The Poly(ADP-ribose) Polymerase-1-regulated Endonuclease DNAS1L3 Is Required for Etoposide-induced Internucleosomal DNA Fragmentation and Increases Etoposide Cytotoxicity in Transfected Osteosarcoma Cells¹

Successful drug treatment in human cancers requires an adequate therapeutic indicator reflecting the specific molecular mechanisms of action of the treatment on target cells. These effects may be enzymatic or structural in nature. Determining the specific factors that are involved in the complex events of cell killing by chemotherapeutic drugs not only provides an understanding of the mechanism by which cancer cells are killed but may also provide important insight for developing new targets and strategies in the treatment of cancer.

Defective induction of apoptosis is implicated in the progression of malignancy and the development of resistance to chemotherapeutic drugs in many types of cancer (1). The chemotherapeutic agent VP-16,³ an epipodophyllotoxin that inhibits DNA topoisomerase II, is thought to induce apoptosis as a consequence of the persistent DNA damage that results from the inability of the inhibited enzyme to mediate the religation of single- and double-strand breaks (2). Apoptosis is characterized by major changes in cell morphology, including chromatin condensation, membrane blebbing, nuclear breakdown, and the formation of membrane-associated apoptotic bodies. At the molecular level, these changes are accompanied by the cleavage of many housekeeping proteins, including PARP-1 and lamins, and ultimately by internucleosomal DNA fragmentation. Regardless of the stimulus, the initiation and execution of apoptosis are mediated by members of the caspase family of aspartate-specific cysteine proteases.

Internucleosomal DNA fragmentation is a terminal step in disposal of the genome in cells undergoing apoptosis. Defective DNA fragmentation has been associated with an increased resistance of cells to apoptosis (3, 4, 5). The mechanism of DNA fragmentation in apoptotic cells is poorly understood, although several endonucleases have been implicated in this process (6, 7). The candidate endonucleases identified to date differ in characteristics such as Ca²⁺ and Mg²⁺ dependence, optimal pH, tissue distribution, and requirement for caspase-3 (7, 8, 9, 10). DFF, also known as caspase-activated DNase, has been suggested to play a major role both in the early transient appearance of DNA strand breaks and in internucleosomal DNA fragmentation during apoptosis (7, 11, 12, 13). DFF is composed of two subunits of M_r 40,000 and 45,000, termed DFF40 (caspase-activated DNase) and DFF45 inhibitor of caspase-activated DNase (ICAD), respectively (5, 11, 12, 13). The endonuclease activity of this enzyme, which is intrinsic to DFF40, is induced on cleavage of DFF45 by caspase-3.

We have recently cloned and characterized DNAS1L3 (14, 15), which is the human homologue of a bovine chromatin-bound and Ca²⁺- and Mg²⁺-dependent endonuclease (16), and also contributes to internucleosomal DNA degradation during apoptosis (15, 17). The activity of DNAS1L3 (15, 17), like that of the bovine endonuclease (16, 18), is reversibly inhibited by poly(ADP-ribosyl)ation. We have also shown that human osteosarcoma cells do not express DNAS1L3, as assessed by reverse transcription and PCR analysis, and that these cells fail to undergo internucleosomal DNA fragmentation after exposure to VP-16 (17). Incubation of purified rat cerebella nuclei with purified DNAS1L3 in the presence of Ca²⁺ and Mg²⁺ results in cleavage of DNA in a pattern identical to that observed in apoptotic cells (14). We showed recently, in an in vitro study, that the inhibition of DNAS1L3 endonuclease activity by PARP-1 is blocked when PARP-1 is neutralized by active recombinant caspase-3 (17). However, the osteosarcoma cells do exhibit internucleosomal DNA fragmentation in response to staurosporine (19) or during confluency-triggered spontaneous apoptosis (20, 21).

We have now investigated this difference between the abilities of VP-16 and staurosporine to induce internucleosomal DNA fragmentation in osteosarcoma cells, specifically exploring the possibility that DNAS1L3 contributes to VP-16 cytotoxicity. In addition, we examined the roles of PARP-1 and Ca²⁺ in the regulation of DNAS1L3-mediated and staurosporine-induced DNA fragmentation.

