Modified Vaccinia Virus Ankara Vector Induces Specific Cellular and Humoral Responses in the Female Reproductive Tract, the Main HIV Portal of Entry Free

Heterosexual intercourse is the major route of HIV-1 transmission (1), and viral entry occurs mainly via the female reproductive tract (FRT) mucosae. One of the attempts to prevent this transmission should focus on inducing mucosal immune responses.

The FRT contains two types of mucosae. The type I mucosal surface is found in the upper genital tract (endocervix, uterus, and Fallopian tubes) and is covered by a monolayer of columnar epithelial cells with tight junctions. The type II mucosal surface is found in the lower genital tract (vagina and ectocervix) and is lined with a stratified squamous epithelium. The boundary between the type I and II mucosae is called the cervical transformation zone. The transformation zone is considered to be more vulnerable to HIV-1 infection (2, 3), owing to the abundance of immune target cells (CD4⁺ T cells, macrophages, and dendritic cells) and the transition of the epithelial phenotype. Macrophages and T cells from the vagina and cervix are permissive to HIV-1 infection in vitro (4, 5). The uterus contains CD4⁺ T cells and macrophages that express HIV-1 coreceptors. Uterine cells and uterine explants are also permissive to HIV-1 infection in vitro (6). Thus, because HIV-1 target cells are present throughout the FRT, an effective vaccine should induce protective responses in all FRT compartments.

Recombinant poxviruses, such as vaccinia virus and canarypox virus, are strongly immunogenic and are often used as vaccine vectors. Partial, but significant, protection against HIV-1 was observed in the Thai phase III trial (RV144) of a canarypox vector expressing HIV-1 Ags plus a trimeric recombinant gp120 protein, used in a prime/boost strategy (7). The modified vaccinia virus Ankara (MVA), another recombinant poxvirus, is widely used in vaccines for infectious diseases (8). Mucosal immunity after MVA vaccination has been studied in the gastrointestinal tract but not in the FRT (9).

Most studies of vaccine responses in the FRT have been performed in mice (10). However, macaques, which exhibit marked immunological and anatomical similarities to humans in contrast to mice (11), are the reference model for HIV-1 research and vaccine studies.

In this study, using an MVA vaccine as a model, we examined whether systemic MVA vaccination induced specific local responses in the macaque FRT. After detailed phenotypic characterization of immune cell subpopulations, we examined specific immune responses in the blood, lymph nodes (LNs), vagina, cervix (endocervix and ectocervix), uterus, and fallopian tubes.

Six sexually mature adult female cynomolgus macaques (Macaca fascicularis) imported from Mauritius were housed in the Infectious Disease Models and Innovative Therapies facilities at Commissariat à l’Energie Atomique et aux Energies Alternatives (CEA). Treatment of nonhuman primates (NHPs) at CEA complies with French national regulations (authorization A 92-032-02; CEA), with the Standards for Human Care and Use of Laboratory Animals (assurance number A5826-01; Office of Laboratory Animal Welfare), and with European Directive 2010/63 (recommendation number 9). Experiments were supervised by veterinarians in charge of the animal facility. This study was approved and accredited by the Comité d’Ethique en Expérimentation Animale of the CEA (A14-080) and by the French Research Ministry. Animals were housed in pairs under controlled conditions of humidity, temperature, and light (12-h light/dark cycles). Water was available ad libitum. The animals were monitored and fed once or twice a day with commercial monkey chow and fruits, by trained personnel, and were provided with environmental enrichment, including toys, novel foodstuffs, and music, under the supervision of the CEA Animal Welfare Officer.

On days 0 and 58, the macaques received two s.c. injections per time point in the right and left side of the upper back, delivering 2 × 1 ml of inoculum containing a total of 4 × 10⁸ PFU recombinant MVA–HIV-1 expressing the Gag, Pol, and Nef proteins from HIV-1 strain LAI (ANRS-MVA HIV-B, MVATG17401; Transgene, Illkirch-Graffenstaden, France). The animals were monitored daily for signs of disease, appetite loss, and lethargy. A physical examination was performed at each blood sampling and each inoculation. All experimental procedures (handling, immunization, blood sampling) were conducted after sedation with ketamine hydrochloride (10 mg/kg; Rhône-Mérieux, Lyon, France). To synchronize their hormonal cycle, an i.m. injection of a synthetic variant of progesterone (Depo-Provera, 30 mg; Pfizer, Paris, France) was given 42 d after the first vaccine injection. The animals were sedated 77 d after the first vaccine injection with ketamine hydrochloride (10 mg/kg) and then euthanized by i.v. injection of 180 mg/kg sodium pentobarbital.

