Dovitinib induces mitotic defects and activates the G2 DNA damage checkpoint.

Dovitinib (TKI258; formerly CHIR-258) is an orally bioavailable inhibitor of multiple receptor tyrosine kinases. Interestingly, Dovitinib triggered a G2/M arrest in cancer cell lines from diverse origins including HeLa, nasopharyngeal carcinoma, and hepatocellular carcinoma. Single-cell analysis revealed that Dovitinib promoted a delay in mitotic exit in a subset of cells, causing the cells to undergo mitotic slippage. Higher concentrations of Dovitinib induced a G2 arrest similar to the G2 DNA damage checkpoint. In support of this, DNA damage was triggered by Dovitinib as revealed by γ-H2AX and comet assays. The mitotic kinase CDK1 was found to be inactivated by phosphorylation in the presence of Dovitinib. Furthermore, the G2 arrest could be overcome by abrogation of the G2 DNA damage checkpoint using small molecule inhibitors of CHK1 and WEE1. Finally, Dovitinib-mediated G2 cell cycle arrest and subsequent cell death could be promoted after DNA damage repair was disrupted by inhibitors of poly(ADP-ribose) polymerases. These results are consistent with the recent finding that Dovitinib can also target topoisomerases. Collectively, these results suggest additional directions for use of Dovitinib, in particular with agents that target the DNA damage checkpoint.


Introduction
Receptor tyrosine kinases (RTKs) are cell surface receptors for polypeptide growth factors, cytokines and hormones [1]. A wide variety of cellular events, including cell growth, differentiation, apoptosis, mobility and angiogenesis are modulated by RTKs. As their dysregulation is implicated in the development and progression of many types of cancer, RTKs have become an important class of molecular targets for anti-cancer therapies [2].
Several small molecule inhibitors of RTKs in clinical use, such as Sorafenib and Sunitinib, in fact target multiple RTKs [3]. The argument for the development of multi-target drugs for cancer treatments is that with the exception of a few malignancies that are driven by mutations of a single gene (such a BCR-ABL in chronic myeloid leukaemia), most cancers are caused by multiple events and involve more than one aberrant signalling pathway [4]. Another rationale of using multi-target drugs is that drug resistance developed in cancer patients treated with single-target drugs is often because of the activation of alternative RTKs pathways. Dovitinib (TKI258; formerly CHIR-258) is an orally bioavailable inhibitor of a number of RTKs [5]. It targets several members of the class III, IV and V RTK family, including vascular endothelial growth factor receptor 1/2 (VEGFR1/2), fibroblast growth factor receptor 1/3 (FGFR1/3), and platelet-derived growth factor receptor beta (PDGFRb). At least in cell line and animal models, Dovitinib is effective in haematological malignancies and solid tumours including acute myelogenous leukaemia [6], multiple myeloma [5], colon cancer [7], pancreatic cancer [8], hepatocellular carcinoma [9,10], renal cell carcinoma [11] and urothelial carcinoma [12]. Clinical trials involving Dovitinib are being conducted for the treatment of haematological malignancies and solid tumours, including Phase III trials for metastatic renal cell carcinoma (ClinicalTrials. gov).
Interestingly, several studies have revealed that Dovitinib could induce cell cycle arrest. Huynh et al. found that Dovitinib triggers a G 2 /M cell cycle arrest in a hepatocellular carcinoma cell line [9]. In contrast, treatment of cells from urothelial carcinoma [12] and lymphoplasmacytic lymphoma [13] appears to trigger an arrest in G 1 phase. These results are somewhat surprising because the RTKs targeted by Dovitinib generally have not been associated with directly controlling the cell cycle engine.
In this study, we aimed to substantiate if Dovitinib indeed induces cell cycle delays. We found that Dovitinib triggered a G 2 arrest in a variety of cancer cell lines. Dovitinib also promoted a delay in mitotic exit in a subset of cells. The G 2 arrest was induced through the activation of the DNA damage checkpoint.

