Transglutaminase 2 Binds to the CD44v6 Cytoplasmic Domain to Stimulate CD44v6/ERK1/2 Signaling and Maintain an Aggressive Cancer Phenotype

Cutaneous squamous cell carcinoma (CSCC) is a UV light–induced disease that is among the most common cancers (1, 2). CSCC is treated by surgical resection, but recurrence of treatment-resistant and aggressive metastatic cancer is a common problem (1, 2). Epidermal cancer stem-like cells (ECS cells) are critical cells that give rise to and drive tumor maintenance (3). ECS cells are extremely aggressive as they self-renew, display high level expression of stem cell markers, are resistant to conventional treatment, and form rapidly growing, highly aggressive and invasive tumors (4). Because they are a highly aggressive cell subpopulation in CSCC, we utilize enriched populations of ECS cells in the current studies. Treatment with agents designed to reduce survival of these cells is an important strategy.

Transglutaminase 2 (TG2) is a unique oncogenic protein that is the subject of intense study relating to its role as a cancer cell survival factor in many tumor types (3, 5–8). TG2 is composed of four domains including an N-terminal β-sandwich, a catalytic core domain that contains the catalytic triad (Cys277, His335, and Asp358) required for the acyl transfer reaction to form covalent ε-(γ-glutamyl) lysine bonds, and two C-terminal β-barrel domains (3, 5). TG2 also has a functionally important guanosine triphosphate (GTP)-binding site that is present in a cleft between the catalytic domain and the penultimate β-barrel that is required for its procancer actions (3, 5, 9–11). Our studies show that TG2 level is increased in ECS cells and drives the aggressive cancer phenotype (3, 9, 10, 12). TG2 has both intra and extracellular roles, and a key feature is its interaction with membrane receptors to trigger the cancer phenotype in a range of cancer types (5, 13, 14). For example, TG2 forms a complex with α6/β4-integrin (12, 15, 16) and GPR56 (17, 18).

Given that TG2 impacts a wide range of signaling cascades, it is likely that TG2 interacts with and activates multiple transmembrane receptors that have not been identified. CD44 is a widely distributed single-chain type I transmembrane adhesion protein and receptor that is encoded by a single-gene on chromosome 11 (19). The gene produces a standard CD44 protein (CD44s) and alternatively spliced variants (CD44v) that have additional sequence inserted on the cytoplasmic side immediately upstream of the transmembrane domain. The CD44 variable isoforms are important cancer stem cell markers, cancer cell survival factors and oncogenes (19, 20). CD44v6 is a particularly important variant that is highly expressed in various cancers (19, 20). The CD44v6 structure includes the N-terminal extracellular region which includes the disulfide bond-stabilized hyaluronan (HA)-binding domain, and the juxta-membrane v6 variable region domain, the transmembrane domain and the intracellular cytoplasmic tail which binds to ezrin, radixin, and moesin (ERM proteins) and ankyrin (20, 21). HA is the CD44v6 ligand and HA/CD44 complex formation triggers increased CD44v6-binding of ERM proteins, ankyrin, F-actin and various signal transduction proteins to the CD44v6 cytoplasmic tail. This binding activates cellular responses that drive cancer cell proliferation, invasion, migration and tumor formation. CD44v6 stimulates oncogenesis in a host of cancers (19, 22). Moreover, CD44v6 is selectively expressed in cancer stem cells (23). CD44v6 is present in skin cancer (24, 25) and is enriched in ECS cells in tumors (24, 26).

In this report, we identify a novel TG2/CD44v6 complex and show that this complex maintains/stimulates CD44v6 activity to enhance cancer cell survival. This involves TG2-binding to the CD44v6 cytoplasmic domain to enhance cancer cell proliferation and invasion. We further show that both TG2 and CD44v6 are required to maintain the cancer phenotype. Mutants of CD44v6 that lack the cytoplasmic domain lack TG2 binding, and display reduced biological activity. In addition, TG2 knockdown/knockout or treatment with TG2 specific inhibitor reduces the CD44v6-related responses. These findings suggest that a novel TG2/CD44v6 interaction is required to drive the CSCC cancer phenotype.

RPMI 1640 (10–040-CV) and DMEM (10–013-CV) were purchased from Corning. l-Glutamine (25030–164), 0.25% trypsin-EDTA (25200–056) and sodium pyruvate (11360–070) were obtained from Gibco. DMEM/F12 (1:1; DMT-10–090-CV) was purchased from Mediatech Inc. B27 serum-free supplement (17504–044) was purchased from Invitrogen.

