|Home | About | Journals | Submit | Contact Us | Français|
Alloreactive T cells are crucial for graft-versus-host-disease (GVHD) pathophysiology, and modulating their trafficking patterns has been efficacious in ameliorating experimental disease. We report here that P-selectin, a glycoprotein found on resting and inflamed endothelium, is important for donor alloreactive T cells trafficking into GVHD target organs such as the intestines and skin. Compared with wildtype recipients of allogeneic bone marrow transplantation (allo-BMT), P-selectin−/− recipients exhibit decreased GVHD mortality and decreased GVHD of the skin, liver and small bowels. This was associated with diminished infiltration of alloactivated T cells into the Peyer's Patches and small bowels, coupled with increased numbers of donor T cells in the spleen and secondary lymphoid organs (SLO). Surprisingly however, donor T cells deficient for PSGL1, the most well-described P-selectin ligand, mediated similar GVHD as WT T cells, and accumulated in SLO and target organs in similar numbers as WT T cells. This suggests that P-selectin may be required for trafficking into inflamed tissues but not SLO, and that donor T cells may utilize multiple P-selectin ligands apart from PSGL1 to interact with P-selectin and traffic into inflamed tissues during GVHD. We conclude that targeting P-selectin may be a viable target for GVHD prophylaxis or treatment.
Alloreactive donor T cells play an important role in the pathophysiology of GVHD, which is a systemic T cell-mediated disease with specific involvement of the intestines, liver, and skin. The infiltration of alloreactive T cells into target organs is an important step in GVHD pathophysiology, and modulation of T cell trafficking represents a promising strategy for GVHD prophylaxis or treatment1-8.
T cells and other leukocytes exit the bloodstream and enter into tissues via a series of regulated steps. In a simplified model, these can be divided into (1) tethering and rolling via the selectins and their ligands, (2) integrin activation upon chemokine ligand/receptor interactions, (3) firm adhesion via high-affinity integrins, and (4) extravasation via molecules including CD31, CD99 and the junctional adhesion molecules (JAMs)9.
P-selectin is one of a family of three glycosylated lectins (E, L, and P-selectin). It is constitutively expressed on vascular endothelium of the skin and bone marrow, and inducibly expressed on other endothelial cells during inflammation. P-selectin is a receptor for PSGL1, a glycoprotein highly expressed on all leukocytes9, as well as other less well-defined ligands, generally believed to be sialyl lewis x (sLex) bearing glycoproteins10-12. Leukocyte interactions with P-selectin are important for tethering and rolling along endothelium, and interference with leukocyte–P-selectin interactions have shown benefit in experimental models of ischemia-reperfusion injury, allergic airway disease, and bleomycin-induced pulmonary fibrosis13-15. This led to us to hypothesize that P-selectin–ligand interactions may be relevant for leukocyte trafficking in acute GVHD.
PSGL1 mRNA is upregulated during GVHD in multiple models16-18. Additionally, donor T cells localizing into GVHD target organs can upregulate PSGL119, although one report by Sykes and colleagues indicated that adoptive transfer of whole splenocytes lacking functional PSGL1 appears to result in intact GVHD20, underlining the complexity and potential redundancy of interactions between P-selectin and its multiple ligands.
Here we show that compared to wildtype (WT) recipients, P-selectin−/− recipients of allo-BMT have improved survival and diminished clinical GVHD morbidity and mortality, as well as attenuated target organ GVHD. Donor T cells in P-selectin−/− recipients were found in greater numbers in the spleen and lymph nodes, but in diminished numbers in the Peyer's Patches and small bowels. However, cognate experiments with the transfer of PSGL1−/− donor T cells into irradiated allo-BMT recipients resulted in comparable GVHD severity, indicating that other P-selectin ligands on donor alloreactive T cells may also be important for trafficking during GVHD.
Our results suggest a requirement for vessel P-selectin for the infiltration of GVHD target tissues by donor alloactivated T cells, and that donor alloreactive T cells during GVHD may utilize multiple P-selectin ligands in addition to PSGL1.
