Caspases promote cell death either by the cleavage-dependent inactivation of survival factors or by the cleavage-dependent activation of proapoptotic factors (Cryns and Yuan, 1998
). The caspase-dependent turn-off of survival pathways can be attempted at a transcriptional level by the cleavage of transcription factors regulating the expression of prosurvival genes (Fischer et al., 2003
). Processing of transcription factors generally results in a drastic decrease of their activity, which suggests that caspases can act as transcriptional corepressors.
HDACs are important regulators of gene expression as part of transcriptional corepressor complexes (Grozinger and Schreiber, 2002
; Peterson, 2002
; Verdin et al., 2003
). Here, we report that caspases can selectively modulate HDACs function during apoptosis. Our evidence demonstrates that HDAC4, but not HDAC1, 2, 3, and 6, is specifically cleaved by caspase-2 and -3 in vitro, whereas HDAC5 is cleaved in vitro by caspase-3, but at a much lower extent with respect to HDAC4.
Experiments in MCF-7 cells expressing caspase-3/CI or caspase-3/WT, and IMR90-E1A cells expressing C9DN, indicate that the apoptosome and the subsequent activation of caspase-3 play the major role in HDAC4 processing after genotoxic stress.
HDAC4 belongs to the class IIa of HDACs and shows the highest expression in heart, skeletal muscle and brain. HDAC4 is part of large multiprotein complexes that mediate its recruitment to specific promoters () (Verdin et al., 2003
). HDAC4 interacts with the MEF2 family of transcription factors and with SRF through the amino-terminal region (Miska et al., 1999
; Youn et al., 2000
; Davis et al., 2003
). In addition, the amino-terminal region is also involved in the binding of the transcriptional repressor C-terminal-binding protein (CtBP) and of BCL6, a sequence-specific transcriptional repressor that is involved in the pathogenesis of non-Hodgkin's B-cell lymphomas (Zhang et al., 2001
; Lemercier et al., 2002
; Verdin et al., 2003
Figure 9. Schematic representation of the effect of caspase activation on HDAC4 and its partners. Cytoplasmic relocalization of HDAC4 carboxy-terminal fragment could trigger the dissociation from the SMRT/N-CoR–HDAC3 complex (Fischle et al., 2002 ).
HDAC4 interacts with two closely related corepressors, silencing mediator for retinoid and thyroid receptor (SMRT) and nuclear receptor corepressor (N-CoR), through the carboxy-terminal region, including the HDAC domain (Huang et al., 2000
). HDAC4 is enzymatically inactive; however, a deacetylase activity arises from the presence of HDAC3 in the SMRT/N–CoR complex (Fischle et al., 2002
). Further HDAC4 partners are represented by heterochromatin protein 1, which mediates transcriptional repression by recruiting histone methyltransferase (Zhang et al., 2002
) and by the recent described p53BP1 (Kao et al., 2003
HDAC4, similarly to other class IIa members, can shuttle in a regulated manner between the nucleus and the cytoplasm, and the phosphorylation-dependent binding to 14-3-3 proteins mediates their cytoplasmic localization (Grozinger and Schreiber, 2000
; McKinsey et al., 2000
; Wang et al., 2000
; Miska et al., 2001
; Zhao et al., 2001
). The Ca2+
/calmodulin-dependent kinase and a still unidentified kinase can mediate phosphorylation of HDAC4, 5, 7, and 9 and promote their nuclear export (Verdin et al., 2003
We demonstrate that HDAC4 nuclear/cytoplasmic shuttling is regulated by caspases. Caspase-dependent cleavage of HDAC4 occurs at Asp 289 and provokes the separation of the amino-terminal region, including the MEF binding sequence and the NLS, from the carboxy-terminal region that includes the HDAC domain and the NES. The corresponding amino-terminal and carboxy-terminal fragments show an exclusively nuclear and cytoplasmic localization, respectively.
Class II HDACs play multiple biological roles being involved in the myogenesis, in the negative selection of thymocyte, in the regulation of Epstain-Barr virus, and probably in neuronal survival (Verdin et al., 2003
). The relationships between the subcellular localization of class IIa HDACs and their biological functions have been characterized more in detail during myogenesis. HDAC4, similarly to HDAC5 and 7, when overexpressed in C2C12 cells can accumulate into the nucleus where it represses MEF2-dependent transcription and muscle differentiation (Lu et al., 2000
; Miska et al., 2001
). However, HDAC4 is mainly cytosolic in proliferating C2C12 cells and relocalizes to the nucleus in myotubes, whereas HDAC5 is prevalently nuclear in myoblasts and translocates into the cytoplasm when cells differentiate (Miska et al., 2001
; Zhao et al., 2001
We similarly observed that HDAC4 is mainly cytosolic in IMR90-E1A and in U2OS cells. In stably transfected U2OS cells, GFP-HDAC4 showed a cytosolic localization in ~70% of the cells; but during UV-induced apoptosis, it translocates into the nucleus in a caspase-dependent manner, coincidentally/immediately before the retraction response, but clearly before nuclear fragmentation. Overall, our data suggest that HDAC4 translocation into the nucleus during cell death is dependent on a caspase cleavage of the amino-terminal region and that this cleavage is an early event during the execution phase of the apoptotic program.
