ABSTRACT
The structure of the transmembrane subunit (TM) of the retroviral envelope glycoprotein (Env) is highly conserved among most retrovirus genera and includes a pair of cysteines that forms an intramolecular disulfide loop within the ectodomain. Alpha-, gamma-, and deltaretroviruses have a third cysteine, adjacent to the loop, which forms a disulfide bond between TM and the surface subunit (SU) of Env, while lentiviruses, which have noncovalently associated subunits, lack this third cysteine. The Betaretrovirus genus includes Jaagsiekte sheep retrovirus (JSRV) and mouse mammary tumor virus (MMTV), as well as many endogenous retroviruses. Envelope subunit association had not been characterized in the betaretroviruses, but lack of a third cysteine in the TM ectodomain suggested noncovalently associated subunits. We tested the Env proteins of JSRV and MMTV, as well as human endogenous retrovirus K (HERV-K)108—a betaretrovirus-like human endogenous retrovirus—for intersubunit bonding and found that, as in the lentiviruses, the Env subunits lack an intersubunit disulfide bond. Since these results suggest that the number of cysteines in the TM loop region readily distinguishes between covalent and noncovalent structure, we surveyed endogenous retroviral TM sequences in the genomes of vertebrates represented in public databases and found that (i) retroviruses with noncovalently associated subunits have been present during all of anthropoid evolution and (ii) the noncovalent env motif is limited to mammals, while the covalent type is found among five vertebrate classes. We discuss implications of these findings for retroviral evolution, cross-species transmissions, and recombination events involving the env gene.
INTRODUCTION
Retrovirus infection is mediated by the envelope (Env) glycoprotein, which is a class I viral fusion protein (VFP) in all of the orthoretroviral genera except the epsilonretroviruses. Class I VFPs are synthesized as precursors that trimerize in the endoplasmic reticulum and are cleaved by cellular furin in the late Golgi apparatus to generate two subunits: a surface subunit (SU), which interacts with the receptor on target cells, and a transmembrane subunit (TM), which mediates fusion and is anchored in the viral envelope near its C terminus. After cleavage, SU and TM remain associated either noncovalently or via an intersubunit disulfide bond (1–3, 82, 83), resulting in trimers of heterodimers that traffic to the cell's plasma membrane to become part of the viral envelope when newly synthesized particles bud from the cell (4).
During infection, the fusion subunits of class I VFPs go through a series of conformational changes that lead from a metastable to a stable state. The first conformational change, which occurs upon binding of SU to the receptor, exposes a hydrophobic region at the N terminus of TM—the fusion peptide—which inserts into the plasma membrane of the target cell. This step then triggers the formation by TM of a highly stable six-helix bundle, a trimeric hairpin structure built around alpha-helical coiled coils that fold back upon one another (5). With the N terminus of TM inserted by its fusion peptide in the membrane of the host cell and the C terminus anchored in the viral membrane, the energy of the conformational change of TM to a stable six-helix bundle drives the fusion of the viral membrane with the cell membrane.
Retroviruses have evolved various mechanisms of SU-TM bonding and dissociation that accompany the fusion and entry steps (2, 6–11). Among retroviruses with covalently bonded Env subunits, the interaction of SU and TM has been most extensively studied in murine leukemia virus (MLV), a gammaretrovirus. Initial activation of Env occurs during maturation of the virion, when the viral protease cleaves a 16-amino-acid peptide (the R peptide) from the C terminus of the cytoplasmic tail. Upon receptor binding, a free thiol on a cysteine in SU isomerizes the single intersubunit bond, resulting in the formation of a new intrasubunit bond in SU and breaking of the bond with TM (3, 11, 12). The subsequent dissociation of SU from TM then allows TM to continue folding into its stable state. The same sequence of steps appears to apply also to the deltaretrovirus human T-lymphotropic virus 1 (HTLV-1) (8). Avian leukosis virus (ALV), an alpharetrovirus, also has subunits that have been shown to be disulfide bonded (7), but no isomerization occurs. Upon binding to a target cell, ALV SU interacts with the receptor to prime TM for its fusion activity; however, fusion is not triggered until the virion is inside an endosome and exposed to low-pH conditions. Therefore, ALV Env needs to retain its fusion-ready conformation until it is in the endosome, and the SU-TM disulfide bond may serve to allow this conformation to be maintained until that time. Indeed, it has been shown that the intersubunit bond in ALV Env is not isomerized during either the entry or fusion process but remains intact indefinitely (10).
