Leader cell myosin 10 controls adhesion dynamics, invasion, and EMT pathway activation

Abstract

Background

The epithelial-to-mesenchymal transition (EMT) relies on cytoskeletal remodeling that drives cell migration and invasion. Leader cells are a highly invasive subpopulation of cancer cells, but our knowledge of how cytoskeletal remodeling influences the EMT, and collective invasion is limited.

Methods

We integrated CRISPR/Cas9 targeting of myosin 10 (MYO10), an actin-associated cytoskeletal remodeling protein, with fluorescence assisted cell sorting (FACS), western blot, Sanger sequencing, and Inference of CRISPR Edits (ICE) analysis to generate MYO10 knockout (MYO10^KO), knockdown (MYO10^KD) and wildtype (MYO10^WT) clones. Bulk RNA sequencing and gene set enrichment analysis were used to assess differential gene expression between MYO10^KO+KD and MYO10^WT clones and epithelial-to-mesenchymal (EMT) pathway and target gene expression. siRNA-mediated MYO10 knockdown was also used to perturb MYO10 expression in A549 and H1975 cell lines. DNA-based tension-gauge tether (TGT) probes and reflection interference contrast and fluorescence microscopy were used to assess integrin-traction forces and cell spread area. Spheroids embedded in Matrigel were used to assess 3D invasion. Western blotting was used to assess levels focal adhesion and EMT target proteins.

Results

Two (2) MYO10^KO, one (1) MYO10^KD and two (2) MYO10^WT clones were generated and validated. MYO10^KO clones demonstrated increased expression of integrin subunits α6 and β4, and decreased integrin force generation, cell spread area, and focal adhesion kinase activity. N-cadherin expression and spheroid compaction were also reduced in MYO10^KO clones while the MYO10^KD clone maintained N-cadherin expression but exhibited reduced 3D invasion relative to MYO10^WT clones. Paradoxically, MYO10^KO and MYO10^KD clones demonstrated significant activation of EMT transcriptional pathways and increased levels of the EMT transcription factor (EMT-TF) SNAIL.

Conclusions

Loss of MYO10 influences several components of the leader cell adhesion machinery by increasing integrin subunit α6 and β4 expression and decreasing N-cadherin expression and integrin force generation while activating EMT transcriptional programs. These findings suggest a novel role for MYO10 in leader cells as a modulator of the integrin profile, traction force, spheroid organization and EMT cell state associated with collective invasion.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02832-0.

Background

The epithelial-to-mesenchymal transition (EMT) is a coordinated transcriptional reprogramming of epithelial cells to alter apico-basal polarity, cell-cell and cell-extracellular matrix (ECM) adhesions towards a motile mesenchymal cell state [1–3]. The shift from an epithelial to a mesenchymal state is classically characterized by changes in expression of cell junctional proteins, cadherins, and intermediate filaments (IFs). In a full EMT state, epithelial cells exhibit phenotypic plasticity and downregulate expression of epithelial markers such as E-cadherin, claudin, ZO-1 and occludin, resulting in a loss of cell-cell adhesion, apico-basal polarity, while enhancing ECM production and becoming motile [4]. As cells undergo EMT, mesenchymal markers like N-cadherin [5, 6], β-catenin, vimentin, fibronectin [4] are upregulated to support changes in cell phenotype. However, discrete shifts from fully epithelial to fully mesenchymal states may be less advantageous to cancer cells, as the retention of both epithelial and mesenchymal features in a partial or hybrid EMT state can aid the formation and migration of circulating tumor cell clusters (CTCs), confer drug resistance, and enhance tumor-initiating potential to promote metastasis [7]. Even so, metastatic cancer cells are observed across the spectrum of epithelial, hybrid or mesenchymal states and utilize co-expression E-cadherin and vimentin to support successful metastases underscoring the importance of dynamic EMT in cancer progression [8].

While EMT is an essential biological process that occurs during embryogenesis, fibrosis and wound healing, it is also implicated in cancer progression as cancer cells use EMT to detach from the primary tumor, invade into surrounding ECM and vasculature and metastasize [9, 10]. However, cancer cells can exist in transitional states during EMT, described as partial EMT, where they exhibit both epithelial and mesenchymal markers [11, 12]. Partial EMT supports metastasis and therapeutic resistance in several carcinoma models by balancing epithelial and mesenchymal features that facilitate collective cell migration [7] and is associated with poor prognosis in breast, pancreatic, cervical, and head and neck squamous cell carcinoma [13–16]. Hence, understanding how genetic drivers of collective invasion regulate EMT transcriptional programs is necessary to elucidate and target mechanisms that drive cancer metastasis and progression to reduce cancer mortality.

Collective invasion occurs when multicellular packs undergo unified coordinated migration. Maintenance of cell-cell junctions, generation of cellular contraction forces and dynamic remodeling of cell-matrix adhesions are key features of collective migration and invasion as cancer cells reorganize to infiltrate surrounding stroma [17]. Leader cells are used to describe a phenotype of an individual cell or group of cells that exists at the leading edge of an invasive cellular front or tumor margin [18]. These cells can exist within a heterotypic cellular pack with trailing follower cells [19] that maintain cell-cell interactions to promote multicellular invasion [18]. Leader-follower organization has been observed within collective invasion packs whereby leader cells are a highly invasive subpopulation that drives invasion along with the less invasive follower subpopulation [20]. Leader cells can display distinct morphologies [20], roles for signaling molecules like RhoGTPases [21], different epithelial marker expression [22], metabolic heterogeneity [23, 24] and transcriptional profiles [25]. While fixed leader-follower phenotypes are reported in collectively migrating models of bladder, breast and lung cancer [20, 22, 26], dynamic leader-follower phenotypes, where leaders and followers change positions leading the invasive front, have also been identified in models of brain cancer [27]. Further, leader cell phenotypes may be context dependent. In models of malignant mammary tissue, migration and invasion patterns appeared to be ECM-dependent [28], which suggests that the emergence and persistence of leader-follower organization in protrusions into the ECM and the leader cell invasive phenotype may differ based on environmental cues. In breast and lung cancer models with leader-follower phenotypes, collective invasion was abrogated by targeting the epithelial and mesenchymal leader cell biomarkers, keratin and fibronectin, respectively [20, 22]. Studies linking collective invasion and EMT in leader cells are limited.

We previously isolated leader and follower subpopulations from the lung cancer, H1299 large cell carcinoma model and identified myosin 10 (MYO10), fibronectin (FN1), and jagged 1 (JAG1) as leader cell biomarkers that drive collective invasion in vitro [20, 25]. In our model, JAG1 activates MYO10 expression which drives cell migration and collective invasion through filopodia-mediated reorganization of fibronectin in leader cells [25]. Myosin 10, an unconventional myosin, is a molecular motor protein that localizes to the tips of filopodia [29, 30]. Myosin 10 is required for formation and transport along filopodia and binds filamentous actin (F-actin) [31, 32]. Dynamic cytoskeletal reorganization is a fundamental component of EMT [33] with the EMT transcriptional profile identified as a regulator of actin remodeling [34]. Furthermore, increased MYO10 mRNA is associated with increased expression of EMT-regulating transcription factors SNAI2 and TWIST1 in lung squamous cell carcinoma [35]. Increased MYO10 expression is also associated with increased metastasis in breast cancer and melanomas [36, 37]. Despite evidence linking collective invasion and cytoskeletal remodeling to EMT, studies on the relationship between the biomarkers that drive collective invasion and EMT gene expression pathways are still limited.

Here, we developed MYO10 knockout H1299 leader clones to determine how stable loss of MYO10 influences leader cell biology and EMT. To generate and validate MYO10^KO and MYO10^WT leader clones, we integrated CRISPR-Cas9 and fluorescence assisted cell sorting (FACS) with Sanger sequencing and indel analysis. We isolated two (2) MYO10^KO and two (2) MYO10^WT leader clones. We also generated one (1) MYO10^KD clone: a compound heterozygote with different mutations on each MYO10 allele and reduced expression of MYO10 at the mRNA and protein levels relative to MYO10^WT clones. We then characterized the cellular morphology, proliferation, and cell cycle progression in leader clones and identified significant reductions in cell proliferation in MYO10^KO clones. RNA sequencing highlighted transcriptional differences between all MYO10^KO+KD clones and MYO10^WT clones including activation of cytoskeleton and extracellular matrix associated pathways in MYO10^KO clones. Cell spread area and cellular integrin force generation were significantly reduced in MYO10^KO+KD leader clones relative to WT controls. Of note, MYO10^KO clones also exhibited increased integrin α6 and β4 subunit expression and decreased N-cadherin and 3D invasion, associated with an epithelial phenotype. Unexpectedly, gene set enrichment analyses demonstrated significant activation of EMT transcriptional pathways and increased expression of EMT transcription factor (EMT-TF) SNAIL in MYO10^KO+KD leader clones relative to MYO10^WT controls, suggesting a role for MYO10 in modulation of EMT gene expression pathways in leader cells. Together, these data suggest that MYO10 may modulate leader cell collective invasion by influencing the EMT cell state and adhesion dynamics.

Methods

Cell culture

Human non-small cell lung cancer (NSCLC) cell lines H1299, H23, H460, H1915, and H1975 (American Type Culture Collection, USA) were cultured in RPMI 1640 Medium while A549 cells (ATCC,, USA) were cultured in Ham’s F-12 medium. All cell culture media was supplemented with 10% (v/v) Fetal Bovine Serum (FBS) (Corning, Cat. No. 35-010-CV), 100 U/ml Penicillin-Streptomycin (Gibco, Cat. No. 15140122) and 2 mM L-glutamine (Gibco, Cat. No. 25030081). Leader and follower cell lines, denoted 3L2 and 3F2 respectively, were derived from H1299 parental cell lines transfected with Dendra2, a photoconvertible fluorophore, as previously described [20, 38]. All cell lines were maintained in a humidified incubator at 37 °C with 5% CO₂. siRNA transfections used human MYO10, SNAIL (Silencer Select siRNA s9224, s9225, s13185 and s13186) or and Negative control 1 siRNA (ThermoFisher Scientific, Cat. No 4392420, 4390843) constructs and Lipofectamine RNAiMAX transfection reagent (ThermoFisher Scientific, Cat. No.13778075). Cells were used in downstream analysis 48 h after transfection.

