Non-canonical dihydrolipoyl transacetylase promotes chemotherapy resistance via mitochondrial tetrahydrofolate signaling

Abstract

Chemotherapy is often a primary treatment for cancer. However, resistance leads to therapeutic failure. Acetylation dynamics play important regulatory roles in cancer cells, but the mechanisms by which acetylation mediates therapy resistance remain poorly understood. Here, using acetylome-focused RNA interference (RNAi) screening, we find that acetylation induced by mitochondrial dihydrolipoyl transacetylase (DLAT), independent of the pyruvate dehydrogenase complex, is pivotal in promoting resistance to chemotherapeutics, such as cisplatin. Mechanistically, DLAT acetylates methylenetetrahydrofolate dehydrogenase 2 (MTHFD2) at lysine 44 and promotes 10-formyl-tetrahydrofolate (10-formyl-THF) and consequent mitochondrially encoded cytochrome c oxidase II (MT-CO2) induction. DLAT signaling is elevated in cancer patients refractory to chemotherapy or chemoimmunotherapy. A decoy peptide DMp39, designed to target DLAT signaling, effectively sensitizes cancer cells to cisplatin in patient-derived xenograft models. Collectively, our study reveals the crucial role of DLAT in shaping chemotherapy resistance, which involves an interplay between acetylation signaling and metabolic reprogramming, and offers a unique decoy peptide technology to overcome chemotherapy resistance.

Fig. 1. Customized acetylation-focused RNAi screen identifies DLAT as a critical driver of cisplatin resistance in cancer.

a An initial screening examining the effect of loss of 50 acetyltransferases and 24 deacetylases on cisplatin response. Each gene was targeted by pooled shRNA virus infection. KB-3-1^cisR cells transduced with lentiviral shRNA were treated with a sublethal dose of cisplatin (5 μg/ml). Candidates exhibiting low cell viability upon knockdown (<85%; gray) were excluded. b Secondary screening utilized the top 11 genes identified from the primary screen in A549^cisR, KB-3-1^cisR, and H1299. Two candidates that induced over 25% cell death upon gene knockdown and cisplatin treatment are highlighted in purple. c Overall survival (OS) ratios for high versus low expression of the top 2 genes, DLAT and KAT2A, in 10 cancer types were determined using TCGA-based Kaplan-Meier Plotter. Pan-cancer RNA-sequencing data from 3565 patients was stratified by medians. d Effect of DLAT knockdown on cisplatin response was determined by cisplatin IC₅₀. Cells were treated with sublethal doses of cisplatin (A549^cisR 2 μg/ml, KB-3-1^cisR 5 μg/ml, PCI-37B 1 μg/ml, H1299 5 μg/ml) for 48 h. e Colony formation potential of cancer cells with cisplatin treatment and DLAT knockdown. f Cisplatin-induced apoptotic cell death (upper) and cell viability (lower) in cancer cells without or with DLAT knockdown. g-j, DLAT knockdown effect on cisplatin-resistant tumor growth. Mice xenografted with KB-3-1^cisR were treated with vehicle control or cisplatin (5 mg/kg) twice a week. Tumor size (g), tumor weight (h), and the expression level of DLAT in tumors (i) are shown. Tumor proliferation is assessed by Ki-67 IHC, and representative images of the staining are shown for each group (j). Scale bars shown in (g) and (j) represent 10 mm and 50 μm, respectively. Data are mean ± SD from 3 technical replicates for (a) and 3 independent biological replicates for (b, d–f). n = 8 per group for (g) and (h), and the error bars indicate SEM for (g) and SD for (h). P values were determined by paired two-tailed Student’s t-test for (c), two-way ANOVA (g), and one-way ANOVA for the rest. Source data are provided as a Source Data file.

Fig. 2. PDC-independent DLAT activity is needed for cisplatin-resistant cancer growth.

