The mouse ovarian surface epithelium model represents the progression of ovarian cancer, encompassing benign (MOSE-E), slow-developing (MOSE-L), and fast-developing disease (MOSE-L_TICv ) (Roberts et al., 2005). Their genotype and phenotype have been extensively characterized, and the expression of fallopian tube markers suggest that the MOSE cells represent serous ovarian cancer development (Creekmore et al., 2011; Cohen et al., 2013; Cohen et al., 2016; Anderson et al., 2013). MOSE-_TICv cells were derived by harvesting the peritoneal fluid from mice injected with MOSE-L cells. This in vivo selection significantly increased the tumorigenicity from 1 × 10⁶ cells injected intraperitoneally into C57BL/6 mice causing lethal disease in ∼100 days (MOSE-L) to 1 × 104 cells causing lethal disease in 21 days (MOSE-L_TICv ) (Roberts et al., 2005; Cohen et al., 2013). MOSE cells were routinely cultured in high-glucose DMEM (HG; 25 mM glucose) (Sigma-Aldrich), supplemented with 4% fetal bovine serum (FBS, Hyclone), 3.7 g/L sodium bicarbonate (Sigma-Aldrich), and 1% penicillin -streptomycin solution (Thermo Fisher Scientific) at 5% CO₂ and 21% O₂ (normoxia, NO). Spheroid formation was achieved as described (Compton et al., 2021; Compton et al., 2022). Briefly, cells were seeded into ultra-low adherent 96-well plates (Corning) in hypoxia (HO, 2% oxygen), and low-glucose (LG, 5 mM glucose) DMEM for 24 h. This represents the oxygen (Kizaka-Kondoh et al., 2009; Emoto et al., 2010; Mutch and Williams, 1994) and glucose levels (Lippi et al., 2013) found in malignant ascites. The benign MOSE-E were not used here because these cells do not form viable spheroids. To mimic spheroid adhesion to the secondary site, MOSE-L, and MOSE-L_TICV spheroids were generated in a LG DMEM in HY for 24 h. To mimic initial attachment at secondary sites, spheroids were transferred to tissue culture-treated plates and kept in LG, HO conditions for 12 h. At this timepoint, the spheroids were attached to the tissue culture plates, slightly flattened and had begun to grow onto the culture plates (outgrowth). Gained access to oxygen and nutrients after adhesion and angiogenesis (reoxygenation) was mimicked by changing the culture conditions to HG medium with 4% FBS at NO for an additional 16 h. Thus, for the subsequent gene and protein analyses, the samples named ‘adherent spheroids’ contained both actively outgrowing cells and the bulk of the spheroid.

Total RNA was isolated from adherent cells, spheroids, and adherent-spheroids cultured in the conditions described above using the RNeasy Mini Kit (Qiagen). Complementary DNA (cDNA) was synthesized using the RT² First Strand Kit (Qiagen), according to the manufacturer’s specifications. Gene expression profiling of adhesion molecules, their regulators, and extracellular matrix was performed using the RT² Profiler PCR Array for mouse extracellular matrix and adhesion molecules (Cat. No. 330231 PAMM-013ZR, Qiagen; gene list in Supplementary Table S1). MOSE-L and MOSE-L_TICV adherent spheroids were also cultured under hypoxic, low-glucose conditions to compare to their reoxygenated adherent spheroid counterpart. Real-time quantitative PCR was conducted using the QuantStudio 6 Pro (Thermo Fisher Scientific). Data normalization and ΔCT (Supplementary Table S2) value comparisons between experimental groups were carried out using Qiagen’s online analysis tools. A heatmap comparing ΔCT values across different stages of ovarian cancer dissemination was generated using Prism 10 (GraphPad). Data shown are the mean of 2-3 replicates.

Western blotting was performed to quantify the protein expression in adherent cells, spheroids, and adherent spheroids. Proteins were extracted with RIPA buffer supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). Protein concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Proteins were separated on an 8% Bioacrylamide SDS-PAGE gel and transferred onto PVDF membranes (Bio-Rad). Membranes were blocked with 5% bovine serum albumin (BSA) in 1× Tris-buffered saline containing 0.1% Tween-20 (TBST).

