Authors : Sophie Ayama-Canden, Rodolfo Tondo, Liliana Piñeros, Noëlle Ninane, Catherine Demazy, Marc Dieu, Antoine Fattaccioli, Tijani Tabarrant, Stéphane Lucas, Davide Bonifazi, Carine Michiels
25 June 2022 – https://www.sciencedirect.com/science/article/pii/S1476558622000434
Abstract
In the context of breast cancer metastasis study, we have shown in an in vitro model of cell migration that IGDQ-exposing (IsoLeu-Gly-Asp-Glutamine type I Fibronectin motif) monolayers (SAMs) on gold sustain the adhesion of breast cancer MDA-MB-231 cells by triggering Focal Adhesion Kinase and integrin activation. Such tunable scaffolds are used to mimic the tumor extracellular environment, inducing and controlling cell migration. The observed migratory behavior induced by the IGDQ-bearing peptide gradient along the surface allows to separate cell subpopulations with a “stationary” or “migratory” phenotype. In this work, we knocked down the integrins α5(β1) and (αv)β since they are already known to be implicated in cell migration. To this aim, a whole proteomic analysis was performed in beta 3 integrin (ITGB3) or alpha 5 integrin (ITGA5) knock-down MDA-MB-231 cells, in order to highlight the pathways implied in the integrin-dependent cell migration.
Our results showed that i) ITGB3 depletion influenced ITGA5 mRNA expression, ii) ITGB3 and ITGA5 were both necessary for IGDQ-mediated directional single cell migration and iii) integrin (αv)β3 was activated by IGDQ fibronectin type I motif. Finally, the proteomic analysis suggested that co-regulation of recycling transport of ITGB3 by ITGA5 is potentially necessary for directional IGDQ-mediated cell migration.
Introduction
The poor prognosis of breast cancer is mainly due to metastasis formation. All stages combined, survival at 5 years is 90% for breast cancer. For triple negative breast cancer the survival rate at 5 years drop to 77% all stages combined and to 12% when a distant metastasis is detected. Metastasis formation is a multistep process in which cancer cells migrate out of the primary tumors and form secondary tumors in distant sites. Cell migration is a complex mechanism crucial for many physiological and pathological processes including tissue regeneration, immune response and cancer progression.
Multiple elements modulate cancer cell migration including the tumor microenvironment composed of immune cells, endothelial cells, fibroblasts, soluble elements (chemotaxis) and interaction with the extracellular matrix (ECM – haptotaxis). Cancer cell migration occurs through different processes, divided in two subtypes: collective cell migration and single cell migration. In collective migration, cells migrate in multicellular strands or in sheets when transformed cells are invasive but not yet metastatic, as observed in some epithelial cancers. In later stages, cell clusters can detach, disseminate and initiate metastasis locally, as observed in melanomas, breast, colon and prostate cancers. Single cancer cell migration occurs as mesenchymal-cell like migration or with an amoeboid behavior. The type of cancer cell migration is influenced by the cancer type, cell-ECM and cell-cell interactions. In general, the structure, composition and physical properties of the ECM highly influence cell migration through durotaxis, haptotaxis and contact guidance. Hence, the interplay and crosstalk in the composition of the ECM also play a major role in the regulation and the behavior of cells, especially during migration.
