Opposing effects of acute versus chronic inhibition of p53 on decitabine’s efficacy in myeloid neoplasms

Decitabine is a DNA methyltransferase inhibitor and is considered a promising drug to treat myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) with p53 mutations. However, whether loss of p53 in fact increases the response of MDS/AML cells to decitabine remains unclear. In this study, we assessed the role of p53 in MDS and AML cells treated with decitabine using mouse models for MLL-AF9-driven AML and mutant ASXL1-driven MDS/AML. CRISPR/Cas9-mediated depletion of p53 in MDS/AML cells did not increase, but rather decreased their sensitivity to decitabine. Forced expression of a dominant-negative p53 fragment (p53DD) in these cells also decreased their responses to decitabine, confirming that acute inhibition of p53 conferred resistance to decitabine in AML and MDS/AML cells. In contrast, MLL-AF9-expressing AML cells generated from bone marrow progenitors of Trp53-deficient mice were more sensitive to decitabine in vivo than their wild-type counterparts, suggesting that long-term chronic p53 deficiency increases decitabine sensitivity in AML cells. Taken together, these data revealed a multifaceted role for p53 to regulate responses of myeloid neoplasms to decitabine treatment.

p53, encoded by TP53 in humans and Trp53 in mice, is the most frequently mutated gene in human cancer 1,2 . p53 is a transcription factor and regulates expression of downstream target genes involved in diverse cellular processes, including apoptosis, cell cycle arrest, senescence, and metabolic regulation. In addition, p53 maintains genomic stability as the "guardian of the genome". Through these functions, p53 plays a central role to prevent tumor initiation and progression. Loss of p53 function, either by mutation, gene deletion, or increased expression of negative regulators, leads to the development of various types of tumors, including hematopoietic neoplasms. Furthermore, p53 mutations are associated with resistance to standard chemotherapy and adverse outcomes in cancer patients.
Interestingly, recent clinical studies have shown that patients with acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS) who had p53 mutations exhibited favorable responses to the treatment with decitabine 3,4 . Furthermore, clonal analyses of the decitabine-treated patients revealed the marked, but not durable, clearance of subclones with TP53 mutations [3][4][5] . Decitabine is a hypomethylating agent that inhibits DNA methyltransferases (DNMTs), and is currently approved for the treatment of MDS and AML 6 . Consistent with the clinical observations, experimental studies have shown that decitabine induces cell death preferentially in p53 null or mutated cells than in p53 wild-type cells [7][8][9] . These findings suggest that decitabine is a promising drug to treat MDS and AML with p53 mutations. However, another report found no significant differences in the response rates of MDS patients with TP53 mutations and those with wild-type TP53 to hypomethylating agents 10 . In addition, several experimental studies have reported conflicting results regarding the relationship between DNA hypomethylation and p53 function. For example, loss of genomic methylation induced by Dnmt1 depletion caused p53-dependent apoptosis in fibroblasts 11 . It was also shown that decitabine treatment provoked p53 activation and apoptosis in colon cancer cells 12 . Thus, the role of p53 in decitabine-treated tumor cells appears to be Division of Cellular Therapy, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, 108-8639, Tokyo, Japan. Correspondence and requests for materials should be addressed to S.G. (email: goyama@ ims.u-tokyo.ac.jp) (2019) 9:8171 | https://doi.org/10.1038/s41598-019-44496-6 www.nature.com/scientificreports www.nature.com/scientificreports/ highly context-dependent. It is therefore important to determine the role of p53 in the regulation of decitabine's efficacy using appropriate models for MDS and AML.
We have developed several mouse models for AML and MDS with MLL fusions or ASXL1 mutations. MLL fusion leukemia is an aggressive leukemia carrying chimeric fusion of the MLL (KMT2A) gene 13,14 . MLL-AF9 is one of the most prevalent forms of MLL-fusion oncogene, and has the ability to transform both human and mouse hematopoietic progenitor cells into AML cells 15 . ASXL1 is an epigenetic regulator that interacts with multiple histone modifying enzymes including BAP1 and EZH2, and is frequently mutated in MDS and AML 6,16,17 . Most ASXL1 mutations exist in exon 12 of the gene, generating C-terminally truncated mutations. We have shown that a C-terminally truncated ASXL1 mutant promotes the development of MDS and AML in concert with NRAS, SETBP1 and RUNX1 mutations 16,[18][19][20][21] .
