|Home | About | Journals | Submit | Contact Us | Français|
Subcellular distribution of calmodulin (CaM) in human immunodeficiency virus type-1 (HIV-1)-infected cells is distinct from that observed in uninfected cells. CaM co-localizes and interacts with the HIV-1 Gag protein in the cytosol of infected cells. Although it has been shown that binding of Gag to CaM is mediated by the matrix (MA) domain, the structural details of this interaction are not known. We have recently shown that binding of CaM to MA induces a conformational change that triggers myristate exposure, and that the CaM-binding domain of MA is confined to a region spanning residues 8–43 (MA-(8–43)). Here, we present the NMR structure of CaM bound to MA-(8–43). Our data revealed that MA-(8–43), which contains a novel CaM-binding motif, binds to CaM in an antiparallel mode with the N-terminal helix (α1) anchored to the CaM C-terminal lobe, and the C-terminal helix (α2) of MA-(8–43) bound to the N-terminal lobe of CaM. The CaM protein preserves a semiextended conformation. Binding of MA-(8–43) to CaM is mediated by numerous hydrophobic interactions and stabilized by favorable electrostatic contacts. Our structural data are consistent with the findings that CaM induces unfolding of the MA protein to have access to helices α1 and α2. It is noteworthy that several MA residues involved in CaM binding have been previously implicated in membrane binding, envelope incorporation, and particle production. The present findings may ultimately help in identification of the functional role of CaM in HIV-1 replication.
During the late phase of human immunodeficiency virus type-1 (HIV-1) infection, the newly synthesized Gag polyproteins are targeted to the plasma membrane (PM)3 for assembly and formation of immature particles (1,–4). During virus maturation, the Gag polyprotein is cleaved into several proteins including myristoylated matrix (MA), capsid, nucleocapsid, and short peptides (SP1, SP2, and P6) (1). One of the major roles of the MA domain of Gag during virus replication is targeting the Gag protein to specific sites on the PM for assembly. Gag association with the PM is critically dependent on the myristoyl group (myr) and a basic patch localized in the N terminus of MA (1, 5, 6). Proper targeting and localization of HIV-1 Gag on the PM is dependent on phosphatidylinositol (4,5)-bisphosphate (7,–9). Efficient binding of HIV-1 Gag and MA to membranes is sensitive to lipid composition and the hydrophobic environment of the bilayer (acyl chains and cholesterol)(10). Structural studies have shown that HIV-1 MA binds directly to phosphatidylinositol (4,5)-bisphosphate and other PM lipids such as phosphatidylserine, phosphatidylcholine, and phosphatidylethanolamine (11, 12). Gag binding to the PM appears to be complex and is likely mediated by a network of interactions between the MA domain and various membrane lipids and/or contents.
Despite the significant progress made in elucidating the molecular determinants of HIV-1 Gag assembly and virus release, the pathway(s) by which Gag is trafficked to the assembly sites in the infected cell and its intracellular interactions are poorly understood. Several cellular proteins have been implicated in Gag intracellular trafficking, virus assembly, and Gag-envelope co-localization. These include the human adaptor protein complexes 1, 2, and 3 (13,–15), tail interacting protein (16), Golgi-localized γ-ear containing Arf-binding protein (17, 18), ADP-ribosylation factor (17), the suppressor of cytokine signaling 1 (19, 20), Lck (a lymphoid specific Src kinase) (21), N-ethylmaleimide-sensitive factor attachment protein receptor (22), Filamin A (23), vacuolar protein sorting-associated protein 18 (24), Mon2 (24), and Lyric (25). The majority of these proteins have been shown to play roles in Gag intracellular trafficking and/or assembly.
We have been particularly interested in understanding the functional role of calmodulin (CaM) in HIV-1 replication. It has been previously shown that the subcellular concentration of CaM in HIV-infected cells is distinct from that in uninfected cells (26). HIV-1 Gag was found to co-localize with CaM in a diffuse pattern in the cytoplasm (27). However, the exact role of CaM in virus replication has yet to be established. CaM is a highly conserved calcium-binding protein expressed in all eukaryotic cells and is implicated in numerous cellular functions (28,–33). It can be localized in various subcellular locations, including the cytoplasm, within organelles, or associated with the PM (28,–33). CaM is also known to interact with many myristoylated and phosphoinositide-binding proteins to regulate their membrane localization (34,–47).
