Cellular strain avoidance is mediated by a functional actin cap – observations in an Lmna-deficient cell model

Adherent cells of tissues forming the human body are interconnected with the surrounding extracellular matrix. Under physiological conditions these tissues and, consequently, cells are continuously exposed to a broad range of physical stimuli resulting from everyday use. These stimuli must be sensed and translated into appropriate cellular responses in order to maintain tissue homeostasis and functionality (Humphrey et al., 2014). This adaptive mechanism, called mechanotransduction, involves structural components, such as focal adhesions (FAs), cytoskeletal components and the nuclear envelope, as well as biochemical events, which transduce the stimuli from the cellular boundaries to the nucleus. In fact, the nucleus itself has been proposed to act as a mechanosensor. Intracellular deformations can alter chromatin conformation and modulate access to transcription factors or transcriptional machinery (Jaalouk and Lammerding, 2009). Upon transduction, the activation of specific genes allows the orchestration of cellular responses to physical cues (Ingber, 2009; Jaalouk and Lammerding, 2009; Vogel, 2006).

While the biochemical pathways of mechanotransduction have been the target of many investigations in the past decade (Chiquet et al., 2009; Hoffman et al., 2011; Vogel, 2006), there is growing evidence that a structural mechanotransduction pathway, physically linking the extracellular matrix to the nucleus, plays an equally important role in regulating cellular adaptive behavior (Chambliss et al., 2013). The possibility of a fast and direct transduction system was first suggested because of observations showing that the timescale of force propagation to the nucleus outmatches the speed of diffusion-based biochemical signals by orders of magnitude (Na et al., 2008; Poh et al., 2009; Wang et al., 2009, 2005). In addition, Wirtz and co-workers demonstrated that, in a two-dimensional (2D) environment, the subset of actin stress fibers running on top of the nucleus (the actin cap), together with the associated FAs, forms a bridge for the direct transduction of extracellular stimuli all the way to the nuclear genome. This is the result of the direct anchoring of actin cap fibers to the nuclear envelope and the underlying lamina meshwork via linker of nucleoskeleton to the cytoskeleton (LINC) complex proteins (Kim et al., 2013).

The nuclear lamina, a meshwork made of type V intermediate filaments called lamins, is essential for maintaining cellular structure and functionality. This lamina is essential in preserving nuclear shape and stiffness (Dahl et al., 2008) and is involved in numerous nuclear functions, such as chromatin organization, gene expression modulation, transcriptional activities and epigenetic chromatin modifications (Dechat et al., 2008). Recently, the role of the nuclear lamina has been extended even further. Discher and co-workers found that the nuclear lamina acts as a ‘mechanostat’ and is able to respond to changes in the mechanical properties of the surrounding cellular environment. In particular, the authors observed that the nuclear lamina shifts its composition to a stiffer meshwork of proteins with increasing stiffness of the surrounding environment, suggesting a key role for the nuclear lamina in the structural mechanotransduction pathway (Swift et al., 2013).

The impact of nuclear lamins on cellular mechanotransduction has become evident ever since mutations in the LMNA gene, encoding a family of nuclear lamins (A-type lamins), have been found to be implicated in the onset of a wide array of disease phenotypes, collectively called laminopathies (Bertrand et al., 2011; Worman and Bonne, 2007). Due to the absence of lamin A/C and LINC complexes, the structural mechanotransduction pathway in laminopathic cells is severely compromised, leading to impaired cellular mechanical properties and an abnormal mechanoresponse, which is particularly crucial in muscular (dys)function (Broers et al., 2004; Emerson et al., 2009; Hale et al., 2008, Lammerding et al., 2005; Lammerding et al., 2004; Schreiber and Kennedy, 2013; Tamiello et al., 2014). Of note, Lmna-deficient cells do not develop a functional actin cap, whereas the conventional basal stress fibers underneath the nucleus remain unaffected (Hotulainen and Lappalainen, 2006).

Here, we use Lmna-deficient mouse embryonic fibroblasts (MEFs) as a cell model to investigate the relevance of the actin cap to the cellular mechanoresponse. In particular, we concentrate on the differential response of actin cap fibers and basal actin fibers to competing physical cues: cyclic uniaxial strain and substrate topography. Intact tissue cells are able to sense and respond to topographical cues at the cellular and subcellular level by aligning the cell body and the actin cytoskeleton along the anisotropy of the substrate; a phenomenon referred to as contact guidance (Dunn and Heath, 1976; Teixeira et al., 2003). In contrast, when subjected to cyclic uniaxial strain of the underlying substrate, cells reorient their body and cytoskeleton away from the strain direction, obeying the so-called strain avoidance response (Faust et al., 2011; Kaunas et al., 2005). Thus, when contact guidance and cyclic strain are applied together and along the same direction, the two cues become competing stimuli for actin cytoskeleton (re)orientation. Recently, we investigated the combined effects of cyclic strain and anisotropic contact guidance – applied along the same direction – on the actin cytoskeleton of myofibroblasts. Interestingly, we observed distinct orientation responses of the actin cap and basal actin fibers within single cells. While actin cap fibers were prone to respond to cyclic strain by strain avoidance, even in the presence of anisotropy along the strain direction, basal fibers predominantly aligned with the direction of topographical anisotropy and did not respond to cyclic strain (Tamiello et al., 2015). We therefore hypothesize that cells without a functional actin cap, such as laminopathic cells, will not respond to cyclic strain but will respond to contact guidance.

