Engagement of monocytes, NK cells, and CD4+ Th1 cells by ALVAC-SIV vaccination results in a decreased risk of SIVmac251 vaginal acquisition

The recombinant Canarypox ALVAC-HIV/gp120/alum vaccine regimen was the first to significantly decrease the risk of HIV acquisition in humans, with equal effectiveness in both males and females. Similarly, an equivalent SIV-based ALVAC vaccine regimen decreased the risk of virus acquisition in Indian rhesus macaques of both sexes following intrarectal exposure to low doses of SIVmac251. Here, we demonstrate that the ALVAC-SIV/gp120/alum vaccine is also efficacious in female Chinese rhesus macaques following intravaginal exposure to low doses of SIVmac251 and we confirm that CD14+ classical monocytes are a strong correlate of decreased risk of virus acquisition. Furthermore, we demonstrate that the frequency of CD14+ cells and/or their gene expression correlates with blood Type 1 CD4+ T helper cells, α4β7+ plasmablasts, and vaginal cytocidal NKG2A+ cells. To better understand the correlate of protection, we contrasted the ALVAC-SIV vaccine with a NYVAC-based SIV/gp120 regimen that used the identical immunogen. We found that NYVAC-SIV induced higher immune activation via CD4+Ki67+CD38+ and CD4+Ki67+α4β7+ T cells, higher SIV envelope-specific IFN-γ producing cells, equivalent ADCC, and did not decrease the risk of SIVmac251 acquisition. Using the systems biology approach, we demonstrate that specific expression profiles of plasmablasts, NKG2A+ cells, and monocytes elicited by the ALVAC-based regimen correlated with decreased risk of virus acquisition.

Introduction Important advances have been made toward the development of a preventive HIV-1 vaccine, but further work is needed to address this global priority. Most notably, the RV144 HIV vaccine trial tested a recombinant Canarypox ALVAC-HIV vCP1521 vaccine administered in combination with the AIDSVAX B/E vaccine containing two monomeric clade B and AE HIV-1 gp120 proteins formulated in alum, a regimen found to decrease the risk of HIV acquisition by 31.2% [1,2]. The ALVAC vector is derived from repeated passages of Canarypox in chicken embryo fibroblasts and demonstrated a high level of safety and tolerability in phase I clinical trials in infants [3,4] and adults [5]. The results of RV144 provided valuable insight into the functionality of the ALVAC vector and revealed that the decreased risk of HIV acquisition correlated with the high serum level of IgG that recognized variable regions 1 and 2 (V1/ V2) of the HIV gp120 envelope protein [6]. Serum envelope-specific IgA, in contrast, correlated with an increased risk of viral infection [7,8]. The presence of ADCC was shown to correlate with a decreased risk of virus acquisition in the presence of low level IgA to gp120 [7]. Notably, only 19.7% of the volunteers enrolled in this trial were found to be positive for CD8 + CTL responses as measured by IFN-γ ELISpot, consistent with the lack of viral replication control in the vaccinees that became infected [1]. Conversely, most vaccinees developed polyfunctional Env-specific CD4 + T cell responses, a secondary correlate of risk of HIV-1 acquisition [9].
The ALVAC-HIV-2 and NYVAC-HIV-2 vaccines afforded protection from the weakly pathogenic HIV-2 in high dose intravaginal (IV) or mucosal challenges in early, preclinical studies in macaques [2,10]. However, a combined vaccine priming of ALVAC-SIV/Gag-Pol and ALVAC-HIV-1/Env boosted with HIV-1/gp120 did not protect rhesus macaques from acquisition of the chimeric Simian-Human Immunodeficiency Virus ku2 (SHIV ku2 ) following a high dose mucosal challenge, though it did limit T cell loss [11]. Similarly, this vaccine modality did not protect against a high dose mucosal challenge of SIV mac251 in adult macaques, and it only transiently reduced plasma virus and CD4 + T cell loss in the vaccinees that became infected [12]. The importance of the challenge dose in the efficacy of ALVAC-SIV based regimens against SIV mac251 was clearly demonstrated by comparing a single high dose to repeated intermediate intrarectal doses of the virus, and by showing that 10 out of 16 animals immunized with ALVAC-SIV/Gag-Pol-Env remained uninfected after oral repeated exposure to low doses of SIV mac251 [13,14]. More recently, the efficacy of the ALVAC-SIV/gp120 regimen was tested against rectal low dose SIV mac251 challenges that infected approximately one third of the control macaques at each challenge while transmitting few virus variants, as notably is the case with HIV in humans. In a comparison of the per-exposure rate of mucosal virus acquisition in vaccinated animals versus controls, this regimen reduced the risk of virus acquisition by 44% [15]. Furthermore, the efficacy of this regimen was modestly augmented to 52% when using a DNA prime [16].
Though the results of RV144 and these macaque studies demonstrate that the ALVAC based vaccines afford some degree of efficacy in humans and macaques, this approach clearly requires improvement. As a possible alternative to ALVAC, we evaluated the vaccinia virusderived NYVAC vector, attenuated through the deletion of 18 genes that regulate virus host range and virulence [17]. NYVAC is an attractive candidate for this purpose as it undergoes an abortive replication in most mammalian cells, is immunogenic, and it demonstrated an overall good safety profile in phase I studies [18]. In rhesus macaques, NYVAC-HIV-2 provided protection from intravenous exposure to the nonpathogenic HIV-2 strain SBL669 [19], and the inclusion of an Env protein boost in the vaccination protocol further increased protection [20,21]. In models of exposure to high doses of SIV mac251 , NYVAC-SIV alone or in combination with the IL-2 and IL-12 cytokines did not protect from virus acquisition, but it delayed disease progression in one-third of vaccinated animals following a single high dose of SIV mac251 [22]. Notably, the vaccine effect demonstrated some degree of durability in this study in animals challenged six months after the last immunization [23]. In the same high dose mucosal challenge SIV mac251 model, the substitution of the NYVAC-SIV prime with a DNA-SIV did not protect against SIV mac251 acquisition, but increased cytotoxic and helper T cell responses that correlated with protection against disease development [23,24].
In the present study, we tested adult female Chinese rhesus macaques to investigate whether the NYVAC-SIV prime/gp120 regimen could decrease the risk of virus acquisition following exposure to repeated low doses of SIV mac251 by the vaginal route. In parallel, we tested the recombinant ALVAC-SIV prime/gp120 boost, since this vaccine modality decreased the risk of HIV acquisition in humans and of SIV mac251 acquisition in macaques following intrarectal exposure [1,15] but was not previously tested against vaginal challenge. Here, we demonstrated that the ALVAC-SIV/gp120/alum vaccine regimen decreases the risk of vaginal SIV mac251 acquisition in Chinese rhesus macaques and confirmed that higher levels of classical monocytes correlate with a decreased risk of virus acquisition. Our data further suggest that this monocyte subset affects the frequency and function of NKG2A + cells in vaginal mucosa. Surprisingly, the NYVAC-based vaccine regimen did not decrease the risk of SIV mac251 acquisition. This vaccine regimen increased the frequency of activated cells, and systems biology analyses show that NYVAC-SIV elicited a different inflammatory profile than the ALVAC-based regimen. All together, these findings highlight a complex interplay between vaccine-induced innate and adaptive immunity in shaping responses to inhibit SIV acquisition. Noting their differences, we contrast the immune responses elicited by these regimens to better understand the correlate of protection.

