Research ArticleHIV

A live-attenuated RhCMV/SIV vaccine shows long-term efficacy against heterologous SIV challenge

See allHide authors and affiliations

Science Translational Medicine  17 Jul 2019:
Vol. 11, Issue 501, eaaw2607
DOI: 10.1126/scitranslmed.aaw2607

Building a safer CMV vector

Vaccine vectors based on cytomegalovirus (CMV) show strong T cell induction and protection against a multitude of pathogens. However, CMV can be harmful to people who are immunodeficient or immunosuppressed. Marshall et al. genetically modified rhesus CMV to allow engagement of host intrinsic immunity. The modified ΔRh110 vector did not spread once administered to nonhuman primates but still induced robust T cell immunity. Hansen et al. showed in a simian immunodeficiency virus (SIV) challenge model that the ΔRh110 vector provided equivalent protection to the parental vector, enabling control and progressive clearance of virus from more than half of the vaccinated primates. Most protected animals that were rechallenged 3 years later were able to control the second challenge, demonstrating the durability of this vaccine. Mutations in the human CMV vector should lead to a potent but restrained CMV that could be used widely in people.


Previous studies have established that strain 68-1–derived rhesus cytomegalovirus (RhCMV) vectors expressing simian immunodeficiency virus (SIV) proteins (RhCMV/SIV) are able to elicit and maintain cellular immune responses that provide protection against mucosal challenge of highly pathogenic SIV in rhesus monkeys (RMs). However, these efficacious RhCMV/SIV vectors were replication and spread competent and therefore have the potential to cause disease in immunocompromised subjects. To develop a safer CMV-based vaccine for clinical use, we attenuated 68-1 RhCMV/SIV vectors by deletion of the Rh110 gene encoding the pp71 tegument protein (ΔRh110), allowing for suppression of lytic gene expression. ΔRh110 RhCMV/SIV vectors are highly spread deficient in vivo (~1000-fold compared to the parent vector) yet are still able to superinfect RhCMV+ RMs and generate high-frequency effector-memory–biased T cell responses. Here, we demonstrate that ΔRh110 68-1 RhCMV/SIV–expressing homologous or heterologous SIV antigens are highly efficacious against intravaginal (IVag) SIVmac239 challenge, providing control and progressive clearance of SIV infection in 59% of vaccinated RMs. Moreover, among 12 ΔRh110 RhCMV/SIV–vaccinated RMs that controlled and progressively cleared an initial SIV challenge, 9 were able to stringently control a second SIV challenge ~3 years after last vaccination, demonstrating the durability of this vaccine. Thus, ΔRh110 RhCMV/SIV vectors have a safety and efficacy profile that warrants adaptation and clinical evaluation of corresponding HCMV vectors as a prophylactic HIV/AIDS vaccine.


Although the advent of antiretroviral therapy and other preventative interventions have greatly reduced the number of new infections and AIDS-related deaths from their peak incidence, epidemiologic modeling suggests that an efficacious HIV vaccine will still be necessary to reduce the annual incidence of new HIV infections to a degree commensurate with ending the epidemic (1, 2). However, despite many decades of concerted effort, vaccine platforms capable of eliciting protective immune responses against HIV or its nonhuman primate counterpart simian immunodeficiency virus (SIV) are few because these viruses are highly immune evasive and either lack susceptibility to natural immunity or rapidly escape immune responses that are initially effective (3). We hypothesized a number of years ago that immune control of these viruses might be possible if infection could be immediately intercepted at portals of viral entry and sites of early spread by preestablished effector-differentiated CD8+ T cell responses either in place in these sites (resident memory cells) or rapidly recruited from the blood (circulating effector-memory cells) without the need for the “too little, too late” process of anamnestic memory T cell expansion, effector differentiation, and trafficking to sites of infection (4). Because human cytomegalovirus (HCMV) and rhesus CMV (RhCMV) naturally elicit and maintain effector-memory T cell responses having these properties, we investigated the possibility of exploiting CMV as a vaccine vector, using RhCMV/SIV vectors based on the 68-1 RhCMV strain in the RM (rhesus monkey)–SIV model as proof of principle (5).

In a series of reports, we demonstrated that a 68-1 RhCMV/SIV vaccine expressing SIV Gag, Rev/Nef/Tat, Pol, and Env was able to super-infect naturally RhCMV-infected RMs and elicit and indefinitely maintain SIV-specific CD4+ and CD8+ T cell responses that closely mimicked the characteristics of responses to RhCMV itself: high frequency, widely distributed in lymphoid and nonlymphoid sites, and effector differentiated (highly effector-memory biased) (68). Insert-specific antibody (Ab) responses were absent in most vaccinated RMs, and unexpectedly, the CD8+ T cell responses elicited by these vectors manifested unconventional epitope targeting characterized by extraordinary breadth and response restriction by either major histocompatibility complex II (MHC-II) or MHC-E but not MHC-Ia, an immunologic feature that was found to be related to genetic changes in the 68-1 RhCMV strain associated with adaptation to in vitro fibroblast culture (9, 10). Although the significance of this unusual CD8+ T cell antigen (Ag) recognition remains an area of active investigation, the hypothesis that preestablished circulating and tissue-based effector-differentiated cellular immune responses might be more efficacious than conventional memory responses is supported by multiple SIV challenge studies, showing that more than half of 68-1 RhCMV/SIV–vaccinated RMs manifested an early and complete control of SIVmac239 infection after mucosal challenge. Protected RMs were found to be definitively infected after challenge, but viral spread appeared to be completely arrested before establishment of a permanent viral reservoir, and the infection was progressively cleared over the ensuing months, until protected RMs became indistinguishable by both virologic and immunologic criteria from animals that were never challenged (8, 11).

This “control and clear” vaccine effect against highly pathogenic SIV has not been reported for any other vaccine modality and, if translatable to humans, has the potential to contribute to control of the HIV epidemic either as a stand-alone vaccine or in combination with Ab-targeted vaccines (12). However, translating CMV-based vectors as a prophylactic vaccine in humans requires careful consideration of safety. Although the vast majority of HCMV infections in people and RhCMV infections in monkeys are clinically inapparent, these viruses have the capacity to cause serious disease in settings of immunodeficiency, with maternal to fetal transmission being of particular concern (13). 68-1 RhCMV, the parent strain of vectors used in the above-described efficacy studies, lacks subunits of a pentameric glycoprotein complex that facilitate viral entry into most nonfibroblast cells (14,15) and, presumably as a result of this restricted tropism, demonstrates reduced viremia, shedding, and horizontal transmission compared to wild-type (WT) RhCMV (16,17). Nevertheless, 68-1 RhCMV retains the ability to disseminate in infected RMs and transmit from one monkey to another and has the potential to cause disease (1820). “68-1–like” HCMV vectors thus still carry some potential risk for vaccine-mediated disease in otherwise healthy populations. The challenge then becomes whether CMV can be further genetically attenuated such that it retains the ability to super-infect, elicit and maintain effector-differentiated T cell responses (including the unconventionally targeted CD8+ T cell responses), and control and clear protective efficacy while losing the ability to widely disseminate in the body, to spread from individual to individual, and to cause disease in settings of immunodeficiency. Moreover, any genetic attenuation must be stable, minimize the likelihood for reversion by mutation or recombination, allow vector manufacture at scale, and involve a virologic mechanism that is conserved between RhCMV, where the concepts will be tested, and HCMV, which will serve as the basis of any clinical vector.