To test this hypothesis, we used Lmna-deficient MEFs (Lmna^−/− MEFs) as a cell model for the most severe laminopathies (Muchir et al., 2003; Sullivan et al., 1999) and compared these to wild-type MEFs (Lmna^+/+ MEFs). The use of Lmna-deficient cells is an attractive alternative to pharmacological agents that inhibit actin cap formation. Inhibitors of myosin or actin polymerization have limited selectivity and generally affect both the cap and the basal actin stress fiber organization (Chambliss et al., 2013; Rehfeldt et al., 2007). In addition, they induce transient changes that are less compatible with prolonged cell remodeling studies. A leading study by Bertrand and colleagues (2014) applied Lmna-mutated myoblasts as a cell model to demonstrate defective strain sensing, but did not discern between subsets of actin fibers.

To apply contact guidance and cyclic uniaxial strain to Lmna^−/− and Lmna^+/+ MEFs in two dimensions (2D), the cells were seeded onto custom-developed stretchable micropost arrays consisting of either circular (isotropic) or elliptical (anisotropic) posts (Tamiello et al., 2015). We first report that both cell types show similar alignment of their cytoskeleton and maturation of their FAs in response to the topographical cues. The subsequent combination of topographical cues and cyclic uniaxial strain, however, revealed distinct responses of the two cell types. While wild-type MEFs displayed the typical strain avoidance behavior of adherent cells, the strain avoidance response of Lmna-deficient MEFs was hampered. We link this observation to the distinct actin cytoskeleton architecture of the latter cell type. The presence of the actin cap interconnected with the nucleus in normal cells triggers a fast and uniform response to cyclic strain, even in presence of topographical cues along the direction of strain. By contrast, the lack of an actin cap and the presence of a basal actin layer in Lmna-lacking cells hinder the strain avoidance response. These results highlight the importance of an intact nucleo-cytoskeletal connection for the correct mechanoresponse to strain and also enhance our understanding of defective mechanotransduction in laminopathies.

In accordance with previous studies (Broers et al., 2004; Khatau et al., 2009, 2010; Kim et al., 2012; Chambliss et al., 2013), the actin cytoskeletal architecture of wild-type MEFs (Lmna^+/+ MEFs) and that of MEFs lacking the Lmna gene (Lmna^−/− MEFs) showed apparent differences. The majority of Lmna^+/+ MEFs (73±5%, mean±s.d. from n=3 with 100 cells per experiment in confocal recordings) displayed a prominent actin cap and, to a lesser extent, an organized layer of basal actin fibers. 12±2% showed a disorganized cap, while 15±5% showed no cap. In contrast, Lmna^−/− MEFs developed an actin cap in only 25±8% of all cells, a disorganized cap in 28±10%, and no cap in 47±11% of all cells (n=3 with 100 cells per experiment). These cells showed a reduced amount of centrally located actin fibers and a noticeable and unaffected basal actin layer (Fig. 1A,B), as reported previously (Broers et al., 2004). Actin cap fibers aligned with myosin in cells containing a cap, while, in cells without an actin cap, myosin had a diffuse staining pattern without evidence of alignment into fibers (Fig. 1C,D).

This distinct actin architecture of the two cell types allowed dissecting of the response of cap and basal layer of actin stress fibers to physical environmental cues. To study the response to topographical cues, we seeded the cells on (anisotropic) elastomeric microposts with elliptical shapes. Cells were allowed to adhere to the microposts for 2, 6 and 24 h. The absence of bending of the microposts due to cellular attachment was verified (Fig. S1). These observations confirmed the rigidness of these posts as calculated (Table S1). Thus, in these experiments, the microposts can be viewed as not being able to be deformed by cells. Next, we examined the actin fiber orientation at the cap and basal layer of cells (Figs 2A; 3A–F) in confocal images of stained actin stress fibers. In Lmna^+/+ MEFs, actin cap fibers had already aligned parallel to the elliptical micropost major axis after 2 h, but lost their alignment within 24 h from seeding, while basal actin fibers appeared to be increasingly oriented along the microposts. For Lmna^−/− MEFs, we observed an increasing trend in alignment of actin cap fibers between 2 and 6 h, which stabilized afterwards, while the basal layer of actin fibers remained aligned with the micropost major axis throughout the course of the experiment.

