Conservation of human immunodeficiency virus type 1 gp120 inner-domain sequences in lentivirus and type A and B retrovirus envelope surface glycoproteins

We recently described a sequence similarity between the small ruminant lentivirus surface unit glycoprotein (SU) gp135 and the second conserved region (C2) of the primate lentivirus gp120 which indicates a structural similarity between gp135 and the inner proximal domain of the human immunodeficiency virus type 1 gp120 (I. Hötzel and W. P. Cheevers, Virus Res. 69:47–54, 2000). Here we found that the seven-amino-acid sequence of the gp120 strand β25 in the C5 region, which is also part of the inner proximal domain, was conserved in the SU of all lentiviruses in similar or identical positions relative to the carboxy terminus of SU. Sequences conforming to the gp135-gp120 consensus for β-strand 5 in the C2 region, which is antiparallel to β25, were then sought in the SU of other lentiviruses and retroviruses. Except for the feline immunodeficiency virus, sequences similar to the gp120-gp135 consensus for β5 and part of the preceding strand β4 were present in the SU of all lentiviruses. This motif was highly conserved among strains of each lentivirus and included a strictly conserved cysteine residue in β4. In addition, the β4/β5 consensus motif was also present in the conserved carboxy-terminal region of all type A and B retroviral envelope surface glycoproteins analyzed. Thus, the antiparallel β-strands 5 and 25 of gp120 form an SU surface highly conserved among the lentiviruses and at least partially conserved in the type A and B retroviral envelope glycoproteins.Lentiviruses are a group of strictly exogenous retroviruses that infect a range of mammalian hosts. One characteristic of this group of retroviruses is the rapid sequence divergence observed between virus strains as well as different lentiviruses, which resulted in the evolution of viruses with large differences in genome organization and sequence (20). Most of the sequence homology between highly divergent lentiviruses is present in the gag and pol gene products (8, 21). Sequence homology between the envelope glycoproteins of different lentiviruses has previously been shown to occur only in the ectodomain of the transmembrane subunit (TM) but not in the surface unit (SU) glycoprotein (3, 8, 21–23). Due to this apparent lack of sequence conservation in lentiviral SU, it has been unclear how the SU of different lentiviruses are structurally related to each other. To address this question, we recently compared SU sequences from the gp120 from primate lentiviruses and the gp135 of small ruminant lentiviruses and found a statistically significant sequence similarity between the second conserved region (C2) of gp120 and a 99-amino-acid region from gp135 (10). Analysis of this gp120-gp135 sequence similarity in the context of the gp120 structure revealed a partial structural similarity between gp120 and gp135.The human immunodeficiency virus type 1 (HIV-1) gp120 core bound to CD4 is composed of two major domains, the inner and outer domains, and a minidomain composed of four antiparallel β-strands, the bridging sheet (13). Sequences from the C2 region form most of the β-strands of a two-helix, two-strand bundle and a five-stranded β-sandwich in the inner domain as well as some β-strands of the outer domain of gp120 (13). Most of the similarity motifs between gp135 and the C2 region of gp120 coincide with sequences corresponding to β-strands 4 through 8 in the HIV-1 gp120 inner domain and β-strands 11 and 12 in the outer domain (10). Significantly, all four cysteines that form two disulfide bonds in the proximal region of the gp120 inner domain as well as the first cysteine of the gp120 V3 loop in β12 (13, 15) are conserved in gp135, indicating a partial similarity between the tertiary structures of gp120 and gp135 (10).The most conserved sequences between gp120 and gp135 correspond to strands β4 and β5 in the five-stranded β-sandwich structure of the proximal region of the inner proximal domain of HIV-1 gp120. Two additional β-strands in this five-stranded β-sandwich are derived from C1 and C5 sequences of HIV-1 gp120 (13). We hypothesized that C1 and C5 sequences, which are part of a structurally conserved SU inner proximal domain, should also be conserved between gp120 and gp135 and possibly in the SU of other lentiviruses. Here we show that two short motifs located in the gp120 C2 and C5 regions which are part of an antiparallel β-sheet in the gp120 inner proximal domain are conserved in the lentiviruses, indicating that a surface of the inner domain of HIV-1 gp120 is conserved in the SU of other lentiviruses. In addition, the C2 motif is also present in the envelope glycoproteins encoded by A-type endogenous retroviral elements and type B retroviruses (type A and B retroviruses), suggesting a local structural similarity between the SU of lentiviruses and type A and B retroviruses.

