PKM2 controls cochlear development through lactate-dependent transcriptional regulation

Significance

The role and mechanism of glucose metabolism in mammalian cochlear development remains elusive. Using cochlear organoids and transgenic mice, we showed that glycolytic metabolism is activated during cochlear sensory epithelium morphogenesis and that the key enzyme of glycolysis, PKM2, serves as an essential regulator of sensory epithelium formation. We further demonstrated that PKM2-dependent lactate accumulation promotes lactylation at histone H3 lysine 9, thus further leading to epigenetic modifications of numerous transcription factors. In addition, our results indicated that PKM2 overexpression and lactate treatment promote HC generation in both mouse and human cochlear explants. These findings reveal the essential role of PKM2-dependent glycolytic metabolism in cochlear development and provide potential targets for HC regeneration.

Abstract

Understanding the role of metabolic processes during inner ear development is essential for identifying targets for hair cell (HC) regeneration, as metabolic choices play a crucial role in cell proliferation and differentiation. Among the metabolic processes, growing evidence shows that glucose metabolism is closely related to organ development. However, the role of glucose metabolism in mammalian inner ear development and HC regeneration remains unclear. In this study, we found that glycolytic metabolism is highly active during mouse and human cochlear prosensory epithelium expansion. Using mouse cochlear organoids, we revealed that glycolytic activity in cochlear nonsensory epithelial cells was predominantly dominated by pyruvate kinase M2 (PKM2). Deletion of PKM2 induced a metabolic switch from glycolysis to oxidative phosphorylation, impairing cochlear organoid formation. Furthermore, conditional loss of PKM2 in cochlear progenitors hindered sensory epithelium morphogenesis, as demonstrated in PKM2 knockout mice. Mechanistically, pyruvate is generated by PKM2 catalysis and then converted into lactate, which then lactylates histone H3, regulating the transcription of key genes for cochlear development. Specifically, accumulated lactate causes histone H3 lactylation at lysine 9 (H3K9la), upregulating the expression of Sox family transcription factors through epigenetic modification. Moreover, overexpression of PKM2 in supporting cells (SCs) triggered metabolism reprogramming and enhanced HC generation in cultured mouse and human cochlear explants. Our findings uncover a molecular mechanism of sensory epithelium formation driven by glycolysis-lactate flow and suggest unique approaches for mammalian HC regeneration.

As a critical worldwide public health issue, deafness is primarily caused by damage to mechanosensitive hair cells (HCs) in the cochlea due to insults, including genetic factors, aging, noise, and ototoxic drugs (1). In mammals, the loss of HCs in the cochlea is irreversible, and HC regeneration serves as a promising strategy for achieving hearing restoration. In contrast to the robust regenerative ability of the intestine and liver, the mammalian inner ear, like the central nervous and cardiac tissues, exhibits limited potential for regeneration after damage (2). Surrounding the HCs in the mouse cochlea, some subtypes of nonsensory epithelial cells (NSECs) have the potential to replace lost HCs by asymmetric cell division or direct transdifferentiation under certain modulations (3, 4). These NSECs have certain progenitor cell features and maintain a short window of regenerative capacity during the first week after birth; notably, this capacity is lost 1 wk after birth. Clues from inner ear development would enable establishing strategies to promote SC dedifferentiation or cell cycle reentry to achieve HC regeneration.

A promising target for driving HC regeneration is metabolism modulation, as recent evidence has identified metabolism as a central regulator of cell fate transition (5). Although the cascade of molecular events that orchestrate inner ear development (6) and extracellular clues guiding HC generation (7) has been extensively explored, the contribution of metabolic regulation to inner ear development or regeneration has received less attention. Among metabolic processes, glycolysis is regarded as the primary source of energy for sustaining cell proliferation, like in most cancer cells (8, 9). During the driving periods of rapid cellular growth, cells favor the glycolytic state over the oxidative phosphorylation (OXPHOS) state; in contrast, oxidative metabolism is preferred in mature cells to support cellular maintenance and homeostasis (10). Research on glycolysis metabolism in the mammalian inner ear has focused on damage protection, such as the overexpression (OE) of the glycolytic enzyme glucose-6-phosphate dehydrogenase (G6PD), which prevents age-related hearing loss progression (11). In vertebrates, recent studies have shown that disrupting the glycolytic gene phosphoglycerate kinase-1 (PGK1) impairs the formation of HCs and neurons in zebrafish otic vesicles (12) and that glucose metabolic flux from glycolysis is required for the specification of HC positional identity and patterning in the developing chicken sensory epithelium (13). However, there may be a difference in the glycolytic metabolic flux of the inner ear between nonmammalian vertebrates and mammals, just as HCs in mammals do not spontaneously regenerate after injury, as in nonmammalian vertebrates.