In order to understand whether cap or basal actin fibers determine cellular orientation, cell alignment was determined by measuring the angle between the micropost major axis and the best-fitted cell ellipses (i.e. ellipses fitted to the outlines of the cell, see Fig. 2B) (Fig. 3A–F). These measurements revealed that cell orientation was parallel to the micropost major axis for both cell types. However, the degree of alignment showed a decreasing trend with time for Lmna^+/+ MEFs and an increasing trend with time for Lmna^−/− MEFs between 2 and 6 h. In addition, it was observed that Lmna^+/+ MEF orientation corresponded to actin cap fiber orientation at 2 and 6 h, while it followed basal actin fiber orientation at 24 h. For Lmna^−/− MEFs, cellular orientation corresponded to the basal layer organization at all investigated time points.

We also evaluated the formation of the actin cap with time by quantifying the anisotropy of the actin cap fibers using the Fibril Tool (Boudaoud et al., 2014) (Fig. 2C). A high anisotropy factor indicated the presence of highly organized actin cap fibers, whereas a low anisotropy factor represented a disrupted or absent cap. For Lmna^+/+ MEFs, the anisotropy increased significantly between 6 and 24 h (P<0.05), while for Lmna^−/− MEFs cap anisotropy remained stable. Lmna^−/− MEFs showed significantly lower anisotropy values than Lmna^+/+ MEFs for the total duration of the experiment (P<0.01, Fig. 3G). These data support the absence or defectiveness of a cap in Lmna^−/− MEFs and indicated that, in Lmna^+/+ MEFs, complete cap formation occurred within 2 h from seeding.

Cellular contact guidance involves the synergistic effects of subcellular tension and FA maturation (Saito et al., 2014). Therefore, we asked whether FA maturation was similar in Lmna^+/+ MEFs and Lmna^−/− MEFs. For this reason, we examined staining patterns of α-actinin 4 and vinculin. α-actinin 4 is recognized to play a major role in stabilizing the FAs (Choi et al., 2008; Feng et al., 2013; Ye et al., 2014), as it bridges vinculin proteins (Wachsstock et al., 1987) with actin stress fibers (Kanchanawong et al., 2010). Representative images of Lmna^+/+ MEFs and Lmna^−/− MEFs on elliptical microposts stained for vinculin and α-actinin 4 are shown in Fig. 4. These immunofluorescence images demonstrated that there were no overt differences between the two cell types. α-actinin 4 was observed on the microposts at the cell periphery. In addition, in Lmna^+/+ MEFs α-actinin 4 was present along the fibers of the actin cap. In Lmna^−/− MEFs, no α-actinin 4 was visible in the cap region, but only along the basal stress fibers. Vinculin, as well, was present at the cell periphery and colocalized with the α-actinin 4 staining. As expected, in contrast to α-actinin 4, vinculin was not found along the fibers of the actin cap of Lmna^+/+ MEFs.

Overall, these data suggest that both Lmna^+/+ MEFs and Lmna^−/− MEFs sense the topographical cues of the elliptical microposts and respond to them by orienting along the major post axis. In contrast to Lmna^+/+ MEFs, Lmna^−/− MEF alignment and actin stress fiber organization remained stable over a time period of 24 h, revealing a sustained sensitivity to topographical cues.

To test our hypothesis that cells with or without a functional actin cap would orient differently in response to contact guidance and cyclic strain applied along the same direction, we seeded Lmna^+/+ MEFs and Lmna^−/− MEFs on elliptical (anisotropic) fibronectin coated elastomeric microposts for 4.5 h (to induce functional cap formation in Lmna^+/+ MEFs) and subsequently applied uniaxial cyclic stretch along the major axis of the microposts. As a control, we performed the same experiment using cells on microposts with a circular (isotropic) cross section.

We found that, on circular microposts, Lmna^+/+ MEFs adapted to the applied cyclic strain by reorienting their cell body, as well as their actin cap fibers, to the (near) perpendicular direction (Fig. 5A). Lmna^−/− MEFs showed a less significant strain avoidance response, which was evident at the basal layer only. Lmna^−/− MEFs did not (re)orient their cell body in response to cyclic strain (Fig. 5B). In general, the cellular reorientation angles coincided with those of the cap layer in Lmna^+/+ MEFs and basal layer in Lmna^−/− MEFs.

When cells were subjected to contact guidance and cyclic uniaxial strain along the same direction (i.e. on the elliptical microposts; strained along their major axis), Lmna^+/+ MEFs responded by strain avoidance at the cap level. Cap fibers reoriented to two mirror-image angles, while the basal actin layer fibers remained aligned along the micropost major axis (Fig. 5C). In Lmna^−/− MEFs, both the basal actin layer and the actin cap remained aligned with the major axis of the microposts. Again, cell alignment corresponded to the main orientation of the actin cap fibers of Lmna^+/+ MEFs and the basal fiber direction of Lmna^−/− MEFs (Fig. 5D). Taken together, these data suggest that actin cap fibers sense and respond to cyclic uniaxial strain and demonstrate a strain avoidance response, even in the presence of competing topographical cues along the direction of strain. When the actin cap is disturbed or absent, as in the Lmna^−/− MEFs, defective strain avoidance is seen on circular microposts and strain avoidance is complete abolished on elliptical microposts (anisotropic contact guidance) Apparently, in these cells the alignment of the actin cytoskeleton, and consequently the cell body, is dictated by topographical cues.

