HIV-1 CD4-induced (CD4i) gp120 Epitope Vaccines Promote B and T-cell Responses that Contribute to Reduced Viral Loads in Rhesus Macaques

Abstract

To target the HIV CD4i envelope epitope, we primed rhesus macaques with replicating Ad-rhFLSC (HIV-1_BaLgp120 linked to macaque CD4 D1 and D2), with or without Ad-SIVgag and Ad-SIVnef. Macaques were boosted with rhFLSC protein. Memory T-cells in PBMC, bronchoalveolar lavage and rectal tissue, antibodies with neutralizing and ADCC activity, and Env-specific secretory IgA in rectal secretions were elicited. Although protective neutralizing antibody levels were induced, SHIV_SF162P4 acquisition following rectal challenge was not prevented. Rapid declines in serum ADCC activity, Env-specific memory B cells in PBMC and bone marrow, and systemic and mucosal memory T cells were observed immediately post-challenge together with delayed anamnestic responses. Innate immune signaling resulting from persisting Ad replication and the TLR-4 booster adjuvant may have been in conflict and reoriented adaptive immunity. A different adjuvant paired with replicating Ad, or a longer post-prime interval allowing vector clearance before boosting might foster persistent T- and B-cell memory.

Introduction

It is widely believed that an effective human immunodeficiency virus (HIV) vaccine will have to elicit both T-cell and antibody responses in order to prevent acquisition and/or control disease progression. In view of the high mutability of HIV and its propensity for immune escape, highly conserved regions of viral proteins, including envelope epitopes, are likely critical in vaccine design. While many vaccine platforms available today induce T-cell responses and binding antibodies, few elicit high levels of neutralizing antibodies –a testament perhaps to the less than optimal design of the immunogens used. HIV entry into a cell involves a complex series of conformational changes (1, 2) initiated by the binding of Env to CD4 and exposure of the highly conserved co-receptor binding site. Antibody responses to this CD4-induced (CD4i) epitope are present in HIV-infected individuals and display broad, potent binding and neutralization of a spectrum of HIV isolates across clades, as well as the highly divergent HIV-2 in the presence of soluble CD4 (3, 4). For these reasons the CD4i epitope constitutes an attractive target for vaccine development. A protein immunogen consisting of the full-length single-chain HIV-1_BaLgp120 linked to the D1 and D2 domains of rhesus macaque CD4 (rhFLSC) was developed (5). Immunization with this construct elicited antibodies to CD4i sites which correlated with control of SHIV_SF162P3 infection following intrarectal challenge of rhesus macaques (6). In spite of these promising results, this rhFLSC immunogen has until now not been incorporated into a viral vector for further vaccine evaluation.

We are pursuing a strategy employing mucosal priming with replicating adenovirus type 5 host range mutant (Ad5hr)-HIV/SIV recombinants, followed by Env protein boosts. This approach has elicited broad cellular immune responses and functional, envelope-specific systemic and mucosal antibodies that correlate with protection from HIV, SIV, and SHIV challenges in chimpanzee and rhesus macaque models (7–12). In an effort to enhance the antibody responses induced by our platform we generated an Ad5hr-recombinant expressing the rhFLSC immunogen, and demonstrated its immunogenicity in mice (13). Here we evaluated whether priming rhesus macaques with replicating Ad-rhFLSC and boosting with rhFLSC protein would elicit broad neutralizing antibodies in serum and cellular immunity against envelope peptides in peripheral blood and at mucosal sites. In a second experimental arm we primed with Ad5hr recombinants encoding rhFLSC as well as SIV Gag and SIV Nef, followed by boosting with rhFLSC protein, expecting that this regimen would elicit broader cellular immunity and equivalent antibody responses. We subsequently assessed protective efficacy following a simian/human immunodeficiency virus (SHIV) intrarectal challenge. Reduced viral loads were observed in vaccinated macaques following challenge, but in spite of elicitation of T-cell responses and antibodies with protective neutralizing activity levels, prevention of SHIV acquisition was not attained. Factors potentially associated with this lack of sterilizing immunity were subsequently explored in depth.

Materials and Methods

Immunogens

Construction of a replication-competent Ad5hr recombinant vaccine encoding rhFLSC (MAd5rhFLSC) was previously described (13). The rhFLSC consists of full length single-chain HIV_BaL gp120, a flexible linker, and the D1 and D2 domains of rhesus macaque CD4 (5). The Ad5hr-SIV₂₃₉gag and Ad5hr-SIV₂₃₉nef_Δ1-13 recombinants were previously described (14, 15).

