The trimeric HIV-1 gp41-derived polypeptide (mgp41) was produced by PX’therapeutics (Grenoble, France) and characterized by biophysical analyses as described in Supplementary Material. Several mgp41 amine residues were substituted with sulfhydryl groups using Traut’s reagent (Piercenet, Rockfort, IL, USA) and then coupled to CTB (Crucell AB, Stockholm, Sweden) using the sulfosuccinimidyl 4-[p-maleimidophenyl]butyrate crosslinker (Piercenet). The CTB-mgp41 conjugate was purified by ultracentrifugation on AMICON filters and its purity checked by migration on SDS-PAGE gel. Its biological activity and the CTB:mgp41 ratio were measured by solid phase ELISA using GM1 gangliosides as capture system and antibodies to CTB or mgp41 for detection. Cholera toxin (CT) from List Biologicals laboratories (USA) was used as adjuvant for SL immunizations and Aluminum hydroxide (Sanofi-Pasteur, Lyon, France) as adjuvant for IL immunizations.

Eighteen adult female cynomolgus macaques (Macaca fascicularis) from Mauritius were housed within CEA animal facilities (Fontenay-aux-Roses, France) according to French national regulations and under the supervision of national veterinary inspectors (CEA Permit Number A 92-032-02). The CEA complies with the Standards for Human Care and Use of Laboratory Animals of the Office for Laboratory Animal Welfare (OLAW, USA) under OLAW Assurance number #A5826-01. This study was approved and accredited under statement number A10-053 by the ethics committee “Comité d’Ethique en Expérimentation Animale du CEA” registered under number 44 by the French Ministry of Research. Animals weighing between 3 and 6 kg, seronegative for SIV, STLV, filovirus, HBV, herpes B, and measles and genotyped for Class I and Class II MHC alleles were used. They were equally distributed between groups based on their genotype (n = 6 per group). Animals were sedated with ketamine chlorhydrate (Rhone-Merieux, Lyon, France) for 1 h. Group 1 received five SL immunizations with CTB-mgp41 (100 μg/dose of mgp41 antigen at W0, 4, and 12 and 50 μg/dose at W8 and 20) with CT (10 μg/dose). Group 2 received three SL immunizations with CTB-mgp41 and CT (SL priming) at W0, 4, and 8 similarly to group 1 followed by IM boosts with 100 µg of mgp41 in Alum (500 μg/dose) at W12 and 20. Group 3 received a SL priming with CT alone (10 µg) at W0, 4, and 8 followed by IM boosts with 100 µg of mgp41 in Alum (500 μg/dose) at W12, 20, and 28. The SL vaccine was applied in a 500 µL volume of PBS under the tongue of sedated animals with their head bending forward to avoid leakage of excess of vaccine and was then rinsed with PBS 15 min later in order to avoid swallowing. For intramuscular immunizations, a 500 µL volume was injected in the right thigh for W12, the left thigh for W20, and the right thigh for W28.

Blood, vaginal, and rectal washes were collected for antibody analysis by ELISA (W0, W6, W10, W14, W22, W24, and W28). Vaginal and rectal secretions were collected using Weck-cel™ sponges pre-moistened with 100 µL of PBS in presence of protease inhibitors as described previously (22).

Fluids were assayed for mgp41-specific IgA and IgG antibodies by a two-step amplified ELISA using biotin-conjugated goat anti-human IgG or IgA antibodies (Southern Biotech, Birmingham, AL, USA) and HRP-conjugated avidin (Sigma-Aldrich, St. Quentin Fallavier, France). Solid-phase bound HRP activity was monitored spectrophotometrically after addition of enzyme substrate. The endpoint titer was the reciprocal of the sample dilution giving an absorbance at least equal to threefold that of background [corresponding pre-immune sample at the lowest dilution (1/50 for sera and 1/3 for mucosal samples)].

