CD98hc (SLC3A2) Loss Protects Against Ras-Driven Tumorigenesis by Modulating Integrin-Mediated Mechanotransduction

Skin squamous cell carcinoma (SCC) is the second most common skin cancer (1). Expression of the type II transmembrane glycoprotein CD98 heavy chain (4F2, SLC3A2) is highly increased in most carcinomas, including SCC, as well as transformed cell lines (2), while in skin, CD98hc is required for proper homeostasis and efficient epidermal wound closure. Via its extracellular domain (C109), CD98hc covalently binds one of several catalytic subunits (the SLC7A5-11 family) to form the functional heteromeric amino acid transporters at the cell surface (3). Besides its role as amino acid transport mediator, CD98hc, by binding to intracellular domains to the adhesion receptor β1 integrins, enhances integrin outside-in signaling (4), thus modulating integrin-dependent cell functions.

CD98hc overexpressing NIH-3T3 fibroblasts developed malignant tumors in athymic mice (5, 6). We previously showed that CD98hc deletion in mouse embryonic stem (ES) cells delays teratocarcinoma formation in mice (4). CD98hc is an integrin-binding protein that enhances signals downstream of β1 integrin therefore regulating integrin-dependent cell behavior, including matrix remodeling in vitro (4, 7, 8). Tumorigenicity of CD98hc-deficient ES cells can be reconstituted by expressing a chimeric form of CD98hc, which is able to interact with β1 integrin (4). This CD98hc/β1 interaction contributes to cell transformation in vitro (9).

Here, we examine the role of CD98hc in a well-established (10) two-step model of SCC. In this Ras-driven cancer model, tumorigenesis begins with the initiation of a single epidermal cell, which occurs following a single subcarcinogenic dose of 7,12-dimethylbenz[a]-anthracene (DMBA). After the initiation stage, the population of mutated cells is promoted to clonally expand during the second stage, referred to as “promotion”. Tumor promotion is elicited by the repeated topical application of chemical agents, such as the phorbol ester, phorbol 12-myristate 13-acetate (TPA) that leads to sustained epidermal hyperplasia as evidenced by an increase in the number of nucleated cell layers and an overall increase in thickness of the epidermis. Tumorigenesis proceeds through the promotion of benign tumors (papilloma) growth and finally the progression of some benign tumors into malignant and potentially metastatic lesions (SCC). This two-stage skin carcinogenesis model enables direct visualization of tumor development and permits evaluation of TPA-induced inflammation response (10).

Here, we combined this model of epidermal tumor formation with conditional deletion of CD98hc in basal keratinocytes to reveal a major contribution of CD98hc in controlling the mechanical properties of the tumor microenvironment, independently of skin TPA-induced inflammation. Specifically, CD98hc-deficient skin was protected against tumor formation, and that CD98hc deletion led to regression of preexisting tumors. We demonstrate that beyond CD98hc-intrinsic proliferation effect within tumor cells, CD98hc-expressing environment is cancer-prone due to modulation of its mechanical properties. This model is known to be sensitive to integrin signaling and rigidity sensing. We now show that CD98hc acts as a gain amplifier for stiffness sensing via integrins. CD98hc is thus a gain amplifier of a positive feedback loop that increases ROCK signaling, matrix stiffness, and YAP/TAZ-driven gene expression.

All procedures were approved by the Institutional Animal Care and Use Committee at the University of Nice-Sophia Antipolis (Nice, France; CIEPAL-AZUR Agreement NCE/2012-66). K14-CreER^T2, CD98hc^fl/fl have been described previously (11). The experiments have been performed on pure FVB/n background mice.

Chemical carcinogenesis experiments were performed on 10-week-old mice (20 animals/group), essentially as previously described (10). Mice received one topical application of 200 nmol of DMBA (Sigma-Aldrich) in 200 μL of acetone followed a week later by biweekly applications of 6.8 nmol of TPA (Sigma-Aldrich) in 200 μL acetone. Papilloma and SCC numbers were scored once a week for up to 32 weeks after the start of promotion. Topical 4-hydroxy tamoxifen (4-OHT) treatment was topically applied when indicated (6 treatments of 1.5 mg 4-OHT in ethanol/back). No difference in benign tumor acquisition was observed in 4-OHT–treated wt mice ruling out any possible effects of 4-OHT treatment (data not shown).

