Germinal centre alloantibody responses are mediated exclusively by indirect-pathway CD4 T follicular helper cells

Abstract

The durable alloantibody responses that develop in organ transplant patients indicate long-lived plasma cell output from T-dependent germinal centres (GCs), but which of the two pathways of CD4 T cell allorecognition are responsible for generating allospecific T follicular helper (T_FH) cells remains unclear. This was addressed by reconstituting T-cell deficient mice with monoclonal populations of TCR-transgenic CD4 T cells that recognised alloantigen only as conformationally-intact protein (direct pathway) or only as self-restricted allopeptide (indirect pathway), and then assessing the alloantibody response to a heart graft.

Recipients reconstituted with indirect-pathway CD4 T cells developed long-lasting IgG alloantibody responses, with splenic GCs and allospecific bone marrow plasma cells readily detectable 50 days after heart transplantation. Differentiation of the transferred CD4 T cells into T_FH cells was confirmed by follicular localisation and by acquisition of signature phenotype. In contrast, IgG alloantibody was not detectable in recipient mice reconstituted with direct-pathway CD4 T cells. Neither prolongation of the response by preventing NK cell killing of donor dendritic cells, nor prior immunisation to develop CD4 T cell memory altered the inability of the direct-pathway to provide allospecific B cell help.

CD4 T cell help for GC alloantibody responses is provided exclusively via the indirect-allorecognition pathway.

Introduction

Cellular immunity has been long regarded as the principal contributor to allograft rejection, but recent clinical data suggests that the humoral arm may be at least as important, in that the presence of donor-specific antibody either before transplantation or that develops afterwards is now clearly associated with failure of kidney (1-5) and heart allografts (6-8). As with conventional protein antigens, the development of effective alloantibody is critically dependent upon the provision of help from CD4 T cells (9-13); interventions that target CD4 T cells may thus disable both the cellular and humoral responses normally responsible for graft rejection. Although modern immunosuppressive agents effectively block cellular alloimmune responses, they act non-specifically and risk life-threatening infection and cancer development. Antigen-specific approaches that obviate these concerns by disabling only those T cells responsible for providing help to allospecific B cells remain frustratingly unrealised, and their development hampered by limited understanding of the interactions between alloreactive T and B lymphocytes that underpin alloantibody production.

Transplantation is unusual because CD4 T cells can recognise alloantigen through two distinct pathways (14-17): in the direct pathway, which is unique to transplantation, alloantigen is recognised as intact protein on the surface of donor APCs; whereas in the indirect pathway, which is akin to recognition of conventional protein antigen, alloantigen is first processed by recipient APCs and then presented as peptide fragments in the context of host MHC class II. Which of these two pathways of alloreactive CD4 T cell activation is responsible for providing help for alloantibody production remains controversial (18, 19), not least because the humoral alloimmune response is complex and composed of several anatomically-distinct components. Thus, simple assay of serum alloantibody may fail to reveal subtle yet important differences in how the helper CD4 T cell allorecognition pathway impacts on the various constituent arms. In this respect, the germinal centre (GC) response requires special consideration (reviewed in (20)), because a recently-described population of highly specialised T follicular helper cells (T_FH) is critical to its development (21-24) and because its output; long-lived plasma cells (LLPC) and memory B cells with high affinity for alloantigen; is likely to hold most relevance for clinical transplantation.

For non-transplant antigens, landmark studies in the 1980s highlighted the requirement for B cells to act as APCs and present processed peptide derived from their internalised target antigen for ‘cognate’ self-restricted interaction with the TCR of antigen-specific helper CD4 T cells (25). This suggests that only indirect-pathway CD4 T cells can provide help to allospecific B cells, because, unlike direct-pathway CD4 T cells, they can interact in a similar cognate fashion with the allopeptide presented by the B cell (Fig. 1A). In support, Auchincloss et al. have demonstrated that mice, in which absence of peripheral MHC class II expression restricted CD4 T cell allorecognition to the direct pathway exclusively, mounted only limited IgM alloantibody responses to skin allografts; IgG isotype-switching did not occur (18). However in that study, the lack of MHC class II expression in recipient mice may have more wide-ranging effects than simply limiting available pathways of allorecognition. Of chief concern, tonic MHC class II signalling, that is non-cognate and peptide-independent, is required for optimal CD4 T cell survival, homeostatic proliferation and reactivity (26, 27); similarly MHC class II signalling provides essential proliferative responses in B cells (28, 29). Host MHC class II expression may thus be required for optimal activation of direct-pathway CD4 T cells and in particular, for the ability of the latter to provide non-cognate help to allospecific B cells. In addition, because the direct pathway lasts only briefly (due to rapid NK-cell mediated lysis of donor dendritic cells (30)), Auchincloss’ findings from a skin graft model may not reflect delivery of help for more rapidly-evolving alloantibody responses to vascularised allografts. This concern also applies to a more recent study (31) , which used adoptive transfer of monoclonal populations of TCR-transgenic CD4 T cells into recipient mice that expressed MHC class II normally, to similarly conclude that only indirect-pathway CD4 T cells were capable of providing help for IgG-switched alloantibody responses against skin grafts. However, in the model adopted, hen egg lysozyme was used as a surrogate alloantigen and the results may therefore not reflect responses against MHC alloantigens.

