Cytomegalovirus Vaccine-induced Unconventional T cell Priming and Control of SIV Replication is Conserved Between Primate Species

Summary

Strain 68–1 Rhesus Cytomegalovirus expressing simian immunodeficiency virus (SIV) antigens (RhCMV/SIV) primes MHC-E-restricted CD8+ T-cells that control SIV replication in 50–60% of vaccinated rhesus macaques. Whether this unconventional SIV-specific immunity and protection is unique to rhesus macaques or RhCMV or intrinsic to CMV remains unknown. Here, using cynomolgus CMV vectors expressing SIV antigens (CyCMV/SIV) and Mauritian cynomolgus macaques, we demonstrate that induction of MHC-E-restricted CD8+ T-cells requires matching CMV to its host species. RhCMV does not elicit MHC-E-restricted CD8+ T-cells in cynomolgus macaques. However, species matched 68–1-like CyCMV/SIV-vaccinated cynomolgus macaques mounted MHC-E-restricted CD8+ T-cells, with half of vaccinees stringently controlling SIV post-challenge. Protected animals manifested a vaccine-induced IL-15 transcriptomic signature associated with efficacy in rhesus macaques. These findings demonstrate that the ability of species-matched CMV vectors to elicit MHC-E-restricted CD8+ T-cells required for anti-SIV efficacy is conserved in nonhuman primates and support development of HCMV/HIV for a prophylactic HIV vaccine.

eTOC Blurb:

Malouli et al. show that the unusual MHC-E-restricted CD8+ T cells required for protection against SIV observed in rhesus macaques vaccinated with rhesus CMV/SIV vectors is a fundamental property of primate CMVs. A protection-associated IL-15 signature is also conserved, highlighting the promise of this vector for prophylactic HIV vaccine development.

Graphical Abstract

Introduction

With 37 million people, the most ever in history, currently living with HIV, the development of a prophylactic HIV vaccine remains a top global health priority. Unfortunately, finding a vaccine for HIV has proven elusive, as recently highlighted by the inability of canarypox- and adenovirus serotype 26-based vaccine modalities to prevent HIV infection in the HVTN 702 and 705 clinical trials (Gray et al., 2021) (Johnson & Johnson, 2021). The lack of a prophylactic HIV vaccine is due in part to the unprecedented immune evasion adaptations and enormous variability of this virus, leading to difficulty in both identifying and effectively targeting viral vulnerabilities. Novel candidate vaccine approaches are therefore urgently needed to slow global HIV incidence.

A pre-clinical HIV vaccine approach based on strain 68–1 of rhesus cytomegalovirus (RhCMV) expressing Simian Immunodeficiency Virus (SIV) antigens (RhCMV/SIV) elicits effector memory (EM) T cell responses that are able to control pathogenic SIV replication in 50–60% of vaccinated rhesus macaques (RM) (Barrenäs et al., 2021; Hansen et al., 2009; Hansen et al., 2011; Hansen et al., 2013a; Hansen et al., 2019; Malouli et al., 2021). Although RhCMV/SIV does not protect against acquisition of infection, RhCMV/SIV-induced control of post-acquisition viral replication is so stringent that viral replication is effectively arrested, and, over time, SIV nucleic acid and replication competent virus are progressively cleared until RhCMV/SIV vector-protected RM are indistinguishable from SIV-naïve RM (Hansen et al., 2013a). This “control and clear” pattern of protection has not been seen with any other T cell-based vaccine reported to date, and accordingly, successful translation of these pre-clinical results into the clinic is an area of intense research that would have a profound and lasting impact on slowing the HIV epidemic.

In contrast to wildtype RhCMV, strain 68–1 RhCMV/SIV vaccines do not induce the expected repertoire of SIV-specific CD8+ T cell responses restricted by MHC-Ia molecules, but instead drive the emergence of a qualitatively distinct CD8+ T cell response that is non-classically restricted by either MHC-II or MHC-E (Hansen et al., 2013b; Hansen et al., 2016). This unconventional T cell response type is due to specific genetic changes in strain 68–1, including deletion or functional inactivation of the pentameric receptor complex components Rh157.5/Rh157.4 and the viral CXC chemokine-like Rh158-Rh161 gene products (orthologues of HCMV UL128/UL130 and UL146/UL147 genes, respectively). Differential repair of either the Rh157.5/Rh157.4 or Rh158-Rh161 gene products in strain 68–1 results in RhCMV/SIV vectors that elicit SIV-specific CD8+ T cells restricted by either MHC-Ia alone or MHC-Ia and MHC-II, but not MHC-E, and these repaired vectors do not protect against SIV (Malouli et al., 2021). Further, deletion of the MHC-E binding VL9 leader peptide encoded by Rh67 (orthologue of the HCMV UL40 gene) results in a 68–1 RhCMV/SIV vector that elicits only MHC-II-restricted CD8+ T cells and a vaccine composed of these vectors also fails to protect against SIV (Verweij et al., 2021). In contrast, inhibition of antigen production by endothelial cells via microRNA-126-p3-mediated restriction in RhCMV/SIV vectors yielded a purely MHC-E-restricted CD8+ T cell response that protected vaccinated RM against SIV similarly to 68–1 RhCMV (Hansen et al., 2022). These results cumulatively demonstrate that MHC-E-restricted CD8+ T cells are an essential component of RhCMV/SIV-mediated control of SIV replication.

Based on these data, it is highly likely that establishment of a human CMV (HCMV) vector-based HIV vaccine able to recapitulate the “control and clear” phenomenon observed in SIV-challenged strain 68–1 RhCMV/SIV-vaccinated RM will require an HCMV vector with the ability to prime MHC-E-restricted CD8+ T cells in humans. It remains unknown if the MHC-E-restricted CD8+ T cell response and the protection from SIV replication induced by RhCMV 68–1 in RM is a species-specific phenomenon due to the unique immunogenetics of RM, or due to intrinsic immunomodulatory properties of CMV itself that are conserved across primate CMV isolates. Defining this unknown parameter is critical to the successful translation of RhCMV into the clinic. Indian-origin RM exhibit remarkably complex MHC-Ia and MHC-II genetics (de Groot et al., 2004; Otting et al., 2005), and this MHC complexity extends to Mamu-E, the MHC-E allele in rhesus macaques. Indeed, in contrast to humans where only two major HLA-E molecules exist in the population, we have identified 30 distinct Mamu-E alleles at the population level, and found that individual RM can express up to four separate Mamu-E alleles simultaneously (Wu et al., 2017). This increased MHC-E allelic diversity in RM compared to humans raises the possibility that the broad targeting by Mamu-E-restricted CD8+ T cells in strain 68–1 RhCMV/SIV vaccinated RM may be due in part to the higher number of Mamu-E alleles expressed. In contrast to RM, Mauritian cynomolgus macaques (MCM), descendants from a population bottleneck event, exhibit simplified MHC-Ia and MHC-II genetics that are more reflective of human HLA immunogenetics (Budde et al., 2010; Burwitz et al., 2009; O’Connor et al., 2007). In line with the reduced MHC complexity in MCM, we have identified only four Mafa-E alleles at the population level, and found that the vast majority of individual MCM express only one or two Mafa-E alleles (Wu et al., 2017). Therefore, the immunogenetics of MHC-E in MCM mirrors that of humans, making MCM the ideal model to determine if the MHC-E-restricted CD8+ T cells induced by RhCMV/SIV in RM is due to the complex immunogenetics of RM or due to conserved, potent immunomodulatory properties of CMV regardless of the primate species.

Here, to parse out the role of host immunogenetics, particularly of the MHC-E and MHC-II loci, and strain-specific CMV mechanisms in the MHC-E-restricted CD8+ T cell response and SIV protection induced by strain 68–1 RhCMV/SIV in RM, we translated the RhCMV vaccine vector approach to MCM. To this end, we cloned cynomolgus macaque CMV (CyCMV) and deleted the CyCMV orthologs of Rh157.5/Rh157.4 and Rh158-Rh161 to generate a “68–1 like” double deleted (dd) CyCMV vaccine vector. Vaccination of MCM with ddCyCMV/SIV vectors recapitulated key immunological features of RhCMV strain 68–1 vaccinated RM, namely an absence of MHC-Ia-restricted CD8+ T cell epitopes that were instead replaced by MHC-II- and MHC-E-restricted EM CD8+ T cells, including responses against the previously described supertopes targeted by every RhCMV/SIV-vaccinated RM (Hansen et al., 2013b; Hansen et al., 2016). Upon repeated low dose intrarectal SIVmac239 challenges half of the ddCyCMV/SIV-vaccinated MCM manifested the early arrest of SIV replication that defines RhCMV-mediated control of SIV. Finally, vaccine-phase whole blood transcriptomic analysis revealed that an IL-15 response signature previously identified in RhCMV/SIV-protected RM (Barrenäs et al., 2021) was present in CyCMV/SIV-protected MCM following vaccination. Cumulatively, these studies recapitulate critical immunological and virological features of RhCMV/SIV vaccination and SIV challenge of RM in MCM, a second nonhuman primate species that possesses immunogenetics remarkably similar to humans, and augur well for translation of pre-clinical CMV vector technology to a HCMV vaccine for prevention of HIV.