In lentiviruses, including human immunodeficiency virus type 1 (HIV-1), SU and TM are noncovalently associated. Although it is unclear whether HIV-1 fusion takes place at the plasma membrane (13, 14) or in an endosomal compartment (15), it is pH independent (16). Following receptor and coreceptor binding, HIV-1 gp120 (SU) undergoes a conformational change that exposes the fusion peptide of gp41 (TM), which buries into the target cell membrane. SU, held in the complex only by noncovalent interactions, then dissociates from TM, which in turn folds into its hairpin trimer, pulling the viral and cellular membranes together to fuse.
Envelope subunit association in the Betaretrovirus genus has not been characterized. This genus includes Jaagsiekte sheep retrovirus (JSRV), the closely related enzootic nasal tumor virus (ENTV), and mouse mammary tumor virus (MMTV), all of which are oncogenic (17, 18) and have endogenous as well as exogenous members (19, 20). Closely related to the betaretroviruses are the numerous betaretrovirus-like endogenous retroviruses (ERVs) that have colonized the genomes of mammals, including the human endogenous retroviruses (HERVs) of the HERV-K(HML-2) group, members of which are estimated to have first infected our ancestors at least 34 million years ago (mya), with new rounds of infection occurring as recently as 200,000 years ago or less (21–24). Additionally, betaretroviruses include a few endogenous and exogenous viral species (e.g., Mason-Pfizer monkey virus [MPMV] and squirrel monkey retrovirus) derived from a recombination event in which an envelope gene was acquired from a gammaretrovirus (25, 26).
This study is concerned only with the Env proteins of the nonrecombinant betaretroviruses. We found that members of the Betaretrovirus genus—including an endogenous betaretrovirus-like retrovirus that first entered the human germ line about 1 million years ago—share with the lentiviruses the feature of noncovalent association between SU and TM. Moreover, distinct patterns of cysteines, along with other features of the TM amino acid sequence, allow easy distinction between retroviruses with covalent and noncovalent Env subunit association. We exploited this feature to examine the ERV fossil record in the genomic DNA of vertebrate species represented in public databases and found that (i) retroviruses with the noncovalent-type Env protein—a feature previously associated only with lentiviruses—have been infecting the genomes of primate ancestors for at least 35 million to 55 million years, alongside retroviruses with covalently bonded Env subunits, and (ii) whereas the covalent-type Env protein is found across five vertebrate classes, the noncovalent type appears to be strictly limited to mammals. Finally, we discuss both evolutionary and biological implications of the two Env types in the context of these findings.
MATERIALS AND METHODS
Cells and expression vectors.293T human embryonic kidney epithelial cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and penicillin-streptomycin (50 mg/ml). The JSRV Env expression plasmid (a gift from Hung Fan) was cloned from wild-type JSRV (27), and expression was driven by the cytomegalovirus (CMV) immediate-early promoter with a FLAG tag inserted at the 3′ end of the envelope reading frame. The MMTV Env expression plasmid contains the MMTV(C3H) (28) envelope-coding region under the control of the CMV promoter and was a gift from Susan Ross. The HERV-K108 envelope-coding region was amplified from genomic DNA using standard PCR methods with the following primer pair: 5′-CACC ATG AAC CCA TCA GAG-3′ and 5′-CTT TGT GGA TTG TAA TTT GGG G-3′; the amplified fragment was ligated into the pcDNA3.1 TOPO mammalian expression plasmid (Invitrogen). The product was cloned into a mammalian expression vector (a gift of Harvard Gene Therapy Initiative) under the control of a CMV promoter. Env protein from amphotropic MLV was expressed from the plasmid pSVA-MLV (29). The ALV Env expression vector was amplified using primers RCG10 (5′-GCCGCTGTGAAAAACAGGGAC-3′) and RCG21 (5′-TGCCTTCCTCGATGGACTCGC-3′) from the RCASBP(B) vector (30) and cloned into the pCB6 mammalian expression vector under the control of a CMV promoter (31). A FLAG tag was introduced at the N terminus of SU, immediately after the signal sequence cleavage site.
