T cell development requires periodic importation of hematopoietic progenitors into the thymus. The receptor-ligand pair P-selectin and P-selectin glycoprotein ligand 1 (PSGL-1) are critically involved in this process. In this study, we examined the expression of functional PSGL-1 on bone marrow hematopoietic progenitors. We demonstrate that functional PSGL-1 is expressed at low levels on hematopoietic stem cells, but upregulated on the cell surface of progenitors that bear other homing molecules known to be important for thymic settling. We found that progenitors able to home to the thymus expressed high levels of PSGL-1 transcripts compared with hematopoietic stem cells. We further demonstrate that hematopoietic progenitors lacking fucosyltransferase 4 and 7 do not express functional PSGL-1, and do not home efficiently to the thymus. These studies provide insight into the developmentally regulated expression of a critical determinant involved in progenitor homing to the thymus.
T cells develop in the thymus; however, the thymus contains no self-renewing progenitors. T lymphopoiesis depends on importation of rare bone marrow (BM) precursor cells, which express chemokine receptors required for efficient thymic homing. These thymus homing progenitors are developmentally downstream of hematopoietic stem cells (HSC) (1–5). HSC are contained within a BM population with the cell surface phenotype of lineage-marker−Kit+Sca1+ (LSK). Expression of the cytokine receptor fms-like tyrosine kinase receptor-3 (Flt3) divides LSK cells further into populations enriched for self-renewing HSC (Flt3−LSK) (6), nonrenewing multipotent progenitors (MPP) (LSKFlt3low) (2), and lymphoid-primed MPP (LMPP) (LSK Flt3high). In this study, LMPP that express CCR9 are referred to as CCR9+ lymphoid progenitors (CCR9+ LP); these cells coexpress CCR7 and are likely to represent efficient thymic homing progenitors (7–10). Further downstream of LMPP, common lymphoid progenitors (CLP) possessing the phenotype Lin−Sca1lowKitlowFlt3highIL-7Rα+ have been shown to express high levels of CCR7, and a subset also expresses CCR9 (8, 10). All of these progenitor types have the potential to generate T cells (8). It has recently been demonstrated that CLP are heterogeneous, and the Ly6D− subset that expresses low level of RAG1 has been suggested to contribute to T cell generation more efficiently compared with downstream Ly6D+ RAG1high cells (11, 12).
Recruitment of BM-derived precursors to the thymus has been suggested to depend on the multistep engagement of adhesion molecules and chemoattractants (13). Progenitors egress the BM, circulate in the bloodstream, and, finally, a select subset of these cells settles in the thymus for T cell development. Although many progenitors possess T cell potential, only a select subset with T lineage potential can physiologically settle in the thymus (8). The most efficient thymic settling progenitors appear to be contained within a small subset of LMPP and CLP progenitors, but not HSC and MPP (8, 14). In the thymus, thymic settling progenitors generate early thymic progenitors or early T lineage progenitor (ETP), the earliest described and most efficient intrathymic T progenitors. ETPs give rise to CD4−CD8− double-negative (DN)2 and DN3 cells. DN3 cells differentiate into CD4+CD8+ double-positive (DP) cells, which subsequently generate CD4, or CD8 single-positive cells, and then exit the thymus to the periphery.
The ability of specific hematopoietic progenitors with T cell potential to settle the thymus reflects their expression of required homing molecules. Recent work has suggested that the chemokine receptors CCR7 and CCR9 each independently support the migration of progenitors into the thymus (8, 10, 15, 16). Parabiosis and competitive thymus reconstitution assays further indicate that P-selectin and its ligand, P-selectin glycoprotein ligand 1 (PSGL-1), also have a role in recruiting progenitors to the thymus (17, 18). Selectins and their oligosaccharide ligands control lymphocyte trafficking to lymph nodes in normal circumstances (19). P-selectin is an adhesion molecule expressed on the surface of activated endothelial cells (20, 21). PSGL-1 is a transmembrane protein that is a ligand for all three selectin molecules (19). PSGL-1 is a 220-kDa disulfide-linked homodimeric sialomucin expressed on the surface of most leukocytes. Posttranslational modifications of PSGL-1 are required for selectin binding. Functional PSGL-1 is a sialylated, fucosylated, core 2 O-glycan clustered with tyrosine sulfate (22). Glycans that contribute to PSGL-1 activity are constructed by a glycosylation reaction in which terminal steps are catalyzed by α1,3 fucosyltransferase (Fuc-T). However, whether expression of functional PSGL-1 is developmentally regulated on hematopoietic progenitors in parallel to the established regulation of CCR7 and CCR9 is unknown.
