Protein kinase C-δ (PKCδ) plays an important role in DNA damage–induced apoptosis. We have previously shown that the PKCδ inhibitor rottlerin protects against cisplatin-induced apoptosis acting upstream of caspase-9. In the present study, we have investigated if rottlerin regulates caspase-2 activation. Knockdown of caspase-2 by siRNA inhibited processing of apical caspase-9 and caspase-8, whereas depletion of caspase-9 had little effect on caspase-2 processing. Rottlerin inhibited activation and processing of caspase-9 and caspase-8 and cleavage of poly(ADP)ribose polymerase. We made a novel observation that rottlerin induced down-regulation of caspase-2 but not of caspase-3, caspase-7, caspase-8, or caspase-9. Pharmacologic inhibitors of PKC, such as Gö 6983 and bisindolylmaleimide, or depletion of PKCδ by siRNA had no effect on the down-regulation of caspase-2 by rottlerin. The proteasome inhibitor MG132 reversed caspase-2 down-regulation by rottlerin, whereas calpain inhibitor had no effect. These results suggest that rottlerin induces down-regulation of caspase-2 via PKCδ-independent but ubiquitin proteasome–mediated pathway. Furthermore, down-regulation of caspase-2 by rottlerin can explain its antiapoptotic function during DNA damage–induced apoptosis. [Cancer Res 2008;68(8):2795–802]
Apoptosis is a genetically determined cell suicidal program that removes unwanted, redundant, and damaged cells and is required to maintain a balance between cell proliferation and cell death (1). Apoptosis is initiated by the activation of caspases, a family of interleukin-1β converting enzyme–like cysteine proteases that specifically cleave proteins after aspartate residues (2–4). Whereas caspase-8 and caspase-9 participate in the initiation phase of apoptosis, caspase-3, caspase-6, and caspase-7 are involved in the execution phase of apoptosis. Activation of these executioner caspases results in the cleavage of critical cellular proteins, including poly(ADP-ribose) polymerase (PARP), DNA-dependent protein kinase, lamin B, and protein kinase C-δ (PKCδ; refs. 3, 5). Caspase-2 is a unique caspase that can serve both as initiator and executioner caspase (6–10).
There are two major pathways of cell death. The extrinsic pathway is triggered by binding of ligand to members of tumor necrosis factor-α receptor superfamily, causing activation of procaspase-8 followed by activation of a caspase cascade (11). DNA-damaging agents are known to activate the intrinsic or mitochondrial pathway by inducing the release of mitochondrial cytochrome c, which allows Apaf-1 to interact with the initiator procaspase-9 to form an active apoptosome complex, resulting in activation of effector caspases, such as caspase-3 and caspase-7 (4). Recent studies from various laboratories have established a role for caspase-2 during genotoxic stress–induced apoptosis (7, 12–25).
PKCδ is a substrate for caspase-3, and it is believed that proteolytic activation of PKCδ is necessary for DNA damage–induced apoptosis (26–29). We have previously shown that rottlerin, a pharmacologic inhibitor of PKCδ, inhibits activation of caspase-3 and caspase-9 induced by the DNA-damaging anticancer agent cisplatin (30, 31), suggesting that PKCδ acts at an early stage of DNA damage–induced apoptosis that precedes activation of caspase-3 and caspase-9. Several studies showed that caspase-2 is activated early in response to DNA damage acting upstream of mitochondria, suggesting that it functions as an apical caspase (7, 13, 21, 32, 33). Because rottlerin inhibits cisplatin-induced processing of caspase-9, we wanted to know if it acts upstream of caspase-2. We made a novel observation that rottlerin-induced down-regulation of caspase-2 via ubiquitin proteasome–mediated pathway. Furthermore, the ability of rottlerin to induce caspase-2 down-regulation was independent of its ability to inhibit PKCδ.