In a companion report (20), we describe a CMV attenuation strategy based on deletion of the Rh110 gene (RhCMV ortholog to HCMV UL82), which encodes pp71, a tegument phosphoprotein that functions to disperse and/or degrade the host intrinsic immunity protein death domain–associated protein (DAXX). DAXX functions in nuclear ND10 bodies to repress transcription of viral immediate early genes, which are critical for early and late CMV gene expression and, thus, viral genome replication, assembly, and egress (2127). In the absence of viral pp71, DAXX represses lytic CMV replication, and the infection becomes and remains latent. Although DAXX can be overcome at high multiplicities of infection in vitro, ΔRh110 (Δpp71) RhCMV is highly spread deficient in vivo, with infection largely restricted to the inoculating dose at the site of inoculation and draining lymph nodes (20). In contrast to parental 68-1 RhCMV, ΔRh110 68-1 RhCMV was not shed in urine, nor transferred to new hosts by close contact or adoptive cell transfer, and this attenuation was stable over time with no signs of reversion in vivo. Despite this attenuation, the SIV insert–specific T cell immunogenicity of ΔRh110 68-1 RhCMV/SIV vectors was similar to their Rh110-intact counterparts in terms of magnitude, durability, effector-memory phenotype, and function and, for the CD8+ T cell responses, in both breadth and unconventional epitope targeting (20). Here, we investigate whether spread-deficient ΔRh110 68-1 RhCMV/SIV vectors can also provide the same control and clear protection against homologous SIV challenge as their spread-competent Rh110-intact counterparts and additionally assess whether these attenuated 68-1 RhCMV/SIV vectors can protect against challenge with a heterologous SIV strain.


Experimental design and ΔRh110 RhCMV/SIV vector immunogenicity

We previously reported vaccination of cycling female RMs with spread-competent (Rh110 intact) strain 68-1 RhCMV/SIV vectors expressing SIV Gag, Rev/Tat/Nef (RTN), 5′-Pol, 3′-Pol, and Env (with inserts primarily based on SIVmac239 gene sequences). This vaccination provided stringent aviremic (except for transient plasma viremia in early infection) long-term (>52 weeks) control of intravaginal (IVag)–introduced SIVmac239 infection in 8 of 16 vaccinated RMs versus 0 of 18 controls (8). Here, we sought to determine (i) whether this homologous efficacy would extend to vaccination with an attenuated (spread-deficient) ΔRh110 68-1 RhCMV/SIV vector set expressing the same predominantly SIVmac239 sequence inserts (ΔRh110/SIVmac239) and (ii) the extent to which a ΔRh110 68-1 RhCMV/SIV–vectored vaccine with heterologous SIVsmE660-sequence inserts (ΔRh110/SIVsmE660) would provide protection against the same SIVmac239 challenge regimen. We elected a heterologous vaccine rather than heterologous challenge approach because available heterologous challenge strains—the SIVsmE660 swarm or SIVsmE543 clone—are sufficiently different in key immunobiologic characteristics from SIVmac239 to make direct efficacy comparisons difficult (28, 29). To this end, two ΔRh110 RhCMV/SIV vector sets were constructed from the parental 68-1 RhCMV bacterial artificial chromosome, each with the coding region of the Rh110 (pp71) gene replaced with the SIV insert (Gag, 5′-Pol, 3′-Pol, RTN, and Env), derived from either the SIVmac239 sequence or an SIVsmE660 consensus sequence (fig. S1A). Divergence between the SIVmac239 and SIVsmE660 amino acid sequences averages 15% across all inserts (fig. S2), differences that approximate the variation between single clade–based HIV vaccines and circulating HIV isolates within that clade (30). Two groups of 14 female RMs each were vaccinated twice (weeks 0 and 14) with the set of five ΔRh110/SIVmac239 (group 1) or ΔRh110/SIVsmE660 vectors (group 2) expressing Gag, RTN, 5′-Pol, 3′-Pol, or Env inserts by subcutaneous administration of 5 × 106 plaque-forming units per vector (Fig. 1A). Immunogenicity was followed for 60 weeks after initial vaccination, at which time repeated, limiting-dose, IVag SIVmac239 challenge was initiated for both vaccine groups and a cohort of unvaccinated controls (group 3; n = 20). Immunogenicity and outcome of groups 1 and 2 were also compared to our previously reported cohort of female RMs (group 4) vaccinated with a set of WT (Rh110 intact) 68-1 RhCMV/SIV vectors (WT 68-1/SIVmac239) expressing the same SIVmac239 sequence inserts that were IVag SIVmac239–challenged by a similar limiting dose protocol (8).

Fig. 1 Immunogenicity of ΔRh110 RhCMV/SIV vectors.

(A) Schematic of the RM groups analyzed in this study. (B) Longitudinal and plateau-phase analysis of the vaccine-elicited, SIV Gag, Rev/Tat/Nef (RTN), Pol, and Env insert–specific CD4+ and CD8+ T cell responses in peripheral blood. In the top panel, the background-subtracted frequencies of cells producing TNF and/or IFN-γ by flow cytometric ICS assay to peptide mixes comprising each of the SIV inserts (SIVmac239 sequence) within the memory CD4+ or CD8+ T cell subsets were summed for overall responses, with the figure showing the mean (+ SEM) of these overall responses at each time point. In the bottom panel, boxplots compare the overall and individual SIV insert–specific CD4+ and CD8+ T cell response frequencies between the vaccine groups at the end of the vaccine phase (each data point is the mean of response frequencies in all samples from weeks 30 to 58 after first vaccination). Two-sided Wilcoxon rank-sum tests were used to compare the significance of differences in plateau-phase response frequencies between group 1 and group 2 (SIVmac239 versus SIVsmE660 inserts in ΔRh110 68-1 vectors) and between group 1 and group 4 (SIVmac239 inserts in WT 68-1 versus ΔRh110 68-1 vectors). (C) Boxplots compare the memory differentiation of the vaccine-elicited CD4+ and CD8+ memory T cells in peripheral blood responding to SIV Gag peptide mix (SIVmac239 sequence) with TNF and/or IFN-γ production at the end of vaccine phase (week 54 for groups 1 and 2; week 60 for group 4). Memory differentiation state was based on CD28 and CCR7 expression, delineating central memory (TCM), transitional effector-memory (TTrEM), and effector-memory (TEM), as designated. Two-sided Wilcoxon rank-sum tests were used to compare the significance of differences in the fraction of responding cells with a TCM phenotype (reciprocal of fraction with effector differentiation − TTrEM + TEM). (D) Same analysis as in (B), but for responses in lung airspace (BAL). Each data point for the boxplots is the mean of response frequencies in all samples from weeks 30 to 54 after first vaccination. (E) Boxplots show plateau-phase analysis (each point is the average of all samples between weeks 24 and 30 after first vaccination) of the vaccine-elicited CD8+ T cell responses to SIV Gag supertopes (SIVmac239 sequence; fig. S1B) in peripheral blood of group 1, group 2, and group 4 RMs by the same ICS assay described above. Gag276–284 (69) and Gag482–490 (120) are MHC-E–restricted supertopes; Gag211–222 (53) and Gag290–301 (73) are MHC-II–restricted supertopes (9, 10). Statistical testing was performed as described in (B). In all panels, n = 14, 14, and 16, respectively, for groups 1, 2, and 4, except group 4 in (E), where n = 10. Analyses were adjusted for multiple comparisons across inserts (B and D), epitopes (C), and supertopes (E) using the Holm method, and P values of ≤0.05 were considered significant. Analyses of total responses (B and D) were not adjusted.