Cells can detect chemical and physical gradients, both on surfaces and in solution, and respond to them with oriented movement. Indeed, cell motility is affected by chemoattractant such as Epithelial Growth Factor (EGF), Transforming Growth Factor (TGF-β), collagen and fibronectin (Fn). It was previously shown, in a microfluidic system, that the oriented cell migration of breast cancer cells depends not only on the gradient profile but also on the range of EGF concentrations. Furthermore, Fn precisely modulates cell migration. This process occurs through its interaction with specific transmembrane proteins, the integrins. The insoluble cellular dimer fibronectin entangles with other ECM insoluble components such as collagen, fibrin and laminin fibers, hence forming a major constituant of the insoluble ECM. It is mainly produced by fibroblasts (stromal cells), but it is also synthetized by cancer cells themselves. Multiple variants of fibronectin exist, expressed from a single pre-mRNA, spliced to generate 20 variants in human. Hence, fibronectin isoforms carry different modules containing peptidic motifs, which can induce specific biological processes via the activation of specific integrins (Figure 1-A). The type and abundance of the three types of Fn in the ECM depend on local and global microenvironment context, promoting or not cell survival, proliferation and migration. The fibronectin is subdivided into domains. Fn domains are subdivided into modules and modules contain specific peptidic motifs (Figure 1-A). The type I Fn module contains IGD units (Q or S), which are motogenic. These motifs can activate integrins and induce cell migration. The type II Fn module mostly participates in cell adhesion to ECM. It also contains domains for fibrin and collagen binding. Type II Fn module is the collagen-binding domain. The minimum domain size contains one module type I Fn and two modules type II Fn. Type III Fn module is the most abundant one and is the most studied until now. It contains fibulin-1, heparin and syndecan binding domains. It also contains the RGD motif which promotes cell proliferation and adhesion by triggering specific integrin activation and cell signaling cascade response, through focal adhesion formation.
Until now, the interplay between specific integrin-mediated ECM adhesion and directional single cell migration is not well understood. In particular, the efforts to understand and guide cellular migration have led to the development of various approaches to generate confined and surface-immobilized chemical gradients able to mimic natural 3D environment and control cell migration in vitro. Several works reported the use of Fn fragments in combination with mixed self assembly monolayers (SAMs) of alkanethiols to promote the biospecific attachment of cells through the integrin-based recognition. Upon demonstrating that low concentrations of GBD (gelatin binding domain) stimulate the migration of human dermal fibroblasts into 3D-gels of native type I collagen fibers, Schor et al attributed the biological activity of GBD to the highly conserved Isoleucine-Glycine-Aspartic acid (IGD) motif of Fn. In order to characterize the oriented migratory phenotype, we previously developed SAMs exposing gradients of IGDQ gradients (Figure 1 B–C). We could thus identify two cell subpopulations associated either with a “stationary” phenotype or with a “migratory” phenotype, the latter determining a considerable cell migration at the sub-cm length scale (Figure 1-C). One of the main limitation to analyze the gene expression of metastatic cells and to identify the pathways involved in metastasis formation was the segregation of the non-migrating cells from those displaying migrating aptitude without losing their phenotype. Our engineered surfaces presented a response to that limitation.
Building on these discoveries, we focused on the understanding of the role of integrins α5β1 and αvβ3 in the recognition with IGDQ, since they are described as essential for cell migration and invasion. Integrins are transmembrane proteins formed by two subunits, one alpha and one beta (Figure 2-A). Each integrin or group of integrins can interact with different elements of the ECM, including fibronectin, collagens or laminins (Figure 2-A). In order to evaluate and characterize the biological migration process induced by the motogenic motif IGDQ through integrin activation, the breast cancer cell line MDA-MB-231 was used. This cell line is epithelial with a mesenchymal-like phenotype that has already undergone the EMT and is frequently used to study cell migration and metastasis formation. Moreover, this cell line constitutively expresses both α5(β1) and (αv)β3 integrins.
With the aim to unravel pathways involved in integrin-dependent cancer cell migration and the motogenic activity of the IGDQ sequence in metastatic processes, we a) studied the role and interaction of α5(β1) and (αv)β3 integrins in a wound healing assay and in single cell migration and b) performed a proteomic analysis coupled to bioinformatics analyses. Altogether, the results identified new integrin-dependent pathways that regulate the migration process using a complete proteomic analysis.
