Fas-mediated apoptosis is an important contributor to contraction of Ag-driven T cell responses acting only on activated Ag-specific T cells. The effects of targeted Fas deletion on selected cell populations are well described however little is known regarding the consequences of Fas deletion on only activated Ag-specific T cells. We addressed this question using the parent-into-F1 (P→F1) model of acute or chronic (lupus-like) graft-vs-host disease (GVHD) as a model of either a CTL-mediated or T-dependent B cell-mediated response, respectively. By transferring Fas-deficient lpr donor T cells into Fas-intact F1 hosts, the in vivo role of Ag-specific T cell Fas can be determined. Our results demonstrate a novel dichotomy of Ag-specific T cell Fas function in that: 1) Fas expression on Ag-activated T cells has costimulatory, helper, and down-regulatory roles in vivo and 2) these roles were observed only in a CTL response (acute GVHD) and not in a T-dependent B cell response (chronic GVHD). Specifically, CD4 T cell Fas expression is important for optimal CD4 initial expansion and absolutely required for help for CD8 effector CTL. Donor CD8 T cell Fas expression played an important but not exclusive role in apoptosis and down-regulation. By contrast, CD4 Fas expression played no detectable role in modulating chronic GVHD induction or disease expression. These results demonstrate a novel role for Ag-specific T cell Fas expression in in vivo CTL responses and support a review of the paradigm by which Fas deficiency accelerates lupus in MRL/lpr lupus-prone mice.
Antigen- driven T cell responses are characterized by clonal expansion and effector generation followed by a sharp reduction in T cell numbers and a return to normal lymphocyte homeostasis (1, 2). T cell contraction is mediated primarily by two processes: 1) activation-induced cell death (AICD),3 a Fas-mediated apoptosis of activated T cells undergoing repetitive TCR stimulation and 2) activated T cell autonomous death, a Fas-independent, Bim-mediated apoptosis of activated T cells that does not involve repetitive TCR stimulation (3, 4, 5, 6). It has been postulated that Fas-mediated AICD is primarily involved in limiting autoreactive T cell responses and in maintaining peripheral tolerance, whereas activated T cell autonomous cell death is involved primarily in limiting foreign Ag responses (7, 8, 9, 10, 11). Although Fas-mediated AICD is not an exclusive mechanism in mediating either peripheral tolerance (10, 11) or AICD (12, 13, 14, 15), its importance in lymphocyte homeostasis and peripheral tolerance is underscored by the occurrence of a syndrome of lymphoproliferation and humoral autoimmunity in mice and humans with genetic defects in either Fas or Fas ligand (FasL) (16, 17, 18, 19).
Interestingly, although defects in Fas/FasL in nonautoimmune-prone mice are associated with humoral autoimmunity, mice do not develop lupus (20, 21, 22), a prototypic humoral autoimmune disease characterized by autoantibody production directed against nuclear Ags and damage to vital organs, particularly the kidneys (23, 24). Fas/FasL defects can however accelerate lupus when bred onto the lupus-prone MRL/+ strain (21). Thus, Fas defects do not cause lupus in nonautoimmune-prone mice; however, in the setting of a loss of tolerance (e.g., MRL/+ mice), Fas-mediated apoptosis is an important down-regulatory mechanism that retards lupus. The mechanism by which Fas-mediated AICD retards lupus is thought to involve a combination of impaired Fas-mediated deletion of autoreactive B cells (25, 26, 27, 28) and impaired Fas-mediated AICD of mature autoreactive CD4 Th cells (29) that together result in unabated T cell help to pathogenic autoantibody-producing B cells (30, 31).
Autoreactive Ag-specific CD4 T cells are central to B cell production of pathogenic autoantibodies in lupus (32, 33, 34, 35, 36); however, foreign-reactive Ag-specific T cells may also be involved, particularly during disease initiation and the initial loss of T cell tolerance (37). Regardless of the exact specificity of the T cells initiating disease, the observation that Fas-mediated AICD affects only clonotypically activated T cells and not bystander activated T cells (38, 39, 40) raises an unresolved question as to whether Fas expression by pathogenic CD4 T cells in lupus plays an important role in disease down-regulation. Specifically, does failure of Fas-mediated AICD on Ag-specific CD4 T cells accelerate lupus and, conversely, would promoting Fas-mediated AICD on Ag-specific CD4 T cells be therapeutically beneficial? Previous studies of the role of T cell Fas have examined the in vivo effect of Fas deficiency on all T cells in either autoimmune-prone or nonautoimmune-prone mice (33, 41, 42) or the in vitro effect of Fas deficiency on autoreactive CD4 T cell lines from MRL/lpr mice (29). It has not been technically feasible to address the in vivo role of Fas on only the Ag-specific CD4 T cells driving lupus in spontaneous models of lupus such as the MRL.
This question can however be addressed in an induced model of lupus, the P→F1 model of graft-vs-host disease (GVHD). In this model, the transfer of homozygous parental strain T cells into unirradiated normal F1 mice leads to activation of donor T cells specific for host alloantigens and results in either 1) acute GVHD, characterized by donor anti-host CD8+ CTL that eliminate host lymphocytes (43) or 2) chronic GVHD in which donor CD4+ T cells provide cognate help to all host B cells resulting in autoantibody production and a lupus-like renal disease (44, 45, 46). Thus, both CTL and T-dependent B cell responses can be analyzed using the same donor and host strains. Using this approach, we transferred Fas-defective donor T cells into Fas-intact normal F1 host mice to test the role of Fas expression on Ag-specific T cells in a CTL-mediated (acute GVHD) or T-dependent Ab-mediated (chronic GVHD) in vivo immune response. This transfer also tests the role of repetitive stimulation of peripheral Ag-specific T cells in the development of lupus due to the ubiquitous presence of host alloantigen. Our results indicate that Fas plays a major role in Ag-specific T cell contraction in cell-mediated responses, particularly for CD8 CTL but plays little if any demonstrable role in directly down-regulating CD4 T cells that provide help for B cells or in CD4 T cell-driven lupus-like renal disease. Additionally, we observed an important costimulatory and helper role for CD4-expressed Fas in the generation of CD8 CTL responses but not in the generation of B cell responses. These results support a review of the paradigm by which T cell Fas defects accelerate lupus and indicate a role for defective CD8 CTL control of autoreactive B cells in disease acceleration.
Materials and Methods
Male mice ages 6–8 wk were purchased from The Jackson Laboratory. Wild-type (WT) C57B6/6J (B6) and B6.MRL-Faslpr/J (lpr) mice were used as a source of donor splenocytes and B6D2F1 (BDF1) mice were used as recipients. In some experiments, B6.SJL mice were used as recipients. All animal procedures were preapproved by the Institutional Animal Care and Use Committee at the Uniformed Services University of Health Sciences and at the University of Maryland School of Medicine.
Induction of GVHD
Single-cell suspensions of splenocytes were prepared in RPMI 1640, filtered through sterile nylon mesh, washed, and diluted to a concentration of 108 viable (trypan blue excluding) cell/ml. Unless otherwise noted, acute GVHD was induced using 50 × 106 B6 WT or lpr donor splenocytes into normal BDF1 recipients.
In some experiments, controls consisted of syngeneic transfers, i.e., B6 WT→B6 WT and WT.lpr→B6 WT using CD45.1 B6.SJL hosts and CD45.2 donors.
In some experiments, acute GVHD was induced using purified donor T cells (12 × 106 CD4+T cells and 6–8 × 106 CD8+T cells) in which donor T cells were negatively selected using a Dynal mouse T cell negative isolation kit (Invitrogen) which depletes B cells, NK cells, monocytes/macrophages, dendritic cells (DC), granulocytes, and erythrocytes using a mixture of rat mAbs for mouse CD45R, CD11b, Ter-119, and CD16/32. Depleted donor cells were typically 83–90% T cells. Both depleted and undepleted donor cells were examined by flow cytometry and adjusted so that recipient mice received equal numbers of B6 WT and B6.lpr T cell subsets and, where indicated, cells were also normalized for total number of naive (CD44low) cells. Chronic GVHD was induced by injecting 8–15 × 106 negatively isolated CD4 T cells from either B6 WT or B6.lpr spleens using a Mouse T cell CD4 Subset Column Kit (R&D Systems) according to the manufacturer’s instructions, which removes B cells, CD8 T cells, and monocytes. The purity of the recovered CD3+CD4+ T cells ranged between 87 and 91%. In some experiments, donor CD4+ T cells for chronic GVHD induction were negatively selected using a Dynal mouse CD4 negative isolation kit (Invitrogen) which depletes CD8+ T cells, B cells, NK cells, monocytes/macrophages, DC, granulocytes, and erythrocytes using a mixture of rat mAbs for mouse CD8, CD45R, CD11b, Ter-119, and CD16/32. Cell suspensions were injected i.v. into the tail vein of normal, unirradiated B6D2F1 recipients. Control mice consisted of uninjected age- and sex-matched F1 mice.
