- Email: [email protected]
Model for the structure of the HIV gp41 ectodomain: insight into the intermolecular interactions of the gp41 loop
Biochimica et Biophysica Acta 1536 (2001) 116^122 www.bba-direct.com
Model for the structure of the HIV gp41 ectodomain: insight into the intermolecular interactions of the gp41 loop Michael Ca¡rey * Department of Biochemistry and Molecular Biology, University of Illinois at Chicago, Chicago, IL 60612, USA Received 26 December 2000; received in revised form 27 February 2001; accepted 8 March 2001
Abstract In human immunodeficiency virus (HIV) the viral envelope proteins gp41 and gp120 form a non-covalent complex, which is a potential target for AIDS therapies. In addition gp41 plays a possible role in HIV infection of B cells via the complement system. In an effort to better understand the molecular interactions of gp41, the structure of the HIV gp41 ectodomain has been modeled using the NMR restraints of the simian immunodeficiency virus (SIV) gp41 ectodomain (M. Caffrey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wingfield, A.M. Gronenborn, G.M. Clore, Solution structure of the 44 kDa ectodomain of SIV gp41, EMBO J. 17 (1998) 4572^4584). The resulting model presents the first structural information for the HIV gp41 loop, which has been implicated to play a direct role in binding to gp120 and C1q of the complement system. ß 2001 Elsevier Science B.V. All rights reserved. Keywords: AIDS; Envelope proteins; gp120; Complement
1. Introduction The human immunode¢ciency virus (HIV) envelope proteins gp41 and gp120 form a non-covalent complex on the viral surface and play critical roles in HIV infection (reviewed in [1,2]). Gp120 directs HIV to the appropriate target cell by binding to the host receptor proteins CD4 and a chemokine receptor. Gp41 mediates fusion of the HIV membrane with the target cell membrane, thereby allowing introduction of the viral genome into the target cell. The essential roles played by gp41 and gp120 make
Abbreviations: HAART, highly active anti-retroviral treatments; HIV, human immunode¢ciency virus; SIV, simian immunode¢ciency virus; TAD, torsion angle molecular dynamics * Fax: +1-312-413-0364; E-mail: ca¡[email protected]
them targets for AIDS therapies. Such therapies would be especially attractive because they are alternative and complementary to current therapies such as highly active anti-retroviral treatments (HAART) [3]. In light of recent evidence that HAART therapies are ine¡ective in 20^50% of treated patients [4], new therapies based on alternative targets are clearly of paramount importance. Structural information on the gp41-gp120 complex is important for the development of new therapies. Firstly, a full understanding of the antigenic properties of the gp41-gp120 complex is of interest for the development of future vaccines against HIV. Secondly, the gp41-gp120 complex is a potential target for novel drug therapies designed to disrupt the complex and hence the function. Unfortunately, there is no structural information available for the gp41-gp120 complex. Consequently, current knowledge is limited to the
0925-4439 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 4 4 3 9 ( 0 1 ) 0 0 0 4 2 - 4
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
structures of various domains of gp41 and gp120 in isolation. In the case of gp120, there is a recent crystal structure for large regions of the HIV gp120 in complex with regions of the CD4 receptor and a monoclonal antibody that inhibits gp120 binding to the chemokine receptor [5^7]. However, mutagenesis and deletion studies have suggested that the C1 and C5 domains of gp120, which are largely missing in the gp120 crystal structure but would be expected to be proximal to one another, are involved in the interaction with gp41 [8,9]. In the case of gp41, there have been numerous crystal structures of the simian immunode¢ciency virus (SIV) and HIV gp41 ectodomains, the extracellular region of gp41 that binds to gp120, in isolation [10^14]. Moreover, the solution structure and dynamic properties of the SIV gp41 ectodomain have been determined by NMR spectroscopy [15^18]. Importantly, the NMR studies were able to characterize the structure and dynamic properties of an approx. 30 residue conserved loop region, which is missing from the available crystal structures, but well de¢ned in the solution state [16]. Mutagenesis and structural studies have suggested that the gp41 loop region binds to gp120 [16,19]. More recently, evidence for a direct interaction between the gp41 loop region and the gp120 C1 and C5 regions has come from a mutagenesis study that demonstrated disul¢de formation between non-native cysteines placed in the gp41 loop and the gp120 C1 and C5 regions [20]. In addition, the gp41 loop region has been shown to interact with C1q of the complement system [21,22]. Interestingly, the gp41-C1q interaction may play an important role in the infection of B cells, which may serve as an alternative reservoir for HIV [23]. Taken together, the intermolecular interactions of the gp41 loop region make it highly desirable to obtain high resolution structural information for this region of gp41. In the present paper, a model for the structure of the HIV gp41 ectodomain is presented, which is based on the solution structure of the SIV gp41 ectodomain [15,16]. 2. Experimental The HIV gp41 ectodomain structure was calcu-
117
lated by the program CNS [24]. The protocol consisted of four steps: (1) high temperature torsion angle molecular dynamics (TAD) starting from an extended conformation; (2) slow cooling TAD; (3) slow cooling Cartesian dynamics; (4) conjugate gradient minimization. For the ¢rst step, the temperature was set to 50 000 K for 2000 steps of 15 fs. In this step, the force constants for NOE, dihedral anî 2, gles, and van der Waals forces were 150 kcal/mol/A î 4 , respectively. 100 kcal/mol/rad2 and 0.1 kcal/mol/A For the second step, the temperature was set to 50 000 K and cooled to 0 K over 1000 steps of 15 fs. In this step, the force constants for NOE, dihedral angles, and van der Waals forces were 150 î 2 , 200 kcal/mol/rad2 and 0.1^1.0 kcal/mol/ kcal/mol/A 4 î A , respectively. For the third step, the starting temperature was set to 2000 K and cooled to 0 K in 3000 steps of 5 fs. In this step, the force constants for NOE, dihedral angles, and van der Waals forces î 2 , 200 kcal/mol/rad2 and 1.0^ were 150 kcal/mol/A 4 î , respectively. In the ¢nal step, 200 4.0 kcal/mol/A steps of conjugate gradient minimization were performed. In this step, the force constants for NOE, dihedral angles, and van der Waals forces were 75 î 2 , 400 kcal/mol/rad2 and 1.0 kcal/mol/A î 4, kcal/mol/A respectively. In steps 2^4, an energy term for noncrystallographic symmetry (NCS) was included with î 2 . In all steps, an a force constant of 300 kcal/mol/A energy term for a conformational database was included with a force constant of 1.0. The P and i dihedral angles within the helical regions were set to the SIV values of 365³ þ 15³ and 345³ þ 15³, respectively. In the loop region, the P angle of conserved residues was set to the SIV gp41 values. The hydrogen bonds of the helical regions were included in the calculation as distance restraints. The long range NOEs were taken from the SIV data set between homologous hydrogens. NOEs between homologous hydrogens were de¢ned as: (1) those between hydrogens of conserved residues (e.g., leucinevaline); (2) those between HN , HK or HL and a hydrogen of a conserved residue; (3) those between HN , HK or HL of two di¡erent residues. The ¢nal data set included 105 P angles, 93 i angles, 93 hydrogen bonds, and 269 long range NOEs per monomer (approx. two per residue). Figures were generated by the molecular graphics program MOLMOL [25].