Clone generation

Chemically modified sgRNA 5’-UCACCUGACCAUAGUCUGUC-3’ targeting Exon 2 of MYO10 was complexed with spCas9 and delivered to H1299 cells via electroporation to generate H1299 edited cell pool (Synthego, USA). Fluorescence activated cell sorting (FACS) was used to isolate single cells from the H1299 edited cell pool. Cells were harvested and stained with APC-conjugated anti-CD70 (Cat. No. 355109), PE-conjugated anti-IL13Rα2 (Cat. No. 360306) and Zombie Yellow™ viability dye (Cat. No. 423103) (all from Biolegend). CD70 + IL13Rα2- single cells were sorted into two 96-well plates (Corning, Cat. No. 3596) using the FACSAria II (BD Biosciences, USA). 96-well plates were assessed every 3 days for clone growth and colony formation. Wells containing two or more distinct colonies of cells based upon microscopy images were excluded from expansion and future analysis. At confluence, clones (denoted as CL) were sequentially expanded to 24-well, 6-well plates then two 10 cm dishes then screened for biomarker status via western blot.

Indel profiling using Interference of CRISPR Edits (ICE) analysis

Genomic DNA (gDNA) was isolated from each clone and the edited locus amplified using Platinum SuperFi II PCR Master Mix (ThermoFisher Scientific, Cat. No. 12368010) and primers forward 5’-GGTTTACATTCTACATTCCCACTGG-3’ and reverse 5’-CCTGTGGATGATGCTTTGCG-3’. PCR proceeded with cycling conditions 1 cycle of 98 °C for 30 s, 35 cycles of 98 °C for 10 s, 65 °C for 10 s and 72 °C for 30 s, then 1 cycle at 72 °C for 5 min. The PCR products were resolved and visualized on a 2% agarose E-gel (Cat. No. G700802) then purified using QIAquick PCR Purification kit (Qiagen, Cat. No. 28106). Amplicons and MYO10 PCR primers were used for bidirectional Sanger sequencing. Comparison of control and edited sample sequences for indel profiling was conducted using the ICE tool (ice.synthego.com).

Western blot

For whole cell lysates, cells were lysed in 2% SDS lysis buffer (50mM Tris-HCl (pH 8.0), 100mM NaCl) with protease and phosphatase inhibitors (ThermoFisher Scientific, Cat. No. 78442). For nuclear and cytoplasmic fraction lysates, cells were harvested using Trypsin/EDTA (Lonza Bioscience, Cat. No. CC-5012) then pelleted via centrifugation and prepared using the NE-PER™ Nuclear and Cytoplasmic Extraction reagents (ThermoFisher Scientific, Cat. No. 78833). Lysates were sonicated using Sonic Dismembrator Model 100 (Fisher Scientific, USA) before protein quantification via bicinchoninic acid assay (ThermoFisher Scientific, Cat. No. 23225). 10 µg of total protein in SDS loading buffer (Boster Bio, Cat. No. AR1112) were boiled at 100 °C for 10 min then resolved on a 4–20% gradient gel (BioRad, Cat. No. 4561096) via SDS-Polyacrylamide gel electrophoresis. Gels were transferred to nitrocellulose membranes (BioRad, Cat. No. 1620112). Membranes were blocked in 5% milk or Bovine serum albumin (BSA) in Tris Buffered Saline with 0.5% Tween 20 (TBS-T) at room temperature for 1 h. The membranes were incubated in primary antibody (see Table S1) diluted in 5% milk or BSA in TBS-T at 4 °C overnight. Membranes were washed then incubated in Horseradish peroxidase-conjugated goat anti-rabbit antibody (1:10,000, Jackson ImmunoResearch, Cat. No. 111-035-003) diluted in TBS-T for 1 h at room temperature then washed in 1X TBS with 0.5% Tween 20 before imaging on ChemiDoc MP Imaging System (BioRad, USA). Uncropped blots shown in Additional Files 2–7.

Cell proliferation assay

Cell proliferation was assessed using WST-8 reduction assay (Dojindo, Cat. No. CK04). Cells (2 ⋅ 10³ cells/well) were seeded into 96-well plates. At 0, 24, 48, 72 and 96 h after seeding, 10 µl of WST-8 reagent was added to each well and incubated in the dark for 3 hrs at 37 °C. Absorbance of each well was read at 450 nm using a BioTek Epoch Microplate reader (Agilent, USA).

Cell cycle analysis

Cells were trypsinized and washed in PBS. 1 ⋅ 10⁶ cells were fixed in ice-cold 70% ethanol at 4 °C. Fixed cells were washed in PBS twice and resuspended in PI-RNAse Staining Buffer: 2 µg/ml propidium iodide (PI) (Millipore Sigma, Cat. No. P4864-10ML), 100 µg/ml RNAse A (Sigma-Aldrich, Cat. No. R4875) with 0.1% Triton-X (Acros Organics Cat. No. 327372500) in PBS at RT in the dark for 30 min. DNA content of 2 ⋅ 10³ events were acquired using the Symphony A1 (BD Biosciences, USA) flow cytometer. Watson (Pragmatic) model in FlowJo version 10.10.0 (FlowJo LLC, USA) was used to assess distribution of cells across the cell cycle phases (G1, S and G2//M).

Spheroid invasion assay

Cells were seeded at a density of 3 ⋅ 10³ in Low Attachment 96-well plates (Corning, Cat. No. 7007), centrifuged at 1500 RPM for 5 min and incubated at 37 °C. Plates were centrifuged every 24 h at 1500 RPM for 5 min for 72 h. Spheroids were collected and resuspended in Matrigel (Corning, Cat. No. 356237) diluted to a final concentration of 5 mg/ml with complete media. Spheroids were plated in 35 mm glass bottom dishes (MatTek, Cat. No. P35G-1.5-14-C) then incubated for 1 h at 37 °C for Matrigel polymerization before addition of cell culture media. Spheroids were imaged using the Olympus IX51 Inverted Microscope 4x objective (0.13 NA air) with an Infinity 2 camera and Infinity Analyze Software version 7.1.1.66 (Teledyne Lumenera, Canada) or Nikon Eclipse Ti Microscope with NIS-Elements Software and Nikon Perfect Focus System (Nikon, Japan).

Integrin adhesion array

Integrin adhesion array kits were procured from Millipore Sigma (Product No. ECM532) to assess the expression of specific integrins on the cell surface of adherent cells. Briefly, assay wells, pre-coated with immobilized monoclonal antibodies against the following human integrin subunits: α1, α2, α3, α4, α5, αV, αVβ3, β1, β2, β3, β4, β6, αVβ5 and α5β1, were rehydrated with PBS for 10 min at RT and then the PBS was discarded. Cells were detached using 5mM EDTA (Fisher Scientific, Cat. No. BP2482100) in PBS then resuspended in Assay Buffer at 1.5 ⋅ 10⁶ cells/ml. 100 µl of cell suspension was added to each anti-α/β array well and incubated at 37 °C in 5% CO₂ for 2 h. The suspension was discarded, and wells were washed in assay buffer thrice. 100 µl of cell stain solution was added to each well for 5 min and wells washed in deionized water 4 times. 100 µl of extraction buffer was added to each well and shaken on an orbital shaker for 10 min and colorimetric change recorded as absorbance of each well (read at 555 nm) using a BioTek Epoch Microplate reader (Agilent, USA). Absorbance reflects the presence of the specific integrin subunit present on the surface of the cell captured by the antibody coated well.

RNA extraction and sequencing

RNA sequencing of monoclonal MYO10 knockout and wildtype cell lines and polyclonal H1299 parental, follower, and leader cells was performed in quadruplicate. The Emory Integrated Genomics Core conducted RNA extraction using RNeasy Mini Kit (Qiagen, Cat. No. 74104) followed by RNA quality and quantity assessments using Nanodrop (Thermo Fisher Scientific, USA) and Bioanalyzer (Agilent Technologies, USA). Library preparation and sequencing were performed at Discovery Life Sciences using the NEBNext^® Ultra™ II DNA Library Prep kit for Illumina (New England Biolabs, Cat. No. E7465) and Novaseq X-Plus (Illumina Inc., USA). Libraries were sequenced to at a depth of ~ 50 million paired reads per sample for 150 bp paired-end (PE 150) sequencing to generate fast.qz files.

RNA-seq data processing and analysis

The Emory Integrated Computational Core performed quality control and analysis of raw sequencing data. Read quality and contamination were assessed using FastQC v0.12.0 and Fastq-Screen v0.15.3 and reports generated with MultiQC v1.24.1 [39–41]. Reads were trimmed using fastp v0.23.4 and were then aligned to the human reference genome hg38 using STAR aligner v2.7.11b [42, 43]. Aligned .sam files were converted to. bam format via samtools v1.19.2 [44]. HTSeq-Count v2.0.5 was used to generate raw count Table [45]. Count tables were combined and analyzed using in-house scripts and DESeq2 [46, 47]. Differential expression analysis assumed table counts follow a negative binomial distribution with Wald Test implementation. Pairwise analyses used either Leader or MYO10 wildtype cell lines as the reference group. Genes with an adjusted P value < 0.05 after Benjamini-Hochberg correction were considered statistically significant [48]. After variance stabilizing transformation of count data, R procomp and pheatmap functions were utilized for Principal Component Analysis (PCA) and Heatmap generation respectively [49]. Volcano plots of log₂ fold-change versus adjusted P values of expression values were created using R package EnhancedVolcano. Gene Set Enrichment Analysis (GSEA) on log₂ fold changes and Over Representation Analysis (ORA) on lists of differentially expressed genes (DEGs) were conducted to assess directionality and significance of changes in gene ontology pathways using the R package clusterProfiler with pathway size restriction between 10 and 500 genes. EMT-core scores for each clone were calculated using gene expression data from a validated list of 130 epithelial to mesenchymal transition-related genes derived from a meta-analysis studying EMT in cancer progression across several carcinoma subtypes [50]. The average gene expression of validated epithelial genes downregulated during EMT was subtracted from average gene expression of upregulated mesenchymal genes for all clones to yield EMT-core scores.