The N- and C-terminal lobes of CaM each possess two helix-loop-helix motifs called “EF-hands” (48,–50). Binding of calcium to EF-hands induces major structural rearrangements in the N- and C-terminal lobes resulting in the opening of large binding pockets on the surface of each domain consisting of hydrophobic residues that are essentially buried in the apoprotein (29, 32, 33). CaM has a “dumbbell-like” architecture with the N- and C-terminal lobes connected by a flexible central linker, thus adopting a semiextended conformation. The central linker can adopt many different conformations allowing the N- and C-terminal lobes to wrap around target sequences in many different orientations. CaM-binding motifs typically consist of a region of ~20 residues forming basic, often amphipathic helix. In many classical CaM-binding targets, hydrophobic residues involved in binding to CaM hydrophobic pockets, usually occupy conserved positions at 1-5-10 or 1-8-14 (30). Additional basic residues are responsible for stabilization of the complex via electrostatic interactions with CaM acidic residues. Although these patterns are found in many CaM-binding proteins, unclassified motifs have also been identified (30).
Attempts to identify specific roles of CaM in viral replication and infectivity are not limited to HIV-1. CaM possesses a functional role in budding of Ebola virus-like particles by interacting with the viral matrix protein VP40 (51). CaM also appears to play a role in simian immunodeficiency virus replication by interacting directly with the MA domain of Gag (27, 34, 37, 52,–55). In vivo and in vitro studies revealed that CaM interacts with additional HIV-1 proteins like Nef, Tat, and gp160 (27, 34, 37, 52, 53, 56). Small-angle x-ray scattering studies have provided a global picture of CaM bound to full-length MA (57) or to short peptides derived from the MA protein (58). Our laboratory (59) and others (57, 60) have shown that CaM binds directly to the MA protein in a calcium-dependent manner. We have shown that binding of CaM induces a conformational change in MA, triggering myr exposure (59). By employing NMR, biophysical, biochemical, and mass spectrometry methods we have identified the CaM-binding region to be a region spanning residues 8–43 (MA-(8–43)) (61). Within MA, residues 11–19 form an α-helix (α1), residues 20–30 form a β-hairpin, and residues 31–45 form a second α-helix (α2) (11, 62,–64). Here, we present the NMR structure of CaM in complex with MA-(8–43). Our data reveal that MA-(8–43) binds to CaM in an antiparallel mode with α1 anchored to the C-terminal lobe of CaM, whereas α2 of MA-(8–43) is docked on the N-terminal lobe. The interaction is mediated by numerous hydrophobic and electrostatic contacts. These findings may help in identification of the precise functional role of CaM in HIV-1 replication.
CaM and MA-(8–43) samples have been prepared as described (61). The CaM·MA-(8–43) complex was prepared by mixing equimolar amounts of CaM and MA-(8–43), which was then passed through a gel filtration column (Superdex 75, GE Healthcare). Fractions of the complex were pooled and concentrated as desired. All samples were stored in a buffer containing 50 mm Tris-d11 (pH 7), 100 mm NaCl, and 5 mm CaCl2.
Isotopically unlabeled and uniformly 13C-, 15N-, or 13C-/15N-labeled protein samples were prepared at ~400–500 μm concentrations. NMR data were collected at 35 °C on a Bruker Avance II (700 MHz 1H) spectrometer equipped with a cryogenic triple-resonance probe, processed with NMRPipe (65), and analyzed with NMRVIEW (66) or CCPN analysis (67). The backbone and side chain atom resonances of the CaM·MA-(8–43) complex were assigned using HNCA, HN(CO)CA, HNCACB, HN(CO)CACB, 15N-edited NOESY- and TOCSY-HSQC, hCCH-TOCSY, and HcCH-TOCSY experiments. Assignments of aromatic signals were confirmed by the (H)CB(CGCC-TOCSY)Har experiment (68). Intramolecular NOE contacts were obtained from three-dimensional 15N-edited NOESY-HSQC, four-dimensional 13C-/15N-edited HMQC-NOESY-HSQC, and four-dimensional 13C-/13C-edited HMQC-NOESY-HMQC (120-ms mixing time). Intermolecular NOEs were detected by three-dimensional 13C-edited/half-filtered NOESY experiments (120-ms mixing time). Assignments of the majority of intermolecular NOEs were cross-validated in the 13C-edited/half-filtered NOESY spectra collected on [12C]CaM·[13C]MA-(8–43) and [13C]CaM·[12C]MA-(8–43) samples. Combined 1H-15N chemical shift differences (Δδ) were calculated as ΔδHN = ((ΔδH)2 + (ΔδN/5)2)0.5 and similarly for 1H-13C as ΔδHC = ((ΔδH)2 + (ΔδC)2)0.5, where ΔδX is a chemical shift change of nucleus X. 1H, 13C, and 15N chemical shifts were referenced to 2,2-dimethyl-2-silapentanesulfonic acid (69).
