At the onset of mitosis, most adherent cells undergo cell retraction characterised by the disassembly of focal adhesions and actin stress fibres. Mitosis takes place in rounded cells, and the two daughter cells spread again after cytokinesis. Because of the well-documented ability of the small GTPase Rap1 to stimulate integrin-dependent adhesion and spreading, we assessed its role during mitosis. We show that Rap1 activity is regulated during this process. Changes in Rap1 activity play an essential role in regulating cell retraction and spreading, respectively, before and after mitosis of HeLa cells. Indeed, endogenous Rap1 is inhibited at the onset of mitosis; conversely, constitutive activation of Rap1 inhibits the disassembly of premitotic focal adhesions and of the actin cytoskeleton, leading to delayed mitosis and to cytokinesis defects. Rap1 activity slowly increases after mitosis ends; inhibition of Rap1 activation by the ectopic expression of the dominant-negative Rap1[S17A] mutant prevents the rounded cells from spreading after mitosis. For the first time, we provide evidence for the direct regulation of adhesion processes during mitosis via the activity of the Rap1 GTPase.
Integrins are the major family of receptors that mediate cell-matrix adhesion (Hynes, 2002). In interphase cells, which are generally well spread, integrins assemble into adhesive structures linked to actin fibres (Geiger et al., 2001). As cells enter into mitosis, focal adhesions and actin stress fibres disassemble, leading to cell retraction along actin retraction cables and cell rounding (Cramer and Mitchison, 1997; Maddox and Burridge, 2003). As cell division nears completion at the beginning of telophase, daughter cells begin to spread on the substratum, as focal adhesions and stress fibres are reformed. To date, the signals and pathways regulating the dynamics of actin fibres and adhesion structures as cells enter and exit mitosis remain unknown. Importantly, in human mammary epithelial cells (HMEs), the level of β1 integrin expressed at the plasma membrane remains unchanged during mitosis, arguing for a regulation of integrin activity (Suzuki and Takahashi, 2003). Nevertheless, the nature of such a regulator of integrin-dependent adhesion, which would itself be regulated during mitosis, remains to be elucidated.
Experimental evidence obtained during the last decade has shown that the small GTPase Rap1 acts as a potent activator of many integrins (Bos, 2005; Caron, 2003), such as α3β1, α4β1, α5β1, αLβ2, αMβ2 and αIIbβ3 (Arai et al., 2001; Bertoni et al., 2002; Caron et al., 2000; de Bruyn et al., 2003; Enserink et al., 2004; Katagiri et al., 2000; Reedquist et al., 2000). To summarise, Rap1 seems to activate all of the integrins connected to the actin cytoskeleton (e.g. from the β1, β2 and β3 families), but not integrins connected to intermediate filaments, such as α6β4 (Enserink et al., 2004). To date, two Rap1 effectors have been clearly linked to integrin activation by Rap1. RAPL (regulator for cell adhesion and polarization enriched in lymphoid tissues), which is mainly expressed in lymphoid organs and tissues, interacts specifically with GTP-bound Rap1. Expression of RAPL is sufficient to stimulate the activation of αLβ2 integrins and polarise their distribution in lymphocytes, in a manner that is dependent on the ability of RAPL to bind both Rap1 and the αL-integrin subunit (Katagiri et al., 2003). Accordingly, lymphocytes from RAPL-knockout mice are severely impaired in their abilities to adhere in vitro and to reach their homing organs in vivo (Katagiri et al., 2004). Nevertheless, the restricted expression of RAPL might indicate that other Rap1 effectors are involved in controlling adhesion in non-lymphoid cells. In particular, RIAM (Rap1-interacting adhesion molecule) is another Rap1 effector whose expression is sufficient to promote αLβ2-integrin- and α4β1-integrin-dependent adhesion of Jurkat cells; moreover, RIAM silencing suppresses the ability of a constitutively active Rap1 mutant to stimulate the adhesion of Jurkat cells to fibronectin and ICAM, or to activate αIIbβ3 integrin when ectopically expressed in Chinese hamster ovary (CHO) cells (Han et al., 2006; Lafuente et al., 2004). Interestingly, talin, a well-described component of focal adhesions, contains in its N-terminal head domain a B4.1 motif whose association with a NPX[Y/F] motif, which is present in many β-subunits, is sufficient to promote integrin activation (Tadokoro et al., 2003). Moreover, talin colocalises and is co-immunoprecipitated with RIAM when both molecules are overexpressed, suggesting that Rap1 activates integrins by promoting the recruitment of talin to integrins through RIAM (Han et al., 2006).
