The DC-HIL ligand syndecan-4 is a negative regulator of T-cell allo-reactivity responsible for graft-versus-host disease


  • Jin-Sung Chung,

    1. Department of Dermatology, The University of Texas Southwestern Medical Center, Dallas, TX, USA
    2. Dermatology Section (Medical Service), Dallas Veterans Affairs Medical Center, Dallas, TX, USA
    Search for more papers by this author
  • Mizuki Tomihari,

    1. Department of Dermatology, The University of Texas Southwestern Medical Center, Dallas, TX, USA
    2. Dermatology Section (Medical Service), Dallas Veterans Affairs Medical Center, Dallas, TX, USA
    3. Department of Clinical Veterinary Medicine, Obihiro University of Agriculture and Veterinary Medicine, Obihiro, Japan
    Search for more papers by this author
  • Kyoichi Tamura,

    1. Department of Dermatology, The University of Texas Southwestern Medical Center, Dallas, TX, USA
    2. Dermatology Section (Medical Service), Dallas Veterans Affairs Medical Center, Dallas, TX, USA
    Search for more papers by this author
  • Tetsuhito Kojima,

    1. Department of Medical Technology, Nagoya University School of Health Sciences, Nagoya, Japan
    Search for more papers by this author
  • Ponciano D. Cruz Jr,

    1. Department of Dermatology, The University of Texas Southwestern Medical Center, Dallas, TX, USA
    2. Dermatology Section (Medical Service), Dallas Veterans Affairs Medical Center, Dallas, TX, USA
    Search for more papers by this author
  • Kiyoshi Ariizumi

    Corresponding author
    1. Dermatology Section (Medical Service), Dallas Veterans Affairs Medical Center, Dallas, TX, USA
    • Department of Dermatology, The University of Texas Southwestern Medical Center, Dallas, TX, USA
    Search for more papers by this author

Correspondence: Kiyoshi Ariizumi, Department of Dermatology, UT Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-9069, USA. Email:

Senior author: Kiyoshi Ariizumi


Acute graft-versus-host disease (GVHD) is the most important cause of mortality after allogeneic haematopoietic stem cell transplantation. Allo-reactive T cells are the major mediators of GVHD and the process is regulated by positive and negative regulators on antigen-presenting cells (APC). Because the significance of negative regulators in GVHD pathogenesis is not fully understood, and having discovered that syndecan-4 (SD-4) on effector T cells mediates the inhibitory function of DC-HIL on APC, we proposed that SD-4 negatively regulates the T-cell response to allo-stimulation in acute GVHD, using SD-4 knockout mice. Although not different from their wild-type counterparts in responsiveness to anti-CD3 stimulation, SD-4−/− T cells lost the capacity to mediate the inhibitory function of DC-HIL and were hyper-reactive to allogeneic APC. Moreover, infusion of SD-4−/− T cells into sub-lethally γ-irradiated allogeneic mice worsened mortality, with hyper-proliferation of infused T cells in recipients. Although there my be little or no involvement of regulatory T cells in this model because SD-4 deletion had no deleterious effect on T-cell-suppressive activity compared with SD-4+/+ regulatory T cells. We conclude that SD-4, as the T-cell ligand of DC-HIL, is a potent inhibitor of allo-reactive T cells responsible for GVHD and a potentially useful target for treating this disease.


Allogeneic haematopoietic stem cell transplantation (HSCT) is a potentially curative option for patients with high-risk haematological malignancies, such as multiple myeloma and leukaemia. The major complication of allogeneic HSCT is graft-versus-host disease (GVHD), in which donor-derived T cells attack recipients’ organs, with potential for lethality, which limiting the use of this option.[1] Donor-derived T cells are considered the main effector cells mediating acute GVHD because they recognize MHC disparities (allo-antigen) between the donor and recipient, which are presented by antigen-presenting cells (APC).

T-cell activation in response to allo-antigen requires two stimulatory signals.[1] The primary signal is delivered through the T-cell receptor (TCR), which recognizes antigens on MHC molecules. This signal is necessary but not sufficient to induce full T-cell activation, which also requires co-stimulation that drives T cells to proliferate and produce cytokines. The co-stimulation signal is mediated by a number of ligand–receptor pairs expressed on APC and T cells, and is a composite or net effect of stimulatory and inhibitory signals mediated between these two cells. The inhibitory TCR include cytotoxic T-lymphocyte antigen-4 (CTLA-4),[2] programmed cell death-1 (PD-1)[3] and B- and T-lymphocyte attenuator (BTLA).[4] Studies using experimental models of acute GVHD have shown that co-stimulatory molecules play a pivotal role in initiating acute GVHD.[5] By contrast, much less is known about co-inhibitory pathways in this process, better understanding of which would make them useful therapeutic targets.

