Nuclear F-actin Formation and Reorganization upon Cell Spreading*♦

Background: Nuclear actin dynamics may function during basic cellular processes such as adhesion, differentiation, cell shape, and motility. Results: Integrin signaling induces nuclear actin polymerization via the LINC complex in response to fibronectin stimulation or cell spreading. Conclusion: Nuclear actin reorganization occurs during cell spreading and fibronectin stimulation. Significance: This may have important implications for understanding mechanotransduction and nuclear plasticity for matrix-directed differentiation. We recently discovered signal-regulated nuclear actin network assembly. However, in contrast to cytoplasmic actin regulation, polymeric nuclear actin structures and functions remain only poorly understood. Here we describe a novel molecular tool to visualize real-time nuclear actin dynamics by targeting the Actin-Chromobody-TagGFP to the nucleus, thus establishing a nuclear Actin-Chromobody. Interestingly, we observe nuclear actin polymerization into dynamic filaments upon cell spreading and fibronectin stimulation, both of which appear to be triggered by integrin signaling. Furthermore, we show that nucleoskeletal proteins such as the LINC (linker of nucleoskeleton and cytoskeleton) complex and components of the nuclear lamina couple cell spreading or integrin activation by fibronectin to nuclear actin polymerization. Spreading-induced nuclear actin polymerization results in serum response factor (SRF)-mediated transcription through nuclear retention of myocardin-related transcription factor A (MRTF-A). Our results reveal a signaling pathway, which links integrin activation by extracellular matrix interaction to nuclear actin polymerization through the LINC complex, and therefore suggest a role for nuclear actin polymerization in the context of cellular adhesion and mechanosensing.

Although actin is one of the most abundant proteins in the cytosol of mammalian cells, its presence in the nuclear compartment is comparably low, which complicates the investigation of its nuclear functions. In addition, actin dynamics in the cytosol are clearly determined by the spatiotemporal control of factors that orchestrate assembly and disassembly of actin filaments, whereas the regulation of nuclear actin is far more complex due to the fact that most proteins including actin regulators are subjected to nucleocytoplasmic shuttling in a highly dynamic manner (1,2). Actin itself is imported and exported by specific mechanisms involving importin 9 and exportin 6 (3,4). Thus, modulation of nuclear actin likely depends and is critically influenced by the signaling context as well as shuttling dynamics of its regulating factors. It is therefore not surprising that the formation of nuclear F-actin has been controversially discussed up to the point that nuclear actin was considered to be entirely or mainly monomeric in its nature (5)(6)(7)(8)(9). Recently, we and others have demonstrated the existence of nuclear F-actin structures in somatic cell nuclei using actin probes, which are specifically targeted to the nucleus (10,11). This enabled us to directly visualize the formation of a highly dynamic endogenous actin network that rapidly formed upon serum stimulation in intact mammalian cell nuclei (10,12). Thus, cells possess the ability to assemble long nuclear actin filaments into organized network-like structures within intact nuclear compartments in response to extracellular ligands. However, whether other and more general cellular processes would involve nuclear actin polymerization remains to be elucidated (13). Here we show that cells can form nuclear F-actin upon adhesion and spreading, which depends on functional integrin signaling and components of the LINC 2 (linker of nucleoskeleton and cytoskeleton) complex. Nuclear actin polymerization can additionally be triggered by direct integrin activation using fibronectin and appears to be mediated by mDia formins. Our data uncover a cell spreading-induced nuclear actin response that involves the nucleoskeleton and components of the nuclear lamina.
Cell Lines, Transfections, and Treatments-NIH3T3 and NIH3T3-nAC cells were maintained in DMEM supplemented with 10% (v/v) serum (FCS) at 37°C in a 5% CO 2 atmosphere. NIH3T3 cells stably expressing nAC where generated by lentiviral transduction. The eGFP in pWPXL-eGFP was replaced by nAC using BamHI and SpeI. Viruses were produced in HEK293T cells by simultaneous transfection of pWPXL-nAC as well as the packaging vector psPAX2 and the envelope plasmid pMD2.G. Supernatant containing lentiviral particles was harvested 48 h after transfection. NIH3T3 cells were transduced with viral particles in the presence of 10 g ml Ϫ1 Polybrene (Sigma).
