We have collected a number of RF tumor and non-tumor samples from different SRV-2 infected macaque species from a variety of sources, including captive macaques from six different US primate research centers and wild-caught animals from Singapore (Table 1). The tissue samples ranged from freshly frozen tissue from recent necropsies to 20–30 year old formalin-fixed paraffin-embedded tissue sections. Genomic DNA was isolated and used in PCR amplification to obtain full-length nucleotide sequences of the SRV-2 env genes. In some cases, sample amount and degradation limited our ability to obtain the full length sequence. We have analyzed the deduced amino acid sequences from these isolates and have compared them with the complete env gene sequences previously identified for an SRV-1 isolate from a rhesus macaque from the California NPRC (D1/RHE/CA; 6), an SRV-3 isolate from a rhesus macaque from the Wisconsin NPRC (D3/RHE/WI; 2), an SRV-2 isolate from a Celebes black macaque (D2/CEL/OR; 14), and two closely related SRV-2 isolates from rhesus macaques (D2/RHE/OR and D2/RHE/OR/V1; 1516) from the Oregon NPRC. Additional partial env gene sequences previously identified from SRV-2 isolates of pig-tailed macaques at the Washington NPRC (D2/MNE/WA) 17 were also included in the comparison.

We determined the complete sequence of the env gene from sixteen different SRV-2 isolates and the sequence of the C-terminal half of the env gene for an additional two SRV-2 isolates. The resulting eighteen env gene sequences were multiply aligned with the SRV-1, SRV-2, and SRV-3 prototype sequences obtained from the NCBI sequence database (Genbank), as indicated above. Using the distantly related simian sarcoma virus (SSV) env gene sequence as outgroup, we performed a phylogenetic analysis using the protein maximum-likelihood method. All of the putative SRV-2 sequences amplified from our tissue samples clustered closely together and were clearly distinct from the SRV-1 and SRV-3 prototype sequences (Figure 1A).

A closer phylogenetic analysis of the SRV-2 sequences revealed the presence of six separate clusters of sequences (Figure 1B). These clusters represent six molecular subtypes of SRV-2. The subtype SRV-2A cluster included the original SRV-2 prototype, D2/CEL/OR, isolated from a Celebes black macaque with SAIDS-RF at the Oregon NPRC in 1985 13, a closely related isolate obtained in early 2000 from a cynomolgus macaque (NM101) with SAIDS-RF from the Lovelace Respiratory Research Institute in New Mexico, and a more distantly related virus, MmMich, obtained in the late1990's from a rhesus macaque at the primate center at the University of Michigan (see Table 1 for a description of viruses and their macaque hosts). The SRV-2B cluster included the previously characterized closely-related SRV-2 isolates obtained in the late 1980's from the rhesus macaque colony at the Oregon NPRC, D2/RHE/OR and D2/RHE/OR/V1 15. In addition, this cluster contained isolates obtained in the early 1980's from two colony-born pigtailed macaques (M78114, T82422) at the Washington NPRC, an isolate obtained in 1995 from a pig-tailed macaque transferred from Indonesia to the Washington NPRC (90167), and an isolate obtained in the mid 1990's from a rhesus macaque (YN91-224) from the Yerkes NPRC, which had all been diagnosed with SAIDS-RF. Also included in this subtype were a number of isolates from additional pig-tailed macaques with SAIDS-RF from the WaNPRC for which only partial sequences have been obtained (D2/MNE/WA) 17. The SRV-2C subtype contained a previously unknown isolate from a pig-tailed macaque (442N) housed at the NIH primate center which had been diagnosed with SAIDS-RF in 1996 24. In addition, SRV-2 virus obtained in the mid-1990's from another pig-tailed macaque (17915) from the NIH, and from a cynomolgus macaque (91048) from the Washington NPRC, both without RF, contained similar sequences and grouped within the SRV-2C subtype. The SRV-2D subtype consisted of three virtually identical isolates obtained in the early 1990's from closely-related healthy pig-tailed macaques (F89336, F90346, and F91249) at the Washington NPRC. The SRV-2E subtype included isolates obtained from five closely related cynomolgus macaques at the Washington NPRC. Finally, the SRV-2F subtype consisted of an isolate obtained in 2003 from a cynomolgus macaque which had been sampled in the wild on the island of Singapore. Closely related isolates were identified in other cynomolgus macaques from the same geographical area (Richard Grant, unpublished data).

