Diabetic kidney disease (DKD) is continuing to impose the highest burden and the strongest correlation with mortality due to the increasing diabetes incidence globally (1). The presentation of DKD in the clinic shows an extreme phenotypic heterogeneity (2). Patients with diabetes will not always develop clinically evident DKD even in the setting of poor glycemic control. This heterogeneous clinical cohorts derived the impetus to uncover endogenous protective factors against the development of DKD (3). The ratio between the tetramer and dimer structure of pyruvate kinase M2 (PKM2) mediates cellular glycolytic reshuffling and metabolic phenotype through its precise allosteric regulation, acting as a keystone in cellular glycolytic remodeling (4, 5). Tetrameric PKM2 has shown a protective role against the progression of DKD and chronic kidney disease (CKD) in type 1 and type 2 diabetes from clinical individuals, partly by rewiring the plethora of carbon compounds to catabolism, reducing toxic glucose metabolites, and thereby circumventing energy consumption and redox stress from synthesis (3, 6). The small-molecule allosteric agent TEPP-46 could transfer the dimeric PKM2 from a loose state (T-state) to a stable and compact state (R-state) to form a tetramer by binding to a pocket of the PKM2 subunit interface (5, 7). PKM2 tetramerization by TEPP-46 possesses high PK enzyme activity and blocks its nuclear translocation (5). TEPP-46 could restore hyperglycemia-induced glomerular and tubular metabolic phenotypes, further inhibiting fibrosis progression in DKD (3, 8).

Recently, the clinical relevance of decreased proximal reabsorption and hyperfiltration by sodium glucose cotransporter 2 inhibitors (SGLT2i) to improve the renal outcomes in patients with diabetes revolutionized the treatment of DKD and CKD, further establishing the core status of the renal proximal tubular center (9). Proximal tubular epithelial cells (PTECs) are metabolically active to impose a high energy demand for its reabsorption and transport processes. Due to its physiologic role, PTECs are susceptible to nutritional stress such as in the diabetic context. Increased reabsorption of glucose and sodium in PTECs under chronic hyperglycemic stress triggers renal hyperfiltration and exacerbates hypoxia, playing a crucial role in preserving the diabetic state and renal injury (10, 11). Compared with other metabolically active cells such as cardiomyocytes or skeletal muscle cells, PTECs show a relatively low metabolic flexibility toward glycolysis and mainly rely on fatty acids as energy source (12). Glycolytic adaptation or maladaptation of PTECs couples metabolic resilience to cell fate and the process of DKD, which remains poorly understood.

In our study, we explored the transcriptomic changes and mechanisms in response to PKM2 activator TEPP-46 in the hyperglycemic human renal proximal epithelial tubular cell line (HK-2 cells) in vitro to reveal the pathophysiological networks involved in the glycolytic reprogramming of hyperglycemic PTECs and thus to update and reshape the prevention strategy for DKD. A total of six independent replicates of HK-2 cells with three replicates in the case/control groups were used. We conducted a genome-wide transcriptome study in HK-2 cells cultured with high D-glucose (HG) for 7 days with (case group) or without (control group) the addition of TEPP-46 for another 1 day in the presence of HG to reveal differentially expressed (DE) genes; meanwhile, the gene co-expression network relationships are constructed by the weighted gene co-expression network analysis (WGCNA) tool (13) because the co-expressed genes with similar expression patterns across samples are controlled by the same transcriptional regulatory programs (14, 15). A DE gene-related biological enrichment analysis was subsequently performed to identify the potential functions in the major module of co-expression networks to reveal the candidate biomarkers to provide new mechanistic opportunities to remodel the cell resilience of PTECs.