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
118
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Fig. 1. Alignment of the SIVsm and HIV-1 gp41 ectodomains. Numbering corresponds to that the of the HIV-1 gp41 ectodomain. Conserved residues are highlighted. The secondary structures are derived from the solution structure of the SIV gp41 ectodomain [16]. The NMR-based data used in the structure calculation of the HIV gp41 ectodomain are summarized as: +, P and i angles; 3, P angle; = , hydrogen bonds; X, long range NOEs between homologous hydrogens (cf. Section 2).
3. Results and discussion As noted above, X-ray crystallography studies have been highly successful in giving information about the helical regions of the gp41 ectodomain; however, in all cases to date the loop region was either removed to facilitate crystallization [10^13] or not visible in the crystal structure [14]. Thus it seems unlikely that the HIV gp41 loop region will be determined by crystallography studies. In contrast, the structure of the SIV gp41 ectodomain containing the loop region was successfully determined by NMR spectroscopy [15,16], which demonstrates the feasibility of determining the HIV gp41 ectodomain structure by NMR spectroscopy. However, the SIV gp41 structure was very challenging for NMR studies due to the large molecular mass of the complex (44 kDa) and the signi¢cant amount of spectral overlap present in a largely helical protein [15]. As a consequence, the NMR study on the SIV gp41 ectodomain required approx. 2 years of full time e¡ort and thus a NMR-based structure determination of the HIV gp41 ectodomain is relegated to some time in the future. As an alternative, the HIV gp41 ectodomain structure has been modeled using the NMR data available for the SIV gp41 ectodomain [15,16]. The rationale for using the SIV NMR data is that
HIV and SIV gp41 share functional, biochemical, and structural similarities, as well as a high degree of sequence identity [26]. The sequence identity can be broken down as follows: 60% in the N-terminal helix, 46% in the loop region, and 49% in the Cterminal helix (Fig. 1). For the present calculation, the structural information consisted of short range and long range information derived from NMR studies of the SIV gp41 ectodomain. As summarized in Fig. 1, three types of data were employed: P and i dihedral angles, hydrogen bonds, long range NOEs between homologous hydrogens (cf. Section 2 for details). An ensemble of ten low energy structures with î and no dihedral angle no NOE violations s 0.5 A violations s 5³ is shown in Fig. 2a. For this ensemble, the RMSD of the backbone and heavy atoms to î , respectively. Due the mean structure is 1.3 and 2.0 A to a smaller number of restraints, the RMSD of the HIV structure is larger than that of the SIV structure (the RMSD of the SIV gp41 ectodomain backbone î , respectively). and heavy atoms was 0.6 and 0.9 A Nonetheless, the ensemble RMSD of the HIV gp41 ectodomain structures suggests that it is a relatively well-determined solution structure. As shown in Fig. 2a, the RMSD of the HIV gp41 loop region is larger than that of the helical regions (the RMSD of the î and 1.2 A î, helix backbone and heavy atoms is 0.6 A
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
respectively). In the case of the SIV gp41 ectodomain, a larger RMSD was also observed for the loop [16]. Based on nuclear relaxation rates and hydrogen exchange properties of the SIV gp41 ectodomain, the loop region is clearly more mobile than the helical regions [15^17]. The increased mobility of the loop region has been suggested to be due to the absence of gp120 interaction domains, which could stabilize the loop structure [16]. As expected, the global fold of the HIV gp41 ectodomain is very similar to that of the SIV gp41 ectodomain. However, there are two di¡erences between the SIV and HIV gp41 ectodomain structures, as highlighted in Fig. 2b. Firstly, the present calculation introduced a disul¢de linked cysteine pair, which is highly conserved among the envelope proteins of retroviridae [26]. In contrast, the SIV gp41 ectodomain structure utilized a double mutant in which the two cysteines were substituted by alanine residues to alleviate possible non-speci¢c intermolecular cross-linking. The substitution of the cysteines clearly has a minimal e¡ect on the loop structure as evidenced by the 15 N-edited HSQC spectra of the wild-type and double mutant, which exhibit virtually
119
indistinguishable spectra [15]. It is important to note, however, that the loop region of the mutant lacking disul¢de bonds may well be less stable and more dynamic than that of the wild-type. In any case, the HIV loop structure presented here presumably represents a structure that is closer to that of the native form. Secondly, the HIV gp41 ectodomain contains a four residue insertion in the loop region (sequence = ASWS, Fig. 1), with respect to that of SIV gp41. In the present model, the four residue insert assumes a small turn conformation, in the absence of NOE or dihedral restraints. Interestingly, the C1 domain of gp120 contains a ¢ve residue insertion with respect to SIV gp120, which is consistent with the notion that the gp41 loop region binds to the gp120 C1 domain [26]. In summary, the model of the HIV gp41 ectodomain lends insight into the structure of the loop region, which has been implicated to be involved with intermolecular interactions, but is missing from all X-ray structures of HIV gp41. In Fig. 2c, the space-¢lling model of the HIV gp41 ectodomain is shown. In this presentation, the positively and negatively charged moieties are colored blue and red, respectively; the hydrophobic moieties
Fig. 2. Model for the HIV gp41 ectodomain. (a) An ensemble of ten low energy structures. The N-terminal helices are colored cyan, the loop region is colored green, and the C-terminal helices are colored red. (b) Ribbon diagram of the minimized mean structure with each presented as a unique color. The locations of the disul¢de bond and amino acid insertion are denoted. (c) Space ¢lling representation of the minimized mean structure with the acidic moieties colored red, basic moieties colored blue, and hydrophobic moieties colored green.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
120
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Fig. 3. (a) Ribbon diagram of the HIV gp41 loop region. gp41 residues that exhibit disul¢de bonds to gp120 C1 and C5 domains, when substituted by non-native cysteines [20] are presented as yellow spheres on one subunit. The size of the sphere represents the relative reactivity of the gp41 mutant. Numbering corresponds to that of the HIV gp41 ectodomain (Fig. 1). (b) Space ¢lling representation of the HIV gp41 loop region. gp41 residues implicated in binding to C1q are colored green.
