Mucosal immune stimulation with HSV-2 and polyICLC boosts control of viremia in SIVΔNef vaccinated rhesus macaques with breakthrough SIV infection

Development of an effective human immunodeficiency virus (HIV) vaccine is among the highest priorities in the biomedical research agenda. Adjuvants enhance vaccine efficacy, but in the case of HIV, strong or inappropriate immune activation may undermine protection by increasing HIV susceptibility. Co-infection with immunomodulatory pathogens may also impact vaccine efficacy. In the rhesus macaque rectal SIVΔNef live attenuated vaccine model, we utilized a low virulence HSV-2 infection and the double-stranded RNA viral mimic polyICLC as tools to probe the effects of distinct types of immune activation on HIV vaccine efficacy and explore novel correlates of protection from wild type SIV. Rectally administered HSV-2 and polyICLC impacted the protection conferred by mucosal SIVΔNef vaccination by favoring partial protection in animals with breakthrough infection following virulent SIV challenge (“Controllers”). However, SIVΔNef persistence in blood and tissues did not predict protection in this rectal immunization and challenge model. Non-controllers had similar SIVΔNef viremia as completely protected macaques, and while they tended to have less replication competent SIVΔNef in lymph nodes, controllers had no recoverable virus in the lymph nodes. Non-controllers differed from protected macaques immunologically by having a greater frequency of pro-inflammatory CXCR3+CCR6+ CD4 T cells in blood and a monofunctional IFNγ-dominant CD8 T cell response in lymph nodes. Controller phenotype was associated with heightened IFNα production during acute SIV infection and a greater frequency of CXCR5+ CD4 T cells in blood pre-challenge despite a lower frequency of cells with the T follicular helper (Tfh) cell phenotype in blood and lymph nodes. Our results establish novel correlates of immunological control of SIV infection while reinforcing the potential importance of T cell functionality and location in SIVΔNef efficacy. Moreover, this work highlights that triggering of mucosal immunity can aid mucosal vaccine strategies rather than undermine protection. AUTHOR SUMMARY An efficacious HIV vaccine is essential to contain the HIV pandemic. Vaccine-mediated protection from HIV may be either enhanced or obstructed by mucosal immune activation; thus, the impact of adjuvants and underlying co-infections that lead to immune activation needs to be evaluated. Using the SIV macaque model, we set out to study the impact of underlying infection with HSV-2 or treatment with the adjuvant polyICLC on rectal immunization with the live attenuated vaccine SIVΔNef. We found that neither stimulus impacted complete protection from SIV; however, the combination of HSV-2 and polyICLC improved control of infection in animals that were not completely protected. Compared with non-controller macaques, controllers had less inflammatory T cells before SIV challenge as well as greater gene expression of IFNα and more functional SIV-specific T cells after infection. The results add to our understanding of the mechanisms of SIVΔNef protection and demonstrate that mucosal immune activation does not necessarily undermine protection in mucosal vaccination against HIV.


134
We evaluated the impact of two mucosal stimuli -HSV-2 and polyICLC -on rectal SIVΔNef-135 mediated protection from SIVmac239 in a cohort of 31 rhesus macaques. The macaques were 136 divided in 4 treatment groups: control (n=10), HSV-2 (n=11), polyICLC (n=4), and HSV-137 2/polyICLC (n=6) (Fig. 1A). We hypothesized that HSV-2 would exert an immune activating 138 effect, undermining SIVΔNef-mediated protection and that polyICLC would boost protection and 139 possibly buffer dysregulating effects of HSV-2. A 10-week regimen of twice-weekly mucosal 140 treatments with 10 7 pfu of HSV-2 and/or 1mg of polyICLC was employed. This regimen was 141 used to mimic physiological conditions, which may include accumulated effects of repeated 142 exposure to immunomodulatory agents such as HSV-2. In addition, we previously found that 143 repeated vaginal exposure to 10 7 pfu of HSV-2 (together with lentivirus) was associated with 144 sustained HSV-2 shedding in the macaques' vaginal mucosa (29, 30). Since we aimed to 145 maximize the effects of stimuli and predicted that frequent shedding would more strongly 146 undermine SIVΔNef-mediated protection, we followed a similar challenge protocol herein.

147
Following the treatment regimens, all macaques were exposed rectally to 1x10 3 TCID 50 148 SIVΔNef. In contrast to higher dose mucosal exposures or intravenous inoculation with 7 151 to the stimuli and vaccine but did not become infected with SIVΔNef (no vaccine take) and to 152 observe effects of the stimuli on vaccine take (15). As we were unable to exclude or fully 153 balance the inclusion of protective MHC alleles between groups (Fig. S1), we present the data 154 concerning SIV protection and pathogenesis for all animals ( Fig. 1-9) as well as for the subset 155 of animals lacking the Mamu A*01, B*08, and B*17 alleles (MHC censored data in Supporting 156 Information).