Animals, immunization, and challenge

Twenty-four Indian rhesus macaques (Macaca mulatta), 12 male and 12 female, were housed and maintained at Advanced BioScience Laboratories, Inc., (ABL, Rockville, MD) according to the standards of the American Association for Accreditation of Laboratory Animal Care. All were negative for SIV, simian retrovirus type D and STLV. Eight macaques per group were immunized at week 0 intranasally and orally, and at week 12 intratracheally, with replication-competent MAd5rhFLSC (group1) or MAd5rhFLSC, Ad5hr-SIV₂₃₉gag and Ad5hr-SIV₂₃₉nef_Δ1-13 (Group 2) at a dose of 5 × 10⁸ PFU/recombinant/route. Group 1 macaques additionally received empty Ad5hr vector to make the total Ad5hr dose 1.5 × 10⁹ PFU. The macaques were boosted intramuscularly at two sites with recombinant rhFLSC protein (Profectus BioSciences, Inc., Baltimore, MD), a total of 100 μg/macaque, in EM005 adjuvant (10 μg in 2% oil; Infectious Disease Research Institute (IDRI) Seattle, WA) at weeks 25 and 36. Four control macaques (Group 3) received empty Ad5hr vector (1.5 × 10⁹ PFU/macaque) and adjuvant alone. Four additional controls were naïve. All macaques were challenged intrarectally at week 42 with a moderate single dose (256 TCID₅₀) of SHIV_SF162p4.

Sample collection

Peripheral blood mononuclear cells (PBMC) obtained throughout the immunization course and postchallenge were purified from whole blood by Ficoll gradient centrifugation and used immediately for intracellular cytokine staining assays. Bone marrow lymphocytes were similarly isolated. Bronchoalveolar lavage (BAL) samples and rectal pinch biopsies were collected and processed as previously described (16). Serum samples were collected, aliquoted, and stored at −70°C until use. Rectal secretions were collected as described previously (17), and stored at −70°C until analyzed.

Intracellular cytokine staining

Freshly isolated PBMC (2 × 10⁶) were stimulated with pools of HIV_BaL gp120, SIV_mac239 Gag, or SIV Nef peptides and stained as described previously (18) except flourochromes for CD4 and gamma interferon (IFN-γ) antibodies were changed to CD4-FITC and IFN-γ-PE (both BD Biosciences). A singlet, followed by live/dead and then lymphocytic gates, were first applied. CD3+ T cells were divided into CD4+ and CD8+ populations, and each population was further subdivided into CD28+ CD95+ central memory (CM) and CD28− CD95+ effector memory (EM) cells. The percentage of cytokine-secreting cells in each memory cell subset was then determined following subtraction of the values obtained with nonstimulated samples. Data were analyzed using FlowJo software (TreeStar Inc.).

Binding, and neutralizing antibodies

Serum binding antibodies to rhFLSC and HIVBaL gp120 Env protein were assessed by enzyme-linked immunosorbent assay (ELISA) as described previously (19). The antibody titer was defined as the reciprocal of the serum dilution at which the optical density (OD) of the test serum was two times greater than that of the negative-control serum diluted 1:50.

Neutralizing antibody titers against SHIV_BaL-P4, SHIV_SF162P3, and SHIV_SF162P4 (all grown in human PBMC) and HIV-2 7312A/V434M (a pseudovirus produced in 293T cells) with and without sCD4 were assayed in TZM-Bl cells as described (20). Sera were also evaluated for neutralization of infectious molecular clones of tier 2 clade B and C IMC.LucR viruses (21) produced in 293T cells using A3R5.7 cells (22) as described (23). Titers were defined as the reciprocal serum dilution at which there was a 50% reduction in relative luminescence units compared to virus control wells which contained no test sample.

Memory B cells, plasma blasts (PB) and plasma cells (PC)

Bone marrow and PBMC lymphocytes were isolated for enumeration of total and rhFLSC-specific IgG and IgA secreting B cells as described (24). Briefly, cells were washed in R10 and aliquots were either assayed directly for antibody secreting cells (ASC) by ELISPOT to quantify PB/PC, or were first stimulated for 3 days in R10 medium supplemented with 1 μg/ml CpG (ODN-2006) (Operon), 0.5 μg/ml recombinant human sCD40L (Peprotech), and 50 ng/ml recombinant human IL-21 (Peprotech) to quantify memory B cells. Assays were carried out in duplicate, and data are reported as the percentage of Env-specific ASC relative to the number of total ASC.