are colored green. There are two striking properties of the HIV gp41 loop region, which are relevant to the proposed interactions with gp120 and C1q. First, the loop region exhibits an absence of charged residues and an abundance of exposed hydrophobic side chains, which are characteristics of protein-protein interaction sites [27,28]. The presence of hydrophobic residues on the surface of the gp41 loop is also observed in the solution structure of the SIV gp41 ectodomain, which was determined by NMR spectroscopy [16]. Moreover, the hydrophobicity of the loop region is consistent with the amino acid sequence of the loop region, for which there are only four charged residues within the 30 residue loop sequence (Fig. 1). Second, the space ¢lling representation of the HIV gp41 ectodomain clearly shows that the loop region forms a cleft, which could be complementary to the binding of globular domains of gp120 and C1q. The structure of the HIV gp41 loop region allows re-examination of previous studies implicating a prominent role in the gp41 interaction with gp120 and C1q of the complement system. With respect to the gp41-gp120 interaction, mutations at positions W57 and V69 of the HIV gp41 ectodomain, which are found in the loop region (Fig. 1), have been shown to disrupt gp120 binding [19]. More re-
cently, in an e¡ort to stabilize the gp41-gp120 interaction, a number of gp41 and gp120 residues were substituted with cysteines. The formation of non-native disul¢de bonds between gp41 and gp120 can then be inferred as evidence for a direct interaction between the respective regions of gp41 and gp120 [20]. As shown by Fig. 3a, the most reactive residues of gp41 are located in the loop region. Importantly, the disul¢de-linked complex exhibits antigenic properties that are similar to the gp41-gp120 complex found in virions [20]. In the case of the gp41-C1q interaction, synthetic peptides corresponding to the gp41 loop region have been shown to bind to C1q [21,22]. The gp41 residues implicated in binding to C1q are highlighted in Fig. 3b. Recently, HIV-1 has been shown to infect B cells via the complement system [23]. In light of the binding of gp41 to C1q of the complement system, the loop region of gp41 may be inferred to play an important role in the infection of B cells. Finally, we note that HIV-1 induced complement activation is stronger upon shedding of gp120 [29]. This observation is consistent with the notion that gp120 and C1q have overlapping binding sites within the gp41 loop region. Taken together, the present model is consistent with mounting evidence that the HIV gp41 loop re-
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
gion is directly involved in binding to gp120 and C1q of the complement system. It is important to note that the present model assumes that the structure of the gp41 loop region represents the gp41 conformation that binds to gp120 and C1q. It is presently being debated whether the helical regions of the gp41 ectodomain undergo large conformational changes during HIV infection (cf. [2,10,11,16]). However, there is no direct evidence that the loop region undergoes conformational changes during HIV infection. In any event, the critical role of gp41 in HIV infection suggests that further studies of gp41 intermolecular interactions are clearly warranted.
[10]
[11]
[12]
[13]
[14]
Acknowledgements The coordinates of the minimized mean structure of HIV gp41 ectodomain have been deposited in the Protein Data Bank (ID 1IF3). This work was supported by RO1 grant AI 47674.
