|Home | About | Journals | Submit | Contact Us | Français|
Circulating lymphocytes continuously enter lymph nodes (LNs) for immune surveillance through specialised blood vessels named high endothelial venules (HEVs)1–5, a process that increases dramatically during immune responses. How HEVs permit lymphocyte transmigration while maintaining vascular integrity is unknown. Here, we report a role for the transmembrane O-glycoprotein podoplanin (PDPN, also known as gp38 and T1α)6–8 in maintaining HEV barrier function. Mice with postnatal deletion of PDPN lost HEV integrity and exhibited spontaneous bleeding in mucosal LNs, and bleeding in the draining peripheral LN after immunisation. Blocking lymphocyte homing rescued bleeding, indicating that PDPN is required to protect the barrier function of HEVs during lymphocyte trafficking. Further analyses demonstrated that PDPN expressed on fibroblastic reticular cells (FRCs)7, which surround HEVs, functions as an activating ligand for platelet C-type lectin-like receptor 2 (CLEC-2)9,10. Mice lacking FRC PDPN or platelet CLEC-2 exhibited significantly reduced levels of VE-cadherin (VE-cad), which is essential for overall vascular integrity11,12, on HEVs. Infusion of wild-type (WT) platelets restored HEV integrity in CLEC-2-deficient mice. Activation of CLEC-2 induced release of sphingosine-1-phosphate (S1P)13,14 from platelets, which promoted expression of VE-cad on HEVs ex vivo. Furthermore, draining peripheral LNs of immunised mice lacking S1P had impaired HEV integrity similar to PDPN- and CLEC-2-deficient mice. These data demonstrate that local S1P release after PDPN-CLEC-2-mediated platelet activation is critical for HEV integrity during immune responses.
LNs are essential sites for immune responses. They are organised into lobules, which are surrounded by lymphatic sinuses that deliver antigens from afferent lymphatic vessels to LNs for identification by naïve lymphocytes that continually home through HEVs (Supplementary Fig. 1). Lymphocyte trafficking is particularly prominent in mucosal LNs, as the majority of foreign antigens enter the body through mucosal epithelium, and in draining peripheral LNs during immune responses1,15. How HEVs accommodate a high rate of lymphocyte trafficking while maintaining their integrity remains unknown.
Platelets support vascular integrity in inflamed tissues by still undefined mechanisms16. Whether and if so, how platelets protect HEV integrity in the LN is unexplored. PDPN, a ligand for the platelet activating receptor CLEC-2, is highly expressed in LNs. We developed mice with tamoxifen (TM)-inducible global deletion of PDPN (Pdpnf/f;CagCre, Supplementary Fig. 2a–d), and focused first on the mucosal LNs of mice around the weaning-age because development of adaptive immunity occurs early and primarily through mucosal LNs17. TM administration from postnatal day 1 (P1) through P5 resulted in ~90% reduction of PDPN at P15 and complete loss at 1 month in both mucosal and peripheral LNs (Supplementary Figs. 2e, 3a–b, d). Beginning at P15 and progressively worsening, Pdpnf/f;CagCre pups exhibited massive bleeding primarily in mucosal LNs including mesenteric LNs (MLNs) and cervical LNs (CLNs) (Supplementary Fig. 3e–f) but rarely in peripheral (inguinal and popliteal) LNs (Fig. 1a, Supplementary Fig. 3a–f). Histology and confocal imaging of MLNs revealed large numbers of extravasated red blood cells (RBCs) around HEVs but not non-HEV vessels of Pdpnf/f;CagCre mice (Fig. 1b–d). PDPN deletion starting at 3-4 weeks of age resulted in a similar mucosal LN bleeding phenotype, suggesting that PDPN is also important for LN vascular integrity in adults (Supplementary Fig. 3c–d).
