Comparative Characterization of Transfection- and Infection-Derived Simian Immunodeficiency Virus Challenge Stocks for In Vivo Nonhuman Primate Studies

Abstract

Simian immunodeficiency virus (SIV) stocks for in vivo nonhuman primate models of AIDS are typically generated by transfection of 293T cells with molecularly cloned viral genomes or by expansion in productively infected T cells. Although titers of stocks are determined for infectivity in vitro prior to in vivo inoculation, virus production methods may differentially affect stock features that are not routinely analyzed but may impact in vivo infectivity, mucosal transmissibility, and early infection events. We performed a detailed analysis of nine SIV stocks, comprising five infection-derived SIVmac251 viral swarm stocks and paired infection- and transfected-293T-cell-derived stocks of both SIVmac239 and SIVmac766. Representative stocks were evaluated for (i) virus content, (ii) infectious titer, (iii) sequence diversity and polymorphism frequency by single-genome amplification and 454 pyrosequencing, (iv) virion-associated Env content, and (v) cytokine and chemokine content by 36-plex Luminex analysis. Regardless of production method, all stocks had comparable particle/infectivity ratios, with the transfected-293T stocks possessing the highest overall virus content and infectivity titers despite containing markedly lower levels of virion-associated Env than infection-derived viruses. Transfected-293T stocks also contained fewer and lower levels of cytokines and chemokines than infection-derived stocks, which had elevated levels of multiple analytes, with substantial variability among stocks. Sequencing of the infection-derived SIVmac251 stocks revealed variable levels of viral diversity between stocks, with evidence of stock-specific selection and expansion of unique viral lineages. These analyses suggest that there may be underappreciated features of SIV in vivo challenge stocks with the potential to impact early infection events, which may merit consideration when selecting virus stocks for in vivo studies.

INTRODUCTION

Repeatable experimental inoculation of naïve animals via a known route using a defined amount of titered, laboratory-generated cell-free virus of known identity is a key feature of in vivo nonhuman primate (NHP) models of AIDS. Because HIV-1 does not productively infect the monkey species used in NHP AIDS models (1), these experiments have relied on the use of the genetically related simian immunodeficiency viruses (SIV) from sooty mangabeys or other Old World primates, with viruses from the SIVmac251 and SIVmac239 lineages serving as the most widely used in studies of viral transmission, dissemination, evolution, and pathogenesis, as well as studies of antiviral immune responses and vaccine-mediated protection (2). There are a number of potential methods for generating primate lentiviruses; however, SIV stocks used for in vivo NHP studies are typically produced in one of two ways: the viruses can be produced by transfection of nonpermissive 293T cells if a molecularly cloned full-length viral genome is available, or, alternatively, cloned viruses and uncloned viral isolates can be produced by infection of permissive primary cells or cell lines with subsequent virus expansion via multiple rounds of viral replication.

Viruses produced in transfected nonpermissive cells package full-length viral genomes that have been transcribed directly from an input DNA plasmid, and they are therefore only subject to the low error rate of cellular DNA-dependent RNA polymerase II. This production method thus offers the advantage of virtual intrastock clonality, thereby allowing for in vivo infections using a virus with a precisely known genetic composition. In addition, different transfection-derived stocks of the same virus will have the same genetic sequence and as a result may exhibit more consistent behavior from stock to stock.

While production of virus in transfected cells requires a full-length molecularly cloned viral genome, infection-based production methods obviate infectious molecular clones and can be used to expand uncloned viral swarms as well as clonal virus stocks generated in transfected cells. Infection-based virus production methods also present several other potential advantages, including greater ease of virus scale-up and the ability to avoid xenoreactivity to producer cell proteins in challenged animals by culturing the challenge virus in challenge host-specific cells, a feature which has proven important for the interpretation of some vaccine protection studies (3). Because they are propagated via multiple rounds of viral replication with reverse transcription errors, infection-derived SIV stocks may contain a large degree of genetic diversity within each stock (4–6). For some experiments, added genetic diversity in the virus challenge stock may be more desirable than stock clonality (7); however, the degree of diversity between different infection-derived stocks of nominally the same virus has not been carefully examined.

These canonical stock characteristics are often considered when choosing an SIV stock for use in vivo in a particular NHP study; however, matched transfection- and infection-derived challenge stocks have not been directly compared, nor have different stocks of the same nominal virus produced in different labs using similar but distinct methodologies. There is also a perception within the NHP AIDS field that infection-derived stocks may be more infectious for in vivo challenges, but this phenomenon has not been formally demonstrated and the specific stock features that may be responsible have not been evaluated. Although the titers of SIV challenge stocks are determined in vitro prior to use in vivo, there may be aspects of SIV stocks that do not affect standard in vitro titer determination assays, such as the presence of proinflammatory cytokines or selection and expansion of specific sequences during culture, which could impact in vivo infectivity at the earliest stages of infection, particularly in the case of mucosal challenges. It is also unclear if different production methods yield viruses with equivalent levels of envelope (Env) glycoprotein incorporation and similar particle/infectivity ratios or how the target cell type used for a titer determination assay might impact the determined titer. Given the cost and scientific value of in vivo NHP AIDS models, we examined a panel of SIV challenge stocks generated in several different labs by both transfection and infection-based methods. Our detailed characterization of several of the more widely used in vivo NHP study SIV challenge stocks reveals key distinctions between transfection- and infection-derived virus stocks and between different infection-derived stocks of the same nominal virus and may provide insight to better inform selection of current SIV challenge stocks and to improve production of SIV challenge stocks in the future.

MATERIALS AND METHODS

Virus stocks.

251-RD (SIVmac251-7/9/2010) was generated by infecting rhesus macaque peripheral blood mononuclear cells (PBMCs) from 15 donors with an animal-titered stock of SIVmac251 (4-19-91) (8) 3 days after stimulation with 1.0 μg/ml phytohemagglutinin (PHA). Infected cultures were maintained in 10% interleukin-2 (IL-2) with a complete medium change 24 h prior to harvest. At day 8 postinfection, virus-containing supernatants were pooled, clarified by centrifugation, aliquoted, and stored in the vapor phase of liquid nitrogen.

