Figure 1.
FAK activity is required for efficient collective cell movement of squamous cell carcinoma (SCC) cells. (A,B) Trajectory and directionality of collective migration of SCC cells in collagen-embedded cell spheroids. Three-dimensional reconstruction of Z-stacked images of spheroids assembled using SCC42B (A) or FRNK-SCC42B (B) cells and embedded into a collagen matrix. Images of freshly assembled spheroids (0 h) and spheroids after a 24 h migration period (24 h) were captured using confocal reflection microscopy. The resulting images represent a stack of 28 sections (Z step of 2.8 μm) with a total physical length of 140 μm (0 h) or 20 sections with a total physical length of 100 μm (24 h). A yellow pseudocolor and an xz projection (24 h far left pictures) of the same spheroid was used to better visualize the entire spheroid’s area. (C–H) Representative images from time-lapse movies of SCC42B (C–E) or SCC38 (F–H) cells embedded as spheroids into collagen matrix and allowed to migrate for 24 h. Ctrl: cells expressing empty vector (C,F) or cells treated with PF-562271 vehicle (E,H); FRNK: FRNK-expressing cells; siCtrl and siFAK: SCC cells after treatment with control or FAK siRNAs, respectively; PF-562271 FAK inhibitor was added to cells at 2 or 5 μM final concentrations (E,H). Graphics at the right of panels (C–H) represent variations over time of the spheroid cross-sectional areas (SCSAs) of the spheroids assembled with the indicated cells. For SCSA quantification, at least 10 spheroids from at least 3 independent experiments were used. Scale bars: 100 µm. **** indicates p < 0.0001.
Figure 1.
FAK activity is required for efficient collective cell movement of squamous cell carcinoma (SCC) cells. (A,B) Trajectory and directionality of collective migration of SCC cells in collagen-embedded cell spheroids. Three-dimensional reconstruction of Z-stacked images of spheroids assembled using SCC42B (A) or FRNK-SCC42B (B) cells and embedded into a collagen matrix. Images of freshly assembled spheroids (0 h) and spheroids after a 24 h migration period (24 h) were captured using confocal reflection microscopy. The resulting images represent a stack of 28 sections (Z step of 2.8 μm) with a total physical length of 140 μm (0 h) or 20 sections with a total physical length of 100 μm (24 h). A yellow pseudocolor and an xz projection (24 h far left pictures) of the same spheroid was used to better visualize the entire spheroid’s area. (C–H) Representative images from time-lapse movies of SCC42B (C–E) or SCC38 (F–H) cells embedded as spheroids into collagen matrix and allowed to migrate for 24 h. Ctrl: cells expressing empty vector (C,F) or cells treated with PF-562271 vehicle (E,H); FRNK: FRNK-expressing cells; siCtrl and siFAK: SCC cells after treatment with control or FAK siRNAs, respectively; PF-562271 FAK inhibitor was added to cells at 2 or 5 μM final concentrations (E,H). Graphics at the right of panels (C–H) represent variations over time of the spheroid cross-sectional areas (SCSAs) of the spheroids assembled with the indicated cells. For SCSA quantification, at least 10 spheroids from at least 3 independent experiments were used. Scale bars: 100 µm. **** indicates p < 0.0001.

Figure 2.
Cancer-associated fibroblasts (CAFs) promote collective invasion of FRNK-SCC cells by acting as leaders of the invasive fronts. (A,C) Representative images of spheroids assembled with FR-SCC cells or with mixed populations of SCC cells and CAFs as indicated and incubated for 20 h (A) or 6 days (C). CAFs were labeled with green 5-chloromethylfluorescein diacetate (CMFDA) to allow their tracking over time. Magnified images in panel (A) (a) and (C) (a–c) are shown to better visualize the leader CAFs (denoted by white arrows) at the tip of the invasive cell tracks. (B) Variations over time of the spheroid cross-sectional area (SCSA) of the spheroids assembled with the indicated cells.
Figure 2.
