Effect of ROCK Pathway Manipulation on Actin Cytoskeleton Architecture: The Third Dimension


 The actin cytoskeleton with its dynamic properties serves as the driving force for the movement and division of cells and gives the cell shape and structure. Disorders in the actin cytoskeleton occur in many diseases. Deeper understanding of the regulation is essential in order to better understand these biochemical processes. In our study, we use metal-induced energy transfer (MIET) as a tool to quantitatively examine the rarely considered third dimension of the actin cytoskeleton with nanometer accuracy. In particular, we investigate the influence of different drugs acting on the ROCK pathway on the three-dimensional actin organization. We find that cells treated with inhibitors have a lower actin height to the substrate while treatment with a stimulator for the ROCK pathway increases the actin height to the substrate. This reveals the precise tuning of adhesion and cytoskeleton tension, which leads to a rich three-dimensional structural behaviour of the actin cytoskeleton. This finetuning is differentially affected by either inhibition or stimulation.


Introduction
The cells cytoskeleton gives mechanical support, allows the cell to deliver cargo and is fundamental for cell division. It consists of three biological polymers: microtubules, intermediate laments and actin. The actin cytoskeleton is predominantly responsible for cellular movement. It is tightly regulated and highly dynamic. Actin is located in a variety of areas of the cell (Fig. 1a). It is found in the cell cortex to give the cell strength and shape [1]. To divide cells, it also forms the contractile ring around the nucleus [2,3]. For motility, it forms microvilli [4], lamellipodia and lopodia [5][6][7][8].
The actin cytoskeleton can create contractile forces in the cell. Contractile forces are generated when actin laments recruit myosin II motor proteins [8,9]. The poly-protein complex of contractile bundles of actin laments with myosin II molecules are called stress bers [10]. In contrast, protrusive forces arise from the coordinated polymerisation of multiple actin laments. Gradients of tension induce local contractions and force cell movements and deformations. [8,9]. These forces are not only important in cell migration, but also play a role in determining cell shape, adapting to the mechanical properties of the environment, determining intracellular movement [11] and morphogenesis of membrane organelles [12].
Filamentous actin (f-actin) is also involved in the generation of mechanical forces which stimulate the differentiation and development of stem cells [13].
Regulated polymerization and depolymerization of actin monomers to polymers leads to a dynamic assembly and disassembly of actin bers, which ensures quick cellular adaptation to changes in environment [11,14].The small Rho GTPase RhoA is a major regulator of the cytoskeleton (Fig. 1B). It activates the Rho-associated coiled coil-containing protein kinase (ROCK) [15]. Through a positive feedback loop, ROCK directly catalyses myosin light chain phosphorylation and inhibits myosin light chain phosphatase [16]. This increases the binding of myosin II to actin and thus the contractility of the cell. ROCK also phosphorylates LIM kinase (LIMK), which then phosphorylates co lin [17]. This deactivates actin depolymerization and the existing actin laments are stabilized and increase in length.
High expression of ROCK/LIMK/Co lin and other members of the ROCK pathway have been identi ed in cancers such as breast, prostate, colorectal and bladder [18][19][20][21]. In tumour cells, increased movement, invasion and metastasis also occur due to the actin cytoskeleton and its adaptation [8]. Viruses modify cells and the actin to best suit their hosts and proliferate [22,23]. Cell movement through the actin cytoskeleton plays a critical role in many pathologies. During chronic in ammatory diseases migration of immune cells is important [24,25] and actin dynamics are a target for therapeutics against chronic kidney diseases [26]. Ageing processes also change the actin expression and dynamics [27]. Therefore, actin modi cations are involved in ageing, cancer, vascular diseases, and neurogenerative diseases like Alzheimer's disease [27,28]. Disrupted functioning of actin is also observed in somatic cells, stem cells and gametes [13,27]. This underlines the necessity to gain a deep understanding of the Rock pathway on actin structure. In the past, a variety of high resolution uorescence studies described the actin organization with high precision in two dimensions [29][30][31]. However, the third dimension, perpendicular to the focal plane, is more di cult to resolve with nanometer resolution. Metal induced energy transfer is a tool that facilitates a high precision analysis of this third dimension.
With metal induced energy transfer (MIET) as a tool, the position of uorescent molecules above a metal surface can be determined with nanometer precision [32][33][34][35]. The principle of MIET compares best to Förster Resonance Energy Transfer (FRET). In FRET the energy of an excited donor dye is transferred to a second dye, the acceptor [36,37]. However, at MIET, the acceptor molecule is replaced by a metal layer.
This results in a distance-dependent energy transfer rate between the metal layer and the donor molecules, which can directly be correlated to the uorescence lifetime [32] and is observable at distances from the metal of more than 100 nm. The effect is based on strong optical near-eld coupling of surface plasmons in a thin metal layer [33]. The uorescence lifetime increases non-linearly with increasing distance to the metal layer and asymptotically reaches the lifetime of a free uorophore without metal layer [32,34]. The uorescence lifetime can hence be converted to a distance value using a calibration curve to obtain super-resolution in the Z dimension [32]. MIET imaging has already been used to study the cell mechanics of human mesenchymal stem cells [34], to better understand human blood platelet spreading[38] and for the examination of microtubules [39] and the nuclear envelope [40].
In this study, we investigate the regulation of the actin cytoskeleton through different targeting sites in the ROCK pathway. To this end, we use Blebbistatin as a myosin II inhibitor [41], Y27632 as an inhibitor for ROCK [42,43] and the Rho Activator II, which increases the activity of RhoA [44,45]. We use MIET to quantitatively observe the height of actin stress bers and the actin cortex within nanometer accuracy. With this, we gain a better understanding of the structure of the actin cytoskeleton in the third dimension and thereby the in uence of the different key points of the ROCK Pathway and their impacts on the threedimensional actin structure. Figure 1A shows a sketch of an adherent cell with its actin cytoskeleton (red). In this work, we analyze both stress bers, which are formed with the aid of myosin II, and the actin edge (blue), which is composed of various actin components such as cortical actin as well as stress bers. In addition, the gradient of actin height along the actin edge is analyzed. Our results show that by treatment with Rho Activator, actin distance to the substrate is higher and distributed on a smaller projected cell area. In contrast to this, the treatment with the inhibitor Y27632 leads to lower actin heights. Blebbistatin, which only affects myosin II activity but does not directly interfere with the actin assembly pathway, does not lead to major changes in height, but attens the actin edge of the cell Our results indicate that there is a precise balance between adhesion tension and cytoskeleton tension of the actin cytoskeleton regulating the three-dimensional architecture. A disturbance of this balance disrupts the nely balanced interplay of forces and leads to attened actin architecture.