Results
Beta 3 integrin depletion influences integrin alpha 5 expression
To investigate the role of α5β1 and αvβ3 in cell migration, a constitutive knock-down strategy for alpha 5 (ITGA5) and beta 3 (ITGB3) integrin subunits, using shRNA-expressing vectors, was used in MDA-MB-231 metastatic breast cancer cells. Indeed, ITGA5 only can dimerize with ITGB1 and ITGB3 with ITGAV. RNA knock-down efficiency was evaluated by measuring mRNA level using qPCR (Figure 2–B) and protein level using western blotting (Figure 2–D–E) and immunofluorescence labeling (Supplemental Figure 1–C–D). ITGA5 and ITGB3 knock-down did not show any significant impact on ITGAv and ITGB1 mRNA expression (Figure 2–C), nor protein abundance as assessed from the mass spectrometry proteomic data (data not shown). shITGA5-126 led to an efficient knock-down with a decrease in ITGA5 mRNA level of 62 % (p < 0.1) and in protein of abundance 72.7 % (p < 0.001). This knock-down was used because we observed that higher knock-down level induced cell detachment due to the well-known involvement of ITGA5 in cell adhesion, thus making impossible the use of such cells for experiments.
For shITGB3-237, a decrease of ITGB3 83 % in mRNA level (p < 0.01) and of 81.1 % in protein abundance (p < 0.001) were obtained. These results were confirmed by immunofluorescence labeling (Supplemental Figure 1–C–D). These results were confirmed using another shRNA for ITGB3 (Supplemental Figure 2).
Interestingly, ITGA5 expression increased in ITGB3-knock-down cells at both RNA and protein levels (Figure 2–B–C–D–Supplemental Figure 1 C). This suggests an indirect negative transcriptional regulation of ITGA5 by ITGB3 expression, while the reverse was not observed.
Regarding proliferation, no significant difference was observed in knock-down cells compared to control cells (shCTL – Supplemental Figure 1–A–B) and no cytotoxic effect of shRNAs was demonstrated (LDH assays – data not shown). This suggests that ITGA5 or ITGB3 depletion has no deleterious impact on cell cycle.
Integrin beta 3 and integrin alpha 5 are implicated in a wound healing assay
Collective migration was evaluated using a wound healing assay during 48h, with cell proliferation blockage (mitomycin C) (Figure 3). The control cells filled 87% of the wound in 48h while it was not the case for ITGA5 and ITGB3 knock-down cells for both shRNAs (Figure 3–A, Supplemental Figure 2E). shITGB3 cells were able to close only 22% of the scratch (Figure 3–B) and shITGA5 cells 33%, with a significant reduction of relative closing compared to shCTL cells (Figure 3–C). These results indicate that ITGA5 and ITGB3 were needed for cell migration in a wound healing assay.
Integrin beta 3 and integrin alpha 5 are both necessary for IGDQ-mediated directional single cell migration
To investigate the role of ITGA5 and ITGB3 on IGDQ-mediated directional cell migration, cells were seeded on the motogenic surfaces (Figure 1) and the migration and proliferation of the cells were monitored by taking holographic images each day for 5 days (Figure 4).
Normal proliferation was observed in shCTL, shITGA5 and shITGB3 cells. Migrating shCTL cells were observed along the IGDQ gradient. However, no migrating cells were observed in both shITGA5 and shITGB3 cells, as shown by the quantification of the number of migrating cells in three independent experiments (Supplemental Figure 3). This was confirmed by using a second shRNA for ITGB3 (Supplemental Figure 4). This inhibition is statistically significant for both knock-down cell lines. The results showed that both ITGA5 and ITGB3 are necessary for the IGDQ-triggered single cell migration.
