Roles of DNA topoisomerase II isozymes in chemotherapy and secondary malignancies

Abstract

Drugs that target DNA topoisomerase II (Top2), including etoposide (VP-16), doxorubicin, and mitoxantrone, are among the most effective anticancer drugs in clinical use. However, Top2-based chemotherapy has been associated with higher incidences of secondary malignancies, notably the development of acute myeloid leukemia in VP-16-treated patients. This association is suggestive of a link between carcinogenesis and Top2-mediated DNA damage. We show here that VP-16-induced carcinogenesis involves mainly the β rather than the α isozyme of Top2. In a mouse skin carcinogenesis model, the incidence of VP-16-induced melanomas in the skin of 7,12-dimethylbenz[a]anthracene-treated mice is found to be significantly higher in TOP2β⁺ than in skin-specific top2β-knockout mice. Furthermore, VP-16-induced DNA sequence rearrangements and double-strand breaks (DSBs) are found to be Top2β-dependent and preventable by cotreatment with a proteasome inhibitor, suggesting the importance of proteasomal degradation of the Top2β-DNA cleavage complexes in VP-16-induced DNA sequence rearrangements. VP-16 cytotoxicity in transformed cells expressing both Top2 isozymes is, however, found to be primarily Top2α-dependent. These results point to the importance of developing Top2α-specific anticancer drugs for effective chemotherapy without the development of treatment-related secondary malignancies.

Anticancer drugs that target DNA topoisomerase II (Top2), including etoposide (VP-16), doxorubicin, and mitoxantrone, are often referred to as Top2 poisons and are among the most effective and widely used anticancer drugs in the clinic. However, life-threatening toxic side effects, including drug-induced secondary malignancies, have been noted in patients receiving Top2-based chemotherapy. An association between infant leukemia and in utero exposure to Top2 poisons has also been reported (reviewed in refs. 1–3). In all cases, the molecular basis underlying carcinogenesis in Top2-based chemotherapy is unclear.

Clinical evidence for a direct link between VP-16 treatment and treatment-related acute myeloid leukemia (t-AML) is particularly strong (1–3). VP-16-induced t-AML is frequently associated with balanced translocations between the mixed lineage leukemia (MLL) gene on chromosome 11q23 and >50 partner genes (the MLL gene is also known as ALL-1, hTRX, or HRX) (4–7). These rearrangements, as well as those found in infant leukemia, cluster within a well characterized 8.3-kb breakpoint cluster region (bcr) (8–16). The bcr of MLL is AT-rich and contains Alu sequences, putative recognition sites of Top2-mediated DNA cleavage, and chromosome scaffold/matrix attachment regions (SAR/MAR) (5, 8–17). There is substantial evidence that chromosome 11q23 translocations in t-AML and infant leukemia are a consequence of drug-induced formation of double-strand breaks (DSBs) (6–9). VP-16 is known to induce DSBs by the formation of a Top2-DNA covalent complex termed the cleavage or cleavable complexes (reviewed in refs. 18 and 19), and mapping of VP-16-induced DSBs to the bcr of the MLL gene has led to the suggestion of a direct link between these DSBs and MLL gene translocations (8, 12–15, 20). However, other studies have also pointed to the involvement of apoptotic nucleases in VP-16-induced DSBs within the bcr of the MLL gene (21–26).

There are two human Top2 isozymes, Top2α and Top2β (27, 28), and VP-16 is known to induce both Top2α and Top2β DNA cleavage complexes (29, 30). The two isozymes share ≈70% sequence similarity but are regulated very differently during cell growth: the α isozyme is a proliferation marker and is greatly elevated in tumor cells, whereas the β isozyme is present in proliferating as well as postmitotic cells (31–34). Top2α has been suggested to function in cell cycle events such as DNA replication and chromosome segregation (35–38), and Top2β has been implicated in transcription (34, 39, 40). It has been unclear, however, whether these two isozymes play different roles in tumor-cell killing and in the development of secondary malignancies during the course of Top2-based chemotherapy.

