|Home | About | Journals | Submit | Contact Us | Français|
Live attenuated SIV vaccines (LAVs) remain the most efficacious of all vaccines in nonhuman primate (NHP) models of HIV/AIDS, yet the basis of their robust protection remains poorly understood. Here, we demonstrate that the degree of LAV-mediated protection against intravenous wildtype SIVmac239 challenge strongly correlates with the magnitude and function of SIV-specific, effector-differentiated T cells in lymph node, but not with such T cell responses in blood or with other cellular, humoral and innate immune parameters. Maintenance of protective T cell responses was associated with persistent LAV replication in lymph node, which occurred almost exclusively in follicular helper T cells. Thus, effective LAVs maintain lymphoid tissue-based, effector-differentiated, SIV-specific T cells that intercept and suppress early wildtype SIV amplification and, if present in sufficient frequencies, can completely control and perhaps clear infection, an observation that provides rationale for development of safe, persistent vectors that can elicit and maintain such responses.
It has been two decades since the first report that LAV administration can completely protect rhesus macaques from subsequent challenge with highly pathogenic, wildtype SIV1, a degree of efficacy that to this day stands in sharp contrast to that achieved with a wide variety of alternative HIV/SIV vaccine strategies2-6. Unfortunately, LAV-mediated protection is inversely proportional to the degree of attenuation, such that the most effective designs retain considerable pathogenic potential, precluding their development as vaccines in humans2,7. While it has long been appreciated that delineation of a common immunologic mechanism for LAV efficacy would be a major advance for the HIV/AIDS vaccine field – providing an immune response target for development of safe and effective vaccine designs3 – the basis of LAV efficacy, particularly the durable, complete, apparently sterilizing control characteristic of the best LAVs, remains controversial. Twenty years of research using various types of LAVs and NHP species, as well as different pathogenic challenge viruses and varying timing and routes of challenge has variously implicated, or argued against, T cell immunity, humoral immunity, innate immunity and viral interference in LAV-mediated protection without the emergence of a unifying, consensus protective mechanism, or clear-cut definition of multiple different mechanisms acting in combination to achieve protection 2,5,6,8-16.
In this study, we sought to clarify the immunologic (or other) basis of LAV-mediated protection against highly pathogenic SIV by statistical correlation of fundamental measures of LAV-elicited immunity with post-challenge efficacy in a large cohort of LAV-vaccinated rhesus macaques. Our approach was to vaccinate groups of macaques with different LAVs that vary in their level of attenuation and degree of amino acid sequence homology with the wildtype SIVmac239 challenge virus with the intent, at the level of the overall cohort, of producing sufficient heterogeneity in both immune responsiveness and protective efficacy to create a statistically powerful immune parameter vs. efficacy matrix. We focused on the most stringent route of SIV challenge – high dose intravenous administration – because 1) LAVs are unique among SIV vaccines in their ability to completely control such systemic SIV challenge1-3, and 2) hematogenous dissemination is a common step in all primary HIV/SIV infections, including those arising from mucosal surfaces17. In this regard, a vaccine that intercepts and completely controls hematogenous HIV spread would be applicable to both mucosal and intravenous routes of HIV entry, the latter a clinically relevant way of acquiring infection that is not addressed by vaccines designed to elicit mucosal immunity. Strikingly, this experimental approach strongly and specifically identified characteristics of the SIV-specific T cell response in lymph node (including its magnitude, effector differentiation, functional properties and kinetics of activation after challenge) as cross-platform predictors of LAV-mediated protection against high dose intravenous SIV challenge. We further linked the maintenance of these protective lymph node-based T cells to the ability of the replication-competent LAV to persist in follicular helper CD4+ T cells in secondary lymphoid tissues. Together, these findings provide further support for the concept that early SIV infection, even the systemically distributed infection following high dose intravenous challenge, is vulnerable to the high frequency, effector differentiated memory T cell responses that are uniquely elicited and maintained over the long term by persistent vectors.
