PKM2 aggregation drives metabolism reprograming during aging process

Abstract

While protein aggregation's association with aging and age-related diseases is well-established, the specific proteins involved and whether dissolving them could alleviate aging remain unclear. Our research addresses this gap by uncovering the role of PKM2 aggregates in aging. We find that PKM2 forms aggregates in senescent cells and organs from aged mice, impairing its enzymatic activity and glycolytic flux, thereby driving cells into senescence. Through a rigorous two-step small molecule library screening, we identify two compounds, K35 and its analog K27, capable of dissolving PKM2 aggregates and alleviating senescence. Further experiments show that treatment with K35 and K27 not only alleviate aging-associated signatures but also extend the lifespan of naturally and prematurely aged mice. These findings provide compelling evidence for the involvement of PKM2 aggregates in inducing cellular senescence and aging phenotypes, and suggest that targeting these aggregates could be a promising strategy for anti-aging drug discovery.

Fig. 1. Lysosomes in senescent cells contain increased glycolytic enzymes than in young cells.

a schematic diagram shows the process of isolating lysosomes from young and etoposide-induced senescent cells and the subsequent identification and verification. b GSEA plot of core components in glycolysis pathway according to the fold change scores in (a). c The Glycolysis pathway in which the increased glycolytic enzymes in lysosomes of senescent cells were marked in different levels of red. d Lysosome immunoprecipitation and subsequent immunoblotting analysis in HEK 293T (TMEM192-3HA) cells treated with 2 μM etoposide (ETO) for indicated days. e, f Representative images (e) and quantification (f) of PKM2 and lysosomes (indicated by HA staining) in HEK 293T (TMEM192-3HA) cells treated with 2 μM etoposide (ETO) for indicated days. Scale bar, 5 μm. n = 4 (four randomly captured images), one-way ANOVA was used, P = 0.0107 (DMSO vs. ETO Day 1). *P < 0.05, ****P < 0.0001. Error bars represent SEM. Source data are provided as a Source Data file.

Fig. 2. PKM2 aggregates accumulate in senescent cells.

a Immunofluorescent imaging of PKM2 in HEK 293T, HeLa, MCF-7, 2BS cells treated with 2 μM etoposide (ETO) for three days. Scale bar, 20 μm for 2BS, 10 μm for the others. b Immunofluorescent imaging of PKM2 in fibroblasts 2BS with different passages. Scale bar, 20 μm. c Immunofluorescent imaging of PKM2 in frozen section of liver from young and aged mice. Scale bar, 20 μm. d Confocal microscopy of purified sfcherry-PKM2 in vitro under different protein concentration and salt concentration. Scale bar, 10 μm. e Immunofluorescent imaging of PKM2 and CS (citrate synthase, mitochondria marker) in senescent HEK 293T cells induced by 2 μM etoposide (ETO). Scale bar, 10 μm. f Schematic diagram showing the five domains of PKM2. g Fluorescent imaging of sfcherry-PKM2 in HEK 293T cells transfected with full length or truncated variants of PKM2. Scale bar, 10 μm. h Fluorescent imaging of purified full length or truncated variants of PKM2 in vitro. Scale bar, 10 μm. i Tests of grip strength (left, n = 9 mice) and rotarod time (right, n = 8 mice) were performed in mice overexpressed with wild-type Pkm2 and aggregate-prone Pkm2 mutant (A2 domain deletion) via AAV9 system for 14 weeks. One-way ANOVA was used. P = 0.0225 (Scramble vs. Del A2), P = 0.0379 (WT vs. Del A2). j ELISA measurement of IL6, Tnfα, Il1α concentration in mouse serum of the mice (i). n = 5 mice, one-way ANOVA was used, P = 0.9789 (IL6: Scramble vs. Pkm2 FL), P = 0.0481 (IL6: Scramble vs. Pkm2 Del A2); P = 0.2388 (Tnfα: Scramble vs. Pkm2 FL), P = 0.0053 (Tnfα: Scramble vs. Pkm2 Del A2); P = 0.5077 (Il1α: Scramble vs. Pkm2 FL), P = 0.0080 (Il1α: Scramble vs. Pkm2 Del A2). NS, not significant, *P < 0.05, **P < 0.01, ****P < 0.0001. Error bars represent SEM.

Fig. 3. PKM2 aggregates comprise many glycolytic enzymes.

a Confocal microscopy in HeLa cells stably expressed with sfcherry or sfcherry-PKM2. Scale bar, 10 μm. b HeLa cells stably expressed with vector, sfcherry, sfcherry-PKM2 were collected and lysed followed by FACS to isolate PKM2 aggregates. c The proteins with high abundance in the MS results of PKM2 aggregates isolated by FACS. d Schematic view of the PKM2 protein-protein interaction network collected from STRING. Proteins were colored according to the phase separation propensity predicted by PhaSePred. The line between two proteins indicated their interaction gathered from high or low throughput. e Immunoblotting of PKM2 and proteins with high abundance in the MS results of PKM2 aggregates. f, g Immunofluorescent imaging of PKM2 and GAPDH (f) or ENO1 (g) in senescent HEK 293T cells induced by 2 μM etoposide (ETO). Scale bar, 10 μm. h Immunofluorescent imaging of PKM2 and GAPDH in scramble or PKM2 knockdown HEK 293T cells treated with 2 μM etoposide (ETO). Scale bar, 10 μm. All the above experiments were repeated thrice on separate days with similar results. Source data are provided as a Source Data file.

