Contribution of Soft Substrates to Malignancy and Tumor Suppression during Colon Cancer Cell Division

In colon cancer, a highly aggressive disease, progression through the malignant sequence is accompanied by increasingly numerous chromosomal rearrangements. To colonize target organs, invasive cells cross several tissues of various elastic moduli. Whether soft tissue increases malignancy or in contrast limits invasive colon cell spreading remains an open question. Using polyelectrolyte multilayer films mimicking microenvironments of various elastic moduli, we revealed that human SW480 colon cancer cells displayed increasing frequency in chromosomal segregation abnormalities when cultured on substrates with decreasing stiffness. Our results show that, although decreasing stiffness correlates with increased cell lethality, a significant proportion of SW480 cancer cells did escape from the very soft substrates, even when bearing abnormal chromosome segregation, achieve mitosis and undergo a new cycle of replication in contrast to human colonic HCoEpiC cells which died on soft substrates. This observation opens the possibility that the ability of cancer cells to overcome defects in chromosome segregation on very soft substrates could contribute to increasing chromosomal rearrangements and tumor cell aggressiveness.


Introduction
Over the last 10 years, it has become evident that cell behaviour not only depends on chemical cues but that mechanical properties of cellular environment play an as important role. This was spectacularly demonstrated by the landmark experiments of Discher's group who showed that mesenchymal stem cells can either differentiate into osteoblasts, fibroblasts or neurons depending upon the Young modulus of the adhesion substrate [1]. It is also well accepted that different cell types need substrates of different Young moduli to properly adhere and proliferate. Osteoblasts require Young moduli in the range of MPa to adhere whereas fibroblasts adhere on softer substrates whose moduli of about 10 kPa [2] and neurons grow on extremely soft substrates of about 1 kPa [1]. These distinctive values are in accordance to the Young moduli that characterize the tissues surrounding these different cell types. These results are of paramount importance for example in tissue engineering to design scaffolds allowing an appropriate growth of cells or in implant integration. Yet adhesion is not the only aspect that characterizes the cell behaviour: cell division is also a crucial aspect for cell fate. Our group started recently to examine the influence of the mechanical properties of the substrate on cell division [3]. These data highlighted that the mechanical properties of the substrate play a critical role in chromosome segregation during mitosis of epithelial cells. Indeed, we observed a progressive increase in chromosomal segregation abnormalities with decreasing substrate stiffness in noncancerous rat kangaroo kidney cells PtK2 [3]. Moreover, soft substrates (below 50 kPa) were described as a physical microenvironment barrier almost completely inhibiting the PtK2 cells [3].
Over the last years, it has been established that tissue stiffness influences tumor progression and can promote the malignant behaviour [4][5][6]. By introducing cancer cells into 3dimensional fibrin matrices, Liu et al. showed that soft matrices of Young modulus about 100 Pa promoted the growth of round colonies with increasing aggressiveness when xenografted in immunodeficient mice [7]. Very recently, Tang et al. revealed the attenuation of cell mechanosensitivity of tumor cells when cultured on soft substrates [8]. In colon cancer, a highly aggressive disease, progression through the malignant sequence is accompanied by increasing chromosomal rearrangements [9][10][11][12]. To colonize target organs, invasive cells cross several tissues of various elastic moduli (as example, 175, 918, 320, 120 and 640 Pa for basement membrane, stroma, lymph, lymph node and liver, respectively) [2,4] and, while most of these cells die during their journey, few resist and can generate metastases [13]. Whether soft tissue increases malignancy or in contrast limits invasive cell spreading remains an open question. Using polyelectrolyte multilayers films (PEM) [14][15][16][17][18], we revealed that human SW480 colon cancer cells displayed increasing frequency in chromosomal segregation abnormalities when cultured on substrates with decreasing stiffness (Figure 1) and [3]. In the present paper, we report that substrates with stiffness of 50 kPa and lower cause massive death of mitotic cells but that few cells resist and achieve mitosis by overcoming abnormal chromosomal segregation. For this purpose, synchronized SW480 cells were seeded on a series of films made of a poly-L-lysine/hyaluronic acid (PLL/HA) 24 stratum capped by poly(styrene) sulfonate/polyallylamine (PSS/PAH) n strata (n = 0, 1 and 2; increasing n increases substrate stiffness [19]) and followed by live-cell imaging.