Previous studies in a mouse skin carcinogenesis model have demonstrated that VP-16 induces carcinogenesis, and it has been suggested that the drug acts as a stage I (convertogenic) tumor promoter (41). In the present study, we have used skin-specific top2β-knockout mice to test the possibility that the Top2α and Top2β isozymes have different roles in the development of secondary malignancies and in tumor-cell killing. Our results suggest that the β isozyme is primarily responsible for VP-16-induced carcinogenesis in this model. Furthermore, in cell-culture models, VP-16-induced DNA sequence rearrangements and DSBs are also found to be primarily Top2β-dependent. By contrast, VP-16 cytotoxicity in tumor cells appears to be primarily Top2α-dependent. These results suggest that the two Top2 isozymes play distinct roles in Top2-based chemotherapy and point to the importance of developing Top2α isozyme-specific drugs for cancer chemotherapy without a high risk of treatment-related secondary malignancies.

Results

VP-16-Induced Melanomas in the Skin of 7,12-Dimethylbenz[a]anthracene (DMBA)-Treated Mice Are Top2β-Dependent.

To evaluate the role of Top2β in VP-16-induced carcinogenesis, skin-specific top2β-knockout mice (K14-Cre top2β^flox2/flox2) and TOP2β⁺ controls (top2β^flox2/flox2, top2β^+/flox2, and K14-Cre top2β^+/flox2) were generated. The top2β^flox2 allele contains two loxP sites flanking a DNA segment encoding the active-site tyrosine region of Top2β. This allele expresses wild-type Top2β, but is converted to a null allele, top2β^Δ2, upon exposure to Cre recombinase expressed from a transgene (34), in the present case, one under the control of the skin-specific promoter K14 (see Materials and Methods). As shown in Fig. 1A, Top2β is absent from both the epidermis (Upper) and hair follicles (Lower) of K14-Cre top2β^flox2/flox2 mice (to be referred to hereafter as the TOP2β⁻ mice), as evidenced by the absence of Top2β immunostaining in DAPI-positive nuclei (see also Fig. 1B for genotyping examples). Cre-mediated deletion of the floxed top2β locus is further evidenced by the appearance of the PCR product corresponding to the top2β^Δ2 allele (to be referred to hereafter as the top2β⁻ allele; see lanes 2 and 3 in Fig. 1B).

Fig. 1.

VP-16 induces melanomas in the skin of DMBA-treated mice. (A) Absence of Top2β in the epidermis (Upper) and hair follicles (Lower) of skin-specific top2β-knockout mice (samples denoted TOP2β⁻). Cryosections of the skin of TOP2β⁺ and TOP2β⁻ mice (8–10 μm thick) were stained with H&E (labeled HE, first column), anti-Top2β antibody (labeled 2β, second column), or DAPI (third column). The merged images of 2β- and DAPI-stained sections are shown in the fourth column (labeled 2β/DAPI). (Scale bars: 10 μm.) (B) PCR-based genotyping of TOP2β⁺ and TOP2β⁻ mice. Genomic DNA samples from tail snippets were genotyped by PCR using primer sets specific for various alleles. Examples are shown here for results with samples from top2β^+/flox2 (lane 1), K14-Cre top2β^+/flox2 (lane 2), and K14-Cre top2β^flox2/flox2 (lane 3) mice. PCR fragments characteristic of the TOP2β⁺, top2β^flox2, top2β^Δ2 (top2β⁻), and K14-Cre alleles are depicted; skin cells of K14-Cre top2β^flox2/flox2 are phenotypically TOP2β⁻, and those from top2β^+/flox2 and K14-Cre top2β^+/flox2 mice are TOP2β⁺ (see the absence of the TOP2β⁺ fragment in lane 3 and the presence of the same fragment in lanes 1 and 2). (C) VP-16-induced melanomas in the skin of TOP2β⁺ and skin-specific top2β-knockout mice (TOP2β⁻). Representative photos of DMBA-initiated mice treated with DMSO (vehicle control), VP-16, or phorbol 12-tetradecanoate 13-acetate (TPA) are shown. The blue arrow points to a typical melanoma. (D) Histological and immunohistochemical analyses of melanomas in the mouse skin. Consecutive sections of skin melanomas were stained with either H&E or melanoma-specific antibodies. Representative pictures of H&E staining (Upper) and melanoma antibody staining (Lower) are shown. The red arrow points to a melanoma mass, the blue arrow points to the epidermis, and the green arrow points to a hair follicle. (Scale bars: 100 μm.)