When indicated, cells were treated with 80 g ml Ϫ1 soluble fibronectin (derived from bovine plasma, Sigma). In Fig. 4C, NIH3T3-nAC cells were preincubated with antibodies (5 g/ml) for 1 h prior to fibronectin (FN) stimulation. The integrin inhibitory antibody (4B4, Beckman Coulter) was compared with nonspecific IgG (Santa Cruz Biotechnology). To observe nuclear actin network assembly upon serum stimulation (see Fig. 1C), cells were starved for 16 h in serum-free medium before stimulation with 20% serum (FCS).
Cell spreading was analyzed by live cell imaging or in fixed samples. Spreading cells are characterized by a more rounded shape, whereas cells with an overall flat profile were considered as control cells. In any case, quantifications were performed from three independent experiments by counting at least 50 nuclei per experiment. Cells that displayed a diffusely distributed nAC signal (i.e. see Fig. 1C, left panel) were quantified as negative, whereas cells displaying nuclear actin filaments of any type or size were quantified as positive.
MRTF-A/SRF Luciferase Reporter Assay-MRTF-A/SRF activity was assayed as described previously (14) using p3DA.luc and pRL-TK reporter plasmids. After transfection, cells were cultured for 24 h, subsequently used in a cell spreading assay, and lysed at the indicated time points.
Immunofluorescence and Image Analysis-Cells grown on coverslips were fixed in 3.7% formaldehyde and permeabilized with 0.3% Triton X-100 before incubation with desired antibodies. Secondary antibodies labeled with Alexa Fluor 488 and Alexa Fluor 555 were from Invitrogen. Slides were mounted with fluorescent mounting medium (Dako). Images were col- lected with an LSM700 confocal microscope (Zeiss) equipped with a 63ϫ, 1.4 NA oil objective. A 10ϫ objective was used in Fig. 5D. Image processing was performed with Fiji (National Institutes of Health).
Visualization of nuclear actin filaments using phalloidin was performed as described previously (10). Confocal z-stacks and DAPI stainings were used to confirm the correct localization of the focus plane. Nuclear MRTF-A translocation during cell spreading was quantified by comparing the fraction of cells with a predominant nuclear signal (confirmed by DAPI staining) to cells with a predominant cytoplasmic signal.
Live Cell Imaging-For live cell imaging, cells were grown, transfected, and treated in -slides (ibidi). Imaging was performed at 37°C in a CO 2 (5%) humidified chamber using an LSM700 confocal microscope and the ZEN software (Zeiss). Maximum intensity projections were calculated with Fiji.
Immunoblotting and Nuclear Fractionations-Nuclear fractionation was performed as described previously (10).
Statistics-Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software). Data are presented as mean Ϯ S.E. unless stated otherwise. Statistical tests were performed as described in the respective figure legends. Statistical significance is defined as p Ͻ 0.05.

RESULTS AND DISCUSSION
We previously generated a nuclear actin probe by fusing Life-Act to an NLS (10). Although this allowed us to reliably detect endogenous nuclear actin polymerization and depolymerization, LifeAct can have some limitations as its expression level needs to be carefully titrated and monitored to prevent any potential stabilization of assembled F-actin structures. We therefore turned to a recently described, antibody-based FIGURE 2. Nuclear actin polymerization occurs during cell spreading and can be detected by different nuclear actin probes. A-C, NIH3T3 cells were transfected with the indicated (nuclear) actin probes. Spreading cells were assessed for nuclear F-actin formation (green) versus non-spreading (control) cells. LifeAct-mCherry (red) was transfected in A to visualize cytoskeletal actin. Scale bar, 10 m. D, spreading and non-spreading (control) NIH3T3 cells were assessed for nuclear F-actin formation. Endogenous nuclear F-actin was visualized using phalloidin (Alexa Fluor 488-conjugated, green). Nuclei were labeled using DAPI (blue). Scale bar, 10 m. approach to visualize endogenous proteins utilizing the Chromobody technology (15). We targeted the Actin-Chromobody-TagGFP to the nucleus, generating Actin-Chromobody-Tag-GFP-NLS, herein termed nAC (Fig. 1A). Transfection of nAC did not affect the abundance of endogenous nuclear actin (Fig.  1B). We validated this approach by inducing nuclear actin network assembly with serum stimulation in living cells ( Fig. 1C; supplemental Movie S1). The observed nuclear F-actin structures as well as the temporal resolution were comparable with our previous findings using LifeAct-GFP-NLS (10). Thus, nAC represents a novel and robust tool to visualize and monitor endogenous nuclear actin dynamics in living cells.
