J.W., Jr., Del Prete, G.Q., Wu, X., Moran, A., Hartnell, L.M., Earl, L.A., Lifson, J.D., and Subramaniam, S. (2015).

10 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Maturation of the HIV-1 core by a non-diffusional phase transition. Nat. Commun. 6, 5854.

surrounding p2: implications for particle assembly and RNA packaging. Protein Sci. 14, 375–386.

Gamble, T.R., Yoo, S., Vajdos, F.F., von Schwedler, U.K., Worthylake, D.K., Wang, H., McCutcheon, J.P., Sundquist, W.I., and Hill, C.P. (1997). Structure of the carboxyl-terminal dimerization domain of the HIV-1 capsid protein. Science 278, 849–853.

Morikawa, Y., Hockley, D.J., Nermut, M.V., and Jones, I.M. (2000). Roles of matrix, p2, and N-terminal myristoylation in human immunodeficiency virus type 1 Gag assembly. J. Virol. 74, 16–23.

Garcia, A.E., Herce, H., and Paschek, D. (2006). Simulations of temperature and pressure unfolding of peptides and proteins with replica exchange molecular dynamics. Annu. Rep. Comput. Chem. 2, 83–95. Han, Y., Hou, G., Suiter, C.L., Ahn, J., Byeon, I.J.L., Lipton, A.S., Burton, S., Hung, I., Gor’kov, P.L., Gan, Z., et al. (2013). Magic angle spinning NMR reveals sequence-dependent structural plasticity, dynamics, and the spacer peptide 1 conformation in HIV-1 capsid protein assemblies. J. Am. Chem. Soc. 135, 17793–17803. Humphrey, W., Dalke, A., and Schulten, K. (1996). VMD—visual molecular dynamics. J. Mol. Graph. 14, 33–38.

Nandhagopal, N., Simpson, A.A., Johnson, M.C., Francisco, A.B., Schatz, G.W., Rossmann, M.G., and Vogt, V.M. (2004). Dimeric Rous sarcoma virus capsid protein structure relevant to immature Gag assembly. J. Mol. Biol. 335, 275–282. Newman, J.L., Butcher, E.W., Patel, D.T., Mikhaylenko, Y., and Summers, M.F. (2004). Flexibility in the P2 domain of the HIV-1 Gag polyprotein. Protein Sci. 13, 2101–2107. O’Carroll, I.P., Soheilian, F., Kamata, A., Nagashima, K., and Rein, A. (2012). Elements in HIV-1 Gag contributing to virus particle assembly. Virus Res. 171, 341–345.

Kaufman, L., and Rousseeuw, P.J. (1990). Finding Groups in Data: An Introduction to Cluster Analysis, Vol. 344 (John Wiley).

Perilla, J.R., Goh, B.C., Cassidy, C.K., Liu, B., Bernardi, R.C., Rudack, T., Yu, H., Wu, Z., and Schulten, K. (2015). Molecular dynamics simulations of large macromolecular complexes. Curr. Opin. Struct. Biol. 31, 64–74.

Keller, P.W., Johnson, M.C., and Vogt, V.M. (2008). Mutations in the spacer peptide and adjoining sequences in Rous sarcoma virus Gag lead to tubular budding. J. Virol. 82, 6788–6797.

Phillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel, R.D., Kale, L., and Schulten, K. (2005). Scalable molecular dynamics with NAMD. J. Comput. Chem. 26, 1781–1802.

Keller, P.W., Huang, R.K., England, M.R., Waki, K., Cheng, N., Heymann, J.B., Craven, R.C., Freed, E.O., and Steven, A.C. (2013). A two-pronged structural analysis of retroviral maturation indicates that core formation proceeds by a disassembly-reassembly pathway rather than a displacive transition. J. Virol. 87, 13655–13664.

Phillips, J.M., Murray, P.S., Murray, D., and Vogt, V.M. (2008). A molecular switch required for retrovirus assembly participates in the hexagonal immature lattice. EMBO J. 27, 1411–1420.

Kingston, R.L., Fitzon-Ostendorp, T., Eisenmesser, E.Z., Schatz, G.W., Vogt, V.M., Post, C.B., and Rossmann, M.G. (2000). Structure and self-association of the Rous sarcoma virus capsid protein. Structure 8, 617–628.

Pronk, S., Pa´ll, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts, M.R., Smith, J.C., Kasson, P.M., van der Spoel, D., et al. (2013). Gromacs 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 29, 1–10.

Kra¨usslich, H.G., Fa¨cke, M., Heuser, A.M., Konvalinka, J., and Zentgraf, H. (1995). The spacer peptide between human immunodeficiency virus capsid and nucleocapsid proteins is essential for ordered assembly and viral infectivity. J. Virol. 69, 3407–3419. Krishna, N.K., Campbell, S., Vogt, V.M., and Wills, J.W. (1998). Genetic determinants of Rous sarcoma virus particle size. J. Virol. 72, 564–577. Larson, D.R., Ma, Y.M., Vogt, V.M., and Webb, W.W. (2003). Direct measurement of Gag-Gag interaction during retrovirus assembly with FRET and fluorescence correlation spectroscopy. J. Cell Biol. 162, 1233– 1244. Li, H., Dou, J., Ding, L., and Spearman, P. (2007). Myristoylation is required for human immunodeficiency virus type 1 Gag-Gag multimerization in mammalian cells. J. Virol. 81, 12899–12910. Liang, C., Hu, J., Whitney, J.B., Kleiman, L., and Wainberg, M.A. (2003). A structurally disordered region at the c terminus of capsid plays essential roles in multimerization and membrane binding of the Gag protein of human immunodeficiency virus type 1. J. Virol. 77, 1772–1783. Ma, Y.M., and Vogt, V.M. (2002). Rous sarcoma virus Gag protein-oligonucleotide interaction suggests a critical role for protein dimer formation in assembly. J. Virol. 76, 5452–5462. Ma, Y.M., and Vogt, V.M. (2004). Nucleic acid binding-induced Gag dimerization in the assembly of Rous sarcoma virus particles in vitro. J. Virol. 78, 52–60.