In LNs, PDPN is expressed by endothelial cells of lymphatic vessels but not by blood vessels including HEVs (Supplemental Fig. 4a). However, PDPN is also highly expressed on FRCs, which surround HEVs and express ER-TR7, αSMA, and PDGFRβ 7,18(Supplementary Fig. 4b–d). To address whether PDPN on FRCs is essential for LN vascular integrity, we generated Pdpnf/f;PdgfrbCre mice, which lack PDPN in FRCs but otherwise exhibit normal FRC organization (Supplementary Fig. 5b, e). Similar to Pdpnf/f;CagCre mice, Pdpnf/f;PdgfrbCre mice developed bleeding in mucosal LNs (Fig. 1e). Pdpnf/f;PdgfrbCre mice also had reduced levels of PDPN on lymphatic endothelial cells (LECs) in LNs (Supplementary Fig. 5b–d). To rule out the contribution of endothelial PDPN to LN bleeding, we developed Pdpnf/f;Tie2Cre mice, which lack PDPN specifically in LECs but not in FRCs (Supplementary Fig. 6a–d). Consistent with the previously described role of PDPN in the separation of blood and lymphatic vessels during embryonic development8, Pdpnf/f;Tie2Cre mice exhibited blood-lymphatic vessel mixing phenotype (Supplementary Fig. 6e, data not shown). However, Pdpnf/f;Tie2Cre mice did not exhibit bleeding around HEVs in the LN (Supplementary Fig. 6e–f). These results indicate that PDPN on FRCs rather than LECs prevents bleeding in LNs.
CLEC-2 is the only known receptor for PDPN9,10,19. To determine its importance for LN vascular integrity, we depleted CLEC-2 in WT neonates using a CLEC-2-specific monoclonal antibody (mAb), INU119. Administration of INU1 resulted in bleeding in mucosal LNs at P15 similar to Pdpnf/f;CagCre mice (Fig. 1f, Supplementary Fig. 7a). To determine the role of CLEC-2 in adult LNs, we made Clec-2−/− bone marrow chimaeras (Clec-2−/− BM chimaera, Supplementary Fig. 7b–c), which, consistent with deleting PDPN in adult mice, developed bleeding in mucosal LNs (Supplementary Fig. 7d). CLEC-2 is primarily expressed on platelets10,19; however, it is also expressed on myeloid and dendritic cells. To verify whether CLEC-2 on platelets is required to protect LN vascular integrity, we developed mice lacking CLEC-2 specifically on platelets (Clec-2f/f;Pf4Cre, Supplemental Fig. 8a–c). Clec-2f/f;Pf4Cre mice developed spontaneous bleeding in mucosal LNs reminiscent of PDPN-deficient mice (Fig. 1g, Supplemental Fig. 8d). Collectively, these results indicate that FRC PDPN and platelet CLEC-2 are required to protect LN vascular integrity.
To explore where platelet CLEC-2 interacts with PDPN in the LN, we stained cryosections of LNs and found frequent associations of platelets with PDPN+ FRCs at the abluminal sides of HEVs of WT but not PDPN-deficient mice (Fig. 2a, Supplemental Fig. 9). Furthermore, intravenously transfused fluorescently labelled CLEC-2-expressing WT platelets were detected on the abluminal sides of HEVs in close association with PDPN+ FRCs of Clec-2−/− BM chimaeras (Fig. 2b), supporting that platelets interact with FRCs around HEVs in a PDPN-CLEC-2-dependent manner.
Mucosal LNs but not peripheral LNs of PDPN- or CLEC-2-deficient mice exhibit spontaneous bleeding, suggesting phenotypic differences between these two types of LNs. We found that mucosal LNs were significantly more permeable to Evans blue dye than peripheral LNs in the steady-state (Fig. 2c). We noted that mucosal but not peripheral HEVs express MAdCAM-1 (Supplementary Fig. 10a–b). However, after an immune challenge (ovalbumin/complete Freund’s adjuvant, OVA/CFA), HEVs in peripheral LNs of WT mice had increased MAdCAM-1 expression and permeability, resembling mucosal HEVs (Supplementary Fig. 10b, data not shown). Importantly, Pdpnf/f;CagCre or Clec-2f/f;Pf4Cre but not WT mice developed bleeding in draining peripheral LNs after OVA/CFA challenge (Fig. 2d). These results suggest a crucial role for PDPN-CLEC-2 interactions in maintaining vascular integrity of steady-state mucosal and immunised peripheral LNs, which share a similar “reactive” functional and molecular phenotype and are specifically vulnerable to bleeding. Consistent with this, bleeding was observed in mucosal LNs of 3-week-old Pdpnf/f;PdgfrbCre mice, but not Pdpnf/f;PdgfrbCre mice bred into a Rag1−/− background that lack lymphocytes (Fig. 2e). In addition, depleting CLEC-2 with INU1 resulted in mucosal LN bleeding in 2-week-old WT but not Rag1−/− mice, highlighting the importance of lymphocytes in promoting mucosal LN bleeding (Supplementary Fig. 11a). We next administered a mAb to L-selectin (Mel-14) that blocks lymphocyte trafficking through HEVs (Supplementary Fig. 11b)20. Mel-14 significantly reduced the bleeding observed in the draining peripheral LN of immunised Pdpnf/f;CagCre mice compared with an isotype control (Fig. 2f, Supplementary Fig. 11c). A previous study shows that activated platelets facilitate lymphocyte adhesion to HEVs21. However, whether platelets transmigrate with or following lymphocytes to the abluminal space of HEVs to fulfill additional function is unknown. We found that draining peripheral LNs from immunised WT mice had a significant increase in the number of platelets at the abluminal sides of HEVs, which was abolished after Mel-14 treatment, demonstrating that platelets migrate across HEVs in a lymphocyte transmigration-dependent manner to interact with FRCs (Supplemental Fig. 11d). These data support that interactions between platelets and FRCs are critical for preventing bleeding in “reactive” LNs during immune responses when lymphocyte trafficking is increased.