Human osteosarcoma cells (143.98.2; ATCC CRL 11226; Refs. 20, 22) were maintained in DMEM supplemented with 10% fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 μg/ml). Cells were transfected with vectors encoding human DNAS1L3 (14) or a caspase-3-resistant human mut-PARP-1 (19) with the use of Mirus TransIT-100 reagent (Panvera, Madison, WI). Both DNAS1L3 and mut-PARP-1 cDNAs were positioned immediately downstream of the coding sequence for six histidine residues and the FLAG epitope, and were cloned in pcDNA3.1 and pCR3.1 mammalian expression vectors (Invitrogen, Carlsbad, CA), respectively. Expression of DNAS1L3 or mut-PARP-1 was tested by immunoblot analysis with antibodies to the FLAG epitope as described below. Apoptosis was induced by exposing cells to 70 μm VP-16 or 1 μm staurosporine (Sigma) for 24 or 36 h at 37°C.

Genomic DNA was isolated from cells as described previously (19), subjected to electrophoresis through a 1.5% agarose gel in Tris-borate EDTA buffer, and stained with ethidium bromide.

After treatment with VP-16 or staurosporine, cell viability was assessed essentially as described (19).

Cells were washed with an ice-cold phosphate buffer saline and then lysed as described (19). A portion (30 μg of protein) of each lysate was fractionated by SDS-PAGE on a 4–20% gradient gel. Alternatively, recombinant DNAS1L3 was precipitated from cell lysates (250 μg of protein) with Ni-NTA magnetic beads (Qiagen, Valencia, CA), and the precipitates were subjected to electrophoresis. Separated proteins were transferred to a nitrocellulose filter, which, after staining with Ponceau S to confirm equal loading and transfer of samples, was probed with antibodies to FLAG (Santa Cruz Biotechnology, Santa Cruz, CA), to caspase-3 (Santa Cruz Biotechnology), to PARP-1 (PharMingen, San Diego, CA), to PAR (Alexis Biochemicals, San Diego, CA), or to DFF45 (kindly provided by X. Wang, University of Texas, Southwestern Medical Center, Dallas, TX). Immune complexes were detected with appropriate secondary antibodies and chemiluminescence reagents (Pierce, Rockford, IL).

Cells were fixed, permeabilized, and stained with a rabbit antibody to FLAG to PAR essentially as described (19). They were then examined with a Nikon fluorescence microscope.

Osteosarcoma cells were incubated in the absence or presence of either 1 μm staurosporine or 70 μm VP-16 for 36 h, after which genomic DNA was extracted and analyzed by agarose gel electrophoresis. Whereas staurosporine induced marked internucleosomal DNA fragmentation, such cleavage of DNA was not detected in the cells treated with VP-16 (Fig. 1,A). Assessment of cell viability by calcein-AM staining revealed that both drugs exhibited substantial cytotoxicity in the osteosarcoma cells, with the effect of staurosporine being slightly greater than that of VP-16 (Fig. 1 B). Thus, internucleosomal DNA fragmentation does not appear to be absolutely required for the cytotoxic effect of VP-16. These results also suggest that the staurosporine-induced endonuclease activity is insensitive to VP-16 and, therefore, independent of inhibition of topoisomerase II.

Activation of caspase-3 and the consequent cleavage of DFF45 are necessary for internucleosomal DNA fragmentation in many cell lines (11, 23, 24). Therefore, we examined whether the lack of internucleosomal DNA fragmentation in VP-16-treated osteosarcoma cells was because of a defect in the activation of caspase-3 or the cleavage of DFF45.

Cells were incubated for 24 h in the absence or presence of staurosporine or VP-16, after which cell extracts were prepared and subjected to immunoblot analysis with antibodies to caspase-3 that recognize both the full-length precursor as well as its M_r 20,000 and M_r 17,000 active fragments. Both drugs induced cleavage of procaspase-3 in a manner indicative of its activation (Fig. 2,A). To confirm that the processed caspase-3 was catalytically active, we examined the status of PARP-1, a caspase-3 substrate, by immunoblot analysis of the same cell extracts with antibodies that recognize both full-length PARP-1 and its M_r 89,000 cleavage product. Both staurosporine and VP-16 induced the cleavage of PARP-1 (Fig. 2 A), verifying that the caspase-3 fragments were catalytically active and that differential activation of this enzyme does not underlie the differential induction of internucleosomal DNA fragmentation by the two drugs.