Flow cytometry analysis
Spleen cells were incubated with anti-murine FcRIII/II mAb (2.4G2) for 20 min and stained with saturating concentrations of FITC-, PE-, PE-Cy5-, biotin-, Alexa Fluor 488- Alexa Fluor 647-, or allophycocyanin-Alexa Fluor 700-conjugated mAb against either CD4, CD8, B220, CD11c, H-2Kd, I-Ad, CD44, Fas (CD95), FasL (CD178), or PD-1(CD279). All mAb were purchased from Invitrogen (Caltag Laboratories), BD Pharmingen, or eBioscience. Intracellular staining for granzyme B was performed using PE-conjugated anti-human granzyme B mAb according to the manufacturer’s protocol (Invtirogen, Caltag Laboratories). The GB12 clone used here has been reported to detect murine granzyme B (47). Annexin V and anti-human Ki-67 sets were purchased from BD Pharmingen and cells were stained according to the manufacturer’s protocol. Four- or five-color flow cytometry was performed using an LSRII flow cytometer (BD Biosciences). Lymphocytes populations were identified by characteristic forward and side scatter using a broad gate and fluorescence data were collected for 10,000 cells. Studies of donor T cells were performed on 5,000 gated cells that were either CD4+ or CD8+ and did not stain positively for the MHC class I of the uninjected parent. For cell sorting, splenocytes from B6.lpr mice were stained for CD3 and B220 as described above and CD3+B220− cells were purified using a BD Biosciences FACSAria flow cytometer cell sorter. Purity of obtained donor T cells was 99%.
Splenocyte suspensions were prepared as described above. CFSE labeling was performed as previously described (48). Briefly, splenocytes were resuspended at 5 × 107 cells/ml in PBS with 0.1% BSA mixed with 5 μM CFSE (Invitrogen) and incubated for 15 min at 37°C. Unbound CFSE was quenched by the addition of 5 initial volumes of ice-cold RPMI 1640 with 10% FCS, after which cells were washed three times with RPMI 1640 before injection into F1 mice. FACS analysis of cells following CFSE labeling indicated a labeling efficiency 98–99%. The extent of the proliferation was analyzed using the ModFit software (Verity Software House) on days 2 and 4.
In vitro proliferation
Splenic T cells were plated at a concentration of 4 × 105 cells/ml in 0.2 ml and were either unstimulated or received one of three concentrations of anti-TCR (1.0, 2.5, 5.0 μg) and 1.0 μg of anti-CD28 in flat-bottom 96-well Costar plates (Corning). Cells were cultured at 37°C in a 5% CO2-containing incubator and pulsed with [3H]thymidine (Amersham Biosciences) for 18 h before harvesting and detection of [3H]thymidine incorporation with a beta scintillation counter.
Detection of anti-host CTL activity ex vivo
Effector CTL activity of freshly harvested splenocytes was tested in a 4-h 51Cr release assay without an in vitro sensitization period as previously described (49). Splenocytes from control and GVHD mice were tested for their ability to lyse Fas-dull p815 cells (H-2d, MHC class I positive, class II negative) (50). Using serial dilutions, effectors were tested in triplicate at four E:T cell ratios, beginning at 100:1 (2 × 106 effectors and 0.02 × 106 targets/well). Percent lysis was calculated according to the formula: (cpm sample − cpm spontaneous)/(cpm maximum − cpm spontaneous) × 100%. Results are shown as the mean percent lysis ± SEM at a given E:T cell ratio for each group.
In vivo cytokine capture assay (IVCCA)
The IVCCA kit (BD Pharmingen) was used to quantitate in vivo production of IFN-γ as described previously (51) using the manufacturer’s instructions. Briefly, mice were injected i.v. with 10 μg of biotin-labeled neutralizing mAb to IFN-γ (R4-6A2) that binds some of the endogenous IFN-γ forming a soluble cytokine-anti-cytokine Ab complex that then accumulates in vivo for a defined period of time (2–72 h). In these studies, mice were bled 1 day after biotin mAb injection and the level of “in vivo-captured” cytokine present within serum samples was measured using ELISA plates coated with a purified mAb directed against a different IFN-γ epitope (AN-18). The ELISA is developed with HRP-streptavidin and tetramethylbenzidine substrate. For quantitation, a standard curve is generated on each plate by adding serial dilutions of a known amount of recombinant mouse cytokine standard that is complexed with a known amount of biotinylated cytokine-binding Ab.
Total RNA was isolated using RNA-STAT-60 isolation kit (Tel-Test). RNA quantity was estimated by spectrophotometric analysis (DU 640; Beckman Coulter) and RNA samples were reverse transcribed using TaqMan Reverse Transcription Reagents (Applied Biosystems). The RT-PCR was conducted on an Applied Biosystems 7500 Real-Time PCR System using TaqMan Gene Expression Assays as previously described (52) using the TaqMan perforin probe/primer set Mm00812512_m1. To verify that equal amounts of RNA were added in each reverse transcription reaction within an experiment, 18S ribosomal RNA was used as an internal control for equal sampling of total RNA. No cDNA template was used as negative control. After performing the validation experiment to demonstrate that efficiencies of targets and internal references are equal, the comparative 2−ΔΔCt method was used to calculate the relative abundance of a target transcript with regard to an internal control (18S RNA). Results were normalized to each individual rRNA value and were calculated as the fold increase over the respective gene expression in control F1 mice according to the ratio of experimental group 2−ΔΔCt value to F1 2−ΔΔCt value.
Measurement of anti-ssDNA Abs
Serum IgG level of anti-ssDNA was determined by ELISA. Plates were coated with heat-denatured calf thymus DNA (Sigma-Aldrich), blocked with 1% BSA, and incubated with serial dilutions of experimental mouse sera beginning at a dilution 1/40 tested in duplicate. The plates were than incubated with alkaline phosphatase-conjugated anti-mouse IgG (Sigma-Aldrich) and OD was quantitated at 405 nm. For each experiment, murine MRL/lpr sera were used as a standard and results converted to arbitrary units using the TITRI program (http://members. tripod.com/∼gestur/programs/titri.html).
Kidney tissue was fixed in 10% buffered formalin and processed for routine paraffin embedding and histologic sectioning. Three-micrometer-thick sections, stained with H&E and periodic acid-Schiff, were blindly scored by a renal pathologist (M.H.). Tubular injury (epithelial vacuolization, tubular dilatation, cast formation, and/or detachment of epithelial cells) was scored on the basis of the estimated fraction of involved cortical tubules, and interstitial inflammation was scored according to the estimated fraction of involved cortex, both according to the following scale: (0, none; 1, 1–10%; 2, 11–25%; 3, 26–50%; and 4, >50%). The severity of proliferative glomerulonephritis (glomerulonephritis score) was determined according to a previously described semiquantitative scoring system developed for a murine model of lupus nephritis (53).
Statistical analysis was performed using Prism 4.0 (GraphPad Software). For all assays shown, mice are tested individually and group mean and SE were calculated. Student’s t test was used to compare mean values between two groups and ANOVA was used for multiple group comparisons. The log rank test was used to compare survival between experimental groups.
Mortality is delayed in lpr→F1 acute GVHD mice
To determine whether Fas expression on Ag-activated T cells has a functional role, we used the P→F1 model to test the ability of Fas-deficient B6.lpr donor T cells to induce lethal acute GVHD in Fas-intact BDF1 hosts. In this model, GVHD results from donor (H-2b) T cell recognition of host (H-2d) alloantigens. As shown in Fig. 1,A, F1 hosts receiving WT B6 splenocytes (WT→F1) exhibited 100% mortality by day 17, whereas F1 mice receiving B6.lpr cells (lpr→F1) exhibited no mortality by day 28. Some lpr→F1 mice however showed clinical signs of acute GVHD, e.g., hunched posture, ruffled fur (54), and the cohort was sacrificed and spleens were analyzed by flow cytometry. Compared with normal uninjected F1 mice, lpr→F1 mice exhibited significant reductions in host B cells (∼99% reduction) (Fig. 1,B), DC (∼74% reduction; Fig. 1,B), and host CD4 T cells (∼93% reduction) and CD8 T cells (∼75% reduction; Fig. 1 C). These data demonstrate that despite the delay in mortality, lpr→F1 mice eventually achieve an acute GVHD phenotype by flow cytometric analysis.