An alignment of the complete env sequence from prototypes of each of the SRV-2 subtypes revealed identical sizes (574 amino acids) and a high degree of conservation throughout the entire protein (Figure 2). The genetic variation between the env genes from isolates within a specific SRV-2 subtype was relatively small, with amino acid identities ranging from 97.3–100% (Table 2). In some cases, few, if any, amino acid differences were detected between the different isolates. This was true for the SRV-2F isolates from an endemically infected group of wild macaques in the same geographical area on the island of Singapore (unpublished data, R. Grant) and for the SRV-2D and SRV-2E subtypes where the different isolates came from a single primate center during the same time period and often consisted of macaques which were related through the dame or sire. The SRV-2E subtype included an isolate from A94040, a cynomolgus macaque that came to the Washington NPRC from Texas and was SRV-2 positive at that time, as well as isolates from descendents or siblings of descendents sharing the same sire. The isolates from A94040 and her child, M96026, were identical in sequence, while only 1–3 amino differences were observed with isolates from her child's half-siblings, M95348 and M96020 who shared the same sire. The SRV-2D subtype isolates from pig-tailed macaques F91249 and F90346 were identical and varied at only one amino acid position from F89336, which shared the same sire. While the isolates from F91249 and F90346 were amplified directly from peripheral blood leukocytes, the isolate from F89336 was obtained from an uncloned Raji cell tissue co-culture which had been maintained in A549 cells since 1996. Isolate F89336 serves as a reference strain within the Washington NPRC.

In the other subtypes, isolates obtained from different primate centers, from different macaque species and at different time periods were remarkably similar. In subtype SRV-2A, the original SRV-2 prototype, D2/CEL/OR, isolated by Raji cell co-culture of PBMC from a Celebes black macaque at the Oregon NPRC in the early 1980's, had only eight amino acid differences with an isolate (NM101) obtained 20 years later from a tissue sample of a cynomolgus macaque from the Lovelace Respiratory Research Institute in New Mexico. An additional SRV-2A isolate was obtained in the mid 1990's from a tissue sample of a rhesus macaque in a primate center at the University of Michigan. Although only the C-terminal half of this sequence was obtained, significant similarity with the other two SRV-2A isolates was noted. In subtype SRV-2B, two isolates (M78114, T82422) from the early 1980's, sequenced directly from PBMCs of colony-born pig-tailed macaques at the Washington NPRC, were identical in sequence. A third isolate (90167) obtained from PBMCs of a pig-tailed macaque at the same site ten years later varied by only one amino acid. This later macaque was captured in Indonesia and transferred to the Washington NPRC, suggesting that it became infected with the SRV-2B subtype already present in the colony. An SRV-2B isolate obtained from RF tissue of a rhesus macaque at the Yerkes NPRC in 1991 had only one amino acid difference compared to the M78114 and T82422 isolates from pig-tailed macaques obtained in 1984 at the Washington NPRC. The SRV-2B prototype, D2/RHE/OR, and the closely related D2/RHE/OR/V1, which were obtained by Raji-cell co-culture from PBMCs, contained four and nine amino acid differences, respectively, with the Washington isolates. Two closely-related SRV-2D isolates were obtained from the NIH primate center in Bethesda, MD. The SRV-2D prototype was obtained in 1996 from RF tissue of pig-tailed macaque 442N while the 17915 isolate was obtained from PBMCs from another NIH pig-tailed macaque. These two sequences varied at four amino acid positions. A partial env sequence was obtained in 1997 from PBMCs from a cynomolgus macaque (91048) which had been transferred to the Washington NPRC from Indonesia. This sequence varied from the 442N prototype by five amino acids within the c-terminal half.