DNA-based tension gauge tether (TGT) probe surface preparation

Glass coverslips were washed in 40 ml ultrapure water then sonicated for 5 min in 40 ml of 50% v/v 200-proof ethanol (Decon Laboratories, Cat. No. 2701) and ultrapure water. The glass surfaces were then washed in 200-proof ethanol and dried in an oven at 100 °C for 15 min to remove ethanol. This was followed by etching in piranha solution (1:3 H₂O₂:H₂SO₄, SupelCo, Cat. No. SX1244, Sigma Aldrich, Cat. No. 216763-500ML) for 35 min. Subsequently, coverslips were washed six times in ultrapure water followed by 200-proof ethanol three times. Cleaned glass surfaces were submerged in 2% v/v 3-Aminopropyltriethoxysilane (APTES) (Sigma Aldrich, Cat. No. 440140) in anhydrous ethanol for 40 min. The surfaces were washed in 200-proof ethanol three times then baked at 100 °C for 1 h. Trans-cyclooctene (TCO)-NHS ester (BroadPharm, Cat. No. BP-22418) was dissolved in anhydrous DMSO (SupelCo, Cat. No. MX1457) to create a solvent concentration of 3 mg/ml. Coverslips were removed from the oven and cooled to room temperature. The TCO-NHS solution (50 µL) was added on the surface of the coverslip and another coverslip added to sandwich the TCO-NHS solution overnight at RT. The resulting TCO-NHS-modified glass surfaces were washed with ethanol and water twice and water droplets gently removed. These surfaces were stored for use within 3 months of preparation. On the day of each experiment, coverslips were applied to the adhesive bottom of a 96-well plate (Grace Biolabs, Cat. No. 204969). Finally, surfaces were functionalized in each well by adding 100 µL of pre-annealed TGT DNA strands (20µM bottom strand and 30µM top stand probes in 1X PBS) producing 56 pN TGT probe surfaces.

Integrin tension imaging

56 pN TGT probe-coated wells were washed once in 1X PBS then in RPMI supplemented with 10% FBS, 1% L-glutamine and 1X penicillin-streptomycin. 1 ⋅ 10⁵ cells were seeded per well and incubated at 37 °C in 5% CO₂ for 90 min. Cells were fixed with 4% formaldehyde (Sigma Aldrich, Cat. No. 47608) in 1X PBS for 20 min at RT protected from light. Plates were stored at 4 °C before imaging. A Nikon Eclipse Ti Microscope with NIS-Elements Software and Nikon Perfect Focus System was used for fluorescent microscopy. Images were acquired with a CFI Apo ⋅100 (NA: 1.49) oil immersion objective using the TRITC Chroma filter cube (Chroma Technology Corp., Cat. No. 49004-ET) and RICM filter cube.

Image analysis

Images from all experiments were processed using ImageJ software [51]. For 3D invasion assays, spheroid circularity and invasive area were quantified as previously described [20]. For integrin tension assays, cell spread area was calculated as the area within the cell membrane of RICM images while cellular tension was described as total fluorescent intensity calculated within the cell spread area. Western blot images were converted to 8-bit format with background subtraction using FIJI (version 2.14.0/1.54f) and integrated density measurements collected [51].

Illustrations and statistical analysis

All illustrations were generated using Microsoft^® PowerPoint Version 16.90.2 (Microsoft, USA) or Biorender (Science Suite Inc., Canada). At minimum, three independent biological replicates (denoted as n) were conducted for each experiment. Statistical analyses were performed using GraphPad Prism Version 10.3.1 (GraphPad Software, LLC, USA). Welch’s t-tests were used for cell spread area and force event comparisons. One-way ANOVA tests with multiple comparisons were performed for all experiments unless otherwise indicated in figure legends. Error bars indicate the mean and standard error. Statistical significance was defined as P values < 0.05.

Results

MYO10 knockout leader H1299 cells can be generated using CRISPR/Cas9

We previously developed Spatiotemporal Genomic and Cellular Analysis (SAGA) to isolate leaders and followers, two subpopulations with distinct invasive phenotypes in 3D spheroid culture, from the H1299 lung adenocarcinoma cell line [20, 38]. MYO10 and JAG1 expression in leader cells are elevated relative to follower cells and drive motility, filopodia formation, and collective invasion [25]. To generate MYO10 knockout leader clones, we used CRISPR/Cas9 with a single guide RNA (sgRNA) targeting exon 2 of MYO10 in H1299 cells (Fig. 1A). This was followed by fluorescence-activated cell sorting of the edited cell pool by CD70 and Interleukin-13 receptor subunit alpha 2 (IL13R⍺2) expression into two (2) 96-well plates to enrich for candidate MYO10^−/− single-cell derived leader clones (Fig. 1A, B). IL13Rα2 was used as a negative selection marker to exclude follower cells from the sorted population, as IL13Rα2 is one of the top differentially-expressed genes (DEGs) between leaders and followers and is not detected in leader cells [25]. Conversely, CD70 was used as a positive selection marker to enrich for leader cells as it is significantly upregulated and overexpressed in leader cells relative to follower cells [52].

Fig. 1.

Loss of MYO10 reduces leader cell proliferation in H1299 leader cells generated using CRISPR/Cas9. A Schematic of clone generation and validation. B Representative Fluorescence assisted cell sorting plot of the CRISPR Cas9 edited H1299 cell pool to isolate MYO10^KO+KD leader cell candidate clones. C Representative screening western blot (n = 1) of MYO10^KO candidate clone whole cell lysate assessing protein levels of leader cell biomarkers, Myosin 10, Jagged1, Interleukin-13 receptor subunit alpha, and HIV-1 Tat interactive protein 2 (HTATIP2). Actin was used as a loading control. Candidate clones 96, 98 and 99 on blot shown were selected for validation via Sanger sequencing. D Chromatograms of genomic DNA (gDNA) from Sanger sequencing with sgRNA target site and base insertions highlighted sequences for validation in leader cells, MYO10^WT and MYO10^KO+KD clones. E Western blot of leader and follower biomarker proteins for Sanger sequence validated MYO10^WT clones 38 and 75 and MYO10^KO+KD clones 84, 98 and 99 and H1299 Parental, follower and leader cells and densitometry analysis of MYO10 protein levels with fold change in protein expression shown relative to MYO10^WT clone 38. Actin was used as a loading control. F Proliferation rates (optical density at 450 nm) were assessed using WST-8 reduction assay. G Brightfield images of confluent cells in 2D culture (20x). Scale bar, 200 μm. H-I. PI-RNAse flow cytometry for cell cycle analysis. Bar charts show mean and SEM (error bars) of % cells in G1, S and G2/M phase (n = 3). For all panels, all experiments conducted in 3 biological replicates with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

Of the 192 wells seeded, 37 candidate clones were isolated and assessed for protein levels of leader-follower biomarkers: MYO10, JAG1, IL13R⍺2, and HIV-1 Tat interactive protein 2 (HTATIP2) (Fig. 1C; Fig. S1A). Isolation and screening of 2 MYO10^WT clones (38 – J2D6 and 75 – J1H2) was also conducted (Fig. S1B). Assessment of leader and follower biomarkers via western blot identified 12 candidate clones negative for MYO10, IL13R⍺2, and HTATIP2 protein on western blot which were then selected for genetic validation of MYO10 status. PCR amplification of the edited region of MYO10 followed by Sanger sequencing was conducted. Subsequently, amplified regions of the MYO10^−/− candidate clones were aligned to leader WT MYO10 reference sequences for indel analysis. Twelve [12] candidate MYO10^KO clones were submitted for indel analysis which identified 2 homozygous MYO10^−/− knockout (MYO10^KO) leader clones (98 – M3F9 and 99 – M3H9) and 1 compound heterozygote MYO10 knockdown (MYO10^KD) leader clone (84 – M3H2) (Fig. 1D; Fig. S1D). Homozygous MYO10^−/− clones 98 and 99 exhibited a + 1-base pair (bp) insertion on both alleles with 99% indel frequencies (Fig. 1D; Fig. S1D). MYO10^−/− clone 84 exhibited a + 1 bp insertion and − 4 bp deletion with 92 and 96% indel frequencies (Fig. 1D; Fig. S1D). The presence of two edited alleles with different mutations designated clone 84 as a compound heterozygote [53]. MYO10^WT clones 38 and 75 had no detectable indels on around the edited region (Fig. S1C). MYO10 loss was stably retained in MYO10^KO clones over more than 10 passages (Fig. 1E). Where necessary, we utilize the annotation of MYO10^KO+KD to describe any result from analyses including clones 84, 98 and 99 while MYO10^KO describes results from analyses using only clones 98 and 99. These data demonstrate that generation of MYO10^−/− H1299 leader clones is possible using the CRISPR/Cas9 genome editing system.