Next we generated stably nAC-expressing NIH3T3 cells to monitor nuclear actin assembly more reliably without transient transfections. While culturing and plating these cell populations, we noticed the formation of distinct nuclear actin filaments during cell spreading (Fig. 2A). These nuclear F-actin structures during spreading were also detectable with a modified nAC probe containing an additional NES (Actin-Chromobody-TagGFP-NLS-NES) to allow for dynamic, nucleocytoplasmic shuttling of the probe (Fig. 2B) or with a comparable LifeAct-GFP-NLS-NES construct (Fig. 2C). Of note, these probes also monitored F-actin structures at the plasma membrane during cell spreading (Fig. 2, B and C). We could confirm the formation of nuclear F-actin during cell spreading using glutaraldehyde fixation and phalloidin staining as the bona fide F-actin marker (Fig. 2D). Taken together, these data reveal the existence of nuclear F-actin formation during cell adhesion and spreading. Of note, spreading-induced nuclear actin filaments appeared to be shorter and thicker than those observed during serum stimulation, indicating that their structure differs signif-icantly from the rapidly and transiently forming network in the serum response.
We subsequently analyzed and quantified spreading-induced nuclear actin assembly over time. As shown in Fig. 3A, nuclear F-actin was detectable within 30 min after plating on precoated coverslips and could persist for 2-3 h before disassembly. On average, the maximum response under these conditions occurred at around 25 min after plating before slowly declining over a time period of 3 h or longer ( Fig. 3B; supplemental Movie S2). Thus, in contrast to the very rapid and transient F-actin burst during the serum response ( Fig. 1C; supplemental Movie S1) (10), spreading-induced nuclear actin assembly is more persistent, and these nuclear actin filaments appear to be shorter and thicker in nature. Of note, these nuclear actin filaments did not colocalize with cofilin (data not shown), which decorates stress-induced nuclear F-actin (16). They were further detectable with LifeAct (Fig. 2, A and B), previously shown not to recognize stress-induced actin-cofilin rods (17). On matrix-coated surfaces, fewer than 80% of all cells displayed detectable nuclear actin filaments during live cell imaging over time, which could indicate that nuclear actin regulation may be dependent on the cell cycle or that it may be influenced by additional, unknown factors.
Cell spreading is a mechanosensing process that involves the formation of integrin-based adhesion to the extracellular matrix (18). Thus, we used soluble FN to directly promote and test ligand-induced integrin activation. Interestingly, stimulation of cells with FN readily triggered nuclear F-actin formation within 30 min (Fig. 4A; supplemental Movie S3). This closely resembles the temporal and structural characteristics of spreading-induced nuclear actin assembly as observed in Fig. 3.  FN-induced nuclear actin polymerization was reversible after removal of FN-containing medium (Fig. 4A; supplemental Movie S3).
To assess whether the observed nuclear actin filaments were due to de novo actin polymerization, we transfected NIH3T3-nAC cells with the nuclear targeted, non-polymerizable form of FIGURE 5. A, quantification of nAC-probed nuclear F-actin formation in NIH3T3 cells at the indicated time points after plating onto glass coverslips. Results are shown as mean Ϯ S.E. from three independently performed experiments. B, knockdown of LINC complex components and mDia1/2 corresponding to panels C and D and Fig.  4, B and E. Left panel, immunoblots for emerin and lamin A/C from siRNA-treated cells as indicated. Right panel, RT-quantitative PCR results relative to control siRNA for Sun1, Sun2, mDia1, and mDia2 from siRNA-treated cells as indicated. C, quantification of nuclear F-actin formation in NIH3T3-nAC cells. Cells were treated with the indicated siRNAs for 48 h prior to plating onto glass coverslips. Error bars represent S.E., ****, p Ͻ 0.001; ns, not significant; data were collected from three independent experiments and analyzed using one-way analysis of variance followed by Tukey's honestly significant difference test. D, bright-field images of spreading NIH3T3 cells treated with the indicated siRNAs at4handϾ12 h after plating. Scale bar,50m. E, representative images of NIH3T3 cells at the indicated time points after plating onto glass coverslips. Cells were stained for endogenous MRTF-A (green). Scale bars, 20 m. Cells were labeled using DAPI. F, quantification of nuclear MRTF-A during cell spreading in dnDia.NLS-or pLink-transfected cells. Error bars represent S.E. G, relative luciferase activity in NIH3T3 cells at the indicated time points after plating onto glass surfaces. Cells were transfected with luciferase reporter plasmids (pRL-TK and p3DA.luc) and either pLink (control) or dnDia.NLS.
actin NLS-Flag-R62D (19,20). These cells did not form FNinduced nuclear F-actin (Fig. 4, D and E), suggesting that they arise from actin polymerization and not merely from bundling of pre-existing nuclear actin filaments.