Transfections and virus pseudotyping.For analysis of retroviral Env expression, 293T cells were transiently transfected with the envelope expression vectors using the calcium phosphate method; at 48 h posttransfection, cells were treated with lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 1% Igepal CA-630 [Sigma]) supplemented with 20 mM N-ethylmaleimide (NEM; Thermo Scientific). Pseudotyped recombinant simian immunodeficiency viruses (SIVs) were made by transient cotransfection of 293T cells with each of the envelope constructs plus the SIV-based vector SIVΔenv (carrying the green fluorescent protein [GFP] gene; a gift from David Evans), using a calcium phosphate transfection kit (Sigma). At 2 days posttransfection, virus was concentrated from cell supernatants by centrifugation at 105,000 × g for 1 h through a 20% sucrose cushion in phosphate-buffered saline (PBS), and pellets were resuspended in low-serum medium (Opti-MEM) or SDS loading buffer.
Western blotting and antibodies.Samples were resolved by SDS-PAGE on 10% or 4 to 15% gradient Tris-HCl gels (Bio-Rad) and transferred onto polyvinylidene difluoride nitrocellulose membranes. JSRV and ALV TMs were probed with an antibody to the FLAG tag (Sigma). MMTV Env was probed with polyclonal antibodies to gp52 (SU) or gp36 (TM) (NCI Repository), both of which were gifts from Susan Ross. MLV Env was probed with a polyclonal antibody to gp70 (SU) (Quality Biotech, Camden, NJ), a gift from Lorraine Albritton. HERV-K Env was probed with a monoclonal antibody to TM (Austral Biologicals). Secondary antibodies were horseradish peroxidase conjugates, and bands were visualized using an ECL Advanced Western blotting detection kit (GE Healthcare).
Data mining for ERV sequences.To identify ERV TM sequences, tBLASTn searches (32) of the National Center for Biotechnology Information (NCBI) genomic databases were performed using a panel of query TM sequences selected to represent all five genera of retroviruses with class I viral fusion proteins, as well as endogenous sequences. Sequences returned by a BLAST search with an E value (probability of a false positive) of ≤0.01 were kept. This threshold was determined by inspecting several search results for an E value below which sequences could always be confirmed visually to be TM sequences.
Sequence analysis.Sequences extracted from the in silico searches were translated in all three reading frames and compared to amino acid alignments of previously characterized, published sequences to reconstruct putative open reading frames. Multiple alignments of the adjusted sequences were then performed using the CLUSTAL-W program (33) in MEGA (version 4.0) software (34). Alignments of the reverse transcriptase (RT) region of pol spanned 295 amino acids and included the conserved motifs characterized by Xiong and Eickbush (35). Alignments of TM sequences of env always included the cysteine region and spanned 150 to 200 amino acids, excluding the cytoplasmic domain, which varies widely in sequence even within genera.
RESULTS
Betaretroviral and lentiviral TMs share sequence features, including number of cysteines.The domain organization of orthoretroviral TMs reflects their common fusion mechanism (Fig. 1). At or near the N terminus is the hydrophobic fusion peptide, which inserts into the target cell membrane following receptor binding, while, near the C terminus, the transmembrane region anchors the protein in the viral membrane. Between these two hydrophobic regions are two heptad repeats (hr1 and hr2) that form the coiled coil that packs upon itself to form the postfusion six-helix bundle. In between hr1 and hr2 is a flexible ectodomain that lies distal to the membrane-anchored ends of the six-helix bundle. In the middle of this flexible region is a rigidly conserved pair of cysteines that mediates an intramolecular disulfide loop that is involved in interaction with SU (6) (Fig. 1). This disulfide loop is a common structural feature not only of retrovirus transmembrane proteins (37) but also of fusion proteins of several other RNA viruses, including Lassa virus, lymphocytic choriomeningitis virus (LCMV), and Ebola virus (38), indicating an important conserved function.
Two types of Env among the class I fusion proteins of retroviruses. (Top) Domain arrangement of retroviral TM, with an alignment demonstrating the conservation of the ISD in the alpha-, gamma-, and deltaretroviruses and the difference in cysteine motifs. FP, fusion peptide; hr1 and hr2, heptad repeats; MPER, membrane proximal external region; TM, transmembrane domain; GALV, gibbon ape leukemia virus; FrMLV, Friend MLV; FIV, feline immunodeficiency virus. Arrows indicate the cysteine residues thought to participate in the intramolecular disulfide loop, conserved among all five genera. (Bottom) The five retroviral genera with class I fusion proteins and their features. *, subunit association has previously been characterized for all genera except the betaretroviruses and is indicated along with the cysteine motif.