In this study, we demonstrate that functional PSGL-1 is expressed on hematopoietic stem cells at very low levels, but upregulated on downstream progenitors that are thought to home to and settle the thymus. We further show that glycosyltransferase enzymes, including Fuc-TVII, specially regulate the formation of functional PSGL-1, thus contributing to T lymphopoiesis.
Materials and Methods
C57BL/6 (CD45.2) and B6.Ly5SJL (CD45.1) mice were purchased from the National Cancer Institute animal facility. PSGL-1 knockout (KO) mice were purchased from The Jackson Laboratory. Rag1-Cre mice (23) were a gift of T. Rabbitts (University of Cambridge). Fuc-TVII/Fuc-TIV double-knockout (DKO) (24, 25) mice were obtained from J. Harris and P. Scott (University of Pennsylvania). Mice were used when 6–10 wk of age. Live animal experiments were performed according to approved protocols of the Office of Regulatory Affairs of the University of Pennsylvania in accordance with National Institutes of Health guidelines.
Cell preparations, flow cytometry, and cell sorting
Flow cytometric (FACS) analysis was performed on a CANTO or LSRII (BD Biosciences). Dead cells were excluded with DAPI. Doublets were excluded through forward scatter height by forward scatter width and side scatter height by side scatter width parameters. Data were analyzed using FlowJo version 4.6.2 (Tree Star). For cell sorting, BM was first enriched for progenitors, then stained and sorted on a FACSAria II (BD Biosciences). Aliquots of sorted cells were reanalyzed to ensure purity, which was usually >90%.
Competitive adoptive transfer
To generate mixed BM chimeras, host wild-type (WT) mice (expressing CD45.1) were lethally irradiated (9.5 Gy) and subsequently injected with T cell-depleted host-type BM mixed with either WT, PSGL-1 heterozygous, PSGL-1 KO, or Fuc-TVII/Fuc-TIV DKO BM (all expressing CD45.2) at a 1:2 ratio. T cells were depleted from BM by incubation with anti-CD4 and anti-CD8 Abs, followed by removal of Ab-bound cells with magnetic beads (Qiagen).
For i.v. transfers into unirradiated mice, 1 × 105 sorted HSC, LMPP, Ly6D− CLP, and Ly6D+ CLP were injected retro-orbitally. To prevent rejection, mice were given 0.1 mg anti-CD4 (GK1.5) the day before BM transfer and every 4 d thereafter (8, 26). Previous work by our laboratory has shown that this treatment does not affect BM or thymic engraftment by donor cells (8, 10).
TaqMan probes used were as follows: GAPDH (Mm99999915_g1), CCR7 (Mm00432608_m1), CCR9 (Mm02620030_s1), PSGL-1 (Mm01204601_m1), Fut4 (Mm00487448_s1), C2GnT1 (Mm02010556_s1), β4galT1 (Mm00480752_m1), and St3gal4 (Mm00501503_m1).
The primers used for St3gal6 were forward, 5′-GCCCTTTCAAAACTGCAGAG-3′ and reverse, 5′-TCCCAACTTCCTCTTCATGG-3′, and for Fut7 were forward, 5′-CAGTGCTGATGCTGTGGTCT-3′ and reverse, 5′-GCGACCGTAGGGTACAAAGA-3′.
OP9 and OP9-DL4 cell culture
Values are expressed as mean ± SEM. The p values were calculated using Microsoft Excel by Student t test.
