Both changes in levels of plasma viral load and CD4⁺ T-cell count followed similar kinetics and range in all macaques. By 11–12 dpi, plasma viral load peaked with a median value of 7.36 log₁₀ SIV RNA copies/ml before decreasing to 5.78 log₁₀ SIV RNA copies/ml on 28 dpi (Figure 1A). As compared to its value before infection, the median blood CD4⁺ T-cell count reached its lowest value on 11–12 dpi (76.9% decrease, p = 0.0004), coincident with the peak in viral load (Figure 1B). In agreement with our previous observations 32, SIV-infection rapidly induced B-cell accumulation in the intestinal mucosa as compared to controls (Figure 1C1D). However, the intensity of B-cell infiltration was variable among SIV-infected animals and even at different sites of the intestinal mucosa within one animal (Figure 1E and 1F). The median number of B-cell areas per μm² of total duodenal mucosa increased by 1.2-fold and 1.8-fold on 14 and 28 dpi, respectively (Figure 1E) whereas it increased by 1.1-fold and 1-fold in the terminal ileum, respectively (Figure 1F). Despite high numbers of isolated B-cell follicles in terminal ileum of SIV-infected animals, the concomitant increase in mucosal thickness lowers the frequency value. (Figure 1D, right panel). The development of GC within B-cell follicles constitutes an independent marker of B-cell activation. Within the intestinal mucosa of SIV-infected animals, B-cell follicles rapidly developed GC, with 39% of duodenal follicles containing GC as compared to 3% in controls. Whereas an average of 21% ileal B-cell areas contained GC in controls, 34% and 44% of them contained GC at 14 and 28 dpi, respectively.

These data evidence a consistent B-cell activation in intestinal mucosa during acute SIV infection, with an early development of GC in intestinal B-cell follicles, which is indicative of the initiation of TD B-cell response. However, SIV-specific antibodies are only detectable in plasma from 21 dpi on in these animals 29.

Because CD20 expression is rapidly lost during terminal B-cell differentiation, human plasma cells and their immediate precursors are generally detected using CD19, CD38 and CD138 mAb. While CD19 expression progressively decreases, CD38 expression begins to be upregulated on human plasmablasts and CD138 expression characterizes late stages of plasma cell maturation 33. The detection of macaque B-cells engaged into terminal differentiation is limited by the lack of suitable reagents: human CD19 and human CD138 mAb possess limited and no cross-reactivity, respectively, while human CD38 mAb are unable to discriminate plasma cells from other simian B-cells subsets (data not shown). We thus evaluated changes in plasma cells by immunohistochemistry using alternative markers: IRF4/MUM-1 and cytoplasmic Ig. High IRF4/MUM-1 protein expression is a hallmark of normal plasma cell differentiation and a major marker of myeloma cells 34.

We found rare IgA plasma cells in GC of the duodenal mucosa of both control and SIV-infected macaques. In contrast, total (stained by IRF4 mAb) and IgG plasma cells per GC were 1.1- and 1.4-fold more numerous in SIV-infected macaques on 28 dpi than in controls (Figure 2A). In terminal ileum, the median number of total plasma cells increased by 1.9- and 3.1-fold on 14 and 28 dpi, respectively. This increase was essentially due to increased numbers of IgG plasma cells per GC, with a 3.8- and 7.2-fold increase on 14 and 28 dpi, respectively (Figure 2C). Therefore, SIV-infection preferentially promotes an IgG response within GC of the small intestine as in spleen and mesenteric lymph nodes 32.

Comparison between median values of each group showed that densities in total plasma cells in the duodenal LP were 1.6- and 2.6-fold higher on 14 and 28 dpi, respectively, than in controls (Figure 3A). However, a variable increase in the density of each class-specific plasma cells occurred on 14 and 28 dpi: 2.6- and 5.9-fold increase for IgM plasma cells, 1.4- and 2.1-fold increase for IgA, and 1.4- and 1.6-fold increase for IgG, respectively (Figure 3B). Therefore, higher proportions of IgM plasma cells were observed on 28 dpi (36% vs. 16% in controls, 2.2-fold) at the expense of IgG (13% vs. 21% in controls) and IgA (52% vs. 63% in controls) plasma cells (Additional file 1: Figure S1A).