To achieve the heterogeneity in both immune responses and protective efficacy necessary for robust statistical delineation of immune correlates, we selected LAV designs for study that vary in their level of attenuation and degree of amino acid sequence homology with the wildtype SIVmac239 challenge virus, and therefore would be expected to differ in the level and persistence of SIV antigens available to elicit adaptive immunity (leading to differences in magnitude, differentiation and durability of the T cell and antibody responses resulting from each vaccine), as well as in the ability of the resulting vaccine-elicited adaptive immunity to cross-react with the challenge virus. The prototype LAV, SIVmac239(Δnef), is only moderately attenuated, and is sequence-matched to the challenge virus. SIVmac239(Δ3) and single cycle SIVmac239 are also sequence-matched with the challenge virus, but are progressively more attenuated than SIVmac239(Δnef)18,19. SHIV89.6 is also only moderately attenuated, but differs from the SIVmac239(Δnef) prototype by virtue of highly heterologous, HIV-derived env, tat, vpu and rev sequences, and its ability to use CXCR4 as a co-receptor5. SIVsmE543(Δnef) is variably attenuated (due to its sensitivity to restrictive TRIM5 alleles), and is heterologous in amino acid sequence to SIVmac239 (~15% overall divergence)13,20. Each LAV was administered to 6-8 macaques to create an overall study group of 32 vaccinated animals (Fig. 1a). Macaques with class I major histocompatibility complex alleles associated with enhanced CD8+ T cell-mediated SIV control (Mamu-A*01, Mamu-B*08, and Mamu-B*17)13 were excluded from this study to focus analysis on novel correlates of protection.
Consistent with the expected characteristics of the selected LAVs, the SHIV89.6-vaccinated macaques in our study group demonstrated the highest peak viremia, highest peak number of LAV RNA copies within peripheral blood mononuclear cells (PMBC), and most acute-phase effector-site CD4+ T cell depletion, followed by the macaques vaccinated with SIVmac239(Δnef), SIVmac239(Δ3), SIVsmE543(Δnef) and single-cycle SIVmac239 (Fig. 1b,c and Supplementary Fig. 1). Differences in these parameters were less apparent in chronic-phase , as the replication of even the least attenuated LAVs was brought under stringent control by week 10 after vaccination, and the modest acute-phase effector-site CD4+ T cell depletion was transient. Notably, however, the peak numbers of PBMC-associated LAV DNA copies established during acute-phase viremia decayed very slowly over time, providing a persistent measure of the amount of peak LAV replication (Fig. 1c).
After a 50-week vaccination period to allow full maturation and stabilization of anti-SIV immunity2, the 32 LAV-inoculated, and six unvaccinated, control macaques were intravenously challenged with 103 infectious units of wildtype SIVmac239 (Fig. 1a). The LAV-vaccinated macaques in this study manifested three distinct outcomes after this challenge: 1) complete protection (n = 19), characterized by no significant net increase in overall plasma viral load (pvl), and either no detection of wildtype SIVmac239 with discriminatory qRT-PCR (n = 16) or transient, low level (< 103 copies ml−1) detection (n = 3); 2) partial protection (n = 7), defined as clear establishment of wildtype SIV infection (> 104 copies ml−1 wildtype SIVmac239 RNA by discriminatory qRT-PCR and persistently elevated overall pvl), but with overall pvl reduced by at least 2 logs in the first 70 days post-challenge compared to unvaccinated controls; and 3) no protection (n = 6), in which post-challenge viral dynamics are indistinguishable from the unvaccinated controls (Fig. 1d). The biologic validity of these different outcome groups was confirmed by the observations that completely protected macaques lacked both an anamnestic T cell response to wildtype SIV challenge (a sensitive measure of increased SIV replication) and any evidence of SIV pathogenesis (e.g., loss of mucosal CD4+ memory T cells), whereas partially protected and non-protected macaques manifested both unequivocal anamnestic responses and mucosal CD4+ memory T cell depletion that was commensurate with their respective pvl levels (Fig. 1e,f). As expected, protection was highest in macaques vaccinated with LAVs that were the least attenuated and most homologous to the challenge virus (Fig. 1d). The peak plasma LAV load (initial 4 weeks after vaccination) and the chronic-phase plasma LAV load immediately before challenge, as well as the number of LAV DNA copies in PBMC immediately before challenge (which, as indicated above, closely correlates with peak LAV load in plasma) – all reflections of LAV attenuation – statistically delineated macaques at each end of the protection spectrum (in particular, completely vs. non-protected macaques; Fig. 1g,h). Notably, however, levels of PBMC-associated LAV RNA at the time of challenge did not significantly correlate with outcome (Fig. 1h).
Prior to challenge, we quantified 11 primary parameters of SIV-specific immunity for assessment as potential correlates of protection, including three humoral immune parameters – SIVmac239 neutralization titres, native SIVenv binding titres (neutralization of tissue culture-adapted SIVmac251), and SIVenv-directed antibody-dependent cell-mediated cytotoxicity (all in plasma samples) – and eight T cell response parameters -- the magnitude of SIV proteome-specific CD4+ and CD8+ T cells in PBMC, bronchoalveolar lavage (BAL) fluid, and lymph node, as well as the breadth of these responses in PBMC. LAV-vaccinated macaques manifested a spectrum of values for each of these parameters, with the exception of SIVmac239 neutralizing antibody titres, which were detectable at low level in only a very small proportion of LAV-vaccinated animals (Suppl. Fig. 2). It is also noteworthy that SHIV89.6-vaccinated macaques lacked antibody responses to the heterologous SIVenv.
