Massive and destructive T cell response to homeostatic cue in CD24-deficient lymphopenic hosts

To study the contribution of CD24 on the host APC to homeostatic proliferation, we transferred CFSE-labeled naive T cells from Thy1.1 congenic mice into sublethally irradiated wild-type (WT) B6 mice and CD24^−/− B6 mice. The spleens were harvested from the recipients 4 d after adoptive transfer. Cell division was measured by CFSE dilution by flow cytometry. As shown in Fig. 1 a, the transferred CD4 T cells from the WT B6 recipients underwent two divisions. At the same time, CD8 T cells divided four times. On the other hand, both populations of the transferred T cells in the CD24^−/− mice underwent much more accelerated (more than seven) divisions. We then performed a kinetic study to substantiate this surprising observation. On day 1, the T cells in both recipients maintained the same level of CFSE fluorescence density (Fig. 1 b). However, on day 2, when little change happened to the transferred T cells in the WT B6 recipients, the T cells in CD24^−/− recipients already started dividing. By day 3, in WT B6 mice, when T cells barely divided, a substantial fraction of T cells had already divided five times in CD24^−/− recipients. Thus, although the same amount of naive T cells was transferred into both groups, T cells started to divide much earlier and faster in the CD24^−/− recipients. Nevertheless, we have not observed an increased accumulation of T cells in the CD24-deficient host (Fig. S1).

Although these studies raised the intriguing possibility that CD24 on the host cells may serve as a key regulator for the rate of homeostatic proliferation, several caveats deserve consideration. First, despite the fact that the CD24^−/− mice were produced with embryonic stem cells from C57BL/6 mice (23), it remains possible that the two hosts are not histocompatible. To test this possibility, we transplanted the skin grafts from the CD24^−/− mice into either WT B6 mice or allogeneic BALB/c mice. As shown in Fig. 1 c, the BALB/c mice promptly rejected the allogeneic grafts, whereas the B6 mice accepted all of the grafts indefinitely. Therefore, the CD24^−/− mice are histocompatible to B6 mice as expected.

As an alternative approach, we adoptively transferred CFSE-labeled T cells from Thy1.1 congenic mice into either C57BL/6 or C57BL/6 CD24^−/− mice. 4 d later, the recipient mice were killed, and proliferation of the donor cells was analyzed by flow cytometry. As shown in Fig. 1 d, the donor cells underwent no proliferation over the 4-d period. These results demonstrate that in the absence of lymphopenic cue, CD24 deficiency in the host cells was unable to induce the proliferation of T cells from congenic mice.

Our extensive analyses have demonstrated that CD24 deficiency does not substantially impact the priming of T cells (19, 20, 24). To test whether CD24 deficiency in the host cells specifically affects homeostatic proliferation, we labeled OT-1 T cells with CFSE and injected them into the host that had been immunized with the specific ovalbumin peptide. As shown in Fig. 2, essentially identical proliferation was observed in the WT and CD24-deficient mice. Again, the OT-1 cells did not proliferate in unimmunized mice. Collectively, our data demonstrate that CD24 deficiency in the host selectively enhanced homeostatic T cell proliferation. Therefore, CD24 on the APC served as a suppressor of homeostatic T cell proliferation.

Previous studies have demonstrated that T cells undergoing homeostatic proliferation have the markers of central memory cells; namely, they express high levels of CD44 but generally fail to down-regulate CD62L (6, 7). Moreover, T cells that underwent antigen-driven proliferation expressed high levels of CD25 and CD69, whereas those that underwent homeostatic proliferation under normal circumstances lacked these two markers (25). As shown in Fig. 3 a, few cells isolated from WT recipients expressed CD25. However, close to 50% of T cells that were recovered from the CD24^−/− host expressed CD25. Regardless of the CD24 status in the host, T cells acquired memory cell phenotype with a high level of CD44 expression (Fig. 3 b, top), which is consistent with previous studies (6, 7). However, the majority of T cells from CD24^−/− recipients but not those from the WT mice had down-regulated 62L (Fig. 3 c). The down-regulation of CD62L was more pronounced at 2 wk after adoptive transfer, when an overwhelming majority of activated T cells displaced a CD44^hiCD62L⁻ phenotype. Thus, removing CD24 from the lymphopenic host converts a weak homeostatic cue into a strong stimulation similar to antigenic-driven proliferation.

