Structure of Equine Infectious Anemia Virus ... - Journal of Virology

JOURNAL OF VIROLOGY, Feb. 2002, p. 1876–1883 0022-538X/02/$04.00⫹0 DOI: 10.1128/JVI.76.4.1876–1883.2002 Copyright © 2002, American Society for Microbiology. All Rights Reserved.

Vol. 76, No. 4

Structure of Equine Infectious Anemia Virus Matrix Protein Hideki Hatanaka,1† Oleg Iourin,1,2 Zihe Rao,3 Elizabeth Fry,1 Alan Kingsman,2‡ and David I. Stuart1,4* Division of Structural Biology, Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford OX3 7BN,1 Department of Biochemistry, University of Oxford, Oxford OX1 3QU,2 and Oxford Centre for Molecular Sciences, University of Oxford, Oxford OX1 3QT,4 United Kingdom, and Laboratory of Structural Biology, School of Life Science and Engineering, Tsinghua University, Beijing 100084, People’s Republic of China3 Received 23 July 2001/Accepted 6 November 2001

The Gag polyprotein is key to the budding of retroviruses from host cells and is cleaved upon virion maturation, the N-terminal membrane-binding domain forming the matrix protein (MA). The 2.8-Å resolution crystal structure of MA of equine infectious anemia virus (EIAV), a lentivirus, reveals that, despite showing no sequence similarity, more than half of the molecule can be superimposed on the MAs of human immunodeficiency virus type 1 (HIV-1) and simian immunodeficiency virus (SIV). However, unlike the structures formed by HIV-1 and SIV MAs, the oligomerization state observed is not trimeric. We discuss the potential of this molecule for membrane binding in the light of conformational differences between EIAV MA and HIV or SIV MA. Equine infectious anemia virus (EIAV) is a nonprimate enveloped virus, classified into the family Retroviridae, genus Lentivirus, along with Human immunodeficiency virus (HIV) (reviewed in reference 1). This pathogen causes equine infectious anemia, the most important infectious disease of horses, and antiviral drugs would be economically important. This is the only lentivirus known to be spread by an insect vector, bloodsucking flies (22). In addition to its direct pharmaceutical interest EIAV also has promise as a retrovirus vector for gene therapy (34); being a lentivirus it can transduce both dividing and nondividing cells (27), and being unable to replicate in human cells (32) it is considered much safer than primate lentivirus vectors. Matrix antigen (MA) is a proteolytic cleavage product of the retrovirus Gag polyprotein (reviewed in reference 13). When expressed in the host cell, Gag is in itself sufficient to produce virus-like particles, which are enveloped with a lipid bilayer and which contain RNA. In typical retroviruses such as HIV, the N terminus of Gag is myristoylated (3, 16) and the Nterminal domain, MA, binds to the inner surface of the host membrane, forming a hexagonal network (35). Upon or just after budding of virus particles from host cells, the Gag polyprotein is cleaved by viral protease into several proteins termed MA, CA, and NC, and the particles experience a dramatic structural change known as maturation (6). In the mature particle CA and NC make a compact RNA-containing conically shaped capsid inside the virus particle, while MA remains at the inner surface of the virus membrane. The role of MA after maturation cleavage is controversial, but, by using its nuclear localization signal, lentivirus MA might help in carrying the viral genome into the nucleus after infection (5,

46); this would require the release of MA from the membrane. The N termini of Gag for some retroviruses, including EIAV, are not myristoylated, and some bear a different chemical modification (19, 43). We infer that this modification is also hydrophobic by analogy with Rous sarcoma virus, where Gag has an acetylmethionyl group at its N terminus (38) and artificial myristoylation of this avian virus enables virus release from mammalian cells (50). Recently Zhou and Resh (52) found that HIV type 1 (HIV-1) MA binds membranes much more weakly than its Gag precursor and proposed that the myristoylated N terminus is sequestered in the isolated MA. They also found that removal of the last helix, residues 97 to 109, of MA enhanced membrane binding. Spearman et al. (44) put forward a similar hypothesis, and, based on cryoelectron microscopy, Fuller et al. (15, 49) suggested that the C-terminal portion of MA is not compactly folded in virus-like particles made of Gag. Finally, in vitro experiments by Hermida-Matsumoto and Resh (20) showed that, in HIV-1, MA is released from the membrane slowly after cleavage. Reil et al. (41) reported that HIV-1 can replicate in the absence of the MA domain in some conditions, provided that the N terminus of Gag is myristoylated. Taken together these results suggest that MA might function as a membrane-binding regulator, like recoverin (1). At present there is no structural information to support such a myristoyl switch model, no structures have been solved for Gag precursors, and all the MA proteins for which the threedimensional structures are available (reviewed in reference 7) have been produced in a heterologous expression system, such that the N termini are not correctly modified. Further, in all the MA structures examined to date, the C termini are spatially far from their N termini, so it is unclear how proteolytic cleavage at the C terminus could be reported to the N terminus. In the HIV-1 and simian immunodeficiency virus (SIV) MA structures (21, 40), the trimeric organization of the molecules is believed to present a bipartite membrane-binding surface, where exposed basic residues (51) cooperate with the N-terminal myristoyl groups to anchor the protein to the inner leaflet of the membrane. In these structures the N termini are

