Peristalsis-Associated Mechanotransduction Drives Malignant Progression of Colorectal Cancer

Abstract

Introduction

In the colorectal cancer (CRC) tumor microenvironment, cancerous and precancerous cells continuously experience mechanical forces associated with peristalsis. Given that mechanical forces like shear stress and strain can positively impact cancer progression, we explored the hypothesis that peristalsis may also contribute to malignant progression in CRC. We defined malignant progression as enrichment of cancer stem cells and the acquisition of invasive behaviors, both vital to CRC progression.

Methods

We leveraged our peristalsis bioreactor to expose CRC cell lines (HCT116), patient-derived xenograft (PDX1,2) lines, or non-cancerous intestinal cells (HIEC-6) to forces associated with peristalsis in vitro. Cells were maintained in static control conditions or exposed to peristalsis for 24 h prior to assessment of cancer stem cell (CSC) emergence or the acquisition of invasive phenotypes.

Results

Exposure of HCT116 cells to peristalsis significantly increased the emergence of LGR5⁺ CSCs by 1.8-fold compared to static controls. Peristalsis enriched LGR5 positivity in several CRC cell lines, notably significant in KRAS mutant lines. In contrast, peristalsis failed to increase LGR5⁺ in non-cancerous intestinal cells, HIEC-6. LGR5⁺ emergence downstream of peristalsis was dependent on ROCK and Wnt activity, and not YAP1 activation. Additionally, HCT116 cells adopted invasive morphologies when exposed to peristalsis, with increased filopodia density and epithelial to mesenchymal gene expression, in a Wnt dependent manner.

Conclusions

Peristalsis associated forces drive malignant progression of CRC via ROCK, YAP1, and Wnt-related mechanotransduction.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12195-023-00776-w.

Introduction

Colorectal cancer (CRC) is estimated to be the third most common cause of cancer death in the United States with the projected number of diagnosed CRC cases over 153,000 in 2023 [93–95].

50% of all diagnosed CRC cases will eventually metastasize, leading to an abysmal 5-year survival rate of a mere 30% [83, 94]. Therefore, expanding our understanding and targeting early CRC progression is vital to improving patient outcomes.

The vast majority of CRC tumors evolve from adenomatous polyps that start as a precancerous clump of cells on the lining of the colon or rectum [28, 63]. Many precancerous adenomatous polyps switch to carcinomas (CRC), especially when they harbor genetic mutations in the KRAS gene [6, 39, 58, 86]. This early malignant progression begins with increased CRC invasiveness into the colon wall, accompanied by increased gene signatures of epithelial to mesenchymal transition (EMT) [39, 52, 60, 76, 77, 88]. Poor prognosis from malignant progression is also associated with cancer stem cell (CSC) enrichment which feeds back into increased EMT gene signatures and motility [9, 70, 91, 115]. Morphologically, motile and invasive cancer cells are linked to increased filopodia, or membrane protrusions [42, 121].

Outside of genetic drivers of malignant progression in CRC, mechanical stimuli, like shear, compression, and tension, play a large role in influencing increased tumor growth and invasion via mechanotransduction [11]. In the CRC tumor microenvironment, cells lining the colonic lumen, including cells in precancerous adenomatous polyps, are continuously exposed to mechanical forces associated with colonic peristalsis [27]. Peristalsis is concurrent multi-axial strain and shear stress that is central to the native mechanics of the colon [32, 118]. The mechanical forces of peristalsis increased leucine-rich, repeat-containing, G-protein-coupled receptor 5 (LGR5) stem cell turnover and proliferation in a non-cancerous mouse intestinal crypt organoid model [66]. In CRC, LGR5 is a cancer stem cell marker [50, 102] and increased LGR5 expression is associated with poor prognosis due to increased tumor growth, invasion, and therapy resistance [36, 38, 44]. We hypothesized that peristalsis will contribute to LGR5⁺ cancer stem cell enrichment and invasiveness in CRC, and to overall malignant progression of CRC.

In many 2D and 3D in vitro models of CRC, mechanical forces like fluid shear or uniaxial strain increased CRC cell adhesion, invasion, and eventual metastasis [35, 100, 105, 119]. In a majority of these studies, the mechanical forces were simplified applications of shear stress or uniaxial strain, but our goal was to understand the holistic effect of peristalsis (concurrent shear stress and multi-axial strain) on CRC progression [18, 47, 67, 74]. We previously developed and validated a peristalsis bioreactor that is capable of mimicking the concurrent multi-axial strain and shear stress associated with peristalsis in vitro [18]. Our device is comprised of a rotating screw-drive combined with a peristaltic pump that delivers multi-axial strain and concurrent shear stress to a biocompatible membrane ‘wall’ made from polydimethylsiloxane (PDMS; Fig. 1). Finite element modeling and experimental measures of strain in the peristalsis bioreactor indicated that the rotation of the screw drive resulted in multi-axial strain on the ‘wall’ that propagated through the ‘wall’, similar to peristalsis. Cells seeded onto the ‘wall’ therefore experienced cyclic multi-axial strain due to the rotation of the screw drive. Concurrently, fluid flow on top of the cell-seeded ‘wall’ was driven by a mini peristalsis pump. Cells seeded on the peristalsis bioreactor ‘wall’ experienced a shear stress of 0.4 Pa and average radial strain of 15% simultaneously, mimicking the range of forces found in colonic peristalsis [5, 51, 100]. We believed that the holistic re-application of the mechanical forces associated with peristalsis was crucial in the study of CRC because our previous work showed that mesenchymal stem cells responded uniquely to peristalsis compared to shear stress or strain alone [18]. Specifically, mesenchymal stem cells exposed to peristalsis had increased proliferation and actin fiber organization compared to cells exposed to shear or cyclic strain alone. Furthermore, mesenchymal stem cells exposed to peristalsis, not shear or strain individually, had gene expression changes strongly indicative of myogenic differentiation, with associated protein expression of α-smooth muscle actin [18]. Therefore, we were motivated in this current work to utilize the peristalsis bioreactor to investigate if peristalsis increased the malignant progression of CRC.

Fig. 1.

Schematic of peristalsis bioreactor. Depiction of peristalsis bioreactor setup, design and the mechanical forces at the cell level. The bioreactor setup is comprised of an Arduino Uno that powers the pump cycling media through the bioreactor. The bioreactor design itself is comprised of a rotating screw drive driven by a DC motor that delivers peristaltic wave patterning to a polydimethyl siloxane (PDMS) membrane with cells seeded on top. The peristaltic pump drives fluid flow over the top surface of the cells

Several mechanotransduction pathways are implicated in malignant progression of CRC. The maintenance and turnover of LGR5⁺ cells in a healthy colon microenvironment are supported by several signaling pathways, including Yes-associated protein-1 (YAP1) [25, 34]. In cancerous settings, YAP1 activation increases EMT gene expression which drives tumor growth and invasion [16, 25]. Importantly, mechanical forces like fluid shear stress also activate YAP1 via the Rho-associated kinase (ROCK) pathway resulting in increased cancer cell motility [54]. ROCK signaling by itself in CRC also modulates morphology and increases invasion, colony formation and self-renewal [124]. Another known driver of malignant progression in CRC is the Wnt pathway, with activating mutations in the Wnt pathway driving CRC growth [4, 22, 73]. In fact, LGR5 is a target gene of Wnt signaling [13, 59]. Interestingly, recent work has shown that Wnt can be activated via mechanotransduction [15, 40]. In this work, we set out to identify these mechanotransduction connections downstream of peristalsis using our novel peristalsis bioreactor [18]. Our goal was to expand our understanding of how peristalsis impacts LGR5⁺ cancer stem cell emergence and invasion of CRC, metrics we defined as crucial to CRC malignant progression, via mechanotransduction.

