Galectin-9 bridges human B cells to vascular endothelium while programming regulatory pathways.

Abstract

Humoral immunity is reliant on efficient recruitment of circulating naive B cells from blood into peripheral lymph nodes (LN) and timely transition of naive B cells to high affinity antibody (Ab)-producing cells. Current understanding of factor(s) coordinating B cell adhesion, activation and differentiation within LN, however, is incomplete. Prior studies on naïve B cells reveal remarkably strong binding to putative immunoregulator, galectin (Gal)-9, that attenuates BCR activation and signaling, implicating Gal-9 as a negative regulator in B cell biology. Here, we investigated Gal-9 localization in human tonsils and LNs and unearthed conspicuously high expression of Gal-9 on high endothelial and post-capillary venules. Adhesion analyses showed that Gal-9 can bridge human circulating and naïve B cells to vascular endothelial cells (EC), while decelerating transendothelial migration. Moreover, Gal-9 interactions with naïve B cells induced global transcription of gene families related to regulation of cell signaling and membrane/cytoskeletal dynamics. Signaling lymphocytic activation molecule F7 (SLAMF7) was among key immunoregulators elevated by Gal-9-binding, while SLAMF7’s cytosolic adapter EAT-2, which is required for cell activation, was eliminated. Gal-9 also activated phosphorylation of pro-survival factor, ERK. Together, these data suggest that Gal-9 promotes B cell – EC interactions while delivering anergic signals to control B cell reactivity.

Introduction

Devising strategies to boost antibody (Ab) B cell responses to vaccines or blunt Ab production associated with autoimmune disorders are under intense investigation. However, the molecular factors driving effective functional transition from naïve B cells to germinal center (GC) B cells, memory B cells and Ab-producing plasma cells are still poorly defined. Our laboratory has obtained data illustrating profound differences in the surface glycomes of human naïve, GC and memory B cell subsets highlighted by blood group i/I-antigen (1). While GC B cells display primarily I-branched poly-N-acetyllactosamines (poly-LacNAcs), naïve/memory B cells express mainly i-linear poly-LacNAcs on their cell surface (1). These linear poly-LacNAcs, particularly on CD45, enable strong binding to the potent immunomodulator, galectin (Gal)-9 (2), and consequent activation of Lyn-CD22-SHP-1 inhibitory axis and blunting of B cell receptor (BCR) signaling (1, 3). Data in Gal-9-deficient mice align with these observations, showing increases in B cell proliferation and enlarged GCs (4) and more robust virus-specific Ab responses (5). Moreover, in vivo treatment with Gal-9 can attenuate vaccine-Ab responses as well as Ab levels in mouse models of lupus (2, 6, 7). Together, these studies indicate that selective Gal-9 binding to naïve/memory B cells could have a major impact on B cell immunity and potentially be leveraged to modulate Ab responses. Knowledge on the spatial, cellular and functional control of Gal-9 on B cells in peripheral lymphoid organs is incomplete and the motivation for this study.

Here, we investigated the cellular localization of Gal-9 in human lymph nodes (LN) and tonsil tissues and found that Gal-9 was not only expressed on naïve B cells (1, 3), but found at a remarkably high level in/on endothelial cells (EC) of high endothelial venules (HEV) and post-capillary venules. Using static and flow cell adhesion assays, we observed adherence of circulating human B cell isolates to cultures of human vascular ECs in a Gal-9-dependent manner. Moreover, in transendothelial migration (TEM) assays, Gal-9 slowed the migration of B cells through vascular ECs. Given the pro-adhesive, migration-restrictive activity conveyed by Gal-9, we investigated transcriptional activities triggered by Gal-9 - human naïve B cell interactions. We observed a conspicuous gene ontogeny (GO) term enrichment in cell regulatory and survival pathways as well as pathways controlling intracellular protein/organelle and plasma membrane dynamics. Of the several B cell genes associated with the regulation of signaling/activation activity, Gal-9 significantly elevated the expression of signaling lymphocytic activation molecule F7 (SLAMF7), which functions as a self-ligand, an immune activator and inhibitor, and a potent immunotherapeutic target for multiple myeloma (8–13). Remarkably, SLAMF7 protein was rapidly induced by Gal-9-binding in the absence of BCR-activation and its immune activating potential was counteracted by concomitant downregulation of SLAM-associated adaptor protein, Ewing’s sarcoma-activated transcript-2 (EAT-2) (11, 14). The absence of EAT-2 confers SLAMF7-mediated inhibitory effects (14). Since CD45 is necessary for SLAMF7-mediated inhibition and CD45 is the main receptor for Gal-9 on B cells (1), Gal-9 may pose a dual threat by binding and evoking SLAMF7-dependent and -independent inhibitory effects on B cell signaling (11). In conjunction with immunoregulatory command, Gal-9 also induced the phosphorylation of pro-survival signaling factor, ERK, suggesting that Gal-9 can actually help maintain B cell viability while managing activation signals.

In all, the collaborative role of Gal-9 in promoting regulatory, survival and cell stabilizing pathways implicate Gal-9 as a putative factor in peripheral tolerance. These findings indicate that Gal-9 and its glycan ligands could potentially have a major impact on the efficiency of localization, retention and function of human B cells in the peripheral LNs.

Results

High endothelial and post-capillary venules are rich in galectin (Gal)-9.

Recent data suggest that Gal-9 has a profound intrinsic and extrinsic effect on the activation and proliferation of naïve B cells (1, 3). Functional observations were obtained using murine B cells deficient in Gal-9 (3) or incubations of soluble recombinant human Gal-9 (Gal-9) with human naive B cells (1). Considering the putative regulatory role for extrinsic Gal-9, the native spatial and cellular expression patterns of Gal-9 in peripheral LN is incomplete. To identify cells and LN structures bearing Gal-9, we performed immunohistochemistry (IHC) of Gal-9 on formalin-fixed, paraffin embedded (FFPE)-human ‘reactive’ tonsils and LN tissues. Gal-9 staining indicated the expected expression of Gal-9 in mantle region of B cell follicles that are predominated by naïve B cells, but there was also robust staining on vascular structures (Fig. 1a). Double IHC staining of Gal-9 (red) and B cell marker PAX5 (brown) indicated that Gal-9 was indeed expressed on endothelial cells (EC) of post-capillary venules and high endothelial venules (HEV) that was distinct from B cell staining (Fig. 1b). IHC of Gal-9 or of peripheral LN addressin (PNAd), a luminal marker of high endothelial venules, on serial sections of LNs revealed co-localization of Gal-9 and PNAd (Fig. 1c).