Results

Generation and characterization of CyCMV vaccine vectors

To determine if CMV vectors can elicit SIV-specific MHC-E-restricted CD8+ T cells in MCM, we first tested the ability of RhCMV vectors to elicit these responses in MCM. However, we previously demonstrated that strain 68–1 RhCMV is unable to productively infect MCM, even in the absence of CD8+ T and NK cells (Burwitz et al., 2016). RhCMV strain 68–1.2, genetically repaired to express Rh61/Rh60 (homologues of the anti-apoptosis HCMV UL36 gene) and Rh157.5/Rh157.4 (homologues of the epithelial cell tropism HCMV UL128 and UL130 genes) which are absent in 68–1, efficiently infected MCM as evidenced by the induction of SIV transgene-specific T cells and virus shedding. However, RhCMV strain 68–1.2 does not elicit MHC-II- or MHC-E-restricted CD8+ T cells in RM (Hansen et al., 2013b; Hansen et al., 2016), and fails to protect RM from SIV replication (Malouli et al., 2021), and thus is not an appropriate vaccine vector for these cross-species studies in MCM. We subsequently demonstrated that a Gag-expressing RhCMV strain 68–1 vector with only Rh61/Rh60 repaired is able to cross the species barrier to infect MCM and elicit SIV Gag-specific T cells (Burwitz et al., 2016). To determine if the minimally altered RhCMV 68–1 Rh61/Rh60 repaired vector would induce MHC-E-restricted CD8+ T cell responses in RM and MCM we first vaccinated two RM with a RhCMV 68–1 Rh61/Rh60 repaired vector expressing Gag to characterize the Gag-specific CD8+ T cell response. We assessed the specificity of Gag-specific CD8+ T cells by stimulating peripheral blood mononuclear cells (PBMC) or leukapheresis product with 15-mer peptides that span the Gag open reading frame (ORF) in an intracellular cytokine staining assay. We then subsequently defined the MHC restriction of all peptides capable of inducing a CD8+ T cell response by performing the above assay in the presence of each of the following: the pan MHC-I-blocking antibody W6/32, the leader sequence-derived MHC-E-blocking VL9 peptide, the MHC-II-blocking G46.6 antibody, or isotype control reagents. Vaccination of RMs with the RhCMV Rh61/Rh60-repaired 68–1 vector induced unconventional MHC-II- and MHC-E-restricted CD8+ T cell responses commensurate with those induced by RhCMV strain 68–1, including targeting of the MHC-II-restricted (Gag_211–222AP11 (Gag 53) and Gag_290–301PF12 (Gag 73)) and MHC-E-restricted (Gag_276–284RL9 (Gag 69) and Gag_482–490EK9 (Gag 120)) supertopes (Figure S1A and Figure 1A, 1B). In contrast, the RhCMV 68–1 Rh61/Rh60 repaired vector failed to induce MHC-E-restricted CD8+ T cells in MCM, and instead elicited predominantly MHC-Ia-restricted CD8+ T cells (Figure 1C). These results indicated that cross-species CMV vaccination is unable to engender SIV-protective, MHC-E-restricted CD8+ T cells even between closely related hosts, and that translation of the RhCMV vector concept to MCM might depend on the use of CyCMV-based vectors.

Figure 1: Generation and characterization of CyCMV vaccine vectors.

(A) Table of CMV vectors used in this study with indicated gene expression. See also Figure S1. (B-G) CD8+ T cell responses to SIV Gag mapped by stimulating PBMC or leukapheresis product with overlapping 15-mer peptides that span Gag ORF. MHC restriction was determined by performing the above assay with an isotype control, W6/32 antibody (anti-MHC-I, red), MHC-II antibody (MHC-II blocking, blue), or VL9 peptide (MHC-E blocking, green) reagents. Response not fully (>50%) blocked by any of these reagents is marked indeterminate (lavender). Boxes depict Gag 15mers recognized by CD8+ T cells in a given animal with colors indicating MHC-restriction of each response as indicated in the legend in (B) RMs or (C) MCM vaccinated with Rh61/R60 repaired RhCMV 68–1/Gag, (D) MCM vaccinated with FL CyCMV 68–1/Gag, (E) MCM vaccinated with ΔCy157.5/Cy157.4 CyCMV/Gag, (F) MCM vaccinated with double deleted (dd)CyCMV/Gag, and (G) RM vaccinated with ddCyCMV/Gag. The MHC-II- (blue) and MHC-E- (green) restricted supertopes are indicated by colored arrows with the amino acid positions in Gag and 15-mer number listed. See also Figures S1, S2, and S6.

To test this possibility, we isolated CyCMV from the urine of a healthy MCM and subsequently captured a full-length, low-passage primary isolate as a bacterial artificial chromosome (BAC) (Figure S2). We first sought to demonstrate that the BAC-cloned, full length CyCMV (FL CyCMV) retained the ability to replicate in MCM. To this end, we introduced SIV Gag into FL CyCMV and vaccinated an MCM. Mirroring the response seen in RM vaccinated with FL RhCMV (Malouli et al., 2021), we detected only conventionally MHC-Ia-restricted CD8+ T cell responses (Figure 1D), confirming that the BAC-cloned CyCMV could infect MCM and that, as expected, additional immunomodulatory gene deletions are required for CyCMV vectors to induce the unconventional immune responses observed in 68–1 RhCMV/SIV-vaccinated RMs. As we previously identified the absence of Rh157.5/Rh157.4 as being critical for the induction of MHC-E-restricted CD8+ T cells in RM (Hansen et al., 2016), we deleted both Cy157.5 and Cy157.4 from the Gag-expressing FL CyCMV BAC to make a ΔCy157.5/ΔCy157.4 CyCMV/Gag vaccine vector (Figure S1B and Figure 1A). Vaccination of two MCM with ΔCy157.5/ΔCy157.4 CyCMV/Gag revealed the priming revealed the presence of MHC-II- and MHC-Ia-restricted CD8+ T cells, but no supertope responses, and a complete lack of MHC-E-restricted CD8+ T cells in ΔCy157.5/ΔCy157.4 CyCMV/Gag-vaccinated MCM (Figure 1E). These results are in line with recent studies in RM, which demonstrated that removal of Rh157.5/Rh157.4 from RhCMV/Gag vectors alone is insufficient to prime MHC-E-restricted CD8+ T cells, and that additional viral factors must also be removed to fully recapitulate the RhCMV strain 68–1 phenotype (Malouli et al., 2021). Indeed, upon further inspection of the genetic changes strain 68–1 and FL RhCMV, we recently identified that the UL146/UL147 HCMV homologue genes (Rh158–161), viral proteins with alpha chemokine function (Oxford et al., 2008; Lüttichau, 2010), were also deleted or functionally inactivated by the original genomic inversion that removed Rh157.5 and Rh157.4 in RhCMV 68–1. Detailed analysis of these genes in RhCMV revealed that expression of any one gene product within the Rh158-Rh161 coding region is sufficient to inhibit the priming of MHC-E-restricted CD8+ T cells, and that removal of both the viral tropism gene products encoded by Rh157.5/Rh157.4 and the CXC chemokine-like proteins encoded by Rh158-Rh161 is required for priming of unconventionally MHC-E-restricted CD8+ T cells in RM (Malouli et al., 2021). Thus, we next deleted Cy158-Cy161 from ΔCy157.5/ΔCy157.4 CyCMV/Gag to generate ΔCy157.5/ΔCy157.4, ΔCy158-Cy161 CyCMV/Gag, a “68–1 like” double deleted (dd) CyCMV vector (Figure S1B and Figure 1A). Vaccination of MCM with ddCyCMV/Gag resulted in both MHC-II- and MHC-E-restricted CD8+ T cells, with epitope targeting similar to 68–1 RhCMV/Gag-vaccinated RM (Figure 1F). Remarkably, we observed targeting of the same MHC-II- and MHC-E-restricted Gag supertopes targeted in every strain 68–1 RhCMV/Gag-vaccinated RM, indicating that ddCyCMV functions analogously in MCM as strain 68–1 RhCMV does in RM. Finally, to determine if ddCyCMV could cross the species barrier and elicit unconventionally MHC-restricted CD8+ T cells in RM, we vaccinated RM with ddCyCMV/Gag. While ddCyCMV elicited MHC-II- and MHC-E-restricted CD8+ T cells when matched to its MCM host, it failed to do so in RM, and instead elicited only MHC-Ia-restricted CD8+ T cells, confirming the requirement for CMV vector matching to its host species in order to elicit unconventionally MHC-restricted CD8+ T cells (Figure 1G).