Fig. 4. PKM2 enzymatic activity and glycolytic flux are impaired in senescent cells.

a, b Immunoblotting of p21, PKM2 (a) and measurement of PKM2 enzymatic activity via LDH-coupled kinetic assay (b) in HeLa cells treated with 2 μM etoposide (ETO) for three days. n = 3 (technical triplicates), Two-tailed unpaired. c, d Immunoblotting of p21, PKM2 (c) and measurement of PKM2 enzymatic activity via LDH-coupled kinetic assay (d) in HeLa cells treated with 10 μM ETO for 24 h and release for indicated days. n = 3 (technical triplicates), one-way ANOVA. e Measurement of PKM2 enzymatic activity with a commercial kit (K709-100) in HeLa cells treated with 2 μM ETO for three days. n = 3 (biological replicates), Two-tailed unpaired t-test. f–h Immunoblotting of p53, p21 and PKM2 (g) and measurement of PKM2 enzymatic activity via LDH-coupled kinetic assay (f) or using a commercial kit (K709-100) (h) in young, middle and senescent fibroblasts 2BS. n = 3 (f, technical triplicates). n = 3 (h, biological replicates), one-way ANOVA. i Quantification of lactate in the culture medium of HeLa cells treated with 10 μM ETO and released for three days. n = 3 (biological replicates), Two-tailed unpaired t-test. j Quantification of lactate in the culture medium of HeLa cells treated with 100 μM 5-Fluorouracil (5-FU) and released for indicated days. n = 3 (biological replicates), one-way ANOVA, P = 0.0023 (DMSO vs. 5-FU R2). k Quantification of lactate in the culture medium HeLa cells treated with 50 nM Camptothecin (CPT) and released for indicated days. n = 3 (biological replicates), one-way ANOVA, P = 0.0011 (DMSO vs. CPT R2), P = 0.0001 (DMSO vs. CPT R3). l Quantification of lactate in the culture medium of HEK 293T cells treated with 10 μM ETO and released for three days. n = 3 (biological replicates), Two-tailed unpaired t-test, P = 0.0008. m Quantification of lactate in culture medium of young, middle and old fibroblasts 2BS. n = 3 (biological replicates), one-way ANOVA, P = 0.0035 (Young vs. Middle), P = 0.0003 (Middle vs. Senescent). **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars represent SEM. Source data are provided as a Source Data file.

Fig. 5. Dispersing PKM2 aggregates impedes cellular senescence.

a Model of screening for compounds dispersing PKM2 aggregates. b Immunofluorescent imaging of PKM2 aggregates in HeLa cells stably expressed sfcherry-PKM2 and treated with 2 μM etoposide (ETO) together with DMSO/K35/K27 for two days. Scale bar, 5 μm. c, d Representative images (c) and quantification (d) of SA-β-gal staining in HeLa or HEK 293T cells treated with 10 μM etoposide for 24 h and released for three days together with DMSO or K35 (50 μM). Scale bar, 100 μm. n = 5 (randomly captured images), one-way ANOVA. e Immunoblotting in HEK 293T cells treated with 2 μM ETO for three days together with concentration gradient of K35. f qPCR analysis in HEK 293T cells treated with 10 μM ETO for 24 h and released for indicated days together with DMSO or K35 (50 μM). n = 3 (biological replicates), Two-tailed unpaired t-test, P = 0.0012 (p21: Day 2), P = 0.0006 (p21: Day 3); P = 0.0055 (IL1A: Day 2), P = 0.0020 (IL1A: Day 3); P = 0.0004 (IL6: Day 2), P = 0.0001 (IL6: Day 3); P = 0.0295 (IL1B: Day 2), P = 0.0169 (IL1B: Day 3). g Immunoblotting in HEK 293T cells treated with 10 μM ETO for 24 h and released for three days together with DMSO, K35 or K35 analogs. h, i Representative images (h) and quantification (i) of SA-β-gal staining in fibroblasts 2BS (P25) exposed to 20 μM ETO for 24 h and released for three days together with DMSO, 50 μM K35/K27. Scale bar, 100 μm. n = 6 (randomly captured images), one-way ANOVA. j Immunoblotting in fibroblasts 2BS constantly treated with DMSO, 25 μM K35/K27. k, l Representative images (k) and quantification (l) of SA-β-gal staining in fibroblasts 2BS cultured with DMSO, 25 μM K35/K27 constantly. Scale bar, 100 μm. n = 4 (randomly captured images), one-way ANOVA. m, n Representative confocal images (m) and quantification (n) of EdU staining proliferation assay in HeLa cells treated with 10 μM ETO for 24 h and released for three days together with DMSO, 50 μM K35/K27. Scale bar, 50 μm. n = 5 (randomly captured images), one-way ANOVA. o Quantification of lactate in the culture medium of HeLa cells exposed to 10 μM ETO for 24 h and released for three days together with DMSO, DMSO, 50 μM K35/K27. n = 3 (biological replicates), one-way ANOVA. p Quantification of lactate in the culture medium of young or senescent fibroblasts 2BS exposed to DMSO, 25 μM K35/K27. n = 3 (biological replicates), one-way ANOVA, P = 0.0016 (Young: DMSO vs. K35), P = 0.0001 (Young: DMSO vs. K27); P = 0.0050 (Senescent: DMSO vs. K35). q Measurement of PKM2 enzymatic activity via LDH-coupled kinetic assay in young or senescent fibroblasts 2BS exposed to DMSO, 25 μM K35/K27. n = 3 (technical triplicates), one-way ANOVA, P = 0.0002 (Senescent vs. Senescent + K35). r Measurement of PKM2 enzymatic activity with a commercial kit (K709-100) in young or senescent fibroblasts 2BS exposed to DMSO, 25 μM K35/K27. n = 3 (biological replicates), one-way ANOVA, P = 0.0090 (Young: DMSO vs. K35), P = 0.0007 (Young: DMSO vs. K27); P = 0.0002 (Senescent: DMSO vs. K35). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars represent SEM. Source data are provided as a Source Data file.