251-CM (SIVmac251-06/2004) (9) was generated by infecting rhesus macaque PBMCs from 6 donors 5 to 7 days after stimulation with 0.5 μg/ml staphylococcal enterotoxin A (SEA) with an SIVmac251 seed stock (1998) (10) grown for 3 to 4 days in CEMx174 cells. Infected cultures were maintained in 50 to 100 U/ml IL-2, with 80% of the virus-containing supernatant volumes collected, clarified by centrifugation, aliquoted, and stored at −80°C every 2 to 3 days postinfection. Following each supernatant harvest, infected PBMCs were resuspended in the remaining supernatant volume plus fresh medium with IL-2. Supernatants from 2 PBMC donors collected on multiple harvest days containing the highest viral p27 content were thawed, pooled, aliquoted, and stored in the vapor phase of liquid nitrogen.

251-DB (SIVmac251-8/27/2008) (11) was generated by infecting human PBMCs from a single donor with SIVmac251 after overnight stimulation with 6.25 μg/ml concanavalin A (ConA). Infected cultures were maintained in 20 U/ml IL-2 with a complete medium change 24 h prior to virus harvest. Virus-containing supernatants collected on days 4 and 7 postinfection were clarified by centrifugation, sterile filtered, pooled, aliquoted, and stored at −80°C.

251-LG (SIVmac251) (12) was generated by infecting rhesus macaque PBMCs with an animal-titered stock of SIVmac251 (4-19-91) (8) 3 days after stimulation with 5.0 μg/ml PHA and culturing infected cells for 21 days after stimulation. The infected culture was maintained in 50 U/ml IL-2, with a half-volume medium change twice a week. Removed medium was tested for SIV p27, and supernatants with a high SIV concentration were pooled (supernatant for days 14, 18, and 21 postinfection). Virus-containing supernatants were clarified by centrifugation, sterile filtered, aliquoted, and stored at −80°C.

251-PM (SIVmac251-06/25/2004) (13) was generated by infecting CEMx174 cells with a SIVmac251 stock (11-19-1995) (14, 15) that had been previously expanded on HUT-78 (16) and then CEMx174 (17) cells. A complete culture medium change was performed on day 20 postinfection. On day 21 postinfection, virus-containing supernatants were collected, clarified by centrifugation, sterile filtered, aliquoted, and stored at −80°C.

766-BK (SIVmac766, a full-length transmitted/founder clone derived from an SIVmac251-infected rhesus macaque) was generated by infecting pooled rhesus macaque CD8-depleted PBMCs from 8 donors with transfected-293T-produced SIVmac766 3 days after stimulation with 2.5 μg/ml PHA. Infected cultures were maintained in 28 U/ml IL-2, and freshly activated CD8-depleted PBMCs were added to the cultures at 3 days postinfection. At day 7 postinfection, virus-containing supernatants were clarified by centrifugation, sterile filtered, aliquoted, and stored in the vapor phase of liquid nitrogen.

766-TFN-BK (SIVmac766) was generated by transfecting 293T cells with the full-length molecular clone plasmid pSIVmac766.4 for 24 h using the TransIT HEK-293 transfection reagent (Mirus Bio) according to the manufacturer's instructions. Culture medium was changed at 24 h posttransfection and again at 48 h posttransfection. At 72 h posttransfection, virus-containing supernatant was clarified by centrifugation, aliquoted, and stored at −80°C.

239-CM (SIVmac239) (18) was generated as described above for 251-CM, except that stimulated rhesus macaque PBMCs were infected with a SIVmac239 seed stock produced in transfected 293T cells.

239-TFN-RD (SIVmac239) was generated by transfecting 293T cells with the full-length molecular clone plasmid p239-FL for 24 h using the ProFection mammalian transfection system calcium phosphate method (Promega) according to the manufacturer's instructions. Culture medium was changed at 24 h posttransfection. At 72 h posttransfection, virus-containing supernatant was clarified by centrifugation, aliquoted, and stored in the vapor phase of liquid nitrogen.

Virus for protein analysis and infectivity thermostability.

Dedicated virus stocks distinct from those described above were generated for protein analysis and infectivity thermostability experiments. Transfection-derived SIVmac239 and SIVmac766 stocks were generated by transfecting 293T cells with the full-length molecular clone plasmid p239SPxFL or pSIVmac766.4 for 24 h using the TransIT HEK-293 transfection reagent (Mirus Bio) according to the manufacturer's instructions. Culture medium was changed at 24 h posttransfection and again at 48 h posttransfection. Infection-derived SIVmac239 and SIVmac766 were generated by infecting pooled rhesus macaque CD8-depleted PBMCs from 12 donors with transfected-293T-produced virus 3 days after stimulation with 2.0 μg/ml PHA. Infected cultures were maintained in 100 U/ml IL-2 with a complete medium change 24 h prior to virus harvest. At day 3 posttransfection or day 8 postinfection, virus-containing supernatants were clarified by centrifugation and sterile filtered. A portion of the filtered supernatants was then aliquoted and stored at −80°C for use in infectivity experiments. To purify and concentrate virus for protein analysis, 30 ml of filtered supernatant was centrifuged through a 30% sucrose cushion, pelleted, and washed as previously described (19). The purified virus pellets were then resuspended in 15 μl PBS for a final concentration factor of 2,000×.

Virus content.

Viral RNA content for each virus stock was determined by running each stock in a quantitative real-time reverse transcriptase PCR (qRT-PCR) assay performed as previously described (20). Each stock was assayed at 3 dilutions, and the means of all 3 dilutions were determined. Viral p27 (CA) content was determined by assaying 8 dilutions of each virus stock in parallel using a commercial SIV p27 antigen capture assay according to the manufacturer's instructions (ABL). At least 3 dilution-corrected p27 values within the linear range of the virus dilution series were averaged to determine the p27 content of each virus stock.

Virus titer.