Immunogenicity and anti-SIV efficacy of ddCyCMV vectors

Having established that the ddCyCMV vaccine vector recapitulated the critical ability of RhCMV 68–1 to prime MHC-E-restricted CD8+ T cells, we generated two additional ddCyCMV vectors encoding either SIV 5’Pol, or a fusion SIV Rev-Tat-Nef protein (RTN), which together with ddCyCMV/Gag, we collectively termed ddCyCMV/SIV (Figure S1C). We vaccinated eight MCM subcutaneously with 1×10⁷ PFU each of all three vectors, and boosted at 14 weeks post-prime (Figure 2A and Table S1). We monitored SIV transgene-specific CD4+ and CD8+ T cells in blood and bronchoalveolar lavage (BAL) fluid and found these responses were detectable throughout the vaccine induction phase (Fig. 2B and 2C). The ddCyCMV/SIV vaccine induced little to no antibodies against the SIV transgenes as anti-Gag antibodies did not consistently increase following vaccination (Fig. S3A). Consistent with RhCMV-induced responses in RM, these SIV transgene-specific CD8+ T cells exhibited a predominantly effector memory (EM) phenotype (Figure S4A). Necropsy analysis of the two ddCyCMV/Gag-vaccinated MCM shown in Figure 1F revealed that Gag-specific T cells were present in the highest frequencies in extralymphoid tissues like liver and lung, in keeping with their EM phenotype (Figure S4B). Thus, these responses largely mirrored the durability and memory phenotype of those detected in 68–1 RhCMV/SIV-vaccinated RM. We did note, however, that the SIV transgene-specific T cells elicited in ddCyCMV/SIV-vaccinated MCM were of overall lower magnitude compared to strain 68–1 RhCMV/SIV-vaccinated RM (Figure S4C).

Figure 2: Immunogenicity of ddCyCMV/SIV vectors.

(A) Schematic overview of vaccine study. (B,C) Longitudinal monitoring of CD8+ (left) and CD4+ (right) T cell responses against Gag, Pol, RTN in (B) PBMC and (C) BAL from ddCyCMV/SIV-vaccinated MCM. Graphs show the mean of eight vaccinated MCM and error bars represent standard error of the mean (SEM). Positive responses shown are values calculated following subtraction of response to no stimulation for each animal at each time point. (D) PBMC from ddCyCMV/SIV vaccinated MCM (blue, n=8) or SIVmac239-infected MCM (red, n=4) were combined with Gag or RTN ORF or the MHC-Ia-restricted minimal-optimal peptides indicated and the CD8+ T cell response assessed by cytokine secretion in ICS. Shapes track with a specific animal across each column. (E) PBMC from the eight ddCyCMV/SIV vaccinated MCM were stimulated with the indicated supertope minimal optimal epitope pulsed onto MCM BLCL (positive control), K562 (negative control, MHC-Ia and MHC-E negative cells), K562 cells transfected with the indicated MHC-E Mafa allele, or (F) RM3 (negative control, MHC-II negative cells) or RM3 cells transfected with the indicated MHC-II Mafa allomorph. Shapes track with a specific animal across each column. (G) PBMC or BAL from the eight ddCyCMV/SIV vaccinated MCM were combined with autologous CD4+ T cells that were infected with SIVmac239 in vitro or left uninfected. Shapes track with a specific animal across each column. See also Figures S3, S4, and S6, and Table S1.

Next, we examined the fine specificity and MHC restriction of these ddCyCMV/SIV-induced T cell responses. Gag-specific CD8+ T cell responses in these eight MCM were similar in breadth and MHC-II- or MHC-E-restriction to RhCMV/SIV vaccinated RM, with MHC-II- and MHC-E-restricted supertopes targeted in every MCM (Figure S4D). Seven of the eight ddCyCMV/SIV-vaccinated MCM possessed the MHC alleles that present the previously described canonically MHC-Ia-restricted CD8+ T cell epitopes Nef_103–111 RM9, Nef_196–203 HW8, and Gag_386–384 GW9 (Budde et al., 2012; Burwitz et al., 2009; Mohns et al., 2015) (Table S1). Despite mounting CD8+ T cell responses against the Gag and Nef ORF, these ddCyCMV/SIV-vaccinated MCM did not target any of these three canonically MHC-Ia-restricted CD8+ T cell epitopes present in these SIV proteins, in contrast to chronically SIV-infected MCM expressing the restricting MHC allele (Figure 2D). We next confirmed the unconventional MHC-E restriction of the Gag_276–284RL9 (Gag 69) and Gag_482–490EK9 (Gag 120) supertopes in MCM by combining PBMC from CyCMV/SIV vaccinated MCM with peptide-pulsed, MHC-deficient cell lines expressing either no MHC-I or only Mafa-E:02:01 molecules (Figure 2E). CD8+ T cells from ddCyCMV/SIV vaccinated MCM recognized these MHC-E restricted supertopes when presented in the context of Mafa-E:02:01 on MHC-null cells transfected to express this allele or on the surface of immortalized MCM B cells (BLCL), confirming MHC-E restriction. We performed similar MHC restriction assays for the Gag_211–222AP11 (Gag 53) and Gag_290–301PF12 (Gag 73) MHC-II restricted supertopes and found that CD8+ T cells from ddCyCMV/SIV vaccinated MCM recognized these epitopes only in the context of Mafa MHC-II allomorphs expressed on the surface of the MHC-II-deficient RM3 cell line or on MCM BLCL (Figure 2F). Finally, ddCyCMV/SIV-induced CD4+ and CD8+ T cells from both PBMC and BAL recognized autologous SIV-infected CD4+ T cells in vitro (Figure 2G). Thus, ddCyCMV/SIV vaccinated MCM recapitulated the immunological phenotype observed in RhCMV/SIV vaccinated RM, including response durability, memory phenotype, absence of conventionally MHC-Ia-restricted CD8+ T cells, presence of MHC-II- and MHC-E-restricted CD8+ T cells, and SIV-recognition, suggesting these animals might exhibit a similar capacity for protection against SIVmac239 replication.

Fifty nine weeks after first vaccination we challenged the eight ddCyCMV/SIV vaccinated MCM and eight unvaccinated controls with an intrarectal escalating low dose of SIVmac239 (Figure 2A). Because RhCMV/SIV-protected RM may not manifest detectable viremia after acquisition of infection (Hansen et al., 2019), we measured SIV infection in ddCyCMV/SIV-vaccinated MCM by monitoring the blood of all animals once to twice weekly for both onset of viremia and the development of de novo responses to SIV Vif and Env, SIV gene products that were not expressed by the vaccine inserts. In contrast to previous studies of strain 68–1 RhCMV/SIV vaccinated RM, ddCyCMV/SIV vaccinated MCM acquired SIV significantly faster than the unvaccinated controls (Figure 3A). Five of eight of the unvaccinated control MCM manifested overt viremia in the first three challenges, with the remaining three animals becoming overtly viremic after the sixth, eighth, and 13th challenge (Figure 3A and 3B, and Table S2). Four of eight of the ddCyCMV/SIV-vaccinated MCM manifested overt viremia within the first two challenges, but strikingly, the other four manifested development of SIV Vif and Env T cell responses in the complete absence of measurable viremia (Figure 3B, 3C, and 3D), in keeping with the phenotype of 68–1 RhCMV/SIV-protected RM. There was no statistical difference in the viral loads between the unvaccinated controls and the four viremic ddCyCMV/SIV vaccinated MCM. All unvaccinated MCM and the four ddCyCMV/SIV vaccinated MCM with detectable SIV viremia seroconverted following infection, while the four ddCyCMV/SIV-vaccinated MCM that controlled post SIV replication did not (Fig. S3B). Although only a minority of bone marrow, spleen, and lymph node biopsies from these four aviremic ddCyCMV/SIV-vaccinated MCM showed the presence of viral nucleic acids (Figure 3E and 3F), cells from these tissues from all four of the animals transferred SIV infection when infused into four SIV-naïve MCM (Figure 3G and 3H), confirming infection with replication-competent SIVmac239 in these animals. These results demonstrate that ddCyCMV/SIV vaccination leads to a similar efficacy (~50%) of protection from SIV replication in MCM compared to 68–1 RhCMV/SIV vaccination in RMs.

Figure 3: Anti-SIV efficacy of ddCyCMV/SIV vectors.