In addition to the generation of ATP energy, glycolytic metabolism can play a critical role in the regulation of gene transcription, protein modification, and even epigenetic modification (14). The versatile regulatory effects of glycolysis stem from related metabolites and enzymes. An important representative is pyruvate kinase (PK), which catalyzes the last step of the glycolytic process that converts phosphoenolpyruvate (PEP) to pyruvate (15). Among the four PK isoforms, PKL and PKR are expressed in the liver and erythrocytes, respectively, and are produced by alternative splicing of precursor PKL RNA. The other two PK isoforms, pyruvate kinase M1 (PKM1), are generally expressed in differentiated cells, while PKM2 is expressed in actively growing cells such as embryonic and proliferating cells (16). Several recent reports have shown the importance of PKM2 in regulating cell proliferation and cell fate transition, including promoting cardiac regeneration (17) and preserving the pluripotency of pluripotent stem cells (18), highlighting the potential of glycolytic enzymes in regenerative medicine. However, the role of glycolytic metabolism in mammalian inner ear development and regeneration remains poorly characterized.

In this study, using cochlear organoid culture systems, we found that elevated canonical glycolysis cascades and the critical enzyme PKM2 are strongly activated during the cell expansion stage. To elucidate the role and mechanism of PKM2 in cochlear development, we generated a PKM2 conditional knockout (CKO) mouse line and crossed with Foxg1-Cre and Sox9-CreER mice to specifically delete PKM2 in cochlear progenitor cells. The absence of PKM2 led to restrained cell proliferative capacity and reduced cochlear size, as well as decreased HC generation. Mechanistically, PKM2-mediated glycolysis triggered lactate accumulation, followed by epigenetic modification of histone lysine 3 (H3) through lactylation (Kla) to sustain cochlear organoid expansion. By combining RNA sequencing (RNA-seq) and CUT&Tag, we revealed lactate-mediated epigenetic modifications prominently altered cochlear gene expression patterns. Finally, we showed that PKM2 OE by AAV vector and lactate administration enhanced HC regeneration in the cochlea.

Results

Elevated Glycolytic Metabolism During Cochlear Organoid Expansion.

To explore potential targets for HC regeneration, we utilized the transcriptomic profiles of cochlear organoids in different culture systems to search for relevant evidence. We used four three-dimensional (3D) culture conditions for Lgr5+ NSEC-derived organoid expansion: EFI (EGF, bFGF, and IGF), EFI-C (EFI and CHIR99021), and EFI-CVP6 (EFI, CHIR99021, VPA, pVc, and 616452) (19), a culture system we previously described (20), EFI-CL (EFI, CHIR99021, and LPA) (Fig. 1A). We found that NSECs achieved the most robust organoid formation capacity under the EFI-CL condition (Fig. 1 B–D). Next, we conducted bulk RNA-seq to compare the genetic profiles of the expanding organoids in EFI-CL and primary NSECs. Among the enriched biological processes from the upregulated genes of the organoid group by Gene Ontology (GO) analysis, the term “Response to glycoside” ranked the highest among all the biological processes (Fig. 1E). Gene set enrichment analysis (GSEA) revealed that canonical glycolysis accompanied by the NADH regeneration process was strongly enriched in the organoid group, while genes related to the mitochondrial respiratory chain complex were strongly downregulated (Fig. 1F). These findings prompted us to determine which glucose metabolism process participates in organoid expansion.

Fig. 1.

Glycolysis metabolism is activated during cochlear organoid expansion. (A) Schematic diagram showing the workflow for cochlear organoid induction. (Scale bar, 100 μm.) (B) Bright-field images of proliferating organoids generated with different expansion media. (Scale bar, 100 μm.) (C and D) Quantification of organoid-forming capacity and organoid viability in different expansion media (n = 3). (E and F) GO term analysis of upregulated genes and GSEA of DEGs from Lgr5+ NSECs and proliferating cochlear organoids. (G) Live images of mitochondria (MitoTracker) in proliferating and differentiating organoids. (Scale bar, 50 μm.) (H) Representative quantification of the ECAR of proliferating organoids cultured in different media. (I) Schematic of the linkage between the cell metabolic phenotype and cell fate. (J and K) Heatmap and columns showing the relative levels of glycolytic and TCA cycle metabolites in proliferating organoids by mass spectrometry (MS) (n = 3). One-way ANOVA and Tukey’s multiple comparison test were performed for C, D, H, and K, and the data are presented as means ±SEMs. ****P < 0.0001, ***P < 0.001, **P < 0.01, n.s., not significant. FACS, fluorescence-activated cell sorting; GLU, glucose; G3P, glyceraldehyde 3-phosphate; GAP/DHAP, glyceraldehyde 3-phosphate/dihydroxyacetone phosphate; 3-PG, 3-phosphoglycerate.

