Melanoma cell adhesion molecule (MCAM) mediates microtentacle generation, homotypic clustering, and reattachment of non-adherent breast tumor cells

Abstract

Background

Melanoma cell adhesion molecule (MCAM) was first identified in advanced primary tumors and metastatic lesions. MCAM expression has been shown in previous studies to promote aggressive cancer phenotypes, including migration, invasion, tumorigenesis, and induction of vimentin and slug, indicative of EMT. Furthermore, overexpression of MCAM has been linked to poor prognosis and aggressive metastatic phenotypes, particularly in triple-negative breast cancer (TNBC). Early work has shown that MCAM can alter the actomyosin cytoskeleton and intermediate filaments; however, the impact of MCAM on the microtubule network and non-adherent cells has not been well studied.

Methods

Utilizing MCF-10A breast epithelial cell line engineered to overexpress MCAM and CRISPR MCAM knockouts in TNBC cell lines, we assessed the impact of MCAM on tubulin-driven cytoskeletal phenotypes in non-adherent conditions. TetherChip technology, xCelligence RTCA, and immunoblotting techniques were utilized to investigate the impact of MCAM overexpression.

Results

Our results show that MCAM overexpression increases vimentin protein levels, α-tubulin post-translational modificaitons (PTMs), microtentacle (McTN) protrusions, homotypic cell clustering, cellular reattachment, and migration. These phenotypic increases can be inhibited with FDA-approved vinorelbine treatment. Conversely, the knockout of MCAM in TNBC cell lines sustains decreases of acetylated α-tubulin, implicating that MCAM expression may alter the microtubule cytoskeleton’s stability directly. Knockout (KO) of MCAM also decreased microtentacle protrusions and McTN-supported phenotypes of homotypic cell clustering and cellular reattachment. Migration was also decreased in TNBC with MCAM KO.

Conclusions

These findings suggest that MCAM can function through the microtubules in TNBC to impact McTN-supported phenotypes of homotypic cell clustering and cellular reattachment in non-adherent breast tumor cells.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13058-026-02346-0.

Introduction

Breast cancer treatments include surgery, radiation, targeted therapy and/or chemotherapy to eliminate the primary tumor. However, approximately 90% of cancer-related deaths are due to metastasis and not the primary tumor, highlighting the need for better therapeutic strategies that target metastasis. Metastasis is a complex, multistep process with several rate-limiting steps. Cancer cells that successfully disseminate to and regrow in distant tissues can be molecularly different from the primary tumor [1, 2]. Therefore, understanding the mechanisms that drive metastasis is crucial to developing better therapies [3–5].

During early metastasis, tumor cells can migrate away from the primary tumor, invade into the surrounding extracellular matrix (ECM), and enter the bloodstream [4–7]. Most circulating tumor cells (CTCs) do not survive the inhospitable environment of the bloodstream due to detachment from the ECM, immune surveillance, and shear stress [4, 6, 7]. Upon CTC entry into the bloodstream platelets are among the first cells for which CTCs interact followed by immune cells. Activated platelets, neutrophils, and double-positive T-cells have been shown to attach to CTCs in part due to cell adhesion molecule interactions, protecting the CTCs from shear stress and natural killer cell surveillance [8–12]. Platelets, neutrophils, and double-positive T-cells also provide growth factors and cytokines to CTCs while in the bloodstream, promoting invasiveness, proliferation pathways, and mesenchymal characteristics [8–12]. Therefore, CTC clusters (2 or more cells) have a higher probability of survival than single CTCs [6, 7, 13, 14]. Specifically, the metastatic potential of homotypic and heterotypic CTC clusters is increased up to 23–50-fold, correlating to a decrease in cancer patient survival [6, 7, 14]. Identifying the mechanisms that promote CTC clustering may aid in identifying targets for future therapeutics to prevent metastatic spread.

The melanoma cell adhesion molecule (MCAM) is an immunoglobulin transmembrane protein that was first identified in advanced primary melanoma and metastatic lesions [15–17]. Although its expression is typically found in embryonic tissues, MCAM is maintained in adult tissues, primarily in smooth muscle, vascular tissue, pericytes, vascular endothelial cells, and subsets of leukocytes [15, 18]. MCAM has been implicated in the metastasis of multiple cancer types [15–17, 19, 20], and its overexpression has been strongly linked to breast cancer, particularly to the triple-negative breast cancer subtype [16, 21, 22]. In breast cancer, increased MCAM expression has been shown to regulate epithelial-to-mesenchymal transition (EMT) and enhance migration, invasion, metastasis, and tumorigenesis [15–17, 19–27]. Increases in MCAM expression are also associated with decreased overall breast cancer patient survival [15, 16, 21, 28].

MCAM expression has been shown to contribute to primary tumor progression and multiple steps in the metastatic cascade [15, 17, 20–22, 24]. However, MCAM’s impact on tumor cells post-intravasation and CTCs in the bloodstream remains unknown. Data suggest MCAM-mediated adhesion and signaling is stimulated through heterophilic interactions with various ECM ligands, cell membrane receptors, and potentially β1 integrins [15, 18, 29, 30]. Therefore, its expression could promote cell-to-cell adhesion, enhancing CTC cluster formation and reattachment to blood vessel walls.

Our lab discovered microtentacles (McTNs), unique tubulin-driven membrane protrusions produced by epithelial cells after ECM detachment [13, 31–41]. McTNs are highly dynamic and consist of post-translational modifications (PTMs) of detyrosinated and acetylated α-tubulin, as well as the intermediate filament and EMT marker, vimentin [31, 34, 38]. We have shown that McTNs mediate homotypic clustering of detached cancer cells, heterotypic clustering of cancer and immune cells (neutrophils), and cellular reattachment from a non-adherent state [31–33, 35–40].

Adhesion molecules are often upregulated in tumors and aid in cluster formation [42]. As MCAM is overexpressed in many cancers, facilitates adhesive interactions and signaling cascades, we hypothesized that MCAM could aid CTC clustering and reattachment, thereby promoting metastatic phenotypes. Using MCAM overexpression and CRISPR knockout systems, we determined, for the first time, that MCAM expression alters microtubules through post-translational modifications and vimentin expression, modulates McTN formation, cell clustering ability and cellular reattachment.

Materials and methods

Cell lines and reagents

All cell lines were purchased from ATCC (Manassas, VA). The MCF-10A cells were maintained as previously described [43]. MDA-MB-468 (MDA-468) cells were maintained in Dulbecco’s Modification of Eagle’s Medium (DMEM, Corning, Cat: 10–013-CV) supplemented with 10% fetal bovine serum (R&D Systems, Cat: S11150H), 1% penicillin–streptomycin (Gemini, Cat: 400–109), and 0.1% human insulin solution (Sigma-Aldrich, Cat: 11,061–68-0). For the Tumor Derived (TD) cell lines, MDA-MB-231 (MDA-231) and MDA-MB-436 (MDA-436) were injected into the 4^th mammary gland of female NSG mice. Cells were extracted from the resulting tumors and regrown in culture as TD cell lines and maintained as previously described [32]. Cells were treated with trypLE Express Enzyme (1X) (ThermoFisher Scientific, Cat: 12,604,013) for cell dissociation. Vinorelbine tartrate (BioVision, Cat: 1957‐5) was resuspended in PBS and used at a final concentration of 10 µM. Okadaic acid (MCE, Cat: HY-N6785) was resuspended in ethanol and used at a final concentration of 1 µM. Cells were exposed to vehicle control, vinorelbine, or okadaic acid and maintained at 37 °C with 5% CO₂ for 1 h. Detailed information on each drug can be found in Table S1 (Supporting Information).