RhCMV vectors are T cell–targeted vaccines with little to no ability to elicit insert-specific Ab responses (7, 8, 31), and in keeping with this, only 3 of 28 RMs in groups 1 and 2 (all in group 2) showed detectable SIV Env–specific Abs after vaccination, and these 3 responses were very low titer (fig. S3A). In contrast, using flow cytometric intracellular detection of CD69 and either or both of tumor necrosis factor (TNF) and interferon-γ (IFN-γ) as the indicator of Ag-triggered T cells responding to pools of consecutive overlapping SIVmac239 sequence 15-mer peptides, all RMs in both ΔRh110/SIV vector–vaccinated groups developed CD4+ and CD8+ T cell responses in blood to all SIV inserts (Fig. 1B). In both group 1 and group 2, the overall response peaked 2 to 4 weeks after initial or boost vaccinations before establishing a stable steady state within ~12 weeks of the second vaccination that was maintained for the duration of the vaccine phase. During the plateau phase of the vaccine response (defined here as weeks 30 to 58 after initial vaccination), total SIV-specific CD4+ and CD8+ T cell response frequencies in peripheral blood, as measured by the response to SIVmac239 sequence peptides, were significantly higher overall in group 1 RMs, given ΔRh110/SIVmac239 vectors, than in group 2 RMs, given ΔRh110/SIVsmE660 vectors (P < 0.001; Fig. 1B, bottom panel). This difference in overall response magnitude in blood was primarily driven by differences in the responses to Env and RTN (Fig. 1B, bottom panel), the SIV inserts with the most divergence between the SIVmac239 and SIVsmE660 sequences (fig. S2).

We also determined the memory differentiation phenotype of the SIV Gag–specific CD4+ and CD8+ T cells at plateau phase in group 1 and 2 RMs by intracellular cytokine staining (ICS), delineating central memory T cells (TCM), transitional effector-memory T cells (TTrEM), and effector-memory T cells (TEM) by their expression of CCR7 versus CD28 (Fig. 1C). This analysis showed a predominance of effector-differentiated cells (TTrEM + TEM) that was similar in both vaccine groups. In keeping with this, SIV-specific CD4+ and CD8+ T cells were enriched in bronchoalveolar lavage (BAL) fluid samples (used as an accessible effector site), and despite using SIVmac239 sequence peptides for this analysis, the magnitude of the overall and individual SIV insert-specific, CD4+, and CD8+ T cell responses in BAL was not different for group 1 and 2 RMs (Fig. 1D).

Both Rh110-intact and Rh110-deleted 68-1 RhCMV vectors elicit CD8+ T cell responses that are entirely unconventional in their MHC restriction (with epitopes presented by MHC-E or MHC-II, not MHC-Ia). Moreover, RMs vaccinated with SIVmac239 Gag-expressing 68-1 RhCMV vectors invariably respond to a set of universal MHC-E– and MHC-II–restricted CD8+ T cell epitopes [so-called supertopes (9, 10, 20)]. All RMs in groups 1 and 2 manifested CD8+ T cell responses to all four of the previously characterized SIVgag supertopes tested (two MHC-E restricted and two MHC-II restricted), and the magnitude of all these universal responses (three of which were sequence identical in both the SIVmac239 and SIVsmE660 inserts and one of which was different by three amino acid substitutions; fig. S1B) was not different between group 1 and group 2 RMs (Fig. 1E).

We next compared the magnitude and phenotype of responses elicited by the ΔRh110/SIVmac239 vectors (group 1) with results from our previously reported cohort of female RMs vaccinated with WT 68-1/SIVmac239 vectors (group 4) (8). As shown in Fig. 1 (B to E), neither the magnitude (blood or BAL) nor the TEM + TTrEM skewing (blood) of the various SIV-specific CD8+ T cell responses, including supertope-specific responses, was different between the two groups. However, the magnitude of plateau-phase SIV-specific CD4+ T cell responses in blood and BAL was significantly higher (P < 0.001 for both), and in blood, the SIVgag-specific CD4+ T cell responses were significantly more TEM + TTrEM biased (i.e., lower %TCM; P = 0.013) in the group 4 RMs compared to the group 1 RMs. These observations suggest that the restricted spread of the ΔRh110 vectors (20), and likely diminished Ag availability, modestly reduced CD4+ T cell immunogenicity and effector differentiation while having little to no effect on CD8+ T cell responses.

To explore heterologous T cell immunity with RhCMV vectors, we directly compared the ability of vaccine-elicited T cell responses of both group 1 and group 2 RMs to recognize and respond to SIVmac239 versus SIVsmE543 sequence peptides. The SIVsmE660 swarm-derived SIVsmE543 clone (29) is 96% identical to the SIVsmE660 consensus amino acid sequence and has a similar 15% overall amino acid sequence divergence from the SIVmac239 (see fig. S2). For CD8+ T cells, we also examined responses to autologous CD4+ T cells infected with the cloned SIVmac239 or SIVsmE543 viruses (Fig. 2). CD4+ T cells from ΔRh110/SIVmac239-vaccinated group 1 RMs showed no difference in their plateau phase responses to matched (SIVmac239) versus mismatched (SIVsmE543) peptide mixes, whereas CD8+ T cells from the same RMs showed a significant reduction (average = 31%; P = 0.017) in the overall frequency of cells able to respond to the mismatched peptides (Fig. 2A). For ΔRh110/SIVsmE660-vaccinated group 2 RMs, both CD4+ and CD8+ T cells recognized mismatched peptides significantly less well than matched peptides, with the reduction in the magnitude of the CD4+ and CD8+ T cell response to mismatched peptides being ~21% (P < 0.001) and ~44% (P = 0.002) less, respectively, than for matched peptides (Fig. 2B). Thus, mismatch between the insert sequence and stimulating peptide sequence modestly reduced the magnitude of ΔRh110/SIVmac239/smE660 vector–elicited CD8+ T cell responses. However, CD8+ T cells from both group 1 and group 2 RMs showed equivalent ability to recognize autologous CD4+ T cells infected with SIVmac239 or SIVsmE543 virus clones (Fig. 2C), suggesting that at the level of SIV-infected cell recognition, the breadth of the CD8+ T cell responses generated by both ΔRh110/SIVmac239 and ΔRh110/SIVsmE660 vector sets was able to overcome sequence mismatch in individual epitopes.

Fig. 2 Cross-recognition by ΔRh110 RhCMV/SIVmac239 and RhCMV/SIVsmE660 vector–elicited T cells.

(A and B) Flow cytometric ICS analysis of SIV-specific CD4+ and CD8+ T cell response frequencies (using TNF and/or IFN-γ readout in memory subset) in the blood of group 1 (n = 14; SIVmac239 inserts) and group 2 (n = 14; SIVsmE660 inserts) RMs in plateau phase (week 44 after first vaccination) comparing recognition of matched versus mismatched peptide mixes (SIVmac239 versus SIVsmE543; see fig. S2), including overall (summed) responses and responses to each SIV insert. Two-sided paired Wilcoxon rank-sum tests were used to compare the significance of differences in matched versus mismatched peptide mix recognition. Unadjusted (total responses) or Holm-adjusted (each insert-specific response) P values of ≤0.05 were considered significant. When significant differences were observed (reduction in response frequencies with mismatched peptide mixes), the median effect size (% reduction with mismatch) is shown. (C) ICS analysis of CD8+ T cell recognition of autologous CD4+ T cells infected with the SIVmac239 versus SIVsmE543 viruses (after background subtraction of the response to mock-infected autologous CD4+ T cells) in plateau phase (between weeks 49 and 57 after first vaccination). Statistical analysis was performed as described above, with n = 12 and 13 for groups 1 and 2, respectively.

