Uncoupling DNA damage from chromatin damage to detoxify doxorubicin

Abstract

The anthracycline doxorubicin (Doxo) and its analogs daunorubicin (Daun), epirubicin (Epi), and idarubicin (Ida) have been cornerstones of anticancer therapy for nearly five decades. However, their clinical application is limited by severe side effects, especially dose-dependent irreversible cardiotoxicity. Other detrimental side effects of anthracyclines include therapy-related malignancies and infertility. It is unclear whether these side effects are coupled to the chemotherapeutic efficacy. Doxo, Daun, Epi, and Ida execute two cellular activities: DNA damage, causing double-strand breaks (DSBs) following poisoning of topoisomerase II (Topo II), and chromatin damage, mediated through histone eviction at selected sites in the genome. Here we report that anthracycline-induced cardiotoxicity requires the combination of both cellular activities. Topo II poisons with either one of the activities fail to induce cardiotoxicity in mice and human cardiac microtissues, as observed for aclarubicin (Acla) and etoposide (Etop). Further, we show that Doxo can be detoxified by chemically separating these two activities. Anthracycline variants that induce chromatin damage without causing DSBs maintain similar anticancer potency in cell lines, mice, and human acute myeloid leukemia patients, implying that chromatin damage constitutes a major cytotoxic mechanism of anthracyclines. With these anthracyclines abstained from cardiotoxicity and therapy-related tumors, we thus uncoupled the side effects from anticancer efficacy. These results suggest that anthracycline variants acting primarily via chromatin damage may allow prolonged treatment of cancer patients and will improve the quality of life of cancer survivors.

Keywords: DNA damage; cardiotoxicity; chromatin damage; doxorubicin; therapy-related tumors.

Fig. 1.

Doxo and Etop, but not Acla, accelerate tumor formation and Doxo causes tissue toxicities in Trp53^+/− FVB mice. (A) Trp53^+/− FVB mice were i.v. injected with Doxo, Acla, Etop, or saline every 2 wk for six times. Drug injections are indicated by arrows. (B) Tumor-free survival is plotted in a Kaplan−Meier curve. Drug injections are indicated by arrows. Log-rank test, ns, not significant; ****P < 0.0001; **P = 0.0093. (C) Tumor-free median survival of mice. (D) Representative microscopic images of the heart from Doxo-treated mouse with thrombosis formation in the left atrium/auricle. Higher magnifications shows thrombi and inflammatory lesions including fibrosis in the left auricle. LA, left atrium; RA, right atrium; LV, left ventricle; RV, right ventricle. (Scale bars, 500 µm and 100 µm, respectively.) (E) Cumulative incidence of thrombosis was analyzed for gender effect. Fisher’s exact test, two-sided. *P < 0.05; **P < 0.01; ****P < 0.0001. (F) The incidence rate of thrombus formation in the heart. Drug injections are indicated by arrows. Cumulative incidence is indicated next to the curve. Two-way ANOVA with repeated measures (RM). (G) Representative Sirius Red staining of the LA from saline- or Doxo-treated mouse. (Scale bars, 100 µm.) (H) Quantification of Sirius Red staining. Kruskal−Wallis test. (I) Incidence rate of depletion of spermatogenesis in male mice. Drug injections are indicated by arrows. Cumulative incidence is indicated next to the curve. Two-way ANOVA with RM.

Fig. 2.

