Deficiency of tumor suppressor Merlin facilitates metabolic adaptation by co-operative engagement of SMAD-Hippo signaling in breast cancer

Abstract

The NF2 gene encodes the tumor and metastasis suppressor protein Merlin. Merlin exerts its tumor suppressive role by inhibiting proliferation and inducing contact-growth inhibition and apoptosis. In the current investigation, we determined that loss of Merlin in breast cancer tissues is concordant with the loss of the inhibitory SMAD, SMAD7, of the TGF-β pathway. This was reflected as dysregulated activation of TGF-β signaling that co-operatively engaged with effectors of the Hippo pathway (YAP/TAZ/TEAD). As a consequence, the loss of Merlin in breast cancer resulted in a significant metabolic and bioenergetic adaptation of cells characterized by increased aerobic glycolysis and decreased oxygen consumption. Mechanistically, we determined that the co-operative activity of the Hippo and TGF-β transcription effectors caused upregulation of the long non-coding RNA Urothelial Cancer-Associated 1 (UCA1) that disengaged Merlin's check on STAT3 activity. The consequent upregulation of Hexokinase 2 (HK2) enabled a metabolic shift towards aerobic glycolysis. In fact, Merlin deficiency engendered cellular dependence on this metabolic adaptation, endorsing a critical role for Merlin in regulating cellular metabolism. This is the first report of Merlin functioning as a molecular restraint on cellular metabolism. Thus, breast cancer patients whose tumors demonstrate concordant loss of Merlin and SMAD7 may benefit from an approach of incorporating STAT3 inhibitors.

Figure 1.

Loss of Merlin is concordant with the loss of SMAD7 in breast cancer. Breast tumor tissue microarrays were stained by immunohistochemistry for the expression of Merlin and SMAD7. The staining intensity is depicted as an IHC score. (A, B) represent the scores of Merlin and SMAD7 with respect to the grade of the tumor when compared to normal breast tissue; (C, D) represent IHC scores for Merlin and SMAD7 with respect to nodal involvement; (E, F) represent staining for Merlin and SMAD7 in relation to the occurrence of metastasis; (G, H) represent staining for Merlin and SMAD7 in relation to the disease stage. Mean Merlin and mean SMAD7 IHC scores were computed with respect to tumor grade. The mean Merlin score for normal breast tissue is significantly higher than that in tumor tissues of all grades (Dunnett’s test, P < 0.0001 for each). SMAD7 staining intensity decreases with advanced grade of the tumor tissue (normal tissue versus grades II and III: Dunnett’s test, P < 0.0001 for each). The difference in SMAD7 intensity with respect to grade I is not significant (Dunnett’s test, P = 0.0553). The staining intensity for Merlin and SMAD 7 are concordantly significantly decreased with nodal involvement (Dunnett’s test, P < 0.0001 for each). Merlin and SMAD7 levels are significantly decreased overall in breast tumor tissues (Dunnett’s test, P < 0.0001 for each) regardless of metastasis. (I) Representative immunohistochemistry images are shown for Merlin and SMAD7 staining. Panels a and c represent normal breast tissues stained for Merlin; panels b and d represent normal breast tissues stained for SMAD7; panels e and g are representative of Merlin staining seen in node negative and metastatic breast cancer tissues; panels f and h depict SMAD7 for the corresponding tissues. (J) Denotes the correlation between Merlin expression and SMAD7. A scatterplot of mean IHC scores of Merlin versus SMAD7 shows a very strong positive linear relation between them (R² = 0.97, ANOVA P = 0.0003). The resulting least squares regression line (or line of best fit) is Mean (Merlin) = −83.15 + 20.17*Mean (SMAD7). It shows that one unit increase in Mean (SMAD7) corresponds on the average to 20.17 units increase in Mean (NF2) levels. Pearson’s correlation coefficient between mean Merlin and mean SMAD7 when analyzed with respect to tumor grade is r = 0.985.

Figure 2.

Merlin deficiency leads to the co-operative activation of YAP/TAZ/SMAD signaling. MCF7 and MCF10AT cells knocked down for NF2 (MCF7 kd and MCF10AT kd), control (MCF7 NT and MCF10AT NT), and SUM159 cells restored for Merlin expression were analyzed for (A) SMAD binding site activity by SBE4 luciferase reporter assay and (B) YAP/TAZ activity with an 8XGTIIC luciferase reporter assay. Merlin deficient cells demonstrate upregulated TGF-β signaling evidenced by increased activity of the Smad-binding element with concomitant inactive Hippo signaling as evident by increased activity of the TEAD-binding element. Merlin restoration reduced SMAD and YAP/TAZ activities. (C) Control and Merlin-deficient MCF7 and MCF10AT cells and SUM159 cells (vector control and Merlin expressing) were immunoblotted for the effectors of TGF-β signaling pathway: total SMAD2/3, phosphorylated SMAD2/3 (Ser465/467 and Ser423/425, respectively), SMAD4 and SMAD7, and for members of the Hippo pathway (total YAP/TAZ, inhibitory phosphorylated YAP and TAZ (Ser127 and Ser89, respectively)). GAPDH was used as a loading control. (D) Merlin deficient MCF7 and MCF10AT cells (kd) demonstrate significantly increased levels of UCA1. In contrast, SUM159 cells re-expressing Merlin show significantly decreased levels of UCA1. UCA1 levels were analyzed by quantitative RT-PCR. β-actin was used as control gene. Relative expression was measured by 2^−∆∆Ct method.

