The transformed and metastasizing BjhtertSV40T-HRasV12 (BJ-Ras) cells were previously constructed by Hahn et al. by inserting three well-defined genetic elements: the telomerase catalytic subunit (hTERT) in combination with two oncogenes (the simian virus 40 large-T oncoprotein, and an oncogenic allele of H-ras) into neonatal human dermal fibroblast (Hahn et al., 1999). The cells were cultured as described previously (Gad et al., 2004). U87 cells were cultured in MEM containing 10% FBS as well as 50 U/ml penicillin and 50 μg/ml streptomycin in 10 cm dishes (Sigma). U118 cells were cultured in DMEM containing 10% FBS as well as 50 U/ml penicillin and 50 μg/ml streptomycin in 10 cm dishes. Transfection was performed by using Lipofectamine 2,000 (Invitrogen) according to the manufacturer’s instructions. For generation of stable Cas9-expressing cell lines, cells were transfected with the Cas9 vector. Stable transfectants were obtained by blasticidin selection. Then, cells were transfected with sgRNA vector followed by doxycycline treatment for 12 h. After 24 , the cells were selected for transduced cells using puromycin. Several single clones were chosen as the vimentin knockout cell lines which were confirmed by PCR and western blotting. Flag-tagged expression plasmids were constructed based on pCMV-Tag 2B (Stratagene). Full-length mouse vimentin cDNA was amplified from C57BL/6J mouse cDNA by RT-PCR and inserted into the vector by BamH1/Sal1, respectively. The vector was confirmed by sequencing. Amino acids were substituted in the vimentin (R273H; Y276F) by mutagenesis using the MutanBEST kit (TaKaRa) and confirmed by DNA sequencing. For FRAP analysis, the BJ-Ras cells were transiently transfected with a plasmid coding for vimentin-EGFP using jetOPTIMUS transfection reagent (Polyplus-transfection) following the manufacturer’s instruction 24 h prior to analysis (Rathje et al., 2014).

For the human codon-optimized Cas9 expression, the doxycycline-inducible plasmid PB-TRE-NLS-linker-Cas9-ZF-IRES-hrGFP-Blasticidin was utilized. Two guide strands to target two segments in the human vimentin gene were designed using the CRISPR Design tool (http://crispr.mit.edu/) and ligated into the pGL3-U6-2sgRNA-ccdB-EF1a-Puromycin plasmid using a BsmBI restriction site.

Vimentin knock down in BJ-Ras cells was performed with vimentin -targeting ON-TARGET siRNA-SMART pool (L-003551-00-0005, Dharmacon) or AllStars Negative Control siRNA (1027280, Qiagen) as control, using INTERFERin transfection reagents (101000028, Polyplus) according to the manufacturer’s protocol.

The vimentin-binding compound used in this study was ALD-R491, (E)-1-(4-fluorophenyl)-3-(4-(4-(morpholine-1-yl)-6-styryl-1, 3, 5-triazinyl-2-amino) phenyl) urea. It has a molecular formula of C28H26FN7O2 and a purity of >98%. The compound was synthesized at Bellen Chemistry Co., Ltd. (Beijing, China), analyzed at Porton Pharma Solutions, Ltd. (Chongqing, China), and provided by Luoda Biosciences, Inc. (Chuzhou, China), and Aluda Pharmaceuticals, Inc. (CA, United States). The cells were treated with 5 μM R491 or DMSO for 4 h, if not stated otherwise in the figure legend. For each experiment, the control used contained DMSO at an equal concentration to the amount of R491 used.

Live cell imaging was performed essentially as described previously (Evans et al., 2022). BJ-Ras cells were seeded at 10³ cells/well in the glass bottom 10 well slide chamber (Greiner Bio-One, 543079), and allowed to attach, spread and grow for 2 days. The cells were then treated with R491 or DMSO for 3 h, in combination with 0.05 μg/ml Hoechst to stain the nuclei. This was followed by live cell imaging, with images taken every 45 min for 17–23 h, using a 10 × magnification on a Zeiss Celldiscoverer7 Widefield Fluorescence Microscope (Carl Zeiss Microscopy, Thornwood, NY). For live cell imaging of siRNA transfected cells, the cells were grown for 3 days after the siRNA transfection to achieve a good KD efficiency, prior to the drug treatment and imaging. Analysis of 12 movies were performed for both the control and for the R491-treated cells for both live cell imaging experiments.