Structure of the CaM·MA-(8–43) complex was calculated with CYANA starting from random initial angles. Distance restraints with upper distance limits of 2.7, 3.3, and 5.0 Å were determined based on the intensities of NOE cross-peaks. Standard pseudo atom corrections were applied to groups of degenerate hydrogen atoms during calculations. and ψ dihedral angle restraints were generated in TALOS+ (70) and used with an uncertainty of ±2 × standard deviation or ±20 degrees, whichever was larger, only for residues with a “good” score. For MA-(8–43), hydrogen bond restraints with O-H and O-N distances of 1.7–2.0 and 2.0–3.0 Å, respectively, were used in α-helical regions (residues 15–19 and 31–39) identified based on the chemical shift index (71). For CaM, hydrogen bond restraints were defined for the β-sheets between residues 27, 63, and residues 100, 136, as well as for several N-capping residues identified from characteristic chemical shifts (72). Positions of calcium atoms were restricted within 2.4 ± 0.1 Å from known Ca2+-binding carboxyl and carbonyl oxygens (73). Low-weight torsion angle restraints were used for a general treatment of side chain rotamers during the initial stages of simulated annealing. A standard protocol for simulated annealing and molecular dynamics minimization in torsion angle space was applied. In the final calculation cycle, 100 random input structures were minimized and 20 structures with the best agreement with the final restraint set were selected. Analysis and visualization of structures were performed using PROCHECK-NMR (74), VMD (75), and PyMOL (The PyMOL Molecular Graphics System, version 18.104.22.168 Schrödinger, LLC.). Electrostatic potential maps were generated using PDB2PQR and APBS software (76, 77).
We have previously shown that the majority of 1H and 15N resonances in the NMR spectrum of MA exhibit significant chemical shift changes upon binding to CaM (59). Extensive loss and/or broadening of the NMR signals precluded determination of the solution structure of the complex. In a subsequent study, we have identified the minimal CaM-binding domain of MA by utilizing a proteolytic digestion assay (61). Analysis of the digestion products by mass spectrometry revealed that the abundant MA species resistant to proteolysis is a peptide spanning residues 8–43 (61). The formation of the complex between MA-(8–43) and CaM has led to substantial chemical shift changes for the vast majority of 1H and 15N signals in the HSQC spectrum of CaM (Fig. 1). These changes are very similar to those observed in the HSQC spectrum obtained for CaM when bound to the full-length MA protein (Fig. 2, A and B), indicating that MA-(8–43) binds to CaM in a manner similar to that observed for the full-length MA protein (61).
Signal assignments were obtained for the vast majority of atoms. No backbone amide signals were detected in the 1H-15N HSQC spectra for Lys30–Lys32 of MA-(8–43) and Lys75–Met76 of CaM due to signal overlaps or possibly due to a conformational and/or chemical exchange on an intermediate NMR scale. Similar broadening of signals for residues within the central linker of CaM has been observed and attributed to the flexible nature of this region (78). The most significant chemical shift changes of backbone amide and side chain methyl signals of CaM have been observed for methionine residues 36, 51, 71, 72, 109, 124, 144, and 145, which are crucial for target binding (30, 33, 79, 80), and for many other residues in the hydrophobic pockets. The largest effects are observed for signals corresponding to CaM residues Ile52–Val55 and Met71–Arg74 located in the N terminus, and residues Met124 and Met144 in the C terminus (Fig. 2, B and C). Despite the observation of substantial 1H, 13C, and 15N chemical shift changes in the NMR spectra of CaM, the intramolecular NOE patterns remain unchanged upon peptide binding, indicating that the structure of CaM is not greatly altered upon binding of MA-(8–43).