179
In contrast, mucosal stimulation by HSV-2, polyICLC, or both prior to vaccination was 180 associated with the appearance of a partial protection "controller" phenotype that was absent in 181 untreated animals ( Table 1, Fig. 1C,D,E). Partial protection with relative control of viral 182 replication ("controller") was defined as SIVmac239 challenge virus infection with rapid post-183 acute containment and maintenance of plasma viremia <10 4 RNA copies/ml (Fig. 1D,E,F). Two 184 macaques that marginally fit these criteria were considered controllers -JG86 had higher peak 185 viral load than the rest but experienced rapid decline in viremia and contained peripheral SIV 186 replication to < 30 copies/mL by 16 weeks post infection (the last time point sampled); IJ50 had 187 low peak viral load and rapid containment to less than 100 RNA copies/ml from weeks 4-8 post 188 infection but then experienced resurgence of SIV plasma viremia at weeks 12 and 16 (Fig. 1D).

189
Nonetheless, plasma SIV viral load in IJ50 was still a Log lower than in most non-controllers and 190 several Logs lower than in unvaccinated macaques at week 16 (Fig. 1F); thus, IJ50 was 191 grouped with the controllers. Controller status associated with protective MHC alleles was only 192 seen in 25-50% of unvaccinated animals. The combination of HSV-2 and polyICLC resulted in 193 the greatest frequency of controller macaques (Table 1, Fig.1C,D) and "any protection", 194 categorized as complete protection or control (Table 1, Fig. 1C). Although the controller 195 phenotype was potentially due in part to MHC haplotype, when macaques with known protective 196 Mamu alleles were censored, the controller phenotype was still most prevalent among HSV-9 197 2/polyICLC-treated macaques ( Table 1, Fig. S1, Fig. S2), indicating a role for the stimuli. The 198 impact of polyICLC alone on the controller phenotype could not be evaluated in the MHC 199 censored dataset as only one polyICLC-treated macaque was negative for all three alleles.

200
However, the controller phenotype was more prevalent within the HSV-2/polyICLC group than 201 the HSV-2 group, suggesting a role for polyICLC (Fig. S2).

202
The study was executed in two parts (Table 1, Fig. S1). The first set of macaques was   Fig. S1). This agrees with the shorter time to maturation of immunity against the 211 homologous SIVmac239 vs heterologous challenge strains (35). Thus, we grouped animals 212 from 13 and 31 weeks for further analyses.

213
Protection groupings based on plasma viral load were corroborated through 214 measurements of SIVmac239 challenge virus DNA in lymphoid and mucosal tissues at the time 215 of euthanasia, 16 weeks post-SIV challenge, assessed by quantitation of SIV Nef DNA (Fig. 2).

216
As expected, completely protected macaques had no detectable SIVmac239 Nef DNA in 217 axillary, mesenteric, or iliac lymph nodes, colorectal tissue, or gut mucosa. Non-controllers and 218 unvaccinated macaques alike had high levels of SIV DNA in all tissues. Controllers varied in 219 their level of tissue SIVmac239 Nef DNA with tissue SIV DNA viral load generally following 220 plasma SIV RNA viral load (Fig. 2, Fig. 1D,F). IJ50 in the HSV-2 group was viremic with 221 SIVmac239 Nef DNA detected at the time of necropsy, but SIVmac239 Nef DNA was only 222 detected in the mesenteric lymph node among the tissues studied. In the MHC-censored dataset, the trends were the same but only one of the controllers had SIVmac239 Nef DNA 224 detected in tissues at necropsy, the HSV-2/polyICLC-treated macaque ID90 (Fig. S3).

225
We further explored the degree of protection by examining CD4 T cell loss. Like 226 completely protected macaques, controllers preserved their peripheral CD4 T cells while non-227 controllers lost over half of them, even more than non-vaccinated animals (Fig. 3A). In jejunum 228 obtained from the SIV infected animals challenged 31 weeks post-SIVΔNef, we found that fewer 229 live cells, fewer CD4 T cells among the live cells, and a more activated CD4 T cell phenotype 230 (CD69+) tended to be present in animals with a lower degree of protection (Fig. 3B).