Cell lysis buffer [9803 (20 mmol/L Tris-HCl (pH 7.5) containing 150 mmol/L NaCl, 1 mmol/L Na₂EDTA, 1 mmol/L EGTA, 1% Triton, 2.5 mmol/L sodium pyrophosphate, 1 mmol/L beta-glycerophosphate, 1 mmol/L Na₃VO₄, 1 μg/mL leupeptin, and 1 mmol/L PMSF)], and anti–E-Cadherin (3195S), anti-ERK1/2 (9102) and anti-ERK1/2-P (4376) antibodies were purchased from Cell Signaling Technology. Anti-CD44v6 (ab78960), anti-Sox2 (ab15830–100) and anti-Twist (ab49254) antibodies were purchased from Abcam. Peroxidase-conjugated donkey anti-rabbit IgG (NA934V) and peroxidase-conjugated sheep anti-mouse IgG (NXA931) antibodies were obtained from GE Healthcare. β-Actin (A5441) antibody, 4′,6-diamidino-2-phenylindole (DAPI; D9542), BSA (B4287), EGF (E4269), insulin (19278), HA (53163), normal Rabbit IgG (NI01), and heat-inactivated FCS were obtained from Sigma-Aldrich. CD44v6-siRNA (SMARTpool, M-009999–03–0005) was purchased from Dharmacon. TG2 antibody (sc-73612), normal mouse IgG (sc-2025), control siRNA (sc-37007), and TG2-siRNA (sc-37514) were purchased from Santa Cruz Biotechnology. pPGK-T7/2-CD44v6(fl; 137816) and pPGK-T7/2-CD44v6(-cyt; 137817) expression plasmids were purchased from Addgene. These encode full-length and cytoplasmic domain deleted CD44v6, respectively (27). Pierce Protein A/G resin (20421) was purchased from Thermo Fisher Scientific. BD BioCoat Millicell inserts (d = 1 cm, 8 μmol/L pore size, #353097) and Matrigel (354234) were obtained from BD Biosciences. MEK1/2 inhibitor U0126 (V112A) was obtained from Promega. TG2 inhibitor (NC9) was prepared as previously described (28). Two-tailed Student t test was used for binary comparison between control and experimental groups. Experiments were repeated a minimum of three times and the results are presented as mean ± SEM.

Cells and tissues were sonicated and then boiled in Laemmli buffer (0.063 mmol/L Tris-HCl, pH 7.5, 10% glycerol, 5% SDS, 5% β-mercaptoethanol), and equivalent amounts of protein were electrophoresed on denaturing and reducing polyacrylamide gels. The proteins were then transferred to nitrocellulose membranes, blocked with 5% nonfat milk for 1 hour and incubated with primary antibody (1: 1,000) overnight at 4°C followed by secondary antibody incubation (1:5,000) for 2 hours at room temperature. The blots were extensively washed, and antibody-specific binding was visualized using a chemiluminescence detection assay.

Meso-1 cells (29) were maintained in RPMI1640 medium containing 2 mmol/L l-glutamine supplemented with 10% heat-inactivated FCS. SCC-13 (30) and HaCaT (31) cells were maintained in DMEM containing 4.5 mg/mL d-glucose, 2 mmol/L l-glutamine, 100 mmol/L sodium pyruvate, and 10% heat-inactivated FCS. Mycoplasma tests were performed at regular intervals to assure an absence of contamination and cell line authenticity was confirmed by short tandem repeat analysis. Enriched populations of ECS cells were grown as spheroids in ultra-low attachment plates in stem cell medium containing DMEM/F12 (1:1), 2% B27 serum‐free supplement, 20 ng/mL EGF, 0.4% BSA, and 4 μg/mL insulin (32). ECS cell levels are enriched to approximately 20% of cells in these cultures, as compared with 0.15% in monolayer cultures. The resulting spheroids were dissociated into single-cell suspensions of ECS-enriched cells that were used in all cell culture and tumor growth experiments. ECS cells represent the most aggressive cell subpopulation present in CSCC tumors (4). For electroporation, 3 μg siRNA or plasmid was suspended in 100 μL of VPD‐1002 nucleofection reagent electroporated into cells using an AMAXA electroporator.