FIGURE 1. Schematic representation of the provision of help by indirect and direct-pathway CD4 T cells to allospecific B cells.

A, Analogous to conventional protein antigen, indirect-pathway CD4 T cells interact cognately with allopeptide presented by B cells following internalisation of their target antigen. B, Donor APCs that expresses both B cell target alloantigen and the allo-MHC class II determinant for CD4 T cell recognition enable formation of a ‘three-cell cluster’ that may provide non-cognate help to allospecific B cells from direct-pathway CD4 T cells.

That direct-pathway CD4 T cells may provide help to allospecific B cells through non-cognate interactions is supported by early in vitro studies demonstrating the provision of cytokine-mediated, contact-independent help for antibody responses against conventional protein antigens (reviewed in (25, 32)). Although physical linkage between the allospecific B and helper T cell is not possible, close proximity is presumably required and is possible through simultaneous interaction with a donor APC that expresses both the B cell target alloantigen and the allo-MHC class II determinant for CD4 T cell recognition. This ‘three-cell cluster model’ (Fig. 1B) was first suggested by Fabre, who noted that blockade of donor MHC class II alloantigen abrogated anti-MHC class I alloantibody responses to a rat heart graft (19).

Assuming the paradox created by Fabre’s and Auchincloss’ work reflects the differences in the experimental systems examined, its resolution requires re-testing using a vascularised allograft model which limits T cell activation to a particular allorecognition pathway, but maintains normal recipient MHC class II expression. Here this is achieved by adoptive transfer of monoclonal populations of TCR-transgenic CD4 T cells, specific for MHC alloantigen via exclusively the direct or exclusively the indirect-pathway, into T cell-deficient, but B cell-replete, murine recipients of heart allografts. This allows a definitive assessment of how the CD4 T cell allorecognition pathway influences alloantibody production; an assessment that includes a detailed examination of the potential for naïve and memory CD4 T cells to provide help for LLPC production following GC development, as well as the role of NK cells in curtailing direct pathway help through rapid killing of donor DCs.

Materials and Methods

Animals

TCR^−/− mice B6.129P2-Tcrb^tm1MomTcrd^tm1Mom/J (H-2^b) (33), MHC class II^−/− mice B6.129S2-H2^dlAb1-Ea/J (H-2^b) (34), μMT B cell-deficient mice B6.129S2-Ighm^tm1Cgn/J (H-2^b) (35), and bm12 mice B6(C)-H2-Ab1^bm12/KhEgJ (H-2^bm12) were purchased from the Jackson Laboratory (Bar Harbor, ME). C57BL/6 RAG2^−/− mice (H-2^b) were gifted by Prof T. Rabbitts (LMB, Cambridge, UK). TCR-transgenic RAG1^−/− TCR75 mice (H-2^b), specific for I-A^b-restricted H-2K^d_54-68 peptide (36), were gifted by Prof P. Bucy (University of Alabama, Birmingham, AL). I-A^bm12-specific TCR-transgenic RAG2^−/− ABM mice (H-2^b) (37) were gifted by Dr T. Crompton (Imperial College, London, UK). C57BL/6 (H-2^b, B6), BALB/c (H-2^d), and (BALB/c × B6) F1 mice were purchased from Charles River Laboratories (Margate, UK). All animals were maintained in specific-pathogen-free facilities and all experiments approved by the United Kingdom Home Office under the Animal (Scientific Procedures) Act 1986.

Generation of Bone Marrow Chimeras

To create mice that lacked MHC class II expression only on their B cells (BCII^−/−), RAG2^−/− mice were sublethally irradiated (2Gy) and reconstituted with 2×10⁷ bone marrow cells obtained from MHC class II^−/− mice. The B lymphocyte compartment in these mice is derived solely from donor MHC class II^−/− bone marrow, whereas other APC lineages are formed from both donor and recipient bone marrow, and are thus a heterogeneous population of MHC class II positive and negative cells. Control chimeric mice (BCII^+/+) were created by reconstituting sublethally-irradiated RAG2^−/− mice with 2×10⁷ B6 bone marrow cells. B cell-replete ABM mice (TCR^−/−.ABM) were created by reconstituting sublethally-irradiated TCR^−/− mice with 2×10⁷ bone marrow cells from RAG2^−/− ABM mice. Chimerism was confirmed by flow cytometric analysis of PBMC at least 4 weeks after reconstitution.

Skin and Heterotopic Heart Transplantation

Full thickness tail skin was sutured as 1cm² grafts onto the recipient’s back. Vascularized cardiac allografts were transplanted intra-abdominally by the technique of Corry and colleagues (38) and rejection, defined as cessation of palpable myocardial contraction, confirmed at explant. T cell-reconstitution was performed by intravenous injection with 10⁷ splenocytes from TCR75 or ABM mice the following day.

NK and CD4 T cell depletion

Recipient TCR^−/−.ABM chimeric mice were treated with 0.5mg anti-NK1.1-depleting monoclonal antibody (mouse IgG2a, clone PK136 - hybridoma purchased from ATCC-LGC Standards Partnership (Middlesex, UK)) on days −2, 0, and +2 in relation to transplantation and weekly thereafter (39). Donor (BALB/c × bm12) F1 mice were treated on days -5 and -3 prior to removal of the hearts with 1mg anti-CD4-depleting monoclonal antibody (rat IgG2b, clone YTS191.1.1.2 - hybridoma purchased from ECACC at the Health Protection Agency (Porton Down, UK)).