Results
The major goal of this study was the examination of the distance from the substrate to the actin cytoskeleton, a parameter we call "actin height". We analyzed the height of a) all actin components together; b) the height of actin stress bers; and c) the height of the actin edge -the outer edge of the actin in the cell -separately (Fig. 2). Further, we analyzed d) the gradient of the actin height along the edge of the cell, a parameter that shows us the roughness of the cell edge. Figure 2 shows in rst column representative images generated via Fluorescence Lifetime Imaging Microscopy (FLIM). Using a previously created calibration curve, the uorescence lifetimes were converted into height values [32,33,35,46,47] to get intensity-weighted height image as described in Methods. In order to also get speci c height information about stress bers, a stress-ber speci c mask was generated unbiased via the TWOMBLI plugin [48]. For further analysis of the actin cortex, the actin edge was extracted from the height image of all components. To examine the regulation of actin by the pathway, untreated cells ( Fig. 2a) were treated with different drugs. Blebbistatin inhibits Myosin II (Fig. 2b), Y27632 inhibits ROCK (Fig. 2c) and the Rho Activator increases the activity of RhoA (Fig. 2d).

Height distribution of all actin components
From the intensity weighted height images (Fig. 2b), the median height of each cell's actin components was evaluated (Fig. 3a). Untreated cells have a median of 114 ± 12 nm. Cells treated with the ROCK inhibitor Y27632 show the lowest median at 97 ± 15 nm, followed by cells treated with the Myosin inhibitor Blebbistatin at 110 ± 17 nm. No signi cant difference between Blebbistatin treated cells and untreated cells can be found. Signi cantly, Rho Activator treated cells show the highest median at 124 ± 12 nm.

Height distribution of stress bers
The median of the stress ber heights of each cell (Fig. 3b) shows a signi cant difference of the Rho Activator population with a median of 132 ± 10 nm compared to all other populations. This is even higher ( + ≈ 10 nm) compared to the median of the total actin of the Rho Activator population (Fig. 3a), indicating a signi cant upshift of the stress bers compared to other actin structures. The height of stress bers of Y27632 treated cells are clearly shifted to lower heights compared to untreated cells.
Despite a higher median of the height of actin stress bers in Blebbistatin treated cells, there is no signi cant difference between Blebbistatin and Y27632 treated cells or between Blebbistatin-treated and untreated cells. The MIET data reveal that untreated cells (Fig. 2, 4a) have arch-like structures between the adhesion points. The Blebbistatin (Fig. 2, 4b) treated cells show an overall similar architecture. The Y27632 treated cell contains very few stress bers and has very limited actin overall. The few stress bers show arch-like structures (Fig. 2, 4c). The Rho Activator cell (Fig. 2, 4d) has actin stress bers at a higher height. The stress bers, however, show no arch-like features.
We further compared the height of the lowest and highest stress ber regions. To this end, we extracted the regions corresponding to the 10th percentile and 90th percentiles for each of the cells (Fig. 3c, d, e, f). This is a measure of the three-dimensional height distribution: highly curved actin structures will display a larger difference between the percentiles than at actin structures. For untreated cells, the 10th percentile is at 90 ± 10 nm, while the 90th percentile is at 147 ± 11 nm. This leads to a rise of the actin arches of 56 ± 7 nm and a relative rise of 38 ± 4 %. Blebbistatin shows no signi cant difference compared to untreated cells. The Rho Activator treated cells have a signi cantly higher 10th percentile at 106 ± 12 nm and 157 ± 9 nm for the 90th percentile. Its actin arcs increase with 50 ± 6 nm and have a relative rise of 32 ± 5 %. Y27632 treated cells, on the other hand, have signi cantly lower heights in the 10th percentile at 106 ± 12 nm, but are not different compared to Blebbistatin treated and untreated cells in the 90th percentile with 146 ± 15 nm. However, they show differences in the rise of actin arches of 59 ± 7 nm and have a relative rise of 42 ± 5%.