Sequence motif of the C5 region of HIV-1 gp120 is present in the SU of all lentiviruses.

As the sequences of three of the five β-strands of the gp120 inner proximal domain β-sandwich are conserved in gp135, we first tried to determine whether the gp120-gp135 sequence similarity extends to the other two β-strands which are part of this structure. One of these strands is β1, located in the C1 region of gp120 (13). Although the sequence of β1 is relatively well conserved among the primate lentiviruses, it is only 3 amino acids long, and a reliable assignation of similar sequences in gp135 could not be done. The other strand of this β-sandwich structure is the 7-amino-acid long β25. This strand is antiparallel to β5, which is the most conserved sequence between gp120 and gp135 (10, 13). Strand β25 is located about 20 amino acid residues upstream from the carboxy terminus of HIV-1 gp120 in the C5 region, and its sequence is highly conserved among strains of primate lentiviruses (sequence KYKVVKI in HIV-1_HXB2; residues conserved between HIV-1 strains are underlined) (12, 24). The last residue of this motif has been shown to be important for anchoring of gp120 on gp41 (9), suggesting that β25 is a functionally important structure of the inner proximal domain of gp120 likely to be conserved in other lentiviral glycoproteins. Sequences similar to the HIV-1 gp120 β25 motif (C5 motif) were visually sought in the gp135 carboxy-terminal region. A similar sequence was found in the caprine arthritis-encephalitis virus (CAEV) and visna virus gp135 between 33 and 34 amino acid residues upstream from the carboxy terminus of gp135 (Fig. ​(Fig.1B).1B). Similar to the C5 motif sequence of primate lentiviruses, the gp135 C5 motif is highly conserved in the gp135 of small ruminant lentiviruses (4, 27, 31, 35, 36). The sequence similarity also included the strictly conserved residue L483 of HIV-1 gp120 in the preceding α-helix 5, which is part of the two-helix, two-strand bundle of the inner domain. Flanking regions of gp120 and gp135 did not show any sequence similarity (not shown). Due to its short length, the significance of the conservation of the C5 motif in gp120 and gp135 was unclear. If this motif is indeed part of a structurally or functionally important domain of SU and not due only to chance, it should also be conserved in the SU of other lentiviruses. Therefore, to establish the relevance of this sequence similarity, we determined whether the C5 motif was also present in the carboxy terminus of the SU of other lentiviruses. Open in a separate windowFIG. 1Alignment of the C2 (A) and C5 (B) motifs of the SU from lentiviruses and type A and B retroviruses. Numbers at the right of the alignments indicate the position of the last residue of the motif from the initiation codon. Letters above the alignment indicate residue positions within each motif. Black backgrounds represent identical amino acids or conservative variations between the lentiviruses and type A and B retroviruses for each position of the motifs. Gray backgrounds represent identical amino acids or conservative variations between the lentiviruses and type A and B retroviruses (but which are nonconservative with the residues in black background) for each position. Numbers in parentheses indicate the number of amino acids between the last position of the C5 motif and the carboxy terminus of SU for each lentivirus. Thick lines indicate sequences which are part of HIV-1 gp120 strands β4, β5, and β25 and helix α5 (13). HIV-1 and HIV-2, human immunodeficiency virus types 1 (strain HXB2, GenBank accession number {"type":"entrez-nucleotide","attrs":{"text":"K03455","term_id":"1906382"}}K03455) and 2 (strain ROD, {"type":"entrez-nucleotide","attrs":{"text":"X05291","term_id":"60146"}}X05291); CAEV, caprine arthritis-encephalitis virus ({"type":"entrez-nucleotide","attrs":{"text":"M33677","term_id":"323294"}}M33677); Visna, visna virus ({"type":"entrez-nucleotide","attrs":{"text":"M10608","term_id":"470313"}}M10608); JSRV, jaagsiekte sheep retrovirus ({"type":"entrez-nucleotide","attrs":{"text":"M80216","term_id":"331338"}}M80216); EIAV, equine infectious anemia virus (AF033820); FIV, feline immunodeficiency virus ({"type":"entrez-nucleotide","attrs":{"text":"M73965","term_id":"323946"}}M73965); BIV, bovine immunodeficiency virus ({"type":"entrez-nucleotide","attrs":{"text":"M32690","term_id":"210706"}}M32690); JDV, jembrana disease lentivirus ({"type":"entrez-nucleotide","attrs":{"text":"U21603","term_id":"733067"}}U21603); HERV-K, human endogenous retrovirus K, type 2 genome ({"type":"entrez-nucleotide","attrs":{"text":"X82272","term_id":"757871"}}X82272); MMTV, mouse mammary tumor virus ({"type":"entrez-nucleotide","attrs":{"text":"X01811","term_id":"61630"}}X01811); MIAE, mouse intracisternal A-type element ({"type":"entrez-nucleotide","attrs":{"text":"M73818","term_id":"194061"}}M73818).Sequences conforming to the C5 motif consensus were also found in the SU of the equine infectious anemia virus (EIAV), feline immunodeficiency virus (FIV), bovine immunodeficiency virus (BIV), and the bovine jembrana disease lentivirus (JDV), 19 to 23 amino acid residues from the carboxy terminus of SU, the same relative position as the C5 motif from the carboxy terminus of gp120 in primate lentiviruses (Fig. ​(Fig.1B).1B). This sequence similarity was clear when considering the chemical similarities of amino acid side chains (Gln/Glu, Tyr/Trp, Lys/Arg/Gln, Val/Leu, or Val/Ile). A survey of lentiviral SU sequences present in GenBank revealed that the C5 motif was also highly conserved between EIAV, FIV, and BIV strains. For example, the C5 motif was found to be strictly conserved in 64 of 69 EIAV gp90 sequences in GenBank and is also stable during in vivo persistent infection (16, 39). However, the little variation that is observed between strains of a given lentivirus follows the same pattern as variation between different lentiviruses, suggesting a common constraint on sequence variation in different lentiviruses. For example, position h of the C5 motif of HIV-1 gp120 can be either of the conservative variations Lys/Arg or Gln/Glu, the same amino acids present at position h in other lentiviruses. Similarly, position h of the C5 motif of CAEV in different strains is either Lys or Arg (35), two of the residues allowed at position h in HIV-1 gp120. In addition, position b in the C5 motif of most FIV gp100 sequences in GenBank is the conservative variation Gln or Glu, the same amino acids present at position b of the C5 motif in EIAV and the small ruminant lentiviruses, respectively. Although the C5 motif is present in all lentiviruses, the flanking sequences were not consistently conserved except for a few amino acids in some pairwise alignments (not shown). Therefore, although conservation of the C5 motif may not be statistically significant in some SU pairwise alignments, the presence of this motif in the same position relative to the carboxy terminus of SU in all lentiviruses indicates that strand β25 of gp120 is an important structural or functional domain conserved in all lentiviruses.

Sequences similar to an HIV-1 gp120 C2 motif are present in the SU of most lentiviruses.