Env-specific secretory IgA (sIgA) in mucosal samples

Rectal secretions were tested for blood contamination using Chemstrips 5 (Boehringer Mannheim). Due to significant amounts of blood in the secretions which would have complicated determination of the origin of IgA and IgG antibodies present, Env-specific sIgA in the secretions was assessed by ELISA using anti-monkey secretory component. Briefly, mucosal samples were 2-fold serially diluted, applied to a half-area 96-well plate (Greiner Bio-One) coated with 1 μg/ml rhFLSC, HIV_BaL gp120, or SHIV_SF162P3 gp120, and incubated at 4°C overnight. HRP-conjugated goat anti-monkey secretory component (GAMon/SC/PO; Nordic) and TMB substrate were used in sequential steps, followed by reading the OD at 450 nm. High-titered sera positive for reactivity against HIV gp120 were negative in this assay at a serum dilution of 1:10. Endpoint titers were defined as the reciprocal of the sample dilution at which the OD of the test sample was equal to twice the mean background OD.

Antibody-dependent cell-mediated cytotoxicity (ADCC)

Serum samples collected at week 38 (2 weeks following the second rhFLSC boost) were assessed for ADCC activity using the RFADCC assay as previously described (25). Briefly, rhFLSC protein was used to coat CEM.CCR5.NKr target cells which were then co-cultured with human PBMC effectors at an E:T ratio of 50:1. Serial dilutions of macaque sera were tested for ADCC activity in a 4h assay. Endpoint ADCC titers are reported as the reciprocal serum dilution at which the background cutoff value (mean ADCC activity over a dilution series of a pool of macaque negative sera plus 3 standard deviations) was reached. ADCC-mediated maximum percent killing of target cells (% ADCC Max killing) was defined for each positive sample as the highest percent killing mediated at any of the dilutions tested. Sera with percent killing below the cutoff value were scored negative. The 50% maximum killing titer was defined as the reciprocal serum dilution at which 50% of the ADCC maximum killing was achieved.

Quantitation of viral RNA

SHIV_SF162P4 RNA in plasma was quantified by the NASBA assay (26, 27). The threshold of detection was 50 SIV RNA copies/ml plasma.

Statistical analysis

Analyses of intracellular cytokine responses, antibody titers, and viral loads used the exact Wilcoxon rank sum test for two-group comparisons. The Spearman rank correlation test was used to assess relationships between immune responses and virologic parameters. Results of statistical analyses were considered significant at P values of ≤0.05.

Results

Favorable levels of vaccine-induced neutralizing antibodies do not prevent SHIV_SF162p4 acquisition

Previously, immunization of macaques with rhFLSC elicited antibody responses to CD4i epitopes which were associated with control of viremia following intrarectal challenge with SHIV_SF162P3 (6). However, all vaccinated macaques became infected and only low-level neutralizing antibodies were elicited against this difficult to neutralize virus. Three of four macaques neutralized SHIV_SF162P3 with ID₅₀ values ranging from 12 to 80, while the best protected macaque had no detectable neutralizing antibody response at all (6). Here, we designed a study for an intrarectal challenge with SHIV_SF162P4, based on the rationale that our prime/boost approach would elicit a higher titer antibody response with neutralizing activity against this more easily neutralized virus, and that we would be able to determine if rhFLSC induced antibodies would be able to prevent SHIV_SF162P4 acquisition. Two weeks after the second rhFLSC boost we evaluated neutralizing antibody responses generated by the vaccine regimen. Neutralizing antibody responses with mean titers ranging from 22 to 50 were observed against tier 2 viruses of HIV clades B and D (data not shown). Mean neutralizing antibody titers induced against SHIV_BALP4, representative of the gp120 portion of the rhFLSC construct, were low (74 and 89 for groups 1 and 2, respectively), whereas intermediate titers were elicited against the SHIV_SF162P4 challenge virus (253 and 266 for groups 1 and 2, respectively; Table 1). Notably, however, these SHIV_SF162P4 neutralizing antibody titers were at protective levels. Bogers et al. previously reported that vaccine-induced protection against SHIV_SF162P4 acquisition was correlated with neutralizing antibody titer. The five protected macaques in that study had titers as low as 1:80 at the time of challenge (11). Moreover, the intrarectal challenge dose in that study was 1800 TCID₅₀, which was higher than the 256 TCID₅₀ dose used here. We also observed neutralizing antibodies against HIV-2 in the presence of CD4, with mean titers of 1247 and 2654 (Table 1), comparable to those observed (ranging from 300 to 2000) following vaccination with rhFLSC alone (6), and indicating that neutralizing antibodies to the CD4i epitope were induced.

Table 1.

Neutralizing antibody titers pre-challenge (week 38).

	ID50
Immunization Groups	SHIVBALP4	SHIVSF162P4	HIV-2 + sCD4
Group 1	74 ± 15	253 ± 63	1247 ± 310
Group 2	89 ± 19	266 ± 59	2654 ± 1024
Groups 3 + 4	0 ± 0	0 ± 0	0 ± 0

Data expressed as means ± SEM. Week 38 = 2 weeks post-boost.

