Immunization of RMs with Rev-Ind Nef¯SIV and immune responses

A total of 8 RMs were infected as neonates (4 RMs) or adults (4 RMs) with the live attenuated, Rev-Ind Nef¯SIV and monitored prospectively for approximately 8 years (Groups 1 and 5, Figure 1). After the initial acute viremia phase, plasma vRNA loads were persistently below the threshold of the assay [25,26]. No signs of immune dysfunction or AIDS were observed, and all animals showed persistent humoral and cellular SIV-specific immune responses, consistent with chronic infection [25,26]. Multicolor flow cytometric analysis demonstrated preservation of the central memory (CM) subset of T cells [25,26]. In efforts to boost immune responses, we re-inoculated these monkeys with a high-dose (1.6 x 10⁴ TCID₅₀) of Rev-Ind Nef¯SIV i.v. at 15 months prior to the pathogenic SIVsmE660 challenges (-15 mos). In parallel, we challenged 4 naïve control monkeys to ensure the infectivity of the vaccine virus inoculum (Group 2 “vaccine 15 mos prior”). The latter 4 monkeys had peak vRNA loads of 10⁴-10⁵ copies/ml at week 2 post-inoculation (p.i.), but viral loads declined rapidly thereafter and became undetectable. In contrast, none of the vaccinees in Group 1 or 5 had measurable viremia upon re-inoculation. All vaccinees in Group 2 seroconverted by week 9 post-inoculation as measured by Western blot analysis; all vaccinees in Group 1 and 5 remained seropositive through 8 years of follow-up. Cellular SIV-specific immune responses were detected in both RM groups as measured by ELISPOT assay for interferon (IFN)-γ-producing cells after stimulation with peptide pools for SIV Gag [27]. Hematological values, including absolute numbers and percentage of CD3⁺CD4⁺cells, CD4⁺/CD8⁺ ratios, the percent CD4⁺CD29⁺ memory T cells, and the number of platelets remained stable in Group 1 and 5 vaccinees during the post-boost period as well as in Group 2 RMs (data not shown). No differences were observed in the frequencies of CM populations of CD4⁺ cells (CD28⁺CD95⁺) and CD8⁺ cells (CD45RA^-CCR7⁺) as measured by multicolor flow cytometric analysis (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Vaccination strategy with Rev-Ind Nef¯SIV and SIVsmE660 challenges.

RMs in Groups 1(low-dose challenge group) 5 and (high-dose only challenge group) had been vaccinated with Rev-Ind Nef¯SIV >8 yrs ago and boosted along with Group 2 monkeys; RMs in Group 2 were vaccinated with Rev-Ind Nef¯SIV 15 months prior to SIVsmE660 low-dose challenges, RMs in Group 3 were vaccinated with Rev-Ind Nef¯SIV 2 weeks prior to SIVsmE660 low-dose challenges; Group 4 consisted of naïve RMs that served as controls for SIVsmE660 challenges; Groups 6, served as control unvaccinated group was challenged with a high-dose of SIVsmE660 only and Group 7 received only Rev-Ind Nef¯SIV and was followed for 2 weeks. Groups 1-4 received the maximum of 17 weekly low-dose i.r. challenges with SIVsmE660 (300 TCID₅₀ as measured by TZM-bl assay). RMs that did not become systemically infected after the 17 low-dose challenges (boxed names) were rechallenged i.r. with a single high dose of SIVsmE660 (990 TCID₅₀) and became infected (rechallenged Group 1-2 and high-dose control Group 6). Blue indicates the live attenuated vaccine strain (Rev-Ind Nef¯SIV). A fifth monkey (RMw-6) did not fulfill the criteria of “vaccine take” (transient low-dose Rev-Ind Nef¯SIV infection with vRNA of 1080 copies/ml at 2 weeks post-vaccination, the expected time of peak vaccine virus viremia); therefore, data from RWm-6 were not included in the analysis. Thin red arrows stand for the weekly low-dose i.r. exposures to the challenge virus, heterologous SIVsmE660; the thick red arrows represent single high-dose i.r. challenges with SIVsmE660.

https://doi.org/10.1371/journal.pone.0075556.g001

Repeated low-dose and high dose SIVsmE660 i.r. challenges

As it is unlikely that any AIDS vaccine recipient will be exposed to an HIV variant exactly matching the immunogens, vaccine efficacy needs to be evaluated using heterologous challenge viruses [31]. Accordingly, we challenged our cohort of Rev-Ind Nef¯SIV-vaccinated RMs with multiple low-dose mucosal exposures of SIVsmE660 as outlined in Figure 1. The overall sequence difference between these two strains ranges from 8-26% at the amino acid level. The highest difference is in Tat (26%), followed by Env (15.1%), Vpr (12%), Pol (8.3%), and Gag (8%) [10]. The pathogenic SIVsmE660 exposure of the different vaccine groups occurred at different time intervals after the initial vaccination and ranged from 8 years to 2 weeks.