Using computer-assisted searches, we were previously unable to find in EIAV, BIV, or FIV the same extensive region of similarity that is observed between the C2 region of gp120 and gp135 (10). However, the presence of the C5 (β25) motif in all lentiviruses suggests that sequences similar to gp120 β5, which is antiparallel to β25 and conserved between gp120 and gp135, are also present in degenerate form in other lentiviruses. Visual examination of SU sequences from different lentiviruses revealed the presence of a similar motif (C2 motif) in EIAV, BIV, and JDV although not in FIV (Fig. ​(Fig.1A).1A). This 12-amino-acid C2 motif encompasses most of gp120 β-strands 4 and 5 and includes a strictly conserved cysteine residue in the β4 region. The C2 motif is highly conserved between strains of EIAV and BIV. In EIAV, the C2 motif is stable during persistent infection, with few conservative changes observed (16, 39). In addition, the C2 motif was found to be strictly conserved in 176 of 179 EIAV gp90 sequences present in GenBank, despite considerable sequence variation in other regions.Although some positions of the C2 motif were not absolutely conserved, we found a common pattern of variation between distantly related lentiviruses. For example, position f of the C2 motif is either Pro in EIAV, HIV-2, and simian immunodeficiency virus or aromatic (Tyr or Trp) in the small ruminant lentiviruses BIV and JDV. Also, position h can be either Phe or Tyr even in closely related lentiviruses (HIV-1/HIV-2, visna virus/CAEV, or BIV/JDV), and position l can be either Arg or Lys in the primate and small ruminant lentiviruses or Gln, which is a common conservative substitution for Arg and Lys, in EIAV, BIV, and JDV. Therefore, the C2 motifs of different lentiviruses appear to have a common constraint on sequence variation, suggesting a structural or functional similarity between the HIV-1 gp120 C2 domain and the SU of EIAV, BIV, and JDV.The other previously described gp120-gp135 conserved motifs outside the β4/β5 region could not be identified in the SU of other lentiviruses, including the sequence of gp120 β8, which has a cysteine forming a disulfide bond with the conserved β4 cysteine. Although the C2 motif was not present in FIV gp100, a similar motif was identified in a location upstream from the FIV gp100 V3 region (sequence SYCTDPLQIPLI, amino acids 318 to 329; conserved residues are underlined), in a similar relative position from gp100 V3 as the C2 motif from the V3 region of HIV-1 gp120. However, some of the highly conserved positions of the motif (positions g, h, and j) were not conserved in FIV gp100, and the significance of this FIV gp100 motif is unclear.

C2 motif is present in type A and B retroviral envelope surface glycoproteins.

The conservation of two short motifs in distant regions of SU that are located close to each other in the tertiary structure of HIV-1 gp120 suggests that this region represents a domain of SU that is of structural or functional importance. The TM ectodomains from lentiviruses and type B retroviruses have been shown to have some sequence similarity (19, 34, 38). Therefore, we asked whether sequence similarity between the Env of lentiviruses and type B retroviruses extends to the C2 and C5 motifs of SU.The type A and B retroviruses have some sequence homology in SU, and most of the sequence homology is located in the carboxy-terminal region of SU (18, 38). Visual examination of SU sequences from the human endogenous retrovirus K (18), mouse intracisternal A-type element (26), the exogenous/endogenous mouse mammary tumor virus (25), and the exogenous/endogenous type B/D jaagsiekte sheep retrovirus (JSRV) and the closely related ovine enzootic nasal tumor virus (which encode type B retroviral envelopes) (6, 38) revealed a sequence closely related to the C2 motif in their conserved carboxy-terminal region (Fig. ​(Fig.1A).1A). This sequence represents one of the most conserved sequences in the SU of this group of retroviruses and is also conserved among different strains or members of endogenous families (not shown). Some positions of the C2 motif, such as positions c, d, and g, are strictly or almost completely conserved between the lentiviruses and type A and B retroviruses. However, more informative than the sequence similarity between lentiviruses and type A and B retroviruses is the lack of distinction between the patterns of sequence variation for