Immunohistochemistry and Immunofluorescence

Splenic 7μm cryostat sections were stained for the presence of B220^+ve B cells; rat anti-mouse B220 (clone RA3-6B2, BD Pharmingen, San Diego, CA) directly conjugated with FITC or detected with Cy3-conjugated goat anti-rat IgG (clone 112-165-143, Jackson Immunoresearch Laboratories, West Grove, PA), peanut agglutinin (PNA)^+ve GC B cells with FITC-conjugated PNA (Vector Laboratories, Peterborough, UK) and CD3^+ve T cells; PE-conjugated rat anti-mouse CD3 (clone 145-2C11, BD Pharmingen, San Diego, CA). Sections were counterstained with 20% Harris’ hematoxylin (Sigma-Aldrich, Poole, UK) and viewed using an IX81 microscope with a 20x 0,70 UplanApo objective lens (Olympus, Japan). Images were photographed using an ORCA-ER digital Camera (Hamamatsu Photonics, Japan) and acquired with Cell^R 2.6 software (Olympus Soft Imaging Solutions GmbH, Germany).

Generation of recombinant soluble H-2K^d

Recombinant soluble H-2K^d was produced using the method previously described for rat MHC class I (40). In brief, pET-22b+ expression plasmids containing the DNA sequences encoding either amino acids 1-280 of the H-2K^d heavy chain (α1, α2 and α3 extracellular domains) or the murine β2-microglobulin (gifted by Prof P. Lehner, University of Cambridge, Cambridge, UK) were transformed into E. coli BL21 (DE3) strain bacteria (Novagen, Merck, UK) and grown in LB broth (Invitrogen, Paisley, UK). Recombinant H-2K^d heavy chain or β2-microglobulin was extracted from inclusion bodies released from E. coli pellets by chemical lysis. Soluble H-2K^d molecules were generated by refolding the purified heavy chain and β2-microglobulin around a synthetic peptide (TYQRTRALV) (ISL, Paignton, UK) using the dilution method of Garboczi et al. (41). Finally, fast protein liquid chromatography purification of the refold mixture was performed (AKTA FPLC, Amersham Biosciences, Buckinghamshire, UK) and the appropriate fraction was collected, pooled, filter sterilized, and stored in aliquots at 4°C.

Determining circulating anti-H-2K^d alloantibody

Serum samples were collected from experimental animals weekly and analyzed for the presence of anti-H-2K^d IgG alloantibody by ELISA. In brief, 96-well ELISA plates (Immulon 4HBX, Thermo, Milford, MA) were coated with recombinant conformational H-2K^d at 5 μg/ml in Na₂CO₃-NaHCO₃ buffer (pH 9.6). Plates were blocked with 1% Marvel dried skimmed milk powder (Premier International Foods, UK), tripling serial dilutions of test sera added and bound IgG antibodies detected by incubating with biotinylated Rabbit F(ab’)2 anti-mouse IgG (STAR11B, AbD Serotec, Oxford, UK) and ExtrAvidin Peroxidase conjugate (Sigma, Poole, UK). Sure Blue substrate (KPL, Gaithersburg, MD) was then added, the reaction stopped by the addition of 0.2M H₂SO₄ and the absorption at 450nm measured in a FLUOstar OPTIMA plate reader (BMG Labtech, UK). For each sample, an absorbance vs. dilution curve was plotted, and the area under the curve was calculated (42). The area under the curve of an experimental sample was expressed as the percentage of positive control (pooled hyperimmune) serum.

B cell ELISPOT assay

Single anti-H-2K^d IgG antibody secreting cells were detected by ELISPOT assay. Briefly, Single cell suspensions from spleen and bone marrow were added onto 96-well MultiScreen Filter Plates (Millipore, Billerica, MA) that were previously coated with recombinant conformational H-2K^d at 2μg/ml in Na₂CO₃-NaHCO₃ buffer (pH 9.6) and blocked with 10% Bovine Serum Albumin (Sigma, Poole, UK). After incubation at 37°C, 5% CO₂ for 20hrs, bound IgG antibodies were detected by incubating with biotinylated Rabbit F(ab’)2 anti-mouse IgG (STAR11B, AbD Serotec, Oxford, UK) and ExtrAvidin Peroxidase conjugate (Sigma, Poole, UK). Spots were developed using 3-amino-9-ethyl-carbazole solution (Sigma, Poole, UK) and read on an AID^TM Elispot Reader version 3.5 (Autoimmun Diagnostika, Strasberg, Germany). Data was expressed as mean number of responders per 10⁶ cells (± SEM).

Autoantibody generation and detection

ABM CD4 T cells (2×10⁶) purified by autoMACS^TM separation using anti-CD4 microbeads (Miltenyi Biotec, Surrey, UK) were injected i.v. into bm12 mice. Mice were bled weekly and circulating autoantibody levels determined by HEp-2 indirect immunofluorescence (The Binding Site Ltd, Birmingham, UK). Serum was incubated on slides coated with HEp-2 cells and bound antibody detected with FITC-conjugated anti-mouse IgG (STAR 70; Serotec, Oxford, UK). For each serum, photomicrographs were taken and the intensity of staining determined by integrated morphometric analysis using MetaMorph software (Molecular Devices Corporation, Downingtown, PA). The fluorescence value was then determined relative to a standard curve of serial diluted pooled hyperimmune serum that was assigned an arbitrary value of 1000 fluorescence units.