Viral RNA (vRNA) load determination was complicated by a need to differentiate the challenge virus (SIVsmE660) from the live attenuated vaccine strain (Rev-Ind Nef¯SIV). Since the latter was based upon the molecular clone of SIVmac239, we developed a set of primers that only recognized Rev-Ind Nef¯SIV but not the challenge virus SIVsmE660 [32]. A second assay based upon primers developed by Cline et al. [33], was then used to measure total vRNA. The difference between the two assays reflects vRNA loads due to SIVsmE660 superinfection.

All vaccinated RMs (Groups 1, 2 and 3) and control Group 4 were challenged up to 17 times i.r. with weekly low-dose SIVsmE660. Animals that remained aviremic after these 17 low doses were given a single high dose of SIVsmE660 (boxed animal names in Figures 1, 2). Remarkably, all 4 RMs of Group 3 challenged with SIVsmE660 2 weeks after receiving the live attenuated Rev-Ind Nef¯SIV (i.e., peak viremia of the vaccine strain, Figure 2G, week 0) became superinfected after only 1 to 3 low-dose SIVsmE660 challenges (Figure 2H). In contrast, some vaccinees of Groups 1 and 2 (Figure 2A-F) as well as control Group 4 remained aviremic (Figure 2J-L). Statistical analysis of this low-dose challenge phase revealed the following for the time-to-first viremia (defined as vRNA >10⁴ copies/ml): Group 3 vs. Groups 1 or 2 (P=0.006 for each comparison by Log rank test, two-sided); Group 3 vs. the combined Groups 1 + 2 (P<0.001). During the low-dose challenge phase, viremia developed in 3 out of 5 Group 4 controls (Figure 2J-L). Time-to-first-viremia was significantly shorter for Group 3 vaccinees compared to Group 4 controls (P=0.022). Clearly, Group 3 vaccinees with acute vaccine strain infection had the highest SIVsmE660 acquisition rate with the fastest time to viremia compared to all other groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Plasma vRNA loads after low-dose (A-B; D-E; G-H; J-K) or high-dose (M-N; P-Q) SIVsmE660 challenges.

Horizontal dashed lines in all panels represent the limit of detection of the RT-PCR assay (50 vRNA copies/ml) [32,33]. RMs in boxes were entered into the single high-dose challenge with SIVsmE660 (thick arrow) shown in M-N and P-Q. The panels in (C, F, I, L, O, R) are mean vRNA loads of each group. Blue, indicates vRNA loads for the live attenuated vaccine strain (Rev-Ind Nef¯SIV), purple indicates vRNA loads both vaccine + challenge virus, and the red line indicates challenge virus-specific vRNA loads.

https://doi.org/10.1371/journal.pone.0075556.g002

The fact that not all naïve controls became viremic prompted the high-dose SIVsmE660 rechallenge, after which all virus-only controls (Group 4 RMs with boxed names and a new control Group 6 of naïve monkeys, Figure 1) became viremic (Figure 2P-R). In addition, the 2 RMs from Group 1 and 3 RMs from Group 2 that remained uninfected after the multiple low-dose challenges (Figure 2A-F) as well as Group 5 monkeys inoculated with Rev-Ind Nef¯SIV at -8 years and -15 months (Figure 1) were also subjected to the single high-dose i.r. SIVsmE660 challenge. After the high-dose rechallenge, all RMs became superinfected with the challenge virus except for vaccinee RSb-3 that remained completely protected (Group 5, Figure 2M-N), outlining the limits of protection from acquisition of a heterologous mucosal challenge afforded by our LASIV. In addition, while all animals from Group 1 and 2 except RAn-8 controlled viremia after resolution of the initial SIVsmE660 infection, vaccinees challenged with a single high dose of SIVsmE660 exhibited persistent viremia, albeit at lower set points than Group 5 control animals (Figure 2). Peak vRNA loads due to SIVsmE660 superinfection were also affected by immunization with Rev-Ind Nef¯SIV. Control RMs of Groups 4 and 5 had peak vRNA loads of 10⁶-10⁸ copies/ml. Peak vRNA loads for SIVsmE660 did not differ between Groups 3 and 4. Group 1 vaccinees had peak vRNA loads ranging from 10⁴-10⁵ copies/ml, except in vaccinee RJi-6 (Group 1, Figure 2N), where several blips of below ~10³ copies/ml were noted. A significant difference in challenge virus peak vRNA loads was noted for the comparison of vaccinees exposed to the pathogenic SIVsmE660 during the chronic phase of live-attenuated Rev-Ind Nef¯SIV infection (Groups 1 + 2) with that of RMs during acute Rev-Ind Nef¯SIV infection (Group 3) (P=0.016 by Wilcoxon rank-sum test, two-sided; Figure 2B and E, versus H).