PSGL-1 in thymic homing
Previous work has shown the importance of PSGL-1 in T cell development (17). We wished to determine the stage at which PSGL-1 is required in T cell development. We generated competitive BM chimeras by intravenously transferring both PSGL-1 KO BM (CD45.2) and WT BM (CD45.1) at 2:1 ratio into lethally irradiated WT hosts. As a control, we injected mixtures of PSGL-1 WT BM (CD45.2) and WT BM (CD45.1) to reconstitute irradiated mice. After hematopoietic reconstitution at 10 wk posttransplant, the BM and thymi of these mice were analyzed by flow cytometry to determine the donor-derived chimerism at successive stages of T cell development (Fig. 1A, 1B). Generation of ETP was almost comparable in control chimeras to the level expected based on the level of reconstitution by BM LSK progenitors (Fig. 1B, top panel). PSGL-1 KO HSC were able to engraft in the BM and differentiate into downstream MPP, LMPP, and CLP efficiently (Fig. 1B, bottom panel). However, the donor chimerism of early thymic progenitors derived from the PSGL-1 KO BM was significantly lower than the chimerism in the BM LSK (Lin−Kit+Sca1+) progenitors (Fig. 1B, bottom panel). These data confirm that the absence of PSGL-1 conferred a disadvantage in the generation of ETP in this competitive situation.
To determine whether PSGL-1 might be required for efficient responses to T-inductive Notch signaling rather than for thymic homing, sorted LMPP from PSGL-1 KO BM were cultured in vitro on OP9-DL4 stromal cells, which expresses the Notch1 ligand delta-like 4 (DL4) that supports T cell development (27). After 2 wk of culture, PSGL-1 KO LMPP and WT control LMPP generated similar numbers of progeny expressing T lineage markers CD25 and Thy1 (Fig. 1C, 1D). These data suggest that the lower number of ETPs in irradiation chimeras made with PSGL-1 KO BM cells was due to impaired thymic settling rather than defective intrathymic development, consistent with previous reports (17).
Expression of functional PSGL-1 is developmentally regulated in BM progenitors
PSGL-1 is expressed by all subsets of leukocytes (19). The binding of P-selectin chimera Ig (IgG) Fc fusion protein was used to measure functional PSGL-1 levels on the cell surface. BM progenitor populations were gated, as indicated in Fig. 2A. Flow cytometric analysis results revealed that expression of functional PSGL-1 was increased in CCR9+LP compared with HSC populations (Fig. 2B). We found that expression of functional PSGL-1 is low in HSC, gradually increases in MPP, and is highest in CCR9+LP and CLP. Functional PSGL-1 expression appeared bimodal in CLP, consistent with earlier reports of heterogeneity in this population (Fig. 2B) (11, 12). Progenitors previously suggested to settle the thymus, CCR9+LP and CLP, were those with the highest level of functional PSGL-1 on their surface. Interestingly, differences were also detected using anti–PSGL-1 Ab, indicating that there may be developmental regulation of PSGL-1 itself (Fig. 2C). Calculation of the mean fluorescence intensity (MFI) for each population showed that MFI of both functional PSGL-1 (Fig. 2D, left panel) and PSGL-1 (Fig. 2D, right panel) was significantly increased in CCR9+LP and CLP, putative thymic settling progenitors, as compared with HSC (Fig. 2D). These data indicate that populations thought to settle the thymus express increased levels of functional PSGL-1.
The expression of functional PSGL-1 separates CLP populations
Downstream of LMPP, CLP also settle the thymus and possess T cell potentials (8, 10, 29–31). Recently, it has been demonstrated that CLP in BM are a heterogeneous population, containing multiple subsets with different lineage potentials (11, 12). Previous work has assessed the in vivo T potential of CLP subsets in irradiated recipients (12). However, hematopoiesis following irradiation is far from physiologic, as irradiation has many secondary effects (8, 32, 33). To investigate the potential of CLP to undergo T cell development in physiological conditions, we assessed T cell development in unirradiated normal adult recipient mice. For these experiments, HSC, LMPP, Ly6D−CLP, and Ly6D+CLP cells were sorted from the BM of B6 donors, and 1 × 105 cells from each population were injected i.v. into each unirradiated B6.Ly5SJL congenic recipient. Two weeks after i.v. transfer of purified progenitors into mice, recipient thymi were analyzed for donor-derived DN3 and DP thymocytes (Fig. 3A). In all experiments, HSC failed to produce any DN3 or DP thymocytes at this time point, as expected, consistent with earlier work (8). Both LMPP and Ly6D−CLP generated donor-derived thymocytes. The number of donor-derived DN3 and DP from Ly6D+CLP progenitor was significantly lower than that generated by LMPP and Ly6D−CLP (Fig. 3A). These results established that the Ly6D+CLP subset is an inefficient T cell progenitor in the unirradiated physiological thymus environment.