Cell line generation

MCF-10A and MDA-468 cells were infected with lentivirus CD146 (MCAM) (Origene, Cat: RC208937L4V) or GFP at an MOI of 10 in the presence of polybrene (8 µg/ml) as previously described [32]. The MCF-10A MCAM OE (overexpressing) cells were maintained in MCF-10A growth media plus 0.5 µg/ml puromycin and the MDA-468 GFP and MCAM OE cells were maintained in their respective growth media plus 5 µg/ml puromycin. UMGCCC Translational Core Facility at the University of Maryland, Baltimore used the MCAM Human Gene Knockout Kit (Origene, CRISPR, Cat: KN208937) to knockout MCAM in the MDA-231TD and MDA-436TD cell lines. Single cell clones were screened to confirm complete knockout of MCAM. Knockout clones 6 (KO6) and 11 (KO11) were selected for the MDA-231TD cells, and clones 12 (KO12) and 13 (KO13) were selected for the MDA-436TD cells.

CellTiter aqueous cell proliferation assay

MCF-10A GFP and MCAM OE cells were plated at 5000 cells/well in 100 µL in a 96‐well plate (Corning Cat: 3474). After 24 h, cells were treated with either 10 µm vinorelbine (1 h pre-treat, kept in media for entire experiment) or its vehicle control, 1 µm okadaic Acid (1 h pre-treat then replaced with fresh media after pre-treat) or its vehicle control, or 1 µm Staurosporine in triplicate. CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega, Cat: G3581) was used according to the manufacturer's instructions. Briefly, at the indicated time points, 20 µL of reagent was added to each well and maintained at 37 °C with 5% CO₂ for 1 h. Absorbance was measured at 490 nm using the CLARIOStar Plus plate reader (BMG Labtech), and values were normalized to t = 0.

Trypan blue exclusion assay

To measure cell viability and proliferation, all cells were plated on day 0 and counted each day for 5 days. On the appropriate day, cells were dissociated from the fish with TrypLE Express Enzyme resuspended in culture media, and mixed 1:1 with Trypan blue dye. Viability and cell number were calculated on a Countess II FL Automated Cell Counter (Thermo Fisher, Cat: AMQAX1000). Values were graphed in GraphPad Prism software and statistical analysis was performed.

McTN counting and perimeter analysis

Cells were seeded onto a TetherChip microfluidic slide (Ibidi µ‐Slide I Luer 0.8, Cat: 80,191) [44]. Cells were incubated for 45 min to allow for tethering. After 45 min, 3.7% formaldehyde/PBS was washed through each channel to fix cells for 15 min. Wheat germ agglutinin (WGA, Alexa Fluor 594 Conjugate, Invitrogen, Cat: W11262) was then added to each channel at a final concentration of 1:100 to allow for the visualization of the cell membrane and Hoechst 33,258 (1:1000) was used to stain the nucleus. McTNs were scored in a population of 100 cells/channel as previously described [32, 33, 44] on a Nikon Ti2‐E inverted microscope at 60 × magnification. Images for perimeter traces were acquired on a Nikon AX‐R confocal system at 60 × magnification. Perimeter image analysis was quantified in ImageJ as previously shown [44] and used as a secondary measure for McTNs.

Immunofluorescence

Cells were loaded into an Ibidi treat 8 chamber slide (Ibidi µ‐Slide 8 well, Cat: 80,826) and incubated for 48 h to allow for growth. The cells were then fixed with a 3.7% formaldehyde/PBS solution for 15 min, followed by two 5-min washes in PBS. Cells were then permeabilized in 0.1% Triton‐X 100 diluted in PBS, blocked in 5% bovine serum albumin (BSA) diluted in PBS. Cells were incubated overnight at 4 °C in 2.5% BSA/PBS with primary antibodies against α‐tubulin (1:1000, Sigma‐Aldrich, Cat: T6199) and vimentin (EPR3776) (1:1000, Abcam, Cat: ab92547). Secondary antibodies, anti‐mouse Alexa Fluor 647 (1:1000, Invitrogen, Cat: A‐11,001) and Hoechst 33,258 (1:1000, Invitrogen, Cat: H1398) were diluted in PBS, added to each well, and incubated for 2 h at room temperature. Finally, wells were washed with PBS and mounted with Fluoromount‐G (Invitrogen, Cat: 00‐4958‐02). Images were acquired using a Nikon Ti2‐E inverted microscope with a Nikon AX‐R confocal system with identical exposure and laser settings in each experiment replicate.

Immunoblotting

Cells were collected in 1:1 ratio of SDS and RIPA buffer containing 0.1% phosphatase inhibitor and 1% protease inhibitor. Lysates were sonicated for 5 s bursts, followed by heating for 5 min at 96 °C. Protein concentration was measured using the BioRad DC protein assay (Cat: 5,000,112) according to manufacturer’s instructions. Total protein (15 µg) was separated by SDS‐PAGE on NuPage 4–12% Bis‐Tris protein gels (Invitrogen, Cat: NP0335BOX) and then transferred to PVDF membranes using the eBlot L1 Fast Wet Transfer System (Genscript, Cat: L00727). Membranes were blocked in 5% BSA or 5% non-fat dry milk in TBS with 0.1% Tween‐20 (TBS‐T) for 1 h at room temperature followed by an overnight incubation at 4 °C in detyrosinated α‐tubulin (1:10,000, RevMAb Clone RM444, Cat: 31‐1335‐00), acetylated α‐tubulin (1:1000, Cell Signaling Technologies, Cat: 5335S), α‐tubulin (1:1000, Sigma‐Aldrich, Cat: T6199), vimentin (EPR3776) (1:1000, Abcam, Cat: ab92547), MCAM (EPR3208) (1:1000, Abcam, Cat: ab75769). or GAPDH (1:5000, Santa Cruz Cat: SC32233) antibody in 5% BSA or 5% non-fat dry milk diluted in TBS‐T. Detailed primary antibody information available in Table S2. Secondary antibodies to IgG conjugated to rabbit and mouse antibodies were used (1:5000, Cat: 711‐035‐152 and 715‐035‐150, Jackson ImmunoResearch Laboratories) and positive antibody binding was visualized using ECL chemiluminescence. Blots were imaged on the iBright Imaging System (Thermo Fisher) using the smart exposure feature. Densitometry analysis was performed with the iBright Analysis Software and transferred to GraphPad Prism for analysis.