Age-matched 7-week-old mice were used for skin carcinogenesis studies. Both TOP2β⁺ and TOP2β⁻ mice received a single application of DMBA, followed by various treatments (see the six treatment groups in Materials and Methods). Under the treatment conditions, these mice developed skin melanomas (Fig. 1C). Papillomas, which appeared at an ≈80-fold lower frequency (data not shown), were not included in the analysis. Histology of a typical melanoma in the mouse skin is shown in Fig. 1D Upper. The expansive dark-brown area showing aggregation of pigmented cells (melanin-expressing melanocytes) is indicative of melanoma. Immunohistochemical analysis of the tumor with mouse melanoma mixture antibody also confirmed the presence of melanoma (Fig. 1D Lower).

The number of melanomas in the skin of each treated mouse was recorded and all data are summarized in Fig. 2A. The average number of melanomas per mouse in various treatment groups is also plotted for each treatment group (Fig. 2 B and C). As shown in Fig. 2B, (open bars), VP-16 treatment of DMBA-initiated TOP2β⁺ mice (see groups 2 and 3 for 10 μmol × 5 applications and 20 μmol × 3 applications of VP-16, respectively) show an increase in the average number of melanomas per mouse (by 10% and 60%, respectively) relative to treatment with DMSO alone (group 1). Strikingly, VP-16 treatment of DMBA-initiated TOP2β⁻ mice decreases, rather than increases, the average number of melanomas per mouse by 50% and 15%, respectively, in groups 2 and 3, relative to the DMSO-treated group 1 controls (Fig. 2B, filled bars). This decrease probably reflects a combination of two factors: the absence of VP-16-induced melanomas, owing to the absence of Top2β, and the antitumor activity of VP-16 (which is largely Top2α-dependent, to be discussed later).

Fig. 2.

VP-16 induces fewer skin melanomas in the absence of Top2β. (A) The number of melanomas in the skin of each mouse is plotted for various treatment groups. The symbols “2β⁺” and “2β⁻” denote, respectively, TOP2β⁺ and skin-specific top2β-knockout (TOP2β⁻) mice. The six groups and their treatment descriptions (see numbers in parenthesis) are indicated at the bottom of the graph. (B and C) The average number of melanomas per mouse for each of the treatment groups denoted by numerals 1–6. Comparing with the DMSO-treated animals (group 1), the differences between the TOP2β⁺ and TOP2β⁻ pairs in the average number of melanomas per mouse are statistically significant for groups 2, 3, 5, and 6 (∗, P < 0.05).

If the above interpretation is correct, increasing the number of VP-16 applications should further reduce the number of melanomas in TOP2β⁻ mice. Indeed, as shown in Fig. 2C (filled bars), increasing the number of VP-16 applications (5 μmol per application) from 5 (group 5) to 10 (group 6) significantly decreases the average number of melanomas in the skin of TOP2β⁻ mice (a decrease of 30% and 87%, respectively, for the two groups, relative to group 1). As a positive control, DMBA-treated TOP2β⁺ and TOP2β⁻ mice were also treated with TPA (see Fig. 2B, group 4). As expected, TPA treatment of the TOP2β⁺ mice greatly increased the average number of melanomas per mouse (by 130%) relative to DMSO treatment (Fig. 2B, open bars). In contrast to VP-16 treatment, exposure to TPA causes a similar degree of increase (150%) in skin melanoma in TOP2β⁻ mice (Fig. 2B, filled bars).