In spite of induction of promising levels of neutralizing antibodies, protection against SHIV_SF162P4 acquisition was not obtained (Fig. 1). All macaques became infected following the moderate single dose intrarectal challenge. A modest but significant reduction in area under the curve (AUC) compared to the controls was observed for group 2 macaques primed with Ad5hr-rhFLSC in addition to Ad5hr-SIVgag and Ad5hr-SIVnef (p = 0.029) but not for group 1, suggesting a benefit derived by the addition of SIV Gag and Nef recombinants. However, there was no difference in AUC values between groups 1 and 2, allowing us to combine the groups. The greater statistical power achieved resulted in a significant difference between both vaccinated groups and the controls (p = 0.018).

Figure 1. Plasma viral loads following SHIV_SF162P4 intrarectal challenge.

Viral loads of individual macaques (panels A–C) and geometric mean viral loads of each immunization group are shown. A: Group 1 macaques primed with Ad5rh-FLSC and boosted with rhFLSC protein; B: Group 2 macaques primed with Ad5rh-FLSC, Ad5hr-SIV₂₃₉gag, and Ad5hr-SIV₂₃₉nef_Δ1-13, and boosted with rhFLSC protein; and C: Control macaques primed with empty Ad5hr-vector and boosted with adjuvant only (n = 4) or unimmunized (n = 4).

Vaccine-elicited binding and neutralizing antibodies

To understand the basis for the lack of protection against acquisition, we first evaluated humoral immune responses over the course of the immunization and challenge. High-titered binding antibodies to both rhFLSC and HIV_BAL gp120 (titers of 10⁶ and 10⁵, respectively) were elicited following the second Env administration for both vaccine groups (Fig. 2). However, the titers were not well-sustained between immunizations. Moreover, the anamnestic response following SHIV_SF162P4 challenge appeared to be delayed. This is illustrated in Table 2 where binding titers 2 weeks post-challenge (week 44) against both rhFLSC and HIV_BaL gp120 remained below the peak titers achieved 2 weeks post-boost (week 38) in both immunization groups. Binding titers to rhFLSC increased by 4 weeks post challenge (week 46), although HIV_BaL gp120-specific binding titers remained below peak levels. We contrast this with previous studies using similar prime/boost vaccine regimens in which significant protective efficacy was achieved. For example, in the study by Patterson et al. (9), immunized with Ad-SIV and boosted with SIV Env protein by a similar schedule to the study here prior to intrarectal SIV challenge, a 3-fold increase in binding titer to SIV gp120 was observed 2 weeks post challenge, with a further increase to ~9-fold at 4 weeks post-challenge. Further, in the study by Demberg et al. (12), immunized with Ad-HIV and boosted with HIV Env protein by the same schedule but challenged at week 50 with SHIV_89.6P, the binding antibody titer measured at 2 weeks post-challenge was equivalent to the week 38 binding titer (Table 2), reflecting the longer time period between immunization and challenge. However, by 4 weeks post-challenge the binding titer to SHIV_89.6P gp120 had increased 10-fold over the peak pre-challenge titer. From these comparisons we conclude that here the humoral anamnestic response to SHIV_SF162P4 exposure was significantly delayed.

Figure 2. Vaccine-induced antibody responses in serum.

Geometric mean serum binding antibody titers against rhFLSC and HIV_BaLgp120 over time. The arrows indicate the time of SHIV_SF162p4 challenge.

Table 2.

Delayed anamnestic response in vaccinated macaques.

		Geometric Mean Binding Titer			Fold Increase (geometric mean)
Macaque Group	Test antigen	2wks post-boost	2wks post-challenge	4 wks post-challenge	2 wkpc/2 wkpb	4 wkpc/2 wkpb
Group 1	rhFLSC	1,222,405	843,217	4,711,167	0.80	3.85
Group 2	rhFLSC	1,138,732	477,305	5,642,405	0.45	5.22
Groups 1 and 2	rhFLSC	1,179,827	646,555	5,222,660	0.58	4.54
Group 1	BaL gp120	163,052	24,335	94,192	0.13	0.54
Group 2	BaL gp120	177,372	16,101	78,199	0.09	0.44
Groups 1 and 2	BaL gp120	170,062	19,794	85,293	0.11	0.49
Patterson et al., 2004	SIV gp120	116,604	395,255	980,155	3.43	8.77
Demberg et al., 2007	SHIV89.6P gp120	27,149	25600	289,631	0.94	10.67

A similar result was obtained upon analysis of neutralizing antibody titers over time (Fig. 3A–C). The profiles of neutralizing titers against SHIV_Balp4, SHIV_SF162P4, and HIV-2 + sCD4 were similar for groups 1 and 2 from 2 weeks after the second boost (week 38) out to 8 weeks post-challenge (week 50). However in all cases, neutralizing titers were lower at week 44, 2 weeks post-challenge, compared to the week 38 titers. This is illustrated in the right-hand panels of Fig. 3A–C, where highly significant differences in titers between the pre-and post-challenge time points were obtained for group 1 and 2 macaques combined. Increases in neutralizing antibody titers only became apparent at week 46, 4 weeks post-challenge, again reflecting a delayed anamnestic response.