As a consequence of the ability of Rap1 to stimulate integrin activation and cell adhesion, expression of the constitutively active Rap1[Q63E] mutant or negative regulators of Rap1 (such as Rap1GAP or Spa-1) respectively stimulates or inhibits the spreading of HeLa cells on fibronectin (Arthur et al., 2004; Tsukamoto et al., 1999). Similarly, knockout of the Rap1 activator C3G in mouse embryo fibroblasts impairs their adhesion and spreading, and accelerates migration (Ohba et al., 2001). Interestingly, overexpression of the Rap1 orthologue similarly stimulates spreading of Dictyostelium discoideum cells (Rebstein et al., 1997; Rebstein et al., 1993). Recently, a genetic screen has led to the identification of the Phg2 kinase (phagocytosis 2) as a Rap1 effector necessary to mediate the stimulation of Dictyostelium cell spreading by Rap1 (Jeon et al., 2007; Kortholt et al., 2006); no mammalian orthologue of Phg2 has been isolated so far.
Because of its apparent ability to regulate all of the different integrins engaged in adhesion, Rap1 constitutes a good candidate for regulation of the turnover of adhesive structures and for control of the retraction as well as spreading that occur during mitosis.
In this work, we show that Rap1 is transiently inactivated during mitosis and that constitutively active or dominant-negative Rap1 mutants respectively inhibit the cell-retraction and cell-spreading phenomena that normally occur during the mitosis of HeLa cells. Our data suggest that Rap1 is a key element in the control of adhesion and spreading dynamics during mitosis.
Modulation of Rap1 activity affects the morphology of HeLa cells
Rap1 activity controls the activation of many integrins, including the α5β1 integrin that mediates cell adhesion and spreading onto fibronectin (Arai et al., 2001). Because integrin activation is a prerequisite for integrin outside-in signalling, Rap1 activity ultimately influences integrin outside-in signalling and thereby regulates cell morphology (Arthur et al., 2004; Tsukamoto et al., 1999). To determine whether Rap1 activity controls HeLa-cell morphology, we expressed a constitutively active mutant (Rap1[Q63E]) or a powerful dominant-negative mutant (Rap1[S17A]) (Dupuy et al., 2005) of Rap1 in cells seeded on fibronectin-coated plates. Transfected cells were easily identifiable by their increased immunofluorescence when stained with an anti-Rap1 antibody (data not shown). As previously described (Arthur et al., 2004), expression of the constitutively active Rap1[Q63E] mutant stimulated the spreading of HeLa cells (Fig. 1A). In comparison with untransfected cells, Rap1[Q63E]-expressing cells exhibited few peripheral actin stress fibres but a very dense network of thin actin stress fibres throughout the cell body. Each fibre was connected to a vinculin-labelled structure that was generally smaller than those found in untransfected cells; the vinculin-rich structures were scattered throughout the cell body. By contrast, cells expressing the dominant-negative Rap1[S17A] mutant were rounded and not spread (Fig. 1A). Actin cables, essentially lining the cell periphery, were connected at their ends to peripheral vinculin-stained structures comparable in size to those displayed by untransfected cells. Hence, despite their rounded morphology, Rap1[S17A]-expressing cells exhibit focal-adhesion structures apparently similar to those of untransfected cells. This explains why these cells are still adherent and do not detach even after performing a shake-off, a technique known to detach poorly adherent cells such as those in mitosis (data not shown).
To determine whether these phenotypes resulted from a long-lasting modulation by Rap1 mutants of HeLa-cell adhesion and spreading capacities, cells expressing Rap1[S17A] or Rap1[Q63E] were trypsinised and replated onto a fibronectin-coated surface. As shown in Fig. 1C, cells expressing Rap1[S17A] indeed had a drastically impaired ability to re-adhere, as only 4% of transfected cells adhered to fibronectin 6 hours after replating. By contrast, Rap1[Q63E]-expressing cells re-adhered more rapidly than untransfected cells. Accordingly, measurements of cell area showed that cells expressing the constitutively active mutant of Rap1 spread more rapidly and to a larger extent than untransfected cells upon replating, whereas cells expressing dominant-negative Rap1 were unable to spread after replating (Fig. 1B,D). No vinculin-rich adhesion structures were observed in cells expressing the Rap1[S17A] mutant, hence explaining their inability to adhere and spread (Fig. 1B). By contrast, cells expressing the constitutively active Rap1[Q63E] mutant displayed the same morphological characteristics upon replating as in Fig. 1A; at all time points observed, replated cells also exhibited numerous thin actin stress fibres and small vinculin-rich structures throughout the cell body (Fig. 1B). Moreover, these cells adopted a much rounder morphology than control cells, which exhibited the polygonal shape characteristic of HeLa cells. In conclusion, Rap1 activity modulates the distribution and size of adhesion structures and actin fibres, and consequently contributes to the maintenance and dynamics of cell morphology in HeLa cells.
Recently, Rac1 was suggested to mediate Rap1-dependent stimulation of cell spreading (Arthur et al., 2004). To investigate whether Rac1 could play a role in our system, we assessed the level of active Rac1 in cells expressing either a dominant-negative or constitutively active mutant of Rap1 (Fig. 1E). Surprisingly, we found that Rac1 was activated in rounded cells expressing dominant-negative Rap1, and that Rac1 was inactivated in overspread cells expressing active Rap1. Because Rac1 and Rap1 activities did not appear to vary in sync, we conclude that Rac1 was not involved in the Rap1-induced changes of cell morphology that we observed.
Rap1 activity is regulated during mitosis
Having observed that modulations of Rap1 activity stimulate or inhibit cell spreading, we hypothesised that Rap1 activity should be tightly regulated during the cell cycle and particularly in mitosis. Indeed, most adherent cells undergo a phase of retraction at the onset of mitosis and, following cytokinesis, the two daughter cells spread to assume their interphasic shape (Cramer and Mitchison, 1993). To test this hypothesis, HeLa cells were synchronised, blocked at the beginning of the S phase by a double thymidine block, and progressively released into the cell cycle. Rap1 activation levels were assessed by pull-down assays whilst progression along the cell cycle was simultaneously controlled by FACS analysis. As shown in Fig. 2A, at 2 hours after release from the thymidine block, cells were in S phase, well spread and exhibited a high Rap1 activity. After a further 7 hours (i.e. 9 hours after release), most of the cells were approaching mitosis: 50% were in G2 or M phase, close to performing mitosis, whereas 39% were ending their replication and 12% of the cells had already undergone mitosis and were back to G1. At this time point, global Rap1 activity was reduced by 50%. After a further 2 hours (i.e. 11 hours after release), while most of the cells (71%) had performed their mitosis and 17% were either in mitosis or late G2 phase, the amount of active Rap1 was at its lower level (reduced by 80% as compared with pre-mitotic cells). After another 2 hours (i.e. 13 hours after release) cells had all returned to the G1 phase and were spread; surprisingly, despite a doubling in Rap1 activity from its minimum, it remained low (40% of the level measured before mitosis) until 15 hours after release (approximately 4 hours after mitosis), after which GTP-Rap1 levels remained high until cells entered another S phase.
To confirm these results obtained from a pool of synchronised cells at a single level, we monitored the changes in Rap1 activity using a Rap1 FRET (fluorescence resonance energy transfer) probe developed in the Matsuda laboratory (Mochizuki et al., 2001). The original Raichu-Rap probe comprises the Ras-binding domain (RBD) of Raf fused in frame to the C-terminal part of Rap1A. In this probe, the YFP and CFP fluorescent domains are fused in frame to the N-terminus of Rap1A and to the C-terminus of the RBD, respectively; the membrane-targeting signal (comprising the last 18 amino acids, including the CAAX sequence) of Rap1A is transferred from the C-terminus of Rap1 to the C-terminus of the FRET probe. Upon activation (GTP loading) of Rap1A, binding of the RBD to Rap1-GTP leads to a conformational change that brings together the two fluorophores and causes an increase in the FRET signal. Because a weak decrease in the intrinsic FRET signal is usually observed during mitosis, as a result of cell morphological changes, irrespective of the probe used (O.G., unpublished observations and Fig. 2B), we constructed a non-responsive Rap1 probe to serve as an internal negative control. The resulting Raichu-Rap[R452L] probe presents a mutation in the RBD that was previously shown to abrogate the interaction of Raf with active Ras (Fabian et al., 1994). As shown in Fig. 2B, entry into mitosis, detected as the breakdown of the nuclear envelope (as assessed by phase contrast and set to t=0 in the graph), was followed by a weak decrease in the FRET signal of the control probe (CTR). By contrast, FRET from the wild-type Raichu-Rap probe dropped abruptly upon entry into mitosis, indicating that Rap1 activity is negatively regulated during the process. Interestingly, whereas the downregulation of Rap1 activity preceded nuclear-envelope breakdown and took place within a few minutes, the post-mitotic recovery of Rap1 was considerably slower, confirming our biochemical results (see Fig. 2A). Taken together, these results show that Rap1 activity is transiently downregulated during mitosis.
Rap1 inactivation is required for the disassembly of adhesion complexes during mitosis
Because endogenous Rap1 is transiently inactivated during mitosis, we investigated the impact of the constitutive activation of Rap1 on the mitotic process by expressing the constitutively active Rap1[Q63E] mutant. Control mitotic cells, as revealed by the typical Hoechst staining of metaphasic chromosomes, exhibited their characteristic rounded morphology (Fig. 3), with F-actin no longer organised as stress fibres, but rather as retraction fibres (Cramer and Mitchison, 1993), and a loss of mature adhesion complexes as attested by the absence of vinculin labelling. By contrast, Rap1[Q63E]-expressing mitotic cells maintained the tight network of thin actin fibres they exhibited in interphase, with a noticeable deformation of the network as it circumvented the area surrounding the chromosomes. Unlike control cells, Rap1[Q63E]-expressing cells also retained vinculin-labelled structures throughout the cell body, and these structures appeared larger at the cell periphery (Fig. 3). We also checked that the phenotype observed was specific to Rap1 by showing that the expression of the constitutively active form of either H-Ras, R-Ras or Rac1 did not phenocopy the inhibition of focal-adhesion disassembly and cell retraction during mitosis induced by active Rap1 (data not shown). Hence, persistence of Rap1 signalling does not prevent cells from undergoing mitosis, but specifically opposes the loss of actin stress fibres and focal adhesions that leads to the characteristic cell rounding observed during the mitosis of HeLa cells.
Misregulation of Rap1 affects the regulation of cell morphology during mitosis
To analyse how the misregulation of Rap1 activity alters the dynamics of the actin cytoskeleton and adhesion sites during mitosis, we performed live video-microscopy experiments on cells expressing either the dominant-negative Rap1[S17A] or the constitutively active Rap1[Q63E] mutants. Because a previous study had shown that Rap1 mutants are not fully active when fused to a N-terminal tag (Dupuy et al., 2005), we co-transfected cells with untagged mutant alleles of Rap1 and a vector encoding red fluorescent protein (RFP) in the ratio 3:1. Under these conditions, more than 95% of RFP-expressing cells showed markedly increased Rap1 levels compared with control untransfected cells, as assessed by immunofluorescence using anti-Rap1 antibodies (not shown). In further experiments, RFP expression was therefore taken as an indicator of the overexpression of the mutant.
Time-lapse video-microscopy was performed between 4 and 48 hours after transfection, a time period sufficient for HeLa cells to accomplish two full division cycles (Fig. 4A; supplementary material Movies 1-3). Cells ectopically expressing Rap1 mutants undergo a first mitosis that is undistinguishable from that of mock-transfected cells (control; CTR): cells round up and go through all of the steps of mitosis with normal morphology and kinetics (Fig. 4A, top panel). However, following mitosis, cells expressing the constitutively active Rap1[Q63E] mutant displayed a more rapid and pronounced post-mitotic spreading than untransfected cells, exhibiting a bigger cell area as well as a large lamellipodium on the opposing side from the sister cell. By contrast, cells expressing the dominant-negative Rap1[S17A] mutant started to spread, but eventually retracted and remained rounded thereafter. These results are reminiscent of those obtained in adhesion and spreading experiments (see Fig. 1). In order to ascertain the temporal correlation of this phenotype with the expression of the dominant-negative Rap1 mutant, the same experiments were performed using RFP-tagged forms of the Rap1 mutants ([S17A] or [Q63E]) (Fig. 4B). Strikingly, expression of the fusion proteins only started to be detectable after cells had accomplished the first mitosis and initiated spreading. In particular, increased RFP-Rap1[S17A] expression correlated with retraction of transfected cells, suggesting that interfering with Rap1 activity resulted in inhibition of post-mitotic spreading and eventually in cell retraction. Despite their rounded morphology, Rap1[S17A]-expressing cells were able to proceed through a further mitosis, again with normal kinetics and morphology except that the cells retained a rounded shape throughout, and the two sister cells remained round and close to one another after cytokinesis (Fig. 4A, lower panel; supplementary material Movie 3). This phenotype was highly reproducible because it similarly affected all cells expressing Rap1[S17A].
The effects of Rap1[Q63E] expression were even more striking. These cells, which were very widely spread following the first mitosis, were unable to round up prior to the second mitosis, despite clear morphological signs of attempted retraction (see supplementary material Movie 2). They proceeded as spread cells through the chromosomal steps of mitosis until cytokinesis, in agreement with the fact that these cells retained actin stress fibres and adhesion sites during mitosis as shown in Fig. 3. As a consequence, 50% of cells expressing active Rap1 were unable to complete cytokinesis, and exhibited a total regression of the cleavage furrow, resulting in large and spread binucleated cells (supplementary material Movie 2 and data not shown). This failure of cytokinesis was also observed in a different batch of HeLa cells as well as with NIH3T3 cells. We then measured the duration of two successive cell cycles, and analysed the results, taking into account the results of Fig. 4B that the mutant Rap1 proteins are only efficiently expressed after the end of the first mitosis. As shown in Fig. 4C, cells expressing the constitutively activated mutant of Rap1 were delayed in their cell cycle, especially during mitosis, which lasted twice as long as in control cells. Despite their rounded morphology, the cell-cycle kinetics of cells expressing dominant-negative Rap1[S17A] were undistinguishable from those of control cells. In conclusion, the regulation of Rap1 activity plays an important role in the cell-shape changes that occur during mitosis. Too much Rap1 activity prevents the cells from de-adhering at the onset of mitosis, eventually resulting in a delayed cell cycle and mitosis as well as a high proportion of binucleated cells, whereas the inability to re-activate Rap1 following mitosis prevents the cells from spreading on the substratum to assume their normal interphasic morphology.
Rap1 activity has been shown to stimulate the adhesive properties of many integrins, as assessed by adhesion assays, ligand-binding assays and using activation-specific reporter antibodies. Less studied is the impact of modulating Rap1 activity on cell morphology, or how Rap1 activity is regulated during the morphological changes that accompany physiological processes. In this article, we have characterised both the effect of modulating Rap1 activity on HeLa-cell spreading and its regulation during mitosis.
Consistent with its role in inside-out integrin activation (Bos, 2005), we show that Rap1 overactivation stimulates cell adhesion and spreading onto a fibronectin matrix (Fig. 1). Cells expressing constitutively active Rap1 displayed a more homogeneous distribution of their adhesion sites connected to a denser network of actin stress fibres. These results show that integrin activation by Rap1 is sufficient for the establishment and maturation of adhesion sites and their connection to the actin cytoskeleton (Fig. 1B,D). Comparatively, cells expressing a dominant-negative Rap1 mutant did not spread and remained rounded (Fig. 1A). These cells are also completely inhibited in their abilities to re-adhere once trypsinised, because no adhesion-related structures could be observed (Fig. 1B). These results clearly show that Rap1 activity is required for cell adhesion onto fibronectin and that it mediates a non-redundant pathway that leads to integrin activation. Because time-lapse microscopy showed that cells retract concomitantly to the expression of dominant-negative Rap1, one hypothesis could be that Rap1 function is also required for the maintenance of preformed adhesion sites (Fig. 4B). However, these cells do not detach but retain some adhesion- and actin-related structures. Rather than Rap1 playing a role in maintaining already formed adhesion sites, it is likely that adhesion sites are subject to constant recycling and that Rap1 activity is required to initiate the formation of new sites. According to this hypothesis, the adhesion sites that remain in cells expressing dominant-negative Rap1 could be explained by the fact that these sites are less prone to recycling, maybe because the tension exerted by the actin cytoskeleton at these sites is somehow different. Further FRAP experiments to study the dynamics of vinculin-rich structures in these cells could be informative.
Correlated with the fact that expression of either constitutively activated or dominant-negative mutants respectively stimulate or inhibit cell adhesion and spreading (Fig. 1), we clearly demonstrated that cell retraction at the beginning of mitosis requires Rap1 inactivation. Indeed, both pull-down and FRET experiments showed that endogenous Rap1 is inactivated during mitosis (Fig. 2); moreover, ectopic expression of constitutively active Rap1 is not compatible with the disruption of adhesion sites and actin fibres that normally occur during this process (Fig. 3). Downregulation of Rap1 activity at the onset of mitosis could be ensured by the inhibition of a guanine nucleotide exchange factor (GEF) activity, stimulation of GTPase-activating protein (GAP) activity or a combination of both mechanisms. In Fig. 4B, the effects of expressing dominant-negative Rap1, which blocks the activation of endogenous Rap1 by titrating GEFs (Dupuy et al., 2005; Feig, 1999), takes place within 2 hours. Comparatively, Rap1 is inactivated within 10 minutes of the onset of mitosis (Fig. 2B). Such a rapid inhibition might indicate that increased GAP activity is required. Interestingly, Rap1GAP has been shown to be directly phosphorylated by Cdc2, the protein that initiates the onset of mitosis (Janoueix-Lerosey et al., 1994). Whether Rap1GAP activity is modulated by phosphorylation is unknown, as is the physiological relevance of its phosphorylation by Cdc2.
