香京julia种子在线播放

All experiments involving the use of live animals were conducted in accordance with the NIH Guide for Care and Use of Laboratory Animals, and all the protocols were approved by the CULA at the Institute of Mechanics, Chinese Academy of Sciences.

Rat bone marrow-derived stem cells (rMSCs) were isolated from 3- to 4-week-old male Sprague–Dawley rats as described previously (Li et al., 2013). Briefly, the rats were sacrificed by cervical dislocation, and the femurs and tibias were then collected. The marrow was flushed out and blown into single cells by an injector, and the collected cells were added into L-DMEM medium (GE Healthcare Life Sciences, Logan, UT, United States) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA, United States) and 1 ng/ml bFGF (R&D systems, Minneapolis, MN, United States). The cells were then maintained in a humidified incubator with 95% air/5% CO₂, 37°C, by refreshing the medium every 3 days. When grown up to 85–90% confluence, the cells were detached using 0.25% trypsin-EDTA for 1 min and passaged into T-25 flask by 1:3 ratio. This procedure was repeated three or four times to collect rMSCs with ∼95% purity.

For CSK staining, FITC-conjugated phalloidin was from Enzo Life Science (Enzo Biochem, Farmingdale, NY, United States), anti-vimentin (Alexa Flour 647-conjugated) and anti-α-tubulin (Alexa Flour 555-conjugated) rabbit monoclonal antibodies (mAbs) were from Cell Signaling Technology (Danvers, MA, United States). For FAC staining, anti-vinculin rabbit mAbs and donkey-anti-rabbit anti-IgG secondary polyclonal antibodies (DyLight 594-conjugated) were from Abcam (Cambridge, Cambridgeshire, UK). For β1 integrin staining, anti-β1 integrin mouse mAbs was from Santa Cruz (Dallas, TX, United States), and the goat-anti-mouse anti-IgG secondary polyclonal antibodies (Alexa Fluor 594-conjugated) were from Abcam. Hochest 33342 for nucleus stain was from Invitrogen (Thermo Fisher Scientific, Waltham, MA, United States). Goat-anti-rat anti-CD11b, CD34, CD45, or CD90 mAbs were purchased from Santa Cruz (United States) and used as biomarkers for identifying the rMSCs as described previously (Li et al., 2013).

Poly-dimethylsiloxane (PDMS) was used to construct the soft gel onto the glass surface via soft contact lithography technique (Whitesides et al., 2001; Cheng and Guo, 2004). Briefly, Sylgard 184 (Dow Corning, United States) was used, and two components of a “base” and a “curing” agent were mixed in 10:1, 33:1, or 70:1 ratio (v/v). The mixture was then poured uniformly on the top of glass coverslip. After additional degassing for 12 h, the PDMS was cured at 65°C for 3 h and at room temperature (∼25°C) for 12 h. The cured PDMS substrate so obtained was then adhered onto the pre-processed, dustless coverslip (Corning, Corning, NY, United States) and treated with O₂ plasma for 1 min to make the surface of the substrate hydrophilic. In the current work, a planar PDMS substrate was used, with a stiffness of 2.5, 0.56, or 0.005 MPa (Wang, 2011; Yang et al., 2011; Sun et al., 2014), while the glass coverslip for tissue culture was used as control, with a stiffness of ∼70 GPa, only for normalizing those data obtained from PDMS substrates with different stiffness levels.

Like those methods described previously (Li et al., 2010; Zhang et al., 2017), rMSCs were placed on three oriented glass coverslips in the presence or absence of PDMS cushion with different stiffness levels (∼2 mm thickness) on the top. Briefly, the cells were seeded on either sterile glass or stiffness-varied PDMS substrate pre-coated with collagen I at 4 μg/cm² overnight. After 24-h pre-growth for steady cell adherence and spread, the substrate was transferred to a custom-made holder and orientated at 180°, 0°, or 90°, respectively (the angle between the substrate outer normal vector and the gravity vector; Figure 1A) for additional 24- or 72-h culture. To reach a similar confluence with the majority of isolated cells at given durations, the cells were seeded at a density of 1 × 10² or 3 × 10² cells/cm² for an additional 72 or 24 h. The effect of hydrostatic pressure among the three orientations was minimized and negligible due to the delicate protocol (Li et al., 2010; Zhang et al., 2017). Triplicate repeats were conducted in each orientation.