Methods

Materials

Cell culture reagents were purchased from ThermoFisher Scientific (Waltham, MA) unless otherwise specified. Cell lines were purchased from American Type Culture Collection (ATCC; Manassas, VA) unless otherwise specified. Polydimethylsiloxane (PDMS) was purchased from DOW Chemical (Midland, MI). All other chemical reagents were purchased from Sigma Aldrich (St. Louis, MO), unless otherwise indicated. Antibodies used for cellular staining were purchased from Santa Cruz Biotechnology (Dallas, TX), unless otherwise indicated. Custom-made oligos were purchased from Integrated DNA Technologies (Coralville, IA). All other molecular biology-grade reagents were purchased from ThermoFisher Scientific (Waltham, MA).

Cell Culture

Two characterized colorectal cancer cell lines, HCT116 (KRAS^G13D and PIK3CA^H1047R mutant) [1, 8] and LoVo (KRAS^G13D,V14A mutant) [1, 8], and two patient derived xenograft cell lines, PDX1 (KRAS^G12C, P53^L103H, and PTEN^G251D mutant) and PDX2 (PIK3CA^E545G and BRAF^V600E mutant), were utilized to assess the effects of peristalsis (PDX lines were gifts from the Kopetz lab at the University of Texas MD Anderson Cancer Center). Dulbecco’s Modified Eagle Medium (DMEM), Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12), and Roswell Park Memorial Institute (RPMI) 1640 Medium supplemented with 10% heat-inactivated fetal bovine serum (Peak Serum, Inc., Wellington, CO) and 1X Antibiotic-Antimycotic solution were used as the primary growth mediums for HCT116, LoVo, and PDX cells, respectively.

Along with the cancer cell lines, one non-cancerous intestinal cell line, HIEC-6 (ATCC-CRL-3266), was assessed in response to peristalsis. OptiMEM Reduced Serum Medium supplemented with 20 mM HEPES, 10 mM GlutaMAX, 10 ng ml⁻¹ Epidermal Growth Factor, 4% heat-inactivated fetal bovine serum, and 1X Antibiotic-Antimycotic solution was used as the primary growth medium for HIEC-6 culture. All cells (cancerous and non-cancerous) were cultured in standard 2D tissue culture plates and treated with 0.25% trypsin to dissociate adherent cells, ahead of seeding on the bioreactor for subsequent experiments.

Preparation and Cell Seeding on Polydimethylsiloxane (PDMS) Membranes

PDMS membranes were prepared at a 10:1 pre-polymer base to crosslinker ratio using previously established protocols [18]. Briefly, in order to maximize seeding of cells on PDMS, Collagen I was used to coat the cell seeding area of the PDMS (1.8 cm²) at 200 µg ml⁻¹. Collagen coated PDMS was refrigerated at 4 °C until use. All cells were seeded on the PDMS at 100,000 cells mL⁻¹. Cells on PDMS were incubated for 4 h at 37 °C to allow for successful attachment. Cell attachment was confirmed via imaging with a Leica DMi8 microscope (Wetzlar, Germany) (Fig. 2A).

Fig. 2.

LGR5⁺ cancer stem cell emergence and proliferation in peristalsis exposed cells. A Representative phase contrast micrographs of HCT116 cells seeded on collagen coated PDMS membranes at time 0 h and after static control or peristalsis exposure for 24 hr. Cells remained adherent even after exposure to peristalsis. Scale bar 100 µm. B Representative LGR5 flow cytometry plots of HCT116 cells maintained in static controls or exposed to peristalsis in the bioreactor. C Box and whisker plot summarizing flow analysis of LGR5⁺ expression (%) after 24 hr maintenance in static controls or exposure to peristalsis. Significant (**p < 0.01, t-test) increase in LGR5⁺ expression was noted in cells exposed to peristalsis compared to static controls. D Representative fluorescence micrographs of HCT116 cells stained with Ki67 antigen (pink fluorescence) and counterstained with nuclear marker, DAPI (blue fluorescence) from static and peristalsis conditions. Scale bar 20 µm. E Quantification of cells expressing Ki67 antigen in the nucleus via box and whisker plots. A mild statistically significant (**p < 0.01, t-test) increase in proliferation was noted in cells experiencing peristalsis compared to static controls. F Representative fluorescent micrographs of HCT116 cells stained with a live (green)/dead (red) stain from both static and peristalsis conditions. Scale bar 100 µm. G Quantification of live/dead images expressed as percentage of live cells. Static and peristalsis conditions were not significantly different in the percentage of live cells in each condition

Bioreactor Assembly and Operation

The peristalsis bioreactor was assembled and operated as previously reported [18]. Briefly, prior to full assembly, the remaining media from seeding on each PDMS membrane was removed and 20 ml of media was added to each bioreactor’s media reservoir. The PDMS membrane was placed into the bioreactor bottom, and the bioreactor top was placed gently over the membrane. The bioreactor was assembled and sealed using commercially available zip ties. The assembled bioreactor, nutrient medium reservoir, and pump were placed into the incubator and connected to an Arduino that ran a pre-programmed code producing the following parameters: 0.4 Pa shear and 15% cyclic strain at 12 rpm [18]. The bioreactor ran for 24 h and cells were then collected for various downstream analyses. A schematic of this set up is shown in Fig. 1. For the static control and static activation conditions, cell seeded PDMS membranes were incubated at 37 °C statically for 24 h. Following 24 h of static or peristalsis exposure, cells were imaged with a Leica DMi8 microscope (Wetzlar, Germany) (Fig. 2A).

ROCK inhibition (Y-27632, 10 µM) [110] and YAP1 inhibition [Verteporfin (VP), 10 µM] [29] were incorporated into the media during bioreactor operation when appropriate. ROCK activation (phorbol 12-myristate 13-acetate (PMA), 100 nM) [116], YAP1 activation (Lysophosphatidic acid (LPA), 20 µM) [12], and Wnt activation (BML-284, 4-N-(1,3-Benzodioxol-5-ylmethyl)-6-(3-methoxyphenyl)pyrimidine-2,4-diamine; 20 µM) [17] were incorporated into the media of static bioreactors for their respective experiments. Experimental conditions are outlined in Table 1.

Table 1.

Experimental conditions tested in the completed studies

Condition	Peristalsis forces	Media additive
Static	−	N/A
Peristalsis	+	N/A
Peristalsis ROCKi	+	+ 10 µM Y-27632
Peristalsis YAP1i	+	+ 10 µM Verteporfin
Static ROCKa	−	+ 100 nM PMA
Static YAPa	−	+ 20 µM LPA
Static WNTa	−	+ 20 µM BML-284

Flow Analysis for Emergence of LGR5⁺ cells

Following static or peristalsis exposure, cells were detached using trypsin and collected from PDMS into single cell suspensions in PBS supplemented with 2% FBS (FACS Buffer). Incubations for flow cytometry were performed using methods and protocols optimized previously for cancer cells [81, 82]. Briefly, cells were incubated with AlexaFluor488-LGR5 antibody, or an isotype matched AlexaFluor-488 antibody for 30 min at 37 °C. After washing and resuspending in fresh buffer, the cells were analyzed on the Attune NxT flow cytometer (ThermoFisher Scientific). Isotype controls were used to establish a gating strategy, cutting off a background gate at 0.5% (gating strategy is demonstrated in Supplementary Fig. 1). Based on the background gate, the percentage of cells expressing LGR5 was determined. Comparisons were drawn between static controls and all other conditions.

Quantification of LGR5 Expression via ELISA

HCT116 cells were exposed to peristalsis or maintained as static controls for 24 h. Following collection, cells were lysed in RIPA buffer supplemented with Halt™ protease inhibitor cocktail to extract protein. Protein concentration was measured using a Pierce BCA assay. LGR5 concentration was quantified using a commercially available LGR5 quantification kit based on sandwich ELISA (MyBioSource, San Diego, CA) following manufacturer’s protocols (25 µg of protein from each sample was used per assay). ELISA assays were performed in technical quadruplicates, from 3 independent bioreactor runs or static controls. A standard curve was generated from LGR5 standards supplied within the kit, and levels of LGR5 were interpolated for all tested conditions. Results were reported as the amount of LGR5 protein (pg ml⁻¹).