Fig. 1.

Gal-9 co-localizes with vascular structures in human LN and tonsil. (a) IHC staining of sections of FFPE-tonsil tissues with anti-Gal-9 (brown) and counterstained with hematoxylin showed strong Gal-9 staining in naïve B cells of the mantle as well as in high endothelial venules (HEV) (black arrows). (b) Dual IHC staining of FFPE-tonsil sections with anti-Gal-9 (red) and anti-PAX (brown) and counterstained with hematoxylin showed high Gal-9 expression on EC of HEVs (white arrows) and spatial localization with parenchymal and circulating B cells undergoing diapedesis (insert). (c) IHC staining of serial sections of FFPE-human LN and tonsil sections with either anti-Gal-9 (brown) or anti-PNAd (brown) and then counterstained with hematoxylin demonstrated co-localization of Gal-9 with PNAd⁺ vessels (black arrows). (d) Dual IF staining of FFPE-tonsil sections with anti-Gal-9 (teal), anti-PNAd (red) and DAPI (blue) showed co-localization of Gal-9 with PNAd⁺ vessels (insert). All experiments consisted of staining with isotype Ab control (Iso) or secondary Ab alone to control for non-specific staining. Images are representative from at least 5 biological and experimental replicates. Micron bars = 100μm.

To further address co-localization, dual immunofluorescence (IF) analysis of Gal-9 (teal) and PNAd (red) on FFPE-human tonsil sections was performed and demonstrated that Gal-9 was conspicuously expressed in and on ECs of HEV, including co-localization with PNAd on the luminal aspect of the HEV (white) (Fig. 1d). Incidentally, high expression of Gal-9 (LGALS9) was also evident from gene mining data (www.ImmGen.org) (15), showing that resting and inflamed skin-draining murine LN expressed LGALS9 at a 3- and 4-times higher level, respectively, than that of naïve B cells (Suppl. Fig 1.). In all, these data suggested that vascular Gal-9 expression in the peripheral LN may be a major source of Gal-9 to bind and modulate naive and memory B cells that express high levels of counter-receptor Gal-9 ligand.

Gal-9 mediates human B cell – vascular EC adhesion while reducing TEM.

Galectins are secreted lectins that bind and modulate the function of leukocytic, epithelial, mesenchymal and cancer cell surface glycoproteins (16–19). By binding β-galactoside-containing glycans, galectins induce glycoprotein lattices/clustering, which modify downstream signaling and survival/death pathways (17, 20). Gal-9, notably, can bind T cell TIM-3, CD44 and/or protein disulfide isomerase, and avidly to linear poly-LacNAc glycans (1) on B cell CD44 and/or CD45, to downregulate activation pathways and/or effector functions (1, 2, 6, 21–23). Interestingly, Gal-9 can also mediate adhesion between human eosinophils and interferon gamma (IFN-γ)-activated fibroblasts and human umbilical vein endothelial cells (HUVEC), indicating that Gal-9 and its dual carbohydrate-recognition domains can bridge cells in trans (24, 25). Whether Gal-9 can mediate B cell adhesion to vascular ECs is not known.

Prior to assaying B cell – vascular EC dynamics, we first evaluated inherent expression of lymphocyte adhesion molecules, lymphocyte (L)-selectin and endothelial (E)-/platelet (P)-selectin-binding moieties sialyl Lewis ^X and ^A involved in lectin-mediated lymphocyte adhesion, juxtaposed to Gal-9 ligand expression on human circulating B cells. Comparative expression of L-selectin and E-/P-selectin ligands, which are known for mediating peripheral LN and inflamed tissue homing and commonly high on human naive and central memory T cells (TCM), respectively, serve as ideal positive controls for assaying relative levels on human circulating B cell subsets. Flow cytometry of circulating T and B cell subsets from PBMC revealed that IgM-memory and memory B cells expressed very little sialyl Lewis antigen and naïve B cells expressed none (Fig. 2a), supporting earlier data showing no human B cell interactions to E-selectin on TNF-α-stimulated HUVEC (26). Moreover, compared with high L-selectin expression on naïve and TCM, circulating naïve B cells expressed 75% less L-selectin (Fig. 2b) (27). As expected, all circulating B cell subsets expressed robust Gal-9-binding activity (Gal-9 ligand) (Fig. 2c and 2d) that was inhibited by lactose (Fig. 2e and 2f) (1). We then confirmed the suitability of HUVEC as a model system by analyzing the inherent EC adhesion molecule expression in conjunction with assessing Gal-9 expression on resting HUVEC, HUVEC incubated with conventional TNF-α, or HUVEC treated with Type 1 IFN-β and/or Type 2 IFN-γ (25, 28). Compared with TNF-α, incubations with IFN-β and/or IFN-γ caused a significant upregulation of HUVEC surface Gal-9 (p<0.001) (Fig. 3a and 3d). Moreover, whereas TNF-α induced E-selectin expression, IFN-β and IFN-γ did not upregulate E-selectin expression (Fig. 3b). Interestingly, other well-described galectins, Gal-1 and Gal-3, were not expressed on the surface of IFN-β/γ-treated HUVEC (Fig. 3e and 3f). Lactose-containing negative controls confirmed carbohydrate-binding activity of Gal-9.

Fig. 2.

Human circulating B cells express low levels of sialyl Lewis^X/A and L-selectin and high levels of Gal-9 ligand. PBMC were stained with antibodies to T cell markers (CD3, CD4 and CD45RO) or to B cell markers (CD19, IgD, CD27) and either anti-sialyl Lewis X/A (sLe^X/A) HECA-452 (a) or anti-L-selectin (CD62L) (b) and analyzed by flow cytometry. PBMC were also stained with anti-CD19, anti-IgD, anti-CD27, Gal-9 and anti-Gal-9; and naïve (IgD⁺/CD27⁻), IgM memory (IgD⁺/CD27⁺) and memory (IgD⁻/CD27⁺) B cell subsets (c) were analyzed for Gal-9 ligand (d) by flow cytometry. Control rhGal-9 staining consisted of 100mM lactose (Lac) in the buffers (e) and (f). Experiments were repeated on at least 3 donors.

Fig. 3.

