Network Pharmacology Reveals MAPK- and RTK-mediated Erythropoietic Signalling as a Shared Mechanism of Dapagliflozin and Empagliflozin
Melike Saday Bozkurt, Ali TurunçIntroduction:
Sodium-glucose cotransporter-2 (SGLT-2) inhibitors consistently raise haemoglobin and haematocrit in patients with type 2 diabetes, heart failure and chronic kidney disease, and post-hoc mediation analyses suggest that this haematological shift accounts for a substantial fraction of their cardiovascular benefit. The molecular targets that link SGLT-2 inhibition to erythroid expansion, however, remain incompletely mapped.
Objective:
To identify the protein targets and biological pathways through which dapagliflozin and empagliflozin modulate erythropoiesis, and to test whether these targets are differentially expressed in human diabetic kidney tissue.
Methods:
Drug targets were predicted with SwissTargetPrediction and PharmMapper. An erythropoiesis gene set (n = 1 293) was assembled from GeneCards using three queries and a gene disease association threshold of ≥10. Drug-target and erythropoiesis sets were intersected; pathway enrichment was performed in Enrichr (KEGG 2026, GO Biological Process 2025, Reactome 2024) and a Protein-Protein Interaction (PPI) network was constructed in STRING v12 (combined score ≥0.4). Hub proteins were ranked by a composite score combining degree, betweenness, closeness and eigenvector centrality. Two diabetic-kidney microarray datasets, GSE30529 (tubulointerstitium) and GSE96804 (glomeruli), were re-analysed with GEO2R to test whether the shared targets were differentially expressed in disease tissue.
Results:
Both inhibitors converged on 33 erythropoiesis-related proteins, including HSP90AA1, CASP3, EGFR, KDR, MAPK14 and HSPA8. KEGG enrichment was led by the MAPK signalling pathway (12 of 33 genes; adjusted p = 4.2 × 10-12), followed by AGE–RAGE signalling in diabetic complications and VEGF signalling. Hub analysis identified HSP90AA1 and CASP3 as the most central nodes (composite scores 0.94 and 0.92). Of the 33 shared targets, 32 were measured in both diabetic-kidney datasets; CASP3 was up-regulated in tubulointerstitium (log2FC = +1.16, adj. p = 1.0 × 10-3) and in glomeruli (logⁿFC = +0.71, adj. p = 1.6 × 10-8), the only target reaching significance in both cohorts.
Discussion:
The convergence of dapagliflozin and empagliflozin on stress-MAPK, receptor tyrosine kinase and chaperone–apoptotic modules positions SGLT-2 inhibition at the interface of three biological programmes already implicated in erythroid expansion. The dual-cohort upregulation of CASP3 in diabetic kidney tissue suggests a CASP3-dependent erythroblast checkpoint that may be relieved by gliflozin therapy, while the centrality of HSP90AA1 supports a chaperone-mediated route to enhanced EPOR competence. AGE–RAGE enrichment anchors these findings in the disease-specific milieu of type 2 diabetes, providing a mechanistic bridge between the metabolic and haematological actions of the class.
Conclusion:
SGLT-2 inhibitors converge on a tight cluster of MAPK-, RTK- and chaperonerelated proteins that overlap with the molecular machinery of erythropoiesis. CASP3 emerges as the most robustly supported candidate hub across both transcriptomic cohorts, and the HSP90AA1-EPOR axis offers a plausible chaperone-mediated route to enhanced erythropoietin signalling. These findings refine the candidate target set for the haematocrit-raising effect of gliflozins and provide a rational basis for targeted experimental follow-up.