Flow cytometry

FITC-conjugated anti-mouse CD4 (clone GK1.5), FITC-conjugated anti-mouse CD11c (clone HL3), FITC-conjugated anti-mouse CD44 (clone IM7), PE-conjugated anti-mouse CD19 (clone 1D3), PE-conjugated anti-mouse CD90.1/Thy1.1 (clone OX-7), PE-conjugated anti-mouse CD279/PD1 (clone J43), PE-conjugated anti-mouse TCR Vβ8 (clone F23.1), allophycocyanin-conjugated anti-mouse CXCR5 (clone 2G8), PE-Cy7-conjugated anti-mouse CD4 (clone L3T4), biotinylated anti-mouse I-A^b (clone 25-9-17), biotinylated anti-mouse CD4 (clone GK1.5), biotinylated anti-mouse CD49b/pan-NK (clone DX-5) and biotinylated anti-mouse CD90.1/Thy1.1 (clone OX-7) were purchased from BD Pharmingen (San Diego, CA). Peripheral blood (depleted of erythrocytes by incubating with 0.17M NH₄Cl red cell lysis buffer) and splenic single cell suspensions were blocked with anti-mouse CD16/CD32 (clone 2.4G2, BD Pharmingen, San Diego, CA), before staining with the relevant antibodies and dead cell exclusion dye 7-AAD (BD Pharmingen, San Diego, CA). Biotinylated antibodies were detected by allophycocyanin -conjugated streptavidin (Invitrogen, Paisley, UK) or allophycocyanin -Cy7-conjugated streptavidin (BD Pharmingen, San Diego, CA) and all cells analysed on a FACSCanto II flow cytometer with FACSDiva software (BD Biosciences, San Jose, CA).

CFSE cell proliferation

Single cell suspensions of splenocytes obtained from TCR75 or ABM mice were stained with 5mM CFSE (Invitrogen, Molecular Probes, Paisley, UK) in the dark for 5mins and then quenched with PBS + 5% FCS. 2-5×10⁶ CFSE stained splenocytes were injected i.v. into recipient mice on the day of grafting, spleens harvested 4 or 7 days later and flow cytometry performed using allophycocyanin -conjugated anti-CD4 plus PE-conjugated anti-CD90.1/Thy1.1 to identify TCR75 T cells and PE-conjugated anti-TCR Vβ8 to identify ABM cells.

Statistical analysis

Alloantibody responses were compared using the Mann-Whitney U test. p < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism version 5.02 (GraphPad Software Inc., San Diego, CA).

Results

Absence of MHC class II expression on B cells abrogates IgG alloantibody responses

B cells lacking MHC class II expression can isotype switch in vitro (43) but to clarify whether switching is possible in vivo, bone marrow chimeric mice were created with a defect in MHC class II expression limited only to B cells (BCII^−/−); MHC class II expression was maintained in the host non-B cell APC and, importantly, CD4 T cell numbers were normal (Fig. 2A). BCII^−/− mice did not produce class-switched IgG alloantibody when challenged with a BALB/c heart allograft, whereas in contrast, control BCII^+/+ chimeric mice, whose B cells expressed MHC class II normally (Fig. 2A), developed anti-H-2K^d IgG responses similar to those seen in naïve wild-type (WT) recipients (Fig. 2B). The absence of alloantibody in the BCII^−/− mice was not due to an inability to mount indirect-pathway CD4 T cell responses, because although only a small proportion of the non B cell APC in the BCII^−/− mice express MHC class II (Fig. 2A; those derived from host bone marrow), TCR-transgenic TCR75 CD4 T cells (that recognize donor H-2K^d antigen as self-restricted allopeptide) nevertheless underwent robust division when adoptively-transferred into BCII^−/− recipients the day after transplantation (Fig. 2C). This response was comparable to that observed in BCII^+/+ recipients (Fig. 2C). In support of a critical role for B cell MHC class II expression in the production of class-switched antibody, we have previously demonstrated that BCII^−/− mice do not develop IgG autoantibody responses upon induction of graft versus host disease (44).

FIGURE 2. MHC class II expression on B cells is necessary for class-switched alloantibody production.