The effect of Top2β on the number of VP-16-induced melanomas in mouse skin is more evident by examining the ratio of the average number of melanomas per mouse in TOP2β⁺ versus that in TOP2β⁻ mice. For the VP-16-treated groups, the ratios are 2.0 (group 5), 2.8 (group 3), 3.3 (group 2), and 13 (group 6). By contrast, the ratios are 1.5 and 1.3, respectively, for groups 1 (vehicle control) and 4 (TPA treatment). The differences in the numbers of VP-16-induced melanomas in TOP2β⁺ and TOP2β⁻ mice are statistically significant (P < 0.05, see groups marked by ∗ in Fig. 2 B and C). These results suggest that VP-16-, but not TPA-promoted, melanomas in the mouse skin are primarily Top2β-mediated.

VP-16-Induced Plasmid Integration and DNA DSBs Are Top2β-Dependent.

VP-16 is known to induce DNA sequence rearrangements and tumors in DMBA-initiated mice (41). To test the possibility that the present finding of a preferential role of Top2β in VP-16-induced skin melanoma is because of a predominant role of the particular isozyme in VP-16-induced DNA sequence rearrangements, an assay that measures the incorporation of a plasmid-borne genetic marker in transfected cells was used to examine VP-16-induced DNA sequence rearrangements in simian virus 40 (SV40)-transformed top2β^+/− and top2β^−/− mouse embryonic fibroblasts (MEFs) (Top2β is expressed in the former but not the latter). As shown in Fig. 3A (open bars), VP-16 (0.5 μM) greatly stimulates (by ≈12-fold) plasmid integration in top2β^+/− MEFs, as compared with DMSO vehicle-alone control. Interestingly, VP-16-induced plasmid integration is dramatically reduced (by ≈7-fold, P = 0.005) in top2β^−/− MEFs as compared with top2β^+/− MEFs (Fig. 3A, VP-16 filled and open bars), suggesting a predominant role of the Top2β isozyme in VP-16-induced DNA sequence rearrangements.

Fig. 3.

VP-16 induces Top2β-dependent plasmid integration and DSBs. (A) Effect of VP-16 on plasmid integration. SV40-transformed top2β^+/− and top2β^−/− MEFs were transfected with linearized pUCSV-BSD plasmid DNA in the presence (0.5 μM) or absence of VP-16, and/or the proteasome inhibitor MG132 (2 μM), as indicated. Integration frequency was measured as described in Materials and Methods. (B) VP-16 induces Top2β-dependent formation of phosphorylated histone H2AX (γ-H2AX), a DNA damage signal for DSBs. Primary TOP2β^+/+ and top2β^−/− MEFs, denoted by +/+ and −/−, respectively, were treated with VP-16 (250 μM, 2 h), and cell lysates were immunoblotted after gel electrophoresis by using anti-γ-H2AX as well as anti-α-tubulin antibody (the latter for loading assessment). The expression levels of Top2α and Top2β in primary TOP2β^+/+ and top2β^−/− MEFs were similarly assessed (Inset). (C) VP-16 induces DNA DSBs as measured by the neutral comet assay. Primary TOP2β^+/+ and top2β^−/− MEFs (denoted, respectively, by +/+ and −/− in the figure) were treated with VP-16 (250 μM, 1.5 h). The neutral comet assay was then performed as described in ref. 45 (Left), and the average tail moments were quantified and plotted (Right) (error bars indicate SEM; ∗, P < 0.001 in comparing the +/+ and −/− data).

The neutral comet assay, which measures the amount of fragmented chromosomal DNA from gently lysed cells and has been extensively used in quantifying chromosomal DSBs, was also used to assess the role of the Top2β isozyme in VP-16-induced chromosome breakage. As shown in Fig. 3C, in wild-type TOP2β^+/+ MEFs, VP-16 induces a significant increase (130%, P = 5 × 10⁻⁸) in fragmented DNA, relative to treatment with the DMSO solvent alone. By contrast, in top2β^−/− MEFs, the corresponding increment of ≈30% is statistically insignificant (P = 0.31). Repeating the experiment by using SV40-transformed TOP2β^+/+ and top2β^−/− MEFs gave essentially the same results (data not shown). These results further support the notion of a predominant role of Top2β in VP-16-induced DSB formation.

Proteasome-Mediated Preferential Degradation of Top2β-DNA Covalent Complexes.