Loss of MYO10 reduces proliferation in leader cells

To assess the effect of MYO10 knockout on leader cell morphology, we used brightfield imaging of 2D cell cultures. The effect of MYO10 silencing on cell proliferation has yielded varying results across several cancer models [54–56]. Hence, we assessed the proliferation rates of MYO10^KO+KD and MYO10^WT leader clones using a WST-8 assay. Homozygous MYO10^KO clones 98 and 99 exhibited significantly reduced proliferation compared to MYO10^WT clone 38 and all other H1299 cell lines (p < 0.05) (Fig. 1F). Interestingly, MYO10^KO clones 98 and 99 appear more rounded in 2D when imaged at 20x relative to MYO10^WT clones, MYO10^KD clone 84, leaders and parental cells (Fig. 1G). The compound heterozygote clone 84 did not exhibit reduced proliferation relative to MYO10^WT clone 38. To assess whether reduced proliferation in MYO10^KO clones 98 and 99 was due to cell cycle alterations, we performed PI-RNAse flow cytometry assay. The percentage of MYO10^KD and MYO10^KO clones in G1 and S phase of the cell cycle were compared to MYO10^WT clone 38 (Fig. 1H, I). Further, the percentage of cells in G2/M phase relative to MYO10^WT clone 38, 7.213%, was only significantly different for compound heterozygote clone 84, 10.90% (Fig. 1I). While the values for % of cells in G1 and S phase were statistically significant when comparing MYO10^KO+KD clones to MYO10^WT clone 38, the magnitude of the difference between these values were minor suggesting that the cell cycle was largely not disrupted due to loss of MYO10 in leader cells. The data imply that stable loss of MYO10 neither influences 2D leader cell morphology nor cell cycle progression but significantly reduces the proliferation in MYO10^KO leader clones compared to the WT clones.

MYO10^KO+KDand MYO10^WT leader cells exhibit distinct transcriptional profiles

To investigate how MYO10 influences the transcriptional landscape of leader cells, we performed bulk RNA sequencing of MYO10^KO+KD and MYO10^WT leader clones. We compared either all MYO10^KO+KD clones (84, 98 and 99) or homozygous MYO10^KO clones (98 and 99) only to MYO10^WT leader clones (38 and 75) to assess differential gene expression (Fig. 2A). Principal component analysis (PCA) indicated that homozygous MYO10^KO clones 98 and 99 clustered together and MYO10^WT clones 38 and 75 clustered together (Fig. 2B; Fig. S2A). Surprisingly, the MYO10^KD compound heterozygote clone 84 existed as an intermediate between MYO10^KO and MYO10^WT populations for Principal component 1 (PC1) (Fig. 2B). Indel analysis revealed that clone 84 had an additional − 4 bp deletion and increased protein levels of MYO10 relative to clone 98 and 99 (Fig. S2A). Volcano plots of gene expression showed significant reductions in MYO10 transcript levels of MYO10^KO clones relative to MYO10^WT clones (Fig. 2C, D). Notably, MYO10 appears in the top 40 DEGs of pairwise comparisons between homozygous MYO10^KO clones (98 and 99) to MYO10^WT leader clones (38 and 75) but not in the top 40 DEGs of pairwise comparisons including the compound heterozygote clone 84 (against MYO10^WT clones) (Fig. 2E). Transcripts with |log₂ fold change| ≥ 2.0 and a p-value < 0.05 across pairwise comparisons were designated as differentially expressed genes (DEGs). Gene expression analysis revealed 1651 DEGs between MYO10^KO clones (98 and 99) and MYO10^WT clones and 1278 DEGs between all MYO10^KO+KD clones (84, 98 and 99) and MYO10^WT clones with 1041 genes overlapping between both sets of pairwise comparisons (Fig. 2F). 610 genes were differentially expressed when only comparing MYO10^KO and MYO10^WT clones compared to 237 genes when comparing MYO10^KO+KD and MYO10^WT clones (Fig. 2F). This significant overlap in DEGs between all MYO10^KO+KD clones (84, 98 and 99) and MYO10^KO clones (98 and 99) indicate that the compound heterozygote clone 84 aligns with the intermediate status of clone 84 between MYO10^KO and MYO10^WT clones observed for PC1 (Fig. 2B).

Fig. 2.

MYO10^KO+KD and MYO10^WT leader clones exhibit distinct transcriptional profiles. A Heatmaps of differentially expressed genes (DEGs) ranked by variance for pairwise comparisons between MYO10^KO+KD clones 84, 98, 99 (left) or MYO10^KO clones 98 & 99 (right) and MYO10^WT clones 38 and 75. Z-scores generated from log₂ normalized RNA expression counts. B Principal component analysis (PCA) plot of RNA sequencing data of MYO10^WT clones 38 and 75 and MYO10^KO+KD clones 84, 98, 99. C Volcano plot of DEGs between MYO10^KO and MYO10^KO+KD versus MYO10^WT leader clone pairwise comparisons. DEGs were depicted as small red dots on a graph with log₁₀ fold FDR-adjusted p-values (y-axis) versus log₂ fold change (x-axis). Data from gene transcripts with absolute values of log₂ fold changes > 1 (green dots), adjusted p-values < 0.05 (blue dots) and transcripts that met neither criterion (gray dots) were also graphed. D RNA counts of MYO10 transcripts represented as a bar graph (E) Heatmap of top 40 DEGs between MYO10^KO+KD (right) and MYO10^KO (right) versus MYO10^WT leader clone pairwise comparisons. F Venn Diagram showing numbers of overlapping and non-overlapping DEGs between MYO10^KO+KD versus MYO10^WT and MYO10^KO versus MYO10^WT pairwise comparisons. G Gene Ontology analysis of top 10 activated or suppressed (H) pathways when comparing MYO10^KO+KD clones (left) or MYO10^KO 98 & 99 clones to MYO10^WT leader clones ranked by -log₁₀ fold FDR-adjusted p-values then Enrichment scores (ES) where -log₁₀ fold FDR-adjusted p-values are equal. Number of genes in each pathway indicated to the right of bar. For all panels n = 4 with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

To characterize biological processes and cellular compartments altered by loss of MYO10 in leader cells, Gene Ontology (GO) analysis of RNA sequencing data was performed. When ranked by -log₁₀ of the adjusted P-value (AdjP) followed by enrichment scores (where -log₁₀(AdjP) values were equal), the top 2 GO activated pathways in MYO10^KO and all MYO10^KO+KD clones relative to MYO10^WT clones were extracellular matrix (GO:0031012) and external encapsulating structure (GO:0030312) gene sets with the third activated pathway for MYO10^KO and MYO10^KO+KD clones being angiogenesis (GO:0001525) and collagen-containing extracellular matrix (GO:0062023), respectively (Fig. 2G). This evidence is consistent with previous reports that MYO10 supports degradation and remodeling of the extracellular matrix in cancer models [55, 57, 58]. In homozygous MYO10^KO clones 98 and 99 only, the top 3 suppressed pathways ranked by -log₁₀(AdjP) were myosin II complex (GO:0016460), muscle myosin complex (GO:0005859), and embryonic neurocranium morphogenesis (GO:0048702) (Fig. 2H). However, the top 3 suppressed pathways when assessing all MYO10^KO+KD clones were natural killer cell chemotaxis (GO:0035747), regulation of natural killer cell chemotaxis (GO:2000501), and CCR2 chemokine receptor binding (GO:0031727) (Fig. 2H). Overall, the results indicate that MYO10 significantly influences the transcriptional profile of leader cells to alter biological processes, like cytoskeletal remodeling and the composition of the extracellular matrix, that define the leader phenotype.

Loss of MYO10 increases integrin α6 and β4 subunit expression in leader cells

Since MYO10 supports integrin binding and activation, we examined the effects of MYO10 KO on integrin expression in MYO10^KO+KD and MYO10^WT leader clones [59, 60]. Gene expression profiling of α and β subunits revealed that Itga6 (α6 subunit) and Itgb1 (β1 subunit) transcripts were the most highly expressed in all H1299 cell lines including leader clones (Fig. 3A). Additionally, Itgb4 expression was significantly increased in all MYO10^KO clones while Itga6 expression was significantly increased in homozygous MYO10^KO clones 98 and 99 but decreased in MYO10^KD clone 84 relative to MYO10^WT clone 38 (Fig. 3B). While Itgb1 expression was increased in MYO10^KO clones 98 and 99, these values were only significant for MYO10^KO clone 99 (Fig. 3B). To determine whether the RNA and protein expression of integrin subunits is consistent, we assessed integrin subunit protein levels via western blot of whole cell lysates. MYO10^KO clones 98 and 99 showed detectable integrin β4 protein levels unlike both MYO10^WT or MYO10^KD clone 84, showing no integrin β4 (Fig. 3C, D). Interestingly, the only other cell line that expressed integrin β4 protein was followers which lack endogenous MYO10 expression (Fig. 3C, D). MYO10^KO clones 98 and 99 expressed increased integrin α6 relative to MYO10^WT clones 38 and 75 or MYO10^KD clone 84 (Fig. 3C, D). STRING database analysis confirmed the primary β subunit binding partners of integrin α6 interacts are integrin β1 and β4 (Fig. 3E). Hence, we sought to confirm cell surface expression of integrin subunits, especially β1 and β4, on leader clones. MYO10^KO clones 98 and 99 not only had significantly higher levels of cell surface integrin subunit β4 when compared to MYO10^WT clones 38 but also surpassed the levels observed in follower cells, which expressed higher integrin β4 on western blot (Fig. 3C, D and F). While clone 98 had higher levels of cell surface integrin β1 subunit, there was no significant difference in β1 subunits when clones 84 and 99 were compared to the WT control clone 38 (Fig. 3F). The surface expression of integrin subunits ⍺1, ⍺2, ⍺3, ⍺4, ⍺5, ⍺5β1, ⍺V, ⍺Vβ5, ⍺Vβ3, β2, β3, and β6 was also quantified using the integrin adhesion array (Supplemental Fig. S3).

Fig. 3.