Virus titers were determined on primary rhesus macaque cells as previously described (21). Briefly, CD8-depleted PBMCs from 3 SIV-naïve rhesus macaques were stimulated with plate-bound anti-CD3 (BD) for 3 days, individually plated at 10⁵ cells/well in 96-well plates, and infected in triplicate with serial 5-fold dilutions of virus. Cells were washed 3 times with PBS 24 h postinfection to remove residual input virus and were then maintained in 100 U/ml IL-2 for 3 weeks. Cell-free culture supernatants were then collected, and a SIV p27 antigen capture assay was used to detect the presence of viral p27 antigen according to the manufacturer's instructions (ABL). The 50% tissue culture infectious dose (TCID₅₀) was then calculated using the Reed-Muench accumulative method (21). Virus titers were also determined on TZM-bl reporter cells (22–24) by counting the number of β-galactosidase-expressing cells per well in triplicate wells infected with serial 3-fold dilutions of virus as previously described (25). Wells containing dilution-corrected blue cell counts within the linear range of the virus dilution series were averaged to generate an infectious titer in infectious units (IU) per ml. For infectivity thermostability experiments, individual 1-ml aliquots of titered frozen virus generated as described above were removed from −80°C and placed in a 37°C water bath. Vials of virus were left at 37°C for a range of times between <2 min (just long enough to thaw) and 8 h and then immediately diluted in DMEM containing 10% fetal bovine serum (FBS), 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 1 μg/ml Polybrene by a dilution factor that yielded 250 IU/0.1 ml for freshly thawed virus, which is within the linear range of the virus infectivity dilution series for all of the viruses used in this study, and then added to TZM-bl cells in triplicate at 0.1 ml per well. β-Galactosidase-expressing cells were then counted 48 h postinfection, as described above.

Protein analysis.

Quantitative measurements of viral p27 (CA) and gp120 protein in virions for determinations of gag/env ratios and estimations of Env trimer spikes per virion were determined using a dual-color fluorescent protein gel analysis as previously described (26, 27). Well-characterized reference preparations of infection-derived HIV-1 BAL, HIV-1 NL4-3, and SIVmac239 cultured in human SupT1-CCR5 T lymphoblastoid cells (provided by the Biological Products Core, AIDS and Cancer Virus Program, SAIC-Frederick, Frederick National Laboratory for Cancer Research, Frederick, MD) were included in the analysis as controls.

SGA.

For single-genome amplification (SGA), for each sample, at least 200,000 viral RNA copies were extracted using the QIAamp viral RNA minikit (Qiagen). RNA was eluted and immediately subjected to cDNA synthesis. Reverse transcription of RNA to single-stranded cDNA was performed using SuperScript III reverse transcriptase according to the manufacturer's recommendations (Invitrogen). In brief, a cDNA reaction of 1× RT buffer, 0.5 mM each deoxynucleoside triphosphate, 5 mM dithiothreitol, 2 U/ml RNaseOUT (RNase inhibitor), 10 U/ml of SuperScript III reverse transcriptase, and 0.25 mM antisense primer SIVEnvR1 (5′-TGT AAT AAA TCC CTT CCA GTC CCC CC-3′; or SIV-R-R1 [5′-CAC TAG CTT ACT TCT AAA ATG GCA GC] for near-full-length [NFL] PCR) was incubated at 50°C for 60 min, 55°C for 60 min and then heat inactivated at 70°C for 15 min followed by treatment with 1 U of RNase H at 37°C for 20 min. Env gene or NFL genomes were then amplified via limiting dilution PCR, where only one amplifiable molecule was present in each reaction. Env PCR amplification was performed with 1× PCR buffer, 2 mM MgCl₂, 0.2 mM each deoxynucleoside triphosphate, 0.2 μM each primer, and 0.025 U/μl Platinum Taq polymerase (Invitrogen) in a 20-μl reaction. First-round PCR was performed with sense primer SIVEnvF1 (5′-CCT CCC CCT CCA GGA CTA GC-3′) and antisense primer SIVEnvR1 (5′-TGT AAT AAA TCC CTT CCA GTC CCC CC-3′) under the following conditions: 1 cycle of 94°C for 2 min and 35 cycles at 94°C for 15 s, 55°C for 30 s, and 72°C for 4 min, followed by a final extension of 72°C for 10 min. Next, 1 μl from the first-round PCR product was added to a second-round PCR that included the sense primer SIVEnvF2 (5′-TAT AAT AGA CAT GGA GAC ACC CTT GAG GGA GC-3′) and antisense primer SIVEnvR2 (5′-ATG AGA CAT RTC TAT TGC CAA TTT GTA-3′) performed under the same conditions used for first-round PCR, but with a total of 45 cycles. NFL PCR was performed similarly, but with unique primers and an increase in extension time from 4 min to 10 min. The NFL first-round PCR primers were SIVU5F1 (5′-AGA AGT AAG CTA GTG TGT GTT CCC ATC TCT) and SIV-R-R1 (5′-CAC TAG CTT ACT TCT AAA ATG GCA GC), while the second-round PCR primers were SIVU5F2 (5′-AGC TAG TGT GTG TTC CCA TCT CTC CTA) and SIV-R-R2 (5′-TAC TTC TAA AAT GGC AGC TTT ATT GAA). Amplicons of the correct size were identified by agarose gel electrophoresis and directly sequenced with second-round PCR primers and SIV-specific primers using BigDye Terminator technology. To confirm PCR amplification from a single template, chromatograms were manually examined for multiple peaks, indicative of the presence of amplicons resulting from PCR-generated recombination events, Taq polymerase errors, or multiple variant templates. Sequences were aligned using ClustalW and hand edited using MacClade 4.08 to improve alignment quality. All trees were constructed by the neighbor-joining method.

Roche 454 pyrosequencing.