To determine whether impaired HEV barrier function is responsible for the observed LN bleeding when PDPN is absent, we intravenously injected FITC-dextran immediately prior to euthanasia. FITC-dextran was contained within HEVs of WT MLNs. However, FITC-dextran leaked from HEVs, but not from non-HEV blood vessels in LNs and other organs, of Pdpnf/f;CagCre mice (Fig. 3a, Supplementary Fig. 12a, and data not shown). Furthermore, intravenously injected, fluorescently labelled RBCs were detected outside of HEVs in Pdpnf/f;CagCre but not in WT mice (Fig. 3b). Ultrastructural analyses indicated that abnormal gaps between endothelial cell membranes were specific to HEVs of Pdpnf/f;CagCre mice, and RBCs were frequently observed between or outside of high endothelial cells in Pdpnf/f;CagCre mice (Fig. 3c, Supplemental Fig. 12b). Together, these data indicate that, in the absence of PDPN, impaired vascular barrier function and defective junctions are restricted to HEVs.
VE-cadherin (VE-cad) is an integral component of endothelial adherens junctions and barrier function22. We found that levels of VE-cad were reduced in HEVs of Pdpnf/f;CagCre MLNs starting at P8, prior to the onset of bleeding (Fig. 3d, Supplemental Fig. 12c–d), consistent with the idea that loss of HEV junctional integrity contributes to the onset of bleeding. Furthermore, following the loss of PDPN, β-catenin, another component of endothelial adherens junctions22, was decreased (Supplemental Fig. 12e) and expression of N-cadherin, which is essential for endothelial-stromal cell junctions22, was reduced around HEVs (Supplemental Fig. 12f). Further analyses indicate that VE-cad was expressed at higher levels in peripheral HEVs than in mucosal HEVs of WT mice (Supplemental Fig. 13a). ZO-122, another junction molecule, was detected on peripheral but not mucosal HEVs. However, HEVs in draining peripheral LNs of immunised WT mice exhibited reduced VE-cad and ZO-1 expression, resembling MLNs of WT mice (Supplemental Fig. 13a, b). Consistent with the bleeding in draining peripheral LNs (Fig. 2d) of immunised PDPN- or CLEC-2-deficient mice, these LNs exhibited a further reduction of VE-cad on their HEVs (Supplemental Fig. 13a), which was normalised by blocking lymphocyte trafficking with Mel-14 (Supplemental Fig. 13c). Furthermore, blocking VE-cad increased permeability in draining peripheral LNs of immunised WT mice (Supplemental Fig. 14). These results support the idea that interactions of PDPN and CLEC-2 preserve HEV barrier function primarily by promoting VE-cad expression on HEVs.
HEVs of Clec-2−/− BM chimaeras exhibited increased permeability to intravenously injected FITC-dextran (Fig. 3e). To test whether this phenotype could be rescued by WT platelets, Clec-2−/−BM chimaeras were transfused daily with WT platelets for 4 days prior to euthanasia. HEVs of Clec-2−/− BM chimaeras that received WT platelets exhibited significantly decreased permeability to FITC-dextran compared to controls (Fig. 3e). Significantly, WT and Clec-2−/− BM chimaeras transfused with WT platelets expressed comparable levels of VE-cad on HEVs (Fig. 3f, Supplemental Fig. 15). Together, these results demonstrate that PDPN and platelet CLEC-2 are essential for preserving adherens junctions of HEVs.
Platelet aggregation is essential for hemostasis16,23. However, blocking aggregation with a mAb to integrin αIIbβ3 did not increase permeability in draining peripheral LNs of immunised WT mice (Supplemental Fig. 14). This suggests that a platelet function mediated by PDPN-CLEC-2 apart from aggregation is required for HEV integrity. Sphingosine-1-phosphate (S1P)13,14, a bioactive lipid, is a strong candidate for mediating this platelet function as it is known to regulate vascular integrity through interactions with its G protein-coupled receptors on endothelial cells13.