We next assessed each of these parameters for correlation with the outcome groups described above using the nonparametric Kruskal-Wallis test, adjusting for multiple comparisons using the Bonferroni method. Strikingly, only two parameters – the magnitude of SIV proteome-specific CD4+ and CD8+ responses in lymph node – statistically delineated the outcome groups using this method (Table 1; Fig. 2a). Moreover, the SIV proteome-specific CD4+ and CD8+ T cell responses in lymph node were the only parameters to statistically delineate each of the three outcome groups from each other by the Wilcoxon rank sum test (Fig. 2a; Suppl. Fig. 3). In partially and non-protected macaques, the lymph node, but not the blood, SIV-specific T cell response magnitude (CD4+ and CD8+) inversely correlated with peak viremia after challenge by Spearman analysis (Fig. 2b), indicating that pre-challenge, lymph node-based T cell responses also predict the degree of SIVmac239 control once infection with the wildtype virus is established. The majority of SIV-specific CD8+ T cells in pre-challenge lymph nodes of subsequently completely protected macaques manifested phenotypic evidence of both recent activation (HLA-DR and PD-1 expression) and effector differentiation (CCR5 expression) (Fig. 2c), suggesting an ability to manifest an immediate anti-viral effector response. To ascertain whether these SIV-specific T cells could directly mediate anti-SIV effector activity, we assessed the ability of sorted populations of memory T cells from PBMC and lymph node to suppress SIVmac239 replication in vitro. Indeed, CD8+ (but not CD4+) memory T cells from lymph node (but not blood) of completely protected macaques manifested potent SIV suppression, and across the overall cohort, the level of this suppression strongly predicted each of the three outcome groups and inversely correlated with the peak post-challenge pvls in those macaques with establishment of wildtype SIVmac239 infection (Fig. 2d).
The ability of efficacious LAVs to maintain protective T cell responses in lymph node appeared to depend on the preferential persistence of LAV replication in this tissue (Fig. 3a), as the magnitude of both the lymph node CD4+ and CD8+ SIV-specific T cell responses closely correlated with the level of cell-associated LAV RNA in lymph node (a measure of ongoing LAV replication; Fig. 3b). As these lymph node T cell responses correlated with protective efficacy (Fig. 2a-d), it is not surprising that the levels of cell-associated LAV RNA in lymph node prior to challenge also correlated with outcome after challenge (Fig. 3c). The preferential persistence of LAV in lymph node is explained by the highly selective replication of LAV in the CD4+, PD-1hiICOShiCD200hiCCR7low follicular helper T cell (TFH) subset (Fig. 3d; Suppl. Fig. 4), which are specifically localized in B cell follicles and therefore found almost exclusively in lymph node and other secondary lymphoid tissues21-23.
Analysis of diverse tissues obtained at necropsy of 10 completely protected LAV-vaccinated macaques confirmed that B cell follicle-rich secondary lymphoid tissues, including lymph nodes, spleen and colonic mucosa, contained significantly higher copy numbers of cell-associated LAV RNA and DNA per 105 cells than primary lymphoid tissues (bone marrow) and extra-lymphoid tissues (jejunal mucosa, liver, BAL) that typically lack such B cell follicles (Fig. 3e). Strikingly, the mean frequencies of SIV-specific CD8+ T cells in each tissue closely correlated with the log mean of cell-associated SIV RNA in that tissue (Fig. 3f), an observation providing further evidence that the extent of ongoing LAV replication determines the frequency at which SIV-specific CD8+ T cells are maintained in any given tissue. Thus, CD4+ TFH cells serve as a sanctuary site for LAV persistence23,24, and concentrate LAV replication (and antigen production) in secondary lymphoid tissues, providing an immune stimulus for production and maintenance of SIV-specific T cells that is proportional to the level of TFH infection.
