Tumor-Stroma Mechanics Coordinate Amino Acid Availability to Sustain Tumor Growth and Malignancy

Abstract

Dysregulation of extracellular matrix (ECM) deposition and cellular metabolism promotes tumor aggressiveness by sustaining the activity of key growth, invasion, and survival pathways. Yet mechanisms by which biophysical properties of ECM relate to metabolic processes and tumor progression remain undefined. In both cancer cells and carcinoma-associated fibroblasts (CAFs), we found that ECM stiffening mechanoactivates glycolysis and glutamine metabolism and thus coordinates non-essential amino acid flux within the tumor niche. Specifically, we demonstrate a metabolic crosstalk between CAF and cancer cells in which CAF-derived aspartate sustains cancer cell proliferation, while cancer cell-derived glutamate balances the redox state of CAFs to promote ECM remodeling. Collectively, our findings link mechanical stimuli to dysregulated tumor metabolism and thereby highlight a new metabolic network within tumors in which diverse fuel sources are used to promote growth and aggressiveness. Furthermore, this study identifies potential metabolic drug targets for therapeutic development in cancer.

Keywords: YAP/TAZ; carcinoma-associated fibroblast; extracellular matrix; mechanotransduction; metabolic crosstalk; metastasis; stiffness; tumor niche.

Figure 1:. ECM stiffening reprograms tumor niche cell metabolism and coordinates amino acid exchange to fuel pro-tumoral activities.

A) Extracellular flux analyses of SCC12 plated on soft or stiff matrix (mean of 9 wells from 3 independent experiments). B) Glucose-derived carbon labeling patterns of metabolites in SCC12 plated on soft (1kPa) or stiff (8kPa) hydrogel and treated with U-¹³C₆-glucose (mean of n=3 independent experiments). C-D) Glutamine (Gln) flux (C) and glutamate (glu) and Aspartate (Asp) flux (D) of SCC12 plated on 1kPa or 8kPa hydrogel (n=3). E) Extracellular flux analyses of CAF plated on soft or stiff matrix (mean of 9 wells from 3 independent experiments). F) Glucosederived carbon labeling patterns of metabolites in CAFs plated on soft (1kPa) or stiff (8kPa) hydrogel and treated with ¹³C-glucose (mean of n=3 independent experiments) G-H) Gln flux (G) and glu and Asp flux (H) of CAFs plated on 1kPa or 8kPa hydrogel (n=3). I) Scheme of the metabolomic experiments for J-K. J-K) Metabolites that were significantly elevated in SCC12-conditioned medium (CM), decreased in double-conditioned medium (SCC12-CM added to CAF and then collected), and elevated intracellularly in CAFs treated with SCC12CM (J, n=3). Metabolites that were significantly elevated in CAF-CM, decreased in double CM (CAF-CM added to SCC12 cells and then collected), and elevated intracellularly in SCC12 cells treated with CAF-CM (K, n=3). Data normalized to 1kPa L) The relative cell numbers of SCC12 cells are shown, cultivated on the indicated substrate in DMEM, 2mM Gln, and the indicated Asp concentration at 72h. Mean of n=9 wells from 3 independent experiments. M) Representative pictures and quantification showing change in percent of proliferative (Ki67+) SCC12 plated on stiff (8kPa) substrate 24h after treatment with the indicated conditioned media (CM; n=3). N) Quantification by traction force microscopy of contractile forces generated by CAF, cultivated on 8kPa hydrogel in DMEM, 2mM Gln, and following treatment with the indicated Glu concentration. Mean of n=6 wells from 3 independent experiments. O) Representative heat map showing contractile forces generate by CAF plated on 8kPa hydrogel following treatment with indicated CM. Mean of n=6 wells from 3 independent experiments. In all panels, data are expressed as the mean ± SD (*P < 0.05, **P < 0.01, ***P < 0.001). Paired samples were compared by 2-tailed Student’s t test, while 1-way ANOVA and post-hoc Tukey’s tests were used for group comparisons. Scale bars 50μm. See also Figure S1 and Figure S2.