Elimination of host cells at 2 wk is impaired in lpr→F1 mice
Mortality in WT→F1 acute GVHD mice is mediated primarily by an attack on the host immune and hematopoetic systems by donor CD8 CTL (55, 56, 57). It is possible that the delayed mortality in lpr→F1 mice is a consequence of delayed or impaired donor anti-host CTL development. Flow cytometric parameters at 2 wk after donor cell transfer can serve as early surrogate markers for longer-term clinical GVHD outcomes such as mortality in acute GVHD and lupus-like renal disease in chronic GVHD (reviewed in Ref. 58). Two-week surrogate markers in acute GVHD mice consist of: 1) engraftment of both CD4 and CD8 donor T cell subsets; 2) elimination of host cells, particularly B cells, that is mediated by 3) donor anti-host CD8 CTL, best seen at 10 days after transfer (49, 55, 59). To determine whether delayed mortality in lpr→F1 mice is preceded by an altered 2-wk GVHD phenotype, F1 mice were assessed at 2 wk after the transfer of 50 × 106 unfractionated donor splenocytes as in Fig. 1. To ensure that the numbers of CD4 and CD8 T cells in the lpr donor cell inoculum were more than or equal to that of the WT donor inoculum, donor cells were normalized for the number of CD4 T cells based on flow cytometry analysis. The donor inoculum of 50 × 106 WT donor cells used in Fig. 2, A and B, resulted in the transfer of 9.2 × 106 CD4 cells and 5.6 × 106 CD8 T cells. Accordingly, lpr→F1 mice received 57 × 106 lpr splenocytes containing 9.2 × 106 CD4 cells and 7.4 × 106 CD8 T cells.
At 2 wk after transfer, WT→F1 acute GVHD exhibited: 1) engraftment of both CD4 and CD8 donor T cell subsets (Fig. 2,A) and 2) significant elimination (54% reduction) of host CD4 T cells (Fig. 2,A) and near complete elimination (∼94% reduction) of host B cells (Fig. 2 B) compared with uninjected normal F1 mice. In this experiment, WT→F1 mice exhibited no significant elimination of host CD8 T cells and host DC were significantly increased (3.8-fold) over control F1 levels.
Compared with WT→F1 mice, the transfer of an equal number of lpr CD4 T cells and a slightly greater number of lpr CD8 T cells resulted in significantly greater engraftment of both lpr donor T cell subsets at 2 wk (CD4 = ∼2.8-fold; CD8 = ∼3.6-fold; lpr→F1 vs WT→F1; Fig. 2,A). Despite the greater engraftment of donor CD8+ T cells, lpr→F1 mice exhibited impaired elimination (i.e., greater survival) of all host cell populations compared with that seen in WT→F1 mice. Compared with control uninjected F1 mice, host CD4 T cells were only reduced by 18% (Fig. 2,A) and host B cells by 58% (Fig. 2,B) in lpr→F1 mice. Interestingly, there was an increase in host CD8 T cells (3.3-fold) and total DC (10.7-fold) in lpr→F1 mice compared with uninjected normal F1 mice. The greater numbers of host T cells, B cells, and DC at 2 wk in lpr→F1 mice compared with WT→F1 mice is consistent with impaired host elimination due to defective donor anti-host CTL effector function. This percentage of cells demonstrating up-regulation of the activation marker CD44 compared with the respective naive donor was >2-fold for WT donor CD4 and 1.8-fold for WT donor CD8 cells, whereas up-regulation was comparable to control (1.1-fold) for lpr donor CD4 cells and 1.5-fold elevated for lpr donor CD8 cells (Fig. 2,C). As reported previously (60), baseline CD44 expression is higher on lpr T cells. Host T cell up-regulation of CD44 (Fig. 2 D) was strong for both GVHD groups with lpr→F1 host CD8 T cells exhibiting significantly greater CD44 up-regulation vs WT→F1 host CD8 T cells. The reduced up-regulation of CD44 on lpr donor CD4 T cells raises the possibility that activation of lpr donor CD4 T cells is impaired in lpr→F1 GVHD mice. Alternatively, it is possible that lpr donor cells (either T cells and/or non-T cells) induce host cell expansion and lymphoproliferation as part of the lpr phenotype as previously described (41) independent of their effects on GVHD.
To exclude a role for lpr non-T cells in the expansion of host DC and CD8 T cells seen in lpr→F1 mice (Fig. 2, A and B), acute GVHD was induced using relatively purified (83–90%), negatively isolated splenic T cells. The donor cell inocula was normalized for CD4 T cells and a stronger GVHD was induced by transferring greater numbers of donor CD4 T cells than used in Fig. 2, A and B. In Fig. 2, E and F, F1 mice received 12 × 106 WT or lpr CD4 cells in conjunction with either 7.7 × 106 WT CD8 or 9 × 106 lpr CD8 T cells, respectively. Engraftment of both donor subsets was observed in WT→F1 mice (Fig. 2,E) and associated with near complete elimination (∼94% reduction) of host B cells (Fig. 2,F), comparable to host B cell elimination using unfractionated WT donor cells (Fig. 2,B). Additionally, enriched WT donor T cells induced greater elimination of host CD4 T cells (∼90% reduction) and CD8 T cells (∼62% reduction; Fig. 2,E) than seen using unfractionated donor cells (Fig. 2,A). Expansion of host DC seen in unfractionated WT→F1 mice (Fig. 2,B) was not seen with purified donor cells (Fig. 2,F); however, DC were not reduced below control either. Thus, transferring greater numbers of WT donor cells increases the strength of donor antihost attack resulting in greater elimination of host splenocytes in WT→F1 mice. Comparing Fig. 2, A and B, to E and F, a hierarchy of host cell elimination can be seen in WT→F1 mice such that host B cells are the most susceptible and exhibit near complete elimination at both donor cell doses, whereas host CD4 cells, CD8 cells, and DC are progressively less susceptible and approach complete elimination only at the higher donor cell dose.
The potentiation of acute GVHD seen with greater numbers of purified WT donor cells was not seen for purified lpr donor T cells. Despite transferring an equivalent number of CD4 T cells and slightly greater number of CD8 T cells than used in WT→F1 mice, lpr→F1 mice exhibit the same general phenotype in Fig. 2, E and F, as seen in A and B, i.e., significantly greater engraftment of donor CD4 (∼2-fold) and CD8 T cells (∼3.2-fold) in conjunction with impaired elimination of host cells compared with WT→F1 mice. Although lpr→F1 mice exhibit elimination of host CD4 (42% reduction) (Fig. 2,E) and host B cells (55% reduction; Fig. 2,E), the reductions are significantly less severe than those seen in WT→F1 mice. As in Fig. 2, A and B, lpr→F1 mice exhibited significant expansion of host CD8 T cells (∼1.8-fold > control F1; Fig. 2,E) and host DC (∼4.7-fold > control F1; Fig. 2 F).
To address the possibility that the expansion of host DC and CD8 T cells in lpr→F1 mice in Fig. 2, E and F, is mediated by the few contaminating non-T cells transferred in the lpr donor inoculum, GVHD was induced using highly purified (>99%) donor T cells obtained by cell sorting. Achieving this level of purity without compromising cell viability due to prolonged sorting permitted a total yield of 8 × 106 CD4 and 3.5 × 106 CD8 lpr T cells which were transferred into a single recipient F1 (Fig. 3,A). The lpr donor cell numbers used are less than those in Fig. 2, A–F, consequently, the strength of the antihost attack was reduced as shown by: 1) reduced donor CD8 engraftment (2.2 × 106) compared with that in Fig. 2,A (3.6 × 106) or Fig. 2,E (4.1 × 106) and 2) impaired elimination of host B cells resulting in no reduction below that of control F1 at day 14 (Fig. 3,A). Despite the weakened GVHD, host CD8 T cells exhibited a 2-fold expansion and host DC exhibited a 9-fold expansion (Fig. 3,A) over control uninjected F1 values similar to those seen in Fig. 2, A and B. Thus, the expansion of host CD8 T cells and DC in lpr→F1 mice seen in Fig. 2, A and B, can be mitigated by transferring more donor T cells (Fig. 2, E and F) or exacerbated by transferring fewer, yet purer (sorted) donor T cells (Fig. 3,A). The greater host DC expansion seen with sorted cells does not support a role for contaminating lpr non-T cells in mediating host T cell and DC expansion in lpr→F1 mice but rather is consistent with an impaired antihost CTL response as indicated by the absence of any host B cell reduction (Fig. 3 A).
