MINIREVIEW The capsid protein of human immunodeficiency virus: intersubunit interactions during virus assembly Mauricio G. Mateu Centro de Biologı ´ a Molecular ‘Severo Ochoa’ (CSIC-UAM), Universidad Auto ´ noma de Madrid, Spain Introduction During HIV-1 morphogenesis [1,2], the capsid protein (CA; or p24) participates in two distinct assembly events. The first occurs inside the cell and involves the Gag polyprotein, of which CA constitutes a part. A spherical capsid comprising up to 5000 Gag subunits is formed through self-association around a dimer of the viral RNA genome, which is encapsidated along with several viral and cellular proteins. Assembly-competent Gag molecules are bound to the plasma membrane and may directly interact with molecules of the viral envelope polyprotein, which are embedded in the membrane. Thus, condensation of the capsid drives its coating by an envelope polyprotein-containing lipid bilayer. As a result of this morphogenetic process, an immature, non-infectious HIV-1 particle buds from the infected cell. Keywords capsid; conformational stability and dynamics; human immunodeficiency virus; molecular recognition; protein association; protein conformation; protein structure– function relationships; virus assembly Correspondence M. G. Mateu, Centro de Biologı ´ a Molecular ‘Severo Ochoa’, Universidad Auto ´ noma de Madrid, Cantoblanco, 28049 Madrid, Spain Fax: +34 91 1964420 Tel: +34 91 1964575 E-mail: mgarcia@cbm.uam.es Website: http://www.cbm.uam.es/ mkfactory.esdomain/webs/CBMSO/ plt_LineasInvestigacion.aspx? IdObjeto=19&ChangeLanguage=2 (Received 23 February 2009, revised 12 August 2009, accepted 20 August 2009) doi:10.1111/j.1742-4658.2009.07313.x The capsid protein (CA) of HIV-1 is composed of two domains, the N-ter- minal domain (NTD) and the C-terminal domain (CTD). During the assembly of the immature HIV-1 particle, both CA domains constitute a part of the Gag polyprotein, which forms a spherical capsid comprising up to 5000 radially arranged, extended subunits. Gag–Gag interactions in the immature capsid are mediated in large part by interactions between CA domains, which are involved in the formation of a lattice of connected Gag hexamers. After Gag proteolysis during virus maturation, the CA protein is released, and approximately 1000–1500 free CA subunits self-assemble into a truncated cone-shaped capsid. In the mature capsid, NTD–NTD and NTD–CTD interfaces are involved in the formation of CA hexamers, and CTD–CTD interfaces connect neighboring hexamers through homodi- merization. The CA–CA interfaces involved in the assembly of the imma- ture capsid and those forming the mature capsid are different, at least in part. CA appears to have evolved an extraordinary conformational plastic- ity, which allows the creation of multiple CA–CA interfaces and the occur- rence of CA conformational switches. This minireview focuses on recent structure–function studies of the diverse CA–CA interactions and interfaces involved in HIV-1 assembly. Those studies are leading to a better under- standing of molecular recognition events during virus morphogenesis, and are also relevant for the development of anti-HIV drugs that are able to interfere with capsid assembly or disassembly. Abbreviations CA, capsid protein of HIV-1; cryoEM, cryoelectron microscopy; cryoET, cryoelectron tomography; CTD, C-terminal domain of CA; EM, electron microscopy; H–D, hydrogen–deuterium; MA, matrix protein; MHR, major homology region; MLV, murine leukemia virus; NC, nucleocapsid protein; NTD, N-terminal domain of CA; RSV, Rous sarcoma virus. 6098 FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS The second capsid assembly event occurs upon bud- ding of the immature virion. The viral protease-medi- ated processing of Gag into several independent proteins and peptides leads to the disassembly of the spherical capsid inside the virion. Some of the folded domains of Gag are then able to reassemble as inde- pendent proteins. The matrix protein (MA; the Gag N-terminal domain), remains associated with the viral membrane, forming a discontinuous inner layer. The nucleocapsid protein (NC) remains associated with the viral RNA to form the nucleocapsid. In addition, some 1000–1500 subunits out of a larger number of CA mol- ecules, released as a two-domain protein, self-assemble into a truncated cone-shaped, hollow structure, namely the mature HIV-1 capsid. The mature capsid and the nucleocapsid that it contains constitute the viral core. As a result of this dramatic structural rearrangement, the immature virion turns into a mature, infectious virion. During the viral cycle, CA is present in different structural environments: as a part of Gag, as an unassembled protein, and as an independent protein forming the mature capsid. Remarkably, the CA poly- peptide appears to have evolved an extraordinary con- formational plasticity, which allows the creation of diverse CA–CA interfaces and other CA–ligand inter- faces during HIV-1 morphogenesis. The present mini- review summarizes the results of recent studies regarding the diverse CA–CA interactions and inter- faces involved either in the assembly of the immature HIV-1 capsid, or in the reassembly of CA into a mature capsid. The detailed description of the structure and conformational dynamics of these interfaces, together with a quantitative dissection of their energet- ics, is providing deeper insights on the molecular recog- nition events responsible for the assembly and stability of the retrovirus capsid, and of viral capsids in general. Moreover, the successful rational or semi-rational design of anti-HIV drugs that are able to impair capsid assembly or disassembly [3] may critically depend on the availability of a sufficiently detailed structural and energetic definition of those interfaces. Identification of CA–CA interfaces in the immature HIV-1 capsid Although the full-length Gag protein can be isolated in soluble form as a homodimer or homotrimer, its atomic structure has not been determined. However, the atomic structures of the separate Gag domains from HIV-1 (and other retroviruses) have been solved by X-ray crystallography and ⁄ or NMR spectroscopy. The N-terminal domain of CA (NTD) and the C-ter- minal domain of CA (CTD) are small, globular and mainly helical. NTD contains a-helices 1–7 of CA, and is connected by a flexible linker to