CD24 is expressed in both hematopoietic and nonhematopoietic cells (26). To determine whether CD24 expressed on the hematopoietic cells is responsible for pacing homeostatic proliferation, we transferred WT and CD24^−/− bone marrow into lethally irradiated CD24^−/− mice. After 4 wk, we transferred CFSE-labeled purified naive T cells from Thy1.1 congenic mice into both recipients, which had been sublethally irradiated right before T cell transfer. We harvested spleens from both groups and analyzed CFSE dilution in the donor T cells (Fig. 4 a). As shown in Fig. 4 b, in the CD24^−/− bone marrow–reconstituted recipients, both CD4 and CD8 T cells had undergone massive proliferation over the 4-d period as expected. In contrast, much less proliferation was observed in the chimera mice reconstituted with WT bone marrow cells. Thus, the expression of CD24 on bone marrow–derived cells is both necessary and sufficient to control the pace of homeostatic proliferation.

It has been established that both mouse and human DCs can induce T cell homeostatic proliferation in vitro when the DC number is substantially higher than T cells (27). To determine whether a lack of CD24 on the DCs promotes their ability to induce homeostatic proliferation, we generated DCs from both WT and CD24^−/− bone marrow and cultured CFSE-labeled purified naive Thy1.1 congenic T cells with the DCs at a suboptimal 1:2 ratio. After 5 d, we analyzed the CFSE dilution to determine T cell proliferation. As shown in Fig. 4 c, T cells cultured with WT B6 bone marrow–derived DCs had barely divided, whereas those cocultured with CD24^−/− DCs divided more than one to five times. Thus, CD24-deficient DCs had a superior stimulatory activity for homeostatic T cell proliferation. Careful analysis of the expression of MHC and costimulatory molecules in the two populations of DCs revealed little difference (Fig. S2). In contrast, the coculture of syngeneic macrophages (Fig. 4 d) and B cells (Fig. 4 e) with naive T cells failed to induce proliferation regardless of whether or not the CD24 gene had been inactivated.

To confirm that CD24 deficiency is sufficient to confer superior homeostatic proliferation in vivo, we adoptively transferred naive T cells together with WT and CD24^−/− DCs into lymphopenic WT hosts and monitored T cell proliferation 4 d later. As shown in Fig. 4 f, the injection of CD24^−/− but not WT DCs led to the strong homeostatic proliferation of T cells. Collectively, our data demonstrate that CD24 expressed on DCs controls the pace of homeostatic proliferation in lymphopenic hosts.

Surprisingly, ∼60% of the CD24^−/− recipients died within 2 wk after T cell transfer, whereas all of the WT recipients remained healthy (Fig. 5). The death was caused by adoptively transferred T cells, as the control mice that received no T cells were healthy. We have systematically examined the host for histological signs of acute graft versus host diseases. As shown in Fig. S3 , no inflammation was observed in the gut or other organs of the CD24-deficient host. Thus, the destruction of the recipient is pathologically distinct from graft versus host diseases. Histology of the spleen suggests extensive hemolysis in the red blood cells in the moribund mice (unpublished data). The high fatality and rapid onset demonstrate the vital role of CD24-mediated regulation of homeostatic proliferation.

As an alternative approach, we used anti-CD4 and anti-CD8 antibodies to transiently deplete T cells in the mice to create a lymphopenic environment and compared WT with CD24-deficient mice for T cell recovery. As shown in Fig. 6 a, a single injection of anti-CD4 and anti-CD8 antibodies caused the rapid depletion of T cells, with a comparable efficacy in WT and CD24 knockout mice. Over time, T cells gradually recovered as expected. Interestingly, the recovery of CD8 T cells was substantially more rapid in CD24-deficient mice than in the WT mice, although the rate of CD4 T cell reconstitution was unaffected by CD24 deficiency. To determine whether the increased T cell number was caused by increased thymic output, the number of recent thymic emigrants were counted by the amount of TCR rearrangement excision circle (TREC⁺) T cells, which was quantified by real-time PCR as described previously (28). Because the frequencies of TREC⁺ T cells was comparable between the two groups (Fig. 6 b), the increased CD8 T cell recovery was therefore not caused by increased thymic output.

To directly demonstrate the contribution of homeostatic T cell proliferation, we pulsed WT and CD24-deficient T cells with BrdU and analyzed the proliferating T cells in the WT and CD24-deficient host. The number of CD4 T cells was reduced in the CD24-deficient mice (Fig. 7 a). However, the percentage of BrdU⁺ T cells was at least doubled in the CD24-deficient mice in both CD4 and CD8 compartments (Fig. 7 b). To reconcile the apparent discrepancy, we stained spleen cells with Annexin V. As shown in Fig. 7 c, the percentage of Annexin V⁺ cells was substantially higher in the CD24-deficient CD4 T cells in comparison with their WT counterpart. Thus, a lag in the accumulation of CD4 T cells was likely caused by increased cell death. In contrast, the percentage of Annexin V⁺ CD8 T cells was reduced in the CD24-deficient mice (Fig. 7 c). As such, both increased proliferation and reduced cell death contributed to a substantially faster reconstitution of CD8 T cells (Figs. 6 and 7).