Integrin (αv)β3 is activated by IGDQ fibronectin type I motif
To investigate the integrin activation on IGDQ-exposing surfaces, shCTL, shITGA5 and shITGB3 cells were seeded on the surfaces for 5 days (Figure 1). The cells were seeded at day 1 where IGDQ peptide concentration is the lowest (Figure 1–A–B). Cells were incubated to permit their proliferation and migration during 5 days and we were able to obtain cells presenting three phenotypes: “static phenotype” for proliferating cells (low IGDQ concentration), “late migratory phenotype” and “migratory phenotype” for migratory cells that followed the IGDQ gradient concentration (Figure 1–C). At this step, cells were fixed and labeled for ITGA5 / F-actin (Figure 4–A–B–C, Supplemental Figure 5) or ITGB3 / F-actin (Figure 4–D–E–F, Supplemental Figure 6). When integrins are activated, focal adhesion complexes are observed at the plasma membrane.
In proliferating cells, no clusters were observed neither for ITGA5 nor for ITGB3 in any of the three cell lines. For both integrins, the labeling showed the presence of the protein at the plasma membrane, well distributed all over the surface of the cells as small dots (Figure 4). Migrating cells were observed only in shCTL cells. In these cells, ITGA5 labeling was observed all over the membrane as in proliferating cells (Figure 4–A). No clusters of ITGA5 were observed. This suggests that integrin α5(β1) was not activated by the IGDQ motif.
ITGB3 labeling in migrating cells showed large clusters at the membrane with a high density of ITGB3, corresponding to focal adhesion complexes due to integrin beta 3 activation (Figure 4–D). A light staining of ITGB3 at the rest of the membrane is also visible. This suggests that integrin (αv)β3 is activated by IGDQ fibronectin type I motif in cells with a migratory phenotype, but not in cells with a proliferative phenotype.
Integrin alpha 5 is activated by RGD type III fibronectin motif
To be sure that integrin alpha 5 kept its capacity to cluster in our model, wild type MDA-MB-231 (WT), shCTL or shITGB3 cells were seeded on classical culture plates or on high concentration fibronectin coated surfaces. On these surfaces, fibronectin exposes the RGD motif. The labeling of ITGA5 was performed 48 hours later (Figure 5).
On classical culture plates, ITGA5 labeling was observed all over the plasma membrane as observed on engineered surfaces, for both shCTL and shIGTB3 cells (Figure 5–A–C). On Fn type III coated surfaces in WT cells, ITGA5 staining showed cross-bars, perpendicularly to the membrane (Figure 5–B–D). This is due to integrin alpha 5 activation and clustering. Small dots of ITGA5, with a light staining, are also visible on the whole membrane surface. These results indicate that ITGA5 is activated by high concentrations of Fn type III, containing RGD motifs. Moreover, the ITGA5 clustering is not affected in shCTL cells, neither by the absence of ITGB3 (Figure 5 D).
In parallel, ITGB3 staining also performed on WT cells. On both uncoated and Fn type III coated surfaces, ITGB3 staining showed the homogenous presence of the protein at the membrane appearing as small dots (Figure 5–E–F). This suggests that integrin (αv)β3 is not activated by high Fn type III module that contains the RGD motifs.
Co-regulation of recycling transport of integrin beta 3 by integrin alpha 5 is potentially necessary for directional IGDQ-mediated cell migration
To deepen the investigation and to understand the interactions between ITGA5 and ITGB3, a whole proteomic analysis was performed in ITGA5 and ITGB3 knock-down cells according to the workflow presented in Supplemental Figure 7. Then for each experiment, proteins significantly differentially expressed were analyzed in a unsupervised hierarchical comparative gene ontology analysis to identify the ones that are enriched. Gene ontology is a bioinformatics tool that permits to cluster genes or proteins under a term, this term describing either a biological process (pathways), a molecular function (protein binding) or a cellular component (protein localization). Gene ontology clusters significantly enriched (FDR > 0.1) were then clustered using ClueGo (Cytoscape tool, consulted on 2020-01). In this study, we choose to globally integrate all significant proteins into cellular processes or subcellular localization clusters. We performed gene ontology (GO) enrichment analysis of the identified differentially expressed proteins to gain insights into the cellular functions and biological processes that are affected by integrin alpha 5 or integrin beta 3.