The vigorous activity of mitochondria in differentiating organoids, especially in HCs, as evidenced by accumulated MitoTracker+ and MitoSox+ signals, suggested that OXPHOS might not participate in the organoid expansion stage (Fig. 1G and SI Appendix, Fig. S1A). Seahorse analysis revealed that organoids in EFI-CL group had the highest extracellular acidification rate (ECAR) after glucose administration (Fig. 1H). The glycolytic capacity did not further increase with oligomycin treatment, which blocks the mitochondrial electron transport process, and treatment with 2-deoxy-D-glucose (2-DG), a nonmetabolizable analog of glucose, prominently reduced the ECAR. We conducted mitochondrial stress tests on expanding and differentiating cochlear organoids (SI Appendix, Fig. S1B). During the differentiation phase, the organoids displayed reduced levels of both Superoxide Dismutase and Glutathione Reductase (SI Appendix, Fig. S1C). The results revealed an increase in mitochondrial oxidative respiration during the differentiation of cochlear organoids, suggesting the metabolic transition from expansion to differentiation (Fig. 1I). The increasing MitoTracker+ and MitoSox+ signals during otic differentiation further indicate that HC differentiation favors OXPHOS (SI Appendix, Fig. S1 D and E). We further detected the substrates and intermediates of glycolysis by MS. Organoids in the EFI-CL group exhibited increased ATP/ADP, NADH/NAD, NADPH/NADP, and IMP levels, suggesting that the EFI-CL group exhibited high energy demand and rapid biosynthesis (Fig. 1J and SI Appendix, Fig. S1 F–H). Furthermore, the majority of glycolytic intermediates, such as fructose-6-phosphate (F6P), fructose 1,6-bisphosphate (FBP), PEP, and pyruvate (PYR), and the final anaerobic glycolytic product, lactate (LAC), accumulated in the organoid culture with EFI-CL (Fig. 1K and Dataset S1). These results suggested that glycolytic metabolism might play a critical role in cochlear organoid expansion.

PKM2 Activity Is Required for Cochlear Organoid Expansion.

Glycolysis is performed by a series of enzymes that convert glucose to pyruvate for the TCA cycle or lactate production (Fig. 2A). To conclusively evaluate whether glycolytic metabolism modulates NSEC proliferation, we treated organoid with pyruvate and lactate. We found that both metabolites boosted organoid formation in a dose-dependent manner, as evidenced by increased organoid size and elevated cell viability (Fig. 2 B–D). Next, we found that Pkm expression was most prominently upregulated among the glycolytic enzymes in the organoids under pro-proliferative conditions (Fig. 2E). Subsequently, the two PKM isoforms were both detected in the expanding organoids and cochlear epithelia from postnatal day 2 (P2) mice, and the PKM2 protein was found that abundantly expressed (Fig. 2 F and G). Furthermore, we confirmed that PKM2 expression was significantly elevated in the robust proliferative culture system (SI Appendix, Fig. S2A).

Fig. 2.

PKM2 controls glycolytic flux and organoid formation. (A) Schematics summarizing glucose flux and the catalyzing enzymes. (B) Bright-field images of organoids after lactate and pyruvate administration. (Scale bar, 100 μm.) (C and D) Quantification of organoid diameter (n = 32) and organoid viability (n = 3) under different conditions. (E) RT-PCR showing mRNA of glycolytic enzymes (n = 3). Actb was loading as control.(F and G) Representative Western blots showing PKM1 and PKM2 expression in the cochlear epithelia and cochlear organoids. (H) Generation of Sox9-CreER; PKM2fl/fl; CKO cochlear organoids. (I) Quantification analysis of ECAR and O₂ consumption rate (OCR) of proliferating organoids upon PKM2 CKO. (J) Representative images and quantification of the diameter of PKM2 CKO organoids (n = 14). (Scale bar, 100 μm.) (K) Representative images and quantification of Ki67+ cells in organoids (n = 12). (Scale bar, 100 μm.) One-way ANOVA and Tukey’s multiple comparison test were performed for C, D, and E, and unpaired Student’s t tests (two-tailed) were performed for J, and K. The data are presented as means ±SEMs. ****P < 0.0001, ***P < 0.001, **P < 0.01, n.s., not significant. HK, hexokinase; ATP, adenosine triphosphate; ADP, adenosine diphosphate; GPI, glucose-6-phosphate isomerase; PFK, phosphofructokinase; ALD, aldolase; TPI, triosephosphate isomerase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PGK, phosphoglycerate kinase; PGAM, phosphoglycerate mutase; ENO, enolase; NAD, nicotinamide adenine dinucleotide; NADH, reduced form of nicotinamide-adenine dinucleotide; LDH, lactate dehydrogenase.

To characterize the physiological role of PKM2 during cochlear organoid formation, we generated a PKM2 CKO mouse line by removing exon 9 of Pkm (Fig. 2H). We crossed PKM2 CKO, Sox9-CreER, and tdTomato mice to generate inducible CKO mice deficient in the PKM2 protein in NSECs. After confirming the deletion of PKM2 in the cochlear organoid by Cre-mediated recombination (SI Appendix, Fig. S2B), ECAR analysis showed that PKM2 deletion led to impaired glycolytic activity and increased OXPHOS (Fig. 2I). The proliferative capacity of NSECs in PKM2-deficient organoids significantly decreased (Fig. 2 J and K and SI Appendix, Fig. S2C). Apart from genetic intervention, we manipulated PKM2 metabolic activity using the agonist TEPP-46 or inhibitor compound 3 K, and we confirmed that PKM2 activity was essential for cochlear organoid expansion (SI Appendix, Fig. S2D).