* Corresponding author. Mailing address: Division of Structural Biology, The Henry Wellcome Building for Genomic Medicine, Roosevelt Dr., Headington, Oxford OX3 7BN, United Kingdom. Phone: (44) 1865 287546. Fax: (44) 1865 287547. E-mail: enquiries @strubi.ox.ac.uk. † Present address: Protein Research Group, RIKEN Genomic Sciences Center, Tsurumi, Yokohama 230-0045, Japan. ‡ Oxford BioMedica, Oxford OX4 4GA, United Kingdom. 1876

VOL. 76, 2002

STRUCTURE OF EIAV MA

1877

TABLE 1. Data collection and processinga Wavelength (Å)

Resolution range (Å)

Total no. of observations

No. of unique reflections

Completeness (%)

R merge (%)b

⬍I/␴(I)⬎

0.9116 0.9781 0.9778 Combined

24–2.8 (2.90–2.80) 24–3.0 (3.11–3.00) 24–3.0 (3.11–3.00) 24–2.75 (2.96–2.75)

19,170 (1,780) 31,433 (2,706) 31,626 (2,822)

6,347 (633) 5,200 (509) 5,188 (521) 6,321 (995)

99 (99) 100 (99) 100 (100) 94 (74)

7.4 (60) 8.7 (68) 9.5 (62) 7.1 (53)

11.4 10.1 11.4 17.2 (1.5)

a

Values in parentheses are for the highest-resolution shell. R merge ⫽ 100 ⫻ ⌺h ⌺ j 储 I h, merged 兩 ⫺ 兩 Ih, j 储 /⌺hN 兩 Ih, merged 兩 , where j ⫽ 1, . . . , N for N data observations, h are unique reflection indices, and Ih, merged is the mean intensity. b

disordered prior to residue 6 or 7, at which point they protrude from the putative membrane-binding surface. The relevance of this mechanism of association to EIAV is unclear since it lacks both the myristate and the linear cluster of basic residues that comprise the HIV-1 membrane-binding motif. Here we report the 2.8-Å resolution crystal structure of EIAV MA. Despite the lack of any apparent sequence similarity with MAs of primate lentiviruses, EIAV MA shows striking overall structural similarity to MAs of HIV-1 and SIV except that it does not exhibit the same trimeric association. There are, however, significant conformational differences between EIAV MA and HIV-1 and SIV MAs that may direct, or be a correlate of, its mode of association. We suggest that the conformation we observe for EIAV MA could correspond to a weakened membrane-binding state and discuss evidence for an alternative membrane-binding mechanism. MATERIALS AND METHODS EIAV MA expression, purification, and crystallization. EIAV MA was expressed using a pET-32a (Novagen) plasmid in Escherichia coli BL21(DE3). SeMet labeling was done using a minimum medium (26) containing selenomethionine instead of methionine. The protein was purified from the lysate initially by using a His-Bind resin column, and the MA portion was cleaved with recombinant enterokinase. The cleavage product was further purified by using a Mono-S (Pharmacia) column with a 0 to 2 M NaCl gradient in 50 mM Tris-HCl (pH 8.0) and then dialyzed against phosphate-buffered saline. The MA coding sequence (45) came from clone CL22, but the resultant protein included a seven-residue extrinsic sequence AMADIGS just N-terminal to the initial methionine of the original MA sequence; this was confirmed by sequencing the 15 N-terminal residues. Mass spectra for the protein with and without SeMet labeling gave molecular weights of 15,004.1 ⫾ 0.2 and 14,864.3 ⫾ 0.5, respectively, which confirmed the protein sequence and labeling of all three methionine residues. The recombinant selenomethionyl EIAV MA protein was crystallized using the sitting-drop vapor diffusion method (18). The reservoir solution contained 30% (wt/vol) polyethylene glycol 5000, 0.2 M ammonium sulfate, and 0.1 M MES (morpholinoethanesulfonic acid; pH 6.5). One microliter of MA protein solution (about 2.5 mg ml⫺1) was mixed with an equal volume of the reservoir solution and set to equilibrate with reservoir solution at 18°C. Crystals were obtained after 5 days. Data collection and phasing. Data were collected at station BM14 of the European Synchrotron Radiation Facility. For data collection at 100 K using a cryostream (Oxford Cryosystem), a crystal was briefly transferred to mother liquor containing 20% (vol/vol) glycerol. The Se absorption edge was located by a fluorescence scan, and data sets were collected at three wavelengths, 0.9778, 0.97806, and 0.91161 Å, corresponding to the peak, inflection, and remote points, respectively, using a 345-mm-diameter MAR-Research (Hamburg, Germany) image plate detector. Data to 2.8 Å resolution were autoindexed, integrated, and scaled using DENZO and SCALEPACK (37). Measured diffraction intensities were converted to structure factor amplitudes using the CCP4 program TRUNCATE (14). Four of the six expected Se atoms were located and refined by using the program SHARP (9), and the maps calculated with the resulting phases were solvent flattened iteratively by using the program GAP (D. I. Stuart and J. Grimes, unpublished data). A comparison of maps with different hands showed

that the crystal belonged to space group P65 with unit cell dimensions of a ⫽ b ⫽ 47.2 Å and c ⫽ 204.8 Å. Model building and refinement. The initial map revealed the presence of two molecules in the asymmetric unit. The noncrystallographic symmetry (NCS) relationship between them was determined and refined with GAP starting from the coordinates of two selenium sites and four marker positions chosen from the map. Iterations of solvent flattening and NCS averaging in combination with the MAD phase information resulted in a map that allowed initial modeling of residues 7 to 109. Model building was carried out using the programs FRODO (23) and O (24). Due to severe anisotropy, structure factors sharpened by the application of anisotropic overall B factors were used for map calculation. Refinement was done using CNS, version 0.5 (2). After several rounds of rebuilding a 2Fo-Fc map (where Fo and Fc are observed and calculated structure factor amplitudes) gave us clear election density for the N-terminal residues of molecule A, and the selenium atoms of the SeMet residues ⫺5 and 1 corresponded well to the two heavy-metal sites of weaker intensity obtained by SHARP. After these residues were modeled and after additional refinement, weak density for the N-terminal residues of molecule B also appeared, and their modeling showed that SeMet residue ⫺6 corresponded to a further site suggested by SHARP. All refinement used a target function including the experimental-phase information in the maximum-likelihood optimization of the model (2), increasing our confidence in the model. Tables 1 and 2 show statistics for the structure determination. The final R factor for the model (27.9%) is high, as was