Recently, we discovered a new co-inhibitory pathway composed of DC-HIL on APC and syndecan-4 (SD-4) on activated T cells.[6, 7] DC-HIL is a highly-glycosylated type I transmembrane receptor (95 000–120 000 molecular weight) expressed constitutively by many APC sub-sets including macrophages, monocytes, epidermal Langerhans cells, CD11c+ CD4+ lymphoid dendritic cells (DC), CD11c+ CD8+ myeloid DC and CD11c+ PDCA-1+ plasmacytoid DC.[8] It is also known as glycoprotein nmb (Gpnmb),[9] osteoactivin[10] and haematopoietic growth factor-inducible neurokinin-1 type (HGFIN).[11] DC-HIL binds to heparan sulphate-like structures on SD-4 expressed by activated (but not resting) T cells, and their binding inhibits strongly the anti-CD3 response of T cells, resulting in cessation of interleukin-2 (IL-2) production and prevention of T-cell entry into the cell cycle.[6, 12] Consistent with a previous finding that SD-4 is expressed primarily by effector/memory (but not recently activated) T cells,[13] infusion of DC-HIL or SD-4 soluble receptor during the elicitation (but not sensitization) phase of contact hypersensitivity effectively blocked the inhibitory function of the endogenous DC-HIL/SD-4 pathway, thereby enhancing ear-swelling responses in this experimental model.[7] Conversely, depletion of SD-4+ T cells by infusion of toxin-conjugated DC-HIL inhibited elicitation (but not sensitization) of contact hypersensitivity.[13] These findings support the concept that binding of DC-HIL to SD-4 inhibits pre-primed T-cell responses.

To determine whether SD-4 is the sole T-cell ligand of DC-HIL and whether its negative regulatory role applies to acute GVHD, we took advantage of SD-4 knockout (KO) mice. SD-4−/− T cells completely lost the ability to bind DC-HIL and to mediate the inhibitory function. SD-4 deficiency had no impact on the intrinsic T-cell response to TCR-induced signals, but enhanced these cells’ responsiveness to APC. Moreover, we showed SD-4 to be a constitutive inhibitor of allo-reactive T cells responsible for GVHD. Hence, SD-4 can be targeted to treat GVHD by increasing the efficacy of allo-HSCT therapy.

Materials and methods


Female BALB/c and C57BL/6 (6–8 weeks old) mice were purchased from Harland Breeders (Indianapolis, IN), and OT-II transgenic mice were purchased from Taconic Farms (Hudson, NY). Pmel-1 TCR transgenic mice (B6.Cg-Thy1a/CyTg(TcraTcrb)8Rest/J) were bought from Jackson Laboratory (West Grove, PA). SD-4-deficient mice were produced by mating SD-4+/− mice bearing a C57BL/6 genetic background.[14] We also produced SD-4-deficient pmel-1 mice by breeding SD-4−/− and pmel-1 transgenic mice. Control groups included mice with wild-type (WT) genotype (SD-4+/+) from the same generation of backcrosses. Following National Institutes of Health guidelines, mice were housed and cared for in a pathogen-free facility and subjected to experimental procedures approved by the Institutional Animal Care Use Center at The University of Texas Southwestern Medical Center (Dallas, TX).

Reagents, antibodies and immunofluorescent staining

Monoclonal antibodies (mAb) against CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD11c (N418), CD19 (eBio 1D3), PD-1 (J43), Foxp3 (FJK-165) and H-2Kb-SIINFEKL (eBio25-D1.16) were purchased from eBioscience (San Diego, CA); mAb against SD-4 (KY/8.2) were from BD Pharmingen (San Diego, CA); secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA); and hgp100 peptide (KVPRNQDWL), ovalbumin(257–264) (OVA257–264) H-2Kb-restricted class I peptide (SIINFEKL), and OVA323–339 H-2Kb-class II peptide (ISQAVHAAHAEINEAGR) were synthesized by the Protein Chemistry Technology Center at UT Southwestern.

For flow cytometry, lymph node cells or T cells (5 × 105 to 10 × 105) were treated with 5 μg/ml Fc blocker (BD Pharmingen) on ice for 30 min and incubated with primary antibody (5–10 μg/ml), followed by addition of secondary antibody (2·5 μg/ml). After washing, cell-bound fluorescence was analysed by FACSCablibur (BD Biosciences, San Jose, CA).

DC-HIL-Fc, comprising the extracellular domain fused to the Fc portion of human IgG1, was produced in COS-1 cells and purified as described previously.[15] Purity of final preparations was high, as judged by a single band in SDS–PAGE/Coomassie Blue staining or in immunoblotting with anti-DC-HIL mAb or goat anti-human IgG antibody.