For each comparison, shCTL vs shITGA5 and shCTL vs shITGB3, the number of proteins detected (Supplemental figure 8–A–C), volcano plot representing the repartition of proteins expressed obtained from quantitative proteomics analysis where green dots represent proteins exhibiting significant fold changes (Supplemental figure 8–B–D), top 10 dysregulated proteins (Figure 6–7–A), top 10 upregulated (Figure 6–7–B) and top 10 downregulated proteins (Figures 6–7–C), the top 15 gene ontology categories impacted according to the percentage of their dysregulated proteins (Figures 6–7–D), the gene ontology categories (GO) impacted by the knock-down (ClueGo trees – Figures 6–7–E) and the repartition of major enriched gene ontology clusters for biological processes (Figures 6–7–F) or for cellular localization (Figures 6–7–G) were obtained using bioinformatics analysis (Figures 6–7). About 2500 proteins were identified in each experiment, with 142 proteins significantly dysregulated in shITGA5 cells and 245 in shITGB3 cells compared to shCTL cells (Supplemental Table 1). It is interesting to notice that the list of the top dysregulated proteins (Figures 6–7–A–B–C) and the list of the top gene ontology categories impacted (Figures 6–7–C) do not always reflect the global impact of the knock-down. Indeed, the main dysregulated proteins are perhaps not the most interesting to study, while the panorama allows to focus both on a specific pathway and related proteins, at the level of biological processes (Figures 6–7–E–F) and cell localization (Figures 6–7–G).
Integrin alpha 5 is linked to vesicle-mediated transport
The invalidation of ITGA5 induced modifications in the expression of proteins involved in multiple biological processes (Figure 6–D–E–F–G; Supplemental table 2). Our analyses revealed in that GO categories and hub proteins significantly enriched were linked to vesicle-mediated transport regulation, catabolic processes, nucleoplasmic transport, cell-substrate junction and cell isolation stress. Interestingly, two GO category clusters impacted by ITGA5 knock-down are similarly ranked in the top 4 modified biological processes (Figure 6–F) and top 2 modified subcellular localizations GO categories (Figure 6–G). This two GO clusters are: endoplasmic reticulum – Golgi apparatus and catabolic process, the latter including proteasome complex, positive regulation of cellular catabolic process and protein deubiquitination.
First, around one-fourth of biological processes and one-third of protein cellular localization GO categories are associated endoplasmic reticulum (ER) to Golgi-vesicle mediated transport (Figure 6 D–E–F). Moreover, “vesicle cargo loading” appears in the top 15 impacted GO categories (Figure 6–D). It is also represented by top downregulated proteins such as the calcium binding protein PDCD6, the GTPase-activating protein ArfGAP3 and the glycogen debrancher enzyme AGL (“Golgi to endoplasmic reticulum retrograde transport regulation Ca2+ dependent signaling pathway”) and top upregulated protein like the valosin-containing protein (VCP)-interacting molecule ANKRD13A (lysosomal trafficking) (Figure 6–A to C). This suggest the involvement of ITGA5 in vesicle-mediated transport regulation.
Second, ITGA5 knock-down cells was also enriched with one-fourth biological processes and one-third protein cellular localization GO categories linked to the catabolic processes at mRNA and protein levels (Figure 6–E–F–G). Moreover, multiple GO categories linked to this hub appears in the top 15 impacted GO categories including “regulation of RNA binding” and “pre-mRNA cleavage required for polyadenylation” (Figure 6–D). It is represented by top downregulated proteins such as the nucleotide excision repair protein RAD23A (proteasome complex), the aminoacyl-tRNA synthase complex auxiliary component EEF1E1 (translation regulation) and the kinase PIK3C2A (endosomal maturation and autophagy).