a Cisplatin response of cancer cells with endogenous DLAT knockdown and rescue expression of human DLAT wildtype (WT) or enzyme-inactive mutant (ΔB; catalytic domain B deleted). DLAT knockdown A549^cisR and KB-3-1^cisR cells expressing DLAT variants were under cisplatin-treated conditions (A549^cisR 2 μg/ml and KB-3-1^cisR 5 μg/ml) for 48 h. Apoptosis and cell viability were assessed by annexin V staining and ATP measurement, respectively. Total DLAT levels in cells with DLAT variants are shown by immunoblotting. b, c DLAT knockdown effect on PDC activity. A549^cisR and KB-3-1^cisR cells with DLAT knockdown were treated with cisplatin for 24 h. PDC activity was determined by monitoring NADH production from PDH reaction (b) and phosphorylation of PDHA1 at S293 (c). d DLAT knockdown effect on PDC assembly. Interaction between E1 (PDH), E3 (DLD), and E2 (DLAT) was determined by E1 co-immunoprecipitation in cells with or without DLAT knockdown and cisplatin treatment. e Effect of PDC inhibition on cisplatin sensitivity. Cells were treated with 5 mM PDK inhibitor dichloroacetate (DCA) and cisplatin for 48 h. Apoptotic cell death and cell viability were measured as described in (a). f Overexpression of ACSS1 in cells lacking DLAT. Acetyl-CoA levels were measured by LC-MS. Apoptotic cell death and cell viability were measured as described in (a). Effect of PDH (g) or DLD (h) knockdown on cisplatin-induced apoptotic cell death and cell viability. Data are mean ± SD from 4 independent biological replicates for apoptosis rates of (a) and cell viability of (g, h left) and 3 independent biological replicates for (b, e, f), cell viability of (a, h right), apoptosis rates of (g, h). P values were determined by one-way ANOVA for all panels. Source data are provided as a Source Data file.

Fig. 3. DLAT contributes to chemotherapy resistance by controlling mitochondrial ROS.

a Effect of targeting DLAT on cellular uptake of cisplatin and cisplatin-mediated DNA damage. Cells were treated with sublethal doses of cisplatin for 24 h, followed by staining with fluorescently labeled antibodies against cisplatin-DNA adducts, phospho-Histone H2A.X, and phospho-53BP1.Effect of DLAT loss and cisplatin treatment on bioenergetics, biosynthesis, and ROS levels. Cells were treated with cisplatin, and ATP, RNA, protein synthesis levels (b) and cellular ROS levels (c) were measured by luminescent assay, metabolic labeling, and DCFDA staining, respectively. d Cisplatin-treated cells with DLAT knockdown were treated with antioxidant NAC (0.5 mM). Cellular ROS (upper), apoptosis (middle), and cell viability (lower) were measured. Effect of mitochondria-targeted antioxidant mito-TEMPO (e) or catalase overexpression (f) on ROS levels, apoptosis, and cell viability in cells with DLAT knockdown and cisplatin. Mito-TEMPO (10 μM) or flag-tagged catalase was introduced in cisplatin-treated cells. Mitochondrial ROS and cytoplasmic hydrogen peroxide were measured by mitoSOX staining and luminescence detection, respectively. g Immunoblotting of apoptosis-associated proteins. Cells with or without DLAT knockdown were treated with sublethal doses of cisplatin for 24 h. h Effect of DLAT loss and NAC treatment on Bcl-xL expression. i Scatter plot of drug sensitivity in KB-3-1^cisR and A549^cisR representing ROS level, apoptotic rate, or growth inhibition vs -log₁₀(P value). Cells were treated with sublethal doses of drugs (2 μg/ml cisplatin, 50 μM carboplatin, 10 nM gemcitabine, 10 μM etoposide, 0.5 μM pemetrexed, 3 nM paclitaxel, 1 μM erlotinib) for 48 h followed by DCFDA staining (left), annexin V staining (middle), and CellTiter-Glo viability assay (right). Fold changes were obtained by comparing the DLAT knockdown group to the control group. Data are mean ± SD from 3 independent biological replicates for (a–f, i). P values were determined by two-tailed Student’s t-test (a, i) and one-way ANOVA for the rest. Source data are provided as a Source Data file.

Fig. 4. DLAT binds to MTHFD2 in the mitochondria.