Collectively, these data suggest that following intravenous challenge, lymph node and other secondary lymphoid tissues are major sites at which LAV-generated, SIV-specific T cells intercept the nascent wildtype SIV infection, and if these T cells are present in sufficient numbers, the wildtype SIV infection can be completely controlled and possibly cleared. To probe the dynamics and characteristics of the immune response at this putative intercept, we used microarray analysis to compare the transcriptional profiles of unfractionated lymph node cells obtained from completely protected, non-protected, and unvaccinated control macaques 7 days prior to, and 4 and 14 days after wildtype SIVmac239 challenge. Prior to challenge, the gene expression profiles of unfractionated lymph node cells from completely protected and non-protected macaques were indistinguishable from each other, but different from unvaccinated control macaques (Fig. 4a,b). Gene pathways that were significant in distinguishing vaccinated from non-vaccinated macaques included metabolic pathways associated with stress responses and survival (EIF2, mTOR, oxidative phosphorylation, protein ubiquitination), as well as pathways (NFAT, PLCγ, PKA and CD28) consistent with a low-level immune response to the LAV (Fig. 4c; Suppl. Table 1). Remarkably, at post-challenge day (PCD) 4 (a time point when cell-associated SIV RNA and DNA were not yet detectable in lymph nodes of the unvaccinated controls), the gene expression profiles of lymph node cells from completely protected and non-protected (but not control) macaques were dramatically different from pre-challenge, involving multiple pathways related to immune activation, but were still indistinguishable from each other (Fig. 4d-f; Suppl. Table 2). A gene-level comparison of PCD 4 vs. pre-challenge lymph node cells from completely protected, non-protected, and control macaques using an immune response gene filter revealed up-regulated T cell, inflammasome and antigen presenting cell gene activation, as well as genes regulating the innate/adaptive immunity interface in the vaccinated macaques compared to controls (Fig. 4g; Suppl. Discussion 1). Since the representation of the major cell subsets in lymph nodes of completely protected and non-protected macaques at PCD 4 was unchanged from baseline (Suppl. Fig. 5), these differences in gene expression profiles between baseline and PCD 4 probably reflect an in situ immune response. Moreover, given that the lymph node transcriptional profiles changed in vaccinated, but not unvaccinated control, macaques, they were almost certainly initiated by an adaptive (LAV-elicited) memory response to challenge. From PCD 4 to 14, the relationship between the transcriptional profiles of lymph nodes from completely protected, non-protected and control macaques changed (Fig. 4g; Suppl. Fig. 6). Consistent with the onset of progression of wildtype SIVmac239 infection in non-protected and control, but not completely protected, macaques, lymph node cells from the non-protected and control macaques showed similar patterns of gene expression associated with enhanced innate IFN and effector T and NK cell responses, whereas expression of immune related genes in lymph node cells from completely protected macaques was largely unaltered over this period (Suppl. Discussion 1).
To specifically compare the activity of memory T cells in the lymph nodes of completely vs. non-protected macaques following wildtype SIVmac239 challenge, we examined the transcriptional profiles of sorted CD8+ and CD4+ memory T cells from lymph nodes obtained at PCD 4 and 14. This analysis revealed that CD8+ memory T cells from PCD 4 lymph nodes of completely protected macaques manifested a relative up-regulation of TCR signaling genes and granzymes, and down-regulation of IL7R compared to non-protected macaques (Fig. 5), reflecting higher T cell activation and early establishment of CD8+ T cell effector responses in lymph nodes from these completely protected animals25-27 (Suppl. Discussion 2). Although fewer gene expression changes distinguished CD4+ memory T cells from the lymph nodes of completely protected compared to nonprotected macaques at PCD 4, there was still significantly higher expression of genes associated with antigen-specific activation of CD4+ memory T cells in completely protected macaques, including CD40LG, LAG3, CTLA4, BTLA, and regulators of T cell survival (NFKB transcriptional pathway)28,29 (Suppl. Fig. 7; Suppl. Discussion 2). By PCD 14, CD8+ memory T cells from non-protected macaques manifested transcriptional evidence of effector function27, e.g., granzymes, as well as innate immune activation downstream of type 1 IFN signaling, protein ubiquitination and cellular proliferation pathways30 (Fig. 5; Suppl. Tables 3,4), consistent with the late development of both adaptive and innate immune responses. CD4+ memory T cells from lymph node of non-protected macaques showed similarly increased expression of type I IFN response genes at PCD 14 (Suppl. Fig. 7). In contrast, CD8+ and CD4+ memory T cells from lymph nodes of completely protected macaques at PCD 14 manifested relative up-regulation of genes associated with long-term memory and cellular quiescence/survival30,31 (Fig. 5; Supp. Fig. 7; Suppl. Table 3; Suppl. Discussion 2). Thus, transcriptional analysis of lymph node memory T cells provides evidence of larger and/or earlier CD8+ and CD4+ T cell response to challenge in completely protected vs. non-protected macaques (in keeping with the higher magnitude SIV-specific T cell responses in the lymph nodes of completely protected animals). This early response is associated with control of infection and prevention of the marked innate and adaptive immune activation that characterized progressive wildtype SIV infection.