Cancer-associated fibroblasts (CAFs) promote collective invasion of FRNK-SCC cells by acting as leaders of the invasive fronts. (A,C) Representative images of spheroids assembled with FR-SCC cells or with mixed populations of SCC cells and CAFs as indicated and incubated for 20 h (A) or 6 days (C). CAFs were labeled with green 5-chloromethylfluorescein diacetate (CMFDA) to allow their tracking over time. Magnified images in panel (A) (a) and (C) (a–c) are shown to better visualize the leader CAFs (denoted by white arrows) at the tip of the invasive cell tracks. (B) Variations over time of the spheroid cross-sectional area (SCSA) of the spheroids assembled with the indicated cells.
Figure 3.
Effect of anti-MMP-2 antibodies on the collective migration of mixed SCC + CAF spheroids. (A,D) Representative images of mixed spheroids containing control (C-SCC42B or C-SCC38) cells and CMFDA-labelled CAFs and incubated for 0, 12 or 24 h in the absence of antibodies (A,C), or in the presence of control anti-SDHB (ctrl ab) or anti-MMP-2 antibodies (B,D). In panels (B–D), bright field and fluorescence images of the same spheroids are shown. (E) Variations over time of the SCSA of the spheroids assembled with C-SCC or C-SCC + CAF. Data are presented as mean ± standard deviation from 3 individual experiments; at least, 15–30 spheroids were used for quantifications. SP, spheroid. Scale bars, 100 µm. **** indicates p < 0.0001.
Figure 3.
Effect of anti-MMP-2 antibodies on the collective migration of mixed SCC + CAF spheroids. (A,D) Representative images of mixed spheroids containing control (C-SCC42B or C-SCC38) cells and CMFDA-labelled CAFs and incubated for 0, 12 or 24 h in the absence of antibodies (A,C), or in the presence of control anti-SDHB (ctrl ab) or anti-MMP-2 antibodies (B,D). In panels (B–D), bright field and fluorescence images of the same spheroids are shown. (E) Variations over time of the SCSA of the spheroids assembled with C-SCC or C-SCC + CAF. Data are presented as mean ± standard deviation from 3 individual experiments; at least, 15–30 spheroids were used for quantifications. SP, spheroid. Scale bars, 100 µm. **** indicates p < 0.0001.
Figure 4.
MMP-2 promotes collective cell invasion of FAK-deficient and proficient SCC cells. (A–D) Spheroids were assembled with the indicated SCC cells infected with empty vector (Ctrl) or overexpressing FRNK, MMP-2 or FRNK + MMP-2 and incubated for 0 or 24 h. (E) Variations over time of the SCSA of control (C), FRNK (FR), MMP-2 (M) or FRNK + MMP-2 (FM) spheroids. Data are presented as mean ± standard deviation from 3–5 individual experiments. Scale bars, 100 µm. * indicates p < 0.01, ** indicates p < 0.001, **** indicates p < 0.0001.
Figure 4.
MMP-2 promotes collective cell invasion of FAK-deficient and proficient SCC cells. (A–D) Spheroids were assembled with the indicated SCC cells infected with empty vector (Ctrl) or overexpressing FRNK, MMP-2 or FRNK + MMP-2 and incubated for 0 or 24 h. (E) Variations over time of the SCSA of control (C), FRNK (FR), MMP-2 (M) or FRNK + MMP-2 (FM) spheroids. Data are presented as mean ± standard deviation from 3–5 individual experiments. Scale bars, 100 µm. * indicates p < 0.01, ** indicates p < 0.001, **** indicates p < 0.0001.
Figure 5.
Mixed spheroids containing unlabeled FR-SCC cells plus either CMFDA-labelled C-SCC or CMFDA-labelled M-SCC cells were incubated for 14 or 20 h in the absence of antibodies or in the presence of control (anti-SDHB) or anti-MMP-2 antibodies. (A,B) Representative bright field and fluorescence images of the indicated mixed spheroids. Dashed white lines are indicated in the bright field images to delineate the periphery of the spheroid. Green arrows point to CMFDA-labelled SCC cells located at the front invasive positions. (C) Variations over time of the SCSA of the indicated spheroids. C: control cells, FR: FR-SCC cells, FR + C and FR + M: mixed spheroids containing unlabeled FR-SCC cells plus CMFDA-labelled C-SCC or CMFDA-labelled M-SCC cells, respectively. C + M: mixed spheroids containing unlabeled C-SCC cells plus CMFDA-labelled M-SCC. Data are presented as mean ± standard deviation from 3 individual experiments. Scale bars, 100 µm.