All newly infected animals and controls seroconverted by week 9 post-challenge as measured by Western blot analysis. We conclude that exposure to pathogenic challenge virus at week 2, the time of peak vaccine strain viremia, not only failed to protect but may even have increased host susceptibility to the heterologous pathogenic virus challenge in comparison with controls. Furthermore, exposure to pathogenic challenge during acute vaccine virus infection also led to significantly higher viral set points and continued replication of both challenge and vaccine strains compared to vaccinees given the live-attenuated virus 15 months to 8 years earlier.

SIV-specific immune responses before and after low-dose challenges

Next, we measured anti-SIV Gag IFN-γ ELISPOT responses before and during the low-dose SIVsmE660 challenges (Figure 3). Only low-level IFN-γ ELISPOT responses were observed among Groups 1, 2, and 3 prior to the repeated low-dose challenges. These responses did not differ significantly among the vaccine groups (P=0.19; Figure 3A). However, during the low-dose challenges, SIV Gag-directed IFN-γ ELISPOT responses increased among all three vaccine recipient groups and differed from controls (P=0.03; Figure 3B), particularly in Group 1, even though the difference between the vaccine groups never reached significant statistical difference. SIV Gag-specific IFN-γ ELISPOT responses on the day of the single high-dose SIVsmE660 rechallenge seen among the vaccinees of Groups 1 and 2 that had not developed persistent challenge virus viremia during the low-dose SIVsmE660 challenge phase (boxed RMs in Figures 1 and 2B,E) had increased above the levels seen prior to the low-dose challenge virus exposures, suggesting some boosting by the repeat exposure (Figure 3C). Of interest was the finding that SIVgag specific responses did not correlate with protection despite the equivalent levels of such responses between animals from Groups 2 and 3 in particular.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Cellular responses and T cell activation.

SIV Gag-specific, IFN-γ-secreting cells were measured by ELISPOT assay (A) before the low-dose SIVsmE660 challenge phase, (B) during the low-dose challenge phase, or (C) after the low-dose phase but before the high dose rechallenge with SIVsmE660 and (D) T cell activation was measured by percentage of CD3⁺ Ki67⁺ cells.

https://doi.org/10.1371/journal.pone.0075556.g003

Given the higher acquisition rate observed in Group 3 animals that were exposed to the pathogenic challenge virus at the time of peak vaccine virus viremia, we sought to test PBMC for evidence of T-cell activation by staining for Ki67, CD69, HLA-DR, and CCR5 by multicolor flow cytometry. Although staining for Ki67 on CD3+ cells was higher for Group 3 animals compared to those of Groups 1+2 and 4, statistical significance was not reached (P=0.16) (Figure 3D). Furthermore, in consideration of the potential inflammation induced by Rev-Ind Nef¯SIV in Group 3 monkeys, we also measured the relative representation of total CD4⁺ T cells, CD4⁺CD45RA^-CCR5⁺, CD45RA^-CCR7⁺ memory and CD4⁺ CD28⁺ CD95⁺ central memory cells before and 2 weeks after post-infection (Group 7). As shown in Figure S1, higher percentages of these populations were noted at 2 weeks post Rev-Ind Nef¯SIV infection relative to baseline confirming the microarray data. However, phenotypic analysis did not reach statistical significance, likely due to the small number of animals.

