Mimicking Cdk2 phosphorylation of Bcl-xL at Ser73 results in caspase activation and Bcl-xL cleavage

Cisplatin is a widely used chemotherapeutic agent, yet its efficacy is limited by nephrotoxicity. The severity of nephrotoxicity is associated with the extent of kidney cell death. Previously, we found that cisplatin-induced kidney cell death was dependent on Cdk2 activation, and inhibition of Cdk2 protected cells from cisplatin-induced apoptosis. Using an in vitro kination assay, we showed that Cdk2 phosphorylated Bcl-xL, an anti-apoptotic member of Bcl-2 family proteins, at serine 73. We also found that this phosphorylated Bcl-xL participated in cell death, as a phosphomimetic mutant of Bcl-xL at the serine 73 site (S73D-Bcl-xL) activated caspases. We now find that S73D-Bcl-xL was cleaved at D61 and D76, which are putative caspase cleavage sites, to generate 15-kDa and 12-kDa fragments. Unlike full-length Bcl-xL, these cleavage products of Bcl-xL were previously reported to be pro-apoptotic. We sought to determine whether these Bcl-xL fragments were necessary for the induction of cell death by S73D-Bcl-xL. Mutation of these caspase cleavage sites prevented the formation of the 15-kDa and 12-kDa Bcl-xL cleavage products, but apoptosis still persisted in a S73D modified Bcl-xL. Our findings show that Cdk2 phosphorylation of Bcl-xL at Ser73, but not the Bcl-xL cleavage products, is necessary and sufficient to induce cell death.


INTRODUCTION
Cis-diamminedichloroplatinum (II) or cisplatin has been widely used to treat various solid tumors, such as ovarian, testicular, head and neck, lung, and bladder cancers. 1 In addition to causing cancer cell death, cisplatin also induces kidney cell death, and this nephrotoxicity is a major dose limiting factor of cisplatin in chemotherapy. 2 Understanding the molecular mechanisms of cisplatin nephrotoxicity is important to find therapy to protect kidneys without compromising its anti-neoplastic effect.
Cyclin-dependent kinase 2 (Cdk2) is primarily known as a cellcycle regulatory protein that participates in cell proliferation. Apart from its role in cell-cycle regulation, its role in cell death pathways is poorly understood but has become increasingly recognized. [3][4][5][6][7] Our laboratory found that cisplatin-induced kidney cell death was dependent on Cdk2 activation, and inhibition of Cdk2 by p21, dominant-negative (DN)-Cdk2, or chemical inhibitors such as roscovitine or purvalanol protected kidney cells from cisplatininduced apoptosis both in vitro and in vivo. [8][9][10] However, the mechanisms of how Cdk2 mediates apoptosis are still unknown. Using an analog-sensitive Cdk2 kinase (F80G-Cdk2), we found that after cisplatin treatment, Cdk2 phosphorylated Bcl-xL, a cell survival protein, at a previously unreported Ser73 site. 11 Although wild-type Bcl-xL (WT-Bcl-xL) inhibits pore formation induced by active Bax/Bak, 12-14 a phosphomimetic Bcl-xL in which Ser73 was replaced by Asp promoted mitochondrial permeabilization, caspase activation, and apoptosis. 11 WT-Bcl-xL was reported to be cleaved after Asp61 and Asp76 (D61 and D76) by caspases. 15,16 The role of Bcl-xL cleavage in apoptosis is not well understood. It is unclear whether Bcl-xL cleavage results in loss of anti-apoptotic function, 16 or whether the cleavage products could actively participate in apoptosis. 15,17,18 In the present study, we report that Cdk2 phosphorylation of Bcl-xL at Ser73 is an upstream event resulting in perinuclear mitochondrial clustering, caspase activation, and subsequent Bcl-xL cleavage after D61 and D76. Furthermore, we show that Ser73 phosphorylation of Bcl-xL, but not Bcl-xL cleavage is necessary and sufficient to trigger apoptosis.

