The binding of herpesvirus entry mediator (HVEM) to B and T lymphocyte attenuator (BTLA) is known to activate an inhibitory signaling cascade in effector T (Teff) cells, but we now report that the HVEM-BTLA pathway is also important to the suppressive function of regulatory T cells (Tregs). Although naive T cells up-regulated BTLA upon TCR activation, Treg expression of BTLA remained low, regardless of TCR activation. Moreover, BTLA−/− CD4+CD25+ Tregs had normal suppressive activity, whereas BTLA−/− Teff cells were more resistant than wild-type Teff cells to suppression by Tregs, suggesting BTLA expression by Teff cells was required for their suppression by Tregs. In contrast to BTLA, HVEM expression was comparable in naive Tregs vs Teff cells, but after stimulation HVEM expression was quickly down-regulated by Teff cells, whereas HVEM was further up-regulated by Tregs. HVEM−/− Tregs had decreased suppressive activity as compared with wild-type Tregs, indicating that Treg expression of HVEM was required for optimal suppression. Consistent with this, T cells from Scurfy mice (FoxP3 mutant) lacked HVEM gene expression, and adoptively transferred wild-type but not HVEM−/− Tregs were able to control alloresponses in vivo by normal Teff cells. Our data demonstrate that Tregs can exert their effects via up-regulation of the negative costimulatory ligand HVEM, which upon binding to BTLA expressed by Teff cells helps mediate the suppressive functions of Tregs in vitro and in vivo.
The molecular interactions responsible for the suppressive activity of CD4+CD25+ regulatory T cells (Tregs)3 remain unclear, but Tregs are known to control immune responses to self-Ags, thereby preventing autoimmune disease, and to regulate responses to nonself molecules in adaptive immunity (1). Tregs have received much attention by transplant investigators because allograft tolerance is not routinely achieved clinically and transplant recipients require lifelong immunosuppression and endure varying degrees of associated drug toxicities. The identification and characterization of Tregs led to the expectation that these cells would be useful for therapy in transplantation. However, although clinical and experimental studies indicate that manipulating the balance between Tregs and responder T cells can dampen host responses posttransplant, little success has been attained using Tregs to achieve long-term allograft survival in normal, immunocompetent hosts (1, 2, 3). Therefore, a better understanding of how Tregs work is essential if progress toward clinical therapy is to be attained. The detailed mechanisms by which Tregs control immune responses in vivo are unknown. In vitro studies often show a requirement for Treg/T effector (Teff) cell contact, although functionally active surface molecules on Tregs such as CTLA4 (CD152), glucocorticoid-induced TNFR family-related receptor (GITR, TNFRSF18), and the programmed death receptor PD-1 (CD279), as well as soluble mediators including IL-10 and TGF-β can also contribute to Treg suppressive functions, depending upon assay conditions and likely other factors (1).
Optimal T cell activation requires the TCR engagement of a cognate MHC-peptide complex and a costimulatory signal. The binding of CD28 to B7-1/B7-2 remains the best-characterized costimulatory pathway, but the persistence of T cell responses in CD28−/− mice led to the discovery of several CD28/B7 homologs, including the components of the ICOS/B7RP-1 and PD-1/PD-L1/PD-L2 pathways (4). The most recently recognized CD28 homolog, B and T lymphocyte attenuator (BTLA, CD272) (5), is unusual in that it interacts with the TNFR superfamily member, herpesvirus entry mediator (HVEM, TNFRSF14) (6). HVEM can promote T cell activation by propagating positive signals from the TNF superfamily member ligand, LIGHT (TNFSF14, CD258), a lymphotoxin-related inducible ligand that competes for glycoprotein D binding to HVEM on T cells (7).
A 2.8-Å crystal structure of the BTLA-HVEM complex shows that BTLA binds the N-terminal cysteine-rich domain of HVEM and uses a unique binding surface compared with other CD28-like receptors (8). The BTLA binding site on HVEM overlaps with the binding site for the HSV type 1 envelope glycoprotein D, but is distinct from the binding site for LIGHT. HVEM has three cysteine-rich domains. Competitive binding analysis and mutagenesis reveals a unique BTLA binding site centered on a critical lysine residue in cysteine-rich domain-1, the most membrane-distal domain, as opposed to LIGHT, which interacts largely through cysteine-rich domain-3 (8, 9, 10).