To exclude the possibility that donor T cells rather than non-T cells promote host expansion due to nonspecific, Ag-independent proliferation, we transferred WT or lpr (both CD45.2) donor cells into syngeneic B6.SJL (CD45.1) hosts. Both GVHD groups exhibited a comparable low-level expansion of host T (Fig. 3,B) and B cells (Fig. 3,C). Moreover, compared with WT→B6 mice, lpr→B6 mice exhibited no significant increase in donor CD8 engraftment (Fig. 3,B) and transferred donor B cells are less than that of WT→B6 mice (Fig. 3,C). lpr→B6 mice exhibited a small but significant increase in donor CD4 engraftment and in donor and host DC compared with WT→B6 mice; however, the magnitude is low-level (≤1.6-fold over control F1) and insufficient to account for the striking increase in host DC seen in Fig. 2,B (10.7-fold) and Fig. 2,F (4.7-fold) over control. Importantly, the striking increase in lpr CD8 T cell engraftment seen in Fig. 2, A and E, was not observed in the lpr syngeneic transfer. Additionally, up-regulation of CD44 in either WT→B6 or lpr→B6 was minimal compared with the respective uninjected control for either host (Fig. 3,E) or donor (Fig. 3,F) CD4 or CD8 T cells and were well below the values seen at day 14 for WT→F1 allogeneic transfers (Fig. 2, C and D). These results indicate that the transfer of WT or lpr T cells into syngeneic hosts results in a small and roughly equivalent nonspecific increase in host cells unaccompanied by evidence of T cell activation (i.e., CD44 up-regulation). Low-level homeostatic proliferation is seen for transferred donor T cells, B cells, and DC and is roughly equivalent.
Lastly, to conclusively address the role of non-T cells in mediating the expansion of host cells in lpr→F1 mice at day 14, we transferred T-depleted WT or lpr donor T cells into WT BDF1 mice and assessed disease phenotype at day 14. As shown in Fig. 4,A, donor CD4 and CD8 engraftment for both lpr→F1 and WT→F1 was not significantly above background values for uninjected F1 mice, confirming the adequacy of donor T cell depletion. Additionally, there were no significant differences between lpr→F1 vs WT→F1 mice for numbers of 1) host CD4 and CD8 T cells (Fig. 4,A), 2) donor or host B cells (Fig. 4,B), or donor or host DC (Fig. 4,C), nor were these values significantly greater than uninjected F1 mice. Lastly, there was no significant increase in the percentage of activated (CD44high) host T cells in either WT→F1 or lpr→F1 mice (Fig. 4,D), indicating that there was no detectable host T cell recognition of transferred donor cells. These results not only confirm the critical role of donor T cells in GVHD pathogenesis (61), but also do not support a role for non-T cells in mediating host lymphocyte expansion at day 14 in lpr→F1 mice seen in Fig. 2.
Taken together, the results of Figs. 2–4 do not support the idea that the expansion of WT host T cells, B cells, and DC seen in lpr→F1 mice at day 14 is a consequence of transferring abnormally or nonspecifically proliferating lpr T cells or non-T cells but instead support a role for impaired elimination by donor antihost CTL.
Fas-deficient donor T cells exhibit delayed expansion and impaired down-regulation
Defective elimination of host cells in lpr→F1 mice (Fig. 2) supports the possibility that donor T cell Fas expression is important in optimal CTL development. To address this idea, we performed a kinetic analysis of donor and host splenocyte subpopulations during the first 2 wk of GVHD. As shown previously (52) and confirmed in Fig. 5, the first 14 days after transfer of WT donor cells into WT F1 mice can be divided into three phases for donor T cells and two phases for host cells. For donor T cells (Fig. 5, A and B), the three phases are: 1) activation and expansion of naive T cells (days 0–7), 2) expansion of mature effector CTL (days 7–10), and 3) down-regulation of effector T cells (days 10–14). For host B cells (Fig. 5,C), the two phases are activation and expansion (days 0–4), followed by elimination by donor antihost CD8 CTL (days 4–14). Host DC (Fig. 4 D) exhibit a similar expansion and elimination phase; however, the kinetics differ slightly from those of host B cells with peak expansion occurring at approximately day 10 followed by the elimination phase from days 10 to 14. Importantly, expansion of host cells is a normal initial feature in WT→F1 acute GVHD mice and is driven by donor B6 (H-2b) CD4 recognition of host allogeneic I-ad on MHC class II (MHC II)-positive F1 host cells (45, 62).
Interestingly, host T cells also exhibit an expansion and elimination phase; however, because mouse T cells do not express MHC II (63), this expansion cannot be explained by donor CD4 recognition of host allogeneic MHC II and although incompletely understood likely represents the host antidonor CD8 CTL response described by Kosmatopolous et al. (64). As shown for WT→F1 mice (Fig. 5, E and F), expansion of host CD4 T cells is seen by day 4, whereas expansion of host CD8 T cells is delayed until day 10 and mirrors the expansion phase of donor effector CD8 CTL. Expansion of host CD4 T cells does not exhibit the striking peak at days 7–10 seen for host CD8 T cells. Despite the difference in expansion kinetics, both host T cell subpopulations decline in WT→F1 mice from days 10 to 14 similar to that seen for host DC, consistent with their elimination by donor antihost CTL that target all MHC class I-bearing host cells.
The kinetics of lpr→F1 mice differ from those of WT→F1 mice in several ways. First, peak day 10 expansion of lpr CD4 T cells is reduced compared with WT CD4 T cells and is preceded by a reduced expansion of lpr CD4 T cells during the initial activation phase (Fig. 5,A, days 4 through 7), consistent with impaired activation of naive lpr CD4 T cells. Second, peak expansion for lpr CD8 T cells, while not reduced compared with WT CD4 T is delayed from days 10 to 14 (Fig. 5,B). Lastly, down-regulation of donor effector lpr T cells from days 10 to 14 is impaired, particularly for CD8+ T cells, resulting in significantly greater numbers of lpr donor T cells at day 14 as shown in Fig. 2, A and C. The delay in expansion of donor effector CD8 T cells in lpr→F1 mice (Fig. 5,B) is associated with a delay in host B cell elimination seen as a shift in the kinetics of B cell elimination curve to the right for lpr→F1 mice compared with WT→F1 mice (Fig. 5,C). We have previously demonstrated that host B cell elimination in acute GVHD is a more sensitive marker for in vivo antihost CTL activity than is analysis by an ex vivo 51Cr release assay (51). Thus, the delay in host B cell elimination in lpr→F1 mice strongly suggests that effector CTL maturation is impaired. Impaired/delayed antihost CTL effectors in lpr→F1 mice could also account for the complete lack of elimination of host DC (Fig. 5,D) and host CD8 T cells (Fig. 5,F) and significantly impaired elimination of host CD4 T cells (Fig. 5,E) as these targets do not exhibit elimination in WT→F1 mice until well after B cell elimination has begun (Fig. 5, C–F).
Taken together, these kinetic results demonstrate that the expansion phase of host cells is equivalent for lpr→F1 vs WT→F1 mice. Specifically, peak expansion of host B cells (day 4), host CD4, CD8, and DC (day 10) did not differ significantly between the two groups. However, the ensuing elimination phase of host cells is clearly delayed in lpr→F1 mice compared with WT→F1 mice. Thus, the expansion of host cells seen at day 14 in Fig. 2, A and B, for lpr→F1 mice does not reflect greater initial expansion of host cells but rather an impaired/delayed elimination phase. Although all MHC class I-bearing host splenocytes are targeted by donor CD8 CTL and undergo significant elimination by day 14 in WT→F1 mice, Fig. 5 confirms the hierarchy of host cell elimination seen in Fig. 2 in that elimination of host B cells has begun by day 7, whereas elimination of host DC and host T cells has not begun until day 10. The delay in the elimination phase for lpr→F1 mice is clearly seen for host B cells (Fig. 5,C), but has not begun by day 14 for host DC and T cells. Host DC and T cells are eventually eliminated as shown in Fig. 1, B and C.
Donor CD8 effector maturation and IFN-γ-dependent Fas/FasL up-regulation is delayed in lpr→F1 mice
Perforin and Fas/FasL pathways are the two major pathways by which donor CD8 CTL eliminate host cells in acute GVHD (49). IFN-γ is absolutely required for both Fas/FasL up-regulation in acute GVHD (49, 51). We have previously demonstrated that peak serum IFN-γ occurs on or about day 7 in acute GVHD mice and is an early marker of CD8 CTL effector maturation preceding maximal CD8 CTL expansion by ∼3 days (65). As shown in Fig. 6,A, peak IFN-γ is seen at day 8 and is equivalent for both lpr→F1 and WT→F1; however, substantial IFN-γ is produced earlier at day 7 in WT→F1 mice vs lpr→F1, suggesting that donor CD8 effector maturation in lpr→F1 mice lags behind that of WT→F1 mice. This one-day lag in IFN-γ production for lpr→F1 mice is functionally significant as shown by a delay in the following IFN-γ-dependent parameters in lpr→F1 mice: a) peak numbers of donor CD8+ T cells that have up-regulated FasL (Fig. 6,B) and peak numbers of host B cells that have up-regulated Fas (Fig. 6 C). Thus, the delay in peak IFN-γ production in lpr→F1 mice is associated with delayed Fas/FasL-mediated host cell elimination as shown by both delayed FasL effector pathway maturation in donor CD8 CTL and a delayed host Fas up-regulation.