We investigated also the degree of cap formation upon stimulation by cyclic uniaxial strain. We observed that Lmna^−/− MEFs scored cap anisotropy values significantly lower than Lmna^+/+ MEFs on both circular and elliptical parallel microposts (Fig. 5E). However, compared to static conditions, cap anisotropy upon cyclic uniaxial strain increased significantly for both cell types (P<0.001), indicating that strain stimulation enhances cap formation.

We wondered whether the lack of strain avoidance response on elliptical horizontal microposts could occur as a result of impaired stress fiber remodeling in Lmna^−/− MEFs. Since α-actinin 4, together with other proteins, such as zyxin and VASP, moves along stress fibers upon mechanical stimulation and enables the repair of strain sites (Smith et al., 2010), we investigated the localization of α-actinin 4 in Lmna^+/+ MEFs and Lmna^−/− MEFs upon mechanical strain. Fig. 6 shows the localization of α-actinin 4 in Lmna^+/+ MEFs and Lmna^−/− MEFs. We observed that α-actinin 4 staining was elongated at the sites of FAs and formed a periodic pattern along the actin cap fibers of Lmna^+/+ MEFs. Instead, in Lmna^−/− MEFs α-actinin 4 remained at the sites of FAs. Strikingly, Lmna^−/− MEFs with a cap were found to have an α-actinin 4 staining pattern that was very similar to that of Lmna^+/+ MEFs.

Since the presence of an actin cap appeared to be relevant for cellular reorientation, we checked whether only cells with a cap would reorient their stress fibers and cell body within the Lmna^−/− MEFs population on elliptical microposts. To this end, we first categorized cells as ‘reoriented’ when their angle relative to the micropost major axis or strain direction (θ) was greater than 20° and smaller than 160°, and cells as ‘remaining’ when their orientation angle was within 20° from the 0° or 180° angle (Fig. 2B). Second, we scored cells as (1) ‘no cap’ if their cap anisotropy score was <1.5, (2) with ‘disrupted cap’ if their cap anisotropy score was between 1.5 and 2.5 and (3) with ‘cap’ if their cap anisotropy score was >2.5. We could not confirm that only cells with a cap had reoriented since within the reoriented Lmna^−/− MEFs (∼40%), ∼12% had a cap, ∼12% had a disrupted cap and ∼16% had no cap (Fig. S2A). These data suggest that cells with disrupted or absent cap can also eventually reorient their cell body in presence of contact guidance and strain avoidance. Thus, the actin cap presence is not a prerequisite for reorientation upon strain and contact guidance but does facilitate cell reorientation.

To obtain further insight in the dynamics of cell and stress fiber (re)orientation, we focused on the cellular aspect ratio. We measured the cellular aspect ratio as the ratio between the long axis and short axis of its best-fitted ellipse; a rounder cell will have an aspect ratio close to 1 and a more elongated cell will have a higher aspect ratio. The aspect ratio of Lmna^+/+ MEFs was lower than the aspect ratio of Lmna^−/− MEFs under static conditions and remained unchanged upon cyclic mechanical strain (Fig. S2B). Further analysis of the Lmna^−/− MEF population showed that the aspect ratio of reoriented cells decreased. Although this change did not attain significance, this implies that cells round-up and lose a degree of polarization. Lmna^−/− MEFs that remained aligned with the elliptical parallel microposts showed a significant increase in aspect ratio compared to Lmna^−/− MEFs in static conditions (P<0.05). These data indicate that Lmna^+/+ MEFs are rounder by nature, while Lmna^−/− MEFs elongate in the presence of topographical cues. In response to mechanical strain, Lmna^−/− MEFs tend to obtain a rounder morphology, like Lmna^+/+ MEFs, suggesting cellular contractility and tension build-up. However, when they do not respond to cyclic strain and remain aligned with the topographical cues, they tend to further elongate.

As well as cellular aspect ratios, changes in cell area and cell height can be the results of processes involving cellular contractility and, as such, may influence the speed and efficiency of cell reorientation. Therefore, we analyzed these aspects for Lmna^+/+ and Lmna^−/− MEFs. Since nuclear lamin staining reveals a very sharp boundary of the nuclear membrane, allowing a relative accurate measurement of the nucleus in the z-direction, we chose to determine nuclear height rather than estimating cellular height (Fig. S3). The results show a moderately but significantly higher nuclear height in Lmna^−/− cells compared to in Lmna^+/+ cells (from 3.17 µm in Lmna^+/+ cells to 4.4 µm in Lmna^−/− cells). This increase is considerably less than previously published by Khatau et al. (2009), who described a 2-fold increase in the height of Lmna^−/− cells based on DAPI staining. Measurents of 2D cell areas of individual cells on round and elliptical microposts (determined manually from confocal images of 30 cells for three experiments) further revealed that the average cell area of Lmna^−/− MEFs was smaller than the cell area of Lmna^+/+ MEFs (Fig. S4). One could argue that a smaller, less spread and higher cell, as in the Lmna^−/− MEFs, can generate less force for cell orientation on substrates, even in the presence of a cap, necessitating the study of cellular contraction potential in cells with and without a cap.