ALVAC-SIV/gp120 but not NYVAC-SIV/gp120 vaccination reduces the risk of SIV mac251 acquisition
Forty female Chinese rhesus macaques were distributed equally in the groups based on age, and weight ( Table 1). Twenty animals were immunized in each vaccine group with either ALVAC-SIV or NYVAC-SIV expressing identical SIV genes (Fig 1A). Animals were first immunized at weeks 0 and 4, and boosted at weeks 12 and 24 with the corresponding viral vector in one limb and a single monomeric native SIV mac251 /gp120 protein formulated in 5 mg of alum in the contralateral limb. A total of 25 macaques were used as controls: two groups of ten macaques each were immunized with the parental viral vectors and adjuvant, while five animals were left naïve (Fig 1A). The number of animals used in each immunization group was not sufficient to compare the relative efficacy of the ALVAC-SIV/gp120 (ALVAC-SIV) and NYVAC-SIV/gp120 (NYVAC-SIV) based vaccines. Rather, this population was adequate to compare each vaccine to mock vaccinated or naïve controls. Four weeks following the last immunization, twelve consecutive challenges were performed with a weekly low intravaginal dose of SIV mac251 using a challenge stock with high genetic diversity, propagated in macaque cells by R. C. Desrosiers [25]. The study was divided into two parts (Part 1 and Part 2) and conducted in two separate animal facilities ( Table 1). Before the initiation of the study, it was decided that if no difference in the rate of SIV mac251 acquisition was observed among the three control groups, vaccine efficacy would be assessed by pooling all controls, and comparing them the vaccinees immunized with each vaccine regimen. The control groups (ALVAC-control, NYVAC-control, and Naïve groups) did not differ significantly from each other following intravaginal challenge exposure to SIV mac251 (S1A Fig). Additional data showing that the ALVAC-SIV group had an estimated vaccine efficacy of 50% at each challenge (Log-rank test: p = 0.0471; Fig 1B) confirmed and built upon the results of one of our prior studies of an equivalent vaccine regimen tested against intrarectal challenges in male and female Indian

PLOS PATHOGENS
ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition rhesus macaques [15]. Unexpectedly, vaccination with NYVAC-SIV did not decrease the risk of virus acquisition in vaccinated animals (Log-rank test: p = 0.2062; Fig 1C). While there was no overall difference in the plasma virus levels of the vaccinated animals that became infected and the controls (Fig 1D), a transient reduction of viremia was observed in the NYVAC group (2 weeks from infection; NYVAC vs. pooled controls p = 0.0019) and in the ALVAC-group (ALVAC vs. control p < 0.001, by the Wilcoxon-Mann-Whitney test, not corrected for multiple comparisons; Fig 1E). To determine whether vaccination affected virus levels in mucosal tissues, we quantified SIV DNA copies in the rectal and vaginal biopsies at 2 weeks post-infection. In rectal tissue, we found a significant difference only between the NYVAC-SIV and control groups (p = 0.019; Fig 1F), whereas both ALVAC-SIV and NYVAC-SIV immunized animals had significantly lower SIV DNA copies than the control group in the vaginal mucosa (p = 0.0093 and p = 0.0015, respectively; Fig 1G).