IFN-treated human vascular EC express a high level of Gal-9. HUVEC were incubated with IFN-β and/or IFN-γ or TNF-α and analyzed with anti-Gal-9 (a) or anti-E-selectin (b) by flow cytometry. Untreated or IFN-β/-γ-treated HUVEC were analyzed with anti-Gal-9 (d), anti-Gal-1 (e) or anti-Gal-3 (f) with or without a control 50mM lactose pretreatment. Experiments were repeated on at least 3 donors and statistical significance was ascertained by a Mann-Whitney test - ***p < 0.001).

Assays were first performed by incubating MACS-sorted CD19+ B cells from PBMC over confluent monolayers of HUVEC in the presence/absence of Gal-9 or with Gal-1, Gal-3, or lactose buffer controls. Following a 1hr incubation at 37°C and gentle washes, B cells were enumerated from several fields of view at 10X magnification. Compared with baseline adhesion, Gal-9 caused a 2.5-fold increase in adhesion (p<0.001) that was completely inhibited in the presence of lactose (Fig. 4a). Moreover, there was no observed increase in B cell adhesion in the presence of Gal-1 or Gal-3. On IFN-stimulated HUVEC, there was a 2-fold increase in B cell adhesion with a decrease in the presence of lactose, suggesting the high Gal-9 levels on IFN-HUVEC may support B cell adhesion (Fig. 4a). Of note, MACS-sorted naive B cells from tonsil tissue were also assayed in this adhesion assay system and similar fold change increases were noted in Gal-9-treated and IFN-induction groups.

Fig. 4.

Human circulating B cells adhere to human vascular EC dependent on Gal-9. (a) MACS-sorted CD19+ B cells from PBMC were assayed for adhesion to confluent monolayers of HUVEC. Where indicated, Gal-1, −3 or −9 were added to B cell – HUVEC incubations, 50mM lactose was added to control for galectin carbohydrate-binding dependency, and IFN-β and -γ pretreatments were imposed to upregulate Gal-9. Experimental data are presented as mean Fold Change (SEM) from baseline B cell adhesion to resting HUVEC from at least 5 independent experiments and donors (Mann-Whitney test - ***p < 0.001, **p < 0.01; Paired t-test - *p < 0.05). (b) MACS-sorted CD19+ B cells from PBMC were assayed for adhesion to confluent monolayers of HUVEC under shear flow conditions. Where indicated, Gal-9 (with or without 50mM lactose) was added to cells, cells were infused over into the chamber, shear stress was increased in 0.1 dynes/cm² every 30 sec from 0.2 – 1 dynes/cm². Data are graphically presented as Adherent Cells (SEM) over a shear stress range from at least 5 independent donors (Paired t-test - **p < 0.01).

In flow chamber experiments under physiologic shear stress, human circulating B cell isolates in the presence or absence of Gal-9 were perfused over confluent monolayers of HUVEC at a shear stress range up to 2 dynes/cm². Compared with cell suspensions containing lactose, we observed a significant 2-fold increase in shear resistance or cells remaining bound in groups containing Gal-9 (Fig 4b) (Suppl. Video 1–3). While we did observe new cell tethers between 0.1 – 1.0 dynes/cm² range with an occasional cell rolling activity, these events were overall rare and transitory, indicative of weak carbohydrate/lectin interactions and not characteristic of selectin - selectin ligand-binding.

To further assess Gal-9 effects on human B cell – EC dynamics, TEM of human circulating B cells was assayed in transwells containing confluent HUVEC monolayers and B cell chemoattractant CXCL13 in the lower chamber. MACS-sorted CD19+ B cells from PBMC, including Gal-9 with or without lactose, were incubated on HUVEC for 16hr, and B cells, which underwent TEM, were enumerated in the bottom chamber. Compared with non-Gal-9-treated controls, the mean fold change (SEM) of B cells that underwent TEM was significantly less (40%) in the Gal-9-treated group (p<0.001) (Fig. 5a). This attenuated migratory effect was reversed in co-cultures containing Gal-9 inhibitor lactose (Fig. 5a). In support of this inhibitory effect, activation of focal adhesion kinase (FAK) was assayed by Western blotting and showed that Gal-9 incubations caused a significant reduction (~65%) in phosphorylated Try-397 on FAK, which is a protein tyrosine kinase associated with integrin-dependent cellular migration (Fig. 5b) (29). These data implicated Gal-9 as a potentially critical determinant in human B cell – EC interactions, serving to both promote adhesion and decelerate B cell TEM.

Fig. 5.

Gal-9 slows human B cell TEM. (a) MACS-sorted CD19+ B cells from PBMC were incubated with Gal-9 with or without lactose over confluent monolayers of HUVEC for 16hr in transwell chambers containing CXCL-13 in the bottom chamber. After 16hr, frequency of B cells in the bottom chamber were enumerated and expressed as Fold Change (SEM) compared with non-Gal-9-treated control. (b) Western blot analysis of pFAK (Tyr397) or of FAK and β-actin controls were performed on unmigrated B cells isolated from the non-Gal-9-treated control or Gal-9-treated wells. Phosphorylated FAK staining intensity data were expressed as Fold Change (SEM) compared with non-Gal-9-treated control. All data are tabulated from at least 3 independent experiments and donors (Unpaired t-test - ***p < 0.001, **p < 0.01).

Gal-9 globally regulates human naïve B cell activation and differentiation pathways.

Gal-9 can suppress human naïve B cell activation, proliferation and signaling under T-dependent and T-independent activation conditions (1). Based on data here showing Gal-9’s role in bridging B cells to ECs, we hypothesized that Gal-9-mediated adhesion could globally impact gene expression programs controlling B cell function.