SCC-13 CD44v6 knockout cell lines (SCC13-CD44v6-KOc1–2-2, SCC13-CD44v6-KOc1–2-3 and SCC13-CD44v6-KOc1–2-5) were generated using CRISPR technology as previously described followed by puromycin treatment and limiting dilution to select stable clones (32, 33). CD44v6-specific CRISPR guide RNA, forward (5′-caccGCTGTGCAGCAAACAACACAGGGG) and reverse (5′-aaacCCCCTGTGTTGTTTGCTGCACAGC), were identified at https://www.genome.gov/genetics-glossary/CRISPR and cloned in the U6-driven pSpCas9(BB)-2A-Puro (PX459) V2.0 vector and 3 μg were electroporated into SCC-13 cells using AMAXA electroporator. At 48-hour postelectroporation, cells were selected with 2 μg/mL puromycin. Cell clones were generated by dilution cloning and CD44v6 knockout was confirmed by immunoblot analysis. We delivered pSpCas9(BB)-2A-Puro (empty vector) to SCC-13 cells and selected with puromycin. These cells were then compared with wild-type (WT) SCC-13 cells and displayed identical Matrigel invasion and spheroid formation rates and biomarker expression (i.e., they display an identical phenotype). For this reason, we have used the WT SCC-13 cells as controls in these experiments.

A total of 500 μg protein was diluted into 400 μL cell lysis buffer and incubated with primary antibody on a microtube rotator at 4°C overnight. Pierce Protein A/G resin was prewashed with cell lysis buffer and 100 μL (packed volume) was added to the mixture for 2 to 4 hours at room temperature. The beads were then washed, the antibody–bead–protein complex was boiled in sample buffer, and the supernatant was electrophoresed and transferred to nitrocellulose for immunoblot analysis (34).

For spheroid formation assay, cells were seeded at a density of 40,000 cells/well into ultra-low attachment plates in stem cell medium. Spheroid number and growth were monitored for three days and measured using ImageJ. For invasion assay, 20,000 ECS cells were seeded into the top chamber of BD BioCoat Millicell inserts precoated with 250 μg/mL Matrigel, and maintained in DMEM containing 1% FCS. ECS cells were enriched to 20% of the total cell population by spheroid growth in stem cell medium and then used in the invasion assay. The bottom chamber contained DMEM supplemented with 10% FCS. The cells were permitted to migrate for 20 hours following the serum gradient and then the insert membrane was fixed and stained with DAPI to visualize the invaded cells.

For tumor studies enriched ECS cells (100,000) were suspended in 100 μL PBS containing 30% Matrigel and subcutaneously injected into each front flank of 8-week-old NSG (NOD/SCID/IL-2Rg^−/−) mice. Five mice (10 tumors) were used per treatment. After 24 hours, treatment was initiated with NC9 dissolved in DMSO/Captisol mixture and intraperitoneally injected at 0 or 20 mg/kg body weight three times per week (Monday/Wednesday/Friday; ref. 12). Tumor growth was monitored for 4 or 5 weeks. Tumor volume was calculated as 4/3π × (diameter/2)³ (4). At 4 or 5 weeks the tumors were harvested, photographed, and extract was prepared for immunoblot analysis. Spheroid-enriched ECS cells, derived from WT SCC-13 and two CD44v6 clonal knockout lines, were injected into each front flank of NSG mice (5 mice/treatment group) and tumor size was measured weekly. At 4 weeks the tumors were harvested, photographed, and extract was prepared for immunoblot analysis. The animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Maryland Baltimore.

The data generated in this study are available upon request from the corresponding author.

We are interested in how the TG2 procancer regulator stimulates the cancer phenotype in squamous cell carcinoma (SCC) and other cancers (4, 32, 33). TG2 is highly elevated in cancer stem cells and so we are interested in signaling proteins that may interact with TG2 to maintain the aggressive cancer phenotype. CD44v6 is an important procancer regulator that is frequently overexpressed in cancer stem cells and is involved in various processes including maintenance of cancer cell stemness (3, 5). We therefore assessed the impact of TG2 on CD44v6 level. We initially treated SCC-13 cells with control or TG2 siRNA and monitored the impact on CD44v6 and ERK1/2 level. Fig. 1A shows that TG2 knockdown cells have reduced CD44v6 and that this is associated with reduced ERK1/2 activity (ERK1/2-P). Moreover, the reduction in CD44v6 level and ERK1/2 activity in TG2-siRNA treated cells is associated with reduced spheroid formation (Fig. 1B) and invasion (Fig. 1C). We repeated these experiments in another epidermis-derived cell line, HaCaT (31). We show that TG2 knockdown in HaCaT cells reduces CD44v6 level and ERK1/2 activity (Fig. 1D) and that this is associated with reduced spheroid formation and invasion (Fig. 1E and F).