Figure 3. Vaccine-induced serum neutralizing antibody titers.

Neutralizing antibody titers to (A) SHIV_BaLp4, (B) SHIV_SF162p4, and (C) HIV-2 with sCD4 added. Values shown are log titers + SEM. The far right-hand plots contrast week 38 and 44 responses of combined group 1 and 2 macaques. Significant differences were evaluated by Wilcoxon matched-pairs signed rank test following log transformation. The arrows indicate the time of SHIV_SF162p4 challenge.

Env-specific memory B-cell responses

The observed delay in humoral anamnestic responses prompted us to examine induction of Env-specific memory B cells and PB/PC. The presence of memory B cells would indicate a readiness to rapidly respond to viral Env, whereas the presence of PB/PC (not distinguished by the ELISPOT assay) would indicate the presence of short-lived ASC and/or the ability to maintain a sustained antibody response. PB/PC levels were evaluated by ELISpot for detection of rhFLSC antibody secreting cells (ASC) on unstimulated lymphocytes of PBMC and bone marrow. Both rhFLSC-specific IgG and IgA ASC were absent in PBMC prior to challenge at week 42 (Fig. 4A), however a low level of both were seen in bone marrow at week 38 after the second Env booster immunization and prior to challenge (Fig. 4B). The levels of these PB/PC were similar between animals of group 1 and 2. In contrast, both rhFLSC-specific IgG and IgA memory B cells were readily observed by ELISpot in lymphocytes of PBMC and bone marrow after three days of polyclonal stimulation, initially appearing after the first Env boost (week 27). Notably, however, the levels of rhFLSC-specific IgG memory B cells declined dramatically after the SHIV challenge by week 50 (8 weeks post-challenge) in both PBMC and bone marrow (Fig. 4A, B). A decline in rhFLSC-specific IgA ASC was less apparent. Because of the similarities of groups 1 and 2 in levels of memory ASC and in their profiles over the course of immunization, we combined the groups for statistical analysis. As indicated by the plots in the far right column of Fig. 4, highly significant reductions in rhFLSC-specific IgG memory B cell levels in PBMC and bone marrow were observed between pre- (week 38) and post-challenge (week 50) time points (p < 0.0001 and p < 0.0042, respectively). It is possible that the reductions in Env-specific IgG memory B cells were in part due to maturation into PB/PC, which were seen to increase over the 38 to 50 week time period (Fig. 4). However, the memory B cell reductions seen here were unusual, as previously we have observed increases in memory B cells in vaccinated macaques (24,28) as well as PB/PC (28) when levels post-challenge were compared to those observed 2 weeks post-Env boosting. Significant differences in rhFLSC-specific IgA memory B cell levels over the same time period were not observed here.

Figure 4. Vaccine-induced IgG and IgA ASC.

(A) PBMC and (B) bone marrow plasma cells/plasma blasts (evaluated by ELISpot on unstimulated cells) and memory B-cells (evaluated by ELISpot on stimulated cells as described in Materials and Methods). Env-specific IgG and IgA ASC are shown as the ratio of Env-specific ASC divided by total IgG or IgA ASC in each sample. The far right-hand plots contrast week 38 and 44 responses. Statistical differences were analyzed by the Wilcoxon signed rank test performed on arcsine-transformed percentages. Arrows indicate the time of SHIV_SF162p4 challenge.