Because Rag-1 is highly expressed in the CLP population (34), we next examined Ly6d expression on CLP cells in Rag-1/YFP mice, which were created by crossing Rag1/Cre mice with mice in which YFP was knocked into the Rosa26 locus. This reporter allows for a more precise analysis of CLP subpopulations (35). Flow cytometric analysis of BM from Rag1-YFP mice showed that CLP expressing high levels of Ly6D also expressed high levels of Rag1-YFP, whereas many fewer Ly6D−CLP expressed YFP (Fig. 3B). About two-thirds of CLP in these experiments were Ly6D+, and, of these, most expressed Rag1-YFP, as expected (11). We sought to analyze the expression of functional PSGL-1 in CLP subsets in the Rag1 reporter mouse. The expression of functional PSGL-1 was decreased by about half in Ly6D+ Rag1-YFP marked cells, compared with Ly6D− Rag1 low cells (Fig. 3C, 3D). The MFI seen with P-selectin fusion protein binding by flow cytometry was significantly decreased as cells progressed from Ly6D−CLP to Ly6D+CLP (Fig. 3E). These results indicate that functional PSGL-1 expression is downregulated within CLP coordinately with the loss of efficient T cell potential.
Requirement of PSGL-1 in thymus settling
Increases in expression of functional PSGL-1 from HSC to CCR9+LP were of the order of 4-fold. Reduction in surface expression as progenitors progressed to Ly6D+CLP stages was similarly modest, 3- to 4-fold. To investigate the biological importance of modest alterations in expression of functional PSGL-1, we examined PSGL-1 heterozygous mice. We first confirmed reduced expression of PSGL-1 transcripts in BM subsets from PSGL-1 heterozygous mice. Quantitative RT-PCR confirmed that PSGL-1 expression was reduced compared with the WT in all BM progenitors examined (Fig. 4A). The MFI of binding by P-selectin fusion protein on PSGL-1 heterozygous CCR9+LP was significantly decreased from WT CCR9+LP cells (Fig. 4B) as determined by flow cytometry. To determine whether a 2-fold reduction could be functionally relevant, we generated competitive BM chimeras by intravenously transferring mixtures of PSGL-1 Het BM (CD45.2) and WT BM (CD45.1) into lethally irradiated WT hosts. As a control, we injected mixtures of PSGL-1 WT BM (CD45.2) and WT BM (CD45.1) to reconstitute irradiated mice. After hematopoietic reconstitution at 10 wk posttransplant, the BM and thymi of these mice were analyzed by flow cytometry to determine the donor-derived chimerism at successive stages of T cell development (Fig. 4C). Generation of ETP in control chimeras was at the level expected based on the level of reconstitution in BM (Fig. 4C, left panel). PSGL-1 Het HSC were able to engraft in the BM and differentiate into downstream MPP, LMPP, and CLP efficiently (Fig. 4C, right panel). However, the donor chimerism of early thymic progenitors derived from the PSGL-1 Het BM was significantly lower than the chimerism in the BM LSK (Lin−Kit+Sca1+) progenitors (Fig. 4C, right panel). These data indicate that a 50% reduction of PSGL-1 conferred a detectable disadvantage in the generation of ETP in this competitive situation, and suggest developmentally regulated differences in expression of functional PSGL-1 are functionally relevant.