NC9 is a highly specific TG2 inhibitor that reduces cancer cell survival (28, 35) and suppresses tumor growth (12, 28, 35) by interfering with TG2 transamidase and GTP-binding activity (3, 9, 10, 12). Other studies show that TG2 GTP-binding activity, but not transamidase activity, is important for TG2 procancer activity in cutaneous SCC cells (3, 9, 10, 12). We therefore examined the impact of TG2 inhibitor treatment on CD44v6 level and ERK1/2 activity in SCC-13 cell tumors. Figure 1G shows that NC9 treatment reduces SCC-13 tumor growth in an NSG xenograft model and that this is associated with a reduction in CD44v6 level and ERK1/2 activity. In addition, we observe a decrease in Sox2 and twist levels and an increase in E-cadherin level indicating that the NC9-dependent loss of CD44v6/ERK1/2 function is associated with reduced tumor cell stemness and EMT. We wanted to examine the impact of TG2 knockout on CD44v6 signaling; however, we have not been able to create a stable TG2 knockout clone in cutaneous SCC cells. We therefore used TG2 knockout Meso-1 cells to study the effect of TG2 absence on CD44v6 and ERK1/2 level and activity. Figure 1H shows that TG2 knockout Meso-1 cells display markedly reduced invasion and express reduced levels of CD44v6, ERK1/2, and ERK1/2-P. Thus, TG2 functions to maintain CD44v6 and ERK1/2 activity in SCC-13 and HaCaT cells and in SCC-13 tumors. In addition, we show that TG2 maintains CD44v6 level and ERK1/2 level and activity in the Meso-1 mesothelioma cancer cells.

We next examined the impact of interfering with CD44v6 function on the cancer phenotype. We treated SCC-13 cells with control- or CD44v6-siRNA and monitored the impact on the cancer phenotype. Figure 2A shows that CD44v6 knockdown reduces ERK1/2 level and ERK1/2 activity, and that this is associated with reduced cell spheroid formation (Fig. 2B) and invasion (Fig. 2C). CD44v6-siRNA also reduced HaCaT cell ERK1/2 level and activity (Fig. 2D) and spheroid formation (Fig. 2E) and invasion (Fig. 2F). As a complimentary approach, we used CRISPR/Cas9 methods to create three SCC-13 CD44v6 knockout cell lines and then compared CD44v6 and ERK1/2 levels and activity. We observe a marked reduction in CD44v6 and ERK1/2 activity in the knockout cells (Fig. 2G) and show that this is associated with reduced spheroid formation (Fig. 2H) and invasion (Fig. 2I).

We note that TG2 knockdown reduces ERK1/2 activity but does not reduce ERK1/2 total level (Fig. 1A and D). In contrast, as shown in Fig. 2A, D, and G, CD44v6 knockdown or knockout reduces both ERK1/2 level and this is associated with reduced ERK1/2-P. We are not sure why this difference is observed. We speculate that interfering with the TG2/CD44v6 pathway at different levels may impact parallel signaling events that influence ERK1/2 level and activity. These studies indicate that ERK1/2 function can be reduced due to a reduction in activity or a reduction in ERK1/2 level. These findings suggest that TG2 and CD44v6 are both required for optimal maintenance of ERK1/2 signaling and stimulation of the cancer phenotype.