TABLE 2. Phasing and NCS averaging and refinementa Parameter

Phasing and NCS averaging Resolution range (Å) .......................................... Figure of merit b Acentric ............................................................. Centric ............................................................... Correlation coefficientc ........................................

Value

24–3.2 (3.5–3.2) 0.396 (0.166) 0.317 (0.063) 0.952 (0.944)

Refinement Resolution range (Å) .......................................... 24–2.8 (3.0–2.8) No. of unique reflections .................................... 6,153 (783) 27.9 (45.8) R factor (%)d ........................................................ R free (%)............................................................. 37.3 (58.7) Protein model .......................................................1,846 atoms, 232 residues Solvent ................................................................... 0 Nonprotein atoms ................................................ 0 e 0.015 r.m.s.d. bond lengths (Å) ................................... 1.7 r.m.s.d. angles (°)e ................................................ r.m.s.d. B (bonded) (Å2) f .................................... 2.1, 2.8 3.6, 4.5 r.m.s.d. B (angle-related) (Å2) f .......................... 79 Mean overall B-factor (Å2) ................................ a Refinement was performed against a composite data set based on the average of all of the data for the three wavelengths. The values shown in parentheses are for the highest-resolution shell. b The figure of merit is derived from SHARP (9) in the absence of solvent flattening or NCS constraints. c The correlation coefficient is for Fs from the end of the NCS averaging. d R ⫽ 100 ⫻ ⌺h 储 Fh, obs 兩 ⫺ 兩 Fh, calc 储 /⌺h 兩 Fh, obs 兩, where Fh, obs is the observed structure factor amplitude and Fh, calc is that which has been calculated from the refined model. e r.m.s.d. from ideal bond lengths or bond angles. f r.m.s.d. of B factors for the bond or angle restraints, for main chain and side chain, respectively.

1878

HATANAKA ET AL.

J. VIROL.

FIG. 1. Stereodiagrams illustrating structure determination and crystal structure of EIAV MA. (A) Omit map calculated by omitting residues of the 310 helical loop (residues 42 to 68). The map contoured at 2 ␴ is shown as a green mesh with the EIAV model superimposed as a red ball-and-stick model. The HIV-1 structure from the corresponding region is drawn as a thinner blue ball-and-stick model. (B) C␣ traces of the two NCS-related molecules superimposed on each other. Molecule A, red; molecule B, green. C␣ atoms are shown by balls every 10 residues. Dashed lines, extrinsic N-terminal segments. (C) Helical supramolecular organization in the crystal. One helix is shown as thick worms, with molecule A white and molecule B grey. Neighboring helices are shown as thin worms with different colors. A unit cell, with its c axis horizontal, is shown. All figures were generated using BOBSCRIPT (11, 12) and Raster3D (33).

observed for HIV-1 (25.9%). The weak (Bmean ⫽ 79 Å2) and markedly anisotropic diffraction (relative B factors B11 ⫽ B22 ⫽ 22 Å2, B33 ⫽ ⫺44 Å2), possible partial order of the omitted residues (30 residues out of 262), and the lack of a model for ordered solvent molecules contribute to the high R factor. Nevertheless, in the light of the satisfactory electron density, stereochemistry (root mean square deviation [r.m.s.d.] bond length ⫽ 0.015 Å) and backbone torsion angles (61.4% of the residues lie in the most favored region of the Ramachandran plot produced by PROCHECK [25] with none in forbidden regions), we consider the model essentially correct. The coordinates have been deposited with the Brookhaven Protein Data Bank, entry code 1HEK.

RESULTS Overall structure: EIAV MA is not trimeric in the crystal. Figure 1A shows a 2.8-Å-resolution electron density omit map (Fobs ⫺ Fcalc [excluding residues 42 to 68], where Fobs and Fcalc are observed and calculated structure factor amplitudes); con-

formational differences between EIAV and HIV MA can clearly be seen. Two molecules (denoted A and B) are found in the crystallographic asymmetric unit (Fig. 1B), and residues ⫺6 to 109 have been modeled for both molecules (residues ⫺6 to 0, AMADIGS, correspond to an extrinsic sequence due to the expression construct). C-terminal residues 110 to 124 are not seen in the electron density map, presumably because of their flexibility. No significant differences between the two molecules were found (Fig. 1B), except that (i) the N-terminal extrinsic residues run in a different way and (ii) the B-factor distributions differ in loop regions (data not shown); therefore we describe only molecule A below unless explicitly mentioned otherwise. Unlike SIV and HIV-1 MAs EIAV MA is not trimeric in the crystal; instead, molecules A and B are organized in a helical