CTD, which contains a small 3 10 -helix, an extended strand and a-helices 8–11 of CA (corresponding to helices 1–4 of CTD) [4–7]. Structural, biochemical and mutational analyses of the immature HIV-1 capsid have been carried out on immature virions, or on capsid-like particles that can be assembled in vitro from full-length or truncated Gag molecules in the presence of nucleic acid and ⁄ or other components [8–16]. Electron microscopy (EM) of negatively stained immature capsid-like particles [8], as well as cryoelectron microscopy (cryoEM) [14,15] and cryoelectron tomography (cryoET) [17,18] of immature capsid-like particles or virions, has revealed a layered organization, which can be interpreted based on bio- chemical evidence and the superposition on the elec- tron density maps of the atomic structure of each domain. The Gag subunits are radially extended, with the N-terminal domain, MA, associated with the inner layer of the viral membrane, and the remaining domains forming the innermost protein layers accor- ding to their positions in Gag: MA, CA-NTD, CA-CTD, the spacer peptide SP1, NC, the spacer peptide SP2 and the C-terminal peptide p6, with the NC domain associated with the viral RNA. CryoEM imaging of immature HIV-1 capsids [14,15] has shown that Gag is organized as a lattice of hexa- mers. Pseudoatomic models obtained by fitting the atomic structures of both CA domains in cryoET maps of immature HIV-1 virions [17,18] indicate that poly- merization of Gag occurs mainly through the forma- tion of an intermediate layer of cup-shaped structures, each made of one hollow hexameric ring of CA domains placed immediately above a stem formed by six SP1 segments (Fig. 1). Mutational analyses are consistent with this view. Several Gag domains and spacer peptides were shown to be involved in the assembly and stability of the immature HIV-1 capsid, but CA plays a major role [19–24]. However, the specific CA–CA interfaces involved in Gag hexamerization and the joining of neighboring hexamers in the immature HIV-1 capsid lattice are still unclear. NTD has been shown to be non-essential for the assembly of the immature capsid [19]. On the other hand, in a comprehensive muta- tional analysis [23], it was found that mutations that impaired immature capsid assembly are located both in NTD (helices 4–6) and in CTD (the loop between helices 7 and 8 and helix 9). Helix 9 forms a major part of the dimerization inter- face in CA and in the free CTD [6,7]. Mutational M. G. Mateu Capsid protein interfaces in HIV-1 assembly FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS 6099 analyses [23] indicate that this interface could be involved in the structural organization of the immature capsid. However, a mutant CTD with a single amino acid deletion has been recently crystallized as a domain-swapped dimer [25]. The domain-swapped interface did not involve helix 9, but instead the major homology region (MHR), a highly conserved, 20- amino acid stretch folded as a strand-turn-helix-8 motif. Because of this unusual conservation, numerous mutational analyses have focused on this stretch, and have revealed that the MHR is important for many different steps in the HIV-1 life cycle, including the assembly of both the mature capsid, where it forms part of the NTD–CTD interface (see below), and of the immature capsid [20,23]. Because the MHR also forms a major part of the domain-swapped dimeriza- tion interface in the mutant CTD structure, it has been proposed that this interface could participate in the assembly of the immature HIV-1 capsid [25]. The structural and functional studies performed to date may be not discriminating enough to clearly favor the involvement of either the domain-swapped or the non- swapped CTD dimerization interfaces (or both) in the assembly and stability of the immature HIV-1 capsid. In summary, in a model [17] of the immature HIV-1 capsid that may be consistent with the experimental results achieved to date: (a) the capsid lattice is formed by Gag hexamers that are stabilized mainly by a six- helix bundle of SP1 peptides; (b) the hexamers are joined through CTD homodimerization (with the inter- face involving either helix 9 or the MHR); and (c) NTD residues participate to some extent in intra- hexamer and ⁄ or inter-hexamer interactions (Fig. 1). This model remains to be validated and extended to higher resolution, and it may be far from complete. In addition to CA, other Gag domains and other proteins are known or suspected to play important roles in the assembly and stabilization of the immature HIV-1 cap- sid. Furthermore, a defect-free, continuous lattice of hexamers could not form spherical particles. In an early EM study of immature capsid-like particles of HIV-1, a fullerene-like icosahedral shell was proposed [8], although later studies showed no evidence of icosa- hedral symmetry [14]. Closing the immature capsid through 12 pentameric CA rings (that can be consid- ered as lattice ‘defects’) cannot be excluded, but sub- stantial regions lacking ordered Gag were visible in detailed electron cryotomographs, suggesting that the immature HIV-1 capsid may be not fully closed [17]. Very recent studies by cryoET suggest that released immature virions include an incomplete but continuous Gag layer covering approximately two-thirds of the virion membrane [18,26]. Curvature in the hexameric lattice is mediated by incorporation of defects of dif- ferent geometries into the lattice, with no evidence of a preference for pentameric defects [18]. It was also sug- gested that late-budding structures with complete Gag capsids may not be normal precursors of extracellular HIV-1 virions [26]. Identification of CA–CA interfaces in the mature HIV-1 capsid CA multimers with the same structural organization as authentic mature capsids have been obtained by self- assembly of CA in vitro, even in the absence of any other biomolecule, and subjected to structural, bio- chemical and mutational analyses [27–37]. CryoEM studies of mature HIV-1 capsid-like particles assembled in vitro revealed that these are composed of an array of CA hexamers. Fitting the atomic structures of NTD and CTD on the cryoEM density map indicated that NTD connects the CA subunits in each hexamer, and CTD connects each hexameric ring to six neighbors through homodimerization [30] (Fig. 1). Similar arrays of CA hexamers were observed