CRISPRi/a screens in human iPSC-cardiomyocytes identify glycolytic activation as a druggable target for doxorubicin-induced cardiotoxicity

Abstract

Doxorubicin is limited in its therapeutic utility due to its life-threatening cardiovascular side effects. Here, we present an integrated drug discovery pipeline combining human induced pluripotent stem cell (iPSC)-derived cardiomyocytes (iCMs), CRISPR interference and activation (CRISPRi/a) bidirectional pooled screens, and a small-molecule screening to identify therapeutic targets mitigating doxorubicin-induced cardiotoxicity (DIC) without compromising its oncological effects. The screens revealed several previously unreported candidate genes contributing to DIC, including carbonic anhydrase 12 (CA12). Genetic inhibition of CA12 protected iCMs against DIC by improving cell survival, sarcomere structural integrity, contractile function, and calcium handling. Indisulam, a CA12 antagonist, can effectively attenuate DIC in iCMs, engineered heart tissue, and animal models. Mechanistically, doxorubicin-induced CA12 potentiated a glycolytic activation in cardiomyocytes, contributing to DIC by interfering with cellular metabolism and functions. Collectively, our study provides a roadmap for future drug discovery efforts, potentially leading to more targeted therapies with minimal off-target toxicity.

Keywords: CRISPR screen; carbonic anhydrase; cardio-oncology; cardiomyoctye; doxorubicin; glycolysis; iPSC.

Figure 1.. Integration of iCMs with CRISPRi/a screens enables identification of novel targets for DIC (doxorubicin-induced cardiotoxicity).

(A) Diagram depicting the iCMs system for large-scale CRISPRi/a screens. The iPSCs were differentiated into iCMs (two biological replicates), which were subsequently transduced with pooled sgRNA libraries at an MOI of 0.3 to ensure each infected iCM received only one sgRNA. Puromycin selection was carried out to eliminate iCMs without sgRNA integration. Then iCMs were transduced with lentiviral dCas9 with either CRISPRi/a machinery at an MOI > 1.5 before cells were treated with vehicle or doxorubicin (0.5 μM) for 72 hours. Surviving iCMs of both groups were harvested to extract genomic DNAs. Screen hits were analyzed based on sgRNA barcode counts and computational analysis through DNA sequencing. The CRISPRi/a screens were conducted on two biological replicates: SCVI273 iCMs and SCVI15 iCMs. Representative images of doxorubicin- and DMSO-treated iCMs demonstrated that sarcomere structure of iCMs was disrupted by doxorubicin (Doxo) as assessed by troponin T (TNNT2) and α-actinin (ACTN2) immunofluorescence staining. (B) Scatter plot depicting CRISPRi screen results for DIC in iCMs. Top ranked sgRNAs targeting such as HPGDS, SCMH1, CA12, ERBB2, and DGKH were enriched (score > 2, P < 0.01). (C) Scatter plot depicting CRISPRa screen results for DIC in iCMs. Top ranked sgRNAs targeting such as CLK2, ATP4A, TWF2, FLT4, and MAP3K4 were enriched (score > 1, P < 0.01). (D) Venn diagram displaying overlapping gene hits between iCMCRISPRi/a screens versus a cancer cell-CRISPR knockout screen after doxorubicin treatment (score > 1, P < 0.05). (E) Gene ontology (GO) biological process and pathway enrichment analysis of iCM CRISPRi/a screen results. See also Figure S1A-G.

Figure 2.. Interference of CA12 attenuates doxorubicin-induced cytotoxicity in iCMs.

(A) Validation of sgRNAs that showed reduced DIC in iCMs from the original CRISPRi/a screens. n = 3. (B) Schematic design of the creation of CRISPRi iPSC lines to validate CRISPRi screen hits. sgRNA targeting the promoters of desired genes was transduced in iPSCs expressing inducible dCas9 fused with KRAB and single clones were picked and expanded to establish specific gene knockdown (Gen2C) iPSC lines. (C) Cell viability in wild-type (WT) and CA12 CRISPRi iCMs treated with vehicle or doxorubicin. n = 3. (D) Cell apoptotic responses in WT and CA12 CRISPRi iCMs treated with vehicle or doxorubicin. Cell apoptosis was assessed by a Caspase-3/7 assay kit. n = 4. (E) Representative immunofluorescent images of ACTN2 and TNNT2 showing sarcomere structure changes in WT and CA12 CRISPRi iCMs treated with vehicle or doxorubicin. (F) CA12 expression levels in WT and CA12 CRISPRi iCMs treated with vehicle or doxorubicin. n = 3. (G) RNA-seq analysis of CA12 expression levels in iCMs with and without doxorubicin treatment from three published GEO databases (GSE157282, GSE106297, and GSE76314). (H-I) Contractile functions measured using video microscopy–based motion vector in WT and CA12 CRISPRi iCMs treated with vehicle or doxorubicin. n = 9. See also Figure S1H.