Ears (both sides) of mice with wild-type (WT)- or CD98hc-deficient skin were topically treated with 20 μL (per side) of either TPA (0.5 μg per side) or vehicle (acetone). Ear thickness was measured with digital calipers and ear swelling was calculated as (thickness at each time point) − (thickness at 0 hour). Paraffin sections were stained with hematoxylin and eosin (H&E) and granulocytes were quantified using ImageJ software.

Samples were collected as described by Montanez and colleagues (12). Histology and immunochemistry were performed as described previously (11). Paraffin-embedded sections were used for all stainings except CD98hc immunostaining, which was performed on frozen sections. Primary antibodies were: rabbit anti human P-cofilin (Cell Signaling Technology), rabbit anti-human P-MYPT1 (Millipore), rabbit anti-human MYPT1 (Millipore), rat anti-mouse CD98 (Clone RL388, eBioscience). Secondary antibodies were used according to manufacturer's instructions, for immunofluorescence Alexa Fluor 594 anti-rat #A-21209 anti-rabbit #A-21207 (Life Technologies) and for immunohistochemistry Vectastain ABC-Kit (Vector #4001) with DAB reagent (Vector #SK4100).

To carry out topography imaging and simultaneous elastic modulus maps, a Bioscope Catalyst operating in Peak Force Quantitative Nanomechanical Mapping mode (Bruker Nano, Inc.), coupled with an inverted optical microscope (DMI 6000B, Leica) was used. The experiments were performed using V-shaped silicon SNL-D probes (nominal spring constant k = 0.06 N/m, side angle 23°; Bruker Nano, Inc.). For each sample (10 μm unfixed frozen sagittal section; ref. 13), a peak force of 5 nN and a tip velocity of 4 μm/second were used (detailed procedure in Supplementary Materials and Methods). For collagen deposition analysis, detailed procedure is mentioned in Supplementary Materials and Methods.

Mouse embryonic fibroblasts (MEF) were derived from CD98hc-conditional knockout homozygote embryos as described previously (7). Similarly, murine keratinocytes were generated and cultured from back skin of adult mice as described in ref. 11 (culture conditions, see Supplementary Data). Both cell types were identified by PCR on genomic DNA [as for mice (7), CD98hc-expressing or -deficient lines], which was performed regularly during their use. MEFs were maintained in DMEM-H (Invitrogen) culture medium containing 10% (v/v) FBS (HyClone), 20 mmol/L HEPES, pH 7.3 (Invitrogen), 0.1 mmol/L nonessential amino acid (Invitrogen), 0.1 mmol/L β-mercaptoethanol (Sigma-Aldrich), and 2 mmol/L l-glutamine (Invitrogen) at 37°C and 5% CO₂. The CD98hc-null MEFs were generated by infecting CD98hc floxed MEFs with adeno-CRE encoding CRE recombinase.

MEFs were replated on the fibronectin-coated gels (PDMS; Supplementary Materials and Methods or Supplementary Fig. S3, polyacrylamide; Matrigen) with indicated stiffness and incubated in DMEM plus 1% (v/v) BSA for 45 minutes or grown on gels for 24 hours before treatment with 1 μg/mL of either Rho inhibitor I or Rho activator II (Cytoskeleton) in serum-free medium for 4 hours. Alternatively, CD98hc-null MEFs infected with retroviruses encoding wild-type CD98hc, wild-type CD69, or chimeras (C69T98E98, C98T69E98, or C98T98E69) were replated on 30 kPa of PDMS gels for 45 minutes in the absence of serum. Cells were fixed with 3.7% formaldehyde in PBS, permeabilized with 0.5% Triton X-100 in PBS at room temperature for 10 minutes, and washed three times with PBS. The cells were then incubated with phalloidin-conjugated rhodamine (Molecular Probes) for 1 hour at room temperature. The cells on coverslips were washed three times with PBS, mounted with a mounting solution (ProLong Gold antifade reagent; Invitrogen), and visualized by fluorescent microscopy (Eclipse TE2000-U, Nikon). Cell spreading was quantified by cell area by using customized software written in MATLAB (MathWorks).