The cells were fixed and permeabilized in PBS containing 3.7% formaldehyde, 0.1% glutaraldehyde and 0.2% Triton-X100, and the immunofluorescent staining was performed as described previously (Gad et al., 2012a). The antibodies and an actin probe used are as follows: anti-mouse-vimentin, clone V9 (V6389, Sigma Aldrich), anti-rat-vimentin (MAB2105, R&D systems), anti-mouse-p-Tyr (sc-7020, Santa Cruz), anti-rabbit-β-tubulin (ab6046, Abcam), Alexa-Fluor-488 phalloidin (Invitrogen), Goat anti-mouse-Alexa-Fluor 555 (Invitrogen), Goat anti-rabbit-Alexa-Fluor 647 (Invitrogen) and Goat-anti-rat-Alexa-Fluor 647 (Invitrogen). The cells were then imaged with ZEISS LSM 980.

For quantification of vimentin filament distribution, the immunofluorescent images of control (n = 23) and treated cells (n = 20) were converted to greyscale using FIJI (v.1.52 g) and then analyzed with the FiNTA software (https://beta.finta-research.com/home, https://github.com/SRaent/FiNTA) (Flormann et al., 2021). We set the fiber diameter in the FiNTA software to 2 and 8 pixels for the control and treated cells, respectively, while all other settings were left as default. We chose these numbers because at these settings the FiNTA program traces the organization of vimentin filaments more faithfully than when we use the same diameter settings for control and drug treated cells. The loop diameters and circumferences were generated using the images of fiber network traces. The average fiber loop circumferences were then plotted as a scatter plot, each point representing a single cell. For quantification of vimentin filament thickness, immunofluorescent images of vimentin were converted to greyscale using FIJI (v.1.52 g). The fiber diameter was then quantified in the greyscale images, using the ‘graphical diameter calculation’ tool in the FiNTA software. In total, 15 control and 15 R491-treated cells were measured (5 from each biological repeat). After identifying the leading edge of the cell, 3 filaments per cell in the region between the leading edge and the nucleus were measured and average diameter per cell was calculated. To reduce bias, the filaments were chosen along a straight line in the region of interest. We could not measure the filament diameter in a totally blinded manner due to the apparent visual difference in the filament thickness between the control and R491-treated cells. For quantification of focal adhesion sizes, immunofluorescent phospho-tyrosine images were converted into greyscale and focal adhesions were highlighted using threshold tool in FIJI (v.1.52 g). Subsequently, the leading edge of the cells were selected and average focal adhesion size per cell was calculated (n = 7 each for control and R491-treated). Average focal adhesion sizes were compared using an unpaired t-test. The individual focal adhesion sizes were transferred to a prism file and the focal adhesion sizes were grouped into bins (bin width 1 μm²), and a frequency histogram was generated. Thereafter, we used a chi-square test to determine the statistical significance in the correlation between drug-treatment and focal adhesion size distributions.

The Fluorescence recovery after photobleaching imaging and analysis were performed as previously described in Warrington et al., 2017 (Warrington et al., 2017). In brief, samples were imaged on an inverted Nikon A1 confocal microscope, with a Nikon 60 × 1.4 NA oil apochromatic objective lens. The images were taken at 1 frame per second, 30 gain, offset 0 with a pixel size of 80 nm, achieved using 5 x zoom, 512 × 512 pixels and with the pinhole at 1.2. For imaging a 488 nm diode laser was used to image at an output of 0.4%, with a band pass 505–550 nm filter. For bleaching the same laser was used at 30% and passed 2 times over a region of interest (ROI) −2 μm² in size at a rate of 8 frames/second, with no image averaging. After bleaching, the fluorescence was 60%–80% of the initial fluorescence. Regions of vimentin near the nucleus were selected for analysis, we avoided regions near the cell periphery which often moved out of focus, as shown in Figure 2. For data acquisition, three pre-bleach images were captured, the area was bleached, followed by image-acquisition every 10 s. Data in which vimentin filaments moved out of focus were discarded.

The soluble and insoluble fractions of proteins were extracted as described previously (Rathje et al., 2014). In brief, to separate the soluble cytosolic fraction of the cells from the non-soluble filamentous fraction, the soluble fraction of the cells was extracted and the 2 cell fractions were analyzed using western blotting (Rathje et al., 2014). Similar extraction procedures have previously been shown to result in the extraction of vimentin tetramers (Soellner et al., 1985).

Proteins were separated using precast polyacrylamide gels (4%–15%) (4568084, Bio-Rad Laboratories Ltd.), and western blot procedures were performed as described previously (Evans et al., 2022). Immunoprecipitation was performed as previously described (Qu et al., 2016). The list of antibodies used is as follows: anti-mouse-vimentin (V6389, Sigma Aldrich), anti-rabbit-β-tubulin (ab6046, Abcam), anti-mouse-β-actin (A-1978, Sigma Aldrich) and anti-mouse-FLAG (Sigma Aldrich, F3165). HRP-conjugated goat anti-mouse/rabbit secondary antibodies (GtxRb-004-DHRPX/GtxRb-003-DHRPX, Immunoreagents). For vimentin monomers/tetramers detection, 20–50 μg of total protein extracts were loaded on the gel. When detecting the vimentin tetramer, β-Mercaptoethanol was not added into the sample buffer. The NIH ImageJ software (version 2.3.0/1.53f) was used for the quantification of the bands from the western blots.