Pairwise comparisons of the different subtype env sequences revealed amino acid conservations ranging from 96.7% between subtypes A and E and between subtypes B and D, to 93.6% between subtypes E and F (Table 2). Seventy-two amino acid positions (13% of the entire sequence) showed differences in at least one of the sequences analyzed, while fifty-five of these differences occurred in more than one of the sequences. A comparison using the seventy-two variant positions visually demonstrated the basis of the sequence differences between the six subtypes (Figure 3). Some of these sequence differences were conserved within a specific subtype. For example, at amino acid position 33 an isoleucine (I) was conserved in all of the SRV-2B isolates, while a leucine (L) was conserved in all of the SRV-2E isolates (Figure 3). On the other hand, some amino acids were conserved across subtypes, as seen in the SRV-2A, SRV-2C, SRV-2D, and SRV-2F sequences which all contained a methionine (M) at aa33. Frequently, there were single amino acid differences in one isolate within a subtype which were not conserved in the other isolate sequences, i.e. glycine (G) at aa29 in the M95348 isolate of subtype SRV-2E. Due to the fact that in most cases PCR amplification products were sequenced directly, without cloning, these differences would not reflect Taq polymerase errors.

In order to determine the genetic variation of SRV-2 within the same animal, multiple clones of a PCR amplification product encoding a 439 aa fragment of the env gene were characterized from two different macaques (F91249, T82422). Eight different clones were obtained from each animal. Sequence analysis of each clone revealed random nucleotide differences between each cloned DNA within an animal (data not shown). However, no nucleotide difference occurred in more than one clone, suggesting that the observed differences were the result of errors induced by Taq polymerase during the PCR amplification step. These data revealed no evidence for the presence of multiple strains of SRV-2 within a single individual.

We further analyzed the genetic variation within an SRV-2 strain from an individual macaque over time. Two tissue samples from macaque M95332 which contained an SRV-2E subtype had been collected at two time points six years apart, thus spanning the evolution of this virus over the time frame from 1997–2003. The complete env gene was amplified from each of these samples and compared. No differences were detected between the two sequences. Furthermore, we compared the isolate in M95332 with the isolates from two animals in the same cohort, M96020 and M96026, and the mother of M96026, A94040, who presumably introduced this SRV-2E isolate into the Washington NPRC colony nine years before the last M95332 isolate was sequenced. This comparison revealed only one amino acid difference between the sequence of M95332 and those of M96020, M96026 or A94040, showing very little variation within this SRV-2E strain even between different animals.

The SRV-2 env gene encodes a precursor polypeptide that undergoes both glycosylation and proteolytic processing during a maturation process that results in the expression of the mature membrane-bound glycoprotein integrated within the virion envelope. The envelope glycoprotein interacts with host cellular receptors to initiate virus adsorption and penetration, and plays an important role in determining cell and tissue tropism. Thus, sequence variation in the env gene can ultimately affect or determine the pathogenic potential of the virus. Interestingly, our data showed that 87% of the amino acids within the 574 amino acid env sequence exhibited no variation in the different SRV-2 isolates examined (Figures 2 and 3). Within the variant amino acid positions, in most cases, only conservative changes were identified. The conservative nature of the env gene variability was further highlighted by the finding that, in many of the positions, the variant amino acid in the SRV-2 isolate was an amino acid found in either or both of the more distantly related SRV-1 or SRV-3 serogroup prototypes. For example, leucine (L) at aa193 was found only in one SRV-2B subtype isolate but in both SRV-1 and SRV-3 prototypes. While for the most part the variant positions were scattered evenly throughout the env sequence, several highly conserved regions were identified. The N-terminal region aa60-aa95 was completely conserved between all SRV-2 isolates examined (Figure 2).