Immunofluorescent Staining and Fluorometry

Cell-seeded PDMS membranes from static, mechanically stimulated, and molecularly stimulated conditions were cut along the cell seeding area for ease of staining. Membranes were rinsed with PBS and fixed in 4% formalin for 15 min at room temperature. Samples were blocked using 0.15% Triton X and 5% Fetal Bovine Serum (FBS) at room temperature for 1 h. Cells were incubated with fluorescently tagged primary antibodies (Ki67-AlexaFluor647, Phalloidin-AlexaFluor488, or YAP1-AlexaFluor488) for 1 h at room temperature. A nuclear counterstain (DAPI) was also included in the antibody incubation. Unbound antibodies were rinsed using PBS, and cell seeded PDMS membranes were mounted using an antifade mounting reagent. Fluorescence was observed using an Olympus Fluoview FV3000 Confocal Laser Scanning Microscope (Tokyo, Japan) with 5 independent, non-overlapping regions for analysis. NIH Image J was utilized to perform fluorometry. Microscopy was performed at the Integrated Microscopy and Imaging Laboratory, a core facility at Texas A&M University.

Quantification of Proliferation

The primary antibody, Ki67 (Alexa Fluor 647) was used to mark active cell proliferation. Ki67⁺ cells were identified by the presence of pink fluorescence within DAPI (blue fluorescence) counterstained nuclei and were quantified manually using previously demonstrated protocols [18]. Analysis was performed to determine the number of actively proliferating cells (Ki67 antigen in the nuclei) versus the total cell number (number of nuclei) to determine the percentage of proliferating cells. Peristalsis exposed cells were compared to static controls.

Quantification of YAP1 Nuclear Translocation

Cells were stained with fluorescently conjugated YAP1 (Alexa Fluor 488) (Cell Signaling Technology; Danvers, MA) to visualize YAP1 distribution in the cell. YAP1 was visualized via green fluorescence, while nuclei were visualized via blue fluorescence. YAP1 activation (i.e. nuclear translocation of YAP1) was analyzed in several conditions including static controls, peristalsis exposed cells, and peristalsis exposed cells where ROCK or YAP1 was inhibited. Ultimately, YAP1 activation was quantified as the mean fluorescent intensity of YAP1 (green) located within the nucleus (blue) area (Supplementary Fig. 2). Mean intensity values from each tested condition were normalized to produce a fold change compared to static control values. Analysis of YAP1 activation was performed using Image J.

As positive controls for YAP1 activation and our quantification, cells in static conditions were also exposed to LPA (a YAP1 activator) prior to fluorescent visualization of YAP1 (Supplementary Fig. 3).

Quantification of Cell Morphology: Elongation and Filopodia Density

HCT116 cells were stained with fluorescently conjugated phalloidin (Alexa Fluor 488) to visualize actin filaments (F-actin). Actin filaments were visualized via green fluorescence and nuclei were visualized via blue fluorescence. Actin filament visualization allowed for the quantification of filopodia density and cell elongation. Fluorescent micrographs were imported into ImageJ and the calibration was set according to the image’s scale bar. The perimeter of the cell and number of filopodia projections were manually quantified to obtain filopodia density (number of projections cell perimeter⁻¹). An example of the manual quantification is presented in Supplementary Fig. 4. Cell elongation was quantified by manually tracing the major axis of each cell to obtain a length value (µm). 1–3 cells per image were quantified from 5 non-overlapping images in 3 experimental runs. All conditions were compared to one another for a complete understanding of filopodia density and elongation changes.

Assessment of Cell Viability

Cell viability was evaluated using a calcein-AM/Ethidium homo-dimer based live/dead fluorescence assay (Thermo Fisher Scientific). HCT116 cells were seeded on PDMS and exposed to peristalsis or maintained in static conditions for 24 h. Cell-seeded PDMS membranes were incubated with the live/dead dyes per manufacturer’s protocols for 15 min, followed by fluorescent imaging on a Leica DMi8 microscope (green live cells at 488 nm; red dead cells at 570 nm). 5 independent, non-overlapping regions were imaged from each membrane. NIH Image J was utilized to perform fluorometry. Live and dead cells were identified by the presence of green or red fluorescence, respectively, and quantified manually. % Live cells were quantified as green fluorescent cells/total cells in a standard area. Cell viability in response to each activator and inhibitor was also assessed (Supplementary Fig. 5).

Gene Expression Analysis

Cells were directly collected from the PDMS using Buffer RLT with β-mercaptoethanol. RNA extraction was performed using the RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA concentration and purity were evaluated using a NanoDrop OneC (ThermoFisher Scientific) and stored at − 80 °C until ready to use. Reverse transcription was performed following manufacturer’s protocols using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). qPCR was performed with a QuantStudio5 (Applied Biosystems) using the Applied Biosystems PowerSYBR Green PCR Mastermix (Thermofisher Scientific) for detection. Genes that were investigated included ROCK1 (Rho-associated protein kinase 1), ROCK2 (Rho-associated protein kinase 2), RhoA (Ras homolog family member A), Piezo1 (Piezo type mechanosensitive ion channel component 1), mTOR (Mechanistic target of rapamycin kinase), AKT (AKT protein kinase), Rac-1 (Rac family small GTPase 1), Rac-2 (Rac family small GTPase 2), CDC42 (Cell division cycle 42), BCAT (Beta-catenin), WNT1 (Wnt family member 1), WNT3a (Wnt family member 3a), WNT4 (Wnt family member 4), WNT5a (Wnt family member 5a), WNT5b (Wnt family member 5b), WNT7a (Wnt family member 7a), WNT7b (Wnt family member 7b), WNT8a (Wnt family member 8a), AXIN1 (Axin 1), AXIN2 (Axin 2), FZD (Frizzled class receptor 1), ZEB1 (Zinc finger E-box binding homeobox1), SNAIL1 (Snail family transcriptional repressor 1), SNAIL2 (Snail family transcriptional repressor 2), ETS-1 (ETS proto-oncogene 1), E-CAD (Epithelial cadherin), MMP2 (Matrix metalloproteinase−2), and MMP9 (Matrix metalloproteinase-9). The primer sequences used for each gene are shown in Table 2. Genes are categorized as follows: Mechanotransduction Genes: ROCK1, ROCK2, RhoA, Piezo1, mTOR, AKT, Rac-1, Rac-2, CDC42; Wnt Pathway Genes: BCAT, WNT1, WNT3a, WNT4, WNT5a, WNT5b, WNT7a, WNT7b, WNT8a, AXIN1, AXIN2, FZD; Epithelial to mesenchymal plasticity (EMP) Genes: ZEB1, SNAIL1, SNAIL2, ETS-1, E-CAD, MMP2, MMP9. Changes in gene expression were calculated using the 2ΔΔC_t method, with GAPDH as the housekeeping control [62]. qPCR experiments were run in triplicates, with 3 independent biological replicates. Comparisons were drawn between static cells that were not exposed to mechanical stimulus and all other conditions.

Table 2.

