Two different strains of the European originated PRRSV1 species were used: the PRSSV1.1 emergent Flanders13 (Fl13) strain (25) and the PRRSV1.3 highly pathogenic Lena strain (29). For in vivo experiments, PRRSV infections were performed at INRA PFIE (Nouzilly, France) for FL13 and ANSES (Ploufragan, France) for Lena infections. The animal experiments were authorized by the French Ministry for Research (authorization no.2015051418327338 and no.2015060113297443, respectively) and approved by the national ethics committee (authorizations no.09/07/13-1 and no.07/07/15-3). Ten-week-old Large White piglets were tested PRRSV free and inoculated intranasally at 5.10⁵ TCID50/animal or mock inoculated. For FL13, 3 pigs were used per group and euthanized 5 days post infection (dpi). For Lena, 4 pigs were used per group and euthanized at 10 dpi. Tracheobronchial LN were collected and processed as described above.

Respiratory, tracheobronchial lymph nodes were collected from Large White conventionally bred sows from Guy Harang slaughterhouse (Houdan, France) and from the controlled UE-PAO-INRA (Nouzilly, France) herd. Tissues were minced and incubated with RPMI 1640 supplemented with 100 IU/ml penicillin, 100 mg/ml streptomycin, 250 ng/ml Fungizone® (Antibiotic-Antimycotic 100X ThermoFisher Scientific, Illkirch, France), 2 mM L-glutamine and 10% inactivated fetal calf serum (FCS, Invitrogen, Paisley, UK). Digestion were performed for 30 min at 37°C with 2 mg/ml collagenase D (Roche, Meylan, France), 1 mg/ml dispase (Invitrogen) and 0.1 mg/ml DNase I (Roche). Filtration on 40 μm cell strainers were performed and red blood cells were lysed using erythrolysis buffer (10 mM NaHCO3, 155 mM NH4Cl, and 10 mM EDTA). Cells were washed in PBS/EDTA and further processed fresh for flow cytometry staining and sorting as much as possible. Alternatively, LN cells were frozen in FCS + 10% dimethyl sulfoxide.

Tracheobronchial LN were frozen in Tissue Teck (Sakura, Paris, France) and conserved at −80°C. Sections of 14 μm were obtained using a cryostat (Leica CM3050S, Nanterre, France) and deposed on Superfrost® glass slides (ThermoFisher scientific). Cryosections were fixed in methanol/acetone (1:1) at −20°C for 20 min. Fixed slides were saturated using PBS supplemented with 5% horse serum (HS) and 5% swine serum (SS) for 30 min at room temperature (RT). When biotinylated antibodies were used, a specific step of endogenous biotins saturation using Avidin/Biotin Blocking Kit (Invitrogen) was added. Primary and secondary antibodies (Table 1) were added at 4°C overnight or 30 min, respectively.

For beads draining experiments, tracheobronchial explants were injected at multiple sites in proximity of the target LN with 0.1 μm red fluorescent beads (FluoSpheres™ Carboxylate-Modified Microspheres, 0.1 μm, red fluorescent (580/605), 2% solids, ThermoFisher Scientific) diluted 1/4 in physiological serum. Explants were incubated 30 min à 37°C to allow the drainage of beads into the LN. The targeted LN was then sampled and frozen in Tissue Teck and processed as above for immunohistochemical staining, except for the fixation, in 4% PFA, 15 min at room temperature, in order to avoid bleaching of the beads' fluorescence by methanol, and the saturation and staining steps done in PBS, 5% HS, 5% SS, supplemented with 0.5% Triton in order to permeabilize the tissue upon PFA fixation.

Slides were analyzed using Zeiss AXIO Observer.Z1 microscope and Zen 2012 Software (Zeiss, Jena, Germany).

Tracheobronchial LN was fixed by overnight incubation in paraformaldehyde 4% at 4°C. Immunohistochemical staining and clearing were performed following the iDISCO+ protocol (30). The sample was acquired on a light-sheet ultramicroscope (LaVision Biotec, Bielefeld, Germany) with a 2x objective. Whole LN and isolated follicles were acquired using 0.63X and 4X zoom factor, respectively. LN was reconstructed in 3D using Imaris 9.1 (Bitplane). The number of follicles was manually quantified.