Primary antibodies were used to detect the integrin monomers ITGα2 (Thermo Fisher), ITGα3 (Thermo Fisher), ITGα4 (Millipore), ITGα5 (Cell Signaling), ITGαV (Millipore), ITGα6 (Thermo Fisher), ITGβ1 (Novus Biologicals), and ITGβ3 (Thermo Fisher). IRDye 680RD and IRDye 800CW secondary antibodies were used for visualizing the protein bands using the LI-COR Odyssey CLx imaging system (LI-COR Environmental). Protein expression levels were normalized to the housekeeping protein α-tubulin. Data are presented as mean ± SEM of at 3 to 4 biological replicates.

Immunofluorescent staining was performed as described (Roberts et al., 2005) in adherent cells and adherent spheroids growing on glass coverslips using primary antibodies directed against ITGα2 (Thermo Fisher), ITGα3 (Thermo Fisher), ITGα4 (Millipore), ITGα5 (Cell Signaling), ITGαV (Millipore), ITGα6 (Thermo Fisher), ITGβ1 (Novus Biologicals), and ITGβ3 (Thermo Fisher) and the appropriate secondary antibodies (Invitrogen). After mounting onto glass slides using ProLong™ Gold Antifade Reagent with DAPI (Invitrogen), images of adherent cells and cell outgrowth around adherent spheroids were acquired with a Nikon 80i epifluorescence microscope and NIS-Elements BR3.0 software. Negative controls are shown in Supplementary Figure S1. Images were analyzed with ImageJ and quantitated with Prism 10 (GraphPad). Data are presented as mean ± SEM from three independent experiments performed in triplicate with approximately 30 images of the cells.

To perform 3D imaging of spheroids, optical clearing technique was performed with the Miltenyi MACS Clearing kit as per the manufacturer’s instructions. Briefly, fixed spheroids were transferred to a polypropylene Eppendorf tube and permeabilized with the MACS permeabilizing solution to allow for the penetration of the primary and secondary antibodies into the spheroids. Nuclei were stained with a DAPI solution (Thermo Fisher). After washing in 1X antibody staining solution, the spheroids were dehydrated in 50%, 70%, and 100% ethanol with 2%Tween 20, cleared in Miltenyi MACS clearing solution and transferred to 35 mm³ glass bottom confocal Petri dishes (Cellvis). The same procedure was used for adherent spheroids, but all the steps were performed in 33 mm² glass bottom confocal Petri dishes. Optical Z stack slices were taken at 40X magnification with a Zeiss LSM880 laser scanning confocal microscope with Airy scan using Zen2 Blue software. Spatial localization of the proteins in the center slice of the non-adherent spheroids was determined by marking the outer rim of the mid-section with a circle of the same width for all spheroids and quantitate the fluorescence in both the outer rim and the center using ImageJ.

Following image acquisition of the adherent spheroids, further image analyses were performed using Imaris 9.8.2 software (Oxford Instruments). In the 3D spheroids, the compartmental fluorescence quantification of nuclear vs. cytoplasmic and cell membrane protein expression was performed. By masking the FITC signal using the DAPI-defined nuclear segmentation, only the cytoplasmic and cell membrane signals were considered in the cytosol compartment, while the nuclear signal was analyzed using DAPI overlapped FITC within the nuclear regions. This approach prevents misclassification of cytoplasmic protein as nuclear. Adherent spheroids were reconstructed by first defining key regions, such as the “site of attachment” and “bottom,” using manual contouring for precise boundary definition. To improve axial resolution and achieve more isotropic 3D data, deconvolution and background subtraction were applied. Segmentation was performed to accurately isolate individual cells from their surroundings. Then object-based measurements were conducted to generate scatter plots correlate to spatial metrics such as X, Y and Z-axis with mean fluorescence intensity values. Experiments were conducted in 3 independent trials with 3-5 spheroids and each of the images provided 20,000 IMARIS data points.