Influence of soft substrate on the tumor mitotic progression
To determine whether tumor cells are able to progress through mitosis on very soft substrates, SW480 cells, synchronized using the mitotic shake-off method, were seeded on PEM films with decreasing stiffness (Young moduli decreasing from 50 down to 0 kPa, Table 1) and followed by live-cell imaging during 2h30. The films were composed of a (PLL/HA) 24 stratum capped by a second (PSS/PAH) n stratum (n = 0, 1 and 2). A typical example of confocal z-section observation of a film composed of PLL/HA stratum and PSS/PAH capping is displayed in Figure 2. Even though the surface chemistry of glass and of the polyelectrolyte multilayers is different, we demonstrated in a previous work that the amount of FBS proteins deposited on the surface does depend neither on the surface chemistry nor on the number of layers constituting the polyelectrolyte multilayers film [16]. Moreover, cells feel essentially the proteins from the serum that adsorb on the surface prior to cell deposition. Thus the main parameter that changes between glass, E0 and E50 / E20 is rigidity and it should be at the origin of the differences in cell behaviour observed in our system. On E 0 , cells rapidly went through a lytic process, as indicated by the release of cytoplasm in the culture medium observed in 100% of the cases ( Figure 3A, arrowhead in row E 0 , Movie S1). However, on E 50 and E 20 substrates, the follow up of chromosome segregation, DNA decondensation and cytokinesis ( Figure 3A and Movie S2-S4) revealed that respectively 60% and 10% of cells were able to achieve mitosis in 2h30 ( Figure 3C). The remaining cells either lyse (18% on E 50 and 88% on E 20 ) or kept blocked in mitosis (22% on E 50 , and 2% on E 20 ). Tilghman et al. showed that cancer cells cultured on soft polyacrylamide gel substrates exhibited a longer cell cycle, due to an extension of the G1 phase of the cell cycle, compared to cancer cells growing on more stiff substrates [20]. In order to demonstrate that the significant proportion of SW480 cells able to progress in mitosis on E 50 was related to the cancerous nature of these cells, their behavior were compared to those of the non-cancerous human colonic epithelial cells HCoEpiC. Production of mitotic HCoEpiC cells by mechanical shakeoff was greatly reduced. Thus, standard asynchronous HCoEpiC cell cultures were used to investigate the influence of E 50 on human colonic epithelial cells. The results show that after 6h of culture on E 50 , asynchronized HCoEpiC cells adopted a round shaped morphology ( Figure 4A) unlike the spread shape of these cells observed on glass ( Figure 4A). On E 50 , HCoEpiC cells went through a lytic process, as indicated by the release of cytoplasm in the culture medium in 100% of the cases ( Figure  4A). Some of these cells showed fragmentation of their nucleus suggesting death by apoptosis ( Figure 4C). Consistent with these observations, no assembly of microtubules and actin filaments could be observed by immunofluorescence experiments using antibody specific for α-tubulin and phalloidin ( Figure 4C). All interphase HCoEpicC cells died on E 50 ( Figure  4B) revealing that these cells are obviously unable to progress in the cell cycle and to re-enter in mitosis. We can point out that our studies were performed on substrates with Young moduli in the range of 1-50 kPa and we observed that non-cancerous cells could not survive on such soft substrates. This result seems at first sight contradictory with the observation of cell division in colon whose Young modulus is 2 -25 kPa [21]. Yet, in colon cells are in a specialized 3D tissue with cell/cell contacts which strongly influence cell behaviour. Such an effect is not reproduced in our 2D in vitro experiments. Nevertheless, our in vitro model is more appropriate for single tumour cells or small groups of cells that escape the tumour mass to invade the stroma and engage in the process leading to metastasis formation after a long journey through tissues of very different Young moduli. Altogether, our results show that the decrease of the stiffness correlates with an increase in cell lethality in vitro for healthy as well as cancerous cells. Nonetheless, very importantly, we observed that some cancer cells can escape from very soft substrates and can achieve mitosis. These findings are highly relevant for both normal homeostasis of solid tissues and for pathological settings.