Why does Top2β play a predominant role in VP-16-induced DSBs and DNA sequence rearrangements? It is unlikely that the difference between the two Top2 isozymes could be because Top2β is the preferential target of VP-16 in DNA cleavage complex formation. Previous studies have shown that VP-16 induces the same amount of Top2α and Top2β DNA cleavage complexes in vivo (30). In a DNA cleavage assay (42) using equal amounts of purified recombinant human Top2α and Top2β isozymes, about equal amounts of Top2α and Top2β cleavage complexes were observed at various concentrations of VP-16 (Fig. 4A). A band-depletion assay measuring the reduction in the amounts of Top2α and Top2β not covalently trapped on DNA (43), after a 15-min VP-16 treatment, also indicated that in SV40-transformed wild-type TOP2β^+/+ MEFs, both Top2α and Top2β bands are depleted to similar extents (Fig. 4B Upper; see Fig. 4B Lower for quantification).

Fig. 4.

VP-16 poisons both Top2 isozymes equally, but Top2β in the trapped Top2β-DNA complexes is preferentially degraded to reveal the hidden DSBs. (A) VP-16 poisons Top2 isozymes equally in vitro. DNA cleavage assays were performed as described in ref. 42; VP-16 concentrations were 0, 0, 2.0, 20, and 200 μM in the five samples from left to right. (B) VP-16 effectively traps both Top2α and Top2β cleavage complexes in vivo. Transformed MEFs expressing both Top2 isozymes were treated with VP-16 (0, 10, 50, and 250 μM) for 15 min, and the amounts of Top2 (2α and 2β) cleavage complexes were measured by the band-depletion assay as described in ref. 43 (Upper). The results are quantified and the percentage free Top2 is plotted for each treatment (Lower). VP-16-induced Top2-DNA cleavage complexes are reversed by a further incubation in VP-16-free medium for 50 min (lane 5). (C) VP-16 induces preferential down-regulation of Top2β. Transformed MEFs expressing both Top2 isozymes were treated with VP-16 (50 μM, 2 h) in the presence or absence of the proteasome inhibitor MG132 (2 μM). The DNA cleavage complexes in the treated cells were reversed by an additional incubation in the absence of VP-16 and MG132 (37°C, 30 min), and then alkaline lysis and S7 nuclease treatment (43) were applied. The amounts of Top2 isozymes were measured by Western blotting.

Previous studies (43, 44) have demonstrated, however, that in VP-16-treated cells, the Top2β isozyme covalently trapped on DNA is preferentially degraded over the DNA-trapped Top2α isozyme through a proteasome-dependent pathway. It has been suggested that this preferential degradation of Top2β-DNA cleavage complexes is a key factor in VP-16-induced DSB formation. In support of this notion, cotreatment with the proteasome inhibitor MG132 is found to abolish DSB induction by VP-16, as evidenced by results of neutral comet assays (data not shown) (44). Cotreatment with MG132 is also found to reduce VP-16-induced plasmid integration in top2β^+/− MEFs by ≈3-fold (P = 0.01; see Fig. 3A and compare the last two sets of open bars). Thus, it appears that the preferential role of the Top2β isozyme in VP-16-induced DSBs and DNA sequence rearrangements is owing to its greater sensitivity to proteasome-mediated degradation when covalently trapped on DNA. Indeed, in SV40-transformed wild-type TOP2 MEFs treated with VP-16, Top2β is found to be preferentially degraded relative to Top2α in a proteasome-dependent manner (Fig. 4C).

Top2β Contributes Minimally to VP-16 Cytotoxicity in Transformed Cells.