To further exclude potential, unspecific effects of fibronectin, we made use of the integrin inhibitory antibody 4B4. Cells preincubated with 4B4 did not display FN-induced nuclear F-actin ( Fig. 4C; supplemental Movie S5). Thus, integrin signaling is an integral and necessary component of this pathway.
We then hypothesized that the observed nuclear actin response involves the nucleoskeleton because this structure has recently been shown to be critically involved in mechanosensing and intracellular force transmission via the LINC complex (21)(22)(23). The LINC complex provides a physical link between the cytoskeleton and the nuclear envelope (24). This is achieved through the nuclear membrane proteins Sun1 and Sun2, which interact with lamins and other components of the nuclear lamina (21). Therefore, we suppressed Sun1 and Sun2 expression by siRNA and again stimulated cells with FN. Notably, this resulted in a significant loss of FN-promoted nuclear F-actin formation (Fig. 4, B and E; supplemental Movie S4), demonstrating a critical role for the LINC complex in integrin-triggered nuclear actin assembly.
Next we assessed the role of the nucleoskeleton and nuclear lamina during cell spreading-induced nuclear F-actin formation. For this, we used glass coverslips, allowing more convenient analyses and subsequent quantifications. Although as may be expected the overall response of cells displaying nuclear actin assembly during spreading was delayed and decreased as compared with extracellular matrix-coated surfaces (Figs. 3 and 5A), we found that both Sun1 and Sun2 were essentially required for nuclear F-actin formation, but not for silencing of either Sun1 or Sun2 individually (Fig. 5, B and C), which indicates an at least partially redundant function for each Sun protein in this context. Moreover, siRNA against emerin or lamin A/C also resulted in a failure of cells to generate nuclear actin filaments (Fig. 5, B and C), further supporting the notion that the spreading-induced nuclear F-actin response involves a functional nuclear lamina. As may be expected, we found that spreading-induced nuclear actin filaments are sensitive to mDia1/2 silencing (Fig. 5, B and C), indicative of a role for these mDia formins. None of these siRNA treatments prevented cell spreading per se (Fig. 5D).
We previously demonstrated that nuclear actin assembly critically regulates the SRF coactivator and actin-binding protein MRTF-A (also called MAL or MKL-1) (10). It has been shown that cell adhesion and spreading require transcriptional regulation of MRTF-A (25,26), which localizes to the nucleus in response to nuclear actin dynamics or F-actin formation and causes the release of monomeric actin from MRTF-A, enabling SRF-mediated transcription (10,27). We therefore analyzed endogenous MRTF-A distribution during spreading and found that nuclear MRTF-A appearance correlated with the detection of nuclear F-actin (Fig. 5, E and F). Importantly, spreadinginduced nuclear MRTF-A localization was blocked after expression of a previously described nuclear dominant negative mDia (dnDia.NLS) (10) (Fig. 5, F and G). This demonstrates that nuclear formin activity is responsible for MRTF-A localization to the nucleus during cell spreading. Nuclear MRTF-A appearance and SRF activity appear to be slightly delayed in comparison with nuclear F-actin detection during spreading, possibly reflecting the notion that MRTF-A/SRF regulation is a consequence of nuclear actin assembly.
Here we identified an adhesion-triggered pathway that promotes the formation of nuclear F-actin during cell spreading (Fig. 6). Interestingly, although the shape of these nuclear filaments differs remarkably from those observed after serum stimulation (10), they appear to be nucleated by the same group of mDia formin regulators. Thus, different pathways may converge at nuclear formin activity to induce linear actin filaments of various length and organization, further suggesting that additional yet unknown actin regulators cooperate. This is consistent with the view that many actin-regulating proteins are detectable in the nucleus (7,28,29). We find that spreadinginduced nuclear actin assembly can regulate MRTF-A similar to the serum-induced response. However, the spreading response is much slower and more persistent in nature than the very rapid network formation, which occurs within seconds upon serum stimulation (Fig. 1C). Thus, it seems tempting to speculate that additional nuclear functions may be regulated as a consequence of actin polymerization in the nucleus; spreading-mediated nuclear actin dynamics could be involved in changes in chromatin organization (5,30) or in the control of nuclear shape and positioning such as reported during cell migration (31,32).