Env-specific T-cell immune responses

It has long been established that viral-specific cellular immunity can control disease progression in SIV-infected macaques (29, 30). More recently this has been dramatically extended to preclinical studies in macaques where vaccine-elicited cellular responses have led to strong control of viremia (31–33). Previously, our replicating Ad-recombinant prime/protein boost strategy yielded viral-specific T-cell responses that correlated with control of set-point viremia (9). Therefore we evaluated T-cell responses elicited by the rhFLSC regimen by intracellular cytokine staining for IFN-γ, IL-2, and TNFα in PBMC, BAL and rectal cells in response to stimulation with Nef, Gag or Env peptides. Robust Gag- and modest Nef-specific CD4 and CD8 responses, both effector memory (EM) and central memory (CM), appeared immediately after the first Ad-recombinant prime in PBMC of Group 2 macaques and were maintained throughout the course of immunization and subsequent challenge period (Fig. 5A). Similar profiles were observed for Gag-specific CD4_EM and CD8_EM cells of BAL and rectal tissue (Fig. 5B, C), where due to insufficient cells, Nef-specific responses were not assayed. Env-specific CD4 and CD8 responses developed more slowly and were observed at lower levels compared to Gag-specific responses in PBMC as well as BAL and rectal tissue (Fig. 5A–C). The low level cellular responses seen at week 14 after the two Ad-recombinant immunizations actually declined significantly with the first rhFLSC boost (week 27) in the CD8_CM PBMC population (p = 0.0076) and in BAL CD4_EM cells (p = 0.0063). Subsequently, Env-specific responses pre-challenge in general peaked at week 38, two weeks after the second rhFLSC boost.

Figure 5. Vaccine-induced cytokine positive cells.

Intracellular cytokine staining of (A) PBMC, (B) BAL, (C) rectal tissue for specific CD4⁺ and CD8⁺ memory T cells secreting IFN-γ, IL-2, and TNF-α. (A to C) Shown are stacked responses to HIV Env, SIV Gag, and SIV Nef peptide pools by CD4⁺ and CD8⁺ CM and EM cells over the course of immunization. The far right-hand plots contrast week 38 and 44 Env-specific cellular responses of combined group 1 and 2 macaques. Statistical analysis was done by the Wilcoxon signed rank test performed on arcsine-transformed percentages. The arrows indicate the time of SHIV_SF162p4 challenge. Error bars represent standard errors of the mean (SEM).

Following SHIV_SF162P4 challenge at week 42, Gag-specific responses in group 2 macaques were in general maintained, or tended to increase. In contrast, Env-specific responses in both immunization groups appeared to decline in most cases. Because of the similarities in levels of Env-specific CD4 and CD8 memory responses in groups 1 and 2 we combined the groups for further analysis. As indicated by the paired pre- (week 38) and post-challenge (week 44) responses (far right columns, Fig. 5) CD4_EM and CD4_CM responses in PBMC significantly declined (Fig. 5A; p = 0.018 and p = 0.013, respectively). Similar significant declines in Env-specific responses over the same time period were also seen in CD4_EM and CD8_EM BAL cells (Fig. 5B; p = 0.0042 and p = 0.0063, respectively) and in CD8_EM cells of the rectum (Fig. 5C; p < 0.0001). A decline in rectal Env-specific CD4_EM cells only approached statistical significance. Similar to the neutralizing antibody titers and Env-specific memory B-cell responses, Env-specific anamnestic responses by CD4 T cells in PBMC and by both CD4 and CD8 T cells of rectal tissue were delayed. None were observed in CD8 T cells of PBMC or in BAL CD4 and CD8 T cells.

Factors influencing differences in viral load

In spite of the failure to prevent SHIV acquisition and the delay in anamnestic responses, both humoral (Table 2, Fig. 3) and in particular the rectal CD4 and CD8 responses (Fig. 5), may have contributed to the reduced viral loads post-challenge. Because the animals were rectally challenged we additionally evaluated levels of Env-specific anti-IgA (secretary component) in rectal secretions. Detection of Env-specific IgA using anti-secretory component is less sensitive than using anti-IgA directly. Nevertheless, Env-specific sIgA titers increased two to four weeks post-challenge (Fig. 6). Thus along with other vaccine-induced humoral and cellular responses, rectal Env-specific sIgA may also have contributed to the reduced viral loads in the immunized macaques as shown by area-under-the-curve values (Fig. 1).

Figure 6. Env-specific secretory IgA in rectal secretions.

Rectal swabs were collected at 36, 44 and 46 weeks post immunization. Time of SHIV_SF162P4 challenge was week 42, marked by the arrow. Graphs show mean secretory antibody titers + SEM.

Non-neutralizing antibodies by interacting with Fc receptors mediate a number of functional activities (34, 35). These have been suggested to inhibit the spread of HIV (36–37) and have also been implicated in the protective efficacy seen in the Phase III HIV vaccine trial, RV144 (38). Vaccine-elicited non-neutralizing antibodies have also been correlated with protection against both SIV (39) and SHIV (17) in rhesus macaques. To determine if antibodies induced by the rhFLSC vaccine regimen impacted the challenge outcome, we measured ADCC activity pre- (week 38) and post-challenge (week 44; Fig. 7). Both groups of vaccinated animals showed comparable ADCC titers and maximum killing percentages at both time points (Fig. 7A, B, D, and E). The post-immunization ADCC titers did not correlate with reduced viremia in the vaccinated macaques (data not shown), perhaps in part due to the fact that the activity levels were not sustained, but dropped significantly between weeks 38 and 44 (Fig. 7C, F). However, by week 44 the macaques with the smallest drop in percent ADCC killing tended to have lower peak viremia, although statistical significance was not reached (Fig. 7G). Nevertheless, the ADCC activity that remained was an important factor as a significant negative correlation was seen between the 50% maximum killing titer at week 44 and subsequent viral loads at week 46 (Fig. 7H). Thus ADCC activity elicited by the rhFLSC vaccine regimen had a measurable effect on the challenge outcome.