Peripheral blood mononuclear cells (PBMCs) were isolated from blood 5 days after the priming (W9) and the boost (W21 for group 1 and 2 or W29 for group 3), 4 weeks after the last immunization (W24, groups 1 and 2 or W32, group 3), and 3 months after the last immunization (W32, groups 1 and 2 or W40, group 3) and assessed for antibody-secreting cells (ASCs) by B cell ELISPOT assay. Macaques were sacrificed 3 months after the last immunization (W33, groups 1 and 2 and W41, group 3) and rectal and vaginal mononuclear cells (MNCs) were prepared by enzymatic digestion of tissues with collagenase II (0.3 mg/mL) in the presence of trypsin inhibitor (0.1 mg/mL) (Sigma-Aldrich) and DNase I (0.1 mg/mL) (Roche) in RPMI medium supplemented with 3% FBS, 25 mM Hepes, antibiotics and Fungizon^® according to supplier’s recommendations (Invitrogen, Saint Aubin, France). Bone marrow (BM) MNC from iliac crest or from humerus were isolated using Ficoll density-gradient centrifugation (MSL2000, Eurobio, Les Ulis, France). PBMC and tissue-derived MNC collected 1 or 3 months after immunization were stimulated in vitro for 6 days with Pokeweed Mitogen (PWM, 1 µg/mL) and Staphylococcus aureus Cowan strain (SAC, 0.1 µg/mL) (Sigma-Aldrich) before ASC enumeration.

Peripheral blood mononuclear cell and MNC from vagina and rectum were assayed for mgp41-specific ASC using an amplified B ELISPOT assay as described (23). Briefly, graded numbers of cells (between 1 × 10⁵ and 8 × 10⁵) were incubated for 16 h at 37°C in nitrocellulose-bottom 96-well plates coated with mgp41 (5 µg/mL) for detection of mgp41-specific ASC or with mouse anti-human kappa and lambda antibodies (5 µg/mL) (Southern Biotech, Birmingham, AL, USA) for detection of total ASC. The antibodies produced were detected by stepwise incubations with biotin-conjugated goat anti-human IgG or IgA (KPL, Gaithersburg, MD, USA), avidin-conjugated peroxidase (Sigma-Aldrich), and AEC-H₂O₂ chromogenic peroxidase substrate (Sigma-Aldrich). Spots were enumerated using CTL immunospot analyzer (Bonn, Germany).

Mononuclear cells were incubated with fluorescent monoclonal antibodies against surface markers in order to identify B cell subsets. Antibodies from BD Biosciences used: anti-CD45 (clone D058-1283), anti-CD3 (clone SP34-2), anti-HLA-DR (clone L243), anti-CD20 (clone 2H7), as well as anti-CD27 (clone M-T271, Miltenyi) and anti-IgD (rabbit polyclonal, AbDserotec). Cells were analyzed on BD LSRII FACS analyzer and Diva Software (BD Biosciences).

Cryosections of SL mucosa were fixed with acetone before hematoxylin/eosin staining according to supplier’s instructions (Sigma-Aldrich) and analyzed using a fluorescence microscope DMD108 (Leica, Nanterre, France). Cryosections were stained with purified antibodies against HLA-DR-DP-DQ (clone CR3/473) and CD1a (clone O10) (Dako, Les Ulis, France) or corresponding isotype controls, revealed with donkey anti-mouse AlexaFluor 594 (Invitrogen), and analyzed with Zeiss LSM780 confocal microscope (Marly Le Roi, France).

Neutralization assays were performed as described (24) using two standard reference strains as Env-pseudotyped viruses (SF162.LS, tier 1 and QH0692.42, tier 2) to infect TZM-bl cells. The 50% inhibitory dose was defined as the sample dilution that caused a 50% reduction in relative luminescence units (RLUs) (25).

Statistical analyses were carried out with Graphpad prism version 6.0 (La Jolla, CA, USA). Pairwise multiple comparisons of experimental groups were performed using the non-parametric Mann–Whitney U-test.

