The TGF-β family comprises many cytokines involved in the control of cell fate 1. Some of these have been shown to control cell death. For instance, at least one of the members of the BMP subfamily of the TGF-β family is required in interdigital death, a classical model of developmentally programmed cell death 2. Another TGF-β family member, TGF-β1, entertains a complicated relationship with cell death, of lymphocytes in particular. When added extracellularly, it has been reported to induce death of B lymphocytes 3,4 and to prevent T lymphocyte death 5,6,7.
A new, and at times somewhat heretical twist, is now reported 8. In mice homozygous for a TGF-β1 null mutation 9,10, thus not expressing TGF-β1, W.J. Chen et al. 8 found that apparently spontaneous apoptosis of both thymic and peripheral T cells was increased compared with wild-type controls. These findings might seem relatively trivial, since TGF-β1 had previously been shown to prevent T cell death (see above). However, not only was spontaneous T cell death impressive in both thymus and peripheral organs, but results were even more striking when T cells were challenged with anti-CD3 antibody (mimicking stimulation through the TCR). In vitro experiments already showed an increased sensitivity of T cells from TGF-β1 knockout mice, resulting in marked activation-induced cell death. The most spectacular results were obtained with in vivo injection of anti-CD3 antibody into TGF-β1 knockout mice, leading to massive apoptosis in both thymus and spleen, visible as terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick end-labeling (TUNEL)-positive cells within hours, and effecting considerable reduction in lymphoid organ cell numbers. This was accompanied by an increase in expression of Fas and Fas ligand in spleen cells, and increase in sensitivity to the Fas pathway and in some situations to the TNFR2 pathway of cell death. These results demonstrated that activated lymphocytes in vivo require TGF-β1 to protect them from death.
More surprising results followed 8. TGF-β1, and more generally all members of the TGF-β family, are known to affect cells from the outside through cognate cell surface receptors, the engagement of which lead to intracellular activation of Smad transcription factors, especially Smad3 in the case of TGF-β1 1,11,12. However, the current work 8 suggests that the protective effects of TGF-β1 might be due to intracellular TGF-β1. Two lines of evidence support this. First, exogenous TGF-β did not correct the increased cell death observed in TGF-β1 knockout cells. Of course, one could imagine that externally applied TGF-β1 could be inactivated by secreted TGF-β antagonists, of which there are many 11. These are very efficient blockers when added externally as shown in the interdigital cell death model mentioned above 13,14, and might be more so in the present TGF-β1 knockout situation where these inhibitors would find no endogenous TGF-β1 to interact with. However, most of the abnormal traits in TGF-β1 knockout mice are reversed or prevented by ingestion of maternal TGF-β1 during suckling, while interestingly the apoptotic T cell phenotype is not. Second, no sensitization to cell death similar to that observed in TGF-β1 knockout mice was observed in mice inactivated for the Smad3 gene, ruling out the participation of a Smad3 transduction pathway. In line with this, Smad3−/− mice showed normal development of lymphocytes and increased proliferation and activation of T cells 15 within an abnormal phenotype which seemed generally milder than TGF-β1−/− (for a review, see reference 11). The authors thus favor the interpretation that TGF-β1 acts not only extracellularly via its cognate receptor and Smad3, but also intracellularly. As an alternative, one may hypothesize that TGF-β1 might act only extracellularly via its cognate receptor but be able to use, say, another Smad than Smad3 16. Definitive arguments in favor of an “intracrine” mechanism similar to that of e.g., fibroblast growth factor 2 17 may require, for instance, retroviral infection of TGF-β1−/− T cells with a construct encoding a nonexportable TGF-β1 molecule. The postulate that TGF-β1 has an intracellular role suggests interesting scenarios. Two sites of action are implied for TGF-β1 molecules, as a ligand for a cell-surface receptor, and as an intracellularly active molecule, as well as two functions, aborting or triggering cell death. One wonders whether the same site on the TGF-β1 molecule would be used in both instances.
The evidence above suggests that in some cases TGF-β1 may act intracellularly to protect cells from death, in the very cells that most probably synthesize it. If protection is intracellular, then where does it take place and how? Mitochondria involvement is appealing, since mitochondria are considered integrators for programmed cell death 18,19. It had previously been shown that TGF-β can be found in mitochondria 20,21. This observation is now confirmed and extended, using several techniques, such as organelle fractionation followed by Western blot analyses and ELISA tests, immunolocalization, and immunogold electron microscopy 8. What then could be the relationship between mitochondrial localization of TGF-β, increased cell death upon lack of TGF-β, and mitochondrial involvement in cell death? Mitochondrial morphology seems to be altered in TGF-β1 knockout cells 8. Also, in dying wild-type cells mitochondria show modifications of their membrane potential, believed to be essential at least in some cases for the release of several proapoptotic factors. Cell death induced through the lack of TGF-β also shows a decrease in mitochondrial membrane potential 8. This is regulated by members of the Bcl-2 family such as the antiapoptotic Bcl–XL molecule. While in wild-type T cells costimulation through the TCR leads to an increase of Bcl–XL mRNA, in TGF-β1 knockout T cells there is no such increase.
Although all these arguments are consistent with a role for TGF-β1 within mitochondria, they are only correlative. In particular, it is difficult to understand what relationship may exist between the presence of TGF-β1 in mitochondria and the level of Bcl–XL mRNA in the cytoplasm; the latter would be more readily explained via the classical effect of TGF-β receptors on Smad transcription factors. Also, even when externally applied, TGF-β1 induces hepatocyte apoptosis through the mitochondrial/Apaf-1 caspase activation pathway 22, downregulates Bcl-XL in a rat prostate epithelial cell line which decreases its death 23, and induces cell death in a variety of cell types in a manner which is enhanced by the mitochondrially located protein ARTS 24, showing that TGF-β1 need not reside within the mitochondria to induce cell death through a Bcl–XL- and ARTS-regulated mitochondrial pathway. True, compared with the work of Chen et al., the latter experiments did not deal with T cells, and externally applied TGF-β induced rather than inhibited cell death. In the situation described by Chen et al., TGF-β1 knockout cells were more sensitive to death, showing that TGF-β in this case protects from cell death 8. It could be that externally applied TGF-β induces cell death by affecting mitochondria indirectly, while TGF-β located in mitochondria might directly protect from cell death. However, it is not excluded that TGF-β1, even if acting within the cell, exerts its protective role primarily elsewhere than in the mitochondria. TGF-β can indeed be located also in other subcellular sites 21.
Altogether, the current work of Chen et al. 8 demonstrates that TGF-β1 strongly protects T lymphocytes from cell death and provides provocative results possibly assigning this protection to intracellular TGF-β1. It is tempting to hypothesize that at least some of the protected cells are those that produce TGF-β1.