Divergent HIV-1-Directed Immune Responses Generated by Systemic and Mucosal Immunization with Replicating Single-Cycle Adenoviruses in Rhesus Macaques

HIV-1 infections usually start at a mucosal surface after sexual contact. Creating a barrier of protection at these mucosal sites may be a good strategy for to protect against HIV-1 infections. While HIV-1 enters at mucosa, most vaccines are not delivered here. Most are instead injected into the muscle, a site well distant and functionally different than mucosal tissues. This study tested if delivering HIV vaccines at mucosa or in the muscle makes a difference in the quality, quantity, and location of immune responses against the virus. These data suggest that there are indeed advantages to educating the immune system at mucosal sites with an HIV-1 vaccine.

ABSTRACT

Most human immunodeficiency virus type 1 (HIV-1) infections begin at mucosal surfaces. Providing a barrier of protection at these may assist in combating the earliest events in infection. Systemic immunization by intramuscular (i.m.) injection can drive mucosal immune responses, but there are data suggesting that mucosal immunization can better educate these mucosal immune responses. To test this, rhesus macaques were immunized with replicating single-cycle adenovirus (SC-Ad) vaccines expressing clade B HIV-1 gp160 by the intranasal (i.n.) and i.m. routes to compare mucosal and systemic routes of vaccination. SC-Ad vaccines generated significant circulating antibody titers against Env after a single i.m. immunization. Switching the route of second immunization with the same SC-Ad serotype allowed a significant boost in these antibody levels. When these animals were boosted with envelope protein, envelope-binding antibodies were amplified 100-fold, but qualitatively different immune responses were generated. Animals immunized by only the i.m. route had high peripheral T follicular helper (pTfh) cell counts in blood but low Tfh cell counts in lymph nodes. Conversely, animals immunized by the i.n. route had high Tfh cell counts in lymph nodes but low pTfh cell counts in the blood. Animals immunized by only the i.m. route had lower antibody-dependent cellular cytotoxicity (ADCC) antibody activity, whereas animals immunized by the mucosal i.n. route had higher ADCC antibody activity. When these Env-immunized animals were challenged rectally with simian-human immunodeficiency virus (SHIV) strain SF162P3 (SHIV_SF162P3), they all became infected. However, mucosally SC-Ad-immunized animals had lower viral loads in their gastrointestinal tracts. These data suggest that there may be benefits in educating the immune system at mucosal sites during HIV vaccination.

IMPORTANCE HIV-1 infections usually start at a mucosal surface after sexual contact. Creating a barrier of protection at these mucosal sites may be a good strategy for to protect against HIV-1 infections. While HIV-1 enters at mucosa, most vaccines are not delivered here. Most are instead injected into the muscle, a site well distant and functionally different than mucosal tissues. This study tested if delivering HIV vaccines at mucosa or in the muscle makes a difference in the quality, quantity, and location of immune responses against the virus. These data suggest that there are indeed advantages to educating the immune system at mucosal sites with an HIV-1 vaccine.

INTRODUCTION

It is estimated that as many as 90% of human immunodeficiency virus type 1 (HIV-1) infections occur by sexual transmission at mucosal surfaces (1), where only one or a few virions infect the host (2). It is hypothesized that creating a barrier to infection at these mucosae may be useful in preventing HIV infection (2). An ideal vaccine not only might generate responses at mucosae but also will likely also need to generate robust systemic immune responses to control viruses that escape the mucosal barrier.

Many studies show that systemic immunization by intramuscular (i.m.) injection can drive mucosal immune responses against HIV-1 (3–6). While this route can generate mucosal responses, there are data to suggest that mucosal immunization may better educate immune cells to target and persist at mucosal surfaces (reviewed in reference 7).

We previously compared systemic and mucosal routes of immunization with helper-dependent adenovirus (HD-Ad) vectors in rhesus macaques (8). Macaques were preimmunized with human serotype 5 Ad (Ad5). They were then vaccinated with HD-Ads expressing JRFL HIV Env gp140 by the i.m. or intravaginal (IVAG) route. Four HD-Ad serotypes (HD-Ad6, HD-Ad1, HD-Ad5, and HD-Ad2) were used for serotype switching without any recombinant protein boosts. This study showed that HD-Ad vaccination by the i.m. route generated stronger systemic T cell responses in peripheral blood mononuclear cells (PBMCs) than vaccination by the IVAG route. In contrast, IVAG vaccination generated stronger CD4 T cell central memory (Tcm) responses in colon samples than the i.m. vaccine route. Both groups repelled high-dose rectal simian-human immunodeficiency virus (SHIV) strain SF162P3 (SHIV_SF162P3) challenge better than control animals. Animals with high colon Tcm responses had lower viral set points. Importantly, more animals in the IVAG group achieved low viral set points than in the i.m. group (8).

Adenoviruses (Ad) are natural mucosal pathogens. They may therefore have utility as mucosal vaccines. Most gene-based adenovirus vaccines are replication-defective Ad (RD-Ad) vectors. These have their E1 gene deleted to prevent them from replicating and causing Ad infections. An E1 deletion Ad vaccine infects a cell, delivers its one copy of an antigen gene, and expresses “1×” of this antigen. They are safe but do not replicate transgenes or their expression.

In contrast, an E1⁺ replication-competent Ad (RC-Ad) vaccine can infect the same cell and replicate the same antigen gene DNA 10,000-fold (9–20). This amplifies antigen production and immune responses to significantly higher levels than for E1 deletion vectors (9–20). While RC-Ad vaccines are more potent, fully replication-competent Ads run the real risk of causing adenovirus infections in humans. Indeed, when live RC-Ad vaccines were used in military recruits, they were encapsulated and given orally primarily to prevent them from causing Ad respiratory infections in nurses and vaccinees (21).