a Top 10 mitochondrial interactors of DLAT beyond PDC in A549^cisR cells with or without DLAT knockdown. DLAT binding proteins were co-immunoprecipitated using DLAT antibody, and peptides were measured by LC-MS/MS. The normalized peak intensity value for each protein is shown. b Co-immunoprecipitation showing candidate interactions with DLAT. c Representative MS spectrum of MTHFD2 peptide fragments is shown. d Cisplatin-induced apoptotic cell death and cell viability in cancer cells without or with knockdown of MTHFD2. Stable knockdown cells were treated with sublethal doses of cisplatin (A549^cisR 2 μg/ml and KB-3-1^cisR 5 μg/ml) for 48 h. Effect of MTHFD2 inhibitors on cisplatin-mediated apoptotic cell death and cell viability. Cells were treated with either 10 μM DS18561882 (e) or 10 μM LY345899 (f) and sublethal doses of cisplatin for 48 h. g Endogenous interaction between MTHFD2 and DLAT, independent of other subunits of the PDC, was determined by MTHFD2 co-immunoprecipitation in cancer cells. h Immunofluorescence assay shows the co-localization of DLAT and MTHFD2 in A549^cisR cells. i Mitochondrial localization of DLAT and MTHFD2 in A549^cisR and KB-3-1^cisR cells is shown by immunoblotting of mitochondria and cytosolic fractions prepared using a mitochondria isolation kit. Tom40 and α-tubulin were used as control markers for mitochondria and cytosol, respectively. Immunofluorescence staining demonstrates the localization of DLAT (j) and MTHFD2 (k) with mitochondrial marker MitoTracker in A549^cisR cells. Scale bars represent 10 μm for (h, j, k). Data are mean ± SD from 3 independent biological replicates for (d–f). P values were determined by one-way ANOVA. Source data are provided as a Source Data file.

Fig. 5. DLAT activates MTHFD2 by acetylating at K44.

a MS spectra of acetyl-lysine peptide fragments of MTHFD2 containing K44 and K50. b A coupled in vitro DLAT acetylation and MTHFD2 activity assay using MTHFD2 wildtype, K44R, and K50R. In vitro acetylation assay was performed using bead-bound DLAT and purified GST-MTHFD2 variants. The acetylated MTHFD2 variants were applied to MTHFD2 activity assay using 0.2 mM tetrahydrofolate as a substrate. c Acetylation at K44 of MTHFD2 was assessed using a specific antibody against acetyl-K44 MTHFD2. MTHFD2 WT and K44R were applied to the in vitro DLAT acetylation assay and immunoblotting. d Effect of DLAT modulation on MTHFD2 K44 acetylation in A549^cisR and KB-3-1^cisR cells. Cells with endogenous DLAT knockdown were rescue expressed with WT or enzyme inactive ΔB DLAT. Acetylation of MTHFD2 at K44 was assessed using acetyl-K44 MTHFD2 antibody. e Left: In vitro DLAT acetylation assay using GST-MTHFD2 and flag-DLAT variants, wild-type (WT), ΔB domain deletion mutant (ΔB), or S475A. Acetylation of MTHFD2 and input were assessed by immunoblotting. Right: Coomassie Brilliant Blue staining of purified flag-DLAT variants. f, g Effect of MTHFD2 acetylation-mimetic mutant K44Q or -deficient mutant K44R expression on apoptosis and cell viability (f) and cellular ROS, NADPH levels, and GSH/GSSG ratio (g) in DLAT knockdown cells treated with cisplatin. DLAT knockdown cells were overexpressed with myc-tagged MTHFD2 K44Q or K44R mutants, and cisplatin resistance and redox status were determined by annexin V staining and bioluminescent assays. Data are mean ± SD from 4 independent biological replicates for cell viability of (f) and 3 for (b, g) and apoptosis rates of (f). P values were determined by one-way ANOVA. Source data are provided as a Source Data file.

Fig. 6. DLAT-Acetyl-MTHFD2 contributes to cisplatin resistance through 10-formyl-THF inducing MT-CO2 expression.

Effect of supplementation with 10-formyl-THF or formate on cisplatin-resistant cell survival and apoptotic cell death in cells with DLAT knockdown. Cells cultured under sublethal doses of cisplatin were treated with 10-formyl-THF (a) or formate (b) at indicated concentrations for 48 h. Apoptosis (top) and cell viability (bottom) were determined. 10-formyl-THF rescues cisplatin-resistant cell growth (c) and MT-CO2 expression (d) in cells lacking K44 MTHFD2 acetylation. Cisplatin-treated cells expressing K44R MTHFD2 were subjected to 10 μM 10-formyl-THF for 24 h followed by annexin V staining, cell viability assay, and assessment of MT-CO1, MT-CO2, and COXIV levels by immunoblotting and quantitative RT-PCR. Cells expressing the acetyl-mimetic mutant form of MTHFD2 K44Q were included for comparison. Effect of MT-CO2 overexpression on cisplatin-resistant cell growth in cells lacking K44 MTHFD2 acetylation (e) or DLAT (f) by K44R MTHFD2 expression or DLAT knockdown, respectively. Data are mean ± SD from 4 independent biological replicates for (d) and 3 for (a–c, e, f). P values were determined by one-way ANOVA. Source data are provided as a Source Data file.