VOL. 76, 2002

supramolecular structure in the P65 crystal, arranged alternately head to tail (Fig. 1C). There are two sets of intermolecular contacts along the helical structure, but these are essentially equivalent, that is, the noncrystallographic symmetry and crystallographic symmetry combine to produce a supramolecular helix which contains 12 molecules in a single turn with a pitch corresponding to the cell constant, c (205 Å). Neighboring helices are bridged by the artificial segment at the N termini of both molecules. Large overall B factors are observed perpendicular to c, probably due to the weak lattice contacts in this direction. Structural differences from HIV and SIV MAs. The crystal structure of the nonprimate lentivirus EIAV MA is strikingly similar to the crystal structures of the primate lentivirus MAs (Fig. 2). Overall, using automatic assignment with the program SHP (47), some 70% of the residues can be aligned, with r.m.s.d.’s of 1.9 and 1.8 Å for SIV and HIV-1 MAs, respectively. This is despite the fact that no sequence similarity between EIAV MA and either of the primate lentivirus MAs could be detected. Nevertheless a structure-based sequence alignment (Fig. 3) identified 10 conserved residues, most of which are within the helical sections of the molecule and some of which are probably responsible for maintaining the characteristic MA architecture. This is reflected in the fact that EIAV MA has five long helices and one 310 helical loop (Fig. 2B and C and 3), each of which has an equivalent in SIV and HIV-1. In particular helices 1, 2, and 4 and their flanking residues superimpose well (Fig. 2A). The C-terminal region following helix 5 is invisible in the EIAV MA structure, while it is observed, albeit with different secondary structures, in the HIV-1 and SIV crystal structures (Fig. 2A). We do not think that this represents a significant difference since the structure of HIV MA determined by nuclear magnetic resonance (NMR) (28, 30, 31) demonstrated that this region is disordered; rather we argue that the EIAV MA structure supports the suggestion that the well-ordered structures seen in the crystals are artifacts of crystallization (29). Like HIV-1 and SIV MAs, EIAV MA resembles broadly the structures of nonlentivirus retroviral MAs (data not shown) in terms of the relative orientations of the corresponding helices and the lengths of the intervening loops (although MAs from nonlentiviruses do not have the fifth helix [7, 8]). Despite the overall similarity there are shifts in the relative positions of the helices compared to the MAs of other lentiviruses. This results in a compacting of the structure in the direction tangential to the membrane surface (orienting the molecule as in the SIV and HIV-1 trimers). Most noticeably comparison with the other structures reveals that the N- and C-terminal helices bend in toward the body of the molecule such that the termini are relatively closer. The N terminus of helix 1 bends down compared to the same helix in HIV-1 and SIV such that the corresponding residues, HIV-1 Gly10 and EIAV Ser8, are 7 Å apart (Fig. 2A). In all structures the N terminus of this helix approaches the C-terminal end of the 310 helical loop, residues 42 to 52 (Fig. 2B). In SIV and HIV-1 the residues prior to residue 6 or 7 are disordered but at this point protrude from the membrane-binding surface, while in EIAV the preceding residues point in the opposite direction such that the first methionine is located on the molecular surface opposite to the loop joining the first and second helices (Fig. 2A).


1879

The extrinsic sequence preceding the natural N terminus runs in different directions in the two NCS-related molecules, but essentially the same conformation is adopted after residue Gly2 (Fig. 1B). There are stabilizing hydrogen bonds between N-terminal residues 3, 5, and 7 and residues 53, 52, and 50 at the junction between the 310 helical loop and the third helix. In the C-terminal or fifth helix in EIAV MA a bending of the chain at Gly97 allows this helix to pack against the body of MA (Fig. 2A and B) such that the C terminus, residue 109, approaches Pro58 of helix 3. The fifth helix also mediates contacts between molecules in the crystal together with helix 2 and the loop between helices 3 and 4 of the adjacent molecule. The segment including the 310 helical loop and the third helix is restructured in EIAV compared to the SIV and HIV-1 crystal structures (Fig. 2A). This region is adjacent to the two termini. The third helix appears to have similar lengths in EIAV and other lentivirus MAs but is actually shifted by three residues (this is clearly seen in Fig. 3). In EIAV MA the loop between the 310 helical loop and the third helix is longer and the loop between the third and fourth helices is shorter, although there is a single-residue insertion compared to HIV-1 MA (Fig. 2A). In addition, the third helix bends at residue Pro58 in EIAV MA. The corresponding residues in HIV-1 and SIV MA are Gly62 (Fig. 3) and Ser62, respectively, which could facilitate a similar bending of the helix by virtue of their flexibility and hydrogen-bonding potential, respectively. Modeling a trimer of EIAV MA based on the SIV and HIV-1 structures leads to a mismatch at the intertrimer interface (Fig. 2C). In SIV and HIV-1 MA there is an intermolecular antiparallel ␤-sheet-like structure with two backbone hydrogen bonds between segments Ala45-Val46-Asn47 and Thr70-Gly71-Ser72 (in HIV-1), corresponding to residues His41-Asp42-Thr43 and Thr66-Leu67-Ser68 in EIAV. The refolding of the 310 helical loop and the loop connecting the third and fourth helices abolishes the potential for such a structure. Indeed the conformation adopted by key residues of the trimer interface more closely resembles the monomeric HIV-1 model determined by NMR than the HIV-1 and SIV trimeric crystal structures. Thus it is quite possible that the EIAV structure could undergo a conformational change in order to make the trimer. An electrostatic depiction of an EIAV monomer reveals a basic surface and a slightly acidic surface. However, a trimer generated using the same symmetry as for the SIV trimer is less strikingly bipartite than either of the HIV or SIV trimers, with an acidic patch in the center of the upper membrane-binding surface instead of a basic surface (Fig. 2D). This negative charge could perhaps be counteracted by the binding of metal ions. A strongly basic patch is observed at the edge of the trimer comprising lysines 9, 12, 13, 16, 24, and 49. In fact the first helix of EIAV (residues 6 to 17) is amphipathic in character. DISCUSSION SIV and HIV-1 MAs form trimeric structures in the crystal, believed to correspond to a membrane-bound conformation. EIAV MA has a helical organization with a novel contact mode (Fig. 1C). Overall the structure is highly similar to those of HIV-1 and SIV MA but is compacted such that the N and

1880

HATANAKA ET AL.