Expression of PSGL-1 and glycosyltransferase genes
We next investigated the molecular basis of dynamic regulation of functional PSGL-1 expression on BM progenitors. It is established that glycosyltransferases regulate several key enzymatic steps for biosynthesis of functional PSGL-1. Important glycosyltransferases for this process include the following: core 2 β1,6-N-acetylglucosaminyltransferases (C2GnT1) (36), Fuc-TVII (25), β1,4-galactosyltransferase (β1,4-GalT1) (37, 38), α2,3 sialyltransferase (ST3GalT4), and ST3GalT6 (39, 40). We examined the expression of glycosyltransferase enzymes in BM and early thymic progenitors. Quantitative RT-PCR confirmed that PSGL-1 and the glycosyltransferase enzyme Fuc-TVII were expressed at higher levels in CCR9+LP cells and Ly6D−CLP (Fig. 5), in agreement with their elevated surface expression of functional PSGL-1 (Fig. 2). In contrast, reduced expression of both of these genes in Ly6D+CLP cells (Fig. 5) suggests that both PSGL-1 and Fuc-TVII influence functional PSGL-1 levels in lymphoid progenitor cells, because functional PSGL-1 declines in the Ly6D+CLP population. The glycosyltransferase enzyme sialyltransferase ST3GalT6 is highly expressed in CCR9+LP cells; and sialyltransferase ST3GalT4, β1,4-galactosyltransferase β1,4-GalT1, fucosyltransferase 4 (Fuc-TIV), and core 2 β1,6-N-acetylglucosaminyltransferases C2GnT1 are expressed in CCR9+LP cells and also Ly6D−CLP lymphoid progenitor cells. We also examined expression of CCR7 and CCR9; consistent with previous reports, we found strong expression of CCR7 in CCR9+LP and CLP subsets (Fig. 5). Interestingly, CCR9 was highly expressed in CCR9+LP, as expected, but also in Ly6D+CLP (Fig. 5).
Fucosyltransferase enzyme 7 regulates thymus settling through formation of functional PSGL-1
PSGL-1 becomes functional (i.e., able to bind P-selectin) only after specific α1,3 fucosylation, which is accomplished by a Fuc-T, Fuc-TVII, with Fuc-TIV playing a supporting role (25, 41). We analyzed progenitor populations in the BM of Fuc-TVII/Fuc-TIV double-deficient mice and found that functional PSGL-1 expression was absent on LSK and LMPP progenitor populations (Fig. 6A, left panels), whereas PSGL-1 protein continued to be expressed on the surface of the indicated populations, as expected (Fig. 6B, left panels). The numbers of hematopoietic precursors were comparable in BM from WT and Fuc-TVII/Fuc-TIV DKO mice (Supplemental Fig. 1A). The absence of functional PSGL-1 on progenitor cells from Fuc-TVII/Fuc-TIV double-deficient mice indicates that expression of functional P-selectin on hematopoietic progenitors is dependent on α1,3 fucosylation catalyzed by fucosyltransferase enzymes Fuc-TVII and/or Fuc-TIV, as has been previously described for neutrophils (41).
It has been previously shown that inactivation of Fuc-TVII by homologous recombination results in >80% decrease in lymphocyte homing to peripheral lymph nodes (25). We tested whether diminished expression of functional PSGL-1 on the cell surface of Fuc-TVII/Fuc-TIV double-deficient BM progenitors resulted in a thymic settling defect. We generated competitive BM chimeras by intravenously transferring both Fuc-TVII/Fuc-TIV DKO BM (CD45.2) and WT BM (CD45.1) at a 2:1 ratio into lethally irradiated WT recipients. As a control, we injected mixtures of CD45.2 WT BM and CD45.1 WT BM. After 10 wk to allow for hematopoietic reconstitution, BM and thymi were analyzed by flow cytometry to determine the donor chimerism within defined progenitor populations (Fig. 6C, upper panel). Fuc-TVII/Fuc-TIV double-deficient HSC were able to engraft in the BM and differentiate into MPP, LMPP, and CLP normally; however, the donor chimerism was significantly decreased in early intrathymic T cell progenitors (Fig. 6C, bottom panel). Indeed, progenitors lacking Fuc-TVII/Fuc-TIV showed a severe defect in thymic settling that was comparable in magnitude to the defect seen with PSGL-1−/− mixed BM chimeras (Fig. 1B). Defects in thymic immigration and immunodeficiency have been linked to increases in thymic receptivity mediated by increased frequencies of P-selectin+ thymic endothelial cells (18). However, frequencies of P-selectin+ thymic endothelial cells appeared unaffected in Fuc-TVII/Fut-TIV DKO mice (Supplemental Fig. 1B, 1C). When cultured on OP9-DL4 stromal cells, Fuc-TVII/Fuc-TIV DKO LMPP generated similar number of CD25 and Thy1.2 DP T cells on per cell basis compared with WT controls, indicating equivalent responses to T-inductive Notch signals (Fig. 6D, 6E). These findings establish an important role for fucosylation-mediated functional PSGL-1 in thymic settling in a competitive situation.