A, Chimeric mice lacking MHC class II expression on B cells (BCII^−/−) were created by reconstituting sub-lethally irradiated RAG2^−/− mice with bone marrow from MHC class II^−/− donors. Four weeks later, flow cytometric analysis of peripheral blood (upper and lower rows) or splenocytes (middle row) from chimeric mice or control chimeric mice expressing MHC class II on B cells (BCII^+/+) was performed, gating on live (7-AAD^-ve) lymphocytes. B cells (CD19) from BCII^−/− mice lack I-A^b MHC class II expression (upper row), whereas I-A^b expression is maintained on host bone marrow-derived CDIIc^+ve DCs (middle row). CD4 T cell numbers are comparable in both groups (lower row). B, Serum levels of anti-H-2K^d IgG alloantibody following challenge with BALB/c heart allograft. The alloantibody response in similarly-grafted WT B6 mice is shown as comparison. Data, representing a minimum of three mice per group, are presented as mean vales +/− SEM. C, Representative flow cytometry plots (of two independent experiments) of CFSE-labelled TCR75 (Thy 1.1^+ve) CD4 T cells 4 days after adoptive transfer into BCII^−/− or BCII^+/+ mice that received a BALB/c heart graft. The indicated levels of cell division were calculated using FlowJo software. A control plot following adoptive transfer of CFSE TCR75 CD4 T cells into BCII^−/− recipients of a B6 syngeneic heart is shown for comparison.

These experiments demonstrate that a definitive comparison of the abilities of direct and indirect-pathway CD4 T cells to provide help for alloantibody production requires a model in which B cell MHC class II expression is maintained. Subsequent experiments thus used TCR^−/− mice - that despite lacking all T cells retain relatively normal B cell homeostasis and compartmentalisation (45) – as recipients of heart allografts; helper function was provided by adoptive transfer of monoclonal populations of TCR-transgenic CD4 T cells.

Indirect-pathway CD4 T cells provide help for class-switched alloantibody responses

To examine the ability of indirect-pathway CD4 T cells to provide help for alloantibody production, TCR^−/− recipients of BALB/c heart allografts were reconstituted with 5 × 10⁶ splenocytes from RAG1^−/− TCR75 mice. These contain a monoclonal population of CD4 T cells specific for one of the dominant peptide epitopes (H-2K^d_54-68) recognised following processing of H-2K^d alloantigen in the B6 strain (36, 46, 47); an epitope that H-2K^d-specific B cells would be expected to present upon internalisation of target alloantigen. In contrast to non-reconstituted TCR^−/− recipients, those reconstituted with TCR75 T cells developed strong anti-H-2K^d IgG alloantibody responses to a BALB/c heart graft, which were even greater than those observed in WT recipients (Fig. 3A).

FIGURE 3. Indirect pathway CD4 T cells provide help for germinal centre alloantibody responses.

A, Serum anti-H-2K^d IgG alloantibody responses to a BALB/c heart allograft in B6, TCR^−/− or TCR^−/− mice adoptively transferred with H-2K^d-peptide-specific TCR75 splenocytes. Data, representing a minimum of three mice per group, are expressed as mean +/− SEM. B, Kaplan-Meier survival curves of BALB/c hearts transplanted into TCR^−/− (n=3), TCR^−/− adoptively transferred with TCR75 splenocytes (n=5), or TCR75 mice (n=3). C, Representative (of two independent experiments) flow cytometry histogram plots of CFSE-labelled TCR75 CD4 T cells 4 days after adoptive transfer into B6 or μMT recipients of BALB/c heart allografts. The indicated levels of cell division were calculated using FlowJo software. D, Representative (of two independent experiments) photomicrographs of immunofluorescence staining of splenic sections from TCR^−/− mice (i-ii) and TCR^−/− mice adoptively transferred with TCR75 splenocytes (iii-viii) harvested 49 days after BALB/c heart transplantation. Panels i, iii and vi depict B220 B cell staining; ii & iv -PNA staining; and vii - CD3 staining. Panel v represents an overlay of iii and iv, while viii, an overly of vi and vii. Of note, PNA^+ve GCs developed in TCR^−/− recipients reconstituted with TCR75 splenocytes (panel iv), but not in TCR^−/− mice, and adoptively transferred TCR75 T cells were detectable within B cell follicles (panel vii). The scale bar shown indicates 100μm. The corner chart depicts the mean (+/− SEM) of PNA^+ve secondary follicles as a percentage of total B cell follicles. E, Number of H-2K^d specific IgG-secreting antibody secreting cells (ASCs) per 10⁶ plated cells within spleen and bone marrow of TCR^−/− and TCR75-reconstituted TCR^−/− recipients 49 days after challenge with BALB/c heart allografts. Data, representing a minimum of three mice per group, is expressed as mean +/− SEM. F, Flow cytometry analysis, gating on live 7-AAD^-ve, CD4^+ve, Thy 1.1^+ve TCR75 cells, reveals differentiation of TCR75 CD4 T cells into PD1^high, CXCR5^high T_FH phenotype, 11 days after transfer into B6 recipients of BALB/c heart allografts, but not upon transfer into either naïve B6 mice or μMT recipients of BALB/c heart allografts. CD44 fluorescence intensity of the gated TCR75 population is also depicted for each group.

Rejection kinetics were also markedly different in the two groups, in that heart grafts were rejected within 8 days in the reconstituted TCR^−/− recipients but continued to beat strongly in the non-reconstituted recipients until termination of the experiment at day 50 (Fig. 3B). Allograft rejection in the recipients reconstituted with TCR75 splenocytes was not simply a consequence of activation of the transferred CD4 T cells, because despite comparably-strong division of TCR75 T cells upon transfer into either WT or B cell-deficient (μMT) B6 recipients of BALB/c heart allografts (Fig 3C), B cell-deficient RAG1^−/− TCR75 mice do not reject BALB/c heart allografts (Fig. 3B). These results suggest instead that indirect-pathway CD4 T cells are unable to mediate acute heart allograft rejection when acting as autonomous cellular effectors, but can do so as a helper population for alloantibody production.