231
Examination of the frequency of jejunum CD4 T cells secreting IFN-γ, IL-17, and IL-22 in 232 response to mitogen stimulation revealed that controllers, but not non-controllers, also 233 preserved their functional CD4 T cells, especially the Th17, Th1Th17, and Th17Th22 subsets in 234 the gut at similar levels to the completely protected macaques (Fig. 3C,D). There was no 235 difference in the frequency of CD8 T cells secreting the same cytokines (not shown). Peripheral

236
T cell loss and gut T cell phenotype followed the same pattern in the MHC-censored dataset.

237
Notably, the Th1 preservation was driven by animals with protective MHC alleles while IL-17 238 and IL-22 secreting cells were preserved in controllers without protective MHC alleles (Fig. S4).

239
Overall, gut T cell function was somewhat reduced in the MHC-censored controllers compared 240 to the total, suggesting a more prominent role for MHC-mediated protection in preserving gut T

249
Although rectal HSV-2 infection increased SIVΔNef infection (vaccine take), it did not reduce 250 protection from wild type SIV and possibly improved protection, especially when administered 251 together with polyICLC. Thus, we investigated HSV-2 shedding and innate immune responses 252 in HSV-2 exposed macaques to appreciate the level of HSV-2 infection and understand if innate 253 responses to HSV-2 could have aided protection from SIV. We detected HSV-2 DNA in rectal 254 swabs from all but one of the HSV-2 exposed macaques (IN31 in the HSV-2/polyICLC group) 255 on at least one of the time points sampled (despite lacking samples during the vaccination 256 phase of those animals challenged 13 weeks post-SIVΔNef) (Fig. 4A). In macaques infected 257 rectally with HSV-2 in the absence of polyICLC, shedding recurred infrequently during the study 258 but over at least 25 weeks after the final HSV-2 inoculation, including in response to rectal 259 mucosa biopsy especially in SIVmac239 infected animals (Fig. 4A).

260
PolyICLC significantly inhibited HSV-2 shedding over the course of the study. Twenty-261 four hours after the 20 th HSV-2 challenge, the level of HSV-2 DNA detected in rectal swabs was 262 not significantly different between the HSV-2 and HSV-2/polyICLC groups, though it tended to 263 be less in macaques treated with polyICLC ( Fig. 4A,B). Analysis of all later time points revealed 264 that animals treated with polyICLC shed HSV-2 significantly less frequently and had less HSV-2 265 DNA in their swabs at shedding times than those not treated with polyICLC ( Fig. 4A,C). In fact, 266 only one of the HSV-2/polyICLC animals (GL42) was shedding at any of the times examined   SIVmac239 seen in HSV-2/polyICLC treatment, shedding appeared to associate inversely with 272 SIV protection status, especially in the non-polyICLC treated animals (Fig. 4A). During the 273 period following SIV challenge, the frequency of shedding trended with the severity of SIV 274 infection though too few macaques were studied to detect a significant effect (Fig. 4E).

275
Further evidence that repeated low dose rectal HSV-2 inoculation resulted in a less 276 virulent infection than high dose inoculation came by examining the impact of HSV-2 on the 277 rectal mucosa. The macaques exhibited little acute rectal inflammatory response to repeated 278 low dose HSV-2 (Fig. S5), which contrasts the response to a single high dose inoculation (17).

279
In particular, CXCL8 was not elevated in swabs taken 24 hours after the final inoculation. Low

312
SIVΔNef persists in lymph nodes of non-controllers but not controllers.

313
In past studies using intravenous immunization with either intravenous or vaginal challenge, the 314 measured levels of LAV, especially in lymph nodes, correlated with protection from virulent SIV 315 challenge (6, 8, 44). We also previously found in our rectal immunization model that polyICLC    (Fig. 6A,B). In contrast, all 324 controllers rapidly controlled SIVΔNef to <30 copies/mL plasma after a peak of replication in 325 plasma (Fig. 6A,B). Non-controllers exhibited overlapping SIVΔNef plasma viral loads to the 326 completely protected animals (Fig. 6A,B). Peak SIVΔNef viral loads were the same among the 14 327 protected, controller, and non-controller groups, diverging in the post-acute period by 8 weeks 328 post-vaccination through to the time of SIV challenge (Fig. 6C). Instead, peak SIVΔNef loads 329 stratified by treatment; HSV-2 alone and not polyICLC (or the HSV-2/polyICLC combination) 330 significantly increased SIVΔNef peak viral load (Fig. 6D). These trends were preserved in the 331 MHC-censored dataset (Fig. S6).

332
Following SIV challenge, some of the completely protected macaques (those that had 333 low to undetectable SIVΔNef viral load prior to SIV challenge) experienced an early resurgence 334 of SIVΔNef RNA in plasma (Fig. 6E). By contrast controllers, which also had undetectable   In PBMCs from the same 8 weeks post-vaccine time point, SIVΔNef DNA levels followed the 354 same pattern (Fig. 6G). These trends were also preserved in the MHC-censored dataset (Fig   355   S6).

