Lipoylation inhibition enhances radiation control of lung cancer by suppressing homologous recombination DNA damage repair

Abstract

Lung cancer exhibits altered metabolism, influencing its response to radiation. To investigate the metabolic regulation of radiation response, we conducted a comprehensive, metabolic-wide CRISPR-Cas9 loss-of-function screen using radiation as selection pressure in human non-small cell lung cancer. Lipoylation emerged as a key metabolic target for radiosensitization, with lipoyltransferase 1 (LIPT1) identified as a top hit. LIPT1 covalently conjugates mitochondrial 2-ketoacid dehydrogenases with lipoic acid, facilitating enzymatic functions involved in the tricarboxylic acid cycle. Inhibiting lipoylation, either through genetic LIPT1 knockout or a lipoylation inhibitor (CPI-613), enhanced tumor control by radiation. Mechanistically, lipoylation inhibition increased 2-hydroxyglutarate, leading to H3K9 trimethylation, disrupting TIP60 recruitment and ataxia telangiectasia mutated (ATM)-mediated DNA damage repair signaling, impairing homologous recombination repair. In summary, our findings reveal a critical role of LIPT1 in regulating DNA damage and chromosome stability and may suggest a means to enhance therapeutic outcomes with DNA-damaging agents.

Fig. 1.. Metabolic CRISPR screen identifies lipoylation as metabolic vulnerability for radiation response in NSCLC.

(A) Schematic illustration of CRISPR-Cas9 screen workflow. (B) Top 10 negatively selected genes ranked by robust rank aggregation (RRA) score in 10 Gy–irradiated group compared to the control. (C) Manhattan plot of the entire 2981 metabolic genes by log₁₀ P value. The top 10 negatively selected genes are highlighted. (D) Schematic illustration of the function of LIPT1, LIAS, DLD, and PDHX, the four top hits involved in lipoylation. (E) Pathway analysis of the 10 most significantly depleted metabolic pathways by Gene Ontology classification in 10 Gy–irradiated cells compared to the nonirradiated cells. (F) Immunoblots of total and lipoylated PDH and α-KGDH subunits (DLAT and DLST, respectively). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), DLAT, and DLST were blotted as loading controls. (G) Clonogenic assay of indicated cell lines after 2, 4, 6, and 8 Gy IR. The surviving fraction was normalized to the corresponding sham control, and survival curves were fitted using the linear-quadratic model.

Fig. 2.. LIPT1 regulates DNA damage repair and chromosome stability.

(A) Representative images and quantification of γH2AX foci by immunofluorescence staining in nonirradiated cells and at 0.5 and 6 hours (h) after 2-Gy IR in WT, LIPT1^−/− H460, and LIPT1^−/− H460 stably expressing Myc-LIPT1 cells. Nuclei were stained with Hoechst 33342. Scale bar, 10 μm. (B) Representative images and quantification of tail moment at specified time points after 10 Gy by neutral comet assay for the evaluation of DSBs in WT, LIPT1^−/− H460, and LIPT1^−/− H460 stably expressing Myc-LIPT1 cells. Scale bar, 50 μm. (C) Representative images and quantification of tail moment at specified time points after 10Gy by neutral comet assay for the evaluation of DSBs in WT and LIPT1^−/− H157 cells. Scale bar, 50 μm. (D) Representative images of metaphases and quantification of chromosome aberrations on mitotic chromosomes by chromosome spread assay at 24 hours post 4-Gy IR in WT, LIPT1^−/− H460, and LIPT1^−/− H460 stably expressing Myc-LIPT1 cells. Scale bar, 5 μm. (E) Representative images of metaphases and quantification of chromosome aberrations on mitotic chromosomes by chromosome spread assay at 24 hours post 4-Gy IR in WT and LIPT1^−/− H157 cells. Scale bar, 5 μm. For (A) to (E), imaging and quantification were performed on 60 to 100 cells per treatment. Two-way analysis of variance (ANOVA) was used for the statistical analyses. ****P < 0.0001. ns, not significant.