Materials and Methods
Materials. Rottlerin, Gö 6983, bisindolylmaleimide (BIM) II, MG132, ALLN, PSI, calpeptin, and protease inhibitor cocktail were obtained from Calbiochem. Cisplatin was from Sigma. siRNA SMARTpool against caspase-2, PKCδ, and nontargeting SMARTpool siRNA were obtained from Dharmacon. Monoclonal antibody to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and polyclonal antibody to PKCδ were from Santa Cruz Biotechnology, Inc. Monoclonal antibody to caspase-2, PARP, and polyclonal antibody to caspase-9 were purchased from BD PharMingen. Mouse monoclonal antibody to caspase-2 was obtained from Imgenex Corp. and rat monoclonal caspase-2 antibody from Alexis. Polyclonal antibody to caspase-2 and caspase substrates were obtained from BioVision. Polyclonal antibody to caspase-3 and monoclonal antibody to caspase-8 were obtained from BioSource/Invitrogen. Horseradish peroxidase–conjugated goat anti-mouse and donkey anti-rabbit antibodies were obtained from Jackson ImmunoResearch Laboratories, Inc. Polyvinylidene difluoride (PVDF) membrane was from Millipore, and enhanced chemiluminescence detection kit was from Amersham. Lipofectamine 2000 transfection reagent was obtained from Invitrogen.
Cell culture. HeLa cells were maintained in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 2 mmol/L glutamine. Ovarian cancer 2008 cells were maintained in RPMI 1640 supplemented with 5% heat-inactivated FBS and 2 mmol/L glutamine. Cells were kept in a humidified incubator at 37°C with 95% air and 5% CO2.
Knockdown of caspase-2 and PKCδ. Control nontargeting siRNA SMART pool or siRNA SMART pool targeted against caspase-2 or PKCδ were introduced into HeLa or 2008 cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Briefly, cells were seeded 1 d before transfection. Forty-eight hours after siRNA transfection, cells were treated as indicated in the text and processed for Western blot analysis as described before (34, 35).
Immunoblot analysis. Equivalent amounts of total cellular extracts were separated by SDS-PAGE and transferred electrophoretically to PVDF membrane. Immunoblot analyses were performed as described previously (30).
Caspase assay. Cells were treated with rottlerin and cisplatin as indicated in the text. Caspase activity of cell extracts treated with or without cisplatin was determined at 37°C using 40 μmol/L caspase substrate z-VDVAD-AFC, z-IETD-AFC, or z-LEHD-AFC in 50 mmol/L HEPES (pH 7.4), 100 mmol/L NaCl, 10 mmol/L DTT, 1 mmol/L EDTA, 0.1% CHAPS, and 5% glycerol. The release of 7-amino-4-trifluoromethylcoumarin (AFC) was measured at excitation wavelength 400 nm and emission wavelength 505 nm using a SPECTRAMAX 340 microplate reader (Molecular Devices) and SOFTmax PRO software.
Rottlerin induces down-regulation of caspase-2. We have previously shown that rottlerin acts upstream of caspase-9 during cisplatin-induced apoptosis (31). Because caspase-2 has been shown to act as an apical caspase during DNA damage–induced apoptosis, we wanted to know if rottlerin acts upstream of caspase-2. Figure 1 shows that caspase-2 siRNA effectively depleted caspase-2 in HeLa cells and ovarian cancer 2008 cells but it did not decrease the levels of other caspases, such as caspase-8, caspase-9, (Fig. 1A), caspase-3, and caspase-7 (data not shown). Because proteolytic processing of caspase-2 is not essential for its activation (36), we also determined caspase activity using a fluorogenic substrate of caspase-2. Both constitutive and cisplatin-induced caspase-2 activity was attenuated in cells transfected with caspase-2 siRNA compared with cells transfected with control siRNA (Fig. 1B). Depletion of caspase-2 was associated with inhibition of cisplatin-induced proteolytic processing of caspase-8 and caspase-9, and cleavage of PARP and PKCδ in both HeLa and 2008 cells (Fig. 1A). The effect of caspase-2 knockdown on the inhibition of caspase activation and cleavage of PARP or PKCδ was more pronounced in HeLa cells compared with 2008 cells, presumably because reduction of caspase-2 level by siRNA was less in 2008 cells compared with HeLa cells. These results suggest that depletion of caspase-2 inhibits cisplatin-induced apoptosis.
Because caspase-9 is believed to be the initiator caspase during DNA damage–induced apoptosis, we compared how depletion of caspase-2 or caspase-9 affects cisplatin-induced processing of each other in HeLa cells (Fig. 1C). Although knockdown of caspase-2 decreased processing of caspase-8 and caspase-9, depletion of caspase-9 did not decrease the processing of these caspases as determined by the amount of cleaved caspases. Knockdown of either caspase-2 or caspase-9 decreased cisplatin-induced cleavage of PARP, although caspase-2 knockdown was more effective. These results suggest that caspase-2 acts upstream of caspase-9 during cisplatin-induced apoptosis.