in cryoEM images of authentic cores isolated from HIV-1 virions [38]. Recently, a CA mutant was used to obtain large, spherical capsid-like particles that collapsed when deposited on EM grids. The flattened particles behaved as 2D CA crystals, and could be used to obtain a detailed 3D map by electron cryocrystallography [39] (Fig. 2A). Fitting the atomic structures of NTD and CTD in this map confirmed and substantially refined the mature HIV-1 capsid lattice model. Three different protein–protein interfaces were observed in the CA Immature Mature Top Side Fig. 1. Proposed models for the organization of the hexameric lat- tices in immature and mature HIV-1 capsids [17]. NTD, CTD and SP1 domains of Gag are respectively colored cyan, yellow and magenta. Scale bar = 8 nm. Figure reproduced with permission [17]. Capsid protein interfaces in HIV-1 assembly M. G. Mateu 6100 FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS lattice: the interface between NTD domains in each hexamer, the dimerization interface between CTD domains belonging to neighboring hexamers, and another interface between CTD and NTD domains belonging to neighboring subunits of the same hex- amer [39] (Fig. 2B–D). This structural model and the CA–CA interfaces involved are supported by muta- tional analyses and biochemical evidence [23,32–34,36] (Fig. 2E, F). In vitro polymerization of CA normally leads to open cylindrical structures, although closed, cone- shaped particles resembling authentic mature capsids can be also obtained. Despite the difference in shape, the cylindrical capsid-like particles are organized with the same hexameric lattice as the cone-shaped, authen- tic mature capsids [30,38,39]. A molecular model for the architecture of the mature HIV-1 capsid has been proposed that follows the principle of a fullerene cone: the body is composed of curved arrays of CA hexa- mers, and the ends are closed by inclusion at defined positions of 12 CA pentamers (acting as lattice ‘defects’) [29,30]. Very recently, cryoEM analyses of two in vitro assembled capsids of another retrovirus, Rous sarcoma virus (RSV), were described [40]. Both capsids are icos- ahedrally symmetric: one is composed of 12 CA penta- mers, and the other of 12 pentamers and 20 hexamers. Pseudoatomic models using the atomic structures of both CA domains revealed three distinct CA interfaces similar to those observed within and between hexamers A B C D EF Fig. 2. Structure of CA and the CA hexamer in the mature HIV-1 capsid [39]. (A) Electron density map obtained by electron cryocrys- tallography of the mature HIV-1 capsid lat- tice; (B, C) Pseudoatomic models of the CA monomer and hexamer, obtained by fitting the atomic structures of NTD (green) and CTD (blue) on the cryocrystallography electron density map. (C) Showing a top view of the hexamer, with one monomer outlined white. (D) View of the hexamer as in (C), but each CA monomer is depicted in a different color; (E, F) Mapping on a CA hexamer (top view and slabbed side view) of mutations to alanine, according to their effect on the assembly of mature HIV-1 capsid-like particles: (i) green or blue dots, mutations in NTD or CTD, respectively, that did not impair assembly; (ii) orange, yellow or red dots, mutations that impaired assembly and are respectively located in the NTD–NTD, CTD–CTD or NTD–CTD interfaces. Figure reproduced with permission [39]. M. G. Mateu Capsid protein interfaces in HIV-1 assembly FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS 6101 in the model of the HIV mature capsid. This study provides support for the fullerene model, and shows how pentamers can be accommodated in the retroviral capsid. Structural and energetic characterization of CA–CA interfaces in the mature HIV-1 capsid From the results reviewed above, it could be concluded that the picture of CA–CA and other protein–protein interfaces in the immature HIV-1 capsid remains blurred because of uncertainties on the structural ele- ments involved, and the possibility of transient interac- tions. By contrast, three different CA–CA interfaces have been clearly identified in the mature HIV-1 cap- sid, and mapped with relative accuracy. Some of the knowledge acquired to date on the detailed structure and energetics of intersubunit interfaces involved in the quaternary structure of the mature HIV-1 capsid is reviewed below. The CTD–CTD dimerization interface A high-resolution structural description of the CTD– CTD interface that connects neighboring hexamers in the mature HIV-1 capsid has been obtained by X-ray crystallography of dimeric CTD [6,7]. This interface is essentially formed by the parallel packing of helix 9 from each monomer, but also involves interactions between residues in the 3 10 -helix of one monomer and residues in helices 9 and 10 of the other monomer. No MHR residues are involved. The structural description of the dimerization interface in the isolated CTD dimer is fully consistent with descriptions of the CTD–CTD interface in the mature HIV-1 capsid. These latter descriptions derive from: (a) pseudoatomic models of capsid-like particles [30,39]; (b) analyses of the effect of CA mutations on the assembly of mature capsid- like particles (Fig. 2E,F) and ⁄ or the formation of viral cores; the results obtained showed that residues located in helix 9 (among others), impaired mature capsid assembly both in vivo and in vitro [23,34]; and (c) hydrogen–deuterium (H–D) exchange experiments analysed by MS. The residues buried in the intersub- unit interfaces could be identified by their slower H–D exchange in the assembled particle, relative to the free CA protein. Again, residues in helix 9 (and others) were involved in intersubunit interactions [33]. In the atomic structure of CTD, the dimerization interface involves some 22 amino acid residues from each monomer, and it would bury approximately 1800 A ˚ 2 of solvent-accessible area if no induced fit had occurred; approximately two-thirds of this area is con- tributed by nonpolar side chains [6,7]. The isolated CTD domain dimerizes with essentially the same affin- ity as full-length CA, and likely exhibits all of the ener- getically significant CTD–CTD interactions in CA [7]. This provided a very unusual opportunity for a quanti- tative thermodynamic dissection of a protein–protein interface in a virus capsid [36,41,42]. The individual energetic contribution