LP, B10.BR, C57BL/6 (B6), B6 Ly5.1+, B6D2F1, BALB/c mice, and PSGL1−/− and P-selectin−/− mice on the B6 background were obtained from Jackson Laboratories (Bar Harbor, Maine). Memorial Sloan-Kettering Cancer Center's (MSKCC) Institutional Animal Care and Use Committee approved all animal protocols. Mice were transplanted as previously described1 and housed in the MSKCC specific pathogen free barrier facilities. All bone marrow transplant recipients were monitored daily for survival and scored weekly for weight loss and signs of clinical GVHD1.
B6 and P-selectin−/− mice received 11 Gy total body irradiation as a split dose three hours apart. B6D2F1 mice received 13 Gy total body irradiation as a split dose.
We used Ly9.1 (CD229.1) to differentiate donor LP T cells (Ly9.1+) from B6 recipient cells (Ly9.1−). Ly5.1 (CD45.1) was used to identify B6 Ly5.1+ cells. The marker H-2Dd was used to identify B6D2F1 host cells. In experiments with B10.BR donor T cells, the marker used was H-2Kk.
Leukocytes were washed, resuspended in staining buffer (PBS + 0.5% BSA + 2mM EDTA). Cells were stained with DAPI, Ly9.1-FITC, Ly9.1-biotin, H-2Dd-FITC, H-2Dd-PE, H-2Kk-FITC, H-2Kk-biotin, Ly5.1-PE, Ly5.1-biotin, CD3ε-APC-Cy7, CD4-Pacific Blue, CD8-PerCP, CD25-PE-Cy7, CD44-Alexa-700, CD62L-PE-Texas Red, CD45- PerCP-Cy5.5, Annexin-V-PE, Ki-67-PE, PSGL-1-PE and FoxP3-PE, and analyzed on an LSR II with DiVA 6.1 (BD Biosciences, San Diego, CA.)
In some experiments, cells were stained with recombinant P-selectin-Fc-fusion and E-selectin-Fc-fusion protein (R&D systems, Minneapolis, MN) and detected with anti-human-IgG-APC (Jackson ImmunoResearch, West Grove, PA).
Anti-CD44 was obtained from Biolegend (San Diego, CA). DAPI and anti-CD62L were obtained from Invitrogen (Carlsbad, CA). Anti-FoxP3 antibody was obtained from eBioscience (San Diego, CA), and intracellular staining performed according to the manufacturer's protocols.
Flow cytometry data was analyzed in FlowJo v8 (TreeStar Software, Ashland, OR).
Magnetically purified splenic T cells (>90% purity by CD3 staining) were labeled with 1μm Carboxyfluorescein succinimidyl ester (CFSE) from Invitrogen for 20 minutes, washed twice, and infused intravenously into irradiated recipients.
Stimulator cells were obtained via CD5+ magnetic bead selection (Miltenyi Biotec, Auburn, CA) and verified to be >90% pure for T cells by flow cytometry after staining with CD3-FITC. Responder cells were obtained from the spleens of non-transplanted young female BALB/c and B10.BR mice, and were depleted of red blood cells via hypotonic lysis, and depleted of T cells via CD5+ magnetic bead selection. Responders were irradiated with 20 Gy from a 60Co source. 105 stimulators were plated with 105 responders in RPMI 1640 + 20% FCS + glutamine + non essential amino acids + penicillin & streptomycin, and incubated at, 5% CO2 for six days, and 1μCi of 3H thymidine added to each well for 24 hours before reading.
Livers were flushed in situ with 5 mL PBS, transferred into 10 mL PBS with 2 mg/mL collagenase D, cut into small pieces, and incubated for 45 minutes at 37 °C on a shaker. The pieces were then gently mashed through a 70 μm cell strainer in PBS with 0.5% BSA, centrifuged, and the pellet resuspended in 7 mL of 30% Histodenz (Sigma-Aldrich, St. Louis, MO) and layered on top of 2 mL RPMI. This gradient was then centrifuged at 1,500 × g for 20 minutes, and the endothelial-cell rich interphase was then obtained and washed before use.
Livers were mashed in PBS + 0.5% BSA, centrifuged at 300×g for 5 minutes, and the pellets resuspended in 40% Percoll in PBS (Sigma-Aldrich), and layered on top of 70% Percoll in PBS for centrifugation at 400 ×g for 30 minutes at 4 °C. The lymphocyte-rich infranate was then obtained and washed with PBS + 0.5% BSA before use.