Although platelets generate and store S1P24, whether PDPN-CLEC-2-mediated platelet activation causes S1P release is unknown. To test this, we stimulated WT and Clec-2−/− platelets with the mAb INU1, which activates CLEC-2 signaling in vitro19, and observed a CLEC-2-dependent S1P release from platelets (Fig. 4a). Furthermore, PDPN+ but not PDPN− melanoma cells induced the release of S1P from WT but not Clec-2−/− platelets (Supplemental Fig. 16a–b), indicating that interactions between PDPN and CLEC-2 induce S1P release from platelets. Next, we found that MLN slices cultured ex vivo with fetal bovine serum (FBS), but not lipid-depleted FBS, retained VE-cad expression on their HEVs (Fig. 4b, Supplemental Fig. 16c), supporting the idea that a lipid, such as S1P, maintains HEV adherens junctions. The addition of WT but not Clec-2−/− platelets rescued VE-cad levels on HEVs of MLN slices cultured in lipid-depleted FBS-containing media. WT platelets were unable to restore HEV VE-cad expression on MLN slices from PDPN-deficient mice (Fig. 4b, Supplemental Fig. 16c), demonstrating that FRC PDPN and platelet CLEC-2 are required for normal VE-cad expression on HEVs. The increase of VE-cad on HEVs in the presence of WT platelets is due, at least in part, to the activation of the S1P receptor 1 (S1PR1), because the S1PR1 antagonist, W14614, blocked the WT platelet-induced VE-cad expression on HEVs (Fig. 4b, Supplemental Fig. 16c). Furthermore, platelets from WT, but not S1P-deficient (S1Pless) mice13, rescued VE-cad expression on HEVs of MLNs (Fig. 4b, Supplemental Fig. 16c), supporting the importance of platelet S1P. S1Pless mice did not develop spontaneous mucosal LN bleeding, likely due to reduced lymphocyte transmigration through HEVs because of lower circulating lymphocytes25 (Supplemental Fig. 17a). However, after OVA/CFA challenge of the hindlimb, the draining peripheral LNs of S1Pless mice exhibited bleeding, increased permeability to FITC-dextran, and decreased VE-cad expression on HEVs, resembling mice lacking PDPN or CLEC-2 (Fig 4c–d, Supplemental Fig. 17b–d). A daily transfusion (for four consecutive days) of WT but not Clec-2−/− platelets resulted in higher levels of VE-cad on HEVs and reduced vascular leak of injected FITC-dextran from HEVs of draining peripheral LNs of immunised S1Pless mice (Fig. 4d, Supplemental Fig. 18). Taken together, these data indicate that PDPN-CLEC-2-dependent local release of S1P from platelets plays a critical role in maintaining HEV integrity during immune responses.
Unlike other venules, HEVs are not circumscribed by typical pericytes and collagen-containing matrix that would activate platelets. Instead, they are surrounded by a perivenular sleeve of FRCs that sequester collagen fibers26,27. Our findings reveal an important new role for FRC PDPN (Fig. 4e), which is well positioned to interact with CLEC-2 on extravasated platelets. S1P in the blood is known to regulate vascular integrity13. Our data demonstrate that PDPN-CLEC-2-dependent platelet activation causes release of S1P in the perivenular space that preserves VE-cad expression on HEVs (Fig. 4e). Thus, cross-talk between FRCs, platelets, and HEVs is essential to maintain HEV integrity in situations of increased lymphocyte trafficking such as chronic inflammation. Recently, components of CLEC-2 signaling pathway have been implicated in preventing inflammation-induced hemorrhage in the skin and lung28. Therefore, PDPN-CLEC-2-mediated local S1P release from platelets may protect vascular integrity in other inflamed tissues.
To generate mice with loxP-flanked Pdpn alleles (Pdpnf/f), a targeting vector was constructed in which exon 2, the major coding exon of Pdpn gene, was flanked by loxP sites (Pdpnf/f), and the neomycin resistance selection cassette (Neo) was flanked by Frt sites (Supplementary Fig. 2a). The Not I-linearized construct was electroporated into C57BL/6-derived embryonic stem (ES) cells, and correctly targeted clones were identified by Southern blots. The Frt-flanked Neo was removed by transient expression of Flp recombinase in the positive ES clones to avoid potential undesirable effects of the Neo cassette. ES cells with normal karyotype bearing a floxed Pdpn allele were microinjected into B6/Tyr blastocytes, which were subsequently implanted into pseudopregnant foster mothers. Male chimaeras were bred with C57BL/6J females for germline transmission. Heterozygous mice were then crossed to generate Pdpnf/f mice (Supplemental Fig. 2a-b).