Actin border analysis
Cell deformations must be precisely controlled as they are key to cell migration, differentiation, division and tissue morphogenesis. These parameters play a signi cant role in carcinogenesis [49,50]. Changes in cell shape are driven by tension gradients in the cellular cortex [51]. This consists of a thin actomyosin network formed by cortical actin and stress bers in addition to actin mesh [49][50][51]. To gain deeper insight into this important outer edge of the cellular cytoskeleton, the height behaviour at the boundary of these actin structures is analyzed.
Due to the high axial resolution, Y27632 treated cells (Fig. 5a) were discovered to have the lowest height (103 ± 15 nm) of all conditions. In contrast, the actin cytoskeleton of Rho Activator treated cells is shifted to higher heights, while the actin edge of untreated (116 ± 13 nm) and Blebbistatin treated cells (114 ± 18 nm) have a similar median.
In addition to analyzing the height of the actin and the in uence of the drugs on it, the derivative of the edge height along the edge (Fig. 5b) was also analyzed. For further analysis of the steepness distribution of the actin edge, we analyzed the slope of the edge according to its gradient angle de ned by where dh is the change in height of a lateral distance dx. Blebbistatin (3.0 ± 0.5°) and Y27632 (3.0 ± 0.8°) treated cells show a similar overall median and signi cantly lower gradient angles compared to untreated and Rho activator treated cells, respectively. There is no signi cant difference between untreated and Rho Activator treated cells. However, untreated cells have a broader distribution and higher variation compared to cells treated with the Rho Activator. Blebbistatin cells display two centres of high density. Both populations show similar angles. Lower angles are also more likely to be found at lower heights. The Rho Activator cells (Fig. 5d), on the other hand, are clearly shifted to the higher heights. The highest density occurs in the range from 110 to 150 nm. They show more evenly distributed angles than untreated cells. Table 1 provides a summary (median ± std) of the analysis of the actin cytoskeleton in z dimension using MIET, as well as the gradient angles along the actin edge. Interestingly, the treatment with the respective drugs leads to changes in total cell area. A signi cantly smaller cell area for the Rho Activator (Fig. 6a) cells compared to the other three populations can be seen.
The cells treated with Y27632 have a signi cantly larger actin cell area than untreated and Blebbistatin treated cells. Furthermore, to see how sparse the actin cytoskeleton is and in which ratio the actin is distributed in the cell, the actin area of the bers obtained using the TWOMBLI plugin was normalised to the cell area (Fig. 6b). The Y27632 treated cells have a signi cantly larger cell area with a low actin content. In contrast the Rho Activator treated cells have a small cell area with a high actin content.
Therefore, a deeper understanding of the impact of this pathway on actin organization is of major concern. Previous structural studies predominantly analyzed the actin regulation and the interaction with the ROCK pathway in the two dimensions of the focal plane [52,53]. Mostly due to technical di culties, the third dimension normal to the focal plane has been examined to a far lesser extent. The threedimensional structure of actin is important for a thorough understanding of cellular physiology and pathophysiology. For this reason, we analyzed here the third dimension in nanometer resolution using MIET. Previously, Chizhik et al. [34] studied the height distribution of actin stress bers of human mesenchymal stem cells on gold-coated glass at speci c time points. In our study, we examined differentiated embryonic mouse broblasts using bronectin as substrate coating. The average actin height appears to be different in the two studies: while in our study, the 10 percentile lowest structures are in the range of 80 nm -120 nm, the earlier study nds lowest actin heights in the order of 20 nm -40 nm. In contrast to this, the actin heights of embryonic mouse broblasts from Kanchanawong et al. [54] compares well to our measured height results. Hence, the height pro le may be cell type dependent.