[15]
[16]
References [1] A.D. Frankel, J.A. Young, HIV-1: ¢fteen proteins and an RNA, Annu. Rev. Biochem. 67 (1998) 1^25. [2] B.G. Turner, M.F. Summers, Structural biology of HIV, J. Mol. Biol. 285 (1999) 1^32. [3] M. Dietrich, J. Butts, R. Raasch, HIV-1 protease inhibitors: a review, Infect. Med. 16 (1999) 716^726. [4] W. Powderly, Current approaches to treatment for HIV-1 infection, J. Neurovirol. 6 (2000) S8^S13. [5] P.D. Kwong, R. Wyatt, T. Robinson, R.W. Sweet, J. Sodroski, W.A. Hendrickson, Structure of an HIV gp120 envelope glycoprotein in complex with the CD4 receptor and a neutralizing antibody, Nature 393 (1998) 648^659. [6] R. Wyatt, P. Kwong, W. Desjardins, R. Sweet, J. Robinson, W. Hendrickson, J. Sodroski, The antigenic structure of the HIV gp120 envelope glycoprotein, Nature 393 (1998) 705^ 711. [7] C. Rizzuto, R. Wyatt, N. Hernandez-Ramos, Y. Sun, P. Kwong, W. Hendrickson, J. Sodroski, A conserved HIV gp120 glycoprotein structure involved in chemokine receptor binding, Science 280 (1998) 1949^1953. [8] E. Helseth, U. Olshevsky, D. Gabuzda, B. Ardman, W. Haseltine, J. Sodroski, Changes in the transmembrane region of the human immunode¢ciency virus type-1 gp41 envelope glycoprotein a¡ect membrane fusion, J. Virol. 64 (1990) 6314^6318. [9] M. Ivey-Hoyle, R.K. Clark, M. Rosenberg, The NH2-terminal 31 amino acids of human immunode¢ciency virus type-1
[17]
[18]
[19]
[20]
[21]
[22]
121
envelope protein gp120 contain a potential gp41 contact site, J. Virol. 65 (1991) 2682^2685. D.C. Chan, D. Fass, J.M. Berger, P.S. Kim, Core structure of gp41 from the HIV envelope glycoprotein, Cell 89 (1997) 263^273. W. Weissenhorn, A. Dessen, S.C. Harrison, J.J. Skehel, D.C. Wiley, Atomic structure of the ectodomain from HIV-1 gp41, Nature 387 (1997) 426^430. K. Tan, J.-H. Liu, J.-H. Wang, S. Shen, M. Lu, Atomic structure of a thermostable subdomain of HIV-1 gp41, Proc. Natl. Acad. Sci. USA 94 (1997) 12203^12308. V.N. Malashkevich, D.C. Chan, C.T. Chutkowski, P.S. Kim, Crystal structure of the simian immunode¢ciency virus (SIV) gp41 core: conserved helical interactions underlie the broad inhibitory activity of gp41 peptides, Proc. Natl. Acad. Sci. USA 95 (1998) 9134^9139. Z. Yang, T. Mueser, J. Kaufman, S. Stahl, P. Wing¢eld, C. Hyde, The crystal structure of the SIV gp41 ectodomain at 1.47 angstrom resolution, J. Struct. Biol. 126 (1999) 131^ 144. M. Ca¡rey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, Determination of the secondary structure and global topology of the 44 kDa ectodomain of gp41 of the simian immunode¢ciency virus by multidimensional nuclear magnetic resonance spectroscopy, J. Mol. Biol. 271 (1997) 819^826. M. Ca¡rey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, Solution structure of the 44 kDa ectodomain of SIV gp41, EMBO J. 17 (1998) 4572^ 4584. M. Ca¡rey, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, 3D NMR experiments for measuring 15 N relaxation data of large proteins: application to the 44 kDa ectodomain of SIV gp41, J. Magn. Reson. 135 (1998) 368^372. M. Ca¡rey, J. Kaufman, S. Stahl, P. Wing¢eld, A. Gronenborn, G. Clore, Monomer-trimer equilibrium of the ectodomain of SIV gp41: insight into the mechanism of peptide inhibition of HIV infection, Protein Sci. 8 (1999) 1904^1907. J. Cao, L. Bergeron, E. Helseth, M. Thali, H. Repke, J. Sodroski, E¡ects of amino acid changes in the extracellular domain of the human immunode¢ciency virus type 1 gp41 envelope glycoprotein, J. Virol. 67 (1993) 2747^2755. J. Binley, R. Sanders, B. Clas, N. Schuelke, A. Master, Y. Guo, F. Kajumo, D. Anselma, P. Maddon, W. Olson, J. Moore, A recombinant human immunode¢ciency virus type 1 envelope glycoprotein complex stabilized by an intermolecular disul¢de bond between the gp120 and gp41 subunits is an antigenic mimic of the trimeric virion-associated structure, J. Virol. 74 (2000) 627^643. N. Thielens, I. Bally, C. Ebenbichler, M. Dierich, G. Arlaud, Further characterization of the interaction between the C1q subcomponent of human C1 and the transmembrane envelope glycoprotein gp41 of HIV-1, J. Immunol. 151 (1993) 6583^6592. I. Quinkal, J.-F. Hernandez, S. Chevallier, G. Arlaud, T.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
122
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Vernet, Mapping of the interaction between the immunodominant loop of the ectodomain of HIV-1 gp41 and human complement protein C1q, Eur. J. Biochem. 265 (1999) 656^ 663. [23] S. Moir, A. Malaspina, Y. Li, T.-W. Chun, T. Lowe, J. Adelberger, M. Baseler, L. Ehler, S. Liu, R. Davey, J. Mican, A. Fauci, B cells of HIV-1 infected patients bind virions through CD21-complement interactions and transmit infectious virus to activated T cells, J. Exp. Med. 192 (2000) 637^ 645. [24] A. Brunger, P. Adams, G. Clore, W. DeLano, P. Gros, R. Grosse-Kunstleve, J.-S. Jiang, J. Kuszewski, M. Nilges, N. Pannu, R. Read, L. Rice, T. Simonson, G. Warren, Crystallography and NMR system: a new software suite for macromolecular structure determination, Acta Crystallogr. Sect. D Biol. Crystallogr. 54 (1998) 905^921.