To understand early transcriptional events conferred by extracellular Gal-9-binding, MACS-sorted human naïve B cells (Suppl. Fig. 2) were cultured overnight with Gal-9 or with controls: buffer only, Gal-9 and lactose, or lactose alone. Before culturing, cells from all groups were pre-incubated with lactose buffer to elute any pre-bound endogenous Gal-9 that could elicit outside-in effects. After removing dead cells, RNA from two biological replicates for each group was extracted and analyzed by RNA-seq. Principal component analysis revealed sample variation as well as sample clustering (Suppl. Fig. 3). Gene clustering of Gal-9 group was distinctive from buffer only and lactose groups, suggesting extracellular Gal-9-binding elicited a specific gene expression program. Gene expression levels from Gal-9 and lactose and lactose alone groups clustered together and, in some cases, were dissimilar from Gal-9 and buffer-only groups, suggesting that lactose treatment nullified extracellular Gal-9-dependent gene regulation in both groups and may have potentially caused other ancillary events. Heatmap and hierarchical clustering revealed differentially-expressed genes upregulated (red) or downregulated (blue) in the Gal-9 group that were distinguishable from controls (Suppl. Fig. 4). There were many significant gene expression differences noted between Gal-9 and control groups (padj<0.05). In some cases, gene comparisons between Gal-9 and buffer only groups were less pronounced than Gal-9 and Gal-9-lactose groups, suggesting that endogenous Gal-9 could be rapidly mobilized and dispersed to induce outside-in signals. Using gene ontogeny (GO) term enrichment analysis, differentially-expressed upregulated or downregulated genes (padj<0.05) were categorized into distinct GO terms (Suppl. Fig. 5). GO terms highlighted conspicuous elevation in pathways that dampen cell activation and improve survival and a depression of pathways necessary for intracellular protein, organelle and cytoskeleton dynamics. Some of the prominent lymphocyte immunoregulatory genes modulated by Gal-9 included: SLAMF7, FCRL4, CCL22, DUSP8, CCL4, SIGLEC10, FCER2, ZEB1, PPP1R26, GAB2, DOK3, PEG10, RIMS3 and GSG2. These results suggested that Gal-9 could collectively promote cell quiescence by simultaneously programming anti-activation, pro-survival and cell stasis pathways.

Among many critical genes up- or downregulated by Gal-9-binding, we focused our assessment on the induction of signaling lymphocytic activation molecule F7 (SLAMF7 also known as CS1, CRACC or CD319) in human circulating and naïve B cells that functionally intersects immunoregulation and cellular adhesion/dynamics effects. SLAMF7 is a type 1 membrane protein that functions as a homotypic adhesion molecule (self-ligand) eliciting cell-cell interactions and profound effects on lymphocyte signaling (8, 9, 11). Depending on expression of SLAMF7-associated adaptor proteins (SAP), namely Ewing sarcoma-associated transcript (EAT)-2, SH-2 domain containing protein phosphatases (SHP)-1 and −2 and SH-2 domain containing inositol phosphatase (SHIP)-1, SLAMF7 homophilic interactions can either enhance or obstruct immune cell signaling (8, 9, 11, 12). Notably, when SLAMF7 associates with EAT-2, immune cell activation and cytokine production are enhanced, whereas, in the absence of EAT-2, inhibitory signals persist (30–34). Here, RT-qPCR, flow cytometric and Western blot analysis of MACS-sorted CD19+ B cells from PBMC treated with Gal-9 confirmed significant upregulation of SLAMF7 message and protein (p<0.05) (Fig. 6a and b). This upregulation was also observed when MACS-sorted tonsillar naïve B cells were incubated with Gal-9 (p<0.01) (Fig. 6c), with Gal-9 and HUVEC cultures (p<0.001) (Fig. 6d), or with IFN-stimulated HUVEC (p<0.01) (Fig. 6e). Incidentally, there was no concomitant increase in expression of RHOA, a GTPase associated with promoting cytoskeleton reorganization and cell motility, which supported Gal-9’s global influence to restrict cell movement (Fig. 6a and c–e).

Fig. 6.

Gal-9 upregulates expression of B cell immunoregulatory factor, SLAMF7. RT-qPCR analysis of SLAMF7 and RHOA was performed on MACS-sorted CD19+ B cells from PBMC incubated with Gal-9 (a). Flow cytometric analysis of SLAMF7 was performed on MACS-sorted tonsillar naïve B cells incubated with Gal-9 (b). RT-qPCR analysis of SLAMF7 and RHOA and Western blot analysis of SLAMF7 were performed on MACS-sorted tonsillar naïve B cells incubated with Gal-9 (c). RT-qPCR analysis of SLAMF7 and RHOA and Western blot analysis of SLAMF7 was also performed on MACS-sorted tonsillar naïve B cells incubated with Gal-9 and a confluent monolayer of HUVEC (d) or with a confluent monolayer of IFN-β/γ-stimulated HUVEC (e). Graphed expression data from at least 5 donors are presented as Fold Change (SEM) of control non-Gal-9-treated B cells (a–c) or of control B cells incubated with untreated HUVEC (d–e) (Unpaired t-test - ***p < 0.001, **p < 0.01, *p < 0.05).

To determine whether SLAMF7 could elicit pro-activation signals, we investigated expression of the required SAP, EAT-2, in human naïve B cells incubated with Gal-9. Compared with untreated control, EAT2 mRNA and EAT-2 protein levels assayed by RT-qPCR and Western blotting were strikingly lower in Gal-9-treated tonsillar naïve B cells by as much as 75% and 60%, respectively (p<0.01) (Fig. 7a). These data, which revealed the putative inhibitory consequence of SLAMF7 upregulation, were underscored by the concomitant elevation in pro-survival factor, phosphorylated ERK (1, 35) (Fig. 7a). Furthermore, human circulating naïve B cells displayed significant downregulation of EAT-2 while upregulating pro-survival ERK signaling depicted by enhanced pERK with Gal-9 treatment (p<0.05) (Fig. 7b). Gal-9-treated unsorted human circulating B cells also resulted in a reduction in EAT2 mRNA and elevated pERK levels, though EAT2 downregulation was less significant likely due to component memory B cell populations (Fig. 7c).

Fig. 7.

Gal-9 ablates EAT-2 while increasing activation of pro-survival factor, ERK. RT-qPCR analysis of EAT2 and Western blotting of EAT-2 were performed on MACS-sorted tonsillar naïve B cells incubated with Gal-9 (a). Western blot analysis of pERK, ERK and β-actin was also performed on MACS-sorted tonsillar naïve B cells incubated with Gal-9 (a). RT-qPCR analysis of EAT2 and western blot analysis of pERK, ERK and β-actin was performed in MACS-sorted circulating naïve B cells (b) and unsorted circulating B cells (c) incubated with Gal-9. Graphed expression data from ≥3 donors are presented as Fold Change (SEM) of control non-Gal-9-treated B cells (Unpaired t-test - **p < 0.01, *p < 0.05).

Together, these data suggested that exogenous and endogenous Gal-9 could induce SLAMF7 expression and simultaneously ablate EAT-2 and elevate pERK levels in human naïve B cells to provoke SLAMF7-mediated inhibitory signals and cell survival independent of BCR ligation.