List of genes and corresponding forward and reverse primer sequences used in qPCR amplification

Gene	Forward sequence	Reverse sequence
GAPDH	CTGGGCTACACTGAGCACC	AAGTGGTCGTTGAGGGCAATG
ROCK1	AACATGCTGCTGGATAAATCTGG	TGTATCACATCGTACCATGCCT
ROCK2	TTGCTCTGGATGCAATACACTC	TCTCGCCCATAGAAACCATCA
RhoA	AGCCTGTGGAAAGACATGCTT	TCAAACACTGTGGGCACATAC
Piezo1	GGACTCTCGCTGGTCTACCT	GGGCACAATATGCAGGCAGA
mTOR	TCCGAGAGATGAGTCAAGAGG	CACCTTCCACTCCTATGAGGC
AKT	TCCTCCTCAAGAATGATGGCA	GTGCGTTCGATGACAGTGGT
Rac-1	ATGTCCGTGCAAAGTGGTATC	CTCGGATCGCTTCGTCAAACA
Rac-2	CAACGCCTTTCCCGGAGAG	TCCGTCTGTGGATAGGAGAGC
CDC42	CCATCGGAATATGTACCGACTG	CTCAGCGGTCGTAATCTGTCA
BCAT	AAAGCGGCTGTTAGTCACTGG	CGAGTCATTGCATACTGTCCAT
WNT1	CGATGGTGGGGTATTGTGAAC	CCGGATTTTGGCGTATCAGAC
WNT3a	AGCTACCCGATCTGGTGGTC	CAAACTCGATGTCCTCGCTAC
WNT4	AGGAGGAGACGTGCGAGAAA	CGAGTCCATGACTTCCAGGT
WNT5a	TCGACTATGGCTACCGCTTTG	CACTCTCGTAGGAGCCCTTG
WNT5b	CATGGCCTACATAGGGGAGG	CTGTGCTGCAATTCCACCG
WNT7a	CTGTGGCTGCGACAAAGAGAA	GCCGTGGCACTTACATTCC
WNT7b	GAAGCAGGGCTACTACAACCA	CGGCCTCATTGTTATGCAGGT
WNT8a	GAACTGCCCTGAAAATGCTCT	TCGAAGTCACCCATGCTACAG
AXIN1	GGTTTCCCCTTGGACCTCG	CCGTCGAAGTCTCACCTTTAATG
AXIN2	CAACACCAGGCGGAACGAA	GCCCAATAAGGAGTGTAAGGACT
FZD	TGCGAGAACCCCGAGAAGT	GGGACCAGAACACCTCGAC
ZEB1	GATGATGAATGCGAGTCAGATGC	ACAGCAGTGTCTTGTTGTTGT
SNAIL1	TCGGAAGCCTAACTACAGCGA	AGATGAGCATTGGCAGCGAG
SNAIL2	CGAACTGGACACACATACAGTG	CTGAGGATCTCTGGTTGTGGT
ETS-1	GATAGTTGTGATCGCCTCACC	GTCCTCTGAGTCGAAGCTGTC
MMP2	TACAGGATCATTGGCTACACACC	GGTCACATCGCTCCAGACT
MMP9	TGTACCGCTATGGTTACACTCG	GGCAGGGACAGTTGCTTCT
E-CAD	CGAGAGCTACACGTTCACGG	GGGTGTCGAGGGAAAAATAGG

Quantification of ROCK Activity

Relative levels of ROCK were quantified using a ROCK Activity Assay Kit (Cell Biolabs, San Diego, CA) following manufacturer’s protocols (25 µg of protein from each sample was used per assay). The assay was performed in technical duplicates, with protein collected from 3 independent bioreactor runs or static controls. A positive control for activated ROCK was diluted to generate a standard curve, from which levels of ROCK activity were interpolated for static, peristalsis, and static with ROCK activator cells. ROCK activity determined in cells exposed to peristalsis or positive control static ROCK activator conditions were normalized to static controls and expressed as a fold change in ROCK activity compared to static controls.

Statistical Analysis

Statistical analysis was performed on GraphPad Prism 9. All reported values are means ± SEM, n and result from 3 to 7 independent biological replicates. All qPCR data was normalized to static conditions within each experimental set and performed in triplicates over at least 3 biological replicates. Image analysis and morphometry included 3–5 non-overlapping fields of view from 3 to 5 biological replicates. ANOVA-based hypothesis testing or t-tests were performed where appropriate, and statistical significance is indicated within each experimental data set with associated p-values. Box and whiskers plots were used to represent collected data. Box values range from the 25th to 75th percentiles with a line at the median and whiskers extend from the smallest value to the largest value.

Results

Peristalsis Enhanced LGR5⁺ Cancer Stem Cell (CSC) Emergence and Proliferation

The HCT116 cell line was exposed to peristalsis or maintained as static controls for 24 h (Fig. 2A). After 24 h, cells were collected for flow cytometry analysis of the colorectal cancer stem cell (CSC) marker LGR5. A gating strategy for flow analysis is provided in Supplementary Fig. 1. Flow analysis for LGR5 indicated that peristalsis robustly and significantly increased LGR5⁺ cells compared to static controls (compare 28.77 ± 3.72%, n = 12 in static controls to 51.65 ± 5.34%, n = 6 in peristalsis; **p < 0.01, t-test; Fig. 2B, C). Proliferation was assessed via Ki67 immunofluorescence in 5 non-overlapping fields of view per experimental set. Peristalsis exposed cells exhibited a minor increase in proliferation compared to static controls (compare 93.27 ± 2.14%, n = 3 in static controls to 100.00 ± 0.00%, n = 3 in peristalsis; **p < 0.01, t-test; Fig. 2D, E). Cell viability was assessed in response to exposure to peristalsis or static cultures using a live/dead fluorescence assay (Fig. 2F). Quantification of live/dead images indicated no significant difference in static or mechanically activated cells with > 96% of cells live in both conditions (Fig. 2G).

ROCK Mechanotransduction and YAP1 Activation Were Driven by Peristalsis-Associated Forces

Once LGR5⁺ enrichment in HCT116 cells was observed with exposure to peristalsis, we investigated if peristalsis associated mechanotransduction was driving these changes (Fig. 3). Gene expression analysis of the Rho/ROCK pathway demonstrated significantly increased expression of Rho pathway effectors (RhoA and CDC42; Fig. 3A). In fact, many other genes associated with the Rho/ROCK pathway and mechanotransduction were also slightly elevated, indicating ROCK activation downstream of exposure to peristalsis (ROCK2, Piezo1, mTOR, and AKT; Fig. 3A). Gene expression data of mechanotransduction involvement was further validated via a ROCK activity assay. Relative to static controls, peristalsis increased ROCK activity by 69% (**p < 0.01, one-way ANOVA; Fig. 3B). Static with ROCK activator (PMA) analysis was also performed as a positive control to validate the use of PMA as a ROCK activator. PMA successfully drove ROCK activation evidenced by a 51% increase compared to static alone (*p < 0.05; one-way ANOVA; Fig. 3B) Combined with the gene expression analysis, the ROCK activity assay confirmed the connection between peristalsis and the Rho/ROCK pathway.

Figure 3.

ROCK and YAP1 mechanotransduction pathways implicated in peristalsis. A Bar graph of mechanosensitive gene expression analysis. Static controls are represented by the black dotted line at 1, with changes in peristalsis a relative fold increase compared to static control. Increases were observed in peristalsis exposed cells compared to static controls in RhoA (****p < 0.0001) and CDC42 (***p < 0.001; one sample t-test). Many of the other genes tested (ROCK2, Piezo1, mTOR, and AKT) were increased but not statistically significant (1.15 to 1.21-fold). B ROCK activity was quantified and plotted as a box and whisker plot normalized to HCT116 maintained as static controls. Significant (**p < 0.01, one-way ANOVA) increase in ROCK activity was observed in cells exposed to peristalsis compared to static controls. Similarly, a positive control for ROCK activation with phorbol ester in static conditions also demonstrated a significant (*p < 0.05, one-way ANOVA) increase in ROCK activity compared to static controls. C Fluorescence micrographs of HCT116 cells stained with YAP1 (green fluorescence) and counterstained with nuclear marker, DAPI (blue fluorescence) from static and peristalsis conditions. Visually, green YAP1 fluorescence appears to be localized within the blue nucleus indicating YAP1 activation. Additional YAP1 staining is also demonstrated in peristalsis conditions where ROCK or YAP1 is inhibited. Inhibition of ROCK (ROCKi) demonstrates more diffuse green fluorescence outside the nucleus; YAP1 inhibition (YAP1i) also similarly results in diffuse YAP1 staining. Scale bar 20 µm. D Box and whisker plots quantifying nuclear YAP1 localization relative to static controls for all tested conditions. Peristalsis increased nuclear localization compared to static controls (****p < 0.0001, one-way ANOVA) while ROCK and YAP1 inhibition decreased localization relative to peristalsis alone (****p < 0.0001 and *p < 0.05, respectively, one-way ANOVA)