Previous studies suggested a general downregulation of cytoskeletal and contractile proteins in laminopathic cells, especially in striated muscle cells (Buxboim et al., 2014; Chancellor et al., 2010; Solovei et al., 2013). We compared the expression levels of actin and myosin in Lmna^+/+ MEFs and Lmna^−/− MEFs, but could not confirm cytoskeletal and contractile downregulation in our laminopathic cells. Moreover, since the potential contraction of fibroblasts is determined by the presence of phosphorylated myosin light chain 2 (hereafter denoted phospho-myosin), we also quantified the expression level of this protein. Both immunofluorescence quantification and western blotting revealed no significant differences in actin expression between the two cell types. In addition, the expression of myosin (non-muscle myosin IIA) was similar in these cells as shown by western blotting (Fig. 7A–C). Ratios between actin and myosin expression were similar at the individual cell level. Comparison of myosin and phospho-myosin levels showed that a larger variation was seen in phospho-myosin levels and that the latter level was slightly enhanced in Lmna^−/− MEFs (Fig. 7D). Overall, these data suggest that Lmna^+/+ MEFs and Lmna^−/− MEFs do not differ in their contraction potential and that the reduced orientation response observed in the Lmna-deficient cells used in this study cannot be explained from a reduced expression of contractile proteins.

We investigated the differential response of actin cap fibers and basal actin fibers to the competing physical cues cyclic uniaxial strain and substrate topography. More specifically, we tested the hypothesis that cells without a functional actin cap, such as laminopathic cells, will not respond to cyclic strain but will respond to contact guidance. Using Lmna-deficient MEFs (Lmna^−/− MEFs) as a model system and intact fibroblasts (Lmna^+/+ MEFs) as control cells, we demonstrate that the absence of an actin cap does not compromise the cellular response to substrate topography. It does, however, impair the response to cyclic strain. We conclude that the presence of an actin cap facilitates the sensing of cyclic strain and mediates a rapid strain avoidance response by reorienting actin cap stress fibers and the cell body. The transmission of topographical cues occurs independently of the actin cap and is mainly governed by the conventional basal actin fibers underneath the nucleus.

By using elliptical elastomeric microposts, we found that the cellular mechanosensing and response to substrate topography under static conditions is independent of the presence of the Lmna gene. Both Lmna^−/− MEFs and Lmna^+/+ MEFs recognize and orient along the major axis of the microposts. In addition, both cell types express α-actinin 4, which colocalized well with vinculin. This is a strong indication of FA maturation (Feng et al., 2013) and is crucial for contact guidance (Saito et al., 2014). In the 2D environment provided by the microposts, Lmna^−/− MEFs and Lmna^+/+ MEFs demonstrated a contact guidance response by aligning the actin cap – if present – and the basal actin layer within 2 h from seeding. In Lmna^+/+ MEFs, the response became less prominent with time, while for Lmna^−/− MEFs, we observed enhanced alignment of the actin cytoskeleton and the cell body with time. These data suggest that A-type lamins are not needed for the alignment of the actin cytoskeleton in response to topographical cues. We speculate that in the case of an intact actin cytoskeleton, the topographic signals are quickly transmitted to the nucleus via the actin cap fibers (Chambliss et al., 2013). Instead, in cells without an actin cap, FA proteins might act as compensatory mechanosensors that trigger signal transmission to the nucleus via slower biochemical pathways (Isermann and Lammerding, 2013), consistent with the longer time scale required to align Lmna^−/− MEF cell bodies. Nevertheless, the molecular machinery that mediates topography sensing remains largely elusive and merits future investigations.

Our analyses of actin stress fibers and cell reorientation on circular microposts subjected to cyclic uniaxial strain revealed that Lmna^+/+ MEFs showed a normal strain avoidance response (Faust et al., 2011; Kaunas et al., 2005), whereas Lmna^−/− MEFs showed a much weaker strain avoidance response. Strikingly, the combination of uniaxial strain and contact guidance (using elliptical posts) along the same direction fully eliminated the strain avoidance response of Lmna^−/− MEFs, but not of Lmna^+/+ MEFs. While it has been observed that knockdown of the nucleo-skeletal connection (via the LINC protein nesprin-1) hinders the strain orientation response of endothelial cells (Chancellor et al., 2010), we demonstrated that the impaired strain avoidance response of Lmna^−/− MEFs is correlated with the distinct actin cytoskeleton of these cells, comprising only a basal layer of actin fibers.

