A phenanthrene derived PARP inhibitor is an extra-centrosomes de-clustering agent exclusively eradicating human cancer cells

We have recently reported the selective eradication of human triple negative mammary cancer cells MDA-231 by phenanthrene derivatives (also acting as potent inhibitors of polyADP-ribose polymerases) [1]. These compounds included the phenanthridines PJ-34 and Phen and the isoquiniline Tiq-A. They were originally designed to protect neuronal cells in the central nervous system from cell death evoked by high activity of PARP-1 in response to DNA damage caused by brain injury, stroke or inflammation [2, 3]. We found that these molecules cause G2/M transition arrest in the cell cycle of both mammary cancer cells MDA-231 and normal epithelial cells MCF-10. However, while G2/M arrest was permanent in the cancer cells, and was accompanied by their massive cell death, normal mammary epithelial cells overcame the cell cycle arrest and continued to proliferate normally in the presence of these phenanthrene derivatives [1]. The most potent compound was PJ-34, which also efficiently prevented the development of MDA-231 xenotransplants in nude mice without inducing detectable toxic effects in the animals [1].

The current results outline a mechanism that apparently underlies the exclusive cytotoxicity of PJ-34 in these human mammary cancer cells. We found that this molecule acts as a centrosomes de-clustering agent in cells with supernumerary centrosomes, which are most abundant in these mammary cancer cells and in most human cancers [4, 5].

Bipolar centrosomes assembly during metaphase is crucial for bipolar spindle formation and accurate chromosomes segregation in cells undergoing mitosis [6, 7]. To accomplish these tasks, cells dividing with more than two centrosomes have developed a yet unresolved molecular mechanism, clustering their extra-centrosomes at two poles [4, 5, 8]. Failure of this bipolar centrosome assembly causes multipolar spindle structures and aberrant chromosomes segregation that prevent normal cell division [5]. This may lead either to 'mitotic slippage' where cells 'slip' out of mitosis to re-enter G1 without satisfying the spindle assembly checkpoint (SAC), or to induction of 'mitotic catastrophe cell death' [5, 9, 10] or 'anaphase catastrophy' [10]. Although cell death induced by failure to accomplish mitosis has been well documented, the underlying molecular mechanisms are still poorly understood [5, 10]

Extra centrosomes are most abundant in cells of human solid cancers and in some of the human hematological malignancies [4, 5], whereas in normal somatic cells the number of centrosomes is restricted, via several control mechanisms, to two centrosomes per cell [8, 9, 11]. This difference between normal and cancer human cells raised the idea of using compounds that interfere with the bipolar clustering of extra centrosomes to achieve selective eradication of cancer cells without harming normal tissues [5, 12].

Our findings confirm this hypothesis by identifying a molecule, which prevents extra centrosomes clustering in mitosis and exclusively eradicates human cancer cells with supernumerary centrosomes without impairing normal proliferating cells.

We examined several potent PARP inhibitors, including the phenanthrene derivatives PJ-34, Tiq-A and Phen and the non-phenanthrene derivatives ABT-888 and BSI-201 (Figure 1) in human cancer cell types with a high occurrence of extra-centrosomal cells (>50%).

In our previous experiments PJ-34 caused G2/M arrest and a massive cell death in MDA-231 mammary cancer cells without impairing normal human mammary epithelial MCF-10 cells [1]. Given that MDA-231 cells have high occurrence of extra-centrosomes, and the interference with bipolar clustering of supernumerary centrosomes causes G2/M arrest [5, 16, 17], we examined the possibility that PJ-34 affects extra-centrosomes clustering in mitosis.

Our experiments disclosed a remarkable activity of PJ-34 on disruption of the bipolar clustering of supernumerary centrosomes in mammary MDA-231 cancer cells, causing distorted multi-polar spindles as early as 6 hours after PJ-34 application, followed by massive cell death within 48-72 hours (Figures 2a, b and Additional File 1). PJ-34 was applied only once to the cells, 24 hours after seeding, at twice to 3-times higher concentration than that completely inhibiting the activity of PARP-1 [3]. The cytotoxic activity of PJ-34 was identified in randomly selected fixed or live MDA-231 cells, scanned by confocal microscopy (Figures 2a, b and Additional File 1; Methods). Also, PJ-34 did not seem to impair the structure or formation of centrosomes, as indicated by centrosomes double labeling with centrin1 and γ-tubuline (Figure 2a) and by confocal imaging of centrosomes in live MDA-231 cells in the interphase (Additional File 2).