Cell shape was quantified using FIJI (v.1.52 g), a scale bar was used to calibrate images at 2, 7, 12, and 17 h where all cells in the frame with a clear border were traced manually with the freehand draw tool allowing measurement of area and shape descriptors; circularity and aspect ratio. For cell shape analysis, the number of cells (indicated as treated/control) were as follows, 2 h (219/284), 7 h (195/285), 12 h (176/256) and 17 h (176/223). Migration speed and persistence of DMSO (n = 394) and R491-treated cells (n = 318) was tracked using the FIJI (v.1.53c) TrackMate plugin, by calibrating to 1.1:1 pixels: µm, setting the frame interval to 45 min and using these settings: Difference of Gaussian (Dog); linking max. distance, 200 µm. The estimated object size threshold were adjusted to track the majority of the cells. All other settings remained in the default setting.

Traction force microscopy experiments were performed using collagen coated 12 kPa hydrogels with 0.2 μm Yellow/green fluorospheres Matrigen SoftTrac™ plates (Cellguidance-Cambridge). Plates were equilibrated with cell culture medium for 30 min at 37°C. BJ-Ras fibroblast cells were seeded at 2500 cells/cm² and allowed to adhere for 1 h prior to 4 h drug treatment and imaging. Phase images of attached cells and fluorescent embedded beads were collected using 40 x magnification on a Zeiss Celldiscoverer7 Widefield Fluorescence Microscope (Carl Zeiss Microscopy, Thornwood, NY). Images of the fluorescent microspheres on a relaxed hydrogel were collected after 5 min treatment with a 0.5% (w/v) Triton-X-100, 20 mM NH₄OH PBS solution to remove cells. Treatment with Triton-X100 is an efficient method to detach cells (Franco-Barraza et al., 2016; Teo et al., 2020). Displacement of beads before and after the removal of cells was tracked by particle imaging velocimetry followed by Fourier transform traction cytometry to estimate the corresponding cell traction force field, using a modified ImageJ macro kindly shared by Dr. S. Lee (Lee and Kumar, 2020) from a previously described method (Martiel et al., 2015). The total traction forces (in N) were measured by integrating traction forces over the cell area. The elastic energy (in J) stored in the gel to produce the observed deformation was calculated by summing the products of displacement with the force over the cell area.

Timelapses obtained as described in the Live-cell imaging section above were analyzed with regard to the number of cells which leave their tips behind during cell migration. For this, two assessors analyzed independently 11 control- and 11 R491-treated timelapses, by manual tracing frame-by-frame of 480 control-treated or 454 R491-treated cells. Cells partially out of frame or too crowded together and overlapping were excluded from the analysis.

Statistical analysis of focal adhesion sizes was performed using unpaired t-test in GraphPad Prism (v. 9.4.0). The focal adhesion sizes and frequencies were used to generate a histogram (bin size, 1 μm²). The statistical analysis of western blots was performed using GraphPad Prism (v.6.0). Data were presented as mean ± SD or mean with SEM. Differences between groups were analyzed by Anova and Student’s t-test. p ≤ 0.05 was considered statistically significant. The mean values of the aspect ratio, cell migration speed and persistence, cells from which debris is detached and left behind during migration, filament distribution, as well as Traction force microscopy data were analyzed using GraphPad Prism (v.9.3.1). And statistical differences were assessed using an unpaired two-tailed parametric t-test or a Mann-Whitney U test when the results were non-parametric. Results were presented as mean ± SD, and a p ≤ 0.05 was considered significant. For the Fluorescence recovery after photobleaching analysis, once processed in NIH ImageJ FIJI software (v.1.53f), the data was transferred to GraphPad Prism (v.9), plotted on an X-Y graph and a one-phase exponential association curve fitted. An extra sum-of-squares F test was performed to compare the plateau and the rates for the control and treatment conditions. To do this the best fit values were obtained from the fitted curve for the plateau (Ymax), the rate (K) and the half-life. Then, the plateau (Ymax), the rate (K) parameters were individually compared to a simpler model whose parameters are shared among the data sets compared. The null hypothesis for this test is that the two samples come from the same population and so share their parameters.

