Crystal Structure of the Oligomeric Form of Lassa Virus Matrix Protein Z

ABSTRACT

The arenavirus matrix protein Z is highly multifunctional and occurs in both monomeric and oligomeric forms. The crystal structure of a dodecamer of Z from Lassa virus, presented here, illustrates a ring-like structure with a highly basic center. Mutagenesis demonstrates that the dimeric interface within the dodecamer and a Lys-Trp-Lys triad at the center of the ring are important for oligomerization. This structure provides an additional template to explore the many functions of Z.

IMPORTANCE The arenavirus Lassa virus causes hundreds of thousands of infections each year, many of which develop into fatal hemorrhagic fever. The arenavirus matrix protein Z is multifunctional, with at least four distinct roles. Z exists in both monomeric and oligomeric forms, each of which likely serves a specific function in the viral life cycle. Here we present the dodecameric form of Lassa virus Z and demonstrate that Z forms a “wreath” with a highly basic center. This structure and that of monomeric Z now provide a pair of critical templates by which the multiple roles of Z in the viral life cycle may be interpreted.

INTRODUCTION

The family Arenaviridae comprises more than 20 different viral species, of which at least nine cause human disease. Of these, Lassa virus (LASV), an Old World arenavirus, poses a particularly acute threat due to its endemicity in western Africa and its propensity to be transported outside Africa. For all arenaviruses, replication takes place in the cytoplasm of infected cells and budding occurs at the plasma membrane. The matrix protein Z of the arenaviruses acts as a bridge between the viral surface glycoprotein (GPC), the viral ribonucleoprotein (NP), and host cell budding machinery (1–3). Z contains an N-terminal myristoylation site for anchoring into the cell membrane, a central zinc-binding RING motif (4, 5), and C-terminal canonical late-domain motifs, which are essential for budding (2, 3). Z has also been shown to inhibit both viral and host cell translation by reducing the affinity of the translation initiation factor eIF4E for the m7GTP cap of mRNA (6, 7). Furthermore, Z negatively regulates the replication and transcription of the viral genome (8). Lastly, arenavirus Z proteins have recently been shown to be involved in the anti-immune response through an interaction between the N-terminal domain of Z and the CARD domains of RIG-i-like receptors (9).

Previous work demonstrated that the Z proteins of both LASV and the related arenavirus lymphocytic choriomeningitis virus (LCMV) are initially expressed as monomers but assemble spherical “body” structures both in vitro (10) and in the cytoplasm and nuclei of infected cells (11).

The structure of monomeric LASV Z has previously been determined by nuclear magnetic resonance (NMR) (12, 13). The NMR structure provided important insights into the overall structure of Z and suggested a mechanism for interaction with eIF4E. However, the structure by which Z monomers self-assemble into oligomeric or “body” forms is still unclear. In order to gain insight into how LASV Z self-assembles, we determined the structure of oligomeric LASV Z by X-ray crystallography to 2.9 Å.

MATERIALS AND METHODS

Plasmids.

Full-length LASV-Z Josiah in pGEX6p-1 (a gift from Juan Carlos de la Torre, The Scripps Research Institute, La Jolla, CA) was engineered with a tobacco etch virus (TEV) protease site in place of the prescission protease site using site-directed mutagenesis to generate LASV glutathione S-transferase (GST)-Z-TEV. LASV GST-Z-TEV mutants were generated using site-directed mutagenesis. Wild-type (WT) Z and mutants were subcloned into phCMV to generate C-terminally tagged LASV Z (Z-HA) for use in budding experiments.

Expression and purification.