We have established that expression of functional PSGL-1 is regulated on BM-derived hematopoietic progenitor cells, and investigated the molecular basis of this regulation. We found that functional PSGL-1 is expressed at low levels on hematopoietic stem cells, but that high levels are found on downstream progenitors that are thought to home to and settle the thymus. Finally, we found that Fuc-TVII/Fuc-TIV–deficient BM progenitors demonstrate thymic settling defects, due to the abrogation of fucosyltransferase-mediated modifications necessary to form functional PSGL-1.
The ability to settle the thymus is selectively achieved by progenitors downstream of HSC, and includes LMPP and CLP (7, 8, 29, 42). One molecular basis of this selectivity was previously shown to be the expression of the chemokine receptors CCR7 and CCR9 on CLP and subsets of LMPP, and progenitors lacking both of these molecules are almost completely unable to contribute to T lymphopoiesis in a competitive situation (8, 10, 15, 16). Earlier work had established that the selectin ligand PSGL-1 also plays a role in facilitating thymic settling and efficient homing of T cells to secondary lymphoid organs (17, 43). We have demonstrated that progenitors lacking PSGL-1 contribute inefficiently to the earliest thymic progenitor populations in competitive irradiation chimeras, thus confirming and extending previous work (17). Interestingly, some donor-derived PSGL-1–deficient cells do still settle the thymus even in this competitive situation, suggesting that other undiscovered molecules may play a role in the absence of PSGL-1.
The formation of functional P-selectin ligand requires the orchestrated action of glycosyltransferases (25, 36, 41). Whereas there are at least five distinct α1,3 fucosyltransferase genes in the human and mouse genome, only two of these, specifically Fuc-TVII and Fuc-TIV, are expressed in leukocytes or their progenitors (38, 44, 45). To investigate the importance of PSGL-1 fucosylation in the migration of progenitors to the thymus, we performed competitive BM transplantation assays using BM from Fuc-TVII/Fuc-TIV double-deficient mice. In competition with WT cells, Fuc-TVII/Fuc-TIV double-deficient cells were clearly disadvantaged in thymic settling. Thus, our data suggest that PSGL-1 is the main substrate of Fuc-TVII/Fuc-TIV involved in thymic homing. However, some progenitor cells can get into thymus without Fuc-TVII, and it remains possible that other Fut genes, including Fuc-TVI (46) or residual PSGL-1 interactions with Fuc-TVII–independent ligands such as E-selectin, may play a role (47). Perhaps for this reason, we did not see a clear increase in frequencies of P-selectin+ thymic endothelial cells, which has been reported to underlie the increased thymic receptivity seen in PSGL-1−/− mice and IL-7R−/− mice (18). However, the phenotype of Fuc-TVII/Fuc-TIV double-deficient cells otherwise largely mirrors the phenotype of PSGL-1–deficient progenitors. The occurrence of some T cell development from PSGL-1–deficient progenitors and also Fuc-TVII/Fuc-TIV double-deficient progenitors suggests that receptor ligand pairs distinct from PSGL-1 and its ligands may also exist that support thymic settling; these remain to be discovered.
In conclusion, the current study established that expression of functional PSGL-1 on BM progenitor is dynamically regulated. Together with the known selective expression of CCR7 and CCR9 chemokine receptors, selective expression of functional PSGL-1 therefore contributes to the regulated basis of selective thymic settling by T cell progenitors.
The authors have no financial conflicts of interest.
We thank Maria Elena De Obaldia, Qi Yang, and Sugata Manna for critical review of this manuscript.
This work was supported by National Institutes of Health Grants HL110741 and AI059621.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- bone marrow
- common lymphoid progenitor
- CD4−CD8− double-negative
- CD4+CD8+ double-positive
- early T lineage progenitor
- fms-like tyrosine kinase receptor-3
- α1,3 fucosyltransferase
- hematopoietic stem cell
- lymphoid-primed multipotent progenitor
- lymphoid progenitor
- mean fluorescence intensity
- multipotent progenitor
- P-selectin glycoprotein ligand 1
- Received April 18, 2011.
- Accepted March 5, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.