We then examined the effect of rottlerin on cisplatin-induced caspase activation. We monitored processing of caspases by Western blot analysis (Fig. 2). Because activation of caspase-2 and caspase-9 may not require proteolytic processing (36), we analyzed samples from the same experiment for caspase activity using fluorogenic peptide substrates (Table 1). Pretreatment of 2008 cells with rottlerin inhibited cisplatin-induced activation of caspase-2, caspase-8, and caspase-9 (Table 1) and processing of caspase-8 and caspase-9 (Fig. 2). Interestingly, rottlerin induced down-regulation of caspase-2. Figure 3 shows that treatment of HeLa cells with rottlerin for 7 hours or longer induced down-regulation of caspase-2 but not of other caspases, such as caspase-3, caspase-7, and caspase-9. We have previously shown that cisplatin or rottlerin had no effect on caspase-2 using a caspase-2 antibody purchased from Transduction Laboratories (30). This antibody was later discontinued by the company. We therefore used several antibodies that recognize different epitopes to monitor down-regulation of caspase-2. These antibodies also showed decrease in caspase-2 in cells treated with caspase-2 siRNA (Fig. 1A and C; data not shown). As shown in Fig. 3, both mouse monoclonal antibody to caspase-2 obtained from BD PharMingen and rat monoclonal antibody from Alexis showed similar time course of caspase-2 down-regulation. Thus, the effect of rottlerin on caspase-2 down-regulation was not due to epitope masking.
|Substrate .||Fold stimulation|
|.||VDVAD-AFC (Cas-2) .||IETD-AFC (Cas-8) .||LEHD-AFC (Cas-9) .|
|CP (5 μmol/L)||2.1||0.8||1.2|
|CP (15 μmol/L)||12||2.8||6.4|
|Rot+CP (5 μmol/L)||0.8||0.2||0.4|
|Rot+CP (15 μmol/L)||2.3||0.7||1.2|
|Substrate .||Fold stimulation|
|.||VDVAD-AFC (Cas-2) .||IETD-AFC (Cas-8) .||LEHD-AFC (Cas-9) .|
|CP (5 μmol/L)||2.1||0.8||1.2|
|CP (15 μmol/L)||12||2.8||6.4|
|Rot+CP (5 μmol/L)||0.8||0.2||0.4|
|Rot+CP (15 μmol/L)||2.3||0.7||1.2|
The effect of rottlerin on caspase-2 down-regulation is independent of PKCδ inhibition. Because rottlerin is an inhibitor of PKCδ, we compared the effects of rottlerin with other inhibitors of PKC on caspase-2 down-regulation. Whereas rottlerin is believed to be a specific inhibitor of PKCδ (37), BIM, and Gö 6983 inhibit all PKC isozymes albeit with different potencies (38). cPKCs are most sensitive and can be inhibited at nanomolar concentrations, whereas nPKCs can be inhibited at submicromolar concentrations in in vitro kinase assays. As shown in Fig. 4A, rottlerin induced down-regulation of caspase-2 as well as proteolytic processing of PKCδ in both HeLa and 2008 cells. Rottlerin alone caused a slight increase in proteolytic cleavage of PKCδ and PARP. This is consistent with our previous report that rottlerin by itself could cause a small increase in cytochrome c release (31). The PKC inhibitors Gö 6983 or BIM caused a modest decrease in the proteolytic activation of PKCδ as determined by the levels of catalytic fragment (CF) of PKCδ. These inhibitors, however, had little effect on caspase-2 down-regulation in HeLa cells.
Because rottlerin is a pharmacologic inhibitor and it may act on additional targets besides PKCδ, we also examined if depletion of PKCδ by siRNA leads to down-regulation of caspase-2. siRNA against PKCδ caused efficient depletion of PKCδ in both HeLa and 2008 (Fig. 4B) cells but did not affect the levels of other PKC isozymes (data not shown). Knockdown of PKCδ failed to induce down-regulation of caspase-2 in these cells. In fact, depletion of PKCδ seems to increase caspase-2 level slightly. Furthermore, rottlerin induced down-regulation of caspase-2 in PKCδ-depleted cells (Fig. 4B and C). These results suggest that the effect of rottlerin on caspase-2 down-regulation was independent of PKCδ.