In the terminal ileum, the density of total plasma cells also increased in the LP by 1.7- and 4.9-fold on 14 and 28 dpi as compared to controls (Figure 3C). As also compared to controls, the density of IgA plasma cells increased by 1.4- and 3.1-fold on 14 and 28 dpi, that of IgG by 2.5- and 15-fold and that of IgM by 2.2- and 8.3-fold, respectively. These changes resulted in higher proportions of IgG plasma cells at 28 dpi (21.5% vs. 7%) at the expense of IgA plasma cells (43% vs. 66%) (Additional file 1: Figure S1B). These data show that SIV-infection favors the generation of IgM (duodenum) or IgG (ileum) rather than IgA plasma cells in the LP of the small intestine.

Taken altogether these data suggest a biased isotype class switching or an impaired survival of IgA plasma cells in the intestinal mucosa of acutely SIV-infected macaques.

We next examined T-cell changes in the LP and inductive sites (PP and ILF) of the small intestine of SIV-infected macaques. In the duodenum of control animals, we observed numerous CD4⁺ (Figure 4B) and CD45RO⁺ T-cells (recognized by the OPD4 clone) (Figure 4C,D) distributed throughout the LP as well as in follicular T-cell zones. As compared to control macaques, CD45RO⁺ T-cell density was 80% lower in the follicular T-cell zones at 14 dpi (Figure 4F). The well-defined line of CD4⁺ and CD45RO⁺ T-cells, visible at the interface between mucosa and muscularis mucosae in controls (Figure 4B, C, left panels, arrow) was no more detectable in SIV-infected macaques, except near and above B-cell follicles (Figure 4C, arrow); indeed CD45RO⁺ T-cell density was decreased by 78% and 88% in muscularis mucosae at 14 and 28 dpi, respectively (Figure 4G). In contrast, the frequency of CD45RO⁺ T-cells within the GC of duodenal B-cell follicles did not decrease before 28 dpi (44% decrease) (Figure 4E). As shown by CD23 and Ki67 staining, the polarization of GC was preserved until 28 dpi (Additional file 2: Figure S2).

In the terminal ileum, the median CD45RO⁺ T-cells density decreased by 59% and 73% respectively at 14 and 28 dpi in follicular T-cell zones, and by 84% in muscularis mucosae at 14 dpi, as compared to controls (data not shown). Concomitantly, a 1.8-fold increase in CD45RO⁺ T-cells was observed within the GC of ileal B-cell follicles on 14 dpi. Thus, in contrast to follicular T-cell zones and muscularis mucosae where the depletion in CD4⁺ T-cells rapidly progressed from 14 dpi on, CD4⁺ CD45RO⁺ T-cells accumulated within GC of mucosal B-cell follicles at 14 dpi.

Because increased apoptosis might contribute to the paucity of IgA plasma cells, we examined the expression of the cleaved caspase-3, a key mediator of apoptosis shared by membrane- and mitochondrial-mediated pathways. In the duodenal mucosa, no cleaved caspase-3⁺ cells were found within GC or follicular T-cell zones of control macaques (data not shown). In the follicular T-cell zone, the median density increased by 6.5-fold between 14 and 28 dpi (Figure 5A-F, 5 G) whereas the median number of cleaved caspase-3+ cells is 10.5 and 3.2 cells/GC at 14 and 28 dpi, respectively (Figure 5H). In the terminal ileum, the median density of apoptotic cells increased by 11-fold in follicular T-cell zones and by only 1.7-fold in GC at 14 dpi, as compared to controls (data not shown). These results indicate that apoptosis progresses with different kinetics in the inductive and effector mucosal compartments of SIV-infected animals.