Perforin-mediated donor CD8+ CTL effector function is defective in lpr→F1 mice
IFN-γ has an important but not exclusive role in donor CD8 CTL perforin/granzyme pathway maturation (51). As shown in Fig. 6,D for WT→F1 GVHD mice, peak numbers of granzyme B-positive donor CD8 T cells occur at day 10 and are significantly greater than those seen in lpr→F1 mice in which a similar peak occurs at day 14. Moreover, at day 10 the time of peak ex vivo antihost CTL activity (59), WT→F1 mice exhibit significantly greater (2- to 3-fold) splenic perforin gene expression (Fig. 6,E) and greater perforin pathway killing activity (Fig. 6,F) vs lpr→F1 mice. The integrity of perforin-mediated donor antihost CTL killing was determined by testing ex vivo cytotoxicity as previously described (59) using Fas-negative targets of host strain. As shown in Fig. 6,F, ex vivo killing by splenocytes from lpr→F1 mice is reduced at least 8-fold compared with WT→F1 mice as estimated by a comparison of the titration curves. These results indicate that in addition to a delay in Fas/FasL pathway maturation (Fig. 6, A–C), lpr→F1 mice also exhibit impaired perforin pathway maturation and function compared with WT→F1 mice.
Costimulatory role of Fas on donor T cells in acute GVHD
It is not clear whether the delay in CD8 CTL effector maturation (Figs. 5 and 6) for lpr→F1 mice reflects a primary defect in lpr CD8 T cells, a secondary defect in lpr CD4 T cell help, or a combination of both. To address this question, the kinetics of donor T cell activation were assessed using CFSE-loaded donor T cells. Controls consisted of syngeneic transfers, i.e., B6 WT→B6 WT and WT.lpr→B6 WT using B6 (CD45.2) donors and B6.SJL (CD45.1) hosts. Donor CD4 and CD8 populations were identified as H-2d negative for GVHD mice (Fig. 7, A and B, respectively) and as CD45.1 negative in syngeneic transfers (Fig. 7, C and D, respectively). No significant differences in donor CD4+ or CD8+ T cell activation were observed at day 2 after transfer (data not shown); however, by day 3, defective activation of lpr donor T cells was observed. Representative day 3 tracings are shown for WT→F1 donor CD4 (Fig. 7,E) and CD8 (Fig. 7,F) and for syngeneic WT→B6 donor CD4 (Fig. 7,G) and CD8 (Fig. 7,H), with the percentage of T cells undergoing at least one division shown to the left and the percentage of undivided cells to the right. The respective tracings for lpr donor cells are shown in Fig. 7, I–L.
A comparison of single tracings from B6 WT donors (Fig. 7, E and H) to B6.lpr donors (Fig. 7, I and L) at day 3 demonstrates that the number of cells undergoing at least one division is reduced in lpr→F1 vs WT→F1 GVHD mice, particularly for CD8 T cells. Donor cells from syngeneic transfers exhibited a small amount of equivalent homeostatic proliferation. The individual tracings are corroborated by group mean values (Fig. 7,M), demonstrating that lpr→F1 mice exhibit a significant increase in the number of undivided donor cells for both CD4 and CD8 T cells compared with WT→F1 mice. This defect in lpr donor CD4 and CD8 T cell division was also seen at day 4 (Fig. 7,N). Donor T cell division was not significantly different for WT→F1 vs lpr→F1 mice in the syngeneic transfers at either day (Fig. 7, M and N). Defective lpr T cell proliferation was also seen in vitro (Fig. 7,O). Stimulation of unfractionated splenocytes with anti-TCR and anti-CD28 mAb resulted in significantly less proliferation at days 2 and 3 for lpr splenocytes compared with WT. The results of Fig. 7 support the conclusion that donor T cell Fas expression has a costimulatory role and is required for optimal activation and expansion in vivo of naive T cells. The defect in lpr donor T cell activation is functionally significant and results in significantly reduced expansion of donor CD4 T cells at days 4 and 7 and CD8 T cells at day 7 (Fig. 5).
Critical role of Fas on CD4 T cells in providing help for CD8 CTL
Our demonstration that initial activation of both CD4 and CD8 T cells is impaired in lpr→F1 mice supports the idea that the ensuing delay in donor CD8 CTL maturation is secondary to defective donor CD4 T cell help. These results do not exclude an additional intrinsic defect in lpr donor CD8 T cells. To address this question, we compared the ability of B6 WT and lpr CD4 T cells to provide help for either B6 WT or lpr CD8 T cells and induce an acute GVHD phenotype at day 14 in normal F1 mice. Donor T cell subsets were first negatively isolated and then recombined before injection into F1 hosts. All four possible recombinations were tested and mice assessed at day 14 after transfer. As shown previously (51) and also in Fig. 5,C, host B cell elimination is a sensitive correlate of in vivo antihost CTL activity in this model and by day 14, host B cells exhibited near complete elimination in B6 WT→F1 mice. Comparing host B cell elimination (Fig. 8,A) and donor T cell engraftment (Fig. 8,B) for the four groups, two important trends were observed. First, isolating and recombining T cell subsets does not alter the day 14 GVHD phenotypes seen in Fig. 2. Specifically, WT→F1 (i.e., WT CD4 + WT CD8→F1) mice exhibited near complete elimination of host B cells, whereas lpr→F1 (i.e., lpr CD4 + lpr CD8→F1) mice exhibited significantly impaired B cell elimination compared with WT→F1 (Fig. 8,A). Similarly, engraftment of lpr CD4 and CD8 T cells is significantly greater than that for WT→F1 (Fig. 8,B) and also seen in Fig. 2, A and E. These results demonstrate that isolation of individual donor T cell subsets and recombination before injection does not alter the day 14 results seen using either unfractionated (Fig. 2, A and B) or copurified (Fig. 2, E and F) donor splenocytes.
Second, lpr CD4 T cells are defective in their ability to provide help for donor CD8 T cell maturation into CTL. This is seen in Fig. 8,A in which defective host B cell elimination seen in lpr CD4 + lpr CD8→F1 mice is corrected by pairing WT CD4 with lpr CD8 T cells, i.e., host B cells exhibit near complete elimination (compare B cell numbers for lpr CD4 + lpr CD8→F1 mice to that of WT CD4 + lpr CD8→F1). Thus, the impaired elimination of host B cells by CD8 CTL seen in lpr→F1 mice (Figs. 2, B and F, 5,C, and 8 A) does not result from an intrinsic defect in lpr CD8 effector function but rather from defective lpr CD4 help as the defect can be corrected by WT CD4 help. The defect in lpr CD4 Th cell function is confirmed when lpr CD4 T cells are paired with WT CD8 T cells. The near complete B cell elimination seen for WT→F1 mice (WT CD4 + WT CD8→F1) is entirely abrogated by the pairing of lpr CD4 T cells for WT CD8. Not only are host B cells not reduced but they are significantly increased over normal uninjected F1 mice (compare B cell numbers for lpr CD4 + WT CD8→F1 to uninjected F1) consistent with a complete abrogation of WT CD8 CTL effector maturation due to defective lpr CD4 help, resulting in expansion of host B cells and a chronic GVHD phenotype.
Fas on CD8 T cells is important in effector down-regulation
If lpr CD4 T cells are defective in their ability to provide help for CD8 CTL effector maturation, why then is host B cell elimination more impaired (i.e., greater numbers survive) using WT CD8 than with lpr CD8 donor T cells? Based on the greater numbers of donor lpr CD8 T cells seen at day 14 in Fig. 2, it is possible that impaired Fas-mediated down-regulation of lpr CD8 T cells compensates for defective CD4 help by reducing the down-regulation phase of acute GVHD, thereby allowing continued elimination of host B cells. By contrast, WT CD8 T cells undergo substantial down-regulation from days 10 to 14. This idea is supported by the data in Fig. 8,B in which engrafted donor lpr CD8 T cell numbers are significantly greater (>3-fold) than engrafted donor WT CD8 T cells regardless of the source of donor CD4 help (compare donor CD8 T cell numbers for lpr CD4 + lpr CD8→F1 or WT CD4 + lpr CD8→F1 to either WT CD4 + WT CD8→F1 or lpr CD4 + WT CD8→F1). Moreover, defective lpr CD8 CTL down-regulation appears to compensate for defective lpr CD4 T cell help by permitting continued CTL host B cell elimination (i.e., after day 10; Fig. 5,C) when WT CD8 normally down-regulate (Fig. 5,B), resulting in significant albeit incomplete host B cell elimination at day 14. Host B cell elimination is not seen when lpr CD4 T cells are paired with WT CD8 T cells (Fig. 8 A, compare lpr CD4 + lpr CD8→F1 to lpr CD4+ WT CD8→F1) and B cells are significantly increased over control consistent with a complete abrogation of donor CD8 CTL maturation and a chronic GVHD phenotype.