Ionizing radiation
Ionizing radiation was delivered with a caesium 137 source from a MDS Nordion Gammacell 1000 Elite Irradiator.

Single cell gel electrophoresis assay
The method for detecting DNA breaks in individual cells was performed as described by Singh et al. [23] by using the pH >13 electrophoresis buffer, with the modification that GelRed Nucleic Acid Gel Stain (Biotium, Hayward, CA, USA; 1/1667) was used instead of ethidium bromide.

Flow cytometry
Flow cytometry analysis after propidium iodide staining was performed as described previously [21].

Live-cell imaging
The setup and conditions of time-lapse microscopy of living cells were as previously described [24].

c-H2AX staining
Cells grown on poly-L-lysine-treated coverslips were fixed by ice-cold methanol for 10 min. The cells were then washed twice with PBS, 5 min. each, before permeabilized and blocked with 3% bovine serum albumin (BSA) and 0.2% Tween-20 in PBS at 25°C for 30 min. The cells were washed twice with wash buffer (0.2% Tween-20 in PBS) before incubated with phosphor-histone H2AX Ser139 antibody (1/500 in 3% BSA, 0.2% Tween-20 in PBS) at 4°C for 16 hrs. The cells were then washed with wash buffer four times, for 5 min. each, and incubated in dark with Alexa Fluor 594 goat anti-rabbit IgG secondary antibodies (Life Technologies) at 25°C for 2 hrs. After washed four times with wash buffer for 5 min. each, the cells were stained with Hoechst 33342 (3 lg/ml in wash buffer) for 5 min., washed three times with wash buffer, before mounted with 100 mM N-propyl-gallate in 9:1 v/v glycerol:PBS.

Dovitinib induces a G 2 cell cycle arrest in different cancer cell lines
The effects of Dovitinib on the cell cycle were examined in cell lines from multiple types of cancers. Nasopharyngeal carcinoma is a rare but highly invasive cancer with effective chemotherapy still to be established. Four NPC cell lines (HONE1, HNE1, CNE2, and C666-1) were treated with different concentrations of Dovitinib. Flow cytometry analysis indicated that Dovitinib induced a G 2 /M arrest (4N DNA content) in a dose-dependent manner (Fig. 1A). A robust accumulation in G 2 /M was stimulated with 5 lM of Dovitinib in these cell lines. The increase of sub-G 1 cells indicated that Dovitinib also triggered apoptosis at the higher concentrations. Accordingly, the apoptotic marker cleaved PARP1 also increased after Dovitinib treatment (see later). In agreement with the stimulation of G 2 /M cell cycle arrest and apoptosis, the number of viable cells, as measured directly using trypan blue exclusion analysis, was reduced by Dovitinib in a dose-dependent manner (Fig. 1B).
Unexpectedly, we found that histone H3 Ser10 phosphorylation was increased after incubation with 2.5 lM of Dovitinib, and returning to basal level after incubation with higher concentrations of Dovitinib (Fig. 1C). Consistent with this, the activity of CDK1, as indicated by the activating phosphorylation (Thr161), was reduced after treatment with 5-10 lM of Dovitinib. These data suggested that Dovitinib could both stimulate mitosis and induce G 2 arrest, both of which are consistent with the 4N DNA contents observed from the flow cytometry analysis. Moreover, the same DNA content could also be resulted from mitotic slippage, essentially generating G 1 cells with tetraploid DNA content [28].
To distinguish these possibilities, we next analysed Dovitinib-treated cells dynamically using time-lapse microscopy. HONE1 stably expressing a histone H2B-mRFP was generated and subjected to livecell imaging to track individual cells. As expected, control cells underwent mitosis randomly over the 24 hrs imaging period ( Fig. 2A). In agreement with the flow cytometry analysis, the cell cycle was not significantly affected by 0.1-1 lM of Dovitinib. In contrast, higher concentrations of Dovitinib prevented cells from entering mitosis (quantified in Fig. 2B). Notably, for cells that could enter mitosis, Dovitinib also induced defects in mitotic exit. We will address this mitotic delay in more detail in the next section.
To ensure that the G 2 cell cycle arrest was not restricted to NPC cell lines, we also investigated the responses of HeLa cells to Dovitinib. Figure 3A shows that Dovitinib also induced a G 2 /M delay in HeLa cells. Single-cell analysis revealed that while most cells exhibited an interphase arrest (Fig. 3B), a subset of cells underwent a protracted and defective mitosis (see below). Entry into mitosis over the 24 hrs period was reduced to~40% in the presence of 2.5 lM of Dovitinib (Fig. 3C). Finally, a G 2 cell cycle was also induced by Dovitinib in Hep3B, a cell line from hepatocellular carcinoma origin (see Fig. 4

