The drug resistance feature of acute myeloid leukemia is related to the cell stiffness

Abstract

Acute myeloid leukemia (AML) is a hematologic cancer. Cytarabine-based chemotherapy is the primary treatment. However, drug resistance presents a significant challenge leading to treatment failure. Our study explores the underlying correlation between AML stiffness and its drug resistance feature. We employed microfluidic technology to measure AML cell deformability, demonstrating that drug-resistant cells exhibit increased stiffness compared to their drug-sensitive counterparts. Transcriptomic analysis revealed that enhanced stiffness in drug-resistant cells is associated with upregulated cytoskeletal protein expression and increased lipid metabolism, particularly the peroxisome proliferators-activated receptor (PPAR) signaling pathway. Mechanistically, we found that knocking down PLIN2 at the genetic level and increasing the cholesterol level promoted the deformation of drug-resistant cells, indicating that intracellular lipid levels are involved in the regulation of cell softness. Our findings suggest that AML cell stiffness could serve as a potential biomarker for drug resistance, providing new insights into the mechanisms underlying AML drug resistance and offering potential therapeutic targets.

I. INTRODUCTION

Acute myeloid leukemia (AML) is an incurable hematologic cancer. Cytarabine-based chemotherapy is the first-line treatment for newly diagnosed AML patients. Unfortunately, the emergence of drug resistance in many patients often leads to treatment failure and mortality.¹ Therefore, elucidating the mechanisms driving drug resistance is crucial for improving therapeutic outcomes and potentially saving patients' lives.

Changes in the cellular architecture contribute to the cell phenotype and differential functions.² Recent studies indicate that biophysical properties, including cell stiffness and elasticity, are crucial determinants of cellular behaviors, influencing survival, proliferation, and responses to external stressors, such as chemotherapeutic agents.^3,4 For instance, tumor or malignant cells are generally more deformable and softer than normal tissue cells.⁵ Culturing tumor cells in a matrix with high hardness can make them resistant to chemotherapeutic drugs.⁶ In addition, ectopic cancer cells in metastatic patients exhibit lower cell stiffness compared to in situ tumor cells, suggesting that tumor cells adjust their physical properties to accommodate their migratory functions.⁷ Given that cell mechanical properties are closely related to the invasion and metastasis potential of cancer cells, cell stiffness has been used as a diagnostic marker for detecting cancer cells in human fluid samples.⁸ Therefore, identifying changes in cell mechanical strength under pathological conditions can serve as a critical indicator for predicting disease progression or drug response rate.

Changes in the cell membrane composition reflect alterations in cell stiffness and participate in the process of alterations in cell stiffness. Reduction in cholesterol content reduces membrane fluidity, thereby increasing the cell stiffness.⁹ More unsaturated fatty acids make the membrane softer and more flexible, thereby reducing cell stiffness.¹⁰ On the contrary, cell adhesion molecules, including integrins and cadherins, mediate the interactions between cells and the extracellular matrix (ECM). These proteins affect the mechanical properties of cells by delivering signaling to the cytoskeleton.¹¹ Cytoskeletal proteins, such as actin, vimentin (VIM), and tubulin, play essential roles in regulating cell behaviors,^12,13 and work in coordination transmitting forces within the cell to the extracellular matrix to regulate intricate cellular activities, such as cell shape and orientation.^14,15 Furthermore, certain proteins with mechanosensitive functions are involved in structural changes. For example, Yes-associated protein (YAP) influences cell mechanics by acting as a mechanosensitive protein that responds to mechanical stimuli such as extracellular matrix (ECM) stiffness, cell shape, and cytoskeletal tension.^16–18 YAP can translocate to the nucleus in response to mechanical cues, promoting gene expression that enhances cell proliferation, survival, and epithelial-to-mesenchymal transition (EMT). These processes contribute to drug resistance in cancer by activating anti-apoptotic and pro-survival pathways.¹⁹