WT and mutant LASV GST-Z-TEV were expressed in Rosetta 2(DE3)pLysS Escherichia coli cells (Novagen) at 37°C and grown in 1-liter cultures to an optical density at 600 nm (OD₆₀₀) of 0.4. Cells were then induced with 500 μM IPTG (isopropyl-β-d-thiogalactopyranoside) (supplemented with 100 μM ZnCl) and incubated overnight at 25°C with shaking at 300 rpm. Expression cultures with harvested via centrifugation and lysed in 50 mM Tris (pH 8.0), 300 mM NaCl, and one EDTA-free complete protease inhibitor tablet (Roche) using an M-110L Laboratory Microfluidizer (Microfluidics). The lysate was cleared by centrifugation, and the supernatant was loaded onto glutathione-Sepharose 4B beads (GE Healthcare). After the resin was washed with 10 column volumes of wash buffer (50 mM Tris [pH 8.0], 300 mM NaCl), the GST fusion protein was eluted with 20 mM reduced glutathione in wash buffer. Purified protein was cleaved with a 1:400 ratio of TEV protease to GST-Z-TEV overnight at room temperature (RT). The cleaved protein was subsequently dialyzed into 25 mM Tris (pH 8.5) and 25 mM NaCl and further purified by anion exchange, where pure LASV Z was collected in the flowthrough. Following ion-exchange purification, the pH of the buffer was adjusted to 7.5, and LASV Z was concentrated to greater than 10 mg/ml, allowed to sit for at least 24 h at 4°C, and then purified on a Superdex-200 GL 10/30 (GE Biosciences) equilibrated with 10 mM Tris (pH 7.5) and 150 mM NaCl. Fractions from the two major peaks of WT LASV Z, corresponding to a monomer and dodecamer, were used in crystallization trials.

Crystallization and data collection.

Purified LASV Z was screened for crystallization using a Topaz microfluidic system (Fluidigm). Hits were translated to hanging-drop vapor diffusion using a 1:1 ratio of well solution to protein at 8 to 12 mg/ml. Both the monomeric and dodecameric fractions crystallized in 300 to 400 mM ammonium sulfate, 100 mM HEPES (pH 7.5), and 17% polyethylene glycol (PEG) 3350, with crystals reaching full size in 1 to 3 days. However, only the dodecamer produced diffraction-quality crystals. Crystals were harvested and cryoprotected in 15% glycerol and flash cooled in liquid nitrogen. Crystals belonging to the space group R32 diffracted to 2.9 Å at Beamline 5.2.1 of the Advanced Light Source. A fluorescence scan was performed to confirm the presence of zinc, and data were collected at 1.2827 Å, corresponding the zinc-K-edge peak.

Data processing and structure determination.

Data were indexed, integrated, and scaled in space group R32 using d*TREK (Rigaku) (14). Structure determination using single-wavelength anomalous diffraction (SAD) and subsequent rounds of refinement were carried out using PHENIX (15, 16). Iterative cycles of model building were performed using COOT (17). Five percent of the reflections were set aside for R_free calculations. The final structure has an R_free of 19.30% and a final R factor of 18.26% using all the data to 2.9-Å resolution.

SEC-MALS.

For size exclusion chromatography coupled with multiangle light scattering (SEC-MALS), purified Z proteins were separated on a Superose6 gel filtration column (GE Healthcare) preequilibrated with buffer (20 mM Tris [pH 7.5], 150 mM NaCl) and coupled in-line with a miniDAWN Treos followed by an Optilab T-rEX refractometer (Wyatt Technologies). Data processing and absolute molecular mass calculations were performed using ASTRA software (Wyatt Technologies).

Deuterium exchange mass spectrometry.

Prior to conducting the exchange experiments, quench conditions that produced an optimal pepsin fragmentation pattern were established as previously described (18, 19). Functionally deuterated, nondeuterated, and equilibrium-deuterated samples were prepared. Functional deuteration of LASV Z was performed by diluting 2 μl of a 5.0 mg ml⁻¹ stock solution of LASV Z (150 mM NaCl, 10 mM Tris, pH 7.5) into 6 μl of D₂O buffer containing 8.3 mM Tris and 150 mM NaCl (pH 7.15) at 0°C. At 10 s and 1,000 s, 12 μl of optimized quench solution (0.8% formic acid, 16.6% glycerol, and 1 M guanidine hydrochloride [GuHCl]) was added, and then samples were frozen at −80°C. Nondeuterated samples were prepared by incubation in H₂O buffer containing 8.3 mM Tris and 150 mM NaCl (pH 7.15) at 0°C. Equilibrium-deuterated samples were prepared by incubation in D₂O buffer containing 0.5% formic acid for 3 days at 25°C.