of each interfacial side chain to CTD–CTD association was determined by alanine scanning mutagenesis on free CTD, using analytical gel filtration chromatography to determine the equilib- rium constant of the association step for each single mutant. It was found that removal of the side chain interactions of any one of almost half the residues at the interface (Ile150, Leu151, Arg154, Leu172, Glu175, Val181, Trp184, Met185 and Leu189) destabilized the CTD–CTD association by over 6 kcalÆmol )1 , leading in each case to essentially monomeric CTD (even at high protein concentrations) [42]. The CTD dimeriza- tion interface is formed by a central area of energeti- cally critical, mostly hydrophobic residues, surrounded by a ring of energetically less important, polar residues (Fig. 3). Those structural and energetic features of the CTD dimerization interface are typical of many protein–pro- tein interfaces, but these generally exhibit affinities higher than those of CA or CTD (K d = 10–20 lm). In CTD, the dimerization affinity is kept low partly because several interfacial side chains contribute each to substantially destabilize the CTD–CTD association [42]. Quantitative thermodynamic double mutant cycles clearly showed that a part of this destabilizing effect is a result of intersubunit electrostatic repulsions at the CTD–CTD interface, including those between Glu180 from both subunits, that may be conserved in HIV [36]. It was suggested that such repulsions could arise as one consequence of a selective pressure to maintain an optimum balance between capsid stability (i.e. for structural integrity in the virion) and instability (i.e. for viral core disintegration and RNA release in the infected cell) [36]. This thermodynamic description of the CTD dimerization interface may also apply to the interface as a part of the HIV-1 capsid: a good correlation was found between the effects on CTD dimerization and on capsid-like particle assembly of mutations that decreased, increased or preserved the affinity, or showed non-additive effects [6,34,36]. The detailed structural and thermodynamic descriptions obtained on the CTD–CTD interface have been already used in the field of anti-HIV research, for example for the design of a helix-9 peptide mimic [3,43], and are Capsid protein interfaces in HIV-1 assembly M. G. Mateu 6102 FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS currently being used in the development of higher- affinity inhibitors of HIV-1 assembly. The NTD–NTD hexamerization interface No hexamers of native CA from HIV-1 (with the CTD inactivated) or its isolated NTD have been obtained yet [32]. However, a high-resolution struc- tural description of the NTD–NTD interface in the mature HIV-1 capsid is available. The homologous NTD from MLV did crystallize in the form of hexa- meric rings, and the X-ray structure of this NTD hexamer could be obtained [44]. The tertiary struc- tures of NTD from MLV and HIV-1 are very simi- lar, and superposition of the atomic structure of monomeric NTD from HIV-1 on the MLV hexamer yielded a detailed model of quaternary interactions in the HIV-1 hexamer [45]. Most recently, the full- length CA from HIV-1 was engineered to form soluble, assembly-competent hexamers, and high-reso- lution crystallographic structures of the modified CA hexamers were obtained [46]. The modifications that yielded the most detailed structures involved the engineering of interprotein disulfide bridges through the introduction of Cys residues at the NTD–NTD interface (by mutations Ala14Cys and Glu45Cys) and the substantial weakening of CA–CA dimeriza- tion through the CTD–CTD interface (by mutations Trp184Ala and Met185Ala). The X-ray model of the NTD hexamer from MLV, the superimposed model of the NTD hexamer from HIV-1, the high-resolution X-ray model of the modified CA hexamer from HIV-1, and the pseudoatomic models of the HIV-1 hexamer obtained from cryo-EM [30] and electron cryocrystallography [39] maps are all in very good general agreement regarding the elements forming the NTD–NTD interface. Helices 1, 2 and 3 of NTD are closer to the central hole of the hexameric ring, forming a 18-helix bundle, with helices 1 lining the hole. The NTD–NTD interface is defined by contacts between residues in helices 1 and 3 from a NTD monomer and residues in helices 1 and 2 from the neighboring mono- mer. These models are also validated by: (a) mutational analysis, which showed that residues located in helix 1 or 2 impaired mature capsid assembly both in vivo and in vitro [23,34] (Fig. 2E,F), and (b) H–D exchange experiments, which identified residues in helices 1 and 2 as involved in intersubunit interactions in mature HIV-1 capsid-like particles [33]. No thermodynamic studies are available on the NTD–NTD interface. However, in the X-ray structure of the MLV hexamer, this interface buries a surface of only 1100 A ˚ 2 . Most of the interactions are weak polar contacts, including some mediated by water molecules, and a substantial hydrophobic central area is absent. All these features differentiate the NTD– NTD interface from the CTD dimerization interface and many other protein–protein interfaces, and may contribute to explain the extremely low intrinsic ten- dency of NTD to homo-oligomerize. Similar to that observed for the CTD dimerization interface, the binding free energy of the NTD–NTD interface may be also limited by intersubunit electrostatic repulsions between charged residues in NTD, although, in this case, those residues do not form a part of the inter- face itself [35]. The NTD–CTD interface The NTD–CTD interface between neighboring subun- its within the hexameric rings that form the mature HIV-1 capsid has been mapped based on different observations, that are summarized below. Fig. 3. Energetic dissection of the CTD–CTD dimerization interface of HIV-1 [42]. A spacefilling model of one of the monomeric subun- its in a CTD dimer is represented. The interfacial residues are color- coded according to the effect on CTD dimerization of removing the interactions of the side chain beyond the Cb (by mutation to ala- nine). The effect is quantitated according to the difference in free energy between nonmutated CTD and each mutant for association of the CTD monomers into a dimer (DDG a ). Red, truncation had a dramatic effect and led to monomeric CTD at all but the highest concentrations (DDG a > +6 kcalÆmol )1 ); orange, truncation had a substantial effect (DDG a = +1.1 kcalÆmol )1 ); green, truncation had no significant effect (DDG a < +0.3 to )0.3 kcalÆmol )1 ), or had only a small negative effect (Lys199; DDG a = +0.5 kcalÆmol )1 ); violet, mutation increased the dimerization affinity (DDG a = )0.4 to )0.8 kcalÆmol )1 ). Figure reproduced with permission [42]. M. G. Mateu Capsid protein interfaces in HIV-1 assembly FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS 6103 First, in the pseudoatomic models of capsid-like particles of the homologous RSV [40], the equivalent NTD–CTD interface involves the MHR, and previous genetic analysis revealed that mutations in the MHR causing a defficiency in assembly were compensated by secondary mutations that also map in the NTD–CTD interface [47–49]. Incidentally, one of these mutations alone eliminated a positive charge from a cluster of basic residues and increased the propensity of CA to assemble. Thus, it has been suggested that charge repulsion, which would be partially relieved by that mutation, may also naturally occur at this interface, at least in RSV [40]. Second, the inhibitory activity of isolated CTD on the in vitro polymerization of CA from HIV-1 was found to be relieved by addition of isolated NTD [41]. Third, crosslinking experiments and MS analysis revealed that, in capsid-like particles, Lys70 of NTD was in close proximity to Lys182 of CTD from a neighboring subunit [33]. Fourth, H–D exchange experiments indicated that the C-terminus of helix 3 and the N-terminus of helix 4 in NTD, which were not a part of the CA–CA inter- faces already identified, were additionally involved in intersubunit contacts [33]. Fifth, the very recent determination of the hexameric form of a modified full-length CA (with a crosslinked NTD–NTD interface and a substantially inactive CTD–CTD dimerization interface) has allowed a more complete analysis of the precise residues and interac- tions involved in the NTD–CTD interface [46]. Most of these involve side chains from helix 8 of the CTD which pack against the N-terminal end of helix 4 of the NTD of a neighboring subunit in the hexamer. Additional contacts involve helix 11 of the CTD and the C-terminal end of helix 7 of the NTD of the neigh- boring subunit. The NTD–CTD interface, similar to the NTD–NTD interface, lacks a substantial hydro- phobic core, and mainly involves polar interactions, including water-mediated hydrogen bonds and inter- domain helix-capping interactions. Finally, the pseudoatomic model of the HIV mature capsid lattice and mutational analysis [39] (Fig. 2E,F) are consistent with the above results. CA–CA interactions and stability of the mature HIV-1 capsid Mutations generally located at or close to CA–CA interfaces, and that either increase or decrease the sta- bility of mature HIV-1 capsid-like particles [34] and ⁄ or authentic HIV-1 cores [50], resulted in a loss of viral infectivity. Thus, the mature HIV-1 cap- sid ⁄ viral core does appear to have evolved an opti- mum, delicate balance between stability inside the virion and instability inside the infected cell. It is remarkable that different electrostatic repulsions between neighboring CA subunits through all three identified interfaces in the mature retrovirus capsid have been shown or suggested to occur [35,36,40]. Furthermore, covariant mutations during HIV evolu- tion may have preserved at least a CTD–CTD elec- trostatic repulsion that was unambiguosly revealed using a thermodynamic cycle [36]. The low oligomeri- zation affinity of CA and the stability balance of the HIV capsid may be a result, in part, of the preserva- tion of intersubunit electrostatic repulsions. Conformational rearrangements of CA and HIV-1 capsid assembly Both CA domains are covalently connected through a flexible linker, and they appear to be unusually flexible themselves. Accordingly, there is substantial evidence for the occurrence of different conformational rear- rangements of CA during the assembly of both the immature and the mature capsid of HIV-1 and other retroviruses. Induced conformational switching in CTD and ⁄ or NTD and ⁄ or alterations in the linker region could explain several observations, including: (a) the different ability of retroviral CAs and their free domains to multimerize or oligomerize in different conditions; (b) the different shapes of the immature and the mature HIV-1 capsid [32], and of the mature capsids of different retroviruses; (c) the different structural organization of CA in the immature and mature capsids, including the arrangement and spacing of the hexameric lattice [12] (Fig. 1); (d) the formation of both hexamers and penta- mers using essentially the same structural elements of CA [40]; (e) the different interactions of CA with several ligands and their effects on capsid assembly [3]; and (f) the possibility of transitions between alternative inter- faces during HIV-1 morphogenesis (see below). Conformational changes in CTD Biophysical and thermodynamic analysis of the process of HIV-1 CTD dimerization (from unfolded monomer to native dimer) [41,51] revealed a transient, folded monomeric intermediate of low conformational stabil- ity. This free monomer lacks a part of the structure of the monomeric subunits in the native, stable homo- dimer. CTD dimerization was found to involve a substantial conformational rearrangement of the inter- acting, transient monomers that was thermodynami- cally characterized. CTD mutant Trp184Ala is unable Capsid protein interfaces in HIV-1 assembly M. G. Mateu 6104 FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS to dimerize [6], even at 1 mm; biophysical and thermo- dynamic analysis of this monomeric mutant indicated that it could provide a good structural model for the transient monomer involved in the dimerization of nonmutated CTD [41]. Recent NMR studies of this CTD mutant were consistent with those results, and revealed that the tertiary structure of the isolated CTD monomer is not identical to that of the monomeric subunit in the dimer. In particular, the dimerization helix 9 is only transiently structured, and the last two helices are rotated by 90° compared to their position in dimeric CTD [52]. Subsequently, the structure of another monomeric CTD mutant, Trp184Ala ⁄ Met185Ala, was reported [53]. In this structure, helix 9 is shortened but