However, how these mechanical properties correlate with drug resistance in AML cells remains poorly understood. In order to reveal the correlation of cell deformation with the drug resistance feature, we assessed the deformation ability of AML cells using microfluidic chip technology and atomic force microscope (AFM). Our data show that drug-resistant AML cells are stiffer than drug-sensitive AML cells. The increased stiffness of drug-resistant cells may be a result of a few activated pathways, including the PPAR pathway, which promotes lipid metabolism, based on the analysis of transcriptome sequencing. Our study demonstrates the correlation of the increase in AML cell stiffness with chemotherapeutic drug resistance, providing a new understanding of the formation of drug resistance of AML cells.

II. RESULTS

A. Cytarabine-resistant AML cells exhibit increased cell stiffness

First, AML cytarabine-resistant cell lines (Ara-C-R) were established by slowly inducing at half-lethal doses of Ara-C drug molecules. The cell viability vs the drug dose is shown in Fig. 1(a) and supplementary material Fig. S1(a). The IC₅₀ value was 0.75 and 6.0 μM for sensitive HL60-S and THP1-S, respectively, while the cell viability of their drug-resistant counterparts was nearly unchanged in the tested concentration range. These data clearly show that two stable resistant AML cell lines were successfully established. Interestingly, we observed significant changes in the morphology of drug-resistant cells in comparison with the sensitive ones, with drug-resistant cells being able to more readily form clusters [Fig. 1(b) and supplementary material Fig. S1(b)], suggesting that drug-resistant cells may highly express cell adhesion molecules on their membrane surface.

FIG. 1.

Cytarabine-resistant AML cells exhibit increased cell stiffness. (a) The drug dose-cell viability profiles of HL60 cells that are sensitive and resistant to cytarabine. (b) The different cell morphologies of sensitive and resistant cells. (c) and (d) Statistical data of the cell deformability and the cell area with the density scatterplot for HL60 and THP-1 cells. A hotter color indicates a higher data density. (e) and (f) The statistics of deformation ability of (c) and (d) panels. The mean deformation values of HL60-S, HL60-R, HL60-R-Arac, THP-1-S, and THP-1-R cells were 0.200, 0.196, 0.186, 0.254, and 0.217, respectively. (g) and (h) The Young's modulus of sensitive and resistant HL60 and THP-1 cells.

Next, a microfluidic device was used to determine the likelihood of a cell to deform and thus assess the cell stiffness. During each microfluidic measurement, a few thousands of cells in a batch were measured for their deformability and cell size [Figs. 1(c) and 1(d)]. As shown in Figs. 1(e) and 1(f), the deformation of both drug-resistant cells (HL60-R and THP1-R) was significantly reduced compared with the sensitive ones (HL60-S and THP1-S). More interestingly, when the established drug-resistant HL60-R were further treated with Ara-C for 2 days, the stiffness became even higher [Figs. 1(c) and 1(e)], implying that the effect of the drug further increased the cell stiffness as well as the drug resistance. We also observed alterations in the cell size. Drug-resistant HL60 cells had a smaller size compared to their sensitive counterparts (158.0 ± 45.4 vs 256.3 ± 60.9 μm²), and further decreased upon drug stimulation (139.0 ± 39.8 μm²) [supplementary material Fig. S1(c)]. In contrast, this size reduction was not observed in the THP-1 cell line. The resistant cells showed a slight increase in size compared to sensitive cells (187.6 ± 50.7 vs 183.2 ± 58.0 μm²) [supplementary material Fig. S1(d)]. These observations suggest that changes in the cell size differ between cell lines, while changes in cell stiffness appear to follow a consistent trend.

The higher stiffness of AML drug-resistant cells was also validated using atomic force microscopy measurement. Both Ara-C-resistant cells exhibited higher Young's modulus in comparison with the sensitive ones [Figs. 1(g) and 1(h)]. These data collectively indicate that both cytarabine-resistant cells exhibit higher mechanical stiffness.