356
Since the presence of SIVΔNef DNA may not reflect ongoing SIVΔNef replication, we

375
We assessed the phenotype of CD4 T cells (Fig. 8A) in PBMCs and LNMCs from 2 weeks 376 before SIV challenge with the goal to identify immunophenotypic patterns that distinguished 377 completely protected macaques from non-controllers and to uncover novel phenotypes 378 associated with control. Tfh cells (PD-1 high CD200 + or PD-1 high CXCR5 + CD4 T cells) within

379
LNMCs have been shown to be a haven for HIV/SIV replication including SIVΔNef and SIV in 380 elite controller macaques (8, 16). In our cohort, the frequency of cells with a Tfh phenotype in 381 blood and lymph nodes (identified as PD-1 high CD200 + CXCR5 + , Fig. 8B,C) correlated with

382
SIVΔNef plasma viral load as expected (Fig. 8D,E). A population of PD-1 low CD200 + CXCR5 + 383 cells having high CXCR5 expression was also identified in lymph nodes (Fig. 8B,F). These PD-384 1 low follicle homing cells, which have been shown to be Tfh precursors that support neutralizing 385 antibody development (45), were present at highest frequency in the completely protected 386 macaques (Fig. 8G). When we looked at all follicle-homing memory CD4 T cells 387 (CD95 + CXCR5 + ) with high CXCR5 expression (Fig. 8H), we found that these cells were more 388 frequent in the blood of controllers than the other groups even though controllers had fewer Tfh 389 cells, fewer PD-1 low Tfh precursors, and fewer CXCR5 high CD4 T cells in lymph nodes (Fig. 8I).

390
These CXCR5 high cells in blood (but not lymph nodes) displayed an increasingly CXCR3 + CCR6 + 391 Th1Th17-like phenotype in macaques with less protection (Fig. 8J,K). There was no apparent 392 difference in the frequency of either CXCR3 + CCR6or CXCR3 -CCR6 + cells between the groups 393 (not shown). Similarly within the whole CD4 T cell population, macaques with less protection 394 tended to have a greater frequency of CXCR3 + CCR6 + cells in blood (Fig. 8L,M). These

396
Altered CXCR5 + CD4 T cell frequencies suggested a potential role for antibody 397 maturation in the controller phenotype. However, when we measured the titer of SIV-specific 398 antibodies in plasma on the day of challenge, we found that antibody titer followed SIVΔNef 399 viremia ( Fig. 9A-C) and did not correlate with peak SIVmac239 viremia in SIV+ macaques ( Fig.   400   9D, Fig. S9).

401
We hypothesized that lack of control, which was associated with elevated Th1Th17 cell 402 frequency, may be due to a dysregulated or inferior T cell response to SIV even though 407 S10), we found in both the CD8 and CD4 T cell compartments that completely protected and 408 non-controller macaques possessed a similar frequency of IFNγ-producing cells while 409 controllers tended to have fewer of these cells, paralleling SIVΔNef persistence (Fig. 10A,B,   410 Fig. S10). However, non-controllers' T cells produced less TNFα than T cells from protected 411 macaques, and IFNγ contributed most to their overall gag/env-specific CD8 T cell cytokine 412 production (Fig. 10C). Similarly in the MHC censored data set, non-controllers had an IFNγ-413 dominant gag/env T cell response. In the full dataset, we noted that CD40L expression was 414 minimal on SIV-specific T cells from non-controllers compared with controllers and completely 415 protected macaques, most notably in the CD8 compartment (Fig. 10B). However, CD40L 416 effects in controllers were related to protective MHC alleles, and when the animals possessing 417 these alleles were censored, controllers and non-controllers both exhibited low CD40L on their 418 antigen-specific T cells in comparison with completely protected animals (Fig. S11). Thus

419
CD40L on CD8 T cells may contribute to the difference between controllers and completely 420 protected macaques. Low CD40L expression in SIV+ animals was not global as 421 PMA/ionomycin-stimulation increased CD40L in all animals similarly (Fig. 10D).

558
There were limitations to our study in terms of specific immune responses and host 559 characteristics that were beyond the scope and so not measured herein. We did not determine 23 560 the tetherin, APOBEC3G, or other alleles that could have promoted certain protection 561 phenotypes. We did not study gene expression over time, but only at a single time point. We did 562 not study antibody functions other than neutralization. And we did not examine SIV-specific T 563 cells in the mucosa or HSV-specific T cells at all. Any of these parameters could have 564 additionally contributed to the phenotypes we identified and should be examined in future 565 studies. In addition, we followed animals out to only 16 weeks post SIV challenge, and the 566 protection groupings were made based on the viral load data during this time period.