Efficacy of ΔRh110 RhCMV/SIV vectors

To determine whether spread-deficient ΔRh110/SIVmac239/smE660 vectors retain the ability to mediate the characteristic control and clear protection demonstrated by spread-competent WT 68-1/SIVmac239 vectors in previous reports (7, 8, 12), we subjected the vaccinated group 1 and 2 RMs, and the unvaccinated group 3 RMs, to repeated (up to 12 challenges at 2- to 4-week intervals), limiting-dose (100 focus-forming units for first six exposures; 300 for last six exposures) IVag SIVmac239 challenge. The goal was to establish infection “take” in each RM (at which time challenges were stopped) and then determine nonprotection versus protection by the presence or absence of progressive SIV infection after infection establishment (7, 8, 12). Because protected RMs may or may not manifest detectable viremia after challenge, SIV infection take is confirmed by the onset of de novo T cell responses to SIV Vif, an SIV Ag not included in any of the RhCMV/SIV vectors (7, 8). In our challenge system, SIV Vif–specific T cells (CD4+ and CD8+) appear in blood 2 to 3 weeks after productive SIV challenge, allowing attribution of successful (infection take positive) challenges when successive challenges are 2 or more weeks apart. With this approach, we were able to establish productive SIV infection in 13 of 14, 14 of 14, and 17 of 20 RMs after up to 12 challenges in groups 1 to 3, respectively, with no statistically significant difference in the rate of infection acquisition in the three challenge groups or in the overall ΔRh110 68-1 RhCMV/SIV vector–vaccinated cohort versus unvaccinated controls (fig. S4).

In keeping with previous observations on WT 68-1/SIVmac239 vector efficacy (7, 8), the outcome of productive SIV challenge was notably different in the vaccinated groups 1 and 2 versus the unvaccinated group 3. Whereas all 17 SIV-infected unvaccinated control RMs manifested typical systemic SIV infection, 7 of 13 group 1 RMs (54%; P = 0.0004) and 9 of 14 group 2 RMs (64%; P < 0.0001) showed the onset of de novo SIV Vif–specific T cell responses in the absence of SIV viremia (n = 3 and n = 5 for groups 1 and 2, respectively) or with plasma viremia positive at only a single time point (n = 4 for both groups 1 and 2; Fig. 3, A and B). To confirm the take of SIV infection in the presumptively protected (SIV Vif response positive) RMs without detectable plasma viremia, we performed adoptive transfer of bone marrow (BM) cells alone or BM cells plus peripheral blood mononuclear cells (PBMCs) obtained after the onset of SIV Vif–specific T cell responses from six of these RMs (three each from groups 1 and 2) into SIV-naïve recipients (Fig. 3C). As shown in the figure, adoptive transfer of cells from all six donor RMs resulted in the onset of typical SIVmac239 infection in recipient RMs, demonstrating the presence of fully replication-competent SIVmac239 in the donor RMs and confirming stringent SIV control in these animals. In addition, in keeping with previous results (7,8), there was no reduction in chronic phase plasma viremia in unprotected, vaccinated RMs relative to unvaccinated controls, consistent with the “all or none” nature of RhCMV/SIV vaccine efficacy. The degree (% protected) and pattern of efficacy observed in groups 1 and 2 were not significantly different from the previously reported efficacy of WT 68-1/SIVmac239 vector–vaccinated RMs (group 4) subjected to a similar challenge protocol [56% with initial stringent control; (8)]. Across all protected versus unprotected group 1 plus group 2 RMs, efficacy was not predicted by the magnitude of overall or individual insert, SIVmac239 peptide–specific CD4+ or CD8+ T cell responses, or supertope-specific CD8+ T cell responses in blood at peak post-prime, peak post-boost, or at vaccine response plateau phase, or by the magnitude of CD8+ T cell recognition of SIVmac239-infected CD4+ T cells at vaccine response plateau phase (fig. S5).

Fig. 3 Efficacy of ΔRh110 RhCMV/SIV vectors.

(A and B) Assessment of the outcome of effective challenge by longitudinal analysis of plasma viral load (A) and de novo development of SIV Vif–specific CD4+ (B, top panel) and CD8+ (B, bottom panel) T cell responses. RMs were challenged until the onset of any above-threshold SIV Vif–specific T cell response, with the SIV dose administered 2 or 3 weeks before this response detection considered the infecting challenge (week 0). RMs with sustained viremia were considered not protected (black); RMs with no or transient viremia were considered protected (red) (8). The fraction of protected RMs in the vaccinated groups (groups 1 and 2, n = 13 and 14, respectively) were compared to that of the unvaccinated group (group 3, n = 17) by Barnard’s exact test of binomial proportions, with the P values shown in (A). (C) BM cells and PBMCs were collected and cryopreserved from ΔRh110/SIVmac239/smE660 vaccine–protected RMs without any detectable viremia (RMs #1 to #3 from group 1; RMs #4 to #6 from group 2) at the indicated time points post-effective challenge (left panel; PID, post-infection day). Cells were thawed and administered intravenously (left panel) to six SIV-naïve RMs to assess the presence of replication-competent SIV, with the plasma viral dynamics in recipient RMs shown (right panel).

As previously shown for protection against SIVmac239 challenge mediated by spread-competent WT 68-1/SIVmac239 vectors, the stringent control of SIVmac239 infection mediated by the ΔRh110/SIVmac239/smE660 vectors occurred in the absence of an increased (boosted) SIV Gag– or SIV Pol–specific T cell response in blood after infection (fig. S6) and without development or boosting of an SIV Env–specific Ab response (fig. S3B). The lack of T cell response boosting was also observed after infection in unprotected (viremic) vaccinated RMs, indicating that the lack of increased T cell responses in protected RMs was not due to limitation in SIV Ag availability. However, vaccinated, unprotected RMs developed high-titer SIV Env–specific Ab responses after challenge (similar to unvaccinated controls), indicating that the lack of such Ab responses in protected vaccinated RMs is almost certainly a function of SIV Ag limitation due to early arrest of infection (keeping Ag levels below the threshold needed for Ab response generation). The conclusion that vaccinated, protected RMs have early arrest of viral spread after initial take of infection, sharply limiting the extent of SIV infection, is also supported by the lack of the activation of circulating monocytes [as measured by increased IFN-induced expression of CD169; (11, 32, 33)] specifically in protected RMs (fig. S7). Together, these results demonstrate that spread-deficient ΔRh110/SIVmac239/smE660 vaccines manifest efficacy equivalent to their spread-competent counterparts, which is not affected by a sequence mismatch between vector insert and challenge strain.

SIV dynamics in ΔRh110 RhCMV/SIV vector–vaccinated, protected RMs

We have previously demonstrated that in RMs protected by WT 68-1/SIVmac239 vector vaccination, the arrest of SIV infection occurs after initial dissemination via both lymphatic and hematogenous routes, the latter including seeding of liver, spleen, and BM. Over extended follow-up, cells harboring SIV slowly disappear from all tissue sites until both virologic and immunologic evidence of SIV infection is lost (8). To determine whether RMs protected by spread-deficient ΔRh110/SIVmac239 or ΔRh110/SIVsmE660 vector vaccination have similar post-infection SIV dynamics, we quantitated cell-associated SIV DNA and RNA in blood and BM of all protected RMs in groups 1 and 2 for up to 60 weeks after SIV infection. As shown in Fig. 4 (A and B), as expected, overtly infected control RMs showed abundant cell-associated SIV RNA and DNA in both blood and BM at all tested time points. In contrast, vaccine-protected RMs in both groups 1 and 2 manifested only sporadic detection of cell-associated virus in blood over 60 weeks of observation (Fig. 4C), consistent with the arrest of progressive SIV infection in these monkeys. Most notable, however, were the SIV dynamics in BM, previously shown to be a common site of early SIV spread in 68-1 RhCMV/SIV vector–protected RMs (8). As shown in Fig. 4D, all but 1 of the 16 protected group 1 and 2 RMs manifested cell-associated SIV RNA in BM 4 weeks after infection, comparable to unvaccinated controls, and cell-associated SIV DNA was also detected in the majority of these BM samples. Similar quantities of cell-associated SIV RNA and DNA were detected in most of the BM samples from these RMs at week 8 as well, but starting at week 12, there was a clear decline in cell-associated SIV in BM, and by week 20, SIV RNA and DNA were below the limit of detection in all BM samples from all RMs. The difference in the number of SIV RNA- and DNA-positive samples from <20 weeks and ≥20 weeks after infection was significant (P < 0.0001, Barnard’s exact test of binomial proportions).