Binding of DC-HIL and T-cell activation assays

CD3+, CD4+ and CD8+ T cells were purified from spleen using pan-T-cell, CD4+ and CD8+ T-cell isolation kits (Miltenyi Biotec, Auburn, CA), respectively, according to the manufacturer's recommendations. For binding of DC-HIL-Fc to T cells,[6] CD4+ or CD8+ T cells (1 × 106) purified from spleens of WT or KO mice were activated by culturing with immobilized anti-CD3 antibody (1 or 3 μg/ml) for 3 days. These cells were incubated with 10 μg/ml DC-HIL-Fc or control immunoglobulin plus 2·5 μg/ml phycoerythrin-conjugated anti-human IgG, followed by flow cytometry.

To assay T-cell responses in vitro, purified CD4+ and CD8+ T cells (2 × 105/well) were cultured for 2 days with increasing doses of hgp100 peptide (for spleen cells of pmel-1 transgenic mice), concanavalin A (Sigma-Aldrich, St Louis, MO), or ELISA plates pre-coated with various doses of anti-CD3 and DC-HIL-Fc or control immunoglobulin. After pulsing with [3H]thymidine (1 μCi/well) in the last 20 hr of the culture period, cells were collected and counted for [3H] radioactivity. Culture supernatant was also harvested and stored at − 85° until required for assaying IL-2, interferon-γ and tumour necrosis factor-α using mouse ELISA kits (BD Pharmingen).

The CD4+ T cells (1 × 106) were labelled with 1 μm carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, OR) in Dulbecco's PBS at 37° for 15 min. After another 30 min of incubation in culture medium, labelled T cells were cultured in ELISA wells pre-coated with anti-CD3 antibody (1 μg/ml). At different time-points thereafter, cells were examined for asynchronous cell division by flow cytometry.[16]

Mixed lymphocyte reaction

Bone-marrow-derived DC (BMDC) were harvested from day 6 cultures of BM cells isolated from BALB/c mice with 10 ng/ml granulocyte–macrophage colony-stimulating factor[17] and used as stimulators. CD4+ T cells purified from KO or WT C57BL/6 mice served as responders.[12] A constant number of BM-DC (5 × 104 cells) was mixed with varying numbers of CD4+ T cells in 96-well plates and cultured for 3 days. T-cell activation was measured by IL-2 production and [3H]thymidine incorporation.

Antigen presentation assay

To examine the impact of SD-4 deletion on the reactivity of CD8+ T cells to APC co-stimulation, BMDC (2 × 104 cells/well) prepared from BM cells of WT mice were pulsed with hgp100 peptide (1 μg/ml) and co-cultuerd with varying numbers of pmel-1 CD8+ T cells (0 × 105 to 2 × 105 cells/well) for 72 hr. To examine the effect of SD-4 deletion on APC capacity of DC, BMDC prepared from KO or WT mice were seeded on 96-well plates (1 × 103 to 40 × 103 cells/well), and pulsed for 3 hr with OVA323–339 peptide (2 μg/ml).[18] Cultured DC were added to CD4+ T cells (1 × 105/well) purified from the spleens of OT-II transgenic mice. One day after co-culture, IL-2 in the culture supernatant was measured by ELISA.

Acute GVHD

Recipient BALB/c mice were treated with antibiotic water (sulfomethoxazole-trimethoprim; Hi-Tech Pharmacal Co., Inc. Amityville, NY) from 3 days before γ-irradiation daily through until day 28. On day 0, recipients were subjected to total body γ-irradiation (6 Gy) and, within 24 hr, were injected via the tail vein with 1 × 107 T-cell-depleted BM cells (from WT mice) and 5 × 105 splenic T cells purified from three KO or WT mice. T-cell depletion was performed using biotinylated anti-Thy1.2 antibody and Dynabeads Biotin Binder (Invitrogen, Carlsbad, CA), and the final preparation contained < 0·1% of Thy1.2+ cells. Control mice only received T-cell-depleted BM cells. Mice were monitored for appearance, body weight and survival on a weekly or daily basis. To examine proliferation of donor-derived T cells in recipients, BM cells and 5 × 105 CD3+ T cells from KO or WT mice were co-injected intravenously into γ-irradiated recipient mice. Five days after injection, cells were prepared from spleens and livers of recipient mice (the latter cells were prepared as described previously[19]) and pulsed with H-2Kb-associated OVA peptide (1 μg/ml) for 30 min. After washing, donor-derived T cells were labelled fluorescently with phycoerythrin-conjugated anti-H-2Kb-SIINFEKL antibody and FITC-conjugated antibody directed against CD4, CD8 or CD69; positive cells were counted by flow cytometry.

Regulatory T-cell assay

To examine SD-4 expression on conventional T (Tconv) cells and regulatory T (Treg) cells, CD4+ T cells were purified from spleen cells of WT C57BL/6 mice using a CD4+ T-cell isolation kit (Miltenyi) and split into two batches: one left untreated and the other cultured for 2 days in 96-well plates (2 × 105 cells/well) pre-coated with anti-CD3 and anti-CD28 antibody (each 1 μg/ml). Cells were surface stained to detect SD-4 (or PD-1) -positive cells and then treated with cell fixation/permeabilization solution (eBioscience), followed by staining with allophycocyanin-conjugated anti-Foxp3 antibody.