Figure 3.

Loss of Merlin induces bioenergetics alterations in breast cancer cells. (A) Protein lysates of MCF7 and MCF10AT cells silenced for Merlin (kd) and control (NT) and SUM159 Merlin restored and vector control cells were immunoblotted for STAT3, phosphorylated STAT3 (Tyr705), total AKT (pan-AKT), phosphorylated AKT (Ser473) and HK2. Cell-free culture media of MCF7 and MCF10AT cells knocked down for NF2 (MCF7 kd and MCF10AT kd, respectively), their control (MCF7 NT and MCF10AT NT, respectively) and SUM159 cells (vector control and Merlin-reexpressing) was assayed for (B) glucose consumption (C) and lactate production. Values were normalized to total protein. (D) Cells depleted for Merlin displayed an increased ECAR indicating increased glycolysis in the cells. SUM159 cells restored for Merlin expression demonstrated decreased ECAR. (E) Merlin deficient cells demonstrated decreased OCR and increased glycolysis, in contrast to SUM159 cells restored for Merlin expression.

Figure 4.

Merlin keeps a check on cellular metabolism and glycolysis mediators. (A) The lysates of MCF10AT cells knocked down for Merlin (kd) and the respective control (NT) and SUM159 cells (vector only and restored for Merlin) were analyzed by a targeted metabolomics assessment for the levels of the indicated metabolites of the TCA cycle. Restoration of Merlin increases the total cellular levels of TCA intermediates while Merlin deficiency has the opposite effect. The steady-state transcript levels of glucose transporters (GLUT) (B), monocarboxylate transporters transporters (MCT) (C) and lactate dehydrogenase (LDHA and LDHB) (D) were analyzed by quantitative RT-PCR from MCF10AT and MCF7 cells (NT and kd) and SUM159 cells restored for Merlin (SUM159 Merlin) and their respective vector control (SUM159 vec). β-actin was used as control gene. *P < 0.05 relative to their indicated respective controls.

Figure 5.

UCA1 directs cell metabolism towards aerobic glycolysis through AKT and STAT3 activation in Merlin-deficient cells. (A) Immunoblotting of protein lysates of Merlin-deficient MCF7 cells silenced for UCA1 (siUCA1) and non-target control (NTC) showed reduced level of phosphorylated AKT (Ser473), phosphorylated STAT3 (Tyr705) and HK2 upon UCA1 silencing. HSP90 was used a loading control. The levels of glucose uptake (B) and lactate export (C) were decreased and the overall amounts of metabolites of the TCA cycle were increased (D) as a result of blunted UCA1 expression. (E) Protein lysates of NF2-deficient MCF7 cells treated with 2 µM of STATTIC inhibitor or DMSO (vehicle control) for 48h were collected. The expression of phosphorylated STAT3 (Tyr705) and HK2 were attenuated upon STAT3 activity inhibition; β-actin was used as a loading control. The levels of glucose consumption (F) and lactate production (G) were also reduced following STATTIC treatment. (H) There was an overall increase in the amounts of metabolites of the TCA cycle in STATTIC treated MCF7 cells deficient for Merlin. *P < 0.05.

Figure 6.

STAT3 inhibition preferentially impacts Merlin-deficient cells. (A) Merlin-deficient cells develop an invasive 3D morphology. The STAT3 inhibitor, STATTIC, appears to not only restore circularity to this structure but also limits its growth. (B) STATTIC inhibits invasive potential of Merlin-deficient MCF7 cells through the modified Boyden chamber. (C) The ability of anchorage-independent growth was assessed by soft agar colony assay and, in a dose-dependent effect, STAT3 inhibition diminished colony formation, being more effective in NF2-silenced cells (kd). *P < 0.05. (D) Deficiency in Merlin expression results in the degradation of SMAD7, which uncouples its inhibitory modulation on the TGF-β signaling pathway. The resulting collaborative activity of TGF-β and Hippo effectors upregulates HK2 expression that dictates a metabolic preference towards aerobic glycolysis.