Fig. 4 Clearance of cell-associated SIV in the BM of ΔRh110 68-1 RhCMV/SIV vector–protected RMs.

(A to D) Longitudinal analysis of PBMC-associated (A and C) and BM cell–associated (B and D) SIV RNA (left panels) and DNA (right panels) from 3 randomly selected unvaccinated RMs with progressive infection (A and B) and all 16 ΔRh110/SIVmac239/smE660 vector–protected RMs in groups 1 and 2 (C and D).

To more globally assess the “total body” SIV infection burden in vaccine-protected group 1 and 2 RMs, we longitudinally followed SIV Vif–specific T cell responses as an in vivo circulating immunologic “biosensor” to detect residual SIV infection–related Ag production in these animals, all of which were aviremic except for rare low-level viral blips before week 34 after infection (Fig. 5A). As noted above, SIV Vif–specific T cell responses are generated and maintained by SIV infection–derived Ag; in WT 68-1/SIVmac239 vector–vaccinated RMs, we have previously associated decline in these responses with progressive clearance of SIV reservoirs (8). All group 1 and 2 protected RMs showed a similar overall pattern of SIV Vif–specific response dynamics characterized by increasing or stable, high frequencies of SIV Vif–specific CD4+ and CD8+ T cells over the first 6 to 12 weeks after infection. Thereafter, there is a slow but unequivocal decline in these frequencies that starts no later than week 20 and continues to extinction (e.g., response below detection limit in blood) over the subsequent 1 to 2 years (Fig. 5B), a pattern that is notably similar to data with the WT 68-1/SIVmac239 vaccine (8). The slope of decline of SIV Vif–specific CD4+ and CD8+ T cell responses in ΔRh110/SIV vaccine–protected group 1 and 2 RMs was not significantly different from that of RMs protected by spread-competent WT 68-1/SIVmac239 vectors (Wald test, F2,582 = 0.097 and 2.10 for CD4+ and CD8+, respectively; Fig. 5C).

Fig. 5 Loss of circulating SIV infection–induced, SIV Vif–specific T cells in ΔRh110 68-1 RhCMV/SIV vector–protected RMs.

(A) Long-term longitudinal analysis of plasma viral load in ΔRh110/SIVmac239/smE660 vector–protected (left and middle panels for groups 1 and 2, respectively) and WT 68-1/SIVmac239 vector–protected RMs [group 4, right panel; (8)]. (B) Long-term longitudinal analysis of SIV Vif–specific CD4+ (top panels) and CD8+ (bottom panels) among the same groups of ΔRh110 and WT 68-1 RhCMV/SIV vector–protected RMs, with the figure showing the mean (+SEM) of these SIV Vif–specific T cell response frequencies in the memory subset at each time point. (C) Wald tests comparing the slope (±95% confidence intervals) of decline of log-transformed SIV Vif–specific CD4+ (left panel) and CD8+ (right panel) T cell response frequencies. Calculation of slopes is described in Materials and Methods. In all analyses, n = 7, 9, and 8 for groups 1, 2, and 4, respectively.

To confirm that the observed loss of SIV Vif–specific T cell responses reflected total body SIV clearance, we selected four of the ΔRh110/SIVmac239/smE660-vaccinated long-term protected RMs (>100 weeks after infection; two RMs each from groups 1 and 2) for detailed virologic and immunologic analysis at necropsy. Three of these four RMs (RMs #7, #9, and #10) had previously manifested a single plasma viral blip early after infection and subsequently remained aviremic, whereas the fourth RM (RM #8) was aviremic throughout its course. All four of these RMs had developed and then lost robust SIV Vif–specific T cell responses while maintaining stable (vaccine maintained) SIV Gag– and SIV Pol–specific T cell responses (fig. S8). At necropsy, all animals had SIV Gag– and SIV Pol–responsive T cells in all tissues examined (Fig. 6A). In contrast, SIV Vif–specific T cell responses were predominantly negative in three of four RMs (RMs #7 to #9), with above-threshold responses in only a few tissues, and, in the other RM (RM #10), were present as low-frequency responses (predominantly CD8+) in multiple sites (Fig. 6B). Cell-associated SIV RNA and DNA were quantified by nested quantitative reverse transcription polymerase chain reaction (RT-PCR)/PCR (8) in extensively sampled tissues from four ΔRh110/SIVmac239/smE660 vaccine–protected RMs (Fig. 6C) and, for comparison, tissues from two ΔRh110/SIVgag vector–vaccinated RMs never exposed to SIV (Fig. 6D) and one unvaccinated RM with progressive SIV infection (Fig. 6E). Both of the ΔRh110/SIVgag-vaccinated unchallenged control RMs were negative for SIV DNA and RNA in all tissues, whereas, as expected, the RM with progressive SIV infection manifested high amounts of both, with SIV RNA ~2 logs higher than DNA. All four ΔRh110/SIVmac239/smE660 vector–vaccinated RMs manifested detectable, albeit low-level, cell-associated SIV DNA in five or more tissues, with 28% (98 of 235) of samples positive overall (versus 0 of 114 samples in vaccinated, unchallenged controls; P < 0.0001 using Barnard’s exact test of binomial proportions). In contrast, cell-associated SIV RNA was detectable in only one RM, 2.6% of overall samples [9 of 345 versus 0 of 114 samples in controls, P = not significant (NS)]. To determine whether this detection of SIV DNA/RNA reflected replication-competent virus, we performed coculture analysis on a total of 1120 tissue specimens sampled from the four protected RMs (Fig. 6F). Only six of these specimens (0.5%), from two of the four RMs, were SIV+ upon coculture (five in RM #2; one in RM #4 versus 270 of 274 SIV+ cocultures in the unvaccinated control RM). We next combined 56 million to 100 million cells from the necropsy tissues of each of these four protected RMs and then adoptively transferred these cells into SIV-naïve recipient RMs and found no transfer of SIV infection in any of the four recipient RMs (Fig. 6G), observations consistent with most of the SIV DNA signals detected in tissues at necropsy representing replication-incompetent proviruses (34).

Fig. 6 Necropsy analysis of ΔRh110 68-1 RhCMV/SIV vector–protected RMs.

(A to C) Analysis of SIV Gag + Pol–specific (A) and SIV Vif–specific (B) CD4+ and CD8+ T cell response frequencies by flow cytometric ICS (using SIVmac239 peptide mixes; see Fig. 1) and tissue-associated SIV DNA and RNA by nested qPCR/RT-PCR (C) in tissues of four ΔRh110/SIVmac239/smE660 vector–protected RMs (RMs #7 and #8 from group 1; RMs #9 and #10 from group 2) taken to necropsy at 713 days (RM #7), 681 days (RM #8), 738 days (RM #9), and 745 days (RM #10) after infection. (D and E) Analysis of tissue-associated SIV DNA and RNA in tissues of two ΔRh110 68-1 RhCMV/SIVgag (SIVmac239 sequence insert) vector–vaccinated RMs that were taken to necropsy 531 and 763 days after vaccination without SIV challenge (negative controls) (D) and one SIVmac239-infected RM with progressive infection taken to necropsy 172 days after infection (positive control) (E). In (C) to (E), each data point indicates an independent tissue sample of the indicated tissue type and the dotted lines indicate the detection threshold. (F and G) Assessment of residual replication-competent SIV in cell suspensions obtained from the indicated tissue samples by in vitro coculture analysis (F) and by adoptive transfer of cells into four SIV-naïve RMs (G).