B. Cytarabine-resistant cells show higher expression of cytoskeletal proteins and lower cholesterol

The mechanical changes in cells are precisely regulated by the fluidity of the cell membrane and the content of intracellular cytoskeletal proteins.^20,21 Therefore, we further evaluated the expression changes of cytoskeleton-related proteins in drug-resistant cells. The RNA-seq data of cytarabine-resistant AML cells in the GEO database (GSE193094) revealed that the protein expression of integrin family (ITGA4 and ITGA11), myosin family (MYO9B and MYO5A), actin (ACTB and ACTN4), actin sequestering protein (TMSB4X), and intermediate filament protein (VIM) were increased [Fig. 2(a)]. This observation also implies that the increase in expression of cytoskeletal proteins may be involved in the process of hardening of drug-resistant cells.

FIG. 2.

Cytarabine-resistant cells show higher expression of cytoskeletal proteins and ABC transporters but lower expression of cholesterol. (a) Expression heatmap of cytoskeleton formation-related genes in Ara-C-sensitive or -resistant HL60 cells from GSE193094. (b) and (c) The total or membrane cholesterol amount in sensitive and resistant cells. (d) Expression heatmap of ABC transporter genes in Ara-C-sensitive or -resistant HL60 cells. (e) Statistical data of the cell deformability and the cell area with the density scatterplot for HL60-R treated by Ctrl or cholesterol. A hotter color indicates a higher data density. (f) The statistics of deformation ability of e panel. The mean deformation values of HL60-Ctrl and HL60-cholesterol cells were 0.280 and 0.309, respectively. (g) and (h) The apoptosis induction result of HL60-R treated by Ctrl or cholesterol in the presence or absence of Arac (5 μM) after 48 h.

We further investigated cell membrane factors related to cellular stiffness. Cholesterol is a special type of lipid component that, together with phospholipids, constitutes the basic membrane structure and plays key roles in influencing the fluidity of cell membranes. We found that the total cholesterol level of both AML Ara-C-resistant cells was significantly lower than that of their sensitive counterparts [Fig. 2(b)]. Consistently, the relative concentration of cholesterol in the cell membrane was also significantly decreased [Fig. 2(c)]. This finding is similar to the previous report that excess cholesterol increases the fluidity of cell membranes while removal of cholesterol induces an increase in cell membrane stiffness.²²

The ATP-binding cassette (ABC) transporter family is an important member affecting the multidrug resistance.^23,24 Therefore, we detected the gene expression levels of ABC transporter family-related proteins and found that the expression of ABCC1, ABCD1, ABCA7, and ABCB7 increased in drug-resistant cells [Fig. 2(d)]. The high expression of these ABC proteins may be one important factor that contributes to drug resistance and lower cholesterol levels in cells because these ABC transporters excrete not only the internalized drug molecules but also cholesterol.²⁵

Therefore, changes in cell stiffness caused by cellular cholesterol levels may be associated with cellular chemoresistance. It is true that we detected the decrease in cell stiffness by adding cholesterol in culture medium [Figs. 2(e) and 2(f)]. However, we did not observe significant changes in apoptosis of these softened cells upon treatment of 5 μM Ara-C [Figs. 2(g) and 2(h)]. This result suggests that reducing the stiffness of drug-resistant cells by increasing cellular cholesterol alone is not sufficient to increase drug sensitivity of these softened cells.

C. Lipid metabolism may be involved in the formation of drug-resistant cells

In order to gain a deeper understanding of the molecular mechanism that causes the increased stiffness of drug-resistant cells, the transcriptome of drug-resistant cells was re-analyzed. First, gene set enrichment analysis (GSEA) was used to globally analyze the gene expression profile of drug-resistant cells. As shown in Fig. 3(a), we observed the gene set of acute myeloid leukemia was activated [normalized enrichment score, (NES) = 1.675, p = 0.0], indicating a stronger AML-promoting signaling transduction. The activation of a gene set of assembly of collagen fibrils and other multimeric structure further shows that the drug-resistant cells had a higher ECM–receptor interaction [Fig. 3(b)]. Interestingly, we also found that the activation of ferroptosis-related gene sets in resistant cells was accompanied by the inhibition of cholesterol synthesis [Figs. 3(c) and 3(d)]. The increase in the levels of ferroptosis-related genes was associated with the upregulation of intracellular lipid peroxidation (LPO). In fact, the LPO level of resistant cells was found to be higher than that of sensitive cells [Fig. 3(e)]. LPO contributes to lipid cross-linking in cell membranes, which may induce an increase in cell membrane stiffness.