Cells were stained in blocking solution, composed of PBS-EDTA (2 mM) supplemented with 5% HS and 5% SS. Staining were made in 4 steps, including PBS/EDTA with 2% FCS washing between each step: Uncoupled primary anti-CD169, anti-IgM and anti-CD172a antibodies were added to the blocking solution for 30 min on ice and then washed. Fluorescent, secondary, mouse isotype specific antibodies were then added, respectively anti-IgG2a-PE-Cy7, anti-IgM-Alexa647 and anti-IgG2b-APC-Cy7 for 20 min on ice and then washed. Then, to saturate the potential unbound IgG1-directed secondary antibodies sites, we incubated cells 30 min with isotype-control IgG1 (10 μg/ml) in blocking solution. Third the fluorochrom-coupled or biotinylated primary IgG1: anti-CD21 coupled to BV510, anti-CD163 coupled to PE and biotinylated anti-CD2 antibodies were added for 30 min on ice and then washed. Finally, streptavidin-coupled to Alexa700 was added for 20 min on ice and washed before resuspension in DAPI containing buffer for sorting (Table 1).

LN cells were stained as for flow cytometry analysis and the different populations were sorted. Since LN cells are fragile the sorting was carried out on fresh cells to maximize cell yield and viability. Dead cells were excluded by DAPI staining (Sigma-Aldrich). Cells were sorted using a Moflo Astrios sorter (Beckman-Coulter, Paris, France) driven by summit 6.2. Puraflow 1X was used as sheath and run at a constant pressure of 25 PSI. Frequency of drop formation was 43 kHz. The instrument used a 100-μm nozzle. Temperature was kept at 4°C during whole sorting. The lasers used for excitation were solid state 405, 488, 561, and 640 nm. Optical filters used for collecting fluorescence were 448/59 nm (DAPI), 546/20 nm (BV510), 579/16 nm (PE), 671/30 nm (Alexa 647), 722/44 nm (Alexa 700), and 795/70 (PE-Cy7) and (APC-Cy7).

FlowJo software (version X.1.0, Tree Star, Ashland, OR, USA) was used for analysis.

Sorted cells were fixed and stained with May-Grünwald-Giemsa methods as previously described (31). Images were acquired and analyzed with a Pannoramic Scan II (3DHISTECH Ltd., Budapest, Hungary).

Total RNAs from sorted cell populations were extracted using the Arcturus PicoPure RNA Isolation kit (ThermoFisher Scientific, St Aubin, France) according to the manufacturer's instructions. The Qiagen RNase-Free DNase Set (Courtaboeuf, France) was used to remove contaminating genomic DNA. Reverse transcription (RT) was made using Multiscribe reverse transcriptase (ThermoFisher Scientific) according to the manufacturer's instructions and qPCR carried out using the iTaq Universal SYBR Green Supermix (Biorad, Hercules, CA). Ribosomal protein S24 (RPS24) was chosen as reference gene as previously described in pig lung (31, 32). Primers used in this publication are reported in Supplementary Table 1. Using the porcine new born pig trachea epithelial cell line NPTR (33) we validated Topoisomerase IIA and Cyclin B2 as transcriptomic markers of cell proliferation, since their expression by RT-qPCR were respectively 30 and 17 times higher in exponential proliferation cultures compared with confluent NPTR cells. Conversely Ki67, which is a good proliferation marker at protein level, was invalidated for transcriptomic studies since its expression only varied by a factor of 2 between proliferating and confluent cells (Supplementary Figure 1a).

All data were analyzed using GraphPad Prism. The unpaired, non-parametric Mann-Whitney Statistical test was used.