To examine the influence of SD-4 deletion on T-cell-suppressive activity of Treg cells, CD4+ CD25neg Tconv cells and CD4+ CD25+ Treg cells were isolated by fractionating purified CD4+ T cells (from spleen cells of WT or KO mice) using anti-CD25 antibody and anti-biotin microbeads (Miltenyi Biotec): Treg cells were collected from eluate of the magnetic column, and Tconv cells from the pass through (purity was > 95%). Tconv cells from WT mice (5 × 105 cells/well) were labelled with CFSE and stimulated with anti-CD3 antibody (5 μg/ml) in the presence of an equal number of γ-irradiated WT spleen cells (as APC). To this culture, varying numbers of Treg cells isolated from WT or KO mice were added. Suppression of Tconv-cellproliferation by Treg cells was determined by flow cytometric analysis of CFSE dilution after 72 hr.

Statistical analysis

Data are presented as means ± SD. The significance of differences between experimental variables was determined using a two-tailed Student's t-test. All data shown are representative of at least two independent experiments.


SD-4 KO mice manifest normal development of leucocytes within lymphoid organs

The absence of published information regarding the impact of SD-4 gene disruption on leucocyte development led us to compare the relative proportions of leucocyte sub-populations (CD4+ and CD8+ T cells, CD19+ B cells and CD11c+ DC) in BM, spleen and lymph nodes of mice aged 6 weeks (Fig. 1a–c). There were no significant differences between WT and KO mice. We also measured ratios of double-positive versus single-positive T cells in thymus and those of CD4+ versus CD8+ T cells in spleen and lymph nodes (Fig. 1d). Again, there was no difference between the two mouse strains. Hence, SD-4 gene deficiency appears to have little to no impact on leucocyte development. Moreover, up to 1 year of age, we observed no morphological nor developmental abnormality.

Figure 1.

No significant changes in leucocyte proportions of lymphoid organs in wild-type (WT) versus knockout (KO) mice: Single cell suspensions prepared from bone marrow (BM) (a), spleen (b) and lymph nodes (LN) (c) of WT and KO naive mice (7-week-old females) were stained fluorescently with marker antibody (CD4, CD8, CD19 and CD11c) and analysed by flow cytometry. Frequency (%) of each leucocyte is shown in dot-blots. (d) These cell suspensions (including thymus) were also analysed for ratios of CD4/CD8-double and single positive cells. Two other pairs (6- and 10-week-old female mice) showed similar results.

T cells from SD-4 KO mice do not mediate the inhibitory function of DC-HIL

Using functional blockade of SD-4 by antibody or Fc-fusion proteins, we showed previously that SD-4 is the ligand through which DC-HIL mediates its inhibitory function.[7] To study the influence of SD-4 expression on the regulation of T-cell function, we first examined the capacity of T cells from SD-4 KO mice to mediate the inhibitory function of DC-HIL (Fig. 2). Specificity of the gene deficiency was confirmed by the inability of T cells to express SD-4 after activation (high expression by WT-T cells, see Supplementary material, Fig. S1), even as they were capable of expressing another inhibitory molecule, PD-1 (Fig. 2a). We then examined the binding of activated T cells to DC-HIL (Fig. 2b), and found that those from WT mice bound strongly to soluble DC-HIL receptor (DC-HIL-Fc), whereas those from KO mice did not. Thereafter, we examined the ability of immobilized DC-HIL-Fc to inhibit T-cell activation triggered by anti-CD3 antibody. CD4+ T cells from WT or KO mice were cultured with immobilized anti-CD3 antibody (increasing doses) and DC-HIL-Fc (constant dose), and their activation was measured as proliferation. DC-HIL-Fc strongly inhibited proliferation of SD-4+/+ CD4+ T cells activated by anti-CD3 antibody at doses < 0·3 μg/ml, although doses > 1 μg/ml rescued the inhibition (Fig. 2c), consistent with our previous results using T cells from BALB/c mice.[6, 7] By contrast, the presence or absence of DC-HIL-Fc had no effect on the proliferation of similarly activated SD-4−/− CD4+ T cells. Loss of responsiveness to DC-HIL was also true for SD-4-deficient CD8+ T cells (Fig. 2d).

Figure 2.

