Dual inhibition of glycolysis and glutaminolysis for synergistic therapy of rheumatoid arthritis

Abstract

Background: Synovial fibroblasts in rheumatoid arthritis (RAFLS) exhibit a pathological aberration of glycolysis and glutaminolysis. Henceforth, we aimed to investigate if dual inhibition of these pathways by phytobiological compound c28MS has the potential of synergistic therapy for arthritis by targeting both glucose and glutamine metabolism.

Methods: The presence of HK2 and GLS across various cell types and associated gene expression in human synovial cells and a murine model of arthritis was evaluated by scRNA-seq. The metabolic profiling of RAFLS cells was done using H¹-nuclear magnetic resonance spectroscopy under glycolytic and glutaminolytic inhibitory conditions by incubating with 3-bromopyruvate, CB839, or dual inhibitor c28MS. FLS functional analysis was conducted under similar conditions. ELISA was employed for the quantification of IL-6, CCL2, and MMP3. K/BxN sera was administered to mice to induce arthritis for in vivo arthritis experiments.

Results: scRNA-seq analysis revealed that many fibroblasts expressed Hk2 along with Gls with several genes including Ptgs2, Hif1a, Timp1, Cxcl5, and Plod2 only associated with double-positive fibroblasts, suggesting that dual inhibition can be an attractive target for fibroblasts. Metabolomic and functional analysis revealed that c28MS decreased the aggressive behavior of RAFLS by targeting both upregulated glycolysis and glutaminolysis. c28MS administered in vivo significantly decreased the severity of arthritis in the K/BxN model.

Conclusion: Our findings imply that dual inhibition of glycolysis and glutaminolysis could be an effective approach for the treatment of RA. It also suggests that targeting more than one metabolic pathway can be a novel treatment approach in non-cancer diseases.

Keywords: Fibroblast-like synoviocytes; Glucose metabolism; Glutaminase; Glutamine metabolism hexokinase; Rheumatoid arthritis.

Fig. 1

HK2 and GLS are associated with specific synovial fibroblast subsets. A scRNA-seq UMAP of all cells sequenced from Torres et al. (2023) [22] along with expression of Hk2, Gls, and the signature of both genes. B Expression of HK2 and GLS in publicly available data from Zhang et al. [25]. C, D Establishment of STIA fibroblast subclusters. E Expression of Gls and Hk2 in STIA fibroblast subclusters. F Expression of GLS and HK2 in previously defined fibroblast sub-cluster in human data from [25]

Fig. 2

Expression of HK2 and GLS is associated with differential gene expression in fibroblasts. A, B Segregation of synovial STIA fibroblasts into Gls-positive, Hk2-positive, double-negative, and double-positive. C Proportion of Gls-positive, Hk2-positive, double-negative, and double-positive cells in STIA fibroblasts. Each dot represents a sample. Statistical significance was completed using one-way ANOVA and Turkey’s post hoc test. *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001. D Heatmap of differentially expressed genes in Gls-positive, Hk2-positive, double-negative, or double-positive cells compared to all other cells. Expression is an average scaled expression of genes with an adjusted P value of < 0.01. E GO term enrichment in Gls-positive, Hk2-positive, double-negative, or double-positive fibroblasts. F Expression of selected marker genes in Gls-positive, Hk2-positive, double-negative, or double-positive fibroblasts from synovial STIA fibroblasts. G Expression of selected marker genes in GLS-positive, HK2-positive, double-negative, or double-positive fibroblasts from [25]