[25] R. Koradi, M. Billeter, K. Wuthrich, MOLMOL: a program for display and analysis of macromolecular structures, J. Mol. Graph. 14 (1996) 52^55. [26] N.W. Douglas, G.H. Munro, R.S. Daniels, HIV/SIV glycoproteins: structure-function relationships, J. Mol. Biol. 273 (1997) 122^149. [27] L. Young, R.C. Jernigan, D.G. Covell, A role for surface hydrophobicity in protein-protein recognition, Protein Sci. 3 (1994) 717^729. [28] S. Jones, J.M. Thornton, Principles of protein-protein interactions, Proc. Natl. Acad. Sci. USA 93 (1996) 13^20. [29] P. Marchang, K. Ute, C. Ochsenbauer, L. Gurtler, A. Hittmair, V. Bosch, J. Patsch, M. Dierich, Complement activation by HIV-1 infected cells: the role of transmembrane glycoprotein gp41, J. Acquired Immune De¢c. Syndr. Hum. Retroviol. 14 (1997) 102^109.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
Model for the structure of the HIV gp41 ectodomain: insight into the intermolecular interactions of the gp41 loop Michael Ca¡rey * Department of Biochemistry and Molecular Biology, University of Illinois at Chicago, Chicago, IL 60612, USA Received 26 December 2000; received in revised form 27 February 2001; accepted 8 March 2001
Abstract In human immunodeficiency virus (HIV) the viral envelope proteins gp41 and gp120 form a non-covalent complex, which is a potential target for AIDS therapies. In addition gp41 plays a possible role in HIV infection of B cells via the complement system. In an effort to better understand the molecular interactions of gp41, the structure of the HIV gp41 ectodomain has been modeled using the NMR restraints of the simian immunodeficiency virus (SIV) gp41 ectodomain (M. Caffrey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wingfield, A.M. Gronenborn, G.M. Clore, Solution structure of the 44 kDa ectodomain of SIV gp41, EMBO J. 17 (1998) 4572^4584). The resulting model presents the first structural information for the HIV gp41 loop, which has been implicated to play a direct role in binding to gp120 and C1q of the complement system. ß 2001 Elsevier Science B.V. All rights reserved. Keywords: AIDS; Envelope proteins; gp120; Complement
1. Introduction The human immunode¢ciency virus (HIV) envelope proteins gp41 and gp120 form a non-covalent complex on the viral surface and play critical roles in HIV infection (reviewed in [1,2]). Gp120 directs HIV to the appropriate target cell by binding to the host receptor proteins CD4 and a chemokine receptor. Gp41 mediates fusion of the HIV membrane with the target cell membrane, thereby allowing introduction of the viral genome into the target cell. The essential roles played by gp41 and gp120 make
Abbreviations: HAART, highly active anti-retroviral treatments; HIV, human immunode¢ciency virus; SIV, simian immunode¢ciency virus; TAD, torsion angle molecular dynamics * Fax: +1-312-413-0364; E-mail: ca¡[email protected]
them targets for AIDS therapies. Such therapies would be especially attractive because they are alternative and complementary to current therapies such as highly active anti-retroviral treatments (HAART) [3]. In light of recent evidence that HAART therapies are ine¡ective in 20^50% of treated patients [4], new therapies based on alternative targets are clearly of paramount importance. Structural information on the gp41-gp120 complex is important for the development of new therapies. Firstly, a full understanding of the antigenic properties of the gp41-gp120 complex is of interest for the development of future vaccines against HIV. Secondly, the gp41-gp120 complex is a potential target for novel drug therapies designed to disrupt the complex and hence the function. Unfortunately, there is no structural information available for the gp41-gp120 complex. Consequently, current knowledge is limited to the
0925-4439 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 4 4 3 9 ( 0 1 ) 0 0 0 4 2 - 4
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
structures of various domains of gp41 and gp120 in isolation. In the case of gp120, there is a recent crystal structure for large regions of the HIV gp120 in complex with regions of the CD4 receptor and a monoclonal antibody that inhibits gp120 binding to the chemokine receptor [5^7]. However, mutagenesis and deletion studies have suggested that the C1 and C5 domains of gp120, which are largely missing in the gp120 crystal structure but would be expected to be proximal to one another, are involved in the interaction with gp41 [8,9]. In the case of gp41, there have been numerous crystal structures of the simian immunode¢ciency virus (SIV) and HIV gp41 ectodomains, the extracellular region of gp41 that binds to gp120, in isolation [10^14]. Moreover, the solution structure and dynamic properties of the SIV gp41 ectodomain have been determined by NMR spectroscopy [15^18]. Importantly, the NMR studies were able to characterize the structure and dynamic properties of an approx. 30 residue conserved loop region, which is missing from the available crystal structures, but well de¢ned in the solution state [16]. Mutagenesis and structural studies have suggested that the gp41 loop region binds to gp120 [16,19]. More recently, evidence for a direct interaction between the gp41 loop region and the gp120 C1 and C5 regions has come from a mutagenesis study that demonstrated disul¢de formation between non-native cysteines placed in the gp41 loop and the gp120 C1 and C5 regions [20]. In addition, the gp41 loop region has been shown to interact with C1q of the complement system [21,22]. Interestingly, the gp41-C1q interaction may play an important role in the infection of B cells, which may serve as an alternative reservoir for HIV [23]. Taken together, the intermolecular interactions of the gp41 loop region make it highly desirable to obtain high resolution structural information for this region of gp41. In the present paper, a model for the structure of the HIV gp41 ectodomain is presented, which is based on the solution structure of the SIV gp41 ectodomain [15,16]. 2. Experimental The HIV gp41 ectodomain structure was calcu-
117
lated by the program CNS [24]. The protocol consisted of four steps: (1) high temperature torsion angle molecular dynamics (TAD) starting from an extended conformation; (2) slow cooling TAD; (3) slow cooling Cartesian dynamics; (4) conjugate gradient minimization. For the ¢rst step, the temperature was set to 50 000 K for 2000 steps of 15 fs. In this step, the force constants for NOE, dihedral anî 2, gles, and van der Waals forces were 150 kcal/mol/A î 4 , respectively. 100 kcal/mol/rad2 and 0.1 kcal/mol/A For the second step, the temperature was set to 50 000 K and cooled to 0 K over 1000 steps of 15 fs. In this step, the force constants for NOE, dihedral angles, and van der Waals forces were 150 î 2 , 200 kcal/mol/rad2 and 0.1^1.0 kcal/mol/ kcal/mol/A 4 î A , respectively. For the third step, the starting temperature was set to 2000 K and cooled to 0 K in 3000 steps of 5 fs. In this step, the force constants for NOE, dihedral angles, and van der Waals forces î 2 , 200 kcal/mol/rad2 and 1.0^ were 150 kcal/mol/A 4 î , respectively. In the ¢nal step, 200 4.0 kcal/mol/A steps of conjugate gradient minimization were performed. In this step, the force constants for NOE, dihedral angles, and van der Waals forces were 75 î 2 , 400 kcal/mol/rad2 and 1.0 kcal/mol/A î 4, kcal/mol/A respectively. In steps 2^4, an energy term for noncrystallographic symmetry (NCS) was included with î 2 . In all steps, an a force constant of 300 kcal/mol/A energy term for a conformational database was included with a force constant of 1.0. The P and i dihedral angles within the helical regions were set to the SIV values of 365³ þ 15³ and 345³ þ 15³, respectively. In the loop region, the P angle of conserved residues was set to the SIV gp41 values. The hydrogen bonds of the helical regions were included in the calculation as distance restraints. The long range NOEs were taken from the SIV data set between homologous hydrogens. NOEs between homologous hydrogens were de¢ned as: (1) those between hydrogens of conserved residues (e.g., leucinevaline); (2) those between HN , HK or HL and a hydrogen of a conserved residue; (3) those between HN , HK or HL of two di¡erent residues. The ¢nal data set included 105 P angles, 93 i angles, 93 hydrogen bonds, and 269 long range NOEs per monomer (approx. two per residue). Figures were generated by the molecular graphics program MOLMOL [25].