Chizhik et al. focus on the time evolution of stress bers, while here, we concentrate on the effect of disturbing the ROCK pathway on actin height distributions. To this end, we just used end-point measurements of cells seeded after 24 h. We nd profound and novel effects of actin regulatory drugs on the height and architecture of the actin structures. In particular, inhibition at an early point in the ROCK pathway leads to a lowering of actin height, while activation leads to an increase in actin height. These differences in height may be caused by re-structuring of the connection between cytoskeleton and extracellular matrix, so called focal adhesions. Integrins expressed on the cell surface bind to the RGD binding motif on bronectin and may exert forces, depending on ROCK intervention. Fibronectin has been shown to dynamically alter its brous structure depending on external forces, revealing cryptic binding sites [55][56][57]. This may lead to force dependent, integrin-mediated re-organization of the cytoskeleton and to recruitment of further proteins to the focal adhesions[58,59] above a certain force threshold [60].
Recruiting more proteins to re-enforce the connection would lead to higher actin structures in line with our results here. Further studies are necessary to understand this mechanism.
The height resolution offered by MIET allows us not only to determine the height of the structures, but also to calculate inclination angles along actin borders. Chizhik at al. nd very low inclination angles of actin stress bers on the order of 0.15° distinct from our values of around 3°. However, we measure different structures: while the previous paper investigates speci c pre-chosen stress bers, we focus on the edge of actin. This edge is comprised of different actin structures like cortical actin and stress bers.
Here, it is interesting to observe mainly an effect of inhibitors: these lead to lower angles, while activators hardly change this angle compared to untreated cells.
In order to understand the behaviour of the actin cytoskeleton in the ROCK pathway (Fig. 1a) in the focal plane as well as in the z-dimension, we adopt a simple model previously described [61,62]. It relates the balance between cytoskeleton tension and adhesion tension. Cytoskeletal tension pulls the cell upwards and rounds it up. Balancing this force is the adhesion tension, which tends to stretch the cell. Figure 7 shows a possible model for the relationship between adhesion tension and cytoskeleton tension to describe the analyzed height behaviour in relation to the ROCK pathway.
Untreated cells form arch like actin bre structures with a ru ed actin edge. When Blebbistatin is added after formation of the actin stress bers, only small changes can be observed. However, the ru ed actin edge becomes attened, showing that the ru ing requires active force generation by the cell, since treatment of the cells with Blebbistatin directly inhibits myosin II activity.
As opposed to Blebbistatin, Y27632 does not directly inhibit myosin II, which appears later in the ROCK pathway, but rather inhibits ROCK, which is an important key point in the pathway. By inhibiting ROCK, less myosin light chain is phosphorylated and thus less myosin II is available for stress ber formation and maintenance of stress ber tension. In addition, inhibiting ROCK prevents co lin phosphorylation by LIM kinase. Non-phosphorylated co lin depolymerizes existing actin and leads to an active degradation of actin. Therefore, there is barely any cytoskeleton tension in the Y27632 treated cells due to the destruction of actin and the lack of stress bers. The adhesion tension predominates in the previously existing equilibrium. This results in signi cant wide spreading of the cells and lowering of the actin structures. Interestingly, the actin stress bers appear to be so weak, that even the reduced myosin II induced tension is su cient for increased arching of the actin stress bers. The comparison of Blebbistatin and Y27632 clearly shows that although both drugs act similarly, the effect on the actin structure is drastically different.