Syndecan-4 (SD-4) -deficient T cells do not mediate inhibitory function of DC-HIL but retain intrinsic responses to stimulation. (a) CD4+ T cells purified from spleen of SD-4−/− mice were activated with immobilized anti-CD3 antibody (0·5 μg/ml) and examined by flow cytometry for surface expression of SD-4 and programmed death-1 (PD-1). Staining with respective antibody or control IgG is shown in open or grey histograms, respectively. (b) Activated CD4+ T cells from wild-type (WT) or knockout (KO) mice were assayed for binding of DC-HIL (open histograms) or control immunoglobulin (grey histograms). CD4+ (c) or CD8+ (d) T cells purified from spleens of WT or KO mice were cultured with immobilized anti-CD3 (varying doses) and control immunoglobulin (open circles) or DC-HIL-Fc (closed circles, at a constant dose). T-cell activation was measured by [3H]thymidine (TdR) incorporation. (e) The supernatant of day 2 culture of T cells stimulated with anti-CD3 antibody (0·3 μg/ml) and DC-HIL-Fc (closed rectangles) or control immunoglobulin (open rectangles) was harvested and assayed for secretion of indicated cytokines. (f) CFSE-labelled CD4+ T cells were stimulated with immobilized anti-CD3 antibody (1 μg/ml) and cultured for the indicated period. CFSE intensity was analysed by flow cytometry and expressed in histograms. (g) CD4+ T cells were also cultured in the presence of concanavalin A at increasing doses and proliferation was assayed. Data (c, d, e, and g) are shown as mean ± SD, n = 3.

We also probed the effect of SD-4 deficiency on cytokine expression by anti-CD3 antibody-activated T cells in the presence or absence of DC-HIL-Fc (Fig. 2e). Interleukin-2 and tumour necrosis factor-α (for CD4+ T cells), and IL-2 and interferon-γ (for CD8+ T cells) were assayed from supernatants of T cells stimulated with anti-CD3 antibody (0·3 μg/ml) plus DC-HIL-Fc or control immunoglobulin. In the absence of DC-HIL (anti-CD3 and control immunoglobulin), there was no significant difference in cytokine production by WT versus KO T cells (CD4+ or CD8+). Consistent with our previous data,[7] co-treatment with DC-HIL markedly inhibited the production of cytokines by SD-4+/+ T cells, whereas it failed to do so for SD-4−/− T cells. Rather, it caused some up-regulation compared with anti-CD3 alone. These results indicate that SD-4 is exclusively responsible for mediating the T-cell-inhibitory function of DC-HIL.

SD-4 deficiency does not alter the intrinsic reactivity of T cells to antigen

SD-4−/− T cells showed similarly strong responsiveness to anti-CD3 antibody stimulation, compared with SD-4+/+ control cells (Fig. 2c,d). To more precisely analyse the proliferative response of these T cells, we examined each round of cell division using the dye CFSE.[16] CD4+ T cells labelled with CFSE were cultured with anti-CD3 antibody (0·5 μg/ml) for 48 or 72 hr (Fig. 2f). At each time-point examined, SD-4+/+ and SD-4−/− T cells showed almost identical patterns of cell division (as reflected from diffusion of CFSE fluorescent intensity). Similar results were also noted with lower concentrations (0·1 and 0·3 μg/ml) of anti-CD3 antibody (see Supplementary material, Fig. S2). We then examined the effect of SD-4 deletion on the intrinsic response triggered by concanavalin A, wihch activates T cells in a non-specific manner (Fig. 2g). Again, there was no significant change in T-cell proliferation. Hence, lack of SD-4 expression does not alter the intrinsic responsiveness of T cells to TCR-dependent or non-specific stimulation. These features distinguish SD-4 from PD-1 and BTLA, whose respective deletions augment T-cell responses to anti-CD3 stimulation.[20, 21]

SD-4 deletion augments DC activation of antigen-specific T cells, but does not alter the T-cell-stimulatory capacity of APC

Using the mixed lymphocyte reaction, we examined the impact of SD-4 deletion on T-cell reactivity in response to allogeneic DC-HIL+ APC (Fig. 3a,b). CD4+ T cells (varying numbers) isolated from WT or KO C57BL/6 mice were co-cultured with DC (constant number) prepared from BM cells of BALB/c mice. T-cell activation was measured by secreted IL-2 (Fig. 3a) or by proliferation (Fig. 3b). SD-4−/− T cells produced IL-2 at a four-fold greater level and proliferated at a two-fold higher level, respectively, than SD-4+/+ T cells.

Figure 3.