To take advantage of transgene DNA replication but avoid the risk of adenovirus infections, we developed single-cycle Ad (SC-Ad) vectors (22–25). SC-Ads retain their E1 genes to allow the virus to replicate its genome but have deletions of their pIIIa gene to block the production of infectious progeny viruses. SC-Ads replicate their genomes and transgenes as well as RC-Ad (up to 10,000-fold) (22). RC- and SC-Ad produce more transgene protein than RD-Ad vectors (22). SC-Ads generate more robust and more persistent immune responses than either RD-Ad or RC-Ads (23). In head-to-head comparisons, SC-Ad produces significantly higher antibody levels and better protection against influenza virus (26). SC-Ads have also shown potency as vaccines against Ebola virus and against Clostridium difficile after single immunization (24; W. E. Matchett, S. S. Anguiano-Zarate, M. A. Barry, submitted for publication).

Given that HIV-1 generally enters the body at mucosal surfaces, we hypothesized that vaccination at mucosal sites might mediate better protection against this type of infection. We hypothesized that mucosal and i.m. vaccination would generate quantitative and qualitative differences in humoral and cellular immune responses in systemic and mucosal immune responses. To test this, rhesus macaques were immunized by a single systemic i.m. immunization or by a single mucosal intranasal (i.n.) immunization with SC-Ads expressing clade B envelope sequences. The animals were then boosted by the same or alternative routes with SC-Ad followed by protein boosts. We describe here how these various SC-Ad immunization strategies affected the generation of HIV binding, antibody-dependent cellular cytotoxicity (ADCC), and neutralizing antibodies as well as their effects on cellular immune responses, including peripheral T follicular helper (pTfh) cells in blood and lymph nodes. We provide data on a pilot SHIV challenge of these animals.

RESULTS

SC-Ad expressing HIV-1 gp160.

It is thought that the use of envelopes that occur early in HIV-1 infections may form structures that favor the production of protective antibodies (27). Given this, we utilized clade B envelope sequences that were isolated before and immediately preceding a peak in the expansion of antibody neutralization breadth (G4 and F8 gp160 Envs, respectively) from HIV patient VC10014 (27). These gp160 sequences expressed under the control of the strong cytomegalovirus promoter were inserted into two single-cycle Ad serotypes: SC-Ad6 and SC-Ad657 (see Fig. S1A in the supplemental material) (28). When used to infect A549 cells, both vectors produced gp160 as determined by Western blotting (data not shown).

Mucosal IVAG immunization mediated better protection than i.m. immunization in our previous HD-Ad studies (8). While the IVAG route was better, this method would be difficult to implement in humans. Testing of SC-Ad-G4 in small animals revealed that IVAG priming generated weak antibody levels, whereas i.n. priming generated strong responses (29). Given this, the i.n. route was used in this study as the mucosal immunization route.

Single mucosal and systemic immunization in rhesus macaques.

A total of 2 × 10¹⁰ virus particles (VP) of SC-Ad6-G4 Env was used to vaccinate groups of 8 female rhesus macaques by single i.m. or i.n. immunization (Fig. 1). This dose is relatively low, approximately 7.5-fold lower than for recently used RC-Ad HIV envelope vaccines delivered by mixed i.n. and i.m. immunization (30). A negative-control vector group was immunized i.n. with SC-Ad6 expressing Ebola virus glycoprotein (gp). Four weeks after single immunization, plasma samples showed significantly higher midpoint enzyme-linked immunosorbent assay (ELISA) binding titers against F8 gp140 only in the i.m. group (P < 0.0001 by one-way analysis of variance [ANOVA] [Fig. 1A]).

FIG 1.

Plasma HIV Env binding titers. Immunizations with different SC-Ads and gp140 proteins are shown above the graph with large arrows. Midpoint F8 gp140 binding titers by ELISA are shown for each animal before and after each immunization. The dotted line indicates the minimal detection limit for antibodies in this assay. Symbols are scattered in the x direction at each time point to allow individual measurements to be observed. SC-Ad6-Ebov is a negative-control Ad vaccine. This group of animals was not boosted with gp140 protein or adjuvant. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (one-way ANOVA after log transformation of the data in comparison to the SC-Ad6-Ebolavirus [Ebov] group).

i.m. versus i.n. boost with SC-Ad6 at week 4.

H. C. Ertl’s group reported that anti-adenovirus neutralizing antibodies produced by i.m. immunization can be avoided by boosting by a different route (31). To test this and enable the reuse of the same Ad serotype in macaques, we divided each SC-Ad6-primed group into 2 groups of 4. These were each boosted with the same SC-Ad6 serotype expressing the alternate F8 Env by either the i.m. or the i.n. route. Midpoint binding titers were elevated in the animals that were prime-boosted by the i.m.-i.m., i.m.-i.n., and i.n.-i.m. routes (P < 0.01 by ANOVA) in samples collected 3 weeks after this boost. No binding antibodies were observed in the i.n.-i.n. group (Fig. 1A). When SF162 viral neutralization activity was tested from these samples, the i.m.-i.m. group showed a mean dilution of plasma resulting in 50% neutralization of the virus (ID₅₀) of 292, significantly different from that of the SC-Ebola virus group (P < 0.05 by ANOVA [data not shown]).

SC-Ad657 boost at week 13.

The animals were then boosted by serotype switching with SC-Ad657 expressing G4 Env at week 13. The same route was used as in the previous boost. Week 15 titers showed that the i.m.-i.m.-i.m., i.m.-i.n.-i.n., and i.n.-i.m.-i.m. groups all had significant antibodies relative to the Ebola virus groups, but the highest levels were in the i.n.-i.m.-i.m. group (P < 0.001 to 0.0001 [Fig. 1A]). The i.n.-i.n.-i.n. group again showed no Env antibodies even after 3 immunizations.