The samples were later thawed at 5°C (20) and passed over an AL-20-pepsin column (16-μl bed volume [Sigma]), at a flow rate of 20 μl min⁻¹. The resulting peptides were collected on a C₁₈ trap (Michrom Magic C18AQ 0.2 × 2) and separated using a C₁₈ reverse-phase column (Michrom Magic C18AQ 0.2 × 50) running a linear gradient of 8 to 48% solvent B (80% acetonitrile and 0.01% trifluoroacetic acid [TFA]) over 30 min with column effluent directed into an LCQ mass spectrometer (Thermo-Finnigan LCQ Classic). Data were acquired in both data-dependent MS1:MS2 mode and MS1 profile mode. SEQUEST software (Thermo Finnigan Inc.) was used to identify the sequences of the peptide ions. The centroids of the isotopic envelopes of nondeuterated, functionally deuterated, and equilibrium-deuterated peptides were measured using DXMS Explorer (Sierra Analytics Inc., Modesto, CA) and then converted to corresponding deuteration levels with corrections for back exchange (21).

Protein structure accession number.

The crystallographic structure factors and coordinates determined in this study have been deposited into the Protein Data Bank with accession number 5I72.

RESULTS

Crystal structure of oligomeric Z.

Size exclusion chromatography coupled with multiangle light scattering (SEC-MALS) demonstrates that LASV Z exists in monomeric and dodecameric forms (Fig. 1). Crystals of the dodecamer grow in 300 to 400 mM ammonium sulfate, 100 mM HEPES (pH 7.5), and 17% PEG 3350 and belong to the space group R32. The structure of dodecameric Z was determined to 2.9 Å using phases obtained from the two zinc atoms inherent to each monomer (Table 1). Five percent of the reflections were set aside for R_free calculations. The final structure has an R_free of 19.30% and a final R factor of 18.26%.

FIG 1.

SEC-MALS analysis of wild-type and mutant Z proteins. Light-scattering analysis of each protein was used to determine the absolute molar mass for each peak (29). Stoichiometries for each protein were determined using the molar mass calculated for the monomeric form. Some mutants yield only one peak (monomer), while the WT and others also yield dodecamers. The exclusion volume of the column is indicated by an arrowhead in panel A.

TABLE 1.

Data collection, phasing, and refinement statistics (determined by SAD) for the crystal structure of dodecameric Z

Parameter	Value(s) for dodecameric Z, Zn peak
Data collection
Space group	R32
Cell dimensions
a, b, c (Å)	116.89, 116.89, 82.53
α, β, γ (°)	90, 90, 120
Wavelength (Å)	1.2827
Resolution (Å)	38.22–2.90 (3.0–2.9)a
Rmerge(%)	0.074 (0.390)
I/σ(I)	10.2 (2.8)
Completeness (%)	99.8 (100.0)
Redundancy	4.8 (4.73)
Refinement
Resolution (Å)	38.22–2.90 (3.66–2.9)
No. of unique reflections	4,923
Rwork/Rfree	18.26/19.30
Ramachandran favored (%)b	96.00
Ramachandran outliers (%)	1.00
No. of atoms
Protein	822
Ligand/ion	4
Water	2
B values
Protein	89.00
Ligand/ion	94.80
Water	69.90
RMSD
Bond lengths (Å)	0.005
Bond angles (°)	1.110

^aThe value for the highest-resolution shell is shown in parentheses.

^bAccording to the criteria of MolProbity.

The asymmetric unit of the crystal structure consists of an antiparallel dimer of Z with well-defined density for residues 26 to 73 in chain A and 26 to 75 in chain B, which correspond to the RING domain of Z (Fig. 2A). Each Z monomer contains two zinc-binding sites, an α helix, two antiparallel β strands, and two loop regions: residues 31 to 40 (loop 1) and residues 61 to 70 (loop 2). The dimer is formed by hydrophobic interactions between the α-helices of its two component monomers, A and B (Fig. 2C).