PKM2 Is Essential for Sensory Epithelium Morphogenesis in Mouse Cochlea.

Next, we evaluate the function of PKM2 during cochlear morphogenesis. Abundant PKM2, not PKM1, expression was found at embryonic day (E) 9.5, the vigorous expansion period in the early development of the cochlea (Fig. 3A). Notably, immunofluorescence analysis confirmed that PKM2 expression gradually decreased from E9 to E16.5, a stage that covers the transition from progenitor cell proliferation to HC differentiation (Fig. 3 B and C). As PKM2 CKO under Sox9-Cre ER led to embryonic lethality, we utilized Foxg1-Cre, a more specific Cre mouse line for the cochlea (Fig. 3D). Under Foxg1-Cre, the otic vesicles, marked by E-CADHERIN, were smaller in the PKM2 CKO mice, as well as the reduced number of SOX2+ progenitor cells (Fig. 3E and SI Appendix, Fig. S3A). At P2, PKM2 deletion reduced the size of the whole cochlea, as evidenced by decreased global cochlear height and width, as well as the spiral length of the cochlear epithelia (Fig. 3 F and G). We traced the effects of PKM2 on cochlear development and found that the overall length of cochlear epithelia notably decreased from E14.5 to P2 in mice with PKM2 deficiency (SI Appendix, Fig. S3B).

Fig. 3.

PKM2 regulates the development of the cochlear sensory epithelium. (A) Representative images showing PKM1 and PKM2 protein expression in the developing cochlear epithelia. (B and C) Representative images and quantitative analysis of PKM2 expression at different developmental time points (n = 6). (Scale bar, 50 μm.) (D) Experimental plan for the generation of the Foxg1-Cre; PKM2 ^fl/fl CKO mouse line. (E) Representative images and quantitative analysis of PKM2 CKO otic vesicles at E9.5 (n = 6). (Scale bar, 50 μm.) (F and G) Representative images and quantitative analysis of P2 PKM2 CKO cochlear size and cochlear epithelial length (n = 6). Scale bar, 200 μm (F), 100 μm (G). (H) Representative images and quantitative analysis of neonatal PKM2 CKO HCs (n = 5). (I) Quantitative analysis of IHCs and OHCs count from P2 PKM2 CKO cochlea. (J) Immunostaining of P2 PKM2 CKO cochlea showing the expression of HC markers MYO7A (red) and POU4F3 (gray). (Scale bar, 100 μm.) (K) Immunostaining of P2 PKM2 CKO cochlea showing F-actin and Parvalbumin (PVALB) staining. (Scale bar, 20 μm.) One-way ANOVA and Tukey’s multiple comparison test were performed for C, and unpaired Student’s t test (two-tailed) was used for E. F. G. H. and I. Data are presented as means ±SEMs. ****P < 0.0001, **P < 0.01, *P < 0.05, n.s., not significant.

Our observations in PKM2 CKO mice revealed a marked reduction in the overall HC population at P2 (Fig. 3H), characterized by a reduced count of inner HCs and a more severe in the absence of outer hair cells (OHCs) at the apex of the cochlea (Fig. 3 I and J). Additionally, distorted stereocilia morphology and orientation were found in PKM2 CKO mice (Fig. 3K). Consistent with the phenotypes of in vivo cochlear tissue, we found that the deletion of PKM2 profoundly affected HC generation in cochlear organoids (SI Appendix, Fig. S3C). To investigate which events affected by PKM2 led to developmental disorders in the sensory epithelium, we collected cochlear organoids for RNA-seq (SI Appendix, Fig. S3D). As shown in the differential expression gene map (SI Appendix, Fig. S3E), PKM deficiency caused 2,634 downregulated genes and 2,744 upregulated genes (Dataset S2). GSEA of the downregulated genes revealed enrichment of several GO terms (SI Appendix, Fig. S3F), including “multicellular organism development” and “cell population proliferation”. Additionally, some enriched GO terms were related to inner ear development, such as “sensory system development”, “hair cell differentiation”, and “sensory perception of sound”. Collectively, our findings support that PKM2-driven glycolysis is required for cochlear sensory epithelium morphogenesis.

Lactate Promotes Cochlear Organoid Expansion.