Blood, serum, and vaginal fluid were collected before and after each vaccine inoculation and at the time of euthanasia. Lymph nodes (LNs) and tissues were collected at necropsy. Serum was isolated by centrifugation at 3000 rpm for 10 min and stored at −80°C. Cervicovaginal fluid was collected with a WECK-CEL spear (Medtronic) placed in the vaginal vault for 2 min. Secretions were recovered from the spears by adding 600 μl of extraction buffer (PBS, NaCl 0.25 M, and protease inhibitor mixture; Merck Millipore, Fontenay-sous-Bois, France) and centrifuging at 13 000 × g for 20 min. Filtered vaginal fluids were stored at −80°C.

PBMCs were isolated in heparin CPT tubes (BD Biosciences, Le Pont-de-Claix, France) after centrifugation for 30 min at 3000 rpm. PBMCs were collected from the top of the CPT gel surface and washed twice. At euthanasia, LNs and FRT tissues were collected. LN cells were obtained by mechanical dissociation. FRT tissues (vagina, cervix, uterus, and fallopian tubes) were isolated and cut into small pieces. Each tissue was digested for 1 h at 37°C with agitation in digestion buffer, consisting of RPMI 1640 (Fisher Scientific, Illkirch, France), collagenase IV (0.3 mg/ml; Sigma-Aldrich, Saint-Quentin-Fallavier, France), FCS (5%; Fisher Scientific), HEPES (0.025 M; Fisher Scientific), DNase (0.1 mg/ml; Roche, Mannheim, Germany), and antibiotics (Fisher Scientific). Undigested pieces were subjected to up to three more digestion steps. Cell suspensions from LNs and FRT tissues were filtered through 70-μm sterile nylon cell strainers (BD Biosciences). The median number of cells recovered from each FRT compartment was 14.2 × 10⁶ cells per gram of tissue (i.e., 64 × 10⁶ cells) in the vagina, 13.8 × 10⁶ per gram of tissue (i.e., 54 × 10⁶ cells) in the cervix, 15 × 10⁶ cells per gram of tissue (i.e., 62 × 10⁶ cells) in the uterus, and 27.7 × 10⁶ cells per gram of tissue (i.e., 18 × 10⁶ cells) in the fallopian tubes.

Whole blood, LN cells, and cells from FRT compartments were analyzed by flow cytometry. The cells were incubated with the Abs listed in Supplemental Table I and then washed and fixed with FACS lysing buffer or BD Biosciences Cell Fix solution. A Fortessa 2-UV 6-Violet 2-Blue 5-YelGr 3-Red laser configuration was used (BD Biosciences), with Diva (BD) and FlowJo 9.8.3 (TreeStar) software. At least 500 events for rare cell populations (i.e., plasmacytoid dendritic cells [pDCs]) were recorded. The gating strategies are illustrated in Supplemental Fig. 1.

Specific cellular immune responses were evaluated with in vitro stimulation assays. The cells were incubated for 5 h at 37°C with medium, 0.3 PFU per cell of live wild-type (wt) MVA, or PMA (5 ng/ml) and ionomycin (500 ng/ml) (Sigma-Aldrich) in DMEM (Fisher Scientific) supplemented with 10% FCS and antibiotics. Brefeldin A (5 μg/ml; Sigma-Aldrich) was added, and the cells were incubated for an additional 10 h at 37°C. For HIV-1 stimulation, cells were incubated with 4 μg/ml overlapping Gag peptide pools in DMEM supplemented with 10% FCS, antibiotics, and costimulatory Abs for 1 h at 37°C and then for an additional 4 h with Brefeldin A (5 μg/ml). The cells were stained with blue dye (LIVE/DEAD Fixable Blue Dead Cell Stain; Thermo Fisher) for viability and then fixed and permeabilized with BD Cytofix/Cytoperm reagent (BD Bioscience). The Abs listed in Supplemental Table II were used for intracellular staining. At least 5000 events in the CD8⁺ T cell gate were recorded. The gating strategy was as described elsewhere (12). Briefly, expression of cytokines and activation markers was evaluated in CD4⁺ and CD8⁺ T cells, and Boolean gate analyses were performed with FlowJo software. The percentages of cells positive for cytokines and activation markers were then compared between unstimulated and MVA- or Gag peptide pool–stimulated cells. The immune response was considered positive against the Ag when the percentage of cells positive for cytokines and activation markers was at least twice as great as the percentage under unstimulated condition.