In contrast to increased plasma IgG and IgM levels observed in these macaques prior to 14 dpi 32, plasma IgA levels progressively decreased from 18% on 14 dpi to 46% on 28 dpi as compared to baseline values before infection (Figure 6A). Considering that BAFF and APRIL are important IgA inducing factors 13353637, we measured both of these cytokines in plasma. The median concentration of plasma APRIL, which was 22 ng/ml before infection, varied between 13 and 18 ng/ml after SIV infection (Figure 6B). In contrast, we observed a sharp increase in plasma BAFF that peaked at 11–12 dpi (1,347 pg/ml versus 298 pg/ml before infection, 4.5-fold increase), and remained elevated at 13–15 dpi (918 pg/ml, 3.1-fold increase) and decreased toward baseline values thereafter (Figure 6C). Throughout the acute phase of infection, BAFF levels correlated with plasma viral load (p < 0.0001) (Figure 6D) and inversely with circulating CD4⁺ T-cell counts (rho = −0.596, p < 0.0001) but not with plasma IgA (data not shown).

In an attempt to correlate increased plasma BAFF levels with its local production in lymphoid tissues, we compared BAFF expression in the intestinal mucosa of acutely SIV-infected and control macaques using Buffy2 mAb and immunohistochemical approach. Both isolated cells and stromal cell network associated with B-cell follicles, and GC were strongly stained by Buffy2 mAb (Figure 6E). The stromal network in T-cell areas immediately adjacent to B-cell follicles was also BAFF⁺. Our present results thus reveal increased BAFF production in intestinal mucosa even if additional experiments in SIV-infected macaques are needed to identify BAFF-producing cells in this microenvironment during SIV infection as well as to quantify the magnitude of BAFF overproduction in specific tissue areas. Thus, plasma IgA progressively decreases during acute SIV infection despite increased systemic and mucosal production of BAFF.

Mauritian adult male Cynomolgus macaques (Macaca fascicularis), weighing 4 to 6 kg, were housed each in single cages within level 3 biosafety facilities. Animals were housed and cared for in accordance with the European Guidelines for Animal Care (“Journal officiel des Communautés Européennes” L358, 18 décembre 1986). A regional Animal Care and Use Committee: “Comité Régional d’Ethique sur l’expérimentation animale Ile-de-France Sud”, reviewed and approved all protocols, with the goal of improving animal welfare and limiting unnecessary suffering. The animals were sedated with Ketamine chlorydrate (Rhone-Merieux, Lyon, France) before virus injection, blood sample collection or euthanasia. The animals were inoculated intravenously with 50 AID50 SIVmac251 and were euthanized at 14 (five animals), 21 (five animals) and 28 (three animals) days post-infection (dpi). Blood samples were collected before infection and every 3 dpi thereafter until euthanasia. Plasma samples were kept at −80°C until use. Changes in plasma viral load, CD4⁺ T-cell counts, circulating and tissue-specific B-cell subsets of these SIV-infected macaques have been previously reported 32. After euthanasia, duodenal and terminal ileum tissue samples (corresponding to a five cm section before the caecum) were formalin-fixed and paraffin-embedded. Control tissues were collected from three non-infected animals.

Plasma BAFF and APRIL levels were respectively quantified using the Quantikine® ELISA kit for human BAFF/BLyS (R&D systems, Abingdon, UK) and APRIL human ELISA from Bender Medsystems (Tebu-Bio, Le Perray en Yvelines, France) according to the manufacturer’s instructions. Plasma IgA concentrations were determined using a monkey IgA ELISA kit (Alpha diagnostic Intl Inc., San Antonio, Tx). Each sample was run in duplicate and results expressed as mean concentration (pg/mL for BAFF, ng/mL for APRIL and mg/ml for IgA) ± SD for each animal and at each time point.