RESULTS
Expression of phosphomimetic Bcl-xL at Ser73 resulted in 15 kDa and 12 kDa cleavage products from caspase activity Using western blot analysis, we determined the expression of Bcl-xL before and after transduction ( Figure 1a). In comparison with the level of endogenous full-length Bcl-xL in untransduced cells (Figure 1a Cleavage of endogenous Bcl-xL is dependent on active caspases downstream of Cdk2 activation As the S73D-Bcl-xL cleavage products were eliminated by the pancaspase inhibitor, zVAD-fmk (Figure 1a, lane 5), we hypothesized that the cleavage of endogenous Bcl-xL observed after cisplatin exposure would similarly be prevented by zVAD-fmk. The cleavage of Bcl-xL was not observed in the membrane fraction of the untreated control cells (Figure 3a Cleavage-resistant phosphomimetic Bcl-xL at Ser73 preserved its ability to induce apoptosis WT-Bcl-xL was reported to be cleaved by caspases after D61 and D76. 15,16 To test whether S73D-Bcl-xL was cleaved at these residues, site-directed mutagenesis was performed to convert aspartic acids at 61 and 76 into alanines (that is, D61A and D76A), resulting in a triple mutated protein (D61A, S73D, and D76A) referred to hereafter as ADA-Bcl-xL. Caspase 3 activation was detected in S73D-Bcl-xLand ADA-Bcl-xL-transduced cells (   ADA-Bcl-xL and S73D-Bcl-xL induce perinuclear clustering of mitochondria Mitochondrial morphology and subcellular distribution change during apoptosis; mitochondria are dispersed throughout the cytoplasm in normal cells, but are condensed and clustered in perinuclear regions in apoptotic cells. 21 Using fluorescent microscopy, we determined the subcellular distribution of mitochondria in cells expressing WT-Bcl-xL, S73D-Bcl-xL, and ADA-Bcl-xL, using 4',6-diamidino-2-phenylindole (DAPI) and Mitotracker Red CMXRos to mark nuclei and mitochondria, respectively. Cells expressing transduced Bcl-xL were identified by GFP fluorescence, as GFP was co-expressed in the adenoviral expression vectors. Mitochondria were dispersed throughout the cytoplasm in WT-Bcl-xL-expressing cells similar to untransduced cells (cells without GFP marker), but were clustered in the perinuclear region in S73D-Bcl-xL-and ADA-Bcl-xL-expressing cells ( Figure 5), consistent with their localization in the early stages of apoptosis.

DISCUSSION
Bcl-xL contains an unstructured loop domain (amino acids, 26-83), which is a regulatory region of Bcl-xL, 22 known to be modulated by phosphorylation 23,24 and cleavage. 15,16 Recently, we reported that in normal kidney cells, Bcl-xL was phosphorylated at Ser73 within the unstructured loop domain in response to cisplatin exposure. 11 Ser73 phosphorylation converted Bcl-xL into a potent pro-apoptotic pore forming molecule, which oligomerized at the mitochondria to release cytochrome c and activate caspase 3. 11  In response to cisplatin exposure, endogenous Bcl-xL was cleaved into fragments similar to S73D-Bcl-xL fragments, suggesting that Bcl-xL cleavage is a relevant biological process in cisplatin-induced apoptosis. However, unlike the S73D-Bcl-xLexpressing cells (Figure 1a, lane 4 and Figure 4b, lane 3), cells exposed to cisplatin generated the 12-kDa cleavage product in higher quantity than the 15-kDa cleavage product (Figure 2, lane 4 and Figure 3a, lane 2). Such differences could be explained by the fact that S73D-Bcl-xL induced cell death represents only a subset of the many cell death pathways induced by cisplatin. It is possible that proteases that process Bcl-xL into the 12-kDa cleavage product were more activated in cisplatin-treated cells than in S73D-Bcl-xL-expressing cells.
