Fig 1.
Detection of a PSGL1-independent P-selectin ligand on in-vivo activated T cells.
(A) Formation of PSGL1-PSelL requires C2GnT1: Spleen cells from B6, PSGL-1null, or C2GnT1null mice were activated in vitro with Concanavalin A for two days, sub-cultured with IL2 supplemented media for two additional days, and then stained with PSel-hIgG chimera. P-selectin geometric mean fluorescence values shown. Figure representative of more than nine independent analyses. (B) A PSGL1-independent P-selectin ligand is detected on CD8 T cells activated in vivo. Geometric mean P-selectin staining of day-3 activated HY male-antigen specific, or OT1 ovalbumin specific, T cell receptor transgenic CD8+ T cells responding to antigen in vitro or in vivo. Responding T cell receptor transgenic cells on wild type (wt), PSGL-1null, or C2GnT1null genetic backgrounds were either stimulated in vitro with splenic dendritic cells (male or Ova-pulsed) from B6 (wt) or PSGL-1null mice, or adoptively transferred into wt or PSGL-1null male recipients. EDTA was included during P-selectin staining in parallel samples of in-vivo generated cells to confirm divalent cation (Ca++) dependent selectin binding. Flow analysis was gated for viable propidium iodidenegative (PIneg), CD8+, responding (CFSE-diluted) donor cells. The data shown are single sample staining values representing biological replicates for the two T cell receptor transgenic models examined. Fig 1B is representative of more than seven independent analyses. (C) PSL2 is preferentially displayed by CD8+ activated T cells responding in peripheral lymph node. P-selectin staining of CFSE-labeled HY-PSGLnull T cells responding in two PSGL1null male recipients (left panel) or CFSE-labeled OT1-PSGLnull T cells responding to ovalbumin antigen in two PSGL1null recipients (right panel). Donor cells were recovered from peripheral lymph nodes or spleen three days after adoptive transfer. Staining of duplicate recipient mice are shown for each donor. Control P-selectin staining in the presence of EDTA shown for respective donor/recipient combinations. Ex-vivo cells were subjected to gating for viable PIneg, CD8+, responding (CFSE-diluted) donor cells. Geometric mean P-selectin values for HY-PSGLnull (left panel) are LN EDTA 4/4, spleen 14/15, lymph node 82/111; for OT1-PSGLnull (right panel) LN EDTA 5/5, spleen 8/10, lymph node 24/27. Fig 1C is representative of three independent analyses.
Fig 2.
PSL2 association with activated T cells requires Ca++.
(A) Cation removal eliminates expression of PSL2. P-selectin staining of ex-vivo CD8+ CFSE-labeled HY or HY-PSGL1null donor T cells responding (CFSElow cloud ‘a’) and non-responding (CFSEhigh cloud ‘b’) at day 3 in PSGL1null recipients (left panels). In right panels, aliquots of the same cells were washed twice in EDTA-containing media, returned to Ca++ replete media, and then stained with P-selectin. Geometric mean values for P-selectin staining of responding (CFSE diluted) donor cells are shown. Fig 2A is representative of five independent analyses. (B) Cation specificity to support PSL2 expression. OT1-C2GnT1null cells responding to Ovalbumin in PSGL1nullThy1.1 recipients were recovered at day 3. The cation-specific dependence of PSL2 expression on these cells was assessed after 5x serial washing in media containing the indicated concentrations of Mn++, Ca++, or Mg++ divalent cations, and returned to Ca++/Mg++ replete media, prior to P-selectin staining. Positive control, (+Ca/Mg), P-selectin staining of responding cells maintained in Ca++/Mg++ replete media. Negative control (EDTA) P-selectin staining of responding cells pre-washed with EDTA-media. Standard deviation of triplicate control stains shown. Fig 2B is representative of four independent analyses. (C) Persistence of PSGL1 P-selectin ligand and loss of PSL2 P-selectin ligand with in vitro culture. HY and HY-C2GnT1null donor cells responding in independent male PSGL1null recipients were harvested on day 3 and either untreated (NT), or pre-washed with EDTA-containing media (PSL2 stripped) and then transferred into Ca++/Mg++ replete media (EDTA). Cells in NT or EDTA groups were then held on ice (0), or cultured at 37°C for 2.5 or 5 hours prior to P-selectin staining. Mean and standard deviation of three P-selectin staining replicates for each condition is shown. P-values determined by two-tailed Student’s T-tests with equal variance; **** P value = 0.0005, ***** P value = 0.00001, *** P value = 0.001, ** P value = 0.008. Fig 2C is representative of three independent analyses. (D) PSL2 rebinding. CD8+ HY-C2GnT1null donor T cells were recovered from peripheral lymph nodes of male PSGL1null Thy1.1 recipients on day 3 and semi-purified (to >50% purity). Responding donor derived cells were either stained for P-selectin (NT) or treated with EDTA-pre-wash to generate donor cells lacking PSL2 (EDTA) and a supernate extract. The supernate extract was made Ca++ replete and compared with a mock supernate, prepared identically except never exposed to cells, for their respective capacity to restore P-selectin ligand expression (SUP vs MOCK) on responding, EDTA pre-washed, donor-derived T cells. Mean and standard deviation of three P-selectin staining replicates for each condition is shown along with P-values determined by two-tailed Student’s T-test. **** P value = 0.00073, *** P value = 0.0018. Fig 2D is representative of eight independent analyses.