Homotypic cell clustering assay

Cells were added (20,000 cells/well) in a low-attach 96-well plate to allow for aggregation for 4 h after corresponding drug treatments. After suspension, cells, including any remaining single cells and aggregates, were collected and tethered onto a TetherChip and incubated at 37 °C for 45 min. Following the incubation, cells were fixed using 3.7% formaldehyde/PBS, and after 15 min stained with Hoechst 33,258 (1:5000). Whole channel scanned images were acquired and stitched using the Nikon Ti2‐E inverted microscope at 4 × magnification. Image analysis was performed in ImageJ (Fiji) using automatic thresholding, followed by particle analysis to quantify cluster number and size (area). Objects smaller than 75 µm² (smaller than the size of a single nucleus) and larger than 200,000 µm² were considered debris and thus not analyzed. From the binarized threshold images, a measured clustering efficiency is obtained by comparing the number of individual clusters at the initial plating time point and at the aggregation end time point.

Cell-electrode impedance attachment and migration assay

Real-time, dynamic monitoring of cellular reattachment from suspension was measured using the xCELLigence RTCA DP analyzer (Agilent) to compare cell attachment rates after pretreatment with vinorelbine or vehicle control [32]. All cells were collected and seeded (40,000 cells/well) in quadruplicate or triplicates into 16-well microelectronic sensor standard plates (E-plates, Agilent) Cell-electrode interactions occur on microelectronic sensor arrays and produce an arbitrary measurement unit, cell index, which indicates the relative amount of cellular reattachment. Impedance was measured every 5 min, for a 24 h period. Values expressed are the mean ± SD of quadruplicate/triplicate wells.

For the migration assays, 5% horse serum for the MCF-10A or 10% FBS for the MDA-231TD and MDA-436TD was added as the chemoattractant to the bottom chamber. Cells were collected, counted, and added to the upper chamber of the CIM plates (Agilent) in serum-free media [32, 33]. Impedance was measured every 15 min for 72 h. Upon completion of the experiment, the CIM cartridges were taken apart, the membranes on the underside of the upper chamber were fixed with 3.7% formaldehyde/PBS solution, stained with Hoechst (1:1000) and WGA (1:100) for 30 min, and then a glass coverslip was mounted on top for imaging.

Time-lapse microscopy

MCF-10A, MDA-231TD, and MDA-436TD OE, KO, and parental cells were plated on 24-well plates (Falcon, Cat: 53,047) at 250,000 cells/well. Phase contrast images of cells reattaching were captured every 5 min for 24 h. The images shown are at the 0- and 4-h time point since most of the cells had reattached by then. Images were taken at 10 × on a Nikon Ti2‐E inverted microscope with a Tokai Hit stage top incubator.

KM plotter survival curves

The breast cancer gene chip mRNA patient data set was selected, and the gene symbol “MCAM 211340_s_at” was used for all survival curves generated. [45] For all breast cancer subtype survival curves, no subtype-specific restriction was applied; each subtype included “all”. Split patients by percentiles of “auto select best cutoff” and OS selected for survival. For survival curves with triple-negative breast cancer subtypes, restrictions were as follows: negative IHC and array ER status, negative PR array status, and negative HER2 array status. Percentile auto-select best cutoff was used for TNBC subtype, with either OS or RFS selected for survival.

Statistical analysis

GraphPad Prism (version 9/10) was used to perform all statistical comparisons. One‐way and two‐way ANOVA tests were performed, with the GraphPad-recommended multiple comparisons post hoc test as indicated. Two‐tailed t‐tests were performed when indicated. A p‐value of 0.05 or less was considered statistically significant.

Results

MCAM expression promotes microtentacle formation

We analyzed patient gene chip data sets from KM Plotter and found that high expression of MCAM correlated to worse patient outcomes. KM Plotter analysis shows breast cancer patients (all subtypes) with high expression of MCAM is associated with decreased overall survival (Fig. 1A). Additional analysis from KM Plotter revealed that high expression of MCAM in breast cancer patients with the triple negative subtype (TNBC) has significantly reduced recurrence-free survival (Fig. 1A) and a decrease in overall survival that is trending towards significance (Supp.Fig. 1A) [45]. Since high MCAM correlates with worse patient outcomes, we hypothesized that MCAM expression increased with breast cancer aggressiveness. Using multiple tumorigenic and metastatic breast cancer cell lines, as well as the non-tumorigenic MCF-10A mammary epithelial cell line, we examined MCAM expression. MCAM was more highly expressed in cells with increased tumorigenicity and aggressiveness, particularly in the TNBC lines MDA-436TD, MDA-157, and MDA-231TD (Fig. 1B). We also observed increases in detyrosinated and acetylated tubulin PTMs, as well as in vimentin, in the more tumorigenic and metastatic breast cancer cell lines, which suggests increased stability, leading to increased McTN formation, and potentially increased functionality (Fig. 1B). We examined McTN production in the same mammary epithelial MCF-10A cell line and tumorigenic and metastatic breast cancer cell lines. As breast cancer cells become more aggressive, we observe an increase in McTN production within the population (Fig. 1C, D). Thus, McTNs are positively correlated with cells that have increased tumorigenicity and metastatic efficiency.

Fig. 1.

Increased MCAM expression is linked to decreased survival in breast cancer patients and correlated to McTN frequency A Gene Chip analysis of altered MCAM mRNA in breast cancer patients from KM Plotter. Overall survival of breast cancer patients, including all subtypes. Auto select best cutoff for percentiles was used (left). Gene Chip analysis of altered MCAM mRNA in breast cancer patients from KM Plotter. High vs. low MCAM mRNA expression and recurrence-free survival in TNBC patients (right). B Immunoblot analysis of breast cancer cell line panel probed for MCAM, detyr α-tubulin, acetyl α-tubulin, vimentin, total α-tubulin, and GAPDH. C Representative McTN images of MCF-10A and MDA-231TD cells tethered onto TetherChips and stained with WGA. Images were taken at 60 × magnification with 3 zoom using Nikon Ti2-E inverted microscope. Images were denoised in a post-processing step using NIS Elements. Scale bar = 10 μm. D McTN quantification done on a panel of breast cancer cell lines tethered onto TetherChips for 45 min. Data represents quantification of McTN frequency from three independent experiments with 100 cells counted for each. Error bars indicate ± standard deviation, n = 3

Since McTN production and MCAM expression increased with breast cancer cell aggressiveness, we examined whether MCAM expression promotes McTNs. We first overexpressed MCAM in the MCF-10A and MDA-468 cells with low endogenous MCAM levels and low McTNs. Exogenous MCAM was expressed an average of 40-fold higher than the GFP expressing control in both MCF-10A (Fig. 2A) and MDA-468 cells (Supp.Fig. 2A). In MCF-10A cells, MCAM overexpression resulted in decreased cellular growth, where an increase was observed in the growth of the MDA-468 MCAM OE cells (Supp.Fig. 1B, C). To determine whether MCAM expression altered McTNs, the microfluidic cell-tethering technology (TetherChip) was used to visualize and analyze McTN positivity in suspended cells [44]. While an average of 20% of the GFP control cells produced McTNs, the percentage of the cell population that produced McTNs in the MCAM OE cell increased to 55% (Fig. 2B–D). The MDA-468 GFP cells average roughly 10% McTN positivity, and MCAM OE significantly increased that to 20% (Supp.Fig. 2B). To measure the increase in McTNs via a secondary method, cell perimeter was also assessed. The MCAM OE cells showed an increase in cell perimeter compared to the GFP controls, indicative of increased McTN production (Fig. 2C). The increase in perimeter indicates that the MCAM OE cells may have either more McTNs per cell, more cells in the population that produce McTNs, and/or longer McTNs. Unfortunately, the membrane of MDA-468 cells exhibits severe blebbing and ruffling when anchored in a suspended state on the TetherChip, posing challenges for perimeter analysis (Supp.Fig. 2C).