We next examined the proteolytic processing of DFF45 in osteosarcoma cells treated with VP-16 or staurosporine. Both drugs induced substantial cleavage of DFF45 into the M_r 32,000 and M_r 11,000 fragments characteristic of apoptotic cells (Fig. 2 B). Thus, neither impaired activation of caspase-3 nor defective cleavage of DFF45 appears to underlie the failure of osteosarcoma cells to undergo internucleosomal DNA fragmentation during VP-16-induced apoptosis.

We (5, 25) and others (26) have shown that either depletion of DFF40 as a result of DFF45 gene disruption or inhibition of endogenous DFF40 by expression of a caspase-3-resistant DFF45 mutant (27) prevents the cleavage of genomic DNA into 50-kb fragments in response to inducers of apoptosis. We next examined whether the failure of osteosarcoma cells to undergo internucleosomal DNA fragmentation in response to VP-16 was associated with impaired generation of such large DNA fragments. Cells were treated with either staurosporine or VP-16 for 24 or 36 h, after which DNA was isolated and subjected to transverse alternating-field electrophoresis as described (19). The two drugs induced the generation of 50-kb DNA fragments to similar extents (Fig. 2 C). however, of potential importance, whereas the abundance of these DNA fragments decreased with time of exposure of cells to staurosporine, it remained essentially unchanged during VP-16 treatment. These results suggest that VP-16-treated osteosarcoma cells fail to process the 50-kb DNA fragments into oligonucleosomal fragments. Thus, osteosarcoma cells appear to lack an endonuclease activity that mediates the additional processing of 50-kb DNA fragments in response to VP-16.

We have shown previously that human osteosarcoma cells, unlike various other cell lines such as U-937 human monocytes, do not express DNAS1L3, as assessed by reverse transcription and PCR analysis with primers specific for human DNAS1L3 cDNA (17). Therefore, we examined the effect of ectopic expression of DNAS1L3 in these cells on their ability to undergo internucleosomal DNA fragmentation in response to VP-16. Osteosarcoma cells were stably transfected with an expression vector encoding DNAS1L3 with a His₆-FLAG tag. The expression and nuclear localization of the recombinant DNAS1L3 were confirmed by immunofluorescence analysis (Fig. 3,A) and immunoblot analysis (data not shown) with antibodies to the FLAG epitope. Exposure of the transfected cells to VP-16 resulted in marked internucleosomal DNA fragmentation (Fig. 3,B), and this effect was blocked by inclusion in the culture medium of BAPTA, a cell-permeable Ca²⁺ chelator. These results suggest that the endonuclease activity of DNAS1L3 mediated internucleosomal DNA fragmentation in the transfected cells and that this activity was dependent on Ca²⁺. BAPTA also blocked staurosporine-induced internucleosomal DNA fragmentation in osteosarcoma cells transfected with empty vector (Fig. 3 B), indicating that this drug also activates a Ca²⁺-dependent endonuclease.

We next determined the effect of expression of DNAS1L3 on the sensitivity of osteosarcoma cells to VP-16 cytotoxicity. Expression of DNAS1L3 potentiated the cytotoxic effect of VP-16, compared with that apparent in cells transfected with the empty vector (Fig. 3 C), suggesting that this enzyme is a determinant of the sensitivity of cells to VP-16. These results also indicate that internucleosomal DNA fragmentation is not merely an end point of apoptosis, but rather contributes to the overall process of cell death, in VP-16-treated osteosarcoma cells.

We have shown recently that the endonuclease activity of DNAS1L3 is negatively regulated by PARP-1-mediated poly(ADP-ribosyl)ation both in vitro and in vivo (15, 17). Therefore, we investigated whether PARP-1 differentially modulates internucleosomal DNA fragmentation mediated by DNAS1L3 in response to VP-16 and that induced by staurosporine by transfecting osteosarcoma cells with vectors encoding DNAS1L3 and a caspase-3-resistant mut-PARP-1, the latter of which was also tagged with the His₆-FLAG sequence. Immunoblot analysis with antibodies to FLAG confirmed the expression of the recombinant proteins in the transfected cells (data not shown). The transfected cells were treated with staurosporine or VP-16 for 24 h, after which DNA was isolated and subjected to agarose gel electrophoresis. Whereas expression of caspase-3-resistant mut-PARP-1 prevented the induction of internucleosomal DNA fragmentation by VP-16 in cells expressing DNAS1L3 (Fig. 4,A), it did not affect that induced by staurosporine in cells not expressing DNAS1L3 (Fig. 4 B). These results suggest that PARP-1 regulates VP-16-induced, DNAS1L3-mediated internucleosomal DNA fragmentation but not such DNA fragmentation induced by staurosporine, additionally supporting the notion that the endonuclease activated by staurosporine is distinct from DNAS1L3. Moreover, cleavage of PARP-1 and concomitant cessation of poly(ADP-ribosyl)ation appears to be necessary for DNAS1L3-mediated internucleosomal degradation of DNA, consistent with our previous in vitro and in vivo observations (17).