The enzymatic role of pyruvate kinase (PK) is to catalyze the irreversible conversion of phosphoenolpyruvate (PEP) to pyruvate by the transfer of a high-energy phosphate from PEP to ADP to form ATP. In mammals, four different isoforms of PK (PKM1, PKM2, PKL, and PKR) were identified (4). All isoforms have a tetrameric form, while only PKM2 has a dimeric form besides the tetrameric form (4). As the terminal and last rate-limiting enzyme in the glycolytic pathway, PKM2 with its unique isoforms plays a crucial role in glycolytic reprogramming, which reconfigures cellular anabolic requirements and thus potentially induces metabolic transformation to alter cell identity and fate (4, 5, 27). The dimer state of PKM2 has low enzymatic activity and shifts the glycolytic flux into branching pathways, such as polyol pathway, pentose phosphate pathway, and uronic acid pathway, thus promoting anabolic function (glycogen synthesis, glycerol synthesis, nucleotide synthesis, or amino acid biosynthesis); the main purpose of the tricarboxylic acid (TCA) cycle is no longer to supply H⁺ and REDOX equivalents to the electron transport chain but to provide carbon source and REDOX equivalents for the subsequent anabolic metabolism. Consistent with this concept, the pro-proliferative role of dimeric PKM2 in proliferating cells, such as embryos or cancer cells, has been widely known (7, 28). Beyond the canonical enzymatic function, dimeric PKM2 could be translocated into the nucleus and act non-glycolysis enzyme function by regulating gene transcription (8), while PKM2 tetramer is located in the cytoplasm with a higher affinity with its substrate PEP to play the enzymatic role of PK that is related to ATP synthesis and catabolism known as PKM2 paradox in the Warburg effect (28). PKM2 interconnects glycolysis together with catabolic and anabolic processes, coordinating with the elaborate complementary metabolism of the TCA cycle intermediates.

The activation of glycolytic PKM2 by TEPP-46, reducing PKM2 dimerization, could prevent against fibrosis development and progression by reducing de novo glycine synthesis to downregulate collagen synthesis and secretion in myofibroblasts (29). Dimeric PKM2 located in the nucleus would stabilize the transcription hypoxia-inducible factor 1-alpha (HIF-1α) and directly phosphorylate the signal transducer and activator of transcription 3 (STAT3) to activate the transcription of several phosphorylated STAT3-targeted genes (8). In diabetic HK-2 cells, TEPP-46 decreased HIF-1α and phosphorylated STAT3 accumulations by decreasing the dimer PKM2 and inhibiting fibronectin, type I collagen α3, and TGF‐β1 expression, there and then preventing their downstream hypoxia and fibrosis (8). Here we intervened HK-2 cells under sustained HG exposition (7 days) with (case group) or without (control group) the addition of TEPP-46 for another 1 day and analyzed the genome-wide transcriptome data from the case and control groups. The results of our bioinformatics analyses showed the downregulation of the endocytosis pathway. Consistent with this, three of the top 10 downregulated DE genes (HSPA8, HSPA2, and HSPA1B) were involved in this pathway. Notably, all of them belong to the ubiquitous heat shock protein 70 (HSP70) family, and the downregulation of these three DE genes had been verified by qTR-PCR experimentsand further by western blotting experiments ( Figure 7 and Supplementary Figure S2). In addition, the four extremely downregulated DE genes (HSPA8, HSPA2, HSPA1B, and ARRB1) predicted 65 microRNAs in the conserved sites ( Figure 8 ), with the context++score percentiles ranging from 20 to 99 ( Supplementary File S3 ). Most of the microRNAs (n = 25) were predicted by HSPA8, while HSPA1B only predicted four microRNAs. We found that 17 and 20 microRNAs were predicted for HSPA2 and ARRB1, where hsa-miR-377-3p was predicted by these two downregulated DE genes with a score equal to 87 and 47, respectively. Two binding sites of hsa-miR-22-3p were predicted by ARRB1 in the positions of 429–436 and 4176–4182 in the 3′UTR region with a score equal to 90 and 70, respectively ( Figure 8 ).

The HSP70 family (Hsp70s) acts in a large variety of cellular housekeeping functions (30). Hsp70s participates in virtually all stages of the life of proteins through the allosteric control of their client substrates, cooperating with various other cellular machineries to perform the processes of protein folding/unfolding, membrane translocation, and aggregation/disaggregation (30, 31). Hsp70s could recognize and bind a KFERQ-like pentapeptide motif in the substrate proteins, making them the key proteins required for chaperone-mediated autophagy (CMA) to deliver cargo to lysosomes. CMA degrades target soluble proteins in a selective manner and is the predominant form of autophagy in the tubular system (32, 33). Hsp70s would regulate autophagic flux through the ubiquitin-mediated proteasome system and also the autophagosome–lysosome fusion system (34). Hsp70s also take part in receptor-mediated endocytosis, impacting their internalization and subsequent delivery to endosomes and lysosomes for ligand processing and receptor degradation or recycling (30). Overall, Hsp70s play the orchestrating and pleiotropic roles in dynamic versatile networks contacting endocytosis, autophagy, and lysosomes.