Fig. 2.

Increased MCAM expression enhances McTN frequency, vimentin, and tubulin PTMs A Immunoblot analysis (left) of MCF-10A GFP and MCAM cell lines probed for MCAM and GAPDH. Densitometry fold change analysis (right) for MCAM protein expression. Error bars indicate mean ± standard deviation, n = 3 (right). ** p < 0.01 (Two-tailed t-test). B McTN quantification of tethered and fixed MCF-10A GFP and MCAM OE cells. Data represents quantification of McTN frequency from three independent experiments with 100 cells counted for each. Error bars indicate ± standard deviation, n = 3. ***p < 0.001 (Two-tailed t-test). C MCF-10A controls and MCAM OE cells analyzed for cellular perimeter altered by McTN positivity of at least 30 individual cells. Error bars indicate ± standard deviation, n = 3. **p < 0.01 (Two-tailed t-test). D McTN images of MCF-10A GFP and MCAM cells stained with WGA. Images were taken at 60 × magnification with 3 zoom using Nikon Ti2-E inverted microscope. Images were denoised in a post-processing step using NIS Elements. Scale bar = 10 μm. E Vimentin and GAPDH (top) in MCF-10A GFP and MCAM OE cells measured via immunoblotting analysis. Densitometric fold change analysis (bottom) for vimentin protein expression. Error bars indicate ± standard deviation, n = 3. **p < 0.01 (Two-tailed t-test). F) Detyrosinated α-tubulin, GAPDH, and α-tubulin (top) in MCF-10A GFP and MCAM OE cells measured via immunoblotting analysis. Densitometric fold change analysis (bottom) of detyrosinated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3. ****p < 0.001(Two-tailed t-test). G) Acetylated α-tubulin, GAPDH, and α-tubulin in MCF-10A GFP and MCAM (top) cells were measured via immunoblotting analysis. Densitometric fold change analysis (bottom) of acetylated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3. *p < 0.05 (Two-tailed t-test).

MCAM promotes vimentin expression and alters microtubule post-translational modifications

Using MCAM OE cells, we have shown that MCAM can regulate McTNs production. Mechanistically, our lab has determined that McTNs are promoted and/or supported by α-tubulin PTMs, specifically detyrosination and acetylation, as well as the intermediate filament and epithelial-to-mesenchymal transition (EMT) marker, vimentin [31, 34, 38]. To elucidate the mechanism by which MCAM promotes McTN formation, α-tubulin detyrosination, α-tubulin acetylation, and vimentin were assessed. Vimentin was significantly upregulated sixfold (Fig. 2E) in the MCF-10A MCAM OE cells, supporting previous work showing that overexpression of MCAM in an alternative breast cancer cell line, MCF-7, promotes vimentin expression and EMT markers [16, 21, 24, 25, 46]. Detyrosinated α-tubulin was significantly elevated by approximately threefold (Fig. 2F) while acetylated α-tubulin decreased (Fig. 2G). Meanwhile, detyrosinated α-tubulin and vimentin were slightly elevated in the MDA-468 MCAM OE cells but did not reach statistical significance (Supp.Fig. 2D, E). Interestingly, acetylated α-tubulin was significantly increased by 0.5-fold in the MDA-468 MCAM OE cells compared to the GFP control (Supp.Fig. 2F) indicating that MCAM effects can be cell type specific.

MCAM-induced microtentacles promote homotypic clustering and cell reattachment

To determine whether the mechanism of MCAM-induced McTNs was due to microtubule stability through increased detyrosinated α-tubulin levels or increased vimentin expression, vinorelbine (VR) and okadaic acid (Oka) were used. Vinorelbine functions through binding to the vinca domain of β-tubulin subunits, which promotes depolymerization of microtubules and is used as a control to verify that the McTNs are microtubule-based (Supp.Fig. 3A, B). Cells treated with vinorelbine are unable to produce McTNs. [32] Okadaic acid is a highly selective protein phosphatase inhibitor of protein phosphatases 1 and 2 that has previously been shown to promote vimentin disassembly (Supp.Fig. 4A, B) [34]. If MCAM increasing vimentin expression or detyrosinated α-tubulin is responsible for elevating McTN production in the MCAM OE cells, we would expect a decrease upon either treatment. As expected, vinorelbine treatment inhibited McTNs in MCAM OE cells, which averaged 58% McTN production to 5% after vinorelbine treatment (Fig. 3A). Treating the MCAM OE and GFP MCF-10A cells with okadaic acid decreased McTN production similarly (Supp.Fig. 5A).

Fig. 3.

Vinorelbine reverses MCAM’s impact on McTN-mediated phenotypes A McTN quantification of tethered in the TetherChip and fixed MCF-10A GFP and MCAM cells treated with vehicle control or vinorelbine (10 μM). The data represent the quantification of McTN frequency from three independent experiments, with 100 cells counted in each. Error bars indicate ± standard deviation, n = 3. **** p < 0.0001(One-way Anova with Tukey post-test). B Clustering efficiency analysis on the number of clusters that formed for vehicle-treated and vinorelbine-treated (10 μM) MCF-10A GFP and MCAM OE cells over the 4 h period in low-attachment 96-well plates compared to those immediately tethered in the TetherChip. T = 0 values were divided by the respective t = 4 h number of clusters formed. Error Bars indicate ± standard deviation, n = 3. * p < 0.05 (Two-way Anova with Tukey post-test). C Representative images of vehicle-treated and vinorelbine-treated MCF-10A GFP and MCAM OE cells that were allowed to cluster for 4 h in 96-well low-attachment plates and then tethered onto TetherChips and stained with Hoechst. Scale bar = 500 μm. D Representative graph of reattachment efficiency assay of vehicle-treated vs. vinorelbine-treated MCF-10A GFP and MCAM OE cells. Error bars indicate ± standard deviation, n = 3. ** p < 0.01 versus GFP and MCAM + VR (Two-way Anova with a Dunnett post-test). E Representative phase contrast still frames of vehicle-treated and vinorelbine-treated MCF-10A GFP and MCAM cells attaching onto 24-well plates, t = 4 h. Images were taken at 20 × magnification on a Nikon Ti2-E inverted microscope with a stage-top Tokai-hit incubator chamber. Scale bar = 50 µm.