Correlation analysis does not delineate cause and effect and, therefore, cannot unequivocally establish whether persistent LAV replication, the associated SIV-specific immune response or an uncharacterized covariate are mechanistically related to the observed protection. However, these data, taken together, lead most logically to a model in which persistent LAV replication elicits and maintains differentiated/activated, anti-viral effector memory T cells in tissues, and if these T cells are present in sufficient frequency prior to challenge, they can completely control (and perhaps clear) wildtype SIV infection in the first days after challenge. Secondary lymphoid tissues appear to be the critical site of viral intercept after intravenous challenge, and CD8+ memory T cells from lymph nodes of optimally LAV-vaccinated animals can strongly suppress wildtype SIV replication in autologous CD4+ T cells. Although CD4+ memory T cells from the lymph nodes of the same macaques lacked this capability, the very strong correlation between the frequency of SIV-specific CD4+ T cells in lymph node tissue and outcome suggests their crucial involvement in protection, perhaps through direct effector activity against SIV-infected macrophages, help of SIV-specific, CD8+ T cell functionality or both32,33. Importantly, lymph node-based, SIV-specific T cell responses also delineated partial controllers from non-controllers, suggesting that when the LAV-elicited SIV-specific T cell responses are below the threshold for complete control, an anamnestic expansion of the LAV-elicited, SIV-specific memory T cells can ensue, with the outcome determined, as with conventional prime-boost vaccines, by a competition between the kinetics and anti-viral activity of the anamnestic response and viral escape4.
Viral interference probably does not substantially contribute to observed protection given that LAV replication is restricted largely to CD4+ TFH cells, a minor subset of the total potential population of CD4+ target cells. Innate immunity elicited by persistent LAV replication is another candidate mechanism for mediating protection. Although our sampling schedule may have missed a transient innate immune response, our detailed transcriptional analysis revealed no evidence of differential innate immune activation in the lymph nodes of protected compared to unprotected macaques before challenge or at 4 days after challenge and thereby argues against this possibility. In addition, the number and activation status of NK cells in lymph node prior to, or following, challenge did not correlate with protection, arguing against involvement of these innate immune effectors (Suppl. Fig. 8). Perhaps the most compelling evidence in support of the T cell protection model of LAV efficacy is the analogous immediate and complete, MHC I allele-independent control of SIV infection that is associated with the high frequency, tissue-based, SIV-specific T cell responses elicited and maintained by SIV protein-expressing cytomegalovirus vectors34. These persistent β-herpesvirus vectors have a completely different biology than LAV (target cell tropism, innate immune interaction, immune evasion and persistence strategies) and do not elicit antibody responses to SIV inserts, factors strongly implicating tissue-based, effector-differentiated memory T cells as the common mediator of protection with these persistent vectors.
In summary, these data identify the magnitude of SIV-specific CD4+ and CD8+ T cell responses in secondary lymphoid tissues (but notably, not blood), as well as the SIV suppression capacity of CD8+ memory T cells in these tissues, as strong (cross-platform) correlates of LAV-mediated protection. These results do not preclude other innate or adaptive immune mechanisms (including antibodies) as potential contributors to LAV efficacy, especially at mucosal surfaces16, or to protection from mucosal challenge afforded by other vaccine modalities, as has been suggested recently in both clinical35 and NHP studies36. However, taken together, these observations provide further support for the conclusion that effector-differentiated, SIV-specific memory T cells, when maintained at a high frequency in tissues important for SIV infection, can intercept and completely control a nascent, highly pathogenic SIV infection without anamnestic expansion15,34. This intercept and control may occur at mucosal sites of entry, lymphoid tissues that are involved in early spread (the latter even when mucosal infection is bypassed by intravenous inoculation) or both, but they always occur before the establishment of a ‘critical mass’ of replication virus that is capable of dynamic evolution and immune escape4. To date, this pattern of protection has been consistently observed only with LAVs or SIV vaccines that are based on persistent cytomegalovirus vectors. These findings therefore support the exploitation of tissue-based, effector-differentiated memory T cell responses and their associated immediate and complete protection by the development of HIV and AIDS vaccines that are based on persistent, nonpathogenic vectors or on other strategies that are capable of eliciting and maintaining such effector memory T cell responses37.
A total of 38 adult male rhesus macaques (Macaca mulatta) of Indian genetic background were used in this study. All macaques were free of cercopethicine herpesvirus 1, simian type D retrovirus, simian T-lymphotropic virus type 1 and SIV infection at the start of the study, and were used with approval of the Oregon National Primate Research Center’s Animal Care and Use Committee, under the standards of the NIH Guide for the Care and Use of Laboratory Animals. Macaques with major histocompatibility complex alleles (Mamu-A*01, Mamu-B*08 and Mamu-B*17), which have previously been associated with spontaneous and/or LAV-associated control of wildtype SIV13, were excluded from this study.