HSPA8 was the most downregulated DE gene [log2(FC) = -1.12] and also one core hub in PPI networks in this study. Located on chromosome 11q24.1, HSPA8 encodes the human soluble constitutive 71-kDa heat shock cognate protein, also known as Hsc70 (35). HSPA8 had been identified as the uncoating ATPase of clathrin-coated vesicles, which would disassemble the clathrin cages to modulate the clathrin-coated vesicle cycle and endocytic trafficking, acting as the essential late step in clathrin-mediated endocytosis (36). Accumulation of CMA-related HSPA8 proteins was observed in aged mutant striatum along with increased GAPDH and clustered lysosomes (32). Wen et al. (2020) reported that HSPA8 was upregulated in exosomes derived from high-glucose-treated renal tubular cells which would stimulate the proliferation and activation of fibroblasts to progress to fibrosis (37). Johnson et al. (2009) reported that HSPA8 could be co-precipitated with pyruvate dehydrogenase, isocitrate dehydrogenase, and the mitochondrial protease OMI in the isolated mitochondria of human embryonic kidney 293 cells, indicating a direct role of HSPA8 in the TCA cycle (38). HSPA1B and HSPA2, also known as HSP70-1/1B and HSP70-2/3, locate on chromosome 6p21.33 and 14q23.3, respectively; both of them encode for the classical Hsp70s proteins. HSPA1B was adaptively increased under salt stress in renal tubular and osmotic stress in collecting duct cells of rodents (39, 40). HSPA2 has prominent roles in the morphological differentiation of male germ cells as well as their functional transformation during post-testicular sperm maturation (41). Zhu et al. (2021) revealed the involvement of HSPA2 signaling pathway in fibroblast pathologies in diabetic wounds through integrated bioinformatic analyses (42). To enable to fulfil their multifarious influences on cell fate, Hsp70s has similarities in structure and redundancy in function; meanwhile, they also exhibit their own structural and biochemical characteristics, exhibiting pleiotropic and context-dependent roles in cells (30, 43, 44). Hsp70s, including HSPA8, HSPA2, and HSPA1B, could be either tumor-promoting or tumor-suppressing depending on cell identity and context (45–47). HSPA8 was involved in AQP2 internalization; coordinately, HSPA1B was likely to play a role in AQP2 trafficking to the apical plasma membrane (48). HSPA1B had been identified as one disease-relevant molecular target in advanced-stage Parkinson’s disease (49). Upregulated HSPA2 exerted a neuroprotective effect through its regulation of endocytosis in a rat middle cerebral artery occlusion model (50), and inhibition of HSPA8 would protect against spinal ischemia–reperfusion injury via astrocyte NF-κB/NLRP3 inflammasome pathway (51).

The endocytosis pathway controls the orchestra of myriad and specialized transport systems in PTECs; thus, it gates the internalization of extracellular substances and cell surface proteins, participates in processes including nutrient uptake, ion transport, junction formation, cellular properties, and signal transduction, and affects renal disease progression and multisystem complications (52, 53). The endocytosis pathway seems to function as a double-edged sword in PTECs: on one hand, endocytosis plays a vital role in the efficient reabsorption of excess ultra-filtered ions and solutes; on the other hand, it leads to the susceptibility of PTECs to excess nutrients and environmental stress. Under diabetic stress, the sustained activation of endocytosis would be uncoordinated with impaired energy metabolism and excessive oxidative stress, which further promotes and sustains dominos of pathological injuries. Excessive mobilization of membrane phospholipids associated with enhanced endocytosis would overwhelm the degradation capacity of the autophagy–lysosome system in PTECs (54).