McTNs promote metastatic phenotypes, including homotypic and heterotypic tumor cell aggregation/clustering and endothelial cell reattachment. [10–14, 19, 32] Since MCAM OE promotes McTN production and McTNs promote cell clustering, we sought to investigate whether MCAM would then increase the formation of homotypic cell clusters. We also sought to determine if alterations to cell clustering due to MCAM expression were dependent on McTNs, so we utilized vinorelbine and okadaic acid. The MCF-10A GFP control and MCAM OE cells were allowed to cluster for 4 h, then loaded onto TetherChips and stained with a nuclear dye (Hoechst) for visualization and analysis. MCAM OE cells had higher clustering efficiency, indicating that increased MCAM expression enhances clustering efficiency (Fig. 3B, C & Supp.Fig. 5B, C). Furthermore, we demonstrate that treatment of MCAM OE cells with either vinorelbine or okadaic acid decreased both homotypic clustering efficiency and average cluster size, indicating that MCAM can mediate clustering through McTNs (Fig. 3B, C & Supp.Fig. 5B, C).

Our lab has determined that McTNs can influence metastatic reattachment of detached tumor cells in vivo.[32, 36, 39] Using the xCelligence platform, we found that MCAM OE cells reattached significantly faster and spread more than GFP control cells (Fig. 3D). Time-lapse phase contrast images were taken in tandem to reinforce that the increase of MCAM expression in MCF-10A corresponds to an increase in attachment (Fig. 3E). Similar to homotypic cell clustering, we sought to determine whether disrupting McTNs with vinorelbine and okadaic acid could reverse any changes in cellular reattachment due to MCAM OE. Inhibition of MCAM-induced McTNs by vinorelbine and okadaic acid were able to reverse and completely inhibit the reattachment seen in the MCAM OE cells (Fig. 3D & Supp.Fig. 5D). This decrease in reattachment is illustrated with time-lapse representative phase contrast images (Fig. 3E & Supp.Fig. 5E). These findings indicate that MCAM can increase McTN production, which can be inhibited by targeting either microtubules or vimentin. Furthermore, vinorelbine and okadaic acid can abrogate increases in MCAM-induced McTN-mediated phenotypes such as homotypic clustering and reattachment.

Previous studies have demonstrated MCAM increases cell migration in cancer [16, 19, 21–24]. We sought to investigate whether our MCAM OE would impact the migratory capabilities of MCF-10A cells. We determined that MCAM OE elevates migration compared to GFP control (Supp.Fig. 6A). The microporous membrane of the migration cartridge was also stained with Wheat germ agglutinin (WGA) and Hoescht to allow for representative imaging which further illustrated the enhanced migration of the MCAM OE cells as the top right panel illustrates a heavy output of membrane staining from the WGA (red), indicating the MCAM OE have formed a monolayer after migrating through the pores whereas the black portions in the images indicates no cells present (Supp.Fig. 6B). The elevated migration was reversible through vinorelbine treatment (Supp.Fig. 6A), indicating a level of reliance upon polymerized microtubules.

MCAM deletion reduces microtentacle frequency

While MCAM OE enhanced McTN production, we next sought to determine whether loss of expression would decrease McTN production. Since MCAM is often overexpressed in breast cancer cells, we utilized CRISPR/Cas9 to delete MCAM in tumor-derived (TD) MDA-231 and MDA-436 cells. The TD TNBC cell lines have enhanced McTN production, cellular reattachment, homotypic cell clustering, and tumor metastasis in mice compared to parental, non-tumor-derived cell lines [32]. MDA-231TD and MDA-436TD cells also express high levels of MCAM protein making them ideal candidates to delete MCAM (Fig. 1 & Supp.Fig. 1). Two single cell clones with complete deletion of MCAM from the MDA-231TD and MDA-436TD cells were isolated, confirmed and expanded, knockout 6 and 11 (KO6 & KO11) for the MDA-231TD (Fig. 4A) and knockout 12 and 13 (KO 12 & KO13) for MDA-436TD cells (Fig. 4B). Interestingly, we saw proliferation of all four individual KOs appeared to be reduced compared to their respective parental cells (Supp.Fig. 7A, B). The reduced proliferation is more apparent in KO12 and KO13, in the MDA-436TD background. Previous work indicates that elevated MCAM expression can increase proliferation, and targeting MCAM with monoclonal antibodies has reduced cell growth in breast cancer [47, 48].

Fig. 4.

TNBC cells McTN positivity decreases upon loss of MCAM expression A Immunoblot analysis (left) of MDA-231TD parental and MCAM KO cells probed for MCAM and GAPDH. Densitometry statistical analysis (right) for MCAM protein expression of the MDA-231TD and knockout cells. Error bars indicate mean ± standard deviation, n = 3. **** p < 0.0001 (One-way Anova with Dunnett post-test). B Immunoblot analysis (left) of MDA-436TD and MCAM KO cells probing for MCAM and GAPDH. Densitometry statistical analysis (right) for MCAM protein expression in the MDA-436TD and knockout cells. Error bars indicate mean ± standard deviation, n = 3. **** p < 0.0001 (One-way Anova with Dunnett post-test). C McTN quantification of tethered, on TetherChips, and fixed (left) MDA-231TD parental cells and MCAM KOs. The data represents the quantification of McTN frequency from three independent experiments, with 100 cells counted per experiment. McTN quantification of tethered and fixed (right) MDA-436TD parental cells and MCAM KOs. The data represents quantification of McTN frequency from three independent experiments with 100 cells counted. Error bars indicate ± standard deviation, n = 3. ** p < 0.01 & * p < 0.05 (One-way Anova with Dunnett post-test. D MDA-231TD parentals and MCAM KOs analyzed for cellular perimeter altered by McTN positivity (left) of at least 30 individual cells. MDA-436TD controls and MCAM KOs were analyzed for cellular perimeter altered by McTN positivity (right) of 30 individual cells. Error bars indicate ± standard deviation, n = 3. *** p < 0.001 & **** p < 0.0001 (One-way Anova with Dunnett post-test).

To determine if the loss of MCAM expression reduced McTNs production, cells were seeded onto TetherChips and visualized using confocal microscopy. As expected, 60–70% of the MDA-231TD and MDA-436TD parental cells produced McTNs, whereas McTN production was reduced to approximately 45% in the MCAM KO clones (Fig. 4C). Cellular perimeter analysis was also calculated and showed a significant decrease in the MDA-231TD MCAM KOs and MDA-436TD MCAM KOs compared to their respective parental cells (Fig. 4D).

MCAM deletion decreased microtubule post-translational modifications

Since MCAM overexpression increased vimentin protein and detyrosinated α-tubulin in MCF-10As, and acetylated α-tubulin in MDA-468 s, we expected that MCAM KO cells would show reduced vimentin and α-tubulin PTMs. Acetylated α-tubulin was significantly decreased in both the MDA-231TD (Fig. 5A) and MDA-436TD (Fig. 5D) KO cells compared to the parentals. Interestingly, detyrosinated α-tubulin was significantly decreased in MDA-231TD KOs (Fig. 5B), without any significant change observed in the MDA-436TD KOs (Fig. 5E). Immunoblot analysis did not reveal any changes in vimentin protein (Fig. 5C, F). However, the MDA-231TD and MDA-436TD cells have undergone EMT and are strongly mesenchymal. It is likely that vimentin is strongly expressed due to upregulation of transcriptional factors that drive EMT, and that the sole deletion of MCAM is unlikely to reduce vimentin. Given the role of detyrosinated and acetylated tubulin in McTN formation [31, 38], the decreased tubulin PTMs following MCAM knockout provide a potential mechanism for the decrease observed in McTNs (Fig. 4C, D).