To generate mice with inducible deletion of PDPN (Pdpnf/f;CagCre), Pdpnf/f mice were crossed with the CAG-Cre-ERT2 Tg mice (B6.Cg-Tg(CAG-cre/Esr1*)5Amc/J, Jackson Laboratories)29. To induce postnatal deletion of PDPN, tamoxifen (TM, MP Biomedical) was dissolved in ethanol/sunflower oil (1:9) and administered orally (p.o.) (20 μg per day) to pups from postnatal day (P) 1-5. Adult deletion was accomplished by administering TM p.o. (1 mg per day) for 5 consecutive days beginning at P21, then once a week thereafter. Wild-type littermates (WT, Pdpnf/w;CagCre or Pdpnf/f) treated with the same regimen were used as controls. Mice deficient for PDPN in pericytes/fibroblasts including FRCs (Pdpnf/f;PdgfrbCre) or in endothelial cells (Pdpnf/f;Tie2Cre) were generated by crossing Pdpnf/f mice with PdgfrbCre Tg mice [Tg(Pdgfrb-cre)9Rha]30 or Tie2Cre Tg mice [Tg(Tek-cre)1Ywa]31, respectively.
CLEC-2 conditional mice were generated in which exons 3 and 4 of the Clec-2 allele are flanked by loxP sites (Clec-2f/f). Deletion of exons 3 and 4 induces a premature stop codon that blocks the expression of CLEC-2’s extracellular domain (Supplementary Fig. 8a). ES clones with correct homologous recombination were microinjected into B6 blastocytes, which were subsequently implanted into pseudopregnant foster mothers. Male chimaeras were bred with ACTB-FLP1 females (B6.Cg-Tg(ACTFLPe)9205Dym/J, Jackson Laboratory) for germline transmission and removal of the Neo cassette. Heterozygous mice were then crossed to generate Clec-2f/f mice. Clec-2f/f mice were crossed with Pf4Cre Tg mice (C57BL/6-Tg(Pf4-cre)Q3Rsko/J, Jackson Laboratory) to generate mice deficient for CLEC-2 on platelets (Clec-2f/f;Pf4Cre mice).
Clec-2−/−9, Rag1−/− (Jackson Laboratories)32, and S1Pless mice13 were previously described. Mice were housed in specific pathogen–free barrier facilities. All mice were of mixed genetic background (129S and C57BL/6J) unless otherwise stated. Sex- and age-matched WT littermate controls were used for all experiments. Mice were included in studies if they exhibited more than 75% for PDPN deficiency and 95% for CLEC-2 deficiency. Random assignment to treatment groups was used in antibody blocking and platelet transfusion studies in vivo. No blinding was used in this study. Animal studies were approved by the Institutional Animal Care and Use Committee of the Oklahoma Medical Research Foundation.
Organs were photographed at autopsy. For histology, samples were fixed in 4% paraformaldehyde (PFA) overnight at 4°C, washed, and embedded in paraffin. Sections (5 μm) were stained with H&E. Confocal microscopy was performed as previously described8. Briefly, tissues were fixed in 4% PFA overnight at 4°C, washed in PBS, cryoprotected in 20% sucrose in PBS at 4°C overnight, embedded in 50% Tissue Freezing Medium/50% OCT, and cryosectioned (20–30 μm). Sections were blocked for 1 hr at room temperature and incubated with primary antibodies overnight at 4°C. Secondary antibodies were added for 1 hr at room temperature. Primary antibodies were to murine thrombocytes (Accurate Chemical & Scientific Corp.), PDPN (clone 8.1.1, Developmental Studies Hybridoma Bank), CD41 (clone MWReg30, eBioscience), PDGFRβ (Santa Cruz), CD31 (clone 2H8, Abcam), CD31 (clone MEC13.3, BD Pharmingen), Ter119 (BD Pharmingen), PNAd (clone MECA-79, Biolegend), ER-TR7 (BioLegend), VE-cadherin (clone 11D4.1, BD Pharmingen), N-cadherin (Abcam), β-catenin (clone 14, BD Pharmingen), MAdCAM-1 (clone MECA-367, from Dr. Eugene Butcher, Department of Pathology, Stanford University School of Medicine, California, USA), as well as biotinylated-goat anti-mouse Lyve-1 (R&D Systems). Corresponding secondary antibodies were labelled with DyLight 488, AlexaFlour 555, or DyLight 649 (Jackson ImmunoResearch). Alternatively, Cy3-conjugated anti-αSMA (clone 1A4, Sigma Aldrich) was also used. Images were collected using an Olympus IX81 with DSU spinning-disk confocal microscope and a Hamamatsu ORCA-R2 camera. Images were analysed using Slidebook 5.0 (Intelligent Imaging Innovations).