below).
As a population of cells containing G 1 DNA content was present after exposure to Dovitinib (in particular 2.5 lM, Figs 1A and 3A), we next examined if entry into S phase was perturbed. Figure 3D shows that BrdU incorporation was not affected following treatment with 2.5 lM of Dovitinib. The lower level of BrdU incorporation at 5 lM of Dovitinib simply reflected the lower percentage of G 1 cells. These results indicated that Dovitinib did not significantly affect G 1 -S transition. Taken together, these results indicated that at micro-molar range, Dovitinib induces a G 2 cell cycle arrest in different cancer cell lines.

Dovitinib induces defective mitotic exit and promotes mitotic slippage
In addition to inducing a G 2 phase arrest, treatment of HONE1 cells with 2.5-5 lM of Dovitinib also resulted in mitosis more protracted than in control cells (Fig. 2B). This is consistent with the increase in histone H3 Ser10 phosphorylation (Fig. 1C). Furthermore, many of the cells were unable to exit mitosis properly and underwent mitotic slippage, resulting in a single daughter cell ( Fig. 2A).
Similar results were obtained with Dovitinib-treated HeLa cells. Increasing concentrations of Dovitinib caused a decrease in mitotic entry and an increase in the duration of mitosis of the remaining cells (Fig. 3C). While the duration from DNA condensation to anaphase was~50 min. in control cells, this was extended to~500 min. in Dovitinib-treated cells. All the cells exited mitosis with mitotic slippage after treatment with 5 lM of Dovitinib (Fig. 3B).
Dovitinib-treated cells could form a metaphase plate similarly as control cells (data not shown), suggesting that early events of mitosis were not affected. However, the metaphase cells were unable to proceed to anaphase properly, indicating that the extension of mitosis was mainly because of a delay from metaphase to anaphase. The DNA eventually decondensed either without a proper anaphase (an example is shown in Fig. 3E) or with defective anaphase that resulted in cytokinesis failure (an example is shown in Fig. 3F).
As mitotic entry was prevented by the Dovitinib-mediated G 2 arrest, we next studied the defective mitosis in more detail using a strategy that circumvented the G 2 arrest. Cells were first trapped in early mitosis using nocodazole before released into fresh medium containing Dovitinib (Fig. 3G). As expected, control cells exited mitosis normally with anaphase and cytokinesis. By contrast, mitotic slippage occurred in cells treated with Dovitinib after a relatively long mitosis.
Notably, although all cell types we examined were arrested at G 2 / M after Dovitinib treatment, not all of them displayed defective mitosis. Flow cytometry analysis indicated that exposure to Dovitinib triggered a G 2 /M arrest in Hep3B cells (Fig. 4A). Live-cell imaging revealed that the mitotic index was reduced in a dose-dependent manner ( Fig. 4B and C). Unlike HONE1 and HeLa cells, however, the length of mitosis was not affected in Hep3B. Instead, a large portion of cells underwent apoptosis directly (Fig. 4B). Collectively, these results indicate that a portion of Dovitinib-treated cells were unable to undergo anaphase properly, resulting in mitotic slippage.