Fig. 5.

MCAM KO decreases TNBC microtubule PTMs A Acetylated α-tubulin, GAPDH, and α-tubulin (top) in MDA-231TD parental and MCAM KOs measured via immunoblot analysis. Densitometric fold change (bottom) analysis of acetylated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3. *p < 0.05. ***p < 0.001 (One-way Anova with Dunnett post-test). B Detyrosinated α-tubulin, GAPDH, and α-tubulin (top) in MDA-231TD parental and MCAM KOs measured via immunoblot analysis. Densitometric fold change (bottom) analysis of detyrosinated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3. *p < 0.05 (One-way Anova with Dunnett post-test). C Vimentin and GAPDH (top) in MDA-231TD parental and MCAM KOs measured via immunoblotting analysis. Densitometric fold change (bottom) analysis of vimentin protein expression. Error bars indicate ± standard deviation, n = 3 (One-way Anova with a Dunnett post-test). D) Acetylated α-tubulin, GAPDH, and α-tubulin in MDA-436TD parental and MCAM KO cells measured via immunoblotting analysis. Densitometric fold change (bottom) analysis of acetylated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3. *p < 0.05 (One-way Anova with Dunnett post-test). E Detyrosinated α-tubulin, GAPDH, and α-tubulin (top) in MDA-436TD parental and MCAM KOs measured via immunoblotting analysis. Densitometric fold change (bottom) analysis of acetylated α-tubulin protein expression. Error bars indicate ± standard deviation, n = 3 (One-way Anova with Dunnett post-test). F Vimentin and GAPDH (top) in MDA-436TD parental and MCAM KOs measured via immunoblotting analysis. Densitometric fold change (bottom) analysis of vimentin protein expression. Error bars indicate ± standard deviation, n = 3 (One-way Anova with a Dunnett post-test).

Loss of MCAM reduces TNBC homotypic clustering and cellular attachment

To investigate the functional impact of the loss of MCAM, MDA-231TD and MDA-436TD parental and KO cells were seeded for the same clustering and reattachment assays as the MCF-10A, with the hypothesis that the loss of MCAM protein will reverse the McTN-mediated phenotypes upregulated by MCAM overexpression. To determine if MCAM’s expression affects McTN functions, the MDA-231TD and MDA-436TD cell lines were allowed to cluster for four hours, followed by tethering onto the TetherChip for analysis and visualization. It was revealed that when MCAM is knocked out in both cell lines, the homotypic cell clustering efficiency was significantly decreased (Fig. 6A, B & Supp.Fig. 8A, B). Reattachment was again measured using xCelligence platform along with phase contrast-timelapse images to visualize reattachment in real time. The MDA-231TD KO cells showed a significant decrease in reattachment compared to the control as early as 1 h (Fig. 6C, D). However, one of the MDA-436TD KOs had a significant decrease in reattachment as early as 2 h, the other clone, KO12, reached a significant reduction after 6 h (Supp.Fig. 8C, D).

Fig. 6.

Loss of MCAM directly decreases McTN-mediated phenotypes in TNBC A Clustering Efficiency analysis (left) of MDA-231 cells over the 4 h period in low-attachment 96-well plates compared to those immediately tethered in the TetherChip. Individual values at t = 0 were divided by the number of clusters at t = 4 h for their respective condition. Error bars indicate ± standard deviation, n = 4. *p < 0.05, **p < 0.01 versus MDA-231TD parental (One-way Anova with a Dunnett post-test). B) Average cluster size of the t = 4 clusters. Error bars indicate ± standard deviation, n = 4. **p < 0.01, ***p < 0.001 versus MDA-231TD parental cells. (One-way Anova with a Dunnett post-test). B Representative images of MDA-231TD parental cells and MCAM KO cells tethered onto TetherChips and stained with Hoechst. Scale bar = 500 μm. C) Representative graph of reattachment efficiency for MDA-231TD parental vs MCAM KO cells. Error bars indicate ± standard deviation, n = 3. *p < 0.05, *p < 0.01 versus MDA-231TD (Two-way Anova with a Dunnett post-test). D Representative phase contrast still frames of MDA-231TD parentals and MCAM KOs. Images show cells attaching onto uncoated 24-well plates at t = 0 and t = 4 h. Images were taken on Nikon Ti2-E inverted microscope at 20 × magnification with a stage-top Tokai-hit incubator chamber. Scale bar = 50 μm. E) Representative graph of migration efficiency assay of MDA-231TD parentals and MCAM KOs on the xCelligence RTCA system. Error bars indicate ± standard deviation, n = 3. ****p < 0.0001 versus MDA-231TD parental cells (Two-way Anova with a Dunnett post-test). F Representative images of the xCelligence CIM plate transwell cartridge that has been stained with WGA and Hoechst to illuminate the migration of MDA-231TD parental and MCAM KO cell lines through the pores. Images were taken on a Nikon Ti2-E inverted microscope at 20 × magnification. Scale bar = 50 μm.

As migration can occur post-clustering and reattachment, we sought to investigate if our MCAM KOs would impact migratory capabilities in TNBC cells, hypothesizing a reversal to the MCF-10A MCAM OE increases. MCAM KO clones exhibited reduced migration. Interestingly, the MDA-231TD KO cells had a delay in their migration (Fig. 6E), whereas the migration of MDA-436TD KOs was almost completely inhibited (Supp.Fig. 8E). This disruption was confirmed with representative images of the underside of the CIM-plates where the empty black spaces are uncovered portions of the nodes in the knockout cell wells and the parental cells had a monolayer of cells (Fig. 6F & Supp.Fig. 8F). These xCelligence assays confirm earlier studies that TNBC migration is decreased in the absence of MCAM. Overall, our data provide new information that MCAM protein loss in TNBC cells yields decreases in MCAM-induced McTN production and reduces homotypic cell clustering and reattachment, possibly mediated by the decrease in McTNs.

Discussion

As epithelial cells enter the bloodstream, cytoskeletal forces become unbalanced, which leads to McTN production. Our lab has demonstrated McTNs aid in non-adherent tumor cell homo- and heterotypic aggregation, which directly impacts the survival of cancer patients [6, 7, 14, 32, 33, 49]. However, our previous work has yet to identify receptors or signaling molecules on the cell surface that may alter McTNs and McTN-supported functions.