A comparison of orthoretroviral TM sequences reveals some notable differences, however, between viruses with noncovalently associated subunits (lentiviruses) and those with covalently bonded subunits (alpha-, gamma-, and deltaretroviruses). First, retroviruses in which SU forms a covalent bond with TM have an additional cysteine adjacent to the loop. There is evidence that this C-terminal cysteine mediates the single intermolecular disulfide bond with a cysteine in SU (3, 7, 8, 11, 39). By comparison, retroviruses of the Lentivirus genus lack this third, C-terminal cysteine, further supporting its role as the mediator of the intersubunit disulfide bond. Second, those retroviruses with covalently bonded subunits also encode an immunosuppressive domain (ISD), lacking in lentiviruses, that has been shown to inhibit lymphocyte proliferation (40) and allow escape from immune effectors of the innate and adaptive arms of the host immune system in a mouse model (41–44). Third, the membrane proximal external region (MPER)—a stretch of 30 residues immediately upstream of the transmembrane region that is thought to be important for Env protein incorporation into virions, as well as membrane disruption during fusion (45–47)—is longer in lentiviruses by 25 to 30 amino acids (48).
The amino acid sequence of betaretroviral TM mirrors that of lentiviruses in its relatively long MPER and its lack of an ISD (Fig. 1). The absence of a third cysteine not only adjacent to the loop but also anywhere within the betaretroviral TM ectodomain strongly suggested that betaretroviruses, like lentiviruses, have noncovalently associated Env subunits. The following studies were performed to test this idea.
The Env subunits of MMTV and JSRV lack intersubunit disulfide bonds.To determine whether the envelope proteins of betaretroviruses possess a covalent bond between the SU and TM, we analyzed Env proteins from two members of the Betaretrovirus genus—MMTV and JSRV—under reducing and nonreducing conditions on a Western blot. We transfected 293T cells with a construct previously shown to generate MMTV Env capable of mediating infection (28) or with a construct expressing JSRV Env with a FLAG tag at the C terminus of TM. As positive controls for demonstrating disulfide bonded Env subunits, 293T cells were transfected with amphotropic murine leukemia virus (amphoMLV) Env (29) or ALV Env with a FLAG tag at the N terminus of SU (see Materials and Methods). Cells were lysed 2 days later in the presence of the alkylating agent NEM, to inactivate any free thiols that could mediate lysis-induced isomerization of disulfide bonds (11). Monomers of MMTV Env and JSRV Env were detected under both reducing and nonreducing conditions (Fig. 2A and B), indicating the absence of covalent bonding between SU and TM. In contrast, monomers of amphotropic MLV Env and ALV Env, both of which have previously been demonstrated to have covalently bonded subunits (3, 7, 12), could be seen only under reducing conditions (Fig. 2C and D).
Monomers of MMTV and JSRV Env can be seen under nonreducing conditions in cell lysates. Lysates of 293T cells transfected with plasmid DNA encoding the envelope proteins of MMTV, JSRV, amphotropic MLV, or ALV were subjected to reducing (with DTT [+]) or nonreducing (without DTT [−]) conditions and analyzed by Western blotting. Env subunit monomers of MMTV (A) and JSRV (B) can be seen under nonreducing conditions in cell lysates. Lane 1 in panel B, lysate from untransfected 293T cells as a control. For MMTV Env, the blot was probed with a polyclonal antibody to SU; the JSRV blot was probed with antibody to a FLAG tag at the C terminus of TM. Env subunit monomers of MLV (C) and ALV (D) Env proteins can be seen only under reducing conditions. The MLV blot was probed with a polyclonal antibody to MLV gp70 (SU). Arrows, the disulfide-bonded 95-kDa species, the 85-kDa Env precursor, and the 70-kDa SU monomer. The ALV blot was first probed with antibody to a FLAG tag at the N terminus of ALV-B SU and then stripped and reprobed with antibody to TM. Arrows, positions of the Env precursors and TM monomers. Molecular mass markers (in kDa) are given on the left.