We further assessed glycolysis-derived metabolites to characterize which metabolites drive cochlear morphogenesis. First, we utilized mouse embryos to monitor glucose uptake by incubating them with 2-NBDG. Strikingly, a remarkable uptake signal was detected in the E9.5 otic vesicle and neural plate (Fig. 4A), and gradually decreased from E9.5 to E16.5 (Fig. 4 B and C), which highlighted the participation of glucose metabolism in early otic lineage establishment and propagation. Similarly, 2-NBDG fluorescence enriched in expanding cochlear organoids and faded along differentiation (Fig. 4 D and E). We next assessed glycolytic metabolites in the organoids by targeted MS and found that overall glucose uptake was hindered, while glycolytic intermediates, including F6P, FBP, and PEP, generated by enzymes upstream of PKM2, accumulated in the PKM2 CKO organoids (Fig. 4F, SI Appendix, Fig. S4A and Dataset S3). In addition, we noticed that PKM2 deficiency led to reduced PYR and LAC production. Reduced levels of citrate (CIT), a marker product of the TCA cycle, were also found in the PKM2 CKO organoids (Fig. 4F and SI Appendix, Fig. S4A). To identify the exact metabolic cascade affected by PKM2, we traced glucose flux using ¹³C-labeled glucose uptake followed by LC–MS (Fig. 4G). The intake of M + 6 GLU significantly decreased upon PKM2 deletion (Fig. 4H), but there was no significant change in the M + 6 GLU/GLU ratio (SI Appendix, Fig. S4B), which indicated that the rate of glucose uptake remained unchanged. No significant changes in the production rates of the PKM2 upstream intermediates F6P, FBP, DHAP/GAP, or 3-PG were detected (SI Appendix, Fig. S4 C and D), while the levels of the PKM2 downstream metabolites PYR and LAC were markedly decreased (Fig. 4 I and J). In addition, the production rate of CIT markedly increased (SI Appendix, Fig. S4E) in the PKM2 CKO organoids, which indicates a shift in metabolic programming upon PKM2 deletion.

Fig. 4.

Glycolytic flux tracing revealed that lactate controls inner ear development. (A) Enrichment of glucose uptake was detected in the midbrain and otic vesicle (arrowheads) by 2-NBDG. (Scale bar, 100 μm.) (B and C) Representative images and quantitative analysis of 2-NBDG fluorescence in the cochlear epithelia during development (n = 6). (Scale bar, 50 μm.) (D and E) Representative images and quantitative analysis of 2-NBDG fluorescence in the cochlear organoids (n = 6). (Scale bar, 50 μm.) (F) Heatmap showing an altered abundance of intermediates after PKM2 CKO in organoids (n = 3). (G) Schematic view of the flux of ¹³C-glucose-derived intermediates in glycolysis. (H) Quantification of ¹³C-glucose uptake after PKM2 knockout (n = 3). (I and J) Incorporation of ¹³C-glucose into PYR and LAC (n = 3). (K and L) Representative images and quantification of the diameter of PKM2 CKO cochlear organoids with or without lactate treatment (n = 15). (Scale bar, 100 μm.) (M and N) Flow cytometry analysis of the cell cycle distribution of PKM2 CKO cochlear organoid cells (n = 3). One-way ANOVA and Tukey’s multiple comparison test were performed for C, E, L, M, and N. Unpaired Student’s t test (two-tailed) was used for H–J. The data are presented as means ±SEMs. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, n.s., not significant.

Exogenous ATP intervention did not rescue the restricted organoid growth caused by PKM2 deficiency (SI Appendix, Fig. S4 F–J), leading us to focus on the role of metabolites. Lactate has long been regarded as an end waste product of glycolysis, but recent studies have shown that lactate accumulation can guide a wide variety of biological events (21, 22). To clarify whether lactate is responsible for PKM2-guided NSEC proliferation, we treated organoids with lactate and found that lactate administration ameliorated the restriction of organoid expansion caused by PKM2 deletion in a dose-dependent manner (Fig. 4 K and L). Furthermore, the S/G2/M cell cycle arrest of NSECs caused by PKM2 deletion was reversed by lactate (Fig. 4M and SI Appendix, Fig. S4 K and L). Furthermore, Oxamate and Galloflavin, the specific lactate dehydrogenase inhibitors, treatment led to the suppressed organoid formation, which further highlights that PKM2-generated pyruvate is mainly converted to lactate during NSEC expansion (SI Appendix, Fig. S5 A–E). These findings suggested that glycolysis-derived lactate serves as a functional metabolic product that regulates the development of the cochlear sensory epithelium.

Lactylation of H3K9 Mediates Genome-Wide Transcriptional Modification.

As lactate was found to control angiogenesis and cardiomyocyte growth through NDRG3-ERK (23, 24), we investigate the possibility of NDRG3 and ERK participation in lactate regulate inner ear development. However, low NDRG3 protein was detected, and no substantial NDRG3 or p-ERK expression was altered upon PKM2 deletion or lactate treatment in inner ear organoids (SI Appendix, Fig. S6 A and B). Growing evidence has revealed that lactate can influence gene expression patterns via lysine lactylation (Kla) (25–27). We found that the PKM2-dependent global protein modifications by Kla were mainly located between the 10 and 17 kDa molecular weight regions, where histones are localized (Fig. 5A). Emerging evidence has shown that histone lactylation serves as a bridge between metabolic changes and transcriptional regulation (28). As shown in Fig. 5B, the lactylation of histone H3 at lysine 9 (H3K9la) prominently decreased after PKM2 deletion but was partly rescued by lactate treatment. However, no substantial change was found in the lactylation of histone H3 at lysine 18 or 27 under PKM2 or lactate intervention (Fig. 5B). In addition, among the reported lactylation sites of histone H4, only the lactylation of histone H4 at lysine 8 (H4K8la) exhibited a trend similar to that of H3K9la (Fig. 5C). We next confirmed that lactate treatment increased Pan-Kla levels and H3K9la, H3K27la, H4K8la, and H4K12la levels in the organoids (Fig. 5D and SI Appendix, Fig. S6C). As the acyltransferase p300 has been identified as a lactylation modifier of histone and nonhistone proteins (29, 30), we treated the organoids with A484, an acyltransferase p300 inhibitor. We detected lactylation of H3K9 and H4K8 (Fig. 5D and SI Appendix, Fig. S6C), confirming that PKM2 promoted lactylation via a p300/CBP-dependent mechanism. The detection of developmental otic tissues revealed that H3K9la strongly decreased during development (SI Appendix, Fig. S6D), and we speculated that H3K9la is mainly responsible for lactate-mediated development in the Organ of Corti.