Third, around one-seventh of biological processes and one-ninth of protein cellular localization GO categories are related to cytoskeleton regulation GO’s hub. This hub includes cell-substrate junction assembly, response to isolation stress (linked to cell detachment from other cells like in single cell migration), integrin complex and adherents junctions (Figure 6–E–F–G). Moreover, “response to isolation stress”, “cell adhesion mediated by integrins” and “cell-substrate junction assembly” appears in the top 15 impacted GO categories (Figure 6–D). It is also represented by top downregulated proteins such as the estrogen resistance protein BCAR3 (cell motility promotion, adhesion remodeling and cytoskeleton reorganization), the nucleotide excision repair protein (reduced characteristics of stemness-EMT) and the apoptotic suppressor proteins XIAP (pro-metastatic phenotype), and top upregulated protein like the cell motility mediator MEMO1 (lamellipodial actin network, adhesion site formation and microtubule outgrowth in cell motility) (Figure 6–A to C). Altogether, this suggests that ITGA5 has a role in cytoskeleton component modulation.
Fourth, ITGA5 knock-down cells were also enriched with one-fourth biological processes and 4% protein cellular localization GO categories, linked to nucleoplasmic transport (Figure 6–E–F–G). In the top upregulated proteins, was ranked the nucleoporin NUP37.
Integrin beta 3 linked to protein expression and metastatic cell capacity
The invalidation of ITGB3 induced modifications in the expression of proteins involved in multiple biological processes (Figure 7–D–E–F–G; Supplemental table 2). Our analyses revealed in ITGB3 knock-down cells GO categories and hub proteins significantly enriched were linked to EMT, mRNA splicing and catabolism, protein production. Interestingly, the two major GO hubs RNA modifications and protein translation represent more than four-fifth of biological processes and half of protein cellular localization (Figure 7–E–F–G).
Fig. 7. Relative differential protein expression analysis for beta 3 integrin knock-down cells. Raw data from mass spectrometry were processed to obtain protein differential expression in shB3-237 knock-down cells compared to control cells (shCTL) using total spectral counting and t-test with p-value < 0.05. A: Top 10 differentially expressed proteins; B: Top 10 upregulated proteins; C: Top 10 downregulated proteins;; D: Top 10 relevant gene onthology categories (GO) – biological processes (BP) modified ranked according to percentage of proteins dysregulated in GO category considered; E: String map of more relevant gene onthology categories (GO) – biological processes (BP) modified (p-value < 0.1, STRING, Cytoscape-GoClue); F: Major gene ontology – biological process clusters modified; G: Major gene ontology – cellular component clusters modified.
First, ITGB3 knock-down cells was enriched with two-fifth biological processes and one-fourth protein cellular localization GO categories linked to RNA modifications and metabolism (Figure 7–E–F–G). Moreover, multiple GO categories linked to this hub appear in the top 15 impacted GO categories including “cleavage involved in rRNA processing”, maturation of SSU-rRNA and RNA splicing. (Figure 7–D). The splicing factor SRRM1 (modifications in splicing of RNA) was found in the top downregulated proteins (Figures 7–A to C). This supports the involvement of ITGB3 in RNA regulation and maturation.
Second, around two-fifth of biological processes and one-fourth of protein cellular localization GO categories are associated to protein metabolism (Figure 7–D–E–F). Moreover, multiple GO categories linked to this hub appear in the top 15 impacted GO categories including ribosomal large subunit assembly and biogenesis (Figure 7–D). It is represented by top upregulated proteins such as the chaperone proteins CLGN and MESD (protein folding) (Figure 7–A to C). This supports the involvement of ITGB3 in protein translation and metabolism.
Third, around one-tenth of biological processes and two-fifth of protein cellular localization GO categories are associated to cytoskeleton organization (Figures 7–D–E–F). Indeed, cell cortex localization is the region beneath the plasma membrane that contains a network of actin filaments and associated proteins. In this hub, all GO categories related to cell motility and stemness are included. Top upregulated proteins such as the GTPase activating protein IQGAP2 (cell contractile regulation and invasiveness inhibition), the kinase TRRAP (stemness) and the demethylase ALKBH5 (invasiveness promotion) and top upregulated proteins like the methyltransferase NNMT (EMT regulation) are linked to these categories (Figures 7–A to C).