Monocyte subsets and MDSCs differently affect the risk of vaginal SIV mac251 acquisition in the ALVAC-SIV regimen
CD14 + classical, CD14 + CD16 + intermediate, and CD14 -CD16 + non-classical monocytes and myeloid derived suppressor cells (MDSCs) play a key role in the modulation of adaptive CD4 + and CD8 + T and B cell responses induced by poxvirus vectors [26] and HIV and SIV infection [27][28][29]. Measurement of the frequency of each monocyte subset in blood before immunization (week 0) and two weeks following the final immunization (week 26) revealed that the two vaccine regimens did not significantly change the overall blood frequency of total or classical monocytes (S1B and S1C Fig), nor of CXCR4 + CD14 + CD16classical monocytes (S1D Fig).
Similarly, the frequency of total, CXCR4 + , intermediate CD14 + CD16 + , or non-classical CD14 -CD16 + monocytes did not differ between the groups (S1E-S1H Fig). Analysis of the frequency of CCR2 + monocyte subsets only demonstrated a significantly higher frequency (p = 0.0269) of the CCR2 + CD14 + CD16 + intermediate monocyte subset in the macaques immunized with NYVAC-SIV (S1I, S1J and S1K Fig). Notably, the frequency of both total and classical monocytes in the ALVAC-SIV group correlated positively with delayed virus acquisition (R = 0.70, p = 0.0277 and R = 0.73, p = 0.0208, respectively; Fig 2A and 2B), confirming our prior results with the ALVAC-based vaccine modality [16]. In contrast, we did not find a correlation between CXCR4 + monocytes and the risk of virus acquisition in the present study. The different vaccine regimens and sample collection times between the current (week 26, ALVAC-SIV prime) and prior (week 27, DNA prime) studies may possibly account for this difference.
In agreement with our prior work, changes in gene expression in the blood of animals in the ALVAC-SIV group (week 26 compared to pre-vaccination) supported the association of classical monocyte cells and inflammasome activation with decreased risk of SIV acquisition (Fig 2C). While no association was observed between SIV acquisition and monocyte frequency in the NYVAC-SIV group, classical monocyte genes associated with the decreased risk conferred by ALVAC-SIV vaccination included AKR1B1 (a marker of M1 macrophage polarization [30]), CCL3 and CDKN1A (monocytic inhibitors of HIV-1 replication [31,32]), CCR2 and CD44 (promotors of monocyte chemotaxis [33,34]), CD14 (the co-receptor of LPS at the surface of monocytes), CLEC7A (an inducer of NLRP3 inflammasome [35]), IL1β (the byproduct of NLRP3 inflammasome activation [36]) and its receptor IL1R2 [37], and F13A1

Classical and intermediate monocytes in blood are associated with cytocidal vaginal NKG2A
In a prior study, we found that ALVAC-SIV/gp120 immunization increased the frequency of NKp44 + cells in the rectal mucosa that in turn correlated with delayed virus acquisition following rectal exposure to SIV mac251 [15]. Furthermore, the frequency of mucosal NKp44 + cells correlated with the plasma level of CCL2, a chemokine produced at a high level by classical monocytes [39]. NK cells are considered to be the first line of defense against viral infections [40] because of their ability to exert cytotoxic activity toward virus-infected cells without the need for MHC-mediated activation, and their regulation of the inflammatory milieu. Consistent with prior reports, we observed low level NKp44 + cells and the prevalence of NKG2A + cell in vaginal mucosa [41,42]. The frequency of NKG2A + cells (defined as CD45 + CD3 -CD20 -CD14 -NKG2A + cells) in vaginal mucosa did not differ among vaccinated and control animals at week 13 (Fig 2D and S2A Fig). The frequency of vaginal NKG2A + cells correlated with the level of intermediate monocytes at the end of immunization (R = 0.69, p = 0.0306, week 26; Fig 2E). Vaginal NKG2A + cells with cytotoxic profiles (CD107a + ) were significantly higher in ALVAC-SIV immunized macaques than in the NYVAC-SIV or control groups (p < 0.0001; Fig 2F). Interestingly, gene expression related to NK cytotoxicity directly correlated with the classical monocyte transcriptomic signature and decreased risk of SIV mac251 acquisition ( Fig  2G). CD14 + monocytes are known to recruit cytotoxic NK cells following inflammasome activation and production of IL-18. We found that cells stimulated with envelope overlapping peptides produced a cytokine profile marked by a trend of IL-18 and the average gene expression of classical monocyte markers (Spearman correlation: R = 0.35, p = 0.266; Fig 2H). IL-18 also for the ALVAC-SIV/gp120, NYVAC-SIV/gp120, and control groups from Part 2 (horizontal line: mean). (E) Correlation between the frequency of intermediate monocytes in blood at week 26 and the cytotoxic function of mucosal NKG2A cells (week 13). (F) Frequency of vaginal NKG2A CD107 + cells in the ALVAC-SIV vaccinated, NYVAC-SIV vaccinated, and control groups from Part 2 of the study at week 26 (horizontal line: mean). (G) Scatterplot of the average of cytotoxic NK cell markers as a function of the average expression of classical monocyte markers. The gene expression of FAS and TNF, two canonical cytotoxic NKs, as function of the expression of classical monocyte markers is indicated by lines. (H) Heatmaps showing the level of expression of transcriptomic markers of classical monocytes in three NHP studies. In the current study (ALVAC/gp120 [Ivag]; left), animals were primed with ALVAC-SIV and boosted with ALVAC-SIV+gp120 formulated in alum. Blood samples were taken 24h after the first boost (week 12) and after the 2nd boost (week 24). In a prior study (DNA/ALVAC/gp120 [IR]; center), animals were primed with DNA and boosted with ALVAC-SIV/gp120 formulated in alum [16]. Blood samples were taken 24h after the first (week 12) and second (week 24) boosts. In another of our prior studies (ALVAC/gp120 [IR]; right), animals were primed with ALVAC-SIV, boosted with ALVAC-SIV+gp120 formulated in alum, and challenged intrarectally [15]. Blood samples were taken 24 h after the first ALVAC-SIV immunization. GSEA enrichment analysis was used to test for enrichment of transcriptomic markers of classical monocytes [80] among genes correlated with challenges in each study (ALVAC/gp120  Fig 2I). Although these individual correlations are weak, together they support the hypothesis that ALVAC-SIV vaccination engages CD14 + monocytes and affects the function of cytocidal vaginal NKG2A + cells, via IL-18 production.
To further investigate functional cytocidal cells in blood, we measured ADCC activity and ADCC titers mediated by plasma from 7 animals from each group in Part 2 of the study using purified SIV 766 gp120 coated target cells [43]. ADCC activity was measured in these ALVAC--SIV and NYVAC-SIV vaccinated animals one week after the last immunization (Fig 3A and  3D). The ADCC titers did not differ at the end of immunization in the two groups (S2B and S2C Fig). There was, however, a trend between both ADCC activity and ADCC titers with delayed SIV mac251 acquisition in the ALVAC-SIV group (Fig 3B and 3C), but not in the NYVAC-SIV group (Fig 3E and 3F).