each position of the motif within and between retrovirus groups, even between closely related viruses. For example, position e of the C2 motif within both the lentiviruses and type A and B retroviruses can be either Pro or basic/Gln; the “dimorphic” position f encodes only Tyr/Trp or Pro (except in HIV-1); position h encodes either Phe or Tyr in all sequences; position i encodes either Ala or a hydrophobic residue in most sequences; position j encodes either Ile, Leu, or Phe in all sequences; position k encodes either Leu, Ile, or Val in all sequences; and position l is preferentially Lys, Arg, or Gln in the lentiviruses and JSRV. Most of these degenerate positions represent very conservative variations (positions a and h through l) or a restricted number of nonconservative variations (positions e and f, in the turn between β4 and β5). The sequence conservation and common pattern of variation between the C2 motifs of lentiviruses and type A and B retroviruses indicate a similar structural or functional constraint on sequence variation in the SU of these two groups of viruses.In contrast to the type A and B retroviruses, sequences similar to the C2 or C5 motifs could not be found in the SU of the Moloney murine leukemia virus, bovine leukemia virus, human T-cell leukemia virus types 1 and 2 (HTLV-1 and HTLV-2), Rous sarcoma virus, feline RD114 endogenous retrovirus, baboon endogenous retrovirus, feline leukemia virus type A, the Mason-Pfizer monkey retrovirus, or any spumaretrovirus even when using the Findpatterns program of the GCG package (7).Here we show that two short SU motifs are highly conserved in the lentiviruses and that one of these motifs is also conserved in the type A and B retroviruses. Many of the pairwise alignments were not statistically significant when tested by the Monte Carlo simulation of the Bestfit program of the GCG package and could therefore be attributed to chance. However, when all lentiviral sequences are included in the analysis and the multiple alignment is interpreted in the context of the X-ray structure of HIV-1 gp120, the conserved C2 and C5 motifs have a clear structural significance. The conservation of these motifs indicates that the region of the HIV-1 gp120 inner proximal domain centered on the antiparallel β-strands 5 and 25 forms a highly conserved lentiviral SU surface and suggests a possible structural similarity between the SU of lentiviruses and type A and B retroviruses in that domain. Although the C2 motif is too short to rule out convergent evolution between the SU of lentiviruses and type A and B retroviruses, their sequence similarity in TM (19, 34, 38) supports a common origin for most or the entire env genes of these two retroviral groups.The reason for the disagreement between the different degrees of sequence similarity in the SU of lentiviruses and the phylogenetic analyses of the pol gene products is unclear but probably reflects differences in evolutionary rates in different lentiviruses or recombination events (19, 20). Precedents for recombination events between env genes of closely or distantly related retroviruses, deduced from phylogenetic analyses, have been described. An exchange of env sequences probably occurred between HTLV-1 and HTLV-2 (19) and between a type C retrovirus closely related to the avian reticuloendotheliosis virus and a type B retrovirus which originated the type D retroviruses (19, 38).Modeling of the trimeric SU complex on the virion surface indicates that strands β5 and β25 form part of the most virion-proximal surface of the gp120 core (14, 37). While none of the residues of the C2 motif was directly tested for interactions with TM, at least one of the residues of β25 in the C5 region of HIV-1 gp120, I491, is important for stable SU-TM interactions (9). Therefore, the conserved lentiviral SU surface may represent a common structure among lentiviruses and possibly type A and B retroviruses for anchoring SU on TM in the envelope glycoprotein complex. It is interesting that the C5 motif region, which forms a β-strand in the CD4-bound gp120 core, is included in a computer-modeled pocket structure postulated to be important in SU-TM interactions (28), suggesting a structural basis for SU shedding upon receptor-induced conformational change.The sequence of the HIV-1 gp120 outer domain, shown as a cross-hatched box in Fig. ​Fig.2,2, is included entirely between the C2 and C5 motifs (13). Our