Behavior of tumor cells bearing chromosome segregation abnormalities
Time-lapse monitoring chromosome segregation showed that SW480 cells seeded on E 50 and on E 20 displayed lagging chromosomes ( Figure 3B, arrows) indicating defects in chromosomal segregation mechanisms. In late telophase, the nuclei reformed and cytokinesis completed ( Figure 3B, 3D and Movie S5 and S6). Among the SW480 cells progressing through mitosis 4.8%, 11% and 13% showed chromosome segregation abnormalities on glass, E 50 and E 20 , respectively. These results showed that despite increasing frequency of abnormalities induced by soft substrates, some SW480 cancer cells are able to progress through mitosis. Which is consistent with observation made on rigid substrates that cells having deficient spindle checkpoint mechanisms may proceed through mitosis even in the presence of chromosomes misconnected to the spindle [22].

The β1-integrin engagement by mitotic tumor cells in response to soft substrates
Interactions with the extracellular matrix through integrins have been shown to influence different aspects of mitotic spindle organization [23,24]. In particular, it is well known that mitotic cells become rounded in preparation for cytokinesis and remain attached to the substrate through retraction fibres via integrin engagement [25]. The retraction fibres provide resistive force that propagates within microtubules to orient the mitotic spindle [23][24][25][26]. We thus investigated the effects the different substrate stiffness on β1-integrin in SW480 cells. In response to soft substrates (E 50 and E 20 ), β1-integrin engagement measured by ligand-induced binding sites (activated β1-integrin LIBS) using conformational-specific antibodies was unchanged compared to glass substrate ( Figure 5A-C, the MAPK lane correspond to control loading). Interestingly, these data are in contrast with a previous study using non-tumor PtK2 cells showing progressively decreases of β1-integrin engagement on soft substrates [3]. These observations are in accordance with other studies demonstrating that murine breast cancer cells retain high levels of active β1-integrin after 24h of culture on a soft substrate [27]. Integrin accumulation in the cell mid-zone is also necessary to induce mitotic cell adhesion and to support cytokinesis by providing mechanical anchoring for the contractile actomyosin ring [28]. To examine the localization of β1-integrin, immunofluorescence performed in cells in telophase revealed β1-integrin accumulation in the mid-body, on E 50 and on E 20 , like on glass ( Figure 5D). On glass, E 50 and Table 1. Apparent elastic modulus of (PLL/HA) 24 -(PSS/ PAH) n with n = 1 and 2.   E 20 actin accumulated also in this cell mid-zone ( Figure 5D). These results suggest continuous β1-integrin engagement during mitosis despite the soft substrates, compatible with the involvement to support cytokinesis.

Mitotic spindle organization of tumor cells in response to soft substrates
We next checked mitotic spindle assembly on soft matrices by immunofluorescence with anti-α-tubulin and DNA staining with Hoechst. The mitotic spindles of SW480 cells, visible on stiff substrates (glass), were preserved on E 50 and even on E 20 ( Figure 6A and B). This result contrasts with that observed for non-cancerous mitotic PtK2 cells seeded on soft matrix since for Young modulus ≤ 50 kPa, the mitotic spindle could not be observed [3]. Consistent with our living cell analysis ( Figure 3B and D), abnormal chromosome segregation events could be observed ( Figure 6B and D), however, without evidence for multipolar or monopolar spindles on the different substrates probably due to β1-integrin engagement maintained on the very soft substrate. Indeed, multipolar spindles were observed for cells bearing mutated β1-integrin [23].