Over the last decade, numerous publications and emerging evidence indicate that MCAM expression is correlated with cancer metastases and reduced cancer patient survival [15–17, 19, 46, 48, 50, 51]. Given the correlation of MCAM expression in TNBC and McTN frequency (Fig. 1), we sought to investigate the impact of MCAM on the cytoskeleton of non-adherent breast epithelial cells and breast tumor cells. Using both overexpression and knockout, we demonstrate that MCAM positively influences McTN frequency in TNBC. With further exploration, we determined that in MCF-10A cells, high expression of MCAM increased vimentin protein expression and detyrosinated α-tubulin (Fig. 2), and TNBC MDA-468 MCAM OE cells saw an increase in acetylated α-tubulin (Supp.Fig. 2). The MCF-10A MCAM OE increase in vimentin and detyrosinated α-tubulin aided in elevated homotypic cell clustering, cellular reattachment, and migration (Fig. 3 & Supp.Fig. 5). This increase in McTNs, McTN-mediated metastatic phenotypes, and migration could be neutralized by treating with the FDA-approved microtubule depolymerizing drug vinorelbine and exposure to okadaic acid to target vimentin. This indicates a reliance on the microtubule cytoskeleton for MCAM to influence clustering, reattachment, and migration. Utilization of CRISPR/Cas9 allowed us to knock out MCAM in two strongly metastatic TNBC lines, MDA-231TD and MDA-436TD, where MCAM deletion significantly decreased McTN frequency (Fig. 4). Importantly, we show that loss of MCAM can decrease the acetylation of α-tubulin (Fig. 5) in two TNBC cell lines, and also decreases detyrosinated α-tubulin in the MDA-231TD, which are both markers for microtubule stability [34, 38, 52, 53]. Functionally, we observed a significant reduction in homotypic cell clustering, cellular reattachment, and migration (Fig. 6 & Supp.Fig. 8).

Importantly, as cancer cells traverse from a solid tumor to a non-adherent environment, such as the bloodstream, the microtubules tip the cytoskeletal forces into their favor [13]. This imbalance of cytoskeletal forces allows for the formation of McTNs through microtubules protruding through the weakened actin cortex [31–33, 37]. Weakening of the actin cytoskeleton, through latrunculin A treatment, Rho-associated protein kinase (ROCK) inhibition, or loss of obscurin all increased McTN frequency and loss of obscurin also led to breast tumor formation and metastasis in mice [31, 35, 37, 54]. We have found McTNs are not only in breast epithelial cells, but also in lung, colon, breast, prostate, and ovarian cancers. [44, 55] Together, these tumors represent over 50% of solid human tumors and may indicate that McTNs are a fundamental epithelial cell behavior since 90% of human solid tumors are epithelial carcinomas [56].

Multiple factors can influence microtubule stabilization, thereby facilitating the production and elongation of McTNs. Tubulin detyrosination and acetylation have both been demonstrated to promote McTN-mediated and metastatic phenotypes [31, 38, 40]. Microtubule-associated proteins, such as tau, can directly bind to microtubules and induce McTNs, and enhance the retention of tumor cells in the lungs of mice [39]. Vimentin has also been shown to increase microtubule stability and the frequency of McTNs [34, 52, 53]. Previous work has shown that MCAM can increase the expression of vimentin and other EMT markers [16]. However, the impact of MCAM on microtubules was previously unknown. Through the utilization of TetherChip technology, we demonstrate that MCAM can modulate McTN frequency in breast cancer cells (Figs. 2, 4 & Supp.Fig. 2). We further emphasize that MCAM’s functions extend beyond previously known interactions with the actin cytoskeleton to include induction of vimentin and detyrosinated tubulin in breast epithelial cells (Fig. 2) and both detyrosination and acetylation of α-tubulin in TNBC (Fig. 5 and Supp.Fig. 2), which in turn aids in McTN production. These data suggest a potential role for MCAM in modulating Vasohibins 1 & 2 and alpha-tubulin N-acetyltransferase (ATAT1) to influence detyrosination and acetylation of α-tubulin in TNBC.

In an early study from Anfosso et al., MCAM engagement was shown to elevate intracellular Ca²⁺ via Fyn, which was required to activate Fak, further supporting a link between MCAM and actin cytoskeleton modulation [57]. Calcium modulation has been linked to alterations in the cytoskeleton. Our group demonstrated that increases in intracellular Ca²⁺ by either thapsigargin to release internal Ca²⁺ stores or ionomycin to allow for the entrance of extracellular Ca²⁺ both decrease McTN formation [35]. Thus, if the mechanism of McTN formation in the MCAM OE cells is solely due to excess and sustained intracellular Ca²⁺, we would expect that the MCAM OE cells would have a decrease in McTNs. However, elevation of intracellular calcium due to MCAM was measured to peak within 20 s and returns to baseline levels by 180 s.[57] Since we consistently measured an increase in McTNs in MCAM OE cells after 45 min in suspension, we hypothesize that the levels of Ca²⁺ or any transient increase would not be sufficient to impair McTNs, however further studies would be required to test this hypothesis.

Metastasis of cancer being responsible for 90% of cancer-related deaths enforces the importance of finding genes and proteins that are crucial in various steps of the metastatic cascade [3, 6, 7, 13, 14]. Although CTCs represent a small fraction of cells in the bloodstream [49], CTC clusters have evidence supporting their role in enhancing CTC survival and metastatic potential [11, 42, 58]. CTCs that survive can utilize microtentacles to enhance reattachment to endothelial cells, thereby increasing the likelihood of seeding and invading at a distant site [31–33]. McTNs have been correlated with clustering, cellular reattachment, migration, and tumor aggressiveness [32, 33, 39, 40]. These data, along with evidence illustrating that 40–50% of metastatic breast cancer patients were discovered to have at least one CTC cluster, suggests that CTC clusters may be a key driver in the progression of the metastatic process [6, 7, 14, 49]. However, the molecular mechanism by which McTNs promote clustering in non-adherent cells remains incompletely understood. As non-adherent tumor cells extend their membranes via McTN formation, this may increase the likelihood of adhesion-molecule interactions and focal adhesion formation, thereby promoting clustering and attachment of non-adherent cells. Immunoglobulin superfamily cell adhesion molecules VCAM-1 and ICAM-1, as well as adhesion molecule CD-44, have been shown to mediate cell–cell cohesion and cell clustering through homophilic interactions and heterophilic interactions with β-integrins or hyaluronan [59–62]. Specifically, VCAM-1 and ICAM-1 facilitate homo- and heterotypic CTC cluster formation mostly through integrin interactions [11, 58, 63].