To test the structure of Env in its fully processed, virion-associated form, we generated pseudotyped particles and assayed these under reducing and nonreducing conditions. SIV virions with an MMTV envelope were generated, based on the success of another group in using SIV as a helper virus in pseudotyping with the HERV-K envelope (49). When the pseudotyped SIVΔenv particles containing MMTV Env protein were analyzed on a Western blot, SU monomers could readily be seen under nonreducing conditions (Fig. 3A, lane 2), indicating the absence of covalent bonding between Env subunits in MMTV pseudovirions, though the rather large amount of precursor suggests inefficient processing (Fig. 3B). Similarly, the blot of FLAG-tagged JSRV Env-pseudotyped SIV revealed the presence of TM monomers in both the reduced and nonreduced samples, further supporting the lack of an intersubunit covalent bond in JSRV Env.
Monomers of MMTV and JSRV TM can be seen under nonreducing conditions in pseudotyped virions. Pseudovirions were produced by cotransfecting 293T cells with an SIV helper construct and an Env-expressing construct from MMTV (A) or JSRV (B). Cell culture supernatant was collected 2 days later and passed through a 20% sucrose cushion. Samples were analyzed by Western blotting under reducing and nonreducing conditions as indicated by the presence (+) or absence (−) of DTT. The MMTV blot was probed with a polyclonal antibody to TM; the JSRV blot was probed with an antibody to the FLAG tag expressed at the C terminus of TM. Arrows, positions of the Env precursor and TM monomers. Molecular mass markers (in kDa) are given on the left.
The Env subunits of an endogenous retrovirus, HERV-K(HML-2), lack an intersubunit disulfide bond.We assayed Env subunit association in HERV-K108, a member of the HERV-K(HML-2) group, which began infecting the germ line of our ancestors an estimated 34 mya (50), with new rounds of infection continuing into the recent past or possibly even ongoing (51–53). HERV-K108 is the only member of the HERV-K group that has been shown to encode an envelope protein capable of mediating infection, albeit very inefficiently (49). Estimates based on mutations acquired in the proviral long terminal repeats (LTRs) place the entry of K108 into the human genome at ∼1.2 mya (54, 55). In phylogenies based on either pol or the TM domain of env, HERV-K(HML2) clusters with betaretroviruses, and members of this group have a two-cysteine motif in the flexible region (Fig. 1), predicting noncovalent association of subunits.
We amplified the HERV-K108 env gene from human genomic DNA and inserted it downstream of a CMV promoter in a mammalian expression vector. Transfection of 293T cells with this construct led to detection in cell lysates of three protein bands on a Western blot probed with a monoclonal antibody against HERV-K TM, including a band at ∼85 kDa and a doublet at ∼37 kDa, consistent with the sizes expected for the precursor molecule and the TM monomer, respectively (Fig. 4A, lane 1). The doublet likely represents different glycosylation forms, as treatment with the deglycosylating enzyme peptide-N-glycosidase F (PNGase F) resolved it to a single band of about the size expected for the predicted 234-amino-acid TM protein (56, 57) (Fig. 4A, lane 2).
HERV-K108 Env lacks an intersubunit disulfide bond. (A) The HERV-K108 Env protein was expressed in 293T cells (lane 1) and treated with PNGase F to remove N-linked glycans (lane 2). The TM doublet in lane 1 is presumably due to different glycoforms. (B) Lysates of 293T cells expressing HERV-K108 Env were electrophoresed under reducing (with DTT [+]) or nonreducing (without DTT [−]) conditions and analyzed by Western blotting. Lane 1, lysate from untransfected 293T cells as a control. (C) Pseudovirions were produced by cotransfecting 293T cells with an SIV helper construct and the HERV-K108 Env-expressing construct and then lysed in SDS loading buffer and electrophoresed under reducing or nonreducing conditions. Lane 1, supernatant from cells transfected with the SIV construct only. All blots were probed with a monoclonal antibody to TM. Arrows, positions of the Env precursor and TM monomers. Molecular mass markers (in kDa) are indicated on the left.