A vaccine containing a gp41 polypeptide conjugated to the CTB mucoadhesive vector was designed to improve its SL delivery (Figure 1). We used a gp41 polypeptide including the gp41 amino acids M24 to S157 from the gp41 sequence of the BRU/LAI isolate in which several mutations were added (Figure 1A) (26). Mutations of seven residues were introduced in the gp41 loop region to increase its net charge and reduce its hydrophobicity, thus reducing the propensity to aggregation of the gp41 ectodomain (27). In addition, eight additional mutations were made at the C-terminal helical region (CHR) to open the six-helix bundle structure and to expose hidden epitopes at the conserved N-terminal helical region (NHR). Dynamic light scattering analyses revealed that the mutated gp41 (called hereafter mgp41) was trimeric in solution. Circular dichroism measurements indicated that around 66% of the mgp41 structure was in alpha-helices and that its thermal stability was strongly reduced compared to the native gp41 ectodomain (Figure S1 in Supplementary Material) (28), suggesting a destabilization of the six-helix bundle. Accessibility of NHR epitopes was confirmed by the significant binding of an exogenous CHR peptide to mgp41 (gp41 sequence 110-141) (29) (Figure S1 in Supplementary Material). The mgp41 polypeptide was chemically conjugated to CTB using a strategy allowing the coupling of one mgp41 per CTB pentamer as illustrated in Figure 1B.

The immunogenicity of the SL CTB-mgp41 vaccine was evaluated in female cynomolgus macaques. Immunohistochemistry analysis confirmed that the macaque SL mucosa was an immunocompetent tissue with a pluristratified epithelium underlined by a thin submucosa containing MHCII⁺ antigen-presenting cells, some of them being CD1a⁺ dendritic cells (DCs) (Figures 2A,B). We compared the efficacy of the SL route of vaccination to the IM immunization and to a combined strategy including SL priming followed by IM boost (SL/IM) to generate serum and mucosal IgG and IgA antibodies against mgp41. As illustrated in Figure 2C, six macaques (group 1, SL) were immunized sublingually with CTB-mgp41 together with CT used as mucosal adjuvant. Six macaques (group 2, SL/IM) received a SL priming similarly to group 1 followed by IM boosts. In order to evaluate the impact of the SL priming in the combined SL/IM vaccination, the six macaques from group 3 (IM) were primed with CT alone by SL immunization and received IM boosts with mgp41 in Alum. Furthermore, macaques from group 3 received a third IM immunization in order to compare SL and SL/IM strategies to optimized conditions for IM vaccination.

To compare the potential of SL immunization to classical IM vaccination, we quantified the mgp41-specific antibodies of IgG and IgA isotypes generated throughout the protocol. Antigen-specific antibodies were measured in sera (Figure 3) and in vaginal and rectal washes (Figure 4) at the indicated time points. The macaques that received five SL immunizations had notable serum mgp41-specific IgG titers up to 2 months after the last immunization (Figure 3A, left panel; Figure 3B, upper panel). Serum mgp41-specific IgG titers reached a significant value after the third SL immunization but the titers did not increase with subsequent SL immunizations (Figure 3A, left panel), suggesting that repeated immunizations with 1-month intervals reduced antigen uptake. In contrast, boost IM immunizations in the SL/IM strategy increased serum mgp41-specific IgG titers by around 10-fold compared to SL immunization (Figure 3A). Interestingly, SL/IM immunization gave significantly higher mgp41-specific IgG antibody levels than the IM strategy 2 weeks and 1 month after the fifth immunization, indicating that SL priming potentiated antibody responses (Figure 3A). No statistical differences in serum mgp41-specific IgA titers were detected between immunization groups (Figure 3).