Figure 7. Serum antibodies mediating ADCC pre- and post-challenge.

Serum samples collected at (A, D) week 38 and (B, E) week 44 (2 weeks post-challenge) were evaluated for ADCC activity. (C) Plots contrasting week 38 and 44 ADCC titers and (F) percent maximum killing responses of combined group 1 and 2 macaques. (G) Week 44 percent maximum killing values were subtracted from those at week 38 and the resulting difference was plotted against peak viral loads. (H) Significant negative correlation of ADCC 50% maximum killing titers at week 44 with viral loads at week 46. Titers less than background were assigned values of 10. Statistical analyses were done by (A, B, D, E) the Mann-Whitney-Wilcoxon test, (C, F) the Wilcoxon signed rank test or (G, H) Spearman’s rank correlation.

Discussion

In this study we evaluated a prime/boost vaccine design targeting CD4i epitopes, asking whether priming with a replicating Ad-recombinant encoding rhFLSC would confer added benefit to immunizations with rhFLSC protein alone, previously shown to result in reduced viremia levels following SHIV_SF162P3 challenge (6). We postulated that the prime/boost regimen would elicit both systemic and mucosal immune responses, including broad neutralizing antibodies and Env-specific cellular immunity, and that these responses would induce sterilizing immunity against intrarectal SHIV_SF162P4 challenge and/or reduce viremia. We observed induction of specific cellular immune responses, both in blood and BAL and in rectal tissue. Even though the neutralization breadth was more limited than we had hoped, the vaccine regimen elicited antibodies capable of neutralizing the challenge virus at titers equivalent to or higher than those previously associated with sterilizing protection following either intrarectal or intravaginal SHIV_SF162P4 challenge (11, 40). Env-specific secretory IgA antibodies were seen in rectal secretions, and antibodies mediating ADCC activity, significantly correlated with reduced viremia, were induced. As favorable levels of systemic and mucosal, cellular and humoral immune responses were achieved, the challenge outcome, including lack of protection from acquisition and only a modest reduction of viral loads in the vaccinated compared to control macaques, was surprising. One factor perhaps contributing to this unexpected outcome was our use of a single moderate challenge dose, used to insure infection of all the control macaques. While repeated low-dose challenges are more representative of natural HIV transmission (41) and permit one to more effectively monitor the protective efficacy of candidate vaccines, here the number of macaques studied did not provide sufficient statistical power to make the use of this challenge method feasible.

Other factors likely impacted the challenge outcome as well. In addition to neutralizing antibody titer, both the avidity of the vaccine-induced antibody (28, 42) and a strong, rapid anamnestic response (9, 28, 42–44) may be important for protective efficacy. Here we observed either delayed or absent anamnestic responses in binding and neutralizing antibodies and in cellular immune responses. Some decline in the measurable antibody response post-challenge may have resulted from immune complex formation, not evaluated here. However the decline contrasted sharply with previous studies in which rapid increases in Env-specific antibodies were observed following both SIV and SHIV challenges (Table 2; 9, 12). We suggest that the delayed antibody response is most likely related to the significant drop in Env-specific IgG memory B cells, seen in both peripheral blood and bone marrow following SHIV exposure (Fig. 4A, B). Loss of memory B cells has been well documented in both HIV (45, 46) and SIV (47, 48) infection. However, as with Env-specific antibody responses, we have previously observed significant increases in Env-specific memory B cells post-challenge in vaccinated macaques (24, 28). Thus, it is not known why the induction of strong Env-specific memory B cells over the course of immunization didn’t offset the subsequent B-cell dysfunction. A better elicitation of Env-specific plasma cells could have provided more sustained antibody responses, obviating the need for rapid antibody production by memory B cells upon virus exposure. It is possible that the SHIV_SF162P4 envelope expressed post-challenge inhibited proliferation and expansion of the memory B cell subset by inducing TGF-β1 production and expression of the B cell inhibitory receptor, FcRL4, as recently described (49). This possibility should be explored. The observed loss of CD4_EM and CD4_CM cells post challenge (Fig. 5A, B) may also have resulted in insufficient T cell help necessary for sustaining memory B cell expansion.