Cells treated with the Rho Activator show an opposite effect to cells treated with inhibitors. The Rho
Activator increases the activity of RhoA, which in turn increases the activity of ROCK. The myosin light chain is increasingly phosphorylated. In addition, a positive feedback loop inhibits myosin light chain phosphatase. More myosin II is provided for stress ber formation and maintenance of existing stress bers. This results in stronger contractions in the cytoskeleton compared to untreated cells. Furthermore, the increased activity of ROCK also causes the LIM kinase to phosphorylate more co lin. As a result, less actin is depolymerized and existing actin structures are stabilized. Thereby, the actin content is increased to an even greater extent. Due to this stabilization of existing actin structures and the stronger contraction, the cytoskeleton tension increases strongly. This creates an imbalance, causing the actin cytoskeleton to be pulled further up and together leading to rounded shape with a small area and high concentration of actin bers. The reduced migration behaviour described in the literature [17,63] can also be explained by this imbalance.
As shown in this study, the regulation of the ROCK pathway has an effect not only on the actin cytoskeleton at the focal plane but also perpendicular to it. However, cell-to-cell variations in shape and size make an even deeper analysis of the effect of these drugs challenging.
In summary, we show that balancing forces in the actin cytoskeleton is required for a proper threedimensional cytoskeletal architecture. Increasing tension leads to an actin cytoskeleton at large height in cells with a small cross-sectional area, while decreasing cytoskeletal tension also attens the actin cytoskeleton but close to the cell's substrate with increased cell area.
To apply bronectin to the gold surfaces, the gold slides were treated with the crosslinker Dithiobis The cells were treated with different drugs the next day. Drug treatment was 50 µM Blebbistatin (Sigma-Aldrich, Darmstadt, Germany) for 40 min, 10 µM Y-27632 (Sigma-Aldrich, Darmstadt, Germany) for 1 h and 1 µg/µL Rho-Activator II (Cytoskeleton, Denver, CO, USA) for 3 h, respectively. After each drug treatment cells were brie y rinsed 3x in PBS, which was immediately removed and then xed in 4% formaldehyde (Polysciences, USA) for 10 min. ps, repetition rate 40 MHz, wavelength 520 nm), which is coupled into a polarization maintaining single mode ber. For both focusing excitation light and collecting uorescence light a high numerical aperture objective (60x1.2 UPlanSApo, Superapochromat, water immersion, WD = 0.28 mm) was used. The collected uorescence has been splitted by a dichroic beam splitter (zt488/861rpc-UF3, AHF/Chroma, Tübingen, Germany), passing the pinhole and is ltered by an emission short-pass ET750sp-2p8 (AHF/Chroma, Tübingen, Germany) and emission long-pass lter BLP01-594R (AHF/Chroma, Tübingen, Germany) to the hybrid-PMT detector. For data acquisition the multichannel picosecond event timer HydraHarp 400 TCSPC module of PicoQuant (Berlin, Germany) was used. FLIM images of each single cell were recorded with the SymPhoTime 64 software (PicoQuant, Berlin, Germany).