Recombinant trimeric Env protein boost at week 24.

Most HIV vaccine studies augment Ad immunizations with protein boosts to amplify antibody responses. For example, in a recent study by Barouch’s group, RD-Ad26 vectors were used twice and boosted three times with adjuvanted gp140 protein (32). Given this, we boosted the SC-Ad-Env groups with 50 μg of recombinant F8 trimeric gp140 protein mixed with Adjuplex adjuvant by the i.m. route (27). The SC-Ad-Ebola virus control group did not receive protein or adjuvant. The protein boosted midpoint binding titers by 2 orders of magnitude in all Env groups (P < 0.0001 [Fig. 1A]). Interestingly, this protein immunization also boosted antibodies in the i.n.-i.n.-i.n. group to levels that were comparable to those in the other groups even though Env binding antibodies were not detected earlier.

Binding and neutralizing antibodies in plasma after a second protein boost.

The animals were boosted with protein a second time at week 38. This increased F8 binding plasma antibody titers to nearly 10⁵ by week 40 (Fig. 1). Neutralizing antibody (NAb) titers against tier 1A SF162 virus increased to 100 to 10,000 at week 40 and were significantly higher than in controls (P < 0.01 to 0.001) (Fig. 2). NAbs against tier 1B virus SS1196 and tier 2 virus JRCSF increased to 100 in most animals, with the exception of two animals in the i.n.-i.n.-i.n. group whose titers were at background levels (Fig. 2). Animals that received mixed i.n. and i.m. immunizations had significantly higher SS1196 neutralizing antibodies than controls (P < 0.05).

FIG 2.

Plasma HIV neutralization titers. Neutralization of the indicated viruses was performed using the TZM-bl neutralization assay. All values were calculated compared to those for virus-only wells. Each dot represents the mean value for each animal. *, P < 0.05; ***, P < 0.001 (one-way ANOVA after log transformation of the data in comparison to those for the SC-Ad6-Ebov group).

ADCC activity after the second protein boost.

Antibody-dependent cellular cytotoxicity (ADCC) activity in week 40 plasma was tested against SHIV_SF162P3-infected cells. ADCC activity was generally higher in animals that had at least one i.n. mucosal SC-Ad immunization (Fig. 3). All animals that received a mucosal immunization had significantly higher maximum percent ADCC than SC-Ad-Ebola virus control animals (P < 0.05, 0.0001, and 0.0001 for i.m.-i.n.-i.n., i.n.-i.m.-i.m., and i.n.-i.n.-i.n., respectively). When compared by 50% ADCC titers, only the i.n. SC-Ad-primed groups were found to have significantly higher ADCC activity than controls (P < 0.05 and 0.001 by ANOVA for the i.n.-i.m.-i.m. and i.n.-i.n.-i.n. groups).

FIG 3.

Plasma ADCC activity. Plasma samples were tested with CD16-KHYG-1 effector cells to kill CEM.NKR.CCR5.CD4⁺-Luc target cells infected with SHIV_SF162P3. Each dot represents the mean value for each animal. Fifty percent ADCC indicates the plasma titers or antibody concentrations required for half-maximal cell lysis (similar to IC₅₀ for neutralization). *, P < 0.05; ***, P < 0.001; ****, P < 0.0001 (one-way ANOVA versus the values for the SC-Ad6-Ebov group).

Antibody responses in saliva and vaginal samples after a second protein boost.

Week 40 saliva and vaginal samples had F8 and SF162 Env binding IgG antibodies as determined by ELISA (Fig. 4). Antibodies were observed at these mucosal sites in most groups, with the exception of the SC-Ad-Ebola virus control group (Fig. 4 and data not shown). Interestingly, there was a distance effect on these mucosal antibodies. All i.m.-i.n.-i.n. and i.n.-i.n.-i.n. immunized animals had Env-binding antibodies in their saliva near the site of immunization. However, only one of these animals had mucosal antibodies at the more distant vaginal site (Fig. 4). When ADCC activity was measured in these mucosal samples, these responses were found to be highly variable (Fig. 5). Only the i.n.-i.n.-i.n. group had significant ADCC activity in saliva compared to that of control animals (P < 0.05 by ANOVA).

FIG 4.

Saliva and vaginal HIV Env binding titrations. ELISA optical densities at 450 nm (OD₄₅₀) are shown for the indicated samples at the indicated dilutions when tested against F8. The low level of antibodies in these mucosal samples prevented reaching saturation of the assay. For this reason, 50% effective concentrations (EC₅₀) could not be reliably calculated for most animals. Rhesus macaque Rh13-091 in the i.n.-i.m.-i.m. group was the only animal for which an EC₅₀ could be calculated (EC₅₀ = 4,580). Similar results were obtained in ELISAs using SF162 gp140.

FIG 5.

Mucosal ADCC activity. Vaginal swabs and saliva samples were tested with CD16-KHYG-1 effector cells to kill CEM.NKR.CCR5.CD4⁺-Luc target cells infected with SHIV_SF162P3. Each dot represents the mean value for each animal. *, P < 0.05 (one-way ANOVA versus the values for the SC-Ad6-Ebov group).

Systemic cellular immune responses after one protein boost.