We also evaluated YAP1, a known mechanosensor (Fig. 3C), via immunofluorescence to localize YAP1 expression. YAP1 was considered activated when green YAP1 fluorescence was located within the blue nuclear counterstain, indicating nuclear translocation of YAP1. Exposure to peristalsis in the HCT116 cell line significantly increased YAP1 nuclear localization by 1.64-fold compared to static controls (****p < 0.0001, one-way ANOVA; Fig. 3D). The small molecule YAP1 inhibitor, Verteporfin, significantly decreased peristalsis induced YAP1 nuclear localization compared to uninhibited peristalsis (*p < 0.05, one-way ANOVA; Fig. 3D). Interestingly, ROCK inhibition via Y-27632 during peristalsis also decreased YAP1 nuclear localization compared to uninhibited peristalsis, suggesting that ROCK activity played an intermediary role between peristalsis mechanotransduction and YAP1 activation (compare 1.64-fold, n = 3 in peristalsis to 1.39-fold, n = 3 in ROCK inhibited peristalsis; a 16% decrease; ****p < 0.0001, one-way ANOVA; Fig. 3D).

LGR5⁺ CSC Emergence was ROCK Dependent but Hippo-YAP1 Independent

With ROCK and YAP1 changes driven by peristalsis mechanotransduction, we examined if ROCK and YAP1 activation ultimately influenced LGR5⁺ CSC emergence (Fig. 4A). When ROCK activity was inhibited, LGR5⁺ enrichment downstream of peristalsis dropped significantly by 83% (***p < 0.001; one-way ANOVA, Fig. 4B). To understand if ROCK activity was sufficient to drive LGR5⁺ enrichment in the absence of mechanical forces, we treated static cells with a ROCK activator (PMA). Flow analysis on static cells stimulated with ROCK agonist (Static ROCKa) indicated a robust enrichment of LGR5⁺ cells (51.13 ± 17.58%, n = 3 in static ROCK activated compared to 28.77 ± 3.72%, n = 12 in static controls, *p < 0.05; one-way ANOVA, Fig. 4B).

Fig. 4.

ROCK and YAP1 involvement in peristalsis induced LGR5⁺ enrichment. A Representative flow analysis plots for all tested conditions: static, static with ROCK activator (ROCKa), static with YAP1 activator (YAP1a), peristalsis, peristalsis with ROCK inhibition (ROCKi), and peristalsis with YAP1 inhibition (YAP1i). Flow analysis indicated that in static conditions, activation of ROCK but not YAP1 enriched LGR5⁺. In peristalsis, inhibition of ROCK alone and not YAP1 resulted in reduced LGR5⁺. B Box and whisker plots quantifying flow cytometry LGR5⁺ expression for all conditions tested (*p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA). C Box and whisker plots of LGR5 ELISA quantification (pg/ml) for all conditions tested (*p < 0.05, one-way ANOVA)

Given the YAP1 nuclear localization and activation observed downstream of peristalsis, we inhibited YAP1 transcriptional activity using verteporfin (Fig 4A; Peristalsis YAP1i). YAP1 inhibition during peristalsis did not reduce LGR5⁺ CSC enrichment. In order to understand if YAP1 activation was involved at all in LGR5⁺ enrichment, we also stimulated static cells with a YAP1 activator (Static YAP1a; LPA), which resulted in no increases in LGR5⁺ expression compared to static cells. Our data indicated that LGR5⁺ enrichment downstream of peristalsis was dependent on ROCK activity, with YAP1 having little to no influence on LGR5⁺ CSC emergence downstream of peristalsis.

LGR5 expression was orthogonally confirmed with a sandwich-based ELISA. Similar results were observed at the protein level relative to static controls across conditions. In the presence of peristalsis, LGR5 was increased by 2.1-fold compared to static controls (*p < 0.05; one-way ANOVA, Fig. 4C). Similar to flow cytometry results, ROCK activation increased LGR5 concentration compared to static controls, though it was not significant. Importantly, ROCK inhibition in peristalsis decreased LGR5 by 0.6-fold compared to peristalsis alone (*p < 0.05; one-way ANOVA, Fig. 4C). Curiously, YAP1 inhibition in peristalsis also decreased LGR5 compared to peristalsis alone while YAP1 activation was not sufficient to drive a significant change compared to static controls (one-way ANOVA, Fig. 4C).

Wnt Activation was Involved in Peristalsis Driven Malignant Progression of CRC

Due to the close connection between LGR5 expression and Wnt activation, Wnt pathway involvement in peristalsis was explored via gene expression and LGR5 flow cytometry analysis (Fig. 5). Gene expression analysis demonstrated significantly increased expression in multiple Wnt ligands and Wnt pathway genes (BCAT, WNT4, WNT5a, WNT5b, WNT7a, WNT7b, and AXIN1) compared to static controls (Fig. 5A, Supplementary Fig. 5A). These gene expression changes confirmed that peristalsis was activating the Wnt pathway, so Wnt influence on LGR5⁺ expression was evaluated. In static conditions, Wnt activation was sufficient to drive a 1.7-fold increase in LGR5⁺ expression compared to static controls (*p < 0.05; t-test; Fig. 5B). Wnt activation in static conditions performed similar LGR5⁺ enrichment to that of peristalsis alone, but in the absence of mechanical stimulation (Fig. 5B). LGR5 expression was confirmed using an ELISA in addition to flow cytometry analysis, where the static activation of Wnt increased LGR5 concentration by 1.9-fold compared to static alone (Fig. 5C). To assess if Wnt activation was dependent on ROCK downstream of peristalsis, Wnt pathway gene expression was evaluated in peristalsis where ROCK was inhibited. Relative to static controls, despite the inhibition of ROCK activity during peristalsis, significant increases were observed in the gene expression of key Wnt pathway ligands and genes (WNT1, WNT4, WNT5a, WNT5b, WNT7a, WNT7b, AXIN1, and FZD; Fig. 5A, Supplementary Fig. 5B). The independent upregulation of Wnt pathway genes (as well as β-catenin nuclear translocation indicating canonical Wnt activity; Supplementary Fig. 6) suggested that peristalsis mediated Wnt activation was likely independent of ROCK activation.

Fig. 5.

Wnt activation in peristalsis driven malignant progression of colorectal cancer. A Heat map of Wnt pathway ligands and genes; static controls are expressed by the color marker corresponding to 1. Gene expression changes in peristalsis and ROCK inhibited peristalsis conditions are demonstrated compared to static controls. Statistical differences between each condition and static controls individually can be found in Supplementary material (Supp. Fig. 6). B Box and whisker plot of flow cytometry LGR5⁺ expression (%) after 24 h exposure to static or static Wnt activator, demonstrated a significant (*p < 0.05, t-test) increase in LGR5⁺ expression with Wnt activation even in the absence of mechanical forces. C Box and whisker plot of ELISA LGR5 quantification (pg/ml) of static and static Wnt activator conditions (*p < 0.05, t-test). When statically activated, Wnt increased LGR5 in both ELISA and flow cytometry analyses

Morphological and Epithelial to Mesenchymal Plasticity Gene Expression Changes in Cells Exposed to Peristalsis

Invasive potential was evaluated via analysis of phalloidin immunofluorescence images (Fig. 6A). Filopodia were visualized as protrusions from the cell, indicated by white arrows in the fluorescent micrographs (Fig. 6A). Changes in cell morphology via filopodia density and elongation were quantified in response to exposure to mechanical forces associated with peristalsis (Fig. 6B, C). Peristalsis exposed cells had significantly higher filopodia density compared to static controls (compare 0.64 ± 0.05 count µm⁻¹, n = 3 in peristalsis vs. 0.24 ± 0.03 count µm⁻¹, n = 3 in static, ****p < 0.0001; one-way ANOVA; Fig. 6B). YAP1 or ROCK inhibition in peristalsis significantly decreased filopodia density compared to peristalsis by 47-50% (****p < 0.0001; one-way ANOVA; Peristalsis ROCKi, Peristalsis YAP1i; Fig. 6B). In the absence of any mechanical stimulation, activation of ROCK, YAP1, or Wnt significantly decreased filopodia density compared to static controls (60–80% lower; ****p < 0.0001; one-way ANOVA; Static ROCKa, Static YAP1a, Static Wnta; Fig. 6B).