Pyrosequencing of viral stocks was done, essentially, as described previously (28). Viral RNA was isolated using the QIAamp MinElute virus spin kit (Qiagen). RNA was then reverse transcribed and amplified in four overlapping amplicons using SIV-specific primers and the Superscript III one-step RT-PCR system with Platinum high-fidelity Taq (Invitrogen). Conditions for the reverse transcription-PCR were as follows: 50°C for 60 min; 94°C for 2 min; 2 cycles of 94°C for 15 s, 60°C for 1 min, and 68°C for 4 min; 2 cycles of 94°C for 15 s, 58°C for 1 min, and 68°C for 4 min; 40 cycles of 94°C for 15 s, 55°C for 1 min, and 68°C for 4 min; and 68°C for 10 min. Each amplicon covered approximately 2.5 kb, and together they spanned the entire viral genome coding sequence. PCR products were gel purified and quantitated using the Qubit double-stranded DNA (dsDNA) HS assay kit (Invitrogen). A total of 12.5 ng of each amplicon was combined to produce 50-ng pools specific to a single viral stock, which were then fragmented using the Nextera DNA sample prep kit (Roche Titanium-compatible; Epicentre) according to the manufacturer's instructions. Following fragmentation, multiplex identifier (MID) tags and adapters for Roche 454 pyrosequencing were added. Products were then cleaned twice with Agencourt AMPure XP beads (Beckman Coulter Genomics). The purified, tagged products were quantitated using the Qubit dsDNA HS assay kit, and their average size was characterized using a high-sensitivity DNA bioanalyzer chip on a 2100 bioanalyzer (Agilent Technologies). Pyrosequencing was performed on a Roche 454 GS Junior instrument according to the manufacturer's protocols (454 Life Sciences).

Roche 454 pyrosequencing data analysis.

SFF files from pyrosequencing were imported into Geneious Pro software version 5.5 (Biomatters). Reads were then separated by MID tag and trimmed to remove the Roche 454 adapter, MID, and transposon sequences as well as low-quality sequence (error probability limit of 0.005). SIVmac239 reads were assembled to a GenBank reference (accession no. M33262), while SIVmac251 and SIVmac766 reads were assembled to their corresponding consensus sequence generated by SGA. Single nucleotide polymorphisms (SNPs) relative to each stock's consensus sequence were found using the SNP calling tool in Geneious Pro by applying the following settings: minimum coverage of 50, minimum variant frequency of 5%, and maximum variant P value of 10⁻⁶⁰. Amino acid variation present in CD8⁺ T cell epitopes was analyzed using a custom data analysis pipeline, as previously described (29). Briefly, FASTQ files extracted from the original SFF file were trimmed in order to remove MID, adapter, and transposon sequences. Low-quality (quality of <19) sequences were masked with an “N” before alignment. FASTQ sequences were then converted to FASTA format, translated in all six reading frames, and aligned against reference SIVmac239 proteins using the BLAST-Like-Alignment-Tool (BLAT). Sequences aligning to epitopes of interest were extracted from the BLAT output table using custom scripts and analyzed for amino acid substitutions. Only variants present at a frequency of 5% or greater were included in our analysis.

Luminex.

Cytokine concentrations in viral preparations were determined by a NHP-specific Luminex assay that detected 36 cytokines, as described previously (30–32). These cytokines included granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage CSF (GM-CSF), GRO-α, alpha interferon (IFN-α), IFN-γ, interleukin-1β (IL-1β), interleukin-1 receptor a (IL-1Ra), IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13, IL-15, IL-17, IL-18, IL-23, IFN-γ-induced protein 10 (IP-10), monocyte chemoattractant protein 1 (MCP-1), macrophage-derived chemokine (MDC), macrophage inflammatory protein 1α (MIP-1α), MIP-1β, Perforin, RANTES, soluble CD40 ligand (sCD40L), soluble FAS ligand (sFASL), transforming growth factor alpha (TGF-α), TGF-β, tumor necrosis factor alpha (TNF-α), TNF-β, and vascular endothelial growth factor (VEGF).

Nucleotide sequence accession numbers.

All 120 new sequences were deposited in GenBank under accession no. KC522134 to KC522253. Reference sequences from previous studies were also included in the analysis (accession no. JQ085996 to JQ086063, GU998976 to GU99002, and JQ677771 to JQ677809).

RESULTS

To initiate these studies, we first assembled a panel of nine SIVmac251 or -239 lineage virus stocks, comprising seven infection-derived stocks and two transfection-derived stocks, generated in six different laboratories (Table 1). Five of the infection-derived stocks contained infected culture expansions of the viral swarm SIVmac251 (33); three of these stocks (251-RD, 251-CM, and 251-LG) were produced in rhesus PBMCs under various target cell stimulation and culture conditions, one was produced in activated human PBMCs (251-DB), and one was produced in the CEMx174 (17) cell line (251-PM). The remaining four stocks comprised paired rhesus PBMC infection-derived and 293T transfection-derived stocks of two molecularly derived viral clones, SIVmac239 (15) and the transmitted/founder clone SIVmac766 derived from a SIVmac251-infected rhesus macaque.

Table 1.

Virus stocks evaluated in this study

Designation	Virus	Lab	Production method	Cell type	Stimulation
251-RD	SIVmac251	R. Desrosiers	Infection	Rhesus PBMCs	PHA
251-CM	SIVmac251	C. Miller	Infection	Rhesus PBMCs	SEA
251-DB	SIVmac251	D. Barouch	Infection	Human PBMCs	ConA
251-LGa	SIVmac251	L. Giavedoni	Infection	Rhesus PBMCs	PHA
251-PMb	SIVmac251	P. Marx	Infection	CEMx174
766-BK	SIVmac766	B. Keele	Infection	Rhesus CD8-depleted PBMCs	PHA
766-TFN-BK	SIVmac766	B. Keele	Transfection	293T
239-CM	SIVmac239	C. Miller	Infection	Rhesus PBMCs	SEA
239-TFN-RD	SIVmac239	R. Desrosiers	Transfection	293T

^aIncluded in the Luminex cytokine and chemokine analysis only (Fig. 6).

^bIncluded in the SGA Env sequencing analysis only (Fig. 2).

Virus content.