Interestingly, this region was quite distinct from the homologous regions in SRV-1 and SRV-3. The C-terminal region from the putative N-linked glycosylation site at aa345 to the C-terminus was extremely well conserved among the SRV-2 isolates with only an occasional amino acid variant. This conserved region within the SRV-2 isolates is homologous to regions within the SRV-3 gp22/20 protein where several domains have been studied in detail. Such domains include the gp70/gp20 proteolytic site (aa382) 25, the known fusion domain (aa384-aa410) 26, the heptad repeat region (aa409-aa462) 27, the immunosuppressive peptide (aa443-aa477) 2528, a membrane spanning domain (aa516-532) 15, and the region spanning the gp22 processing site (aa558) 25(see Figure 2). In contrast to the conserved N-terminal region, the C-terminal region was highly conserved not only between SRV-2 subtypes but also with the SRV-1 and SRV-3 sequences, with one region (aa419-aa485) completely conserved between the isolates from the different SRV serogroups. All twenty-two cysteine (C) residues were conserved between SRV-2 isolates and the SRV-1 and SRV-3 prototypes (Figure 3) indicating that all SRV serogroups maintain the same basic disulfide-linked three-dimensional env structure. SRV-1 and SRV-3 sequences contained three additional conserved cysteine residues which were not found in the SRV-2 isolates. Eleven putative N-linked glycosylation sites (NXS/T) were conserved in all the SRV-2 isolates and the SRV-1 and SRV-3 prototypes.

To further explore the conservation detected in the C-terminal region of the different SRV-2 subtypes and the SRV-1 and SRV-3 serotypes, we searched the existing structural databases for similar three-dimensional structures with 3D-PSSM 29 using the D2/CEL/OR subtype 2A sequence as probe. 3D-PSSM is a program that uses a threading algorithm to map the input sequence onto known 3-dimensional structures based on several parameters including amino acid sequence, multiple alignments, and secondary structure predictions. A region of the SRV-2 env polypeptide between amino acids 420–485 that was completely conserved in all SRV-2 isolates and in the SRV-1 and 3 prototypes was found to have strong structural homology to the C-terminal domains in the envelope proteins of two other retroviruses, Moloney murine leukemia virus (MMLV) and human T-lymphotropic virus (HTLV-1), as well as to a domain in the envelope protein of Ebola virus, a ssRNA virus. The known crystal structures of the MMLV, HTLV-1 and Ebola envelope proteins revealed that these domains form a highly conserved hairpin loop structure stabilized by a disulfide bond 303132. This loop structure is believed to be responsible for viral fusion with cellular membranes in several virus species, some of which share little or no obvious evolutionary relationship. Such viruses include the above mentioned oncogenic retroviruses, the orthomyxovirus influenza 33, the lentiviruses HIV-1 and SIV 3435, the paramyxovirus SV5 3637, and filoviruses 38. Three-dimensional structural predictions, using the Cn3D structure viewing program, revealed an almost perfect alignment between the structure predicted for the SRV-2 domain and the known crystal structures of the other proteins, even though only nine amino acids in a 45 aa stretch were conserved (Figure 4). The region corresponding to the conserved domain within the SRV-2 isolates has been previously identified as an immunosuppresive peptide in several oncogenic retroviruses capable of inducing an immuosuppressed state in their hosts (FeLV, MuLV, REV-A) 39. In addition, the entire gp20 protein of SRV-3 has also been found to have immunosuppressive properties 40. Studies with SRV-3 have revealed that residues in the immunosuppressive peptide found within the gp20 protein subunit are responsible for binding to the gp70 protein subunit and are crucial for virus-cell fusion 2528.

Tissue samples were obtained from a variety of macaque species naturally infected with SRV-2 that were housed at different primate research centers in the U.S. over the last twenty years. Some of these macaques were asymptomatic, while others had been diagnosed with SAIDS-RF (see Table 1). The tissue samples included RF tissue from RF positive animals, tongue, spleen, tonsils and PBMCs or PBMC/Raji cell co-cultures. Samples had been collected over a wide time-range (1983–2003), starting at the height of the SRV-2 epidemic in 1982–83. Macaque species included pig-tailed macaques (Macaca nemestrina (Mne)), Celebes black macaques (Macaca nigra (Mni)), cynomolgous macaques (Macaca fascicularis (Mfa)) and rhesus macaques (Macaca mulatta (Mmu)). Tissue samples were obtained from various primate research centers including the Washington NPRC (Seattle, WA; C.-C. Tsai and M.E. Thouless), Yerkes NPRC (Atlanta, GA; Harold McClure), National Institutes of Health (Bethesda, MD; Riri Shibata), Lovelace Respiratory Research Center (Albuquerque, NM; Carole Emerson), University of Michigan (Ann Arbor, MI; Nina Woodford); as well as one wild cynomolgous macaque whose blood was drawn during a catch-and-release process in the forests of Singapore (University of Washington; Lisa Jones-Engel).