The cytotoxicity of the integrin inhibitors BT3033, E7820, ITG-IN-TFA, GLPG0187, TC I-15, AIIB2, Cyclo (RGDyK), and K34c was tested using the alamarBlue assays as described (Grieco et al., 2023b) in MOSE-L and MOSE-L_TICv monolayers and spheroids. Drug properties are shown in the Supplementary Table S3. Determination of drug concentrations that do not overtly induce apoptosis and controls are shown in the Supplementary Figure S1. Spheroids were seeded in ULA 96-well plates and maintained for 24 h under HO LG conditions, while monolayers were plated in 96-well plates under NO HG for 24 h. Cells were then treated with drugs at the indicated concentrations in DMEM containing 2% FBS for 24 h and 48 h before AlamarBlue (10 µL/well) was added to wells. Fluorescence was measured with a Cytation 5 plate reader (540/590 nm). The highest growth-inhibitory but non-toxic concentrations were used for aggregation and outgrowth assays.

For the spheroid aggregation assay, MOSE-L and MOSE-L_TICV cells were incubated with BT3033 (0.5, 5 µM), E7820 (15, 20 µM), ITG-IN-TFA (25, 30 µM), GLPG0187 (1.3, 5 nM), TC-I 15 (20, 25 µM), AIIB2 (20, 30 µM), Cyclo (RGDyK) (5, 15 µM), and K34c (10, 15 µM) in ultra-low adherence plates for 24 h under HO LG conditions. Aggregation was assessed qualitatively at 20X magnification using a Nikon Eclipse Ts2R microscope with NIS Elements BR5.43 imaging software. For adhesion assays, the same drug treatments were applied after spheroid formation. Images were taken following adhesion under HO LG conditions (24 h) and following reoxygenation under NO HG conditions (48 h).

All statistical analyses were performed using GraphPad Prism 10. Data are presented as mean ± SEM, with biological replicates of at least n = 3. Comparisons between groups were analyzed using one-way or two-way ANOVA or paired t-test as indicated. Intensity measurements for both spheroid and adherent spheroid conditions were obtained using Imaris software which provided the means of 20,000 data points per trial. Spatial and intracellular differences were calculated with paired t-tests. Results were considered statistically significant at p < 0.05.

Given the critical role of integrins and integrin regulators in mediating cell–cell and cell-ECM adhesion, we first assessed changes in integrin expression associated with ovarian cancer progression. We compared fold changes in gene expression levels between the benign MOSE-E, MOSE-L (representing slow-developing disease), and MOSE-L_TICv (aggressive, fast-developing disease). In general, the expression levels of individual integrin subunits were very different between the cancer stages (Supplementary Table S2). As shown in Figure 1A, highly expressed in both cancer cell lines were ITGα3, ITGα5, ITGαV, and ITGβ1. No, or very low expression (CT > 35) was noted for ITGαE, ITGαL, ITGαM, and ITGαX. This was expected since these integrins are expressed mostly in immune cells (Dotzer et al., 2021). Most integrins were lower in MOSE-L compared to the benign MOSE-E (black bars) which was partially reversed in the highly aggressive MOSE-L_TICv (green bars). Only ITGα2 expression further decreased in the aggressive disease (Figure 1A). Other adhesion molecules and regulators such as CD44, CTNNA1, CTNNB1, NCAM1 and VCAM1 were highly expressed in the MOSE cells while others were either low (CDH2, CTGF, ICAM1) or were not expressed (Figure 1B; Supplementary Table S2). A similar expression pattern as observed in the integrin expression with a lower expression in the MOSE-L than MOSE-E but higher in MOSE-L_TICv was observed for CD44, CTGF and ICAM1 (Figure 1B). This elevation correlates well with their silencing in earlier stages of ovarian cancer but increased expression in later stages (CTGF) (Cannistra et al., 1995; Kikuchi et al., 2007), and contribution to an aggressive phenotype (CTGF, NCAM) (Cannistra et al., 1993; Zecchini et al., 2011), and stemness (CD44) (Zhang et al., 2008; Strobel et al., 1997).