Fourth, around one-tenth of protein cellular localization GO categories are associated to melanosome, which is a vesicle that contains melanin in melanocytes (Figure 7–F). Top upregulated proteins such as the heat-shock co-chaperone DNAJC13 endosomal trafficking regulation) and top upregulated proteins like the GTPase ARL8B (lysosomal transport) and the exosome component EXOSC2 (exosome regulation) are included in this category (Figure 7–A to C). This suggests an implication of ITGB3 in the trafficking of multiple vesicle types.
We also analyzed if there were common differentially expressed proteins found in shITGA5 cells and in shITGB3 cells as compared to shCTL cells (Supplemental table 2). Eighteen proteins were identified of whose 17 were similarly up-regulated or down-regulated in both cell lines.
Discussion
Integrins are major players of metastatic process. In this work, we focused on cancer cell migration and we studied the role of alpha 5 and beta 3 integrin subunits. Both integrins are described to be activated by fibronectin. In previous other studies, using siRNA or shRNA knock-down, it was shown that both integrins are implied in cell migration. ITGB3 is described to be linked to directional cell migration and ITGA5 to random cell migration. Cell migration is regulated by ECM elements either in tissues or on artificial coated surfaces. Indeed, ITGA5 was described to be implied in regulating random cell migration on high-fibronectin coated surfaces while ITGB3 is implied in directional migration on low-fibronectin coated surfaces. Except in studies on embryo development or on integrin trafficking, all other works studied these integrins separately regarding their impact on cell migration. In this work, we studied in parallel the impact of both integrins αvβ3 and α5β1 in a triple breast cancer cells in in vitro model, and more specifically regarding their implication in IGDQ-mediated cell migration. To achieve this objective, we knocked down each one of the two subunits using shRNA. The collective migration was evaluated using a wound healing assay on uncoated surfaces and single cell migration performed on the surface exposing an IGDQ gradient. A complementary full proteomic analysis was performed to determine the interplay between both integrins and their role in cancer cell migration. Using bioinformatics analyses, an overview of the potential roles of the integrins of interest has been unraveled. An experimental validation will be necessary to confirm this hypothesis.
In this study, we showed that ITGB3 and ITGA5 are necessary for cell migration in a wound healing assay. Moreover, we discovered that ITGB3 expression induced a negative control on ITGA5 expression at mRNA and protein levels. To the best of our knowledge, no previous study demonstrated a mutual regulation between beta 3 and alpha 5 integrin expression. This might be explained by a compensation system due to the loss of ITGB3, while the reverse is not true, i.e. alpha 5 integrin does not seem to influence integrin beta 3 expression. The study of miRNAs can be a good path to explain for the overexpression of ITGA5 observed in integrin beta 3 knock-down cells.
The motogenic effect of IGD(Q) motif from fibronectin type I module was first described in 1999 by Schor & al. showing an activation of fibroblast migration using soluble IGD peptides. This activity depends on functional integrin αvβ3, the phosphorylation of FAK at the tyrosine position 125 and the inhibition of α5β1 mediated signaling. Moreover, IGDQ was linked only to cell migration and not to cell adhesion onto Fn, unlike RGD motif fibronectin type III module. Our results showed that IGDQ fibronectin type I motif activates ITGB3 during single cell migration and not during cell proliferation while ITGA5 clustering is induced by Fn type III containing RGD motif. Interestingly, the absence of ITGA5 induced the loss of cell migration capacity on IGDQ-exposing surfaces, despite the fact that ITGB3 is expressed and localized at the cell membrane. It was previously shown that the transduction cascade following the IGDQ exposure to fibroblast is inhibited by signaling mediated through integrin α5β1. Coupled with our results, this suggests a necessity to have this α5β1 and αvβ3 signaling co-regulation to permits the IGDQ-mediated cell migration. It has to be noted that RhoA and Rac1 protein abundance was not affected by ITGA5 or ITGB3 knock-down (data not shown), indicating that the modification observed in cell migration was not due to a change in the expression of these two proteins.