The numbers of host CD4 and CD8 T cells remaining at day 14 (Fig. 8,C) confirm the two critical roles of Fas discussed above, i.e., a helper role on donor CD4 T cells and a down-regulatory role on donor CD8 T cells. As shown in Fig. 5, E and F, expansion of host T cells from days 0 to 10 is a characteristic of WT→F1 acute GVHD and is abrogated at day 10 by the strong donor antihost cytolytic response resulting in elimination of host CD4 and CD8 T cells by day 14. This can also be seen in Fig. 8,C as a significant reduction of host CD4 and CD8 T cells below uninjected F1 controls for WT CD4+ WT CD8→F1. By contrast, host T cell elimination is impaired for lpr CD4+ lpr CD8→F1 mice, especially for host CD8 T cells which are significantly greater than control F1 levels. These results confirm the day 14 results seen in Fig. 2, A and E. Pairing of WT CD4 with lpr CD8 corrects antihost CD8 effector function seen for lpr CD4 + lpr CD8→F1 mice (Fig. 8,A) and induces complete elimination of host B cells (Fig. 8,A) and significantly greater elimination of host CD4 T cells than that seen for WT→F1 mice (Fig. 8,C). Because host B cell elimination is nearly complete in WT→F1 mice (Fig. 8,A), no further increase in this parameter can be measured in WT CD4 + lpr CD8→F1 mice; however, due to hierarchical differences in host cell elimination, a further significant reduction in host CD4 T cells (compared with WT→F1 mice) is now seen for WT CD4 + lpr CD8→F1 mice. Thus, lpr donor CD8 T cells are more potent than WT CD8 T cells when provided normal CD4 T cell help, likely due in part to their greater numbers at day 14 (Fig. 8 B, compare WT CD4 + lpr CD8→F1 vs WT CD4 + WT CD8→F1),which in turn may reflect impaired Fas-mediated AICD compared with WT CD8 T cells.
Of note, host CD8 T cells are significantly elevated in lpr CD4 + lpr CD8→F1, WT CD4 + lpr CD8→F1, and lpr CD4 + WT CD8→F1 mice; however, the explanations for the first two are different from the latter. As shown in Fig. 5,F, expanded host CD8 T cells for lpr→F1 mice at day 14 do not represent abnormal expansion but rather a failure of elimination by delayed donor antihost CD8 CTL (Figs. 5,B and 6 A–F). Similarly, expanded host CD8 T cells in lpr CD4 + WT CD8→F1 mice occurs in conjunction with expanded host B cells and is thus part of a chronic GVHD phenotype, indicating a complete failure of antihost CD8 CTL to eliminate expanded host cells. In contrast, WT CD4 + lpr CD8→F1 mice exhibit expanded host CD8 T cells in the setting of substantial elimination of other host cell populations, thus precluding defective antihost CTL function as an explanation in this combination. More likely, the expanded host CD8 T cells represent a host-vs-graft down-regulatory response as previously reported (64, 66, 67, 68).
Peak (day 7) serum IFN-γ results (Fig. 8,D) corroborate the defect in lpr CD4 Th cell function. As shown previously (51, 65), peak IFN-γproduction correlates with the subsequent strength of the donor CD8 CTL response as measured by peak donor CD8 engraftment at day 10. Defective B cell elimination seen for lpr→F1 mice at day 14 (Fig. 8,A) is accompanied by significantly reduced peak IFN-γ levels (Fig. 8,D); however, substituting WT CD4 for lpr CD4 T cells (WT CD4 + lpr CD8→F1) corrects both the antihost CTL response (i.e., host B cells are completely eliminated at day 14) and the peak IFN-γresponse. Thus, lpr CD8+ T cells are capable of a normal IFN-γ response (Fig. 8,C) and normal CTL effector function, i.e., host B cell elimination (Fig. 8,A) when paired with a normal source of CD4+ T cell help (i.e., WT CD4 + lpr CD8→F1). Conversely, the defective peak IFN-γ seen in lpr CD4 + lpr CD8→F1 mice is worsened when lpr CD4 T cells are paired with WT CD8 T cells such that IFN-γ levels for lpr CD4 + WT CD8→F1 mice are not only significantly reduced vs lpr CD4 + lpr CD8→F1, but they also do not differ significantly from uninjected F1 mice. Abrogation of the IFN-γ response is associated in turn with a complete absence of host B cell elimination at day 14 (Fig. 8 A), consistent with a failure of donor antihost CTL effector maturation, all of which are observed in the chronic GVHD phenotype (51).
Lastly, host DC numbers at day 14 corroborated the foregoing observations and exhibited a pattern similar to that seen for host B cells and host CD4 T cells; however, the differences did not reach statistical significance due to the wide SE (data not shown).
Fas-defective donor CD8 T cells exhibit prolonged proliferation despite undergoing Fas-independent apoptosis
The greater numbers of lpr cells vs WT donor T cells at day 14 (Figs. 2, A and E, and 5, A and B) indicate that Fas plays an important role in down-regulation of activated donor T cells, especially CD8 T cells. Additionally, as shown in Fig. 8, Fas-defective lpr CD8 T cells mature into effector CTL and eliminate host B cells at levels of CD4 help that are insufficient for WT CD8 CTL maturation, suggesting that defective down-regulation permits persistent activation and killing by lpr CD8 T cells. To confirm these indirect observations, apoptosis and proliferation were compared for WT and lpr donor CD8 T cells. As shown in Fig. 9, A and B, lpr CD8 T cells exhibited significantly reduced apoptosis at day 10 as measured by the percentage of total annexin V-expressing cells compared with WT donor CD8 T cells; however, this difference was lost by day 14. Moreover, there was no significant difference at day 10 in the numbers of annexin V-expressing donor CD8 T cells; however, by day 14, lpr donor CD8 T cells were significantly increased compared with WT. These results indicate that Fas-independent mechanisms induce apoptosis in lpr donor CD8 T cells, an idea that is supported by the data in Fig. 9, C and D, demonstrating that up-regulation of PD-1, a molecule important in T cell down-regulation (69), is not defective in lpr CD8 T cells as shown by comparable or increased PD-1 up-regulation in lpr CD8 T cells measured as either the percentage (Fig. 9,C) or total numbers (Fig. 9,D) at either day 10 or 14. Additional evidence of defective lpr CD8 down-regulation is provided by studies of proliferation (Fig. 9,E), demonstrating that lpr CD8 T cells exhibit a significant increase in the number of proliferating (Ki-67-positive) donor CD8 T cells at day 14, a time when WT donor CD8 T cells have undergone significant down-regulation (Fig. 5,B). Increased numbers of proliferating lpr donor CD8 T cells at day 14 is associated with a reciprocal increase in the numbers of proliferating host (WT) CD8 T cells in lpr→F1 mice compared with WT→F1 mice (Fig. 9,F). As discussed above for Fig. 8 C, this increase cannot be explained solely by failure of CD8 CTL elimination and likely represents the well-described role of perforin and CD8 T cells in down- regulating CTL responses (66, 67) as part of a host antidonor response (64, 68).
Fas-defective donor CD4 T cells do not exhibit excessive help for host B cells
Fas-defective, MRL/lpr mice exhibit an acceleration of lupus-like disease compared with Fas-intact MRL/+ mice and defects in both T cells and B cells have been implicated in this acceleration (33, 34, 70). It has been postulated that failure of Fas-defective CD4 T cells to undergo AICD could result in prolonged help for autoreactive B cells (31). Conversely, our demonstration that lpr CD4 T cells exhibit defective help for CTL responses could indicate that the helper function of lpr CD4 T cells is globally defective and indicating that help for B cells is also impaired. To assess the role of Fas expression by Ag-specific CD4 T cells in both B cell help and lupus-like disease severity, we compared lupus-like disease in the P→F1 model by transferring equal numbers of either WT or lpr-purified naive (CD44low) donor CD4 T cells into WT recipients. If defective Fas-mediated down-regulation of Ag-activated CD4 T cells prolongs help for B cells, lpr→F1 mice should exhibit worsening of disease parameters. We therefore used relatively low numbers of donor CD4 T cells (12 × 106) to induce a mild disease and to better show disease worsening by lpr CD4 T cells. As shown in Fig. 10, no differences were observed in host B cell expansion (Fig. 10,A), a marker of CD4 help for B cells (45) or in donor CD4+ T cell engraftment (Fig. 10 B) at 2 wk, indicating that, in contrast to acute GVHD, initial CD4 help for B cells is not defective. Serum IFN-γ values at day 7 after donor cell transfer for WT→F1 (747 ± 85 pg/ml) and lpr→F1 (890 ± 59 pg/ml) were not significantly different and neither value was significantly greater than uninjected normal F1 (643 ± 92 pg/ml). These IFN-γ results are in agreement with previous work (51) and demonstrate that in contrast to the high- serum IFN-γ seen in acute GVHD, chronic GVHD is not necessarily characterized by striking IFN-γ elevations.