Syndecan-4 (SD-4) deletion renders T cells hyper-reactive to allo-antigen in mixed lymphocyte reaction, without changes in antigen-presenting cell capacity. CD4+ T cells purified from spleens of wild-type (WT) (open rectangles) or knockout (KO) C57BL/6 mice (closed rectangles, 1 or 2 × 105 cells/well) were co-cultured with bone-marrow-derived dendritic cells (BMDC) (5 × 104 cells/well) from BALB/c mice for 3 days. Interleukin-2 (IL-2) (a) and [3H]thymidine ([3H]TdR) incorporation (b) were assayed (mean ± SD, n = 3). (c) Spleen cells isolated from SD-4+/+ (WT) or SD-4−/− (KO) pmel-1 mice were cultured with various concentrations of gp100 peptide for 2 days. [3H]TdR incorporation was measured (mean ± SD, n = 3). (d) Increasing numbers of CD8+ T cells isolated from WT- or KO-pmel-1 mice were activated by BMDC pulsed with hgp100, and IL-2 secretion was assayed (mean ± SD, n = 3). (e) Various numbers of BMDC prepared from KO or WT mice were co-cultured with syngeneic CD4+ T cells from OT-II mice for 1 day in the presence of ovalbumin (OVA) peptide. T-cell activation was measured by IL-2 production (mean ± SD, n = 3).

We next used a defined antigen model of gp100 (melanoma-associated antigen).[22] SD-4 gene deficiency was introduced into the pmel-1 TCR transgenic mice (in which all CD8+ T cells express the same TCR specific to a particular gp100 antigen peptide).[23] With respect to relative proportions of leucocyte sub-populations in lymphoid organs, there was no significant difference between SD-4+/+ and SD-4−/− pmel-1 mice (data not shown). We then assayed the reactivity of T cells to gp100 peptide-loaded APC. Spleen cells isolated from SD-4+/+ or SD-4−/− pmel-1 mice were stimulated by increasing doses of antigen and measured for proliferation (Fig. 3c). SD-4+/+ pmel-1 spleen cells proliferated and produced IL-2 in response to gp100 antigen in a dose-dependent manner. Similarly, SD-4−/− pmel-1 spleen cells responded to antigen, but with significantly elevated levels (more than twofold greater responses by SD-4−/− pmel-1 T cells) at almost every single dose of antigen. To more rigorously examine the impact of SD-4 deletion, BMDC were prepared from WT mice and allowed to stimulate SD-4+/+ or SD-4−/− CD8+ T cells (Fig. 3d). SD-4−/− CD8+ T cells produced greater levels of IL-2 than SD-4+/+ CD8+ T cells (up to twofold), consistent with the previous data (Fig. 3c).

As SD-4 is also expressed by DC (unpublished data), we examined the possibility that contaminant APC in the T-cell preparation from KO mice contributed to hyperactivation (Fig. 3a). SD-4−/− BMDC (increasing numbers) were pulsed with OVA peptide and then allowed to stimulate CD4+ T cells (constant number) from OT-II mice (Fig. 3e). At all the doses tested, there was no significant difference in IL-2 production by T cells activated by SD-4+/+ versus SD-4−/− DC. Altogether, SD-4 deletion had no impact on T-cell responses in the absence of accessory signals delivered by DC, but it augmented the DC-induced response (enhanced co-stimulatory signals resulting from lack of the inhibitory function of DC-HIL/SD-4 between APC and T cells).

SD-4 inhibits acute GVHD by negatively regulating allo-reactive T cells

Since SD-4−/− T cells were hyper-reactive to allo-antigen in the mixed lymphocyte reaction (Fig. 3a), we examined their effect on acute GVHD (Fig. 4). BALB/c mice were γ-irradiated at a sub-lethal dose and then infused with T-cell-depleted allogeneic BM cells (from C57BL/6 mice) with or without CD3+ T cells isolated from KO or WT mice. Body weight was noted weekly and survival was noted daily through to day 100. All mice lost about 30% of initial body weight within a week after BM transplantation, but recovered some weight during the 2nd week. Thereafter, differentially treated mice displayed disparate outcomes (Fig. 4a). Mice that received BM cells alone completely recovered their weight 3 weeks post-BM transplantation and survived for at least 100 days. Mice that received BM cells plus SD-4+/+ T cells partially recovered their weight, with 50% dying by day 32, and the rest survived for at least 100 days (Fig. 4b). By contrast, mice that received BM cells plus SD-4−/− T cells lost weight progressively (up to 40%) due to severe diarrhoea, with 50% dying by day 14, and all dead by day 32. We also examined proliferation of infused T cells in recipients, by measuring the number of donor-derived T cells (H-2Kb+) in spleen and liver of mice at day 5 post-BM transplantation (Fig. 4c,d). In spleen (Fig. 4c), there was twofold to threefold greater CD4+ and CD8+ SD-4−/− T cells than SD-4+/+ T cells, and also more CD69+ (activated) cells than in recipients of SD-4+/+ T cells. Similar results were observed in liver, which is another major target of acute GVHD (Fig. 4d).[1] These results indicate that infusion of T cells devoid of SD-4 worsens morbidity and mortality of acute GVHD, most likely through hyper-reactivity to allo-antigen.

Figure 4.