FIG. 3.

RNA sequencing analysis revealed lipid metabolism may be involved in the formation of drug resistant cells. (a)–(d) the GSEA data of cytarabine-resistant AML cells. (e) The LPO content in sensitive and resistant cells. (f) The volcano plot of gene expression differences. (g) The KEGG enrichment analysis data of upregulated genes. (h) The protein–protein interaction analysis network of upregulated genes. (i) Statistical data of the cell deformability and the cell area with the density scatterplot for HL60-R and HL60-R shPLIN2 cells. A hotter color indicates a higher data density. (j) The statistics of deformation ability of HL60-R and HL60-R shPLIN2 cells. The mean deformation values of HL60-R, HL60-shPLIN2-1, and HL60-shPLIN2-2 cells were 0.275, 0.304, and 0.369, respectively. (k) and (l) The apoptosis induction result of HL60-R-shNC and HL60-RshPLIN2 in the presence or absence of Arac (5 μM) after 48 h.

Next, the limma R package (version 3.58.1) was used to re-analyze the GEO dataset (GSE193094) and 348 genes were found to upregulate [Fig. 3(f)]. Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis of these upregulated genes indicates that the ECM-receptor signaling pathways were enriched [Fig. 3(g)]. The mechanical sensing and response of cells to ECM can regulate their stiffness through the ECM-receptor signaling pathways,²⁶ where integrins and other receptors play a key role in sensing the stiffness of ECM and transmitting signals to the internal structure of cells (such as the cytoskeleton).²⁷

Moreover, the PPAR signaling pathway was found to be enriched in drug-resistant cells [Fig. 3(g)]. It is known that activated PPAR-α promotes fatty acid uptake and β-oxidation in mitochondria and peroxisomes by upregulating the gene expression of fatty acid transporters and oxidases.²⁸ Protein–protein interaction (PPI) analysis based on the STRING database further reveals the enrichment of PPAR and ECM-related pathways, with specific molecules involved [Fig. 3(h)]. In addition, we observed upregulation of genes related to the arachidonic acid metabolic pathway and lipoxygenase, which was centered around 5-lipoxygenase (ALOX5) as well as 15-lipoxygenase (ALOX15) [Fig. 3(h)]. This observation is also consistent with the higher LPO level in drug-resistant cells. ALOX5 is one iron-containing non-heme dioxygenase that catalyzes the peroxidation of polyunsaturated fatty acids such as arachidonic acid.²⁹ Its upregulation may lead to high oxidative cross-linking of unsaturated fatty acids in the membrane and increase the cell stiffness.³⁰ Furthermore, we observed that the lipogenic differentiation-related proteins PLIN2 and lipoprotein lipase (LPL) were the central proteins connecting ECM and focal adhesion [Fig. 3(h)], further suggesting a close connection between lipid metabolism and cell stiffness. To confirm this connection, we used shRNA to knockdown the expression of PLIN2 in drug-resistant cells (supplementary material Fig. S2). We found that PLIN2 knockdown significantly increased the cell deformability [Figs. 3(i) and 3(j)], and moreover, knocking down PLIN2 with shPLIN2 significantly increased Ara-C-induced apoptosis [about 16% for shPLIN2-1 and 20% for shPLIN2-2, Figs. 3(k) and 3(l)]. This result reveals the role of PLIN2 in the cell deformation and suggests that affecting cell hardness by regulating lipid metabolism may improve the drug sensitivity of drug-resistant cells to some certain extent.