SIVmac239(Δnef)1, SIVmac239(Δ3)18, SHIV89.65, and SIVsmE543(Δnef) were given to 6-8 macaques each intravenously at a dose of 3 × 104 focus forming units (ffu), 2 × 105 ffu, 2 × 103 ffu, and 3 × 104 ffu, respectively (all replicating SIV stocks quantified using the sMAGI cell assay41). SIVsmE543(Δnef) has a truncated Nef protein (the first N-terminal 58 amino acids flanked by C-terminal 16 amino acids VSFYKRKGGTGRDLLQ derived from SIVmac239). Single cycle SIVmac23919 was given intravenously three times at a dose of 5μg p27 each (by ELISA) at weeks 0, 4, and 8. All macaques were intravenously challenged with PBMC-expanded SIVmac239 at a dose of 1000 ffu at 50 weeks after LAV vaccination, with the exception of two SIVsmE543(Δnef)-vaccinated macaques that were challenged at 73 weeks post-vaccination using the same stock of wild-type SIVmac239 challenge virus.
Mononuclear cell preparations were obtained from blood, bone marrow, BAL, lymph nodes, spleen, liver, jejunum and colon as previously described 38,42,43. For cell-associated viral load determination, 1×106 cells were pelleted and frozen immediately in liquid nitrogen and then stored at −80°C until use. For the transcriptional analysis of lymph node, unfractionated lymph node mononuclear cells (1 × 106) or sorted lymph node cells (harvested at 7 days before and 4 and 14 days after challenge) were resuspended in RLT lysis buffer (Qiagen) and stored at −80°C until use. CD4+ and CD8+ memory T cells were isolated by sorting on a FACS Aria II (BD Biosciences) on the basis of cell surface staining patterns of the lineage markers CD3, CD4, CD8, and memory markers CD95 and CD28, using previously described criteria38. In some studies, CD4+ memory T cells were further sorted based on expression levels of PD-1 using mAb eBioJ105 (eBioscience).
Quantitative PCR and RT-PCR assays targeting a highly conserved nucleic acid sequence in gag were used for standard measurement of plasma SIV RNA and cell-associated SIV RNA/DNA in PBMC and cells isolated from tissues, as previously described44,45. The primers and probes used to discriminate SIVmac239 from SIVmac239(Δnef) and SIVmac239(Δ3) were described previously46. Potential mixtures of SIVsmE543(Δnef) and SIVmac239 were assayed using the SIVsmE543 gag-specific primers 5′-GCAGGGCAACTTAGAGAGCCAAG-3′ and 5′-TGGTAT(G/A)GGGTTTTGTTGCCTG-3′ and probe 5′CalFluor®Red610-TAGTTCCTGCAATGTCTGAT C-BHQ®2 and the SIVmac239 gag-specific primers 5′-GAGGCCCTCGCACCAGTG-3′ and 5′-CATGGTCCATTTTTCCACATTTC-3′ and probe 5′Quasar® 670-CAATCCCTTTTGCAGCAGCCCAACAG-BHQ®2. Potential mixtures of SHIV89.6 and SIVmac239 were assayed using the SHIVenv parental construct vpu-specific primers 5′-CTTTCATTGCCAAGTTTGTTTCATAAC-3′ and 5′-TCGTCGCTGTCTCCGCTTC-3′ and probe 5′Quasar®670-AAAGCCTTAGGCATCTCCTATGGCAGGA-BHQ®2 and the SIVmac239 env-specific primers 5′-GCApTTCAGAGGpGGATGC-3′ and 5′-CCCAAGACATCCCATACATTACTTGT-3′ and probe 5′CalFluor® Orange-ACTCCAGAATCGGCCAkpCTGGk-BHQ®1 where p and k bases (Glen Research) refer to pyrimidine and purine analogs, respectively, that permit pairing at base-divergent positions. All assays were optimized for maximal sensitivity of specific target detection of 5 copy equivalents per test reaction (95% assurance) and validated for minimum discriminate ratios of 105:1 for the indicated potential mixtures of virus sequences. The presence of inducible, replication competent LAV was detected in whole lymph node mononuclear cell preparations (or sorted cell populations obtained as described above) by co-cultivation of these cells with CEMx174 cells40, using flow cytometric analysis of intra-cellular SIVgag p27 (using SIVmac251 p27-specific mAb, ABL, Inc.) to detect infection47.