Dovitinib induces DNA damage and promotes inhibitory phosphorylation of CDK1
As part of the effects of Dovitinib resemble that of DNA damage, we next investigated if the G 2 DNA damage checkpoint is activated after Dovitinib challenge. Figure 5A shows that exposure of HONE1 cells to Dovitinib resulted in an increase in CDK1 Tyr15 phosphorylation. Interestingly, p53 was also activated by treatment with Dovitinib. However, the fact that HONE1 cells were arrested mainly at G 2 phase (Fig. 1A) suggested that the p53-dependent G 1 DNA damage checkpoint may not be as effective as the G 2 DNA damage checkpoint in causing a cell cycle arrest.
Dovitinib also promoted CDK1 Tyr15 phosphorylation in HeLa cells (Fig. 5B). Phosphorylation of another inhibitory site of CDK1 (Thr14) was similarly increased. Although Dovitinib promoted the accumulation of cyclin B and the phosphorylation of the activating T-loop (Thr161; because of the accumulation of G 2 cells), these results indicated CDK1 was inactivated by inhibitory phosphorylation.
To see directly if Dovitinib could induce DNA damage, we analysed whether DNA breaks were generated by single cell gel electrophoresis assay (comet assay). Figure 5D shows that DNA damage was induced by Dovitinib similarly as DNA damaging agents including ionizing radiation and Adriamycin. We also use phosphorylation of histone H2AX (c-H2AX) as an indicator of DNA double-stranded breaks [32]. An increase of c-H2AX foci as detected with immunostaining ( Fig. 5E) or c-H2AX signal as detected with immunoblotting ( Fig. 5F) was induced by Dovitinib.     Taken together, these data indicate that the Dovitinib-mediated G 2 cell cycle arrest was associated with an activation of the DNA damage checkpoint and an increase in CDK1 inhibitory phosphorylation.

Dovitinib-mediated G 2 arrest can be abrogated by disrupting the DNA damage checkpoint
To test the hypothesis that Dovitinib activates the G 2 DNA damage checkpoint, we next examined if small molecule inhibitors of the checkpoint could overcome the Dovitinib-mediated arrest. A CHK1 inhibitor UCN-01 [33] was initially used to disrupt the checkpoint. As expected, Dovitinib stimulated the phosphorylation of CDK1 Tyr15 , thereby preventing cells from entering into mitosis (as indicated by the decrease in histone H3 Ser10 phosphorylation). Addition of UCN-01 reversed the phosphorylation of CDK1 Tyr15 and stimulated mitotic entry in both HeLa (Fig. 6A) and Hep3B (Fig. 6B) cells, indicating that CHK1 inhibition could abrogate the Dovitinib-mediated G 2 arrest. Similar uncoupling of the Dovitinibmediated G 2 arrest was induced by a more specific CHK1 inhibi-  tor AZD7762 [34] in both HeLa (Fig. 6C) and Hep3B (Fig. 6D) cells.
We also examined if inhibition of WEE1 (a downstream target of CHK1) could abrogate Dovitinib-mediated arrest. A WEE1 inhibitor MK-1775 [35] was added to Dovitinib-treated Hep3B cells followed by tracking with time-lapse microscopy (Fig. 6E). As expected, Dovitinib two daughter cells at the lower concentration of Dovitinib. At higher concentration, however, nearly all the cells underwent protracted mitosis and mitotic slippage, generating tetraploid G 1 cells. These results suggested that even when the Dovitinib-mediated G 2 DNA damage checkpoint is bypassed, the resulting mitosis is still defective. To see if these results could be reproduced in another cell line, HeLa cells were subjected to similar analysis (Fig. 7A). As described above, treatment of HeLa cells with 2.5-5 lM of Dovitinib induced a G 2 arrest. Inhibition of CHK1 with UCN-01 or AZD7762 uncoupled the arrest and promoted mitosis. As controls, UCN-01 or AZD7762 alone did not affect the cell cycle. Interestingly, although the WEE1 inhibitor MK-1775 could also uncouple the G 2 arrest induced by 2.5 lM of Dovitinib, it was unable to overcome the checkpoint induced by a higher concentration of Dovitinib. It is possible that at higher concentrations of Dovitinib, inhibition of WEE1 alone could not bypass the checkpoint without activating the CDC25s. As CHK1 lies upstream of both WEE1 and CDC25s, its inhibition could be more effective in uncoupling the checkpoint. Finally, Dovitinib-mediated cell cycle arrest in another cell line, HONE1, was also reduced after inhibition of WEE1 or CHK1 (Fig. 7B).