With expression on cancer cells and endothelial vasculature cells, MCAM-ligand/coreceptor interactions could promote adhesion, leading to cluster formation and attachment to the endothelial cells lining the vasculature. In this study, our current data suggests that MCAM increases McTN frequency and promotes homotypic cell clustering and reattachment through reliance on microtubules and vimentin in the MCF-10A (Fig. 3). Conversely, as we knock out MCAM in TNBC cell lines, we show that the decrease in McTNs and acetylated α-tubulin (Fig. 5) resulted in a direct reduction in clustering and in reattachment (Fig. 6 & Supp.Fig. 8). Previous research has demonstrated that, in melanoma cells, adhesion to human dermal microvascular endothelial cells was dependent upon MCAM as loss of MCAM reduced adhesion and migration across the endothelial cells [64]. Unfortunately, no specific binding mechanism was determined, and further investigation is required. MCAM interacts with cell-surface receptors, integrins, and extracellular matrix ligands to mediate cell–cell interactions, including β1 integrins, Galectins, and Laminins [30, 65–67]. These interactions have been found to recruit Src tyrosine kinases and the Ezrin, radixin, and moesin (ERM) proteins, which activate signaling pathways involved in cell adhesion and actin rearrangement through focal adhesion kinase (FAK) and protein tyrosine kinase 2-beta, similar to VCAM and ICAM [15, 18, 57, 59]. This evidence could support a model that, as cancer cells intravasate into the bloodstream, MCAM aids in microtubule stabilization through increased PTMs, McTNs extend the surface of the cell where MCAM or other adhesion molecules on the surface of the specific tumor cell type have a greater chance to engage with like cells, or immune cells, to enable cluster formation and attachment to the vasculature via focal adhesion formation and actin reorganization, thus enhancing metastasis.

Given that CTC clusters increase the survival and metastatic potential of cancer cells, identifying proteins that regulate their clustering would be highly impactful. We show that the knockout of MCAM disrupts TNBC ability to form homotypic clusters and reduces reattachment and migration (Fig. 6 & Supp.Fig. 8). This prompts the question of whether personalized treatment for MCAM-positive cancers is possible. Proteolytic cleavage of membrane-bound MCAM results in a soluble form (sMCAM) that can be detected in plasma, serum, urine, and cerebrospinal fluid [51]. Furthermore, soluble MCAM and membrane-bound MCAM have been identified in plasma and have been recently used and are currently used as circulating tumor or outcome biomarkers in clinical trials for kidney cancer[68], non-small-cell lung cancer [69], glioblastoma [70], and breast cancer [18, 71]. Recent data have shown two MCAM antibodies that do not bind to physiologically normal tissue expressing MCAM and have been shown to reduce metastatic phenotypes [20, 48, 72]. Using sMCAM or a tumor-specific MCAM (TsMCAM) antibody to confirm MCAM expression via liquid biopsy, combined with antibody treatment targeting tumor-specific MCAM, would provide researchers and clinicians with leverage over MCAM-positive cancers in personalized treatment.

As antibody treatment targeting tumor-specific MCAM requires further investigation, our study showed that, alternatively, MCAM may stabilize the microtubule cytoskeleton by elevating tubulin PTMs, thereby promoting McTNs and McTN-mediated phenotypes. Interestingly, one study that used a drug panel to assess compounds capable of dissociating CTC clusters found that inhibitors targeting tubulin consistently reduced CTC cluster formation and size [11]. Although okadaic acid is not used as a drug treatment, it provides valuable data to determine that without the support of vimentin, the stabilization of microtubules was not strong enough to form McTNs in the MCAM OE cells (Supp.Fig. 5). These data indicate that MCAM overexpression in the MCF-10A enhances McTN-mediated phenotypes through the stabilization of microtubules via vimentin. Exposing our MCF-10A MCAM OE cells to vinorelbine and okadaic acid resulted in a direct decrease in McTNs compared to the vehicle control, and thus in a decrease in McTN-mediated phenotypes (Fig. 3 & Supp.Fig. 5). Vinorelbine is a vinca alkaloid agent currently FDA-approved for treating breast and non-small cell lung cancer, which works by depolymerizing microtubules. These data suggest that if microtubules are not polymerized or stabilized by PTMs, MCAM's effects on clustering, reattachment, or migration are reduced. The MDA-231 TD TNBC cells have high MCAM protein levels (Fig. 1, 4), and our previous work has shown that vinorelbine disrupts clustering, reattachment, and metastasis in mice injected intravenously with MDA-231 TD cells [32]. This indicates that MCAM may promote metastatic phenotypes by stabilizing the microtubule cytoskeleton through α-tubulin PTMs and vimentin. Thus, targeting the microtubule cytoskeleton with vinorelbine could be a viable alternative for cancer patients with high MCAM expression to disrupt CTC clusters, reattachment, migration, and metastasis, until further advances in MCAM-specific targeting via antibody therapy are ready for clinical use.

Overall, these experiments demonstrate that MCAM increases tumor cell metastatic phenotypes through stabilizing the microtubule cytoskeleton in non-adherent environments, as encountered in the bloodstream or lymphatics during metastasis. Moreover, McTN-mediated metastatic phenotypes promoted by MCAM expression could be abolished through destabilizing the microtubule cytoskeleton using FDA-approved therapies.

Supplementary Information and Data available in Supp file 1.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ATAT

Alpha tubulin acetyltransferase 1

ATCC

American type culture collection

BSA

Bovine serum albumin

Ca²⁺

Calcium

CAM

Cell adhesion molecule

CD146

MCAM

CRISPR

Clustered regularly interspaced short palindromic repeats

CTC

Circulating tumor cells

DMEM

Dulbecco’s modified eagle’s medium

ECM

Extracellular matrix

EMT

Epithelial-to-mesenchymal transition

ERM

Ezrin, Radixin, and Moesin complex

FDA

Federal and drug administration

FAK

Focal adhesion kinase

ICAM

Intracellular adhesion molecule 1

KO

Knockout

MCAM

Melanoma cell adhesion molecule

McTNs

Microtentacles

OE

Overexpression

Oka

Okadaic acid

PTM

Post-translational modification

ROCK

Rho-associated protein kinase

sMCAM

Soluble MCAM

TD

Tumor derived

TNBC

Triple-negative breast cancer

TsMCAM

Tumor-specific MCAM

TBS-T

Tween-20

VCAM

Vascular cell adhesion molecule

VR

Vinorelbine

WGA

Wheat germ agglutinin

Author contributions

The conceptualization and design of the study were carried out by D.A.A., S.S.M., and M.I.V. Data acquisition was conducted by D.A.A., while analysis and interpretation involved contributions from D.A.A., K.N.T., J.A.J., M.L.M., D.E.G., D.O.O., J.L.C., S.S.M., and M.I.V. Cell line generation was conducted by D.A.A. and the translation core laboratory at University of Maryland, Baltimore. D.A.A. was responsible for drafting the manuscript, with edits made by D.A.A. along with K.N.T., J.A.J., M.L.M., D.E.G., D.O.O., J.L.C., S.S.M., and M.I.V. All authors read and approved the final manuscript.

Funding

This work was supported in part by the METAvivor Foundation and grants to S.S.M. from the National Institutes of Health (R01-CA154624 and R01-CA124704). J.A.J. was supported by 5T32CA154274-13 from the National Cancer Institute. M.L.M. was supported by 5T32GM008181-33,34 from the National Institute of General Medical Sciences. The Greenebaum Comprehensive Cancer Center was supported by P30-CA134274 and the Maryland Department of Health’s Cigarette Restitution Fund Program by CH-649-CRF.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing Interests

The University of Maryland School of Medicine owns patents on the subject of microtentacles and microfluidic cell tethering that list one of the authors on the current manuscript (Stuart S. Martin) as an inventor.