Cytokines were analyzed via Cytometric Bead Array (BD Biosciences) per the manufacturer's directions. Linear regression curves were analyzed in Microsoft Excel.
Blood was analyzed on a Hemavet (Drew Scientific, Waterbury, CT).
Survival was calculated with the Mantle-Cox log-rank test. We made all other comparisons with the Mann-Whitney U. Calculations were performed in Prism (Graphpad Software, La Jolla, CA).
We chose a clinically-relevant, well-defined MHC-matched minor-antigen-mismatched strain combination LP→B6 (H-2b) to study the role of P-selectin in GVHD. We first assessed P-selectin express in vessels of the skin and liver in mice with and without GVHD (allogeneic: LP→B6; syngeneic: B6→B6). This revealed that on day 7 post-transplant, endothelial cells (CD45-Ter119-CD11b-CD31+) from the spleen and liver in both recipients of syngeneic and allogeneic BMT displayed P-selectin (Figure 1A-B). By contrast, endothelium from non-transplanted B6 mice displayed less P-selectin (Figure 1C). Moreover, hepatic endothelium also displayed substantial levels of E-selectin on day 7 post-transplant (Figure 1D). These data suggest that BMT-associated inflammatory processes, potentially related to the conditioning regimen, can upregulate selectin molecules on vessel endothelium post-transplant.
To test the importance of P-selectin in GVHD development, we transplanted WT or P-selectin−/− recipients with LP T cell-depleted bone marrow (TCD-BM) and T cells, and observed that P-selectin−/− recipients had attenuated GVHD morbidity and mortality (Figure 2A-B), suggesting that P-selectin in allo-BMT recipients could be important for GVHD pathophysiology. We also noted intact engraftment of TCD-BM in P-selectin−/− recipients (Figure 2A and not shown).
PSGL1 is an important (though not sole) ligand for both E- and P-selectin9,11. We therefore assessed the ability of PSGL1−/− donor alloreactive T cells to cause GVHD in WT allo-BMT recipients. When we performed cognate experiments with WT or PSGL1−/− donor T cells in the parent → F1 model B6 (H-2b) → B6D2F1 (H-2b/d) with 5×106 B6 TCD-BM + 2×106 WT or PSGL1−/− T cells, we were surprised to note similar survival and GVHD morbidity between recipients of WT and PSGL1−/− donor T cells (Figure 2C-D). These observations however, correspond well to a report suggesting that functional inactivation of PSGL1 in donor splenocyte allografts does not significantly ameliorate GVHD20.
We next tested whether PSGL1−/− T cells could still bind P-selectin. When we stained WT and PSGL1−/− splenic T cells from donor mice for P-selectin and E-selectin ligands, we found that PSGL1−/− donor T cells (as expected) did not express PSGL1, but still expressed substantial levels of P-selectin ligands, at levels comparable with those of WT T cells (Supplemental Figure 1). By contrast, we found no appreciable levels of E-selectin ligands (Supplemental Figure 1).
We assessed the functionality of donor T cells in WT and P-selectin−/− allo-BMT recipients, by adoptively transferring purified (CD5+) CFSE-labeled LP splenic T cells into irradiated WT and P-selectin−/− recipients. On day 5, we observed increased numbers of CFSElo fast-proliferating allo-activated CD4 T cells in the spleens of P-selectin−/− recipients as compared with WT recipients (Figure 3A), suggesting that alloreactive T cells selectively accumulate in the spleens of P-selectin−/− recipients. Rapidly-proliferating CFSElo alloactivated T cells in P-selectin−/− and WT recipients displayed similar levels of CD25, CD44, and CD62L (Figure 3B), suggesting that T cell alloactivation is intact in P-selectin−/− recipients.
We evaluated donor T cell apoptosis in WT and P-selectin−/− recipients in two models, B10.BR (H-2k) → B6 (H-2b) as well as LP→ B6. We adoptively transferred CFSE-labeled LP or B10.BR donor splenic T cells into irradiated allogeneic B6 or P-selectin−/− recipients, and then analyzed recipient spleens on day 3 (B10.BR T) or day 5 (LP T). These experiments revealed similar levels of T cell apoptosis (not shown).