As previously described8, tissues were collected, homogenized, and total protein concentration was calculated using a spectrophotometer (Eppendorf). Tissue lysates containing 20 μg total protein were resolved on a 10% SDS-PAGE resolving gel and transferred to an Immobilon-P membrane (EMD Millipore). Membranes were probed with a Syrian hamster anti-mouse PDPN (clone 8.1.1), rat anti-mouse VE-cad (clone BV13, Abcam), or mouse anti-GAPDH mAbs. Membranes were washed and probed with horseradish peroxidase-conjugated secondary antibodies. Signal was developed using ECL reagents (Thermo Scientific).
Bone marrow cells from both femurs and tibias were collected from WT and a limited number of Clec-2−/− mice that survived after weaning in our facility by flushing the marrow cavity with 5 ml of HBSS containing 1 mM EDTA, as previously described8. The cell suspension was then run through a 100 μm mesh cell strainer to remove aggregates, centrifuged at 300 x g for 8 min, and resuspended in sterile saline. Cells (5 × 106) in 200 μl saline were injected retro-orbitally into WT recipient mice that had been irradiated with 1100 rad. Engraftment efficiency was determined using flow cytometry analysis on a FACSCalibur (Becton Dickinson) with an antibody against CLEC-2 (clone 17D9, from Dr. Caetano Sousa, Immunobiology Laboratory, Cancer Research U.K., London Research Institute, London, U.K.)36.
PDPN expression on LN cells was analysed by flow cytometry. Briefly, LNs were digested in digestion buffer (RPMI-1640 (Invitrogen) containing 0.2 mg/ml collagenase P (Roche), 0.1 mg/ml DNase I (Invitrogen), and 0.8 mg/ml Dispase (Roche)) at 37°C for 20 min. Digested LNs were vigorously mixed to ensure disruption of capsule and release of leukocytes. Cell suspensions were placed in ice-cold PBS containing 2% FCS and 5 mM EDTA and centrifuged at 300 x g for 10 min at 4°C. Pellets were resuspended in digestion buffer. Washed cells (5×106) were blocked for 15 min on ice (in HBSS containing Ca2+ and Mg2+, 2 % horse serum, 20 μg/ml anti-CD16/CD32 (2.4G2, BD Pharmingen)). Isolated cells were incubated with antibodies to PDPN (clone 8.1.1) and biotinylated CD31 (clone MEC13.3, BD Pharmingen) for 30 min on ice and washed. Cells were then incubated with fluorescent dye-conjugated antibodies: AF488-conjugated goat anti-Syrian hamster IgG (Jackson ImmunoResearch), PerCP-conjugated streptavidin (BioLegend), and PE-conjugated rat anti-mouse CD45 (clone 30-F11, BD Pharmingen) for 20 min on ice. Flow cytometry analysis was performed on a FACSCalibur (Becton Dickinson).
WT and Rag1−/− mice were intraperitoneally (i.p.) injected with 8 μg/g body weight of a mAb to CLEC-2 (clone INU1)19 at P1, 6, and 11. These mice were then sacrificed on P15.
For erythrocyte labelling, whole blood from WT mice was collected into EDTA-coated Microvette 500 K3E tubes (Sarstedt), washed in HBSS (without Ca2+ and Mg2+), and centrifuged at 1200 x g for 10 min. Cells were resuspended in HBSS (without Ca2+ and Mg2+) at a 1:5 dilution of the original volume and incubated with 5 μM CM-DiI (Invitrogen). Washed, labelled RBCs were resuspended at a 75% hematocrit, and 200 μl were intravenously injected into mice. For platelet labelling, platelet-rich plasma was obtained by collecting whole blood into an Eppendorf tube containing 20 IU heparin, adding 500 μl modified Tyrode’s buffer (137 mM NaCl, 0.3 mM Na2HPO4, 2 mM KCl, 12 mM NaHCO3, 5 mM HEPES, 5 mM glucose, pH 7.3) containing 0.35% BSA, and centrifuging for 8 min at 100 x g at room temperature. Platelets were collected by centrifuging for 10 min at 1000 x g. Washed platelets were incubated with 5 μM CellTracker Green (Invitrogen) for 30 min at 37°C. Labelled platelet-rich plasma was washed in HBSS (without Ca2+ and Mg2+) and resuspended in HBSS (without Ca2+ and Mg2+) prior to intravenous injection. Efficiency of labelling was confirmed using a FACSCalibur (Becton Dickinson) and analyzed using CellQuest Pro (Becton Dickinson).