Morphological change and cytoskeletal expression were determined by immuno-cytochemistry. At given durations, cells grown on oriented glass or PDMS were rinsed with phosphate-buffered saline (PBS), fixed into 4% paraformaldehyde within 1 min for 15 min, and washed and permeated with 0.1% Triton 100-X for 10 min at room temperature. The collected cells were incubated in 1% bovine serum albumin/PBS for 60 min at 37°C to block non-specific staining. Filamentous actin, vimentin, and tubulin were stained with a mixture of phalloidin at 5 μg/ml, anti-vimentin mAbs at 1:800, and anti-tubulin mAbs at 1:50 for 60 min at 37°C. The cells were then washed and incubated with Hoechst 33342 for 10 min to stain the nuclei. Images of the stained cells were examined using a laser confocal microscope (LSM 710, Carl Zeiss AG, Oberkochen, Germany) with a 63× oil-immersion objective at a slicing height of 0.65 μm in a stepwise interval of 0.322 μm for three-dimensional (3D) imaging. Triplicate repeats were done in each case.

Several parameters were obtained from these images: 1) Nucleus translocation, defined as the longitudinal distance between the nucleus centroid and the substrate, was determined by Imaris software (Bitplane, Zurich, Switzerland) through the 3D reconstructed images of stained actin and nucleus; 2) CSK expression was quantified as the mean relative fluorescence intensity (arbitrary unit or AU) of stained actin, vimentin, or tubulin using ImageJ software (National Institutes of Health, Bethesda, MD, United States). To compare these values in distinct orientations in repeated experiments, a calibration curve was built at systematically varied laser power and PMT gain for the same fluorescent probes. To further test the expression difference between the two types of substrates, a ratio of each value onto PDMS to that onto glass was calculated in all the cases. Perinuclear actin stress fibers and vimentin cords, defined previously (Zhang et al., 2017), were used for quantitative analysis; 3) Anisotropy of perinuclear actin, defined in a recent protocol (Boudaoud et al., 2014), was also adopted to quantify the behavior of this type of fibrillar structures. Here a residual eigenvalue, q, calculated from the pixel intensity array of a given region of interest, denotes a completely isotropic fibril structure when q = 0 or a completely anisotropic fibril structure when q = 1. Noticing that the fibrillar isotropy and anisotropy respectively represent the randomized and aligned actin network, this eigenvalue was estimated in the region of the nucleus contour; 4) Cell morphological analysis was simply conducted on the cell contour identified by stained actin. Cell projected area, circularity (= 4πA/perimeter²), and aspect ratio (= long-axis length/short-axis length) was determined using ImageJ.

FAC immunostaining was similar to the procedure detailed above for cytoskeleton staining. Briefly, after washing and fixing, the cells were co-incubated with phalloidin at 5 μg/ml and anti-vinculin mAbs at 1:200 for 60 min at 37°C, rinsed and incubated with secondary antibodies at 1:200 for 60 min at 37°C, and finally incubated with Hoechst 33342 for 10 min. FACs were visualized using confocal microscopy by collecting 0.65-μm-thickness information at the focal plane. The number of total FACs was counted for ∼45 cells in each case, and the area of total FACs was calculated, respectively, using Matlab software. To exclude the potential impact of a different cell size, the resulting FAC number and area for cells were normalized per 1,000 μm² cell area. Similar protocol was applied for β1 integrin immunostaining.