The above studies suggest that Top2β is primarily responsible for VP-16-induced DSBs and DNA sequence rearrangements. To test whether Top2β is also important for VP-16 cytotoxicity, we determined in triplicate the IC₅₀ of VP-16 for two pairs of transformed top2β-knockout/knockdown cells by using a colorimetric assay based on cleavage of the tetrazolium dye 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) by mitochondrial dehydrogenase from viable cells. The IC₅₀ values of VP-16 are not significantly different for top2β^+/− and top2β^−/− MEFs (0.038 ± 0.007 vs. 0.040 ± 0.006 μM; P = 0.43, t test). Furthermore, in PC12 cells expressing top2β short hairpin RNA (shRNA), whereas the level of Top2β, but not Top2α, is much reduced (by ≈90%) relative to the same cells expressing the control shRNA. IC₅₀ values of VP-16 for the same pair of samples are not significantly different (1.6 ± 0.1 and 1.9 ± 0.1 μM, respectively; P = 0.19, t test). These results indicate that whereas Top2β rather than Top2α plays a major role in VP-16-induced carcinogenesis, the opposite is true in terms of VP-16 cytotoxicity.

Discussion

Previous studies have demonstrated that VP-16 induces papillomas on the skin of DMBA-treated mice in a classical two-stage carcinogenesis model; furthermore, switching the order of VP-16 and DMBA applications has no effect on papilloma incidence, indicating that the drug behaves as a stage I (convertogenic) tumor promoter, presumably through its induction of DNA sequence rearrangements (41). In the present study, we have used skin-specific top2β-knockout mice to evaluate the roles of the two isozymes of DNA, Top2α and Top2β, in VP-16-induced carcinogenesis. Melanomas, rather than papillomas, are the main tumor type detected in the mouse skin in the present study, however, plausibly because of genetic background differences of the mouse strains used: previous studies used albino mice that probably produce no visible melanoma because of a lack of melanin expression in their skin, whereas mice used in the present study have a mixed genetic background, including 129SvEv (>75%), various degrees of BALB/c and C57BL/6, and express melanin in the skin. VP-16 is shown to induce 2- to 13-fold more melanomas, depending on the dose and schedule of VP-16 treatment, in the skin of DMBA-treated TOP2β⁺ mice than in that of similarly treated skin-specific top2β-knockout mice. By contrast, the classical tumor promoter, TPA, induces about the same number of skin melanomas in DMBA-treated mice whether Top2β is expressed in the skin or not. These results suggest that it is the Top2β isozyme that plays a predominant role in VP-16-induced carcinogenesis.

The above conclusion is further supported by studies in tissue culture models. By using a plasmid integration assay to monitor DNA sequence rearrangements, VP-16-stimulated integration of plasmid DNA is shown to be Top2β-dependent: stimulation of integration frequency by VP-16 is much more significant in SV40-transformed MEFs derived from top2β^+/− mice, which express Top2β, than in SV40-transformed MEFs derived from top2β^−/− mice, which do not. Furthermore, the proteasome inhibitor MG132 blocks VP-16-stimulated plasmid integration, suggesting that VP-16-induced DNA sequence rearrangements involve the proteasome pathway. The last result is consistent with that of a recent study implicating the involvement of the proteasome pathway in processing VP-16-induced Top2β-DNA covalent complexes into DSBs (44).

The predominant role of the Top2β isozyme in mediating VP-16-induced carcinogenesis and DNA sequence rearrangements can be attributed to its involvement in DSB formation upon VP-16 treatment. Neutral comet assay indicates that VP-16-induced DSBs are Top2β-dependent in both primary and SV40-transformed MEFs. Furthermore, the predominant role of the Top2β isozyme in VP-16-mediated DSB formation is likely the result of a greater sensitivity of the DNA cleavage complexes of Top2β, relative to that of Top2α, in proteasome-mediated degradation. Whereas the two isozymes are comparable in their ability to form covalent complexes, the Top2β-concealed DNA breaks in the covalent complexes appear to be more easily converted to DSBs by the proteasome degradation pathway (44).

Based on these and other results, a model for the role of Top2β in VP-16-induced DSBs, DNA sequence rearrangements, and carcinogenesis is proposed (see Fig. 5 for a schematic). In this model, VP-16 stabilizes reversible Top2β-DNA cleavage complexes, which are converted into nonreversible Top2β-DNA covalent complexes, in part through transcriptional collisions (43). The nonreversible covalent complexes then undergo proteasomal degradation, baring the hidden DSBs in them; processing of these DSBs through nonhomologous end-joining finally leads to DNA sequence rearrangements and carcinogenesis. It is unclear why the DNA cleavage complexes of Top2β are more sensitive to proteasome-mediated degradation than their Top2α counterparts. Because proteasomal degradation of Top2 cleavage complexes is partially transcription-dependent (43), however, the preferential sensitivity of the Top2β complexes to proteasomal degradation might be related to the preferential involvement of Top2β in transcription (39, 40). Further studies are necessary to establish the molecular pathways in processing the Top2-DNA covalent complexes.