The potential influence of host restriction factors on viral loads

The animals were enrolled and grouped years before the influence of MHC class I or TRIM5α alleles on viral loads became known. However, it is now generally recognized that low-dose SIVsmE660 challenges may be influenced by TRIM5α alleles [7,34]. We therefore typed all vaccinees and controls enrolled in this study for their MHC Class I and TRIM5α genotypes (Table 1; RMs in Group 6 are not included in this discussion since they did not undergo SIVsmE660 challenge). While the distribution of TRIM5a alleles was not perfectly equal among groups, it remains comparable with only 1 sensitive genotype in Groups 2, 3 and 5 while the other groups comprised a blend of moderately to highly restrictive genotypes. Of note, neither TRIM5α nor the MHC B*008 or B*017 appeared to greatly affect the susceptibility of infection within any defined group except perhaps for Group 2 (Table 1). Of note though, the low-dose control Group 5 comprised four out of five “restrictive” genotypes, which may have lead to two RMs requiring 17 weekly low-dose SIVsmE660 challenges. However, the other 2 animals possessing a restrictive genotype as well as the fifth, with a moderate genotype, were infected rapidly during the low dose challenge series, suggesting that perhaps other yet to be defined factors played a role. Of note, all controls, regardless of their TRIM5α or MHC genotype, became infected after the single high-dose SIVsmE660 challenge. Finally, we also performed statistical analysis by combining restrictive, moderate, susceptibility to infection of TRIM5α alleles of all the vaccinees vs. control RMs for virus acquisition. We noted that vaccinees and control RMs experienced comparable acquisition rates over the first 4 challenges and after that, vaccinees showed slower acquisition compared to controls (P=0.045; based on Mantel-Cox test) which is in agreement with recently published studies [35,36], suggesting that TRIM5α played a minor role at best in acquisition of infection in this study.

Gene expression analysis of RMs 2 weeks after i.v. vaccination with Rev-Ind Nef¯SIV (Group 7)

To understand the potential mechanism that led to the increased susceptibility of RMs to challenge virus acquisition 2 weeks after the inoculation of Rev-Ind Nef¯SIV, three naïve RMs were enrolled for PBMC and rectal tissue gene-expression profiling before and after vaccination with Rev-Ind Nef¯SIV (Group 7, Figure 1 and Table 1); these additional RMs were vaccinated similar to Group 3 animals and monitored for plasma vRNA loads (Figure 4A). At week 2, all three Group 7 RMs were euthanized, and PBMC and rectal biopsies were collected using identical procedures pre-vaccination, thus allowing us to compare paired samples from each RM, with pre-vaccination samples as autologous controls for microarray analysis (Methods; Figure 4B). Unsupervised analysis using PCA showed a tight grouping of PBMC samples, while rectal tissues exhibited greater heterogeneity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. vRNA load measurements, principal component analysis (PCA) and heatmap analysis.

(A) vRNA loads were measured at 2 weeks after Rev-Ind Nef¯SIV vaccination (Group 7; 3 RMs) The RMs were subsequently euthanized to collect rectal tissues and PBMC. (B) PCA was performed on log₂ expression measures for autologous PBMC and rectal biopsy specimens collected before and two weeks after inoculation with the live attenuated vaccine strain, Rev-Ind Nef¯SIV. Infected and naïve PBMC samples were centered and grouped together, indicating that the samples were nearly homogenous. In contrast, we observed heterogeneity in rectal tissues. (C) Hierarchical clustering of differentially expressed genes in PBMC. Supervised analysis identified approximately 100 differentially expressed genes between PBMC of the RMs collected before and after vaccination using a significance threshold of a false discovery rate) <5% and log₂ fold-change >1 was analyzed using average-linkage hierarchical clustering with a Pearson correlation coefficient distance metric. We noted a clear upregulation of expression of a set of chemokines and chemokine receptors, most notably CCL3, and CCL5 (highlighted).

https://doi.org/10.1371/journal.pone.0075556.g004

Supervised analysis identified approximately 100 individual genes that were differentially expressed in PBMC (Figure 4C) when comparing pre- and post-vaccination samples but none in rectal biopsies. We noted a clear upregulation of expression of a set of chemokine and chemokine receptors, most notably CCL3, also known as macrophage inflammatory protein-1α (MIP-1α), and C-C chemokine receptor 5 (CCR5). We further validated the increased expression of CCR5 by quantitative qPCR (data not shown). Increased expression was seen also for several genes involved in different cellular processes, including cell cycle, signal transduction and chromatin remodeling (Figure 4C). In contrast, gut biopsies exhibited no such differences, suggesting that relatively few changes occurred in the gut during acute infection with Rev-Ind Nef¯SIV or that tissue sample heterogeneity, the relatively few lymphocytes residing in the mucosal tissues, and the low number of RMs available did not permit observation of significant differences.