As pan-caspase inhibitor, zVAD-fmk, prevented the endogenous cleavage of Bcl-xL (Figure 3a, lane 3), the cleavage of Bcl-xL is downstream of caspase activation. Cdk2 activation is upstream of caspase activation because inhibition of Cdk2 either by DN-Cdk2 (Figures 3a and b, lane 5) or by purvalanol (Figures 3a and b, lane 7) prevented caspase activation (caspase 3) and subsequent Bcl-xL cleavage. Expression of phosphomimetic S73D-Bcl-xL resulted in caspase activation, 11 suggesting that Cdk2 phosphorylation of Bcl-xL at Ser73 is upstream of caspase activation, but downstream of Cdk2 activation.
The 15-kDa cleavage product of Bcl-xL has a similar molecular weight as Bcl-xS, a proapoptotic isoform of Bcl-xL, and we investigated whether the observed cleavage product could be Bcl-xS. Bcl-xS is produced from Bcl-x gene with BH1 and BH2 domains truncated due to alternative splicing of Bcl-x mRNA. 25 On the other hand, Bcl-xL cleavage products are produced post-translationally by activated caspases, resulting in the loss of the BH4 domain. 15,16 Since the formation of both the 15-kDa and 12-kDa Bcl-xL fragments could be prevented by zVAD-fmk, we concluded that those fragments were Bcl-xL cleavage products rather than Bcl-xS.
Mitochondria are normally distributed throughout the entire cytoplasm. 21 Perinuclear mitochondrial clustering has been reported in apoptotic cells induced by various stimuli such as TRAIL, 26 exogenous expression of t-Bid 27 and Bax, 21 aluminum maltolate, 28 and staurosporin. 28 Although perinuclear mitochondrial clustering is a phenotype consistently associated with apoptosis, it is still not clear whether mitochondrial clustering facilitates cell death. It is speculated that condensation of mitochondria could mediate the release of apoptogenic factors from mitochondria, and perinuclear localization hastens the delivery of pro-apoptotic molecules from the mitochondria into the nucleus, thereby increasing the rate of apoptosis induction. 21 Here, we show that ADA-Bcl-xL induces perinuclear clustering of mitochondria, similar to S73D-Bcl-xL ( Figure 5). This suggests that Cdk2 phosphorylation of Bcl-xL was sufficient to cause pBcl-xL at Ser73 activates caspases to cleave Bcl-xL NS Seng et al mitochondrial redistribution and caspase activation, independent of Bcl-xL cleavage. Our data suggest that Cdk2 phosphorylation of Bcl-xL at Ser73 is an important part of cisplatin-induced kidney cell death pathways ( Figure 6). Caspases, activated by phosphomimetic S73D-Bcl-xL, proteolyzed Bcl-xL into the 15-kDa and 12-kDa cleavage products. Elimination of the caspase cleavage sites resulted in the absence of the Bcl-xL cleavage products, but did not prevent mitochondrial redistribution, caspase activation, and subsequent apoptosis, indicating that the cleavage products were not essential for apoptosis induction. However, the differences in the intensities of the Bcl-xL cleavage products between cisplatin-induced apoptosis and S73D-Bcl-xL-induced apoptosis suggest that the mechanisms of how the cleavage products are generated could be regulated by more than one pathway. It is known that cisplatin triggers multiple cell death pathways, 2,29 and Cdk2 phosphorylation of Bcl-xL represents one of these modes. Further investigations on how the cleavage products are generated, especially in cisplatin-treated cells, could offer insights into mechanisms of cisplatin-induced cell death.