The endocytic cycle, together with HSP70s, integrates pleiotropic and dynamic cellular roles in response to the internal and external environment. In Drosophila melanogaster, clathrin-uncoating ATPase Hsc70-4 could interact genetically with the activated mechanistic target of rapamycin (mTOR) signaling, leading to an increase in bulk endocytosis and a decrease in the targeted endocytic degradation of excess nutrient importers in cells and together increasing the intracellular nutrient burden (55). mTORC1 regulated endocytosis and nutrient transport in rodent proximal tubular cells and played a central role in the pathophysiology of DKD (9, 56). In diabetic proximal tubular cells, the increased mTORC1 activity and relevance of enhanced amino acid transport and probably decreased branched-chain amino acid degradation accelerated fibrogenesis (9). Recently, the activation of mTORC1 had been tied to intracellular glycolysis state. Orozco et al. (2020) reported that mTORC1 sensed metabolites downstream of aldolase and upstream of the GAPDH-catalyzed steps of glycolysis and pinpointed dihydroxyacetone phosphate (DHAP) as the key and sufficient metabolite to activate mTORC1 even in the absence of glucose (57). Thus, here we deduced that decreased mTORC1 activity by reducing the accumulation of glycolytic intermediate DHAP responding to TEPP-46 might be partly tied to the downregulation of Hsp70s (HSPA8, HSPA2, and HSPA1B) network and certain endocytosis pathways in hyperglycemic HK-2 cells, which would circumvent metabolic stress and remodel homeostasis between enhanced intracellular transport and impaired metabolic capacity to the natural potency.

The coordination of cellular metabolic remodeling and endocytic pathways represents an important feature of tubular cellular plasticity and adaptive homeostasis (52). Our bioinformatics analyses showed that the epidermal growth factor receptor (EGFR) was simultaneously involved in four downregulated pathways (adherens junction, regulation of actin cytoskeleton, endocytosis, and EGFR tyrosine kinase inhibitor resistance). EGFR trafficking is a key regulator in cell proliferation, differentiation, division, and survival, and EGFR internalization occurs via endocytosis by its recruitment into clathrin-coated pits (58, 59). Numerous studies had demonstrated that transient activation of the EGFR pathway was required for promoting kidney recovery from acute injury, whereas sustained activation of EGFR in PTECs under diabetogenic stimuli was sufficient to induce epithelial–mesenchymal transition and cause spontaneous and progressive renal tubulointerstitial fibrosis (10, 60). Here we proposed that PKM2 activator TEPP-46 could alter the intracellular trafficking of EGFR signaling cascade partly through its impact on endocytosis, in which the downregulated Hsp70s (HSPA8, HSPA2, and HSPA1B) were involved, presenting a new perspective on remodeling cellular mechanics and dynamics through metabolic phenotypes. However, functional research is still required for the candidate genes in this study to validate their molecular mechanism of in vitro reprogramming under hyperglycemic stress that orchestrates HK-2 cell functions.

Our study treated hyperglycemic HK-2 cells with PKM2 activator TEPP-46 and conducted genome-wide transcriptome analysis for the case/control group. Finally, we identified 2,902 DE genes including 87 and 56 extremely downregulated and upregulated DE genes (adjusted P-value ≤ 0.05 and |log₂(FC)| ≥ 1). In addition, 2,509 DE genes among them (86.46%) were co-expressed in the key module. Four extremely downregulated DE genes (HSPA8, HSPA2, HSPA1B, and ARRB1) and three extremely upregulated DE genes (GADD45A, IGFBP3, and SIAH1) enriched in the downregulated endocytosis (hsa04144) and upregulated p53 signaling pathway (hsa04115), respectively, were validated by the qTR-PCR experiments, where the relative expression levels of four extremely downregulated DE genes were significantly different (P-value < 0.05) between the case and control groups. Here we shed light that Hsp70s (HSPA8, HSPA2, and HSPA1B) are the key transcriptomic biomarkers responding to the PKM2 activator TEPP-46. Our results together highlight a possibility that PKM2 tetramerization induced by TEPP-46 in hyperglycemic HK-2 cells would reshape the interplay among endocytic trafficking, dynamics of protein folding–unfolding, and autophagy–lysosome system through the versatile networks of Hsp70s, rewiring the crosstalk between EGFR signal transduction circuits and metabolic stress to promote resilience.

The raw RNA sequencing data were deposited in the Gene Expression Omnibus (GEO) of National Center for Biotechnology Information (NCBI) with the accession number GSE205674 at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE205674.

XW and ZW conceived and designed the experiments. JY, XL, and ZW performed the experiments. ZW, JY, DH, and XW analyzed the data and wrote the manuscript. ZW, DH, XL, and XW improved the manuscript. ZW and JY contributed equally to this manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the Union Program of the Key Scientific and Technological Project of Henan Province (grant no. 2018020107) and the National Natural Science Foundation of China (grant nos. 82100896).

XW was employed by the company Konge Larsen ApS. DH was employed by Shijiazhuang Zhongnongtongchuang (ZNTC) Biotechnology Co., Ltd.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.