(A) Kaplan-Meier curve displaying number of challenges needed for SIV acquisition in the eight ddCMV/SIV vaccinated MCM (black) and eight unvaccinated controls (blue). (B) Measurements of SIV RNA per mL plasma in ddCMV/SIV vaccinated and control MCM post-infection. The last challenge prior to a positive viral load and/or detection of a T cell response against SIV proteins not in the vaccine was determined as the challenge leading to successful infection. Red lines depict protected, vaccinated, MCM; black lines depict unprotected, vaccinated MCM; and blue lines depict unvaccinated, control MCM. Dashed line = limit of quantification at 15 copies/mL. (C) SIV Env- and (D) SIV Vif-specific CD8+ T cell responses post-infection. Responses are following subtraction of response to no stimulation for each animal at each time point. Colors match those in panel B, but for clarity only display the mean response magnitude for the unprotected vaccinated MCM and unvaccinated control MCM. (E) Cell associated viral DNA and (F) viral RNA loads from bone marrow, peripheral lymph node (axillary or inguinal), mesenteric lymph node, and spleen biopsies. Values above the limit of detection are filled while values below the limit of detection are not. Non-circle symbols indicate the specified protected MCM. DNA Minimum Threshold: 0.28 copies DNA/10⁶cells. RNA Minimum Threshold: 0.51 copies RNA/10⁶cells. (G) Anatomical source and number of cells from protected, ddCyCMV/SIV vaccinated MCM that were used for adoptive transfer studies to SIV naïve MCM. (H) Plasma viral loads in four MCM that received the cells listed in panel G. Viral loads depict time post-infection (post-transfer) and each donor/recipient pair is listed on the right. Dashed line = limit of quantification at 50 copies/mL. See also Figure S6 and Table S2.

Transcriptomic signature of ddCyCMV/SIV protection

We recently defined a RhCMV/SIV transcriptomic signature of protection in RM (Barrenäs et al., 2021). To determine if a similar signature was present in ddCyCMV/SIV-protected MCM, we conducted functional genomics analyses using RNA sequencing (RNAseq) of whole blood from the eight ddCyCMV/SIV-vaccinated MCM immediately prior to vaccination and one, three, seven, and 14 days after prime and boost vaccination (Figure S5A). We initially assessed the major sources of variation among all expressed genes with a principal components analysis (PCA). Although animal sex is the dominant source of variation, animals also separated by outcome status within sex along PC1, with protected animals converging in the center of PC1 (Figure 4A). We then identified vaccine responsive genes stratified by outcome status via differential expression testing to define differentially expressed (DE) genes from baseline. We found 2,407 DE genes in at least one time-point post-vaccination, with the majority (98.8%: 2,378/2,407) being observed in protected MCM only (Figure 4B). Protected animals exhibited robust induction (up-regulation) of T cell receptor signaling and innate antiviral immunity pathways, including interferon signaling, pattern recognition receptor signaling, and death receptor signaling, consistent with the previously reported RhCMV/SIV vaccine protection signature (Barrenäs et al., 2021). Additionally, protected animals showed reduced expression (down-regulation) of natural killer cell signaling and B-cell signaling pathways relative to nonprotected animals (Figure S5B).

Figure 4: Transcriptomic signature of ddCyCMV/SIV protection.

(A) Principal component analysis bi-plot showing transcriptomic variation in the dataset. See also Figure S4. (B) Total number of differentially expressed genes at each vaccine phase time point (D1-D14 = days 1, 3, 7, and 14 post ddCyCMV/SIV prime, and BD0-BD14 = days 0, 1, 3, 7, and 14 post ddCyCMV/SIV boost) relative to baseline in SIV protected and unprotected MCM. (C) Time series heatmap of the 68–1 RhCMV/SIV protection signature genes enriched in ddCyCMV/SIV protected MCM. Plotted are log2 fold change values of the 122 leading edge genes across the vaccination time series in MCM (left) and RM (right), with RM whole blood one day post exogenous IL-15 treatment in the middle. RM data are from Barrenas et al. (D) Pre-vaccination baseline normalized expression Z-score boxplots of the 84 up-regulated leading-edge genes identified in panel C distinguishes SIV protection outcome (top plot). Pre-vaccination baseline normalized expression Z-score boxplots of the 38 down-regulated leading edge genes identified in panel C distinguishes outcome (bottom plot). See also Figure S5.

We previously identified a core set of genes that were associated with IL-15 signaling and RhCMV/SIV vaccine-induced protection in RM (RM IL-15 x DDE) (Barrenäs et al., 2021). To determine the possible conservation of this gene signature in ddCyCMV/SIV-protected MCM, we compared the MCM gene signature to the RhCMV/SIV vaccine protection signature to distinguish outcome in MCM. We first performed a differences of differential expression (DDE) analysis to directly compare the response to vaccination in protected animals to the same response in nonprotected animals. We then tested if this response differential was enriched for the RhCMV/SIV protection signature and stratified into up- and down-regulated gene sets using gene set enrichment analysis (GSEA). We observed enrichment across the entire time series in MCM, with the strongest responses at one day following prime vaccination (Prime Day 1), just before the boost vaccination (Boost Day 0), and one day after the boost vaccination (Boost Day 1) (Figure S5C). The leading-edge genes (i.e., top contributing genes) identified from this enrichment analysis revealed strong conservation between MCM and RM and highlight the importance of IL-15 response pathways, including death receptor signaling, immune cell signaling programs, pattern recognition receptor signaling, and NK cell response in CMV/SIV vaccine-mediated protection (Figure 4C and Figure S5C). Lastly, we interrogated the ability of the RhCMV/SIV protection signature to distinguish outcome from baseline gene expression profiles, as observed previously (Barrenäs et al., 2021). Using the RhCMV/SIV protection signature leading-edge genes enriched in MCM, we observed stratification by outcome status at the pre-vaccination baseline time point using z-score transformed normalized expression values of IL-15 response genes (Figure 4D). Collectively, these genes serve to distinguish the protected and nonprotected outcome groups based on baseline expression levels.

Discussion

The recent failure of the HVTN 702 and 705 clinical trials underscores the urgent need for novel modalities in HIV vaccine development (Gray et al., 2021) (Johnson & Johnson, 2021). While the 68–1 RhCMV/SIV vaccine approach has demonstrated a unique “control and clear” vaccine-mediated effect, it remained unknown if this MHC-E-restricted CD8+ T cell-based protection mechanism was exclusive to RhCMV and/or RM only due to the complex immunogenetics of RM or other unknown factors. The results presented here, whereby key immunological features of the RhCMV vaccine approach were successfully transposed to MCM, a separate nonhuman primate with MHC genetics more similar to humans, demonstrate that the ability to elicit MHC-E-restricted CD8+ T cells and protect against SIV replication post-acquisition is likely a conserved intrinsic ability of CMV itself based on its co-evolved mastery of host immunomodulation. Indeed, we failed to observe induction of MHC-E-restricted CD8+ T cells in both RM and MCM unless the appropriately modified CMV matched to its host species was used. Further, in ddCyCMV/Gag-vaccinated MCM, we observed elicitation of CD8+ T cells universally targeting the identical MHC-E-restricted Gag supertopes previously identified in RhCMV/Gag-vaccinated RM, likely due to the conserved peptide-binding motif and function of MHC-E across RM, MCM, and humans (Wu et al., 2017). Although HLA-E-restricted CD8+ T cells are rarely present in humans, HCMV infection can prime HLA-E-restricted CD8+ T cells targeting UL40 leader peptide mimics mismatched in sequence to the host’s HLA leader peptides (Hoare et al., 2006; Sullivan et al., 2017). Additionally, we recently demonstrated the ability to in vitro prime human CD8+ T cells targeting an HLA-E-restricted HIV Gag epitope with sequence homology to the Gag_276–284RL9 (Gag 69) supertope targeted in both RhCMV-vaccinated RM and CyCMV-vaccinated MCM (Yang et al., 2021). Cumulatively, these results support clinical translation of HCMV-based vaccines capable of eliciting HIV-specific HLA-E-restricted CD8+ T cells.