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
118
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Fig. 1. Alignment of the SIVsm and HIV-1 gp41 ectodomains. Numbering corresponds to that the of the HIV-1 gp41 ectodomain. Conserved residues are highlighted. The secondary structures are derived from the solution structure of the SIV gp41 ectodomain [16]. The NMR-based data used in the structure calculation of the HIV gp41 ectodomain are summarized as: +, P and i angles; 3, P angle; = , hydrogen bonds; X, long range NOEs between homologous hydrogens (cf. Section 2).
3. Results and discussion As noted above, X-ray crystallography studies have been highly successful in giving information about the helical regions of the gp41 ectodomain; however, in all cases to date the loop region was either removed to facilitate crystallization [10^13] or not visible in the crystal structure [14]. Thus it seems unlikely that the HIV gp41 loop region will be determined by crystallography studies. In contrast, the structure of the SIV gp41 ectodomain containing the loop region was successfully determined by NMR spectroscopy [15,16], which demonstrates the feasibility of determining the HIV gp41 ectodomain structure by NMR spectroscopy. However, the SIV gp41 structure was very challenging for NMR studies due to the large molecular mass of the complex (44 kDa) and the signi¢cant amount of spectral overlap present in a largely helical protein [15]. As a consequence, the NMR study on the SIV gp41 ectodomain required approx. 2 years of full time e¡ort and thus a NMR-based structure determination of the HIV gp41 ectodomain is relegated to some time in the future. As an alternative, the HIV gp41 ectodomain structure has been modeled using the NMR data available for the SIV gp41 ectodomain [15,16]. The rationale for using the SIV NMR data is that
HIV and SIV gp41 share functional, biochemical, and structural similarities, as well as a high degree of sequence identity [26]. The sequence identity can be broken down as follows: 60% in the N-terminal helix, 46% in the loop region, and 49% in the Cterminal helix (Fig. 1). For the present calculation, the structural information consisted of short range and long range information derived from NMR studies of the SIV gp41 ectodomain. As summarized in Fig. 1, three types of data were employed: P and i dihedral angles, hydrogen bonds, long range NOEs between homologous hydrogens (cf. Section 2 for details). An ensemble of ten low energy structures with î and no dihedral angle no NOE violations s 0.5 A violations s 5³ is shown in Fig. 2a. For this ensemble, the RMSD of the backbone and heavy atoms to î , respectively. Due the mean structure is 1.3 and 2.0 A to a smaller number of restraints, the RMSD of the HIV structure is larger than that of the SIV structure (the RMSD of the SIV gp41 ectodomain backbone î , respectively). and heavy atoms was 0.6 and 0.9 A Nonetheless, the ensemble RMSD of the HIV gp41 ectodomain structures suggests that it is a relatively well-determined solution structure. As shown in Fig. 2a, the RMSD of the HIV gp41 loop region is larger than that of the helical regions (the RMSD of the î and 1.2 A î, helix backbone and heavy atoms is 0.6 A
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
respectively). In the case of the SIV gp41 ectodomain, a larger RMSD was also observed for the loop [16]. Based on nuclear relaxation rates and hydrogen exchange properties of the SIV gp41 ectodomain, the loop region is clearly more mobile than the helical regions [15^17]. The increased mobility of the loop region has been suggested to be due to the absence of gp120 interaction domains, which could stabilize the loop structure [16]. As expected, the global fold of the HIV gp41 ectodomain is very similar to that of the SIV gp41 ectodomain. However, there are two di¡erences between the SIV and HIV gp41 ectodomain structures, as highlighted in Fig. 2b. Firstly, the present calculation introduced a disul¢de linked cysteine pair, which is highly conserved among the envelope proteins of retroviridae [26]. In contrast, the SIV gp41 ectodomain structure utilized a double mutant in which the two cysteines were substituted by alanine residues to alleviate possible non-speci¢c intermolecular cross-linking. The substitution of the cysteines clearly has a minimal e¡ect on the loop structure as evidenced by the 15 N-edited HSQC spectra of the wild-type and double mutant, which exhibit virtually
119
indistinguishable spectra [15]. It is important to note, however, that the loop region of the mutant lacking disul¢de bonds may well be less stable and more dynamic than that of the wild-type. In any case, the HIV loop structure presented here presumably represents a structure that is closer to that of the native form. Secondly, the HIV gp41 ectodomain contains a four residue insertion in the loop region (sequence = ASWS, Fig. 1), with respect to that of SIV gp41. In the present model, the four residue insert assumes a small turn conformation, in the absence of NOE or dihedral restraints. Interestingly, the C1 domain of gp120 contains a ¢ve residue insertion with respect to SIV gp120, which is consistent with the notion that the gp41 loop region binds to the gp120 C1 domain [26]. In summary, the model of the HIV gp41 ectodomain lends insight into the structure of the loop region, which has been implicated to be involved with intermolecular interactions, but is missing from all X-ray structures of HIV gp41. In Fig. 2c, the space-¢lling model of the HIV gp41 ectodomain is shown. In this presentation, the positively and negatively charged moieties are colored blue and red, respectively; the hydrophobic moieties
Fig. 2. Model for the HIV gp41 ectodomain. (a) An ensemble of ten low energy structures. The N-terminal helices are colored cyan, the loop region is colored green, and the C-terminal helices are colored red. (b) Ribbon diagram of the minimized mean structure with each presented as a unique color. The locations of the disul¢de bond and amino acid insertion are denoted. (c) Space ¢lling representation of the minimized mean structure with the acidic moieties colored red, basic moieties colored blue, and hydrophobic moieties colored green.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
120
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Fig. 3. (a) Ribbon diagram of the HIV gp41 loop region. gp41 residues that exhibit disul¢de bonds to gp120 C1 and C5 domains, when substituted by non-native cysteines [20] are presented as yellow spheres on one subunit. The size of the sphere represents the relative reactivity of the gp41 mutant. Numbering corresponds to that of the HIV gp41 ectodomain (Fig. 1). (b) Space ¢lling representation of the HIV gp41 loop region. gp41 residues implicated in binding to C1q are colored green.