Last, we repeated the adoptive transfer experiment using cells collected at late time points from four different ΔRh110/SIVmac239/smE660 vaccine–protected, always aviremic RMs (RMs #1 and #2 from group 1; RMs #4 and #5 from group 2) that were previously shown (early after the onset of protection) to harbor replication-competent SIV by adoptive transfer. A total of 108 pooled cells from BM, lymph node, or blood collected at 60 to 102 weeks after infection from these RMs were administered to four SIV-naïve recipients, with no take of SIV infection detected in the recipient RMs (Fig. 7). Together, these results provide compelling evidence that replication-competent SIV declines over time in ΔRh110 68-1 RhCMV/SIV–vaccinated, long-term protected RMs such that, after ~2 years, lymphoid cells infected with replication-competent SIV are very rare or undetectable.

Fig. 7 Loss of transferable SIV in long-term ΔRh110 68-1 RhCMV/SIV vector–protected RMs.

Second assessment of replication-competent SIV by adoptive transfer of cells from four long-term ΔRh110/SIVmac239/smE660 vector–protected RMs (RMs #1 and #2 from group 1; RMs #5 and #6 from group 2) that were previously shown to harbor replication-competent SIV by the same assay.

Rechallenge of RhCMV/SIV vector–protected RMs

We next addressed the question of whether 68-1 RhCMV/SIV vector–vaccinated RMs retain the capacity to clear a second SIV challenge after control and progressive clearance of an initial challenge and, if so, whether RMs vaccinated with spread-competent versus spread-deficient vectors differ in this regard. To this end, we followed 8 protected RMs from our previously reported cohort of WT 68-1/SIVmac239 vector–vaccinated animals (group 4) (8) and 12 ΔRh110/SIVmac239/smE660 vaccine–protected RMs from this study (5 from group 1 and 7 from group 2) for at least 2 years after initial SIV infection. All RMs developed and then lost SIV Vif–specific CD4+ and CD8+ T cell responses in blood during this follow-up while retaining stable frequencies of (vaccine-elicited) SIV Gag– and SIV Pol–specific CD4+ and CD8+ T cell responses (fig. S9). We then initiated the same repeated, limiting-dose IVag SIVmac239 challenge protocol used in the first challenge. All RMs were infected by this challenge protocol, as indicated by the redevelopment of SIV Vif–specific CD4+ and CD8+ T cell responses (Fig. 8A). Four of five group 1 RMs, five of seven group 2 RMs, and seven of eight WT/ 68-1 SIVmac239 vector–vaccinated RMs were protected after this second SIV challenge, again showing either no viremia or only transient viremia (Fig. 8B). The twice-protected group 1 and 2 RMs included RMs #1, #2, #4, and #5, which were previously shown to lack transferable SIV before second challenge. BM and/or PBMC samples from these four RMs were collected after the (second) onset of SIV Vif–specific T cell responses and were inoculated into four additional SIV-naïve RMs. All of the four recipient RMs became SIV infected (Fig. 8C), indicating the presence of replication-competent SIV in these aviremic animals and thereby confirming a second, stringently controlled SIV infection. Overall, 16 of the 20 rechallenged RhCMV/SIV vector–vaccinated RMs were protected a second time. Although this degree of efficacy (80%) is higher than the overall efficacy of initial challenge (58%), this difference did not quite achieve statistical significance (P = 0.06). These data confirm that both spread-competent and spread-deficient (ΔRh110) RhCMV/SIV vectors are able to maintain efficacy for ~3 years after last vaccination and can provide protection against more than one SIV challenge.

Fig. 8 Resistance of ΔRh110 68-1 RhCMV/SIV vector–protected RMs to repeat SIV challenge.

(A and B) Outcome of repeat SIVmac239 challenge of long-term 68-1 RhCMV/SIV vector–protected RMs (n = 5, 7, and 8 for groups 1, 2, and 4, respectively) by longitudinal analysis of de novo SIV Vif–specific CD4+ and CD8+ T cell responses (A) and plasma viral load (B) with protected and nonprotected RMs defined as described in Fig. 3. (C) Third assessment of replication-competent SIV by adoptive transfer of cells from RMs #1, #2, #5, and #6 after effective rechallenge (reinduction of SIV Vif–specific T cell responses) with repeated aviremic protection.


HCMV infection is ubiquitous, especially in resource-poor settings, and although HCMV persists for life in infected individuals, the vast majority of these individuals will never develop CMV disease due to immune control of viral spread after primary infection and upon reactivation from latency (3537). The vast majority of such HCMV+ individuals would not be expected to develop symptomatic infection upon administration of an HCMV-based vaccine, even one with WT replication and spread capacity. However, in the setting of prophylactic vaccination of large populations, non–HCMV-infected individuals, potentially including immunocompromised subjects, would possibly be exposed to such a WT HCMV-based vaccine, through either direct administration or potentially spread from a vaccinated subject, and a subset of such individuals would be at risk of developing overt HCMV disease (38, 39). To mitigate this risk, we have sought to make a CMV-based vaccine safer by identifying an attenuation strategy that would substantially limit vector spread within and between hosts and thereby preclude disease in vaccinated individuals, spread to (and within) the fetus of pregnant subjects, and shedding in secretions (to prevent person-to-person spread). The strategy should preserve the ability of the vector to super-infect CMV+ individuals, productively infect sufficient numbers of cells to prime robust T cell responses, and persist long term to provide the antigenic stimulation needed for maintaining effector-memory differentiation. In a companion paper (20), we provide evidence that ΔRh110 RhCMV may strike such a balance in RMs, showing a ~1000-fold reduction in vector spread in vivo, no vector shedding in secretions, and no animal-to-animal spread with close contact or leukocyte transfusion. This vector still retains the ability to elicit insert-specific T cell responses that are comparable in magnitude, phenotype, function, epitope targeting, and durability as Rh110-intact RhCMV vectors. Furthermore, by insertion of the SIV Ags into the Rh110 locus, we also eliminate the possibility of reversion to WT by homologous recombination with the endogenous virus present in CMV-infected hosts.

Here, we performed a large vaccination and challenge trial of ΔRh110 RhCMV/SIV vectors to extend the immunogenicity analysis to a larger cohort of RMs vaccinated with these attenuated vectors and to determine protection from highly pathogenic SIVmac239 challenge. We also expanded our analysis of ΔRh110 RhCMV/SIV vectors to include determination of the extent to which mismatch between the vector insert sequence and SIV challenge strain would affect SIV-infected cell recognition by vector-elicited T cells and vaccine efficacy, as such mismatch will be invariably present in any clinical application of this vaccine. These results confirm that ΔRh110/SIVmac239 vectors elicit insert-specific CD8+ T cell responses that are essentially indistinguishable in magnitude, phenotype, and durability from that of WT 68-1/SIVmac239 vectors. CD8+ T cell responses elicited by ΔRh110 RhCMV vectors expressing SIVmac239 versus SIVsmE660 sequence inserts in blood were reduced in magnitude by 30 to 40% when tested on mismatched sequence peptides, but these responses were equivalent in their ability to recognize SIVmac239-infected and SIVsmE543-infected autologous CD4+ T cells. Thus, although epitope recognition by the unconventionally (MHC-E and MHC-II) restricted (9, 10) CD8+ T cells elicited by ΔRh110 RhCMV/SIV vectors can be modestly compromised by sequence divergence, the breadth of these responses is sufficiently great to ensure equivalent recognition of cells infected by divergent SIV strains. ΔRh110 RhCMV/SIV vector–elicited CD4+ T cells were largely unaffected by insert-target sequence mismatch (0 to 20% reduction) but were significantly reduced, albeit modestly, in both magnitude and effector-memory bias after the boost vaccination relative to Rh110-intact RhCMV/SIV vector–elicited responses. This modest reduction is consistent with the interpretation that RhCMV vector–elicited CD4+ T cell responses may be more sensitive to reduction in overall Ag availability than the corresponding CD8+ T cell responses.