Figure 3.. Genetic deletion and pharmacological inhibition of CA12 reduce cytotoxicity and improve functions of doxorubicin-treated iCMs.

(A-B) Expression levels of CA12 in wild-type (WT) and CA12 knockout (KO) iPSCs detected by quantitative real-time PCR and western blot (SCVI273 line). n = 3. (C) Cell viability in WT and CA12 KO iCMs treated with vehicle and doxorubicin. n = 3. (D) Cell apoptosis in WT and CA12 KO iCMs treated with vehicle and doxorubicin. n = 4. (E) Contraction velocity in WT and CA12 KO iCMs treated with vehicle and doxorubicin. n = 5. (F) Normalized relative calcium decay in WT and CA12 KO iCMs treated with vehicle and doxorubicin. Values were normalized to that of the WT group. n = 35–43. (G) Representative MEA recordings showing that rate-corrected field potential duration (FPDc) was prolonged in doxorubicin-treated iCMs compared to vehicle-treated iCMs. (H) MEA analysis showing FPDc changes among WT and CA12 KO iCMs treated with vehicle and doxorubicin. n = 4. (I) Schematic overview of drug discovery using an integrated platform combining iCMs, CRISPRi/a screens, drug screen, and molecular docking. (J) Docking of indisulam into CA12 protein using Schrödinger molecule docking software showing indisulam as a potential CA12 inhibitor. (K) Indisulam functioned as a CA12 inhibitor to reduce doxorubicin-induced iCM death. n = 8. (L) Indisulam downregulated doxorubicin-induced activation of carbonic anhydrase activity. n = 3. (M) Indisulam rescued doxorubicin-induced impairment of contraction velocity in iCMs. n = 4. (N) Indisulam improved doxorubicin-induced prolonged calcium decay in iCMs. n = 35–56. (O) MEA analysis showing FPDc changes among WT and CA12 KO iCMs treated with vehicle and doxorubicin. n = 12. All data were obtained in SCVI273 iCMs. See also Figure S1I-J, Figure S2A-H, and Figure S3A-G.

Figure 4.. Genetic deletion and pharmacological inhibition of CA12 reduce DNA damage and improve sarcomere arrangement in doxorubicin-treated iCMs.

(A) Western blot analyses showing CA12 and γH2A.X (a DNA damage marker) expression levels in wild-type (WT) and CA12 knockout (KO) iCMs (SCVI273 iPSC line) with and without doxorubicin treatment. (B) Quantification of the CA12 and γH2A.X protein levels in all groups shown in A. (C) Western blot analysis showing TOP2B expression in CA12 KO and WT iCMs after vehicle or doxorubicin treatments (SCVI273 iCMs). (D) Densitometric quantification of the TOP2B protein levels in different treatment groups shown in C. (E-F) CA12 and γH2A.X protein levels in iCMs derived from two iPSC lines (SCVI273 and SCVI15) after treatment with vehicle, doxorubicin (Doxo), and Doxo + indisulam (Indi). (G) Representative immunofluorescent images and analysis of WT iCMs and CA12 KO iCMs after DMSO, doxorubicin, or doxorubicin + indisulam treatment for 72 hr from SCVI273 and SCVI15 lines. Micro-patterned iCMs were stained for ACTN2 and TNNT2 to evaluate the sarcomere structural integrity. (H) Quantification of continuous z-line lengths of WT and CA12 KO iCMs treated with vehicle, doxorubicin, or doxorubicin + indisulam shown in G. Analyses were conducted in iCMs derived from SCVI273 and SCVI15 iPSC lines, and 6–9 cells from each group were used for quantification. (I) Quantification of sarcomere orientational order parameter (OOP) in SCVI273 and SCVI15 iCMs from different groups. n = 3. (J) Diagram showing the experimental design of using iPSC-derived human engineered heart tissues (EHTs) to assess the rescue effect of indisulam on doxorubicin-induced cardiotoxicity in a 3D format. (K) Cell viability analysis of EHTs treated with doxorubicin and doxorubicin + indisulam (SCVI20 iCMs) using a CellTiter-Glo 3D cell viability assay. (L) Contractile function changes of EHTs after 72 hr treatment with vehicle, doxorubicin, and doxorubicin + indisulam. n = 4. See also Figure S1I-J, Figure S2I-J, and Figure S4.

Figure 5.. Transcriptional analysis of the role of CA12 in doxorubicin-induced cardiotoxicity.