To determine whether, as with other betaretroviruses, HERV-K SU and TM were noncovalently associated, lysates (Fig. 4B) and SIV pseudovirions (Fig. 4C) from 293T cells transiently transfected with the HERV-K Env construct were analyzed by Western blotting using a monoclonal antibody to TM with and without dithiothreitol (DTT) treatment. In both cases, TM monomers were present in samples that were not reduced (Fig. 4B and C, lanes 3). In both cases, treatment with DTT led to significant sharpening of the ∼80-kDa precursor band (Fig. 4B and C, lanes 2), suggesting that it is present in a variety of disulfide-bonded forms. These results show that, like other betaretroviruses, the Env subunits of this approximately 1 million-year-old betaretrovirus are noncovalently associated, as predicted by the two-cysteine motif that it carries in its TM sequence.
Retroviruses with noncovalently associated subunits have a long evolutionary history within the primate lineage.To gain insight into how far back in evolutionary time that these two envelope types have been conserved, we examined the cysteine motifs of the various groups of HERVs in the human genome, which represent infections going back 60 to 70 mya (58), corresponding with the time of emergence of primates (59). Consensus sequences for the HERV-H, HERV-K(HML-2), HERV-K(HML-3), and HERV-K(HML-5) groups were generated previously (60–64); consensus motifs for the remaining groups were determined by aligning sequences culled from the NCBI nr/nt database (Table 1). Also included were two recently discovered endogenous lentiviruses from the genomes of the European rabbit (65, 66) and the gray mouse lemur, a prosimian (67, 68).
Cysteine motifs in TMs of various ERVs
The covalent and noncovalent motifs are both represented, and both appear to have a long history in the primate lineage. HERV-H, HERV-FRD, HERV-R, ERV-9, HERV-W, and HERV-T all have a three-cysteine motif in the TM ectodomain, suggesting that the Env subunits were covalently bonded when these retroviruses were active. The oldest elements with the three-cysteine motif are estimated to have infected the primate lineage over 40 million years ago (58). Note that, despite their great age, the HERV-W and HERV-FRD Env proteins are still functional and are posited to mediate formation of the placental syncytiotrophoblast layer in primates (69–71). Characterization of HERV-W Env protein processing indicates that the subunits are indeed covalently bonded, as predicted by the three-cysteine motif in TM (72). Three of the groups in the human genome—all of them members of the HERV-K(HML) supergroup—have a two-cysteine motif in TM, implying that the Env subunits of these elements were noncovalently associated, as we found for HERV-K108. Not unexpectedly, the two endogenous lentiviruses RELIK and pSIVgml also have two-cysteine motifs. The oldest known members of the HERV-K(HML5) group are estimated to have first integrated into the primate lineage as far back as 55 mya (62), suggesting that retroviruses with noncovalently associated Env proteins, previously thought to be a feature only of lentiviruses, have been present throughout the span of primate evolution.
A comparison of the motifs of betaretroviruses and lentiviruses further reveals a conserved threonine residue C terminal to the cysteine pair that is common to both genera and present in exogenous members as well (feline immunodeficiency virus and bovine immunodeficiency virus were the only exceptions found). This conserved residue is of interest because in mutational analysis of the interheptad repeat region of HIV-1 to determine which residues participate in the noncovalent association of subunits, mutation of this threonine had the most pronounced effect on subunit association of any of the mutations tested (73). Moreover, a variant of HIV-1 Env has been engineered with a disulfide bond between TM and SU by the mutation of the threonine to cysteine at the same position (T605C); this new cysteine formed an intersubunit bond with a cysteine that the researchers added to the C-terminal domain of SU (74). Although the engineered protein was incapable of mediating fusion, it appears that the site has been conserved as a key region of intersubunit association by retroviral Env proteins of both structures for at least 55 million years.
The two Env types have distinct species distributions.The Betaretrovirus genus encompasses members with either of the two Env types; MPMV, for example, has the covalent-type Env, derived from an ancient recombination with a gammaretrovirus, while MMTV and JSRV have the noncovalent type. However, the numerous studies of the species distribution of betaretroviruses have been based on the RT region of the pol gene, thus obscuring any host range differences associated with the Env type. In order to assess any significance of the Env type on host range, we performed an extensive search of the genomic sequences offered through the NCBI databases, using a panel of TM query sequences selected to represent a diverse range of species and including representatives from both exogenous and endogenous retroviruses.