MEFs were replated on PDMS gels as described above. When mentioned, CD98hc-null MEFs were transiently transfected with wild-type CD98hc, extracellular domain-deleted CD98hc or CD69 for 36 hours, and then replated on 30 kPa PDMS substrates and incubated as above. Whole cell lysates were prepared using modified RIPA buffer (50 mmol/L Tris, pH 7.5, 150 mmol/L NaCl, 50 mmol/L NaF, 1 mmol/L sodium pyrophosphate, 0.1% sodium deoxycholate, 1% Nonidet P-40, protease inhibitors cocktail, and 1% CHAPS). Protein lysates were quantified using the bicinchoninic acid method (Thermo Scientific Pierce), loaded on SDS-PAGE, and analyzed by Western blotting. The primary antibodies for Western blotting were anti-phospho-Y⁵⁷⁶FAK (Invitrogen), anti-FAK (Santa Cruz Biotechnology), anti-phospho-S⁴⁷⁶Akt (Cell Signaling Technology), anti-Akt (Cell Signaling Technology), and anti-α-tubulin (Sigma-Aldrich).

RNAs were extracted from back skin samples (bearing or not tumors) using TRIzol reagent (GIBCO), or from MEFs plated on gels using RNeasy Plus Kit (Qiagen). Reverse transcription was performed on total RNA using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Sets of specific primers (see Supplementary Table S1) were used for amplification using 7900HT Real Time PCR System (Applied Biosystems). Samples were normalized to GAPDH (tumors) or rplp0 (MEFs) using the ΔC_t method. Statistical significance was determined with Student t test.

CD98hc expression is linked to malignancy (2, 14, 15) and CD98hc can regulate tumor-associated processes such as anchorage independence (4, 8), nutrient transport (3, 16), and mTOR (17). To assess CD98hc function in skin tumorigenesis, we generated mice with conditional deletion of CD98hc in the basal layer of the epidermis (K14CreER^T2;CD98^fl/fl; ref. 11) and applied a two-stage chemical skin carcinogenesis protocol. Topical treatment with the carcinogen DMBA-induced oncogenic Ras mutations and subsequent repeated treatments with the phorbol ester TPA promoted outgrowth of initiated cells, resulting in benign papillomas in control mice about 8 weeks after initiation (average of 3 papillomas/mouse at 10 weeks and 10 papillomas/mouse by 16 weeks; Fig. 1B and C). Strikingly, papilloma formation was delayed by 8 weeks in CD98hc-deficient mice. Even after 25 weeks of TPA treatment, only 2 papillomas per mouse, on average, were detected on CD98hc-deficient skin, whereas 10 papillomas per mouse were observed in control mice (Fig. 1A and B). Tumor multiplicity was strongly reduced (Fig. 1A), suggesting that CD98hc promotes Ras-induced tumor formation and growth. Typical papillomas in vehicle-treated mice showed exophytic growth of squamous cells. In contrast, 4-OHT–treated skin only occasionally manifested hyperplasia of the epidermis (Fig. 1D). CD98hc deletion was induced by 2-week pretreatment with 4-OHT compared with vehicle-treated skin (Fig. 1A and E). The rare papillomas or hyperplastic area, observed on skin treated with 4-OHT, expressed CD98hc (Fig. 1F), suggesting that they were formed by keratinocytes that had escaped Cre-mediated recombination. These data strongly suggest a tumor-promoting function of CD98hc in Ras-induced papillomas.