The above studies suggest that interfering with TG2 or CD44v6 function attenuates the aggressive cancer cell phenotype. HA is the major ligand that binds to the CD44v6 extracellular domain to activate CD44v6-related signaling (36–38). We therefore wanted to examine if interfering with TG2/CD44v6 function could attenuate HA-stimulation of the cancer phenotype. We first examined the impact of TG2 knockdown on the ability of HA to stimulate SCC-13 and HaCaT cell invasion. As shown in Fig. 3A, treatment of SCC-13 cells with HA increases cell invasion through Matrigel, but treatment with TG2-siRNA reduces basal invasion rate and HA stimulated invasion. HaCaT cells display a similar response to these treatments (Fig. 3B). We next determined if interfering with CD44v6 function reduces the response to HA. Figure 3C shows that HA stimulates invasion of cells treated with control siRNA and that basal and HA stimulated invasion is reduced under CD44v6 knockdown conditions. We repeated the HA challenge experiments in CD44v6 knockout cells (SCC13-CD44v6-KOc1–2-2 and SCC13-CD44v6-KOc1–2-3 cells), which shows that CD44v6 knockout reduces invasion compared with control cells and that invasion is not restored by HA treatment (Fig. 3D). Analysis of cell extracts confirm the absence of CD44v6 in the knockout cell lines and show that this is associated with reduced ERK1/2 and ERK1/2-P (Fig. 3D) TG2 levels were not reduced by CD44v6 knockout (Fig. 3D).

We next examined the role of TG2/CD44v6-dependent ERK1/2 signaling. Fig. 3E shows that HA treatment increases invasion, and that both basal and HA-stimulated invasion are reduced in the presence of U0126, a MEK1/2 inhibitor. We repeated the HA challenge experiment in CD44v6 knockout cells (SCC13-CD44v6-KOc1–2-2 and SCC13-CD44v6-KOc1–2-3 cells) in the presence or absence of U0126 (Fig. 3F) and showed that CD44v6 knockout reduces invasion, and that the invasion is not further suppressed by U0126 treatment. Analysis of cell extracts confirm the absence of CD44v6 in the knockout cell lines and show that this is associated with reduced ERK1/2 and ERK1/2-P (Fig. 3F). These findings implicate ERK1/2 as a key downstream target of TG2/CD44v6.

The above findings suggest that both TG2 and CD44v6 are required to maintain ERK1/2 activity and the cancer phenotype, and that TG2 is required to maintain CD44v6 level. We next assessed the mechanism of TG2/CD44v6 action. We hypothesized that this may involve formation of a TG2/CD44v6 complex. To test this hypothesis, we prepared SCC-13 cell extracts for immunoprecipitation with IgG control serum, and anti-TG2 and anti-CD44v6 antibodies. We find that both anti-TG2 and anti-CD44v6 specifically precipitate ERK1/2 and that anti-CD44v6 also precipitates TG2 (Fig. 4A) which suggests formation of a TG2/CD44v6/ERK1/2 complex. We next repeated this experiment using HaCaT cell extracts. We show that anti-TG2 precipitates CD44v6 and ERK1/2 and that anti-CD44v6 precipitates TG2 and ERK1/2 (Fig. 4B).

An important issue is identifying the CD44v6 domain requirements for TG2-binding. To test this, we electroporated SCC-13 CD44v6 knockout cells with expression vectors encoding full-length CD44v6, CD44v6(fl), or CD44v6 lacking the intracellular cytoplasmic domain, CD44v6(-cyt). Extracts were then prepared for immunoprecipitation. Figure 4C (left) shows that anti-TG2 precipitates TG2, CD44v6(fl) and ERK1/2 in CD44v6(fl) expression cells, but that anti-TG2 does not precipitate CD44v6(-cyt) or ERK1/2 in CD44v6(-cyt) expressing cells. The right panel shows the reciprocal experiment demonstrating that anti-CD44v6 precipitates CD44v6(fl), TG2 and ERK1/2 in CD44v6 expressing cells, but does not precipitate TG2 or ERK1/2 in CD44v6(-cyt) expressing cells. These findings suggest that TG2 and ERK1/2 bind to the CD44v6 cytoplasmic domain to form a complex.

The findings in Fig. 1G show that treatment with TG2 inhibitor reduces tumor formation and that this is associated with reduced CD44v6 level and reduced ERK1/2 activity. We wanted to determine if interfering with CD44v6 phenocopies these responses. To achieve this, we compared the growth of WT and CD44v6 knockout SCC-13 cells in a NSG mouse xenograft model. Figure 5A and B show that two separate clonally derived CD44v6 knockout cell lines display markedly reduced tumor growth as compared with WT SCC-13 cells. Monitoring of tumor extracts revealed that the reduction in tumor growth is accompanied by reduced CD44v6, ERK1/2, and ERK1/2-P levels (Fig. 5C). In addition, Sox2 and Twist levels are reduced, and E-cadherin level is increased in CD44v6 knockout tumors indicating a reduction in tumor cell stemness and EMT (Fig. 5C). These findings suggest that both TG2 maintains CD44v6 level (Fig. 1G) and that TG2 and CD44v6 cooperate to maintain ERK1/2 signaling and the cancer phenotype. Moreover, these results suggest that TG2 and CD44v6 activity are both required to drive optimal tumor growth.