FIG 2.

Structure of the oligomeric form of LASV Z. (A) The asymmetric unit is comprised of a dimer of Z, in which one monomer is colored yellow and the other blue. Although the construct used for crystallization encodes full-length Z, only residues 26 to 73 are ordered in monomer A and residues 26 to 75 are ordered in monomer B. Bound zinc atoms, two per monomer, are illustrated by gray spheres. (B) Top, crystallographic symmetry generates the dodecameric “wreath,” comprised of six dimers of Z. Bottom, the electrostatic surface potential of oligomeric Z calculated using APBS (30) shows that the center of the wreath is highly basic. Positive surface is colored blue; negative surface would be colored red with significant limits of ±10 k_BT/e.c., where k_B is Boltzmann's constant, T is temperature, and e.c. is the charge of an electron. (C) The four interfaces of oligomeric LASV Z. The underlying structure of the basic inner surface of the wreath includes residues K32, W35, and K68. Dimerization of Z is mediated through hydrophobic interactions between the alpha helix of each monomer. A hydrogen bond network forms the hydrophilic interface between related monomers (A to A′ and B to B′). Interactions between adjacent dimers (A to B and B′; B to A and A′) are mediated through a hydrophobic interface.

The dodecamer is assembled from six dimers related by crystallographic symmetry. Overall, the dodecamer resembles a “wreath” with neighboring dimers in alternating orientations: one up and one down (Fig. 2B). Thus, there does not appear to be a “top” and “bottom” to the ring, but rather there are two equivalent bifunctional faces, each having six flexible N termini and six flexible C termini. The wreath of six A-B dimers is stabilized by a hydrogen bond network between adjacent and crystallographically equivalent chains (i.e., A to A′ and B to B′), and also by a hydrophobic interface surrounding zinc-binding sites I and II between adjacent, nonequivalent chains from different asymmetric units (i.e., A to B′) (Fig. 2C). The outside of the wreath is 70 Å in diameter and uncharged. The central chamber of the ring is 38 Å in diameter and is ringed by repeating triads of K32, W35, and K68. In each triad, the lysine pair sandwiches the tryptophan rings (Fig. 2C), making the inner portion of the dodecamer extremely basic (Fig. 2B).

Comparison to the monomeric form of Z.

Z from the oligomeric (crystal) structure and Z from the monomeric (NMR) structure superimpose with a root mean square deviation (RMSD) of 2.4 Å. The primary differences can be mapped to three areas: (i) loop 1 residues K32, W35, and F36 adopt different orientations and locations in the oligomer and monomer; (ii) the α helix in the oligomer of Z is extended (residues 51 to 60, versus 51 to 57 in the NMR monomer); and (iii) the C-terminal tail lies in close apposition to zinc site 2 (Zn2) in the oligomer but is more extended in the monomer (Fig. 3A). Repositioning of the C-terminal tail upon oligomerization coincides with isomerization of P72 (Fig. 3A, inset) and results in a shift in the position of Zn2. Proline isomerization is catalyzed by prolyl isomerases in the cell and without such acceleration is a slow process. Indeed, we find that oligomerization of Z is a concentration and time-dependent process.

FIG 3.

Comparison of the monomeric and oligomeric Z structures. (A) The monomeric NMR structure of Z (magenta, PDB code 2KO5) is superimposed on one copy of Z in the dodecamer (blue). The primary differences can be mapped to the backbone and side chains of loop 1 residues K32, W35, and F36 (right), an extension of the alpha helix (left), and the location of the C-terminal tail and position of the second Zn-binding site (left). Repositioning of the tail results in isomerization of P72 (inset). (B) Left, superimposition of four copies of the monomeric NMR structure of LASV Z (magenta and green) onto the dodecamer reveals that the conformation of monomeric Z is incompatible with dodecamerization. W35 and F36 of one monomer (A and A′; magenta) sterically clash with the alpha helix and C-terminal tail, respectively, of the opposing monomer (B and B′; green). These regions are indicated with arrows. Right, four copies of the X-ray structure of LASV Z (blue and yellow) are shown for comparison. Surface representation is shown for only one “dimer” of the four copies for clarity.