Specific Abs were measured by enzyme immunoassay in serum and vaginal fluid collected by WECK-CEL spears. Ninety-six–well MaxiSorp microplates (Nunc; Thermo Fisher) were coated overnight with 10⁵ PFU per well of wt MVA (Transgene, Illkirch, France) in NaHCO₃/Na₂CO₃ buffer or with 1 μg/ml p24 Ag (a kind gift from B. Verrier, Laboratoire de Biologie Tissulaire et d'ingénierie Thérapeutique UMR 5305) in PBS. The plates were then blocked for 1 h with PBS containing 3% (w/v) BSA (Sigma-Aldrich) or with PBS containing 10% skimmed milk. The plates were washed five times with PBS containing 0.1% Tween 20 and 10 mM EDTA and then incubated with 2-fold serial dilutions of macaque fluids diluted in PBS containing 1% BSA for 1 h at room temperature (to detect anti-MVA IgG/IgA) or in PBS containing 1% skimmed milk and 0.05% Tween 20 for 1 h at 37°C (to detect anti-HIV IgG), starting at 1:50 for serum and 1:20 for vaginal fluid. The plates were washed five times and incubated for 1 h with a 1:20,000 dilution of HRP-conjugated goat anti-monkey H+L chain IgG (Bio-Rad, Marne-la-Coquette, France) or with a 1:5,000 dilution of HRP-conjugated goat anti-monkey IgA (Alpha Diagnostic, San Antonio, TX). The plates were washed five times and then 100 μl of O-phenylenediamine dihydrochloride (Sigma-Aldrich) was added and incubated for 30 min at room temperature in the dark. The reaction was stopped by adding 2 N H₂SO₄. Absorbance was measured at 492 nm with a spectrophotometer, and data were analyzed with Magellan software (both from Tecan, Lyon, France). Ab titers were calculated by extrapolation from the OD as a function of a serum-dilution curve and were defined as the dilution of the test serum reaching 2 OD of the corresponding preimmune serum or vaginal fluid, tested at 1:50 and 1:30, respectively.

Tissue biopsies collected at euthanasia were immediately immersed in RLT–2-ME 1/100 lysis buffer (QIAGEN, Courtaboeuf, France) and then disrupted and homogenized with a TissueLyser LT (QIAGEN, Redwood City, CA). RNA was purified with a QIAGEN RNeasy Micro Kit. Contaminating DNA was removed using the RNA Cleanup step of the RNeasy Micro Kit. Purified RNA was quantified with a ND-8000 spectrophotometer (NanoDrop Technologies, Fisher Scientific) before being checked for integrity on a 2100 Bioanalyzer (Agilent Technologies, Massy, France). cDNA was synthesized and biotin labeled using the Ambion Illumina TotalPrep RNA Amplification Kit (Applied Biosystems/Ambion, Saint-Aubin, France). Labeled cRNA was hybridized on Illumina Human HT-12 v4 BeadChips, which target 47,323 probes corresponding to 34,694 genes. The manufacturers’ protocols were followed.

Microarray data were analyzed with R/Bioconductor software. Gene expression values were quantile normalized. Differentially expressed genes were identified with a paired nonparametric t test (q-value < 0.05), based on a fold-change cutoff of 1.2. Functional enrichment analysis used QIAGEN’s Ingenuity Pathway Analysis https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/). Hierarchical clustering presented in the heat maps was generated with the Euclidian metric and complete linkage methods. Microarray raw data have been submitted to the European Bioinformatics Institute ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-5663. The transcriptomic and cellular coexpression network was generated using the Spearman correlation coefficient, based on the abundance of cell populations and normalized gene expression values across the entire dataset. Significant correlations (R > +0.70 and p value < 0.01) were restricted to correlations between cell populations and between cell populations and gene expression levels.