To confirm that DNAS1L3-mediated internucleosomal DNA fragmentation in response to VP-16 was inhibited as a consequence DNAS1L3 poly(ADP-ribosyl)ation, we precipitated DNAS1L3 from VP-16-treated osteosarcoma cells expressing this endonuclease in the absence or presence of mut-PARP-1. The precipitates were then subjected to immunoblot analysis with antibodies to PAR. VP-16 induced a moderate increase in the extent of poly(ADP-ribosyl)ation of DNAS1L3 in cells expressing the endonuclease alone, but this effect was markedly potentiated by coexpression of the mut-PARP-1 (Fig. 4 C). Thus, these results confirm the regulation of DNAS1L3 by poly(ADP-ribosyl)ation and are consistent with the DNA fragmentation data for the transfected cells. They additionally suggest that removal of PAR moieties attached to DNAS1L3 by the action of PAR glycohydrolase is required for induction of the endonuclease activity of this protein in cells undergoing VP-16-induced apoptosis.

We have shown that expression of DNAS1L3 is required for VP-16-induced internucleosomal DNA fragmentation in human osteosarcoma cells, thereby identifying a specific relation between a proapoptotic chemotherapeutic drug and an apoptotic endonuclease. Ectopic expression of DNAS1L3 also increased the rate of VP-16-induced apoptosis in these cells, suggesting that the resistance of certain cancer cells to VP-16 might result from DNAS1L3 deficiency. Furthermore, our data show that osteosarcoma cells fail to undergo internucleosomal DNA fragmentation during VP-16-induced apoptosis despite their harboring an endonuclease capable of mediating such DNA cleavage in response to staurosporine, and despite their exhibiting such apoptotic events as caspase-3 activation, cleavage of PARP-1 and DFF45, and the consequent degradation of genomic DNA into 50-kb fragments in response to VP-16. Failure to undergo internucleosomal DNA fragmentation was also observed in other cell lines such as the cervical carcinoma cell line, HeLa, as well as the non-Hodgkin’s lymphoma cell line, Daudi.⁴

The fact that osteosarcoma cells undergo internucleosomal DNA fragmentation during staurosporine-induced apoptosis indicates the presence of an endonuclease activity that is capable of catalyzing this reaction but which is not responsive to VP-16. These cells gained the ability to degrade their DNA into oligonucleosomal fragments in response to VP-16 when transfected with an expression vector encoding DNAS1L3, suggesting that this enzyme is required for VP-16-induced internucleosomal DNA fragmentation.

The reason that VP-16 activates DNAS1L3 but not the endonuclease activated by staurosporine is unclear. Clearly, the two drugs act in different fashions. Staurosporine is a bacterial alkaloid that was initially described as an inhibitor of protein kinase C (28) but has subsequently been shown to inhibit many additional protein kinases (29, 30). Moreover, staurosporine induces apoptosis in virtually all of the cell types examined to date (19, 31, 32, 33, 34). Staurosporine-induced cell death is associated with a rapid and prolonged increase in the intracellular Ca²⁺ concentration and accumulation of reactive oxygen species (35, 36). VP-16 is a potent inhibitor of topoisomerase II; it stabilizes a complex between the enzyme and DNA, and thereby induces DNA strand breakage (2). In contrast to that elicited by staurosporine, cell death induced by VP-16 is not associated with the intracellular generation of reactive oxygen species (37), suggesting that oxidative stress might contribute to the induction of internucleosomal DNA fragmentation by staurosporine. The generation of reactive oxygen species has been associated with the release of Ca²⁺ from intracellular stores (38, 39), which results in the activation of Ca²⁺-dependent enzymes such as certain endonucleases. However, intracellular Ca²⁺ release is not likely a factor in the differential induction of internucleosomal DNA fragmentation by staurosporine or VP-16 in osteosarcoma cells or in the preferential activation of DNAS1L3 by VP-16. Both internucleosomal DNA fragmentation mediated by recombinant DNAS1L3 in response to VP-16 and that induced by staurosporine in osteosarcoma cells were dependent on Ca²⁺, given that both were inhibited by BAPTA. Moreover, these results suggest that Ca²⁺ was, in fact, released from intracellular stores in cells treated with either staurosporine or VP-16, consistent with previous data showing that both drugs induce intracellular Ca²⁺ release in other cell types (36, 40, 41, 42). However, the release of Ca²⁺ induced by VP-16 likely occurs independently of the generation of reactive oxygen species.