Numerous intermolecular NOEs between CaM and MA-(8–43) residues were detected in the 13C-edited/half-filtered NOESY spectra. As shown in Fig. 3A, the 13C-edited/half-filtered NOESY spectrum obtained on [13C]CaM·[12C]MA-(8–43) shows NOEs between Trp16 of MA-(8–43) and methyl groups of several hydrophobic residues in the C-terminal lobe of CaM such as Ile100, Met124, and Met144. Unambiguous NOEs between hydrophobic residues in the C-terminal helix (α2) of MA-(8–43) and hydrophobic residues in the N-terminal lobe of CaM were also detected in a spectrum obtained for the reciprocally labeled [12C]CaM·[13C]MA-(8–43) sample (Fig. 3B). MA-(8–43) residues Leu21–Tyr29 and CaM residues Met76–Asp80 lack medium-range intramolecular NOE contacts indicative of a regular secondary structure. Backbone amide signals of residues Gly24M–Gln28M and Thr79C–Asp80C (M and C indicate MA-(8–43) and CaM, respectively) exhibit moderately strong NOEs with water resonance, suggesting increased solvent accessibility and lack of secondary structure, consistent with the dynamic nature of these regions. No NOE contacts have been observed between the N- and C-terminal helices of MA-(8–43), indicating the absence of intramolecular contacts between the two domains. Furthermore, no NOEs have been observed between the N- and C-terminal domains of CaM.
A summary of the NOE distance restraints is shown in Table 1. A total of 187 intermolecular NOEs detected between MA-(8–43) and CaM residues were used in the structure calculations. An ensemble of the lowest-energy 20 structures calculated for the complex shows a good convergence and correspondingly low positional root mean square deviation values within the structured regions (Fig. 4 and Table 1). Previous NMR and circular dichroism data have shown that the free MA-(8–43) peptide exists in a random coil conformation, but adopts an α-helical conformation when bound to CaM (61). Consistent with these observations, the CaM·MA-(8–43) structure shows that residues Asp14–Arg20 and Leu31–Glu40 of MA-(8–43) are α-helical, whereas residues Leu21–Gln28 connecting the two MA α-helices, residues Leu8–Leu13 and Leu41–Arg43, lack a regular secondary structure (Fig. 5A). The linker regions were not restrained during structure calculations and the two domains in the resulting structures adopt pseudo-random orientations with respect to each other, limited only by steric factors.
The CaM protein interacts with MA-(8–43) in an anti-parallel mode (Fig. 5). Surface representation of the CaM·MA-(8–43) complex shows a large interaction interface (Fig. 5B). The CaM protein preserves its semiextended conformation when bound to MA-(8–43). Furthermore, an overall electrostatic compatibility between highly positively charged MA-(8–43) and negatively charged CaM is clearly observed (Fig. 5C). The hydrophobic faces of α1 and α2 helices of MA insert into the C- and N-terminal hydrophobic grooves of CaM, respectively (Fig. 6). The major stabilizing factor of the interaction between α1 of MA-(8–43) and CaM is the deep insertion of Trp16 and Ile19 side chains into the C-terminal pocket of CaM (Fig. 6A). As indicated by strong NOE cross-peaks (Fig. 3A), the indole ring of Trp16 is in a close proximity to CaM residues Ile100, Met124, Ile125, Ala128, Val136, and Met144 (Fig. 6, A and B). Additionally, the side chain of Ile19M interacts with CaM residues Phe92, Val105, Met109, Leu112, Met124, Phe141, and Met145. Weak intermolecular NOEs were also observed between Leu13M and Leu21M and hydrophobic CaM residues; Leu13M interacts with Met109C and Met124C, whereas Leu21M interacts with Ala88C and Met144C. Hydrophobic interactions between α2 of MA-(8–43) and CaM N terminus involve MA-(8–43) residues Tyr29, Leu31, Ile34, Val35, Ala37, and Leu41, and CaM residues Leu18, Phe19, Val35, Leu39, Met36, Met51, Val55, Phe68, Met71, and Met72 (Figs. 3B and and66C). Surprisingly, the aromatic side chain of Trp36, positioned in the middle of MA helix α2, is mostly solvent-exposed and does not participate in the interactions (Fig. 6, C and D).