J. VIROL.

FIG. 2. Structural comparison of EIAV MA with other lentivirus MAs. (A) Stereo superimposition of MAs of EIAV (color coded by B factor from blue to red), HIV-1 (green), and SIV (yellow), all shown as worms. Thick worms, regions used for superimposition (residues 7 to 20, 23 to 41, and 69 to 98 in EIAV MA), with the termini labeled. L and L1, loop regions exhibiting greatest deviation (310 helical loop and mobile loop between helices 1 and 2, respectively). (B) Stereo cartoon representation of EIAV MA molecules shown modeled in a trimeric organization, based

VOL. 76, 2002

C termini are in closer proximity. Both the N- and C-terminal regions are involved in crystal contacts, but these are unlikely to induce large-scale reorientation of the N- and C-terminal helices. Crystal contacts are more likely to trap a structure close to a minimum energy state of a molecule (which may or may not be biologically relevant). The packing down of the N-terminal helix leads to a rearrangement in the 310 helical loop, which prevents trimerization; therefore, the state we observe is not assembly competent and is unlikely to interact with a membrane. We suggest that the state we observe may be closer to the post-membrane-bound conformation of the molecule. In the HIV-1 and SIV crystal structures the residues for the fifth helix are detached from the body of MA. We propose that in the Gag precursor this detachment is facilitated by the tension between the MA and CA domains. Fuller et al. (15, 49) hypothesized that the C-terminal one-third of MA is included in the low-density layer of cryoelectron microscope images of the virus-like particle, suggesting that in the Gag precursor the sequence C-terminal to the fourth helix is unfolded. This is in line with the experiments of Zhou and Resh (52), who gradually truncated HIV-1 MA from its C terminus and found that the removal of helix residues 97 to 109 dramatically enhanced binding to membranes. The proximity of the termini in the EIAV structure suggests a direct mechanism for communication between the two termini via a refolding of the 310 helical loop-third helix region with which both interact. Thus, the fifth helix attaches to the body of MA, as would be predicted to occur following cleavage, and the ensuing repacking of the 310 loop redirects the N terminus. The affinity of Phe101 (Ile104 of HIV-1) to the most hydrophobic region of the third helix, namely, 56-VIPLL-60 (60-ILGQL-64 in HIV-1) may cause the helix to slide along. Then one turn of this helix unwinds into an extended loop at the N-terminal end, allowing the 310 helical loop to be translated upwards toward the membrane, where Val46 (Leu51 in HIV-1) blocks the N terminus and orientates it away from the membrane. Thus, by comparing HIV and EIAV matrix proteins, we find differences in the oligomeric state correlated with conformation. Cooperativity between the changes in the oligomeric state and the conformational changes modifying the association with membranes would be in line with the severalminute time lag for MA release from the membrane after proteolytic cleavage (20). Mutation R55W in Mason-Pfizer monkey virus converts this type D retrovirus phenotype to type C (the same as that for HIV-1), i.e., redirecting Gag assembly from a cytoplasmic site to the plasma membrane (42). A structural comparison shows that this mutation occurs at a site corresponding to Gly56 in HIV-1 (Gln52 in EIAV) located in


1881

the first pitch of the third helix, where we would expect structural rearrangements to affect the assembly competence and capacity for membrane binding. As far as we are aware differential membrane binding of MA and Gag has been examined only in HIV-1. We therefore cannot be certain that it occurs in EIAV and other retroviruses. Since it is common for the N termini of MA molecules to have a hydrophobic modification, it is expected that the N terminus might be sequestered in the post-membrane-bound state (52). However in the crystal structure, the N terminus of EIAV MA is not correctly modified and is involved in bridging interactions. Two totally hydrophobic cavities are found by the program GRASP (36) in both of the two NCS-related MA molecules (Fig. 2B); in contrast no cavities are found in either of the trimeric MA structures. While these cavities are too small to accommodate a large hydrophobic group (e.g., myristoyl or acetylmethionyl) without further conformational change, they do suggest a loose packing of the hydrophobic core, which might allow the cavities to inflate or conflate into a single cavity. The EIAV structure determined contains additional, extrinsic residues prior to the natural N terminus, which are observed in two different conformations in the crystal. Thus the extrinsic residues cannot mimic the conformation of a natural N-terminal modification; however, conversely, they do not by virtue of a single erroneous conformation distort the structure at the natural N terminus. It is possible that by providing some bulk adjacent to the natural N terminus they might, in some generic way, emulate the effect of an N-terminal modification. Membrane-binding assays with His-tagged and untagged EIAV show that unnatural modification of the N terminus is possible without significantly affecting the biological properties (39). The EIAV MA structure, however, suggests that it may use a different mechanism for membrane targeting. The amphipathic character of the first helix (Fig. 2C) is ideally suited for wedging between the phospholipid head groups in a membrane. Further, examination of the B factors (Fig. 2B and C) for the EIAV structure reveals a highly mobile hinge (residues 20 to 23), which adopts a conformation more similar to that observed in the HIV-1 MA solution structure (determined by NMR) (28, 30, 31) than that observed in the HIV-1 and SIV crystal structures. It is thus possible that the N-terminal portion can swivel round, possibly permitting anchorage in the membrane. This may represent a purer form of the myristoyl switch, achieved through the protein rather than myristate. Provitera et al. (39) determined that the binding of EIAV to negatively charged (POPS) bilayers protected amino-terminally located Arg and Lys residues, which would be in line with this suggestion and with their report that there is no significant penetration of the membrane surface on binding. The same