Discussion

Efficient B cell immune surveillance in peripheral lymphoid organs is essential for development of effective Ab responses to infectious pathogens. On the contrary, dysregulated B cell surveillance or aberrant B cell proliferation/function in the periphery are hallmarks of persistent infections, autoimmune diseases and B cell malignancies. We and others have recently published data indicating that Gal-9 resides at the crosshairs of these homeostatic and pathological processes (1–4, 6, 7, 21, 23). While best known for its role in suppressing effector T helper cell function (36–38), Gal-9 can also have a profound effect on human naïve B cells by binding CD45 to induce Lyn-CD22-SHP-1-dependent suppression of BCR-mediated signaling and dampen cell activation and proliferation, while elevating levels of pro-survival factor, pAKT in the absence of BCR crosslinking (1). Due to the robust expression of Gal-9 and Gal-9-binding glycans on human circulating and naïve B cells, Gal-9 is postulated as a critical determinant in B cell function. Our prevailing premise is that Gal-9 – Gal-9 ligand interactions on naïve B cells help govern a variety of cell tuning activities required for survival and, importantly, for managing activation thresholds and signaling vital for peripheral B cell tolerance.

Here, we investigated the spatial expression of Gal-9 in human peripheral LNs and tonsil tissues and interrogated Gal-9’s role in human B cell – vascular EC dynamics and related effects on B cell transcriptional programming. Our results suggested that vascular structures in LNs (resting and inflamed) express a remarkable amount of Gal-9. Since human circulating and naïve B cells express an abundance of Gal-9 ligands, the conspicuous expression of Gal-9 in/on LN vascular ECs implicated a potential adhesion and/or regulatory cell partner relationship. Accordingly, our studies included assessments of human B cell – human vascular EC interactions and, to our surprise, we observed Gal-9-dependent B cell adhesion to ECs under both static and low physiologic shear flow conditions. Moreover, Gal-9 reduced human B cell TEM and related RhoA and pFAK migration signals, suggesting that Gal-9 could serve to help bridge for adhesion and decelerate TEM during a methodical vascular exit. This is remarkable in that, to our knowledge, there have been no studies demonstrating human B cell – human vascular EC adhesion interactions (26).

TEM of B cells requires a complex orchestrated array of cell attachment, degradative, pliability and related signaling networks that are still incompletely defined. These events are exquisitely coordinated by expression of the appropriate cell surface glycoprotein receptors, lectins and cell signals that help prime a B cell’s fate as it enters LN parenchyma. In mice, dogma suggests that the LN homing receptor, L-selectin, is the canonical HEV targeting lectin predominantly by binding 6-sulfo sialyl Le^X glycans on HEV (39). While L-selectin is required for T cell homing to peripheral LNs (40–43), the role of L-selectin in B cell recruitment seems less absolute. Since B cells express significantly less L-selectin (27, 42, 44) and B cell pools and Ab responses persist in peripheral LNs of immunized L-selectin-deficient mice (44), there may be other non-L-selectin adhesive mechanism(s) contributing to B cell homing to and/or retention in peripheral LNs. Interestingly, compared with T cells, B cell tethering, rolling and sticking to HEV lacking core-2 O-glycans, which can bear Gal-9-binding poly-LacNAcs (45), are reduced (46); and Gal-9-deficient mice exhibit elevations in circulating lymphocytes (4). Moreover, recent data suggest that mouse B cells lacking O-glycans exhibit severe deficits in homing to peripheral lymphoid organs due to defects in TEM activity (47). In these studies, the deficit of LN homing appears attributable to a weakened TEM capacity of O-glycan-deficient B cells in response to chemotaxis signals conferred by chemokine CCR7 (47). In our studies here, the deficit in human B cell TEM activity is due, in part, to robust Gal-9 – Gal-9 ligand binding interactions. This observation implicates Gal-9 - glycan interactions as critical regulators controlling the entry as well as exit of human B cells to and from peripheral LNs. As L-selectin - glycan ligand interactions typically feature high bond association (k_on) and dissociation (k_off) rates which permit tethering and rolling over a broad range of shear forces, we believe that Gal-9 – Gal-9 ligand interactions may play an ancillary pro-adhesive role in retaining B cells in LNs while delivering immunoregulatory signals as they enter LN parenchyma.

Due to the identified adhesive role of Gal-9 bridging human B cell and vascular ECs and delaying B cell TEM, we sought to determine the impact of extracellular Gal-9-binding on cellular pathways in human circulating/naïve B cells. In whole transcriptome assays of human naïve B cells incubated with Gal-9, we observed global induction of regulators controlling protein serine/threonine kinase activity, cell differentiation and cell activation and of stress-activated signaling cascade pathways. On the other hand, genes governing membrane, cytoskeletal and organelle dynamics, and protein modification, processing and organelle targeting were significantly downregulated. The collective activation of immunoregulatory and cell survival pathways and the lowering of genes controlling the physical cell activity traits indicate that Gal-9 binding to naïve B cells can impose quiescence and cell stasis, while maintaining vitality. Gal-9’s role, in this regard, may help maintain B cell homeostasis in the periphery until encountering cognate antigen.

More focused assessments on the immunoregulators triggered by Gal-9-binding highlight the upregulation of cell surface receptor SLAMF7, which is capable of transmitting inhibitory signals in lymphocytes (11, 48, 49). SLAMF7 is a self-ligand expressed on activated B cells and plasma cells and, at high levels, on multiple myeloma (MM) cells, in which SLAMF7 is a promising immunotherapeutic target (9, 10). Our data confirmed that SLAMF7 levels on human circulating and tonsillar naïve and memory B cells that possess endogenous Gal-9 expression (50) are, in fact, detectable. However, when Gal-9 was presented to B cells in an extracellular soluble or homo-/heterotypic cell manner, SLAMF7 was markedly upregulated while its adaptor EAT-2 was concomitantly downregulated to bestow SLAMF7’s function as an inhibitory signaling receptor. In conjunction with this effect, there was significant upregulation of ERK phosphorylation, which is indicative of an anti-apoptotic state (35). So, while SLAMF7 and its inhibitory potential are elaborated, there is simultaneous upregulation of pro-survival signaling pathways. Additionally, ERK signaling has been implicated in B cell adhesion, (51), which further highlight its activity in the context of Gal-9-binding. Because naïve B cells and ECs lining HEV and post-capillary venules are abundantly interspersed in peripheral LNs, it is likely possible to achieve local extracellular concentrations capable of delivering outside-in signals (7, 52, 53). To this end, SLAMF7 ligation can downregulate BCR signaling via a CD45, Src kinase (Fyn and Lyn), and SHIP-1 inhibitory axis (11). So, Gal-9 induction of SLAMF7 may function as a parallel BCR inhibitory pathway to the other Lyn-CD22-SHP-1 inhibitory circuit invoked by Gal-9 binding to CD45. We speculate that Gal-9 engagement of CD45 could trigger both SLAMF7-dependent and -independent inhibitory pathways while promoting survival of B cells (Fig. 8).