Fig. 5.

A model for VP-16-induced carcinogenesis. In this model, VP-16 stabilizes the Top2β isozyme covalently trapped on chromosomal DNA, and the trapped isozyme is then preferentially degraded relative to the trapped 2α isozyme by a proteasome pathway; the proteasomal degradation exposes topoisomerase-concealed DSBs for repair by nonhomologous end-joining (NHEJ), which in turn results in DNA sequence rearrangements and carcinogenesis.

Whereas Top2β, rather than Top2α, is shown to have a predominant role in VP-16-induced carcinogenesis, our studies of Top2β-knockout and knockdown cells suggest that the opposite is the case in VP-16 cytotoxicity against transformed cells. The importance of Top2α in VP-16 cytotoxicity is consistent with results from previous studies that the TOP2α gene is mutated in cell lines selected for lower levels of resistance to Top2 drugs, and the TOP2β gene is mutated only in top2α mutant cells selected for higher levels of resistance to Top2 drugs (45, 46). It has been suggested that the collision between the replication forks and Top2 cleavage complexes plays a major role in VP-16 cytotoxicity (47); consequently, the predominant role of Top2α in DNA replication may lead to more frequent collisions between this isozyme and the replication fork, and hence its higher cytotoxicity.

There are several important clinical implications of our findings. First, Top2-targeting drugs such as VP-16 are known to be associated with the development of secondary malignancies, such as t-AML from MLL gene translocations (1–3). The predominant role of the Top2β isozyme in VP-16-induced DNA sequence rearrangements and carcinogenesis suggests that the action of VP-16 on Top2β is the major reason for the development of secondary malignancy in patients receiving Top2-based chemotherapy. Accordingly, the development of Top2α isozyme-specific anticancer drugs may offer the clinical advantage of reducing secondary malignancies. Second, our findings suggest that chemotherapy with anticancer drugs specific to Top2α may also reduce other types of tissue toxicity, because Top2β, and not Top2α, is present in nonproliferating tissues such as the adult heart. Third, based on the proposed mechanism outlined in Fig. 5, other strategies are conceivable for preventing the development of secondary malignancies associated with Top2-based chemotherapy. Cotreatment with the proteasome inhibitor bortezomib, for example, is predicted to prevent proteasomal processing of Top2β-DNA covalent complexes into DSBs and may thus prevent the development of malignancies such as t-AML in patients receiving Top2-based chemotherapy. Proteasome inhibitors may also increase cytotoxicity of Top2 drugs and hence enhance treatment efficacy, because unprocessed Top2β-DNA covalent complexes are expected to contribute to cytotoxicity.

Materials and Methods

Mouse Strains.

Skin-specific deletion within the floxed top2β allele is achieved by crossing the top2β^flox2 lines (34) with mice expressing Cre recombinase from the keratin 14 promoter (kindly provided by A. P. McMahon, Harvard University). The K14-Cre transgenic mouse line expresses Cre in keratinocytes of the epidermis as well as hair follicles during prenatal and postnatal development (48, 49). Mice with the genotype K14-Cre top2β^flox2/flox2, K14-Cre top2β^+/flox2, top2β^flox2/flox2, and top2β^+/flox2 were generated and used in this study; with the exception of K14-Cre top2β^flox2/flox2 mice, which specifically lack Top2β in skin cells, all of the others are phenotypically TOP2β⁺ in all tissues. The K14-Cre top2β^flox2/flox2 skin-specific top2β-knockout mice exhibit a normal lifespan and show no skin abnormality other than cyclic alopecia (data not shown). Genotyping of the various alleles was done as described (in refs. 34 and 49).