Inflammation facilitates cancer development and growth. Because intestinal epithelia CD98hc is required for inflammation-induced tumorigenesis in the gut, promoting colitis-associated cancer in mice (15), we analyzed the effect of epidermal CD98hc loss in the inflammatory responses stimulated by topical TPA application, the known inflammatory inducer in skin model. First, we quantified the extent of skin edema by treating the ears with TPA and measuring their thickness. A strong reaction was observed following TPA application (vs. vehicle-treated skin; Fig. 2A and B); histologic analysis of the ears, prepared 6 hours after last TPA application, revealed no gross difference in ear thickening and swelling of WT versus CD98hc knockout (KO) mice (Fig. 2A–C). Besides, marked spongiosis and extensive infiltration of leukocytes were observed in the edematous dermis independent of CD98hc expression (Fig. 2A). Similar TPA-induced infiltration of inflammatory leukocytes was detected between CD98hc-expressing and -deficient mice (Fig. 2D). Thus, TPA-induced inflammatory response, ultimately leading to skin tumorigenesis, is CD98hc-independent. Altogether, these data strongly suggest that the dramatic effect of CD98hc loss on skin tumorigenesis relates to a cell-autonomous effect.

Because CD98hc has been shown to act intrinsically as a cell proliferation enhancer both in tumoral and nontumoral cells, Ras-induced tumorigenesis inhibition in CD98hc-deficient skin was expected. Thus, we analyzed the effect of CD98hc deficiency once tumors were already formed. After either 11 or 23 weeks of DMBA/TPA [inducing, respectively, early or large papillomas defined as a size of over 6 mm (before SCC conversion)], we treated mice with topical 4-OHT or vehicle, and monitored succeeding tumor progression (Fig. 3A and B). In both cases and within 2 weeks (4-OHT treatment period), not only further papilloma growth or conversion to SCC were totally abrogated, in spite of continued TPA application, but papillomas fully regressed, irrespective of papilloma initial size (Fig. 3A and B, regression within 5–7 weeks). In sharp contrast, in the first set-up, vehicle-treated mice exhibited progressive papilloma formation such that, at 19 weeks, they manifested an average of 18 papillomas per mouse (Fig. 3A). In the second set-up, by 27 weeks post-DMBA, SCC conversion occurred only in CD98hc-expressing mice, with an average of one SCC per mouse, as judged by their flattened morphology and downward appearing growth (Fig. 3B). Besides, these SCCs exhibited markedly increased CD98hc expression as compared with normal skin and a similar increased expression was observed in human SCC (Fig. 3C). The development of SCC in control mice was confirmed by histologic analysis (Fig. 3D). Furthermore, increased CD98hc expression was a general feature of human SCC, being present in all 12 human SCC studies (GEO profiles for SLC3A2), compiled over 310 samples. Thus, while CD98hc ablation in skin does not modulate apoptosis (11), it induces a drastic regression of Ras mutation-bearing tumors, preventing progression to SCC.

To assess how CD98hc deficiency impacts tumorigenesis, beyond its cell-autonomous effect on proliferation and absence of extrinsic effect on TPA-induced inflammation, we analyzed mechanical features of the tumor microenvironment. A stiff and organized three-dimensional microenvironment is important in tumor formation and tumor cell invasiveness (18) and we previously showed that, in fibroblasts, CD98hc can regulate RhoA (7), an important regulator of skin stiffness (13). Interestingly, CD98hc-deleted skin exhibited an atomic force microscopy (AFM) force map skewed toward low MPa values, indicating more compliant tissue (Fig. 4A and B and Supplementary Fig. S1A). The AFM nanoindentation of the epidermal CD98hc-deficient skin revealed an elastic modulus of 0.063 ± 0.01 MPa (n = 4), whereas wt skin, elastic modulus in the papillary (upper) dermis reached 0.53 ± 0.03 MPa (mean, n = 3; Fig. 4C). Although other methods have reported varying E values (19–21), our data are in good agreement with previous AFM studies (22) performed on skin. Thus, CD98hc deficiency decreases epidermal tissue stiffness. Highly aligned ECM molecules, in particular collagen fibers, increase matrix strength and stiffness in vivo. Skin sections of control and 4-OHT–treated mice stained with Masson trichrome (Supplementary Fig. S1B) and Sirius red (Fig. 4D and E) highlight a major defect in fibrillar collagen, as seen using polarized filter, with a less organized collagen-rich dermis in CD98hc-deficient skin (Supplementary Fig. S1C and S1D). Similarly, abnormal fibronectin deposition was observed in KO skin. Therefore, we conclude that the loss of epidermal CD98hc results in more compliant skin due to drastic in vivo changes on collagen and fibronectin organization.