In addition to favorable hydrophobic interactions, the complex is likely stabilized by several salt bridges as indicated by strong NOEs between the side chains of Lys CϵH and Glu CγH groups. As indicated by the NOE cross-peaks, Lys15M is in close proximity to Glu120C, whereas Lys18M is close to Glu114C (Fig. 6, A and B). An analogous structural arrangement is observed at the beginning of MA-(8–43) α2 helix with the Lys30M side chain in close proximity to Glu47C, similar to the Lys15M/Glu120C pair (Fig. 6C). Salt bridges can also form within the central part of the complex as Lys26 and Lys27, located in the linker between α1 and α2 of MA-(8–43), are favorably close to the highly acidic region in the central linker of CaM domains (Asp78–Glu87). However, unambiguous assignments of NOEs between the respective side chains have been precluded due to signal overlaps. Taken together, there is a pronounced complementarity in terms of the sizes of MA-(8–43) and CaM interacting domains. The smaller helix α1 of MA binds the relatively small C-terminal domain of CaM, whereas helix α2 of MA, which is about twice as long as α1, is fully engaged in binding the larger N-terminal CaM domain. Moreover, the length of the linker between the two MA helices is sufficient to allow full contact of the interaction interfaces while retaining a flexibility of CaM and MA central linker regions.
We have previously shown that CaM interacts directly with the MA protein, inducing a conformational change that triggers myr exposure (59). We have identified the minimal CaM-binding domain of MA to a region spanning residues 8–43 (MA-(8–43)) (61). The 36-residue region of MA is the largest CaM-binding motif discovered to date because the CaM-binding domain of target proteins is typically ~15–20 amino acids long (30). The positioning of the hydrophobic residues involved in binding to CaM indicates that MA-(8–43) contains a novel CaM-binding motif, which does not belong to any of the previously characterized motifs.
The presence of two highly adaptable hydrophobic surfaces on the N- and C-terminal domains of CaM is a prominent feature that, together with the flexibility of the protein central region, allows it to bind to numerous targets (33). Even though the CaM-interacting proteins are characteristic by a high variability of amino acid sequences within binding regions, several motifs have been identified based on the relative positions of anchoring hydrophobic residues. Accordingly, MA residues Trp16 and Ile19 constitute a 1-4 binding motif, whereas Leu31, Ile34, and Val35 make a 1-4-5 motif. To the best of our knowledge, such motifs have not been previously identified. A tryptophan residue is, however, found as the first anchoring residue in the majority of peptides binding to the C-terminal lobe of CaM.
Besides target protein sequences, CaM-protein complexes exhibit high variability also in terms of overall structures. CaM exclusively binds to α-helical motifs but several distinct binding modes have been identified differing in the involvement and relative positions of CaM binding lobes. Depending on the sequence and length of the target protein, CaM interacts with the protein via either one or two lobes simultaneously. When both CaM lobes are involved, CaM typically adopts a compact structure by wrapping around an α-helix of the target protein (29, 33). The binding mode represented in the CaM·MA-(8–43) complex, in which one molecule interacts with CaM via two well separated helical motifs, is distinct from the vast majority of reported structures. Thus, MA-(8–43) represents a rare target sequence that is substantially different (in length and composition) from almost all known CaM-binding domains, which confirms the enormous versatility of CaM·target complex formation. Among the ~40 structures of CaM·protein/peptide complexes deposited in the Protein Data Bank, only one structure possesses features highly similar to those observed in the CaM·MA-(8–43) structure. The bipartite binding motif has been identified for CaM bound to Munc13-1, a regulator of synaptic vesicle priming (81). Munc13-1 binds the C-terminal lobe of CaM via its long N-terminal helix that is connected by a long flexible linker to a very short α-helix, which binds the N-terminal lobe of CaM (81). In another structure of the CaM complex with a peptide derived from vacuolar calcium-ATPase BCA1 protein, two CaM-interacting α-helices in the BCA1 peptide are connected by a very short and rather rigid linker (82); the central part of the CaM·BCA1 complex lacks the flexibility observed in the structures of CaM bound to Munc13-1 or MA-(8–43). Taken together, our structural findings show a rare CaM interacting motif and, together with Munc13-1, suggest a new class of CaM complexes adopting a modular architecture.