Cell-cycle analysis TKPTS cells were trypsinized and collected by centrifugation at 800 × g for 7 min. The cells were processed for propidium iodide (PI) staining as previously described. 8 Cell-cycle analysis was performed using FACSCalibur for 10 000 cells per experiment, and analyzed by FlowJo (Ashland, OR, USA). Percent apoptosis was quantified by the cell population in sub G0/G1 region 31  Histone kinase assay Untransduced TKPTS cells and TKTPS cells either transduced with 100 MOI DN-Cdk2 adenovirus for 24 h or treated with 9 μM purvalanol for 1 h were lysed in NP-40 lysis buffer (150 mM NaCl, 50 mM Tris pH 7.5, 1% NP-40 and 0.5 mM EDTA) supplemented with protease inhibitor cocktail. In vitro kinase assay was performed using histone H1 (Upstate Biotechnology, Billerica, MA, USA) as a Cdk2 substrate as previously described. 32 Subcellular fractionation Cells were harvested using CHAPS lysis buffer (50 mM HEPES, pH 7.4, 5 mM CHAPS, and 5 mM DTT) supplemented with protease inhibitor cocktail. Cell lysate was homogenized by douncing and centrifuged at 16 000 × g to separate membranes from supernatant. To analyze membrane-associated proteins, the membrane fraction was further sonicated in RIPA lysis buffer, supplemented with protease inhibitor cocktail.

Site-directed mutagenesis
Site-directed mutagenesis was performed by using the Q5-Site-directed mutagenesis kit (E0554S, New England Biolabs, Ipswich, MA, USA) according to the manufacturer's protocol. D61A mutagenesis was generated by using the forward and reverse primers 5′GCACCTGGCGGCTA GCCCGGCCG3' and 5′CAGGATGGGTTGCCATTGATGGCACTGG3′, respectively. D76A mutation was generated by the forward and reverse primers, 5′CAGCAGTTTGGCTGCGCGGGAGG3′ and 5′CTGTGGCCAGTGGCTCCATTC3′, respectively. To create D61A, D76A double mutation, D61A mutation was created after D76A mutation was performed in S73D-Bcl-xL plasmid. All mutations were confirmed by DNA sequencing, performed at the University of Arkansas for Medical Sciences core facility.
Fluorescent microscopy TKPTS cells transduced for 48 h with WT-Bcl-xL, S73D-Bcl-xL, and ADA-Bcl-xL were labeled with 100 nM Mitotracker Red CMXRos (Molecular Probes, Eugene, OR, USA) for 1 h at 37 ºC, fixed, and then processed as previously described. 11 DAPI-containing mounting media (Vector, Burlingame, CA, USA, H1500) was used to label nuclei. Images were acquired using Nikon Eclipse Ti-S (Nikon, Melville, NY, USA), apochromatic × 60 oil lens and NIS Elements imaging software (Nikon). Figure 6. The proposed model of Cdk2 phosphorylation of Bcl-xL in cisplatin-induced apoptosis. Cdk2 activated in response to cisplatin treatment phosphorylated Bcl-xL at Ser73. Bcl-xL, when phosphorylated at Ser73, promoted self-oligomerization to engage in caspasedependent mitochondrial pathway of apoptosis. 11 In the process, active capases cleaved Bcl-xL at previously reported sites, D61 and D76. However, the resulting cleavage products of Bcl-xL were not necessary for apoptosis. Phosphorylation of Bcl-xL at Ser73 was sufficient to induce cell death. Inhibition of Cdk2 using DN-Cdk2 or purvalanol and inhibition of active caspases using zVAD-fmk blocked caspase activation, and subsequent Bcl-xL cleavage.

Adenovirus
Adenoviruses for WT-Bcl-xL, S73D-Bcl-xL, and DN-Cdk2 used in this study were generated as described. 10,11,33 Using a similar strategy, ADA-Bcl-xL adenovirus was generated for this study. In brief, site-directed mutagenesis was performed to generate D61A and D76A in the S73D-Bcl-xL plasmid as described above. The cDNA fragment containing the triple mutation (D61A, S73D, and D76A) was excised from the plasmid using restriction enzymes HindIII and EcoRI to replace a similar fragment in pAd-Track-CMV-WT-Bcl-xL. ADA-Bcl-xL adenovirus was generated in AD-293 cells, amplified in HEK-293 cells, and purified using CsCl banding as described. 34 Statistical analysis All quantitative measures were expressed in mean ± S.E.M. Probabilities of statistical significance were determined using the two-tailed t-test for independent variables with data from at least three separate experiments. Statistical significance was defined as Po0.05.