Dovitinib-mediated G 2 arrest can be promoted by inhibition of poly(ADP-ribose) polymerases
Given that Dovitinib could activate the DNA damage checkpoint, we predicted that its effect could be increased if the DNA damage repair pathway is compromised. One of the earliest responses to DNA damage is the binding of poly(ADP-ribose) polymerases (PARPs) to damaged sites. The binding of PARPs to the site of DNA damage activates its catalytic activity and promotes local poly(ADP-ribose)-dependent recruitment of DNA repair enzymes required for efficient DNA repair. Cells were treated with Dovitinib alone or in combination with a poly (ADP-ribose) polymerase inhibitor Olaparib (AZD2281). While a partial G 2 /M arrest was induced in HeLa cells after incubation with 5 lM of Dovitinib for 32 hrs, a more robust arrest was induced by combining Olaparib with Dovitinib (Fig. 7C). This suggested that inhibition of PARP-mediated DNA repair promoted the activation of the G 2 DNA damage checkpoint by Dovitinib. Longer incubation with Dovitinib induced apoptosis (as indicated by sub-G 1 population), which also appeared to be promoted by Olaparib (Fig. 7C). This was confirmed by the increase of apoptosis-mediated cleavage of PARP1 (Fig. 7D). These data indicate that inhibition of PARPs increased the Dovitinibmediated cell cycle arrest and apoptosis, further supporting the idea that Dovitinib could induce DNA damage.