To characterize the vaccine responses at the mucosal level, an extensive identification of the cell subpopulations present in the different compartments was performed. The phenotypes of immune cells collected from FRT compartments and LNs were analyzed by flow cytometry. The proportions of immune cell subsets were characterized in four sites of the FRT (vagina, cervix, uterus, and fallopian tubes) by comparison with whole blood and proximal and distal LNs (iliac, axillary, and inguinal). Leukocytes were present in all FRT compartments (Fig. 1A).

Two subpopulations of innate lymphoid cells (ILCs) were identified on the basis of NKG2A (NK cells) and NKp44 (ILC-3) expression (Supplemental Fig. 1B). NK cells accounted for <4.0 ± 2.0% (median ± SD) of leukocytes in blood and LNs. This percentage was higher in the FRT compartments, especially in the uterus (13.9 ± 5.2%) (Fig. 1B). The NK phenotype differed between the mucosae and blood; mucosal NK cells did not express CD16 FcγR but did express CD69. In the LNs and FRT, NKp44 was expressed by ILC-3 cells but not by NK cells (Fig. 1C, Supplemental Fig. 1B). Although few in number, ILC-3 cells were found in all LN and FRT compartments and preferentially in the fallopian tubes (0.5 ± 0.4%) (Fig. 1D).

The distribution of three main APC populations (CD14⁺ APCs, CD123⁺ pDCs, and CD11c⁺ myeloid dendritic cells [mDCs]) is shown in Supplemental Fig. 1C, according to the gating strategy.

CD14⁺ APCs were the main APC subtype in all FRT compartments (Fig. 1E) (1.5 ± 1.0–5.3 ± 5.6% of total leukocytes) and were located primarily in the uterus. The proportions of these cells were similar in the lower FRT and blood, but only mucosal CD14⁺ cells expressed the activation marker CD69 and the FcγR CD16, particularly in the vagina (Fig. 1F). The distribution of mDCs (CD11c⁺) was similar to that of CD14⁺ cells, and mDCs were primarily found in the uterus (3.4 ± 2.2% of total leukocytes) (Fig. 1G). The proportion of pDCs (CD123⁺) was similar in all compartments, where they represented <1% of total leukocytes (Fig. 1H).

Neutrophils were the main leukocyte subtype in blood (43.8 ± 14.1% of CD45⁺ cells). In the FRT, neutrophils were primarily found in the cervix and upper compartments (8.9 ± 6.8% in cervix, 3.4 ± 1.8% in uterus, and 2.3% ± 4.0 in the fallopian tubes) (Fig. 1I).

T lymphocytes were the main leukocyte subpopulation in the LN and FRT compartments. The percentage of CD8⁺ T cells in the FRT mucosae was higher than the percentage of CD4⁺ T cells, in contrast with blood and LNs (Fig. 2A, 2D). CD4⁺ T cells represented ∼50% of total leukocytes in LNs, nearly 20% in the lower FRT, and 10% in the upper FRT (uterus and fallopian tubes) (Fig. 2A). The T cell memory phenotype was defined by CD28, CD95, and CD45RA expression (Supplemental Fig. 1D). In contrast to blood and LNs, the majority of CD4⁺ T cells in the FRT exhibited the central memory phenotype (CD28⁺/CD95⁺), whereas naive cells (CD28⁺/CD95^neg) were rare (Fig. 2B). In the FRT, up to 40% of leukocytes were CD8⁺ T cells (Fig. 2D). These cells expressed markers of central memory (CD28⁺/CD95⁺) and effector memory (CD28^neg/CD95⁺), in contrast to blood and LNs, in which most CD8⁺ T cells were naive (Fig. 2E). Mucosal CD4⁺ and CD8⁺ T cells frequently expressed CD69 (Fig. 2C, 2F).

Between 6.7 and 15.9% of blood and LN leukocytes were B cells (CD20⁺) (Supplemental Fig. 1D), whereas B cells were infrequent in all FRT compartments (0.1 ± 0.1 to 1.0 ± 1.4% of leukocytes) (Fig. 2G).

Thus, cells involved in initiating immune responses, as well as effector cells, were present throughout the macaque FRT, with specific distribution according to the compartment.