To directly assess the alloreactivity of donor T cells in WT vs. P-selectin allo-BMT recipients, we recovered purified T cells from the spleens and livers of mice with GVHD on day 28 post-transplant (5 × 106 LP TCD-BM + 1×106 LP T → B6 or P-selectin−/−), and performed mixed lymphocyte reactions with irradiated B6 (allogeneic) and BALB/c (third-party) T cell-depleted stimulators. Donor alloreactive T cells in WT vs. P-selectin−/− transplant recipients had similar alloreactivity in vitro (Figure 3C); this suggests that the large population of alloreactive T cells found in the spleen in P-selectin−/− recipients of CFSE-labeled T cells (Figure 3A) reflects the accumulation of these T cells, and not differences in activation or proliferation.
We also found similar numbers of splenic donor CD4+CD25+FoxP3+ regulatory T cells on day 14 after allo-BMT in WT and P-selectin−/− recipients (not shown). Additionally, serum TNF and IFNγ levels were similar on day 14 post-transplant (not shown).
PSGL1 may interact with other receptors in addition to P-selectin, and we therefore assessed PSGL1 expression on donor CFSElo alloactivated T cells in WT and P-selectin−/− recipients, to evaluate whether donor T cells may have a compensatory upregulation of PSGL1 when placed into a P-selectin−/− recipient. This revealed that donor CFSElo fast-proliferating alloactivated T cells in P-selectin−/− recipients had similar or decreased levels of cell surface PSGL1 compared with those in WT recipients (Figure 3D).
To assess the alloreactivity and proliferation of WT vs. PSGL1−/− T cells, we adoptively transferred WT and PSGL1−/− splenic T cells into irradiated allogeneic (B6D2F1) recipients. We observed that WT and PSGL1−/− alloreactive T cells in the spleen showed similar patterns of proliferation and activation (CD25, CD44, CD62L). (not shown).
We also assessed the in vivo expansion and trafficking of PSGL1−/− T cells in GVHD in the model B6 TCD-BM + B6 WT vs. PSGL1−/− donor T → B6D2F1. On day 14, we found similar numbers of donor CD4 and CD8 T cells in the spleen, liver, MLN, and PLN in recipients of WT vs. PSGL1−/− T cells (not shown).
We also evaluated levels of P-selectin ligand on donor WT and PSGL1−/− T cells on day 14 post-transplant, and noted that PSGL1−/− alloreactive T cells in the spleen had a modest decrease in levels of cell surface P-selectin ligand, but that PSGL1−/− alloreactive T cells in the liver, MLN, and PLN had similar levels of cell surface P-selectin ligand as WT T cells (Supplemental Figure 2). These P-selectin ligands may be relevant for the trafficking of PSGL1−/− T cells during GVHD.
As donor alloreactive T cells appeared to display similar activation, proliferation, apoptosis, and PSGL1 expression in irradiated WT and P-selectin−/− allo-BMT recipients, we next assessed their accumulation in lymphoid and non-lymphoid tissues. Non-transplanted P-selectin−/− mice have similar numbers of CD4 and CD8 T cells in the spleen, MLN, PLN, and Peyer's Patches (PP) as corresponding WT animals, but decreased numbers of CD4 T cells in the liver (not shown). Twenty-four hours after lethal radiation (11 Gy, split dose), we observed comparable numbers of T cells in the spleen, liver, MLN, PLN, and PP of WT and P-selectin−/− animals (not shown).
Finally, upon enumerating donor infiltrating T cells in lymphoid tissues and GHVD target organs of allo-BMT recipients on day 14 post-transplant, we found increased numbers of CD4 and CD8 effector (CD44+CD62L−) and central memory (CD44+CD62L+) T cells in the spleen, mesenteric lymph nodes (MLN), and peripheral lymph nodes (PLN) of P-selectin−/− recipients (Figure 3E). This was associated with decreased numbers of donor T cells in the Peyer's Patches (PP) and epithelium of the small bowels (IEL) in P-selectin−/− recipients (Figure 3E). Complete blood counts also revealed increased numbers of circulating lymphocytes in the blood of P-selectin−/− recipients on days 7 and 14 post-transplant (Figure 3E).