When ectopically expressed, the amino-terminal fragment of HDAC4 (HDAC4ΔC) induced apoptosis by activating the mitochondrial pathway, as demonstrated by the dependence on caspase-9, the release of cytochrome c
, and the processing of PARP. Class IIa HDACs contain multiple, independent repressive domains (Verdin et al., 2003
). The amino-terminal region of HDAC4 but also that of HDAC-7 and HDAC-9/MITR are able to repress transcription in the absence of their deacetylase domains (Dressel et al., 2001
; Wang et al., 1999
; Sparrow et al., 1999
; Chan et al., 2003
). We observed that the proapoptotic function of HDAC4ΔC was correlated to an efficient repression of MEF2C transcriptional activity. Surprisingly, HDAC4ΔC showed a stronger MEF2-repressive activity compared with the full-length HDAC4. Different explanations can be evoked: 1) HDAC4ΔC shows an exclusively nuclear localization, whereas HDAC4 was nuclear in ~60% of the cells. 2) The subnuclear localization of the two proteins was different because HDAC4-FLAG, when localized into the nucleus, was present in speckle-like structures (Wu et al., 2001
), whereas HDAC4ΔC-FLAG showed a diffuse nuclear staining. 3) From our studies, it seems that HDAC4 lacking the carboxy-terminal region is more stable of the full-length protein. It is possible that all these aspects account for the increased repressive activity of the caspase-cleaved amino-terminal segment of HDAC4.
Removal of aa 166-289 from HDAC4ΔC abolished its repressional activity and the proapoptotic function. These data further support the link between the repressional and the cell death activities of HDAC4. Curiously, the deletion HDAC4ΔC/Δ166-185, which should remove the previously reported MEF2 binding site (Wang and Yang, 2001
), was still able to repress MEF2C transcription. Repression of MEF2C by the HDAC4 fragments lacking this putative binding site has already been observed (Wang et al., 1999
). Two possible explanations can be evoked: 1) An additional MEF2C binding site could exist in HDAC4ΔC. 2) It could be possible that HDAC4ΔC/Δ166-185 is recruited, through different interactors, to MEF2 promoter independently from the binding to MEF2. Interestingly within the amino-terminal region of HDAC4 an oligomerization domain has been mapped (Wang and Yang, 2001
; Kirsh et al., 2002
). Therefore, HDAC4ΔC/Δ166-185 could interact with endogenous HDAC4 and be recruited in the complex containing MEF2C.
Diverse cellular decisions are controlled by the MEF2 family of transcription factors in different cell types (McKinsey et al., 2002
), including proapoptotic (Youn et al., 1999
) and prosurvival functions (Mao et al., 1999
). Protection from apoptosis has been observed in postmitotic neurons. RNA interference on MEF2A, revealed a critical role of this factor in the neuronal activity-dependent survival of granule neurons (Gaudilliere et al., 2002
). Moreover, when the same cells were challenged to apoptosis by K+
withdrawal, MEF2A and MEF2D underwent a caspase-mediated processing (Li et al., 2001
). Caspase-dependent processing of MEF2 family members has also been investigated in mature cerebrocortical neurons in response to excitotoxic insults. In this cellular system, the caspase-dependent cleavage of MEF2 transcription factors generates fragments that act in a dominant-interfering manner to abrogate MEF2-dependent neuroprotection (Okamoto et al., 2002
). Similarly caspase-dependent cleavage of SRF, another prosurvival factor that is regulated by HDAC4, inhibits its transcriptional activity (Bertolotto et al., 2000
; Drewett et al., 2001
This evidence suggests that caspases act in coordinated manner to suppress the survival pathways regulated by SRF and MEF2 transcription factors both in “cis” by the direct cleavage of the transcription factors and in “trans” by regulating HDAC4 function. In this scenario, when a limited level of caspases is activated by mitochondria, HDAC4 cleavage could sustain the apoptotic signal by acting on mitochondria in a sort of amplificatory loop.
Interestingly ectopically expressed HDAC5 promotes apoptosis (Huang et al., 2002
), and in a transgenic mouse model the inducible expression of a signal resistant form of HDAC5 in cardiomyocytes resulted in sudden death of the mice. Cardiomyocytes death and dramatic changes in mitochondrial morphology were observed (Czubryt et al., 2003
). MEF2 and HDAC5 regulate in an opposite manner peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) expression, a master regulator of mitochondria biogenesis (Puigserver et al., 1998
; Wu et al., 1999
). Because HDAC4ΔC controls cell survival through the mitochondrial pathway, it will be interesting to evaluate whether PCG-1α expression is modulated by HDAC4ΔC and whether PCG-1α can counteract apoptosis induced by this deletion.