To examine whether the transferred indirect-pathway CD4 T cells mediated development of a GC alloantibody response, histological and immunohistochemical analysis was performed on recipient spleens harvested fifty days after heart transplantation. Over half of the follicles had undergone secondary maturation to become PNA (48) positive GCs and notably, transferred TCR75 T cells were detectable within the follicles (Fig. 3D). In addition, approximately three-fold greater anti-H-2K^d plasma cells were present within the bone marrow than within the spleen (Fig. 3E), consistent with LLPC output from a GC response (49).

Given the recent identification of a select and highly-specialised subset of CD4 T_FH cells as pivotal for providing essential helper signals for GC development (50-52), we sought to confirm whether the transferred TCR75 T cells acquire T_FH characteristics. One of the key attributes of the T_FH cell is that only precursors with high affinity for target antigen are selected; these presumably out-compete those antigen-specific cells of lower affinity. For these experiments, congenically-labelled TCR75 T cells were therefore transferred into WT, rather than TCR^−/−, recipients of BALB/c heart grafts, as this permitted competition with the endogenous B6 population for differentiation to T_FH status. As shown in Fig. 3F, approximately 6% of the transferred TCR75 T cells acquired the CD4^+ve CD44^high CXCR5^high PD1^high (53) signature-phenotype of the T_FH subset. This population did not develop upon transfer either to naive B6 mice or, more notably to transplanted μMT mice (Fig 3F), despite up-regulation of activation marker CD44 in the latter comparable to WT recipients (Fig. 3F). This accords with the recent demonstration that B cell antigen presentation is required for differentiation into T_FH status (54).

Naïve direct-pathway CD4 T cells are unable to provide help to allospecific B cells

A similar adoptive-transfer model into TCR^−/− recipients was developed to examine the role of direct-pathway CD4 T cell help in alloantibody production. TCR^−/− mice were engrafted with BALB/c × bm12 F1 hearts and reconstituted the next day with 10⁷ splenocytes from TCR-transgenic RAG2^−/− ABM mice. These mice contain a monoclonal population of CD4 T cells that recognise I-A^bm12 alloantigen of the F1 donors exclusively via the direct-pathway. Although the transferred transgenic CD4 T cells could potentially provide help via the three cell cluster model for development of anti-MHC class I alloantibody (Fig. 4A), anti-H-2K^d IgG alloantibody was not detectable at any time point after heart transplantation (Fig. 4B). However all hearts were still beating 50 days after transplant and thus to exclude the possibility that reduced viability caused by the transfer of ABM CD4 T cells contributed to the failure to develop humoral alloimmunity, the experiments were repeated using TCR^−/−.ABM bone-marrow chimeric recipients in which the ABM CD4 T cell population was self-renewing from hematopoietic precursors (Fig. 4C). Again, allaoantibody did not develop (Fig. 4B). The failure of ABM CD4 T cells to provide help for alloantibody was not due to their inability to mount a functional allospecific response, because: ABM CD4 T cells underwent robust division following adoptive transfer into TCR^−/− recipients of BALB/c × bm12 F1 heart allografts (Fig 4D); TCR^−/−.ABM bone marrow chimeric mice rejected bm12 skin grafts rapidly (MST 10 days, n= 4); and ABM CD4 T cells provided help through graft-versus-host recognition of I-A^bm12 alloantigen for autoantibody production (44, 55) upon transfer into naïve bm12 recipients (Fig. 4E).

FIGURE 4. Direct-pathway CD4 T cells cannot provide help for class-switched alloantibody responses.

A, Schematic representation of a three cell cluster between (BALB/c × bm12) F1 donor APCs, anti-H-2K^d specific B cells and direct-pathway ABM CD4 T cells. B, Anti-H-2K^d IgG alloantibody response to a (BALB/c × bm12) F1 heart transplant in TCR^−/− mice (n=5), TCR^−/− mice reconstituted with ABM splenocytes (n=5), and TCR^−/−.ABM bone marrow chimeric mice (n=3). Values from un-grafted naive mice (n=7) and B6 recipients of BALB/c hearts (n=4) are shown for comparison. C, Representative flow cytometry analysis of peripheral blood from TCR^−/−.ABM bone marrow chimeric mice, four weeks after creation. A representative plot of naïve TCR^−/− mice are shown as comparison. D, Representative (of three independent experiments) flow cytometry histogram plots of CFSE-labelled ABM CD4 T cells 7 days after adoptive transfer into TCR^−/− recipients of (BALB/c × bm12) F1 heart allografts. The response following a syngeneic B6 heart graft is shown for comparison. E, Autoantibody responses, calculated from Hep-2 indirect immunofluorescence staining, following the adoptive transfer of either 2×10⁶ ABM (n=4) or syngeneic (n=3) CD4 T cells into WT bm12 mice. Values from naïve mice are shown as comparison (n=3). F, Representative (of two independent experiments) flow cytometry histograms, gating on CD4^+ve, Thy1.1^+ve, live (7-AAD^-ve), CFSE-labelled TCR75 CD4 T cells, four days after adoptive transfer into bm12.TCR^−/− mice challenged at transfer with either a BALB/c or (BALB/c × B6) F1 heart graft. G, Anti-H-2K^d IgG alloantibody responses in bm12.TCR^−/− mice, reconstituted with TCR75 splenocytes and challenged with (BALB/c × B6) F1 heart grafts (n=3). Values from naïve mice (n=3) and d21 sera from B6 recipients of BALB/c hearts (n=4) are shown for comparison.