Fig. 5.

PKM2-lactate cascade-driven H3k9la modification. (A) Western blot analysis showing PKM2-dependent overall lactylation levels. (B and C) Screening of PKM2-dependent histone lactylation sites. (D) Western blot analysis showing changes in histone 3 lactylation levels after lactate and p300 inhibition. (E) Experimental workflow for evaluating H3K9la targets. (F) Differential peak distribution of H3K9la target genes after PKM2 knockout. (G) A Venn diagram reveals genes whose expression is dependent on PKM2 and lactate. (H) Differential peaking calling and motif analysis of H3K9la binding genes. (I) RT-PCR analysis showed the relative expression of selected genes. Results were normalized to Actb in the same sample and then normalized to the control group (n = 3). (J and K) Visualization of the alternations in peaks of selected genes. ChIP-qPCR confirmed the promoter sequence expression of the corresponding genes (n = 3). The data are presented as means ± SEMs. Unpaired Student’s t tests (two-tailed) were performed for H. One-way ANOVA was performed for I–M. ***P < 0.001, **P < 0.01, *P < 0.05.

To determine whether gene modifications by H3K9la occur in a PKM2- and lactate-dependent manner, we collected expanding cochlear organoids, PKM2 CKO organoids and PKM2 CKO organoids treated with lactate for CUT&Tag assays to monitor H3K9la genome-wide modifications (Fig. 5E). Among the gene regions, H3K9la was the main genetic modification within the promoter regions and binding patterns were altered by PKM2 CKO and lactate (Fig. 5F). Further, 418 genes binding with H3K9la were reduced by PKM2 deletion and restored by lactate administration, which was functionally enriched in developmental processes, such as brain development and sensory organ morphogenesis (SI Appendix, Fig. S6 E and F). Combined with the restricted transcriptome changes by PKM2 CKO (SI Appendix, Fig. S6G), we overlapped the genes whose mRNA expression was downregulated by PKM2 deletion, H3K9la-binding genes whose expression was reduced by PKM2 deletion, and H3K9la-binding genes whose expression was rescued by lactate administration (Fig. 5G). Among the 80 genes shared by the three sections, de novo motif analysis for peak regions with notable differential peaks of several TFs that contribute to cell fate determination, including Sox transcription factor family members Sox15 and Sox13, as well as Wnt9a, Elf3, and Ebf3 (Fig. 5H, SI Appendix, Fig. S6H and Dataset S4).

In addition, we verified the conspicuously downregulated genes, Sox family, from RNA-seq data and noticed the expression of Sox2, Sox4, Sox9, and Sox21 were significantly downregulated (Fig. 5I). Among these genes, Sox2 and Sox9 peaks in promotor regions binding with H3K9la were diminished upon PKM2 deletion and reversed by lactate, and the variation in promoter was validated by Chromatin Immunoprecipitation followed by qPCR (ChIP-qPCR) (Fig. 5 J and K). Peaks altered at the promoter regions of the shared genes, such as Aif1l and Rflna, were shown in SI Appendix, Fig. S6I. Alternated peaks in exon and noncoding binding regions could also be found in Sox21, Sox15, and Elf3 (SI Appendix, Fig. S6J). Collectively, these results indicated that H3K9la is correlated with the chromatin states of PKM2-regulated genes, which might account for the glycolysis-lactate cascade–triggered morphogenesis of the sensory epithelium.

PKM2 OE and Lactate Promote HC Generation in Human and Mice.

This developmental mechanism inspired us to evaluate the potential of the PKM2-driven glycolytic cascade in promoting HC regeneration. Coupled with LY411575 (LY), a selective small gamma-secretase inhibitor that is capable of promoting SCs transdifferentiation into HCs via inhibiting the Notch signaling (31, 32), the cultured neonatal mouse cochlear explants were treated with AAV particles or lactate (Fig. 6A). The number of newly generated HCs, defined as SOX2+/MYO7A HCs, triggered by a single LY was significantly augmented by PKM2 OE or LAC addition (Fig. 6 A and B). Through EdU tracing, we identified limited mitotic generation of HCs upon the treatments, indicating PKM2 OE/LAC promotes HC generation mainly by transdifferentiation (SI Appendix, Fig. S7 A and B). Besides, both IHCs and OHCs were extra generated by PKM2 OE or LAC (SI Appendix, Fig. S7 C and D). We then collected PKM2-overexpressing cochlear explants after 10 d of culture, and LC–MS analysis confirmed that PKM2 promoted PYR and LAC production (Fig. 6 C and D). The results suggest that metabolic reprogramming by glycolytic activation potentially promotes HC regeneration.