In this study, we assessed the role of p53 in the regulation of decitabine's efficacy using the above described mouse MDS/AML models and human cord blood cells. Our study clearly showed that acute inhibition of p53 did not increase, but rather decreased sensitivity of MDS/AML cells to decitabine. In contrast, AML cells generated from bone marrow progenitors of Trp53-deficient mice showed hypersensitivity to decitabine, which implies that long-term chronic inhibition of p53 function gradually increases sensitivity of AML cells to decitabine through secondary genetic and/or epigenetic changes.

Results
Acute inhibition of p53 confers resistance to decitabine in MLL-AF9-driven AML cells. We first examined the effect of acute depletion of p53 on decitabine's efficacy using a mouse model for MLL-AF9 leukemia. We transduced MLL-AF9 (coexpress GFP) into mouse bone marrow progenitor cells, and transplanted the cells into sublethally irradiated recipient mice. Consistent with our previous reports 15,22 , all recipient mice developed AML around 2 months after transplantation. We collected GFP + leukemia cells from spleens of the moribund mice, and enriched for leukemia stem cell activity by serial transplantation through secondary, tertiary, and quaternary recipient mice. These MLL-AF9-expressing cells with strong leukemogenicity were then transduced with Cas9 together with a non-targeting (NT)-single guide RNA (sgRNA) and two independent sgRNAs targeting Trp53 (Fig. 1a,b). sgTrp53-(2) induced nearly complete depletion of p53 protein, while sgTrp53-(1) induced expression of aberrant p53 protein that migrated faster than wild-type p53 protein in MLL-AF9 cells (Fig. 1c). MLL-AF9 cells transduced with the Trp53-targeting sgRNAs became resistant to a p53-Mdm2 interaction inhibitor DS-5272 23 (Fig. 1d), indicating that both sgRNAs depleted functional p53. We then compared sensitivity of MLL-AF9 cells transduced with these sgRNAs to decitabine using Cell Viability Assay. Decitabine inhibited the growth of control MLL-AF9 cells more efficiently than that of p53-depleted MLL-AF9 cells (Fig. 1d).
To characterize the decitabine-induced suppression of cell growth, we analyzed cell cycle status and apoptosis in vehicle-and decitabine-treated MLL-AF9 cells. Decitabine treatment induced G2/M arrest in both p53-intact and p53-depleted MLL-AF9 cells (Fig. 1e), suggesting that p53 is dispensable for decitabine-mediated cell cycle arrest. However, decitabine-induced cell death was not evident in p53-depleted MLL-AF9 cells (Fig. 1f), which probably accounts for their reduced response to decitabine.
To assess the role of p53 on decitabine's efficacy in vivo, we then transduced tRFP-coexpressing sgRNAs [NT and sgTrp53-(1)] into MLL-AF9/Cas9 cells. These MLL-AF9/Cas9/sgRNA-tRFP cells were transplanted into recipient mice, and the mice were treated with vehicle or decitabine. Although Trp53-depletion did not alter leukemic progression of MLL-AF9 cells in non-treated mice, it decreased sensitivity of MLL-AF9 cells to decitabine, as evidenced by the increase of Trp53-depleted (tRFP + ) cells in decitabine-treated mice (Fig. 1g). Thus, acute inhibition of p53 function by CRISPR/Cas9-mediated Trp53 depletion reduced responsiveness of MLL-AF9 cells to decitabine both in vitro and in vivo.
Next, we transduced vector or a dominant-negative p53 fragment (p53DD 24 ) (each coexpresses NGFR) into MLL-AF9 cells and assessed the effect of decitabine on these cells both in vitro and in vivo (Fig. 2a). MLL-AF9 cells transduced with p53DD grew normally in the presence of DS-5272 (Fig. 2b), indicating the efficient inhibition of p53 function by p53DD in them. Similar to the results of p53-depletion, p53DD-transduced cells were relatively resistant to decitabine compared with vector-transduced cells (Fig. 2c). We then transplanted vector or p53DD-transduced MLL-AF9 cells into recipient mice, and treated these mice with vehicle or decitabine. Flow cytometric analysis of NGFR + (vector/p53DD-transduced) cells in peripheral blood at day 16 revealed a tendency of increase of p53DD-transduced cells only in decitabine treated mice (Fig. 2d). These data suggest that forced expression of p53DD also conferred resistance to decitabine in MLL-AF9 cells, as Trp53 depletion did.