Important for ovarian cancer is the expression of ECM proteins such as collagens, FN1, TNC, and their regulators (MMPs, SPP1, TIMPs). We saw similar changes in their expression during MOSE progression with a reduced expression in the MOSE-L compared to the benign cells, but a higher expression in the MOSE-L_TICv (Col2A1, LAMA2, LAMB3, THBS2, MMP3,9,10,13, and SPP1). However, most of these genes were still highly expressed (Figures 1C,D; Supplementary Table S2) and we have shown previously that both COL1A1 and FN1 are rapidly secreted by MOSE spheroids (Shea et al., 2023). MMP3 expression was higher in the MOSE-L_TICv than in the benign MOSE-E (Figure 1D), which correlates well with studies in human tissue that show an elevation of MMP3 in higher-grade ovarian cancer tissues (Wang et al., 2019). SPP1 (osteopontin), a glycoprotein overexpressed in many cancer types and associated with a poor prognosis (Wei et al., 2020), was overexpressed in the MOSE-L_TICv cells while the expression of SPARC, an inhibitor of ovarian cancer, growth and invasion (Said et al., 2007) was progressively reduced during MOSE progression albeit still highly expressed (Figure 1D). LAMA1 and LAMC1 were sequentially reduced in the cancer cells, while LAMB2 was higher in the cancer cells. This correlates well with reports of their expression patterns in human ovarian tumor tissues (Kuhn et al., 2012). These data suggest that while the changes in the adhesion molecules, ECM components, and their remodeling enzymes were modest, together they may represent complex, stage-specific shifts that support cancer progression. Similar shifts higher expression of these adhesion-related molecules in MOSE-L_TICv may be contributing to their increased metastatic potential and aggressive behavior.

To mimic the transcoelomic metastasis of ovarian cancer rather than using cell monolayers we generated spheroids of the MOSE-L and the aggressive MOSE-L_TICv cells in hypoxic and low glucose culture conditions as observed in ascites (Schielzeth et al., 2020). We then let them adhere in the same culture conditions (adherent spheroids) and then mimicked gaining access to oxygen and nutrients (reoxygenation-reox) by changing the culture conditions to normoxia and high glucose media (NOHG) (Figure 1E). Highly expressed in all culture conditions in both MOSE-L and MOSE-L_TICv were ITGβ1, ITGα3, ITGαV, and ITGα5. This is in agreement with their expression in human tumors (Ahmed et al., 2002; Ahmed et al., 2005). As shown in Figure 1F, several integrins were higher expressed after aggregation (ITGβ2 and ITGβ3) while only ITGα3 was downregulated by aggregation in both cell lines. Other integrins changed in a cell type-specific manner. In MOSE-L_TICv , aggregation increased the expression of ITGα2, ITGα4, ITGβ2, and ITGβ4, suggesting a contribution to cell-cell adhesion and survival. ITGα3, ITGα5, and ITGβ1 were reduced in MOSE-L_TICv spheroids. In contrast, integrin expression either did not change in MOSE-L or in the opposite direction from MOSE-L_TICv . Adhesion of the spheroids in HO LG conditions had a limited effect on integrin expression in MOSE-L_TICv where we observed an increased expression of ITGβ1 and ITGα5 while ITGα2 and ITGα4 expression levels were lower (Figure 1F). In MOSE-L, ITGα2, ITGα4, ITGβ2, and ITGβ3 were increased while ITGα5 was decreased after spheroid adhesion. In contrast, exposure of the adherent spheroids to reoxygenation caused an increase in most integrins in both cell lines. The fast-developing disease was characterized by more pronounced increases in all integrins but ITGβ2, ITGβ3, and ITGβ4 which decreased after reoxygenation. These results imply differential roles of individual integrins in aggregation and adhesion and at different stages of the disease.