Because PKCδ functions as a proapoptotic protein and rottlerin is believed to inhibit PKCδ, we examined the consequence of PKCδ knockdown on the activation of caspases and cleavage of PARP in HeLa cells. Figure 4D shows that cisplatin caused a concentration-dependent increase in the processing of caspase-2, caspase-8, caspase-9, and caspase-3, and cleavage of PARP. PKCδ siRNA effectively depleted PKCδ in HeLa cells. Although cisplatin caused a concentration-dependent increase in the levels of catalytic fragment of PKCδ in control siRNA–transfected cells, we were unable to detect PKCδ-CF in cells transfected with PKCδ siRNA. Depletion of PKCδ by siRNA, however, attenuated but did not prevent processing of caspases and cleavage of PARP. PKCδ knockdown had little effect on the processing of caspases in 2008 cells (data not shown). Thus, the antiapoptotic effect of rottlerin may be independent of PKCδ.
Rottlerin induces caspase-2 down-regulation via the proteasome-mediated pathway. We then examined the mechanism by which rottlerin induces down-regulation of caspase-2. Rottlerin induced down-regulation of caspase-2 in both HeLa (Fig. 5A) and 2008 (Fig. 5B) cells but it did not induce down-regulation of caspase-9, caspase-3, caspase-8, or caspase-7. We compared the ability of several protease inhibitors in preventing caspase-2 down-regulation by rottlerin in HeLa cells (Fig. 5A). Two different proteasome inhibitors, MG132 and PSI, inhibited rottlerin-induced down-regulation of caspase-2 in HeLa cells. ALLN, an inhibitor of the proteasome, cathepsins, and calpains, also inhibited caspase-2 down-regulation by rottlerin. In contrast, the calpain inhibitor calpeptin failed to prevent rottlerin-induced down-regulation of caspase-2. Figure 5B shows that pretreatment of 2008 cells with MG132 before rottlerin treatment also blocked rottlerin-induced down-regulation of caspase-2. These results suggest that rottlerin triggers caspase-2 down-regulation via the proteasome-mediated pathway.
Because MG132 inhibited caspase-2 down-regulation by rottlerin, we examined if MG132 could reverse the effect of rottlerin in protecting cells against cisplatin-induced apoptosis. Figure 6A shows that pretreatment of HeLa cells with MG132 before cisplatin treatment enhanced caspase-2 processing and cleavage of PARP, whereas pretreatment with rottlerin induced caspase-2 down-regulation and inhibited PARP cleavage. Cisplatin-induced processing of caspase-3 was not inhibited by MG132 but was apparently enhanced. The increase in caspase-3 processing by MG132 was associated with an increase in proteolytic processing of PKCδ (Fig. 6B). Although pretreatment of cells with MG132 before rottlerin treatment inhibited caspase-2 down-regulation, it failed to enhance cisplatin-induced caspase activation or PARP cleavage (Fig. 6A).
We also examined the effect of MG132 on caspase activity using a fluorometric assay. Consistent with the Western blot analysis (Fig. 6A), MG132 enhanced cisplatin-induced activation of caspase-2, but rottlerin inhibited caspase-2 activation even when cells were pretreated with MG132 (Fig. 6C). Thus, rottlerin may also inhibit caspase-2 activation and processing.
In the present study, we made a novel observation that rottlerin induces down-regulation of caspase-2, and this may explain the ability of rottlerin to protect against DNA damage–induced apoptosis. Down-regulation of caspase-2 by rottlerin was triggered by the proteasome-mediated pathway. Furthermore, the effect of rottlerin on caspase-2 down-regulation was independent of its ability to inhibit PKCδ.
Rottlerin was originally identified as a specific inhibitor of PKCδ (37). Since the demonstration that PKCδ is a substrate for caspase-3, there have been numerous studies that used rottlerin to link PKCδ with DNA damage–induced apoptosis (39). We have previously shown that rottlerin inhibited cisplatin-induced caspase activation and cell death, suggesting that PKCδ not only acts as a substrate for caspase-3 but also regulates caspase activation (30). Rottlerin not only inhibited activation of caspase-3 but it also inhibited processing of caspase-9 (31). We proposed that rottlerin acts at an early stage before caspase-9 activation to regulate cisplatin-induced cell death (31).