www.nature.com/scientificreports www.nature.com/scientificreports/ generated by combined expression of two MDS-associated mutations; SETBP1-D868N and ASXL1-E635RfsX15, in mouse bone marrow progenitor cells 20,21 . cRAM (combined expression of RUNX1 and ASXL1 Mutations) cells were generated by transducing a C-terminally truncated RUNX1 mutant (RUNX1-S291fsX300 25 ) into bone marrow progenitors derived from the conditional knock-in mice expressing mutant Asxl1 mimicking human ASXL1-E635RfsX15 mutation 19,21 . To examine the effect of p53 inhibition on the growth of MDS/AML cells, we transduced Cas9 and Trp53-targeting sgRNAs into cSAM and cRAM cells (Fig. 3a). Both cSAM and cRAM cells transduced with sgTrp53-(1) and (2) did not express full-length p53 (Fig. 3b), and were completely resistant to DS-5272 (Fig. 3c,d), indicating the efficient depletion of functional p53 in them. Similar to the results of MLL-AF9 cells, p53-depleted cSAM and cRAM cells became relatively resistant to decitabine (Fig. 3c,d). We then assessed the relationship between p53 and decitabine in human cord blood CD34 + cells. We transduced vector or p53DD (each coexpresses NGFR) into cord blood CD34 + cells, and cultured these cells in the absence or presence of decitabine. We observed increase of p53DD-transduced NGFR + cells only in the culture containing decitabine (Fig. 3e), indicating again the important role of p53 to enhance the growth-inhibitory effect of decitabine. Thus, acute inhibition of p53 conferred resistance to decitabine in both normal and malignant hematopoietic cells.

AML cells generated from bone marrow progenitors of Trp53(−/−) mice show hypersensitivity to decitabine.
Several studies have revealed phenotypic differences between transient and long-term depletion of particular genes in cells, partly due to genetic compensation in response to gene knockout 26 . To assess the influence of chronic p53 deficiency in AML cells, we next established a mouse AML model using Trp53-deficient mice. Because a compensatory network for p53 deficiency should have been established during embryonic development, cells derived from Trp53-deficient mice are suitable for this purpose. We transduced MLL-AF9 (coexpress GFP) into bone marrow progenitor cells derived from wild-type or Trp53(−/−) mice to generate MLL-AF9 leukemia cells through serial transplantations. We then transplanted these wild-type and Trp53-deficient MLL-AF9 leukemia cells into recipient mice, and treated them with vehicle or decitabine  www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 4a). Interestingly, in contrast to earlier results, decitabine treatment significantly prolonged survival of mice that received Trp53-deficient MLL-AF9 cells, while it showed only a marginal effect on those receiving wild-type MLL-AF9 cells (Fig. 4b). To directly compare sensitivity of wild-type and Trp53-deficient MLL-AF9 cells to decitabine, we next performed a competitive transplantation assay. We transduced mCherry into wild-type MLL-AF9 cells as a marker, mixed wild-type (mCherry + ) and Trp53-deficient (GFP + ) MLL-AF9 cells, and transplanted them to recipient mice. The recipient mice who received both wild-type (mCherry + ) and Trp53-deficient (GFP + ) MLL-AF9 cells were then treated with vehicle or decitabine (Fig. 4c). Again, treatment with decitabine prolonged survival of these recipient mice (Fig. 4d). Trp53-deficient (GFP + ) cells always became the major population in mice treated with vehicle control, indicating their stronger leukemogenic activity than wild-type counterparts. On the other hand, the ratio of GFP + /mCherry + cells was highly variable between individual mice, and importantly, wild-type MLL-AF9 (mCherry + ) cells often became dominant in decitabine-treated mice (Fig. 4e). These data suggest the increased in vivo sensitivity of Trp53-deficient MLL-AF9 cells to decitabine. Thus, contrary to the effect of acute p53 inhibition, chronic inhibition of p53 function appears to increase sensitivity of AML cells to decitabine treatment.