We first identified swine LN MΦ according to their LN localizations with regard to B and T cell areas. The anti-CD21b antibody B-Ly4 mainly stains immature pre-class-switched B cells (34) and was used to localize B cell follicles. Anti-CD8α expressed on CD8 T cells and memory/activated CD4 T cells in swine (35) was used to identify the T cell areas (Figures 1a,c). As previously observed in swine (26), T cell areas are diffused, whereas B cell follicles are precisely delineated. CD169 and CD163 markers previously used to identify swine LN MΦ (27) and mouse LN MΦ were used [for review (8)]. As expected LN from conventionally reared animals (Houdan slaughterhouse) frequently presented several well-developed B cell follicles, whereas control-reared animals from INRA heard presented fewer and smaller follicles (data not shown). We observed CD169 expression at the periphery of the LN (Figure 1a) in an extended subcapsular location (Figure 1b). These CD169 subcapsular staining colocalized with CD163 staining (Figures 1a,c,d). Because of the reverse structure of the pig LN, these peripheral cells are localized next to the efferent vessels. Interestingly, CD169 was also expressed in close association with the B cell follicles (Figure 1a). CD169^pos/CD163^neg/CD21^neg cells were situated on a thin band at the periphery of the CD21-positive follicles (Figure 1a, green perifollicular crescent) whereas CD169^pos/CD163^neg/CD21^pos cells were identified in a more diffuse “crescent shape” intrafollicular area (Figure 1a, yellow intra-follicular staining). Finally, CD169^neg/CD163^pos cells were present in the medulla, tightly associated with the lymphatic cords, both in the LN parenchyma as well as in the sinus (Figure 1d, red staining).

A flow cytometry analysis of enzyme-digested LN was performed to isolate the 4 populations identified using microscopy. The 4 populations were distinguished using a CD163/CD169/CD21 and FSC gating and sorted as described in Figure 2A, followed by MGG staining (Figure 2B) or by RT-qPCR (Figure 2C). CD163^pos/CD169^pos cells presented a rounded or slightly indented nucleus with coarse chromatin and abundant clear vacuolated cytoplasm, in agreement with a macrophagic phenotype (Figure 2B). Moreover, RT-qPCR analyses revealed that these cells expressed the macrophagic CSF1R, MAFB, and MerTK genes (Figure 2C). Because of the reverse structure of the pig LN, these cells located at the periphery are next to the efferent vessels. Thus, we called them efferent MΦ (effMΦ). CD163^pos/CD169^neg cells displayed abundant clear vacuolated cytoplasm and a large deeply indented nucleus with lacy chromatin similar to blood monocytes and to the MCM from murine LN (11). They expressed similar levels of CSF1R and MerTK than effMΦ but higher levels of MAFB (Figure 2C). According to their association with the cord vessels, these cells were called cord MΦ (cordMΦ).

CD163^neg/CD169^pos/CD21^neg cells displayed a blue-gray cytoplasm with a grainy texture surrounding a large vesicular or slightly indented nucleus with coarsely clumped chromatin (Figure 2B). These cells expressed CSF1R, MAFB, and MerTK genes (Figure 2C). According to their morphology and their localization at the periphery of the follicle, these cells were called perifollicular MΦ (PFMΦ). CD163^neg/CD169^pos/CD21^pos cells displayed a high nucleo-cytoplasmic ratio with a large round nucleus with coarse and dense chromatin and a nucleolus surrounded by a perinuclear halo and a rim of basophilic cytoplasm, looking very similar to porcine bone marrow preB cells (36), in strong support with an immature B cell identity. Moreover, their localization in the periphery of the follicle (Figures 1a,d), reminds the dark zone localization of centroblasts in mouse and human [for review see (37)]. In agreement with their immature B cells microscopic profile, these cells expressed CD19, a component of the B cell receptor complex, and PAX5, a regulatory gene involved in the somatic hypermutation process (38) (Figure 2D) but none of the macrophagic markers tested (Figure 2C). The cells were subsequently named CD169^pos B cells. Although CD169 protein expression was detected by FACS (Figure 2A) and microscopy (Figures 1a,d), CD169^pos B cells did not express detectable levels of CD169 mRNA (Figure 2D). Conversely, CD169 mRNA quantification in effMΦ (CD169^pos), cordMΦ (CD169^neg) and PFMΦ (CD169^pos), was in agreement with cytometry and microscopy data (Figure 2D).

To investigate the relationship between the PFMΦ and the B cell follicles we proceeded to whole LN transparisation (Figure 3a), combined with CD169/CD21 immunostaining and imaging using fluorescent light sheet microscope. This allowed to reconstitute the whole LN 3D structure (Figure 3b and Supplementary Movie 1). This tracheobronchial LN, sampled from a conventionally reared animal, contained 539 B cell follicles (Supplementary Movie 2), each of them being individually associated with a PFMΦ area (Supplementary Movie 1). The afferent central entry appeared composed of a collection of smaller afferent vessels which separated from each other while entering deeper in the LN parenchyma (Supplementary Movie 1). One of this afferent vessel and its continuous sinus was tracked (in red, Figure 3c and Supplementary Movie 2). The sinus appeared interconnected with other sinuses along its parenchymal trail to finally pour in the efferent subcapsular sinus.