on the SIV and HIV crystal structures. Molecules are viewed from the side with the putative membrane-binding surface uppermost. Five helices (H1 to H5) and a helical loop (L) are depicted as rods and colored semitransparently from blue to red with increasing B factor. Side chains of some residues are also shown (yellow and green). Intramolecular cavities found in both NCS-related molecules are shown in mesh for one monomer. Cavities found in molecule A are drawn in green, cyan, and blue, and those found in molecule B are drawn in magenta and red. (C) Same representation as for panel B except that it is shown opaque and is viewed looking down onto the putative membrane-binding surface. Cyan spheres, residues that make the intertrimer contacts in the other lentiviruses; the steric clash between the molecules in these regions can be seen. The basic residues on the amphipathic helix 1 and adjacent 310 helical loop are blue. (D) GRASP representation of the electrostatic properties of the upper putative membrane-binding surface of the EIAV (left) and HIV trimers. The electrostatic potential is shown on the accessible surface and colored on a similar scale for each trimer, with positive charge in blue and negative charge in red. Images were generated using BOBSCRIPT (11, 12) and Raster3D (33).

1882

HATANAKA ET AL.

J. VIROL.

FIG. 3. Structure-based sequence alignment of EIAV MA with HIV-1 MA. Residues conserved perfectly among the MA sequences of EIAV, HIV-1, and SIV strains used for crystal structure determinations are boxed with a black background. Helices (squiggles along the sequence) were assigned manually. The image was produced by ESPript (17).

authors observe that EIAV MA exists as a multimer in solution whose protein-protein interactions are destabilized by membrane binding. They suggest that the protein can form dimers, possibly related to the intermolecular contacts found in our crystal. Bukrinskaya et al. (4) suggested that serine phosphorylation of MA disrupts membrane binding in the early stages of the virus cycle, facilitating translocation to the nucleus. Structurebased sequence alignment (Fig. 3) confirms that the nuclear localization signal found at residues 25 to 33 of HIV-1 MA (5) is not conserved in EIAV MA, and it remains to be determined whether EIAV MA is actively imported into the nucleus, although an EIAV-derived vector system can transform nondividing cells (34). Moreover, the location of the nuclear export signal (NES) in HIV-1 MA (10) is occupied by a leucine-rich type NES consensus sequence (LX2-3LX2-3LXL) in EIAV MA (11-LKKLEKVTV-19). One of the phosphorylated serines in HIV-1, Ser111, lies on the C-terminal helix. This corresponds to residue Tyr108 in EIAV, which interacts with Arg54 (Arg58 in HIV-1), the only basic amino acid in the third helix (Fig. 2B). Ser111 phosphorylation would therefore favor the state observed in EIAV. It would be interesting to know if Tyr108 of EIAV is ever phosphorylated. In summary, combining our observations on MA with those on CA (48), we suggest that the following sequence of events occurs on retrovirus maturation. In immature virions, the Gag precursor binds tightly to the virus envelope using MA and makes a hexagonal network by lateral oligomerization at the MA and CA regions, which are spaced by an unfolded region. On maturation, proteolytic cleavage by retrovirus protease releases the C terminus of MA and the N terminus of CA. The positive charge of the latter induces the refolding of CA Nterminal residues, and a new CA-CA interface converts the CA network into the condensed capsid. In contrast, the C-terminal residues of MA refold into the fifth helix, attach to the MA body, and, by refolding the 310 helical loop-third helix, reorient the N terminus away from the virus membrane. MA remains attached to the envelope of the mature virion simply due to its high concentration; it is released into the cytoplasm once the virus fuses with a new cell. Its switching and dissociation are enhanced by a cooperative mechanism and phosphorylation. Finally, some MA molecules are transported into the nucleus,