Evaluation of these results with the Ingenuity Pathway Analysis (IPA) showed that the top network genes regulated by Rev-Ind Nef¯SIV include functional groups responsible for hematopoiesis, cellular movement, and immune cell trafficking. Several functionally linked genes were upregulated as shown in Figure 5. Network analysis revealed that a set of upregulated chemokines and chemokine receptors formed clusters/cores with NF-ĸB and other signaling molecules (Figure 5). Collectively, these data indicate Rev-Ind Nef¯SIV vaccination induced immune activation within 2 weeks of infection that was clearly discernible in PBMC but not in rectal tissues. In turn, immune activation caused by the live attenuated vaccine virus may have increased host susceptibility to the heterologous SIVsmE660 during the low-dose challenges and facilitated its dissemination.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Network pathway analysis.

Network pathway analyses were performed using the Ingenuity Pathway Analysis Network tool. Node colors show the average gene expression levels of RMs vaccinated with Rev-Ind Nef¯SIV relative to the average before vaccination (red, relative increase in expression; green, relative decrease in expression). CCR5, CCL3 and other signaling molecules were upregulated and were directly and indirectly involved with other immune system pathways. The numbers underneath the nodes are 1) log₂-fold change, and 2) FDR-adjusted P values.

https://doi.org/10.1371/journal.pone.0075556.g005

Animals

Indian RMs were housed at the Yerkes National Primate Research Center (YNPRC, Emory University, Atlanta, GA) and handled according to the National Institutes of Health guidelines on the care and use of laboratory animals. Animal experiments were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the U.S. Public Health Services/National Institutes of Health, as well as according to the recommendations in the Weatherall report on ‘‘The Use of Non-human Primates in Research’’ (http://www.acmedsci.ac.uk/images/project/nhpdownl.pdf). The protocol was approved by the Committee on the Ethics of Animal Experiments of Emory University (IACUC ID: YER-256-2008Y; Emory University Animal Welfare Assurance Number A3180-01). The animals were housed indoors with a 12 hour light/dark cycle, in individual cages but in visual and auditory contact with other RMs, were fed monkey chow (Purina) ad libitum supplemented daily with fresh fruit. Standard enrichment was provided by the YNPRC enrichment staff including perches, rubber toys and varied treats such as peanuts and cereals. Blood was collected under ketamine or Telazol anesthesia from the femoral vein. Animals were closely monitored and observed for development of disease at least twice daily. Animals determined to be under stress or in discomfort, were administered appropriate anesthetics and/or analgesics as directed by the clinical veterinary staff. Monkeys showing signs of disease or distress or reaches IACUC endpoints, such as pain or stress that could not be alleviated using standard analgesics and/or chemotherapy were humanely euthanized with an intravenous overdose of pentobarbital sodium according to the guidelines of the American Veterinary Medical Association. Distribution of RMs into the various groups is illustrated in Figure 1; Determination of the major histocompatibility complex (MHC) or Tripartite-motif 5 alpha (TRIM5α) genotyping were done after the initiation of the study given the fact that such tools were not available at the study onset.

Virus stocks

Virus stocks were generated in peripheral blood mononuclear cells (PBMC) collected from healthy, pathogen-free RMs and were titrated in vitro using CEMx174-GFP cells [58]. Rev-Ind Nef¯SIV was generated in our laboratories [25] and SIVsmE660 was obtained from the AIDS Research & Reference Reagents Program (ARRRP), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), Bethesda, MD, USA. Immunizations with Rev-Ind Nef¯SIV were given by the i.v. route.

Mucosal SIV challenges

Ketamine-anesthetized RMs received either low-dose or high-dose i.r. inoculations of SIVsmE660 in a total volume of one ml through a gastric feeding tube, which was, inserted 5 cm into the rectum with the aid of a lubricant. Virus administration (300 50% tissue culture infectious doses (TCID₅₀) for each low and 990 TCID₅₀ for the high-dose challenges) was followed by two ml of supplement-free RPMI-1640 medium and one ml of air. The animals’ pelvis were left elevated for 15 min before the RMs were returned to their cages.

