In this study we demonstrate a direct link between pro-apoptotic TGF-β signaling and transcriptional up-regulation of the apoptosis activator PUMA in c-Myc-driven B-cell lymphomas. PUMA was required for the effective, early induction of apoptosis in both human and murine lymphomas, but its induction does not require c-Myc activity. The involvement of PUMA in this process was demonstrated by stable knockdown of PUMA in human BL, but to be confident that we could assess the effect of complete lack of PUMA on TGF-β-induced apoptosis, we extended our analysis to compare Eμ-Myc
lymphomas derived from Puma
null and wild type mice. We believe this is the first demonstration that murine Eμ-Myc
lymphomas are susceptible to TGF-β-induced apoptosis. Stroma-derived TGF-β from infiltrating immune cells has previously been reported to mediate a tumor suppressor function in vivo
by inducing senescence of Eμ-Myc
lymphoma cells (32
). Our data suggest that in addition to senescence, TGF-β expressed in the tumor microenvironment would also be likely to induce ALK5-dependent apoptosis of malignant cells.
We report that a complete loss of PUMA in knock-out Eμ-Myc
cells and knockdown of PUMA in human BL cells delay cell death but do not ultimately prevent cells from undergoing apoptosis. This observation is consistent with the idea that PUMA has a role in priming cells for apoptosis as part of a polygenic apoptotic response that may have a role in “sensing” the apoptotic signal (33
). As well as PUMA activation, the apoptotic program involves the early activation of BIK in human cells followed by down-regulation of BCL-XL
). Such changes in expression levels of the other BCL-2 family members presumably are sufficient to induce apoptosis, albeit with slower kinetics. As predicted previously, we have seen no TGF-β-dependent activation of BIK in mouse cells because the Smad-binding element identified within the human BIK promoter is not conserved (Ref. 3
and ). BCL-XL
, however, is down-regulated in response to TGF-β in murine Eμ-Myc
This suggests that loss of BCL-XL
function may ultimately be sufficient in murine lymphoma to induce apoptosis. but we cannot exclude the possibility that other BH3-only sensitizers might also be induced in response to TGF-β in the mouse.
The mechanism of PUMA induction by TGF-β in this system requires some further analysis. We have determined that PUMA is induced even in the presence of protein synthesis inhibitors, demonstrating that de novo protein synthesis is not required for PUMA transcriptional up-regulation in BL cells and also that in vivo, activated Smad3/Smad4 is recruited directly to a SBR within the endogenous promoter. Constitutive activity of the isolated putative Smad-binding region was dependent on consensus SBE sequences in transient transfection assays, and it was partially inhibited by blocking endogenous TGF-β signaling and lowering Smad4 levels; however, exogenous addition of ligand did not induce its activity. These data suggest that chromatin structure and/or transcriptional repressors may be important for correct regulation of the endogenous gene.
The induction of PUMA-mediated apoptosis by TGF-β superfamily members has been described previously in oligodendrocytes (induced by activin A but not by TGF-β) and by TGF-β in a gastric cancer cell line. In both cases, however, elevated PUMA levels were dependent on p53 family members. A p53 family inhibitor pifithrin-α blocked activin A-induced apoptosis of oligodendrocytes (34
), whereas TGF-β-induced apoptosis of SNU-16 gastric cancer cells was due to the induction of p73, which presumably bound p53 consensus sequences on the PUMA
). In these studies, there was no evidence that Smads were recruited directly to the PUMA
promoter as detected in BL cells. However, it is possible that a p53 family member might still be involved in TGF-β-induced regulation of PUMA in our system by acting as co-activators with the Smad proteins. Smads are regulated by numerous post-translational modifications and cooperate with many other transcription factors (including the p53 family) to regulate gene transcription, thus ensuring that their function is highly context dependent (6
). p53 and p73, for example, directly bind Smads 2 and 3 and can stabilize Smad-DNA complexes on specific gene promoters (36
). It would seem unlikely that p53 is involved in TGF-β/Smad activation of the PUMA
promoter because all but one of the BL cell lines used in this study express mutant p53 (BL2 cells express wild type p53 (37
)). Transcripts of p73 are expressed in BL cells but are not increased by TGF-β treatment.6
So far, direct activation of PUMA
by TGF-β has been demonstrated in B-cell subsets (lymphomas and human primary centroblasts) but not in murine embryonic fibroblasts or HaCaT epithelial cells. The reason for the selective response in certain cell types is currently unclear. It is possible that the proteome of B-cells uniquely allows PUMA activation in response to TGF-β. The panel of BL cells examined here lack expression of SLUG, a transcription factor involved in the repression of PUMA transcription.6
Expression of SLUG in hematopoietic progenitor cells is responsible for protecting progenitor cells from DNA damage-induced apoptosis by repressing p53-mediated PUMA
). TGF-β may be able to induce PUMA
in B-cells by virtue of its lack of repression. Another possibility is that PUMA
is regulated by a B-cell-specific transcription factor in conjunction with TGF-β-activated Smads.
A study published during the preparation of this manuscript implicates the activation status of B-cells as a factor in PUMA production, with PUMA being involved in the regulation of antigen specific memory B-cells both in vitro
and in vivo
. PUMA protein expression was detected in histological sections of human lymph nodes, potentially as a result of antigenic stimulation, because stimulation of cells in vitro
with the mitogens Staphylococcus aureus
Cowan or LPS, or alternatively following ligation of the B-cell receptor or CD40, all resulted in up-regulation of PUMA. Provided sufficient pro-survival signals are received to maintain BCL-XL
levels, the effect of PUMA induction in this system is negated (26
). The mechanism of PUMA induction was not identified by Clybouw et al.
) but was determined to be independent of p53 because p53−/−
B-cells produced similar levels of PUMA to wild type cells.
Given our finding that PUMA
is induced in vitro
by TGF-β treatment of centroblasts () and that active, phosphorylated Smad2 is detectable throughout the germinal center (3
), we can speculate that PUMA detected in the tonsil sections could result from TGF-β signaling. This may arise as a result of the establishment of an activation-induced TGF-β-autocrine feedback loop (39
). There are several lines of evidence to support this hypothesis. TGF-β mRNA is detectable in both resting and activated B-cells (40
). Stimulation of human tonsil cells with S. aureus
Cowan does not significantly affect TGF-β mRNA levels but does result in a 7-fold increase in TGF-β protein secretion (41
). Similarly, LPS treatment of murine splenocytes induces the production of bioactive TGF-β (42
). The induction of TGF-β in activated B-cells is thought to limit the expansion of the activated population. Without this autoregulatory feedback loop severe autoimmunity develops (43
). Our data presented here indicate that regulation of PUMA plays a role in the apoptotic response of B-cells exposed to TGF-β and suggest that this may play a role in the efficient control of activated B-cell proliferation. This notion may provide an explanation for the elevated levels of IgA observed in Puma
null mice (45
) because IgA class switching is promoted by TGF-β (46
). The intriguing possibility that activation-induced PUMA production occurs because of an autocrine TGF-β/ALK5/Smad pathway and the relative role, mechanisms, and kinetics of regulation of BCL-XL
and other BCL-2 family members in B-cell survival warrants further investigation.