Week 38 PBMCs collected just prior to a second F8 Env protein boost were assayed for T cells against Env peptides and adenovirus. All Env-immunized animals had Env-specific gamma interferon (IFN-γ)-secreting cells in their PBMCs (Fig. 6A). The levels of Env-specific IFN-γ spot-forming cells (SFCs) were generally increased in animals that received at least one mucosal immunization. However, IFN-γ SFCs were found to be significantly higher in the i.n.-i.n.-i.n. SC-Ad group only when they were compared to SC-Ad Ebola virus-immunized control animals (P < 0.05 by ANOVA). Anti-Ad SFCs were relatively low in all groups compared to anti-Env SFCs at this time point. However, these T cells were stimulated by infectious adenovirus and not with Ad peptides, so this may underestimate anti-Ad responses.

FIG 6.

IFN-γ-secreting cells from PBMCs and lymph nodes. PBMCs and lymph node cells were analyzed by ELISPOT assay by staining for IFN-γ. “Anti-Env” indicates cells that were stimulated with conserved HIV Env peptides and SC-Ads. The total number of spot-forming cells (SFCs) in each of the stimulated wells was counted and adjusted to control medium as background. Each dot represents the mean value for each animal. *, P < 0.05 (one-way ANOVA).

Systemic cellular immune responses after a second protein boost.

At week 40, PBMCs and inguinal lymph node cells were assayed for Env-specific IFN-γ SFCs by enzyme-linked immunosorbent spot (ELISPOT) assay (Fig. 6B). This protein boost increased Env-specific SFCs in PBMCs and in lymph nodes to similar levels in all of the Env-immunized animals.

Mucosal cellular trafficking.

Flow cytometry was performed on cells from week 40 rectal biopsy specimens without any antigen stimulation. These data showed similar numbers of α4β7 CD4 and CD8 cells at rectal sites (Fig. 7A). Cells were counterstained for CD69 to identify activated T cells that might be at risk of SHIV infection. The numbers of activated CD69⁺ CD4⁺ cells in rectal tissues were similar between the groups (Fig. 7B). Cells were stained for FoxP3 to identify regulatory T (Treg) cells that might downregulate protective immune responses. In this case, FoxP3⁺ CD4⁺ cells at this mucosal site were not appreciably different between the different vaccination route groups (Fig. 7B).

FIG 7.

Mucosal T cell trafficking and activation. T cells were harvested from rectal biopsy specimens collected after the second protein boost and analyzed by flow cytometry for CD4, CD8, α4β7 integrin, CD69, and FoxP3. Each dot represents the mean value for each animal.

HIV envelope-specific Tfh cell distributions.

Vaccine routes also impacted the distribution of T follicular helper (Tfh) cells in the animals. Tfh cells are uniquely specialized in providing help for B cell selection and maturation in the germinal centers (GCs) by locally supplying needed factors such as the cytokines interleukin 21 (IL-21) and IL-4 (33, 34). Peripheral Tfh (pTfh) cells can also be detected in the blood, and higher numbers of these cells may be associated with better responses in the RV144 trial (34). Vaccine testing in macaques has shown that animals with the highest NAb titers also have the highest Tfh responses in lymph nodes (35). Pauthner et al. reported that the best tier 2 NAb responses correlated with strong Tfh and B cell responses in draining lymph nodes after subcutaneous protein immunizations (36). Conversely, it may be that observation of pTfh cells in the blood may be a sign of poor Tfh trafficking to lymph nodes and subsequent GC responses.

Given the role of Tfh cells, week 40 PBMCs and lymph node cells were stimulated with recombinant gp140 envelope protein and CXCR5⁺ IL-21⁺ CD4⁺ Tfh cells were measured by flow cytometry (Fig. 8). Animals that were immunized with Ad and protein by only the i.m. route had significantly higher peripheral Tfh cells in PBMCs than other groups (Fig. 8). In lymph nodes, Tfh cells were lowest in the control and i.m.-only groups. In contrast, approximately half of the animals that received at least one i.n. mucosal immunization have detectable Tfh in their lymph nodes after the last protein boost (Fig. 8).

FIG 8.

Tfh cell response in the blood and in lymph nodes. PBMCs and lymph node cells collected at week 40 were stimulated with HIV-1 Env protein and then examined for coexpression of CD3, CD4, CXCR5, and IL-21. Each dot represents the mean value for each animal. *, P < 0.05 (one-way ANOVA).

Rectal challenge with SHIV_SF162P3.

This was an immunogenicity study to evaluate immune responses generated by varied vaccine routes, rather than a challenge study. Dividing the SC-Ad-primed groups into different boost groups in the immunogenicity study reduced each group size to 4, which is underpowered for robust examination in a SHIV challenge study. In addition, only HIV Env was used as an immunogen to test the effects of vaccine route on antibody responses. No SIV Gag, Pol, or other antigens were delivered. While there were these caveats, the immunized macaques were challenged rectally with SHIV_SF162P3 to make the most use of the animals and as a pilot study for future challenge studies.

Four unimmunized control animals were added to the study, and each group was challenged weekly by rectal inoculation with 1 ml of a 1:300 dilution of SHIV_SF162P3 challenge stock provided by NIH. The rectal route was chosen as a mucosal challenge route based on guidance that it might provide more reliable infection than the vaginal route with SHIV_SF162P3. This challenge equaled 4.3 50% tissue culture infective doses (TCID₅₀) on rhesus PBMCs and 137 TCID₅₀ on TZM-bl cells.