III. DISCUSSION AND CONCLUSION

Understanding how cell stiffness is correlated with drug resistance features could uncover novel mechanisms and potentially reveal new diagnostic potentials. The aim of this study is to explore this relationship by systematically characterizing the stiffness of AML cells and correlating it with their drug resistance. It is hypothesized that alterations in the mechanical properties of AML cells could provide protection against drug-induced cytotoxicity by affecting drug uptake and excretion, intracellular trafficking, and/or survival signaling pathways. This research has shown that cytarabine-resistant AML cells possessed enhanced cellular rigidity (Fig. 1). Through transcriptome sequencing, we found that drug-resistant AML cells upregulate the gene expression of cytoskeletal proteins and activate lipid metabolism pathways (Fig. 2), which may correlate the cell stiffness with the drug resistance of these AML cells.

Stiff cells might resist apoptosis triggered by chemotherapeutic drugs by altering mechanical transduction pathways.^31–33 Our study identified PLIN2 as a protein highly expressed in drug-resistant AML cells [Fig. 3(h)], linking lipid metabolism and extracellular matrix (ECM) receptor signaling pathways. This suggests that PLIN2 may be a participant in mechanical force sensing and transmission, and drug resistance in AML cells may result from the activation of PPAR and lipid metabolism pathways. Specifically, activated PPAR-α and PPAR-δ increase the expression of ABCA1 (ATP-binding cassette transporter A1) and CETP (cholesterol ester transfer protein) and thus promote the clearance of cholesterol and drug molecules from the cytosol.^34,35 Meanwhile, the upregulation of ALOX5 and ALOX15 leads to elevated lipid peroxidation (LPO) in the cell membrane [Fig. 3(e)]. Supporting our data, other studies have shown that reduced cholesterol and increased LPO in the cell membrane both decrease the membrane fluidity.³⁶ It is believed that upregulated cytoskeletal proteins may decrease the cell flexibility. Overall, these changes result in increased stiffness of drug-resistant AML cells, which may reduce membrane permeability, limit the uptake of drug molecules, and contribute to drug resistance. Consistently, inhibiting the highly expressed PLIN2 increased the deformability of drug-resistant cells and alleviated drug resistance [Figs. 3(i)–3(l)]. The increased stiffness of drug-resistant cells appears to result from the cumulative effects of continuous drug treatment, probably through regulating the expression of various proteins and lipid metabolism. In turn, this stiffness may regulate the resistance of these cells to chemotherapeutic agents. However, how cells' stiffness affects drug sensitivity is context-dependent. For instance, knocking down PLIN2 (softening the cells) improved drug sensitivity, whereas increasing cholesterol levels in the cell membrane, i.e., softening the cells, did not yield the same effect. Thus, cell stiffness and drug resistance are correlated in some way, but they do not represent a direct cause-and-effect relationship.

In summary, our research has provided some new insights into the role of cell mechanics in drug resistance, which suggest new strategies for overcoming drug resistance of AML cells.

IV. MATERIALS AND METHODS

A. Cells and reagents

THP-1 and HL60 cell lines were purchased from Cellcook (Guangzhou, China). Cytarabine (Ara-C) and Cell-Counting-Kit-8 (CCK-8) were purchased from MCE (USA). The RPMI 1640 medium was purchased from Gibco (USA) and penicillin–streptomycin was purchased from Yeasen (Shanghai, China).

B. Establishment of drug-resistant cells

AML drug-resistant cell lines, including THP-1 and HL60 Ara-C-resistant cells, were established by 8 week induction at a semi-lethal dose (IC₅₀). To maintain the resistant capacity, the cells were cultured in a medium containing a semi-lethal dose of Ara-C for AML cells at each culture stage.