To elucidate the potential mechanotransductive pathways, rMSCs were incubated with anti-β1 integrin, blocking mAbs at 1.5 μg/ml for 1 h per 24 h (Anguiano et al., 2017) in a total of 72 h of culture (Abcam, Cambridge, United Kingdom) or with 50 ng/ml F-actin depolymizer cytochalasin D (Sigma-Aldrich) for 24 h.

The cultured rMSCs at 72 h were collected on various stiffness levels and orientations. Their total RNA was harvested using RNA extraction kit (Tiangen, Beijing, China) with an in-column DNase digestion step, according to the instructions of the manufacturer. The corresponding cDNA was generated using ReverTra Ace-a (Toyobo, Osaka, Japan) with 1 μg of RNA per reaction in a total volume of 20 μl. A reverse transcriptase-polymerase chain reaction was carried out using GoTaq^® qPCR Master Mix with a two-step method as per the user manual (Promega, Madison, WI, United States) and then measured by a quantitative real-time amplification system (QuantStudio 7, Thermo Fisher). The optimized primers for PCR tests are summarized in Supplementary Table S1.

Three-way ANOVA was performed to test the statistical significance of differences among the three factors of stiffness, orientation, and duration. Two-way ANOVA test, followed by Holm–Sidak test, was used to test the statistical significance of differences in the measured parameters between the two factors of orientation and duration or of orientation and stiffness. For comparisons between any two groups, Student’s t-test or Mann–Whitney test was also performed upon data passing the normality test or not.

Different cell types present diverse phenotypes together with varied ECM stiffness in a pericellular microenvironment. Here we first tested nucleus longitudinal translocation in rMSCs onto typical PDMS substrates when glass coverslip served as the control (Figure 1A). The orientation dependence of nucleus translocation for rMSCs onto glass was presented with a higher value at 0° for 24 h (Figure 1B), consistent with that previously described for MC3T3-E1 cells (Zhang et al., 2017). When rMSCs were placed onto 2.5 MPa PDMS, nucleus translocation was relatively higher at 90° for 24 or 72 h (Figure 1B). These data indicated that the orientation dependence of nuclear translocation likely appeared for rMSCs on 2.5 MPa PDMS in a time-dependent manner. We further tested the combined impacts of substrate stiffness and orientation and found that both factors presented a significant difference in regulating nucleus longitudinal translocation at 2.5 MPa PDMS (symbol # in Figure 1B). Specifically, the translocation onto 2.5 MPa PDMS was lower at 0° for 24 h but higher at 90° for 24 and 72 h compared with those onto glass, potentiating the different mechanisms of substrate stiffness in regulating rMSC stability onto an oriented substrate.

It is well known that PDMS and glass substrates have distinctive properties in both elasticity and chemistry. Thus, we added two more PDMS stiffness levels to further test their mechanical remodeling under gravity vector. On either 0.56 or 0.005 MPa PDMS (Figure 1B), the rMSCs presented no orientation dependence of their nucleus translocation for 24 or 72 h, which is different from the above-mentioned observations onto 2.5 MPa PDMS. This was further verified by testing the statistical differences for combined effects of stiffness and orientation. No significant distinctive translocation in three orientations was observed on these softer PDMS (^# p ≥ 0.408) even with slightly lower values at 0° for 24 or 72 h. Taken together, these results indicated that the orientation dependence of nuclear translocation is also stiffness-dependent and only presented onto stiffer PDMS substrate.