The changes in vimentin filament distribution can affect cell migration speed and directionality (Ronnlund et al., 2013; Rathje et al., 2014; Danielsson et al., 2018). Therefore, we aimed to determine how the R491-treatment would affect the distribution of vimentin filaments. To this end, we analyzed the organization of vimentin IFs in BJ-Ras cells, as described in the Materials and Methods section. We observed that in the R491-treated cells, the vimentin filaments were distributed in a more disorganized manner throughout the cytoplasm, as compared to the control (Figure 1). In the control cells, the vimentin filaments radiated out from the perinuclear position to the leading edge of the cells in an organized and more parallel arrangement. In contrast, the vimentin filaments in the R491-treated cells were more disorganized with increased loop circumferences and in thicker filament bundles (Figure 1).

Vimentin filaments are dynamic, with constant exchange of subunits, as previously shown using FRAP analysis ((Yoon et al., 1998), Figure 2). Because R491 binds vimentin at the ²⁷³RxxY motif and inhibits the formation of vimentin tetramers in U87 and U118 cells [(Zhang et al., 2021), Supplementary Figure S1], we hypothesized that the drug would change the vimentin filament dynamics. Using FRAP analysis, we observed that vimentin filaments in BJ-Ras cells treated with R491 at 5 µM or 10 µM for 3 h prior to imaging showed increased stability but showed no change in the rate of fluorescence recovery as compared to the controls (Supplementary Figure S2; Supplementary Table S1). No change was observed with 1 µM concentration of the drug (Supplementary Figure S2). Taken together, this suggests that the drug increases the stability of vimentin filaments in the cell not by changing the rate of delivery of vimentin subunits to the filaments but by altering the filament subunit assembly and/or disassembly dynamics.

In migrating cells, the ways in which different assembly statuses of vimentin are distributed affect the shape, polarity and motility of cells (Helfand et al., 2011). Non-filamentous vimentin is shown to be enriched within the lamellipodia, whereas mature filaments inhibit lamellipodia formation. As R491 reduces the production of tetramers and also the mobile fraction of vimentin (Supplementary Figure S1, Figure 2), we speculated that the assembly and incorporation process of ULFs into the mature filament might be affected by R491 treatment, and we would expect an increase in the amount of soluble vimentin. We therefore isolated and quantified the fold change of the vimentin soluble fraction in BJ-Ras cells after drug treatment. In line with previous reports, the soluble fraction of vimentin in fibroblasts was much lower than that of actin or tubulin [(Rathje et al., 2014), Figure 3]. However, there was an increased soluble fraction of vimentin in R491-treated cells, with an average 3.2-fold increase, as compared to the control. R491-treatment resulted in a slight, but not significant, trend of increased soluble fractions of actin and tubulin (Figure 3).

Cell motility is an essential part of cancer cell metastasis and it depends on the cytoskeleton and cell-ECM adhesion. Since the distribution of vimentin was changed in response to R491-treatment, we analyzed if R491 also changed the adhesion, migration and shape of BJ-Ras cells. At all timepoints analyzed, there was a trend to decreased cell migration speed and directionality, and an increase in cell elongation of the R491-treated cells, as compared to control, with significant differences at the 17 h time point (Figure 4, Supplementary Figure S3). The R491-but not control-treatment resulted in an erratic formation of lamellipodia in all regions at the cell periphery during cell migration, and a non-coordinated movement of the different parts of the cells (Figure 5). The drug also resulted in very thin and long trailing edges of migrating cells which often broke off from the cell body during migration, leaving the rear end of the cell behind and attached to the underlying surface (Figure 6). In multi-cell clusters, the control-treated cells moved past each other in a flat and elongated shape, apparently adapting to the adjacent cells, while the R491-treated cells did not elongate, or adapt their shape. Instead, the drug-treated cells passed by other cells without a clear leading edge, and with flat, lamellipodia-like protrusions forming simultaneously in different directions (Figure 5). To determine if the effect of R491 on cell motility is due to an interaction with vimentin, we analysed the migration of vimentin knockdown cells, in the presence of drug or control. We observed no effect of the drug in these cells, suggesting that the effect of the drug depends on vimentin (Supplementary Figure S4).

Cross talk between vimentin and other cytoskeletal proteins has been observed to regulate the contractile forces of cells (Eriksson et al., 2009; Liu et al., 2015). To determine if R491-treatment had an effect on the contractile forces of single cells, we compared the forces between drug-treated and control BJ-Ras cells. We observed that the R491 increased the total contractile force, stored elastic energy and maximum force of cells, as compared to the control (Figure 7).

To understand the basis for the drug-dependent effect on the contractile forces of cells, we compared the size and numbers of cell-matrix focal adhesions at the leading edges between R491-treated and control-treated BJ-Ras cells. We found that R491-treatment decreased the average size of the focal adhesions. We also observed and there was a significant increase in the number of small focal adhesions in the R491 treated cells when compared to control-treated cells (Figure 8).