These findings were confirmed via histopathological analysis at a later time-point (day 35). When we analyzed the pathological sub-scores for lymphocytic infiltrates, P-selectin−/− recipients exhibited significantly decreased lymphocytic infiltrates into the small bowels (Figure 3F, left). At this later time point, we also observed a non-significant trend towards decreased lymphocytic infiltrates in the liver. Finally, we also observed decreased neutrophilic infiltrates into the small bowel, and a trend towards decreased neutrophilic infiltrates into the liver, at day 35 after allo-BMT (Figure 3F, right), which may be due to the function of P-selectin in neutrophil tethering and rolling21; consequently P-selectin-deficient endothelium may also impair neutrophil trafficking.
We assessed damage to GVHD target organs in WT and P-selectin−/− allo-BMT recipients by histopathology, and observed a decrease in hepatic and small bowel GVHD on day 35 post-transplant (Figure 4A-C). On day 35 post-transplant, P-selectin−/− recipients also showed significantly decreased numbers of apoptotic cells in the skin, indicating diminished cutaneous GVHD (Figure 4D-E).
In this manuscript, we demonstrate that P-selectin−/− allo-BMT recipients are resistant to the development GVHD, which suggests that P-selectin of recipient origin is an important molecule for GVHD pathophysiology. Furthermore, in cognate experiments, we observe that donor T cells deficient for PSGL1, the most well-described ligand for P-selectin, had a surprisingly intact potential to cause GVHD, suggesting that donor T cells can use multiple ligands, in addition to PSGL1, to mediate their interactions with P-selectin on vessel endothelium during the inflammatory processes associated with acute GVHD.
First, we observed that alloreactive T cells in WT vs. P-selectin−/− allo-BMT recipients demonstrate comparable levels of alloreactivity (activation markers in vivo and T cell proliferation in vitro), apoptosis, and expression of PSGL1. However, upon enumerating donor T cells in mice with GVHD, we noted that compared to the situation in WT recipients, donor alloreactive T cells in P-selectin−/− recipients accumulated in increased numbers in lymphoid tissues, and decreased numbers in GVHD target tissues such as the small intestine. By contrast, non-transplanted WT B6 and P-selectin−/− animals have fairly similar numbers of T cells in lymphoid and non-lymphoid organs.
The observation that P-selectin−/− recipients had more donor alloactivated T cells in the spleen, SLO, and peripheral blood after allogeneic BMT was initially surprising, in light of the enhanced survival of these mice. Yet while P-selectin has been implicated in the trafficking of T cells into inflamed organs11,22,23, no data exist regarding its involvement in lymphoid tissue trafficking, which is instead mediated by the peripheral lymph node addressins24 (PNAds), L-selectin25, CCR726 and LFA-127. We therefore believe that P-selectin may be important for the trafficking of alloreactive T cells into non-hematopoietic tissues such as the gut or liver during GVHD, but relatively dispensable for the trafficking of T cells into lymphoid organs.
We were surprised to note that despite a difference in overall mortality, there was no difference in colonic GVHD pathology or degree of lymphocytic or neutrophilic infiltrates in the colons of WT and P-selectin−/− allo-BMT recipients. However, a number of reports may explain this finding. The first is the observation that in murine models of chronic colitis, T cells require CD18, but not PSGL1, to cause disease, suggesting that in our model systems, there simply may not be a direct requirement for P-selectin for colonic GVHD28. The second are reports on ulcerative colitis29 and mouse models of dextran sodium sulfuate (DSS) colitis30, which indicate that P-selectin, as expressed on circulating platelets, is important for the co-recruitment of leukocytes to the colon. Platelets utilize P-selectin to bind endothelium expressing PSGL1, and leukocyte/platelet aggregates are then subsequent required for leukocyte adhesion in colonic vessels. Indeed, neutralization of platelets reduced leukocyte adhesion in mouse DSS colitis models. Since allo-BMT recipients with GVHD establish full donor chimerism, circulating platelets in our model systems are expected to be of donor origin, and thus P-selectin+/+; consequently, they would not be expected, according to these reports, to mediate defective leukocyte trafficking to the colon.
Taken together, P-selectin−/− recipients of allo-BMT exhibit diminished systemic, cutaneous, and gastrointestinal GVHD, coupled with increased numbers of donor alloactivated T cells in the spleen and SLO, and decreased numbers of infiltrating donor T cells in the small bowels. This increased cellularity in lymphoid tissues could be due to either a defect in the exit of alloreactive T cells from these SLO, or a defect in their trafficking into GVHD target organs in P-selectin−/− recipients.