After the first challenge, 2 animals in the unimmunized control group and 2 animals in the i.m.-i.m.-i.m. group became infected (Fig. 9). One animal in each of the mixed-route groups (i.m.-i.n.-i.n. and i.n.-i.m.-i.m.) became infected after one challenge. None of the animals in the i.n.-i.n.-i.n. group were infected after the first challenge. Grouping the animals by i.m. or i.n. priming did not reveal significant differences in infection (Fig. 10A). Viral loads in plasma indicated that all animals except the Ebola virus group reached high viral loads after 3 challenges (Fig. 10B). Animals in the i.n.-i.n.-i.n. group had a delay in reaching these high viral loads. As challenges continued, animals in all groups became infected, with the exception of one animal in the Ebola virus group that remained uninfected after 7 challenges. Trim5α and major histocompatibility complex (MHC) alleles were examined retrospectively (Table 1). This analysis did not reveal overtly protective genes in the resistant Ebola virus group animal. Most animals could not be classified with alleles that might keep them moderately protected, but most groups had at least one animal with a higher likelihood of protection by virtue of these alleles. It should be noted that 2/4 animals in the i.m.-i.m.-i.m. and i.n.-i.n.-i.n. groups had Trim5α and MHC alleles that might predict a higher likelihood of innate protection against SIV_smm and perhaps SHIV (Table 1). While these may modulate interpretation of the results, recent studies suggest that protective Trim5 alleles may not affect repeated rectal challenges with SHIV_SF162P3 (37).

FIG 9.

Protection against repeated rectal SHIV_SF162P3 challenge. The indicated groups were challenged rectally with 4.3 TCID₅₀ (on rhesus PBMCs) of SHIV_SF162P3 on a weekly basis. Plasma samples were analyzed for SHIV RNA copies. Animals with RNA copies above 10 were considered infected, and the numbers of challenges required to infect that animal were used as events for Kaplan-Meier analysis.

FIG 10.

SHIV_SF162P3 acquisition and viral loads. (A) Animals from Fig. 8 were grouped by their initial SC-Ad priming route (i.m. or i.n.), yielding groups of 8, and Kaplan-Meier analysis was performed. (B) Plasma SHIV_SF162P3 RNA levels over the course of the challenge study.

TABLE 1.

Retrospective screening for SIV protective gene alleles

Vaccine group	Animal no.	MHC typing	Trim5α alleles	Degree of viral protection
Unimmunized	RHJ663	Not done	CypA/TFP	High
	RH3-39	Not done	Q/TFP	Moderate
	RHJ403	Not done	CypA/Q	Moderate
	RHJ791	Not done	Q/TFP	Moderate
SC-Ad-Ebov	RH13-005	A11, B01, B17	Q/TFP	Moderate
	RH13-007	A08, A11, B01, B17	Q/TFP	Moderate
	RH13-043	A08, A11, B17	Q/TFP	Moderate
	RH13-135a	A08, A11, B17	Q/TFP	Moderate
i.m.-i.m.-i.m.	RH13-027	A11, B01, B17	TFP/TFP	High
	RH13-031	A08, A11, B01	CypA/Q	Moderate
	RH13-051	A08, A11, B17	CypA/Q	Moderate
	RH13-139	A08, A11, B17	TFP/TFP	High
i.m.-i.n.-i.n.	RH13-039	A11, B01, B17	CypA/Q	Moderate
	RH13-045	A08, A11, B01, B17	Q/Q	Susceptible
	RH13-095	A08, A11, B17	Q/TFP	Moderate
	RH13-159	A08, A11	CypA/TFP	High
i.n.-i.m.-i.m.	RH13-013	A11, B17	CypA/Q	Moderate
	RH13-067	A08, A11, B01, B17	Q/TFP	Moderate
	RH13-091	A11, B01, B17	TFP/TFP	High
	RH13-121	A08, A11, B17	Q/TFP	Moderate
i.n.-i.n.-i.n.	RH13-025	A11, B17	CypA/Q	Moderate
	RH13-033	A08, A11, B17	CypA/TFP	High
	RH13-087	A08, A11, B01, B17	TFP/TFP	High
	RH13-125	A08, A11, B17	Q/TFP	Moderate

^aRH13-135 resisted all SHIV challenges.

Postmortem viral loads in tissues.

This pilot challenge study was terminated 9 weeks after first challenge. PBMCs and gut tissues were isolated, and RNA was purified and evaluated for SHIV genomes (Fig. 11). Postmortem samples had varied levels of SHIV RNA, with lower levels in the i.n.-i.n.-i.n. group in the colon. When the SC-Ad-Ebola virus animal that resisted infection was included in analysis, no groups reach statistical significance. If this resistant animal’s RNA level was censored, the i.n.-i.n.-i.n.-2IM PRO group reached a P value of <0.05 by one-way ANOVA compared to the SC-Ad-Ebola virus group.

FIG 11.

SHIV loads in tissues. RNA from PBMCs and postmortem tissues were collected and qPCR was performed to detect SHIV RNA.

DISCUSSION

This immunogenicity study examined how different routes of gene-based vaccination influence the production of systemic and mucosal immune responses against HIV-1 using new replicating single-cycle Ad (SC-Ad) vaccines (22–25). Most Ads in clinical vaccine testing are E1 deletion replication-defective Ads (RD-Ads). Our new SC-Ads retain their E1 genes to allow them to replicate antigen transgenes up to 10,000-fold like a replication-competent Ad (RC-Ad) (22–25) but do not produce infectious viruses.

We immunized rhesus macaques with relatively low doses of SC-Ad expressing clade B HIV-1 gp160 by the i.n. and i.m. routes to compare mucosal and systemic routes of vaccination. SC-Ad by itself generated significant antibodies against Env after a single immunization by the i.m. route. Varying the route of second immunization allowed the same Ad6 serotype of the vaccine to be used for priming and boosting in the animals, consistent with observations in mice (31).