Fig. 6.

PKM2 OE and lactate treatment increase human HC generation. (A) Experimental plan for evaluating PKM2 and lactate in mouse cochlear regeneration. Representative images of regenerated HCs after PKM2 OE and lactate treatment. (Scale bar, 100 μm.) (B) Quantitative analysis of extra generated HCs after PKM2 OE and lactate treatment from (A) (n = 6). (C and D) Heatmap and columns showing the abundance of glycolytic intermediates after PKM2 OE. (E and F) Representative images and quantitative analysis of PKM2 expression in the human auditory epithelia (white box shows the prosensory domain) from 9 PCW-13 PCW abortus (n = 6). (Scale bar, 50 μm.) (G and H) Representative images and quantitative analysis of 2-NBDG fluorescence in the human fetal cochlear organoids (n = 6). (Scale bar, 100 μm.) (I) Schematic diagram showing the workflow for evaluating HC generation in the human cochlear embryonic explants. Representative images of PKM2 OE and lactate-treated human explants with LY. (Scale bar, 100 μm.) (J) Quantitative analysis of MYO7A+/SOX2+ cells in human cochlear explants from (I). One-way ANOVA was performed for B, F, H, and J. Unpaired Student’s t tests (two-tailed) were performed for D. The data are presented as means ± SEMs. ****P < 0.0001, **P < 0.01, *P < 0.05, n.s., not significant.

Moreover, we delivered PKM2 carried by AAV2/DJ into the inner ear of newborn Pou4f3-diphtheria toxin receptor mice via a round window approach (SI Appendix, Fig. S7 E and F) and found an increased number of newly generated HCs, as determined by colabeled MYO7A+, SOX2+, and EdU+ cells, in the combination of PKM2 OE and LY than in single LY treatment upon diphtheria toxin treatment (SI Appendix, Fig. S7 G and H). Using cochlear samples from aborted human embryos, we first assessed the expression pattern of PKM2 during the critical stages of HC generation. As shown in Fig. 6E, POU4F3-labeled HCs were initially detected at 11 wk postconception (PCW), and the protein expression of PKM2 gradually decreased as gestational age increased (Fig. 6F). We also found that the human cochlear prosensory epithelium had a significantly greater glucose uptake capacity at 9 PCW than at the HC generation stage (Fig. 6 G and H). We further constructed human PKM OE viruses based on the AAV2/DJ-CMV-pA backbone and cultured human cochlear explants 12 PCW to assess the HC generation (Fig. 6I). Similar to the results of the experiments in mice, more HCs were generated after PKM2-OE or supplementation with LAC than single LY treatment (Fig. 6J). Thus, an intervention targeting the PKM2-lactate cascade is a promising therapeutic approach for HC regeneration.

In summary, we revealed that PKM2-dependent glycolysis regulates the formation of cochlear sensory epithelium via the lactylation of H3K9la and an altered chromatin state for transcription initiation (SI Appendix, Fig. S8). PKM2-triggered metabolic reprogramming promotes HC generation in the human cochlea, providing insight into regenerative hearing restoration.

Discussion

In this study, our work provided the evidence that glycolytic metabolism modulates cochlear sensory epithelium formation through lactate-dependent epigenetic modification. Specifically, the lactylation of H3K9la led to alteration cochlear transcriptome. No previous study has attempted to intervene in metabolic pathways to regenerate HCs, and we showed that the OE of the key glycolytic enzyme PKM2 or lactate supplementation led to increased HC generation in both mouse and human cochlear epithelium.

PKM2 Is Indispensable for Auditory Sensory Epithelium Formation.

We found that PKM2 is the dominant isoform in proliferating organoids and the cochlear primordium, and gene KO in mice and organoids confirmed its critical role in controlling cochlear epithelium formation. In contrast to those of the constitutively enzymatically active PKM1 isoform, the polymeric state and activity of PKM2 are subject to allosteric regulation by intracellular metabolites, including FBP, serine, and succinyl-aminoimidazolecarboxamide ribose-5’-phosphate (33, 34) and studies have highlighted the critical role of PKM2 in regulating the Warburg effect (35, 36). Consistent with these studies, we showed that PKM2 controls sensory epithelial development mainly by regulating downstream metabolite production.

Other glycolytic enzymes, such as ENO1, have been shown to direct germ layer differentiation from embryonic stem cells (37), and activated PFKFB2 supports to support stem cell differentiation during alveolar regeneration (38). It is worth noting that glycolytic flux not only affects cell proliferation but also HC generation in this study. Our results are similar to PKM2 promoting the proliferation of adult cardiomyocytes (17), PKM2 also promotes inner ear postnatal SCs reentry into the cell cycle, and it needs further study to determine whether glycolytic flux regulates hair cell fate-related genes. The levels of metabolites were not monitored in the study. Generating glycolytic intermediate-specific probe coupled with advancements in imaging techniques to dynamically monitor the level of metabolites (39), such as lactate and pyruvate, which would further facilitate progress on promoting metabolic reprogramming in inner ear development and regenerative therapy. In addition, future studies are needed to further elucidate the interaction between glycolysis and other metabolic pathways and to explore the possibility of applying glycolysis-related regulatory strategies to regenerative medicine and cell therapy.