are colored green. There are two striking properties of the HIV gp41 loop region, which are relevant to the proposed interactions with gp120 and C1q. First, the loop region exhibits an absence of charged residues and an abundance of exposed hydrophobic side chains, which are characteristics of protein-protein interaction sites [27,28]. The presence of hydrophobic residues on the surface of the gp41 loop is also observed in the solution structure of the SIV gp41 ectodomain, which was determined by NMR spectroscopy [16]. Moreover, the hydrophobicity of the loop region is consistent with the amino acid sequence of the loop region, for which there are only four charged residues within the 30 residue loop sequence (Fig. 1). Second, the space ¢lling representation of the HIV gp41 ectodomain clearly shows that the loop region forms a cleft, which could be complementary to the binding of globular domains of gp120 and C1q. The structure of the HIV gp41 loop region allows re-examination of previous studies implicating a prominent role in the gp41 interaction with gp120 and C1q of the complement system. With respect to the gp41-gp120 interaction, mutations at positions W57 and V69 of the HIV gp41 ectodomain, which are found in the loop region (Fig. 1), have been shown to disrupt gp120 binding [19]. More re-
cently, in an e¡ort to stabilize the gp41-gp120 interaction, a number of gp41 and gp120 residues were substituted with cysteines. The formation of non-native disul¢de bonds between gp41 and gp120 can then be inferred as evidence for a direct interaction between the respective regions of gp41 and gp120 [20]. As shown by Fig. 3a, the most reactive residues of gp41 are located in the loop region. Importantly, the disul¢de-linked complex exhibits antigenic properties that are similar to the gp41-gp120 complex found in virions [20]. In the case of the gp41-C1q interaction, synthetic peptides corresponding to the gp41 loop region have been shown to bind to C1q [21,22]. The gp41 residues implicated in binding to C1q are highlighted in Fig. 3b. Recently, HIV-1 has been shown to infect B cells via the complement system [23]. In light of the binding of gp41 to C1q of the complement system, the loop region of gp41 may be inferred to play an important role in the infection of B cells. Finally, we note that HIV-1 induced complement activation is stronger upon shedding of gp120 [29]. This observation is consistent with the notion that gp120 and C1q have overlapping binding sites within the gp41 loop region. Taken together, the present model is consistent with mounting evidence that the HIV gp41 loop re-
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
gion is directly involved in binding to gp120 and C1q of the complement system. It is important to note that the present model assumes that the structure of the gp41 loop region represents the gp41 conformation that binds to gp120 and C1q. It is presently being debated whether the helical regions of the gp41 ectodomain undergo large conformational changes during HIV infection (cf. [2,10,11,16]). However, there is no direct evidence that the loop region undergoes conformational changes during HIV infection. In any event, the critical role of gp41 in HIV infection suggests that further studies of gp41 intermolecular interactions are clearly warranted.
[10]
[11]
[12]
[13]
[14]
Acknowledgements The coordinates of the minimized mean structure of HIV gp41 ectodomain have been deposited in the Protein Data Bank (ID 1IF3). This work was supported by RO1 grant AI 47674.