Fig. 6.

Morphological and invasive potential gene expression changes in mechanically exposed cells. A Representative micrographs of HCT116 cells exposed to all conditions (static, static ROCK activator (ROCKa), static YAP1 activator (YAP1a), static Wnt activator (Wnta), peristalsis, peristalsis ROCK inhibition (ROCKi), and peristalsis YAP1 inhibition (YAP1i) stained with Phalloidin (green) and Dapi (blue). Magnified images (bottom) contain arrows in white to identify filopodia protrusions. All scale bars 10 µm. Visually, increased protrusions were observed in peristalsis compared to static controls. B Box and whisker plot quantifying average filopodia density (count µm⁻¹) for all tested conditions. Peristalsis increased average filopodia density compared to static controls (****p < 0.0001, one-way ANOVA). ROCK and YAP1 inhibition decreased filopodia density compared to peristalsis (****p < 0.0001, one-way ANOVA). ROCK, YAP1, or Wnt activation in static conditions decreased filopodia density compared to static controls (****p < 0.0001, one-way ANOVA). C Box and whisker plot quantifying cell elongation as cell length (µm) for all tested conditions. Cells exposed to peristalsis were significantly elongated compared to static controls (***p < 0.001, one-way ANOVA). Cells were less elongated when ROCK or YAP1 were inhibited during peristalsis compared to uninhibited peristalsis (****p < 0.0001, one-way ANOVA). Activation of various pathways had mixed results in cell elongation (****p < 0.0001, **p < 0.01, one-way ANOVA). D Bar graph representing epithelial to mesenchymal plasticity (EMP) gene expression where static controls are shown by the black line at 1. Gene expression changes in peristalsis, peristalsis ROCK inhibition (ROCKi), and peristalsis YAP1 inhibition (YAP1i) are shown as a fold change relative to static controls. Exposure to peristalsis increased expression of several genes by 1.13- to 2.25-fold relative to static controls (ZEB1, SNAIL1, SNAIL2, ETS-1, MMP2, and MMP9). E-CAD expression was also decreased 0.62-fold relative to static controls. ROCK and YAP1 inhibition during peristalsis resulted in mixed increases in EMP associated genes. Importantly, ROCK inhibition in peristalsis resulted in increased E-CAD expression compared to peristalsis alone (*p < 0.05, two-way ANOVA). E Bar graph representing EMP gene expression in static Wnt activated cells. Static controls are represented by the black dotted line at 1, with changes in static Wnt activation a relative fold increase compared to static controls. Wnt activation in static conditions in the absence of peristalsis significantly increased EMP related genes (MMP9 (*p < 0.05), SNAIL1 (**p < 0.01), SNAIL2 and ETS-1 (***p < 0.001) and ZEB1 (****p < 0.0001)) compared to static controls (t-test)

Cells exposed to peristalsis were also significantly more elongated compared to static controls (compare an elongation of 37.07 ± 2.57 µm, n = 3 in peristalsis to 29.60 ± 1.33 µm, n = 3 in static, ***p < 0.001; one-way ANOVA; Fig. 6C). When ROCK or YAP1 activity was inhibited in cells experiencing peristalsis, their ability to elongate was removed (****p < 0.0001; one-way ANOVA, Fig. 6C). Interestingly in static conditions, activating ROCK in the absence of a mechanical stimulus did not elongate cells at all, evident by their rounded morphologies (length of cells in Static vs. Static ROCKa; Fig. 6A). This elongation was also not significantly altered with YAP1 activation in static conditions (compare static cells at 29.60 ± 1.33 µm, n = 3 vs. 30.81 ± 1.66 µm, n = 3 when YAP1 was activated). Our data suggested that while cell elongation was a consequence of peristalsis, activation of ROCK and YAP1 pathways in the absence of mechanical stimulus/peristalsis were not sufficient to drive cell elongation. While Wnt activation in the absence of peristalsis did not drive increased filopodia density, significantly increased cell elongation was observed in static Wnt activated conditions, similar to peristalsis stimulated cells (compare 35.90 ± 2.78 µm, n = 3 in static Wnt activated to 37.07 ± 2.57 µm, n = 3 in peristalsis).

To further corroborate the increased invasive potential evidenced by increased filopodia density, epithelial to mesenchymal plasticity (EMP) gene expression was assessed. Notably, peristalsis had 1.13- to 2.25-fold increased expression compared to static controls in many EMP genes (ZEB1, SNAIL2, ETS-1, MMP2, and MMP9) and decreased expression in E-CAD (Fig. 6D). Additionally, there were minimal differences between ROCK inhibited and YAP1 inhibited peristalsis compared to each other. Relative to peristalsis alone, ROCK inhibition during peristalsis only significantly increased MMP2 (*p < 0.05, two-way ANOVA) out of all the genes tested. Importantly the loss of E-CAD gene expression in peristalsis was completely reversed with ROCK inhibition (*p < 0.05, Fig. 6D). Therefore, ROCK inhibition in peristalsis led to partial reversal of epithelial to mesenchymal plasticity gene expression.

On the other hand, YAP1 inhibition increased SNAIL2 (**p < 0.01) and ETS-1 and MMP2 (*p < 0.05) expression, along with a lower E-CAD expression (Fig. 6D). These results indicated that YAP1 inhibition during peristalsis did not fully reverse peristalsis-induced epithelial to mesenchymal plasticity.

In order to test if Wnt activation on statically cultured cells influenced epithelial to mesenchymal plasticity, we evaluated the expression of EMP genes. Compared to static controls, Wnt activation increased gene expression of MMP9 (*p < 0.05), SNAIL1 (**p < 0.01), SNAIL2 and ETS-1 (***p < 0.001), and ZEB1 (****p < 0.0001, t-test; Fig. 6E). However, Wnt activation only demonstrated a minimal decrease in E-CAD, still suggesting strong Wnt involvement in epithelial to mesenchymal plasticity development.

Verification of LGR5 Cancer Stem Cell Emergence Due to Peristalsis-Associated Mechanotransduction in Alternate Cell Lines

To verify the findings from the HCT116 cell line, four alternate lines were tested: the KRAS mutant cell line LoVo, two patient-derived xenograft lines, PDX1 (KRAS mutant) and PDX2 (KRAS WT), and the normal intestinal epithelial cell line, HIEC-6. Exposure of LoVo cells to peristalsis resulted in 2.4-fold enrichment in LGR5 expression compared to static controls (*p < 0.05, t-test; Fig. 7A, B). Similar to LoVo, KRAS mutant PDX1 cells also expressed significant 1.6-fold increases in LGR5 when exposed to peristalsis compared to static controls (*p < 0.05; t-test; Fig. 7C, D). In contrast to LGR5 enrichment in KRAS mutant lines, the KRAS WT PDX2 cell line demonstrated no change in LGR5⁺ stem cell emergence between static and peristalsis conditions (Fig. 7E, F).

Fig. 7.