Previously, we demonstrated that the combination of anisotropic topographical cues and cyclic strain along the same direction results in differential orientation responses of the actin cap stress fibers and the basal actin stress fibers within the same cell (Tamiello et al., 2015). Here, we used cells with or without an actin cap to dissect and interrogate the behaviors of the two actin layers in response to substrate topography and strain. It is known that the actin cap plays a crucial role in mechanosensing of environmental stiffness (Kim et al., 2012) and mediates fast mechanotransduction in response to fluid shear stress (Chambliss et al., 2013). Its contribution to strain mechanosensing and mechanotransduction, however, was not elucidated previously. Similar to stiffness and shear sensing, the high contractility and dynamics of actin cap fibers make them a candidate for strain mechanosensing. Nevertheless, one could argue that the basal actin layer is simply slower in responding to cyclic strain. The numerous interconnections of the basal layer with the underlying substrate, the exertion of traction forces to the substrate (Chancellor et al., 2010), and the reduced dynamics of the basal layer (Chambliss et al., 2013) could reduce its ability to quickly respond to the changing microenvironment. Whether the response of Lmna-deficient cells to cyclic strain is impaired or simply delayed due to altered actin dynamics (Ho et al., 2013) needs to be demonstrated, as we have only tested a limited number of time points and strain magnitudes.

Despite extensive investigations of cellular strain avoidance behavior, there is no consensus about the underlying mechanism of this behavior. Previous work has shown that cell reorientation might include a phase in which cells obtain a rounder morphology and subsequently enter a new phase of elongation at an angle to the strain direction (Jungbauer et al., 2008). In our study, we did not find a significant decrease in the cellular aspect ratio for Lmna^+/+ MEFs, but we did observe a decrease in the aspect ratio of Lmna^−/− MEFs that actively reoriented, even in the presence of topographical cues along the strain direction. This suggests that Lmna^−/− MEFs may need to acquire a rounder morphology in order to reorient away from the strain direction. In addition to this, our results make us speculate that, in healthy cells with a functional actin cap, reorientation is initiated and guided by the cap itself. Subsequently, via the interconnection of the actin cap with the nucleus, the strain signal is transduced to the nucleus, which regulates the reorientation of the cell body. However, this still does not explain the mechanism of stress fiber remodeling during reorientation. It remains to be elucidated whether stress fiber turnover (Lee et al., 2010) or stress fiber rotation and associated FA sliding (Chen et al., 2012; Goldyn et al., 2009) is the cause of fiber reorientation. Based on our observations, we propose that, in the actin cap, the rapid remodeling of stress fibers is caused by stress fiber rotation and FA sliding. At the basal layer, on the other hand, given the numerous adhesion sites, stress fiber turnover is the candidate mechanism for actin remodeling. This is in line with the lower actin remodeling dynamics observed at the basal layer. It is proposed that, at this layer, stress fibers need to disassemble and form de novo at a more favorable angle. To verify these assumptions, follow-up experiments involving the real-time imaging of stress fiber remodeling during mechanical strain are required to monitor the dynamics of the cap, basal actin stress fibers and FA reorientation.

Our results point out that cap anisotropy is enhanced upon mechanical straining. Both cell types (i.e. those with and without the Lmna gene) tended to produce and align more actin cap stress fibers on top of the nucleus upon mechanical straining. This can be regarded as an adaptive cellular mechanism for tension build-up and to promote cellular reorientation via stress fiber contraction. A similar adaptive response has been observed in cells migrating from soft to stiff matrices (Raab et al., 2012), while reduction of matrix stiffness reduced tension build-up and rescues nuclear disturbances in laminopatic cells (Tamiello et al., 2014). However, although cyclic strain triggered Lmna-deficient fibroblasts to develop thin actin stress fibers on top of their nucleus, the low contractility associated with these thin fibers and their low numbers did not seem enough to induce stress fiber and cell reorientation. The defective strain response could not be related to an absence or reduction of contractile proteins in Lmna^−/− MEFs in the present study. In fact, there was a trend towards an increase of contractile proteins in Lmna^−/− MEFs compared to their wild-type counterparts. Incorrect alignment or organization of actin or myosin fibers can be a cause of loss of contractility in laminopathic cells (Ho et al., 2013; Lanzicher et al., 2015). It is especially the attachment of actin fibers to the nucleus that has been shown to be impaired in Lmna^−/− cells (Folker et al., 2011), resulting in their inability to direct nuclear orientation (see e.g. Houben et al., 2009). In cardiomyocytes of Lmna^−/− mice, it has been shown that absence of lamin A/C causes disorganization of nesprin, leading to loss of the interconnection between actin and the nucleus (Nikolova-Krstevski et al., 2011).

Defective mechanosensing and/or mechanoresponses may be key mechanisms in the development of a variety of diseases associated with evading harmful mechanical forces on cells and tissues. While many studies do not differentiate between mechanosensing and mechanoresponses, our study indicates that, based on the initial attachment and alignment of actin fibers, mechanosensing in Lmna-deficient cells seems to be largely intact while the response is not. These data are in line with previous studies showing a defective mechanoresponse due to Lmna mutations or ablation of the LINC complex in 2D and 3D culture systems (Bertrand et al., 2014; Brosig et al., 2010). From these and other studies, it has become clear that both lamin A/C proteins and LINC complex protein disturbances affect mechanoresponsiveness, although a more pronounced effect is seen in cells lacking lamin A and C (Chambliss et al., 2013).