Here, beginning with a urine sample from an MCM to generate CyCMV and ending with an SIV challenge study of MCM vaccinated with a vector based on this virus, we demonstrate that central features of the CMV vaccine vector system are conserved between viral isolates and species. Indeed, ddCyCMV/SIV-vaccinated MCM successfully recapitulated the priming of MHC-E-restricted CD8+ T cells, IL-15 vaccine signature, and immediate SIV arrest phenomenon that defines RhCMV/SIV-mediated protection in RM. However, we did observe noteworthy differences between the two species, including overall magnitude of the ddCyCMV/SIV-specific T cell response and rate of SIV acquisition. The SIV transgene-specific CD8+ and CD4+ T cell responses observed in our ddCyCMV/SIV-vaccinated MCM are of overall lower magnitude than seen in strain 68–1 RhCMV/SIV-vaccinated RM (Hansen et al., 2009; Hansen et al., 2011). We, and others, have previously noted lower frequency T cell responses in MCM, including responses against CyCMV itself during reactivation (Budde et al., 2012; Burwitz et al., 2009; Burwitz et al., 2016). Although the reason is unclear, it may depend upon differences in immune cell populations between the two species (Sibley et al., 2021). Regardless of the difference, neither the magnitude of the SIV-specific CD4+ T cell response nor the magnitude of the SIV-specific CD8+ T cell response in 68–1 RhCMV/SIV vaccinated RM predicted vaccine efficacy in RM (Barrenäs et al., 2021). Thus, we surmise that the lower frequency T cell responses present in ddCyCMV/SIV vaccinated MCM are reflective of the species as a whole, is not specific to the ddCyCMV vector, and does not impact ability to contain SIV replication following infection. The faster acquisition of SIV observed in ddCyCMV/SIV-vaccinated MCM is at odds with previous observations in RhCMV/SIV-vaccinated RM, as 68–1 RhCMV/SIV-vaccinated RM do not acquire SIV at a rate significantly different from unvaccinated controls. Specifically, in three separate studies no statistically significant difference in acquisition was found between 28 RhCMV/SIV vaccinated RM versus 20 unvaccinated RM (Hansen et al., 2019), 16 RhCMV/SIV vaccinated RM versus 14 control RhCMV vaccinated RM versus 12 unvaccinated RM (Hansen et al., 2013a), and 12 RhCMV/SIV vaccinated RM versus 12 RhCMV/SIV + Ad5/SIV vaccinated RM versus 9 DNA/SIV + Ad5/SIV RM versus 28 unvaccinated RM (Hansen et al., 2011). The reason for the divergence in rate of acquisition between RM and MCM is unclear, and may relate to MCM-specific differences in immune subsets as noted above, but it should be noted that our current study was not designed to be adequately powered to address acquisition and that subsequent studies are required to definitely address this topic.

In sum, we demonstrated here that the defining features of the CMV vaccine vector platform, namely induction of unconventionally MHC-E-restricted CD8+ T cells and immediate intercept and inhibition of SIV replication, could be translated to a second nonhuman primate species. These results help in extending the RhCMV vaccine vector into the clinic. Indeed, our finding that the UL146/UL147 genes are inhibitors of MHC-E restricted CD8+ T cell induction resulted in the removal of these genes from prototype HCMV vaccine vectors now entering phase I clinical trials (Vir Biotechnology). Given the complexity of CMV, other unanticipated roadblocks may likely be encountered before full successful clinical translation of RhCMV-induced protection against replication is achieved, but our findings here establish the feasibility of such an endeavor.

STAR Methods

Resource Availability

Lead contact:

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jonah Sacha (sacha@ohsu.edu).

Materials availability:

There are restrictions on the CyCMV vaccine vector BACs described in this manuscript. However, these materials are available to researchers at academic institutions via material transfer agreements with OHSU.

Data and code availability:

• Transcriptomics data sets are deposited at the Gene Expression Omnibus under accession number GSE205080.

• The R markdown code applied to these analyses can be accessed at https://github.com/galelab/Malouli_Signature_of_ddCyCMV-SIV_protection.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and subject details

Cynomolgus and Rhesus Macaques.

A total of 26 Mauritian cynomolgus macaques (MCM, Macaca fascicularis from the island of Mauritus, 9 males, 17 females, age range: 4–16 years old) were purchased from accredited vendors and shipped to the Oregon National Primate Research Center (ONPRC) for use in this study. MCM were MHC typed as we have previously described in detail (Burwitz et al., 2017). A total of four rhesus macaques (RM, Macaca mulatta, 3 males, 1 female, age range: 4–7 years old) were obtained from the ONPRC colony for use in this study. All study MCM and RM were free of cercopithicine herpesvirus 1, D-type simian retrovirus, simian T-lymphotrophic virus type 1, and Mycobacterium tuberculosis. All study MCM and RM were housed at the ONPRC in Animal Biosafety level (ABSL)-2 and ABSL-2+ for the vaccine and SIV challenge phase, respectively. MCM in the vaccine group arrived at ONPRC approximately two years prior to SIV challenge, while, MCM in the control group arrived at ONPRC, underwent a 30-day quarantine period, then proceeded immediately to low dose SIV challenge alongside the vaccinated MCM. Rooms had autonomously controlled temperature, humidity, and lighting. Study MCM and RM were both pair-cage housed whenever possible, and all MCM and RM had visual, auditory and olfactory contact with other animals. MCM and RM received commercially prepared primate chow twice daily and received supplemental fresh fruit or vegetables daily. Fresh, potable water was provided via automatic water systems. Physical exams including body weight and complete blood counts were performed at scheduled protocol time points. MCM and RM were sedated with ketamine HCl with the addition of Dexmeditomidine and Atipamezole as a reversal agent for some procedures, including subcutaneous vaccine administration, venipuncture, and SIV challenge. Apheresis was performed as described in detail previously (Wu et al., 2020a). At scheduled endpoints, MCM and RM were euthanized with sodium pentobarbital overdose (>50 mg/kg) and exsanguinated via the distal aorta, and tissue collection at necropsy was performed by a certified veterinary pathologist. MCM and RM care and all experimental protocols and procedures were approved by the ONPRC Institutional Animal Care and Use Committee (IACUC). ONPRC is a Category I facility and the Laboratory Animal Care and Use Program at the ONPRC is fully accredited by the American Association for Accreditation of Laboratory Animal Care (AAALAC) and has an approved Assurance (#A3304–01) for the care and use of animals on file with the NIH Office for Protection from Research Risks. The IACUC adheres to national guidelines established in the Animal Welfare Act (7 U.S.C. Sections 2131–2159) and the Guide for the Care and Use of Laboratory Animals (8th Edition).

Method details

Capture of CyCMV as a bacterial artificial chromosome (BAC) and construction of CyCMV/SIV Vectors.

CyCMV strain 31908 was isolated from an MCM urine sample and minimally cultured on primary embryonic rhesus fibroblasts (RFs) generated at the ONPRC as described previously (Burwitz et al., 2016). Viral DNA was isolated using a modified Hirt prep protocol to determine the full genome sequence of the isolate (GenBank accession number KX689265). To capture the viral genome as a bacterial artificial chromosome (BAC), plasmid pHA1 (Messerle et al., 1997) (Schumacher et al., 2000) which contains a BAC cassette encompassing a xanthine-guanine phosphoribosyl transferase (gpt) selectable marker flanked by loxP sites and PacI restriction sites, was used. As RhCMV 68–1 derived from a BAC introduced into the genome between ORFs Rh181 (US1) and Rh182 (US2) retained genome stability and infectivity after prolonged tissue culture (Chang and Barry, 2003), the BAC cassette was introduced into the same location in CyCMV 31908. Through alignment of the 68–1 and 31908 genomes, the equivalent insertion site was identified and a capture cassette was designed containing 1500bp homology to the CyCMV genome upstream and downstream of the target location divided by a PacI restriction site and flanked by SacI restriction sites. To avoid oversizing the genome after BAC capture during in vitro selection, the downstream homology arm was designed to target the genome just upstream of Cy186 (US8), effectively deleting Cy182 (US2) through Cy185 (US6) in the BAC captured genome. PacI digestion was used to linearize the capture cassette and clone the PacI digested pHA1 BAC cassette between the two homology arms. SacI digestion of the final capture plasmid generated a linear DNA fragment with terminal homology arms targeting the viral genome.

To capture the viral genome as a BAC, 2×10⁶ RFs were seeded in a T175 culture flask and harvested the next day by trypsination. The fibroblasts were pelleted, washed twice with PBS, nucleofected with 2ug of linearized BAC capture plasmid using an Amaxa Nucleofector (Lonza), and seeded into a T75 culture flask. On the following day, nonadherent cells were removed and the fibroblasts washed with PBS. The RFs were then superinfected with CyCMV 31908 at an MOI of 5. To enrich for recombinants between the viral genome and the BAC capture cassette containing a gpt selectable marker, 100 µM mycophenolic acid was added to inhibit inosine-5′monophosphate dehydrogenase (IMPDH) and 25 μM xanthine was added to allow the synthesis guanine nucleotides in the presence of gpt. Hence, only cells infected by CyCMV containing a BAC cassette including the gpt selection marker should be able to synthesize guanine nucleotides and replicate DNA. Both the mycophenolic acid and the xanthine were replenished every other day by addition to the culture medium and the selection was continued until the entire cellular monolayer was infected by CyCMV. The supernatant was then harvested and two more rounds of selection were performed on RF in T175 culture flasks using a low MOI for more stringent selection. At the end of the third round the virus grew readily in the presence of the selection agents, so circular viral DNA was harvested from CyCMV 31908 BAC infected RF and this DNA was used to electroporate electrocompetent DH10β E.coli cells, which were plated on chloramphenicol containing agar plates as the BAC cassette contained a chloramphenicol acetyltransferase (CAT) gene conferring chloramphenicol resistance. All resulting colonies were grown in LB medium with appropriate selection and the isolated DNA was analyzed by restriction digest with XmaI to identify clones carrying a complete CyCMV 31908 BAC genome. Two positive clones were analyzed by next generation sequencing to determine the presence of an intact genome and to exclude off target mutations acquired during prolonged tissue culture during in vitro selection. Finally, the clone showing the least nucleotide changes compared to the published genome sequence was used to create a repaired full-length BAC of 31908 (see Figure S2).