Pornillos, O., Ganser-Pornillos, B., and Yeager, M. (2011). Atomic-level modelling of the HIV capsid. Nature 469, 424–427.

Purdy, J.G., Flanagan, J.M., Ropson, I.J., and Craven, R.C. (2009). Retroviral capsid assembly: a role for the CA dimer in initiation. J. Mol. Biol. 389, 438–451. Reynolds, A.P., Richards, G., de la Iglesia, B., and Rayward-Smith, V.J. (2006). Clustering rules: a comparison of partitioning and hierarchical clustering algorithms. J. Math. Model. Algorithms 5, 475–504. Sali, A., and Blundell, T.L. (1993). Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234, 779. Scheifele, L.Z., Kenney, S.P., Cairns, T.M., Craven, R.C., and Parent, L.J. (2007). Overlapping roles of the Rous sarcoma virus Gag p10 domain in nuclear export and virion core morphology. J. Virol. 81, 10718–10728. Schur, F.K.M., Hagen, W.J.H., Rumlova, M., Ruml, T., Muller, B., Krausslich, H.G., and Briggs, J.A.G. (2015). Structure of the immature HIV-1 capsid in intact virus particles at 8.8 A˚ resolution. Nature 517, 505–508. von Schwedler, U.K., Stray, K.M., Garrus, J.E., and Sundquist, W.I. (2003). Functional surfaces of the human immunodeficiency virus type 1 capsid protein. J. Virol. 77, 5439–5450. Sindelar, C.V., and Downing, K.H. (2010). An atomic-level mechanism for activation of the kinesin molecular motors. Proc. Natl. Acad. Sci. USA 107, 4111– 4116. Sindhikara, D. (2010). Exchange often and properly in replica exchange molecular dynamics. J. Chem. Theor. Comput. 6, 2804–2808.

de Marco, A., Davey, N.E., Ulbrich, P., Phillips, J.M., Lux, V., Riches, J.D., Fuzik, T., Ruml, T., Kra¨usslich, H.G., Vogt, V.M., and Briggs, J.A.G. (2010). Conserved and variable features of Gag structure and arrangement in immature retrovirus particles. J. Virol. 84, 11729–11736.

Sugita, Y., and Okamoto, Y. (1999). Replica-exchange molecular dynamics method for protein folding. Chem. Phys. Lett. 314, 141–151.

Meng, X., Zhao, G., Yufenyuy, E., Ke, D., Ning, J., Delucia, M., Ahn, J., Gronenborn, A.M., Aiken, C., and Zhang, P. (2012). Protease cleavage leads to formation of mature trimer interface in HIV-1 capsid. PLoS Pathog. 8, e1002886.

Von Schwedler, U.K., Stemmler, T.L., Klishko, V.Y., Li, S., Albertine, K.H., Davis, D.R., and Sundquist, W.I. (1998). Proteolytic refolding of the HIV-1 capsid protein amino-terminus facilitates viral core assembly. EMBO J. 17, 1555–1568.

Morellet, N., Druillennec, S., Lenoir, C., Bouaziz, S., and Roques, B.P. (2005). Helical structure determined by NMR of the HIV-1 (345-392)Gag sequence,

Waheed, A.A., and Freed, E.O. (2012). HIV type 1 Gag as a target for antiviral therapy. AIDS Res. Hum. Retroviruses 28, 54–75.

Sundquist, W.I., and Kra¨usslich, H.G. (2012). HIV-1 assembly, budding, and maturation. Cold Spring Harb. Perspect. Med. 2, a006924.

Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved 11

Please cite this article in press as: Goh et al., Atomic Modeling of an Immature Retroviral Lattice Using Molecular Dynamics and Mutagenesis, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.017

Wriggers, W. (2010). Using situs for the integration of multi-resolution structures. Biophysical Rev. 2, 21–27. Wright, E.R., Schooler, J.B., Ding, H.J., Kieffer, C., Fillmore, C., Sundquist, W.I., and Jensen, G.J. (2007). Electron cryotomography of immature HIV-1 virions reveals the structure of the CA and SP1 Gag shells. EMBO J. 26, 2218–2226. Yu, F., Joshi, S.M., Ma, Y.M., Kingston, R.L., Simon, M.N., and Vogt, V.M. (2001). Characterization of Rous sarcoma virus Gag particles assembled in vitro. J. Virol. 75, 2753–2764.

Zaccai, N.R., Chi, B., Thomson, A.R., Boyle, A.L., Bartlett, G.J., Bruning, M., Linden, N., Sessions, R.B., Booth, P.J., Brady, R.L., and Woolfson, D.N. (2011). A de novo peptide hexamer with a mutable channel. Nat. Chem. Biol. 7, 935–941. Zhao, G., Perilla, J.R., Yufenyuy, E.L., Meng, X., Chen, B., Ning, J., Ahn, J., Gronenborn, A.M., Schulten, K., Aiken, C., and Zhang, P. (2013). Mature HIV-1 capsid structure by cryo-electron microscopy and all-atom molecular dynamics. Nature 497, 643–646.

12 Structure 23, 1–12, August 4, 2015 ª2015 Elsevier Ltd All rights reserved