The importance of the HVEM-BTLA pathway in control of Teff cell functions was shown in several animal models. For example, compared with controls, HVEM−/− mice had increased susceptibility to Con A mitogen-induced, T cell-dependent autoimmune hepatitis (11), and BTLA−/− mice had greater acute allergic airway inflammation (12), worse myelin oligodendrocyte glycoprotein peptide-induced experimental autoimmune encephalitis (4), and were unable to accept partially MHC-mismatched cardiac allografts (13). In the current study, we propose a new model in which Tregs exert their suppressive effect via up-regulation of the negative costimulatory ligand HVEM that, upon ligation with BTLA expressed on the Teff cell side, enhances the suppressive function of Tregs and helps control immune response in vitro and in vivo.
Materials and Methods
HVEM−/− and other mice
A targeting vector was constructed using a 7.8-kb genomic fragment containing exons 3–8 of the HVEM gene, and a 0.8-kb sequence around exon 3, including ATG, was replaced by pMC1 neo (14). The targeting vector was linearized, electroporated into ES cells from 129 mice, HVEM+/− ES cell clones were screened by Southern blot analysis, and chimeric mice (B6/129) were derived by blastocyst injection. HVEM+/− mice were crossed to produce HVEM−/− mice and backcrossed at least eight generations on a C57BL/6 background before study. BTLA−/− mice on the C57BL/6 background were previously described (5). We purchased wild-type (WT) C57BL/6 (H-2b), BALB/c (H-2d), C57BL/6/DBA2 F1 (B6D2F1), congenic Thy1.1, Scurfy and RAG2−/− C57BL/6 mice (The Jackson Laboratory). Studies were performed with a protocol approved by the Institutional Animal Care and Use Committee of Children’s Hospital of Philadelphia.
We purchased CFSE (Molecular Probes), PE-conjugated rat anti-mouse Foxp3 mAb (eBioscience) and additional mAbs for flow cytometry (BD Pharmingen). Biotinylated BTLA tetramer was made as reported (6) and conjugated with streptavidin-PE or streptavidin-FITC at a molar ratio of 1:4 before each staining. For Foxp3 studies, cells were first labeled with cell surface markers, fixed, permeabilized, and stained with PE-conjugated rat anti-mouse Foxp3 mAb (13).
RNA extracted using RNeasy kits (Qiagen) was reverse transcribed with random hexamers (ABI PRISM 5700; Applied Biosystems). Primer and probe sequences for target genes were used for quantitative PCR amplification of total cDNA (ABI PRISM 5700 and TaqMan Predeveloped Assay Reagents; Applied Biosystems). Relative quantitation of target cDNA was determined by arbitrarily setting the control value to 1, and changes in the cDNA content of a sample were expressed as fold increases above the set control value. Differences in cDNA input were corrected by normalizing signals obtained with specific primers to ribosomal RNA, and nonspecific amplification excluded by performing quantitative PCR in the absence of target cDNA.
In vitro cellular assays
CD4 T cells were negatively selected using magnetic beads (Miltenyi Biotec) and fractionated into CD25+ and CD25− populations (>95% purity by flow cytometry). We used soluble CD3 mAb plus gamma-irradiated syngeneic APC, CD3 and CD28 mAbs, or PMA/ionomycin plus IL-2 to activate T cells in vitro. Treg function was assayed using MACS-purified CFSE-labeled CD4+CD25− T cells as effector cells and CD4+CD25+ cells or retrovirally transduced Foxp3+ CD4 T cells; CD3 mAb-induced CD4+CD25− cell proliferation was assessed by flow cytometric analysis of CFSE dilution after 72 h (13).
Retroviral transduction of primary T cells
Murine Foxp3 was cloned into a murine stem cell virus-based bicistronic retroviral vector, MinR1, containing a 5′ and 3′ long terminal repeat site, and an internal ribosomal entry site (IRES) cassette downstream of the cloning region and upstream of a cell surface-expressed nonsignaling nerve growth factor receptor (15). Plasmid mHVEM-FL-GFP-RV was made from two PCR products, with primers 5′-BglII mHVEM and mHVEM/GFP using mHVEM-FL-IRES-GFP-RV as template, and primers mHVEM-GFP and 3′-GFP plus Sal using mHVEM-FL-IRES-GFP-RV as template; PCR products were annealed, amplified with primers 5′-BglII mHVEM and 3′-GFP plus Sal, digested with BglII and SalI, and ligated into IRES-GFP-RV that was digested with BglII and SalI. The Phoenix ecotropic packaging cell line was used for cotransfection with MinR1 Foxp3 and mHVEM retroviral constructs in the presence of Lipofectamine 2000. Retroviral supernatants were added to CD4+CD25− T cells activated overnight with PMA, ionomycin, and IL-2; transduction efficiency was 90–95% as determined by flow cytometry, and cells were centrifuged over Ficoll before use.