Our proteomic results also pointed out a modification in RNA translation, splicing and metabolism in shITGB3 cells. Links between RNA translation and focal adhesion complexes formed by integrins activation were already described since 20 years. Chicurel & al. showed the localization of ribosomal complexes near focal adhesion complexes, suggesting a new type of gene regulation by integrins and by mechanical stress. The spatial localization of translation facilitate local protein enrichment at their sites of function. This permits to have a physical proximity of translated proteins and their binding partners. The co-localization / compartmentalization of translation related proteins (ribosomes, EIF4E – EIFE4G), microtubules, microfilaments and focal adhesion (vimentin) at the lamelipodia of migrating fibroblasts was evidenced by Willett & al. These data showed a direct link between cell migration and this focal adhesion/integrin-mediated translation. The same team demonstrated a preferential association between integrin beta 3 and 40S ribosomal subunits, all along the lamelipodia in migrating fibroblasts. In our integrin beta 3 knock-down model, EIF5B (Eukaryotic translation initiation factor 5B), one of the proteins necessary for the translation initiation was found downregulated with 0.8 fold-change (p-value: 0.0044). Interestingly, EIF5B binds to EIF4E and can be connected to ITGB3 via the mTOR-EGFR pathway. This suggests a potential integrin-mediated translation dependent of ITGB3 and EIF5B.
The loss of migration capacity on IGDQ-exposing surface can be linked to a defect of ITGB3 clusterisation in ITGA5 knock-down cells, which is dependent on integrin recycling trafficking. These results, added to the results of our proteomic analysis, suggest that ITGA5 regulates ITGB3 trafficking, more specifically the recycling one, that allows to form ITGB3 clusters after its activation. Moreover, ITGB3 is found at the membrane indicating that the membrane addressing is not modified. On the other hand, the reverse compensating regulation was not observed (Figure 5). Indeed, the loss of ITGB3 did not modify ITGA5 clustering upon α5(β1) activation on Fn coated surfaces (RGD motif). It has been shown that the recycling transport of α5β1 and αvβ3 is co-dependent. The transport of α5β1 inhibits the transport αvβ3 and reciprocally. But from our knowledge, no study indicates this potential necessary presence of α5β1 to allow the αvβ3 recycling transport. It was shown that αvβ3 blockage in rich-fibronectin environment promotes α5β1 dependent random cell migration via filopodia. At the same time αvβ3 activation inhibits α5β1-dependent cell migration and induces directional migration in low-fibronectin environment. All those processes are Rab-dependent and also implied the lysosomal regulation and trafficking. It would be interesting to study the trafficking of both integrins during cell migration on IGDQ-exposing surfaces. Furthermore, since this study has been performed using MDA-MB-231 cells, it would also be interesting to study other cell lines, with different in vitro migration properties and/or different metastatic capacities when injected in vivo.
The trafficking of integrins was described, at different levels, as essential for cellular migration and cancer metastasis promotion. Indeed, it was shown that exosome containing integrins are implied in cell migration through a crosstalk with stromal fibroblast cells, but also in organotropy where exosomes containing α6β4 and α6β1 were linked to lung metastasis, while exosomes containing integrin αvβ5 were linked to liver metastasis. Furthermore, the endocytosis of integrins, permitting their recycling, plays a major role in cell migration. Using surface gradients allowed to highlight the mechanisms involved in integrins trafficking in cell migration triggered by IGDQ and, in the future, will help to highlight potential targets to prevent metastasis in cancer patients.