Long-term studies were then performed to determine whether lupus-like renal disease was more severe in F1 recipients receiving lpr donor CD4 T cells. Two cohorts were injected: 1) F1 mice receiving 8 × 106 naive donor CD4 T cells and assessed at 8 wk of disease and 2) F1 mice receiving 15 × 106 donor CD4 T cells and assessed at 12 wk of disease. No significant differences in host B cell numbers, donor CD4 engraftment, or host T cell expansion were observed between lpr→F1 and WT→F1 mice in the 8-wk cohort (data not shown) or the 12-wk cohort (Fig. 10, C and D), with the exception of a small but significantly greater number of host CD8 T cells for WT→F1 vs lpr→F1. Serum anti-ssDNA Ab were similar for WT→F1 vs lpr→F1 mice and no significant differences were observed for the 8-wk cohort at any time point (Fig. 10,E). Both groups in the 12-wk cohort exhibited greater anti-ssDNA Ab levels overall compared with the 8-wk cohort due to the greater number of donor CD4 T cells injected. Significantly greater serum anti-ssDNA was observed for lpr→F1 mice at 4 wk and, to a lesser extent, at 12 wk. These differences did not translate into more severe renal disease as histologic scores of tubular epithelial injury (Fig. 10 F) and of mesangial hypercellularity (data not shown) were not significantly different between WT→F1 vs lpr→F1 mice. Similarly, there was no significant difference in renal disease for the 8-wk cohort receiving 8 × 106 CD4+ T cells (data not shown). These results indicate that Fas- defective CD4 T cells do not exhibit excessive help for B cells or induce more severe lupus-like renal disease in the P→F1 model used here.
Contraction of activated T cells following peak effector generation is mediated by both Fas-dependent and Fas-independent mechanisms (3, 4, 5, 6). Because Fas-dependent AICD involves only the Ag-specific T cells involved in an immune response (38, 39, 40), we chose to determine the in vivo effect of selective Fas deficiency on only the activated Ag-specific T cells involved in either a CTL or T-dependent Ab-mediated response. This question can be readily addressed in the P→F1 model by transferring Fas-defective lpr donor T cells into normal (unirradiated) Fas-intact F1 hosts. The P→F1 model is used in these studies not for its relevance to human GVHD but instead for its ability to model an in vivo T cell-driven immune response. Our results demonstrate that Fas expression on Ag-activated T cells has costimulatory, helper, and down-regulatory roles in vivo; however, these roles were observed only in a CTL response and not in a T-dependent B cell response. For example, the costimulatory and helper roles were seen in acute GVHD in which Fas-defective lpr donor CD4 T cells exhibited defective help for the maturation of donor CD8 T cells into effector CTL. Fas-defective lpr CD4 exhibited no defect in providing help to host B cells and in the induction of IgG autoantibodies in chronic GVHD. Although a costimulatory role for T cell-expressed Fas has been previously described in vitro (71, 72, 73), this in vivo dichotomy of Fas dependence for CD4 help to CD8 cells but not for B cells is novel.
Similarly, the down-regulatory role of Fas was also best seen in acute GVHD in that Fas-defective donor CD8 T cells (and to a lesser extent, Fas-defective donor CD4 T cells) exhibited impaired down-regulation as demonstrated by significantly greater donor engraftment and greater proliferation compared with down-regulated WT donor T cells at day 14. In chronic GVHD mice, there were no significant differences in engraftment for Fas-defective vs WT donor CD4 T cells at either 2, 8, or 12 wk of disease, indicating that Fas expression on Ag-activated CD4 T cells played no detectable down-regulatory role in T cell-driven B cell responses. The lack of demonstrable Fas function in chronic GVHD may relate to relatively lower IFN-γ production. Acute GVHD mice exhibit striking elevations of serum IFN-γ (102 to 103-fold > normal) compared with that seen in chronic GVHD mice (2- to 3-fold > normal) (51). In acute GVHD mice, serum IFN-γ levels peak at day 7 of disease (65) and are followed at day 10 by IFN-γ-dependent Fas up-regulation on both donor and host cells (49) and significantly increased donor T cell apoptosis (approximately >3-fold vs normal) (52). By contrast, the lower IFN-γ response in chronic GVHD mice is associated with minimal donor T cell Fas up-regulation and low-level donor T cell apoptosis (≤2-fold vs normal) (49, 52, 65). Based on the well-recognized proapoptotic effect of IFN-γ and its role in AICD (74, 75), it is reasonable to hypothesize that IFN-γ-dependent AICD is an important down-regulatory mechanism for donor T cells in acute but not chronic GVHD. The results of the present study formally confirm this idea by demonstrating that in the presence of high-serum IFN-γ, Fas expression on donor T cells is essential for optimal down-regulation and contraction of CTL responses. Lpr donor T cells nevertheless undergo Fas-independent apoptosis (Fig. 9) however, this was not sufficient to reduce donor CD8 T cell proliferation and cell numbers to levels comparable for WT donor cells at day 14. By contrast, we were unable to demonstrate any role for Fas expression in CD4-driven B cell responses in the setting of low- serum IFN-γ elevations typical of chronic GVHD mice.
Although the lpr phenotype is associated with massive lymphoproliferation (21), we were unable to demonstrate a role for lpr non-T cells in mediating the expansion of host cells at day 14 seen in lpr→F1 mice. Instead, host expansion is a consequence of delayed donor CD8 CTL effector function. It has been previously demonstrated that the first 2 wk of chronic GVHD consist of donor CD4-driven expansion of host cells, whereas acute GVHD consists of an initial donor CD4-driven expansion phase followed by a second donor CD8 CTL-mediated elimination phase (59, 65, 76). Additionally, we have demonstrated that different host cell populations exhibit different elimination kinetics in WT→F1 acute GVHD mice, with B cells exhibiting the earliest elimination followed by host CD4 T cells and then later host CD8 and DC (65). This hierarchy of host cell elimination in WT→F1 mice is also seen in Fig. 5. Lastly, we have previously demonstrated that interventions that prevent donor CD8 CTL maturation will convert acute GVHD to chronic GVHD phenotype (i.e., host expansion) at 2 wk (55, 77, 78) and, conversely, interventions that promote CD8 CTL maturation will convert chronic GVHD to acute GVHD (i.e., host elimination) (65, 79). Thus, impaired host cell elimination (particularly B cells) in GVHD mice is a surrogate marker for impaired CD8 CTL function.
The kinetic data in Fig. 5 clearly demonstrate both expansion and elimination phases in both WT→F1 and lpr→F1 acute GVHD. Importantly, there was no significant difference in peak host expansion for any of the four host cell populations measured. Thus, it is not the expansion phase that is abnormal in lpr→F1 but rather the elimination phase that is delayed that in turn is secondary to delayed and/or impaired maturation of donor antihost CD8 CTL effectors as shown by the following in lpr→F1 mice: 1) delayed peak donor CD8 T cell numbers (Fig. 5,B), 2) delayed donor CD8 FasL up-regulation (Fig. 6,B) and granzyme B expression (Fig. 6,D), 3) reduced peak day 10 perforin gene expression (Fig. 6,E), and 4) reduced peak day 10 ex vivo killing of host targets (Fig. 6 F).