TG2 is an important multifunctional procancer factor and stem cell survival protein that impacts a host of signaling cascades to drive the cancer phenotype in many cancer types (3, 5). We have described important roles of TG2 in epidermal SCC where it maintains VEGF (13), NRP1 (39), GIPC1/SYX/RhoA/p38 (39), α6/β4-integrin (9, 12), YAP1/TAZ (12), HGF/MET (32), and mTOR (33) signaling to stimulate tumor growth.

In the current study, we examine TG2 regulation of CD44v6. CD44v6 is an important oncogenic protein and stem cell marker (22, 40) that is expressed in epidermis (41) and in epidermis-derived cancer (24). Our studies show that TG2 knockdown reduces CD44v6 level in SCC-13 and HaCaT cells and reduces spheroid formation and Matrigel invasion. In addition, treatment with TG2 inhibitor reduces SCC-13 cell CD44v6 level and tumor growth. We also tested the impact of TG2 knockout on CD44v6. We previously created TG2 knockout Meso-1 cells that display markedly reduced spheroid formation, Matrigel invasion, migration, and tumor growth (42). Meso-1 is a mesothelioma cancer cell line derived from peritoneal mesothelioma (43). Characterization of WT and TG2 knockout Meso-1 cells reveals a marked reduction in CD44v6 level in TG2 knockout cells that is associated with attenuation of the cancer phenotype. These findings show that TG2 maintains CD44v6 level in SCC-13, HaCaT and Meso-1 cells, indicating that this regulation is a general phenomenon. In this study, we did not focus on the mechanism whereby TG2 maintains CD44v6 level; however, an elegant study by Matei and colleagues showed that treatment of ovarian cancer cells with recombinant TG2 increases mRNA levels of CD44 and that this is inhibited by treatment with RelB siRNA (44). They also showed that TG2 maintains CD44 level in ovarian cancer cell xenografts by increasing CD44 mRNA level (44). Thus, NFκB signaling plays an important role in this model. TG2 may regulate CD44v6 mRNA level in SCC-13 cells, although it is also possible the TG2 interaction with CD44v6 acts to stabilize the complex and maintain CD44v6 level. Future studies will examine the mechanism whereby TG2 controls CD44v6 level.

We next examined the effect of CD44v6 loss on the cancer phenotype. We show that CD44v6 knockdown reduces SCC-13 and HaCaT cell spheroid formation and Matrigel invasion. This is also observed in SCC-13 CD44v6 knockout cells which display a marked and stable reduction in spheroid formation and Matrigel invasion. CD44v6 regulates a host of downstream signaling cascades (45). These include Src, Ras, MEK1/2 and ERK1/2, PI3K/Akt, Rac1, Rho and Cdc42, and ERM proteins and ankyrin (19, 46). In addition, CD44v6 interacts with various transmembrane receptor proteins including VEGFR, EGFR, and MET (46). Of particular interest is the impact of TG2 and CD44v6 on ERK1/2 signaling. Our studies reveal a marked reduction in ERK1/2 activity in TG2 compromised cells that phenocopies the ERK1/2 reduction in CD44v6 knockdown and knockout cells. In addition, treatment with the MEK1/2 inhibitor (U0126) reduces ERK1/2 activity and suppresses the cancer phenotype.

These findings suggest that a TG2/CD44v6/ERK1/2 pathway drives the cancer phenotype. However, it could be more complicated than this, since CD44v6 also interacts with receptors such as EGFR which can contribute to ERK1/2 activation (46). An important feature is that TG2 knockdown reduces CD44v6 level, and ERK1/2 activity. In contrast, CD44v6 knockdown does not impact TG2 level, suggesting that TG2 may initiate the signaling cascade.