We then examined immunophenotypes of these MLL-AF9 cells. Acute p53 depletion using CRISPR/Cas9 resulted in an increase of immature cell populations (c-Kit + Gr-1 − ) that are enriched for self-renewing leukemia stem cells (LSCs) 28 . Interestingly, the c-Kit + Gr-1 − LSC-enriched population was not increased in Trp53-deficient MLL-AF9 cells (Fig. 7a,b). Although decitabine induced differentiation of all these MLL-AF9 cells (Fig. 7a-c), the distinct immunophenotypes between sgTrp53-transduced and Trp53-deficient MLL-AF9 cells in steady state could underlie the opposing effects of acute vs chronic p53 depletion on decitabine's efficacy in vivo. We also assessed expression of LC3-II, a marker of autophagy 29 , in vehicle or decitabine-treated MLL-AF9 cells. The LC3-II expression was highest in Trp53-deficient MLL-AF9 cells, which was further increased by decitabine treatment (Fig. 7d). These data suggest that chronic p53 deficiency may lead to increased autophagy signaling, which could also contribute to the enhanced sensitivity of Trp53-deficient MLL-AF9 cells to decitabine.

Discussion
Development of drugs that target p53-mutated tumor cells has been an important challenge. Decitabine is a promising candidate with such function. However, it is still under debate whether loss of p53 in fact increases the response of tumor cells to decitabine. Our data strongly indicate that acute inhibition of p53 does not increase sensitivity, but rather confers resistance to decitabine in MDS and AML cells. Therefore, co-inhibition of DNMTs and p53 will not be a therapeutic option for p53-intact tumors. Instead, combination therapy with hypomethylating agents and p53 activators could show synergistic inhibitory effects in myeloid neoplasms with wild-type p53.
In contrast to the effect of acute p53 inhibition, long-term deficiency of p53 appears to increase sensitivity of AML cells to decitabine. We found that MLL-AF9-driven AML cells generated from bone marrow progenitor cells of Trp53(−/−) mice were more sensitive to decitabine treatment than their p53-intact counterparts in vivo. Consistent with our data, a previous report also showed that decitabine induced differentiation of MLL-AF9 cells through p53-independent mechanisms 30 . Thus, these findings suggest that AML cells with chronic p53 deficiency/mutations retain sensitivity, or even become more sensitive to decitabine treatment.
Mechanisms for the phenotypic differences between acute and chronic inhibition of p53 in MDS/AML cells warrant further investigation. Several recent studies have revealed genetic compensation in response to gene knockout can explain the different results obtained by acute and chronic inhibition of a specific gene 26 . For example, conditional depletion of tumor suppressor RB1 enables cell-cycle reentry of quiescent MEFs, while quiescent MEFs derived from Rb1-mutant mice are unable to reenter the cell cycle, partly due to the compensatory upregulation of p107 31 . Similar genetic compensation may occur in AML cells generated from Trp53(−/−) bone marrow cells. In our MLL-AF9 leukemia model, acute p53 depletion using CRISPR/Cas9 increased the frequency of LSC-enriched population, whereas chronic p53 deficiency in MLL-AF9 cells rather decreased the LSC fraction. We also found the substantial enhancement of autophagy signaling only in MLL-AF9 cells with chronic p53 deficiency. These differences between acute and chronic p53 inhibition could contribute to their opposing effects on decitabine's efficacy in leukemia cells. How the chronic loss of p53 induces these distinct changes in myeloid neoplasms remains to be elucidated. Given that p53 plays crucial roles to maintain genomic and epigenetic treated with decitabine. (right) Representative FACS plots. (left) Percentages of Trp53-deficient (GFP + only) MLL-AF9 cells in all (both GFP + and mCherry + ) MLL-AF9 leukemia cells. N = 5 (vehicle) or 6 (decitabine). *P = 0.0303, **P = 0.0152, ***P = 0.0087, Mann-Whitney U test.
www.nature.com/scientificreports www.nature.com/scientificreports/ homeostasis 32 , long-term p53 deficiency may induce secondary genetic and/or epigenetic changes that increase the response of MDS/AML cells to decitabine. Key factors involved in these changes need to be identified in future studies.