The anti-MVA T cell response was monitored in blood using an in vitro restimulation assay. Ag-specific CD4⁺ T cells were identified as CD154⁺ cells. Their percentage increased in blood 2 wk after the first and second vaccine injections (mean, 0.84 and 0.85% among total CD4⁺ T cells, respectively) (Fig. 3A). Three weeks after the second vaccine injection, the anti-MVA response was analyzed in all compartments. Ag-specific CD4⁺ T cells were significantly detected in PBMCs and LNs (Fig. 3B). The largest percentage of MVA-specific CD4⁺ T cells was found in the axillary LNs (0.23–3.24% of total CD4⁺ T cells). No MVA-specific CD4⁺ T cell response was detected in FRT tissue, because the percentage of CD154⁺ CD4⁺ T cells did not change significantly after MVA restimulation (Fig. 3B). In all compartments, CD154⁺ cells represented a large percentage of CD4⁺ T cells after stimulation with PMA and ionomycin (data not shown). The anti–HIV-1 response was also analyzed after in vitro stimulation with Gag peptide pools and costimulatory Abs (anti-CD28 and anti-CD49d mAbs). The addition of costimulatory Abs induced nonspecific activation of T cells and, thus, increased background CD154 expression, even in nonstimulated conditions. Because this could have masked weak responses, we measured the HIV-1–specific CD4⁺ T cell response by analyzing the percentage of CD4⁺ T cells that expressed CD154 and produced IFN-γ (i.e., only specific T cells). The Gag-specific CD4⁺ T cell response was very weak and was only detected in PBMCs (Fig. 3C).

Thus, MVA-specific CD4⁺ T cells were primarily found in blood and in LNs draining the vaccine inoculation site.

Like the CD4⁺ T cell response, the vaccine response mediated by CD8⁺ T cells was monitored by in vitro restimulation assays in blood over time, as well as in all compartments at euthanasia. The percentage of MVA-specific CD8⁺ T cells that produced IFN-γ increased in blood after the first vaccine inoculation and increased sharply after the second injection (mean, 1.79 and 5.38% of total CD8⁺ T cells, respectively) (Fig. 4A). A similar profile was observed for MIP-1β and TNF-α production by Ag-specific CD8⁺ T cells over time (data not shown). MVA-specific CD8⁺ T cells were detected in PBMCs and LNs of all animals 3 wk after the second vaccine injection. They represented 0.15–7.16% of total CD8⁺ T cells, depending on the animal (Fig. 4B, 4C). Interestingly, MVA-specific CD8⁺ T cell responses were also detected in all FRT compartments (Fig. 4B, 4D), and especially in the vagina (0.04–1.08% of total CD8⁺ T cells for the IFN-γ response). The anti-MVA response mediated by CD8⁺ T cells was polyfunctional. Importantly, most of the CD8⁺ T cells that had only one function (produced one cytokine/chemokine) secreted MIP-1β, whereas those with two functions produced MIP-1β and IFN-γ, and those with three functions produced MIP-1β, IFN-γ, and TNF-α (Fig. 4E). MVA-specific CD8⁺ T cells from the FRT produced less TNF-α than did their blood and LN counterparts (Fig. 4C–E). 

HIV-1 Ag-specific CD8⁺ T cells were also measured in the different compartments. The background signal was high in all conditions, as noted for CD4⁺ T cell responses. To detect HIV-1 Ag-specific CD8⁺ T cells, the analyses focused on CD8⁺ T cells that produced both MIP-1β and IFN-γ. Anti-Gag CD8⁺ T cells were detected above background in one animal’s PBMCs (triangle), LNs, and uterus (Fig. 4F).

Thus, MVA-specific CD8⁺ T cell responses were polyfunctional and found in all FRT compartments, in addition to blood and LNs.

To analyze humoral responses, vaccine-specific Ig titers were serially determined in serum and vaginal fluid by ELISA. MVA-specific IgG was detected in all of the animals’ sera 2 wk after the first and second vaccine inoculations (1,216 ± 430 and 50,079 ± 11,780, respectively; mean titer of six animals) (Fig. 5A) and in vaginal fluid only after the second inoculation (634 ± 386, mean titer of six animals) (Fig. 5B). Similarly, MVA-specific IgA was detected in serum after the two vaccine inoculations (282 ± 250 and 1960 ± 1532, respectively; mean titer of six animals) and in vaginal fluid only after the second inoculation (titer of 231 ± 132, mean of six animals) (Fig. 5C, 5D). The anti–HIV-1 humoral response was estimated by measuring Gag-specific IgG (Fig. 5E, 5F). These Abs were detected in the serum of four of six animals after the first vaccine inoculation and in all six animals’ serum after the second inoculation. In contrast, Gag-specific IgG was not detected in vaginal fluid (Fig. 5F).