SIV-specific T cells were enumerated in mononuclear cells isolated from blood and tissues by flow cytometric intracellular cytokine analysis, as previously described in detail48. Briefly, mixes of sequential (11 amino acid overlapping) 15-mer peptides (AnaSpec) spanning the SIVmac239 gag, env, pol, nef, rev/tat, vif/vpr/vpx were used as antigens in conjunction with anti-CD28 (BD Biosciences) and CD49d co-stimulatory monoclonal antibodies (BD Biosciences). For the transcriptional analysis of lymph node, unfractionated lymph node mononuclear cells (1 × 106) or sorted lymph node cells (harvested at 7 days before and 4 and 14 days after challenge) were resuspended in RLT lysis buffer (Qiagen) and stored at −80 °C until use. Co-stimulation in the absence of peptides served as background control. Cells were stained with fluorochrome-conjugated monoclonal antibodies, and data were collected on an LSRII (BD Biosciences) and analyzed using the FlowJo software program (version 8.8.6; Tree Star, Inc.). Response frequencies (CD69+TNF-α+ and/or CD69+IFN-γ+) were first determined in the overall CD4+ and CD8+ population and then memory corrected (memory subset populations were delineated on the basis of CD28 and CD95 expression)48. The sum of net response frequencies determined for each peptide mix described above was considered the overall (pan-proteome) SIV-specific response frequency. To measure breadth of SIV-specific T cell responses in PBMC, we determined the frequency of CD4+ and CD8+ T cells that were responsive to 82 separate pools of ten 15mer peptides consisting of SIVmac239 proteins (env, pol, gag, nef, rev, tat, vif, vpr, and vpx with 22, 26, 13, 6, 2, 3, 5, 2, and 3 pools, respectively).Breadth was reported as the number of 10-peptide mixes with an above-threshold (0.05%) response. To determine the cell surface phenotype of SIV-specific CD8+ T cells, mononuclear cells were stimulated as described above, except that the CD28 co-stimulatory mAb was used as a fluorochrome conjugate to allow CD28 expression levels to be assessed by flow cytometry. After incubation, cells were stained on the surface for lineage markers (CD3, CD4, CD8) and memory/activation markers (CCR5, PD-1, HLA-DR) prior to fixation/permeabilization and then intracellular staining for response markers (IFN-γ, TNF-α). Viral suppression assays were based on the approach of Martins, et al.47. Briefly, autologous CD4+ T cells were isolated from PBMC by NHP CD4+ T cell isolation kits (Miltenyi Biotec), activated in vitro with staphylococcal enterotoxin B (1 μg ml−1; Sigma), anti-CD3, anti-CD28, and anti-CD49d antibodies, infected with SIVmac239 stock by centrifugation (1200g for 2 h, 30°C) and labeled with CFSE (2.5 μM) for 10 min at 37°C. These target cells were incubated with (or without) effector cells (sorted CD4+ or CD8+ memory T cells, obtained as described above) at an effector:target ratio of 1:1 memory for 72 h, and the percentage of SIVgag p27+ cells in the CFSE+ subset was determined by flow cytometric analysis of intra-cellular SIVgag p27 staining (mAb 55-2F12, NIH AIDS Research and Reference Reagent Program). The %viral suppression was calculated by the formula; 100 - (%p27 with effectors - %p27 in uninfected target cells) / (%p27 without effectors - %p27 in uninfected target cells) × 100. T cell phenotype was performed by flow cytometry as previously described49, using established criteria for memory T subsets, activated T cells and follicular helper T cells21,22,39,46,50,51.
Neutralizing antibodies against SIVmac239 and tissue culture-adapted SIVmac251 were quantified in luciferase reporter gene assays using the TZM-bl and M7 Luc cell lines, respectively52. SIVenv-specific antibodies with antibody-dependent cell-mediated cytotoxicity activity were measured as described53.