Figure 5.
Mixed spheroids containing unlabeled FR-SCC cells plus either CMFDA-labelled C-SCC or CMFDA-labelled M-SCC cells were incubated for 14 or 20 h in the absence of antibodies or in the presence of control (anti-SDHB) or anti-MMP-2 antibodies. (A,B) Representative bright field and fluorescence images of the indicated mixed spheroids. Dashed white lines are indicated in the bright field images to delineate the periphery of the spheroid. Green arrows point to CMFDA-labelled SCC cells located at the front invasive positions. (C) Variations over time of the SCSA of the indicated spheroids. C: control cells, FR: FR-SCC cells, FR + C and FR + M: mixed spheroids containing unlabeled FR-SCC cells plus CMFDA-labelled C-SCC or CMFDA-labelled M-SCC cells, respectively. C + M: mixed spheroids containing unlabeled C-SCC cells plus CMFDA-labelled M-SCC. Data are presented as mean ± standard deviation from 3 individual experiments. Scale bars, 100 µm.
Figure 6.
FAK over-expression not only induces collective cell invasion but favors the emergency of cells that detach from the spheroids. (A) Representative data showing the efficiency of GFP-FAK expression in transiently transfected cells. Representative immunoblot analysis of FAK and pFAK in mock- and GFP-FAK-transfected cells is shown in the middle. The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratio FAK or pFAK/β-actin was estimated by densitometry. Graphics show relative quantification of FAK mRNA, and FAK and pFAK proteins in the indicated cells. (B) Representative images of spheroids assembled with the indicated cells after 24 h of migration in the absence of any antibody or in the presence of control or anti-MMP-2 antibody. (C) Variations over time of the SCSA of the indicated spheroids. (D) Variations of the α factor of the indicated spheroids after 24 h of migration. (E) Representative bright field and fluorescence images of the indicated spheroids after 24 h of incubation in the presence or absence of the indicated antibodies. White dashed lines indicate the periphery of the spheroids. White arrows point to cells that escaped from the spheroids and migrated as single cells, most of which were GFP-negative. Scale bars, 100 µm. * indicates p < 0.01, **** indicates p < 0.0001.
Figure 6.
FAK over-expression not only induces collective cell invasion but favors the emergency of cells that detach from the spheroids. (A) Representative data showing the efficiency of GFP-FAK expression in transiently transfected cells. Representative immunoblot analysis of FAK and pFAK in mock- and GFP-FAK-transfected cells is shown in the middle. The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratio FAK or pFAK/β-actin was estimated by densitometry. Graphics show relative quantification of FAK mRNA, and FAK and pFAK proteins in the indicated cells. (B) Representative images of spheroids assembled with the indicated cells after 24 h of migration in the absence of any antibody or in the presence of control or anti-MMP-2 antibody. (C) Variations over time of the SCSA of the indicated spheroids. (D) Variations of the α factor of the indicated spheroids after 24 h of migration. (E) Representative bright field and fluorescence images of the indicated spheroids after 24 h of incubation in the presence or absence of the indicated antibodies. White dashed lines indicate the periphery of the spheroids. White arrows point to cells that escaped from the spheroids and migrated as single cells, most of which were GFP-negative. Scale bars, 100 µm. * indicates p < 0.01, **** indicates p < 0.0001.

Figure 7.