[15]
[16]
References [1] A.D. Frankel, J.A. Young, HIV-1: ¢fteen proteins and an RNA, Annu. Rev. Biochem. 67 (1998) 1^25. [2] B.G. Turner, M.F. Summers, Structural biology of HIV, J. Mol. Biol. 285 (1999) 1^32. [3] M. Dietrich, J. Butts, R. Raasch, HIV-1 protease inhibitors: a review, Infect. Med. 16 (1999) 716^726. [4] W. Powderly, Current approaches to treatment for HIV-1 infection, J. Neurovirol. 6 (2000) S8^S13. [5] P.D. Kwong, R. Wyatt, T. Robinson, R.W. Sweet, J. Sodroski, W.A. Hendrickson, Structure of an HIV gp120 envelope glycoprotein in complex with the CD4 receptor and a neutralizing antibody, Nature 393 (1998) 648^659. [6] R. Wyatt, P. Kwong, W. Desjardins, R. Sweet, J. Robinson, W. Hendrickson, J. Sodroski, The antigenic structure of the HIV gp120 envelope glycoprotein, Nature 393 (1998) 705^ 711. [7] C. Rizzuto, R. Wyatt, N. Hernandez-Ramos, Y. Sun, P. Kwong, W. Hendrickson, J. Sodroski, A conserved HIV gp120 glycoprotein structure involved in chemokine receptor binding, Science 280 (1998) 1949^1953. [8] E. Helseth, U. Olshevsky, D. Gabuzda, B. Ardman, W. Haseltine, J. Sodroski, Changes in the transmembrane region of the human immunode¢ciency virus type-1 gp41 envelope glycoprotein a¡ect membrane fusion, J. Virol. 64 (1990) 6314^6318. [9] M. Ivey-Hoyle, R.K. Clark, M. Rosenberg, The NH2-terminal 31 amino acids of human immunode¢ciency virus type-1
[17]
[18]
[19]
[20]
[21]
[22]
121
envelope protein gp120 contain a potential gp41 contact site, J. Virol. 65 (1991) 2682^2685. D.C. Chan, D. Fass, J.M. Berger, P.S. Kim, Core structure of gp41 from the HIV envelope glycoprotein, Cell 89 (1997) 263^273. W. Weissenhorn, A. Dessen, S.C. Harrison, J.J. Skehel, D.C. Wiley, Atomic structure of the ectodomain from HIV-1 gp41, Nature 387 (1997) 426^430. K. Tan, J.-H. Liu, J.-H. Wang, S. Shen, M. Lu, Atomic structure of a thermostable subdomain of HIV-1 gp41, Proc. Natl. Acad. Sci. USA 94 (1997) 12203^12308. V.N. Malashkevich, D.C. Chan, C.T. Chutkowski, P.S. Kim, Crystal structure of the simian immunode¢ciency virus (SIV) gp41 core: conserved helical interactions underlie the broad inhibitory activity of gp41 peptides, Proc. Natl. Acad. Sci. USA 95 (1998) 9134^9139. Z. Yang, T. Mueser, J. Kaufman, S. Stahl, P. Wing¢eld, C. Hyde, The crystal structure of the SIV gp41 ectodomain at 1.47 angstrom resolution, J. Struct. Biol. 126 (1999) 131^ 144. M. Ca¡rey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, Determination of the secondary structure and global topology of the 44 kDa ectodomain of gp41 of the simian immunode¢ciency virus by multidimensional nuclear magnetic resonance spectroscopy, J. Mol. Biol. 271 (1997) 819^826. M. Ca¡rey, M. Cai, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, Solution structure of the 44 kDa ectodomain of SIV gp41, EMBO J. 17 (1998) 4572^ 4584. M. Ca¡rey, J. Kaufman, S.J. Stahl, P.T. Wing¢eld, A.M. Gronenborn, G.M. Clore, 3D NMR experiments for measuring 15 N relaxation data of large proteins: application to the 44 kDa ectodomain of SIV gp41, J. Magn. Reson. 135 (1998) 368^372. M. Ca¡rey, J. Kaufman, S. Stahl, P. Wing¢eld, A. Gronenborn, G. Clore, Monomer-trimer equilibrium of the ectodomain of SIV gp41: insight into the mechanism of peptide inhibition of HIV infection, Protein Sci. 8 (1999) 1904^1907. J. Cao, L. Bergeron, E. Helseth, M. Thali, H. Repke, J. Sodroski, E¡ects of amino acid changes in the extracellular domain of the human immunode¢ciency virus type 1 gp41 envelope glycoprotein, J. Virol. 67 (1993) 2747^2755. J. Binley, R. Sanders, B. Clas, N. Schuelke, A. Master, Y. Guo, F. Kajumo, D. Anselma, P. Maddon, W. Olson, J. Moore, A recombinant human immunode¢ciency virus type 1 envelope glycoprotein complex stabilized by an intermolecular disul¢de bond between the gp120 and gp41 subunits is an antigenic mimic of the trimeric virion-associated structure, J. Virol. 74 (2000) 627^643. N. Thielens, I. Bally, C. Ebenbichler, M. Dierich, G. Arlaud, Further characterization of the interaction between the C1q subcomponent of human C1 and the transmembrane envelope glycoprotein gp41 of HIV-1, J. Immunol. 151 (1993) 6583^6592. I. Quinkal, J.-F. Hernandez, S. Chevallier, G. Arlaud, T.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart
122
M. Ca¡rey / Biochimica et Biophysica Acta 1536 (2001) 116^122
Vernet, Mapping of the interaction between the immunodominant loop of the ectodomain of HIV-1 gp41 and human complement protein C1q, Eur. J. Biochem. 265 (1999) 656^ 663. [23] S. Moir, A. Malaspina, Y. Li, T.-W. Chun, T. Lowe, J. Adelberger, M. Baseler, L. Ehler, S. Liu, R. Davey, J. Mican, A. Fauci, B cells of HIV-1 infected patients bind virions through CD21-complement interactions and transmit infectious virus to activated T cells, J. Exp. Med. 192 (2000) 637^ 645. [24] A. Brunger, P. Adams, G. Clore, W. DeLano, P. Gros, R. Grosse-Kunstleve, J.-S. Jiang, J. Kuszewski, M. Nilges, N. Pannu, R. Read, L. Rice, T. Simonson, G. Warren, Crystallography and NMR system: a new software suite for macromolecular structure determination, Acta Crystallogr. Sect. D Biol. Crystallogr. 54 (1998) 905^921.
[25] R. Koradi, M. Billeter, K. Wuthrich, MOLMOL: a program for display and analysis of macromolecular structures, J. Mol. Graph. 14 (1996) 52^55. [26] N.W. Douglas, G.H. Munro, R.S. Daniels, HIV/SIV glycoproteins: structure-function relationships, J. Mol. Biol. 273 (1997) 122^149. [27] L. Young, R.C. Jernigan, D.G. Covell, A role for surface hydrophobicity in protein-protein recognition, Protein Sci. 3 (1994) 717^729. [28] S. Jones, J.M. Thornton, Principles of protein-protein interactions, Proc. Natl. Acad. Sci. USA 93 (1996) 13^20. [29] P. Marchang, K. Ute, C. Ochsenbauer, L. Gurtler, A. Hittmair, V. Bosch, J. Patsch, M. Dierich, Complement activation by HIV-1 infected cells: the role of transmembrane glycoprotein gp41, J. Acquired Immune De¢c. Syndr. Hum. Retroviol. 14 (1997) 102^109.
BBADIS 62030 30-5-01 Cyaan Magenta Geel Zwart