Fluorescence lifetime data evaluation
Fluorescence photons were detected in time-tagged, time-resolved mode, which makes it possible to collect all photons of single pixels and sort them into a histogram according to their arrival time after the last laser pulse. A multi-exponential deconvolution t with a self-recorded IRF was applied to the data sets in the SymPhoTime 64 software (PicoQuant, Berlin, Germany).

Converting lifetime to height values
The MIET GUI programme of the Enderlein group (University of Göttingen) in MATLAB was used to generate the calibration curve to get the height data from the lifetime data. The theory of how to generate the height data from the lifetimes was described in detail previously [32,33,35,46,47]. and PBS (n = 1.33 [68]) as mounting medium. The calibration curve was then used to calculate the height values in a custom-written MATLAB code, which was adapted from the MIET GUI [32,35].

MATLAB height analysis
For further analysis of the calculated heights the MATLAB ver. R2019b software (The MathWorks, Inc, Natick, Massachusetts, USA) was used. In order to determine the edge of the cell, the image was binarized and the largest object was selected. The edge of the largest object corresponds to the actin edge and was calculated with pixel accuracy. To avoid binarization artefacts, the cell was eroded by 2 pixels. Areas of cells larger than the image section were not included in the edge analysis. For the determination of the gradient angle along edge, the cell height along the edge line was derived and the angle was calculated for every edge pixel accordingly. To normalize the delta of the 90th percentile to the 10th percentile of stress ber height, a parameter we call relative stress ber rise, the difference was divided by the 90th percentile.

Extract stress bers
To extract the stress bers from the intensity weighted height images, these images were analyzed using the TWOMBLI plugin [48] in Fiji [69,70]. The resulting masks were applied to the existing datasets in MATLAB.

Quanti cation and statistical analysis
The populations used for the analysis of the entire study are composed of the following number of single cells: Untreated N = 39, Blebbistatin N = 38, Y27632 N = 39, Rho Activator N = 48. Plots in Fig. 2, 3, 6 are generated using MATLAB. Boxplots are generated using PlotsOfData [71]. For statistical analysis Wilcoxon-rank-sum test was performed in MATLAB. ns: p > 0.05, *: p < = 0.05, **: p < = 0.01, ***: p < = 0.001, ****: p < = 0.0001. Figure 1 a. Scheme of a cell with actin cytoskeleton (red) and nucleus (grey) shown in 3D. The further analyzed actin edge consists of different types of actin and is shown in blue. The image section shows the parameter angle along the edge, which describes the differentiation of the edge. b. Schematic of the investigated ROCK Pathway and the in uence of certain drugs on key points in the pathway. RhoA activates the Rho-associated coiled coil-containing protein kinase (ROCK). ROCK catalyzes the phosphorylation of the myosin light chain (MLC) to myosin light chain phosphate (MLC-P) and inhibits myosin light chain phosphatase (positive feedback loop). This increases the binding of myosin II to actin and thus the contractility of the cell increases. ROCK phosphorylates LIM kinase (LIMK). This phosphorylates co lin. The existing actin laments stabilize and increase as actin depolymerization is deactivated. Rho activator enhances the action of RhoA, Y27632 directly inhibits ROCK in the pathway and Blebbistatin inhibits the action of myosin II.