Previous studies utilizing small-angle neutron scattering methods have suggested that CaM adopts an extended open-clamp conformation when bound to MA (60). Tryptophan fluorescence data additionally suggested that interactions are possibly mediated by insertion of Trp16 and Trp36 in hydrophobic pockets of CaM (57). Although these findings agree with most of our previous (59, 61) and current data, the structure of CaM·MA-(8–43) clearly shows that Trp36 is exposed to solvent and is not involved in any interaction with CaM; the orientation of the helix positions the indole group of Trp36 away from the binding pocket (Fig. 6). Typically, changes observed in tryptophan fluorescence signal are induced when the indole ring is involved in binding or upon occurrence of conformational changes in the protein, which involve a change in the environment of the tryptophan residue (e.g. protein unfolding) (83). Indeed, CaM-induced MA unfolding leads to exposure of both Trp16 and Trp36 rings and is likely to contribute to the observed changes in the fluorescence signal.
An alternative interpretation of the fluorescence data are that Trp36 is involved in the binding of full-length MA but not MA-(8–43). Besides the structural evidence, this possibility is also ruled out based on the striking similarity of chemical shift changes observed for CaM residues upon binding to full-length MA and MA-(8–43). As shown in Fig. 2, chemical shift changes are almost identical for CaM residues located in the hydrophobic pocket (residues surrounding Phe19, Val35, Val55, and Met72) in proximity to the MA Trp36 ring. If the indole group of Trp36 were inserted in the pocket, the ring current effect would have caused significantly different chemical shift changes in some of these CaM residues. Thus, we conclude that Trp36 is not involved in the binding of CaM to MA and that the CaM·MA-(8–43) structure is a true representative model of the CaM·MA complex.
Unfolding of the MA protein appears to be one of the main requirements for the formation of CaM·MA complex (57, 59,–61). The vast majority of the structural studies on CaM·protein complexes were conducted with short peptides that constitute the minimal binding domain of target proteins. It is not clear how these helical peptides that are often part of the overall fold of the protein become accessible to CaM. CaM-induced unfolding of protein targets has been shown to be important for the biological function of target proteins like in enzymatically driven cleavage reactions (84). Although the consequences of MA unfolding upon binding to CaM are not known, it is reasonable to hypothesize that the CaM-Gag interaction is regulated intracellularly and is dependent on calcium signaling events during virus replication. Among the multiple functions of CaM is regulation of activity of myristoylated and phosphoinositide-binding proteins to regulate their membrane localization (34,–47).
The N-terminal region of MA is critical for diverse Gag functions including Gag-membrane interactions, regulation of the myr switch mechanism, and binding to several cellular constituents implicated in Gag trafficking and/or gp160 incorporation (8, 9, 11, 16, 62, 85,–88). Several residues in helix α1 of MA (Leu13, Trp16, Glu17, and Lys18) are important for gp160 incorporation (89,–91), which led to the suggestion that the MA domain of Gag interacts directly with the gp160 protein via helix α1 (89, 92). Previous studies have shown that CaM concentration in cells infected with HIV-1 is markedly increased upon expression of the HIV-1 gp160 protein (26). The CaM-binding region was identified as a helical peptide in the gp41 protein and deletion of this region led to diminished virus infectivity (26, 52). Thus, the interplay between gp160, CaM, and Gag could be an important and underestimated event in the virus replication cycle. We believe that characterization of the interactions between MA and CaM will facilitate future studies and help to identify the exact role of CaM in HIV-1 replication.
*This work was supported, in whole or in part, by National Institutes of Health Grant 1R01AI087101 (to J. S. S.) and intramural funding from the University of Albama at Birmingham Center for AIDS Research.
Chemical shift assignments of the CaM·MA-(8–43) complex have been deposited in the Biological Magnetic Resonance Data Bank and assigned accession number 19604.
3The abbreviations used are: