SH4-domain-induced plasma membrane dynamization promotes bleb-associated cell motility

A subset of cytosolic proteins reside at the inner leaflet of the plasma membrane (PM) by virtue of an N-terminal targeting signal consisting of at least one acylation (myristoylation or palmitoylation) and a second signal (e.g. an additional acylation or a stretch of basic amino acids). These motifs, initially discovered as membrane anchors of the Src kinase family and therefore also referred to as SH4 domains, are necessary for membrane association of these proteins and confer membrane anchorage to heterologous cargo (McCabe and Berthiaume, 1999; Resh, 1994). The Src protein family comprises non-receptor tyrosine kinases with important roles in the regulation of intracellular signal transduction networks (Thomas and Brugge, 1997). Oncogenic forms of Src kinases and the prototypic member Src in particular cause cell transformation by deregulation of cell migration, proliferation and survival (Frame, 2004). Appropriate subcellular localization and membrane attachment are prerequisites for the biological activity of Src kinases in, for example, oncogenic transformation (Kaplan et al., 1988; Sigal et al., 1994). Roles of SH4 domains beyond this targeting function, however, have not been described to date. Moreover, the transport pathways towards and across the PM used by SH4-domain-containing proteins, as well as the determinants for PM specificity of these domains, have remained largely elusive.

The PM of eukaryotic cells is subject to dynamic reorganizations that are mediated by the submembranous cortical cytoskeleton under the control of specific small Rho GTPase-regulated signalling cascades. Depending on the GTPase involved, such PM dynamization leads to the active generation of cell protrusions, such as, for example, lamellipodia and filopodia (Hall, 1998). In addition, localized destabilization of the cortical actin meshwork results in the formation of rounded cell protrusions generated by the osmotic pressure of the cell interior; these protrusions are referred to as PM blebs (Charras et al., 2005; Cunningham, 1995; Sheetz et al., 2006). Following bleb formation, contractile acto-myosin force is generated to drive bleb retraction. Multiple, mechanistically and functionally distinct types of PM blebs have been described, including blebs induced during apoptosis or necrosis (Coleman et al., 2001; Lane et al., 2005; Mills et al., 1998; Sebbagh et al., 2001; Torgerson and McNiven, 1998), cytokinesis (Fishkind et al., 1991) and cytokine secretion (MacKenzie et al., 2001). Furthermore PM blebs have been suggested to facilitate cell motility of tumour cells in 3D matrices (Sahai and Marshall, 2003).

Here, we studied the SH4 domains of several Src kinases and of the Leishmania parasite virulence factor hydrophilic acylated surface protein (HASPB) (Denny et al., 2000; Stegmayer et al., 2005). Unexpectedly, we found that expression of SH4 domains causes the reorganization of the PM to produce highly dynamic non-apoptotic membrane blebs. PM blebbing was found to be a conserved activity of Src kinase SH4 domains and depends on endogenous Src activity. Because SH4-domain-induced PM blebbing correlated with enhanced cell invasion in 3D matrices, these results suggest an active role of SH4 domains in the oncogenic properties of Src kinases.

To study SH4-domain biology, we initially analyzed the SH4 domain of the Leishmania parasite virulence factor HASPB using our previously described CHO cell lines that stably express GFP fusion proteins of the 18-amino acid SH4 domain of HASPB (N18-HASPB-GFP), or mutants thereof, in a doxicycline (dox)-inducible manner (Stegmayer et al., 2005). Confocal microscopy of live and fixed cells revealed efficient PM targeting of GFP by the SH4 domain of HASPB (Fig. 1Aa,b). However, to our surprise, expression of N18-HASPB-GFP drastically altered the PM morphology of these cells, which displayed extensive membrane blebbing with N18-HASPB-GFP residing in the limiting membrane of these bulky protrusions. Titration of the amounts of dox used for induction demonstrated a correlation between N18-HASPB-GFP expression levels and PM blebbing (Fig. 1C, see also supplementary material Fig. S1 for a quantification of N18-HASPB-GFP expression levels). Although even low dox concentrations were sufficient to induce moderate PM blebbing, N18-HASPB-GFP expression levels induced by at least 300 ng/ml dox resulted in efficient membrane blebbing in over 80% of the cells. Equally efficient PM blebbing was observed upon transient expression of full-length HASPB fused to GFP (Fig. 1Ba) or a non-tagged version of full-length HASPB (Fig. 1Bb). HASPB-induced PM blebbing, although less pronounced than in CHO cells presumably due to lower levels of expression, was also observed in HeLa cells (data not shown). Two N18-HASPB-GFP mutants in which the acylation acceptors were disrupted were also analyzed: the myristoylation- and palmitoylation-deficient N18-Δmyr-HASPB-GFP localized diffusely in the cytoplasm and the nucleus in a pattern reminiscent of GFP alone (compare Fig. 1Ae,f with Fig. 1Bc) (see also Denny et al., 2000; Stegmayer et al., 2005). By contrast, the non-palmitoylated N18-Δpal-HASPB-GFP (which retains its myristoylation) accumulated at perinuclear membranes [most probably the Golgi apparatus (Denny et al., 2000; Stegmayer et al., 2005)] and was excluded from nuclei (Fig. 1Bd). Because both mutants failed to induce PM blebbing (Fig. 1D), we conclude that HASPB requires association with the PM to induce cell surface blebbing. Together, the SH4 domain of Leishmania HASPB potently induces PM blebbing in a manner that depends on its association with the inner leaflet of the cell-limiting membrane.

To further analyze the mechanism of HASPB-induced membrane bleb formation and to gain insight into the putative biological relevance of the blebs, we next performed a confocal real-time imaging analysis (Fig. 2 and supplementary material Movie 1). This analysis revealed that bleb formation and retraction occurred on average within 44 seconds (supplementary material Fig. S2A). Of note, SH4-domain-induced membrane blebbing could be observed over hours without apparent damage to the cells. Consistent with this observation, these cells were strictly non-apoptotic, as judged by TUNEL analysis, and SH4-domain overexpression did not alter susceptibility of the cells for induction of apoptosis following treatment with etoposide and cycloheximide (Fig. 3). Similar results were obtained when chromatin condensation or cell surface exposure of phosphatidylserine was analyzed, and SH4-domain-induced PM blebbing was insensitive to the caspase inhibitor zVAD-Fmk (data not shown). A 4D analysis revealed that PM blebbing occurred at the entire surface of N18-HASPB-GFP-expressing cells with the notable exception of cell-cell or cell-substratum contacts. This blebbing activity was most pronounced at the edges of the cell (supplementary material Movie 2 and Fig. S3A). PM dynamics were markedly reduced in N18-Δmyr-HASPB-GFP-expressing control cells (supplementary material Movie 3 and Fig. S3B). In line with previous characterizations of PM blebbing in M2 melanoma cells (Charras et al., 2005; Cunningham, 1995), the life span of individual blebs involved three distinct steps: fast expansion (average 6 seconds, 0.332 μm/second) followed by an extended stationary phase (average 12 seconds, 0.012 μm/second) and a subsequent slow retraction period (in average 26 seconds, 0.071 μm/second) (Fig. 2E,F and supplementary material Fig. S2A,C). Although most of the quantified blebs formed at and retracted to the identical position at the PM, a subpopulation of blebs displayed lateral mobility with accelerated apparent bleb expansion and retraction. Both types of blebs reached a similar size of approximately 1.7 μm in diameter at maximum extension and retracted completely (Fig. 2E,F and supplementary material Fig. S2B). During retraction, the previously round shape of the blebs typically turned into a less homogenous, wave-like morphology. In contrast to blebbing observed during ATP-induced secretion of IL-1β (MacKenzie et al., 2001), our real-time microscopy analysis revealed that HASPB-mediated PM blebbing in 2D does not cause significant shedding of HASPB-positive vesicles into the cell-culture supernatant (supplementary material Movies 1, 2 and Fig. 2). We conclude that SH4-induced membrane blebs are highly dynamic, display distinct kinetics during bleb formation and retraction, and are not indicative of the induction of apoptotic programmes.

To gain insight into the cellular signal transduction pathways that govern SH4-domain-mediated PM blebbing, we addressed the involvement of the small GTPases Rho, Rac and Cdc42. The SH4 domain of HASPB fused to the fluorescent mCherry protein (N18-HASPB-mCherry) was co-expressed with GFP fusion proteins of wild-type GTPases, dominant-negative (N17) Rac or Cdc42, or C3 transferase, to inhibit the activity of Rho. Only C3-mediated inhibition of Rho significantly reduced bleb formation (Fig. 4). Even though no blebbing occurred, N18-HASPB-GFP still localized to the PM in the presence of C3 (data not shown), indicating that SH4 domains are targeted to the PM in a Rho-independent manner but use a signalling cascade downstream of Rho to trigger membrane blebbing at the PM.

The Rho effector kinase Rock provides acto-myosin contractility required for most types of PM blebbing described (Leverrier and Ridley, 2001; Morelli et al., 2003; Sahai and Marshall, 2003). Consistently, blocking Rock activity by the specific inhibitor Y-27632 potently interfered with HASPB-mediated PM blebbing (Fig. 5A,C). Similar results were obtained using blebbistatin (Straight et al., 2003), a drug that specifically interferes with the ATPase activity of the myosin II motor protein. Analysis of the F-actin network demonstrated some enrichment of F-actin at the limiting membrane and, in particular, at the neck of the blebs. By contrast, no specific association of microtubuli with the blebs was observed (Fig. 5B). Nevertheless, depolymerization of either F-actin or microtubules by cytochalasin D or nocodazole, respectively, potently abrogated HASPB-induced PM blebbing (Fig. 5A-C). Together, these results show that the induction of PM blebs by the HASPB SH4 domain depends on the integrity of both F-actin and microtubule networks, and requires myosin II function. These results further reveal a novel link between Rho-Rock activity and SH4 domains.

PM blebbing is associated with cell motility during tissue invasion of rounded tumour cells (Sahai and Marshall, 2003; Wolf et al., 2003). To test whether SH4-domain-induced blebbing is able to promote cell invasion, we analyzed whether the formation of HASPB-induced membrane blebs correlates with cell migration in a 3D matrix. Using a well-established matrigel transwell assay (Hooper et al., 2006), the percentage of cells that invaded the matrigel and crossed the transwell membrane was determined. Following induction of transgene expression, invasion of GFP-expressing control cells was only occasionally observed. By contrast, N18-HASPB-GFP but not N18-Δmyr-HASPB-GFP cells invaded into the matrigel with significantly higher efficiency (Fig. 6). The invasiveness of N18-HASPB-GFP cells closely matched that of metastatic MDA-MB-435 human breast carcinoma cells, which served as a positive control. Thus, SH4-domain overexpression promotes cell invasion. Importantly, imaging within the matrigel revealed a rounded morphology with prominent cell protrusions for all SH4-domain-expressing cells, most of which displayed pronounced dynamic PM blebbing (see supplementary material Movies 4, 5). Although we cannot exclude that other SH4-domain effects are involved in cell invasion, these results suggest that the induced blebbing is instrumental for SH4-domain-induced 3D cell motility.

We next tested whether PM blebbing is a conserved function of SH4 domains of protooncogenes of the Src kinase family and created stable CHO cells lines expressing the SH4 domains of Fyn, Yes, Lck and Src. All these SH4 domains potently induced PM blebbing, with slight variations in their morphology, size and bleb number per cell (Fig. 7A) as well as the efficiency of bleb induction (Fig. 7B). Transient expression of full-length Lck-GFP also efficiently induced membrane blebbing in CHO cells (Fig. 7A,B). SH4 membrane blebs were non-apoptotic, required Rock and myosin II activity as well as F-actin and microtubule integrity, and, as tested for N18-Yes-GFP-expressing cells, correlated with enhanced 3D motility (data not shown). In addition, scanning electron microscopy was performed on CHO cells after transfection with expression constructs for N18-Yes-GFP, or for N18-Yes-Δmyr-GFP as a control (Fig. 7C,D). Control cells displayed some membrane ruffling at the periphery as well as at the dorsal surface and were decorated by numerous microvilli-like protrusions, but no PM blebs were observed (Fig. 7D). These cells were indistinguishable from mock-transfected control cells (data not shown). By sharp contrast, an abundant fraction of cells in cultures transfected with the N18-Yes-GFP expression construct was positive for a large number of blebs directly emanating from the PM at cell edges as well as at the dorsal surface, which additionally contained microvilli-like structures (Fig. 7C). Consistent with the above-presented fluorescence microscopy, blebs were variable in size and number, ranging from a few up to more than 100 per cell. Peripheral membrane ruffling also appeared slightly increased in such bleb-positive cells. Thus, PM blebbing is a conserved activity of Src kinase SH4 domains and the resulting blebs can cover large areas of the dorsal surface of SH4-domain-expressing cells.

To test whether membrane blebbing activity via SH4 domains involves the activity of the Src kinases, we analyzed the effects of inhibition of endogenous Src on SH4-domain-mediated PM blebbing (Fig. 8). The two well-established, structurally distinct specific Src kinase inhibitors PP1 and SU5565 were added to CHO cells expressing N18-HASPB-GFP (Fig. 8, left) or N18-Fyn-GFP (Fig. 8, right). Treatment with both inhibitors at 1 μM already markedly reduced PM blebbing. When 10 μM inhibitor was used, blebbing inhibition was within the range of that observed with addition of the Rock inhibitor Y-27632. Similar results were obtained for the SH4 domains of Src, Yes and Lck (data not shown). We conclude that endogenous Src activity is crucial for PM blebbing through SH4 domains.

The above results revealed a role of Rho, Rock and endogenous Src activity in the formation of SH4-domain-induced membrane blebbing. We next asked whether these factors act locally during bleb formation and analyzed their presence in PM blebs induced by the Yes SH4 domain. Confocal microscopy revealed significant amounts of RhoA as well as Rac1 in the limiting membrane but not the lumen of the blebs. By contrast, Cdc42 could only be detected in marginal amounts in some of the blebs analyzed (Fig. 9). Similarly, analysis of the distribution of Src revealed a modest but significant localization of the kinase in the membrane of PM blebs (Fig. 10A), and myosin light chain (MLC), a major substrate of Rock, was found prominently at the bleb base (Fig. 11A). No significant relocalization within the cells was observed upon expression of SH4 domains for RhoA, Rac1, Src and MLC (Fig. 10A, Fig. 11A and data not shown), suggesting that their incorporation into blebs might reflect their genuine partial PM localization (e.g. for Rac1, activity of which is not required for blebbing). This included the lack of displacement of Src from the PM upon SH4-domain overexpression, which indicates that, under these experimental conditions, the expressed SH4 domain does not saturate PM docking sites of endogenous Src. By contrast and in line with the inhibitory effects on bleb formation of the Src inhibitor PP1, the use of a phospho-specific antibody to specifically detect active Src revealed a marked accumulation in the lumen of SH4-domain blebs (Fig. 10B). A similar enrichment in the bleb lumen was found for pMLC, which was used as a surrogate for local Rock activity (Fig. 11B). Of note, we did not observe an overall increase in pSrc or pMLC levels in SH4-domain-expressing cells by western analysis (supplementary material Fig. S5A,B). Together, these results suggest that Rho-Rock and Src signalling cascades act at sites of bleb formation to facilitate SH4-domain-induced PM blebbing.

The presented functional analysis of SH4 domains revealed PM bleb formation as a novel and unexpected activity of these membrane-targeting motifs. According to bleb size and morphology, to the three-step mode of bleb dynamics as well as to the speeds of bleb expansion and retraction, SH4 blebs are remarkably similar to non-apoptotic blebs observed in M2 melanoma cells (Charras et al., 2005; Cunningham, 1995). Both modes of blebbing share the sensitivity to actin depolymerization and inhibition of Rho-Rock or myosin II. According to a model of bleb formation that integrates these requirements, local destabilization of cortical actin triggers ballooning of the PM driven by the hydrostatic pressure of the cytoplasm and possibly acto-myosin contractility at the neck of the bleb. Subsequent initiation of de novo actin polymerization at the bleb cortex and myosin II contraction then stops bleb expansion to ultimately cause retraction of the blebs (Charras et al., 2005). The mechanisms of bleb retraction thus appear to be similar in SH4-domain-expressing and M2 cells that lack functional filamin A expression. SH4-domain expression, however, did not affect filamin A expression levels or its serine phosphorylation (supplementary material Fig. S5C), suggesting that bleb initiation occurs via distinct pathways in these cell systems. The role of microtubule integrity in this process, which has not been analyzed in M2 cells, requires further clarification but could reflect microtubule-mediated transport of factors essential for bleb formation towards the PM. In line with this hypothesis, we observed that blocking Golgi transport at 20°C potently interferes with SH4-domain-induced membrane blebbing (data not shown).

Non-apoptotic cells generally fail to display PM blebbing under regular cell-culture conditions. In addition to the previously described M2 cells (Charras et al., 2005; Cunningham, 1995; Sheetz et al., 2006), the present study introduces SH4 domains as novel regulators that induce constitutive PM blebbing. Assuming that similar blebbing programmes are activated in SH4-domain-expressing and M2 cells, different cues can activate common machineries with efficiencies that probably depend on the cellular environment. Physiologically, such induction is observed with tumour cells that do not display membrane blebs when cultured in 2D but activate blebbing motility in a 3D environment (Sahai and Marshall, 2003). SH4-domain overexpression, however, resulted in the constitutive activation of blebbing in 2D that was maintained in a 3D environment, possibly by mimicking the constitutive activation of a cellular pathway. Because full-length SH4-domain-containing proteins such as HASPB and Lck also triggered PM blebbing in a manner dependent on SH4-domain integrity, we favour the hypothesis that such pathways rely on the function of SH4 domains from endogenous proteins.

Because SH4-mediated bleb induction correlated with increased cell invasion and 3D cell cultures are largely refractory to in-depth molecular analyses, we took advantage of SH4-expressing cells to study the molecular mechanisms of PM blebbing. Our attempts to define the determinants for PM blebbing in SH4 domains revealed that bleb induction requires PM localization of the SH4 domain. Based on the variety of SH4 domains capable of PM bleb induction, the mode of acylation [e.g. myristoylation only (Src) or in combination with palmitoylation (Yes)] does not determine blebbing activity. Sequence comparison of the various SH4 domains analyzed failed to reveal a common signature motif (supplementary material Fig. S4) and, thus far, we could not generate a SH4-domain mutant that does not induce blebbing but is transported to the PM (data not shown). The identification of such a mutant will be an important aim of future studies.

In terms of cellular machinery, our results identify endogenous Src activity as a factor that is crucially involved in bleb formation. Because inhibition of Src did not affect PM localization of the SH4 domains (data not shown), it probably represents a downstream effector. This also implies that SH4 domains might exert biological activities beyond their membrane-targeting function. Our localization analyses suggest that Rock and Src act locally at the sites of blebbing to promote bleb formation. Both, active Src and active Rock (as evidenced by MLC phosphorylation) were clearly enriched in the bleb lumen. Because we failed to detect a bulk induction of pSrc or pMLC in SH4-domain-expressing cells (supplementary material Fig. S5A,B), SH4 domains might induce blebbing by recruiting these factors via direct or indirect mechanisms. Alternatively, SH4-domain expression might cause local activation of these kinases without elevating their global levels of activity. Irrespective of the detailed mechanism, our data suggest that local kinase activities govern bleb dynamics.

How might Src signalling regulate SH4-domain-induced PM blebbing? A recent study on the dynamics of molecular composition of M2 blebs provided evidence for the successive recruitment of ezrin, actin, bundling proteins and contractile proteins to retracting blebs (Charras et al., 2006). Although not analyzed specifically, these observations suggest multiple possible scenarios of how Src could be involved in SH4-domain-induced blebbing. First, filamin is a substrate of Src kinases such as Lck and Src that can regulate its activity (Pal Sharma and Goldmann, 2004). Our analysis excluded global changes in filamin expression or its serine phosphorylation (supplementary material Fig. S5C); however, SH4 domains might downmodulate filamin activity via tyrosine phosphorylation by recruited Src. In this scenario, SH4 domains would use a different molecular mechanism to create a situation that is functionally equivalent to that in M2 cells. Second, ezrin, which stabilizes membrane-actin crosslinks in M2 blebs (Charras et al., 2006), is also subject to regulation by Src (Elliott et al., 2004; Srivastava et al., 2005). Finally, bleb retraction might be governed by actin nucleation events mediated by members of the Diaphanous-related formin protein family (Faix and Grosse, 2006; Eisenmann et al., 2007; Kitzing et al., 2007), proteins that also functionally cooperate with Src (Gasman et al., 2003; Koka et al., 2005; Tominaga et al., 2000). Current work in our laboratory attempts to differentiate between these possibilities.

Interestingly, Src is, in addition to the SH4-domain-induced blebbing motility described herein, also involved in the formation of invadopodia and podosome cell protrusions that facilitate cell invasion (Artym et al., 2006; Buccione et al., 2004; Hauck et al., 2002; Linder and Aepfelbacher, 2003; Tarone et al., 1985; Wyckoff et al., 2006). These structures mediate cell motility in 3D by promoting degradation of the extracellular matrix in a Rock-independent manner and are morphologically distinct from the PM blebs observed herein. Our study therefore suggests that Src kinases can promote 3D cell motility via several, fundamentally different, mechanisms. Together, the results presented here show that SH4 domains support bleb-associated cell motility by modulating PM dynamics in an Src-dependent manner. This novel activity of SH4 domains warrants future investigations on their role in bleb-associated motility of tumour cells.

All CHO and HeLa cell lines were cultivated in MEM alpha or DMEM medium (Invitrogen), respectively supplemented with 10% FCS and antibiotics. For the induction of transgene expression, 1 μg/ml doxicycline (dox) was routinely added for 24 hours. CHO-GFP, CHO-N18-HASPB-GFP, CHO-N18-Δmyr-HASPB-GFP and CHO-N18-Δpalm-HASPB-GFP cell lines have already been described along with the description of quantitative protein expression-level analysis by western blotting and flow cytometry (Engling et al., 2002; Stegmayer et al., 2005). For the expression of GFP fusion proteins of the SH4 domains of Yes (MGCIKSKENKSPAIKYRP), Fyn (MGCVQCKDKEATKLTEEN), Lck (MGCGCSSHPEDDWMENID) and Src (MGSNKSKPKDASQRRRSL), and mutants thereof, the respective coding sequences were cloned into pREV-TRE2 (Stratagene), which contains a transactivator/doxicycline-responsive element. Stable CHO and HeLa cell lines were generated by retroviral transduction and FACS sorting as described previously (Engling et al., 2002). MDA-MB-435 human breast carcinoma cells were received from ATCC. The expression constructs for fusion proteins of wild-type or activated GTPases were described previously (Gasteier et al., 2003; Moreau et al., 2000). The C3-GFP plasmid, as well as expression plasmids for myc-tagged GTPases, were kind gifts of Stefan Offermanns (University of Heidelberg, Germany). The expression plasmids for N18-Yes-wt-GFP and N18-Yes-ΔMyr-GFP as well as the full-length HASPB, HASPB-GFP and N18-HASPB-mCherry were generated by subcloning the respective coding sequences into pREV-TRE2. Expression plasmids for mCherry (Shaner et al., 2004) and mRFP-actin (Pacholsky et al., 2004) were kindly provided by Roger Tsien (UCSD, San Diego, CA) and Ulrike Engel (Nikon Imaging Center, Germany), respectively. The expression plasmid for the GFP fusion of Lck was generated by subcloning the Lck ORF into pEGFP-N1.

Antibodies used were as follows: mouse anti α-tubulin clone B-5-1-2 (Sigma), mouse anti c-Src clone B-12 (Santa Cruz), rabbit anti Src (pY⁴¹⁸) (Biosource), rabbit anti MLC2 (Cell Signaling), rabbit anti MLC (pS19) (Cell Signaling), mouse anti filamin 1 clone PM6/317 (Santa Cruz), rabbit anti filamin A (S²¹⁵²) (Cell Signaling), goat anti mouse IgG Alexa-Fluor-568 (Invitrogen), goat anti rabbit IgG Alexa-Fluor-568 (Invitrogen), goat anti mouse IgG Alexa-Fluor-660 (Invitrogen). Reagents used were as follows: Alexa-Fluor-660 phalloidin (Invitrogen), blebbistatin (Calbiochem), cytochalasin D (Calbiochem), DMSO (Merck), Doxicyline (Sigma), Hoechst 33258 (Sigma), LinMount (Linaris E6004) Nocodazole (Calbiochem), phalloidin-TRITC (Sigma), PP1 (Calbiochem), SU6656 (Calbiochem) and Y-27632 (Calbiochem).

For F-actin staining, cells grown on coverslips were fixed for 10 minutes at 4°C with 3% PFA, washed in TBS (50 mM Tris, pH 7.5, 150 mM NaCl), permeabilized with TBS/0.1% Triton X-100 (1 minute, room temperature) and incubated for 1 hour with 0.5 μg/ml phalloidin-TRITC and 1 ng/ml Hoechst 33258 in TBS. For staining of Src, pSrc, MLC, pMLC and Myc-tagged GTPases, cells were fixed and permeabilized in the same way, but for staining of α-tubulin, cells were fixed and permeabilized simultaneously with 3% PFA/0.1% Triton X-100 (10 minutes, 4°C). Subsequently, cells were washed with TBS, blocked with TBS/1% BSA (20 minutes, room temperature) and incubated with anti-Myc or -α-tubulin (1 hour, room temperature) or anti-Src, pSrc, MLC or pMLC antibody (over night, 4°C) in TBS, respectively. Cells were subsequently washed three times with TBS and incubated for 1 hour with the respective fluorescently labelled secondary antibody (Alexa-Fluor-568 or -660) together with phalloidin (TRITC or Alexa-Fluor-660) and Hoechst 33258 in TBS. After these staining steps, coverslips were washed with TBS before mounting on glass slides with LinMount and stored at 4°C. Apoptosis was analyzed by TUNEL assay using the in situ cell death detection kit, TMR red (Roche Applied Science) according to the manufacturer's instructions. Pictures of living cells in Lab-Tek chamber slides (nunc) and fixed cells on coverslips were taken from a bleb-rich middle section. Images were taken with a LSM 510 confocal laser scanning microscope (Zeiss) using 63× and 100× oil immersion objectives, respectively, and processed using Adobe Photoshop.

2D and 3D real-time sequences were acquired using a PerkinElmer Life and Analytical Sciences Ultraview LCI spinning disc confocal (Boston) mounted on a Nikon Eclipse TE200 microscope stand equipped with a Nikon Plan Apo 60× 1.4NA oil objective. The microscope was kept at 37°C by a microscope incubator box (EMBL Heidelberg). Time-lapses of single sections were acquired for up to 10 minutes with 150 ms exposure time (6.6 fps) using binning 1×1. 4D series were acquired for up to 15 minutes with 50 ms exposure time per section and a step size of 0.5 μm between sections covering the whole cell body. Kymographs of the 2D real-time sequences were created using ImageJ (http://rsb.info.nih.gov/ij/index.html) plugins as described (Pepperkok et al., 2005) using only steps 6-8 in the section `Analyzing the Kinetics of Transport Carriers'. Visualization of 4D data was done using Imaris (Bitplane).

For scanning electron microscopy (SEM), cells were fixed with 2.5% glutaraldehyde (v/v) in 0.1 M PBS with 1% sucrose (w/v), dehydrated in a graded series of ethanol and critical-point dried using carbon dioxide. Afterwards, cells were rotary-coated by electron beam evaporation using a BAF 300 freeze-etching device (Bal-Tec) with a single layer of platinum-carbon (coating thickness 3 nm). The samples were imaged in a Hitachi S-5200 in-lens field emission scanning electron microscope at an accelerating voltage of 4 kV using the secondary electron signal.

All GTPases used were expressed as fusion proteins with GFP. In order to analyze HASPB-dependent blebbing under these conditions, N18-HASPB was co-transfected as a fusion protein with the red fluorescent protein mCherry (Shaner et al., 2004). Cells were seeded on six-well plates and grown to a confluency of about 80%. For transient expression, cells were transfected with the constructs indicated using the calcium phosphate method (MBS transfection system; Stratagene). Transfected cells were passaged 24 hours post-transfection and transferred to microwell culture dishes (MatTek) to allow for in vivo imaging (48 hours post-transfection).

Transgene expression was induced with 1 μg/ml dox in N18-HASPB-GFP cells growing on glass coverslips for 24 hours. To synchronize PM bleb formation, membrane blebbing was efficiently abrogated by treatment with 90 μM Y-27632 for 2 hours. After extensive washing of the cells, de novo bleb formation was allowed for 3 hours in dox-supplemented medium containing a solvent control or 90 μM Y-27632, 100 μM blebbistatin, 1 μM cytochalasin D, 1-10 μM PP1, 1-10 μM SU6656, 100 μM Nocodazole. Subsequently, the cells were fixed and stained for F-actin or α-tubulin. Apoptosis was induced by incubation of cells with Etoposide (200 μg/ml) and cycloheximide (200 μg/ml) overnight.

Transwell assays were performed as described previously (Hooper et al., 2006). Briefly, transwell membranes (pore size 8 μm) (Greiner Bio-One) were coated with 20 μl of undiluted growth-factor-reduced matrigel (BD Biosciences). For the analysis of inducible CHO cell lines, cells were kept in the presence of 1 μg/ml dox throughout. Following 12 hours dox induction in the presence of 10% FCS, cells were starved for 12 hours, harvested and counted. 10⁶ cells were seeded in the upper chamber of the transwell in medium supplemented with 0.5% serum. The lower chamber was filled with medium supplemented with 10% serum. After 48 hours incubation at 37°C, cells in the upper chamber were carefully removed with a cotton swab and the cells that had traversed the membrane were fixed with 4% PFA, stained with Hoechst 33258 and counted at the microscope (Leica DMIRE2). To account for potential differences in cell seeding and/or proliferation during the 48 hours invasion period, cells were also plated in parallel in culture plates without matrigel, fixed at the time-point of harvest of the invasion samples and stained with Hoechst 33258 for chromatin. The number of nuclei in five representative areas was quantified and the number of invaded cells was normalized to the total number of cells. Maximum differences in cell numbers observed were 1.5-fold in parallel assays and increased cell proliferation observed in individual experiments was not correlated with the expression of any of the transgenes analyzed.

3D analyses were performed as described (Kitzing et al., 2007). Briefly, CHO cells stably expressing N18-HASPB-GFP were transfected with an mRFP-actin expression plasmid. 24 hours later, cells were counted and seeded directly into matrigel (BD Biosciences) and placed into μ-Slide VI chambers (ibidi). Cell morphologies were analyzed by taking z intervals of 1 μm with a 40× objective using confocal microscopy (Leica TCS SP2). 3D reconstructions and animations of the z-sections were performed using Leica Confocal Simulator Software.

6×10⁵ cells seeded in a 6 cm dish were treated with 1 μg/ml dox for 24 hours. The cells were lysed in 250 μl Rock lysis buffer (Coleman et al., 2001), the total protein amounts were detected using a BCA detection kit according to the manufacturer's instructions and, after addition of 50 μl 6×SDS sample buffer, the lysates were boiled at 100°C for 5 minutes. For SDS-PAGE, 40-50 μg of total protein was loaded. After transferring the proteins to a PVDF membrane, the membranes were blocked for 30 minutes with 5% milk in TBS-T (TBS + 1% Tween-20) before they were incubated over night with antibodies directed against the protein of interest. Secondary antibodies were either coupled to horse radish peroxidase for ECL detection or to Alexa-Fluor-688 antibodies for detection with the Odyssey infrared imaging system (LI-COR Biosciences).

This work was supported by grants from the Deutsche Forschungsgemeinschaft (grants Fa 378/6-1 and GK1188 to O.T.F., SFB 544-B15 and GK1188 to W.N., and Emmy Noether fellowship GR2111/1-2 to R.G.) and by a fellowship from the CHS Stiftung to O.T.F. We thank PerkinElmer for continuous support of the EMBL Advanced Light Microscopy Facility. We are grateful to OliverKeppler and Rainer Pepperkok for stimulating discussion and critical comments on the manuscript, to Roger Tsien, Michael Way, Ulrike Engel, Claudia Haller and Stefan Offermanns for providing reagents, and to Eberhard Schmid for expert technical assistance.