Figure 2:. Metabolic reprogramming is dependent on glutamine metabolism and GLS1 expression and sustains CAF and SCC pro-tumoral activities.

A-B) Gln-derived carbon labeling patterns of metabolites in SCC12s (A) and CAFs (B) plated on soft (1kPa) or stiff (8kPa) hydrogel and treated overnight with U-¹³C₅-Gln (mean of n=3 independent experiments). C) Representative image of immunoblot analysis confirming the knockdown of GLS1 by 2 independent siRNA sequences in SCC12s. D-E) Intracellular Gln and Glu levels (D) and extracellular Glu level compared to t0 (E) of SCC12 48h after GLS1 inhibition by either siRNAs (siGLS1) or pharmacological inhibitor (CB839; n=3). F) Representative images of immunoblot analysis confirming the knockdown of GLS1 by 2 independent siRNA sequences in CAFs. G-H) Intracellular Gln and Glu levels (G) and extracellular Asp level compared to t0 (H) in CAFs 48h after GLS1 inhibition by either siRNAs (siGLS1) or a pharmacological inhibitor (CB839; n=3). I) Representative pictures and quantification showing change in percent of proliferative (Ki67+) SCC12 cells plated on stiff (8kPa) substrate 24h after treatment with the indicated conditioned media (CM; n=3). J) Representative heat map of traction force microscopy experiments showing contractile forces generate by CAF plated on 8kPa hydrogel following treatment with indicated CM. Mean of n=6 wells from 3 independent experiments. K) Representative pictures and quantification showing invasion in three-dimensional co-culture assay after the indicated treatment. Mean of at least n=12 spheroid from three independent experiments. In F and I, mean expression in control groups (si-NC) was assigned a fold change of 1 to which relevant samples were compared. Data are expressed as the mean ± SD (*P < 0.05, **P < 0.01, ***P < 0.001) of at least 3 independent experiments. Paired samples were compared by 2-tailed Student’s t test, while 1-way ANOVA and post-hoc Tukey’s tests were used for group comparisons. Scale bars: In I and J 50μm; in K 400μm See also Figure S3.

Figure 3:. SLC1A3 enables Aspartate-dependent nucleotide biosynthesis to sustain proliferation in SCCs, while SLC1A3 promotes glutamate-dependent glutathione synthesis to sustain cell contractility in CAFs.

A) Asp-derived carbon labeling patterns of metabolites in SCC12 plated on soft (1kPa) or stiff (8kPa) hydrogel and treated overnight with ¹³C₄-Asp (mean of n=3 independent experiments). B) Glu-derived carbon labeling patterns of metabolites in CAFs plated on soft (1kPa) or stiff (8kPa) hydrogel and treated overnight with ¹³C₅-Glu (mean of n=3 independent experiments). C) Intracellular level of total glutathione (GSHt) and ratio of GSHt/reduced glutathione (GSH) of CAFs plated on soft (1kPa) or stiff substrate (8kPa) 24h after treatment with the indicated conditioned media (CM). Mean of n=9 wells from three independent experiments. D) Representative pictures and quantification showing the intracellular level of ROS and superoxide of CAFs following treatment with the indicated CM. Mean of 50 cells from 3 independent experiments. Scale bar: 20μm E) Representative pictures and quantification showing the F-actin cytoskeleton rearrangement as well as the P-MLC2 level of CAFs plated on stiff hydrogel (8kPa) following treatment with the indicated CM. Mean of 50 cells from 3 independent experiments. Scale bar 50μm F) RT-qPCR analyses of SLC1A family expression in SCC12 plated of soft or stiff substrate (n=4). G) RT-qPCR analyses of SLC1A family expression in fibroblasts (n=5) and CAFs (n=5). H) Percent of proliferative (PCNA+) SCC12 cells plated on stiff hydrogel 48h after the indicated treatments (n=3). I) Quantification by traction force microscopy of contractile forces generated by CAF plated on 8kPa hydrogel following the indicated treatment. Mean of n=6 wells from 3 independent experiments. J-K) Representative pictures and quantification showing invasion in three-dimensional co-culture assay after the indicated treatment. Mean of at least n=12 spheroids from three independent experiments. Scale bars 400μm. Data are expressed as the mean ± SD (*P < 0.05, **P < 0.01, ***P < 0.001) of at least 3 independent experiments. Paired samples were compared by 2-tailed Student’s t test, while 1-way ANOVA and post-hoc Tukey’s tests were used for group comparisons. See also Figure S4.

Figure 4:. Metabolic reprogramming is coordinated by a YAP/TAZ-dependent mechanotransduction pathway.

A-B) Quantification of SLC1A3 and GLS1 protein expression level in SCC12s (A) and CAFs (B) following siRNA-mediated YAP/TAZ knockdown or forced expression of YAP by cells infection with a lentiviral vector containing the YAP coding sequences (pYAP) and compared with cells transfected with a control siRNA (siNC) or with cells infected with a control vector (pGFP), respectively. Representative images of 3 independent experiments were shown. C-H) SCC12 cells (C-E) and CAFs (F-H) were plated on soft hydrogel (1kPa) and incubated with magnetic beads coupled to collagen. Representative confocal imaging (C and F) showing YAP/TAZ localization following the indicated treatments (n=3). Red dots: Paramagnetic beads. RT-qPCR revealed increased expression of GLS1 and SLC1A3 after magnet stimulation, but not GAPDH (negative control, D and G). This effect was blunted by YAP/TAZ siRNA knockdown (E and H). I) Sequence analysis predicted the presence of TEAD binding sites in the promoter regions (1,500 bp upstream to the start codon) of GLS1 and SLC1A3. J-K) ChIP-qPCR confirmed the presence of TEAD/YAP binding sites in the GLS1 and SLC1A3 promoter regions in both SCC12s (J) and CAFs (K). Results are expressed as percentage of total input DNA prior to immunoprecipitation with anti-YAP or anti-IgG control. Means of 3 independent experiments performed in triplicate. CTGF and CYR61, two known YAP targets, were used as positive controls. Data are expressed as the mean ± SD (*P < 0.05, **P < 0.01, ***P < 0.001) of at least 3 independent experiments. Paired samples were compared by 2-tailed Student’s t test, while 1-way ANOVA and post-hoc Tukey’s tests were used for group comparisons. Scale bars: 50μm. See also Figure S5.

Figure 5:. Alteration of mechanotransduction affects metabolic reprogramming and tumor progression in vivo.

A) Following cancer cell implantation, mice were treated with daily BAPN (100 mg/kg; n = 14); with daily i.p. injections of verteporfin (20mg/kg; n = 26) or vehicle, (4T1 cells n=26). 67NR cells were used as control (non-invasive tumor) B) Atomic force microscopy revealed decreased tumor stiffness in BAPN- and verteporfin-treated mice. Data are represented by Tukey boxplots. Median represents at least 70 measures from n=3 mice per group. C) Intratumoral GLS activity was measured. Each dot represents a mouse. D-E) Tumor cells isolated from the indicated tumors were analyzed for the levels of Asp and glu (data normalized to the 67NR group; each dot represents a single mouse). F-G) Representative picrosirius red images (F) and quantification (G) showing the collagen deposition and remodeling as well as representative IHC images (F) and quantification (G) showing GLS1, SLC1A3 and PCNA staining. Each dot represents a mouse. H-K) Tumor progression was analyzed as assessed by the tumor volume (H), lung macro metastasis analysis (I-J) and survival outcome (K). Each dot represents a mouse. Data are expressed as the mean ± SEM (*P < 0.05, **P < 0.01, ***P < 0.001). Group comparisons were performed by 1-way ANOVA and post-hoc Tukey’s tests. Scale bars: 50μm. See also Figure S6.

Figure 6:. Inhibition of metabolic reprogramming in either CAF or cancer cells blunts tumor progression.

