FAK acts as a suppressor of RTK-MAP kinase signalling in Drosophila melanogaster epithelia and human cancer cells

Abstract

Receptor Tyrosine Kinases (RTKs) and Focal Adhesion Kinase (FAK) regulate multiple signalling pathways, including mitogen-activated protein (MAP) kinase pathway. FAK interacts with several RTKs but little is known about how FAK regulates their downstream signalling. Here we investigated how FAK regulates signalling resulting from the overexpression of the RTKs RET and EGFR. FAK suppressed RTKs signalling in Drosophila melanogaster epithelia by impairing MAPK pathway. This regulation was also observed in MDA-MB-231 human breast cancer cells, suggesting it is a conserved phenomenon in humans. Mechanistically, FAK reduced receptor recycling into the plasma membrane, which resulted in lower MAPK activation. Conversely, increasing the membrane pool of the receptor increased MAPK pathway signalling. FAK is widely considered as a therapeutic target in cancer biology; however, it also has tumour suppressor properties in some contexts. Therefore, the FAK-mediated negative regulation of RTK/MAPK signalling described here may have potential implications in the designing of therapy strategies for RTK-driven tumours.

Figure 1. Genetic interactions between RET and FAK.

(A–F) Confocal images of wing disc epithelia. Control tissues (A–C) and experimental tissues (D–F) expressed GFP driven by ptc-gal4, shown in A′ and D′. Experimental tissues (D–F) also expressed dRET^CA. Immunostaining against pSrc (A″ and D″), pMAPK (B and E) and pFAK proteins (C and F; see methods), as a proxy for probing their activation levels, are shown in grayscale panels. Note increased phosphorylation of Src, MAPK and FAK after dRET^CA expression within the ptc domain, indicated by red arrows. Scale bars, 50 µm. (G–P) Images of adult eyes with indicated relevant genotypes; full genotypes are listed in supplemental material. GMR (glass multimer reporter) is an eye specific promoter. GMR-gal4 was used to drive UAS-dFAK transgene expression. GMR-dRET^WT and GMR-dRET^CA are fusion recombinant constructs. Wild type (G) and dFAK^CG1 (L) animals displayed a normal eye pattern; note that the dFAK^CG1 is in a white background (see Figure S1B). Expression of dRET^WT caused a mild eye miss-patterning phenotype (H), and lowering the genetic dose of dFAK gene in these animals either enhanced eye roughness (I–J) or completely disrupted patterning and decreased eye size (J). Reciprocally, suppression of both effects was observed by co-expression of dFAK (K). (M–N) A similar enhancement was observed by halving the dose of dFAK gene after expression of dRET^CA. (O) Doubling dRET^CA dose caused a very rough and small eye, comparable to (J), which was partially suppressed when dFAK was also expressed (P). Eye size quantifications of panels G, H, J, K, O and P are shown in Figure S1D. Scale bars, 100 µm.

Figure 2. FAK suppresses RET-driven effects in different fly tissues.

(A–D) Eyes expressing dFAK displayed a normal adult eye phenotype, while dRET^WT expression perturbed the normal pattern. Co-expression of dRET^WT and dFAK supressed dRET-driven mis-patterning defects. Scale bar, 100 µm. (E–H) dFAK-expressing wings via ptc-gal4 showed no detectable defects similarly to control wings. Expression of dRET^CA led to disappearance of anterior cross veins in all adult escapers (arrow in inset box), which was rescued by simultaneous expression of dFAK with full genetic penetrance. Scale bar, 500 µm. (I–L) ptc-driven dRET^CA expression also led to incomplete rotation of the male genitalia in all adult escapers (arrows). dFAK co-expression rescued this phenotype with full penetrance and it did not affect the normal development of the genitalia by itself. Scale bar, 100 µm. (M) Quantification of the penetrance of adult eclosion for the indicated genotypes, note that dFAK co-expression rescued significantly the developmental lethality of ptc>dRET^CA animals. Error bars are standard deviation in this and all plots; ‘ns’ stands for non-statistically significant, **** means p<0.0001 in this and all plots (see methods). (N) Conversely, dFAK loss, which by itself had no effect in viability, enhanced to almost full penetrance the developmental lethality of dpp>dRET^CA animals. (O–R) Confocal images from wing discs with the indicated genotypes. Note that dFAK mutation enhanced the size and shape defects associated with ectopic expression of dRET^CA within the dpp stripe. For a detailed characterisation of the dFAK mutant alleles used here, please see Figure S1A–B. Scale bars, 50 µm.