Plasma vRNA levels

RNA was isolated using the QiaAmp Viral RNA Mini-Kit (Qiagen, Valencia, CA), and vRNA levels were measured by quantitative RT-PCR; the assay sensitivity was 50 vRNA copies/ml [32,33].

Differential PCR to distinguish between Rev-Ind Nef¯SIV and SIVsmE660

We used differential PCR to detect the vaccine (Rev-Ind Nef¯SIV) and challenge (SIVsmE660) viruses. SIVmac251-specific primers and probes developed by Cline et al. [33] were used to detect vRNA from both Rev-Ind Nef¯SIV (vaccine virus) and SIVsmE660 (challenge virus). Differentiation of the Rev-Ind Nef¯SIV vaccine virus from the challenge virus SIVsmE660 in the same vRNA sample was performed with primers and probes specific for SIVmac239 developed by Hofmann-Lehmann et al. [32].

The differences between these two assays reflect vRNA loads of SIVsmE660 superinfection.

ELISPOT assays

PBMC were tested for virus-specific responses using overlapping Gag peptide pools of SIVmac239 (ARRRP) as described previously [27,59]. For long-term ELISPOT assays, cells were cultured at a final concentration of 5 × 10⁶ cells per ml in 24-well plates for 10 days with SIVmac239 overlapping Gag peptide pools at a final concentration of 2 µg/ml of each peptide. Interleukin-2 (IL-2) was added to the cultures every 3 days at a final concentration of 10 U/ml.

MHC class I and TRIM5 α typing of macaques

All RMs were typed for MHC and TRIM5α after completion of the study since genotyping was not routinely performed at the time of initiation of the current study. However, it has since been reported that select MHC class I alleles are associated with delayed SIV disease progression [60,61] and therefore, typing for Mamu-A*001, B*008 and B*017 (Table 1) was performed to evaluate the distribution of these alleles into the various groups. TRIM5α genotypes of RMs challenged with SIVsmE660 were determined as published [38]. In brief, DNA was isolated from cryopreserved PBMC and genotyping was performed at the Genetics Core of the New England Primate Research Center.

PBMC and rectal biopsy collections

For microarray analysis, blood was collected in sodium citrate CPT tubes and centrifuged immediately according to the manufacturer’s instructions to separate PBMC and plasma; PBMC were stored in RNAlater solution (Ambion, Austin, TX) at -80°C. Similarly, rectal pinch biopsies were collected 2 weeks before and after the administration of Rev-Ind Nef¯SIV. Rectal pinch biopsies were collected and immediately placed into RNAlater solution and stored -80°C until further processing. We allowed 2 weeks for the rectal tissues to heal before i.v. inoculation with Rev-Ind Nef¯SIV, the live attenuated vaccine strain.

RNA extraction and microarray analysis

Total RNA from RMs PBMC and rectal biopsies was extracted using RNAeasy extraction kits (Qiagen, Valencia, CA). cDNA labeling, hybridization, staining and scanning were performed according to the manufacturer’s instructions (Affymetrix, Santa Clara, CA).

Microarray data analysis

Array quality was assessed using the R/Bioconductor package [62]. Affymetrix CEL files were processed and normalized using the robust multiarray average (RMA) algorithm [63]. Principal component analysis (PCA) was performed on the normalized expression values to identify phenotypically similar groups by visualizing the top three principal components. To identify genes correlating with the phenotypic groups, the data were fitted to a linear model using the limma package [64] and tested for differential gene expression. Results were adjusted for multiple testing using the Benjamini and Hochberg (BH) method [65], and significance was determined using a false-discovery-rate cutoff of <5% and a log fold-change cutoff of >1. Expression values of significant probe sets were loaded into the MultiExperiment Viewer (MeV) [66] and analyzed using average-linkage hierarchical clustering with a Pearson correlation coefficient distance metric. Gene interaction network analysis and visualization were performed on significant probe sets using the Ingenuity pathway analysis (IPA) software package. Microarray data was deposited into the gene expression omnibus database (GSE49663).

Statistical analyses

The log-rank test was used to compare the time-to-peak viremia distributions between the groups. The Wilcoxon-rank sum test was used to compare peak viremia between groups. The Montel-cox test was used to compare difference between TRIM5α and viral acquisition. All P-values reported are based on two-sided tests.