The 3D rendering of follicles acquired at higher resolution delineated a PFMΦ area forming a semi-spherical structure that cap the B cell follicle on one of its sides (Supplementary Movie 3).

To identify the MΦ taking in charge particulate antigens drained through afferent lymphatics, we injected ex vivo red fluorescent 0.1 μm beads in the tissue surrounding a tracheobronchial LN. Injected beads were allowed to drain for 30 min at 37°C. The draining LN was then sampled and immunostained for CD169 and CD21 expressions. Beads were present in effMΦ and PFMΦ but not in cordMΦ (Figure 3d). Zooming on B cell follicles, beads were mostly associated with PFMΦ although some signal was observed deeper in the follicle associated with CD169^pos B cells (Figure 3e). By referring in DAPI staining, PFMΦ appeared clearly situated in the space between the follicle and the LN parenchyma, reminiscent of mouse LN subcapsular sinus space (Figures 3f,g). To precisely set the limit between the CD21^pos cells and the PFMΦ, a view of the interface between intrafollicular B cells and PFMΦ is depicted in Figure 3h with saturated CD21 staining. Once this limit affixed on the CD169/CD21 co-staining (Figure 3i), several beads appeared situated at the contact between PFMΦ protrusions and intrafollicular CD169^pos B cells (Figure 3i). Thus, PFMΦ and CD169^pos B cells closely interact with antigen drained from the peripheral tissue, at the frontier of the B cell follicle.

The CD21, IgM, and CD2 markers, previously proposed by Sinkora et al. (39) for porcine bone marrow B cell development analysis, were used to better integrate CD169^pos B cells across the LN B cell differentiation steps. The CD169^pos/CD21^pos follicular B cells (Figure 4A, purple cells) did not express IgM and were mixed with CD21^pos/CD169^neg/IgM^neg B cells (red cells) in the intrafollicular area contiguous with PFMΦ (blue cells). IgM^pos/CD21^pos cells (yellow cells) were localized in the center of the follicle. Finally, IgM^pos/CD21^neg cells (green cells) were rarely present in the center of the follicles, but well represented in extrafollicular area, as well as in the LN periphery, in direct contact with effMΦ (Figure 4A). Thus, this first overview of LN B cells phenotypic localization is in agreement with CD21^pos/IgM^neg cells (CD169 positive and negative) being dark zone-localized centroblasts, IgM^pos/CD21^pos cells being light zone centrocytes and plasmablasts and IgM^pos/CD21^neg cells being mostly extrafollicular mature plasma cells.

We completed this microscopic study by FACS analysis using the panel utilized for MΦ identifications, which was complemented with the myeloid marker Sirpa/CD172a and the B cell markers CD21, IgM and CD2 (Figure 4B). The different cell-types were sorted as described in Figure 4B and stained using MGG coloration (Figure 4C). Like CD169^pos B cells, centroblasts displayed a large, round nucleus with multiple peripheral nucleoli surrounded by a rim of basophilic cytoplasm. Centrocytes displayed a small, round nucleus with clumped chromatin and with scant cytoplasm. Plasmablasts displayed some heterogeneity; their size was intermediate to large with scant to moderate slightly basophilic cytoplasm and a vesicular nucleus with reticular chromatin. Plasma cells demonstrated an intermediate-sized nucleus with reticular chromatin and a rim or just a crescent of clear cytoplasm with a prominent juxtanuclear archoplasm.

Using Ki67 staining we identified proliferating B cells that appeared mostly localized in the follicle area contiguous to CD169^pos PFMΦ, in agreement with the phenotype and dark zone-localization of CD169^pos B cells and centroblasts (Figure 4D).