Fig. 3

Therapeutic inhibition of glycolysis and glutaminolysis. A Representative western blot showing that the expression of HK2 and GLS1 in glucose and glucose-free media remains unaffected by the action of c28MS in RAFLS (n = 3). B Color scale encoded heat map illustrating variations in the concentration of key metabolites in RAFLS (n = 3) between control and treatment groups in glucose medium (glucose medium; 25 mM of glucose and 6 mM of glutamine). C Color scale encoded heat map illustrating variations in the concentration of key metabolites in RAFLS (n = 3) between control and treatment groups in low-glucose medium (2 mM of glucose and 6 mM of glutamine). D One-dimensional H¹ nuclear magnetic resonance spectrum fold change in glucose and lactate levels in glucose medium (25 mM of glucose and 6 mM of glutamine) and glutamate and glutamine in low-glucose medium (2 mM of glucose and 6 mM of glutamine) of treated and untreated RAFLS (n = 3). Values are expressed as mean ± standard error mean. p < 0.05 was considered statistically significant

Fig. 4

Functional assays to establish inhibitory effects of c28MS and standard treatments in RAFLS in glucose and glucose-free/low-glucose medium (CB839: 300 nm, 3BrPy: 25uM, c28MS: 2uM). A–D, H–K RAFLS (n = 3) were plated for matrigel invasion and migration scratch assay in glucose and glucose-free medium. Quantification of the results (A, C, H, J) and representative images of RAFLS invasion and migration (B, D, I, K) are shown. E, L RAFLS (n = 3) were plated and starved overnight with 0.1% fetal bovine serum (FBS) and inhibitors added the following day in high-glucose (25 mM) and low-glucose (2 mM) medium. MTT was added on day 5 after the addition of conditions. RAFLS (n = 3) were starved overnight 0.1% FBS after plating. Compound and standard treatments were added in 1% and 10% FBS media under high-glucose (25 mM) and low-glucose (2 mM) and left to proliferate for 4 days in the presence of EdU and consequently counterstained with Hoechst 33,342 (blue) dye to assess the total proportion of cells. F, M Results were quantified as the percentage of EdU-positive cells among the total number of Hoechst 33,342-positive cells and representative images shown in G and N. Horizontal lines and error bars show the mean ± SD. * = p < 0.05; **** = p < 0.0001

Fig. 5

Effect of treatment on the levels of disease markers identified by scRNA-seq. A Volcano plots exhibiting the presence of IL-6, CCL2, and MMP3 between the indicated stages of disease in Hk2- and Gls- (double-positive) positive cells. Each dot represents a gene. Genes in red: adjusted p value < 0.05 and foldchange > 0.5 and <  − 0.5. p value calculated using FindMarkers() (Seurat).RAFLS (n = 3) were incubated with vehicle and inhibitors (CB839 (300 nM), 3BrPy (25 μM), and c28MS (2 μM)) for 1 h followed by stimulation with TNF (5 ng/mL) for 24 h. The production of IL-6, CCL2, and MMP-3 in the culture supernatants of RAFLS (n = 3) was measured using commercially available ELISA kits in the presence (B–D) or absence (E–G) of glucose. Values are mean ± standard error mean. p < 0.05 was considered statistically different

Fig. 6

Treatment with c28MS decreases clinical scores and histologic scores in the K/BxN mouse model of arthritis. A Clinical scores were determined in vehicle-treated animals (n = 5) and c28MS-treated animals (n = 5) at a dose of 2.5 mg/kg. One hundred fifty microliters of K/BxN mouse serum was injected at day 0 and treatment was administered intraperitoneally daily from day 0 and continued until day 10. B Histologic scores were determined on day 10 after serum transfer in vehicle-treated or c28MS-treated mice. C Sections of the ankle joints of vehicle or c28MS-treated mice were stained with hematoxylin and eosin (H&E) or Safranin O on day 10 after arthritis induction. Black arrows indicate joint inflammation and hypertrophy, and yellow arrows highlight cartilage damage, which was reduced in c28MS-treated mouse ankles. D, E Induction of arthritis and treatment (c28MS and MTX) prognosis in treated and untreated K/BxN mice model (n = 5 per group) D clinical score and E paw swelling. Horizontal lines and error bars show the mean ± SD. * = p < 0.05; **** = p < 0.0001