To assess the virion contents of stocks produced via different methods in different laboratories, we first determined the viral RNA and p27 CA protein content in seven of the virus stocks (Fig. 1A). All of the virus stocks evaluated contained between 5.67 × 10⁸ and 1.3 × 10¹⁰ RNA copies/ml and between 50 and 876 ng p27/ml, with good agreement between the two measurements that were consistent with previous results indicating roughly 10¹⁰ viral particles per μg of CA protein for other retroviruses (34). The two 293T transfection-derived virus stocks (766-TFN-BK and 239-TFN-RD) contained the highest virion content at ∼10¹⁰ RNA copies/ml and >850 ng p27/ml, whereas the infection-derived stocks ranged between 5.67 × 10⁸ and 7.60 × 10⁹ RNA copies/ml and 50 to 585 ng p27/ml. We next evaluated infectivity by determining the titers of the stocks in parallel on the TZM-bl reporter cell line (22–24) (Fig. 1B) or on CD8-depleted PBMCs from 3 rhesus macaque donors (Fig. 1C). Although the relative differences in titers between stocks were similar for a given PBMC donor animal, there were apparent donor-dependent effects on the determined 50% tissue culture infective dose (TCID₅₀) values, with 10-fold or greater differences in calculated TCID₅₀ between two PBMC donors noted for 4 of the 7 evaluated stocks (Fig. 1C). For all 7 virus stocks, the determined TZM-bl titer was within 6-fold of the mean of the titers determined on primary cells (Fig. 1B and C).

Fig 1.

Virus content of SIV in vivo challenge stocks. (A) Quantification of viral RNA copy number by qRT-PCR and p27^CA protein content by antigen capture enzyme-linked immunosorbent assay (ELISA). RNA copy data represent the mean values for each sample assayed in duplicate at 3 different dilutions within the linear range of the dilution series. Eq, equivalents. The p27 data represent mean values for each sample at 3 or more dilutions within the linear range of the dilution series. (B) Stock infectivity titers as determined on TZM-bl reporter cells expressed as dilution normalized infectious units (IU) per milliliter. Titers represent mean values for all dilution-normalized titers that fell within the linear range of the dilution series. (C) Stock infectivity titers as determined on CD8-depleted rhesus PBMCs from 3 donors expressed as 50% tissue culture infectious dose (TCID₅₀). Mean titers for all 3 donors are also shown. TCID₅₀ values were determined by the Reed-Muench accumulative method. (D) Specific infectivity of each stock based on TZM-bl or mean primary cell titer and viral RNA copy equivalents.

The two 293T transfection-derived SIV stocks had the highest infectivity in both TZM-bl and rhesus PBMCs (Fig. 1B and C), with titers of 2.4 × 10⁶ IU/ml and 7.5 × 10⁵ IU/ml on TZM-bl for 766-TFN-BK and 239-TFN-BK, respectively. We utilized the TZM-bl and mean primary cell titers to calculate the specific infectivity for each stock to determine if there were any differences in the infectivity of viruses generated by different means on a per virion basis. As shown in Fig. 1D, the specific infectivities for each of the seven stocks were similar, ranging between 2.2 × 10⁻⁵ and 4.2 × 10⁻⁴ IU/RNA copy, with the two transfection-produced stocks at the higher end of this range, suggesting that quantitative virion content rather than qualitative differences in the viruses is largely responsible for differences in the in vitro infectious titer.

Sequence diversity and analysis.

Virus stocks generated by productive infection of permissive cells are likely to contain diverse viral sequences due to the accumulation of mutations generated by multiple rounds of viral reverse transcription; however, different stocks of the same nominal virus might contain different amounts of viral diversity and different specific viral sequences. We therefore performed single-genome amplification (SGA) sequencing analysis of the full-length Env gene or utilized published sequences of four different SIVmac251 stocks (251-RD, 251-DB, 251-CM, and 251-PM). Sequencing analysis revealed different degrees of Env diversity within each stock, with as much as 2.1% maximum sequence diversity in 251-RD and as little as 0.9% maximum sequence diversity in 251-DB (Fig. 2). There was also clear evidence for selection and expansion of unique Env sequences within each of the SIVmac251 stocks, with all of the Env sequences from 251-DB, 251-CM, and 251-PM clustered on separate branches of the phylogenetic tree with no overlap, suggesting that the predominant sequences within each SIVmac251 stock are distinct from those present in the other SIVmac251 stocks. We note that the Env sequences in two other stocks of SIVmac251, a 1991 progenitor stock (6) and a 2006 reexpansion of the 1991 stock (35), produced in the same lab as 251-RD, contained similar maximum diversity and phylogenetic tree structure, as shown here for 251-RD (not shown).

Fig 2.

SIVmac251 stock Env phylogeny. Unrooted neighbor-joining tree constructed from SGA-derived full-length Env sequences from the indicated SIVmac251 stocks, as well as the infectious molecular clones SIVmac239 and SIVmac766. Branch colors indicate the stock from which each sequence was derived. Percentages in parentheses indicate the maximum genetic diversity of Env sequences within each stock.

Differences between stocks in overall sequence diversity were also noted when we expanded our sequencing analysis to full-length viral genomes using both SGA sequencing analysis (Fig. 3, top panels) as well as 454 pyrosequencing (Fig. 3, bottom panels). In agreement with the results shown for Env sequences in Fig. 2, 251-RD (Fig. 3A) contained greater sequence diversity than 251-CM (Fig. 3B) or 251-DB (Fig. 3C) throughout the viral genome, with 1.4%, 0.9%, and 0.5% maximum sequence diversities for the three stocks, respectively. Evidence of selection and expansion of unique sequences within each infection-derived virus stock was also observed in the full-length viral sequences. To evaluate the potential functional impact of different virus genetic contents within the different SIVmac251 stocks, we examined the preexisting frequency of several well-characterized CD8⁺ T cell escape mutations in four different epitopes within 251-RD, 251-CM, and 251-DB (Table 2). Marked differences were noted between the stocks, with 251-RD containing low but detectable levels (<50%) of preexisting escape mutations in all four of the examined epitopes, whereas >80% of sequences within 251-CM contained preexisting escape mutations to the two Mamu B*17 major histocompatibility complex (MHC)-restricted epitopes, while <5% contained escape mutations to the Mamu A*01 and Mamu B*08 restricted epitopes. In contrast to 251-RD, 251-DB contained <5% escape mutations to 3 of the 4 examined epitopes, but did contain a detectable preexisting escape variant within the Env_832-840FW9 epitope.