The NMR-determined structure of monomeric Z is unable, as a static unit, to fit into the oligomer observed via X-ray crystallography due to numerous clashes of both the backbone and side chains. Both backbone and side chains must rotate and translate to form the dodecahedral ring (Fig. 3B). Hence, conformational changes are essential for Z oligomerization.

The differences between the monomeric and dodecameric Z structures are located primarily in the alpha helix, loop 1, and the C-terminal tail. To determine whether these regions “breathed” or showed flexibility in their positions in solution, we analyzed the LASV Z dodecamer by deuterium exchange mass spectrometry. This method measures the ability of peptide amide hydrogens to freely and reversibly exchange with solvent deuterium. Hydrogens for which mobility is restricted (by conformational anchoring and/or ligand binding) exchange more slowly. Hydrogens for which mobility is unrestricted (conformationally mobile) exchange more rapidly. The three regions that change conformation in dodecamer assembly (extension of the α-helix, loop 1, and the C-terminal tail) show slow H/D exchange kinetics. In contrast, the first four ordered residues of LASV Z (residues 26 to 30) and the loop around Zn2 (residues 64 to 68) show moderate to higher H/D kinetics (Fig. 4). The extreme N and C termini of Z (residues 1 to 25 and residues 76 to 99) showed very high H/D exchange rates, indicating that these regions are highly flexible. Indeed, residues 1 to 25 and 76 to 99 are disordered in the crystal structure and in solution studies by NMR. These results suggest that the dodecamer is a stable unit once formed and does not interconvert between the monomer and dodecamer on the time scales we studied.

FIG 4.

Deuterium exchange mass spectrometry analysis of the Z oligomer. Top, Ten- and 1,000-second amide hydrogen-deuterium exchange maps for LASV Z. Exchange rates are colored from blue (slowly exchanging amides) to red (rapidly exchanging amides). Bottom, cartoon representation of one-third of the LASV Z dodecamer, colored based on the 10-s exchange rates.

Dimerization and the K-W-K triad are important for oligomerization.

Wild-type Z elutes as monomers and SDS-resistant dodecamers when purified via size exclusion chromatography. Although the crystallographic asymmetric unit and basic building block of the dodecamer is a dimer of A and B protomers, such dimers are not observed in solution. The dimer is likely a short-lived assembly intermediate.

To determine which residues might regulate formation of the Z ring, we made mutations to LASV Z and produced these proteins in E. coli. Purified Z was subjected to SEC-MALS analysis to determine the average molar mass of each mutant (Fig. 1B). We focused our attention on the α-helical dimeric interface and on the residues that shift conformation between the monomer and dodecameric Z structures: the K32-W35-K68 triad that forms the center of the ring and P72, which isomerizes in the C-terminal tail. As expected, WT Z elutes as two peaks that correspond to monomeric and dodecameric Z. The mutation K32A (in the KWK triad) produces Z protein that also elutes as monomer and dodecamer. However, K32A-bearing Z consistently oligomerizes more quickly and at lower concentrations than the WT. In contrast, W35A-bearing Z elutes as a single peak, shifted ∼1 ml from that of the WT monomer. While the elution volume of W35A compared to size exclusion standards alone would suggest that the Z W35A mutant forms a dimer, the more accurate MALS analysis demonstrates that W35A produces only monomeric Z (Fig. 1B). The shift in elution volume may result from differing interaction with the column or a change in shape. Mutation L56R (A-B dimer interface) or K68A (KWK triad) produces Z that elutes as only monomer, with elution volumes consistent with WT- and K32A-generated monomers. Lastly, mutation of P72 to alanine results in almost exclusively oligomeric Z. SEC-MALS analysis shows that the P72A-bearing Z oligomer has a stoichiometry similar to that of WT Z, although this mutant elutes slightly before WT Z and the peak is broader, indicating a more polydisperse sample. In addition, the P72A mutation renders oligomeric Z sensitive to SDS. Based on these results, we conclude that both the KWK triad and the A-B dimer interface are important for oligomerization. Further, isomerization of P72 may provide an energy barrier to control formation of the ring.