HA is the major ligand that binds to the CD44v6 extracellular domain to activate CD44v6-related signaling (36–38). We therefore wanted to know if inhibiting TG2 or CD44v6 signaling (by knockdown or knockout) would attenuate the ability of HA to activate CD44v6-dependent responses. These studies show that treatment of WT CSCC cells with HA stimulates ERK1/2 activity and the cancer phenotype. In contrast, loss of TG2 or CD44v6 impedes the ability of HA to stimulate these responses. This suggests that both proteins are required to maintain optimal HA stimulation of downstream signaling and the cancer response. It is interesting that HA treatment has been reported to increase TG2 expression (47). This finding, coupled with our current results, suggest that TG2 and CD44v6 may form a positive feedback system that drives the cancer phenotype.

We next wanted to explore the mechanism of action of TG2 and CD44v6. The N-terminal CD44v6 extracellular domain contains motifs that serve as docking sites for various extracellular ligands including extracellular matrix glycoproteins and proteoglycans that activate CD44v6 signaling (48); however, HA is the major ligand that binds to the CD44v6 extracellular ligand binding domain to activate CD44v6 function (36–38). The signal is conveyed via the transmembrane domain to the intracellular cytoplasmic domain (45) which binds to downstream effector proteins to activate cell responses. These effector proteins include ERM proteins and ankyrin, which bind to the CD44v6 cytoplasmic domain and F-actin (46). Rho family proteins and Src, Ras, MEK1/2, and ERK1/2 also bind to the CD44v6 cytoplasmic domain (46).

TG2 exists in the cell cytoplasm in the closed/folded GTP-binding conformation and in the extracellular environment as the open/extended transamidase active conformation (49). This localization suggests that TG2 could bind to CD44v6 on the extracellular or intracellular domains. However, given that GTP bound TG2 has been strongly implicated in the pathogenesis of cancer (9), we surmised that the closed/folded GTP-binding form of TG2, present in the intracellular environment, would most likely interacts with CD44v6. We therefore expressed CD44v6(fl) and CD44v6(-cyt) in CD44v6 knockout cancer cells and studied the interaction of these proteins with TG2 and ERK1/2. Biochemical experiments show that anti-TG2 coprecipitates TG2, CD44v6 and ERK1/2 in cells expressing CD44v6(fl), but that anti-TG2 cannot precipitate CD44v6 or ERK1/2 in cells expressing CD44v6(-cyt), suggesting that TG2 and ERK1/2 bind to the CD44v6 intracellular cytoplasmic domain to form a TG2/CD44v6/ERK1/2 complex. However, we do not yet know if this is a direct TG2/CD44v6 interaction or if other proteins are involved.

On the basis of these findings, we propose that TG2 and ERK1/2 bind to the CD44v6 cytoplasmic domain and that this interaction results in activation of ERK1/2 signaling to stimulate the cancer phenotype and tumor growth (Fig. 5D). This is consistent with findings showing the CD44v6 activates ERK1/2 signaling in thymoma cells (50). The fact that TG2/CD44v6/ERK1/2 complex formation is required to optimally activate ERK1/2 signaling is an important finding. We have described important roles of TG2 in epidermal SCC where it maintains VEGF (13), NRP1 (39), GIPC1/SYX/RhoA/p38 (39), α6/β4-integrin (9, 12), YAP1/TAZ (12), HGF/MET (32), and mTOR (33) signaling to stimulate tumor growth. Among these targets, TG2 has been shown to form a complex with α6/β4-integrin (12, 15, 16) and GPR56 (17, 18). This study adds CD44v6 to the group of TG2 interacting proteins. It is interesting that TG2 can bind to these highly varied target proteins. For this reason, future studies should focus on the TG2 structural constraints that permit these interactions and identifying specific TG2 regions involved in partner formation.

R.L. Eckert reports grants from NIH during the conduct of the study. No disclosures were reported by the other authors.

X. Chen: Conceptualization, investigation, writing–review and editing. G. Adhikary: Conceptualization, supervision, investigation, methodology, writing–review and editing. J.J. Newland: Investigation, methodology. W. Xu: Investigation. J.W. Keillor: Resources. D.J. Weber: Resources, investigation. R.L. Eckert: Conceptualization, resources, supervision, funding acquisition.

This work was supported by NIH grants (R01 CA211909) to R.L. Eckert and utilized the facilities of the Greenebaum Comprehensive Cancer Center (P30 CA134274) at the University of Maryland School of Medicine.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.