Fig 3.

SIVmac251 stock whole-genome sequence diversity. Highlighter plots (www.hiv.lanl.gov) derived from SGA sequences (top) and polymorphism frequency plots from 454 deep sequencing (bottom) indicating point mutations relative to the stock consensus sequence across the entire viral genome are shown for 251-RD (A), 251-CM (B), and 251-DB (C). 454 deep sequencing polymorphisms detected in fewer than 5% of sequences are not shown. Numbers below highlighter plots denote the genomic nucleotide number for both SGA and 454 sequences. Corresponding genomic regions are represented below the polymorphism frequency plots. LTR, long terminal repeat.

Table 2.

Variation in epitopes restricted by common MHC I alleles^a

Epitope	MHC restriction	Amino acid sequenceb	% frequency of epitope inc:
			251-RD	251-CM	251-DB
Tat28–35TL8	Mamu A*01	TTPESANL
		S-------	29.7 (47)	—	—
Vif172–179RL8	Mamu B*08	RRDNRRGL
		---H----	7.6 (0)	—	—
Env832–840FW9	Mamu B*17	FQEAVQAGW
		-H-------	31.1 (27)	89.6 (82)	45.0 (25)
		------V--	11.1 (33)	—	—
Nef195–203MW9	Mamu B*17	MQPAQTSKW
		-H-----Q-	9.0 (13)	6.1 (9)	—
		-H----AQ-	—	82.3 (82)	—

^aOnly nonsynonymous mutations detected at a frequency of >5% by 454 deep sequencing are shown.

^bWild-type amino acid sequence derived from the 251-RD majority consensus.

^cValues in parentheses represent the frequency detected by SGA sequencing. −, absence of variant (<5% frequency).

To evaluate genetic differences between infection- and transfection-derived virus stocks as well as to assess the amount of diversity generated by infection-based production of a molecularly cloned virus, we compared the SGA-generated Env sequence for 239-TFN-RD with that of 239-CM (Fig. 4A) and the SGA-generated Env sequence for 766-TFN-BK with that of 766-BK (Fig. 4B). As expected, the two transfection-derived stocks contained clonal virus, with very rare point mutations identified, consistent with random error introduced during in vitro reverse transcription for sequence analysis. The two matched infection-derived stocks contained limited amounts of viral diversity, with the majority of Env sequences identical to each other and a small number of sequences containing between 1 and 5 random point mutations (Fig. 4).

Fig 4.

Env sequence diversity of transfection- and infection-derived molecular clone viruses. Shown are highlighter plots of SGA-derived Env sequences indicating point mutations relative to the stock consensus sequence for matched pairs of transfection- and infection-derived stocks of SIVmac239 (A) and SIVmac766 (B). Numbers below highlighter plots indicate the Env gene nucleotide number.

Virion Env content.

Because the SIV envelope (Env) glycoprotein is not required for the production of particles with an intact lipid bilayer envelope, there could be substantial producer cell-dependent variability in Env incorporation per virion. We therefore comparatively assessed the protein contents of matched infection- and transfection-derived virions. For this experiment, new SIVmac239 and SIVmac766 stocks were generated to accommodate the large amounts of concentrated viral material required for Env assessment. The stocks generated for these analyses had infectivity titers of between 3 × 10⁵ and 7 × 10⁵ IU/ml on Tzm-bl cells, with the transfection-derived stocks demonstrating 2- to 3-fold-higher overall titers (data not shown), consistent with our findings for the in vivo challenge stocks produced in other laboratories (Fig. 1). These new viruses, produced in infected CD8-depleted rhesus PBMCs or in transfected 293T cells were purified, concentrated 2,000×, and examined by SDS-PAGE using a dual-color fluorescent imaging method (Fig. 5A) (26, 27). Virion p27^CA and gp120^SU contents were quantitated using linear regression analysis by generating densitometry standard curves with dilution series of purified p27^CA and gp120^SU of known concentration (Fig. 5A and B). Total protein was stained using SYPRO Ruby, allowing for determination of p27^CA content, and glycoprotein was stained with SYPRO Pro-Q Emerald, allowing for determination of gp120^SU content. As shown in Fig. 5A, the 293T-generated stocks contained markedly lower levels of p27^CA-normalized gp120 than the matched infection-derived stocks. Using these data, we were able to calculate the molar Gag/Env ratio for each virus preparation and, based on an estimated 1,400 gag molecules per particle (26, 36), estimate a mean number of trimer spikes per virion. Each infected-PBMC-generated stock contained ∼13- to 15-fold more Env trimers than the matched transfection-generated stocks, with SIVmac239- and SIVmac766-producing PBMCs containing 9 and 6 trimers per virion, respectively, while the two transfection-derived stocks contained less than 1 trimer per virion on average. In agreement with previously published results for HIV-1 virus stocks (27), these results indicate that transfected-293T-cell-produced SIV stocks contain fewer Env trimer spikes per virion than infection-produced stocks.

Fig 5.

Virion-associated Env content analysis. (A) SDS-PAGE gel stained with two SYPRO dyes to detect and quantitate glycoprotein (green) and total protein (red). Purified SIVmac239 and SIVmac766 stocks generated in infected PBMCs or transfected 293T cells were lysed and run on the same gel with well-characterized reference viruses and serial dilutions of quantified, purified gp120 and p27 for the establishment of standard curves for densitometry analysis. M.M. Stds, molecular mass standards (kDa). (B) Densitometry standard curves established using the p27^CA standards (red) and the gp120 standards (green) in panel A for linear regression analysis of the p27 and gp120 contents of the purified SIV stocks.

Thermostability of infectivity.