DISCUSSION

The structure of the Lassa virus matrix protein in a stable, wreath-like dodecameric assembly as presented here illustrates conformational changes required for oligomerization. Previous NMR and mutational studies of LASV Z and eIF4E demonstrated that residues F30, K32, S33, W35, N38, and K39 of Z are involved in interaction with eIF4E (12). These residues are inaccessible in the oligomeric crystal structure but accessible in the monomeric NMR structure. Curiously, analytical ultracentrifugation, SEC, and electron microscopy studies of the closely related LCMV Z protein all indicate that eIF4E can interact with both oligomeric and monomeric forms of LCMV Z (22). Further, assembly of LCMV Z into oligomeric bodies greatly increases its ability to regulate eIF4E (22). It is worth noting, however, that analytical ultracentrifugation studies of LCMV Z demonstrate that it forms 24-member “bodies” (10), as opposed to the 12-member oligomer observed here for Lassa virus Z. Hence, either LCMV Z may form a different oligomeric structure to interact with eIF4E or eIF4E may bind to some yet-to-be-determined oligomeric structure of arenavirus Z that is distinct from the structure presented here.

Of particular interest is the highly basic nature of the inside of the dodecameric ring. A negatively charged binding partner has not been previously identified for any arenavirus Z. One potential partner is RNA, as it has been observed bound to the center of the oligomeric ring forms of the matrix proteins VP40 from Ebola virus (23) and M from Borna disease virus (24). The dodecamer illustrated here assembled from purified monomer. Both the purified monomer and its dodecamer have A_260/280 ratios of 0.74, suggesting that very little, if any, nucleic acid copurifies with the protein. Further, incubation of the ring with a 6-nucleotide single-stranded RNA and repurification via size exclusion did not result in a shift in the A_260/280 value of the eluting protein (data not shown). Previous work by Kranzuch and Whelan demonstrated that Machupo virus (MACV) Z forms a 1:1 interaction with the polymerase to inhibit RNA synthesis and that Z does not directly bind to RNA to achieve this function (25).

The RNA-binding ring form of Ebola virus VP40 appears to be involved in transcriptional regulation rather than matrix assembly (26). However, work by Capul et al. demonstrated that the K68A mutation in LASV Z (which renders purified Z monomeric and reduces basic charge at the center) results in greater-than-wild-type levels of negative regulation of an LCMV minigenome, while the P72A mutation (which accelerates ring formation) impairs the ability of Z to negatively regulate the expression of the minigenome (27). Our findings that the K68A mutation results in monomeric Z and the P72A mutation results in oligomeric Z suggest that for LASV, like for MACV, the regulatory role of Z is driven by monomeric Z and that the monomer does not bind to RNA. Thus, the basic interface of the LASV Z ring may point to another, as-yet-unknown, binding partner and/or function for Z, such as interaction with the negatively charged phospholipids of the cellular bilayer.

The exact role or roles of this conformationally rearranged dodecamer of Z remain unknown. However, the fact that this oligomeric ring is built through rearrangement of tertiary and quaternary structure is not surprising, as Z may follow in the same vein as other viral matrix “transformer” proteins (23, 26, 28). Indeed, the four-gene genome of the arenaviruses has an even greater need for highly plastic polypeptides that can adopt different structures to achieve multiple functions. The structure of Z presented here, combined with that of the previously determined monomer, provide a pair of templates by which the separate and critical roles of Z may be interpreted. These conformations of Z and point mutations which accelerate (K32A) or prevent (W53A, L56R, and K68A) oligomerization will serve as a guide for further biological experimentation to explore how the structures of Z confer its multiple functions in the virus life cycle.

Funding Statement

Erica Ollmann Saphire was additionally supported by the Burroughs Wellcome Fund.