Comment on “Cutting Edge: Regulatory T Cells Do Not Mediate Suppression via Programmed Cell Death Pathways”

We read the article by Szymczak-Workman et al (1) with great interest. They have concluded that apoptosis of effector CD4 T cells is not required for CD4+Foxp3+CD25+ regulatory cell (Treg) mediated suppression. Several studies have demonstrated the pivotal role of Treg mediated apoptosis in suppression of immune responses (2-6). It is surprising that this study by Vignali's group did not observe cell death in responding CD4+ T cells (Tresp) in vitro. Some of the reasons they have given for the discrepancy in their findings include different experimental systems, involving different stimulation conditions and using FACS versus MACS purified Treg cells. We speculated that in their in vitro experimental conditions, they have tested for apoptosis too early (68 hr), which is probably why they have failed to observe cell death. Their cells showing more proliferation indicate that they have stronger stimulation conditions than those in our previous study that showed apoptosis in Tresp cells (4).

Different T cell activation conditions can yield varying amounts of proliferation suppression/apoptosis of Tresp cells (7) (Fig 1A). Under lower stimulation conditions, we found that the onset of apoptosis was about 48 hours after co-culture but occurred at 0.5 - 2 μg/ml α-CD3 antibody concentrations, consistent to our previous findings. (Fig.1B). We then determined the fate of CD4 Tresp cells under stronger stimulation conditions, using soluble α-CD3 antibody along with α-CD28 antibody and APC and using FACS sorted Tregs. We found that although the apoptosis of Tresp cells was slightly delayed and was not observed on day 1 and 2 (data not shown), the CD4 Tresp cells did undergo apoptosis on days 3 and 4. Moreover, using the CD4 Tresp cells from Bim KO (BKO) mice obtained from Jackson mice facility (BKO1) and from Vignali's group (BKO2), we found that those cells underwent substantially less apoptosis (Fig.1C). Although Bim-KO cells were reduced in numbers on day 4 in Treg co-cultures, they were largely rescued from proliferation suppression (Fig.1C). These data were consistent to our previous findings (4). Although these new data may point out to the role of apoptosis independent components in Treg mediated suppression in vitro and in vivo, apoptosis of Tresp CD4 T cells clearly plays an important role in Treg suppression. Our data here demonstrate that even under stronger activation conditions, Tregs do induce Bim dependent apoptosis rather in a delayed manner. We believe that Vignali's group determined apoptosis at 68 hr after co-culture, which is too early for their activation conditions and thus have failed to detect apoptotic cells. However, it is interesting that they observed Treg induced suppression without detecting apoptosis in Tresp cells. It is also possible that they excluded apoptotic/dead Tresp cells using a smaller forward scatter gate, when analyzing for Annexin V -, PI - live cells in their Fig 2B.

Fig.1.

Responder or control CD4+ CD25- (Tresp or Tcon) and CD4+ CD25+ (Treg) cells were isolated from spleen and stimulated in cultures as in (4). Tresp cells were cultured alone or in co-cultures with Tcons or Tregs at 1:1 ratio. (A) The cells were stimulated using soluble α-CD3 antibody (0.5 μg/ml) along with α-CD28 antibody (2 μg/ml) (a), α-CD3 antibody along with APC (b) or α-CD3 antibody along with α-CD28 antibody and APC (c). The proliferation of live (left) and the percentage of Treg induced apoptosis (right) of carboxyfluorescein succinimidyl ester (CFSE) labeled Tresp cells, as detected 4 days after culture. (B) The cells were stimulated using indicated concentrations of soluble α-CD3 antibody and 2 μg/ml of α-CD28. IL-2 was added at 20ng/ml concentration where indicated. (C) The cells were stimulated using indicated concentrations (0.5 or 5 μg/ml) of soluble α-CD3 antibody, 2 μg/ml of α-CD28 and APC for 66 or 96 hours. Percentage Tresp apoptosis (upper panel) or CFSE+ propidium iodide- Tresp live cell counts (lower panel), as measured by flow cytometry in equal volumes of cells and at constant times. ***P <0.005, ** P< 0.05.

Also, we have consistently observed that Tregs did not suppress IBD caused by Bim KO cells in vivo (4). The authors have shown that Tregs suppressed IBD caused by Bim KO cells, and speculated that the differences may arise from mouse facilities and the commensal microbiota in mice. Because severity of IBD is strongly dependent on gut commensal microbiota, this possibility cannot be ruled out. However, to induce IBD, they have used naïve T cells from C57BL/6 (WT) or Bim KO mice and intravenously injected the cells in to the tail vein of the mice. 3–4 wk after initial transfer, upon clinical signs of sickness, they have injected Tregs intraperitoneally. In the IBD model, injection of naïve cells is followed by their cytokine dependent homeostatic proliferation, and also their expansion in the gut in response to gut microbiota. Therefore we believe that Tregs should be present early on to limit the cytokine dependent homeostatic expansion of effector cells, by causing Bim dependent effector cell death. Whereas in our experiments, we injected Tregs on d5-d7 after initial transfer, they have injected Tregs only after the onset of the disease. i.e after the effector cells have responded to gut microbiota. By injecting Tregs much later, the authors have missed the window at which the effector cells are sensitive to cytokine deprivation. Also, they have not determined the frequency of apoptotic cells in vivo. Thus, based on their in vivo experiments, it is not appropriate to interpret that Tregs suppress T effector cells independently of Bim dependent apoptosis.

After the onset, the disease involves reactions to microbiota in the gut, activation by TLR ligands and excessive inflammatory cytokines, where Tregs cannot cause cytokine deprivation. In our initial study we had speculated this and noted “Treg cells might initially decrease the size of the effector CD4+ T cell population by apoptosis, and this process may permit apoptosis-independent control of the smaller T cell population..”(4). We did not rule out the possibility that Tregs may employ other mechanisms. Indeed we have shown that Tregs can suppress IBD independently of causing apoptosis on effector cells under Th17 milieu (8). However, in the IBD model, where naïve cells are transferred with and without Tregs, Bim dependent cytokine deprivation is one of the suppressive mechanisms employed by Tregs (4). Thus the authors were not justified in concluding that apoptosis was not involved in Treg mediated suppression, based on the experiments performed in completely different conditions.

Another minor issue is that, in their Fig.2A, the legends “Tconv alone” and “Tconv + WT Treg” appear to be switched. “% Undivided cells” also appear to show that conventional cells are proliferating better with Tregs.

In general, speculating on different mechanisms of Treg suppression just by using different cell culture conditions and different in vivo experimental settings has caused a great deal of confusion in Treg literature. It is interesting that in the same issue of Journal of Immunology, Vignali's group has also shown TRAIL dependent programmed cell death mediated by Tregs (9). In light of those findings from the same group, it is unclear how they could conclude that Tregs do not mediate suppression via programmed cell death pathways.