Of primary importance, we found that the extent and pattern of protection afforded by ΔRh110/SIV vector vaccination, irrespective of sequence match versus mismatch between vector insert and challenge virus, was essentially identical to that of Rh110-intact vectors. In our previous analysis of WT 68-1/SIVmac239 vector vaccination, 56% of RMs were protected after initial challenge and 50% after 1 year (8). This degree of efficacy was not significantly different from 59% overall efficacy of the ΔRh110/SIV vectors observed in the present study, with all these protected RMs showing both initial and long-term protection. The percentage of protected RMs was actually higher for monkeys given the challenge-mismatched ΔRh110/SIVsmE660 vectors (64%) compared to RMs given the challenge-matched ΔRh110/SIVmac239 vectors (54%). Although this difference was not statistically significant, the finding that heterologous efficacy is as good as or better than homologous efficacy is an encouraging sign for clinical translation. Moreover, the characteristics of protection after ΔRh110/SIVmac239/smE660 vaccination were very similar to that of the WT 68-1/SIVmac239 vaccine. Animals acquired SIV, but except for rare viral blips, there was complete elimination of viremia, which is consistent with replication arrest. SIV was stringently controlled before systemic immune activation, as demonstrated by a lack of monocyte activation, before anti-Env Ab production, and in the absence of boosting of the vaccine-stimulated T cells. Together, these data indicate that vaccine-elicited immune protection can be achieved with substantially reduced levels of RhCMV vector spread.

We have previously demonstrated that WT 68-1/SIVmac239 vector–protected RMs show progressive loss of SIV infection over time, and this viral clearance process is particularly well documented in the current analysis of ΔRh110/SIVmac239/smE660 vector–vaccinated RMs. We demonstrate loss of detectable cell-associated SIV RNA/DNA detection in BM over the first 20 weeks after infection and decline in SIV Vif–specific CD4+ and CD8+ T cells in blood to below the threshold of detection over 1 to 2 years. RMs that were able to transmit infection to naïve recipients by transfer of cells obtained early after the onset of protection no longer transmitted infection 1 to 2 years later. Four ΔRh110/SIV vector–protected RMs [two each given vectors with matched (group 1) versus mismatched (group 2) SIV inserts] were taken to necropsy ~2 years after infection for extensive tissue analysis. Although PCR analysis demonstrated that these four RMs had more SIV DNA than two ΔRh110/SIVgag-vaccinated, but never SIV-challenged controls, SIV RNA and co-culturable virus was largely undetectable, and adoptive transfer of cells from tissues from these RMs did not transfer SIV infection. Together, these results suggest a vanishingly small amount of residual infectious SIV in these ΔRh110/SIVmac239/smE660 vector–protected RMs. The residual SIV DNA in these four necropsied RMs did, however, appear to be somewhat more than in our previous analysis of WT/SIVmac239 vector–protected RMs. This finding and the more frequent detection of low-frequency SIV Vif–specific T cells in tissues of the current RMs relative to the previously studied RMs (8) is consistent with the conclusion that viral clearance was not quite complete in these animals. This does not necessarily indicate a difference in the extent or kinetics of viral clearance between WT 68-1/SIVmac239 and ΔRh110/SIVmac239/smE660 vector–protected RMs because the WT 68-1/SIVmac239 vector–vaccinated RMs studied previously at necropsy were males infected by intrarectal challenge (as opposed to females being infected via IVag challenge), and four of these six previously studied animals were taken to necropsy after >1000 days after infection compared to ~700 days in the current study. Given the apparent dependence of SIV Vif–specific T cell responses on SIV Vif Ag production by SIV-infected cells, the observation that the slope of decline of SIV Vif–specific T cell responses in WT 68-1/SIVmac239 and ΔRh110/SIVmac239/smE660 vector–protected RMs was not significantly different (both CD4+ and CD8+) suggests that the rate of SIV infection clearance was broadly similar with both vaccines. This is in keeping with our previous hypothesis that SIV clearance in RhCMV/SIV vector–protected RMs predominantly results from arrest of infection before seeding a long-lived SIV reservoir and the subsequent decline of the less durable reservoir that is initially seeded (11). Although this implies that vaccine-elicited T cell responses are not actively clearing the viral reservoir, these responses very likely contribute to maintaining stringent replication control while the residual viral reservoir spontaneously declines, and if this is the case, WT 68-1/SIVmac239 and ΔRh110/SIVmac239/smE660 vector–elicited responses appear to be equivalent in this activity.

Last, we directly compared the outcome of a second round of SIV challenge in RMs that were previously protected by WT 68-1/SIVmac239 or ΔRh110/SIVmac239/smE660 vector vaccination and subsequently cleared the initial infection, as assessed by extinction of their SIV Vif–specific T cell responses over ~2 years. Remarkably, 80% of these rechallenged RMs, across all vaccine groups, were able to control this second challenge, with reinfection and aviremic control demonstrated in four protected RMs by conversion of the adoptive transfer assay of SIV infection from negative before the second challenge to positive after, in the absence of viremia. These data indicate that WT/SIVmac239 and ΔRh110/SIVmac239/smE660 vectors can maintain efficacy for up to ~3 years after last vaccination with the marked stability of the SIV-specific T cell responses elicited by these vectors suggesting that the potential for efficacy might extend for considerably longer periods, perhaps lifelong. However, note that SIV-specific T cell response magnitude in blood did not correlate with outcome in the first challenge for the ΔRh110/SIVmac239/smE660 vector–vaccinated RMs and that 20% of previously RhCMV/SIV vector–protected RMs were not protected after the second challenge, despite maintaining stable SIV-specific T cell responses. Thus, there is either an element of stochasticity to protection, or some unmeasured aspect of the innate or adaptive immune response to vaccination that is required for efficacy, and this parameter(s) can vary over time.

The control and clear protection against highly pathogenic SIVmac239 challenge afforded by RhCMV/SIV vectors is unique and offers an alternative mechanism for a clinically useful prophylactic HIV/AIDS vaccine either alone or in combination with an Ab-targeted vaccine designed to reduce HIV acquisition (12). The ability to substantially limit vector spread while preserving both the extent (%protected) and durability of efficacy is a critically important step in clinical translation of the CMV vector platform, as is the demonstration that RhCMV/SIV vector efficacy can tolerate the equivalent of an intra-clade sequence mismatch between vaccine insert and challenge virus strain without loss of efficacy. However, a major limitation of this study is that CMVs are species-specific viruses and a clinical vector for vaccination of humans against HIV will be based on HCMV, not the orthologous, but distinct, RhCMV. The pp71 protein is encoded by UL82 in HCMV, and although the RhCMV and HCMV pp71 proteins have similar function, UL82 deletion in HCMV results in a more pronounced growth defect in vitro than Rh110 deletion in RhCMV, suggesting that a ΔUL82 HCMV might be more attenuated in vivo in people than ΔRh110 RhCMV in monkeys (20). While this additional attenuation increases the margin of safety for clinical testing, it might also reduce immunogenicity, or more likely, increase the dose required to achieve full immunogenicity—issues that can only be resolved through human testing. Despite this potential concern, the results presented in this study strongly support the further development of pp71-deleted, 68-1–like HCMV/HIV vectors as prophylactic vaccines for HIV/AIDS.