To date, cell rounding has been shown to require a RhoA-Rock pathway, probably by acting on actin-myosin contractility through phosphorylation of myosin light chain (MLC) (Maddox and Burridge, 2003). More recently, two different studies have suggested a role for moesin during mitosis of Drosophila S2 cells. Moesin belongs to the ERM (ezrin, radixin, moesin) family of proteins that link the plasma membrane to the actin cytoskeleton. Moesin is the only member of this family expressed in Drosophila, in which it is required for complete cell retraction and correct orientation of the mitotic spindle (Carreno et al., 2008; Kunda et al., 2008). How Rap1 and moesin pathways are coordinated to implement cell retraction during mitosis presently remains unclear. However, it is unlikely that a structural protein such as moesin could directly regulate the activity of the signalling molecule Rap1; rather, it is conceivable that two pathways, one directly regulating adhesion and another pathway promoting cell contractility, could cooperate to induce cell rounding during mitosis. To support this idea, signs of contractility remain visible in cells inhibited in their ability to retract upon expression of the constitutively active Rap1 mutant as well as in cells expressing dominant-negative Rap1 (Fig. 4A; supplementary material Movies 1-3). These observations suggest that several pathways regulate cell rounding during mitosis.
Surprisingly, once mitosis has been completed, Rap1 activity only increased slowly, with maximal activation only reached after cells had fully spread (4 hours after mitosis, Fig. 2A). However, Rap1 activity is required to allow cell spreading, because the inhibition of Rap1 by expression of the dominant-negative Rap1[S17A] mutant completely blocks cell spreading (Fig. 4A). This delayed activation raises questions about the role of Rap1 in cell spreading after mitosis. It seems that a weak Rap1 activation overall is sufficient to initiate spreading, maybe because a subset of Rap1 molecules is activated at specific locations or because active Rap1 is targeted to specific regions of the cell. We did not observe any specific local increase in FRET signal with the Raichu-Rap1 probe, possibly owing to its overexpression. Nevertheless, because levels of active Rap1 are higher in cells plated on fibronectin compared with cells maintained in suspension, it is likely that Rap1 in turn is activated by integrin engagement (Ohba et al., 2001). Such an auto-stimulatory loop could help reinforce and secure newly formed adhesion sites, and might also allow cells to spread directionally.
Finally, we show that expression of the dominant-negative Rap1[S17A] mutant has no effect on cell-cycle progression. Despite remaining rounded, cells progress along the cell cycle and in particular into mitosis within the same time course as control cells (Fig. 4C). These results suggest that neither mitosis nor cytokinesis require fully developed adhesion sites in order to proceed. Compared with control cells, interphase and mitosis last longer in cells expressing the constitutively active Rap1[Q63E] mutant. In particular, mitosis lasts 100% longer than in control cells. This could be related to the finding that cell rounding helps to stabilise spindle assembly (Kunda et al., 2008). Interestingly, Kunda et al. showed that the unstable metaphase spindles observed in non-fully retracted cells lacking moesin could be rescued by artificially rounding the cells. A possible explanation is that, in non-rounded cells, astral microtubules can no longer contact the cortex to separate the two asters of the spindle (Rosenblatt, 2008). This might also occur in cell expressing constitutively active Rap1, in which the mitotic spindle is not as stable as in control cells, turning from side to side before chromosome segregation (data not shown). Nevertheless, another simple explanation could be that the lack of actin-fibre disruption disturbs the organisation of the mitotic spindle. Because cytokinesis is a process mainly dependent of acto-myosin contractility (Matsumura, 2005), maintenance of the actin-fibre network might also explain the cytokinesis defects and the fusion of the cells expressing constitutively active Rap1 (Fig. 4A).
To conclude, we provide strong evidence that cell-adhesion dynamics play an important role in regulating the progression of mitosis, and that activity of the integrin activator Rap1, which is finely tuned during this process, is involved. Further challenges will be to unveil the signals from the cell cycle that trigger cell rounding, and decipher the pathways upstream and downstream of Rap1 that act in this process.