Positive selection ensures that all mature T cells have the ability to recognize self-ligands (29). It is becoming clear that positive-selecting ligands continue to play an important role after T cells have matured. In the lymphopenic environment, the same ligands drive naive T cells to proliferate spontaneously (5, 10). Although homeostatic proliferation is polyclonal in nature, it is slow paced and does not lead to deleterious consequences under physiological conditions. In this study, we report evidence that the pace of homeostatic proliferation is set by CD24 expressed on the DCs and that this regulation is critical for the survival of the host under a lymphopenic environment.

Our data demonstrate that WT T cells adoptively transferred into CD24-deficient mice underwent massive homeostatic proliferation that ultimately led to the death of the majority of the recipients. The death was associated with massive T cell activation but not infiltrations of organs, which is inconsistent with the activation of organ-specific T cells. Although this may be associated with so-called “cytokine storm,” our analysis did not reveal an obvious rise in the percentage of T cells capable of producing TNF-α, a cytokine frequently associated with cytokine storm (Fig. S4). Therefore, the death was unlikely caused by excessive TNF production. Histological analysis of the spleen revealed extensive hemosiderin deposits in red pulps, which is indicative of hemolysis in moribund mice (unpublished data), although its cause and significance remains to be determined.

In addition to the adoptive transfer model, we also observed a substantially increased homeostatic proliferation in CD24-deficient mice after CD4 and CD8 T cells were deleted by a single injection of a low dose of anti-CD4 + anti-CD8 antibodies. These results are noteworthy for three reasons. First, because this assay involved no adoptive transfer, it helped to further rule out the possibility that increased proliferation was caused by histoincompatibility between donor and recipient cells even though the CD24-deficient mice were produced in embryonic stem cells derived from C57BL/6 mice. Second, because the hosts were not irradiated, the increased proliferation was not associated with changes in the irradiated recipient. Third, although CD24 expression in T cells was essential for optimal homeostatic proliferation in the WT host, such a requirement can be bypassed by a lack of CD24 on host APCs. Thus, negative regulation mediated by CD24 on APCs plays a dominant role over the positive regulatory role for CD24 on T cells.

We have presented several lines of evidence that the lack of CD24 in the DCs explains the massive homeostatic proliferation. Thus, the CD24-mediated suppression of homeostatic proliferation is mediated by bone marrow–derived cells as the reconstitution of CD24^−/− mice with the CD24^+/+ bone marrow restored the pace of homeostatic proliferation. In an in vitro model of homeostatic proliferation, DCs from CD24^−/− mice were more potent in inducing the proliferation of syngeneic T cells, whereas B cells and macrophages were unable to induce homeostatic proliferation. Most importantly, the adoptive transfer of bone marrow–derived DCs from CD24^−/− mice led to massive homeostatic proliferation in the lymphopenic WT host. Thus, defective CD24 expression on DCs plays a dominant role in driving massive homeostatic proliferation.

We have demonstrated that a targeted mutation of CD24 exacerbated immune deficiency associated with the targeted mutation of CD28 (19, 20), which revealed CD24 as a redundant costimulatory molecule for antigen-driven proliferation. In contrast, CD24^−/− DCs have a drastically stronger activity in inducing homeostatic proliferation. The apparently opposite function of CD24 in the two processes highlights the distinct requirements of T cell response to antigen and homeostatic cue. In this context, it is worth noting that anti-CD24 antibody 20C9, which blocks the costimulatory function of CD24, has no effect on the DC-mediated costimulatory function (unpublished data). It is likely that distinct structures on CD24 are involved in T cell costimulation and in homeostatic proliferation.

CD24 is a highly glycosylated cell surface molecule that is anchored through a glycosylphosphatidylinositol tail. Mouse CD24 consists of 27 amino acids that include eight potential O-linked glycosylation sites and three potential N-linked glycosylation sites (30). The heterogeneity of CD24, which is presumably caused by differential glycosylation, has made it difficult to identify the counter receptor involved in pacing homeostatic proliferation. In spite of this, our work provides the first evidence that the pace of T cell homeostatic proliferation is actively maintained by CD24 expressed on host APCs to avoid deleterious consequences associated with the uncontrolled polyclonal activation of T cells.

C57BL6/J and BALB/c mice were purchased from Charles River Laboratories through a contract with the National Cancer Institute. Transgenic mice OT-1 that express TCR specific for ovalbumin peptide SIINFEKL and B6.Thy1.1 mice were purchased from The Jackson Laboratory. Mice with the targeted mutation of CD24 were produced using embryonic stem cells from C57BL/6 mice as described previously (23). All of the mice were maintained under specific pathogen-free conditions in the animal facility at the Ohio State University. All studies involving animals were approved by the Ohio State University Institutional Laboratory Animal Care and Use Committee.