To determine the vascular permeability in MLNs, 250 μg of lysine-fixable FITC-dextran (2,000 kDa, Invitrogen) in 40 μl total volume physiologic saline was injected into the superior mesenteric artery that feeds MLNs immediately before sacrificing animals. Frozen sections of MLNs were stained with anti-CD31 or PNAd antibodies and analysed for the presence of dextran outside the blood vessels. Alternatively, 500 μg of lysine-fixable FITC-dextran (2,000 kDa, Invitrogen) or fluorescently labelled RBCs resuspended at a 75% hematocrit were intravenously injected into mice 1 min or 10 min, respectively, prior to sacrificing mice. Frozen sections of LN were stained with an anti-CD31 antibody and analysed for the presence of fluorescently labelled RBCs outside the blood vessels. Evans blue dye (2% in saline) was intravenously injected 5 min prior to sacrificing the mice. Mice were perfused with heparanized saline and LNs were placed in formamide (Fisher Scientific) overnight at 55°C. Tissues were removed from formamide, and absorbance was measured at 620 nm in a spectrophotometer33.
LNs were removed and fixed overnight at 4°C in 2% gluteraldehyde in 0.1 M cacodylate buffer (pH 7.3). Tissues were washed in 0.1 M cacodylate buffer and post-fixed in 1% osmium tetroxide (1.5 hrs) and 1% tannic acid (1 hr). Tissues were dehydrated through graded alcohols and embedded in Epon resin (EMS). Semi-thin sections (~300 nm) were cut using an ultramicrotome (RMC 7000, RMC) equipped with a diamond knife and stained with toluidine blue, some of which were imaged. Ultrathin sections (~70 nm) were stained with uranyl acetate and lead citrate and visualised using a Hitachi H-7600 electron microscope with a 4 megapixel digital monochrome camera and AMT-EM image acquisition software (Advanced Microscopy Techniques).
Platelets were isolated as described above and were intravenously injected into Clec-2−/− BM chimaeras once a day for 4 consecutive days. For S1Pless mice, WT or Clec-2−/− platelets were isolated as before and resuspended in HBSS without Ca2+ and Mg2+. Either WT or Clec-2−/− platelets were administered once a day for 4 consecutive days. One minute prior to sacrifice, mice were injected (i.v.) with 500 μg of FITC-Dextran (2,000 kDa, Invitrogen) in 100 μl HBSS without Ca2+ and Mg2+.
Three- to four-week-old mice received subcutaneous injections of 250 μg Ovalbumin/Complete Freunds’ Adjuvant (OVA/CFA) into the hindlimb34. Contralateral hindlimbs were injected with equivalent volume of saline as a control. Mice were sacrificed 1 week after challenge and popliteal LNs were processed for immunofluorescence staining.
Three- to four-week-old mice were challenged with 250 μg OVA/CFA into the hindlimb. Mice were then administered i.p. 30 μg of an anti-mouse L-selectin antibody (Mel-14)20 every other day for one week. Mice were sacrificed and popliteal LNs were dissected and analysed.
To determine the blocking efficiency of Mel-14, an in vitro lymphocyte rolling assay was used. Briefly, 100 μg/ml streptavidin was coated on 35-mm polystyrene dishes overnight at 4°C. After washing three times with HBSS, the dishes were blocked with 1% human serum albumin in HBSS for 2 hr at 4°C and then incubated with biotin conjugated 6-sulfo-sLex, a ligand for L-selectin, for 2 hr at 4°C35. Isolation of peripheral leukocytes was described previously36. Briefly, heparinized blood was obtained from WT or Pdpnf/f;CagCre mice that were treated with Mel-14 or rat IgG. After lysis of red blood cells, leukocytes were centrifuged at 100 x g for 10 min. Leukocytes (0.1 × 106/ml) were perfused over 6-sulfo-sLex coated surface under shear stress at 1.0 dyn/cm2. The number of rolling cells was counted using the Element software (Nikon).
After hindlimb challenge with OVA/CFA, mice were i.p. injected with 30 μg every other day of a mAb against VE-cad (clone 11D4.1, BD Pharmingen) or against GPIIbIIIa (clone JON/A)37. One week after immune challenge, Evans blue dye permeability assays were performed as described above.