These findings associated the remarkable cytotoxic activity of PJ-34 in MDA-231 cells with de-clustering of their extra-centrosomes in mitosis, causing multi-polar distorted spindles and aberrant chromosomes segregation that lead to 'mitotic catastrophe cell death' [5].

Other phenanthrene derived molecules acting as PARP inhibitors, Phenanthridinon (Phen), and Tiq-A (Figure 1; [18, 19]), similarly interfered with centrosomes clustering in MDA-231 cells. However, their effect was milder and accompanied by a lower percentage of cell death (Figure 3).

In contrast, non-phenanthrene derivatives acting as potent PARP inhibitors, BSI-201 and ABT888 [18, 19], did not share this capability. Even when used at high concentrations (2-4 times higher than concentrations completely inhibiting the activity of PARP-1) they did not impair the bifocal clustering of extra centrosomes in these cells (Figure 3).

To further test the suggested contribution of PJ-34 to extra-centrosomes de-clustering, we examined its activity in other human cancer cell-lines with high occurrence of supernumerary centrosomes.

PJ-34 similarly eradicated the tested cells, including lung (H1299), colon (DLD-1), ovarian (HeyA8), and pancreatic (Panc-1) cancer cells. PJ-34 (20-30 μM) caused G2/M arrest in these cells and a subsequent massive cell death, observed 48-96 hours after PJ-34 application (Figures 4 and 5). In accordance, PJ-34 prevented the bipolar clustering of supernumerary centrosomes in these cells, causing multipolar spindles and aberrant chromosomes alignment detected within 18 hours after PJ-34 application (Figure 6). For the tested five types of cancer cell-lines, there was a significant correlation between the fraction of cells with multi-polar spindles and the reduction in cell viability (as indicated by the reduction in ATP level) of the treated cells (Figures 2a, 4 and 6; r = 0.92; p < 0.0001). Thus, cell death could be attributed to de-clustering of their extra-centrosomes.

In contrast, normal human proliferating cells were resistant to the cytotoxic activity of PJ-34, Tiq-A and Phen in human cancer cells. Even at the high concentrations used and the long incubation periods, the most potent compound, PJ-34, did not interfere with the cell cycle of benign human proliferating cells. The tested cells included human mammary epithelial cells MCF-10, mesenchymal cells prepared from human thymus and human umbilical vein endothelial cells (Huvec) (Figure 7). As for Huvec, PJ-34 did not affect their cell cycle within 48 hours incubation (Figure 7). These cells did not survive longer incubation periods even without treatment (> 5% cell death was measured in untreated Huvec cells after 72 hours incubation).

In these three types of benign cells, PJ-34 did not interfere with the formation and assembly of centrosomes (two centrosomes at the poles of bifocal spindles), nor with chromosomes segregation. Bipolar assembly of centrosomes, bifocal spindle formation and chromosomes segregation were not impaired in the normal proliferating cells, even after long incubation with high concentrations of PJ-34 (Figure 8).

Our results suggest that the exclusive eradication of human cancer cells with extra centrosomes by the phenanthridine derivative PJ-34 is attributable to its extra-centrosomes de-clustering activity in mitosis (Figures 2, 4, 5, 6, Additional File 1 and Additional File 2).

According to our observations, PJ-34 did not impair centrosomes in both human normal and cancer cells, nor did it interfere with the formation of centrosomes in the interphase of multi-centrosomal cancer cells or impaired the bi-polar assembly of centrosomes in mitosis of normal proliferating cells with two centrosomes (Figures 2a, 8, Additional File 1 and Additional File 2). These features may underlie its outstanding capability to exclusively eradicate human cancer cells while keeping normal proliferating cells un-affected (Figures 5 and 7).