There is an intimate and bidirectional relationship between matrix organization/stiffness and RhoA-driven cellular contractility mediated by its effector Rho Kinase (ROCK; refs. 13, 19): cellular contractility stimulates matrix assembly (23) and a stiffer matrix stimulates RhoA activity (24). We previously showed that, loss of CD98hc induces a reduction of RhoA activation in unchallenged skin (11). We therefore assessed the effect of CD98hc deletion in keratinocytes on RhoA signaling in papillomas as judged by the ROCK-mediated phosphorylation of the regulatory subunit of myosin phosphatase (Mypt; Fig. 5A). Papillomas exhibited extensive phospho-Mypt in the epithelium, particularly in suprabasal layers. In contrast phospho-Mypt was nearly absent in the few papillomas derived from 4-OHT–treated Slc3a2^fl/fl-K14CreER^T2 mice (Fig. 5A). Similarly, phosphorylation of cofilin, an indirect downstream target of ROCK, was lost in 4-OHT–treated compared with vehicle-treated skins (phospho-cofilin staining, Supplementary Fig. S2). Thus, in contrast to wound healing with a soft provisional matrix (11), CD98hc deletion markedly reduced ROCK activity in papillomas associated with a reduction in matrix stiffness.

ROCK regulates YAP/TAZ (Yes-associated protein/transcriptional coactivator with PDZ-binding motif), which is an important transcriptional mechanism that controls production of fibrogenic factors, such as (connective tissue growth factor) CTGF, in addition to controlling cell size, proliferation, and malignant transformation (25). Moreover, recent work shows that cells can “read” ECM cues as a function of the activity of YAP/TAZ through RhoA/ROCK signaling (25). We, therefore, analyzed the effect of CD98hc depletion on YAP/TAZ transcriptional activity by assaying expression of endogenous target genes. Real-time PCR on papillomas revealed that there was a major reduction in transcripts of both mANKRD1 and mCTGF (∼5- and ∼3-fold reduction, respectively), two well-characterized YAP/TAZ targets (25), in 4-OHT–treated compared with vehicle-treated mice (Fig. 5B). As expected, mANKRD1 and mCTGF were even more highly expressed in SCC reaching levels about 5-fold greater than in papillomas (Supplementary Fig. S3A). Thus, loss of basal keratinocyte CD98hc markedly suppresses both RhoA/ROCK signaling and expression of YAP/TAZ-regulated genes.

Integrins play a central role in stiffness sensing (26, 27). We previously reported that CD98hc expression modulates integrin signaling, which can influence the growth of teratocarcinomas (4). This suggests that CD98hc might alter the cellular responses to the tumor microenvironment. Cell spreading is controlled by extracellular matrix (ECM) stiffness (28). CD98hc-null cells exhibited a striking defect in stiffness sensing. Specifically, when adhering to fibronectin covalently bound to silicone gels of varying compliance, CD98hc-null fibroblasts failed to spread on 3.5 kPa substrates, an elasticity in the range of those found in skin (Fig. 6A and B; ref. 19). Similarly, activation of two integrin-regulated promoters of cell survival, proliferation, and tumor cell invasion, pp125^FAK and AKT, was reduced (Fig. 6C). These defects were specifically rescued in cells expressing full-length human CD98hc (Fig. 6D). Remarkably, adhesion to stiffer substrates (30 and 300 kPa), resulted in partial restoration of both cell spreading and biochemical signaling in the CD98hc-null fibroblasts. Loss of CD98hc in vivo markedly suppresses RhoA/ROCK signaling, resulting in decreased transcription of YAP/TAZ-regulated genes (Fig. 5 and Supplementary Fig. S3A). Direct activation of Rho with cytotoxic necrotizing factor-1 (CNF-1) in CD98hc-deficient cells rescued YAP/TAZ activation (Supplementary Fig. S3B), conversely, its inhibition by exoenzyme C3 in WT cells blocked YAP/TAZ-driven gene expression (Supplementary Fig. S3C), thus we propose that Actin-Rho/ROCK-matrix stiffness-YAP/TAZ cascade is critical downstream of CD98hc.