Because inherent differences in the activation-induced cytokine profiles of the ABM and TCR75 T cells may influence their ability to provide B cell help, a further experimental model was devised that incorporated TCR75 T cells but limited their response to the direct pathway. TCR75 CD4 T cells were adoptively transferred into TCR^−/− mice that had been crossed onto a bm12 background (TCR^−/−.bm12). TCR75 CD4 T cells do not respond to processed H-2K^d when restricted on the MHC class II I-A^bm12 antigen (Fig. 4F) and thus when transferred into TCR^−/−.bm12 recipients of a BALB/c × B6 F1 heart can only recognise target I-A^b/K^d_54-68 epitope directly on donor APCs (Fig. 4F); indirect recognition is not possible. When restricted in this fashion, adoptively-transferred TCR75 T cells were unable to provide help for alloantibody production (Fig. 4G).

Neither prevention of donor DC killing nor memory reactivation enable direct pathway CD4 T cells to provide allospecific B cell help

Although the above experiments suggest strongly that direct-pathway CD4 T cells are unable to provide help to alloreactive B cells, we thought it important to examine two further considerations before the three-cell cluster model could be refuted definitively: that rapid termination of the direct pathway response, rather than an intrinsic defect in its helper potential, prevented alloantibody formation; and that direct-pathway memory CD4 T cells may differ from naïve cells in their ability to provide B cell help. NK cells kill donor DCs within days of transplantation (30, 56, 57) and to obviate this killing, TCR^−/−.ABM chimeric mice were depleted of NK cells by treatment with anti-NK1.1 antibody prior to and following transplantation with BALB/c × bm12 F1 heart grafts. In addition, to prevent graft-versus-host recognition from triggering recipient humoral immunity (44, 55) donor CD4 T cells were depleted by administering anti-CD4 antibody several days before heart graft removal (Fig. 5A). However, despite effective depletion of recipient NK cells (Fig. 5B), anti-H-2K^d IgG alloantibody was not detected (Fig. 5C).

FIGURE 5. Neither prolongation of direct pathway responses nor memory generation influences humoral alloimmunity.

A, Representative flow cytometry dot plots of peripheral blood from (BALB/c × bm12) F1 at heart graft procurement following CD4 T cell depletion, compared to control, untreated mice. B, Representative flow cytometry dot plot of splenic NK cells (CD49b) in TCR^−/− mice 7 days after treatment with 0.5mg anti-NK1.1 antibody, compared to control untreated TCR^−/− mice. C, Serum levels of anti-H-2K^d IgG alloantibody in TCR^−/−.ABM chimeric mice, depleted of NK cells and challenged with a (BALB/c × bm12) F1 heart transplant (n=3). Values from control, naïve mice (n=3) and d21 sera from B6 recipients of BALB/c hearts (n=4) are shown for comparison. D, Kaplan-Meier survival curves of first (1^st graft, n=6) and second (2^nd graft, 6 weeks after first, n=4) bm12 skin grafts in TCR^−/−.ABM bone marrow chimeric recipients. E, Anti-H-2K^d IgG alloantibody responses to a (BALB/c × bm12) F1 heart allograft in TCR^−/− .ABM chimeric mice (un-primed CD4 T cells, n=3) and TCR^−/−.ABM chimeric mice primed with a bm12 skin graft 6 weeks previously (memory CD4 T cells, n=6). Values from naïve mice (n=7) and d21 sera from B6 recipients of BALB/c hearts (n=4) are shown for comparison.

Memory CD4 T cell responses differ from naïve in costimulation requirements for activation, the degree of proliferation and in the spectrum of cytokines secreted (58-60). To assess whether memory direct-pathway CD4 T cells could provide help for alloantibody production, TCR^−/−.ABM chimeric mice were primed with a bm12 skin graft, 6 weeks before challenge with a BALB/c × bm12 F1 heart graft. In principle, the MHC class II-mismatched skin graft provides target alloantigen for memory ABM CD4 T cell development, but not for anti-MHC class I alloantibody responses; the H-2K^d-allospecfic B cell population remains antigen-inexperienced until second challenge with the heart graft. However, despite the presence of direct-pathway CD4 T cell memory, as evident by accelerated rejection of a second skin graft six weeks later (Fig. 5D), no anti-H-2K^d alloantibody responses were detectable (Fig. 5E).

Discussion

The results of this study address the long-standing controversy regarding the relative roles of the direct and indirect CD4 T cell allorecognition pathways in the alloantibody response to organ transplants. We show that only indirect-pathway CD4 T cells can differentiate into T_FH for production of LLPC from GC alloantibody responses. In contrast, direct-pathway CD4 T cells are unable to provide help for class-switched alloantibody responses, even when CD4 T cell memory has been established by prior immunisation or when the duration of the direct-pathway responses is extended by depletion of recipient NK cells.