PLOS PATHOGENS
ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition weeks following the last immunization (week 26), animals vaccinated with NYVAC-SIV had a significantly higher percentage of vaccine-induced (Ki67 + ) activated CD38 + and gut-homing α 4 β 7 + CD4 + T cells in blood than did the ALVAC-vaccinated animals (p = 0.0293 and p = 0.0010, respectively, Mann-Whitney test; Fig 4B and 4C). In contrast, macaques vaccinated with ALVAC-SIV had significantly higher vaccine-induced Ki67 + CXCR3 + CCR6 -Th1-type CD4 + T cells (p = 0.0076; Fig 4D and S2F Fig) and significantly lower total CXCR3 -CCR6 -Th2-type CD4 + T cells (p = 0.0176; Fig 4E) in blood than the NYVAC-vaccinated animals. The frequency of Th1 CD4 + cells in ALVAC-vaccinated animals was strongly associated with both decreased risk of SIV mac251 acquisition (R = 0.97, p = 0.0111; Fig 4F) and the level of classical monocytes (R = 0.94, p = 0.0167; Fig 4G). The levels of CCR5 on CD4 + T cells were similar in both vaccines (S2G Fig). Conversely, there was a negative trend between the frequency of CD4 + Th2 cells and early SIV mac251 acquisition in the same group (R = -0.79; p = 0.100; Fig  4H). Together, these data demonstrate that ALVAC and NYVAC vectored vaccines induce functionally different CD4 + T cell subsets and that classical monocytes play a key role in the induction of CD4 + Th1 cells associated with reduced risk of SIV mac251 acquisition. The importance of CD4 + Th1 cells is further supported by the finding that the frequency of CD14 + HLA-DRcells (MDSCs), whose frequency is associated with an increased risk of virus acquisition (S1M Fig), is also associated with the suppression of envelope-specific Th1 CD4 + IFN-γ producing cells (R = -0.71, p = 0.0268; Fig 4I). No differences were found in the percentage of other phenotypically defined CD4 + T cell subsets, such as Th17 and T follicular helper cells, nor was a correlation found between any of the CD4 + T cell subtypes and the rate of SIV mac251 acquisition in the animals immunized with NYVAC.

Frequency of gut-homing α 4 β 7 + plasmablasts correlates with decreased risk of SIV mac251 acquisition
To assess the effect of vaccination on B cells, we analyzed blood plasmablasts (PBs) [15] and the α 4 β 7 and CXCR3 homing markers on the surface of PBs before immunization (week 0) and one week after the last immunization (week 25). The α 4 β 7 integrin mediates lymphocyte migration to mucosal sites by binding to MAdCAM-1, where it is expressed on the inner surface of the mucosal venules [44]. As a chemokine receptor, CXCR3 binds chemokines CXCL9 and CXCL10, generally released at the site of inflammation where CXCR3-expressing cells are recruited [45].Vaccination with NYVAC-SIV, but not ALVAC-SIV, induced a significant increase in the percentage of total circulating PBs after vaccination (p = 0.0428; Fig 5A and  5B). In the NYVAC group, the PB level rose due to an increase in CXCR3 + plasmablasts (p = 0.0324; Fig 5C) and a smaller decrease in α 4 β 7 + PBs (p = 0.0106; Fig 5D).
Significant changes in total, α 4 β 7 + , or CXCR3 + plasmablasts were not observed in the ALVAC-SIV group (Fig 5E and 5F). Analysis of antibody responses to the envelope protein revealed different kinetics of induction of antibodies to gp120 that were faster in NYVAC-SIV, but the level of envelope-specific systemic antibodies or mucosal IgG to the gp70 V1/V2 scaffold did not differ between the two groups at the end of immunization (S3A and S3B Fig).  Fig 5G), whereas no differences were observed in NYVAC-SIV immunized animals (p = 0.5331; Fig 5H). Transcriptomic profiling of PBMCs from ALVAC-SIV/gp120 identified genes (Fig 5I) whose expression trended with a decreased risk of virus acquisition (R = 0.43, p = 0.0766; Fig 5J), including TNFRSF13B (a promoter of B cell proliferation and plasma cell differentiation [46]), CCR10 (encoding a chemokine receptor allowing plasmablast homing to mucosal Ab sites [47]), MZB1 (a chaperone essential for plasma cell differentiation [48]), TNFRSF17 (a plasma cell pro-survival gene [49]), CD27 (a marker of mature memory B cells [50]), CD38 (a marker of long-lived plasma cells [51]), and both FCGR2B and CD19 (two inhibitors of B cell differentiation to plasma cells [52]).