Experiments mixing CD4 and CD8 donor T cells from WT and Fas-defective lpr donor mice underscore our two main conclusions that 1) CD4-expressed Fas plays a major role in help for CD8 CTL and 2) CD8-expressed Fas plays a major role in CTL down-regulation. Specifically, recombining individually purified CD4 and CD8 T cells and injecting them into F1 mice (either WT CD4 + WT CD8→F1 or lpr CD4 + lpr CD8→F1) reproduces the phenotypes shown in Fig. 1 using unfractionated lpr donor T cells. That is, lpr→F1 mice exhibit impaired antihost CTL elimination of host B cells, reduced peak serum IFN-γ levels, and impaired donor CD8 down-regulation. Defective elimination of host cells by lpr CD8 T cells is corrected when paired with WT CD4 T cells, indicating that defective lpr CD8 CTL function in lpr→F1 mice is primarily a consequence of impaired CD4 T cell help. This conclusion is further supported by the reciprocal combination. Pairing lpr CD4 T cells with WT CD8 T cells (lpr CD4 + WT CD8→F1) results in not only a failure to reduce host B cells below values of uninjected F1 mice but values were increased. This increase in host B cells is consistent with conversion to chronic GVHD and indicative of a major impairment of donor CD8 CTL function (50, 55, 80). Thus, the greater host B cell elimination seen when Fas-defective lpr CD4 T cells are paired with Fas-defective CD8 T cells (lpr CD4 + lpr CD8→F1) vs WT CD8 donor T cells (lpr CD4 + WT CD8→F1) can be explained by the impaired down-regulation of lpr CD8+ T cells that not only compensates for the reduced lpr CD4+ Th cell help but also prolongs CTL action by delaying down-regulation such that CTL function eventually results in host B cell elimination. The profound defect in lpr CD4+ Th cell function is unmasked when paired with WT CD8 T cells that undergo normal down-regulation and demonstrates that lpr CD4 T cell help is insufficient to induce detectable elimination of host B cells by WT CTL. Not only do these experiments illustrate the importance of CD4 Fas expression in providing help to CD8 CTL, they also emphasize the importance of CD8 Fas expression in down-regulation by demonstrating that Fas-defective donor CD8 T cells have such a severe impairment in down-regulation that they can persist and mediate host B cell elimination despite minimal CD4 T cell help. Nevertheless, qualitative defects in CTL function likely exist in lpr→F1 mice as demonstrated by the delay of >2 wk in mortality (Fig. 1,A) that is out of proportion to the approximate 4-day delay in near complete host B cell elimination for lpr→F1 vs WT→F1 acute GVHD mice (Fig. 5 C).
Although Fas is best known for its role in apoptosis induction, a costimulatory role has been previously reported for T cell-expressed Fas in vitro. TCR-stimulated T cells receiving either anti-Fas mAb (71, 72) or FasL (73) in vitro exhibited increased T cell proliferation rather than apoptosis. These in vitro studies support our in vivo and in vitro observation that CD4 Fas expression provides an important costimulatory signal rather than an apoptosis signal to CD8 CTL early in T cell activation. The defect in proliferation of lpr CD4 cells seen both in vitro and in vivo is associated with a functional defect in help for CTL but not for B cells. Moreover, reverse signaling of CD8-expressed FasL can result in costimulation and optimal CD8 expansion following Fas binding of CD8-expressed FasL in vitro (81, 82, 83),raising the possibility that T cell-T cell binding of Fas-FasL contributes to optimal CD8 CTL induction. By contrast, we were unable to demonstrate a costimulatory role for CD4-expressed Fas in promoting help for B cell responses in chronic GVHD mice in the P→F1 model. Although T cell Fas expression is known to contribute to down-regulation of B cell autoantibody production (84), we found no reports of a costimulatory role for T cell Fas expression in B cell autoantibody production.
Our results have implications for the role of T cell- expressed Fas not only in lymphocyte homeostasis following a normal immune response as modeled by P→F1 GVHD, but also in lupus pathogenesis with separate implications for CD4 and CD8 T cells. Regarding CD4 T cells, their critical role in driving autoantibody production in both human and murine lupus is well known (34, 36, 61, 85, 86, 87). CD4 T cells from MRL/+ lupus-prone mice exhibit an intrinsically lowered threshold of activation that has been hypothesized to lead to enhanced autoreactive T cell proliferation and enhanced helper functions for autoimmune B cells (88, 89). This hyperresponsiveness likely provides the initial loss of tolerance to self-Ags required for lupus expression. Disease acceleration in Fas-defective MRL/lpr mice is multifactorial and reflects contributions from T cells, B cells, and DC (33, 41, 42, 70, 90, 91). Regarding the T cell contribution, it has been demonstrated that MRL CD4 T cell lines exhibit defective apoptosis (29) and that lpr T cells hyperproliferate (30) and provide cognate help to B cells in vivo (70), leading to the conclusion that impaired Fas-mediated down-regulation of autoreactive CD4 T cells accentuates the hyperresponsiveness and the loss of T cell tolerance characteristic of MRL/+ mice. As a result, T cell help for autoreactive B cells is prolonged and disease is accelerated (29, 31). Although the P→F1 model breaks tolerance by transferring foreign-reactive CD4 T cells rather than autoreactive T cells, a lupus-like phenotype nevertheless ensues (44). Using this model, our study failed to demonstrate that F1 mice receiving Fas-defective lpr CD4 T cells exhibit greater Ag-specific (donor) CD4 T cell expansion, greater CD4 help to B cells, consistently greater autoantibody production, or induce more severe lupus-like renal disease in F1 mice compared with mice receiving WT donor CD4 T cells. Extrapolating these results to MRL/lpr lupus, our data raise the possibility that Fas deficiency on lpr CD4 T cells is insufficient by itself to account for excessive T cell help to B cells characteristic of MRL/lpr lupus. Our data however do not exclude a role for defective CD4 Fas in MRL/lpr acceleration as disease is characterized by high IFN-γ levels (vs low IFN-γ in chronic GVHD) and is improved by IFN-γ blockade (92, 93). It is possible that the proapoptotic effect of elevated IFN-γ retards disease in MRL/+ mice by up-regulating Fas expression on hyperresponsive CD4 T cells, resulting in AICD, and that this effect is reduced in MRL/lpr mice despite high IFN-γ levels, thereby accelerating disease.
Regardless of the exact role of Fas on lupus CD4 T cells, our results demonstrating that Fas is more important in CD8 CTL responses than in CD4-mediated B cell responses raise the novel possibility that defects in CD8 CTL function are an unrecognized contributor to disease acceleration in MRL/lpr mice. CD8+ CTL are well known for their roles in viral immunity, tumor surveillance, allograft rejection, and lymphocyte homeostasis (94, 95). It has been previously demonstrated in the chronic GVHD model of induced lupus that CD8 CTL limit the expansion of all B cells (both autoreactive and foreign reactive) using both perforin and Fas pathways (49) and that the loss of a single pathway can tip the balance toward expansion of autoreactive B cells and autoimmunity (50). Similarly, both Fas and perforin are important in controlling lupus in MRL/lpr mice. As shown by Peng et al. (96), humoral autoimmunity in perforin-intact, Fas-intact MRL/+ mice can be accelerated with perforin gene inactivation indicating a major role for perforin, separate from Fas, in controlling autoimmunity in this model. These results strongly implicate an important down-regulatory role for CD8 CTL in MRL lupus given that perforin and Fas are the two major pathways by which CTL lyse targets (97) and CD8+ T cells, along with Th1 CD4+ T cells, are a major FasL-bearing cell population (98). Thus, while defects in Fas or perforin do not cause lupus, in the setting of a loss of tolerance and lupus initiation (e.g., MRL/+ mice or chronic GVHD mice), defects in one or both of these pathways can accelerate disease.
Defective in vitro CTL function is well described in human and murine lupus (99); however, it is unresolved whether this represents a primary defect, a secondary defect resulting from the altered immunoregulation of lupus, or a combination of both. Supporting the existence of a secondary defect is the demonstration that MRL/lpr mice exhibit an age-related and disease-related defect in in vitro CD8 CTL function (100) and that a similar defect in in vitro CTL function is induced in chronic GVHD mice at 2 wk of disease (55). In both models, defective in vitro CD8 CTL generation is secondary to defective in vitro IL-2 production similar to that reported in humans (101). In MRL/lpr mice, defective in vitro CTL and IL-2 responses are mediated by a CD4+ T suppressor cell, possibly a regulatory T cell (100). Our results in this study indicate that defective CD8 CTL function in MRL/lpr mice may be due not only to possibly enhanced regulatory T cell function, but also to defective CD4 costimulation that not only impairs maturation of CD8 CTL but may also result in a qualitative defect in those CTL that do mature. The functional consequences in vivo may be mitigated somewhat by delayed down-regulation of those CD8 CTL that manage to become activated.
Although the nature and meaning of the lupus-associated defect in CD8 CTL function described above are not fully understood at present, it is possible that therapeutic enhancement of CD8 CTL function may be beneficial in lupus patients. Supporting this idea is work by Fan et al. (102) demonstrating that promoting CTL specific for autoreactive B cells is both feasible and beneficial in the NZB/W model of spontaneous lupus. Our results in this study support further work aimed at identifying the mechanisms that are important in mediating in vivo CD8 CTL responses, particularly as they relate to controlling autoreactive B cells.
We thank Dr. Keith Elkon for helpful discussions and Karen Wolcott for technical assistance.
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.
These studies were supported by National Institutes of Health Grant AI 47466 (to C.S.V.). R.P. and A.S. are recipients of an Engelicheff Fellowship Award from the Maryland Chapter of the Arthritis Foundation.
Abbreviations used in this paper: AICD, activation-induced cell death; GVHD, graft-vs-host disease; IVCCA, in vivo cytokine capture assay; P→F1, parent-into-F1; WT, wild type; FasL, Fas ligand.