Sykes et al. have shown that the sphingosine-1-phosphate receptor agonist FTY720 can sequester T cells in SLO and away from target organs, thus attenuating GVHD5. Yet the differential accumulation of donor alloactivated T cells in lymphoid vs. non-lymphoid organs in the present study may be due to different requirements in the selectins for entry into these two types of organs. The inability of T cells to enter GVHD target organs without P-selectin, and the requirement for L-selectin/PNAd interactions to enter lymphoid tissues, may explain why P-selectin−/− allo-BMT recipients had diminished infiltrates into the small bowels coupled with increased numbers of donor T cells in the spleen, peripheral and mesenteric lymph nodes.
In parallel experiments, we assessed the importance of PSGL1, the most well-described leukocyte ligand for P-selectin, for GVHD. Surprisingly, we observed that PSGL1−/− donor T cells caused not only similar GVHD morbidity and mortality as WT T cells, but also that these T cells had similar proliferation, alloactivation, and infiltration into the SLO and liver as WT T cells.
Despite being ablated for PSGL1, PSGL1−/− T cells appear to still express substantial levels of other P-selectin ligands (Supplemental Figures 1 and and2),2), and these other ligands may interact with P-selectin to compensate for PSGL1 deficiency in the setting of acute GVHD. Indeed, non-transplanted WT and PSGL1−/− mice displayed similar leukocyte cellularity in lymphoid and non-lymphoid tissues, and numbers of WT and PSGL1−/− donor T cells were also comparable in lymphoid and non-lymphoid tissues post-transplant in mice with GVHD.
In conclusion, our report suggests although recipient P-selectin is an important molecule for the pathophysiology of GVHD, in strongly inflammatory settings such as acute GVHD, multiple P-selectin ligands on donor T cells may be important for their trafficking and tissue infiltration. Consequently, ablation of PSGL1 alone on donor T cells may not be sufficient to abrogate their interactions with recipient P-selectin. Our results suggest that targeting P-selectin in allo-BMT recipients, or multiple P-selectin ligands on donor T cells and leukocytes, may represent a novel therapeutic strategy for GVHD prophylaxis or treatment.
The authors thank the staff of the Memorial Sloan-Kettering Cancer Center Research Animal Resources Center for excellent animal care.
Financial Support: This research was supported by National Institutes of Health award numbers RO1-HL069929 (MvdB), RO1-CA107096 (MvdB), RO1-AI080455 (MvdB), PO1-CA33049 (MvdB) and R01-HL095075 (MvdB). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Support was also received from the US Department of Defense: USAMRAA Award W81XWH-09-1-0294 (MvdB), the Ryan Gibson Foundation (Dallas, TX), the Elsa U. Pardee Foundation (Midland, MI), the Byrne Foundation (Etna, NH), the Emerald Foundation (New York, NY), and The Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center funded by Mr. William H. Goodwin and Mrs. Alice Goodwin (New York, NY), the Commonwealth Foundation for Cancer Research (Richmond, VA), The Bobby Zucker Memorial Fund (Phoenixville, PA) and The Lymphoma Foundation (New York, NY).
O.A. is the recipient of an Amy Strelzer Manasevit Scholar Award from The National Marrow Donor Program (NMDP) and The Marrow Foundation and is supported by grant P20-CA103694 from the National Institutes of Health. J.L.Z. is the recipient of a fellowship grant from the Lymphoma Research Foundation. I.K.N. is supported by the Deutsche Krebshilfe, Mildred-Scheel-Stiftung. O.P. is supported by the Deutsche Forschungsgemeinschaft (DFG).
Author contributions: S.X.L. and A.M.H. designed and performed experiments, analyzed data, and wrote the paperI.K.N. and T.H.T. designed and performed experiments
O.A. designed experiments
J.L.B., O.M.S., D.S., J.L., C.K., A.K., V.M.H., U.K.R., N.Y., R.J., J.L.Z., O.P., L.D., and K.B. performed experiments
C.L., A.C.L., and G.M. performed experiments and analyzed data
L.P., B.F., and B.F. provided valuable reagents
M.vdB. designed experiments, analyzed data, and wrote the paper