While i.n. primed animals initially lagged in Env antibody production compared to i.m. primed animals, changing the boost to the i.m. route pushed antibody levels highest in the i.n.-i.m.-i.m. group. In contrast, animals immunized with SC-Ad by only the i.n. route generated no detectable Env-binding antibodies.

SC-Ads by themselves generated significant antibody responses at relatively low doses. However, these responses were stunted by the first protein boost with gp140, which amplified Env binding antibody 100-fold in all of the animals. This dramatic boost effect was also observed in the refractory i.n.-i.n.-i.n. SC-Ad group. This demonstrated that these nasally immunized animals had indeed been primed at their mucosa and that these mucosally educated B cells responded to systemic gp140 protein nearly as strongly as all of the other groups. The vaccine routes and combinations generated similar levels of tier 1 and 2 virus neutralization, with weaker cross-reactivity in the i.n.-i.n.-i.n. group. Strongest plasma ADCC activities were observed in i.n. primed animals. The weakest ADCC activity was in animals that were immunized by only the conventional i.m. route. This suggests that the route of immunization does indeed have an impact on the quality of immune responses generated against HIV.

Mucosal antibody levels were evaluated in the saliva and vaginal swabs from the animals after the last protein boost. It was interesting that Env-binding antibodies were higher in saliva, near the site of i.n. vaccination, and lower in the distant vaginal site. In contrast, i.m. immunization generated similar antibody binding responses at both mucosal sites. This runs counter to the concept that mucosal immunization at any site mediates efficient crossover of immune responses at other mucosal sites. It should be noted that these ELISA binding titers measured IgG antibodies. Therefore, some of the better balance in responses in the i.m. groups may be due to transudation of IgG from the systemic compartment. IgA responses were not measured due to the limited amounts of samples.

Low-level ADCC activity was observed in vaginal samples and saliva but was significant only in the saliva of the i.n.-i.n.-i.n. SC-Ad group. This again indicates that there is value in mucosal immunization but again suggests that a distance effect limits delivery of effector antibodies to distant mucosal sites.

The vaccine route also impacted the distribution of T follicular helper (Tfh) cells in the animals. Animals that were immunized by only the i.m. route had high pTfh cell counts in the blood but low Tfh cell counts in lymph nodes. Conversely, some of the animals immunized by the i.n. route had high Tfh cell counts in lymph nodes, but all had low Tfh cell counts in the blood. This suggests that systemic immunization by the i.m. route may distract Tfh cells away from germinal centers, where they are needed to form productive germinal centers.

Data from the STEP and HVTN505 HIV vaccine trials in humans suggested that Ad vaccines might have facilitated increased virus acquisition rather than protection (38, 39). Subsequent data indicate that this side effect is not unique to Ad vaccines. Similar effects are observed after vaccination with many vaccines, including virus-like particles (VLPs) (40), attenuated varicella-zoster virus (VZV) (41), and live attenuated simian immunodeficiency virus (SIV) (42).

While mucosal immunization may theoretically provide a better barrier of protection at the mucosal site of entry, it may also increase the number of activated CD4 cells at the same site to increase the likelihood of vaccine-mediated HIV acquisition. To probe the effects of vaccine route on these types of cells, rectal tissues from the vaccinated macaques were examined for increases in mucosal trafficking α4β7 cells and activated CD4 T cells. Activated CD69⁺ cells were similar in all of the animals regardless of vaccine route. α4β7 cells were similar in all of the animals.

These measurements were made from rectal samples 2 weeks after a second protein boost, a time when immune cells are presumably in a highly activated state. Exposure to HIV at such a time point might increase the risk of virus acquisition. While this concern is now routinely considered, this acquisition side effect may actually be highly time dependent, as has been suggested in SIV vaccine studies using live attenuated Rev-Ind Nef SIV vaccine (42). In this study, macaques were immunized with attenuated SIV vaccine at varied times before exposure to repeated low-dose rectal SIV_smE660 challenges. Animals that were vaccinated 2 weeks before challenge became superinfected faster than unimmunized controls. In contrast, animals that were immunized up to 15 months before the same challenge were protected by the same vaccine. Thus, the same vaccine increased acquisition if SIV exposure occurred too soon after vaccination or protected if exposure was delayed longer after vaccination.

Although this study was originally intended to only test immunogenicity against HIV Env immunogens, the animals were also challenged rectally with SHIV_SF162P3 to maximize the use of this precious resource. All of the animals were ultimately infected and protection was unimpressive. The absence of any SIV antigens in the vaccines likely limited protection in this pilot challenge study.

While overall protection was not observed, there were notable reductions in SHIV RNA in gastrointestinal tissues in animals that were vaccinated with SC-Ad by only the i.n. route. Considering that the gut is ravaged by HIV early infection, this suggests that mucosal immunization strategies may provide favorable responses in at-risk tissues.

We previously compared i.m. versus intravaginal (IVAG) immunizations with HD-Ad vectors expressing only HIV Env (8). The IVAG route was tested since this was regionally close to the site of later rectal challenge. Under these conditions, IVAG immunization mediated better protection against SHIV_SF162P3 than the traditional i.m. route (8). While the IVAG route was better, this route seemed impractical to translate to humans for large-scale vaccination against HIV-1. We therefore switched to the more practical i.n. route for this study based on the ability to immunize at one mucosal site to drive immune responses at distant mucosal sites (reviewed in reference 7). While the i.n. route did drive favorable responses, it is concerning that mucosal antibody responses remained local and did not spread to more distant vaginal sites. It is possible that vaccinating at the vaginal or rectal site may mitigate this distance effect, but it is unclear how these routes could be implemented as a general vaccine strategy. Efforts to target vaccines or antigens to rectal and vaginal mucosal sites may be an alternate strategy to overcome this potential distance effect.