formed, and the last two helices are placed as in the dimer. It has been proposed that the structural differences found for the two monomeric mutants may be a result of the different pHs used in the NMR experiments [54]. The structure of the monomer with the transiently formed helix was determined at neutral pH, whereas that of the monomer with the structured helix was determined at acidic pH and, when the pH was raised, the reso- nances belonging to that helix disappeared [53]. These pH-dependent changes in tertiary structure are not unusual in proteins [55,56]. In addition to its propensity to local rearrangements and dynamics, the free CTD monomer appears to be highly flexible overall [54]. Thus, both in vitro and dur- ing HIV-1 morphogenesis in vivo, the energetic balance between alternative CTD conformations could be altered by even subtle changes in the environment [12,31], mutation and ⁄ or ligand binding to CTD (such as that of capsid assembly inhibitor peptides CAI and NYAD-1) [3]. This model is consistent with the several modes of association observed for CTD in crystal form, including the balance between a domain- swapped CTD–CTD interface [25] and a nondomain- swapped interface, proposed to have a role during HIV-1 morphogenesis. In the context of the unusual conformational plastic- ity of CTD, it should be noted at this point that muta- tion of certain CTD residues, including residues belonging to the MHR, had substantial effects on CTD dimerization in solution [41], even though those residues and the MHR are located away from the rele- vant dimerization interface. Thus, it may be worth considering that, in addition or alternatively to a direct role in intersubunit interactions, some CA residues and the MHR could indirectly participate in HIV-1 morphogenesis; for example, by stabilizing an assem- bly-competent CA conformation and ⁄ or facilitating interactions between other residues. The lack of a perfect fitting of the closely-matching atomic structure of the CTD dimer [6,7] on the elec- tron cryocrystallography map of mature HIV-1 capsid- like particles has led to the suggestion that small conformational changes could also occur in the CTD dimerization interface, and in the tertiary structure of CTD itself, during assembly of the mature HIV-1 cap- sid [39]. The detailed atomic structure and quantitative thermodynamic description available for the CTD dimerization interface could be then considered to define the sterically unconstrained, minimum free energy conformation. Steric constraints in the mature HIV-1 capsid lattice could distort and destabilize somewhat the ‘ideal’ CTD–CTD interface (and per- haps other CA–CA interfaces). Such constraints could contribute, in addition to electrostatic repulsions and other effects, to establish the appropriate balance between stability and instability of the viral core. Conformational changes in NTD During maturation, the proteolytic cleavage of Gag at the linker between CTD and SP1 could destabilize the proposed SP1 six-helix bundle, facilitating disassembly of the immature capsid [17,57]. In addition, processing of the MA-NTD linker allows the folding as a b-hair- pin of the NTD N-terminal segment, leading to a local conformational change in NTD. This rearrangement has been proposed to destabilize the immature capsid and ⁄ or create the NTD–NTD interface observed in the mature HIV-1 capsid, thus promoting core assembly [58]. In addition, alterations in the H–D exchange pro- tection pattern when immature and mature virus-like particles were compared have provided evidence for a maturation-induced formation of the NTD–CTD inter- face observed in the mature HIV-1 capsid [59]. Ligand (e.g. CypA [60]) binding to NTD can also lead to conformational rearrangements. To summarize, conformational rearrangements allowed by the unusual structural plasticity of both CA domains, and their connection through a flexible linker, may shift their association equilibria during HIV-1 morphogenesis. The rearrangements proposed and their detailed molecular description remain to be further substantiated in future structural studies. Macromolecular crowding effects on CA–CA association and HIV-1 capsid assembly Both NTD and CTD of retroviruses have an intrinsi- cally low or undetectable tendency to oligomerize in solution, and conformational rearrangements in CA M. G. Mateu Capsid protein interfaces in HIV-1 assembly FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS 6105 (including those reviewed above) may not be sufficient to allow the establishment of the relatively stable CA– CA interactions observed during HIV-1 assembly. For example, formation of a b-hairpin in NTD during maturation may not provide an explanation for the apparent inability of unmodified CA and its NTD to hexamerize in solution because the hairpin is already formed in them [45]. Similarly, the induction of a con- formational rearrangement in NTD through interac- tion with CTD is unlikely to explain why CA with the dimerization interface inactivated through mutation is unable to hexamerize in the same conditions where nonmutated CA is able to polymerize into capsid-like particles [33]. Indeed, there is evidence that, in addition to conformational changes in CA, physicochemical conditions having an effect on the chemical activity (‘effective concentration’) of CA may also play a major role in the associative properties of NTD and CTD to form the HIV-1 capsid, as reviewed below. The concentration of CA inside the mature HIV-1 virion may be at least 3.5 mm [38] (approximately 8mm if an estimation of close to 5000 CA molecules per virion [16] is accepted). However, in the very lim- ited space available inside a mature HIV-1 virion, thousands of molecules of MA, CA, SP1, NC, SP2 and p6, as well hundreds of other viral and cellular protein molecules and two long RNA molecules, are found. Thus, in the virion, as in the cell, macromolecu- lar crowding effects must be in operation as a result of the exclusion of water molecules from the large frac- tion of internal volume occupied by the macromole- cules themselves [61]. The chemical activity, or ‘effective concentration’ of CA in the HIV-1 virion (upon release from Gag during maturation) must be not a few millimolar, but much higher. Under these conditions, protein association reactions, such as CA assembly, must be strongly favored [61]. The available experimental evidence summarized below is consistent with this prediction. In vitro polymerization of CA into mature HIV-1 capsid-like particles in the virtual absence of macro- molecular crowding (at maximum protein concentrations in the order of 500 lm, which is far lower than the CA effective concentration in the virion) was achieved only in nonphysiological conditions, such as: (a) a very high ionic strength [32] that could screen electrostatic repul- sions [35,36] or (b) the use of a CA–NC fusion protein in the presence of nucleic acid [9,28,29], which could increase the CA local concentration through multiple NC-mediated interactions with CA–NC. Similarly, it has been suggested that putative interactions between CA and a NC–RNA complex could initiate HIV-1 core assembly in vivo [38]. However, this possibility may have to be re-evaluated if the chemical activity of CA inside the maturing virion (as a result of macromolecu- lar crowding) is taken into account. By contrast to previous in vitro assembly procedures, very efficient assembly of mature HIV-1 capsid-like particles was recently achieved using free CA at moder- ate concentrations, in the absence of any other bio- molecule, and at physiological ionic strength [37]. The procedure was simply based on the addition of inert macromolecular crowding agents to increase the CA effective concentration to very high values, closer to those present in the HIV-1 virion. The capsid-like tubu- lar and, occasionally, cone-shaped particles formed were indistinguishable by EM from those obtained at very high salt concentrations (and known to be orga- nized similarly to authentic mature HIV-1 capsids). The capsid-like particles formed at close-to-physio- logical CA effective concentration and ionic strength were kinetically much less stable than those formed at high ionic strength [37]. This reinforces the view that electrostatic repulsions could contribute to the observed HIV-1 core instability in infected cells. In addition, the release of the viral core and many free CA molecules from the confined space in the virion into the very large volume within the cell would also facilitate uncoating by ‘dilution’ (i.e. through a dra- matic decrease in the CA effective concentration) [50]. These observations indicate that the very high chemical activity of CA inside the virion as a result of macromolecular crowding may be critical for the assembly and stability of the mature HIV-1 capsid. The very high chemical activity of CA in the maturing virion may promote association through the CA–CA interfaces identified, even though both retroviral CA domains have a low or negligible tendency to oligo- merize in solution. Similarly (and in addition to con- densation mediated by MA–plasma membrane and NC–RNA binding), macromolecular crowding in the cell may have a strong influence on the assembly of the immature HIV-1 capsid, by promoting CA–CA and other weak Gag–Gag interactions. Acknowledgements The author acknowledges J. L. Neira for collaboration and critical reading of the manuscript, and M. del A ´ lamo, R. Bocanegra and A. Rodrı ´ guez-Huete for excellent experimental work. Current work in the author¢s laboratory is supported by grants from FIPSE (36557 ⁄ 06), the Spanish Ministry of Science (BIO2006- 00793) and the Madrid Regional Government (S-0505⁄ MAT ⁄ 0303), and an institutional grant from Funda- cio ´ n Ramo ´ n Areces. The author is an associate Capsid protein interfaces in HIV-1 assembly M. G. Mateu 6106 FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS member of the Instituto de Biocomputacio ´ nyFı ´ sica de los Sistemas Complejos, Zaragoza, Spain. References 1 Adamson CS & Jones IM (2004) The molecular basis of HIV capsid assembly-five years of progress. Rev Med Virol 14, 107–121. 2 Ganser-Pornillos BK, Yeager M & Sundquist WI (2008) The structural biology of HIV assembly. Curr Opin Struct Biol 18, 203–217. 3 Neira JL (2009) The capsid protein of human immuno- deficiency virus: designing inhibitors of capsid assembly. FEBS J 276, doi:10.1111/j.1742-4658.2009.07314.x. 4 Gitti RK, Lee BM, Walker J, Summers MF, Yoo S & Sundquist WI (1996) Structure of the amino-terminal core domain of the HIV-1 capsid protein. Science 273, 231–235. 5 Momany C, Kovari LC, Prongay AJ, Keller W, Gitti RK, Lee BM, Gorbalenya AE, Tong L, McClure J, Ehrlich LS et al. (1996) Crystal structure of dimeric HIV-1 capsid protein. Nat Struct Biol 3, 763–770. 6 Gamble TR, Yoo S, Vajdos FF, von Schwedler UK, Worthylake DK, Wang H, McCutcheon JP, Sundquist WI & Hill CP (1997) Structure of the carboxy-terminal dimerization domain of the HIV-1 capsid protein. Science 278, 849–853. 7 Worthylake DK, Wang H, Yoo S, Sundquist WI & Hill CP (1999) Structures of the HIV-1 capsid protein dimerization domain at 2.9 resolution. Acta Crystallogr 55, 85–92. 8 Nermut MV, Hockley DJ, Jowett JB, Jones IM, Gar- reau M & Thomas D (1994) Fullerene-like organization of HIV gag-protein shell in virus-like particles produced by recombinant baculovirus. Virol 198, 288–296. 9 Campbell S & Vogt VM (1995) Self-assembly in vitro of purified CA-NC proteins from Rous sarcoma virus and human immunodeficiency virus type 1. J Virol 69, 6487–6497. 10 Campbell S & Rein A (1999) In vitro assembly proper- ties of human immunodeficiency virus type 1 Gag protein lacking the p6 domain. J Virol 73, 2270–2279. 11 Campbell S, Fisher RJ, Towler EM, Fox S, Issaq HJ, Wolfe T, Phillips LR & Rein A (2001) Modulation of HIV-like particle assembly in vitro by inositol phos- phates. Proc Natl Acad Sci USA 98, 10875–10879. 12 Mayo K, Huseby D, McDermott J, Arvidson B, Finlay L & Barklis E (2003) Retrovirus capsid protein assembly arrangements. J Mol Biol 325, 225–237. 13 Huseby D, Barklis RL, Alfadhli A & Barklis E (2005) Assembly of human immunodeficiency virus precursor gag proteins. J Biol Chem 280, 17664–17670. 14 Fuller SD, Wilk T, Gowen BE, Kra ¨ usslich H-G & Vogt VM (1997) Cryo-electron microscopy reveals ordered domains in the immature HIV-1 particle. Curr Biol 7, 729–738. 15 Wilk T, Gross I, Gowen BE, Rutten T, de Haas F, Welker R, Kra ¨ usslich H-G, Boulanger P & Fuller SD (2001) Organization of immature human immunodefi- ciency virus type 1. J Virol 75 , 759–771. 16 Briggs JAG, Simon MN, Gross I, Kra ¨ usslich H-G, Fuller SD, Vogt VM & Johnson MC (2004) The stoichiometry of Gag protein in HIV-1. Nat Struct Mol Biol 11, 672–675. 