Following in vivo inoculation, particularly for low-dose mucosal challenges, some time may elapse before a virion finds a suitable target cell to initiate infection. Although transfection-derived SIV stocks were found to have equivalent or better specific infectivity in vitro than infection-derived stocks, despite smaller amounts of virion-incorporated Env (Fig. 1D), we sought to determine if infection- and transfection-based production methods lead to differences in the thermostability of infectivity not captured by standard titer determination assays. To address this question, we evaluated the stability of infectivity at 37°C for 1× infectious stocks of SIVmac239 and SIVmac766 harvested from the same culture supernatants used for protein analysis (Fig. 6). Individual 1-ml aliquots of frozen virus-containing supernatant were placed at 37°C for various amounts of time prior to being added to TZM-bl cells. Although some small differences were noted in the thermostability of the virus stocks, with rhesus PBMC-produced SIVmac239 showing the greatest reduction in infectivity after 8 h at 37°C, there were no consistent differences between the infection- and transfection-produced viruses, and all of the stocks maintained 60% or greater infectivity, with 3 of 4 stocks maintaining >80% of infectivity after 8 h at 37°C.

Fig 6.

Thermostability of infectivity. Shown are the percentages of infectivity of SIVmac239 and SIVmac766 stocks generated in infected PBMCs or transfected 293T cells measured on TZM-bl cells after incubation at 37°C for the indicated amounts of time. The percentage of infection for each virus was normalized to 100% for freshly thawed virus calculated by determining the mean number of infected cells in triplicate wells. The values shown are means of two independent experiments.

Cytokine and chemokine content.

In addition to virion-incorporated proteins, nonviral soluble factors present in virus stocks could potentially modulate initial infection events. In particular, the presence of cytokines and chemokines could potentially impact the recruitment, identity, and activation state of early cellular targets. We examined the cytokine and chemokine contents of both infection- and transfection-derived stocks using a 36-plex Luminex assay with cross-reactivity for human and rhesus analytes (30–32). As shown in Fig. 7, the infection-derived SIV stocks contained a markedly greater abundance and variety of cytokines and chemokines than did the two transfection-derived stocks (766-TFN-BK, 239-TFN-RD). Although the infection-derived stocks contained consistently higher levels of cytokines and chemokines, there was a large degree of variability in the specific content of each stock. While the hierarchical clustering of 239-CM and 251-CM, two different stocks generated in the same lab, suggest that cell type, stimulation, and culturing techniques may lead to some degree of predictability in the cytokine and chemokine content of a stock, there were still substantial differences between these two stocks, including ∼73-fold-greater levels of IL-10 and ∼2-fold-greater levels of IFN-γ in 251-CM (Fig. 7; see Table S1 in the supplemental material).

Fig 7.

Cytokine/chemokine content of SIV in vivo challenge stocks. A heat map shows the results of a quantitative 36-plex Luminex assay with cross-reactivity for human and rhesus macaque analytes. A clustergram shows hierarchical clustering of virus stocks based on similarities in cytokine/chemokine content.

In the transfection-derived virus stocks, only TGF-β and VEGF were present at >100 pg/ml, with 988 and 660 pg/ml TGF-β and 2,124 and 4,225 pg/ml VEGF detected in 766-TFN-BK and 239-TFN-RD, respectively (Fig. 7; see Table S1 in the supplemental material). All other cytokines and chemokines were at relatively low or undetectable levels in these two stocks. In contrast, each infection-derived stock contained at least 9 analytes at >100 pg/ml and at least 1 analyte at >1,000 pg/ml, with 5 of 6 infection-derived stocks containing 3 or more analytes at >1,000 pg/ml. Perforin was present at >1,000 pg/ml in all of the infection-derived stocks, except 251-RD, with 251-DB containing the highest concentration at 12,444 pg/ml. Several other analytes were detected at notably high levels, including CD14 at 106,376 pg/ml in 766-BK, IL-2 at 18,905 pg/ml in 251-DB, and MCP-1 at 44,699 pg/ml in 251-RD. 251-CM also contained markedly high levels of IFN-γ (7,272 pg/ml) and IL-10 (11,752 pg/ml).

DISCUSSION

Well-characterized virus stocks of known identity are critical for the design and interpretation of in vivo SIV infection studies conducted in nonhuman primates. Historically, however, characterization of SIV challenge stocks has been limited, despite the potential for substantial stock-to-stock variability in both viral and nonviral features. While uncloned viral swarms are produced by infection of permissive cells in culture, infectious molecular clones are produced by transfection of an irrelevant, suitable cell type with a viral plasmid containing a full-length DNA viral genome, with or without subsequent expansion by infection of permissive cells. Differences between infection- and transfection-based production systems have the potential to markedly influence the character of the virus stock produced, and indeed there is a perception in the field that infection-derived virus stocks may be more infectious in vivo, particularly for mucosal challenges. Ample variability also exists in infection-based production methods, including differences in seed virus culture history, cellular substrate, cell stimulation method, culture duration, harvest procedure, and viral evolution in culture, all of which have the capacity to considerably alter the properties of the generated virus stock (Table 1 and Materials and Methods). Given questions about potential differences between different SIV challenge stocks, we comparatively characterized viral and nonviral parameters for a panel of SIVmac239 or -251 lineage stocks representative of viral swarms and infectious molecular clones produced in several different cell types by infection- and transfection-based methods (Table 1).

Although the titers of individual SIV challenge stocks are typically determined in vitro prior to use in an in vivo NHP study, we first determined how the various stocks compared head to head by performing parallel evaluations of viral parameters, including virion content, infectivity on different target cells, specific infectivity, and Env incorporation. Intriguingly, the two transfection-produced virus stocks contained the highest virion content and the highest overall stock infectivity in both TZM-bl and primary rhesus PBMC target cells (Fig. 1). When titers of viruses were determined in the same assay in parallel, the relative titers between stocks were similar, regardless of the target cell type or PBMC donor. However, for several viruses, the determined titer differed by >10-fold from one PBMC donor to the next, suggesting that the common practice of reporting in vitro titers for in vivo viral inoculum size may be relatively uninformative and that caution should be exercised when comparing infectivity titers derived in different labs on different cell types or on cells from different donors. One potential solution to this confounding variable would be for the field to report infectivity titers using a common cell-line-based assay, such as the TZM-bl assay used here, which provided a consistent and predictive value for in vitro infectivity.