Fig. 3.. Metabolomic profiling of LIPT1^−/− H460 cells reveals decreased abundance of TCA intermediates and increased 2HG.

(A) Principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA) of metabolomic profiles in WT and LIPT1^−/− H460 cells. (B) Metabolite set enrichment analysis comparing WT and LIPT1^−/− H460 cells. (C) Heatmap analysis of top 50 differential metabolites WT and LIPT1^−/− H460 cells. (D) Relative abundance of the indicated TCA cycle metabolites in WT and LIPT1^−/− H460 cells. (E) Relative abundance of 2-hydroxyglutarate (2HG) in WT and LIPT1^−/− H460 cells. (F) Impact of LIPT1 deficiency on enzymes and metabolites related to the TCA cycle. For (D) to (E), error bars represent SD. Unpaired t tests were used for the statistical analyses. ****P < 0.0001, **P < 0.01.

Fig. 4.. LIPT1^−/− cells exhibit elevated baseline H3K9me3 and impaired TIP60 recruitment to DNA damage sites.

(A) Representative images and quantification of H3K9me3 by immunofluorescence staining in WT, LIPT1^−/− H460, and LIPT1^−/− H460 cells stably expressing Myc-LIPT1 cells. (B) Representative images and quantification of H3K9me3 by immunofluorescence staining in WT and LIPT1^−/− H157 cells. (C and D) Representative images and quantification of in situ proximity ligation assay (PLA, green dots) of interactions between histone H3 and TIP60 interaction (C) and between γH2AX and TIP60 (D) in WT and LIPT1^−/− H460 cells, with or without 10-Gy IR. (E and F) Representative images and quantification of in situ PLA (green dots) of interactions between histone H3 and TIP60 interaction (E) and between H2AX and TIP60 (F) in WT and LIPT1^−/− H157 cells with or without 10-Gy IR. For (A) to (F), nuclei were stained with Hoechst 33342. Scale bar, 10 μm. Imaging and quantification were performed on >100 cells per treatment. One-way ANOVA was used for the statistical analyses for (A), unpaired t test was used for (B), and two-way ANOVA was used for (C) to (F). ****P < 0.0001.

Fig. 5.. LIPT1^−/− cells have impaired activation of ATM and its downstream DNA damage repair signaling cascade.

(A) Representative images and quantification of in situ PLA (green dots) of ATM and TIP60 interaction in nonirradiated control and 1 hour after 4 Gy in WT, LIPT1^−/− H460, and LIPT1^−/− H460 reconstituted Myc-LIPT1 cells. Nuclei were stained with Hoechst 33342. Scale bar, 10 μm. (B) Quantification of in situ PLA (green dots) of ATM and TIP60 interaction in nonirradiated control and 1 hour after 4 Gy in WT and LIPT1^−/− H157 cells. (C and D) Immunoblotting analysis of ATM-pS1981, Ku70, γH2AX, and histone 3 in the soluble nuclear and chromatin fractions of WT and LIPT1^−/− in H460 (C) and in H157 (D) cells, with or without IR (0.5 hours post 10 Gy). Histone H3 and γH2AX served as chromatin markers. LE, long exposure; SE, short exposure. (E and F) Immunofluorescence images and quantification of colocalized ATM-pS1981 (red) and γH2AX (green) foci at 0.5 hours post-4 Gy in WT, LIPT1^−/− H460, and Myc-LIPT1–reconstituted H460 cells (E), as well as WT and LIPT1^−/− H157 cells (F). Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Scale bar, 10 μm. (G and H) Immunoblot analysis of ATM-pS1981, total ATM, Chk2-pT68, total Chk2, lipoyl-DLAT/DLST, and γH2AX in H460 (G) and H157 (H) cells at 0.5 hour post-10Gy. (A), (B), (E), and (F), quantification was performed on >100 cells per treatment. Two-way ANOVA was used for (A) and (B), one-way ANOVA was used for (E), and unpaired t test was used for (F). ****P < 0.0001.