Fig. 6. Compounds K35 and K27 delay aging of naturally and prematurely aged mice.

a, b Body weight (a) and lifespan (b) of C57BL/6 J mice injected intraperitoneally with doxorubicin (10 mg/kg) twice at day 0 and day 7 and intragastrically administrated with vehicle or 50 mg/kg K35/K27 every two days from day 2. n = 7 mice for K35 groups, n = 8 mice for the others, one-way ANOVA (a), P = 0.0122 (DOX vs. DOX + K35), P = 0.0179 (DOX vs. DOX + K27); Log-rank (Mantel-Cox) (b), P = 0.0001 (DOX vs. DOX + K35), P = 0.0002 (DOX vs. DOX + K27). c Grip strength (left) and rotarod time (right) of the mice (a) were tested at day 20. n = 12 mice (left), n = 8 mice (right), one-way ANOVA, P = 0.0064 (Rotarod: DOX vs. DOX + K35), P = 0.0361 (Rotarod: DOX vs. DOX + K27). d Examinations of ALT (left) and AST (right) in serum of mice (a) at day 20. n = 12 mice, one-way ANOVA. e–g Immunoblotting (e) and qPCR analysis (f, g) at day 20. n = 6 mice, one-way ANOVA, P = 0.0008 (Cxcl2: DOX vs. DOX + K35), P = 0.0004 (Cxcl2: DOX vs. DOX + K27); P = 0.0004 (IL6: DOX vs. DOX + K35), P = 0.0007 (IL6: DOX vs. DOX + K27); P = 0.0026 (Tnfα: DOX vs. DOX + K35), P = 0.0007 (Tnfα: DOX vs. DOX + K27). h, i Representative images (h) and quantification(i) of SA-β-gal staining in frozen liver tissue sections from mice (a) at day 20. Scale bar, 100 μm for 20× magnification images. n = 8 mice, one-way ANOVA. j Lifespan of 18-month-old C57BL/6 J intragastrically administrated with vehicle or 50 mg/kg K35/K27 every five days. n = 14 mice per group, log-rank (Mantel-Cox), P = 0.0153 (Vehicle vs. K35), P = 0.0169 (Vehicle vs. K27). k Tests of grip strength (left, n = 15 mice) and rotarod time (right, n = 13 mice) of mice (j) after four months. One-way ANOVA, P = 0.0035 (Rotarod: Vehicle vs. K35), P = 0.0022 (Rotarod: Vehicle vs. K27). l, m Representative images (l) and quantification (m) of SA-β-gal staining in frozen lung sections from mice (j) after six months. Scale bar, 100 μm for 20× magnification images. n = 6 mice, one-way ANOVA. n Immunoblotting in lung tissue from mice (j). o qPCR analysis in liver tissue of mice (j). n = 4 mice, one-way ANOVA, P = 0.0107 (p21: Vehicle vs. K35), P = 0.0215 (p21: Vehicle vs. K27); P = 0.0073 (IL6: Vehicle vs. K35), P = 0.0172 (IL6: Vehicle vs. K27). p ELISA analysis in serum of mice (j). n = 5 mice, one-way ANOVA, P = 0.0360 (IL6: Vehicle vs. K35), P = 0.0043 (IL6: Vehicle vs. K27); P = 0.0257 (TNFα: Vehicle vs. K35), P = 0.0185 (TNFα: Vehicle vs. K27); P = 0.0176 (IL1α: Vehicle vs. K35), P = 0.0042 (IL1α: Vehicle vs. K27). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars represent SEM. Source data are provided as a Source Data file.