In agreement with previous findings for HIV-1 (27), our quantitative Gag:Env analysis revealed markedly lower levels of gp120 in purified SIV virions produced in transfected 293T cells than in virions produced in infected PBMCs (Fig. 5), although a study examining transfection-produced SIV prepared and analyzed using different methods found higher total numbers of Env trimers (37). Given the minimal shedding of gp120 from virions (38), this result likely reflects less efficient Env incorporation into virions produced in transfected 293T cells than in infected T cells. Despite these substantially lower levels of virion-incorporated Env, transfection-produced SIVs surprisingly had particle/infectivity ratios similar to or higher than those produced in infected PBMCs (Fig. 1D). While this may demonstrate that only a limited number of Env spikes are required to drive fusion of the viral and cellular membranes (39), we note that the Gag:Env results presented here only reveal mean per particle Env levels and do not describe the distribution of Env spikes across virions, which may influence infectivity based on the stoichiometry of Env-CD4-coreceptor interactions and membrane fusion (40, 41). In addition, although these data may indicate that higher levels of virion-associated Env do not confer any measurable increase in viral infectivity for SIV as measured by in vitro assays, Env-dependent increases in infectivity for infection-derived viruses may be counterbalanced by the accumulation of deleterious mutations introduced during viral replication.

Reduced Env levels for transfection-produced viruses may have implications for studies assessing antibody-mediated reductions in virus acquisition (27); however, the SIVmac239 or -251 lineage viruses assessed here have been shown previously to be highly neutralization resistant (6, 42) and thus may be less susceptible to this effect than more neutralization-sensitive viral lineages. Apart from neutralization sensitivity, it is not clear if reduced Env content might also influence overall in vivo infectivity titer in a way that is not captured by standard in vitro titer determination assays. Once applied to a mucosal surface, individual virions may be subjected to physiological temperatures for several hours before encountering permissive target cells. We therefore also compared the stabilities of infectivity at 37°C for infection- and transfection-derived viruses, but again found that lower levels of virion-associated Env were not associated with any measurable decrease in the thermostability of infectivity (Fig. 6). It nevertheless remains possible that other factors at the in vivo site of infection, including differences in pH and the presence of proteases, could differentially impact high- and low-Env-containing virions.

For some SIV NHP studies, the use of a diverse viral swarm, such as one of the infection-derived SIVmac251 stocks examined in this study, is favored because it allows for enumeration of the transmitted and founder viral variants that initiate systemic infection (4, 9, 11, 43). By challenging animals with a viral dose that results in transmission of one or only a few variants, researchers are able to more closely model human sexual transmission of HIV-1 (44, 45). However, using SGA sequencing and 454 deep sequencing, we show in Fig. 2, 3, and 4 that for multiple stocks designated “SIVmac251,” the amount of sequence diversity and the specific sequences within each stock differ. This finding has implications for limiting dose challenge studies where a small number of viral variants are transmitted, as the available pool of viral variants from which one or few can be selected may be quite different in identity and extent of diversity from one stock to the next, despite the stocks having the same name. The genetic distinction between stocks could potentially lead to different study outcomes, depending on which virus stock is used, particularly for studies where selection for or against a specific viral variant or mutation is important. This is illustrated by the presence of preexisting variants within several known CD8⁺ T cell epitopes in the SIVmac251 stocks (Table 2). In addition to influencing the reactivity of specific CD8⁺ T cell responses, vaccines that elicit CD8⁺ T cell responses against a specific viral epitope may be less effective against SIVmac251 stocks that harbor preexisting epitope variants. In order to avoid the potential confounding effects of unpredictable viral swarm constituents while maintaining the ability to enumerate transmitted viral variants by sequence analysis, we are developing a “synthetic viral swarm” containing a defined number of genetically tagged but otherwise isogenic and phenotypically identical clones (unpublished data).

In addition to viral parameters, we also evaluated nonviral constituents of each virus stock. We focused on cytokines and chemokines as measurable analytes with the potential to influence early infection events and in vivo viral infectivity by modulating target cell identity, recruitment, and activation state. As expected, the infection-produced virus stocks, grown in cultures of activated PBMCs, contained higher levels and greater numbers of cytokines and chemokines than did the transfection-produced stocks. Although further studies are required to determine what effect, if any, increased cytokine and chemokine content has on early in vivo infection events, one might speculate that elevated levels of proinflammatory cytokines and chemokines could result in a non-virally induced local immune activation and enhanced in vivo viral infectivity. Although the amounts and identities of the cytokine and chemokine present in the inoculum vary significantly, it is unclear whether any of these stock constituents are physiologically relevant for sexual virus transmission. A study conducted by Politch and coworkers (46) found that the semen of healthy, fertile men contains elevated levels of several cytokines; however, the specific constituents were substantially different from those we detected in the SIV challenge stocks. Although it is not clear how the cytokine and chemokine content of other genital secretions or from the semen of HIV-1-infected individuals might compare with those present in the challenge stocks, it is clear that understanding what else besides virus is present in any given challenge stock is important for more accurately modeling sexual HIV-1 transmission.

With an emphasis on production methods as determinants of stock characteristics that may be relevant for in vivo NHP challenge studies, these studies highlight key distinctions both between infection- and transfection-derived virus stocks and between different infection-produced stocks of the same nominal virus. Each infection-derived stock contains a distinct combination of viral sequences and nonviral constituents, whereas transfection-derived stocks, while genetically consistent and containing lower levels of cytokines and chemokines, also contain lower levels of virion-associated Env. While it is tempting to consider how individual virus stock features described here might correlate with the reported in vivo infectivity of the stocks, well-controlled, head-to-head in vivo studies that carefully examine one stock variable at a time using the same in vivo titration protocol will be required to address this important question. In addition, although this study is to our knowledge the most extensive of its kind, it is important to note that our analyses were not comprehensive and there are other possible virus stock characteristics, including differential Env glycosylation profiles (47), stock endotoxin levels, the incorporation of cellular adhesion molecules, and microRNA content, that we did not assess here, but which could potentially influence viral infectivity and early infection events in vivo. Nevertheless, our findings highlight a number of underappreciated features of SIV challenge stocks and indicate differences between production methods that may be relevant for in vivo infectivity and the dynamics of early infection. These findings underscore the importance using well-defined virus stocks and argue for detailed characterization of SIV challenge stocks in the future.