Insight into mechanoresponse of cells can aid in assessment of disease severity, both as a diagnostic and prognostic tool. In laminopathies, genetic mutation analyses are insufficient for diagnosis, since they do not predict the disease penetrance (Vytopil et al., 2002). New tools to measure mechanosensing and – perhaps even more important – the mechanoresponse, could aid in making important decisions in patient treatment and advice on behavior: people with a defective mechanoresponse are prone to generating excessive tissue damage upon heavy exercise.

In conclusion, we have demonstrated that sensing is not impaired in Lmna-deficient fibroblasts lacking an actin cap topography, while strain sensing and response are compromised. We therefore suggest that the actin cap in adherent cells plays an important mediating role in the structural mechanotransduction of cyclic mechanical strain. Cyclic strain and matrix topography are presented to tissue cells in a myriad of situations, and the ability of cells to respond to these cues is crucial for maintaining tissue functionality. The findings of this study broaden our understanding of cellular mechanotransduction, and shed light on the mechanisms and consequences of diseases of mechanotransduction, such as laminopathies.

Wild-type embryonic fibroblasts (Lmna^+/+ MEFs) as well as Lmna-knockout mouse embryonic fibroblast (Lmna^−/− MEFs) were obtained as described previously (Sullivan et al., 1999). Cells were cultured as described (Tamiello et al., 2014). Cell seeding density on micropost substrates for experiments was between 2000 and 2500 cells/cm².

The elastomeric micropost arrays were fabricated using standard photolithography, according to previous protocols (Tamiello et al., 2015; Yang et al., 2011). The microposts were characterized by a radius (r) of 1 µm in the case of circular post cross section (circular microposts) and a semi-major axis a of 1.5 µm and semi-minor axis b of 0.87 μm in case of an elliptic post cross section (elliptical parallel microposts). The center-to-center distance was 4 µm. Micropost height was 3 µm. Circular and eliptical microposts showed the same bending stiffness in the direction orthogonal to the applied stress. For static experiments, micropost arrays were bonded to glass coverslips (Menzel), while for dynamic experiments, microposts were bonded to the flexible bottom of six-well plates (Uniflex Series Culture Plates, Flexcell FX 5000, Flexcell International) using a corona discharger. To allow cell adhesion, the tops of PDMS microposts were functionalized with fibronectin as described previously (Tamiello et al., 2015).

Before cell seeding, the microposts were equilibrated in medium at 37°C for at least 15 min. Next, cells were seeded on top of the fibronectin-coated elastomeric micropost arrays and allowed to adhere under optimal culture conditions. At 2, 6 and 24 h after seeding, cells were fixed in 3.7% formaldehyde in PBS for 15 min at room temperature and prepared for analysis.

The elastomeric microposts arrays were subjected to uniaxial cyclic stretch using the FX-5000 Flexcell system (Flexcell Corp.; Mc-Keesport), applying a maximum strain level of 10% at a frequency of 0.5 Hz as described previously (Tamiello et al., 2015). Before applying the loading protocol, cells were seeded on top of the microposts for a time period of 4.5 h to allow cell adhesion. Next, the cells were cyclically loaded on the posts for 3.5 h, followed by cell fixation as described above.

To prepare for immunofluorescence studies, formaldehyde-fixed cells were permeabilized with 0.1% Triton X-100 (Merck) in PBS for 10 min and incubated with 3% bovine serum albumin (BSA) in PBS in order to block non-specific binding. The following antibodies or conjugates were used: Atto-488-conjugated Phalloidin (1:500, Phalloidin-Atto-488, cat. #49409, Sigma) or Texas-Red-conjugated Phalloidin (1:100, #T7471, Molecular Probes); monoclonal β-actin antibody (IgG1, 1:5000; #MUB0110S NordicMUbio, Susteren, The Netherlands); rabbit antibody anti-α actinin 4 (IgG1, 1:250; #EPR2533, Abcam); monoclonal mouse antibody hVIN-1 to vinculin (IgG1, 1:200; #V9131, Sigma-Aldrich); monoclonal mouse non-muscle myosin IIA antibody (IgG2B, 1:1000; #ab55456 Abcam); polyclonal rabbit Lamin B1 antibody (1:1000; #ab16048 Abcam); monoclonal phosphomyosin-light chain 2 antibody (1:200, IgG1; #3675, Cell signaling, Danvers, MA). All antibodies were applied for 1 h at room temperature. As secondary antibodies, either goat anti mouse-Ig conjugated to Alexa Fluor 555 (A21127, Molecular Probes, for vinculin) and goat anti rabbit-Ig conjugated to Alexa Fluor 488 (A11008, Molecular Probes, for α-actinin 4) were used. Alternatively, the combination of goat anti-mouse-Ig conjugated to Cy5 (ab97037, Abcam, for vinculin) and goat anti-rabbit-Ig conjugated to FITC (for α-actinin 4) with Phalloidin–Texas-Red allowed a triple labeling of the cell adhesion and actin structures. For double labeling of actin and myosin, and double labeling of myosin and phospho-myosin, goat anti-mouse-IgG1 conjugated to FITC and goat anti-mouse-IgG2B conjugated to TxRd (ITK, Southern Biotechnology) were used. For nuclear counterstaining staining, DAPI was added. Cells were mounted on glass coverslips using Mowiol or a glycerol mounting solution (Broers et al., 2004).