Analysis of LGR5 cancer stem cell emergence due to peristalsis in alternate cell lines. A Representative LGR5 flow analysis plots of KRAS Mutant LoVo cells maintained as static controls or exposed to peristalsis in the bioreactor. B Box and whisker plots summarizing LGR5⁺ expression (%) after 24 h maintenance in static controls or exposure to peristalsis. LoVo cells exposed to peristalsis had a significant increase in LGR5⁺ enrichment compared to static controls (*p < 0.05). C Representative LGR5 flow analysis plots of static controls and peristalsis conditions in KRAS WT PDX1 cells. D Box and whisker plots quantifying LGR5⁺ expression (%) after 24 h maintenance in static controls or exposure to peristalsis. PDX1 cells exposed to peristalsis had significantly (*p < 0.05, t-test) increased LGR5⁺ enrichment compared to static controls. E Representative LGR5 flow analysis plots of static controls and peristalsis conditions in KRAS WT PDX2 cells. F Box and whisker plots quantifying LGR5⁺ expression (%) after 24 h maintenance in static controls or exposure to peristalsis. There was no significant difference between PDX2 cells exposed to peristalsis and static controls. G Representative LGR5 flow analysis plots of static controls and peristalsis conditions in the non-cancerous intestinal cell line, HIEC-6. H Box and whisker plots summarizing LGR5⁺ expression (%) after 24 h maintenance in static controls or exposure to peristalsis. Peristalsis exposed cells showed no significant difference compared to static controls in LGR5 emergence

In order to further understand the consequences of peristalsis in a non-cancerous setting, the HIEC-6 cell line was exposed to peristalsis or maintained as static controls for 24 h. Flow analysis for LGR5 indicated no significant differences between static and peristalsis exposed cells (Fig. 7G, H).

Discussion

Mechanical forces like shear stress and strain have been shown to drive tumor progression in many different cancer types. In vitro studies linking mechanical forces to CRC progression, however, relied on simplified applications of forces like laminar shear stress or compressive stress [5, 26, 30]. Peristalsis on a chip models have recently been developed in the study of colonic motion [51, 55] but still exhibit force simplification in their application to CRC progression [27]. This work relied on the use of our novel peristalsis bioreactor, which produces physiological values of multi-axial strain and shear forces (Fig. 1) [18, 51, 53]. In the colon, peristalsis is a complex motion of wave propagation that generally runs at a multitude of frequency cycles. Our model (12 cycles min⁻¹) adopted a high frequency based cyclic system to heighten the effects of peristalsis over a condensed period of time (24 h) [21, 97]. With our established parameters (0.4387 Pa shear and 15.9% cyclic strain at 12 rpm) [18] the effects of peristalsis associated forces were evaluated in respect to the malignant progression of colorectal cancer.

We defined malignant progression by evaluating two independent metrics reported in the colorectal cancer literature as associated with poor prognosis or tumor progression. These metrics included:

• (i)Increase or enrichment in Leucine-rich repeat-containing G-protein-coupled receptor 5 positive (LGR5⁺) cells; a marker of CRC-specific cancer stem cells (CSCs) [50, 102]. In fact, LGR5 overexpression is linked to tumor initiation, invasion, and therapy resistance in CRC [36–38, 44, 108].
• (ii)Changes in cellular invasive potential via morphology changes and epithelial to mesenchymal plasticity (EMP) programming. CRC progression past its early stages relies on its invasion into the colon wall itself prior to wider metastasis which is characterized by key morphological changes [61, 101]. Furthermore, EMP is associated with poorer outcomes and increased invasion and metastasis in CRC [2, 10, 45].

With these invasive potential metrics, we leveraged our peristalsis bioreactor to understand how CRC cells responded to the unique mechanical forces associated with peristalsis. Our primary findings indicated that peristalsis increased LGR5⁺ cells compared to static controls, with only minimal changes in proliferation (Fig. 2B–E). Our focus on LGR5 was due to its important role in the carcinogenesis of CRC [7] and its ongoing role in CRC invasion and resistance to therapy [36–38, 44, 108]. The LGR5⁺ population originates from epithelial cells and is linked to tumor progression, thereby making it an ideal CSC biomarker in our study [44, 92].

With the emergence of LGR5⁺ cancer stem cells upon exposure to peristalsis, we evaluated if the ROCK pathway and YAP1 mechanotransduction were involved downstream of mechanical forces. Gene expression analysis of Rho/ROCK pathway and common mechanotransduction genes and the ROCK assay findings confirmed that peristalsis activated ROCK (Fig. 3A, B). YAP1 nuclear translocation, i.e., activation, was also observed downstream of peristalsis. However, ROCK inhibition in peristalsis also inhibited YAP1 activation, indicating that ROCK activity was a necessary intermediary in peristalsis-associated YAP1 activation (Fig. 3C, D). ROCK activation leading to YAP1 activation is reported in several cells, downstream of mechanical forces like shear stress [16, 54]. Recent work has specifically demonstrated the direct connection between ROCK activation and subsequent YAP1 activation in the context of mechanical stimulation [24, 64, 114]. Therefore, it was unsurprising that peristalsis resulted in a similar ROCK-dependent activation of YAP1. In fact, ROCK inhibition decreased the LGR5⁺ enrichment observed with peristalsis, while YAP1 inhibition alone did not in our flow cytometry analysis (Fig. 4A, B). Further, in static conditions, ROCK activation using a small molecule activator enriched LGR5⁺, while YAP1 activation using a Hippo pathway activator did not (Fig. 4C). These findings emulated results found in literature indicating ROCK’s involvement in CSC emergence and proliferation and influence over YAP1 [99, 103, 125]. This confirmed that ROCK activity was an important component of peristalsis associated mechanotransduction leading to LGR5⁺ enrichment.

The involvement of the Wnt pathway was also evaluated (Fig. 5). Peristalsis alone enriched various Wnt ligands associated with both the canonical and non-canonical Wnt pathways (Fig. 5A) [31, 79]. The static stimulation of canonical Wnt signaling (BML-284) also increased the LGR5⁺ CSC population compared to static controls alone, implying that Wnt-based LGR5⁺ enrichment downstream of peristalsis was a likely outcome (Fig. 5B, C). Interestingly, ROCK inhibition did not alter changes observed in Wnt pathway genes, implying that peristalsis can activate Wnt independent of ROCK. This was expected, given that many studies found that Wnt is involved in mechanotransduction and mechanosensing in other cell types [15, 23, 48].

After the establishment of ROCK, YAP1 and Wnt in peristalsis associated mechanotransduction, we explored how cell morphology was impacted via filopodia density and elongation (Fig. 6A–C). Filopodia protrusions are directly correlated with cancerous invasion where an increased filopodia density increases invasion [41, 43, 90]. Our results indicated that peristalsis increased filopodia density compared to static controls. Importantly, when ROCK or YAP1 were inhibited in peristalsis, this effect was muted (Fig. 6B). Similar to filopodia density trends, cell elongation was increased in peristalsis and muted when ROCK or YAP1 were inhibited in peristalsis (Fig. 6C). Cells invade in two distinct phenotypes: amoeboid-like or mesenchymal. In the amoeboid-like migration, cells exhibit a round cell-body phenotype with minimal protrusions and in mesenchymal migration, cells are characterized by an elongated cell body and longer filopodia protrusions [85, 87, 113]. Our filopodia density and cell length results indicated that peristalsis likely triggered mesenchymal-like migration, with more elongated cells (Fig. 6C). With the inhibition of ROCK and YAP1 reversing these effects, it is clear both ROCK and YAP1 are involved in peristalsis-associated morphology changes.

Interestingly, when ROCK or YAP1 were statically activated, there was a clear reduction in both filopodia density and cell length (Fig. 6B, C). A fluorescence live/dead assay confirmed that cell viability was not affected with activator treatments (Supplementary Fig 7). In literature, PMA, the chosen activator of ROCK for this study, is shown to alter the actin cytoskeleton and reduce polygonal geometry in various cell types [65, 98]. Additionally, PMA will induce the formation of membrane ruffles in other cancer cell types, as opposed to filopodia protrusions, which explains the limited number of filopodia and rounding observed in the HCT116 cells in this work [3, 20]. While the effect of the YAP1 activator, LPA, on cell length was minimally different from static alone, there was still a significant decrease in filopodia density. LPA is a known stimulator of YAP/TAZ and numerous Rho GTPases, including CDC42 which regulates filopodia formation [75, 107]. Thus, it is interesting that even in the presence of LPA, there was a decrease in filopodia. Future evaluations of invasive morphologies downstream of peristalsis call for gene and protein knock-down to understand the involvement of YAP1 activation. Collectively, our results indicate that in the case of filopodia density, the mechanical stimulation of peristalsis is vital to the formation of filopodia and the acquisition of an invasive morphology.