C. Assessment of drug resistance

The assessment of drug resistance was evaluated using the CCK-8 and apoptosis flow cytometry. To evaluate the drug IC₅₀ in sensitive and resistant cells, we first seeded HL60/THP-1 cells in 96-well plates at 10 000 cells per well and added cytarabine in a gradient dilution method. Three replicate wells were set up. After 48 h of culture, 10 μl of CCK-8 reagent was added to each well and incubated for 1–2 h. The absorbance at 450 nm was measured using a microplate reader. The cell viability was determined by comparing the absorbance of the treated cells with the control ones.

To detect the apoptosis of sensitive and resistant cells using flow cytometry, the cultured cells were washed twice with PBS, and then resuspended in 1× binding buffer at a concentration of 1 × 10⁶ cell/ml. Annexin V-allophycocyanin (2 μl) and propidium iodide (2 μl) were added to 100 μl of each cell suspension and incubated in the dark for 15 min, followed by dilution with 400 μl of 1× binding buffer and immediate analysis using a flow cytometer to determine the percentage of apoptotic cells.

D. Lentivirus infection and stable cell line construction

To generate shRNA lentivirus against PLIN2, we co-transfected HEK293T cells with the recombinant packaging plasmid and shPLIN2 plasmid (IGE Biotechnology, Guangzhou). The culture supernatant containing the virus was collected 72 h after transfection and the cell debris was removed after centrifugation. For infection of HL60-R cells, we added the harvested lentiviral solution to the cell culture medium and added polybrene (Sigma) to a final concentration of 8 μg/ml, and replaced it with fresh medium after 24 h. Finally, cells stably expressing the target plasmid were selected using puromycin (Thermo Fisher) at a concentration of 3 mg/ml. The plasmid information of shPLIN2-1 and shPLIN2-2 are shown in the supplementary material (Table S1).

E. Microfluidic measurement of cell deformability and cell size

The polydimethylsiloxane microfluidic channel was fabricated based on previously reported techniques,³⁷ where inertial focusing and cross-channel hydrodynamic stretching were employed to measure the deformability and size of cells in suspension with high throughput [Scheme 1(A)]. The channel height was 30 μm and the channel width before and after the stretching region was 60 μm. This microfluidic system was mounted on an inverted microscope (Cnoptec BDS500) and visualized using a high-speed camera (Photron Nova S12) through a 20× objective lens. Approximately 15–20 s after the infusion of the cell solution (2.5 × 10⁵–7.5 × 10⁵ cells/ml) via a syringe pump (KDS Legato 130) at an optimized flow rate of 600 μl/min (for the flow to reach a steady state), high-speed imaging commenced. Images were captured at a frame rate of 288 000 frames per second (fps) with a 0.5 μs exposure time, covering a field of view of 128 × 128 μm.

SCHEME 1.

Approaches for measurement and quantification of the cell size and deformation. (A) A schematic diagram of the experimental setup, working principle, and definition of the cell size and cell deformation. (B) Selected sequential images for a single cell traveling through the stretching extensional flow region. The right image shows the extraction of the cell contour, major (a), and minor (b) semiaxis. The red circles denote the cell boundary detected using the active contour method.

F. Image processing and data analysis

A home-built MATLAB script was used to process the recorded images from hydrodynamic stretching for cell tracking and extracting the cell contour [Scheme 1(B)].³⁷ The precise contour of the cell shape (red), and the major (a) and minor (b) axis were obtained with the active contour method.³⁸ A density scatterplot and box plot of the cell deformation [as defined in Scheme 1(A)] vs cell area for each cell phenotype were obtained and compared between different cell groups.

G. Measurement of cell Young's modulus with atomic force microscope (AFM)

Briefly, drug-resistant and -sensitive HL60, THP-1 cells were seeded in 35 mm culture dishes coated with 0.1 mg/ml of poly-D-lysine (Yeasen, 60715ES08). The mechanical properties of cells were acquired with NanoWizard ULTRA Speed 2 at 37 °C in culture medium. The obtained data were processed by JPKSPM Data Processing software using the Hertz model. The following equation was used to obtain the Young's modulus:

$$F=\frac{4}{3}\frac{E}{1-{\nu }^2}\sqrt{R}{\delta }^{\frac{3}{2}},$$

where F is the applied force, δ is the indentation, ν is the Poisson's ratio, E is Young's modulus of the cell, and R is the radius of the spherical probe. Here, Poisson's ratio of cell is 0.5 and the spherical probe SAA-SPH-10 has a radius of 10 μm.