previous sequence analysis indicates that the gp135s of small ruminant lentiviruses have a similar inner/outer domain organization: most strands of the inner domain β-sandwich as well as β12, located in the outer domain immediately upstream from the gp120 V3 loop, are conserved between gp135 and the gp120 of primate lentiviruses (10). The identification of a homologue of gp120 β25 in gp135 about 290 amino acid residues downstream from the C2 motif provides further support for a similar domain organization in the SU of primate and small ruminant lentiviruses. Consistent with this interpretation, the putative outer domain of gp135, located between the C2 and C5 motifs, is highly glycosylated and contains more than 80% of the potential N-linked glycosylation sites of gp135 (11), similar to the heavy glycosylation of the gp120 outer domain (37). In this gp135 domain model, the distance between the C2 and C5 motifs in the primary structure of SU would indicate a larger relative size of the putative outer domain of gp135 than gp120 outer domain. The presence of the C2 and C5 motifs in EIAV, BIV, and JDV would also suggest an analogous inner/outer domain organization for the SU of these lentiviruses. However, the shorter sequence between the C2 and C5 motifs in EIAV, BIV, and JDV may indicate either a much smaller or absent outer domain in the SU of these viruses (Fig. ​(Fig.2).2). The conserved C2 motif of EIAV gp90 was shown to be part of a minor neutralization epitope recognized by a murine monoclonal antibody (1), suggesting that the EIAV C2 motif is better exposed on the virion surface than the C2 motif of gp120, compatible with a smaller or absent outer domain in gp90. Interestingly, the C2 motif of type A and B retroviruses is located in the carboxy terminus of SU (Fig. ​(Fig.2)2) and C5 appears to be absent, indicating that the surface glycoproteins of type A and B retroviruses, although possibly structurally related to the SU of lentiviruses, probably lack an outer domain homologue and have a different domain organization than the SU of lentiviruses. Open in a separate windowFIG. 2Location of the C2 and C5 motifs in retrovirus envelope glycoproteins. The Env glycoproteins (excluding the amino-terminal leader peptide) are drawn to scale and aligned by the SU-TM cleavage sites conserved in all retroviruses (dotted line). The SU and TM domains of Env are indicated by double arrows. The boundaries of the C2, V3, and C5 regions of HIV-1 gp120 are indicated by thick lines above the alignment, and the location of the HIV-1 gp120 outer domain sequence is shown by a cross-hatched box. The black and gray boxes in the SU domain indicate the positions of the C2 and C5 motifs, respectively. Asterisks represent the described PNDs of EIAV (1), visna virus (29), FIV (17), and T-cell-adapted strains of HIV-1 (33) and HIV-2 (2).The two conserved colinear motifs of lentivirus SU could be useful as structural points of reference for comparative structural studies of SU from different lentiviruses. Variable domains of SU are important in the mechanisms of host cell invasion, tropism determination, and immune evasion. In HIV-1, the third variable loop V3 of gp120 is the main target of neutralizing antibodies in tissue culture-adapted strains and also determines coreceptor usage and tropism (5, 30, 32, 33). Whether sequences in variable regions of SU in other lentiviruses that are functionally equivalent to the gp120 V3 loop are also structurally related to the gp120 V3 loop is not clear. The position of variable domains relative to the C2 and C5 motifs could therefore indicate their structural relationship. For example, the principal neutralization domain (PND) of EIAV gp90, postulated to be functionally equivalent to the gp120 V3 loop (1, 16), is located upstream from the C2 motif instead of downstream, as the V3 loop in gp120 is (Fig. ​(Fig.2),2), suggesting that the gp90 PND and the gp120 V3 loop, while having similar roles in evasion of humoral immune responses, may not be structurally related to each other. A similar situation also occurs in visna virus, whose PND, located in the carboxy-terminal region of gp135 (29), was previously shown to be structurally unrelated to the HIV-1 gp120 V3 loop (10). This would indicate that different lentiviruses may have evolved different regions of a primordial lentivirus surface glycoprotein to perform similar functions important in virus-host interactions.