Loss of MYO10 increases integrin α6 and β4 subunit expression in leader cells (A) RNA counts of integrin ⍺ and β subunit genes (B) RNA counts of ITGA6, ITGB4 and ITGB1 in MYO10^KD, MYO10^KO and MYO10^WT leader clones, H1299 parental, follower and leader cells (n = 4). C-D Representative western blots and densitometry assessing protein levels of integrin subunits ⍺4, ⍺5, ⍺6, ⍺V, β1, β3, β4, β5 with fold change shown relative to MYO10^WT clone 38. E STRING Database analysis of proteins interacting with integrin ⍺6. F Integrin adhesion array quantifying β4 and β1 subunit cell surface expression (n = 2). G Representative western blots and densitometry showing MYO10 expression in whole cell lysate of siMYO10 or siNegCtrl treated MYO10^WT leader clones 38 and 75, H1299 leader cells, A549 and H1975 cells (n = 2) and (H) integrin subunit ⍺6, β1, and β4 expression in siMYO10 or siNegCtrl treated A549 and H1975 (n = 2). Actin was used as a loading control. For all panels, experiments conducted in 3 biological replicates with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

To assess whether our findings on the effects of MYO10 could be reproduced in other models of NSCLC, we screened A549, H23, H460, H1915, and H1975 cell lines via western blot for and determined that A549 and H1975 expressed endogenous MYO10 (Supplemental Fig. S4). We performed siRNA-mediated MYO10 knockdown in MYO10^WT clones 38, 75, leader cells, as well as A549 and H1975 cells (Fig. 3G and Supplemental Fig. S4). Integrin subunit expression was assessed as MYO10 depletion altered expression patterns in MYO10^KO. In A549 cells, MYO10 depletion slightly increased integrin α6, decreased integrin β1, and had no effect on integrin β4 (Fig. 3H). Interestingly, integrin subunit α6, β1, and β4 was reduced with MYO10 knockdown in H1975 cells (Fig. 3H). In sum, we observed that loss of MYO10 remodels leader cell integrin profile, significantly increases the levels of integrin subunits α6 and β4, and increases cell surface expression of integrin β4 in H1299 cells but this was not observed in other NSCLC cell lines. MYO10 depletion has the opposite effect on integrin α6 and β4 subunit expression in H1975 cells.

MYO10 controls integrin expression, traction forces and cell spreading via focal adhesions in leader cells

Fibronectin activation of the integrin-focal adhesion kinase (FAK) pathway drives force generation in leader cells [20]. Given that loss of MYO10 altered integrin expression in leader clones, we hypothesized that MYO10^KO+KD clones would have decreased force generation and diminished integrin-FAK pathway activation. Force generation in leader clones was visualized and quantified using DNA-based 56 pN traction gauge tether (TGT) probes that present fibronectin mimetic peptide, cyclic RGDfK (cRGD). These probes shear when integrin-cRGD forces exceed their force tolerance threshold (F_tol). These shearing events are detected using a fluorophore-quencher system where dissociation of the quencher strand eliminates quencher signal and produces a fluorescent signal (Fig. 4A). This mechanism allows us to report the integrin forces that exceed F_tol of TGT probes by quantifying increased fluorescence intensity [61]. Due to their altered integrin expression, we first assessed whether MYO10^KO and MYO10^WT clones could generate a detectable fluorescent signal using cRGD-presenting 56 pN TGT probes. We observed fluorescent signals in both MYO10^KO and MYO10^WT clones which was necessary for experimental validation as our previous work detected differences in leader and follower cell traction force generation using a 36 pN TGT probe which sheared under less force than the current probe (56 pN) [20]. Quantification of cell spread area, and integrated fluorescence intensity revealed that MYO10^KO+KD clones had a significantly smaller spread area when compared to MYO10^WT clones (Fig. 4B, C). Follower cells also exhibited a larger cell spread area than leader cells, suggesting that the MYO10-dependent factors which contribute to cell adhesion necessary for cell spreading are specific to leader cells not followers lacking endogenous MYO10. Further, MYO10^KO+KD clones generated significantly lower integrin traction force than MYO10^WT clones and the MYO10^KD clone 84 did not differ significantly in integrin force signals generated relative to MYO10^KO clones (Fig. 4D). These findings suggest that MYO10 supports force transmission via actomyosin contractility and regulates mechanotransduction in leader cells. Furthermore, leader cells have increased expression of subunits α6 and β4, which may serve as a compensatory mechanism in the absence of MYO10.

Fig. 4.

MYO10-dependent integrin expression controls traction force and cell spreading via focal adhesions in leader cells (A) Schematic of force generation assay using irreversible 56 piconewton (pN) DNA-based TGT probe and representative RICM and fluorescence imaging (TRITC) to detect traction force signal of probe shearing in MYO10^KD, MYO10^KO and MYO10^WT leader clones. Images were acquired using 100x oil immersion objective and Chroma filter cubes on Nikon Eclipse Ti Microscope. B Representative RICM, TRITC and overlay tension images of MYO10^KO and MYO10^WT leader clones, H1299 parental, follower and leader cells showing cells spread on tension probe-coated glass and tension signal using irreversible 56pN TGT probe after 90 min incubation and formaldehyde fixing. Scale bar, 20 μm. C-D Quantification of (C) cell spread area and (D) total fluorescence intensity of tension signals for MYO10^KO and MYO10^WT leader clones, H1299 parental, follower and leader cells where each dot represents a single cell, N > 20 cells for each experiment. Welch’s t-test were used to compare means. E Western blots of focal adhesion complex components phosphorylated focal adhesion kinase (p-FAK), focal adhesions kinase (FAK), paxillin, ⍺-actinin, vinculin and tensin 2 in of whole cell lysate assessing of in MYO10^KO and MYO10^WT leader clones, H1299 parental, follower and leader cells (n = 3). Actin was used as a loading control. F Densitometry of FAK, phosphorylated FAK (pFAK^Y397) and ratios of pFAK^Y397 to FAK. G Densitometry of Paxillin, phosphorylated Paxillin (pPaxillin^Y118) and ratios of pPaxillin^Y118 to Paxillin. Western blot densitometry fold changes were normalized to target protein expression in MYO10^WT clone 38. H Representative TRITC and tension images of of siMYO10 or siNegCtrl treated MYO10^WT leader clones 38 and 75, H1299 leader cells, A549 and H1975 cells showing fluorescent tension signal using irreversible 56pN TGT probe after 90 min incubation and formaldehyde fixing. Scale bar, 10 μm. I Quantification of mean (left) and total (right) fluorescence intensity produced by tension signals where each dot represents a single cell, N > 20 cells for each experiment. Welch’s t-test were used to compare means. For all panels, all experiments conducted in 3 biological replicates with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

To determine whether spread area and traction forces in MYO10^KO+KD clones were reduced due to diminished activity of FAK, we assessed protein levels of phosphorylated focal adhesion kinase (pFAK^Y397) and total FAK. All MYO10^KO+KD clones had less pFAK^Y397 when compared to MYO10^WT clones despite MYO10^KO clones 98 and 99 having increased levels of total FAK (Fig. 4E, F). Normalized ratios of the fold change pFAK^Y397/FAK show that MYO10^KO clones have less active FAK relative to WT controls which could be due to reduced activation of the integrin-FAK pathway (Fig. 4E, F). Western blot also indicated that all MYO10^KO+KD clones had mildly increased expression of paxillin, a focal adhesion associated protein essential for focal adhesion turnover, while MYO10^KO clones had decreased phosphorylated Paxillin^Y118 (pPax^Y118) compared to MYO10^WT clones (Fig. 4E, G). All MYO10^KO+KD clones had reduced normalized ratios of pPaxillin^Y118/Paxillin which may be due to reduced activity of FAK (Fig. 4E, G).

To assess whether our observations on MYO10 supporting force transmission in leader cells were reproducible in different NSCLC cell lines, we repeated siRNA-mediated MYO10 knockdown followed by tension probe force generation assay. In wildtype clones and H1975 cells, MYO10 depletion significantly reduced the total force events and mean force events (Fig. 4H, I). In leader cells and A549 cells, there were no significant differences traction force generation in MYO10 depleted and non-depleted cells (Fig. 4H, I). These findings show that in H1299 leader cells MYO10-dependent integrin expression supports force generation and cell spreading and MYO10-depletion decreases levels of pFAK^Y397 and pPax^Y118 in leader clones. However, MYO10 depletion affects traction force generation and α6, β1, and β4 integrin expression differently in distinct NSCLC lung cancer models, like A549 and H1975 cells.

MYO10 supports spheroid compaction, collective invasion, and N-cadherin expression in leader cells

Leader cell integrin-mediated forces are essential for 3D invasion [20]. We next investigated how stable loss of MYO10 influences 3D spheroid invasion as our MYO10^KO+KD clones generate significantly less force than their wildtype controls (Fig. 4D). Spheroids formed from homozygous MYO10^KO clones (98 and 99) were more diffuse and had a significantly larger area than spheroids formed from the compound heterozygote homozygous MYO10^KD clone 84, MYO10^WT clones, H1299 parental cells, followers and leaders (Fig. 5A). The apparent increase area of spheroids formed from MYO10^KO clones 98 and 99 is associated with reductions in adhesion molecules resulting in decreased cell-cell adhesion. Inability of cells to adhere in spheroids has been linked to loss of cell-cell adhesion molecules like E- and P-cadherin [62]. Previous reports indicated that our H1299 derived cell lines expressed N-cadherin and did not express E-cadherin [20]. Volcano plots showing DEGs between MYO10^KO clone 98 and MYO10^WT clones revealed that CDH2 (N-cadherin gene) was significantly downregulated (Fig. 5B). E- and P-cadherin proteins were not detected in any of our H1299 derived cell lines (Fig. 5C). Interestingly, MYO10^KO clones 98 and 99 showed low N-cadherin expression on western blot relative to clone 84, MYO10^WT clones, H1299 parental cells, followers and leaders indicating that loss of MYO10 influenced N-cadherin expression in leader clones (Fig. 5C). This result is consistent with significant reduction in MYO10^KO clones 98 and 99 of CDH2 expression from our transcriptional data (Fig. 5B). From this data we can infer that MYO10 supports N-cadherin expression which may increase spheroid compaction in leader clones.