Intraabdominal vascularized mouse cardiac transplantation across a full MHC-mismatch was performed using BALB donors and C57BL/6 recipients, with anastomoses of donor ascending aorta and pulmonary artery end-to-side to recipient infrarenal aorta and inferior vena cava, respectively (13). Grafts were assessed daily by abdominal palpation; rejection was defined as cessation of cardiac contraction and confirmed by histology. For adoptive transfer studies of Treg function, BALB/c hearts were grafted into RAG2−/− recipients (B6 background), followed by i.v. transfer of 0.5 × 106 CD4+CD25− T cells plus 0.25 × 106 WT or HVEM−/− CD4+CD25+ Tregs; each cell population was isolated using magnetic beads for in vitro assays.
Allograft survival was used to generate Kaplan-Meier survival curves, and comparison between groups was performed by log-rank analysis.
BTLA and HVEM mRNA expression by resting and activated T cells
We used quantitative real-time PCR to examine the expression of BTLA, HVEM, and LIGHT by purified CD4+CD25− (non-Tregs) vs CD4+CD25+ (Tregs) cells. CD4+CD25− T cells and naturally occurring CD4+CD25+ Tregs were isolated using magnetic beads and stimulated in vitro with PMA/ionomycin plus IL-2. Cell activation led to BTLA up-regulation by non-Tregs but not by Tregs (Fig. 1,a). In contrast, cell activation induced higher expression of HVEM (p < 0.05) by Tregs than non-Tregs (Fig. 1,b). Expression of a second HVEM receptor, LIGHT, was transiently up-regulated by 4 h of stimulation of CD4 T cells in vitro but returned to baseline levels thereafter (Fig. 1,c). In addition, T cells from Scurfy mice, which have a truncated form of Foxp3 (16), lacked HVEM expression even after activation, suggesting the involvement of Foxp3 in regulation of HVEM expression (Fig. 1 d). Hence, upon activation, activated Teff cells express increased BTLA mRNA but down-regulate HVEM mRNA, whereas Tregs up-regulate HVEM but not BTLA mRNA.
BTLA and HVEM protein expression by resting and activated T cells
Flow cytometry showed BTLA expression by 10–15% of naive CD4+ and CD8+ T cells and >90% of B cells (Fig. 2,a). Adoptive transfer of CFSE-labeled T cells into F1 mice led to alloactivation-induced up-regulation of BTLA protein by almost all CD4+ and CD8+ T cells by 72 h (Fig. 2,b). Similarly, most CD4+Foxp3− T cells highly expressed BTLA protein after in vitro TCR ligation or TCR plus CD28 costimulation, but the extent of BTLA expression by CD4+Foxp3+ T cells remained low (Fig. 2,c). Hence, Teff cells, but not Tregs, up-regulated BTLA upon activation. Flow cytometric analysis of HVEM expression using a PE-labeled BTLA tetramer showed HVEM was mainly expressed by resting T cells (Fig. 2,d); although some CD11b+ monocytes expressed HVEM, there was no expression by B cells or dendritic cells. HVEM expression by CD4+ or CD8+ T cells was down-regulated by in vitro activation (Fig. 2,e). Activation of Foxp3+ Tregs led to progressively increased surface expression of HVEM upon activation, in contrast to an almost complete lack of HVEM expression by Foxp3− cells in the same wells (Fig. 2 f). These matching quantitative PCR and flow cytometry data led us to propose that ligation of BTLA on Teff cells, by HVEM on activated Tregs, contributed to the suppressive function of Tregs. To test this proposition, we generated HVEM−/− mice for in vitro and in vivo studies.
Differing mechanisms by which HVEM−/− and BTLA−/− mice have impaired Treg suppression
We used homologous recombination to disrupt exon 3 of the murine HVEM gene (Fig. 3). Our studies of homozygous HVEM−/− mice backcrossed for over eight generations to the C57BL/6 strain showed the mice were normal in appearance, growth and fertility, had a normal number of T cells, B cells, monocytes, NK cells, and granulocytes and normal lymphoid architecture (data not shown). Moreover, analysis of 4- to 6-wk-old WT, BTLA−/− and HVEM−/− mice showed a comparable number of CD4+CD25+ Foxp3+ Tregs in lymph nodes and spleen (Fig. 4 a), as well as in thymus and bone marrow samples (data not shown).