To assess whether effects on stiffness sensing were mediated through integrin interactions, we tested the capacity of reconstitution of the null cells with wild-type CD98hc or with mutants (Fig. 7A and B). Both wild-type and a mutant (C98T98E69) that couples only to integrins rescued the defect in stiffness sensing in the CD98hc fibroblasts. In contrast, mutants that support amino acid transport but do not associate with integrins (C69T98E98 and C98T69E98) failed to rescue (Fig. 7A and B). Skin tumors arise from the underlying basal layer cells, thus we tested the reconstitution of CD98hc-deficient keratinocytes (Supplementary Fig. S4) and found similarly that only mutant coupling with integrins (C98T98E69) rescued the defect in stiffness sensing. These data show that CD98hc, via its interaction with integrins, acts as a gain amplifier of stiffness sensing.

In this study, we examined the sensitivity of the epidermal CD98hc-deficient mouse to Ras-driven skin carcinogenesis. Mice lacking epidermal CD98hc were protected against chemically induced papilloma formation, even though TPA-associated inflammation response was preserved. Beyond this protection linked to CD98hc cell-autonomous effect on cell proliferation, we demonstrate, for the first time, that when CD98hc deletion is induced after papillomas are formed, tumor regression is observed. Stiffness of the tumor microenvironment is crucial during cancer growth and integrins are involved in stiffness sensing. We reveal a new role for CD98hc as an amplifier of integrin-mediated cellular stiffness sensing leading to both ROCK-mediated increased matrix stiffness and nuclear relay of mechanical signals via YAP/TAZ signaling. Thus, CD98hc is a central gain amplifier in a self-reinforcing feedback loop that regulates both matrix stiffness and the cellular responses to stiffness.

CD98hc is highly expressed in many tumors of different origins as well as in established tumor cell lines. Our results showing that loss of CD98hc protects against tumor formation is in good agreement with recent studies on more invasive and nonaccessible tumors from the intestinal epithelium (15, 29). Indeed, Nguyen and colleagues showed that CD98hc role in the gut tumorigenesis is partly due to an effect on epithelial intrinsic cell proliferation. We previously showed that the integrin-interacting function of CD98hc can mediate cellular proliferation in teratocarcinomas and lymphocytes (4, 30). Notably, we report that, conversely to the gut, CD98hc in the skin does not perturb TPA-induced immune responses, involved during tumorigenesis.

Furthermore, we show for the first time that not only CD98hc acts on tumor progression, but that its loss causes tumor regression. Importantly, CD98hc also forms the heavy chain of an amino acid cotransporter that can support the nutrient requirements of tumors (31) and consequent mTOR activation (17). Thus, it is likely that the capacity of CD98hc to support metabolic reprogramming combined with its capacity to amplify mechanotransduction contribute to its critical role both in the clonal expansion of lymphocytes (30, 32) and in tumorigenesis.

The tumor microenvironment is a mechanically tunable meshwork comprising ECM proteins, a heterogeneous population of stromal cells, and secreted soluble factors. The modifications of the mechanical properties of this microenvironment, which are monitored by ECM receptors such as integrins, play an important role during tumor progression (33). Here, we provide a novel function of CD98hc independent of its required role in immune response linked to epithelial tumorigenesis. Our data are in good agreement with studies on breast carcinoma, showing the critical role of integrin signaling in mechanosensing during tumorigenesis. CD98hc function as a coreceptor of integrins probably participates in integrin signaling in mechanosensing in many cancer types.

SCC is the second most common skin cancer (1) and its pathogenesis is tied to integrin expression (34, 35). Elegant studies point to an important role for keratinocyte α5β1 integrin rather than α3β1 or α2β1 in SCC formation (36). Most importantly, these studies show that the tumors arise from the underlying basal layer cells (37). Newer studies point to dysfunction in environmental sensing, an integrin-dependent process, as a contributor to tumor formation and progression (19, 38, 39). We now find that CD98hc loss drastically compromises environment stiffness sensing and conversely leads to a marked reduction in matrix stiffness.