Rac1 expression of tumor cells in response to soft substrates
Rac1, proteins of the small GTPase Ras family, are involved in the orientation of the mitotic spindle and its activity increases at the plasma membrane of polar sides during cytokinesis [29]. We further synchronized a collection of SW480 cells and analyzed Rac1 expression through Western blot experiments [29]. There were no differences in Rac1 expression for mitotic SW480 cells seeded either on soft substrates (E 50 and E 20 ) or on stiff substrate (glass) (Figure 7A-C). Our results showed that mitotic tumor SW480 cells maintain Rac1 expression on soft substrates. On the contrary, we previously observed that noncancerous PtK2 cells progressively decreases this expression on soft substrates [3]. Importantly, β1-integrin engagement and Rac1 expression maintained on soft substrates (E 50 , E 20 ) were not sufficient to allow massive division of tumor SW480 cells.  Indeed, on these substrates only 60% (E 50 ) and 10% (E 20 ) of cells are able to progress in mitosis

DNA replication activity of tumor cells in response to soft substrates
To determine whether SW480 cells that were able to progress through mitosis were capable of reenter cell cycle, we investigated their capability to undergo DNA replication 4h after being seeded on different substrates. SW480 cells seeded on E 50 and E 20 show site of replication uniformly distributed in the nucleus ( Figure 8A) with respectively 60% and 23% of cells with nuclear BrdU signal ( Figure 8B). It is noteworthy that the percentages of SW480 cells incorporating BrdU correlated with the percentage of cells achieving mitosis on E 50 and E 20 ( Figures 3C and 8B), suggesting that SW480 cells able to complete chromosome segregation on soft substrates are further able to undergo a new cycle of DNA replication.

Conclusion
In conclusion, we report that despite massive cell death on extremely soft substrates (E 0 ), tumor cells like the SW480 colon cancer cells, even when bearing abnormal chromosome segregation, resist to the very soft substrates (E 20 and E 50 ) and achieve mitosis. These findings might be highly relevant for the pathophysiology of cancer and the dissemination of colon tumor cells. Indeed, their cancerous nature at least some of the tumor cells might help them to overcome chromosomal segregation abnormalities linked to the change in substrate stiffness and therefore escape the soft substrates to pursue their journey up to the site of metastasis formation. Moreover, this ability to overcome segregation abnormalities could result in more chromosomal rearrangements, which may contribute to increasing tumor cell aggressiveness. Further investigating the response of tumor cells to physical environmental changes may help to identify new potential targets for anticancer therapy.  (Table 1) [19]. The short hand notations E 0 , E 20 and E 50 are for (PLL/HA) 24 , (PLL/HA) 24 -(PSS/ PAH) n films with n = 0, 1 and 2, respectively.

PEM characterization
CLSM observations were performed with Zeiss LSM 510 microscope using x40/1.4 oil immersion objectif. FITCfluorescence detected after excitation at 488 nm with cutoff dichroic mirror 488 nm and emission band-pass filter 505-530 nm. Rho-fluorescence detected after excitation at 543 nm, dichroic mirror 543 nm, and emission long pass filter 585 nm.

Live-cell imaging
For dividing assays, cells were incubated 20 min with Hoechst 33242 (0.1 µg/mL, Sigma) prior to mechanical shakeoff, replated at 1.2x10 4 per cm 2 on film-coated coverslips and mounted in a Ludin Chamber (Life Imaging Services, Basel Switzerland) at 37°C, 5% CO 2 , on a Leica DMIRE2 microscope equipped with a 40× HCX PL APO PH2 (0.75 NA) objective and a Leica DC350FX CCD (Leica FW4000 software). Images were acquired every 5 min for 2h30, by fluorescence and phase contrast.

Supporting Information
Movie S1. Movie corresponding to Figure 3A