Carcinogenesis Assay with a Mouse Skin Model.

Seven-week-old skin-specific top2β-knockout mice and their TOP2β⁺ controls were used. The back of each mouse was shaved 2 days before treatment. The tumor initiator DMBA (1 μmol in 100 μl of DMSO) was applied once in the first week, and then various treatments (2 applications per week) were applied for six groups of animals: group 1, DMSO (100 μl), 5 applications; group 2, VP-16 (10 μmol in 100 μl of DMSO), 5 applications; group 3, VP-16 (20 μmol in 200 μl of DMSO), 3 applications; group 4, the tumor promoter TPA (17 nmol in 100 μl of DMSO), 8 applications; group 5, VP-16 (5 μmol in 100 μl of DMSO), 5 applications; group 6, VP-16 (5 μmol in 100 μl of DMSO), 10 applications. Mice were examined every week for appearance of melanomas on their skins. The number of melanomas visibly notable was scored at the end of the 16th week. The average numbers of tumors induced in different treatment groups were compared by using Student's t test.

Histochemical and Immunohistochemical Analyses.

Immunohistochemical analysis and melanin bleaching of mouse skin sections were performed as described (50, 51). Mouse melanoma mixture antibody (Abcam, Cambridge, MA), rabbit anti-Top2β antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and Cy3- or Cy2-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) were used in these experiments.

Cells.

Primary MEFs were isolated from day 13.5 TOP2β^+/+, top2β^+/Δ2 and top2β^Δ2/Δ2 mouse embryos, as described in ref. 44, and SV40-transformed MEFs were obtained by transformation with pAN2 DNA (44). PC12 cells were first clonally selected and then used to generate top2β-shRNA and control-shRNA knockdown cells. A rat TOP2β-shRNA sequence (5′-GCCCCCGTTATATCTTCAC-3′) was generated based on partial rat TOP2β cDNA sequence (GenBank accession no. D14046). Duplex DNA of the sequence (5′-TGCCCCCGTTATATCTTCACTTCAAGAGAGTGAAGATATAACGGGGGCTTTTTC-3′) was made and cloned into the LentiLox 3.7 vector (obtained from L. van Parijs, Massachusetts Institute of Technology, Cambridge, MA). The control-shRNA sequence (5′-GCGCGCGTTAAATCTTCAC-3′) was created by altering three nucleotides in the rat TOP2β-shRNA sequence (underlined), and duplex DNA of the sequence (5′-TGCGCGCGTTAAATCTTCACTTCAAGAGAGTGAAGATTTAACGCGCGCTTTTTC-3′) was cloned into the LentiLox 3.7 vector. The shRNA-expressing vectors were then inserted with the phosphoglycerate kinase (PGK)-driven Neo^r gene. Lentiviral stocks were prepared and virus-infected PC12 cells were selected from 2-week-old cultures in the presence of 700 μg/ml G418. Single colonies were isolated and characterized and cultured in a 37°C incubator with 5% CO₂, in RPMI medium 1640 supplemented with 10% horse serum, 5% FetalPlex animal serum complex (Gemini Bio-Products, West Sacramento, CA), 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin in flasks coated with collagen type I (BD Biosciences, San Jose, CA).

Plasmid Integration Assay.

Assays with transformed MEFs were performed as described in ref. 26. VP-16 was added at the time of transfection, and where indicated, the proteasome inhibitor MG132 (2 μM) was added 30 min before and during transfection. Integration frequencies were calculated as the ratios of the numbers of blasticidin-resistant colonies and surviving cells.

Abbreviations

DMBA

7,12-dimethylbenz[a]anthracene

TPA

phorbol 12-tetradecanoate 13-acetate

VP-16

etoposide

Top2

DNA topoisomerase II

TOP2β⁺

mouse with phenotype of wild-type Top2β

TOP2β⁻

mouse with phenotype of mutant Top2β

DSBs

double-strand breaks

MEFs

mouse embryonic fibroblasts

bcr

breakpoint cluster region

t-AML

treatment-related acute myeloid leukemia

SV40

simian virus 40

shRNA

short hairpin RNA.