The results of the database search are shown in Table 2 and reveal a striking difference in the species distribution of the covalent- and noncovalent-type env sequences. Represented in the survey are 78 species spanning five vertebrate classes: mammals, birds, reptiles, amphibians, and bony fishes. The mammalian samples are further divided into three subclasses corresponding to birth mechanisms—the placental, egg-laying, and marsupial subclasses—which could have relevance in terms of the level of accessibility of germ cells to retrovirus infection.
Covalent- and noncovalent-type TM sequences represented in NCBI databasesa
The noncovalent type of Env is widespread among the mammalian class, being found in numerous primates, as well as insectivores, rodents, domestic and wild herd species, bats, dolphins, and domestic cats and dogs. Furthermore, it is found among all three reproductive types (although it is unknown whether germ line integrations occurred prior to or after emergence of the reproductive traits). However, the noncovalent-type sequences are notably absent outside mammals, despite an even more extensive search using query sequences from some of the more divergent noncovalent sequences found in mammals during the initial search (see Materials and Methods). Although there are generally fewer sequence data available for nonmammalian species, we were able to recover covalent TM sequences from 26 nonmammalian species, none of which had noncovalent TM sequences. Thus, it is unlikely that our result is an artifact of sampling bias.
The covalent Env type, in contrast, has a very wide taxonomic range, being present in the genomes of organisms from all five of the vertebrate classes represented in the survey. The covalent type of TM was also found in the genome of every species of mammal that was positive for the noncovalent type, indicating that the mammalian lineage has been susceptible to infection by retroviruses with either Env type at various times throughout its evolution.
DISCUSSION
This study was the first to characterize Env subunit association in the Betaretrovirus genus, demonstrating that members of this genus share with lentiviruses a noncovalent association of SU and TM. With the Env subunit association of all five orthoretrovirus genera with class I fusion proteins now characterized, two clearly distinct Env structures emerge: the alpha-, delta-, and gammaretroviruses (and a few betaretroviruses that acquired gammaretrovirus-like env genes via recombination) have an Env protein with covalently bonded subunits, a shorter MPER, and an ISD sequence; lentiviruses and betaretroviruses are characterized by Env proteins with noncovalently associated subunits, a longer MPER, and the absence of an ISD sequence. Our finding that betaretroviruses, which have a two-cysteine motif in TM, lack an intersubunit covalent bond is in agreement with the structural and biochemical studies that indicate that two cysteines within the flexible portion of TM mediate a strictly conserved intramolecular disulfide loop and that the third, C-terminal cysteine, when present, mediates the intersubunit disulfide bond (3, 7, 8, 11, 39).
Based on the presence of a two-cysteine motif (CX7C) in the TM sequence, we predicted that members of the HERV-K(HML2) group, which initially infected the primate germ line about 34 mya (50), would have a noncovalent Env structure. This prediction was confirmed by our finding that the Env subunits encoded by HERV-K108, a member of the HML2 group, are noncovalently associated. Moreover, examination of the TM sequences of human endogenous retroviruses revealed typical two-cysteine motifs in betaretrovirus-like elements, including members of the HERV-K(HML5) group, which initially infected the primate germ line some 55 mya (62). Similarly, the two-cysteine motif is also found in two endogenous lentiviruses—pSIVgml and RELIK—estimated to be at least 4 million years old (67, 68) and 12 million years old (66, 75), respectively. Although the recent discovery of these sequences indicates that the lentivirus genus is much older than previously thought, the appearance of the two-cysteine motif in the ancient betaretrovirus-like HERV-K(HML5) group suggests that the noncovalent Env structure arose with the betaretrovirus lineage, perhaps before the emergence of the lentiviruses.
Unlike lentiviruses, betaretroviruses are widely represented among endogenous proviruses. The observation that they have a noncovalent Env structure led to the idea that perhaps the difference between noncovalent and covalent association would have implications for the host range of the antecedent viruses. Indeed, TM sequences showed a sharp distinction in species distribution, in that while retroviruses with the covalent type of Env are found among multiple orders, noncovalent types are limited to mammals. This pattern mirrors what is known of exogenous retrovirus infections, in that there are no documented cases of infections of nonmammalian species with lentiviruses or nonrecombinant betaretroviruses. Thus, our study has uncovered an important correlation of Env subunit association and host range, revealing a previously unrecognized aspect of the evolutionary dynamics of retroviruses.