Infusion of syndecan-4-deficient (SD-4−/−) T cells in sublethally γ-irradiated mice exacerbates graft-versus-host disease. (a and b) After total body γ-irradiation, BALB/c mice were injected with T-cell-depleted bone marrow (BM) cells prepared from C57BL/6 mice (BM alone); or BM cells plus 5 × 105 CD3+ T cells purified from wild-type mice (BM+WT) or knockout mice (BM+KO). Body weight (a) and survival rate (%) (b) were monitored weekly and daily, respectively. Combined data from two independent experiments (n = 9) are shown. (c and d) Five days following infusion of BM cells and T cells, spleen (c) or liver (d) was procured from mice (n = 5), and number of CD4+, CD8+, or CD69+ donor-derived cells was counted by flow cytometry and expressed as cell number per 1 × 105 of liver or 1 × 104 of spleen cells. *< 0·00005.

SD-4 deletion has no effect on the T-cell-suppressive activity of Treg cells

Because donor-derived Treg cells are known to play a pivotal role in preventing GVHD induced by co-injection of BM cells and T cells isolated from C57BL/6 mice into total body γ-irradiated BALB/c mice,[24] we studied the influence of SD-4 deletion on the T-cell-suppressive activity of Treg. We examined expression of SD-4 on conventional CD4+ Foxp3 T cells (Tconv) versus CD4+ Foxp3+ Treg cells (Fig. 5). The Tconv and Treg cells freshly isolated from naive WT mice represented 90% and 10%, respectively, and neither expressed SD-4. In contrast, PD-1 was expressed by a minuscule fraction of Tconv cells (4·6%) and by some Treg cells (22% of Foxp3+ cells) (Fig. 5a). The Tconv and Treg cells were activated by culture for 2 days with immobilized anti-CD3/CD28 antibody. This activation induced SD-4 expression in both Tconv and Treg cells, and PD-1 expression (Fig. 5b): 36% of activated Treg cells expressed SD-4, with more Treg cells (53%) expressing PD-1.

Figure 5.

Syndecan-4 (SD-4) expression by regulatory T (Treg) cells following activation. CD4+ T cells freshly isolated from spleens of wild-type (WT) naive mice were left untreated (a) or cultured for 2 days with anti-CD3/CD28 antibody (b). These cells were then surface-stained fluorescently with anti-SD-4 or programmed death-1 (PD-1) antibody (or control IgG), followed by intracellular staining of Foxp3. Flow cytometric data are shown in dot-plots with % of each population. Conventional T (Tconv) (Foxp3) and Treg (Foxp3+) cells were gated separately and examined for SD-4 or PD-1 expression in histograms (staining with experimental or control antibody is indicated as open and filled histograms, respectively).

Finally, we assayed the ability of SD-4+/+ versus SD-4−/− Treg cells to suppress T-cell activation (Fig. 6). Varying numbers of CD4+ CD25+ Treg cells purified from spleens of naive WT or KO mice were co-cultured with CFSE-labelled CD4+ CD25neg Tconv cells in the presence of anti-CD3 antibody and irradiated APC. T-cell proliferation was assayed by CFSE dilution. Without Treg cells, 60% of Tconv cells proliferated. As expected, SD-4+/+ Treg cells inhibited this proliferation in a dose-dependent manner (down to 13% proliferation), and SD-4−/− Treg cells exhibited similar inhibitory capacity at every dose tested. These results show that SD-4 deficiency has little or no influence on Treg-cell function, thereby supporting the idea that exacerbation of GVHD by infusion of SD-4−/− T cells is primarily the result of augmented reactivity of Tconv cells to APC co-stimulation.

Figure 6.

Effect of syndecan-4 (SD-4) deletion on T-cell-suppressive activity of regulatory T (Treg) cells. Treg cells from wild-type (WT) or SD-4 knockout (KO) mice were added to the culture of CFSE-labelled CD4+ CD25 Tconv cells at various cell ratios in the presence of anti-CD3 antibody (5 μg/ml) and irradiated antigen-presenting cells. T-cell activation was assessed by flow cytometric analysis of CFSE dilution, and data are shown in histograms with % of proliferated cells. Untreated or activated Tconv cells without Treg cells are also shown.


SD-4 belongs to the SD family of transmembrane receptors heavily laden with heparan sulphate chains consisting of alternating disaccharide residues.[25] Because these heparan sulphate chains bind to a variety of proteins, including growth factors, cytokines, chemokines and extracellular matrices,[26] SD-4 can participate in a wide range of physiological and pathological conditions. Indeed, SD-4 is known to play important roles in cell matrix-mediated and growth factor-mediated signalling events.[27] SD-4-deficient mice may appear normal, but respond to intentional wounding with delayed repair, impaired angiogenesis, and poor focal adhesion of cells to matrix.[28] SD-4 also regulates immune responses: when given endotoxin, SD-4 KO mice succumb more readily to shock than WT controls;[29] SD-4 on B cells triggers formation of dendritic processes, which facilitate these cells’ interaction with other immune cells.[30] Our studies constitute the first evidence showing SD-4 on T cells to regulate the activation of allo-reactive T cells in GVHD.