The morphology changes in peristalsis indicating increased invasive potential were further corroborated by epithelial to mesenchymal plasticity (EMP) gene expression. Increases in genes associated with epithelial to mesenchymal plasticity accompanied by a loss of E-CAD was observed in peristalsis exposed cells compared to static controls (Fig. 6D). The loss of E-CAD is key to epithelial to mesenchymal plasticity in combination with increases in genes like ZEB1, SNAIL1, SNAIL2, ETS-1, MMP2 and MMP9 [56, 78, 80, 109]. Based on cell elongation, filopodia density and EMP gene expression, peristalsis increased mesenchymal-like invasive potential.

When ROCK was inhibited in peristalsis, the loss of E-CAD expression was reversed, indicating an incomplete EMP transition in the presence of ROCK inhibition (Fig. 6D) [49, 72, 96]. This finding is consistent with literature indicating that Y-27632 (ROCK inhibitor), is shown to reduce EMP, specifically with the induction of E-CAD expression [111, 117, 123].

Further, YAP1 inhibition during peristalsis did not change E-CAD loss. Curiously, YAP1 inhibition did not reduce peristalsis-induced EMP gene expression at all (Fig. 6D). Even though verteporfin (YAP1 inhibitor) is documented to reduce EMP gene expression [46, 122], our findings indicate that YAP1 inhibition alone was not sufficient to reduce peristalsis-induced EMP. The discrepancy is likely because the role of YAP1 and its inhibition was tested previously in contexts that did not include a physical mechanical stimulus. In the presence of mechanically induced EMP, such as ours, YAP1 inhibition did not reverse EMP. Ultimately, only ROCK inhibition in peristalsis led to partial reversal of epithelial to mesenchymal plasticity gene expression, as the loss of E-CAD is a key component in EMP [78, 109].

Along with ROCK and YAP1, Wnt activation in static conditions was explored in connection to invasive morphology and EMP gene expression changes. Wnt activation demonstrated decreased filopodia density but a minor increase in cell length demonstrating some involvement in cellular morphology changes. These findings are expected as canonical Wnt activation, such as in the tested case, is not directly linked to modulation of the actin cytoskeleton. Rather, non-canonical Wnt signaling is known to play a role in cytoskeletal functions [68, 89, 112]. Further, Wnt activation in static conditions markedly increased EMP gene expression trends, as expected from literature (Fig. 6E). Wnt activation is established in the connection to EMP in CRC cells using in vitro Transwell and wound healing systems and in vivo tumorigenicity mouse models [57, 104]. Collectively, our results demonstrate that HCT116 cells respond to peristalsis uniquely via ROCK, YAP1, and Wnt to drive CRC malignant progression via cell morphology and EMP gene expression changes.

We also explored this connection in two alternate cells: the cell line LoVo, and a patient-derived xenograft line, PDX1, both sharing the KRAS mutation with HCT-116 (Fig. 7). With the increases in LGR5⁺ expression seen in LoVo and PDX1 cells, we determined that alternate cell lines were influenced by peristalsis in similar fashion as HCT116 cells (Fig. 7). This demonstrated that our findings were not cell-type specific or restricted to HCT116 cells alone. Additionally, we tested a KRAS WT PDX line (PDX2) to determine the relevance of KRAS mutation status in our major metric of invasive potential. Interestingly, no increase in LGR5 expression in peristalsis was observed in the WT line. Our work with cancer cell lines was further corroborated by the findings in the non-cancerous intestinal cell line, HIEC-6. In the non-cancerous setting, peristalsis exposed cells demonstrated no change in LGR5 enrichment compared to static controls (Fig. 7).

This work supports the hypothesis that peristalsis will not drive all intestinal cells to become cancerous. We hypothesized that cells with active somatic mutations (i.e. KRAS) may sense and respond uniquely to mechanical forces. KRAS mutations specifically are known to drive tumorigenesis and stimulate cancer stem cell-like emergence but their connection to peristalsis mechanics has yet to be established [33, 71, 106, 120]. Our future work will explore if and how KRAS mutational status alters peristalsis associated mechanotransduction in the malignant progression of colorectal cancer.

The implications of this work further the field of CRC by expanding the rationale for modulating peristalsis mechanics for therapeutic gain to curb early CRC progression and invasion. Several modifiable factors regulate peristalsis including diet, nutrition, and circadian rhythm [14, 19, 69, 84]. Peristalsis is also well regulated therapeutically via pro-kinetic agents that are currently not indicated in the treatment of CRC. By positively linking peristalsis to CRC malignant progression, modifiable factors like diet and nutrition can now be utilized to modulate peristalsis, and thereby arrest CRC progression.

Conclusion

Peristalsis associated mechanotransduction plays an important role in the malignant progression of colorectal cancer (CRC). As evidenced by the work using our novel peristalsis bioreactor, peristalsis-associated forces increase LGR5⁺ cancer stem cell enrichment, epithelial to mesenchymal plasticity gene expression, and invasive morphology changes compared to static controls. Peristalsis mechanotransduction was largely driven by ROCK and Wnt activation, with some mixed involvement of YAP1. The results in both KRAS mutant and WT cell lines indicate that KRAS mutation status may be a potential driver involved in peristalsis associated mechanotransduction and malignant progression. KRAS mutation status would ultimately be an interesting point of future analysis to add to the importance of peristalsis mechanotransduction in the malignant progression of CRC. The results from this work illustrate that the modulation of peristalsis with diet and nutrition, and current colonic motility therapeutics may be a promising avenue in the treatment of CRC.

Supplementary Information

Below is the link to the electronic supplementary material.

Shreya A. Raghavan

Dr. Shreya Raghavan is an Assistant Professor in the Department of Biomedical Engineering at Texas A&M University. She earned her MSE in Biomedical Engineering at the University of Michigan, and a PhD in Biomedical Engineering from Wake Forest University/Virginia Tech. Her graduate training centered around regenerative engineering and gastrointestinal physiology. Dr. Raghavan completed NIH funded postdoctoral training at the University of Michigan in cancer tissue engineering and translational immuno-oncology. Starting Spring 2020, Dr. Raghavan directs the Stem Cell, Cancer and Immune Tissue Engineering lab at Texas A&M University. Her group builds mechanically competent microenvironments to study how and why cancers have specific patterns of metastatic spread. Her approaches integrate mechanobiology, biomaterials and regenerative engineering to ask questions that intersect the cancer stem cell/immune axis. Her work is funded by the NIH/NCI, the Department of Defense, and the Cancer Prevention and Research Institute of Texas. Dr. Raghavan is an award-winning teacher, recognized for her inclusive pedagogy in the undergraduate classroom by a Montague Scholars Award from Texas A&M University. Dr. Raghavan believes that accessibility and intentional inclusion are both required to foster diversity and equity in all STEM spaces, be it the research lab, the classroom, or the broader academy.

Data Availability

Data supporting the findings of this study are deposited into a Texas Data Repository with the following https://doi.org/10.18738/T8/TJTOZN.

Declarations

Conflict of interest

Abigail Clevenger, Maygan McFarlin, Claudia Collier, Vibha Sheshadri, Anirudh Madyastha, John Paul Gorley, Spencer Solberg, Amber Stratman, and Shreya Raghavan declare that they have no conflicts of interest.

Human Studies Statement

No human studies were carried out by the authors for this article.

Animal Studies Statement

No animal studies were carried out by the authors for this article.