Finally, these data suggest that replicating SC-Ad vectors may be a robust and safe platform for vaccination against HIV-1 and other infectious diseases. We believe that HIV vaccines should be transitioned to vaccine platforms that amplify HIV antigen genes. For Ads, this could be either SC-Ad or RC-Ad vectors. Whether the improved safety of SC-Ad vaccines has merit over RC-Ad will need to be tested in humans since most animal models underreport adenovirus replication and amplification.

MATERIALS AND METHODS

Single-cycle adenovirus expressing HIV-1 envelope.

Clade B gp160 envelope sequences that arose before (G4) and immediately preceding (F8) a peak in the expansion of antibody neutralization breadth from HIV patient VC10014 (27) were used as immunogens. Motif-optimized G4 and F8 gp160 sequences were recombined into SC-Ads based on human Ad serotypes 6 and 657 (22–24, 28). An SC-Ad6 expressing Ebola virus glycoprotein was used as a negative-control vaccine. Viruses were rescued and purified as described previously (22–24).

Envelope proteins.

The envelope gene F8 cloned from an HIV clade B-infected subject used in the SC-Ad vector was motif optimized and modified by site-directed mutagenesis to express uncleaved, trimeric gp140. Details of expression, purification, and antigenic characterization have been described previously (27).

Animals.

Female adult rhesus macaques (Macaca mulatta) of Indian origin were maintained at the Michael Keeling Center for Comparative Medicine and Research at the University of Texas M.D. Anderson Cancer Center, Bastrop, TX, in the specific-pathogen-free breeding colony. All animal handling was carried out in accordance with the policies and procedures of the Mayo Clinic and the University of Texas M.D. Anderson Cancer Center, the provisions of the Animal Welfare Act, PHS Animal Welfare Policy, and the principles of the Guide for the Care and Use of Laboratory Animals (43). All the surgical and blood collection was performed when the animals were anesthetized using standard anesthetic and aseptic techniques. Meloxicam (0.2 mg/kg of body weight on the first day and then 0.1 mg/kg on days 2 and 3 postoperatively) was used to prevent pain after lymph node biopsies.

Immunizations.

Macaques were anesthetized by the intranasal (i.n.) or intramuscular (i.m.) route with 2 × 10¹⁰ virus particles (VP) of the indicated SC-Ad vaccine. Animals were boosted with 50 μg of purified, trimeric recombinant F8 gp140 combined with Adjuplex adjuvant by the i.m. route as described previously (27, 35). i.m. immunization was performed into the quadriceps of the animals. For i.n. immunization, anesthetized animals were placed in either right or left lateral recumbency depending on the side to be inoculated. A 250-μl inoculum was introduced in the nostril slowly, over 1 min, into the both nasal cavities. Following infusion, the animal was observed at 1 and 24 h after immunization.

Sample collections.

Samples were collected as described previously (44). Briefly, peripheral venous blood samples were collected in EDTA. Before isolation of peripheral blood mononuclear cells (PBMCs), plasma was separated and stored immediately at −80°C. PBMCs were prepared from the blood on Ficoll-Hypaque density gradients. Saliva and vaginal swabs collected with Wek-Cel spears in 1 ml of phosphate-buffered saline (PBS) containing protease inhibitors and vortexed, and supernatants were collected after centrifugation at 2,000 rpm as described previously (45). Samples were kept frozen at −80°C until further use. Prior to performing rectal biopsies, feces was manually removed from the rectum if needed. A large-bore otoscope was use to visualize the site of biopsy in the rectum. Then a flexible endoscopic forcep biopsy tool was inserted through the otoscope to obtain a small (∼1- to 2-mm) tissue sample. The procedure was repeated, spacing the sampling sites apart from each other with visual confirmation. Approximately 10 to 15 biopsy samples were collected and immediately placed in PBS. The samples were rinsed with PBS and were digested with collagenase and DNase for 30 min at 37°C in a rotary shaker. Mononuclear cells were isolated from these digests by separation on Percoll gradients (35 and 60%) and were used for immune assays. For lymph node biopsies, nodes were identified in either the axillary or inguinal area by palpation. After removing the hair and aseptically preparing the site, a small incision was made over the lymph node using a scalpel. The node was removed with scissors, and hemostasis was achieved with manual pressure. The skin and subcutaneous layers were closed with absorbable monofilament sutures. After removal, the lymph node was immediately placed in PBS. Cells were isolated from lymph nodes by homogenization and passage through 40-μm filter and then used for immune assays.

ELISPOT assay for detecting antigen-specific IFN-γ-producing cells.

Freshly isolated PBMCs were stimulated with either F8 gp140 protein (1 μg/ml) or heat-inactivated Ad6 (7.0 × 10⁸ VP/well) to determine the numbers of IFN-γ-producing cells by the enzyme-linked immunosorbent spot (ELISPOT) assay using the methodology reported earlier (46–48). Briefly, aliquots of PBMCs (10⁵/well) were seeded in duplicate wells of 96-well plates (polyvinylidene difluoride-backed plates, MAIP S 45; Millipore, Bedford, MA) precoated with the primary IFN-γ antibody, and the lymphocytes were stimulated with either concanavalin A (ConA), F8 gp140 protein, or heat-inactivated Ad6. After incubation for 30 to 36 h at 37°C, the cells were removed and the wells were thoroughly washed with PBS and developed as per the protocol provided by the manufacturer. Results are expressed as IFN-γ spot-forming cells (SFCs) per 10⁵ PBMCs after subtraction of the duplicate wells with medium only (negative control) and are considered positive if greater than twice the background and greater than 5 SFCs/10⁵ PBMCs.

Antibody ELISAs.