Fig 3.
PSL2 is T cell-extrinsic, dependent on recipient C2GnT1 expression, and docks on L-selectin.
(A) PSL2 display requires C2GnT1 expression in recipient and, together with PSGL1, constitute the total P-selectin ligand on responding donor-derived CD8+ T cells. OT1 or OT1C2GnT1null donor cells were transferred with ovalbumin into PSGL1null or C2GnT1null recipients and analyzed on day 3. P-selectin staining was compared on these (untreated) cells and on corresponding EDTA-pre-washed samples (post-EDTA). The anticipated ligands ‘PSelL’ contributing to P-selectin binding based on data shown above is indicated. Triplicate P-selectin stains for each sample are plotted. Fig 3A shown is representative five independent analyses. (B) L-selectin-independent entry of CD8 T cells into tracheo-bronchial LN. OT1-PSGLnull-LSel+/+ and OT1-PSGLnull-LSel-/- donor cells labeled with CTV or CFSE tracking dyes were co-injected iv into PSGL-Thy1.1 recipients together with ovalbumin. On day 3, spleen, axillary lymph nodes, and tracheo-bronchial lymph nodes were separately harvested and assessed for relative numbers of donor-derived responding T cells (left column graph) and their expression of P-selectin ligand (right column graph). Mean and standard deviation of triplicate samples used for counting and P-selectin staining for each condition is shown. P-values for comparing levels of P-selectin staining of L-Sel+/+ vs LSel-/- donor cells were determined by two-tailed Student’s T-tests with equal variance; * P value = 0.07, **** P value = 0.0007, ***** P value = 0.00006. Fig 3B shown is representative of three independent analyses. (C) L-selectin expression is gene dose dependent. Mel 14 (anti-L-selectin) staining of naïve CD8+ T cells from peripheral lymph node of OT1-PSGL1-/-LSel+/- versus OT1-PSGL1-/-LSel+/+ mice. Geometric mean P-selectin values are indicated. (D) PSL2 expression on activated CD8+ T cells responding in PerLN correlates with L-selectin gene dose. Naïve OT1-PSGL1-/-LSel+/+ and OT1-PSGL1-/-LSel+/- (L-selectin heterozygous) donor cells were labeled with CFSE or CTV tracking dyes, combined, and co-injected with ovalbumin into PSGL1-/- Thy1.1 recipients. P-selectin staining is shown at comparable cell division numbers for donor cells obtained from the same recipient at day 3. Recipient #1, OT1-PSGL LSel+/+ (CFSE, green) and OT1-PSGL1 LSel+/- (CTV, black). Recipient #2, same donor cells but dyes reversed. Mean and standard deviation of geometric mean P-selectin staining at comparable division number is shown. P-values for comparing levels of P-selectin staining at each division number were determined by two-tailed Student’s T-tests with equal variance; recipient #1: division #6 P value = 0.00003 *****, division #7 P value = 0.00001*****, division #8 P value = 0.0001****; recipient #2: division #6 P value = 0.002 ***, division #7 P value = 0.0002****, division #8 P value = 0.05** Fig 3D shown is representative of three independent analyses. (E) Mel14 and P-selectin co-staining of activated CD8+ T cells. CFSE-labeled donor OT1C2GnT1-/- cells were transferred with ovalbumin into PSGL1-/- recipients. Responding donor-derived cells in peripheral lymph node were assessed at day 3 for P-selectin binding and L-selectin expression; analysis gated on responding (CFSE-diluted), viable, CD8+ donor cells. Fig 1E is representative of five independent analyses. (F) Meca 79 staining of CD8+ recipient cells versus donor-derived activated T cells. CTV-labeled OT1C2GnT1-/- donor cells were transferred into PSGL1-/- recipients with ovalbumin. On day 3 both recipient-derived CD8+ T cells and responding (CTV diluted), donor-derived, CD8+ cells were stained for Meca79 expression either before (top panels) or after (bottom panels) applying EDTA-wash conditions used to strip PSL2. Geometric mean P-selectin staining values for technical replicates are shown. Fig 3F is representative of three independent analyses.
Fig 4.