All the results using SD-4 KO mice unambiguously indicate SD-4 on T cells to be the sole DC-HIL ligand responsible for mediating its T-cell-inhibitory function (SD-4−/− T cells did not bind DC-HIL nor did they react to DC-HIL's inhibitory function), with one exception: DC-HIL-Fc treatment up-regulated cytokine production by SD-4−/− CD4+ T cells (compared with SD-4+/+ CD4+ T cells) following in vitro anti-CD3 stimulation (Fig. 2e). Because DC-HIL binds not only to a peptide sequence of SD-4 but also to saccharide (probably heparan sulphate or other structurally related saccharides),[6, 12] we speculate that absence of SD-4 and APC may restrict DC-HIL interaction exclusively to saccharides on T cells, thereby producing effects independent of SD-4. To be sure, we do not think that this mechanism accounts for the enhanced response of SD-4−/− T cells to co-stimulation by DC-HIL+ APC (Fig. 3). Rather, we consider that lack of the DC-HIL/SD-4 pathway (inability to induce SD-4-linked inhibitory signals) leads to an enhanced T-cell response, most likely through DC-HIL co-stimulation (DC-HIL-Fc versus the native form of DC-HIL). Our recent finding that APC from DC-HIL-knockout mice become more potent T-cell stimulators (unpublished data) is consistent with this concept.

Compared with WT, SD-4-deleted T cells produced no change in T-cell response to non-specific stimuli (e.g. concanavalin A), similar to responses of PD-1-deleted or BTLA-deleted T cells.[20, 31, 32] In contrast, the T-cell response to anti-CD3 antibody resulted in different outcomes in the absence of APC: SD-4-deleted T cells were as responsive as the WT, whereas PD-1-deleted or BTLA-deleted T cells were hyper-reactive. This is an interesting disparity that may be related to the fact that PD-1 and BTLA associate directly with the TCR/CD3 complex, localizing within the immunological synapse formed by the interface between T cells and APC,[33, 34] whereas SD-4 does not interact directly with the synapse.[35] Hence, absence of more proximally located co-inhibitors (PD-1 or BTLA) but not a distal one (SD-4) may directly reduce the threshold for CD3 reactivity. Note that these assays are devoid of APC.

Several co-inhibitory receptors can regulate the allo-reactivity of T cells, including CTLA-4 and PD-1, which have been evaluated in GVHD. CTLA-4 acts along with the CD28–CD80/CD86 stimulation pathway to inhibit T-cell allo-reactivity.[2] Its marked influence has been suggested by a report that polymorphisms in the CTLA-4 gene in the donors are associated with morbidity of acute GVHD.[36] In mouse models, infusion of CTLA-4-Fc, which prevents T cells from being activated by co-stimulatory signals delivered by binding of CD28 to CD80/CD86, ameliorated the lethality of GVHD.[37] However, this effect was not impressive, and this strategy was not intended to block the intrinsic regulatory function of CTLA-4. PD-1 on T cells inhibits T-cell activation by binding to the ligands (PD-L1 and PD-L2) on APC. PD-1 expression is up-regulated in the infiltrating cells on GVHD target organs (e.g. intestine and liver) in mouse models with full MHC disparate T cells.[38] PD-1 blockade by infusion of anti-PD-1 antibody resulted in accelerated GVHD and enhanced mortality, mostly mediated by IFN-γ secretion from donor T cells.[38] Akin to our data, studies using T cells from PD-1 KO mice documented an enhanced capacity to induce GVHD. Collectively, like CTLA-4 and PD-1 receptors, SD-4 may serve as a novel target to prevent GVHD.

Another difference from CTLA-4 and PD-1 is the effect on Treg-cell function. CTLA-4 on Treg cells down-regulates the expression of CD80 and CD86 on DCs, thereby making DC less activated or more tolerogenic.[39] PD-1 on naive Treg cells can convert naive T cells to inducible Treg cells in the presence of APC.[40] By contrast, SD-4 is probably unrelated to the suppressive activity of Treg cells, although its expression is induced upon in vitro activation with anti-CD3 antibody.

We conclude that SD-4 is a negative regulator of T-cell allo-reactivity responsible for acute GVHD in animal models. SD-4 differs from CTLA-4 and PD-1 by an inability to alter the intrinsic ability of T cells to respond to TCR-activation signals and by lack of influence on Treg-cell function. These attributes support the concept of SD-4 as a new therapeutic mechanism for treating GVHD by blocking allo-reactivity of effector T cells while sparing Treg-cell activity.


We thank Irene Dougherty and Megan Randolph for technical and secretarial assistance. This research was supported by National Institutes of Health grant (AI064927-05) and a Pilot and Feasibility Study Grant from Galderma.

Conflict of interest disclosures

The authors declare no competing financial interests.