Fig. 8.

Gal-9-dependent B cell adhesion and regulation in the periphery. Human circulating B cells, LN-resident naïve B cells and LN ECs express an abundance of Gal-9, which bind cell surface carbohydrate ligands (a). Gal-9 can bind cell surface ligands on intravascular circulating B cells and LN-resident naive B cells in the extracellular matrix (ECM) to help bridge B cells to ECs (b). Gal-9 binding to naïve B cells not only dampens BCR activation via CD45-Lyn-CD22-SHP-1 pathway (1), but also globally elevates cell survival and immunoregulatory genes, while suppressing cell activation and mobility genes (c). Among several immunoregulators induced by Gal-9, SLAMF7 is upregulated while pro-activation signaling SLAMF7-adapter, EAT-2, is lowered, which confers inhibitory pathways. (Adapted from cartoon model in Ref. (1)).

Taken together, our data suggest that Gal-9 can have a dual function as an adhesion molecule bridging Gal-9 ligands on circulating naïve and memory B cells to ECs and as a immunoregulator transmitting signals that can modulate activation, survival and membrane dynamics as cells traverse the endothelium, enter B cell follicles and await cognate pathogenic antigen. Whether the adhesive effects translate to the efficiency of B cell homing to peripheral LNs is still under investigation. Nonetheless, these activities may be important for retention of naïve B cells in LN and potentially more critical in controlling autoreactive B cell activity in the periphery. This hypothesis is supported by observations that Gal-9-deficient mice exhibit increased GC size and Ab responses to infection and that Gal-9 treatment can inhibit vaccine-Ab responses and ameliorate a lupus model in a mouse (2, 4–7). These findings continue to advance the premise that differential Gal-9-binding to B cell subsets can have a major impact on B cell immunity and manipulating levels of Gal-9 and its ligands can strategically be applied to boost or blunt Ab-responses during vaccination and autoimmunity.

Materials and Methods

Cells.

Peripheral blood mononuclear cells (PBMCs) were isolated from normal healthy donor leukopacks (Children’s Hospital Blood Donor Center, Boston, MA; OneBlood, Inc., Miami, FL) and tonsillar mononuclear cells were isolated from discarded tonsil specimens provided by Nicklaus Children’s Hospital (Miami, FL) using Histopaque-1077 (Sigma-Aldrich, Milwaukee, WI) gradient centrifugation and then stored in liquid nitrogen for future use in flow cytometry and in B cell isolation, adhesion and transcriptional analyses (1, 54). To isolate resting naïve B cells, PBMC aliquots were thawed, washed and subjected to a combination of immunomagnetic bead cell separation kits: Human Naïve B cell Isolation Kit II and CD27 MicroBeads Kit (Miltenyi Biotec, Auburn, CA). To validate B cell and naïve B cell isolation, PBMCs and resting naïve B cell isolates were stained with a combination of anti-CD3-APCCy7, anti-CD19-PerCP, anti-CD27-PECy7, anti-IgD-APC and Zombie Green viability dye (Biolegend, San Diego, CA) and analyzed by flow cytometry, confirming purity of IgD⁺/CD19⁺/CD27⁻ naïve B cells of >95%.

Flow cytometry of L-selectin, sialyl Lewis X and Gal-9 ligands.

PBMCs were stained with a T cell or B cell antibody panel as follows. T cell panel: anti-CD3-PE Cy7 (Biolegend), anti-human CD4-PerCP (Biolegend), anti-CD45RO-APC Cy7(Biolegend), anti-CD62L-PE (Biolegend) and either biotin-anti-sialyl Lewis X (Clone HECA-452) (BD Biosciences, San Jose, CA) and streptavidin-APC (BD Biosciences) or recombinant human Gal-9 (rhGal-9) (R&D Systems, Minneapolis, MN) ± 100mM lactose (ThermoFisher Scientific, Haverhill, MA) and anti-human Gal-9-APC (Biolegend). B cell panel: anti-human CD19-PerCP (Biolegend), anti-CD3-APC Cy7 (Biolegend), anti-CD14-APC Cy7 (Biolegend), anti-CD27-PE Cy7 (Biolegend), anti-human IgD-FITC (Biolegend), anti-CD62L-PE and either biotin-anti-sialyl Lewis X (Clone HECA-452) and streptavidin-APC or rhGal-9 ± 100 mM lactose and anti-human Gal-9-APC. For panels assessing sialyl Lewis X and L-selectin, PBMCs (10⁶/50μl), biotin-anti-sialyl Lewis X (Clone HECA-452) and anti-CD62L-PE were incubated for 45 min to 1hr at 4°C in a low binding 96-well plate, washed with FACS buffer (HBSS, 2% FCS, 10mM HEPES, 2mM EDTA), incubated with streptavidin-APC for 30 min at 4°C, washed with FACS buffer and then transferred and fixed in PBS/1% paraformaldehyde. For panels assessing Gal-9 ligand, PBMCs (10⁶/100μl) and rhGal-9 ± 100mM lactose in PBS/1% BSA in a low-binding 96-well plate were incubated for 45 min to 1hr on ice, washed, incubated with anti-human Gal-9-APC 45 min on ice, washed, incubated with other primary antibodies 45 min on ice, washed and then transferred and fixed and stored in PBS/1% paraformaldehyde until analysis. Please note that all incubations and washes for Gal-9 ligand assay were ± 100mM lactose. Flow cytometry was performed on a FACSCanto (BD Biosciences, San Jose, CA) and analyzed using either Diva (version 8.0.1, BD Biosciences) or FlowJo software (version 7.6.5, FlowJo LLC, Ashland, OR).

Flow cytometry of Gal-1, Gal-3, Gal-9 and SLAMF7.