Evaluation of the DNA- and chromatin-damaging activities of anthracyclines. (A) Structures of Topo II poisons used in this study, with the critical amine group in red. (B) K562 cells were treated for 2 h with 10 µM indicated drug. γH2AX levels were examined by Western blot. (C) Quantification of the γH2AX signal normalized to actin. (D) DSBs were analyzed by CFGE. (E) Quantification of relative broken DNA in D. (F) Part of the nucleus from MelJuSo-PAGFP-H2A cells was photoactivated. Photoactivated PAGFP-H2A was monitored by time-lapse confocal microscopy for 1 h in the absence or presence of indicated drug at 10 µM. Lines in Pre column define the regions of cytoplasm (C), nucleus (N), and activated area (A). (Scale bar, 10 µm.) (G) Quantification of the release of fluorescent PAGFP-H2A from the photoactivated region after drug administration. Two-way ANOVA, ****P < 0.0001. (H) Endogenously tagged scarlet-H2B U2OS cells were treated with 10 µM indicated drugs. Cells were fractionated, and the nuclear versus cytosolic fraction of H2B was examined by Western blot. Calnexin was used as cytosolic, and lamin A/C was used as nuclear marker. (I) The fraction of cytosolic versus nuclear H2B upon histone eviction by the drugs indicated is plotted. Two-way ANOVA, ***P < 0.001; ****P < 0.0001; ns, not significant. (J) Cell viability in K562 cells. Two-way ANOVA, Amr vs. Doxo, diMe-Doxo or Acla, **P < 0.01. (K) Relative IC₅₀ values of each drug compared to Doxo in different cell lines.

Fig. 3.

Both diMe-Doxo and Acla are effective anticancer drugs. (A) Overall survival of de novo geriatric AML patients treated with a drug regimen including Acla or Ida. Log-rank (Mantel−Cox) test. (B and C) Dose-dependent cell viability of human AML samples, shown as mean ± SD of technical duplicates. (D) Schematic overview of AML PDX mouse experiment. (E and F) The engraftment of human AML cells in the BM of the first cohort at week 9: absolute counts of (E) myeloid blast cells (CD45⁺CD33⁺ blasts) and (F) HSCs and LSCs (CD34⁺CD38⁻ blasts). Each symbol represents one mouse. Student’s’ t test. (G) The engraftment of human AML cells of the second cohort: the percentage of human HSCs and LSCs in PB at week 13. Student’s’ t test. (H) C57BL/6 mice were s.c. injected with MC38 cells. One week after tumor challenge, mice were treated with indicated drugs every week. Drug injections are indicated by arrows. (I and J) MC38 tumor growth following (I) experiment 1 procedure or (J) experiment 2 procedure. One-way ANOVA, saline vs. treatment, *P < 0.05; **P < 0.01.

Fig. 4.

N,N-dimethylation of Doxo prevents cardiotoxicity in mice and hiPSC-derived cardiac microtissues. (A–C) Wild-type (WT) FVB mice were i.v. injected with indicated drug every 2 wk: Doxo or Acla for 8 times and diMe-Doxo for 15 times. (B) The body weight of mice, shown as floating bars with maximum−median−minimum values. Two-way ANOVA with RM. (C) The incidence rate of cardiotoxicity. Arrows indicate drugs injections. Two-way ANOVA. (D–G) Cardiac function assessed by echocardiography 12 wk posttreatment start. FVB mice were treated for eight times with saline (5 mL/kg, n = 5), Doxo (5 mg/kg, n = 8), diMe-Doxo (5 mg/kg, n = 6), or Acla (5 mg/kg, n = 9). (D) A 3D reconstruction of the diastole heart by echocardiography. In the sagittal section, the left ventricle (cyan) and left atrium (magenta) are highlighted. (E–G) Quantification of echocardiography, (E) FS, (F) left ventricular EF, and (G) cardiac output. For E–G, ordinary one-way ANOVA, Doxo vs. saline, diMe-Doxo, or Acla, *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. (H) Drug toxicity on cardiac microtissues 24 h posttreatment. Contraction amplitude of microtissues treated with 20 µM different drugs. Krushal−Wallis test, ***P < 0.0002. (I–K) Drug toxicity on cardiac microtissues treated with single drugs or a combination of Amr and Acla. (I) Contraction amplitude of microtissues treated with the indicated single (20 µM) or combination drugs (10 µM + 10 µM). (J) Maximum velocity in micrometers per second is indicated for a represented microtissue for the different treatments. (K) Quantification of the maximum velocity. For I and K, ordinary one-way ANOVA, Ctr vs. treatments; *P < 0.05; ***P < 0.001; ns, not significant.