Caspase-9 is believed to be the apical caspase during DNA damage–induced apoptosis, although recent studies suggest that caspase-2 may act upstream of caspase-9 (7, 13, 21, 24, 32, 33). Our results show that knockdown of caspase-2 inhibited the processing of caspase-9 and caspase-8. In contrast, knockdown of caspase-9 failed to inhibit processing of caspase-2 and caspase-8. These results suggest that caspase-2 may indeed act upstream of caspase-9 during cisplatin-induced apoptosis. Although rottlerin inhibited activation and processing of apical caspase-8 and caspase-9, we made an unexpected observation that rottlerin induced down-regulation of caspase-2. Rottlerin, however, did not influence the levels of other caspases, such as caspase-3, caspase-7, caspase-8, and caspase-9. It is conceivable that down-regulation of caspase-2 by rottlerin may inhibit activation and processing of caspase-9 and downstream caspases.
Because PKCδ plays an important role in DNA damage–induced apoptosis, it is conceivable that rottlerin induces caspase-2 down-regulation via PKCδ-dependent pathway. However, several studies reported that rottlerin acts on additional targets besides PKCδ (40–46). For example, rottlerin acts as a mitochondrial uncoupler independent of its effects on PKCδ (40, 43, 46). We therefore compared the effects of rottlerin with other pharmacologic inhibitors of PKC. Rottlerin inhibits PKC isozymes with IC50 values for PKCδ (3–6 μmol/L), cPKCs (30–40 μmol/L), and PKCε, PKCη, and PKCζ (80–100 μmol/L; ref. 37). BIM inhibits all PKC isozymes with rank order of potency α > βI > ε > δ > ζ (38). Although cPKCs are inhibited at nanomolar concentrations, nPKCs are inhibited at submicromolar concentrations. One of the caveats with the use of these broad specificity PKC inhibitors is that they inhibit multiple PKC isozymes, which may have opposite effects on apoptosis. For example, we have shown that inhibition of PKCα by cPKC-specific inhibitor Gö 6976 or knockdown of PKCα by siRNA enhanced cisplatin-induced apoptosis (47). Thus, depending on the concentrations of these inhibitors, they may increase or decrease activation of caspases and proteolytic cleavage of PKCδ and PARP. We have found that although 1 μmol/L BIM or Gö 6983 attenuated proteolytic cleavage of PKCδ, they did not induce caspase-2 down-regulation. Furthermore, silencing of PKCδ by siRNA failed to induce caspase-2 down-regulation. These results suggest that the effect of rottlerin on caspase-2 down-regulation was PKCδ independent.
PKCδ may be activated by cofactors or by proteolytic activation. During apoptosis, cleavage of PKCδ at the hinge region by caspase-3 removes the autoinhibitory regulatory domain leading to its proteolytic activation (39). Because down-regulation of caspase-2 prevents activation of downstream caspases, including caspase-3, it also decreases proteolytic activation of PKCδ, which is cleaved by caspase-3. To determine if inhibition of PKCδ processing by caspase-2 knockdown was responsible for the antiapoptotic effect of rottlerin, we examined the effect of PKCδ knockdown on caspase activation and cisplatin-induced apoptosis. Silencing of PKCδ had modest effect on the processing of caspase-2 and caspase-9, and cleavage of PARP in HeLa cells, but it had little effect on the processing of caspases in 2008 cells (data not shown). Because rottlerin effectively inhibited processing of these caspases in both cell types, these results suggest that rottlerin acts on additional targets besides PKCδ.
To understand how rottlerin induces caspase-2 down-regulation, we compared the effects of several protease inhibitors, including calpain inhibitor calpeptin and proteasome inhibitors MG132 and PSI. Only proteasome inhibitors were able to prevent rottlerin-mediated down-regulation of caspase-2, suggesting that rottlerin triggers caspase-2 down-regulation via proteasome-mediated pathway. Because MG132 prevents caspase-2 down-regulation by rottlerin, we expected that pretreatment with MG132 would reverse the protective effect of rottlerin on cisplatin-induced cell death. However, MG132 had no effect on rottlerin-mediated protection against cisplatin-induced cell death. One possibility is that caspase-2 is not important for cisplatin-induced cell death, and thus, inhibition of caspase-2 down-regulation has no effect on cisplatin-induced cell death. This is unlikely because knockdown of caspase-2 by siRNA inhibited cisplatin-induced cell death. Rottlerin also prevented activation of caspase-2 even when caspase-2 down-regulation was prevented by MG132. Thus, rottlerin not only induces caspase-2 down-regulation but it also inhibits caspase-2 activity.