Theoretically, in cancer cells with high occurrence of extra-centrosomes (>50%) when mitosis is accompanied by eradication of multi-centrosomal cells, an exponential reduction of the fraction of bi-centrosomal cancer cells is expected. In this case, the higher their proliferation rate the more rapidly will these cancer cells be eradicated by extra-centrosomes de-clustering agents.

The indicated interference of the tested phenanthrene derivatives with the bipolar clustering of supernumerary centrosomes was not shared by non-phenanthrene derivatives acting as potent PARP inhibitors as well (Figure 3). This questions an exclusive role of PARP inhibition in extra centrosomes de-clustering. Nevertheless, these results could also reflect a possible, yet un-identified selective inhibition of PARP isoforms by phenanthridines.

Results obtained from genome-wide RNAi screen in multi-centrosomal cells, including embryonic Drosophila S2 cells and human malignant cells, identified genes required to suppress multipolar mitosis. Proteins involved in the organization and regulation of the cytoskeleton, including Kinesin HSET/Ncd (which is not required for polar organization in normal somatic cells) and the four chromosomal passenger complex (CPC) components, Aurora-B, INCENP, Survivin, and Borealin were identified among the main proteins required for bifocal clustering of extra-centrosomes in these cells [12, 17]. Evidence for the necessity of tankyrase-I and a putative human PARP-16 homolog in centrosomes clustering has been also provided [17, 20]. PolyADP-ribosylation of tankyrase-1 may contribute to spindle bipolarity by providing a static matrix, anchoring microtubule-associated motor proteins and spindle proteins [17, 20]. A specific role for PARP-16 in mitosis has not been identified yet [17].

PARP proteins are present in centrosomes [21‐24], and both Aurora-B and INCENP are targets for polyADP-ribosylation [22, 24]. In addition, supernumerary centrosomes were found in PARP-1 deficient cells [25]. Thus, in view of the exclusive interference of PJ-34 with extra centrosomes de-clustering in mitosis, a possible synergism between PARP activity and other mechanisms underlying extra-centrosomes clustering should be further investigated.

The indicated cytotoxic activity of the phenanthrene derivatives in human cancer cells with high occurrence of extra-centrosomes links for the first time between two well known, but poorly understood phenomena; Some anti-tumor effects of natural phenanthridines were first described in the early fifties [26‐28], and a higher cytotoxicity of some phenanthridines in cancer cells as compared to normal cells has been reported before [29, 30], although the underlying mechanisms have never been understood. In addition, cell-cycle arrest resulting from failure of centrosomes to accomplish bipolar clustering in multi-centrosomal cells has been acknowledged for nearly a century [6, 31, 32]. Nevertheless, the causal relationship between multiple centrosomes and malignancy has only recently been clarified [32‐35].

Mitotic-spindle microtubules are among the most effective targets for anti-cancer therapy [5, 10, 17, 36‐39]. Therefore efforts are invested in trying to find mechanisms exclusively targeting mitosis in cancer cells [40‐42].

Unlike the exclusive interference of PJ-34 with extra centrosomes bi-polar clustering in mitosis (Figures 2, 3, 4, 5, 6, Additional File 1 and Additional File 2), the currently known small molecules interfering with mitosis in cancer cells target microtubule polymerization (vinca alkaloids, taxanes, colchicines, and also griseofulvin). These compounds therefore impair normal cells as well, and cause severe side effects [36‐39].

The discovered potency of PJ-34 in preventing bipolar clustering of extra centrosomes in mitosis may pave the way for a new efficient therapy lacking intolerable side effects, which is based on compounds that exclusively target the unique dependence of many human cancer cells on their supernumerary centrosomes clustering for their survival.

Since many human cancer cells have high occurrence of multicentrosomal cells, the exclusive cytotoxic activity of PJ-34 in multi-centrosomal human cancer cells holds promise as a therapeutic tool for a selective chemotherapy in a wide range of human cancers, beyond the currently used selective therapeutic tools targeting specific genes/proteins that affect few cancer cell types [43‐45].

The identification of a molecule exclusively targeting the unique dependence of most human cancer cells on their supernumerary centrosomes bi-polar clustering for their survival, paves the way for a new highly selective cancer-targeting therapy, efficient in a wide range of human cancers.