Fig. 5.

MYO10 supports leader cell spheroid compaction and collective invasion by modulating N-cadherin expression. A Brightfield images (4x) of 3D spheroids and quantification of spheroid area. Scale bar, 500 μm (n = 3). B Volcano plot of DEGs between MYO10^KO clone 98 and MYO10^WT leader clone pairwise comparisons with CDH2 (N-cadherin) gene transcript highlighted as large red dot and RNA counts of N-cadherin gene transcript CDH2 in MYO10^KO and MYO10^WT leader clones, H1299 parental, follower and leader cells. Data presented as mean with SEM (n = 4). C Representative western blot of N-cadherin, P-cadherin and E-cadherin of whole cell lysates (n = 3). Actin was used as a loading control. Densitometry shown for N-cadherin expression. D-F Brightfield images (4x) of 3D spheroid invasion assay and outlined invasive area after 48 h with quantification of spheroid invasive area and circularity MYO10^WT clones 38 and 75, MYO10^KD clones 84, 98 and H1299 Parental, follower and leader cells. For all panels, all experiments conducted in 3 biological replicates with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

Spheroids of MYO10^KO clones did not retain their 3D structure and fragmented once embedded in Matrigel (Fig. 5D). MYO10^KD clone 84 spheroids retained their 3D structure and were successfully embedded in Matrigel like MYO10^WT clones, H1299 parental cells, followers and leaders (Fig. 5D and Supplemental Figure S6A). MYO10^KD spheroids were significantly less invasive than MYO10^WT clone 38 (Fig. 5E). MYO10^WT clone 38 had a significantly larger invasive area than MYO10^WT clone 75 suggesting that WT clones have different invasive capacity. Spheroid circularity, an indirect measure of sheet-like invasion demonstrated by follower cells (as opposed to leader cell chain-like invasion), was also assessed (Fig. 5F). When compared to MYO10^WT clone 38, only cell lines with followers, H1299 Parentals, and 3F2 follower cells, demonstrated significant differences in spheroid circularity (Fig. 5F). Mixed spheroids containing 30% MYO10^KO clone 98 or MYO10^WT clone 38 and 70% followers were made to recapitulate the mixed subpopulations found in H1299 parental population. The inclusion of follower cells did not rescue the lack of spheroid compaction (Supplemental Figure S6B) which may suggest that targeting MYO10 may influence the 3D structure of tumor cell clusters in other contexts. Our results suggest that while MYO10 supports collective chain-like invasion in leader cells, the absence of MYO10 does not fully convert leader cells to the sheet-like invasive pattern observed in follower cells which do not express MYO10.

MYO10 depletion activates EMT transcriptional pathways, increases SNAIL and reduces nuclear β-catenin in leader cells

As MYO10^KO leader clones demonstrated significantly reduced N-cadherin, we aimed to assess the effects of MYO10 depletion on EMT transcriptional pathways. Unexpectedly, GSEA of RNA sequencing data comparing MYO10^KO+KD versus MYO10^WT clones showed significant activation of EMT-associated pathways including neural crest cell differentiation (GO:0014033), mesenchymal cell differentiation (GO:0048762), regulation of epithelial to mesenchymal transition (GO:0010717) and epithelial to mesenchymal transition pathways (GO:0001837) (Fig. 6A). Gene expression of key mesenchymal proteins fibronectin (FN1), vimentin (VIM), β-catenin (CTNNB1), epithelial proteins zonula occludens-1 (TJP1) and claudin 1 (CLDN1) were assessed and their protein levels validated via western blot to confirm activation of EMT in MYO10^KO+KD leader cells. Fibronectin was significantly increased at the transcript level in MYO10^KD clone 84 and MYO10^KO clone 98 and protein levels increased in all MYO10^KO+KD clones relative to MYO10^WT clone 38 (Fig. 6B, D). Vimentin was significantly increased at the transcript level in MYO10^KD clone 84 but protein levels were increased in all MYO10^KO+KD clones relative to MYO10^WT (Fig. 6B, D,E). Although transcript levels of β-catenin were significantly increased in MYO10^KO clones 98 and 99 relative to WT clone 38, the protein levels were moderately reduced in MYO10^KD clone 84 and MYO10^KO clone 98 and increased in MYO10^KO clone 99 (Fig. 6B, D,E). Transcript levels of TJP1 were not significantly different when compared to MYO10^WT clone 38 and protein levels of its gene product zona occludens-1 (ZO-1) also showed no detectable difference (Fig. 6C, D). While CLDN1 was significantly increased in MYO10^KO clones 98 and 99, the protein product claudin 1 was not detectable via western blot (Fig. 6C, D).

Fig. 6.

MYO10 modulates hybrid EMT features, increases SNAIL and reduces nuclear β-catenin in leader cells (A) Gene set enrichment analysis of MYO10^KO+KD vs. MYO10^WT clones for EMT pathways. B-C RNA counts of FN1, VIM, CTNNB1, TJP1, CLDN1 in MYO10^KO+KD and MYO10^WT leader clones, H1299 parental, follower and leader cells (n = 4). D-E Representative western blots and densitometry assessing canonical EMT protein levels fibronectin, vimentin, beta catenin, claudin 1 and ZO-1 (n = 3) with fold change in protein expression shown relative to MYO10^WT clone 38. F Western blot of nuclear and cytoplasmic fractions for EMT-TFs and densitometry of SNAIL showing fold change. G (left) Representative western blots and densitometry showing SNAIL expression in whole cell lysate of siMYO10 or siNegCtrl treated MYO10^WT leader clones 38 and 75, H1299 leader cells, A549 and H1975 cells (n = 2). Actin was used as a loading control. And (right) STRING Database analysis of proteins interacting with SNAIL. H Western blot of nuclear and cytoplasmic fractions of β-catenin with densitometry showing fold change in nuclear β-catenin (left) and cytoplasmic β-catenin (right). I EMT-core score calculation of MYO10^KO+KD and MYO10^WT clones, mean, standard deviation and standard error of the mean and 95% confidence interval (CI). For all panels, all experiments conducted in 3 biological replicates with one-way ANOVA with Dunnett’s post hoc test for multiple comparisons of all means compared to MYO10^WT clone 38 mean values and error bars indicate standard error of the mean (SEM) with p-values *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, unless otherwise indicated. ns indicates no significance

To determine which transcription factors were involved in the activation of EMT transcriptional networks in MYO10^KO+KD clones relative to MYO10^WT clones, protein levels of EMT transcription factors (EMT-TFs) SNAIL, Twist 1, Zeb 1 and Slug were evaluated. Nuclear and cytoplasmic fraction of cell lysates were isolated and immunoblot revealed dramatic increases in the for EMT-TF SNAIL in MYO10^KO clones with slight increases in SNAIL expression in MYO10^KD clone 84 relative to WT controls (Fig. 6F). Zeb 1 and Twist 1 TFs showed no clear pattern linking its expression to leader clone MYO10 status (Fig. 6F). Western blot of siMYO10 versus siNegCtrl incubated MYO10^WT clones 38, 75, leader cells, as well as A549 and H1975 cells, shows that MYO10 depletion only enhanced SNAIL expression in wildtype clones 38 and 75 relative to non-depleted samples (Fig. 6G). MYO10 depletion had the opposite effect in H1975 cells resulting in a slight decrease in SNAIL, while substantive changes in SNAL expression were not observed regardless of MYO10 status in H1299 leader and A549 cell lines (Fig. 6G). STRING database queries of the transcription factor SNAIL shows that it interacts with epigenetic regulatory enzymes like DNA methyltransferase 1 (DNMT1), histone deacetylase 1 and 2 (HDAC1/2) and euchromatic histone lysine N-methyltransferase 1 (EHMT2) which can change gene expression (Fig. 6G). Increased expression of integrin subunits α6 and β4 was observed in MYO10^KO clones 98 and 99 (Fig. 3A-C). In models of colorectal cancer, splice variants of integrin α6A were shown to inhibit Wnt/β-catenin pathway activation [63]. This led us to examine whether changes in expression of integrin subunits α6 and β4 observed in MYO10^KO clones influenced β-catenin signaling. We therefore evaluated nuclear and cytoplasmic levels of β-catenin via western blot and show that MYO10^KO clones 98 and 99 had reduced nuclear β-catenin relative to WT but not MYO10^KD clone 84 which could explain the retained epithelial features observed in these two clones (Fig. 6H). Only MYO10^KO clone 98 had reduced cytoplasmic β-catenin relative to WT clone 38.

To evaluate whether EMT pathway activation observed in MYO10^KO+KD clones resembles that seen in EMT-driven cancer progression, EMT-core scores were calculated from RNA sequencing data of MYO10^KO+KD and MYO10^WT clones. EMT-core scores are generated from gene expression analysis of 130 EMT-associated genes identified by metanalysis of several cancer subtypes [50]. Positive gene expression values were assigned to genes which drive mesenchymal phenotypes and negative values assigned to genes which drive epithelial phenotypes. MYO10^KO clones 98 and 99 had EMT-core scores of 12.783 and 10.992, respectively while MYO10^KD clone 84 had a score of 14.100 (Fig. 6I). MYO10^WT clones 38 and 75 had EMT-core scores of 9.723 and 10.777, respectively, lower than MYO10^KO+KD leader clones. MYO10^KO+KD and MYO10^WT clones had a mean EMT-core score and standard deviation 12.62 ± 1.560 and 10.25 ± 0.7450. Unpaired t-tests of EMT-core scores revealed that the difference between MYO10^KO+KD and MYO10^WT clones was not significant, p = 0.1484 despite our GSEA indicating statistically significant activation of EMT transcriptional pathways (Fig. 6A, I). These observations support a model where both knockdown and knockout of MYO10 increase nuclear levels of SNAIL to activate EMT transcriptional networks while MYO10 knockout reduces nuclear β-catenin.