There were little changes in the other highly expressed adhesion molecules in MOSE-L after aggregation but an increased expression of CD44, CDH2, and NCAM1 in MOSE-L_TICv (Figure 1G). Adhesion of the spheroids caused an increase in the expression of CTGF, ICAM1, and CTNNA1 in both cell lines while all other shifts were cell type specific. Notably, NCAM1 was decreased in in MOSE-L_TICv but still higher expressed than in MOSE-L which expressed higher levels of VCAM1 after aggregation. In addition to adhesion, both proteins guide immune cell trafficking and recruitment. Reoxygenation increased adhesion molecules mostly in the MOSE-L_TICv (CD44, CDH1, CTNNA1) while only CDH3 and CTGF were upregulated and ICAM-1 was downregulated in both cell lines (Figure 1G).

There was also very little overlap in ECM genes that responded to aggregation with an increase (ECM1) or a decrease in expression levels (COL3A1, COL4A1) between the slow- and the fast-developing disease. The aggressive MOSE-L_TICv showed robust increases in several collagens, FN1, HAPLN1, LAMB2, LAMC1, POSTN, TNC, and THBS2 after aggregation but reduced levels of LAMB2 and TNC (Figure 1H), suggesting a role of these genes in cell-cell adhesion. After adhesion of the spheroids, several collagens (notably COL1A1 and COL3A1) were upregulated in the MOSE-L_TICv but reduced in the MOSE-L spheroids. While still highly expressed in all stages of dissemination, FN1 was highly elevated after reoxygenation in both cell lines as were COL1A1, COL4A2, COL5A1, and TNC.

ECM regulators that were highly expressed in both cell lines (TIMP2, TIMP3 and SPP1) were reduced after aggregation but higher in adherent spheroids (Figure 1I). In contrast, SPARC expression was reduced after aggregation with a gradual increase after aggregation and adhesion of MOSE-L_TICv spheroids and reoxygenation. MMPs were mostly upregulated by aggregation and further by adhesion of the spheroids in MOSE-L. We observed little changes in these genes in MOSE-L_TICv after aggregation and reduced expression after adhesion of the MOSE-L_TICv spheroids. Reoxygenation of the adherent spheroids increased several of the highly expressed ECM regulators in both cell lines. These data suggest that the interactions of specific integrins, adhesion molecules, ECM and their regulators play a role at specific stages during ovarian cancer dissemination. Our data also show a distinct difference between the responses of the cell types to these culture conditions that may contribute to the aggressive phenotype of the MOSE-L_TICv .

Integrins are recognized as key players in cancer, contributing significantly to processes such as migration, tumor invasion, and metastasis (Hamidi and Ivaska, 2018; Haake et al., 2024). Altered integrins expression is commonly linked to the aggressive nature of cancer cells by enhancing their capacity to invade nearby tissues and metastasize to distant locations (Hamidi and Ivaska, 2018). Thus, we focused next on integrins and compared changes in their protein expression during the metastatic spreading of ovarian cancer (Figure 1B) to the qPCR results above. Our results revealed distinct patterns of integrin expression at each stage: a gradual increase (ITGα3) or decrease (ITGβ3) in integrin expression as ovarian cancer progresses to advanced stages during dissemination, a decreased expression after spheroid formation but increased after spheroid adhesion (ITGα2, ITGαV, ITGß1), or an increased expression in spheroids but decreased after spheroid adhesion (ITGα4, ITGα5 and ITGα6) (Figures 2A,B). These results are similar to qPCR data above, confirming expression changes at both the gene and protein levels.