All repairs needed to restore the captured CyCMV 31908 genome to its predicted original sequences were performed using en passant homologous recombination (Tischer et al., 2010) to avoid leaving residual DNA sequences in the final constructs. The CyCMV BAC was transferred into E.coli strain GS1783 since these bacteria can express the lambda (λ) phage derived red recombination genes after heat shock induction, while also expressing the I-SceI homing endonuclease upon arabinose induction (Tischer et al., 2010). Recombination inserts were generated by synthesizing recombination primers introducing a homology region upstream and downstream of an I-SceI homing endonuclease recognition site followed by an aminoglycoside 3phosphotransferase gene conferring kanamycin resistance (KanR), which was used to select for successful recombinants. In a second step, I-SceI expression was induced resulting in a DNA double strand break and the subsequent removal of the selection marker by homologous recombination of the introduced homology regions. To confirm BAC integrity and to ensure repair of the targeted regions, XmaI restriction digest analysis of purified BAC DNA was performed and the engineered genomic loci was assessed by Sanger sequencing analysis. After all repairs had been performed, the final CyCMV 31908 BAC was assessed by next generation sequencing to confirm the absence of off target mutations and purified BAC DNA was used to electroporate RFs (250V, 950μF) to reconstitute CyCMV. Viral stocks were generated by purifying infectious virus from clarified culture medium through a sorbitol cushion (20%D-sorbitol, 50 mM Tris [pH 7.4], 1 mMMgCl2), and the virus was stored at −80°C until use. The GenBank accession number for the CyCMV 31908 BAC is ON109535.

To generate viral vectors carrying immunologically traceable markers for in vivo experiments and to generate vector sets for challenge experiments SIVmac239 derived transgenes (Gag, 5’ Pol, RTN) were used as described by us previously in published studies using RhCMV based vaccine vectors (Sturgill et al., 2016; Hansen et al., 2016; Hansen et al., 2013a; Hansen et al., 2013b; Hansen et al., 2009) (Hansen et al., 2017). To insert these genes into the CyCMV BAC, a homology region flanking an I-SceI site and a KanR selection marker were inserted into the selected SIV gene. The transgenes were amplified by PCR and the product recombined into the Cy13.1 (RL13) locus in the vector backbone replacing the endogenous ORF as previous data has indicated that the presence of an unaltered RL13 homologue results in genome instability (Stanton et al., 2010; Taher et al., 2020). The KanR cassette was subsequently removed scarlessly as described above and all recombinants were characterized by XmaI restriction digests and Sanger sequencing across the modified genomic locus. Lastly all vectors were fully analyzed by next generation sequencing to exclude off-target mutations and to confirm full accordance of the generated with the predicted full genome sequence. All BAC derived recombinants were reconstituted and expanded in RFs, viral titers were determined using TCID₅₀ assays, and viral stocks were stored at −80°C.

CyCMV vaccinations and SIVmac239 challenge.

CyCMV vectors were delivered subcutaneously at 1×10⁷ PFU per vector. All ddCyCMV/SIV vaccinated and control MCM were challenged with repeated intrarectal low dose SIVmac239 as previously described with an escalating dose as shown in Table S2. The SIVmac239 stock 20082 was produced by the Virology Services Unit of the Wisconsin National Primate Research Center using SIVmac239 hemi-genome plasmids obtained from the NIH AIDS Research and Reference Reagent Program. These plasmids were transfected into 293T cells, and the supernatant was propagated on mitogen-activated PBMC isolated from SIV naïve RM. The titer of this stock was 90,000 50% tissue culture infective doses (TCID₅₀)/mL. The desired challenge dose (Table S2) was aliquoted into 1 mL of sterile PBS and administered rectally as previously described (Sheppard et al., 2014). Once an MCM developed a positive plasma viral load and/or T cell response against Vif or Env (antigens not in the vaccine) it was considered infected and no longer challenged.

Cell Processing.

PBMCs were isolated from EDTA-treated whole blood and apheresis product purified using Ficoll-Paque (GE Healthcare) density centrifugation as previously described (Wu et al., 2020b; Wu et al., 2020a; Wu et al., 2017). Cells were resuspended in RPMI 1640 containing 10% FBS (R10; Hyclone Laboratories, Logan, UT). Single cell suspensions from lymph node and spleen were prepared by dicing in a large Petri dish and then pressing over a 70 μM cell strainer. Bone marrow was shaken for five minutes in a 50mL conical with 1xPBS + 2mM EDTA. Tubes were spun at 830g x 5min at room temperature and then the supernatant was removed. Cells were resuspended in Hanks Balanced Saline Solution (HBSS) and then layered over Ficoll-Paque. The buffy coat was removed and counted for downstream assays or freezing. Lysing matrix tubes were prepared by collecting small (~100ug) lymph nodes, similarly sized pinch biopsies from spleen, or 100μl aliquots of bone marrow directly into the tubes. These were then flash frozen at −180°C.

Immunological Assays

CD4+ and CD8+ T cells specific for SIV were detected as previously described (Sacha et al., 2008; Hansen et al., 2019; Hansen et al., 2016; Hansen et al., 2013b; Fujita et al., 2014; Sturgill et al., 2016) using flow-cytometric intracellular cytokine analysis. Sequential 15-mer peptides that overlap by 11 amino acids comprising the SIVmac239 Gag, Pol, Rev/Tat/Nef, Vif, and Env proteins were combined with PBMC or leukapheresis product and co-stimulatory antibodies anti-CD28 and anti-CD49d. Cells were combined with antigen and incubated for 1hr at 37°C before the addition of Brefeldin A and an additional eight-hour incubation. Co-stimulation without antigen served as a negative control. Cells were then stained with fluorochrome conjugated antibodies listed below and data was acquired on an LSRII (BD Biosciences) and analyzed using FlowJo software (BD Biosciences). CD4+ and CD8+ T cell responses were assessed for CD69+, TNFα+ and/or CD69+, IFNγ+ frequencies. For longitudinal analysis of SIV-specific responses shown in Figures 2B, 2C, 3C, and 3D, the values are background subtracted. For background subtraction, a no stimulation control was included for each animal at each time point and this value was subtracted from the values obtained with the experimental stimulation of interest (Gag, Pol, TRN, Env, or Vif peptide pools). MHC blocking ICS were performed as above except peptide stimulation was preceded by the addition of one of each the following specific inhibitors: 1) the pan anti-MHC-I mAb W6/32 (10mg/mL), 2) the MHC-II-blocking mAb G46.6 (10mg/mL), or 3) the MHC-E blocking VL9 peptide (VMAPRTLLL; 20μM) for one hour before peptide was added, as previously described (Hansen et al., 2016). To be considered MHC-E restricted by blocking, the individual peptide response must have been blocked by both anti-MHC-I clone W6/32 and MHC-E-binding peptide VL9, and not blocked by the anti-MHC-II clone G46.6. To be considered MHC-II-restricted by blocking, the individual peptide response must have been blocked by the anti-MHC-II clone G46.6 and not blocked by either the anti-MHC-I clone W6/32 or the MHC-E-binding peptide VL9. To be considered MHC-Ia-restricted by blocking, the individual peptide response must have been blocked by the anti-MHC-I clone W6/32 and not blocked by either the MHC-E-binding peptide VL9 or the anti-MHC-II clone G46.6. Individual peptide responses failing to meet any of these criteria is marked as indeterminant. For MHC restriction assays, antigen presenting cells consisting of either MCM BLCL, K562, K562 transfected with pCEP4 plasmid encoding Mafa-E*02:01, RM3, or RM3 dual transfected with pCEP4 plasmids expressing Mafa-DRB*w501 + Mafa-DRA*0103, DRB*w2102 + Mafa-DRA*0103, DRB*1002 + DRA*0102, or DRB5*0301 + DRA*0102 were pulsed with peptide as cocultured with PBMC in ICS assays as we previously described (Wu et al., 2017; Hansen et al., 2016; Hansen et al., 2013b). Longitudinal analysis of responses were memory corrected as previously described using CD28 and CD95 markers to define memory populations (Hansen et al., 2013a). For target cell recognition assays, CD4 T cells were enriched from PBMC or apheresis product using paramagnetic beads that positively select CD4+ T cells. These cells were activated by overnight incubation with staphylococcus enterotoxin B (SEB), CD3, CD28, and CD49d in R15 supplemented with 200 units/ml IL-2. Activated CD4 T cells were washed three times with HBSS approximately 24 hours later. Cells were then infected with SIVmac239 using magnetofection protocols as previously described (Sacha and Watkins, 2010; Vojnov et al., 2010). After three days, infected and uninfected CD4+ T cells were counted and resuspended. SIV+ versus SIV- T cells were then incubated with PBMC at an effector:target ratio of 40:1 in conditions matching previously described peptide-stimulations and then intracellular cytokine stained as described above. Representative flow plots are presented in Figure S6.