Fig. 7. DLAT and MTHFD2 K44 acetylation correlates with poor clinical response in cancer patients receiving chemotherapy-containing regimens.

DLAT and acetyl-MTHFD2 staining of tumors collected from head and neck squamous cell carcinoma (HNSCC) patients treated with cisplatin or carboplatin-containing regimens. NED: no evidence of disease for 2 years after treatment (therapy sensitive), recurrent: disease recurred within 2 years of treatment (therapy resistant). DLAT expression (a) and MTHFD2 K44 acetylation (b) levels in treatment-sensitive and -resistant patient tumor specimens collected after the treatment. Representative images for each staining score (0 ~ + 3) are shown. c DLAT and acetyl-K44 MTHFD2 correlation in cisplatin/carboplatin-treated HNSCC patient samples. DLAT and acetyl-MTHFD2 staining of tumors collected from non-small cell lung carcinoma (NSCLC) patients receiving chemoimmunotherapy-containing regimens. PD: progressive disease with at least a 20% growth in the size of the tumor or spread of the tumor since the beginning of treatment (resistant); SD: stable disease or PR: partial response (sensitive). DLAT (d) and acetyl-MTHFD2 K44 (e) levels and treatment responses. f Correlation between DLAT and acetyl-K44 MTHFD2 in chemotherapy-treated NSCLC patient samples. Scale bars shown in (a, b) and (d, e) represent 50 μm. Results are presented as mean ± SD for (a, b, d, e). n = 18 (sensitive), n = 30 (resistant) for (a, b), n = 48 for (c), n = 9 (sensitive), n = 13 (resistant) for (d, e), n = 22 for (e). P values were obtained using unpaired two-tailed Student’s t-test (a, b, d, e) and two-tailed Pearson correlation (c, f). Source data are provided as a Source Data file.

Fig. 8. Decoy peptide DMp39 targets DLAT-mediated MTHFD2 acetylation and increases tumor chemosensitivity in vitro and in vivo.

a The structure of MTHFD2. K44 containing α-helix in MTHFD2, is marked in purple. The amino acid sequence of DMp39 is shown at the top. b HPLC-based in vitro serum stability assay of DMp39. DMp39 (0.2 mg) was incubated for indicated times with pre-warmed 50% serum and applied to HPLC. The remaining DMp39 is shown by the heights of the respective elution peaks. c Cells were treated with the indicated concentrations of DMp39 before detection of MTHFD2 K44 acetylation. d, e Effect of DMp39 on cisplatin sensitivity. DMp39 (20 μM) and cisplatin (2 or 5 μg/ml) were administered for 48 h. MTHFD2 acetylation, apoptotic cell death, and cell viability were determined. f Organ toxicity of the cisplatin and cy7-DMp39 combination in mice. Mice were injected with 0.1 mg/kg DMp39 and 5 mg/kg cisplatin from 7 days post xenograft 2 times/week for 22 days. The positive control group was administered 10 mg/kg/day of cisplatin for 3 consecutive days. Liver, kidney, and spleen toxicity were assessed by serum alanine aminotransferase, cystatin C, and spleen index (spleen weight (mg)/body weight (g) x 10), respectively. g Efficacy test of Cy7-DMp39 in vivo. Cy7 or cy7-DMp39 and cisplatin were administered as described in (f), and fluorescence was imaged and quantified in mice with KB-3-1^cisR-derived xenograft tumors (g; left) and with lung cancer patient-derived xenograft tumors (g; right). MTHFD2 and Ac-K44 MTHFD2 levels in tumors were assessed by co-immunoprecipitation of MTHFD2. Effect of DMp39, cisplatin, and the combination on tumor growth (h, k) and weight (i, l) in KB-3-1^cisR and lung cancer PDX tumor-bearing mice. Representative images of the tumors at the endpoint are presented in (h) and (k). Ac-K44 MTHFD2, MT-CO2, and DLAT levels in both xenograft mouse tumors were assessed by IHC staining (j, m). Scale bars in (h) and (k) represent 10 mm and 25 μm in (j) and (m). Results are presented as mean ± SEM for (h, k) and mean ± SD for the rest. n = 3 per group for (e, f), n = 6 for (g) left, n = 5 for (g) right, n = 6 for (h–j), and n = 5 for (k–m). P values were obtained by one-way ANOVA. Source data are provided as a Source Data file.