Thus, systemic MVA vaccination induced detectable vector-specific IgG and IgA in vaginal fluid after the second vaccine injection.

To better characterize the events at the molecular level, we compared the transcriptomic profiles from vaginal, cervical, and uterine tissue samples. The numbers of differentially expressed genes (DEG) in each comparison are represented in Fig. 6A. We found that 3810 and 4800 genes were differentially expressed in the cervix and uterus compared with the vagina, respectively. We identified 3804 DEGs in the uterus versus the cervix. The Venn diagram in Fig. 6B represents the common DEGs between the comparisons, showing that 624 genes were shared between the three comparisons (i.e., cervix versus vagina, uterus versus vagina, and uterus versus cervix). The relative expression of genes identified as DEGs in at least one comparison is represented by the heat map in Fig. 6C. Four main branches (gene clusters) were identified by hierarchical clustering. The clustering branch 2 was primarily driven by DEGs overexpressed in the uterus. Similarly, branch 1 was driven by DEGs overexpressed in the vagina, whereas branches 3 and 4 were driven by DEGs overexpressed in the cervix. Each gene set was analyzed using Ingenuity Pathway Analysis. Branches 3 and 4 were merged for the enrichment analysis. Functional enrichment analysis of canonical pathways and upstream regulators are represented on the side of each branch. Upstream regulator analyses highlighted that β-estradiol (branch 1 p = 9.69 × 10⁻³⁰; branch 2 p = 2.7 × 10⁻²¹; branches 3 and 4 p = 8.39 × 10⁻⁸), progesterone (branches 3 and 4 p = 1.16 × 10⁻¹⁰), and estrogen receptor (ESR1) (branch 2 p = 8.77 × 10⁻¹⁷; branches 3 and 4 p = 1.45 × 10⁻¹⁰) constituted top upstream regulators in at least one of the four branches, confirming that hormones are involved in central regulation pathways in the FRT.

Enrichment analyses were then filtered to reveal immune-related pathways. Significant immune-related canonical pathways and associated p values are shown in Fig. 6D. The results showed that numerous pathways are associated with the uterus (branch 2) and are linked to NK cells and APCs (NK cell signaling p = 8.51 × 10⁻⁴; IL-15 signaling p = 2.04 × 10⁻³; cross-talk between dendritic cells and NK cells p = 2.04 × 10⁻³; production of NO and reactive oxygen species by macrophages p = 9.77 × 10⁻⁴). Similarly, in the vagina (branch 1), immune canonical pathways are associated with macrophages/monocytes and T cells (CTLA4 signaling in cytotoxic T cell p = 2 × 10⁻⁴; TCR signaling p = 4.47 × 10⁻³; FcγR-mediated phagocytosis in macrophages and monocytes p = 9.12 × 10⁻⁴; CCR5 signaling in macrophages p = 4.57 × 10⁻³). No significant immune-related pathway was identified for the cervix (clustering branches 3 and 4).

To integrate flow cytometry and transcriptomic data, we generated a coexpression network. We restricted the correlations to DEGs associated with the pathways in bold type in Fig. 6D. The coexpression network revealed that frequent immune populations in the vagina (T and B cells) correlated positively with these branch 1 immune pathways and negatively with branch 2 immune pathways (Fig. 6E). Conversely, frequent immune populations in the uterus (APCs and ILCs) correlated positively with branch 2 immune pathways and negatively with the branch 1 immune pathways.

Thus, transcriptome and correlation analyses highlighted the specificity of each FRT compartment and the compartmentalization of FRT immune cells.