Our study also revealed that p53 plays a critical role to regulate therapeutic resistance in MLL-fusion-driven AML and mutant ASXL1-driven MDS/AML. Although TP53 mutations rarely coexist with MLL-fusions and ASXL1 mutations, nonmutational p53 dysfunction could contribute to the development of myeloid neoplasms with MLL-rearrangements or ASXL1 mutations. Indeed, several studies have shown that MLL-fusion oncoproteins suppress p53 function [33][34][35] . Furthermore, we previously showed that mutant ASXL1 derepressed miR-125a 18 , www.nature.com/scientificreports www.nature.com/scientificreports/ which was shown to downregulate p53 expression 36 . We also found that the MDM2 inhibitor DS-5272 has potent antileukemia effects against MLL-AF9 leukemia (Hayashi Y et al., unpublished data). These findings suggest that multiple nonmutational mechanisms can disturb p53 function in these myeloid neoplasms. In addition to the well-known somatic TP53 mutations, germ line mutations of TP53 occur in a small fraction of AML patients and www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ are particularly frequent in therapy-related AML 37 . The MLL-AF9-driven leukemia model using Trp53(−/−) cells could be a good experimental model for AML with germ line TP53 mutations. Our data also raise the possibility that AML patients with germline TP53 mutations, who often show resistance to chemotherapies, might respond well to decitabine treatment.
In summary, we showed the opposing effects of acute and chronic p53 inhibition on decitabine's efficacy in MDS and AML cells. Acute inhibition of p53 confers resistance to decitabine, while long-term chronic inhibition of p53 increases sensitivity to decitabine in myeloid neoplasms. Our results imply that the favorable clinical response of patients with TP53 mutations to decitabine is not the direct consequence of p53 dysfunction, but is likely due to secondary genetic/epigenetic alterations provoked by p53 dysfunction in MDS/AML cells.
Mice. C57BL/6 (Ly5.2) mice (Sankyo Labo Service Corporation, Tokyo, Japan) were used for bone marrow transplantation assays. Trp53(−/−) mice, in which 5′ part of exon 2 including translation initiation site of Trp53 gene was replaced with Neomycin resistance gene, were provided from the RIKEN BioResource Center (Ibaragi, Japan) 40 . All animal studies were approved by the Animal Care Committee of the Institute of Medical Science at the University of Tokyo (approval number: PA13-19, PA16-31, PA17-75), and were conducted in accordance with the Regulation on Animal Experimentation at University of Tokyo based on International Guiding Principles for Biomedical Research Involving Animals.
Transplantation assay and Decitabine treatment. Mouse bone marrow cells derived from wild-type or Trp53(−/−) mice were transduced with MLL-AF9, and were transplanted intravenously into sublethally irradiated (525 cGy) recipient mice. MLL-AF9-expressing leukemia cells were harvested from spleens of moribund mice, and were serially transplanted into recipient mice. The serial transplantation was subsequently repeated several times to generate MLL-AF9 cells with strong leukemogenicity. The MLL-AF9 cells were then transduced with Cas9 and sgRNAs, or transduced with vector or p53DD, and were injected intravenously into non-irradiated recipient mice (5 × 10 5 /body). For experiments in Fig. 4, wild-type and Trp53(−/−) MLL-AF9 cells (1-5 × 10 5 cells/body) were injected intravenously into non-irradiated recipient mice. For competitive transplantation, we transduced mCherry into wild-type MLL-AF9 cells as a marker, and mixed wild-type and Trp53(−/−) MLL-AF9 cells before transplantation. For treatment with decitabine, mice were subcutaneously administered with vehicle control or decitabine (0.6 mg/kg/day) every other day from day 1. Decitabine (HY-A0004) was purchased from MedChemexpress.
Human cord blood cell culture. Human cord blood (CB) were obtained from the Japanese Red Cross Kanto-Koshinetsu Cord Blood Bank (Tokyo, Japan). Mono nuclear cells (MNCs) were isolated from CB by density gradient centrifugation using Lymphoprep TM (density 1.077; Alere Technologies AS, Oslo, Norway). The CD34 + cell fraction was then isolated from the MNCs using the MidiMACS system (CD34 + Microbead Kit; Miltenyi Biotec; Bergisch Gladbach, Germany) according to the manufacturer's protocols. CB CD34 + cells were incubated in StemSpan TM SFEMII (STEMCELL Technologies) supplemented with 50 ng/ml mouse Flt-3, 50 ng/ml human TPO, 50 ng/ml human SCF, 25 ng/ml human IL-6 (R&D Systems) and 750 nM stemregenin-1 41 . Cell viability assay. Cells were seeded in a 96-well plate at a density of 1 × 10 4 cells per well, and were cultured with DS-5272 or decitabine for 72 h. Cell viability was measured by Cell Counting Kit-8 (Dojindo,