(A) Activated T cells expressing PSGL1 and/or PSL2. P-selectin binding profiles of activated T cells generated using donor/recipient combinations specified expressing PSGL1 and PSL2, PSL2 alone, or neither. Staining with P-selectin was performed on untreated cells or on cells after EDTA pre-washing revealing the independent contributions of PSGL1 (EDTA-resistant) and PSL2 (EDTA-strippable). Independent triplicate staining profiles are shown. Fig 4A is representative of five independent analyses. (B) Activated platelets expressing P-selectin can bind activated T cells via PSGL1 and/or PSL2. Platelets from P-selectin+/+ and P-selectin-/- mice were isolated, labeled with anti-CD41-APC, activated with thrombin, paraformaldehyde fixed, and washed. Cells shown in (A) were then assessed for capacity to bind CD41-APC labeled platelets by flow cytometry. Free platelets were excluded from analysis by forward/side light scatter characteristics. Platelets from P-selectin-/- mice were assayed in parallel to resolve P-selectin dependent platelet adhesion to responding T cells. Expected P-selectin ligand displayed in parentheses. Mean and standard deviation of percentage CD41-APC+ (platelet bound) cells indicated for P-selectin+/+ platelets and P-selectin-/- platelets. P-values for differences in P-selectin+/+ platelet binding determined by Student’s T test, with confidence levels reaching 2, 3 or 4 SD as indicated by **,***,**** respectively. Fig 4B is representative five independent analyses. (C) PSL2 supports adhesive interaction with immobilized P-selectin. Tracking dye-labeled donor T cells responding in OT1→ PSGLnull, OT1→ C2GnT1null, OT1 C2GnT1null→ PSGLnull, and OT1 C2GnT1null → C2GnT1null adoptive transfers yielded activated cells expressing PSGL1 and PSL2, PSGL1, PSL2, or neither ligand respectively. These cells were compared for adhesion to P-selectin immobilized on V-bottom wells under shear generated by centrifugation. Wells were coated with blocking agent alone (no P-selectin) or with PSel-hIgG chimera (immobilized P-selectin). Cells were also centrifuged in wells pre-coated with P-selectin but in the presence of EDTA to prevent selectin mediated binding (immobilized P-selectin + EDTA). Non-adherent cells pelleting to the nadir were harvested directly and cells adhering to the tapered well sides were subsequently harvested using EDTA elution (Adherent). Donor derived activated T cells in non-adherent vs adherent fractions were quantified by flow cytometry-based counting and each normalized to input cell numbers. Triplicate counts of pooled quadruplicate samples were used to generate mean and standard deviation values of cell yields as shown. The significance of differences in cell yields were determined by applying two-tailed Student’s T-tests with equal variance. P-values reaching 1–5 sigma levels of significance were indicated by *, **, ***, ****, ***** respectively. The results shown are representative of four independent analyses.
Fig 5.
PSL2 transfer from resident lymph node cells to activated T cells.
(A) CD8+ T cells responding in C2GnT1null recipients lack PSL2 but can be stained with P-selectin after in vitro mixing with C2GnT1+ lymph node cells. OT1C2GnT1null donor cells responding in peripheral lymph node of PSGL1null recipients at day 3 were stained with P-selectin to detect PSL2 (left panel). The same donor cells responding in C2GnT1null recipients fail to stain with P-selectin (center panel, ie. lack PSL2) but can be stained with P-selectin (right panel) after mixing with whole peripheral lymph node cells from the OT1C2GnT1null → PSGL1null adoptive transfer (re: Methods). The contribution of recipient:donor, or donor:donor, doublets was excluded during flow cytometry by use of doublet discrimination and distinct fluorescent tracking dyes for each responding donor cell preparation. Triplicate P-selectin staining is shown along with the average geometric mean fluorescence and standard deviation. Fig 5A is representative of five independent analyses. (B) L-selectin expression on activated CD8+ T cells is required for acquisition of P-selectin ligand. P-selectin staining of OT1-C2GnT-/- (L-selectin+/+) CD8+ T cells responding in peripheral lymph node of PSGL1null recipients (OT1C2GnT1null L-selectin+/+ → PSGL1null) detects PSL2 (left panel). Center panel: P-selectin staining of donor T cells responding in spleen after adoptive transfer of OT1-C2GnT1null L-selectin+/+ → C2GnT1null (‘PSL2-deficient’), before and after mixing/pelleting with whole peripheral lymph node cells from the ‘PSL2-competant’ OT1C2GnT1null L-selectin+/+ → PSGL1null adoptive transfer. P-selectin staining is enhanced after mixing when responding CD8+ donor T cells are L-selectin+/+ (center panel) but not when they are L-selectin-/- (right panel). As in Fig 5A, the potentially confounding possibility of recipient:donor, or donor:donor, doublet formation in the analysis was excluded during flow cytometry using doublet discrimination and since each