Because male-to-female transmission via the FRT mucosae is the main route of HIV-1 transmission, it is essential to study vaccine responses in the FRT. We conducted a detailed characterization of the immune cells involved in MVA–HIV-1 vaccine responses in the cynomolgus macaque, as well as the vaccine responses themselves in all FRT compartments during the luteal phase, by comparison with blood and draining LNs. Previous studies of mucosal responses to MVA vaccination have been limited to the gastrointestinal tract (9), whereas the FRT is the main portal of entry for sexually transmitted pathogens. To our knowledge, we show for the first time that s.c. MVA injections induce specific Ig (IgG and IgA) and polyfunctional CD8⁺ T cells in the FRT of macaques. This study reveals that each FRT compartment has its own characteristics, as shown by immune cell phenotyping and transcriptomic analyses.

The first part of this study clearly shows that immune cells are compartmentalized. Two subpopulations of ILCs were identified according to their phenotypes. Thus, NK cells, defined by NKG2A expression, were primarily found in the upper FRT and expressed low levels of CD16 FcγR, as previously described (13), and no NKp44, in contrast to gut mucosae (14) (M. Cavarelli, personal communication). Localization of NK cell activity within the uterus was confirmed by transcriptomic analyses, because NK cell–related pathways were associated with the uterus (branch 2) and correlated with uterine immune cell populations (coexpression network, Fig. 6E). A second subtype of ILCs, called ILC-3, which expressed NKp44 but not NKG2A (14), were also found primarily in the upper FRT.

We identified three main populations of professional APCs, based on their phenotypic markers: CD14⁺ APCs, mDCs, which expressed CD11c⁺, and pDCs, which expressed CD123⁺. These APCs were distributed throughout the FRT, but differences in their distribution and phenotype were noted. The largest percentage of mDC CD11c⁺ cells and CD14⁺ cells was found in the uterus. Moreover, phenotypic analyses of APC subtypes confirmed that CD14⁺ APCs from the vagina expressed CD16 FcγR, contrary to intestinal CD14⁺ APCs (15). This characteristic of vaginal CD14 APCs was supported by transcriptomic analysis, because the “FcγR-mediated phagocytosis in macrophages” pathway was enriched in the vagina.

B lymphocytes were the main leukocyte subtype in LNs, whereas they were detected at very low percentages throughout the FRT, primarily in the vagina. This low frequency of B cells in the FRT was not due to the enzymatic digestion procedure, because very few B cells were detected by immunostaining of tissue biopsies (data not shown). In this study, we detected B cells by their CD20 expression rather than the CD19 marker that is usually used in humans, because the available Abs do not cross-react in cynomolgus macaques. Because CD20 is not expressed by all B cell subsets (16), this could explain the low percentage of B cells found in our samples. The distribution of CD20^neg B cells, such as plasma cells and Ab-secreting cells, in FRT compartments will require further studies with specific markers.

T lymphocytes were the main immune cell populations in all FRT compartments. CD8⁺ T cells were more abundant than CD4⁺ T cells within the mucosae, in contrast to blood and LNs. In particular, the vagina exhibited the largest percentage of CD8⁺ T cells, as well as a specific transcriptomic signature related to T cell pathways. Coexpression networking showed that T cell abundance correlated positively with these T cell pathways (branch 1) (Fig. 6E). We confirm that CD4⁺ and CD8⁺ T cells express memory markers in the FRT (17), whereas naive cells were primarily found in blood and LNs. Among the memory T cells, resident memory lymphocytes have been described in tissues such as the vagina (18). Our analyses performed in one vaccinated animal showed that CD8⁺ resident memory T cells were primarily present in the vagina (39.8% among effector memory T cells) and in the cervix (22.2% among effector memory T cells) and were less frequent in the uterus (2.5% among effector memory T cells) (data not shown). Together, these data show that immune cells exhibit tissue specificity in the macaque FRT.

Our results match some published data on the human FRT, including the observations that mucosal NK cells exhibit a unique phenotype and are primarily found in the uterus (19); ILC-3 are detected in the upper FRT (20); APCs are distributed throughout the FRT, particularly in the upper tract (21) and ectocervix (22); CD20⁺ B cells are rare throughout the FRT (23); and memory CD8⁺ T cells represent a large proportion of immune cells in all compartments of the human FRT. Thus, the localization and phenotype of immune cell subtypes are similar in the macaque and human FRT, validating the cynomolgus macaque as a model for human reproductive biology and genital immunity, including FRT mucosal immune responses to vaccination.

Therefore, we vaccinated female cynomolgus macaques s.c. with MVA–HIV-1, selected as a vaccine model, and analyzed specific responses in the FRT, LNs, and blood.