It was indicated previously that the orientation dependence of nuclear translocation is mainly relying on mechanical pathways via cytoskeleton remodeling and focal adhesion reorganization (Zhang et al., 2017). Next, we tested how CSK remodeling is associated with differential nucleus longitudinal translocation onto 2.5 MPa PDMS. Compared with high actin expression (Figure 2A) but low vimentin expression (insert in Figure 2A) onto glass, these CSK protein expressions seemed reversed onto 2.5 MPa PDMS, that is, with low actin expression but high vimentin expression at 24 h. Quantitative analyses further supported these observations. Indeed actin expression normalized to the one onto glass was ∼50% reduced onto 2.5 MPa PDMS in all cases, except of the one at 90° for 72 h (Figure 2B and insert for vimentin), where it yielded ∼1.5-fold higher onto 2.5 MPa PDMS (Figure 2C). Intriguingly, the exceptional difference in extremely high actin expression was positively related to the higher nucleus translocation at 90° (dotted bars in Figure 1B). By contrast, vimentin expression is ultimately reversed onto PDMS, that is, 3.0–5.0-fold higher for 24 h or 1.2–2.5-fold higher for 72 h in all cases (Figure 2D). These data indicated that placing rMSCs onto 2.5 MPa PDMS reduced the actin expression but fostered the vimentin expression as compared with those onto glass, implying that the mechanical stability of rMSCs could be achieved relying more on vimentin and less on actin onto 2.5 MPa PDMS or more on actin and less on vimentin onto stiff glass. More importantly, the differential distributions of actin and vimentin onto 2.5 MPa PDMS and glass were consistent with those above-mentioned orientation dependences of nucleus longitudinal translocation. Compared with those onto glass, the low nucleus translocation onto 2.5 MPa PDMS at 0° for 24 h was associated with a relatively low actin and high vimentin expression, suggesting that high vimentin expression at 0° stabilizes the location of the nucleus when actin is reduced significantly by a short duration. By contrast, high nucleus translocation at 90° for 72 h was associated with high actin and high vimentin expressions, indicating that long-duration maintenance at 90° needs more cytoskeletons to resist the nucleus translocation or hold the nucleus steadily (also seen in those perinuclear vimentin distributions below). Additionally, the normalized tubulin expression fluctuated around unity (0.7–1.3) in all cases (Supplementary Figure S1), implying that tubulin contributes much less to this differential mechanism between the two types of substrates. Collectively, the orientation difference of nucleus translocation and global CSK presentation exists onto 2.5 MPa PDMS, with high vimentin expression at 0° for 24 h or high actin and vimentin expressions at 90° for 72 h. Meanwhile, those absolute fluorescence intensities of cytoskeletal protein expression also supported the above-mentioned observations using normalized ones. Onto 2.5 MPa PDMS, low actin, high vimentin, and low tubulin expressions were observed in three orientations for 24 or 72 h, even with a few exceptional cases of high actin expression at 90°, indifferent vimentin expression at 0°, and reversely high tubulin expression at 180° or 0° (Supplementary Figure S2).

We further tested global CSK expressions on softer PDMS substrates. On 0.56 and 0.005 MPa PDMS, CSK protein expressions presented different orientation-dependent patterns from the one on 2.5 MPa (Figures 2E,F and inserts). On 0.56 MPa PDMS, the cells yielded a high actin expression at 0° for 24 or 72 h, all of which were comparable with those on glass except of one case at 180° for 24 h (Figure 2C). By contrast, vimentin expression was relatively high at 90° for 24 or 72 h (Figure 2D) with 3.1–3.8-fold enhancement than those onto glass, consistent with the high nuclear translocation at 90° at long duration (cf. Figure 1B) and also implying a compensatory role of vimentin to support nucleus stability. On 0.005 MPa PDMS, the cells presented a relatively low actin expression, especially at 0° for 24 or 72 h, compared to those onto glass (Figure 2C). Vimentin expression was still lowered for 24 h but enhanced for 72 h especially at 90° (Figure 2D).

Finally, we tested typical mechanosensitive gene expressions at 72 h. Data indicated that the expression of actin, vimentin, or α-tubulin was indifferent on varied substrate stiffness and orientation, implying that the cytoskeleton is favored to maintain their gene level in a conservative way on the current settings (Supplementary Figure S3).