FAK over-expression in SCC cells favors epithelial to mesenchymal transition (EMT) of adjacent non-transduced SCC cells. (A) Representative immunoblot of E-cadherin (EC), N-cadherin (NC), vimentin (VIM) and ZEB1 in mock- or GFP-FAK transfected SCC cells. The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratio EC, NC, VIM or ZEB1/β-actin was estimated by densitometry. Quantification of the relative protein levels is shown in (B). (C–E) Representative images of immunostaining for EC, NC, VIM and CK in the indicated cells. Insets in C are shown to highlight that GFP-positive cells at the periphery, but not in the middle, of the migrating nests lose EC immunostaining. The inset shown in panel (D) is an over-exposed image to highlight the membrane NC immunostaining in GFP-FAK-SCC38 cells. **** p < 0.0001.
Figure 7.
FAK over-expression in SCC cells favors epithelial to mesenchymal transition (EMT) of adjacent non-transduced SCC cells. (A) Representative immunoblot of E-cadherin (EC), N-cadherin (NC), vimentin (VIM) and ZEB1 in mock- or GFP-FAK transfected SCC cells. The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratio EC, NC, VIM or ZEB1/β-actin was estimated by densitometry. Quantification of the relative protein levels is shown in (B). (C–E) Representative images of immunostaining for EC, NC, VIM and CK in the indicated cells. Insets in C are shown to highlight that GFP-positive cells at the periphery, but not in the middle, of the migrating nests lose EC immunostaining. The inset shown in panel (D) is an over-exposed image to highlight the membrane NC immunostaining in GFP-FAK-SCC38 cells. **** p < 0.0001.
Figure 8.
Expression of epithelial/mesenchymal biomarkers in association with FAK/MMP-2 in SCC cells (A) Western blot analysis showing expression of EC, CK and VIM in SCC cells infected with empty vector (C) or in SCC cells expressing FRNK (FR), MMP-2 (M) or FRNK + MMP-2 (FM). The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratios EC, CK or VIM/β-actin were estimated by densitometry (B). (C,D) Representative images of immunostaining of the indicated cells for CK (red) plus VIM (green) (C) or EC (red or green) (D). Insets in panel (C) highlight the loss of CK and gain of VIM in M-SCC38 cells compared with FR-SCC38 cells.
Figure 8.
Expression of epithelial/mesenchymal biomarkers in association with FAK/MMP-2 in SCC cells (A) Western blot analysis showing expression of EC, CK and VIM in SCC cells infected with empty vector (C) or in SCC cells expressing FRNK (FR), MMP-2 (M) or FRNK + MMP-2 (FM). The membranes were stripped and reprobed with anti-β-actin antibody to assure even loading of proteins in each lane. The ratios EC, CK or VIM/β-actin were estimated by densitometry (B). (C,D) Representative images of immunostaining of the indicated cells for CK (red) plus VIM (green) (C) or EC (red or green) (D). Insets in panel (C) highlight the loss of CK and gain of VIM in M-SCC38 cells compared with FR-SCC38 cells.
Figure 9.
pFAK is highly expressed at the borders of cancer cell nests matching with areas of loss of E-cadherin expression. (A) Representative image of pFAK immunofluorescence in head and neck SCCs showing high expression levels at the border of the tumor. (B) Double immunofluorescence labeling of a head and neck SCC tissue for E-cadherin (red) and pFAK (red) showing that pFAK is predominantly expressed at the border of the cell nest where E-cadherin is lost. Insets show the boundary between the tumor nest and the surrounding stroma (dashed white line). Scale bars, 50 µm.
Figure 9.
pFAK is highly expressed at the borders of cancer cell nests matching with areas of loss of E-cadherin expression. (A) Representative image of pFAK immunofluorescence in head and neck SCCs showing high expression levels at the border of the tumor. (B) Double immunofluorescence labeling of a head and neck SCC tissue for E-cadherin (red) and pFAK (red) showing that pFAK is predominantly expressed at the border of the cell nest where E-cadherin is lost. Insets show the boundary between the tumor nest and the surrounding stroma (dashed white line). Scale bars, 50 µm.
Figure 10.