Analyses of humoral responses confirmed that MVA vaccination induces strong anti-MVA IgG responses in serum, whereas anti-HIV IgG was not detected in all animals. In contrast to other mucosal fluids, cervicovaginal fluid contained more IgG than IgA. MVA-specific IgG and IgA were detected in vaginal fluid after the second vaccine injection. Local anti-MVA IgG titers were lower than in serum, because the vaccine was administered s.c. Cervicovaginal IgG has been shown to originate primarily from the systemic compartment (24). Anti-HIV IgG titers in serum were much lower than anti-MVA IgG titers, which could explain why anti-HIV IgG was not detected in vaginal fluid, given their systemic origin.

Anti-MVA responses mediated by CD4⁺ T cells were weaker than those mediated by CD8⁺ T cells in blood and LNs, and they were not detectable in FRT mucosae. Because the peak of CD4⁺ T cell responses precedes the peak of CD8⁺ T cell responses (25), immune responses analyzed 77 d after the prime may not be optimal to detect a strong CD4⁺ T cell response. The largest anti-MVA CD4⁺ T cell responses were measured in axillary LNs (i.e., those draining the vaccine injection site [upper back]). However, MVA-HIV vaccination may induce nonspecific CD4⁺ T cell activation, because CD4⁺ T cells exhibited high CD154 expression in some animals in unstimulated conditions.

Our study clearly shows that MVA vaccination induces strong specific CD8⁺ T cell responses in blood, LNs, and all FRT compartments. In the FRT, they were primarily localized in the vagina, but specific responses were also detected in the cervix, uterus, and fallopian tubes. Responses mediated by CD8⁺ T cells were polyfunctional, because specific CD8⁺ T cells positive for two or more cytokines were detected. Boolean gating analyses were used to sort specific CD8⁺ T cells according to the number of functions that they displayed (single, double, triple, or quadruple cytokine producers). The majority of single-producer cells were MIP-1β⁺, double producers were MIP-1β⁺IFN-γ⁺, and triple producers were MIP-1β⁺IFN-γ⁺TNF-α⁺. These findings correspond to reports of blood CD8⁺ T cell responses (26). Thus, TNF-α and IL-2 are produced only by highly polyfunctional cells. We noted that the percentage of triple-producer cells was lower in the FRT than in blood and LNs. Together, these results demonstrate that TNF-α⁺-specific CD8⁺ T cells are less abundant in the FRT than in blood and LNs.

A previous study has demonstrated that the MVA–HIV-1 vaccine induced T cell responses primarily against Gag and Pol peptides (27). Because of the limited amount of mucosal cells recovered to perform the different experiments and Ag stimulations, cellular and humoral immune responses induced against the HIV-1 insert in our present study were focused on anti-Gag responses. HIV-1 (Gag)-specific responses were mediated by CD4⁺ T cells and were detected primarily in blood. These responses were weaker than MVA-specific CD8⁺ T cell responses and were not detectable in the FRT, with the exception of the uterus from one animal. The MVA–HIV-1 vaccine was used in this study as a model, and animals were vaccinated s.c. with two injections of the same vaccine construct. Therefore, specific responses primarily targeted the immunogenic vector. To enhance insert-specific responses, it will be essential for the boost or the prime to use another type of vaccine construct, such as a DNA vaccine, in addition to MVA (27).

The local environment of the FRT mucosae is under the influence of several factors, including hormones during the menstrual cycle, semen during intercourse, and sexually transmitted pathogens (28–30). Because these factors impact mucosal immune cells and their environment, it will be crucial to study their possible influence on mucosal vaccine responses.

We thank Dr. Anne-Sophie Beignon, Dr. Antonio Cosma, Dr. Mireille Centlivre, and the B Cell and Mucosae division members for scientific discussions, D. Young for critical editing of the manuscript, Transgene for providing the wt MVA strain, l'Agence nationale de recherche sur le sida et les hépatites virales /INSERM and the Vaccine Research Institute for providing the MVA–HIV-B vaccine, the Infectious Disease Models and Innovative Therapies Core Facilities for animal intervention and sampling, Rahima Yousfi for technical assistance, and Dr. Bernard Verrier for providing the p24 HIV-1 Ag.

The authors have no financial conflicts of interest.