Binding IgG antibody responses in macaque plasma and secretions were measured by kinetic ELISAs against F8 gp140 or SF162 gp140 as described previously (27, 35). Briefly, half-well assay plates (Corning) were coated with 50 ng of recombinant gp140 in 0.2 M sodium carbonate-bicarbonate buffer, pH 9.4, overnight at 4°C. Plates were washed in dilution buffer 1× PBS plus 0.1% Triton X-100, followed by a blocking step with 150 μl per well of 1× PBS, 1% normal goat serum, and 5% bovine serum albumin (BSA) for 1 h at room temperature (RT). Plasma samples were heat inactivated at 56° for 30 min. Serially diluted samples were transferred to the assay plate and incubated for 1 h at room temperature. Plates were washed 3 times and bound antibodies were detected with peroxidase AffiniPure goat anti-human IgG, Fcγ fragment specific (Jackson ImmunoResearch), diluted to 1:5,000, and incubated for 1 h at RT. After a washing, 50 μl per well of tetramethylbenzidene (TMB) one-component substrate (Southern Biotech) was added and incubated for 10 min. The reaction was stopped with the addition of 50 μl per well of 1 N sulfuric acid. The plates were read at 2 wavelengths, 450 nm and 650 nm, in a SpectraMax190 plate reader (Molecular Devices, Sunnyvale, CA). Assays were standardized with positive- and negative-control monoclonal antibodies and naive plasma, which were included with each assay. The 50% titers of Abs were determined by measuring the dilutions of samples required for 50% maximal binding by nonlinear regression in Prism v7.0.

Neutralization assay.

HIV neutralization was performed using the TZM-bl neutralization assay as described previously (27, 35). All values were calculated compared to those for virus-only wells.

ADCC.

CEM.NKR.CCR5.CD4⁺-Luc target cells were infected with 50 ng of SHIV_SF162P3 and cultured for 4 days as described in reference 49. Two-fold serial dilutions of each sample were added to the infected targets for 20 min at room temperature. CD16-KHYG-1 effector cells were added at a 10:1 effector-to-target ratio and incubated for additional 8 h. The cells were lysed and luciferase activity was measured on a Bio-Tek plate reader. Fifty percent antibody-dependent cellular cytotoxicity (ADCC) indicates the plasma titers or antibody concentrations required for half-maximal cell lysis (similar to 50% inhibitory concentration [IC₅₀] for neutralization).

Flow cytometry.

Cells collected from rectal and lymph node biopsies were incubated overnight with 0.2 μg of gp140 or media alone in the presence of Golgi Plug for the last 4 h. After culture, cells were harvested and incubated on ice for 45 min with a panel of human antibodies that cross-react with rhesus macaque samples. For Tfh cell analyses, cells were stimulated with recombinant gp140. For all other flow cytometry assays, the cells were not stimulated with antigen. The panels included the following fluorochrome-labeled antibodies: CD8 (Qdot655), α4β7 (phycoerythrin [PE]), and CXCR5 (PE), all obtained from the Nonhuman Primate Reagent Resource; CD69 (BV737, clone FN50) and FoxP3 (PECy5, clone PCH101), obtained from eBioscience; IL-21 (BV421, clone 3A3-N2.1), CD45 (BV786, D058-1283), and CD3 (clone SP34-2, PE-Cy7 labeled), all from BD Bioscience (San Jose, CA); and CD4 (Pacific Blue, clone OKT4), from Thermo Fisher Scientific (Waltham, MA). Dilutions for antibodies were determined by following the manufacturers’ recommendations. Dead cells were excluded by using the LIVE-DEAD fixable dead cell stain kit, obtained from Invitrogen (Carlsbad, CA). Subsequently, the cells were washed twice with PBS containing 2% fetal bovine serum (FBS) and 2 mM EDTA and then fixed and permeabilized with FoxP3 Fix/Perm kit (Thermo Fisher Scientific). The intracellular markers FoxP3 and IL-21 were stained in permeabilization buffer. Both compensation controls (OneComp eBeads; Thermo Fisher Scientific) and fluorescence minus one (FMO) controls were utilized. All the samples were collected on an LSR Fortessa X-20 analyzer (BD Biosciences, San Jose, CA) and were analyzed using FlowJo software (FlowJo, LLC, Ashland, OR). Approximately 2 × 10⁵ to 1 × 10⁶ events were collected per sample.

SHIV_SF162P3 rectal challenge.

SHIV_SF162P3, R157 derived, from harvest 3 (16 March 2012) was generously provided by Nancy Miller, NIAID. This stock had a P27 content of 66 ng/ml and an RNA content of log ∼9.35. Its TCID₅₀ in Indian origin rhesus PBMC was 1,288/ml, and its TCID₅₀ in TZM-bl cells was 4.1 × 10⁴/ml. One milliliter of a 1:300 dilution of the stock was used as recommended by Nancy Miller. This equaled 4.3 TCID₅₀ on rhesus PBMCs and 137 TCID₅₀ on TZM-bl cells. This dose was used for weekly intrarectal (i.r.) challenge. Plasma samples were analyzed for SHIV RNA copy numbers by Leidos Biomedical Research, Inc., Frederick National Laboratory. Animals with RNA copies above 10 were considered to be infected, and the numbers of challenges required to infect that animal were used as events for Kaplan-Meier analysis. Once infected, the animal was no longer challenged. Plasma viral loads were monitored periodically by the same method until the end of the study.

SHIV load in tissues.

At the end of study, PBMCs and postmortem tissues were collected. PBMC and gut samples were analyzed for SHIV RNA by quantitative PCR (qPCR).

Statistical analysis.

Prism 7 graphical software was used for all statistical analyses.