Discussion

Leader-follower dynamics have been observed in models of collectively migrating lung, ovarian, breast and bladder cancer [18, 20, 22, 26, 64, 65]. However, the molecular mechanisms that define and maintain leader-follower phenotypes remain unclear. Our lab previously identified myosin 10 as a top DEG in H1299 leaders relative to followers and demonstrated that this motor protein drives collective invasion, motility, and filopodia-dependent ECM remodeling [25]. Here, we utilized CRISPR/Cas9 gene editing system to generate MYO10^KO and MYO10^KD leader clones from the H1299 lung large cell carcinoma model. We demonstrate that stable knockdown or complete knockout of MYO10 reduces cell spread area and traction force generation (Fig. 4B-D), activates EMT transcriptional networks (Fig. 6A) and increases nuclear levels of SNAIL, a transcription factor identified to initiate and sustain EMT, (Fig. 6F). Ablation of MYO10 in leader clones increased expression of integrin subunits α6 and β4 (Fig. 3B-D) and decreased N-cadherin expression (Fig. 5B, C). Together, these observations provide evidence that in leader cells MYO10 opposes the activation of EMT transcriptional profiles. The regulation of EMT-associated transcriptional programs by MYO10 may reflect the molecular motor protein’s role in organizing many of the cytoskeletal components and adhesion machinery that support a key feature of the leader cell phenotype, collective invasion.

MYO10 has been extensively studied for its role in filopodia induction and dynamics [30, 60]. Gene ontology analysis of RNA sequencing data revealed that the extracellular matrix pathway (GO:0031012) was enriched between the transcriptional profiles of MYO10^KO+KD and MYO10^KO clones when compared to MYO10^WT leader clones (Fig. 2G). All MYO10^KO+KD leader clones had upregulated expression of fibronectin (FN1) at both the RNA and protein levels (Fig. 6B, D,E). These observations align with other work which demonstrates that MYO10 depletion in breast carcinoma increased the expression of ECM genes and induced an EMT-like state in cancer cells at the periphery of tumors [66]. The increased expression of ECM components in MYO10^KO+KD cell lines may, in part, be due to altered integrin activation and binding affinities at filopodia tips resulting in defective filopodial-ECM sensing [59, 67].

Next, we show increased expression of integrin subunits α6 and β4 was at the transcriptional and protein levels in homozygous MYO10^KO leader clones 98 and 99 (Fig. 3B-D). Although we detected increased cell-surface integrin β4 via the integrin adhesion array (Fig. 3F), we could not detect integrin α6 using this method as the assay does not capture it. This observation indicated that MYO10 is critical for the leader phenotype, as highly invasive leader cells exhibit mesenchymal characteristics but integrin α6β4 provides stable adhesions between epithelial cells and the basement membrane. MYO10^KO clones increased expression of integrin subunits α6 and β4, which may be an adaptation to improve cell-ECM adhesion as MYO10^KO+KD clones exhibit reduced pFAK^Y397, and lower normalized ratios of pFAK^Y397 to total FAK and diminished traction force (Fig. 4D-F). Another plausible hypothesis for the increased expression of α6 and β4 subunits may be a compensatory mechanism for Myo10 loss in the MYO10^KO clones, as integrin α6β4 influences the expression of motility and invasion-associated genes in breast carcinoma [68]. Integrin α6β4 has also been identified as a mediator of cancer cell migration by stabilizing actin-based projections like lamellae and filopodia which supports the hypothesis that the upregulation of α6 and β4 subunits observed in our MYO10^KO leader clones is a compensatory mechanism in the absence of MYO10-dependent actin cytoskeletal remodeling [69].

In the current study, we utilized DNA-based tension probes with cRGD motifs to assess traction force generation in leader cells. The RGD (arginine-glycine-aspartic acid) motif is an amino acid sequence mimicking the integrin binding site on fibronectin and fibrinogen. Depending on the surrounding amino acids and chemical geometry (i.e. linear versus cyclic) of the RGD peptide ligand, the different integrin subtypes display varying binding affinity [70]. Our rationale for including tension probes with the cRGD motif is based on our earlier finding that leader cells directed collective invasion through fibronectin repatterning via MYO10-dependent filopodia [25]. We hypothesized that loss of MYO10 should influence leader cell ability to bind and generate forces on surfaces mimicking the integrin binding site of fibronectin. Our assessment of traction force revealed that MYO10^KO+KD clones have significantly reduced force generation from integrin-ligand binding, which may have resulted from reduced MYO10-dependent activation of FAK (Fig. 4E, F). This observation that both MYO10 knockdown and knockout clones display reduced traction force generation supports the idea that MYO10-dependent filopodia organize fibronectin to promote effective force transmission and facilitate collective invasion.

When compared to follower cells, H1299 leader cells demonstrated increased traction force mediated by the activity of FAK [20]. In our previous work, we detected significant differences in traction forces exceeding 36pN at the periphery of leader cells, not observed in followers, and attributed these differences to increased forces produced by focal adhesions [20]. Notably, when using a TGT probe with a 56pN threshold we do not detect a significant difference in leader and follower traction previously observed (Fig. 4D). This may indicate that we are approaching the maximum integrin-ligand binding force for the heterogenous H1299 leaders such that fewer probes are being sheared and less fluorescent signal is being produced and detected. This discrepancy may also be due to a change from using linear RGD to cyclic RGD probes in this study, as integrins are known to have different binding affinity to each variant [70]. Further, a moderate reduction in integrin subunit β3, which is imperative for cell migration and mechanosensing [71], was observed in the MYO10^KD clone 84 and MYO10^KO clone 99. While the tension generation assay demonstrates changes in mechanical activity of MYO10 KO clones, it may not necessarily reflect the cells’ behavior in 3D, as changes in stiffness of the cell substrate and polarization dynamics in three dimensions may alter mechanotransduction. The increase in integrin subunits α6 and β4, reduction in mechanotransduction markers, and decreased force generation in MYO10^KO clones we report are consistent with prior research demonstrating that loss of integrin α6β4 promotes focal adhesion assembly [72], increases cell spreading, and traction force generation [73].

Initially, we hypothesized that spheroid compaction failure was a consequence of reduced traction force generation due to altered actin filament dynamics in MYO10^KO clones 98 and 99, as force generated by the actin cytoskeleton maintains and stabilizes intercellular cadherin complexes [74] (Fig. 5A). However, this hypothesis failed to explain why MYO10^KD clone 84 formed compact spheroids but exhibited similar traction forces as MYO10^KO clones 98 and 99 (Figs. 4D and 5A). Assessment of cadherin transcription and protein levels demonstrated that MYO10^KO clones 98 and 99 expressed significantly less N-cadherin than MYO10^WT leader clones and all other H1299-derived cell lines, but the underlying mechanism is unknown (Fig. 5B). Significant reduction of N-cadherin may not limit the invasive capacity of H1299 leader cells but limits cell-cell adhesion to abolish the ability for cells to invade as a “collective” unit. Our recent work highlighted that the 4T1 murine breast cancer cell line recapitulated leader-follower invasive dynamics in 3D-culture and exhibited a stable transcriptionally distinct subpopulation termed “singles” lacking E-, P-, and N-cadherin [64]. Singles are invasive but lack the cell-cell connections to support collective invasion and also utilized secreted factors from collectively invading cells to enhance their diminished invasive potential [64]. Therefore, the dramatic reduction of N-cadherin in MYO10^KO clones 98 and 99 (which do not express E- or P-cadherin), may have shifted these leader clones toward a “single-cell-like” phenotype with diminished invasive potential and an absence of collective cell organization.

A plausible explanation for reduced N-cadherin in MYO10^KO clones 98 and 99 was reduced nuclear localization of Twist 1, a transcription factor which drives EMT, as increased nuclear Twist 1 has been shown to enhance N-cadherin expression in prostate cancer in a β1 integrin-dependent manner [75]. However, nuclear protein levels of Twist 1 were increased in MYO10^KO clone 98 and decreased in MYO10^KO clone 99 relative to MYO10^WT clones 38 suggesting that nuclear localization of Twist 1 may not be the primary driver of N-cadherin expression in our model (Fig. 6F; Fig. S5). In endothelial cells, MYO10 transports VE-cadherin along filopodia to initiate cell-cell junction formation [76]. Another potential mechanism explaining this observation is that loss of MYO10 reduced intrafilopodial N-cadherin transport which disrupted the formation of cell-cell junctions, but this requires further investigation. We also observed that MYO10 depletion reduced spheroid invasive area (Fig. 5D, E). Our H1299 leader cells were initially isolated from Matrigel embedded spheroids in 3D and maintained in Matrigel-free 2D culture [20]. ECM cues are observed to initiate feedback loops in basal-like breast cancer cells that support In the studies here, altering the composition of the ECM in our spheroid invasion assays may enhance or suppress the chain-like invasive leader phenotype, The cRGD motif on our tension probes is fibronectin-mimetic, therefore the differential force generation and mechanotransduction in MYO10^KO and MYO10^WT clones may be dramatically different in ECM substrates lacking fibronectin.

Targeting MYO10 in leaders, to reduce N-cadherin expression, may limit metastasis as heterotypic cellular models of NSCLC and cancer associated fibroblasts (CAFs) show that E-cadherin/N-cadherin adhesion facilitates CAFs-guided cancer cell invasion [77]. However, additional in vitro and in vivo studies are needed to confirm this hypothesis and elucidate the relationship between MYO10 and N-cadherin. We confirmed activation of EMT transcriptional pathways and demonstrated enhanced expression of canonical EMT markers fibronectin and vimentin with no significant changes in epithelial markers ZO-1 and claudin 1 in MYO10^KO+KD clones relative to MYO10^WT controls (Fig. 6A, D,E).