Fig. 6.. LIPT1^−/− cells are functionally deficient in HR repair.

(A) Recruitment of EXO1–yellow fluorescent protein to laser-generated DSBs in single living WT and LIPT1^−/− H157 cells. Relative fluorescence intensity of the local accumulation of EXO1 at laser-induced DNA damage sites was quantified for over 10 min and displayed as mean ± SD in WT and LIPT1^−/− H157 cells. Simple linear regression was used for the statistical analyses. ****P < 0.0001. Scale bar, 10 μm. (B) Representative images and quantification of RAD51 foci by immunofluorescence staining in nonirradiated control and at 6 hours after 4 Gy in WT, LIPT1^−/− H460, and LIPT1^−/− H460 stably expressing Myc-LIPT1 cells. Nuclei were stained with DAPI. Scale bar, 10 μm. Imaging and quantification were performed on >100 cells per treatment. Two-way ANOVA was used for the statistical analyses. ****P < 0.0001. (C) Representative immunoblots demonstrate knockout of LIPT1 in U2OS and HEK-293 DR-GFP HR cells using the CRISPR-Cas9 gene editing system. Lipoylation was detected on DLAT and DLST, with GAPDH serving as the internal control. (D) Schematic illustration of the DR-GFP HR reporter assay. (E) HR activity assay by quantifying the population of cells positive for both DsRed and GFP in WT and LIPT1^−/− U2OS (top) and WT and LIPT1^−/− HEK-293 DR-GFP HR cells (bottom). Data were represented as means ± SD, unpaired t tests were used for the statistical analyses. **P < 0.01, ****P < 0.0001.

Fig. 7.. Inhibition of KDM4B disrupts the H3K9me3-TIP60-ATM signaling axis and HR repair.

(A) Representative immunoblots validating siRNA suppression of KDM4B in WT and LIPT1^−/− H460 cells, with GAPDH used as a loading control. (B) Representative images and quantification of H3K9me3 by immunofluorescence staining in Ctrl and siKDM4B WT and LIPT1^−/− H460 cells. (C and D) Representative images and quantification of in situ PLA (green dots) of interactions between γH2AX and TIP60 interaction (C), and between ATM and TIP60 (D) in Ctrl and siKDM4B WT and LIPT1^−/− H460 cells, with or without 10- and 4-Gy IR. (E) Representative images and quantification of γH2AX foci by immunofluorescence staining in nonirradiated cells and at 6 and 24 hours after 4-Gy IR in Ctrl and siKDM4B WT and LIPT1^−/− H460 cells. (F) Representative images and quantification of RAD51 foci by immunofluorescence staining in nonirradiated cells and 6 hours after 4-Gy IR in Ctrl and siKDM4B WT and LIPT1^−/− H460 cells. For (B) to (F), nuclei were stained with Hoechst 33342. Scale bar, 10 μm. Imaging and quantification were performed on >100 cells per treatment. Two-way ANOVA was used for the statistical analyses. *P < 0.05, **P < 0.01, ****P < 0.0001.

Fig. 8.. L2HGDH overexpression normalizes H3K9me3 and reverses DNA damage phenotypes in LIPT1^−/− H460 cells.

(A) Representative immunoblots of Myc-L2HGDH overexpression in WT and LIPT1^−/− H460 cells, with GAPDH used as a loading control. (B) Relative abundance of L-2-hydroxyglutarate (2HG) in nontransfected control cells and flow sorted GPF-L2HGDH–positive LIPT1^−/− H460 cells. Data were represented as mean ± SD, and unpaired t tests were used for the statistical analyses. **P < 0.01. (C) Representative images and quantification of H3K9me3 by immunofluorescence staining in nontransfected control and Myc-L2HGDH–positive WT and LIPT1^−/− H460 cells. (D) Representative images and quantification of γH2AX foci by immunofluorescence staining in nonirradiated cells and at 6 and 24 hours after 4-Gy IR in Ctrl and Myc-L2HGDH–positive WT and LIPT1^−/− H460 cells. (E) Representative images and quantification of RAD51 foci by immunofluorescence staining in nonirradiated control and at 6 hours after 4 Gy in Ctrl and Myc-L2HGDH–positive WT and LIPT1^−/− H460 cells. For (C) to (E), nuclei were stained with Hoechst 33342. Scale bar, 10 μm. Imaging and quantification were performed on >80 to 100 cells per treatment. Two-way ANOVA was used for the statistical analyses. *P < 0.05, ****P < 0.0001.

