2-Amino-ethoxydiphenyl borate (2-APB, 100 µM), Nifedipine (10 µM), LaCl₃ (100 µM), Cytochalasin D (1 µM), Blebbistatin (20 µM), Nocodazole (1 µM), ML-7 (20 µM), the ROCK inhibitor Y27632 (20 µM), and phalloidin-TRITC were purchased from Sigma-Aldrich. Thapsigargin (TG, 10 µM) was purchased from Abcam, U73122 (10 µM) from MedChemExpress, and fibronectin from Corning. ITGB1 (Beta1) siRNA (Integrin siRNA, 30 nM) was purchased from Thermo Fisher Scientific.

Primary ASM cells were isolated from the tracheas of 6-8-week-old Sprague Dawley rats as described previously (approved by the Ethics Committee of Changzhou University on Studies Ethics, Grant No. NSFC 11532003) (Wang et al., 2016). The cells were maintained in low-glucose Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS, Thermo), 100 μg/ml penicillin, and 100 unit/ml streptomycin at 37°C with 5% CO₂ in a humidified incubator.

To make nuclear localized ERK biosensor, the nuclear-exporting signal (NES) peptide in the construct was replaced with double nuclear-localized signal (NLS) peptide (2x Pro-Lys-Lys-Lys-Arg-Lys-Val). Briefly, ECFP DNA fragment with 2xNLS sequence at the C-terminal was amplified by PCR. Both the original construct and amplified ECFP fragment were digested by NotI and SalI restriction enzymes, followed by ligation of the two digested products together to generate Nuc-ERK FRET construct.

DNA plasmid and siRNA were transfected into ASM cells by using Lipofectamine3000 according to the manual protocol (Invitrogen). Briefly, cells were passaged in medium without antibiotics into 6-well plates the day before transfection. By following the protocol, 2.5 µg biosensor DNA, 4 µl P3000, and 4 µl Lipofectamine3000 were mixed to assemble Lipid-DNA particles for each well before adding it into the cell culture. The medium was changed in 8 h and cell imaging experiments were performed 40–64 h later.

For siRNA co-transfection, when cells reached 60%–80% confluency, siRNA (30 nM) was transfected into the cells with 4 µl Lipofectamine3000 reagent. The medium was changed in 8 h, and after siRNA transfection for 24 h, the ERK FRET biosensor DNA was further transfected into the cells.

The efficiency of ITGB1 siRNA transfection was assessed with mRNA expression using Real-Time Quantitative PCR (qPCR) assay. Total RNA from cultured ASM cells was extracted using the TRI Reagent RNA Isolation Reagent (#T9424, Sigma). Total RNA weighing 500 ng was applied to generate 1st strand cDNA by using the Revert Aid First Strand cDNA Synthesis Kit (#K1622, Thermo, MA). The sequences of associated qPCR primers for rat ITGB1 were derived from the previous report including GAA​TGG​AGT​GAA​TGG​GAC​AGG​AG (ITGB1 forward), CAG​ATG​AAC​TGA​AGG​ACC​ACC​TC (ITGB1 reverse), and the control GAPDH primers AGG​TCG​GTG​TGA​ACG​GAT​TTG (forward) and GGG​GTC​GTT​GAT​GGC​AAC​A (reverse) (Luo et al., 2018). The primers were synthesized from General Biosystems (Anhui, China), and PowerUp SYBR Green Master Mix (#A25742, Applied Biosystems, CA) was used for PCR amplification. The reaction was run in the qRT-PCR system (StepOnePlus, Applied Biosystems) with 1 µL of the cDNA in a 10 µL reaction according to the manufacturer’s instructions. Calibration and normalization were done using the 2^−∆∆CT method, where ∆∆CT = CT (target gene) -CT (reference gene), and CT referred to the PCR cycle number by reaching the defined fluorescence intensity. Fold changes in mRNA expression were calculated based on the resulting CT values from three independent experiments.

After DNA transfection for 24 h, cells were detached using Accutase solution (Sigma), and transferred to collagen I and fibronectin (40 μg/ml) double-coated 6-well BioFlex plates (Flexcell International Corporation, Hillsborough, NC, United States). After which the cells were cultured in the medium containing 1% FBS for 12–16 h and reached 60–80% confluence before the mechanical stretch. Cyclic stretch was applied with a 0.5-Hz sinusoidal curve at 12% elongation by using a Flexcell^® FX-5000^TM Tension System (Flexcell^® International Corporation). The computer program-controlled bioreactor used vacuum and positive air pressure to apply cyclic strain to cells cultured on flexible-bottomed BioFlex plates. After the durations of cyclic stretch, the BioFlex plates with cells were moved from the Flexcell incubator to Zeiss microscopy for FRET imaging. The images of a group of cells were acquired within 10 min for each well.