It has been shown that the proteolytic fragment of PKCδ generated by caspase-3–mediated cleavage is necessary for DNA damage–induced apoptosis (26–29). It has also been reported that PKCδ is a substrate for caspase-2 because recombinant PKCδ was cleaved by recombinant caspase-2 in vitro (19). The authors suggested that caspase-2 is primarily responsible for doxorubicin-induced cleavage of PKCδ because 5 μmol/L caspase-3 inhibitor zDEVD partially blocked doxorubicin-induced PKCδ cleavage, whereas 10 μmol/L caspase-2 inhibitor zVDVAD completely blocked it. Thus, inhibition of caspase-2 may inhibit proteolysis of PKCδ. We, however, failed to detect any cleavage of PKCδ in MCF-7 cells (48) that lack functional caspase-3 (49), although these cells express caspase-2 (20). Furthermore, we have shown that caspase-2 acts upstream of caspase-3. Thus, inhibition of caspase-2 by zVDVAD is expected to inhibit caspase-3 and, thus, cleavage of PKCδ. Regardless of whether PKCδ is cleaved by caspase-3 or caspase-2, rottlerin can still inhibit apoptosis by directly inhibiting the catalytic fragment of PKCδ generated by these caspases. Consequently, even when caspase-2 down-regulation is inhibited by proteasome inhibitors, rottlerin can still prevent cisplatin-induced apoptosis. Thus, one possibility is that rottlerin not only acts upstream of caspase-2 to regulate its level, but it also acts downstream of caspase-2 and caspase-3 to directly inhibit PKCδ catalytic fragment.
There are, however, controversies whether rottlerin inhibits PKCδ or not. Gschwendt et al. (37) have shown that rottlerin directly inhibits PKCδ catalytic activity, whereas Davies et al. (50) failed to show inhibition of PKCδ activity by rottlerin in vitro. Our results show that knockdown of PKCδ, which inhibited cisplatin-induced generation of PKCδ catalytic fragment, had modest effect on the processing of caspases, suggesting that rottlerin may act on additional targets besides PKCδ. Several studies have shown that mitochondrial uncoupling effect of rottlerin could contribute to apoptosis in a PKCδ-independent manner (41, 46). These studies, however, do not explain how rottlerin exerts its antiapoptotic function during DNA damage–induced apoptosis. We have found that rottlerin inhibited processing of caspase-2 as well as caspase-9, caspase-3, and caspase-7 even when caspase-2 down-regulation was inhibited by MG132, suggesting that rottlerin acts upstream of caspase-9 in the presence of the proteasome inhibitor.
The mechanism of caspase-2 activation remains elusive. Caspase-2 can function both as an apical as well as an effector caspase. It has been reported that caspase-2 is processed by caspase-3, and this processing could result in an amplification loop (20). Because caspase-3 is a substrate for PKCδ and phosphorylation of caspase-3 by PKCδ results in its activation (51), inhibition of PKCδ by rottlerin may inhibit caspase-3 and further processing of caspase-2. Blockage of caspase-2 processing may in turn result in inhibition of caspase-3 and processing of PKCδ, forming a negative feedback loop. Thus, proteolytic activation of PKCδ may be dependent on the generation of the processed form of caspase-2.
In summary, we have shown that caspase-2 functions as an apical caspase during DNA damage–induced apoptosis, and rottlerin acts upstream of caspase-2 via PKCδ-independent pathway. We made a novel observation that down-regulation of caspase-2 by rottlerin rather than inhibition of PKCδ activity can explain the antiapoptotic function of rottlerin.
Grant support: CA85682 and CA71727 (A. Basu) from the NIH/National Cancer Institute.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Usha Sivaprasad, Shalini Persaud, and Soumya Krishnamurthy for critical reading of the manuscript, and Jiyoung Lee and Rohini Dhar for their assistance with some experiments.