Not only the global presentation of actin within entire cell but also its localized distribution at the vicinity of the nucleus is crucial in manifesting CSK remodeling and maintaining nucleus stability. We further compared the distribution of perinuclear actin stress fibers onto glass or PDMS substrate. The fibers are likely uniformly aligned with high intensity onto glass (Figure 3A) but randomly oriented with low intensity onto 2.5 MPa (Figure 3B) or even unmeasurable onto 0.56 (Figure 3C) and 0.005 MPa (Figure 3D) PDMS, presenting a significant difference in perinuclear fibers between the two types of substrates. This observation was confirmed by the quantitative analyses that the number of perinuclear fibers was higher onto PDMS than those onto glass for 72 h (Figure 3E). At this duration, the fiber number onto 2.5 MPa PDMS was extremely higher at 90°, further supporting the consistency between global actin expression and nucleus translocation onto PDMS. We also compared the anisotropy of perinuclear fibers on all three PDMS substrates using the residual eigenvalue described previously (Boudaoud et al., 2014). It yielded lower values for cells onto PDMS (∼0.05–0.15) than those onto glass (∼0.15–0.25) in three orientations at two durations (Figure 3F), consistent with the above-mentioned observations of randomized fibers onto PDMS and aligned fibers onto glass from confocal images (Figures 3A,B). We also counted the number of cells with branched perinuclear actin network from the total cells observed. The fractioned number was again increased with time onto either substrate and yielded higher values onto 2.5 MPa PDMS than those onto glass (Table 1), further confirming the occurrence of high randomization or low anisotropy of the actin network for the cells onto PDMS. Similar to those orientation dependences of global actin expression (Figure 2C) and perinuclear fiber presentation (Figure 3E), the number of actin fibers was higher at 90° than that at 180° or 0° onto the same PDMS substrate for 72 h. These results implied that the relatively branched and stronger fibers are favorable in this orientation to maintain the nucleus stability. Taken together, mechanical support for rMSC stability is mostly attributed to the perinuclear stress fibers with a large number, low anisotropy, and low intensity onto PDMS as compared to those with a small number, high anisotropy, and high intensity onto glass. Onto a softer PDMS substrate at 0.56 MPa, a large number of tiny actin filaments (Figure 3C) tended to take over those stress fibers across the nucleus that appeared onto glass, with a relatively high expression and anisotropy but low mechanical support to nucleus translocation (Figure 3F). This substrate seemingly served as a transition one between the 2.5 and 0.005 MPa PDMS substrates since there was no visible filament with quite low anisotropy values onto 0.005 MPa PDMS (Figure 3F).

We also compared the perinuclear vimentin cords onto the two types of substrates. Onto 2.5 MPa PDMS (those images at 0.56 and 0.005 MPa were not able to be reconstructed for quantification), the cords were formed surrounding the nucleus (Figure 4A). The number of vimentin cords yielded higher values at 90° for 24 or 72 h (Figure 4B), partially supporting the orientation dependence of nucleus translocation and global vimentin expression at 90° for 72 h. By contrast, few vimentin cords were visible onto glass (Figure 4B). As indicated, the distribution of vimentin cord number was narrowed down to 0 or 1 onto glass but centered around 2 or 3 onto PDMS (Supplementary Figure S4). Moreover, vimentin onto PDMS tended to be dispersedly distributed from the nucleus to the cell edge, specifically extending into those lamellipodia (Figure 4A) where the actin presentation is low (cf., Figure 2B). These results implied that the vimentin network also provided structural bases for supporting the cell as actin does.