All paraffin-embedded tissues were cut into (3 μm) sections, perpendicularly to the intestinal wall so that each intestinal compartment could be examined. After antigen retrieval in sodium citrate (pH6) (CD20, Ki67, Buffy2), EDTA (pH 9) (CD3) or EDTA (pH 8) (other Ab), sections were then labeled with optimized concentrations of monoclonal or polyclonal antibodies against: CD20 (L26), IRF4 (MUM-1P), IgA (rabbit F(ab’)₂, IgG (rabbit F(ab’)₂, IgM (rabbit F(ab’)₂, CD45R0 (OPD4), and CD68 (clone KP1) (all from Dako, Glosturp, Danemark), CD4 (1 F6) and CD23 (1B12) (from Novocastra, Newcastle, UK), Ki67 (MIB5, Beckman Coulter, Fullerton), CD3 (SP34-2, Becton Dickinson, Franklin Lakes, NJ) and cleaved caspase-3 (rabbit serum, Cell Signaling Technology Inc., Danvers, MA). Polyclonal Ig and mouse isotype controls were from Dako or R&D systems. Antibody binding was visualized with the Novolink anti-rabbit/mouse secondary Ab Polymer kit (Novocastra Laboratories, Newcastle upon Tyne, UK) according to the manufacturer’s instructions. Binding of Buffy2 mAb (Enzo life Sciences, Villeurbanne, France) was visualized by the EnVision™ + Dual Link Kit from Dako. Nuclei were counter-stained with Hematoxylin QS (Vector Laboratories, Burlingame, CA). Digital images of tissue sections were captured without manipulation using a Zeiss Microscope (Axiophot 2) coupled to a Microfire camera (Optronics, CA) and using the MorphoLite software (Explora Nova, La Rochelle, France).

Image analysis was performed with the Mercator 4.42 software (Explora Nova) on tissue sections from three non-infected animals as controls and from SIV-infected animals euthanized at either 14 or 28 dpi. Epithelium was outlined by hand and excluded in each field. Area from muscularis mucosae to tips of villi was referred to as “total mucosa” whereas CD3⁺ T-cell areas surrounding B-cell areas, including isolated lymphoid follicles and B-cell follicles in PP, were referred to as “follicular T-cell zones” throughout the manuscript. The LP was defined as total mucosa minus follicular T-cell zones and B-cell areas. The number of B-cell areas was determined by counting CD20⁺ areas in total mucosa at 100X magnification, corresponding to surface areas of 9,467 to 308,392 (duodenum) and 8,000 to 12,409 (ileum) μm². Data are expressed as the average number of B-cell areas per μm² of total mucosa. Immunoreactive cells were counted in all GC present in every section at 100X magnification. By using the Novolink anti-rabbit/mouse secondary Ab Polymer kit, we experienced limited difficulties in discriminating Ig-producing cells from the reticular background possibly observed in GC or sub-epithelial areas. Accordingly, only cells with a clear cytoplasmic staining were taken into account for quantification. Data are expressed as mean (±SD) number of positive cells per GC. Immunoreactive cells were counted in follicular T-cell zones, muscularis mucosae (MM) or LP by analyzing as many fields as possible at 100X magnification. Surface areas of 7,900 to 112,172 (duodenum) and 6,167 to 161,140 (ileum) μm² for follicular T-cell zones; 1,182 to 71,742 (duodenum) and 29,000 to 51,500 (ileum) μm² for muscularis mucosae; 147,000 to 900,000 (duodenum) and 235,000 to 750,000 (Ileum) μm² for LP were analyzed. Data are expressed as the mean density of immunoreactive cells (number of positive cells (±SD)/μm²).

Non-parametric Wilcoxon‘s test or Mann-Whitney’s test (two-tailed unless otherwise indicated) and correlations (Spearman’s rank test) were assessed using the GraphPad Prism 5 software (La Jolla, CA). p values < 0.05 were considered as significant.