The processes of FRET imaging and quantification were similar to our recent descriptions (Ouyang et al., 2019; Yao et al., 2020). Briefly, the Zeiss microscopy system (Zeiss Cell Observer) was equipped with the functions of multi-positions, fine auto-focusing, and automatic-switchable dichroic rotator. The scope stage was supplemented with an incubator box (Zeiss) to maintain temperature at 37°C and 5% CO₂ for live cell samples. During FRET image acquisitions through ECFP and FRET (YPet) channels, the parameters for excitation filter and dichroic mirror were 436 ± 10 and 455 nm, respectively, and the emission filters of ECFP and FRET (YPet) channels were 480 ± 20 nm and 535 ± 15 nm, respectively.

To take the FRET images before and after the cyclic stretch, we made an accessory holder with acrylic plate to load the BioFlex plate on the microscopy stage, as shown in Figure 1. Before starting the stretch, 15–20 positions of fluorescent cells were selected per well, and FRET images were acquired as the pre-stretch condition. After the duration (0.5–1 h) of cyclic stretch, the same positions of cells were reloaded with minor adjustments, and the FRET images were taken within 10 min for each well before the next cycle.

FRET quantifications were processed using the Wang Lab (UCSD)-developed software package FluoCell in MATLAB (available on http://github.com/lu6007/fluocell) (Qin et al., 2019). Fluorescence signals from ECFP and FRET (YPet) images were measured after background subtractions, and the ratio of the two channels was calibrated in the pixel-to-pixel manner. Data procession and statistical analysis were done by the software of Graphpad Prism 6, and Excel. The quantified FRET data from a group of cells was expressed in curves (Mean ± S.E.M.), and scattering dots (Mean ± S.D.). *, **, *** and **** indicate p < 0.05, 0.01, 0.001, and 0.0001 from Student’s t-test, which was applied for significant difference analysis. Multiple times of t-test analysis done between the control and one experimental group were carried for variable experimental conditions using Graphpad Prism 6. All described FRET experiments have been repeated independently on different days with similar conclusions, and statistical quantifications were performed based on the data acquired from different time.

To understand how mechanical stimulation from cyclic stretch activates biochemical signals in live cells, we applied FRET biosensor to measure ERK kinase activity in ASM cells. According to the previous study (Takahara et al., 2014), 12% strain deformation with a 0.5-Hz sinusoidal curve was within the physiological scale and was applied on the cells seeding on the elastic membrane by the Flexcell^® Tension System, which could induce alignment of cells perpendicularly to the stretch direction in 7.5 h (Supplementary Figure S1). In regarding the inapplicability of visualizing the FRET signals on the microscope simultaneously during the cyclic stretch, the procedures were carried out in the following process: the FRET images were taken after 0.5 h-stretch followed by another 0.5 h-stretch and imaging, and then continued with one more 1 h-stretch and imaging (depicted in Figure 2A). The imaging was usually done within 10–30 min between the three times of stretch, and images of similar cell positions were acquired each time.

Intracellular ERK in the cytoplasm was activated by the mechanical stretch, as seen by the FRET changes (Figure 2B). FRET quantification from a group of cells showed apparent activity increase with statistical significance (Figure 2C). As control, no apparent FRET change was observed in the similar imaging procedure without stretch stimulation (Supplementary Figure S2). The platelet-derived growth factor (PDGF) regulates ERK activity in ASM cell proliferation and also inflammation in asthmatic condition (Lee et al., 2001; Kardas et al., 2020). The magnitude of ERK FRET change (∼15–20%) from the mechanical stretch was comparable to PDGF-stimulated FRET change in ASM cells (Supplementary Figure S3). In consideration of ERK shuttling between the cell cytoplasm and nucleus, we modified the ERK biosensor to generate nuclear localized version Nuc-ERK FRET. As shown in Figures 2D,E, nuclear ERK was also activated by the mechanical stretch, which indicates the cytoplasmic increase is not due to nuclear ERK transportation. Hence, we used cytoplasmic FRET measurement to represent the whole cells in the following study. The ERK activation reported by FRET signals was verified by increased phosphorylation level of ERK with phospho-ERK antibody immunoblotting under the cyclic stretch (Supplementary Figure S4). The 12% strain deformation of ASM cells is within the physiological range of stretch magnitude (Takahara et al., 2014), which is also seen by the maintained cell shape (Figure 2B) and actin stress fibers (Supplementary Figure S5) after stretch. We further pretreated the cells with ERK inhibitor PD98059, which abolished stretch-induced FRET signals (Figures 2F,G), hence the FRET changes truly reported ERK activity.