NYVAC and ALVAC based vaccines differently affect gene expression
Data analysis of whole blood obtained following the immunizations revealed that ALVAC-SIV induced a much stronger innate transcriptomic response than NYVAC-SIV. The response primarily occurs early enough that most of the differentially expressed (DE) genes can be identified within 24 hours of immunization (

Discussion
Our study demonstrates that the canarypox based ALVAC-SIV/gp120/alum regimen decreases the risk of vaginal acquisition of SIV mac251 in Chinese rhesus macaques with a vaccine efficacy of 50%. These results reproduce the efficacy afforded by the ALVAC-SIV/gp120/alum regimen in male and female Indian rhesus macaques (44% efficacy) following intrarectal exposure to the same virus stock [15] and reaffirm the efficacy of this vaccine strategy. In contrast, an identical vaccine regimen vectored with the vaccinia-derivative NYVAC was surprisingly not efficacious. The immune responses induced by these two vaccine regimens were marginally higher in NYVAC-SIV immunized animals as also observed in macaque studies using HIV clade C immunogens [54]. Importantly, similar results were reported from the recently completed HVTN096 HIV trial in humans [55], demonstrating that the NYVAC-HIV vector expressing clade C gp140 in combination with the bivalent heterologous boost used in RV144 (clade B MNgp120 and clade AE A244gp120) elicited 2-4-fold higher antibody responses to fewer V1/V2 scaffolds than in RV144. As in RV144, the antibody response to V1/V2 was not sustained. The modest difference between the antibody responses of RV144 and HVTN096 trials may have also been influenced by the prime, the immunogen expressed in the poxvirus vectors (gp120 in RV144; gp140 in HVTN096), and by the relation of the prime envelope to the boost envelope (homologous in RV144; heterologous in HVTN096). The DNA prime in HVTN906 was also tested in combination with NYVAC-HIV and found to elicit marginally higher CD4 + and CD8 + T cell responses than the NYVAC-HIV prime, although the responses were not sustained. Similar observations were reported in macaques using a DNA-SIV/ NYVAC-SIV vaccine regimen without the gp120 protein boost [24]. By integrating phenotypic cell subset analyses and functional immunological assays with systems biology in the present study, we demonstrate that the ALVAC regime induces a higher level of durable interferon responses during priming than NYVAC-SIV that support the distinct kinetic of induction of cytokines and chemokines within the first 24 hours from immunization [56].
The differences between ALVAC and NYVAC may derive from their different host ranges. The Canarypox-based ALVAC only replicates effectively in avian species [18,57], and it therefore did not evolve the necessary genetic determinants to hijack or inhibit the more complex immune systems of mammals. In contrast, poxviruses such as the vaccinia derivative NYVAC, smallpox, and monkeypox co-evolved with mammals, allowing for the selection of multiple genes to counteract mammalian immune responses [58]. We demonstrate here that ALVAC is more effective than NYVAC in harnessing innate responses against the heterologous genes expressed by the recombinant ALVAC vectored vaccines. Accordingly, recombinant ALVA-C-HIV vaccines have induced high levels of ALVAC-specific CD8 + T cell responses in humans, but negligible cytotoxic CD8 + T cells to the HIV insert [59]. ALVAC-SIV induces protracted interferon production during priming, as demonstrated by systems biology. Importantly, a strong interferon signature was also observed in RV144 vaccinees [53]. ALVAC (but not NYVAC) possesses the ability to preferentially infect CD14 + monocytes [60], activate the inflammasome, and induce the release of both IL-1β and IL-10. Within the first 24 hours of vaccination with ALVAC, the plasma levels of IL-1β and IL-10 increase by 20-50-fold [56]. The finding that NYVAC-SIV immunization did not reduce the risk of SIV mac251 acquisition was somehow unexpected given the overall similarity of SIV specific responses elicited by the two vaccine regimens, including antibodies to V1/V2, neutralizing antibody titers, and ADCC.
A more in-depth investigation of vaccine-induced immuneresponses revealed significant qualitative differences in T cell responses and homing markers on plasmablasts. We found that CD4 + Th1 cells were higher in the ALVAC-SIV group and correlated with a decreased risk of PLOS PATHOGENS ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition SIV mac251 acquisition. In contrast, CD4 + T Th2 cells were higher in the NYVAC-SIV group and correlated with an increased risk of SIV mac251 acquisition. Oddly, the CD95 + Ki67 + CD4 + Th2 cell subset correlated with a decreased risk of SIV mac265 acquisition in a study whereby the ALVAC-SIV prime was substituted with a DNA prime [16]. In addition, NYVAC-SIV induced significantly higher levels of gut-homing Ki67 + CD4 + α 4 β 7 + and Ki67 + CD38 + activated T cells.
Both CD4 + Ki67 + subsets correlated with an increased number of transmitted virus variants in vaccinated animals that became infected in a prior study, and the Ki67 + CD4 + α 4 β 7 + T cell subset also correlated with an increased risk of SIV mac251 acquisition [15]. Analysis of CCR5 expression in vaccine induced CD4 + T-cells did not reveal significant differences between the two groups. Plasmablasts expressing the homing marker CXCR3 for inflammatory sites were increased in the NYVAC-SIV group, and those expressing the α 4 β 7 homing marker for mucosal sites were decreased as expected, suggesting differential migration to antibody-producing cell tissues. However, we could not detect differences in serum and mucosal binding antibody responses and functional serum antibody responses, such as neutralizing antibodies and ADCC, in the two vaccinated groups. Importantly, we demonstrated that inflammasome activation in classical monocytes is a strong correlate of reduced risk of SIV mac251 acquisition, not only following intrarectal exposure to SIV mac251 as oberved previously [16], but also following intravaginal exposure to the same virus stock [25]. Collectively, our data on the ALVAC-based vaccine suggest the role of monocyte mediated (trained) immunity, an ancient response to pathogens linked to emergency myelopoiesis and durable epigenetic changes in monocytes [61].
Our findings raise the question of how monocytes influence the decreased risk of virus acquisition. We found here an association between vaccine induced CD14 + monocytes and vaginal NKG2A + CD107 + cells and, in the prior study, between CCL2 (a chemokine produced largely by CD14 + classical monocytes) [39] and rectal NKp44 + cells [15] (Fig 6). Both of these NK subsets correlated with a decreased risk of intravaginal or intrarectal SIV mac251 acquisition (Fig 6). In addition, we found a trend between blood ADCC activity and ADCC titers with a reduced risk of SIV mac251 acquisition in a subgroup of ALVAC-SIV vaccinated macaques, whose plasma was available. NKp44 + cells are important for maintaining gut homeostasis [42,62]. NKG2A + cells express the inhibitory receptor that limits the magnitude and duration of antiviral cytotoxic responses, possibly curbing mucosal tissue inflammation [63]. However, the antibody used to identify the NKG2A + cells population has been shown to cross-react with NKG2C, an activating receptor [64] that precludes, at present, a definitive characterization of these cells. Interestingly, we found in a prior study that mucosal NKG2A -NKp44cells producing IFN-γ were associated with increased SIV mac251 acquisition [15]. Further work will thus be necessary to define the role and function of NK cells subsets and monocytes in vaccine protection.
Our data suggest that systemic immunization with the ALVAC-SIV/gp120/alum regimen influences the function of mucosal NK cell responses, likely via monocytes and/or the cytokines and chemokines produced by monocytes. This begs the question of what the functional features of these monocytes are, and whether they are pro-inflammatory or antiinflammatory.
We therefore favor the hypothesis that monocytes induced by the ALVAC-SIV regimen may be predominantly anti-inflammatory. This hypothesis is supported by the finding that ALVAC is a potent inducer of interferon responses. Recent work in a mouse model study of T. Gondii infection demonstrated that interferons are cytokines that can "educate" monocytes to became anti-inflammatory [66]. In that study, the production of interferons by bone marrow PLOS PATHOGENS ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition NK cells during the first few hours of infection was able to prime monocytes to become antiinflammatory before their egress from bone marrow [66]. Thus, the current and prior studies suggest the hypothesis that the ALVAC-SIV vaccine engages NK cells by inducing an early burst of IFN-γ and infecting CD14 + monocytes, activates the inflammasome, and shapes innate responses at mucosal sites associated with a reduced risk of virus acquisition (Fig 6). Thus, CD14 + monocytes would have an indirect effect on vaccine efficacy by orchestrating other protective responses. Indeed, the frequency of CD14 + monocytes correlated with the frequency of vaccine-induced CD4 + Th1 in this study, as well as with Th2 negative for CCR5 expression in a prior one where the ALVAC-SIV prime was substituted with a DNA prime [16]. In turn, the number of both CD4 + Th1 and Th2 (CCR5 -) T cell subsets and CD14 + monocytes correlated with a reduced risk of virus acquisition (Fig 5). The contribution of antibodies to cyclic V2 to the protection observed earlier [15] could not be fully assessed in the current work because of insufficient samples. In sum, the present study reproduced the efficacy of RV144 in Chinese rhesus macaques following challenge exposure by the vaginal route. The efficacy of the ALVAC-SIV/gp120/ alum vaccine regimen is linked to innate responses, such as interferon and myeloid cells during priming, CD4 + Th1 responses, NKG2A + cells, and ADCC. This work thus defines novel correlates of risk and demonstrates the reproducibility of the efficacy observed in the ALVACbased regimen in the rigorous SIV mac251 macaque model, thereby establishing a benchmark for the future improvement of HIV vaccine candidates.