where they fulfil a further function at an early stage of the retrovirus life cycle. ACKNOWLEDGMENTS We thank K. Harlos, J. Grimes, and K. Maenaka for advice; E. Y. Jones, A. Sharma, and S. Ikemizu for help with data collection; and R. Esnouf for computing support. L. Vincent helped with manuscript preparation, R. Aplin performed the mass spectrometry, and T. Willis performed N-terminal sequencing. This work was supported by the UK Medical Research Council. H.H. was supported by the Uehara Foundation. REFERENCES 1. Ames, J. B., R. Ishima, T. Tanaka, J. I. Gordon, L. Stryer, and M. Ikura. 1997. Molecular mechanisms of calcium-myristoyl switches. Nature 389:198– 202. 2. Brünger, A. T., P. D. Adams, G. M. Clore, W. L. DeLano, P. Gros, R. W. Grosse-Kunstleve, J.-S. Jiang, J. Kuszewski, M. Nilges, N. S. Pannu, R. J. Read, L. M. Rice, T. Simonson, and G. L. Warren. 1998. Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D 54:905–921. 3. Bryant, M., and L. Ratner. 1990. Myristoylation-dependent replication and assembly of human immunodeficiency virus 1. Proc. Natl. Acad. Sci. USA 87:523–527. 4. Bukrinskaya, A. G., A. Ghorpade, N. K. Heinzinger, T. E. Smithgrail, R. E. Lewis, and M. Stevenson. 1996. Phosphorylation-dependent human immunodeficiency virus type 1 infection and nuclear targeting of viral DNA. Proc. Natl. Acad. Sci. USA 93:367–371. 5. Bukrinsky, M. I., S. Haggerty, M. P. Dempsey, N. Sharova, A. Adzhubel, L. Spitz, P. Lewis, D. Goldfarb, M. Emerman, and M. Stevenson. 1993. A nuclear localization signal within HIV-1 matrix protein that governs infection of non-dividing cells. Nature 365:666–669. 6. Coffin, J. M. 1996. Retroviridae: the viruses and their replication, p. 1767– 1847. In B. N. Fields, D. M. Knipe, and P. M. Howley (ed.), Fields virology, 3rd ed. Lippincott-Raven Publishers, Philadelphia, Pa. 7. Conte, M. R., and S. Matthews. 1998. Retroviral matrix proteins: a structural perspective. Virology 246:191–198. 8. Conte, M. R., M. Klinova, E. Hunter, T. Ruml, and S. Matthews. 1997. The three-dimensional solution structure of the matrix protein from the type D retrovirus, the Mason-Pfizer monkey virus, and implications for the morphology of retroviral assembly. EMBO J. 16:5819–5826. 9. de la Fortelle, E., and G. Bricogne. 1997. Maximum-likelihood heavy-atom parameter refinement in the MIR and MAD methods. Methods Enzymol. 276:472–494. 10. Dupont, S., N. Shanova, C. DéHoralius, C.-M. A. Virbasius, X. Zhu, A. G. Bukrinskaya, M. Stevenson, and M. R. Green. 1999. A novel nuclear export activity in HIV-1 matrix protein required for viral replication. Nature 402: 681–685. 11. Esnouf, R. M. 1997. An extensively modified version of Molscript which includes greatly enhanced colouring capacities. J. Mol. Graph. 15:133–138. 12. Esnouf, R. M. 1999. Further additions to Molscript version 1.4, including reading and contouring of electron-density maps. Acta Crystallogr. D 55: 938–940. 13. Freed, E. O. 1998. HIV-1 Gag proteins: diverse functions in the virus life cycle. Virology 251:1–15.

VOL. 76, 2002 14. French, S., and K. Wilson. 1978. On the treatment of negative intensity observations. Acta Crystallogr. A 34:517–525. 15. Fuller, S. D., T. Wilk, B. E. Gowen, H.-G. Kräusslich, and V. M. Vogt. 1997. Cryoelectron microscopy reveals ordered domains in the immature HIV-1 particle. Curr. Biol. 7:729–738. 16. Göttlinger, H. G., J. G. Sodroski, and W. A. Haseltine. 1989. Role of capsid precursor processing and myristoylation in morphogenesis and infectivity of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 86:5781– 5785. 17. Gouet, P., E. Courcelle, D. Stuart, and F. Métoz. 1999. ESPript: multiple sequence alignments in Postscript. Bioinformatics 15:305–308. 18. Harlos, K. 1992. Micro-bridges for sitting-drop crystallizations. J. Appl. Crystallogr. 25:536–538. 19. Henderson, L. E., R. C. Sowder, G. W. Smythers, and S. Oroszlan. 1987. Chemical and immunological characterizations of equine infectious anemia virus gag-encoded proteins. J. Virol. 61:1116–1124. 20. Hermida-Matsumoto, L., and M. D. Resh. 1999. Human immunodeficiency virus type 1 protease triggers a myristoyl switch that modulates membrane binding of Pr55 gag and p17MA. J. Virol. 73:1902–1908. 21. Hill, C. P., D. Worthylake, D. P. Bancroft, A. M. Christensen, and W. I. Sundquist. 1996. Crystal structures of the trimeric human immunodeficiency virus type 1 matrix protein: implications for membrane association and assembly. Proc. Natl. Acad. Sci. USA 93:3099–3114. 22. Joag, S. V., E. B. Stephens, and O. Narayan. 1996. Lentiviruses, p. 1977– 1996. In B. N. Fields, D. M. Knipe, and P. M. Howley (ed.), Fields virology, 3rd ed. Lippincott-Raven Publishers, Philadelphia, Pa. 23. Jones, T. A. 1985. Interactive computer graphics: FRODO. Methods Enzymol. 115:157–171. 24. Jones, T. A., J. Y. Zou, S. W. Cowan, and M. Kjelgaard. 1991. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47:110–119. 25. Laskowski, R. A., M. W. MacArthur, D. S. Moss, and J. M. Thornton. 1993. PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26:283–291. 26. LeMaster, D. M., and F. M. Richards. 1985. 1H-15N heteronuclear NMR studies of Escherichia coli thioredoxin in samples isotropically labeled by residue type. Biochemistry 24:7263–7268. 27. Lever, A. M. L. 1996. HIV and other lentivirus-based vectors. Gene Ther. 3:470–471. 28. Massiah, M. A., M. R. Starich, C. Paschall, M. F. Summers, A. M. Christensen, and W. I. Sundquist. 1994. Three-dimensional structure of the human immunodeficiency virus type 1 matrix protein. J. Mol. Biol. 244:198– 223. 29. Massiah, M. A., D. Worthylake, A. M. Christensen, W. I. Sundquivst, C. P. Hill, and M. F. Summers. 1996. Comparison of the NMR and X-ray structures of the HIV-1 matrix protein: evidence for conformational changes during viral assembly. Protein Sci. 5:2391–2398. 30. Matthews, S., P. Barlow, J. Boyd, G. Barton, R. Russell, H. Mills, M. Cunningham, N. Meyers, N. Burns, N. Clark, S. Kingsman, A. Kingsman, and I. Campbell. 1994. Structural similarity between the p17 matrix protein of HIV-1 and interferon-␥. Nature 370:666–668. 31. Matthews, S., P. Barlow, N. Clark, S. Kingsman, A. Kingsman, and I. Campbell. 1995. Refined solution structure of p17, the HIV matrix protein. Biochem. Soc. Trans. 23:725–728. 32. Maury, W. 1998. Regulation of equine infectious anemia virus expression. J. Biomed. Sci. 5:11–23. 33. Merrit, E. A., and M. E. P. Murphy. 1994. Raster3D version 2.0. A program for photorealistic molecular graphics. Acta Crystallogr. D 50:869–873.