responding donor cell preparation had been labeled with distinct fluorescent tracking dyes. Triplicate P-selectin staining is shown along with the average geometric mean fluorescence and standard deviation. Fig 5B is representative of four independent analyses. (C) Graphical summary of geometric mean P-selectin staining from flow cytometry data shown in Fig 5B. with corresponding color code. Standard deviation of triplicate stains shown by error bars. The significance of differences in P-selectin staining of responding donor-derived LSel+/+ and LSel-/- CD8+ T cells from spleen were determined before and after mixing with ‘PSL2-competant’ cells by applying two-tailed Student’s T-tests. P-values reaching 4 (****, P value .00004) and 2 (**, P value .01) sigma levels of significance are indicated. (D) P-selectin ligand detected on resident lymph node cells is removed with EDTA. P-selectin staining of whole lymph node cells 3 days after adoptive transfer of CFSE-labeled donor HY-C2GnT1null into a male PSGLnull recipient. CD8+ and CD8- cells of recipient and donor origin were analyzed either directly (NT) or after applying the same EDTA washing conditions used to strip PSL2 (post-EDTA). As also shown in Fig 2A, responding CFSElow donor cells Fig 5D (b) express PSL2 while non-responding CFSEhigh donor cells Fig 5D (a) lack it. A fraction of recipient CFSE-negative CD8+ and CD8- lymph node cells (c) and (d) respectively, were stained with P-selectin (left panels). P-selectin staining of these recipient cells was lost if pre-washed with EDTA (post EDTA, right panels), as observed for donor-derived activated T cells (subset ‘b’ versus ‘e’. Fig 5D is representative of more than three independent analyses. (E) Memory phenotype CD8+ cells in ‘naïve’ peripheral lymph node cells. Gated analysis of the CD8+ peripheral lymph node cells from untreated PSGL1null mice for expression of CD44 and Ly6c. Fig 5E is representative of three independent analyses. (F) Within the CD8+ subset of untreated lymph node cells, only memory phenotype cells express PSL2. P-selectin staining of CD8+ lymph node cells with gating as shown in Fig 5E for CD44highLy6chigh, CD44lowLy6chigh, and CD44lowLy6clow subsets from naïve PSGL1null or C2GnT1null mice, either before or after EDTA pre-washing (EDTA). Standard deviation of triplicate stains shown by error bars. The significance of differences in P-selectin staining of the indicated CD8+ T cell subpopulations were determined by applying two-tailed Student’s T-tests with equal variance. P-values for specific comparisons were marked (* indicating P values ≤ 0.0001). Fig 5F is representative of three independent analyses.
Fig 6.
Model of P-selectin ligand expression and activity on primary activated CD8+ T cells consistent with prior literature and data.
PSL2 structure: PSL2 association with activated T cells requires calcium and is quantitatively dependent on L-selectin. PSL2 is likely an L-selectin ligand produced by lymph node HEV and bearing 6-O-sulfo sLex, the L-selectin-binding structure shared by all known L-selectin ligands. 6-O-sulfo sLex is also likely present on the exposed, unengaged, face of PSL2 since Meca79 staining correlated with PSL2 expression on both recipient and responding-donor CD8+ T cells (Figs 3F and 4D); Meca79 recognizes 6-O-sulfo sLex on core 1 extensions. P-selectin recognition of the canonical P-selectin ligand PSGL1 requires both sLex and proximal tyrosine sulfation; whether PSL2 bound to L-selectin on T cells also presents sLex/Tyr-SO4 for P-selectin binding or whether the externally exposed 6-O-sulfo sLex detected by Meca79 can substitute as P-selectin ligand is unknown. In either case PSL2 must present (multiple?) selectin ligands on two faces, available for concurrent binding of L-selectin and P-selectin, and most consistent with PSL2 being a mucin-type glycoprotein. It is possible that PSL2 can only serve as a P-selectin ligand after binding/stabilization by L-selectin. PSL2 loading onto activated T cells. PSL2 loading onto CD8+ T cells occurs most clearly after their activation in lymph nodes consistent with available literature that activation is accompanied by enhanced functional affinity of L-selectin for its ligands, perhaps through L-selectin dimerization. L-selectin on activated cells effectively acquires PSL2 by extraction from either resident lymphocytes (Fig 5) or from the fluid phase after its presumed secretion by HEV. Contribution of PSL2 to P-selectin driven signaling and adhesion. Both L-selectin and PSGL1 are thought to co-reside at tips of microvilli. Once bound to L-selectin, sLex glycans on PSL2 may support L-selectin relocation to lipid raft signaling platforms and in proximity to PSGL1. Upon engagement of P-selectin on activated platelets or activated endothelia, the two proximate P-selectin ligands may provide avidity enhancements along with distinct signaling inputs through PSGL1 and L-selectin supporting integrin activation and chemokine responses respectively.