Study design

The primary objective of this study was to determine whether attenuated (spread-deficient) ΔRh110 68-1 RhCMV/SIV vectors expressing homologous or heterologous SIV Ag inserts would, relative to unvaccinated controls, provide cycling female RMs stringent post-acquisition control of SIVmac239 infection, administered by repeated, limiting-dose IVag challenge. On the basis of previous experience with WT 68-1 RhCMV/SIV vectors (68), we randomly assigned n = 48 cycling female RMs assigned to one of three vaccine groups as follows: n = 14 ΔRh110/SIVmac239 (group 1), n = 14 ΔRh110/SIVsmE660 (group 2), and n = 20 unvaccinated (group 3). This group size was anticipated to allow us to resolve 20% protection at 90% power, pooling the vaccine groups. Although only the RMs with take of infection (animals with SIVvif T cell response induction and either cell-associated SIV in tissue or plasma viremia post-challenge) were considered for evaluation of protection (group 1, n = 13; group 2, n = 14; group 3, n = 17), our criteria for stringent SIV control (aviremic infection) were met in 16 of 27 vaccinated RMs (59%), allowing us to proceed to our secondary objectives of determining the extent of viral clearance over time in these protected RMs, and the ability of previously protected RMs to control a second infection. At the end of an ~2-year observation period, during which time SIVvif responses in blood in all protected RMs decayed to below the level of detection and all virologic assays reverted to (or remained) negative, the 16 protected RMs were arbitrarily assigned to either necropsy for comprehensive tissue analysis of residual SIV (n = 4; two each from groups 1 and 2) or to repeat SIV challenge (n = 12; 5 from group 1 and 7 from group 2). The latter analysis was also performed on long-term protected RMs vaccinated with WT 68-1 RhCMV/SIV vaccine from our previous report (8). All the described RM experiments were performed once, and all results from these experiments are included in the presented data (no data were excluded as outliers). All plasma and cell-associated viral load assays were assayed by blinded analysts; however, due to logistical constraints, other staff were not blinded to treatment assignments. Primary data are reported in data file S1.

Statistical analysis

We compared the fraction of protected RMs between treatment groups and challenges using Barnard’s exact test of binomial proportions. To compare time-to-event data, we used the Mantel-Haenszel log-rank test. To examine SIV dynamics in vaccinated and protected RMs, we fit linear models of T cell responses with time and treatment group as independent variables. Slope analyses of SIV Vif–specific T cell responses are described below. For all other comparisons, we used nonparametric Wilcoxon rank-sum tests for both paired and unpaired comparisons and Kruskal-Wallis for comparisons of more than two groups. Neutralizing Ab titers were log10-transformed and normalized to baseline before computation of the area under curve (AUC) and compared using Wilcoxon rank-sum tests. For log transformations when zeros were present, a small positive constant smaller than any nonzero value was added to all values before log transformation. For comparisons of AUC for percent responses, data were also baseline-subtracted before AUC calculation. For all analyses of SIV dynamics including those described below for SIV Vif–, SIV Gag–, and SIV Pol–specific T cell responses, we also fit confirmatory models to account for variation among individual RMs using linear mixed models, which confirmed our analysis in each case. All statistical analyses were conducted in R version 3.2.2 using the following R package versions: lmtest 0.9.34, zoo 1.8.1, survival 2.42.3, Exact 1.7, and lme 1.1.17. All P values are based on two-sided tests and unadjusted except where noted. Adjusted P values were computed using the Holm procedure for family-wise error rate control. Boxplots in Fig. 1 and fig. S5 show jittered points and a box from first to third quartiles [interquartile range (IQR)] and a line at the median with whiskers extending to the farthest data point within 1.5*IQR above and below the box, respectively.

For analyses of SIV-specific T cell responses, we log-transformed the responses before fitting to account for variance over time. For Vif-specific responses, we used Wald tests to compare models with and without specific time/group interaction terms to determine whether SIV clearance rate differed by vaccine. Models were fit using all data points in range after defining start and stop time points for each analysis according to the following predetermined procedure: We determined the start point as the time when the relevant mean response over all RMs reached its first peak before declining. The end point was the first time point after the start point at which the mean response was below the threshold for “return to baseline,” which we determined by taking the mean plus three SDs of all response values in the plateau phase (beyond 96 days after infection). For SIV Gag– and SIV Pol–specific CD4+ and CD8+ T cell responses, we used Wald tests to compare individual slopes to 0.


Materials and Methods

Fig. S1. ΔRh110 RhCMV/SIV vector design and supertope amino acid sequences.

Fig. S2. Comparison of the amino acid sequence of SIVmac239 versus SIVsmE660/smE543 vaccine inserts.

Fig. S3. Analysis of SIV Env–specific Ab responses after vaccination and after acquisition of SIV infection.

Fig. S4. Acquisition of SIV infection by groups 1, 2, and 3 RMs with repeated, limiting-dose IVag SIVmac239 challenge.

Fig. S5. Immune correlates analysis.

Fig. S6. Analysis of SIV Gag– and SIV Pol–specific CD4+ and CD8+ T cell responses after SIV infection.

Fig. S7. Analysis of circulating monocyte activation after SIV infection.

Fig. S8. Analysis of plasma viral load and SIV Vif–, SIV Gag–, and SIV Pol–specific CD4+ and CD8+ T cell responses in vaccine-protected, necropsied RMs from groups 1 and 2.

Fig. S9. Analysis of SIV Gag– and SIV Pol–specific CD4+ and CD8+ T cell responses in vaccine-protected and subsequently rechallenged RMs from groups 1, 2, and 4.

Data file S1. Primary data.

References (4046)


Acknowledgments: We thank A. Sylwester, A. Okoye, C. Kahl, S. Hagen, R. Lum, Y. Fukazawa, E. McDonald, L. Silipino, N. Whizin, K. Randall, A. Selseth, Z. McWatters, I. Cardle, E. Cangemi, and L. Boshears for technical or administrative assistance; B. Keele for providing SIVmac239 challenge virus; D. Montefiori for nAb assays; and A. Townsend for figure preparation. Funding: This work was supported by the Bill & Melinda Gates Foundation–supported Collaboration for AIDS Vaccine Discovery (OPP1033121 to L.J.P.); the National Institute of Allergy and Infectious Diseases (NIAID) (P01 AI094417 and R37 AI054292 to L.J.P.; R01 AI059457 to K.F.); the NIH Office of the Director (P51 OD011092); and the National Cancer Institute (contract HHSN261200800001E to J.D.L.). Author contributions: S.G.H. planned and performed animal experiments and immunologic assays, assisted by A.B.V., C.M.H., E.A., J.C.F., D. Morrow, R.M.G., and J.Y.B. J.B.S. and B.J.B. planned and performed infected cell recognition assays, assisted by J.S.R. J.D.L. planned and supervised SIV quantification by PCR/RT-PCR, assisted by K.O., R.S., B.B., W.J.B., and M.H. K.F., D. Malouli, J.W., and E.E.M. designed, constructed, and quality-tested RhCMV/SIV vectors, assisted by J.W. A.W.L., and M.K.A. managed the animal care and procedures. J.S. and P.T.E. performed all statistical analyses. L.J.P. conceived the RhCMV vector strategy, supervised all experiments, analyzed and interpreted data, and wrote the paper, assisted by S.G.H., J.D.L., K.F., P.T.E., and J.B.S. Competing interests: OHSU and L.J.P., E.E.M., S.G.H., and K.F. have a substantial financial interest in Vir Biotechnology Inc., a company that may have a commercial interest in the results of this research and technology. L.J.P., S.G.H., and K.F. are also consultants to Vir and coinventors of patent PCT/US2011/036657 “Recombinant RhCMV and HCMV vectors and uses thereof” licensed to Vir. J.B.S. has received compensation for consulting for Vir. The potential individual and institutional conflicts of interest have been reviewed and managed by OHSU. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. The computer code used to perform statistical analysis is available at RhCMV/SIV vectors can be obtained through a material transfer agreement.

Stay Connected to Science Translational Medicine

Navigate This Article