Enzyme-linked immunosorbent assay (ELISA)

Assays were performed essentially as described previously (Malherbe et al., 2020). To assay plasma antibody responses, half-well ELISA plates (Costar) were coated with recombinant SIVmac251 Gag (NIH HIV Reagent Program) or recombinant monomeric SIVmac239 gp120 at a concentration of 1 μg/ml (Gag) or 2 μg/ml (Env) in 0.2M H₂CO₃ buffer pH 9.4 at 4°C overnight. Plates were washed in binding buffer (PBS pH 7.4 + 0.1% Triton X-100) and blocked with 150 μl PBS containing 5% dried milk and 1% goat serum for 1 hr at room temperature. Blocking buffer was discarded and eight 3-fold serial dilutions of plasma or control antibodies were added to unwashed cells in 50 μl binding buffer. Purified polyclonal anti-SIV immunoglobulin was used as a positive control and naïve macaque immunoglobulin was used as the negative control. After 1 hr at RT, plates were washed 3x and then probed for 1 hr with 50 μl 1:5000 dilution of a horseradish peroxidase-conjugated goat anti-human IgG Fc fragment-specific polyclonal antibody (Jackson ImmunoResearch). Plates were then washed 5x, and bound Ab was visualized by the addition of 50 μl tetramethylbenzidine (Southern Biotech) for 10 min before stopping the reaction with 50 μl 1N H₂SO₄. Optical density was immediately quantified on a SoftMax Pro 5 microplate reader (Molecular Devices) at 450 nm.

Antibodies

To define the memory vs. naïve subsets, the following antibodies were used: SP34–2 (CD3; Alexa700, BD Biosciences), L200 (CD4; AmCyan, BD Bioscienes), SK-1 (CD8; PerCP-cy-5.5, BD Biosciences), MAB11 (TNFα; FITC, eBioscience), B27 (IFNγ; APC, BD Pharmingen), FN50 (CD69; PE, BD Biosciences), CD28.2 (CD28; PE-TexasRed, BD Biosciences), 150503 (CCR7; R&D Systems), and SA (PacBlue, BD Biosciences). CCR7 was biotinylated using the EZ-link Malemide-PEO Solid Phase Biotinylation kit from Fisher Scientific, as per manufacturer instructions. For cell recognition and T cell response assays, the following antibodies were used: anti-CD3 (clone: SP34–2, Pacific Blue; BD Biosciences), anti-CD8 (clone: SK1, TruRed; BD Biosciences), anti-CD4 (clone: L200, PE-Cy7, BV395; BD Biosciences), anti-CD28 (clone: CD28.2, PE; BD Biosciences), anti-CD95 (clone: DX2, FITC; BD Biosciences), anti-CD69 (clone: FN50, ECD, Biolegend), anti-CCR7 (clone: 150503, Pacific Blue, R&D Systems), anti-IFNγ (clone: B27, FITC; BD Biosciences), anti-TNFα (clone: MAb11, Alexa 700; BD Biosciences), and LIVE/DEAD Fixable Yellow Dead Cell Stain (Life Technologies) was used to assess cell viability.

SIV DNA and RNA quantitation

For the SIV challenge studies in Figure 3, plasma SIVmac239 RNA levels were determined using a gag-targeted quantitative real-time/digital RT–PCR format assay, essentially as previously described, with 6 replicate reactions analyzed per extracted sample for assay thresholds of 15 SIV RNA copies ml⁻¹ (Hansen et al., 2019). For the adoptive transfer experiments, SIVmac239 RNA levels were performed by the ONPRC Molecular Virology Core. Detection of SIV nucleic acids was performed as previously described (Chang et al., 2021b; Chang et al., 2021a). This assay has a threshold of 50 copies/mL. The following primers were used in these assays: forward primer, sGAG21: 5′-GTCTGCGTCAT(dP)TGGTGCATTC-3′; reverse primer, sGAG22: 5′-CACTAG(dK)TGTCTCTGCACTAT(dP)TGTTTTG-3′. Probe, psGAG23: 5′-FAM-CTTC(dP)TCAGT(dK)TGTTTCACTTTCTCTTCTGCG-BHQ1–3′. (BHQ1, black hole quencher-1; FAM, 6-carboxyfluorescein.) The gag sequence between the SIVmac239 challenge virus and the ddCyCMV/Gag vaccine vector is fully conserved allowing for the possibility of detection of ddCyCMV/Gag in plasma. However, previous exploration of this topic found that 68–1 RhCMV/Gag is not detected in plasma of vaccinated RM (Hansen et al., 2017), and in line with these results, this assay only detected SIV RNA in MCM plasma samples following SIVmac239 challenge.

Quantitative assessment of SIV DNA and RNA in cells and tissues was performed using gag-targeted, nested quantitative hybrid real-time/digital RT–PCR and PCR assays, as previously described (Hansen et al., 2019; Hansen et al., 2017; Hansen et al., 2013a). The nested primers utilized were: SIVnestF01: 5′-GATTTGGATTAGCAGAAAGCCTGTTG-3′, SIVnestR01: 5′-GTTGGTCTACTTGTTTTTGGCATAGTTTC-3′. The outer reverse primer (SIVnestR01) perfectly matches SIVmac239 gag, but contains seven mismatches in this codon-optimized portion of the vaccine gag insert sequence, essentially precluding detection of vaccine-derived gag with this assay (Hansen et al., 2017). SIV RNA or DNA copy numbers were normalized based on quantitation of a single copy rhesus genomic DNA sequence from the CCR5 locus from the same specimen, as described, to allow normalization of SIV RNA or DNA copy numbers per 10⁶ diploid genome cell equivalents. Ten replicate reactions were performed with aliquots of extracted DNA or RNA from each sample, with two additional spiked internal control reactions performed with each sample to assess potential reaction inhibition. Samples that did not yield any positive results across the replicate reactions were reported as a value of ‘less than’ the value that would apply for 1 positive reaction out of 10. Threshold sensitivities for individual specimens varied as a function of the number of cells or amount of tissue available and analyzed; for graphing consistency values are plotted with a common nominal sensitivity threshold.

Adoptive Transfer

Single cell suspensions were thawed in HBSS and counted. Tissues were then combined and filtered over a 70uM cell strainer. Cells were combined and counted in HBSS and then transfused into recipient animals via the saphenous vein.

RNAseq

Whole blood was collected from MCM into PAXgene RNA tubes (PreAnalytiX) according to the manufacturer’s instructions. RNA was isolated using RNAdvance Blood Kit (Beckman) following the manufacturer’s instructions. mRNAseq libraries were constructed using KAPA RNA HyperPrep Kit with RiboErase (HMR) Globin in conjunction with the KAPA mRNA Capture Kit (Roche Sequencing). Libraries were sequenced on an Illumina NextSeq500 sequencer using Illumina NextSeq 500/550 High Output v23 (150 cycles) following the manufacturer’s protocol for sample handling and loading. Raw sequencing reads were de-multiplexed with bcl2fastq. Residual adapters and low quality bases were then removed with Trim Galore, a wrapper for cutadapt, followed by globin read removal with bowtie2 (Langmead and Salzberg, 2012). Filtered reads were mapped to the RM genome (Mmul10, Ensembl v100, Warren et al. 2020) with STAR v2.7.5 (Dobin et al., 2013) and gene expression was quantified with htseq-count v0.12 (Anders et al., 2015). The Rhesus Macaque genome was used because of its superior annotation relative to the MCM genome. Mapping to the RM genome resulted in 1,707 more genes for analysis and in general, it is better to use a higher quality, slightly divergent genome in NHP studies (Gaska et al., 2019). Gene counts were imported into the R statistical software for subsequent analyses. We first removed lowly expressed genes, then normalized the filtered counts using TMM normalization (Robinson and Oshlack, 2010) followed by voom transformation (Law et al., 2014).