Confocal imaging of stained cells was performed as described (Tamiello et al., 2014). Z-series were generated by collecting a stack consisting of optical sections using a step size of 0.45-1.00 µm in the z-direction, while a minimum pinhole opening (1 AU) was used.

Actin stress fiber confocal Z-stacks of each analyzed cell were divided in two sub-stacks, one on top of the nucleus (cap) and one at the bottom, near the culture substrate (basal). For each subset, the maximum intensity projection images were used. Subsequently, the nuclear outline was tracked by thresholding the maximum intensity projection of the DAPI Z-stack. This outline was used as a mask for the cap and basal projections. Directionality analysis of the confocal images of cap and basal actin stress fibers was conducted using the Directionality plug-in in Fiji (http://fiji.sc/Fiji). This method exploits the image fast Fourier transform (FFT) algorithms. The 2D FFT determines the spatial frequencies within an image in radial directions and the output of the plug-in is a normalized histogram that reports the amount of structures at angles between 0° and 180° with a bin size of 2°. The reference angle was chosen as the major axis of the elliptical microposts in static conditions, and along the strain direction in dynamic conditions. Finally, the values for the average fiber fraction for each angle were calculated.

To quantify the extent of cap formation, the alignment of actin stress fibers on top of the nucleus was measured. This was carried out by using the FibrilTool plug-in of ImageJ on the confocal slice of actin stress fibers located on top of the nucleus. The nuclear outline was used as a region of interest, in order to measure the features of the cell actin cap only. The cap anisotropy score is given as a value between 0 (for no order in actin cap fibers orientation meaning a absent or disrupted cap) and 1 (perfectly ordered, parallel actin cap fibers).

Outlines of cells and the maximum intensity projections of actin stress fibers were visualized and measured for each experimental condition. Cell orientation was measured relative to the major axis of the elliptical microposts in static conditions and the strain direction in dynamic conditions. The orientation of the best-fit ellipse to the outline of each cell was measured using ImageJ software. Orientation angles were reported as a histogram for each experiment with bin size of 20°. Finally, the average cell orientation for each bin was calculated from three independent experiments. The cell aspect ratio was measured from the best-fit ellipse of cell outline. It is defined as the ratio between the long axis and short axis of the best-fitted ellipse. The aspect ratio is close to 1 for rounded cells and higher for more elongated cells.

Confocal recordings of actin, myosin and phospho-myosin stainings were made with identical settings for each antibody staining with a low axial resolution (20× lens, NA 0.5) and increased opening of the pinhole, resulting in a theoretical z-section slice of 6.1 µm (i.e. covering the thickness of an average cell in a single confocal slice). Care was taken to avoid pixel saturation in any recording. For each antibody staining, three different fields of view per cell line were analyzed. Double labeling (β-actin and myosin, myosin and phospho-myosin) was performed to compare fluorescence signals within individual cells. For each manually outlined cell the raw integrated optical density was calculated and divided by the number of cells, resulting in an average integrated optical density per cell.

Total cell lysates were made from cells that had reached a low (40% confluency) or high (80% confluency) density at 24 h after seeding. Samples were loaded onto an 8% or 12% SDS-polyacrylamide gel (Bio-Rad Laboratories, Berkeley, CA). Antibodies used were a non-muscle myosin IIA antibody, a lamin B1 antibody, a β-actin antibody and a monoclonal mouse anti-phospho-myosin-light chain 2 antibody. For details of antibodies, see above section on immunofluorescence. Immunoblot dilutions were: for the lamin B1 antibody, 1:15,000; for myosin IIA, 1:10,000; for β-actin, 1:5000; and for phosphomyosin light chain 2, 1:200. Secondary antibodies were horseradish peroxidase (HRP)-linked anti-mouse or rabbit-IgG antibody (1:3000; Cell Signaling, cat# 7076S & 7074S, respectively) Peroxidase activity was detected using an enhanced chemiluminescence western blotting detection kit (SuperSignal West Pico Chemiluminescent Substrate, Life technologies; and Western Lightning Plus, ECL).

Wild-type and Lmna-lacking MEFs were a gift from Dr Brian Burke (Nuclear Dynamics and Architecture Group, Institute of Medical Biology, Immunos, Singapore, Singapore). We thank Lynn Poolen (Maastricht University) for her help in performing pilot experiments for this study, and Gitta Buskermolen (Eindhoven University of Technology) for suggestions for revision.