Importantly, the physiologic elasticity of skin has been long evaluated, using other types of technologies, to range between 5 and 10 kPa (20–22), gel compliance we used in our in vitro studies. In the AFM settings we utilized, this value reaches 0.53 MPa on average. This apparent discrepancy can be explained by recent work from Akhtar and colleagues (40) showing that elastic modulus of the constituent ECM molecules is higher than the one of the connective tissues themselves. The dermis is of fibrillar nature, and the nanoscopic AFM probe we used in vivo is likely to find very distinct areas, such as fibers, upon which to indent. This leads to measurement of elastic modulus reaching values in the MPa range. Finally, in good agreement with the explanation above, our data suggest that, in the absence of CD98hc, the ECM components responsible for structural rigidity, such as collagen content, might be modulated, leading to an overall decrease in stiffness and altering the skin mechanical properties

Integrin-mediated adhesions are intrinsically mechanosensitive and a large body of data implicates integrins in sensing mechanical forces (39, 41). Piccolo's laboratory pioneered the notion that mechanical signals and cell shape are converted into biochemical responses by two related transcription factors, YAP and TAZ (26, 42). In particular, YAP and TAZ are nuclear relays of mechanical signals exerted by ECM rigidity (25) and mCTGF, a well-characterized YAP/TAZ target, induces collagen deposition. CD98hc loss reduces ROCK activity in vivo, resulting in decreased signaling of YAP/TAZ, and concomitantly a decrease of mCTGF expression, suggesting a role for CD98hc in collagen deposition. Importantly, cells deficient for CD98hc lose their capacity to “read” physical and mechanical cues. The absence of CD98hc makes cells on a stiff ECM behave as if they were on a soft one. Thus, CD98hc tunes cellular stiffness sensing during tumorigenesis via the integrin/ROCK/YAP/TAZ pathway.

In summary, our data establish a central role for CD98hc in chemically induced tumor development and progression in skin and show that removal of CD98hc can induce tumor regression. We demonstrate that CD98hc mediates this process by acting as an amplifier of integrin-mediated mechanotransduction and that the gain in stiffness sensing provided by CD98hc forms the core of a positive feedback loop that increases the stiffness of the microenvironment and the cellular responses. Finally, we implicate YAP/TAZ as a RhoA/ROCK-regulated pathway that contributes to CD98hc-mediated tumorigenesis. These studies also highlight the potential synergic effect of blocking CD98hc, in an anticancer therapy, allowing targeting both tumor cell proliferation and tumor environment matrix stiffness.

No potential conflicts of interest were disclosed.

Conception and design: S. Estrach, E. Boulter, M.H. Ginsberg, C.C. Féral

Development of methodology: S. Estrach, F.S. Tissot, M.H. Ginsberg, C.C. Féral

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Boulter, S. Pisano, A. Errante, L. Cailleteau, M.H. Ginsberg, C.C. Féral

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Estrach, E. Boulter, S. Pisano, F.S. Tissot, M.H. Ginsberg, C.C. Féral

Writing, review, and/or revision of the manuscript: S. Estrach, S.-A. Lee, E. Boulter, M.H. Ginsberg, C.C. Féral

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.-A. Lee

Study supervision: M.H. Ginsberg, C.C. Féral

Other (performed experiments): C. Pons

The authors thank the IRCAN core facilities (microscopy and animal facility) for technical help.

This study was supported by grants from INSERM InCa/AVENIR (R08227AS), from Association pour la Recherche sur le Cancer (ARC R10159AA), from Agence Nationale de la Recherche (ANR R09101AS), through the “Investments for the Future” LABEX SIGNALIFE: program reference # ANR-11-LABX-0028-01, as well as from from NIH grant HL 078784 and HL 31950. E. Boulter was the recipient of a postdoctoral fellowship from the Ligue Nationale Contre le Cancer (RAB 12007 ASA) and a Marie Curie IRG from the European Union FP7 (agreement 276945).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.