Figure 3. Requirement of the N-terminal FAK FERM domain.

(A) Linear representation of dFAK mRNA, its derivatives UAS-transgenes and their resultant protein isoforms: a full-length dFAK isoform; an N-terminal deletion mutant that lacks the first 400 amino acids residues of dFAK including its FERM domain; and a point mutant isoform that bears a replacement of the Tyrosine⁴³⁰ residue for a Phenylalanine residue, which impairs the auto-phosphorylation site and consequently the kinase activity of dFAK. (B) Expression profiles of each UAS-dFAK transgene in the eye (driven by GMR-gal4) as determined by quantitative (q) PCR of RNA samples (see methods). We used a pair of primers (3F and 3L) flanking a 200 bp region corresponding to the C-terminal domain (FAT: Focal adhesion targeting domain), which is a common region to all the isoforms. (C) Eye size quantification of the indicated genotypes, shown in D-G. Eye sizes on the Y-axis are represented as relative values to the mean of GMR>dRET^CA (‘ns’: not statistically significant; **** = p<0.0001; n = 8–10 for each genotype). (D–G) Eye micrographs correspond to the indicated genotypes. Note that while the auto-phosphorylation mutant version of dFAK was expressed at lower levels than the N-terminal mutant isoform (B), it was still able to rescue the size of dRET^CA-expressing eyes (G), to a similar extent as the full-length dFAK isoform (E). However, the N-terminal deletion mutant isoform did not suppress the small eye size of dRET^CA animals (C, D and F). Scale bar, 100 µm.

Figure 4. Moderate relative RET/FAK levels lead to inhibition of programmed cell death.

(A–C) Armadillo immunostaining revealed cell outlines of wild type (A), dFAK^CG1 (B), and GMR-dRET^WT (C) retinas at 42 hs after puparium formation (APF). The boxed areas were traced to highlight their cellular composition (A′–C′). Each ommatidium is composed of 4 cone cells (red), 2 primary pigments cells (yellow), 6 secondary and three tertiary cells (white), and three-bristle cells (green) make the hexagonal lattice. Note that dFAK^CG1 eyes display normal patterning (B′) while GMR-dRET^WT retinas displayed normal ommatidial cores but additional interommatidial cells (white cells in C′). Scale bars, 10 µm. (D–G) TUNEL labelling of retinas at 28 h APF. Note that the developmental programmed cell death observed in wild type and dFAK^CG1 retinas were suppressed in GMR-dRET^WT retinas. Co-expression of dFAK rescued this inhibition of cell death (G). Scale bar, 50 µm. (H–K) Hid overexpression (GMR-hid) gave a small eye phenotype, which was suppressed by dRET^WT co-expression (J). This dRET-dependent inhibition was also suppressed by dFAK co-expression (K) while dFAK itself did not suppress Hid-mediated effects in the eye (I). Scale bar, 100 µm. (L) Eye size quantification of the indicated genotypes (as depicted in panels H–K) represented as relative values to the wild type mean value (‘ns’: not statistically significant; **** = p<0.0001; n = 8–10 for each genotype).

Figure 5. High relative levels between RET and FAK induce ectopic cone cell differentiation in the eye.

We examined the cellular patterning of the pupal retinas in correspondence to the adult eye phenotypes shown in panels A–D. Scale bars, 100 µm. (E–H) Merged images of stainings for nuclei (DAPI, blue), Dlg (cell outlines, red) and Cut (cone cells, green), from retinas at 42 h APF. Bottom panels show Dlg (E′–H′) and Cut (E″–H″) immunostainings individually. (E) Note the symmetric hexagonal array, and four Cut⁺ cone cells per ommatidium (white arrows) in control retinas. (F and G) Note the change in cellular composition of these retinas with high RET/FAK ratios, primarily composed of Cut⁺ cone-like cells. (H) dFAK expression within a 2X GMR-dRET^CA background suppressed this phenotype (also see S1D); some normal four-cone cell clusters (white arrows) can be identified and interommatidial cells reappeared (yellow arrows). Scale bars, 10 µm.

Figure 6. FAK inhibits RTK signalling by impairing Ras/MAPK pathway.