Discussion
In this study, we demonstrated that Dovitinib promoted cell cycle arrest in both G 2 and mitosis. A G 2 /M arrest was observed in multiple cell lines, including several from NPC (HONE1, HNE1, CNE2, and C666-1; Fig. 1A), cervical carcinoma HeLa (Fig. 3A), hepatocellular carcinoma Hep3B (Fig. 4A), and several other cell lines from colon and fibroblasts origin (our unpublished data). Using live-cell imaging of histone H2B-GFP-labelled cells, we found that Dovitinib triggered delays in mitosis and G 2 in a concentration-dependent manner. While interphase (G 2 ) arrest was induced after incubation with relatively high concentrations of Dovitinib, a delay in mitosis was also triggered with the lower concentrations (Figs 2 and 3). These cells exhibited a delay in metaphase-anaphase transition and exited mitosis with mitotic slippage or cytokinesis failure. We also noted that different cell lines displayed different sensitivity to Dovitinib. For example, Dovitinib induced extensive apoptosis in Hep3B, which may explain why no mitotic delay was observed (Fig. 4).
Similar to Dovitinib, several small molecule inhibitors of multiple RTKs have been reported to trigger G 2 /M cell cycle arrest. For example, Sorafenib can induce a G 2 /M arrest in cells from hepatocellular carcinoma [36]. Foretinib, a VEGFR-2 and c-MET inhibitor, has also been shown to induce a G 2 /M arrest in cells from hepatocellular carcinoma [37] and ovarian cancer [38]. However, the underlying mechanisms of these cell cycle arrests are not known.
Why Dovitinib promoted delays in G 2 or mitosis? It is conceivable that as a non-specific kinase inhibitor, Dovitinib could also target protein kinases involved in cell cycle control. Indeed, the metaphaseanaphase delay caused by Dovitinib (Fig. 3E) was reminiscent to that caused by inhibitors of the mitotic kinase PLK1 [39,40]. However, Dovitinib does not have high affinity for serine/threonine-specific kinases [7]. A more probably explanation is based on the recent finding that Dovitinib could also target topoisomerases [41]. Hasinoff et al. realized that Dovitinib structurally resembles the DNA minor groove binding dye Hoechst 33258. Because DNA binding compounds are often topoisomerase inhibitors, Hasinoff et al. also investigated the ability of Dovitinib to interact with topoisomerases and found that Dovitinib can inhibit the catalytic activity of topoisomerase I and topoisomerase II [41].
Topoisomerase II is able to catenate and decatenate DNA by introducing a transient double-strand break into DNA and then pass a second DNA strand through the break before resealing it. Together with proteins including condensin, Topoisomerase II contributes to the compaction of the chromosome, thereby ensuring normal mitotic chromatin resolution [42]. Many topoisomerase inhibitors are wellknown anti-cancer drugs. These topoisomerase inhibitors are either classic topoisomerase poisons, which cause DNA strand breaks through stabilization of intermediate cleavable complexes, or true catalytic inhibitors that prevent decatenation and can induce mitotic slippage [43]. Hasinoff et al. found Dovitinib can stabilize the cleavable complexes and acts as a topoisomerase II poison [41]. This is in agreement with our finding that Dovitinib could induce DNA damage as measured by comet assays (Fig. 5D) and c-H2AX (Fig. 5E and F). Accordingly, Dovitinib could also activate the G 2 DNA damage checkpoint. Phosphorylation of both CHK1 and CDK1 was promoted by Dovitinib (Fig. 5C). In support of this, the inhibition of CDK1 and prevention of mitosis could be abrogated with CHK1 inhibitors (Figs 6  and 7). Our data are in accordance with the possibility that Dovitinib also could act as a catalytic inhibitor of topoisomerase II, in particularly at relatively lower concentrations, because of the defective mitotic exit (Fig. 3). Hence it is possible that while relatively low concentrations of Dovitinib inactivated topoisomerase II's catalytic activity, higher concentrations acted as a classic topoisomerase poison and stabilized intermediate cleavable complexes.
Activation of the G 2 DNA damage checkpoint by Dovitinib has several implications to potential anti-cancer therapies. One possibility is that Dovitinib may be able to act synergistically with other DNA damaging agents. Combined therapy may promote a stronger G 2      arrest than Dovitinib alone without the side effects of using high dosages of drugs. As the Dovitinib-mediated G 2 arrest could be uncoupled with small inhibitors of CHK1 or WEE1, another possibility is that Dovitinib can be used in combination with reagents that uncouple the checkpoint (Fig. 7A and B). Forcing damaged cells into premature mitosis triggers mitotic catastrophe and has been developed as a strategy for anti-cancer therapies. Along the similar line, inhibition of DNA repair by, for example, inhibitors of poly(ADP-ribose) polymerases, may promote the DNA damage induced Dovitinib (Fig. 7C and  D). In this connection, cancer cells that already have defects in the DNA damage checkpoints or DNA repair pathways may be sensitized to Dovitinib in a synthetic lethal manner.
It should be noted, however, that the concentrations of Dovitinib that induced the G 2 or mitotic delay was relatively high (1-5 lM) compare to that required to inhibit RTKs (IC 50 <10 nM) [7]. In mouse xenograft models, dose levels associated with anti-tumour efficacy are between 10 and 30 mg/kg, with the corresponding C max between 0.4 and 1.9 lM [7]. The clinical exposure of Dovitinib is less than that used in animal models. In one study, the C max was found to be between 0.01 and 0.3 lM [44]. Whether a G 2 or mitotic delay could be achievable with the doses of Dovitinib used in clinical trials should be considered.
In conclusion, Dovitinib induced mitotic exit defects and DNA damage, resulting in mitotic slippage and activation of the G 2 DNA damage checkpoint, respectively. These results suggest additional directions for use of Dovitinib.