We also observed a significant drop in CD8_CM cells in peripheral blood and in CD4_EM BAL cells between the second Ad-recombinant immunization and the first Env boost. This result, as well as the less than optimal induction of B cell memory, may reflect conflicts in induced innate immune responses elicited by the Ad immunizations and the adjuvant/protein boost. Precedence for the reorientation of immune responses by TLR ligands has been established in allergy research, where they have been shown to reverse Th2 skewing and restore the Th1/Th2 balance (50). This shift derives from the dependence of Th1 and Th2 cell differentiation on their cytokine environment. IFN-γ and IL-12 are the dominant factors for development of Th1 cells, while the presence of IL-4 leads to Th2 cells (51). These cytokines, also produced by Th1 and Th2 cells respectively, are cross-regulatory, so that IL-4 diminishes priming by IFN-γ and IFN-γ diminishes priming by IL-4. Thus, TLR agonists as adjuvants, by interacting with TLR on a variety of cell types and triggering production of different cytokines, can have potent effects on T-helper cell induction, leading to modulation of immune responses. These effects are not always predictable. For example, a TLR4 agonist, as contained here in the EM005 adjuvant (52), is expected to promote Th1 responses, yet it has recently been shown that TLR-4 stimulation of human basophils leads to IL-4 production. In fact in the presence of IgE, such stimulation can result in Th2 skewing (53).

Innate immune responses to Ad are mediated by numerous signaling pathways, including not only multiple MyD88-dependent TLR signaling pathways (54) but also lectin receptors, inflammasome signaling comprising AIMs-like receptors, NOD-like receptors and RIG-1-like receptors (55). Moreover, Ad, either because of immune-stimulatory effects of the viral structural antigens or other gene products, itself acts as a potent adjuvant (56, 57). Further, we have shown that replication-competent Ad is broadly distributed in macaque tissues, and following mucosal immunization targets macrophages and mDC in BAL and rectal mucosa. It can persist at least 25 weeks in rectal tissue (58). Therefore we cannot exclude that residual adenovirus may have still been present at the time of the Env booster immunizations, leading to combined effects of the Ad recombinants together with the administered adjuvant. Moreover, it has been shown that adjuvants differ in their effects depending on the length of time following administration, such that short-term effects might differ from long-term effects (59). A thorough analysis of cytokine responses over the course of replicating Ad priming and Env/adjuvant boosting would be extremely valuable in addressing these issues.

The EM-005 adjuvant used here contains GLA, a synthetic lipid A derivative and TLR-4 agonist (52). Pattern recognition receptors (PRR) can induce a variety of responses depending on the cell type activated, the ligand recognized, and the history of the responding cell (60). Pairing of the replicating Ad-recombinant with the TLR-4 agonist may not have provided the optimal outcome: maintenance and boosting of cellular memory while inducing a potent anti-Env antibody response. The latter may have benefitted from a TLR2 agonist, shown to induce anti-Ad neutralizing antibody and anti-transgene responses (61) following Ad vector administration.

The yellow fever vaccine, one of the most successful, has been reported to elicit a spectrum of innate and adaptive immune responses including a mixed Th1/Th2 cytokine profile (62). The authors indicate the importance of defining specific pattern recognition receptors that promote such a Th1/Th2 balance and elicit both long-lived neutralizing antibody and memory T cells. Thus, further investigations to define a different pairing of adjuvant with the replicating Ad vector are warranted. Alternatively, a longer interval between the Ad-recombinant priming and Env boost, which could allow clearing of the vector, might lead to enhanced memory responses following a subsequent Env-EM-005 booster immunization.

As confirmed here, CD4i epitopes elicit antibodies with neutralizing activity (3, 6, 63) and effector functions (4) and contribute to control of HIV infection, indicating their continued value in vaccine design. Nevertheless, further investigation of approaches able to enhance vaccine-elicited T- and B-cell memory and avoid B-cell dysfunction following viral exposure are of critical importance. The choice of a replicating Ad-vector for priming immune responses was based in part on its targeting of mucosal sites, as well as the need for persistent antigen expression. However, in view of this persistence, exploration of effects of combined innate signaling molecules and/or timing of immunizations and/or adjuvant appropriateness should be explored.

Highlights.

• Rhesus macaques were vaccinated with rhFLSC targeting HIV CD4i envelope epitopes

• Strong mucosal and systemic cellular and humoral immune responses were induced

• Protective neutralizing antibody levels did not prevent SHIV_SF162P4 acquisition

• Memory T and B cells declined rapidly post-challenge with delayed anamnestic response

• Different vector/adjuvant pairing might foster persistent T- and B-cell immune memory