Changes in the expression of integrins such as ITGβ1 or ITGα2 were less pronounced than expected after review of the current literature (Masi et al., 2023; Strobel and Cannistra, 1999; Dotzer et al., 2021). We hypothesized that integrin expression changes may occur only at specific sites of the spheroids and may be masked by the unresponsive bulk of the spheroid when quantitated by qPCR or Western blotting Therefore, we next investigated spatial differences in integrin expression in the spheroids, selecting integrins that showed a change in expression after aggregation and adhesion. Here, we employed a 3D reconstruction model of laser-scanning confocal microscopy images. Shown in Figure 3 are distinct differences in the distribution of highly expressed integrins and CD44 in the center section of non-adherent MOSE-L (Figure 3A) and MOSE-L_TICv (Figure 3B) spheroids. Most integrins were significantly higher expressed in the periphery of the spheroids as demonstrated by a ratio of protein expression in the periphery over expression in the core >1 while ITGαV and ITGα6 were uniformly expressed throughout the spheroids (ratio periphery/core ∼1) (Figure 3C). The ratio of ITGα2, ITGα3, and ITGβ3 expression in the periphery versus core was significantly higher in the MOSE-L_TICv than in the MOSE-L while there was no difference observed for the other integrins. Interestingly, in addition to a different spatial expression of the integrins in the spheroids, we also observed differences in the intracellular location of the integrins. Therefore, we measured the mean fluorescence intensity of integrin expression in the nucleus or the cytoplasm and plasma membrane with the Imaris software as described in the method section. In both cell lines, only ITGα2 was expressed mostly in the cytosol and plasma membrane (not significant for MOSE-L_TICv ) while all other integrins were significantly higher expressed in the nucleus (Figure 3D). In general, cells in the periphery had a higher expression of integrins in the cytosol, while cells localized at the core of the spheroid showed a predominant nuclear location (see Figures 3A,B, right column). While investigations into the function of nuclear integrins were beyond the scope of this investigation, the potential impact of the integrins on gene expression and, thereby, on the phenotype of the cancer cells is of great importance. Similarly, the key adhesion and signaling molecule CD44 was higher expressed in peripheral regions of the spheroids with an intracellular location shift from nuclear expression in the spheroid core to membrane/cytosolic in the peripheral cells (see z-stack in Supplementary Video S1).

We then determined the impact of adhesion on integrin expression and distribution using laser scanning microscopy and 3D reconstruction of the Z-stack images. We focused specifically on ITGα2, ITGα3, ITGαV, and ITGβ1 expression and localization since these integrins were upregulated in adherent spheroids as determined by Western blot (Figure 2) albeit this provided only a bulk expression profile without spatial distribution analysis. The Imaris software allowed us to visualize integrin distribution and color mapping the mean intensity levels of integrin expression at the adhesion site (Figure 4A) or throughout the adherent spheroid (Figure 4B). As shown in Figure 4A, ITGα2 and ITGαV showed a strong expression at the site of adherence and in the outgrowth in both MOSE-L and MOSE-L_TICV models while ITGα2 and ITGβ1 were found highly expressed in the cells growing out from the spheroid after attachment. ITGβ1 expression was also found on the site of adherence but that was more variable between spheroids. The 3D reconstruction of the attached and somewhat flattened spheroids in Figure 4B (cut off vertically at the middle of the attached spheroid with outgrowth areas on the left side) shows that adhesion differentially affected integrin expression. While there was a low ITGα2 expression in the outer layer of the floating spheroids and very little expression in their core (see Figures 3A,B), after adhesion ITGα2 expression was apparent at the site of adhesion. Importantly, this was not restricted to a few cell layers but an intermediate intensity extended up to 40–50 μm into the spheroid core from the adhesion site. This change in expression was similar to the adhesion effects on ITGαV expression. In contrast, ITGα3 was mostly expressed at the site of spheroid outgrowth and at the spheroid surface while ITGβ1 was expressed at the adhesion sites but was highest in the outgrowing cells (Figure 4B). These results indicate that integrin expression during spheroid formation and attachment is not uniform, but rather spatially regulated, likely depending on each integrin’s role in anchoring, invasion, and interaction with the surrounding tumor microenvironment.