Fig. 9.. α-KG supplementation restores DNA damage repair and enhances survival of LIPT1^−/− H460 cells after radiation.

(A) Representative images and quantification of in situ PLA (green dots) of interactions between γH2AX and TIP60 in WT and LIPT1^−/− H460 cells with or without 1 mM dimethyl–α-KG and 10-Gy IR. Nuclei were stained with Hoechst 33342. Scale bar, 10 μm. Imaging and quantification were performed on >100 cells per treatment. Two-way ANOVA was used for the statistical analyses. ****P < 0.0001. (B) Immunoblotting analysis of ATM-pS1981 and histone H3 in the chromatin fractions of WT and LIPT1^−/− in H460 cells with or without 1 mM dimethyl–α-KG, 1 mM α-KG, and 10-Gy IR. Histone H3 was used as marker and internal control for the chromatin fraction. Protein levels of ATM-pS1981 were normalized with internal control histone H3. (C) Clonogenic assay of WT and LIPT1^−/− H460 cells with or without 1 mM dimethyl–α-KG after 2, 4, and 6 Gy. The surviving fraction was normalized to the corresponding sham control and survival curves were fitted using the linear-quadratic model. (D) Schematic illustrating how LIPT1 deficiency impairs the IR-induced, TIP60-ATM–mediated HR damage repair pathway due to a deficiency in α-KG–dependent demethylation.

Fig. 10.. Inhibition of lipoylation enhances NSCLC’s radiation response in vivo.

(A) Schematic illustrating timeline of WT and LIPT1^−/− H460 tumor inoculation and treatments. (B) Tumor growth rate of WT and LIPT1^−/− H460 xenografts in athymic nude mice with or without 10 Gy IR. n = 8 to 10 tumors. (C and D) Representative images (C) and quantification (D) of immunoblotting analysis of lipoylated-DLAT and DLST using pooled tumors from H460 tumor–bearing athymic nude mice that either received vehicle or CPI-613 (20 mg/kg) for 14 doses. Intensity was quantified by ImageJ and normalized by GAPDH, n = 4. Data were represented as means ± SD, and two-way ANOVA was used for the statistical analyses. *P < 0.05, ****P < 0.0001. (E) Schematic illustrating timeline of tumor inoculation and treatments. Mice received daily treatment with either CPI-613 (20 mg/kg) or vehicle daily for 14 days. In the irradiated group, 10-Gy IR was administered to tumors after the first two doses of CPI-613. (F) Tumor growth rate of WT H460 xenografts in athymic nude mice treated with vehicle, CPI-613, 10-Gy, or combined treatment. n = 7 to 10 tumors. (G and H) Representative images (G) and quantification (H) of Ki67 and γH2AX-positive nuclei by immunohistochemistry staining. Scale bar, 100 μm. Data were represented as means ± SD, and two-way ANOVA was used for the statistical analyses. *P < 0.05, ****P < 0.0001. (I) Schematic illustrating timeline of KP9-3 tumor inoculation and treatments. C57B/L6 mice received daily treatment with either CPI-613 (20 mg/kg) or vehicle every other day for a total of eight doses. In irradiated group, 10-Gy IR was administered to tumor between the first two doses of CPI-613. (J) Tumor growth rate of KP9-3 xenografts in C57B/L6 mice treated with vehicle, CPI-613, 10-Gy, or combined treatment. n = 8 to 12 mice. For (A), (C), and (G), error bars represent the SEM.