Vaginal and rectal antibody levels induced by the different vaccine strategies were low (Figure 4). All the macaques had mgp41-specific IgG in vaginal washes (Figure 4A), suggesting that these antibodies were produced systemically. Accordingly, rare mgp41-specific IgG ASCs were detected in the genital tract of macaques (data not shown). Interestingly, the SL/IM immunization induced significantly higher levels of gp41-specific IgG antibodies compared to the IM immunization 2 weeks after the fourth and fifth immunizations both in the genital tract and the rectal mucosa (Figures 4A,C). Furthermore, mgp41-specific IgA antibodies were detected in the genital tract and in the rectal mucosa after SL and SL/IM vaccination, but not after IM immunization (Figures 4B,D). Of note, the individuals who developed vaginal responses were the same who developed rectal responses (data not shown). Furthermore, vaginal gp41-specific IgA antibodies detected in SL and SL/IM groups were associated with higher numbers of vagina-associated mgp41-specific IgA ASC in these groups compared to IM immunization (SL: 7 ± 3 gp41-specific IgA ASC; SL/IM: 8 ± 2 gp41-specific IgA ASC; IM: 4 ± 2 gp41-specific IgA ASC).

To further analyze the antibody responses induced after SL and SL/IM immunizations compared to IM immunization, the frequency of circulating mgp41-specific ASC of IgG and IgA isotypes was determined 5 days postimmunization after priming (W9), boost (W21), and third IM immunizations (W29 only for IM group) (Figure 5). Variable numbers of mgp41-specific IgG ASC between macaques of a same group were detected after the priming with no significant differences between groups. The macaques from IM group had significantly higher numbers of circulating IgG ASC after the third IM immunization (week 29) compared to the other groups (Figure 5A), which can be explained by the increased bioavailability of the antigen after IM immunization. In contrast, both SL and SL/IM immunizations induced higher numbers of mgp41-specific IgA ASC after the last immunization compared to the IM group (Figure 5B), confirming in macaques the importance of the route of immunization to induce antibody responses of IgA isotype.

To further evaluate the persistence of the gp41-specific responses generated by the SL mgp41-CTB vaccine, circulating gp41-specific ASC were also quantified 1 month and 3 months after the last immunization after in vitro stimulation in order to enumerate antigen-specific memory B cells. As shown in Figure 6A, the macaques of the IM group had significantly higher numbers of circulating IgG ASC 3 months after the third IM immunization compared to other groups (Figure 6A). In contrast, there seemed to be a trend toward higher numbers of mgp41-specific ASC of IgA isotype 1 month after SL/IM immunization compared to IM immunization as indicated by a p-Value of 0.08 using the non-parametric Mann–Whitney U-test (Figure 6B), consistent with the observation that SL/IM immunization, but not IM immunization generated mucosal gp41-specific IgA antibodies (Figure 4). We also determined the frequency of total CD20⁺CD27⁺ IgD⁻ memory B cells in the blood and BM at the end of the protocol by flow cytometric analysis. SL and combined SL/IM immunizations strongly increased the frequency of memory B cells both in blood and BM compared to IM immunization (Figures 6C,D).

Altogether, these data indicated that the SL immunization increases the frequency and the duration of IgA antibody responses.

To further evaluate the functional characteristics of the antibodies produced by SL immunization compared to antibodies generated after IM immunization, we determined their neutralizing activity against clade B virus primary isolates. The neutralizing activity of sera was performed using a TZM-bl assay with the QHO692.42 virus primary isolate, a clade B transmitted founder virus classified as Tier 2. Sera collected 2 weeks (W22) and 3 months (W32 for SL and SL/IM groups and W40 for IM group) after the last immunization were tested at several dilutions. Pre-immune sera from each macaque were used as control (Figure 7). Dose–response curves indicated that a modest but antigen-specific neutralizing activity was detected in the three immunization groups at week 22 compared to pre-immune samples (Figure 7A). Interestingly, immune sera from SL and SL/IM immunization groups had persistent neutralizing activity 3 months postimmunization contrary to the IM group. Analysis of the sera of individual macaques at the lowest dilution confirmed that the SL priming with the mgp41-CTB antigen improved the functional characteristics of the antibodies produced (Figure 7B). These data indicated that the SL immunization favors the generation and persistence of functional antibodies.