The impact of humoral alloimmunity on clinical transplant outcome is increasingly emphasised, and the development of anti-HLA alloantibody following transplantation is now clearly associated with early graft failure (1, 8). In addition, approximately one-third of patients on the waiting-list for a kidney transplant have detectable levels of serum alloantibody that has developed either following pregnancy or from previous transplant or blood transfusion. These ‘sensitised’ patients frequently wait excessively long, and sometimes indefinitely, for a transplant. The mechanisms by which the anti-HLA alloantibody continues to be produced many years after the initiating challenge have received surprisingly little attention, but such durable responses most likely reflect seeding of LLPC to the bone marrow and thus indicate prior GC activity. Germinal centres are a characteristic, albeit not exclusive (61), feature of thymic-dependent responses and the characterisation of the T follicular helper cell as a phenotypically distinct population of CD4 T cells that is uniquely capable of providing help to GC B cells has only become apparent over the last decade (21-23). Restricting help to a highly-select and limited helper population (52) may be a critical feature in developing the necessary competitive environment to ensure preferential selection of high-affinity B cells and thus drive affinity maturation (62). Allospecific T_FH have not been detailed previously, but such is their critical role in production of high-affinity and long-lasting alloantibody that their characterisation may hold important clinical potential, not least because their distinct phenotype may enable development of specific strategies that deplete T_FH in the expectation that graft survival is improved.

Although previous studies have highlighted the ability of indirect-pathway CD4 T cells to provide help for class-switched alloantibody responses (11, 12, 18, 31), this does not necessarily indicate GC activity, because isotype-switching can occur at extrafollicular foci. Thus the differentiation of the transferred TCR75 T cells into a CD44^high CXCR5^high PD1^high signature phenotype, allied to their anatomical location within the follicle and the presence of allospecific LLPC in the bone marrow provide the first definitive evidence of indirect-pathway TFH CD4 T cells. Notably, the T_FH subset did not develop upon transfer into B cell deficient mice, supporting the concept that cognate interaction with the B cell delivers unique signals to helper T cells to guide differentiation and migration to the follicle (54). This may also explain why only a relatively small percentage of the transferred TCR75 T cells acquired T_FH status; our on-going studies indicate that the entire population divides several times after transfer, and thus the majority that did not acquire T_FH phenotype presumably made contact with non-B cell APCs.

Although previous publications, including our own, have concluded that direct-pathway CD4 T cells cannot provide help for class-switched alloantibody responses (18, 31, 63), we felt it important to re-examine this concept, because as discussed in the introduction and as suggested by Fabre’s findings (19, 64), these studies may obscure a role for direct-pathway help either because the B cell compartment lacked MHC class II expression (18) or because non-vascularised skin graft models were studied (31). Of more significance, definitive rebuttal of direct-pathway help for alloantibody production requires specific examination of the memory CD4 T cell subset. A major consideration of direct-pathway help is that, unlike the indirect-pathway, it theoretically enables recognition of the helper T cell epitope to be dissociated from that of the allospecific B cell epitope. For example, direct-pathway CD4 T cells target the allogeneic MHC class II on the surface of donor APCs, whereas the B cell may target a different alloantigen, typically the MHC class I, on the surface of the same cell. Hence it would be possible for direct-pathway CD4 T cells, that have developed memory from prior challenge with target MHC class II alloantigen, to provide help to naïve B cells recognising an additional, but previously un-encountered, MHC class I alloantigen on the graft. Given the resistance of memory cellular responses to costimulation-blockade (58), such memory direct-pathway CD4 T cells could potentially provide effective help when naïve indirect-pathway T cells are inhibited by adjuvant immunosuppression.

It was similarly necessary to exclude NK cell recognition of donor DCs, rather than any inherent defects in the helper function of direct-pathway CD4 T cells, as the reason for the inability of the adoptively-transferred ABM T cells to promote alloantibody production, because although NK cells limit the duration of direct-pathway responses by rapidly killing donor DCs (30, 57), their contribution to graft rejection (65-67), raises the possibility of developing strategies that improve graft survival by depleting NK cells. Thus despite confirming that ABM CD4 T cells can prime humoral autoimmunity through graft-versus-host allorecognition of B cell MHC class II, we have been unable to demonstrate an ability to function as direct-pathway helper cells for alloantibody production, regardless of either the duration or the degree of antigen experience of the direct pathway response.

In summary, we provide conclusive evidence that indirect-pathway CD4 T cell responses deliver help for sophisticated germinal centre responses that result in the production of long-lasting alloantibody, whereas the direct pathway is entirely unable to provide helper function for isotype switching. Targeting the indirect-pathway, for example by administering allopeptide-specific regulatory T cells (13), may thus hold clinical potential for preventing alloantibody-mediated graft damage.

Nonstandard Abbreviations

B6

C57BL/6

BCII^−/−

lack of MHC class II expression restricted to B cells

BCII^+/+

normal MHC class II expression

GC

germinal center

LLPC

long-lived plasma cell

PNA

peanut agglutinin

TCR^−/−.ABM

chimeric mice that contain B cells and a monoclonal population of ABM CD4 T cells

TCR^−/−.bm12

TCR^−/− mice crossed onto a bm12 background

TFH

T follicular helper cell

WT

wild-type