All conjugated antibodies used were purchased from either eBioscience (Thy1.1, CD80, CD86, CD44, and CD62L) or BD Biosciences (CD4, CD8, H-2K^b, I-A^b, CD25, CD40, Annexin V, IFN-γ, TNF-α, and isotype control).

T lymphocytes were purified from pooled spleen and lymph node cells by negative selection. In brief, pools of spleen and lymph node cells were incubated with a cocktail of antibodies specific for CD11b (Mac-1), Fc receptor (2.4G2), B220, and CD11c. The Dynal beads coated with goat anti–rat IgG were used to negatively select T cells. The purity of the cells was checked by flow cytometry to be >95%. The purified T cells congenic in the Thy1 locus (B6 and Thy1.1⁺) were labeled with CFSE and injected intravenously into irradiated (600 R) recipient mice that were WT or CD24^−/− (B6 and Thy1.2⁺) at a dose of 5 × 10⁶/mouse. At given times after adoptive transfer, spleen cells were harvested and analyzed for the intensity of CFSE dye and other cell surface markers.

Bone marrow cells from WT B6 mice and CD24^−/− mice were transferred back to 1,000 rad–irradiated CD24^−/− mice. 4 wk later, CD24 expression was detected in both groups to determine the reconstitution efficiency.

Bone marrow cells from WT B6 mice or CD24^−/− mice were cultured with 10% RPMI and recombinant mouse GM-CSF as described previously (31). LPS was added into the culture after day 10 at a final concentration of 1 μg/ml for 2 d. The nonadherent cells harvested from the culture were mature DCs.

For suspension cultures, CFSE-labeled Thy1.1⁺ T cells (usually 500,000/well) were mixed with given numbers of matured DCs in 24-well plates in RPMI 1640 medium. Cells were cultured at 37° in a humidified 5% CO₂ incubator for 5 d before analysis.

WT or CD24^−/− mice were injected with 100 μg anti-CD4 (GK1.5) and 100 μg anti-CD8 (2.4.3) antibody. The mice were bled, and percentages of CD4 and CD8 T cells in the peripheral blood were determined by flow cytometry. 7 d after the antibody injection, some of the mice were given three injections of BrdU (1 mg/injection i.p.) with 12-h intervals. 12 h after the last BrdU injection, spleen cells were analyzed for division by BrdU incorporation and cell death by Annexin V staining.

The cell surface markers, including CD4, CD8, CD24, CD25, CD44, CD62L, and Thy1.1, were analyzed by three- or four-color flow cytometry using fluorochrome-conjugated monoclonal antibodies purchased from BD Biosciences. To assess intracellular cytokine production, spleens were harvested. The splenocytes were stimulated with PMA for 4 h and stained for cell surface markers CD4 and CD8 followed by intracellular staining for IFN-γ/TNF-α and/or isotype control using the CytoFix/CytoPerm kit (BD Biosciences).

Copy numbers of the sjTREC were used to determine the frequency of recent thymic emigrant (28). In brief, total DNA was purified from total splenocytes with the DNA Easy kit (QIAGEN). Real-time PCR was performed using the Quantitative PCR kit (Invitrogen). The primers used were as follows: mδRec primer (5′-GGGCACACAGCAGCTGTG-3′), ψJα primer (5′-GCAGGTTTTTGTAAAGGTGCTCA-3′), mδRec-ψJα fluorescent probe (5′-FAM-CACAAGCACCTGCACCCTGTGCA-TAMRA-3′), and CD8β forward primer (5′-CAGGACCCCAAGGACAAGTACT-3′), reverse primer (5′-CACTTTCACCATACAAAACTCCTTTG-3′), and probe (5′-FAM-TGAGTTCCTGGCCTCCTGGAGTTCTTC-TAMRA-3′). The number of TREC/million T cells was calculated by: 2 × (number of TREC/number of CD8β × percentage of T cells) × 10⁶.

Fig. S1 shows that despite a major difference in the rate of homeostatic proliferation, the numbers of donor T cells were comparable in WT and CD24^−/− recipients. Fig. S2 shows that CD24 deficiency does not alter the expression of MHC and costimulatory molecules on DCs. Fig. S3 shows the histology analysis of healthy WT mice and moribund CD24-deficient mice. No inflammation was observed in the moribund mice. Fig. S4 shows the high and comparable percentages of TNF-α– and IFN-γ–producing cells in WT and CD24^−/− recipients.

We thank Lynde Shaw for secretarial assistance.

This study was supported by grants from the National Institutes of Health and by the State of Ohio Biomedical Research Technology Transfer Grant.

The authors have no conflicting financial interests.