Frozen tissue samples or cell pellets were quickly thawed in a standard proteinase K extraction buffer containing 0.1% SDS and vortexed. Paraffin-embedded formalin fixed samples were first treated with xylenol to remove the paraffin before extraction. Samples were digested over night at 50°C and DNA was isolated by standard phenol/chloroform extraction and ethanol precipitation.

To obtain the complete sequence of the different SRV-2 env genes, template DNA was amplified by PCR using SRV-2 specific primers. Primers srv2-env a (5'-CCTGAGATCACTCCTTTTCTTTGCTCAT-3') and srv2-env1 b (5'-CCGTCATTGGCTGACCAGTTTAG-3') were used to obtain a 1,796 bp PCR product containing the complete SRV-2 env sequence. In some cases, the primers srv2-env a and srv2-env b (5'-CAGTTGAGACGGCAGTGGTT-3') were used to obtain a 1,235 bp fragment which overlapped with a 1,041 bp fragment obtained using srv2-env1 a (5'-GAGTGCTGGCTATGCTTACCATC-3') and srv2-env1 b. PCR fragments were purified with the GeneClean III Kit (Q-Biogen) and directly used as templates in a sequencing reaction.

To search for SRV-2 variants in the same animal, a 439 bp fragment of the env gene was PCR-amplified from genomic DNA of two macaques (T82422, F91249) using Platinum Taq polymerase and the primers: SRV2-env 1a (5'-GAGTGCTGGCTATGCTTACCATC-3') and SRV2-env b (5'-CAGTTGAGACGGCAGTGGTT-3'). The fragments were subcloned into the pCR-Blunt vector (Invitrogen Life Technologies, Carlsbad CA) using standard procedures. Eight clones were obtained from each animal for sequence analysis.

Sequencing was performed on an ABI model 310 automated sequencer with Prism Big Dye terminator cycle sequencing ready reaction kit with Amplitaq DNA polymerase FS (Applied Biosystems). DNA sequences were analysed using Sequencher 4.1.4 (GeneCodes).

Multiple nucleotide and encoded amino acid alignments were done with Vector NTI 9.0.0 (InforMax) based on the ClustalW algorithm (EMBL, Heidelberg, Germany) or with the ClustalW program 55 directly. The location of the signal peptide was determined by SignalP 3.0 56, SigCleave 57, and iPSORT 58. The location of transmembrane domains was determined with MEMSAT2 59 and Tmpred 60. Putative glycosylation sites were determined with NetNGLyc 1.0 61.

The program 3D-PSSM 62 was used to identify proteins with structural similarities to the SRV2 env prototype (D2/CEL/OR). The structures for the six highest scoring proteins were obtained from the Molecular Modeling Database (MMDB)63 which contains experimentally determined three-dimensional biomolecular structures obtained from the Protein Data Bank. Matching structures were aligned with SRV2 env using NCBI's 3D-structure viewer Cn3D v4.1 64. Using NCBI's vector alignment search tool (VAST)65, we determined further protein structure neighbors by direct comparison of 3-dimensional protein structures stored in MMDB. This allowed us to further evaluate possible structural and functional domains of the SRV2 env protein.

Multiple sequence alignments were performed with ClustalW 55. Phylogenetic analysis of amino acid sequences was done with protein maximum-likelihood (ProML) method from the PHYLIP package version 3.62 of phylogenetic analysis programs (.)66 (University of Washington, Seattle). Bootstrap analysis was performed using the programs Seqboot and Consense from the PHYLIP package. The treefiles were displayed with TreeView 67. Identical tree topologies were obtained by using neighbor-joining analysis of the protein distance matrices and by a parsimony method as implemented in the PHYLIP package. All of the major branch points were strongly supported by boot-strap analysis and statistical evaluations within ProML.