Focal adhesion complex is required to anchor the cell onto the substrate mechanically. Thus, we compared the FAC reorganization onto glass (Figure 5A) or 2.5 (Figure 5B), 0.56 (Figure 5C), or 0.005 (Figure 5D) MPa PDMS in three orientations at two durations. Global differences were again found in normalized number (Figure 5E) and area (Figure 5F) of total FACs between the two types of substrates, presenting lower values onto PDMS (especially on softer substrates with 0.56 and 0.005 MPa) than those onto glass. Meanwhile, both the values of FAC number and area were higher at 0° onto glass for 24 h, consistent with the orientation dependence found for MC3T3-E1 cells (Zhang et al., 2017). By contrast, the value was indifferent onto three PDMS substrates in three orientations at two durations, implying that FAC formation happens in a stiffness-dependent but orientation- and time-insensitive manner onto these relatively soft substrates. This should not be surprising since the mechanical strength of the existing FACs is sufficiently enough to stabilize the cells (Zhang et al., 2017).

Lastly, we compared the morphological alterations of rMSCs since the cell is stabilized onto the substrate via reorganized FACs. Here the cell contour was identified by actin staining images (Supplementary Figure S5), and the projected area, circularity, and aspect ratio were then determined on three PDMS and glass substrates. Global differences of the three parameters were observed between the two types of substrates or the two durations (Figure 6). At 24 h, there was a striking unanimity in all three orientations, that is, the area gradually increased with decreased stiffness, the circularity onto PDMS was slightly higher at 0° but comparable at 180° or 90°, and the aspect ratio on PDMS was significantly lower in all three orientations. These data indicated that the cells tended to become less long and narrow onto PDMS even of a similar size with those onto glass. At 72 h, both the area and circularity were relatively higher, but the aspect ratio was lower, onto 2.5 MPa PDMS; while the area gradually decreased, the circularity was higher, but the aspect ratio was maintained onto 0.56 and 0.005 MPa PDMS, suggesting that the cells became large in size with a spherical shape at 2.5 MPa but small in size with a spherical shape at 0.56 and 0.005 MPa. It was also noted that the decrease of the area at 72 h onto 0.56 and 0.005 MPa PDMS was correlated to the high vimentin increase (cf. Figures 2D, 4B). Additionally, the area and circularity were increased, but the aspect ratio was decreased with time for the cells onto PDMS, similar to the time-dependent morphological alterations onto glass except of the case of time-independent cell circularity. Finally, stiffness dependence and orientation independence were found in cell morphology, consistent with our previous observations for MC3T3-E1 cells (Zhang et al., 2017).

The above-mentioned results indicated that optimizing substrate mechanics and orientation represents a critical step for maintaining the efficient longitudinal translocation of cell nucleus. To elucidate the mechanotransductive pathways involved in affecting rMSC nucleus longitudinal translocation, rMSCs grown typically at 90° on 2.5 MPa PDMS for 72 h were incubated with blocking mAbs against β1 integrin or F-actin de-polymerizer cytochalasin D. The results indicated that β1 integrin expression was significantly decreased, and the fluorescence dots originally existing between rMSCs and the substrate in normal control (Figure 7A) disappeared after blocking (Figure 7B), which is positively correlated with the reduced nucleus longitudinal translocation (Figure 7D). Meanwhile, vinculin expression was lowered (Figure 7C), the number (Figure 7E) and area (Figure 7F) of total FACs were significantly reduced, and actin stress fibers became smaller and thinner (lower panels in Figures 7A–C). Similar observations were found by treating F-actin with cytochalasin D, resulting in remarkably reduced nucleus longitudinal translocation (Figures 7G,H).

To further test if the orientation effect is reversible, the rMSCs originally placed at 90° for 72 h were re-placed horizontally for an additional +24 or +72 h. The results indicated that the nucleus longitudinal translocation was still visualized at +24 h but significantly lower at +72 h, implying that, in addition to substrate stiffness, substrate orientation is also involved in maintaining the nuclear longitudinal translocation of rMSCs and that this orientation effect seemed to be reversible (Figure 7I). Taken together, these results suggested that the typical integrin–FACs–actin mechanotransductive axis plays a key role in regulating rMSC nucleus longitudinal translocation in response to substrate stiffness and orientation (Figure 7J).