Materials and Methods
pRK5 empty vector, pRK5-Rap1[S17A] and pRK5-Rap1[Q63E] were previously described (Dupuy et al., 2005). pEmRFP-C2 was generated by replacing the GFP-encoding BamHI-NotI fragment from the pEGFP-C2 vector (Clontech) with the sequence encoding the monomeric RFP (a kind gift of Roger Y. Tsien, University of California at San Diego, La Jolla, CA), amplified by PCR. pRaichu-Rap1[R452L] was built by targeted mutagenesis of pRaichu-Rap1 (a precious gift of Michiyuki Matsuda, Kyoto University, Kyoto, Japan) with the QuikChange kit (Stratagene).
Cell culture and transfection
HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% foetal bovine serum (FBS) on plastic dishes or glass coverslips that had been precoated with fibronectin (10 μg/ml). Plasmids were transfected with Lipofectamine Plus (Invitrogen). For video microscopy, a 35-mm culture dish was first pre-coated with fibronectin (10 μg/ml), and then filled with polydimethylsiloxane polymer (Dow Corning) in which three wells were subsequently cut out to enable the simultaneous recording of three distinct experiments (see below).
Cells were prepared for indirect immunofluorescence using a method that preserves the actin retraction fibres observed during mitosis (Mitchison, 1992). Briefly, cells seeded on glass coverslips were incubated for 1 minute with cytoskeleton buffer [10 mM 2-(N-morpholino) ethanesulfonic acid buffer, pH 6.1, 138 mM KCl, 3 mM MgCl2, 2 mM EGTA] containing 0.32 M saccharose and 0.1% Triton X-100, then fixed for 20 minutes in a 4% solution of paraformaldehyde in cytoskeleton buffer containing 0.32 M saccharose. Coverslips were then washed three times in Tris-buffered saline (TBS: 20 mM Tris-HCl, pH 7.4, 0.15 M NaCl) and paraformaldehyde was quenched for 20 minutes with 0.1 M NH4Cl. After three washes in TBS, cells were permeabilised for 10 minutes with TBS containing 2% BSA and 0.2% Triton X-100 (TBS-BT). After three washes in TBS containing 0.02% Triton X-100 (TBST), samples were incubated for 1 hour with primary antibodies: affinity-purified rabbit polyclonal anti-Rap1 antibodies (Beranger et al., 1991), mouse monoclonal anti-vinculin VII F9 (a kind gift of Marina Glukhova, Institut Curie, Paris, France), diluted in TBST. After four washes in TBST, cells were incubated for 45 minutes with Alexa-Fluor-488-coupled phalloidin (1/250 in TBST; Molecular Probes) and the fluorescent-coupled secondary antibodies Cy3-coupled goat anti-mouse IgG (1.3 μg/ml in TBS-BT; Jackson) and Alexa-Fluor-647-conjugated goat anti-rabbit IgG (2 μg/ml in TBS-BT; Molecular Probes). Cells were washed four times in TBS and incubated for 2 minutes in 1 μg/ml Hoechst 33258 in TBS-T. Finally, cells were washed in TBS and coverslips were mounted with Gold Prolong reagent (Molecular Probes).
Adhesion and spreading assays
At 1 day after transfection with vectors encoding mutants of Rap1 (dominant negative or constitutively active) or RFP as a control, HeLa cells were trypsinised for 5 minutes at 37°C, washed once with PBS, incubated for 20 minutes in a rotating tube at 37°C in DMEM containing 10% FBS, and finally seeded on coverslips coated with fibronectin (10 μg/ml). At the indicated times, cells were gently washed with PBS and processed for immunofluorescence to label Rap1 and F-actin according to the immunofluorescence protocol mentioned above. At least five fields for each condition were photographed with a 40× lens. Adhesion was measured as the number of adherent cells expressing either Rap1 mutant (as assessed by immunofluorescence with an anti-Rap1 antibody) or RFP (control), expressed as a percentage of the total number of cells adhering after 6 hours. From pictures acquired, the cell periphery of more than 200 cells per condition was drawn by hand according to the F-actin staining and cell surface was measured with the ImageJ freeware (http://rsb.info.nih.gov/ij).
Pull-down of GTP-bound Rac1
Assays were performed with the Rac1 activation assay Biochem Kit (Cytoskeleton) according to the manufacturer's instructions.