High MMP2 mRNA levels are significantly associated with poor prognosis independently of EMT transcriptional signature or PTK2 mRNA levels. (A) A heatmap showing positive (red) and negative (blue) correlations of mRNA levels of PTK2, MMP2, and the indicated mesenchymal (CDH2, ZEB1, ZEB2, SNAI1, TWIST1, CDH11, CDH13, TGFB2, VIM) and epithelial (KRT13, CLDN7, CLDN4, CDH28, CDH1) biomarkers in head and neck SCCs included in the TCGA dataset (n = 566 tumors). (B) Graphic representation of the epithelial/mesenchymal gene expression signature of head and neck SCC included in the TCGA dataset. Tumors were dichotomized into high (red) or low (blue) gene expressors taking into account whether the mRNA levels of MMP2, PTK2, CDH1 and the epithelial (Ep) genes, CLDN7, CLDN4, CDH26, KRT13, or mesenchymal (Msc) genes, CDH2, ZEDB1, ZEB2, SNAI1, SNAI2, TWIST1, CDH11, CDH13, TGFB2, and VIM, were, respectively, above or below the median value of mRNA levels of the corresponding gene. (C) Kaplan–Meier curves for head and neck SCC patients (TCGA database) according to the MMP2 mRNA levels (median value = 12.3). The 95% confidence intervals (shaded areas) are also represented; permutated log-rank p-value for the hypothesis testing of equality of curves between groups is also reported. (D) Forest plot of primary outcome indicating genes associated with prognosis in head and neck SCCs.
Figure 10.
High MMP2 mRNA levels are significantly associated with poor prognosis independently of EMT transcriptional signature or PTK2 mRNA levels. (A) A heatmap showing positive (red) and negative (blue) correlations of mRNA levels of PTK2, MMP2, and the indicated mesenchymal (CDH2, ZEB1, ZEB2, SNAI1, TWIST1, CDH11, CDH13, TGFB2, VIM) and epithelial (KRT13, CLDN7, CLDN4, CDH28, CDH1) biomarkers in head and neck SCCs included in the TCGA dataset (n = 566 tumors). (B) Graphic representation of the epithelial/mesenchymal gene expression signature of head and neck SCC included in the TCGA dataset. Tumors were dichotomized into high (red) or low (blue) gene expressors taking into account whether the mRNA levels of MMP2, PTK2, CDH1 and the epithelial (Ep) genes, CLDN7, CLDN4, CDH26, KRT13, or mesenchymal (Msc) genes, CDH2, ZEDB1, ZEB2, SNAI1, SNAI2, TWIST1, CDH11, CDH13, TGFB2, and VIM, were, respectively, above or below the median value of mRNA levels of the corresponding gene. (C) Kaplan–Meier curves for head and neck SCC patients (TCGA database) according to the MMP2 mRNA levels (median value = 12.3). The 95% confidence intervals (shaded areas) are also represented; permutated log-rank p-value for the hypothesis testing of equality of curves between groups is also reported. (D) Forest plot of primary outcome indicating genes associated with prognosis in head and neck SCCs.

Table 1.
Correlations between PTK2 and MMP2 mRNA levels and clinical variables of the patients with head and neck squamous cell carcinomas included in the TCGA database.
Table 1.
Correlations between PTK2 and MMP2 mRNA levels and clinical variables of the patients with head and neck squamous cell carcinomas included in the TCGA database.
| | PTK2 mRNA Levels | MMP2 mRNA Levels |
---|
| N | Low | High | p | Low | High | p |
---|
pT classification | | | | | | | |
T1–T2 | 209 | 111 | 98 | 0.094 | 124 | 85 | <0.0001 |
T3–T4 | 334 | 152 | 182 | | 141 | 193 | |
pN classification | | | | | | | |
N0 | 193 | 102 | 91 | 0.005 | 105 | 88 | 0.021 |
N+ | 347 | 159 | 188 | | 159 | 188 | |
Distant metastasis | | | | | | | |
M0 | 190 | 96 | 94 | 0.450 | 97 | 93 | 0.192 |
M+ | 66 | 37 | 29 | | 39 | 27 | |
Disease stage | | | | | | | |
I–II | 119 | 67 | 52 | 0.098 | 70 | 49 | 0.007 |
III–IV | 374 | 178 | 196 | | 167 | 207 | |