For measuring peripheral lymphocyte counts, whole blood obtained in EDTA-coated tubes was used to obtain a complete blood count using a Hemavet.
Parental murine melanoma cell line B16-F0, which contains PDPN+ and PDPN− cells, was purchased from American Type Culture Collection (ATCC). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% heat-inactivated fetal bovine serum (FBS) and 1% L-glutamine/penicillin/streptomycin (Cellgro) at 37°C in a humidified atmosphere of 5% CO2. B16-F0 cells were stained with anti-murine PDPN mAb (clone 8.1.1) and a DyLight488-conjugated secondary antibody. Stained cells were then sorted with a FACSAria III cell sorter (BD Biosciences) to generate the PDPN+ and PDPN− B16-F0 cells.
Platelets were isolated from WT or CLEC-2 deficient mice as before. After being washed with modified Tyrode’s buffer containing 0.35% fatty acid-free BSA, platelets (1 × 108 in 100 μl) were then incubated with a mAb against CLEC-2 (clone INU1 at 10 μg/ml) for 10 min. at room temperature. S1P concentration was determined using an S1P ELISA kit (Echelon Biosciences) according to manufacturer’s protocol. Alternatively, platelets were incubated with 5×105 PDPN+ or PDPN− cells for 10 min. at room temperature. Then, the supernatant was collected and S1P concentration was quantitated by electrospray ionisation tandem mass spectrometry (ESI-MS/MS) using stable isotope dilution with heptadeuterated S1P (Avanti Polar Lipids) as the internal standard as described previously38.
MLNs were dissected from WT and PDPN-deficient mice and embedded in low melting point agarose (Invitrogen) at 37°C. Blocks were cooled on ice and sectioned (~250 mm) using a Vibratome (McIlwain tissue chopper). LN slices were then coincubated with washed platelets (2 × 107 in 100 μl of RPMI 1640 containing 10% charcoal stripped FBS; Invitrogen) and isolated from WT, Clec-2−/−, or S1Pless mice, for 1.5 hrs at 37°C while shaking gently. In some experiments, normal FBS or the S1PR1 antagonist, W146 (5 μM, Cayman Chemical), was included. MLN slices were then fixed in 4% PFA in PBS for 30 min and processed for cryopreservation. Cryosections were stained with antibodies against CD31 (clone 2H8, Abcam) and VE-cad (clone 11D4.1, BD Pharmingen).
PDPN levels were calculated using ImageJ software to compare the mean intensities between the groups from a minimum of 5 low magnification (10x) images. For quantification of FITC-dextran leakage, six cryosections were cut per mouse and each section was analysed for the total number of HEVs (~50/section of MLNs) and non-HEVs based on CD31 or PNAd staining and morphology. The number of vessels that exhibited dextran outside the vessel was determined and the percentage of “leaky” vessels compared to the total number of that vessel type was determined. Similarly, the percentage of VE-cad expression on vessels from MLN cryosections was determined by comparing the number HEV and non-HEV that were VE-cad positive to the total number of vessel type. Alternatively, VE-cad intensities on HEVs were determined using image J analysis software.
Statistical tests were performed using Prism software (GraphPad). Two-sided, Student t-tests were performed after the data were confirmed to fulfill the criteria of normal distribution and equal variance. Differences were considered statistically significant when P < 0.05.
We thank Paul Kincade and Linda Thompson for critical reading of the manuscript, Ralf Adams for providing PdgfrbCre mice, and Michael Kinter and Caleb Marlin for technical assistance. Work was supported by grants from the National Institute of Health (GM103441, GM097747, HL085607, HL093242, HL103432, HL65590, HL112788), VA Merit Award (BX001984), The Deutsche Forschungsgemeinschaft (SFB688), Chinese National Natural Science Foundation (30928010), Jiangsu Provincial Special Program of Medical Science (BL2012005), Jiangsu Province’s Key Medical Center (ZX201102), and American Heart Association (SDG7410022).
Author contributionsB.H.H. and J.F. designed and performed experiments, analyzed results, and drafted the manuscript. J.M.M., Y.P., M.S., T.Y., R.S.-M., F.L., A.J.M. and S.M. performed experiments. S.J.W., S. R. C., F.M., B.N., A.S., and P.R.H. supplied key reagents and mice. R.P.M., H.C., and M.L.K. helped analyse results and commented on the manuscript. L.X. designed and supervised research and wrote the manuscript.
The authors declare no competing financial interests.