We then investigated the mechanosensitive components in transducing the mechanical stimulation into biochemical signals in cells. Ca²⁺ channels located on plasma membrane or endoplasmic reticulum (ER) are sensitive to mechanical stimulations (Kim et al., 2015). By switching the culture medium to Ca²⁺-free one before cyclic stretch, ERK activation was inhibited with a delayed response (Figures 3A,B). This suggests that Ca²⁺ channels on the plasma membrane had significant contributive role while not essential to the mechanical activation of ERK. However, by selective chemical inhibition of IP₃R Ca²⁺ channel with 2-APB, or SERCA pump with Thapsigargin on ER membrane (Wang et al., 2021), the mechanical activation of ERK was almost blocked (Figures 3C,D), indicating that Ca²⁺ release from the ER store had an essential role. In contrast, there was little effect on the ERK activation by inhibition of L-type Ca²⁺ channel with nifedipine, or store-operated Ca²⁺ channel (SOC) with LaCl₃ on plasma membrane (Parekh and Putney, 2005; Wang et al., 2021) (Figures 3E,F). From statistical comparison of the FRET changes (in percentage) after 0.5 h stretch and total 1 h stretch (Figure 3G), inhibition of IP₃R or SERCA pump on ER membrane had the most significant impact on the mechanical activation of ERK, followed by inhibiting extracellular Ca²⁺ uptakes with reduced delayed response, while L-type Ca²⁺ channel and SOC on the plasma membrane were not actively involved into the ERK activation.

Since Ca²⁺ release through IP₃R channel on ER membrane was essential in the stretch activation of ERK (Figure 3), we further looked at whether inositol 1,4,5-triphosphate (IP₃) production was necessary in turning on the IP₃R channel. As a classic pathway, phospholipase C (PLC) is an enzyme that hydrolyzes PIP2 on the plasma membrane to generate IP₃, which results in opening the IP₃R Ca²⁺ channel (Bartlett et al., 2020). In our experiments, inhibition of PLC activity with specific inhibitor U73122 had little effect on the cyclic stretch-induced ERK activation in comparison to the control group (Figures 4A–E). This indicates that the PLC-IP₃ signaling pathway was not essential in mechanical stimulation of IP₃R Ca²⁺ channel for ERK activation.

Actin cytoskeleton often serves for mechanical transmission within cells (Wei et al., 2020). We hence checked the role of cytoskeleton in the mechanical activation of ERK by FRET measurements. After loss of actin cytoskeleton integrity with Cytochalasin D (Cyto D) treatment which inhibited actin filament polymerization (Supplementary Figure S5), cyclic stretch-induced ERK activation was almost completely inhibited (Figures 5A,B). Contrarily, inhibition of microtubule cytoskeleton with Nocodazole treatment did not show much effect (Figure 5C). The small GTPase Ras acts upstream of ERK through Ras-Raf-MEK-ERK signaling (Rudzka et al., 2021). The stretch-induced ERK activation was suppressed by pretreatment with Ras inhibitor Salirasib (Figure 5D), supporting that mechanically induced Ca²⁺ signal activated ERK through Ras pathway.

Integrins are mechanosensitive molecules on plasma membrane (Sun et al., 2016; Potla et al., 2020). Integrin β1 (ITGB1) is a predominant subunit in forming functional heterodimers with integrin α1, α2…α11, and α_v subunits (Cai et al., 2021), so we tried to regulate the representative β1 to check its role in the cyclic stretch-induced activation of ERK. After knockdown of ITGB1 expression with siRNA in cells, there was no obvious impact on the cyclic stretch-induced ERK activation in comparison to the control groups (Figures 6A–C). The reduced mRNA expression of ITGB1 was confirmed by qPCR measurements (Figure 6D). Statistical quantifications confirmed no significant change in the mechanical stretch-activated ERK by ITGB1 knockdown (Figure 6E). These data indicate that integrin α_(x)β1 was not the primary mechanosensor in the mechanical activation of ERK.