A characteristic of DNAS1L3 not described for other endonucleases is its regulation by poly(ADP-ribosyl)ation. Although VP-16 induces DNA strand breakage by inhibiting topoisomerase II, the strand breaks are not accessible to PARP-1 because they remain associated with topoisomerase II (43). Furthermore, topoisomerase II is a significant substrate of PARP-1 and is inhibited in the poly(ADP-ribosyl)ated state (44). However, the strand breaks therefore cannot likely contribute either to the activation of PARP-1, which is triggered by binding of the enzyme to the ends of DNA strands or to the consequent inhibition of DNAS1L3. The staurosporine-specific induction of endonuclease activity in osteosarcoma cells likely requires factors that are not targeted by VP-16.

Activation of caspase-3 has been shown to be required for internucleosomal DNA fragmentation (23, 45). However, it is unlikely that caspase-3 activation is a factor that underlies the differential induction of internucleosomal DNA fragmentation by staurosporine or VP-16 in osteosarcoma cells, given that activation of this protease (as evidenced both by cleavage of the proenzyme and by proteolysis of its substrates, PARP-1 and DFF45) was detected in cells treated with either drug. Clearly, additional experimentation is required to fully delineate the mechanisms responsible for the difference in inducing internucleosomal DNA fragmentation between the two drugs in osteosarcoma cells.

The activation of DFF, which results from the cleavage of DFF45, also did not appear to differ between osteosarcoma cells exposed to VP-16 and those exposed to staurosporine. We and others have shown recently that expression and cleavage of DFF45 are required for the processing of DNA into both 50-kb and oligonucleosomal fragments (5, 25, 26, 27). The generation of 50-kb DNA fragments was observed in osteosarcoma cells treated with either staurosporine or VP-16. However, whereas in staurosporine-treated cells the 50-kb DNA fragments were processed additionally with time into oligonucleosomal fragments, those in VP-16-treated cells remained unchanged. These results suggest that DFF may contribute to the generation of 50-kb DNA fragments but is not responsible for internucleosomal DNA cleavage in osteosarcoma cells.

The novel observation concerning the molecular mechanism of an anticancer drug is the fact that ectopic expression of DNAS1L3 increased the sensitivity of osteosarcoma cells to VP-16-induced cell death. This effect might be attributable to an increased generation of DNA breaks in the transfected cells. These additional DNA breaks may enhance nuclear poly(ADP-ribosyl)ation and thereby result in an increased rate or extent of NAD depletion, which culminates in collapse of the mitochondrial membrane potential and release of cytochrome c. Resistance of cancer cells to chemotherapeutic drugs has been associated with altered drug metabolism (which has been attributed to increased expression of the P-glycoprotein encoded by the MDR gene), with increased redox detoxifying action of glutathione, with increased hepatic cytochrome P450 activity, and with mutation of topoisomerase II. Resistance of cells to apoptosis has been attributed to increased expression either of antiapoptotic factors, such as members of the Bcl-2 family of proteins, or of proliferative factors, such as phosphoinositide 3′-kinase and the protein kinase Akt (1, 46). Our results now suggest that loss of expression or inactivation of endonucleases also might contribute to reduced sensitivity of cells to drug-induced apoptosis. Internucleosomal DNA fragmentation has been considered merely as an end point of apoptosis, as a means to dispose of genomic DNA. However, our present data suggest that endonucleases such as DNAS1L3 might also play an active role in apoptotic cell death. The demonstration that DNAS1L3 is directly associated with VP-16-induced cell death may facilitate the development of new approaches to anticancer therapy.