Animals, vaccines and SIV mac251 challenge
All animals included this study were female rhesus macaques (Macaca mulatta) imported from China and obtained from the Washington National Primate Research Center (Seattle, WA). Animals were not tested for MHC-I expression. The care and use of the animals were in compliance with all relevant NIH institutional guidelines. A total of 65 female rhesus macaques were randomized into five groups (the five groups are referred to in the text as ALVAC-SIV/ gp120 [ALVAC-SIV], NYVAC-SIV/gp120 [NYVAC-SIV], ALVAC-control, NYVAC-control, and naïve). The animals in the ALVAC and NYVAC-SIV groups were immunized at weeks 0, 4, 12, and 24 with intramuscular inoculations at 10 8 PFU of either ALVAC-SIV (vCP180) or NYVAC-SIV (VP1071) carrying the identical Env-Gag-Pol genes from SIV K6W [67]. At weeks 12 and 24, the animals from these groups received 200 μg of native SIV mac251 gp120 protein [12] formulated with 5mg of alum, as a boost in the opposite thigh of the vector immunization.
The control groups included 10 animals each, which received either parental ALVAC-SIV or NYVAC-SIV vectors and alum. Five naïve control animals were included before the challenge phase. Animals were challenged four weeks after the last immunization (week 28) with SIV mac251 [25] at 120 TCID 50 for each challenge as previously described [15,68]. Animals that tested negative for SIV-RNA in plasma were rechallenged with up to a maximum of 12 weekly administrations.

Measurement of viral RNA and DNA
Plasma SIV mac251 RNA levels were quantified by nucleic acid sequence-based amplification [69]. SIV/DNA levels in mucosal biopsies from week 2 post-infection were quantified by a real-time qPCR with sensitivity set at ten copies x 10 6 cells, as previously described [70].