A) Following 67NR co-implantation with CAFs stably transfected with indicated doxyxycline inducible shRNA, mice were treated with doxyxycline (n=10 per group). B) Graphic representation of tumor formation induced by 67NR cells alone or in the presence of CAFs stably transfected with indicated shRNA. Total numbers of mice bearing tumors or invasive tumors after treatment are shown. C) Representative images of primary tumors isolated from mice (left panels) showing 67NR cells invading from the primary tumor (T) into the adjacent tissue (aT). Picrosirius red staining visualized by both parallel (upper panels) and orthogonal (middle panels) light display tumor ECM remodeling at the areas invaded by the tumoral 67NR cells. Representative IHC images show PCNA staining and the overall percentage of PCNA+ tumor cells. D) Following implantation of stably transfected 4T1 cells, mice were treated with doxycycline in order to induce inhibition of either GLS1 (n=10), SLC1A3 (n=10), or both GLS1 and SLC1A3 together (n=11). E-H) Tumor progression was assessed by the tumor volume (E), percentage of PCNA+ tumor cells (F), lung macro metastasis number (G) and survival outcome (H). Each dot represents a mouse. I) Following 4T1 cells implantation, mice were treated with daily i.p. injections of CB839 (n=10); with daily i.p. injections of TFB-TBOA (n=10); with daily i.p. injections of both TFB-TBOA and CB839 (n=10) or vehicle (n = 10). J-M) Tumor progression was assessed by the tumor volume (J), lung macro metastasis number (K), survival outcome (L), and percentage of PCNA+ tumor cells (M). Each dot represents a mouse. Data are expressed as the mean ± SEM (*P < 0.05, **P < 0.01, ***P < 0.001). Group comparisons were performed by 1-way ANOVA and posthoc Tukey’s tests. Scale bars: 50μm. See also Figure S7.

Figure 7:. Additive effects among pharmacological inhibitors of metabolic reprogramming in primary HNSCC tumors.

A-D) Analysis of GLS1 (A) and SLC1A3 (C) mRNA levels in human head and neck squamous carcinoma tumors (HNSCC, n=497) and normal head and neck tissues (n=44). Kaplan-Meier survival analysis is shown for patients with HNSCC by log-rank test. Patients were divided by the expression levels of GLS (B) or SLC1A3 (D) mRNA. High expression: upper quartile (n=130); Low/medium expression: below the upper quartile (n=389). E) Representative images of primary HNSCC tumors are shown as isolated from a cohort of 48 patients stratified by their invasiveness score. Picrosirius red staining visualized by both parallel (upper panels) and orthogonal (middle panels) light displays tumor ECM remodeling, and representative IHC images show GLS1 (red) and SLC1A3 (gray) staining (lower panels). F) Quantification of collagen remodeling, GLS1, and SLC1A3 protein expression levels in low, middle, and high invasive HNSCC patients. G-H) Correlation between GLS staining and invasiveness and between SLC1A3 staining and invasiveness are shown. Pearson correlation coefficient (R²) are indicated. I-J) Representative images are shown of patient-derived spheroids (PDS) treated 72h with the indicated drugs and quantification of the surrounding invasion (n=7 HNSCC patients; each dot represents a PDS (n>5), while each color represents a single patient. K-M) In the HNSCC PDX mouse model, following tumor implantation, mice were treated with daily i.p. injections of CB839 (N=3; n=6); with daily i.p. injections of TFB-TBOA (N=3; n=6); with daily i.p. injections of both TFB-TBOA and CB839 (N=3; n=6) or vehicle (N=3; n = 6). HNSCC progression was quantified by tumor volume (K) measurement, PCNA (red) staining (L) and overall percentage of PCNA+ tumors cells (M). Data are expressed as the mean ± SEM (*P < 0.05, **P < 0.01, ***P < 0.001). Paired samples were compared by 2-tailed Student’s t test, while 1-way ANOVA and post-hoc Tukey’s tests were used for group comparisons. Scale bars:100μm.