Pull-down of GTP-bound Rap1 in cells synchronised by a double thymidine block
To synchronise cells, a double thymidine block was performed by incubating cells for 19 hours at 37°C in normal medium (DMEM containing 10% FBS) supplemented with 2 mM thymidine; cells were washed three times with normal medium, then incubated for 10 hours in pre-warmed normal medium containing 30 μM deoxycitidine. Finally, cells were blocked again by incubating them for at least 19 hours at 37°C in normal medium supplemented with 2 mM thymidine. Cells were sequentially released into the cell cycle by replacing the medium with normal medium containing 30 μM deoxycitidine. The levels of Rap1-GTP were assessed using a pull-down assay as previously described (Franke et al., 1997). Briefly, cells were washed with cold PBS and lysed in a buffer containing 50 mM Tris-HCl, pH 7.5, 15 mM NaCl, 20 mM MgCl2, 5 mM EGTA, 1% Triton X-100, 1% N-octylglucoside, 100 μM PMSF, 10 μM leupeptin and 10 μM aprotinin. Lysates were cleared by centrifugation, and Rap-GTP complexes were recovered on glutathione-Sepharose beads precoupled to GST fused with the Ras/Rap-binding domain of RalGDS. Precipitates were washed three times with lysis buffer and solubilised in SDS sample buffer. Aliquots of cleared cell lysates were kept for detection of total Rap1 contents. Following western blotting, Rap1 was detected with affinity-purified polyclonal antibodies (Béranger et al., 1991). Actin was detected using the monoclonal AC-74 anti-β-actin antibody (Sigma). Proteins were visualised on western blots by chemiluminescence (SuperSignal West Femto, Pierce); images were captured with a FUJI LAS-1000 CCD camera and quantified using FUJI Image Gauge software. In parallel to the pull down, the cell-cycle stage was assessed by flow cytometry. Briefly 2×106 synchronised cells were trypsinised, washed with PBS, fixed with ice-cold 70% ethanol, washed again twice with PBS, incubated for 5 minutes with a 1 mg/ml RNAse-A solution in PBS, centrifuged and then resuspended in 400 μl of PBS, 50 μg/ml propidium iodide. Fluorescence was measured at 620 nm with excitation at 488 nm.
Live video-microscopy and immunofluorescence microscopy
Live video recordings were acquired on a Leica DMIRBE microscope equipped with a 37°C thermostated, 5% CO2-equilibrated motorised stage. Immunofluorescence pictures were acquired on a Leica DM6000B Epifluorescence microscope with a 63× Apochromat lens (Leica Microsystems). All acquisitions were made with a Photometrics CoolSnap HQ camera and analysed with the Metamorph software (Universal Imaging).
HeLa cells (3×106) were electroporated (250 V, 1500 μF, infinite resistance) with 25 μg of FRET probe DNA using an Easyjet plus machine (Equibio). Then, cells were seeded in glass-bottom dishes (Intracel, Royston) pre-coated with fibronectin (10 μg/ml). At 7 hours after transfection, cells were synchronised in S phase with 2.5 mM thymidine for 24 hours. Cells were then released in fresh DMEM. For time-lapse experiments, the culture medium was replaced by Leibovitz's L-15 medium (Invitrogen), containing 10% FBS. DIC and fluorescence images were acquired with a Deltavision microscope equipped with a cascade II EMCCD camera (Photometrics) using a binning of 1×1 and a gain of 140. A 40×/1.35 UApo oil objective was used for all images. Filter configuration was 436/10 for the CFP excitation filter, and 465/30 and 535/30 for the CFP and YFP emission filters, respectively. The exposure time was 200 milliseconds for CFP and YFP images using a 32% ND filter. ImageJ software was used to quantify fluorescence signals. Briefly, after background subtraction, the YFP:CFP ratio of the whole cell was calculated and used to represent the FRET efficiency.
The authors are indebted to M. Bornens and M. Théry for stimulating discussions and help with video-microscopy, and to M. Gluckhova and M. Matsuda for providing anti-vinculin antibodies and Raichu-Rap FRET probe, respectively. This work was funded in part by a grant from the Association pour la Recherche contre le Cancer. A.G.D. was successively supported by post-doctoral fellowships from the Cancer and Solidarity Foundation (Geneva) and the `Fondation pour la Recherche Médicale'. V.T.D. was the recipient of a post-doctoral fellowship from the Institut Curie Genhomme project.
The authors are extremely saddened by the loss of their collaborator Emmanuelle Caron. She was a young talented scientist, senior lecturer at Imperial College London, recognised by her peers and loved by her students. Emmanuelle was also a great person, full of energy and joy of life, always available to others. She supervised Aurélien Dupuy for two years and helped him considerably while he was finishing the work for this manuscript. Emmanuelle was more than a supervisor; she was a mentor. She will be terribly missed. This work is dedicated to her memory.