(A) Geneset UMAP, clustering of genesets by distance of gene-gene covariance. Genes are clustered nearby by high covariance (colors indicate standard deviations) (left panel). UMAP clustering of genes by covariances (colors indicate relative log-expression of each sample) (right panel). iCMs from wild-type group (WT), CA12 knockout group (KO), wild-type + doxorubicin group (WT + Doxo), wild-type + doxorubicin + indisulam group (WT + Doxo + Indi), and CA12 knockout + doxorubicin group (KO + Doxo) were all derived from SCVI273 line. Doxorubicin and indisulam altered pathways were highlighted in circles. (B) Venn diagram of differentially expressed genes across paired groups (WT + Doxo vs. WT, WT + Doxo vs. KO + Doxo, and WT + Doxo vs. WT + Doxo + Indi) (FDR < 0.01 and log₂ fold-change > 1). (C) Hallmark pathway analysis of the 559 overlapped genes between paired groups in B. (D) The top downregulated genes by doxorubicin that were rescued by genetic (knockout) and pharmacological (indisulam) inhibition of CA12 in iCMs. (E) The top upregulated genes by doxorubicin that were repressed by genetic (knockout) and pharmacological (indisulam) inhibition of CA12 in iCMs. (F) Quantitative real-time PCR analysis of identified top DEGs in SCVI273 and SCVI15 iCMs from vehicle, doxorubicin, and doxorubicin + indisulam treatment groups. n = 3. (G) The basal glycolysis and compensatory glycolysis levels detected by glycoPER using a Seahorse XF Glycolytic Rate Assay in DMSO and doxorubicin-treated WT and CA12 KO iCMs derived from SCVI273 and SCVI15 lines. n = 5–20. (H) A glycolysis inhibitor 2-deoxy-d-glucose (2-DG) improved cell viability in doxorubicin-treated iCMs. n = 6. (I) Cell viability of iCMs from groups of wild-type control, PDK4 CRISPRi, wild-type + doxorubicin, PDK4 CRISPRi + doxorubicin. n = 3. (J) Effects of PDK4 CRISPRi on contraction velocity of doxorubicin-treated iCMs. n = 9. (K) Effects of PDK4 inhibition on calcium decay in doxorubicin-treated iCMs measured by calcium imaging. The data were obtained using iCMs from groups of wild-type control, wild-type control + doxorubicin, and PDK4 CRISPRi + doxorubicin. n = 9. See also Figure S5 and Figure S6A-F.

Figure 6.. Inhibition of CA12 attenuates doxorubicin-induced cardiotoxicity in vivo.

(A) Schematic showing experimental designs of evaluating the rescue effect of indisulam on doxorubicin-induced cardiotoxicity in mice. (B) Echocardiogram analysis showing changes in left ventricular ejection fraction in mice receiving vehicle (n = 5), doxorubicin (n = 10), and doxorubicin + indisulam treatments (n = 10). (C) Echocardiogram analysis showing changes in left ventricular fraction shortening in mice receiving vehicle (n = 5), doxorubicin (n = 10), and doxorubicin + indisulam treatments (n = 10). (D) Heart weight to tibia length ratios with representative images of mouse hearts from treatment groups of vehicle, doxorubicin, and doxorubicin + indisulam (n = 14 mice/group). (E and F) Wheat germ agglutinin (WGA) staining and quantification of myocyte cross-section areas in mice receiving vehicle, doxorubicin, and doxorubicin + indisulam treatments (> 142 slides were analyzed per group). (G) Representative images showing primary mouse cardiomyocytes isolated using a Langendorff-free enzymatic perfusion approach. (H) Representative calcium traces of isolated primary mouse cardiomyocytes from mice receiving vehicle, doxorubicin, and doxorubicin + indisulam treatment. (I) Normalized calcium decay measured by calcium imaging in isolated primary mouse cardiomyocytes from mice receiving vehicle, doxorubicin, and doxorubicin + indisulam treatment. n = 38–47. (J) Normalized calcium peak amplitude detected in isolated primary mouse cardiomyocytes from mice receiving vehicle, doxorubicin, and doxorubicin + indisulam treatment. n = 38–47. (K) Additive effect of indisulam and an RARG agonist (CD1530, 10 μM) on improving doxorubicin-induced cell death in iCMs derived from cancer patients who are more sensitive (DoxTox) to DIC. n = 3. (L) The effect of indisulam on iCM (SCVI273) viability following treatment with mitoxantrone (1 μM), epirubicin (1 μM), daunorubicin (1 μM), and idarubicin (5 μM). n = 6. See also Figure S2C, Figure S3H-K, Figure 4N-Q, and Figure S6G-N.