CFSE-labeled CD4+CD25− T cells from WT or HVEM−/− mice were stimulated in vitro with CD3 mAb and syngeneic APC, increasing numbers of Tregs from WT mice were added, and effector T cell proliferation assessed by CFSE-based division profiles. Regardless of the presence or absence of Tregs, the proliferation of HVEM−/− Teff cells was significantly increased compared with proliferation of cells in WT controls (first panel, Fig. 4,b), suggesting that HVEM is critical to down-regulation of T-T interactions. However, when comparing the suppressive function of WT and HVEM−/− Tregs, HVEM−/− Tregs were significantly less suppressive than WT Tregs (second panel, Fig. 4 b), indicating that the increased expression of HVEM on Tregs after activation is of functional significance. The presence or absence of HVEM on DC used in the Treg assays did not affect Treg suppression when assayed using WT Tregs, nor did they affect outcomes using HVEM−/− Tregs (data not shown). Collectively, our data suggest that HVEM is not only required for controlling T-T interactions but is important to Treg function.
In concurrent studies of the role of BTLA expression on Teff cells vs Tregs, we found that BTLA−/− Teff cells proliferated as well as WT Teff cells in the absence of Tregs (third panel, Fig. 4,b), consistent with the lack of HVEM expression by Teff cells after 48 h of in vitro stimulation (Fig. 2,b). However, as an increasing number of Tregs was added, the lack of Teff cell expression of BTLA led to Teff cells being significantly more resistant to Treg suppression by Tregs than WT Teff cells (third panel, Fig. 4,b). Conversely, and consistent with the low expression of BTLA on Tregs in either naive or activated states, BTLA−/− Tregs showed normal suppressive activity compared with WT Tregs (fourth panel, Fig. 4 b), indicating BTLA deficiency does not affect Treg suppressive function.
Overexpression of HVEM enhances Treg suppressive function
We used retroviral transduction, to test whether overexpression of HVEM would modify the function of Tregs. We have shown that transduction of murine CD4+CD25− cells with murine Foxp3 can efficiently convert Teff cells to Tregs (15), and we now transduced Foxp3 plus HVEM, or Foxp3 alone, into CD4+CD25− T cells. Efficiency of Foxp3 and HVEM transduction, monitored using human nerve growth factor receptor and GFP reporters, respectively, was >90% in each case (data not shown). Functions of CD4+Foxp3+ Tregs, with or without associated HVEM overexpression, were assessed by the in vitro suppression assay. As the retroviral vector used to transduce HVEM contained a GFP reporter, we used CFSE-labeled Thy1.1 congenic CD4+CD25− T cells as Teff cells, and the suppressive effect was determined by the total number of Thy1.1+ T cells recovered after 72 h of incubation, and by their corresponding CFSE profiles. Forced expression of HVEM by CD4+Foxp3+ T cells significantly enhanced their suppressive function, resulted in a decreased total number of Thy1.1+ Teff cells at 72 h, as a result of fewer Teff cells entering the cell cycle (Fig. 5 a). These data show that HVEM is an important effector molecule for Treg function.
BTLA−/− and HVEM−/− mice are resistant to induction of allograft tolerance
Given our data on the importance of the HVEM-BTLA pathway in mediating the suppressive effects of Tregs in vitro, we tested the impact of HVEM or BTLA deficiency on Treg function in vivo. We used a model in which fully MHC-mismatched murine cardiac allografts survive permanently in recipients receiving 2 wk of therapy with the histone deacetylase inhibitor, trichostatin A, plus rapamycin, used at a subtherapeutic dosage (17). We had found that trichostatin A enhanced the function of Tregs in vivo, and that a small dose of rapamycin was necessary to limit residual effector T cell proliferation, in the early posttransplantation period, which was not controlled by Tregs alone (18). As previously, trichostatin A plus rapamycin therapy achieved permanent allograft survival using the stringent BALB/c→C57BL/6 cardiac transplant model, whereas depletion of central and peripheral CD4+CD25+ Tregs by pretransplant thymectomy and CD25 mAb caused rejection of allografts (Fig. 5,b). However, we now found that long-term allograft survival could not be achieved in recipients deficient in BTLA or HVEM (Fig. 5,b), consistent with a role for the HVEM-BTLA pathway in the optimal function of Treg in vivo. Consistent with this, adoptive transfer of WT but not HVEM−/− B6 Tregs to RAG2−/− recipients of BALB/c allografts was able to suppress allograft rejection induced by cotransfer of WT Teff cells (Fig. 5 c).