Resting naïve B cells were stained separately with either anti-human Gal-9-APC (Biolegend), anti-Gal-3-PE Cy7 (Biolegend) or anti-Gal-1 (Invitrogen/Thermo Fisher Scientific, Waltham, MA) plus goat anti-mouse IgG-FITC (Southern Biotech) ± 100mM lactose (Sigma-Aldrich). Human umbilical vein endothelial cells (HUVEC), including those incubated for 24–72 hr with TNF-α (10ng/ml), IFN-β (60ng/ml), IFN-γ (20ng/ml) or IFN-β (60ng/ml) and IFN-γ (20ng/ml), were stained for Gal-1, Gal-3 or Gal-9 as follows: Using a low-binding 96-well plate, HUVEC (10⁶/50μl PBS/1%BSA) harvested with 0.5mMEDTA and washed in HBSS were incubated with anti-human Gal-1 (Invitrogen/Thermo Fisher, Waltham, MA) or mouse IgG1 (Biolegend) with anti-Gal-3-PE Cy7 (Biolegend) or control rat IgG2a-PE Cy7 (Biolegend); or with anti-human Gal-9-APC (Biolegend) or control mouse IgG1-APC (Biolegend) for 45 min on ice and washed. To control for endogenous carbohydrate-dependent Gal binding, control groups were prepared in parallel, wherein all incubations and washes included 100mM lactose. Samples stained with conjugated antibodies were transferred and fixed and stored in PBS/1% paraformaldehyde until analysis. Samples stained with unlabeled antibodies (anti-human Gal-1 and mouse IgG1) were resuspended in goat anti-mouse IgG-FITC (Southern Biotech, Birmingham, AL), incubated 45 min on ice, washed and then fixed and stored in PBS/1% paraformaldehyde until analysis. Flow cytometry was performed on a FACSCanto as above.

For detection of cell surface SLAMF7, resting naïve B cells, including cells incubated for 16hr with 4μg/ml hrGal-9 (R&D Systems), were washed and stained with PE-anti-human SLAMF7 (Abcam). Flow cytometry was performed using BD FACSCelesta^™ (BD Biosciences).

Immunohistochemical and immunofluorescence staining.

FFPE-human LN and tonsil specimens were obtained through the Dana Farber/Harvard Cancer Center Specialized Histopathology and Pathology Specimen Locator Services and stained for Gal-9, PAX5 or peripheral node addressin (PNAd) using a Leica automated staining platform as follows. For immunohistochemical (IHC) staining of Gal-9, sections were subjected to antigen retrieval on Lecia Bond H1(30) (Citrate) for 30 min, incubated with mouse anti-human Gal-9 antibody (10 μg/mL; Clone 9M1–3; Biolegend) or isotype control (10 μg/mL; Biolegend) diluted in Leica antibody diluent for 30 min, incubated anti-mouse secondary antibody-HRP conjugates and developed with DAB (Leica Bond Refine Detection Kit). For dual IHC staining of PAX5 following Gal-9 staining, sections were stained with rabbit anti-PAX5 antibody (1:50; clone D19F8; Cell Signaling Technology) or isotype control (10 μg/mL; Biolegend) and detected using the Leica Bond Polymer Refine Red Detection Kit, post primary alkaline phosphatase (AP) for 15 min at room temperature, and polymer-AP for 20 min at room temperature. For PNAd IHC staining, sections were subjected to EDTA antigen retrieval for 30 min, incubated with rat anti-PNAd antibody (1:100; Clone MECA-79; Biolegend) or isotype control (10 μg/mL; Biolegend) diluted in Leica antibody diluent for 30 min, incubated anti-rat secondary antibody-HRP conjugates, developed with DAB (Leica Bond Refine Detection Kit). All IHC stained sections were counterstained in hematoxylin.

For immunofluorescent analysis of PNAd and Gal-9, FFPE tissue sections were sequentially subjected to EDTA antigen retrieval and incubated with rat anti-PNAd (1:100; clone MECA-79) using the Leica Biosystems Refine Detection Kit, secondary anti-rat Ab HRP conjugate (Vector Labs) and Alexa Fluor^™ 594 Tyramide Reagent (Invitrogen/Thermo Fisher) and then subjected to citrate antigen retrieval and incubated with mouse anti-human Gal-9 (10μg/mL; clone 9M1–3; Biolegend) using the Leica Biosystems Refine Detection Kit and OPAL^™ - 690 Tyramide Reagent (Perkin Elmer). PNAd was imaged in Texas red with red pseudo color and Gal-9 was imaged in Cyanine 5 with teal pseudo color on the Thermo Fisher EVOS microscope.

Gene expression profile analysis.

Data mining and gene expression comparative analysis of murine LN stromal and immune cells was performed using the GEO dataset (GSE15907; www.ImmGen.org) (15). Data demonstrated that murine ECs in uninflamed and inflamed skin-draining LN express a high level of LGALS9 relative to those levels in murine naive B cells (3). Box plots illustrated 3–4-fold higher LGALS9 levels in skin-draining LNs (SLN) compared with Gal-9 in naïve B cells.

Global transcriptome analysis.

Resting human naïve B cell isolates were treated with 50mM lactose on ice for 30 min, washed twice with media and then incubated overnight in 24-well plates using RPMI-1640/10% FCS/2mM L-glutamine, 1,000 Units/ml penicillin/streptomycin. There were four treatment groups included in the overnight cultures: 1.) Buffer alone, 2.) 1μg/ml rhGal-9 (R&D Systems), 3.) 1μg/ml rhGal-9 + 50mM lactose and 4.) 50mM lactose alone. After 16hrs, cultured cells were harvested and subjected to a Dead Cell Removal kit (Miltenyi Biotec) prior to staining for flow cytometric analysis and isolating RNA using RNeasy Plus Mini kits (Qiagen, Hilden, Germany). RNA samples were submitted to the Dana-Farber Cancer Institute Molecular Biology Core Facilities (Boston, MA) for RNA-seq analysis.

For library preparation, libraries were prepared using Roche Kapa mRNA HyperPrep sample preparation kits from 100ng of purified total RNA according to the manufacturer’s protocol. The finished dsDNA libraries were quantified by Qubit fluorometer, Agilent TapeStation 2200, and RT-qPCR using the Kapa Biosystems library quantification kit according to manufacturer’s protocols. Uniquely indexed libraries were pooled in equimolar ratios and sequenced on an Illumina NextSeq500 with single-end 75bp reads. For RNAseq analysis, sequenced reads were aligned to the UCSC hg19 reference genome assembly and gene counts were quantified using STAR (v2.5.1b) (55). Differential gene expression testing was performed by DESeq2 (v1.10.1) (56) and normalized read counts (FPKM) were calculated using cufflinks (v2.2.1) (57). RNA-seq analysis was performed using the VIPER snakemake pipeline (58). Expression raw data can be accessed at the NCBI site in the GEO repository with a GEO accession number of GSE160678.