17 Wright ER, Schooler JB, Ding HJ, Kieffer C, Filmore C, Sundquist WI & Jensen GJ (2007) Electron cryoto- mography of immature HIV-1 virions reveals the structure of the CA and SP1 Gag shells. EMBO J 26, 2218–2226. 18 Briggs JAG, Riches JD, Glass B, Bartonova V, Zanetti G & Kra ¨ usslich H-G (2009) Structure and assembly of immature HIV. Proc Natl Acad Sci USA 106, 11090– 11095. 19 Borsetti A, O ¨ hagen A ˚ &Go ¨ ttlinger HG (1998) The C-terminal half of the human immunodeficiency virus type 1 Gag precursor is sufficient for efficient particle assembly. J Virol 72, 9313–9317. 20 Mammano F, O ¨ hagen A ˚ ,Ho ¨ glund S & Go ¨ ttlinger HG (1994) Role of the major homology region of human immunodeficiency virus type 1 in virion morphogenesis. J Virol 68, 4927–4936. 21 Dorfman T, Bukovsky A, O ¨ hagen A ˚ ,Ho ¨ glund S & Go ¨ ttlinger HG (1994) Functional domains of the capsid protein of human immunodeficiency virus type 1. J Virol 68, 8180–8187. 22 Reicin AS, O ¨ hagen A ˚ , Yin L, Ho ¨ glund S & Goff SP (1996) The role of Gag in human immunodeficiency virus type 1 virion morphogenesis and early steps of the viral life cycle. J Virol 70, 8645–8652. 23 von Schwedler UK, Stray KM, Garrus JE & Sundquist WI (2003) Functional surfaces of the human immuno- deficiency virus type 1 capsid protein. J Virol 77, 5439–5450. 24 Ako-Adjei D, Johnson MC & Vogt VM (2005) The ret- roviral capsid domain dictates virion size, morphology and coassembly of gag into virus-like particles. J Virol 79, 13463–13472. 25 Ivanov D, Tsodikov OV, Kasanov J, Ellenberger T, Wagner G & Collins T (2007) Domain-swapped dimerization of the HIV-1 capsid C-terminal domain. Proc Natl Acad Sci USA 104, 4353– 4358. 26 Carlson L-A, Briggs JAG, Glass B, Riches JD, Simon MN, Johnson MC, Mu ¨ ller B, Gru ¨ newald K & 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. M. G. Mateu Capsid protein interfaces in HIV-1 assembly FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS 6107 [...].. .Capsid protein interfaces in HIV-1 assembly M G Mateu 27 Ehrlich LS, Agresta BE & Carter CA (1992) Assembly of recombinant human immunodeficiency virus type 1 capsid protein in vitro J Virol 66, 4874–4883 28 Gross I, Hohenberg H & Krausslich H-G (1997) ¨ In vitro assembly properties of purified bacterially expressed capsid proteins of human immunodeficiency virus Eur J Biochem 249,... extensive thermodynamic characterization of the dimerization domain of the HIV-1 capsid protein Protein Sci 14, 2387–2404 Alcaraz LA, del Alamo M, Barrera FN, Mateu MG & Neira JL (2007) Flexibility in HIV-1 assembly subunits: solution structure of the monomeric C-terminal domain of the capsid protein Biophys J 93, 1264–1276 Wong CH, Shin R & Krishna NR (2008) Solution structure of a double mutant of the. .. immunodeficiency virus- 1 capsid protein J Mol Biol 318, 519–531 del Alamo M, Neira JL & Mateu MG (2003) Thermodynamic dissection of a low affinity protein- protein interface involved in human immunodeficiency virus assembly J Biol Chem 278, 27923–27929 Garzon MT, Lidon-Moya MC, Barrera FN, Prieto A, Gomez J, Mateu MG & Neira JL (2004) The dimerization domain of the HIV-1 capsid protein binds a capsid protein- derived... Sexton J, Sakalian M & Prevelige PE Jr (2002) Kinetic analysis of the role of intersubunit interactions in human immunodeficiency virus type 1 capsid protein assembly in vitro J Virol 76, 6900–6908 33 Lanman J, Lam TT, Barnes S, Sakalian M, Emmet MR, Marshall AG & Prevelige PE Jr (2003) Identification of novel interactions in HIV-1 capsid protein assembly by high-resolution mass spectrometry J Mol Biol 325,... homology region in mature retroviral capsid assembly J Virol 82, 5951– 5961 Lokhandwala PM, Nguyen TL, Bowzard JB & Craven RC (2008) Cooperative role of the MHR and the CA dimerization helix in the maturation of the functional retrovirus capsid Virol 376, 191–198 Forshey BM, von Schwedler U, Sundquist WI & Aiken C (2002) Formation of a human immunodeficiency virus type 1 core of optimal stability is crucial... of full-length HIV-1 CA: a model for the mature capsid lattice Cell 131, 70–79 40 Cardone G, Purdy JG, Cheng N, Craven RC & Steven AC (2009) Visualization of a missing link in retrovirus capsid assembly Nature 457, 694–699 41 Mateu MG (2002) Conformational stability of dimeric and monomeric forms of the C-terminal domain of 6108 42 43 44 45 46 47 48 49 50 51 52 53 54 human immunodeficiency virus- 1 capsid. .. effects at the dimerization interface of the HIV capsid protein J Mol Biol 345, 893–906 37 del Alamo M, Rivas G & Mateu MG (2005) Effect of macromolecular crowding agents on human immunodeficiency virus type 1 capsid protein assembly in vitro J Virol 79, 14271–14281 38 Briggs JAG, Simon MN, Gross I, Krausslich H-G & ¨ Fuller SD (2003) Structural organization of authentic, mature HIV-1 virions and cores... (2004) Assembly properties of the human immunodeficiency virus type 1 CA protein J Virol 78, 2545–2552 35 Douglas CC, Thomas D, Lanman J & Prevelige PE Jr (2004) Investigation of N-terminal domain charged residues on the assembly and stability of HIV-1 CA Biochemistry 43, 10435–10441 36 del Alamo M & Mateu MG (2005) Electrostatic repulsion, compensatory mutations, and long-range nonadditive effects at the. .. double mutant of the carboxy-terminal dimerization domain of the HIV-1 capsid protein Biochemistry 47, 2289–2297 Alcaraz LA, del Alamo M, Mateu MG & Neira JL (2008) Structural mobility of the monomeric C-terminal FEBS Journal 276 (2009) 6098–6109 ª 2009 The Author Journal compilation ª 2009 FEBS M G Mateu 55 56 57 58 domain of the HIV-1 capsid protein FEBS J 275, 3299–3311 Nguyen PA, Soto CS, Polishchuck... (2009) X-ray structures of the hexameric building block of the HIV capsid Cell 137, 1282–1292 Bowzard JB, Wills JW & Craven RC (2001) Second-site suppressors of Rous sarcoma virus CA mutations: evidence for interdomain interactions J Virol 75, 6850–6856 Purdy JG, Flanagan JM, Ropson IJ, Rennoll-Bankert KE & Craven RC (2008) Critical role of conserved hydrophobic residues within the major homology region . MINIREVIEW The capsid protein of human immunodeficiency virus: intersubunit interactions during virus assembly Mauricio G. Mateu Centro. involved either in the assembly of the immature HIV-1 capsid, or in the reassembly of CA into a mature capsid. The detailed description of the structure and