The biological significance of Env subunit association is uncertain. One outcome of the noncovalent Env structure is that SU is easily shed from the virion surface or from cells in culture (76, 77). In HIV-1 infection, gp120 shedding results in the exposure of regions that normally interact with gp41 and appears to elicit high titers of nonneutralizing antibodies, leading some researchers to propose an immune decoy function for the shed gp120 (77). Another result of shedding is that the released SU can bind its cognate receptor at the cell surface and affect virus-receptor interaction. HIV-1 has evolved two accessory proteins that independently mediate downregulation of CD4 expression: Nef, which causes internalization of CD4 from the cell surface and its subsequent degradation, and Vpu, which interacts with CD4 in the endoplasmic reticulum, targeting it to the proteasome. Highlighting the importance of receptor downregulation to the HIV-1 life cycle are studies showing that Env incorporation into virions and virus release are both more efficient when interaction between Env and CD4 is inhibited (78, 79). The ability of shed SU to bind receptor thus presents the possibility that SU shedding may provide an additional means of blocking interaction between Env and receptor, allowing efficient release of mature virions.
Phylogenetic trees based on the RT region of the five genera considered in this study resolve into two major branches, defining the class I and class II ERVs (Fig. 5, left). Trees based on TM similarly reveal two major branches, corresponding to the retroviruses that have an ISD-positive/covalent TM (alpha-, gamma-, delta-, and type D betaretroviruses), on one side, and the lentiviruses and betaretroviruses (both of which are ISD negative) on the other side (Fig. 5, right). Viruses marked with an asterisk show discordance between RT-based and TM-based trees, suggesting recombination events that have allowed these class II retroviruses to acquire gammaretroviral env genes (48). Examples are the betaretrovirus Mason-Pfizer monkey virus (MPMV) (4, 25), the alpharetrovirus Rous sarcoma virus (RSV) (48), and human T-lymphotropic viruses (HTLVs) and bovine leukemia virus (BLV) (48, 80), members of the Deltaretrovirus genus. That betaretroviruses have a noncovalent Env structure implies that the documented recombination events as well as those suggested by RT/TM tree discordance all involve replacement of a noncovalent env sequence by a covalent one. In contrast, no combination of a class I RT and a noncovalent Env has ever been identified, despite the finding that all but 2 of the 52 mammalian species represented in our survey had both types of env in their genomes. This directional bias supports the idea of host range expansion afforded by acquisition of a covalent env in place of a noncovalent env as a driver of pol/env recombination. Such an event appears to have occurred, for example, in the case of Python molurus endogenous retrovirus (PyERV), which has a betaretrovirus-like pol and a mammalian env (81).
Comparison of phylogenetic trees based on RT and TM protein-coding sequences. Simplified neighbor-joining trees depict the evolutionary relationships among exogenous retroviruses from different genera. The tree on the left is based on the RT region and depicts the class I and II designations applied to ERVs. The tree on the right is based on TM sequences from the same set of retroviruses with the motif of the cysteine loop, and the presence or absence of ISD is indicated. α, β, γ, δ, and λ, alpha-, beta-, gamma- and deltaretroviruses and lentiviruses, respectively; asterisks, viruses that have been characterized as recombinants; animal symbols, vertebrate classes in whose genomes each of the env types was found in this study; scale bars, numbers of substitutions per site; REV, reticuloendotheliosis virus; BaEV, baboon endogenous virus; PERV, porcine endogenous virus; XMRV, xenotropic murine leukemia virus-related virus; FLV, feline leukemia virus; CAEV, caprine arthritis encephalitis; TgERVF, Taeniopygia guttata endogenous retrovirus F.
Our comparison of the ectodomain sequences of betaretroviruses and lentiviruses revealed a conserved threonine residue C terminal to the cysteine pair that is common to both genera and, at least in HIV-1, appears to play a key role in subunit association (73). Along with the two-cysteine motif, long MPER, and lack of an ISD, the shared features of lentiviral and betaretroviral Env proteins point to a deeper evolutionary relationship between these genera than previously recognized. Combined with more sequences from endogenous lentiviruses, comparative analyses of the distribution within mammals of noncovalent TM sequences could uncover the roots of the lentivirus-betaretrovirus relationship, offering a chance to further our knowledge of the long-term dynamics between an important human pathogen and its host.
FOOTNOTES
- Received 8 June 2012.
- Accepted 21 November 2012.
- Accepted manuscript posted online 5 December 2012.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