RT-qPCR Analysis.

To confirm Gal-9-dependent gene regulation identified by RNASeq analysis, we performed additional experiments incubating 4μg/ml rhGal-9 with sorted human naïve B cell isolates. B cells were incubated with 4μg/ml rhGal-9 (R&D Systems), 50mM lactose or buffer control for 16hr, washed 2x with PBS, pelleted and lysed for RNA extraction in Buffer RLT (Qiagen). Alternatively, cultures of resting or IFN-γ/β-activated HUVEC (72hrs) were incubated for 16hr with sorted naïve B cells and buffer control. B cells were carefully aspirated, washed 2x with PBS, pelleted and lysed for RNA extraction in Buffer RLT (Qiagen). RNA was isolated per manufacturer protocol. SuperScript^™ VILO^™ cDNA synthesis kit (Invitrogen) was used to covert isolated RNA to cDNA. Real Time-quantitative PCR (RT-qPCR) was then performed using TaqMan Fast master mix (ThermoFisher Scientific) and TaqMan primers to amplify genes, SLAMF7 (AssID: Hs00904275_m1), EAT2 (AssID: Hs01592483_m1), RHOA (AssID: Hs00357608_m1) and internal control 18s (AssID: Hs03003631_g1). Assays included 20ng cDNA/sample and TaqMan master mix was used according to manufacturer’s protocol. For every donor analyzed, each gene was run in triplicate.

Adhesion Assays.

Early passage HUVEC were grown in T-75 flasks using R&D Endothelial Cell Growth Base Media with growth supplement (R&D Systems). Cells were harvested with trypsin/EDTA (0.53 mM), washed and replated at 200,000 cells/well/1.5 ml in 12-well plates. HUVEC were incubated at 37°C for 1–2 hours and, where indicated, then incubated for 24–72 hr with fresh media containing IFN-β (180ng/ml; Peprotech) and/or IFN-γ (60ng/ml; Peprotech). Prior to the assay, 1μg/ml rhGal-9 (R&D Systems) with or without 50mM lactose (ThermoFisher Scientific) was added to HUVEC cultures for 30 min. MACS-sorted human B cells suspended at 10⁶/ml in 1μg/ml hrGal-9 were added to confluent HUVEC cultures for 1hr, wells were washed 2X, and bound cells were fixed in 3% glutaraldehyde and counted under phase contrast microscopy. Cell counts were tabulated from a minimum of 6 fields from a field of view at 20X magnification per well/3 wells over ≥4 experiments and graphed as mean±SEM.

For assays conducted under physiologic shear flow conditions, early-passaged HUVEC were plated into microchambers (IbiTreat μ-Slide VI0.4, Ibidi, Germany) at 50,000 cells/channel/30 μL. Microchambers were incubated 24 – 48 hours with fresh HUVEC media (R&D systems). Microchamber channels were loaded with fresh HUVEC media containing 1μg/ml rhGal-9 (R&D Systems) with or without 50mM Lactose for 1hr prior to assay. MACS-sorted human B cells suspended at 2 × 10⁶ cells/mL were mixed with HUVEC media containing 1μg/ml rhGal-9 (R&D Systems) with or without 50mM lactose for 30 min prior to loading into chamber channels. Microchambers were secured using an AMEP-VH021 dual-slide vessel holder (Thermo Fisher Scientific) and visualized with an EVOS M7000 imaging system (Thermo Fisher Scientific). B cells were drawn into the chamber channels using a Harvard Apparatus PhD 2000 Series pump with 10cc syringe and allowed to sediment for <4 min. B cells were then subjected to pre-programmed increasing shear stress video-recorded at 10X with adherent B cells tabulated from a minimum of duplicate channels over ≥5 experiments and graphed as mean±SEM.

Migration Assays.

To assess B cell migration, HUVEC were cultured on transwells (Corning) with 3μm pore size for 72 hours. MACS-sorted CD19 B cells from PBMC were plated in the top chamber and B cell chemoattractant CXCL13 (R&D Systems) was added to the bottom chamber. Where indicated, 4μg/ml hrGal-9 (R&D Systems) was added to the top chamber with or without 50mM lactose (inhibitor of galectin-glycan binding). B cells were allowed to migrate for 24 hours and the B cells that migrated to the bottom chamber were collected and counted using a hemocytometer.

Immunoblotting.

MACS-sorted CD19+ B cells from PBMC or tonsillar naïve B cell isolates were incubated with 4μg/ml hrGal-9 with/without HUVEC or incubated with IFN-stimulated HUVEC monolayer. After 24 hours, B cells were collected and lysed in RIPA buffer (Pierce, Inc.) with protease and phosphatase inhibitors (Thermo-Fisher Scientific, Inc.). Cells were incubated on ice for 30 mins and centrifuged at 10000 RPM for 10min at 4°C. The supernatant was collected, and protein concentration was quantified using BCA (ThermoFisher Scientific). Equal amounts of protein from each sample was prepared with denaturing sample buffer (ThermoFisher Scientific). Samples were boiled for 5 minutes and loaded on a 4–12% gradient SDS-PAGE gel (Bio-Rad) and subjected to electrophoresis. Separated proteins were then transferred to PVDF membrane (Merk Millipore, Inc.), blocked for 1hr and incubated with primary antibodies to pFAK (Cell Signaling), FAK (Cell Signaling), human SLAMF7 (Abcam), human EAT-2 (Abcam), pERK (Cell Signaling), ERK (Cell Signaling) and β-actin (Abcam) overnight. Membranes were washed and incubated with IRDye^®-conjugated anti-rabbit or mouse secondary antibodies (LI-COR) for 1h at RT. Blots were then analyzed using a LI-COR Imager (LI-COR Biosciences, Lincoln, NE).

Statistics.

Statistical analysis was done using GraphPad Prism: Mann-Whitney test, unpaired two-tailed t-test and/or paired t-test.

HIGHLIGHTS.

• Human naïve B cells use galectin-9 to bind vascular endothelium and restrict TEM

• Galectin-9-binding triggers global immunoregulatory pathways in human naïve B cells

• Galectin-9-binding globally restricts cellular dynamics pathways in human naïve B cells

• Galectin-9-binding promotes human naïve B cell survival

• Galectin-9-binding engages a SLAMF7-inhibitory pathway in human naïve B cells