H. Quantification of cholesterol and lipid peroxidation

The cellular cholesterol content was measured using the Amplex Red Cholesterol Assay Kit (Beyotime Biotechnology, S0211S, China) according to the operating instruction. Specifically, 100–200 μl of BeyoLysis Buffer A for Metabolic Assay per 0.2–1 × 10⁶ cells was pipetted appropriately, and then homogenized using a hand-held grinder. The mixture was then centrifuged at 12 000 g for 3–5 min, and the supernatant was collected for subsequent quantification using a microplate spectrophotometer (BioTek Epoch2, USA).

The quantification of lipid peroxidation (LPO) was detected using the LPO detection kit (Solarbio, BC5245, China). According to the operating instruction, the cell lysate was extracted, mixed with the detection reagent, and placed in a metal bath at 100 °C for 60 min. After cooling on ice, 200 μl of supernatant was aspirated into a 96-well plate, and the absorbance of each sample at 532 and 600 nm was measured using microplate spectrophotometer (BioTek Epoch2, USA). The LPO value of each cell sample was calculated based on the concentration–optical density curve of the standard tube.

I. RNA-seq data analysis

In this study, the gene expression matrix of GSE193094 was retrieved from the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE193094). Gene set enrichment analysis (GSEA) was first performed using the GSEA software (version 4.3.3). Gene sets from Reactome, KEGG, and WikiPathways were selected for the enrichment analysis to identify pathways and biological processes associated with the gene expression patterns in the dataset.

Following the initial enrichment analysis, differential expression analysis was conducted using the limma package in R. The resulted differentially expressed genes (DEGs) were filtered based on a threshold of adjusted p-value <0.05 and a log₂ fold change (logFC) >1 or < −1. To explore the biological significance of these DEGs, clusterProfiler (version 4.14.4) was used for KEGG pathway enrichment analysis. The significantly enriched KEGG pathways were selected based on an adjusted p-value threshold of <0.05. The enriched pathways were visualized using ggplot2 package to facilitate a clear understanding of the biological processes involved.

J. Statistical analysis of the cell deformation and cell size

For statistical comparisons between two experimental groups, a two-tailed Student's t-test was employed. In cases involving three or more groups (typically three or four), one-way analysis of variance (ANOVA) was conducted. If significant differences were detected by ANOVA, Tukey's honestly significant difference (HSD) test was subsequently applied for pairwise multiple comparisons. Statistical significance was indicated as follows: ^*: p < 0.05; ^**: p < 0.01; and ^***: p < 0.001.

SUPPLEMENTARY MATERIAL

See the supplementary material for data that support the findings of this study.

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflicts to disclose.

Ethics Approval

Ethics approval is not required.

Author Contributions

Yu Wang and Hao Jiang contributed equally to this paper.

Yu Wang: Data curation (equal); Investigation (equal); Methodology (equal); Software (equal); Writing – original draft (equal). Hao Jiang: Data curation (equal); Formal analysis (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Zhenwei Su: Investigation (equal); Methodology (equal); Visualization (equal); Writing – review & editing (equal). Ran Wang: Data curation (equal); Investigation (equal); Validation (equal); Writing – review & editing (equal). Xinyuan Luo: Investigation (equal); Visualization (equal). Lingxiao Zhang: Funding acquisition (equal); Supervision (equal). Zhi Ping Xu: Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Writing – original draft (equal); Writing – review & editing (equal). Fenfang Li: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Chao He: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

DATA AVAILABILITY

The data that support the findings of this study are available within the article.