1883

34. Mitrophanous, K. A., S. Yoon, J. B. Rohll, D. Patil, F. J. Wilkes, V. N. Kim, A. J. Kingsman, and N. D. Mazarakis. 1999. Stable gene transfer to the nervous system using a non-primate lentiviral vector. Gene Ther. 6:1808– 1818. 35. Nermut, M. V., D. J. Hockley, J. B. Jowett, I. M. Jones, M. Garreau, and D. Thomas. 1994. Fullerene-like organization of HIV gag-protein shell in viruslike particles produced by recombinant baculovirus. Virology 198:288–296. 36. Nicholls, A., K. A. Sharp, and B. Honig. 1991. Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11:281–296. 37. Otwinowski, Z., and W. Minor. 1997. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276:307–326. 38. Palmiter, R. D., J. Gagnon, V. M. Vogt, S. Ripley, and R. N. Eisenman. 1978. The NH2-terminal sequence of the avian oncovirus gag precursor polyprotein (Pr76 gag). Virology 91:423–433. 39. Provitera, P., F. Bouamr, D. Murray, C. Carter, and S. Scarlata. 2000. Binding of equine infectious anemia virus matrix protein to membrane bilayers involves multiple interactions. J. Mol. Biol. 296:887–898. 40. Rao, Z., A. S. Belyaev, E. Fry, P. Roy, I. M. Jones, and D. I. Stuart. 1995. Crystal structure of SIV matrix antigen and implications for virus assembly. Nature 378:743–747. 41. Reil, H., A. Bukovsky, H. R. Gelderblom, and H. G. Göttlinger. 1998. Efficient HIV-1 replication can occur in the absence of the viral matrix protein. EMBO J. 17:2699–2708. 42. Rhee, S. S., and E. Hunter. 1990. A single amino acid substitution within the matrix protein of a type D retrovirus converts its morphogenesis to that of a type C retrovirus. Cell 78:77–86. 43. Schultz, A. M., L. E. Henderson, and S. Oroszlan. 1988. Fatty acylation of proteins. Annu. Rev. Cell Biol. 4:611–647. 44. Spearman, P., R. Horton, L. Ratner, and I. Kuli-Zade. 1997. Membrane binding of human immunodeficiency virus type 1 matrix protein in vivo supports a conformational myristyl switch mechanism. J. Virol. 71:6582– 6592. 45. Stephens, R. M., J. W. Casey, and N. R. Rice. 1986. Equine infectious anemia virus gag and pol genes: relatedness to visna and AIDS virus. Science 231: 589–594. 46. Stevenson, M. 1996. Portals of entry: uncovering HIV nuclear transport pathways. Trends Cell Biol. 6:9–15. 47. Stuart, D. I., M. Levine, H. Muirhead, and D. K. Stammers. 1979. Crystal structure of cat muscle pyruvate kinase at a resolution of 2.6 Å. J. Mol. Biol. 134:109–142. 48. von Schwedler, U. K., T. L. Stemmler, V. Y. Klishko, S. Li, K. H. Albertine, D. R. Davis, and W. I. Sundquist. 1998. Proteolytic refolding of the HIV-1 capsid protein amino-terminus facilitates viral core assembly. EMBO J. 17:1555–1568. 49. Wilk, T., I. Gross, B. E. Gowen, T. Rutten, F. de Haas, R. Welker, H. G. Krausslich, P. Boulanger, and S. D. Fuller. 2001. Organization of immature human immunodeficiency virus type 1. J. Virol. 75:759–771. 50. Wills, J. W., R. C. Craven, and J. A. Achacoso. 1989. Creation and expression of myristylated forms of Rous sarcoma virus Gag protein in mammalian cells. J. Virol. 63:4331–4343. 51. Zhou, W., L. J. Parent, J. W. Wills, and M. D. Resh. 1994. Identification of a membrane-binding domain within the amino-terminal region of human immunodeficiency virus type 1 Gag protein which interacts with acidic phospholipids. J. Virol. 68:2556–2569. 52. Zhou, W., and M. D. Resh. 1996. Differential membrane binding of the human immunodeficiency virus type 1 matrix protein. J. Virol. 70:8540–8548.