The current data show that TCR activation leads to BTLA up-regulation by Teff cells and HVEM up-regulation by Tregs, and that as a result, the HVEM-BTLA pathway is important to Treg-dependent control of T cell responses in vitro and in vivo. These data provide a new perspective on the BTLA-HVEM pathway and show that HVEM can be added to the armamentarium of Treg effector molecules relevant to immune therapy. Several TNFR family members have costimulatory effects on T cells, including CD27, 4-1BB (CD137), OX40 (CD134), CD30 and GITR, and GITR and OX40 are also implicated in control of Treg function (19, 20, 21, 22, 23). GITR is expressed at high levels on CD4+CD25+ Tregs and its ligation by GITR ligand expressed by APC revokes Treg suppression and provides a costimulatory signal for Ag-driven proliferation of naive T cells and polarized Th1 and Th2 clones (22, 23). Likewise, although both naive and activated Tregs express OX40, triggering OX40 on Tregs using agonist Abs inhibited their capacity to suppress, and restored Teff cell proliferation and IL-2 cytokine production (24). However, in contrast to CTLA4, GITR, or OX40, which are expressed by both Tregs and Teff cells after activation, the inhibitory receptor BTLA is up-regulated in Teff cells but is expressed at only low levels by Tregs, whereas HVEM is mainly expressed by Tregs after T cell activation, resulting in negligible competition for binding to HVEM between Tregs and Teff cells. This differential pattern of expression for receptor and ligand on resting vs activated T cells provides an additional and hitherto unexpected means for fine tuning and regulation of T cell responses via the BTLA-HVEM pathway.
Given that the binding of HVEM and BTLA was identified in 2005 (6, 9), the questions arise as to how the connection to Treg function was not previously recognized, and what are the therapeutic implications of this expression? With regard to the first question, LIGHT knockout mice showed enhanced cardiac allograft survival using the same model as used in the current studies, and use of HVEM-Ig in WT recipients also prolonged survival (25). These data were interpreted as showing the importance of the LIGHT/HVEM pathway in T cell costimulation, though in retrospect a role for HVEM-Ig binding to BTLA and promoting negative signals within T cells may also have contributed. Likewise, a previous report of an HVEM knockout mouse emphasized an unexpected enhancement of T cell responses, with increased susceptibility to autoimmunity including Con A-induced hepatitis and myelin oligodendrocyte glycoprotein peptide-induced experimental autoimmune encephalopathy (11), though no data were generated relating to Treg functions. Clearly the complexity of interactions involving HVEM and its ligands, as well as the ability of LIGHT itself to bind HVEM, lymphotoxin-β receptor, and the soluble TNFR decoy receptor 3, provide a considerable number of permutations, even before taking into account differential expression upon cell activation.
With regard to therapeutic aspects, agonistic mAbs that cross-link BTLA and suppress the proliferation and production of cytokines such as IL-2 and IFN-γ by human (26) and murine (27) T cells are reported, and administration of HVEM-Ig can also dampen immune responses (25, 28, 29, 30). Our data suggest that the enhancement of HVEM expression by gene therapy or other approaches may be an additional tool for immune therapy, as may consideration of the BTLA-HVEM pathway during in vitro efforts to scale-up, for eventual in vivo application, large numbers of Tregs (31, 32).
Lastly, our data are relevant to the interpretation of two reports published during revision of this work. First, the immunogenicity of vaccines was markedly improved when HVEM-BTLA interactions were blocked using HSV glycoprotein D-coupled Ag (33). Our data suggest that this increased potency would result at least in part from impairment of host Treg responses. Second, in addition to the inhibitory HVEM/BTLA interaction, HVEM was shown to bind to CD160 on T cells and inhibit T cell activation (34). In this case, our data would suggest that HVEM expression by Tregs can exert inhibitory effects via ligation of BTLA or CD160 on Teff cells. Future studies will address whether these two inhibitory pathways function independently and whether they are equally susceptible to regulation by HVEM-expressing Tregs.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by Grant R01-AI54720 from the National Institutes of Health (to W.W.H.).
Abbreviations used in this paper: Treg, regulatory T cell; Teff, effector T; GITR, glucocorticoid-induced TNFR family-related receptor; BTLA, B and T lymphocyte attenuator; HVEM, herpesvirus entry mediator; IRES, internal ribosomal entry site; WT, wild type.