RNA was isolated using RNeasy Micro Kits (Qiagen) and the quantity and quality confirmed using a NanoDrop 2000c (Thermo Fisher Scientific) and an Experion Electrophoresis System. Samples (50 ng) were amplified using Illumina TotalPrep RNA amplification kits (Ambion). Microarray analysis was conducted using 750 ng of biotinylated cRNA hybridized to HumanHT-12_V4 BeadChips (Illumina) at 58°C for 20 h. The arrays were scanned using Illumina’s iSCAN and quantified using Genome Studio (Illumina). Analysis of the GenomeStudio output data was conducted using R (R Development Core Team)54 and Bioconductor55 software packages. Quantile normalization was applied, followed by a log2 transformation. The LIMMA package56 was used to fit a linear model to each probe and to perform (moderated) t-tests or F-tests on the compared groups. To control the expected proportions of false positives, the FDR for each unadjusted P value was calculated using the Benjamini and Hochberg method implemented in LIMMA. Multidimensional scaling was used as a dimensionality reduction method in R to generate plots for evaluation of similarities or dissimilarities between data sets. Ingenuity Pathway Analysis software (IPA, Ingenuity Systems) was used to annotate genes and rank canonical pathways. An immune response gene filter, constructed from innate and adaptive immune response gene queries of Gene Ontology (www.geneontology.org), T cell activation and complement pathway genes from IPA, and our own IFN- and inflammasome-response gene lists57 was used where noted to reduce datasets for visualization prior to FDR estimation. The microarray data are available through the National Center for Biotechnology Information Gene Expression Omnibus (GEO superseries GSE40006).
The goal of the primary statistical analysis was to predict post-challenge outcome from pre-challenge immune response variables (responses to the LAVs). The primary measure of outcome was the ordinal protection variable, the categories of which – completely protected, partially protected and non-protected – were determined and validated as described in the Results section. The eleven primary independent potential immune response predictors are summarized in Table 1. In all cases, we used immunologic values obtained from samples collected at the closest available time point before the challenge. To determine which, if any, of these immune response parameters predict outcome, we used univariate, nonparametric Kruskal-Wallis tests to assess each of the 11 primary predictor (immunologic) variables vs. the three protection categories. We considered a parameter to be a correlate of predictor of outcome if the P value resulting from this analysis was less than 0.05 after Bonferroni-correction for multiple comparisons. In follow-up analyses, we conducted Wilcoxon rank sum tests to compare predictors across pairs of the three protection categories. In addition, to analyze the degree of protection in those macaques with establishment of the wildtype SIVmac239 virus after challenge (partially protected and non-protected groups), we used the Spearman test to conduct nonparametric correlation analysis of immune parameters with a quantitative measure of outcome, the “maximum ΔPlasma Viral Load” (max. Δ pvl), which is the log10-transformed difference between the maximum of the post-challenge peak pvl (from PCD 4, 7, 10, 14, 21, 28) and the average baseline pvl (mean of pre-challenge time points: day −14, −4, and 0). We conducted all follow-up and secondary or exploratory statistical analyses without multiplicity correction, using nonparametric tests: Kruskal-Wallis, Wilcoxon, and Spearman tests for categorical, dichotomous, and correlation tests, respectively58. When comparing cell-associated viral load across tissue types, we applied Wilcoxon signed rank tests to the within-animal differences across tissues. We used R for all statistical analyses54.
This work was supported by the Bill & Melinda Gates Foundation (grant # 41185), the International AIDS Vaccine Initiative (IAVI), the National Institute of Allergy and Infectious Diseases [including NIH grant R37 AI054292 (Picker), contract #HHSN272200900037C and the Center for HIV-AIDS Vaccine Immunology (CHAVI) program], the NIH Office of Research Infrastructure Programs (P51 OD011092), and the National Cancer Institute (contract #HHSN261200800001E). The authors gratefully acknowledge R. Desrosiers (Harvard University) for providing SIVmac239Δnef and SIVmac239Δ3; P. Johnson and T. Lui (University of Pennsylvania) for SIVsmE543Δnef; C. Miller (University of California, Davis) for SHIV89.6 and wildtype SIVmac239; D. Evans (Harvard University) for a single cycle SIVmac239; N. Letvin for TRIM5 allele typing; and R. Wiseman and D. Watkins for MHC typing. We thank N. Winstone, A. Leon, J. Clock, A. Nogueron, L. Pan, M. Cartwright, A. Filali and P. Wilkinson for technical assistance, and J. McElrath, S. Self, W. Koff, A. Okoye, J. Schmitz, and J. Ahler for helpful discussion and advice.
AUTHOR CONTRIBUTIONS: YF, RL, NC, EM, SH and SGH performed experiments and analyzed data assisted by MDR and JYB. HP managed the project and analyzed data, assisted by AS. TS, AWL and MKA managed the animal protocols. MP Jr. and JDL provided SIV/LAV quantification, assisted by RS, YL and KO. DM and GF provided neutralizing Ab and cytotoxic Ab quantification, respectively. MJC, FL, ATS, PS and RPS carried out the microarray analysis and interpreted the results. PTE performed the statistical analysis. LJP conceived this study, supervised experiments, analyzed data, and wrote the paper, assisted by YF, HP, PTE, AM, RPS and JDL.
AUTHOR INFORMATION: The authors have no competing financial interests to declare.