To investigate the plasticity of integrin expression, specifically their response to adhesion and proliferation, we compared the integrin expression in cells growing out from the adherent spheroids (outgrowth) to the original adherent monolayer cells to determine if the expression levels returned to the original levels. The expression levels of ITGα2, ITGα3, and ITGβ1 were higher in the outgrowth while only CD44 was lower in the outgrowth in both cell types (Figures 5A,B). Other integrins were either higher (ITGαV), lower (ITGα5, ITGβ3) in the outgrowth of MOSE-L or were not different (ITGαα4). In contrast, in MOSE-L_TICv we observed a significant increase in ITGα5 in the outgrowing cells and a reduced expression of ITGα4 while the expression levels of ITGαV, ITGα6, and ITGβ3 was not different from the original monolayers. Interestingly, the intracellular localization did not change for most integrins. Most integrins were expressed both in the cytosol and in the nuclei while ITGα2I and ITGβ3 were mostly expressed in the cytosol. Only ITGα3 and ITGβ3 expression changed from mostly cytosolic to mostly nuclear in MOSE-L_TICv outgrowth (Figure 5A). In MOSE-E cells, integrins were expressed in the cytosol or at the focal adhesions but not in the nucleus. This focal adhesion localization was not observed in the cancer cells in either the monolayers or the outgrowths. This correlates well with the low focal adhesion number and size in the cancer cells we have observed before (Creekmore et al., 2013). These results suggest a role of the nuclear localization of integrins in the transformation and progression of ovarian cells and point to both an adaptation to culture conditions and the return to the expression patterns of the original monolayer for some integrins but a differential role of specific integrins in the outgrowing cells from the attached spheroids in a cell type-specific manner.

To determine which integrin expression is critical for aggregation and adhesion, we used commercially available inhibitors against differentially expressed integrins in concentrations that stunted growth but did not affect cell viability (Supplementary Figure S1) and monitored MOSE aggregation and adhesion. As shown in Figure 6A, the MOSE cells form a single, tight spheroid after 24 h incubation in low oxygen and glucose conditions. Inhibition of ITGα2β1 with BT3033 prevented aggregation in both cell types in concentrations from 0.5 to 5 μM (Figure 6A and data not shown) while multiple small spheroids still formed in MOSE-L_TICv . However, neither TC-I, also an ITGα2β1 inhibitor, nor inhibition of the ITGα2 (with E7820) or ITGβ1 (with AllB2) monomers suppressed aggregation. Similarly, inhibition of ITGα5β1, ITGαVβ3, or ITGαV did not prevent aggregation. Only CycloY (RGDyK), a specific ITGαVβ3 inhibitor, suppressed aggregation in MOSE-L at 15 μM while the fusion to a single spheroid but not aggregation per se was observed in MOSE-L_TICv . The multiple spheroid formation was also apparent in the treatment with the other integrin inhibitors (Figure 6A). These results suggest that either only the ITGα2β1 heterodimer is critical for aggregation, or the inhibitors do not effectively suppress integrin signaling.

We next investigated the effect of these inhibitors on adhesion and outgrowth. Inhibition of ITGα2β1 with BT3033 at 0.5 μM had no effect in either MOSE line while 5 μM (the dose that prevented aggregation) inhibited adhesion and, subsequently, outgrowth and viability of the MOSE spheroids (Figures 6B,C). In contrast, neither TC-I nor E7820 nor AII2B inhibited adhesion and had little impact on viability. However, while the outgrowth of MOSE-L_TICv spheroids was not inhibited, the cells growing from MOSE-L spheroids had a migratory phenotype rather than the outgrowth as a sheet as observed in the controls. Inhibition of ITGαVβ1 with ITG-INT-FA or ITGαVβ3 with (Cyclo (RGDK)) did not suppress adhesion but reduced drastically the outgrowth in MOSE-L and decreased the viability with less effect on MOSE-L_TICv (Figures 6B,C). Inhibition of ITGαV using GLPG0187 completely prevented adhesion and outgrowth in both MOSE-L and MOSE-L_TICv . Inhibition of ITGα5β1 did not prevent adhesion but spheroid outgrowth and viability were reduced in MOSE-L. Thus, ITGαV may play a dominant role in cancer cell adhesion and outgrowth while inhibition of other integrins may reduce the activation of signaling pathways that induce proliferation, often in a cell-type specific manner. This is currently being investigated in our lab.