Differential expression analyses were performed with limma, testing for changes in expression at each time point post-vaccination relative to baseline (adj p value<0.05, absolute log₂ fold change > 1.5). We first identified differentially expressed genes in protected and nonprotected animals separately, then directly compared protected and nonprotected animals using a differences of differential expression (DDE) contrast for each post-vaccination time point. Differentially expressed genes were binned into co-expression modules using heatmap.2 and WGCNA (Langfelder and Horvath, 2008). Functional enrichment for each co-expression module was performed with Ingenuity Pathway Analysis (Krämer et al., 2014).

To identify gene sets that track with protection status, we ranked each gene by its DDE moderated t-statistic and used this pre-ranked gene list as input for gene set enrichment analysis (GSEA) (Subramanian et al., 2005). We ran GSEA in WebGestaltR (Liao et al., 2019) against MSigDB Hallmark Gene Sets (Liberzon et al., 2015) along with the previously defined RhCMV/SIV protection signatures, separated into up- and down-regulated gene sets (Barrenäs et al., 2021). A gene set was considered significantly enriched at FDR < 0.05. We further explored the ability of the RhCMV/SIV protection signature to distinguish animals by protection status at their pre-vaccination baseline time point. We extracted the RhCMV/SIV protection signature leading edge genes from the GSEA, generated a z-score heatmap of normalized expression, and performed a Wilcoxon rank sum test to compare z-scores between protected and nonprotected animals.

Quantification and Statistical analysis

All statistical analysis was performed using Graphpad Software Prism v9.3.1. A Log-rank (Mantel-Cox) test was performed on the acquisition curve shown in Figure 3A (n=8 ddCyCMV/SIV-vaccinated MCM versus n=8 unvaccinated control MCM). An area under the curve analysis was performed on the SIV transgene-specific T cell response in Figure S4C (n=9 strain 68–1 RhCMV/SIV-vaccinated RM versus n=8 ddCyCMV/SIV-vaccinated MCM).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Alexa Fluor 700 mouse anti-human CD3, clone SP34-2	BD Biosciences	Cat#557917; RRID:AB_396938
AmCyan mouse anti-human CD4, clone L200	BD Biosciences	Cat#561488; RRID:AB_10693557
PerCP-Cy5.5, mouse anti-human CD8α, clone SK1	BD Biosciences	Cat#341051; RRID:AB_400209
FITC, mouse anti-human TNFα, clone MAb11	eBioscience	Cat#11-7349-82; RRID:AB_465424
APC, mouse anti-human IFNγ, clone B27	BD Biosciences	Cat#554702; RRID:AB_398580
PE, mouse anti-human CD69, clone FN50	BD Biosciences	Cat#557050; RRID:AB_396566
PE-TexasRed, mouse anti-human CD28, clone CD28.2	BD Biosciences	Cat#562296; RRID:AB_11151918
mouse anti-human CCR7, clone 150503	R&D systems	Cat#MAB197-100
Pacific Blue, mouse anti-human CD3, clone SP34-2	BD Biosciences	Cat#558124; RRID:AB_397044
PE-Cy7 mouse anti-human CD4, clone L200	BD Biosciences	Cat#560644; RRID:AB_1727474
PE-Cy7 mouse anti-human CD4, clone L200	BD Biosciences	Cat#564107; RRID:AB_2738596
PE, mouse anti-human CD28, clone CD28.2	BD Biosciences	Cat#561793; RRID:AB_396072
FITC, mouse anti-human CD95, clone DX2	BD Biosciences	Cat#555673; RRID:AB_2100496
ECD, mouse anti-human CD69, clone FN50	Biolegend	Cat#310942; RRID:AB_2564277
Pacific Blue, mouse anti-human CCR7, clone 150503	BD Biosciences	Cat#562555; RRID:AB_2728119
FITC, mouse anti-human IFNγ, clone B27	BD Biosciences	Cat#554700; RRID:AB_395517
Alexa Fluor 700, mouse anti-human TNFα, clone MAb11	BD Biosciences	Cat#557996; RRID:AB_396978
mouse anti-human CD28, clone CD28.2	BD Biosciences	Cat#555726; RRID:AB_396069
mouse anti-human CD49d, clone L25	BD Biosciences	Cat#340976; RRID:AB_400198
mouse anti-human MHC-I, clone W6/32	Lab-generated from hybridoma	N/A
mouse anti-human HLA-DR, clone G46.6	BD Biosciences	Cat#556642; RRID:AB_396508
PerCP-eFluor710, mouse anti-human CD8a, clone SK1	Life Tech	Cat#46-0087-42; RRID:AB_1834411
BV510, mouse anti-human CD4, Clone L200	BD Biosciences	Cat#563094; RRID:AB_2738001
PE, mouse anti-human CD69, Clone FN50	eBioscience	Cat#12-0699-42; RRID:AB_10733526
FITC, mouse anti-human TNFa, Clone MAB11	BD Biosciences	Cat#552889m; RRID:AB_394518
Bacterial and Virus Strains
SIVmac239 (in vitro)	Lab-generated	N/A
SIVmac239 (in vivo)	Lab-generated	Stock 20082
E. coli DH10β	ThermoFisher Scientific	Cat#EC0113
Chemicals, Peptides, and Recombinant Proteins
Pacific Blue, Streptavidin	BD Biosciences	Cat#563259, RRID:AB_2869475
SIVmac239 Gag, Pol, Rev/Tat/Nef, Vif, Env peptides (15mer with 11 a.a. overlap)	NIH HIV Reagent Program	Gag=ARP-6204 Pol=ARP-6443 Rev=ARP-6448 Tat=ARP-6207 Nef=ARP-8762 Vif=ARP-6205 Env=ARP-6883
VL9 peptide (VMAPRTLLL)	Genscript USA incorporated	N/A
Recombinant SIVmac251 pr55 Gag protein	NIH AIDS Reagent Program	ARP-13384
Recombinant SIVmac239 gp120 Env protein	lab generated	N/A
Recombinant interleukin-2	Genscript USA incorporated	Cat# Z00368-1
Deposited Data
whole blood mRNA-seq data	GEO (Gene Expression Omnibus)	GSE205080
Experimental Models: Cell Lines
Primary embryonic rhesus fibroblasts	Lab-generated	N/A
K562 cell line	ATCC	CCL-243
RM3 cell line	Shared from the laboratory of Dr. David I. Watkins.	Described in publication: PMID 3116083
Experimental Models: Organisms/Strains
Mauritian-origin Macaca fascicularis	Accredited vendors for each MCM listed in Table S1	N/A
Macaca mulatta	Oregon National Primate Research Center	N/A
Oligonucleotides
sGAG21: 5′-GTCTGCGTCAT(dP)TGGTGCATTC-3′	Biosearch Technologies	N/A
sGAG22: 5′-CACTAG(dK)TGTCTCTGCACTAT(dP)TGTTTTG-3′	Biosearch Technologies	N/A
psGAG23: 5′-FAM-CTTC(dP)TCAGT(dK)TGTTTCACTTTCTCTTCTGCG-BHQ1-3′	Biosearch Technologies	N/A
SIVnestF01: 5′-GATTTGGATTAGCAGAAAGCCTGTTG-3′	Biosearch Technologies	N/A
SIVnestR01: 5′-GTTGGTCTACTTGTTTTTGGCATAGTTTC-3′	Biosearch Technologies	N/A
Recombinant DNA
CyCMV 31908 bacterial artificial chromosome	Lab-generated	GenBank accession #ON109535
pCEP4 mammlian expression vector	ThermoFisher	Cat# V04450
Mafa-E*02:01:02	Lab-generated	Described in publication: PMID 29150562 PMCID: PMC5736429 DOI: 10.4049/jimmunol.1700841
Mafa-DR Alleles	Lab-generated	Described in publication: PMID 17384942 PMCID: PMC2836927 DOI: 10.1007/s00251-007-0209-7
Software and Algorithms
FlowJo v10.8.1	BD Biosciences	N/A
Prism v9.3.1	GraphPad Software	N/A
FastQC	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/	PMC4243306
STAR v2.7.5	https://github.com/alexdobin/STAR	PMC3530905
R markdown code	https://github.com/galelab/Malouli_Signature_of_ddCyCMV-SIV_protection	Malouli_Signature_of_ddCyCMV-SIV_protection
Bowtie 2	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml	PMC3322381
HTSeq v0.12	Anders et al., Bioinformatics. 2015	PMC4287950

Highlights:

• Unconventional CMV vaccine immunogenicity is conserved in primate CMV

• MHC-E-restricted CD8+ T cells primed only when CMV is matched to parent species

• Cynomolgus CMV vaccine vectors induce protection against SIV replication in cynos

• IL-15 transcriptomic signature of SIV protection is present in CMV-vaccinated cynos