PLOS PATHOGENS
ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition Gaithersburg, MD). The reaction was blocked for 2 hours at 37˚C with PBS containing 5% FBS. After washing the plates three times with 0.25% Tween 20 Dulbecco's PBS, they were rinsed with 10% FBS-RPMI 1640, and incubated in triplicate with 5 × 10 5 PBMCs/well in a 100-μl reaction volume with peptide at a concentration of 8 μg/ml. After an 18-hour incubation, Dulbecco's PBS containing 0.25% Tween 20 was used to wash the plates five times, and once with distilled water. After a 16-hour incubation with 75 μl/well 5 μg/ml biotinylated rat anti-mouse IFN-γ, the plates were washed six times with Coulter wash (Coulter, Miami, FL), and incubated with a 1/500 dilution of streptavidin-AP (Southern Biotechnology Associates, Birmingham, AL) for 2.5 hours. Later, the plates were washed five times with Coulter wash and once with PBS, then developed with nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate chromogen (Pierce, Rockford, IL) and stopped by washing with tap water. Lastly, the plates were air-dried and read using an ELISpot reader (Hitech Instruments, Edgemont, PA).

Neutralizing antibodies
Serum neutralizing activity was measured as reduction in expression of luciferase reporter gene after a single round of infection in TZM-b1 cells, as described previously [72]. TZM-b1 cells were obtained from the NIH AIDS Research and Reference Reagent Program as contributed by J. Kappes and X. Wu. Briefly, 200 TCID 50 of pseudoviruses were incubated with serial 3-fold dilutions of test sample in duplicate, in a final volume of 150 μl for 1 out of 18 h at 37˚C in 96-well flat-bottom culture plates. Freshly trypsinized cells (10,000 cells in 100 μl of growth medium containing 75 μg ml −1 DEAE-dextran) were then added to each well. Two sets of control wells were included as controls: the one set to receive cells and virus (positive control) and the other set to receive cells only (background control). For measurement of luminescence, cells were then transferred to 96-well black solid plates (Costar) and signal was detected using the Britelite luminescence reporter gene assay system (PerkinElmer Life Sciences). Neutralization titers were defined as the dilution at which relative luminescence units were reduced by 50% compared to that in positive control wells minus the background signal detected in negative control wells. Stocks of Env-pseudo-typed viruses (SIV mac251.6 and SIV mac251.30 ) were prepared by transfection in 293T cells and titrated in TZM-bl cells, as previously described [73].

SIV Env-specific serum IgG binding antibody assay
The total macaque IgG were measured by macaque IgG ELISA and a custom SIV bAb multiplex assay (SIV-BAMA) was used to quantify SIV Env-specific IgG antibodies in serum as previously described [74,75]. Purified IgG (DBM5) from a SIV-infected macaque (kindly provided by M. Roederer, VRC, NIH) was used as the positive control to calculate SIV antibody concentration. A Levy-Jennings Plot was used to track positive controls for each group. Specific activity was calculated from the total macaque IgG levels and the SIV specific concentrations. The quantitation of antibodies against native V1/V2 epitopes was performed through binding assays against native SIV V1/V2 antigens expressed as gp70-fusion proteins related to the CaseA2 antigen used in the RV144 correlate study (provided by A. Pinter). These synthetic proteins contain the glycosylated, disulfide-bonded V1/V2 regions of SIV mac239 , SIV mac251 , and SIVs mE660 (corresponding to AA 120-204 of HXB2 Env), linked to residue 263 of the SU (gp70) protein of Fr-MuLV.

ADCC assay
ADCC activity was assessed as previously described using a constitutive GFP expressing EGFP-CEM-NKr-CCR5-SNAP cells as target [43]. Briefly, one million target cells were PLOS PATHOGENS ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition incubated with 50 μg of SIV gp120 wild type protein for 2 h at 37˚C, washed, and labeled with SNAP-Surface Alexa Fluor 647 (New England Biolabs, Ipswich, MA; cat.# S9136S) as recommended by the manufacturer for 30 min at RT. Heat inactivated plasma samples were serially diluted (7 ten-fold dilutions starting at 1:10) and 100 μl were added to a 96-well V-bottom plate (Millipore Sigma). Following this, 5000 target cells (50 μl) and 250,000 human PBMCs (50 μl) as effectors were added to each well to give an effector/target (E/T) ratio of 50:1. The plate was incubated at 37˚C for 2 h followed by two PBS washes. The cells were re-suspended in 200 μl of a 2% PBS-paraformaldehyde solution and acquired on an LSRII equipped with a high throughput system (BD Biosciences, San Jose, CA). Specific killing was measured by loss of GFP from the SNAP-Alexa647 + target cells. Target and effector cells cultured in the presence of medium were used as negative controls. Anti-SIVmac gp120 monoclonal antibody, KK17 (NIH AIDS reagent program), was used as a positive control. Normalized percent killing was calculated as the following: (killing in the presence of rectal secretion-background)/ (killing in the presence of KK17-background) ×100. The ADCC endpoint titer is defined as the reciprocal dilution at which the percent ADCC killing was greater than the mean percent killing of the negative control wells containing medium, target and effector cells, plus three standard deviations.

IgG linear epitope mapping in serum
The first week after the last immunization, 1:20-diluted sera were added to ELISA plates coated with overlapping peptides encompassing the entire SIV K6W gp120 amino acid sequence [76]. Linear peptide mapping of serum was performed by microarray (PepStar) [77]. Briefly, JPT Peptide Technologies GmbH (Germany) produced array slides designed by Dr. B. Korber (Los Alamos National Laboratory) by printing a library of overlapping peptides (15-mers overlapping by 12) covering full-length gp160 of SIV mac239 and SIV smE660 onto epoxy glass slides (PolyAn GmbH, Germany). One printing area of each quad-slide contained three identical sub-arrays, each containing the full peptide library. After hybridizing the slides using a Tecan HS4000 Hybridization Workstation, the samples were incubated with DyLight 649-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch, PA). Later, fluorescence intensity was measured with a GenePix 4300 scanner (Molecular Devices) and analyzed with the GenePix software. The background value was subtracted from the binding intensity of the postimmunization serum to each peptide, defined as the median signal of the pre-bleed serum for that peptide plus three-times the standard error among the three sub-arrays on slide. The total IgG concentration measured was used to normalize the values for each peptide as reported above (Unit = signal intensity/μg/ml total IgG).