(A–D) Phosphorylated (active) MAPK immunostainings from wing discs with the indicated genotypes. (A–B) When dFAK was expressed in the ptc-compartment (green), pMAPK staining was unchanged compared to GFP-only expressing cells. (C–D) dRET^CA expression increased pMAPK staining in the ptc domain but co-expression with dFAK suppressed this dRET^CA-induced activation of MAPK. Scale bars, 50 µm. (E) Quantification of pMAPK immunostaining within the ptc stripe (see methods). Intensity of pMAPK signal is represented as relative values to the mean intensity of control tissues (A) (‘ns’: not statistically significant; **** = p<0.0001; n = 4–6 for each genotype).

Figure 7. FAK suppression of RTK signalling is conserved.

(A–B) Adult eyes images of animals expressing Drosophila EGFR (dEGFR) alone or in combination with dFAK. Note that dFAK expression suppressed the rough, small eye phenotype driven by dEGFR. Scale bar, 100 µm. (C–C′) Expression of dEGFR within the ptc domain resulted in increased MAPK phosphorylation, and co-expression of dFAK rescued the ectopic pMAPK staining within the ptc stripe (D–D′). Scale bar, 50 µm. (E) Quantification of pMAPK immunostaining within the ptc stripe (see methods). Intensity of pMAPK signal is represented as relative values to the mean intensity of control tissues (Figure 6A; **** = p<0.0001; n = 4–6 for each genotype). (F) Quantification of the penetrance on adult eclosion for the indicated genotypes. Note that dFAK co-expression significantly rescued the developmental lethality associated to ptc-driven dEGFR expression (* = p<0.05). (G) Western blots from protein extracts from MDA-MB-231 cells after 48 or 72 h transfection with FAK siRNA. FAK protein levels were effectively knocked down. While total levels of EGFR and ERK were not changed at 48 h, there was a marked upregulation in phosphorylated ERK1/2 upon FAK knockdown, which was more apparent at 48 h after siRNA transfection. Actin levels were probed as an additional loading control. (H) MDA-MB-231 cells were transfected with either non-targeting (siNT) or FAK-specific siRNA (siFAK) and serum starved prior to addition of EGF. Note that FAK knockdown resulted in increased phosphorylation of ERK1/2 in response to EGF treatment (30 µM, 15 minutes).

Figure 8. FAK decreases EGFR at the plasma membrane via reduced recycling.

(A–B) MDA-MB-231 cells transfected with non-targeting (NT) siRNA or FAK siRNA were immunostained with anti-EGFR antibody (green, A″ and B″), Rhodamine-phalloidin (red, A′ and B′) and DAPI (blue). Note the differential localisation of EGFR; while in siNT cells the receptor is distributed in plasma membrane and internal vesicles, FAK downregulation leads to an increase of EGFR levels at the cellular membrane. Scale bar, 10 µm. (C) Quantification of relative EGFR membrane levels, values are expressed as relative levels of the receptor against the mean value of siNT cells; four confocal fields for each condition were analysed: n = 347 cells (siNT), and n = 414 cells (siFAK). p<0.0286 in a Mann-Whitney test. (D) MDA-MB-231 cells were transfected with either non-targeting (siNT) or FAK-specific siRNA (siFAK) and deprived of serum prior to addition of 80 µM Dynasore. siNT-transfected cells showed an increased pERK1/2 level in response to both Dynasore treatment (80 µM, 30 minutes) and FAK knockdown. Total levels of EGFR and ERK were not changed and actin levels were probed as an additional loading control. (E) The internalization of membrane EGFR (top panel) and recycling of internalised EGFR (bottom panel) were determined in MDA-MB-231 cells transfected with non-targeting siRNA (siCTR) or FAK siRNA (siFAK). Values are means ± Standard Deviation (SD) of two independent experiments with four to eight replicates of each time point per genotype. See materials and methods for more details. FAK knockdown did not affect receptor internalization but increased the recycling of the internalised EGFR pool. (F) A working model for the regulatory mechanism of FAK. Ectopic expression and/or hyperactivation of RTKs activate FAK and Ras among other signalling pathways. FAK mediates a negative regulation of receptor recycling; when FAK is reduced or absent, there are more RTKs molecules at the plasma membrane, thus enabling a higher flux of signalling through Ras/MAPK pathway. See the text for more details.