PKM2-Mediated SC Plasticity Remodeling Enhances HC Generation.

In contrast to previous studies limited to HC regeneration in zebrafish (40), chickens (41), and mice (42), we made a pioneering attempt to overexpress PKM2 to promote HC generation in human tissues. In the cardiovascular system, PKM2 controls the cardiomyocyte regeneration (17) and PKM2 agonist TEPP-46 has shown a significant therapeutic effect on dilated cardiomyopathy (43). Similar therapeutic effects of PKM2 in inducing cell proliferation and expediting tissue recovery have been demonstrated in hair follicles (44), skeletal muscle (45), and liver injury caused by carbon tetrachloride (46). These similar results for PKM2 for tissue regeneration support the importance of focusing on PKM2 as a transformation target for regeneration research.

Inspired by the hearing recovery induced by LY treatment in a noise-damaged mouse model (47), a launched phase I/IIa clinical trial recently announced that LY3056480, another gamma-secretase inhibitor similar to LY, failed to restore hearing in adults with mild-moderate sensorineural hearing loss (48). The Gamma-secretase inhibitors transdifferentiate SCs into HCs via the inhibition of the NOTCH pathway, and the failure of this study may be attributed to insufficient consideration of issues related to SC consumption and plasticity. After a series of rigorous experiments involving an organoid model, mouse cochlear explants, and human cochlear explants, we confirmed that the PKM2-lactate cascade has promising therapeutic value when combined with LY. Considering the shortcomings in current drug delivery systems, in future research, we hope to combine scientific drug delivery systems to solve the problems of drug and metabolite degradation and efficiently cross the blood–labyrinthine barrier, which might improve the uptake of PKM2 agonists or lactate-bound LY by SCs.

PKM2-Dependent Histone Kla and Epigenetic Modification of Critical TFs.

Initially, we hypothesized that NDRG3-ERK is responsible for the lactate-regulated inner ear development, the differences between our data with those in other cell types might be attributed to lactate stabilize NDRG3 and ERK under oxygen-limited conditions (24), further study should be to investigate whether lactate regulates NDRG3 in inner ear systems under different oxygen level. Our data provide insight into the role of H3K9la lactylation in inner ear development and regeneration. Histone methylation and acetylation have been reported to regulate cochlear gene expression during inner ear development and in response to ototoxic insult (49, 50). In contrast to lactylation, H3K9 acetylation (H3K9ac), which is related to an accessible chromatin state, is abundant in the neonatal mouse cochlea and is then gradually converted to H3K9 methylation (H3K9me1), leading to reduced gene transcription (50, 51). Consistent with our results, these studies highlight the crucial role of H3K9 modification in inner ear development. In addition, H3K4me1 levels are correlated with SC regenerative potential, and the inhibition of H3K4me1 promotes HC regeneration via SC transdifferentiation (51, 52). Our results support that manipulating epigenetic modifications might remove repressive epigenetic barriers that silence gene networks and increase the efficiency of HC regeneration.

Furthermore, we found potential genes that promote HC regeneration. CUT&Tag analysis revealed H3K9la binding peaks in the promoter region of a series of genes including the Sox family members Sox2 and Sox9. While TFs including the Sox family members play significant roles, their influence is not confined to the inner ear. Additionally, PKM2 deficiency hinders HC generation; whether PKM2 interacts with other inner ear development genes, such as Cdh7, Tbx2, and Atoh1, needs to be further investigated. Among the novel potential targets, Aif1l was reported to orchestrate actomyosin contractility and the extension of filopodia (53), and the function Rflna remains to be explored. Future studies are needed to clarify the function of the above targets in PKM2-lactate cascade–guided cochlear development and evaluate their potential capacity in HC regeneration.

Materials and Methods

The detailed description of the material and method utilized in this study is provided in SI Appendix. Animals: All experiments were approved by the Institutional Animal Care and Use Committees of Fudan University and strictly followed the national experimental animal standards and welfare requirements. Before abortion, pregnant women were fully informed and signed an informed consent form to conduct the fetal tissue experiments. This study was reviewed and approved by the Ethics Committee of the Eye, Ear, Nose and Throat Hospital Affiliated with Fudan University, China (project number: 2022166). Mouse lines, genetic manipulation, immunohistochemistry, cochlear explant and organoid culture, and sequencing methods in this study are described in SI Appendix. CUT&Tag and RNA-seq data have been uploaded to the NCBI database under Sequence Read Archive (SRA) ID codes: PRJNA1107354 and PRJNA1107371.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

CUT&Tag and RNA-seq data have been deposited in NCBI database Sequence Read Archive (SRA) ID codes: PRJNA1107354 (54) and PRJNA1107371 (55). All other data are included in the manuscript and/or supporting information.