RNA isolation and microarray processing
Total RNA was isolated from Paxgene tubes by the use of Paxgene Blood RNeasy minikits (Qiagen) following the manufacturer's protocol. RNA quality was assessed on an Agilent 2100 bioanalyzer, using the nanochip format, and only intact RNA was used for microarray analysis. Fifty nanograms of each RNA sample were hybridized to one Agilent 8×60 rhesus macaque array (Agilent, G4102A). Raw array intensities were read, background corrected with the norm-exp method, quantile-normalized and log2 transformed for variance stability, using functions implemented in the R package limma. For each array, intensities of replicated probes were averaged. Two arrays (for samples P168_R808_NYVAC-SIV_w4_0h and P168_A06028_ALVAC-SIV_w0_6h) with low overall raw intensities were considered outliers and excluded from downstream analysis. Technical replicates arrays (2 replicates for 4 samples) were averaged prior to the differential gene-expression analysis. To identify genes differentially expressed between ALVAC-SIV/gp120 and NYVAC-SIV/gp120 and their respective empty vector control (ALVAC-control and NYVAC-control) adjusting for pre-vaccination transcriptomic profile, we first subtracted the pre-vaccination expression from all post-vaccination samples separately for each donor (i.e. to obtain fold-change post/pre-vaccination) and fitted a linear model with the treatment as independent variable for each gene. A moderated ttest, as implemented in the R package limma, was used to assess the statistical significance of the difference between vaccine and empty vector control. Benjamini & Hochberg correction was used to adjust for multiple testing.
Gene Set Enrichment Analysis (GSEA) was used to identify pathways that were enriched in the set of genes that distinguished the treatment groups, where one treatment group (ALVAC-SIV) showed a reduced rate of SIV mac251 acquisition. In GSEA, the most varying probes across samples was used to remove redundant probes annotated to a same gene. The gene list ranked by LIMMA moderated t-statistic were used as input for the GSEA analysis. The pathways (i.e. genesets) database used for all GSEA analysis were the Molecular Signatures Database (version 6.1) genesets and blood cells markers 48 . The GSEA Java desktop program was downloaded from http://www.broadinstitute.org/gsea/index.jsp and the default parameters of GSEA preranked module (number of permutations: 1000; enrichment statistic: weighted; seed for permutation: 101, 15 � gene set size � 500) were applied for analyses.

Statistical analysis
Continuous factors between the two groups were compared using the Wilcoxon rank-sum test. The Spearman's rank correlation was used to perform correlation analyses with the calculation of exact permutation p values. The Log-rank test of the discrete-time proportional hazards model defined the number of challenges before acquisition of infection. The changes in plasmablast levels from pre-to post-vaccination were evaluated by the paired Wilcoxon signed-rank test.

Code availabilities
Code used to generate the figures is available at https://github.com/sekalylab/p168 PLOS PATHOGENS ALVAC-SIV but not NYVAC-SIV vaccines decrease SIV mac251 acquisition

Ethics Statement
All animals included in this study were female rhesus macaques (Macaca mulatta) imported from China and obtained from the Washington National Primate Research Center (Seattle, WA). The care and use of the animals were in compliance with all relevant NIH institutional guidelines. Animals were cared for in accordance with the American Association for the Accreditation of Laboratory Animal Care (AAALAC) standards in AAALAC-accredited facilities (Animal Welfare Assurance A4149-01). All animal care and procedures were carried out under protocols approved by the NCI or NIAID Animal Care and Use Committees (ACUC; Protocol P168) and/or the University of Washington Institutional Animal Care and Use Committee (UW protocol #4266-04). The studies adhered to the regulations/guidelines as outlined in the Guide for the Care and Use of Laboratory Animals, Eighth Edition; The Animal Welfare Act; CDC "Biosafety in Microbiological and Biomedical Laboratories"; and Public Health Service Policy on Humane Care and Use of Laboratory Animals. The Animal Care and Use Committee approved all experiments and met all applicable federal and institutional standards. Animals were closely monitored daily for any signs of illness, and appropriate medical care was provided as needed. Animals were housed individually during the challenge phase to reduce the risk of transmission of SIV or other viruses. All clinical procedures, including biopsy collection, administration of anesthesia and analgesics, and euthanasia, were carried out under the direction of a laboratory animal veterinarian. Steps were taken to ensure the welfare of the animals and minimize discomfort of all animals used in this study. Animals were fed daily with a fresh diet of primate biscuits, fruit, peanuts, and other food items to maintain body weight or normal growth. Animals were monitored for mental health and provided with physical enrichment including sanitized toys, destructible environments (cardboard and other paper products), and audio stimulation. Heatmap of interferon geneset associated with the number of SIV challenges to infection in at least one vaccine/ immunization/timepoint. GSEA was used to assess the enrichment of the 31 interferon genesets in the MSigDB databases. The Normalized Enrichment Score (NES) of the genesets is depicted in the heatmap with a blue-white-red color gradient; NES < 0 indicates that the geneset is associated with increased risk of acquisition, while NES > 0 means that the interferon geneset is associated with lower risk of acquisition (i.e. protection). Enrichments associated with FDR > 0.05 are shown in grey. The x axis records the number of weeks and hours from vaccination (eg., "w12.24" = 12 weeks, 24 h post-vaccination). (PPTX)