The sorted B cells were characterized by analyzing B cell-transcription factors genes expression by RT-qPCR (40) [for review see (37)]. CD169^pos B cells and centroblasts expressed significantly more BCL-6 (Figure 5) than all the other B cells. BCL-6 is a master transcription factor expressed specifically in centroblasts (40) and involved in the inhibition of their differentiation into plasma and memory B cells (40, 41). Centrocytes expressed median levels of BCL-6, PAX5, and IRF4 as expected for these cells at an intermediate differentiation stage (Figure 5). As expected, plasmablasts and plasma cells expressed the lowest levels of BCL-6 and PAX5, whereas plasma cells expressed the highest levels of IRF4 (42), XBP1 (43, 44), and Blimp1 (45, 46) which are known markers of terminally differentiated B cells (Figure 5). CD169^pos B cells and centroblasts (CD21^pos/IgM^neg) expressed the proliferation markers Topoisomerase IIA (Figure 5) and Cyclin B2 (Supplementary Figure 1b) at significantly higher levels than all the other B cells, which is consistent with the expression of Ki67 in the follicular area contiguous with PFMΦ.

In conclusion, the phenotype, the localization, the expression of commonly used transcription factors and the proliferation status consistently defined five B cell maturation stages. To note, we confirmed here that the CD169^pos B cells were strongly related to centroblasts, and might be an early step of this differentiation stage.

In vivo infections were performed to explore the mechanisms used by PRRSV to persist in the porcine LN, and its impact on the follicular B cell maturation process.

In a first experiment, 4 animals were infected with Lena PRRSV and tracheobronchial LN were collected 10 dpi. At the time of this preliminary experiment we did not yet distinguish PFMΦ from CD169^pos B cells, we thus sorted one single CD169^pos/CD163^neg population. We also sorted LN cDC2 as a myeloid negative control of infection (23) as described in Supplementary Figure 2a. The sorting strategy was validated by RT-qPCR of key specific genes (Supplementary Figure 2b). As previously validated (23), we measured cell-associated PRRSV by RT-qPCR on viral RNA normalized using RPS24 reference gene (2^−ΔCt). PRRSV RNA was detected in effMΦ (10 dpi) (Supplementary Figure 2c) in agreement with effMΦ population decrease proportion among total LN live cells (Supplementary Figure 2d), whereas a weaker expression was detected in the pooled PFMΦ/CD169^pos B cells (Supplementary Figure 2c), with no significant decrease of this last composite population (Supplementary Figure 2d).

Realizing that the CD163^neg/CD169^pos population was composed of mixed B and macrophagic cells we then performed a second in vivo infection in which we discriminated PFMΦ from CD169^pos B cells. Moreover, to ascertain that our first results could be extended to different strains and times post-infection, we then used Flanders 13 strain and sacrificed the animals 5 dpi. Three infected animals were sacrificed, tracheobronchial LN cells were isolated and analyzed using the same gating as in Figure 4B. Upon infection, a proportional decrease of effMΦ and an increase of plasmablasts were observed while no change in the proportions of the other cell types occurred (Figure 6A). The populations were sorted and viral RNA content was measured by RT-qPCR. In agreement with the preliminary experiment, PRRSV RNA was detected in effMΦ (Figure 6B) [Ct = 29.0 ± 1.3 (SD)] but not in the cordMΦ (Ct = 35.6 ± 0.7). Interestingly PFMΦ (Ct = 27.8 ± 1.5) but not CD169^pos B cells (Ct = 35.8 ± 2.6), were positive. No PRRSV RNA was consistently detected in the B cell compartment. Thus, 10 days post-Lena infection and 5 days post-FL13 infection we observed cell depletion and viral RNA in effMΦ as well as viral RNA in PFMΦ.

The expression of cytokines involved in B cell survival/maturation, IL-10, IL-21 and BAFF were measured in LN MΦ upon FL13 infection. PRRSV infection induced the transcriptomic upregulation of BAFF in effMΦ (Figure 7A) but did not modify IL-10 and IL-21 transcriptional expressions.

We then tested if PRRSV infection impacted the B cell differentiation program by measuring the expression of 5 B cell transcription factors (BCl6, Pax5, IRF4, XBP1, Blimp1). Strikingly, CD169^pos B cells but not CD169^neg centroblasts or other B cell differentiation stages, presented a 2.4-time increase in BCL-6 expression, paralleled with a 1.8-fold increase in Topoisomerase IIA (Figure 7B).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.