SNAIL regulates EMT at the epigenetic level [78, 79]. Our earlier work in the H1299 leader-follower model described significant global epigenetic rewiring of leader cells and integrated RNA sequencing data to identify MYO10 as key mediator of epigenetic heterogeneity and collective invasion in leader cells [25]. Increased nuclear SNAIL observed in MYO10^KO+KD leader clones may drive the transcriptional rewiring in part due to cytoskeletal changes in leader cells as SNAIL-mediated induction of EMT caused altered cytoskeletal gene expression and remodeling in breast cancer cell lines (Fig. 6F) [80]. Therefore, it is reasonable to hypothesize that knockout of MYO10, a cytoskeletal-associated protein, could activate EMT transcriptional networks in MYO10^KO clones via a SNAIL-dependent mechanism. Further, increased expression of SNAIL enhanced the EMT-features of human melanoma cells with decreased cell proliferation, adhesion and increased fibronectin, consistent with our observations in MYO10^KO clones (Figs. 1G, 5A and 6D) [81]. These observations in conjunction with global transcriptional changes in shown in GSEA of MYO10^KO+KD clones activating regulation of epithelial to mesenchymal transition (GO:0010717) and epithelial to mesenchymal transition (GO:0001837) pathways suggest that MYO10 may activate EMT transcriptional pathways in a SNAIL-dependent manner (Fig. 6A).

Relative to MYO10^WT leader clones, the MYO10^KO+KD clones 84, 98 and 99 had increased EMT-core scores in alignment with observed activation of EMT transcriptional networks and a shift towards a more mesenchymal state along the EMT spectrum (Fig. 6I). The EMT transcriptional activation and requisite increase in EMT-core scores observed in MYO10^KO+KD clones may result from compensatory upregulation of mesenchymal genes (and downregulation of epithelial genes) to restore the homeostatic invasive leader state. Interestingly, the gene list used to generate EMT-core scores included genes which clustered with significantly EMT gene expression signatures but currently have undefined roles in EMT and cancer potentially expanding what we currently define as EMT-related genes [50]. A partial epithelial-to-mesenchymal transition (EMT) state enhances invasion and cancer progression [82]. The enhanced epithelial features (decreased N-cadherin and increased α6 and β4) observed in MYO10^KO clones 98 and 99 suggest a novel role for MYO10 suppression of epithelial cell mechanisms that limit mesenchymal cell-cell adhesion necessary for motile pack organization and oppose formation of dynamic focal adhesions required for collective invasion. As increased nuclear β-catenin is associated with EMT activation, retained epithelial features in MYO10^KO clones may be explained by the reduction in nuclear β-catenin observed in the MYO10^KO clones but this warrants further investigation [83]. Recent studies show that EMT is not only increased at the periphery of tumor borders [84] but also associated with reduced CD4 + and CD8 + T cell infiltration in NSCLC [85] underscoring the necessity for deeper investigation of factors that drive EMT activation in cancer settings. Understanding how MYO10 influences the EMT state of leader cells beyond the transcriptional profile may provide insight into mechanisms driving tumor invasion and could help identify therapeutic targets that modulate EMT-related features in this invasive subpopulation.

Despite rigorous experimental design and analysis, there are several limitations of this study exist. First, the leader-follower model used to interrogate the role of MYO10 was isolated from a single large cell lung carcinoma line (H1299) [20], which may not capture the molecular or phenotypic heterogeneity of leader-follower dynamics either within large cell carcinoma or among other lung cancer subtypes. Leader and follower subpopulations have been reported in breast [23, 64], bladder [26], and ovarian cancers [65], and new biomarkers for leader and follower cells across several lung cancer cell lines have been are being defined [86, 87]. Additional comparative analyses across multiple lung cancer models and subtypes are essential to determine whether MYO10 expression defines or supports the leader phenotype and to identify the drivers of collective invasion that underlie lung cancer metastasis.

Second, although the CRISPR-Cas9 system used to perturb MYO10 expression was designed to minimize off-target activity, potential off-target effects were not formally assessed. Utilizing unbiased double-strand break identification tools like GUIDE-Seq [88] could strengthen causal inference of the data by determining whether CRISPR nucleases introduced additional genomic mutations beyond MYO10 disruption. In an attempt to minimize this limitation, the gRNA against MYO10 was designed and optimized to minimize, if not eliminate, off-target effects [89] reducing the likelihood off-target effects influencing our observed phenotype in the setting of MYO10 depletion. Further, the use of single-cell derived leader clones allowed comprehensive functional interrogation of MYO10, however, the heterogeneity inherent to the H1299 parental population, and subsequently leader cells, may have contributed to the clonal variation observed between MYO10 knockdown and knockout clones [90]. We attempted to mitigate this issue by analyzing multiple independent clones per MYO10 knockout and wildtype conditions and incorporating pooled analyses in this work.

While our in vitro analyses identified MYO10-dependent mechanisms that define the leader cell phenotype, a key future direction will be to determine whether these observations in vitro translate to differences in metastatic behavior in vivo. Developing luciferase-expressing MYO10^KO, MYO10^KD and MYO10^WT clones for bioluminescent imaging of orthotopic tumor grafts could be visualized for primary tumor progression and metastatic dissemination. Given that MYO10 depletion disrupts collective invasion, reduces proliferation and N-cadherin expression, alters the integrin profile and diminishes leader cell traction forces, we anticipate that in vivo MYO10-deficient and wildtype tumors will exhibit different metastatic patterns, tumor burdens and survival rates. These studies facilitate deeper understanding of how both partial and complete loss of MYO10 affects collective invasion and in the future will expand our knowledge on the role of leader cell MYO10 in metastatic patterning, tumor burden and survival in vivo.

Conclusions

Integrating CRISPR-Cas9, FACS, western blot and Sanger sequencing to generate and isolate MYO10^KO+KD clones was feasible since H1299 leader-follower subpopulations stably and binarily express either MYO10 or IL13Rα2 biomarkers. However, expanding this methodology across several cancer cell lines with subpopulations that demonstrate distinct invasive capacities may be limited as these subpopulations may not express stable, discrete markers to facilitate their separation from the parental cell line. Further, clonal cell lines facilitated detailed investigation of MYO10’s role in leader cell biology. Nevertheless, minimizing heterogeneity within the H1299 leader cell population may not fully recapitulate the effects of targeting MYO10 in more complex models.

In summary, we generated MYO10^KD, MYO10^KO, and MYO10^WT leader clones within the H1299 large cell carcinoma model. We show that MYO10 knockdown leader cells exert decreased cellular traction force, diminished cell spread area and activation of dynamic cell adhesion machinery while activating of EMT transcriptional networks. Furthermore, MYO10 knockout clones also demonstrate increased integrin α6 and β4 subunit expression, decreased N-cadherin and impaired spheroid compaction (Fig. 7A). We also show paradoxical activation of EMT transcriptional networks in MYO10^KO leader clones are influenced, at least in part, by alterations in nuclear SNAIL and β-catenin which suggest a novel role for MYO10 as a potential modulator of EMT transcriptional programs within our model (Fig. 7A). Our earlier models of invasion in H1299 cells proposed that MYO10 enhanced leader cell collective invasion through filopodia-driven ECM remodeling and atypical VEGF-signaling [20, 25]. The results of this study build on those existing models and assert that MYO10 suppresses transcriptional activation of EMT pathways and integrin α6 and β4 subunit expression and maintains N-cadherin expression to support collective invasion in H1299 leader cells.

Fig. 7.

Proposed model of MYO10 function in maintaining leader cell phenotype

Abbreviations

EMT

Epithelial–to–mesenchymal transition

MYO10

Myosin 10

MYO10^KO

Myosin 10 knockout

MYO10^KD

Myosin 10 knockdown

MYO10^WT

Myosin 10 wildtype

MYO10^KO+KD

Myosin 10 knockout and knockdown

ECM

Extracellular matrix

IF

Intermediate filaments

JAG1

Jagged 1

gRNA

guide RNA

sgRNA

single guide RNA

EMT

TFs–epithelial–to–mesenchymal transition transcription factors

NSCLC

Non–small cell lung cancer

IL13Rα2

Interleukin–13 receptor subunit alpha–2

FACS

Fluorescence–activated cell sorting

gDNA

Genomic DNA

TGT

Tension gauge tether

TRITC

Tetramethylrhodamine isothiocyanate

RICM

Reflection interference contrast microscopy

CRISPR/Cas9

Clustered regularly interspaced short palindromic repeats/CRISPR–associated protein 9

HTATIP2

HIV–1 Tat Interactive Protein 2

PCA

Principal component analysis

DEGs

Differentially expressed genes

GO

Gene ontology

cRGD

Cyclic RGD

FAK

Focal adhesion kinase

WT

Wildtype

KO

Knockout

KD

Knockdown

ECM

Extracellular matrix

Authors’ contributions

Conceptualization: R.G., M.V., H.O., A.I.M.; Methodology: R.G., M.V., H.O.; Data Analysis: R.G., M.V., H.C., R.A.; Software: R.G., H.C., R.A.; Investigation: R.G., M.V., H.O.; Writing – original draft: R.G.; Writing – review and editing: R.G., .M.V., H.O., A.A., H.C., R.A., S.Y., A.P., K.S., A.I.M; Visualization: R.G., M.V.; Supervision: R.G.; H.O., K.S., A.I.M; Project administration: R.G., C.T.B., K.S., A.I.M.; Funding acquisition: A.I.M and K.S.

Funding

This project is funded by National Institutes of Health (NIH) grants R01CA250422 (to A.I.M.), R01CA247367 (to A.I.M.), and P01CA257906 (to A.I.M.). This work was also supported by the National Science Foundation Graduate Research Fellowship Program (to M.V.).

Data availability

The RNA sequencing data supporting the conclusions of this manuscript were submitted to the GEO data repository under accession number GSE297208 and are available at the following link: [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE297208](https:/www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE297208) .

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.