Decitabine mildly attenuates MLL‐rearranged acute lymphoblastic leukemia in vivo, and represents a poor chemo‐sensitizer

Abstract MLL‐rearranged acute lymphoblastic leukemia (ALL) represents a highly aggressive ALL subtype, characterized by aberrant DNA methylation patterns. DNA methyltransferase inhibitors, such as decitabine have previously been demonstrated to be effective in eradicating MLL‐rearranged ALL cells in vitro. Here, we assessed the in vivo anti‐leukemic potential of low‐dose DNA methyltransferase inhibitor decitabine using a xenograft mouse model of human MLL‐rearranged ALL. Furthermore, we explored whether prolonged exposure to low‐dose decitabine could chemo‐sensitize MLL‐rearranged ALL cells toward conventional chemotherapy as well as other known epigenetic‐based and anti‐neoplastic compounds. Our data reveal that decitabine prolonged survival in xenograft mice of MLL‐rearranged ALL by 8.5 days (P = .0181), but eventually was insufficient to prevent leukemia out‐growth, based on the examination of the MLLAF4 cell line SEM. Furthermore, we observe that prolonged pretreatment of low‐dose decitabine mildly sensitized toward the conventional drugs prednisolone, vincristine, daunorubicin, asparaginase, and cytarabine in a panel of MLL‐rearranged cell lines. Additionally, we assessed synergistic effects of decitabine with other epigenetic‐based or anticancer drugs using high‐throughput drug library screens. Validation of the top hits, including histone deacetylase inhibitor panobinostat, BCL2 inhibitor Venetoclax, MEK inhibitor pimasertib, and receptor tyrosine kinase foretinib, revealed additive and moderate synergistic effects for the combination of each drug together with decitabine in a cell line‐dependent manner.


INTRODUCTION
Rearrangement of the Mixed Lineage Leukemia (MLL, or KMT2A) gene is a cytogenetic aberration highly prevalent in infants (<1 year of age) diagnosed with acute lymphoblastic leukemia (ALL), where it constitutes ∼80% of the cases. MLL rearrangements mark a very aggressive ALL subtype. Despite highly intensified treatment protocols, eventfree survival (EFS) chances for MLL-rearranged infant ALL only reach 35-40%, falling well short of survival rates of infants and older children with ALL carrying other cytogenetic aberrations (70-90%) [1][2][3].
Hence, novel treatment strategies based on the specific molecular pathobiology are crucial.
The main oncogenic hit of MLL-rearranged ALL is the in-frame fusion of the MLL gene with one of multiple fusion partner genes, generating MLL fusion genes that encode chimeric proteins that drive leukemogenicity and disease maintenance [4][5][6]. MLL itself functions as a histone methyltransferase, and the most recurrent fusion partner genes, AF4 (AFF1), ENL (MLLT1), and AF9 (MLLT3), all encode proteins that are part of complexes regulating epigenetic mechanisms. As truncated parts of the MLL fusions these proteins interfere with and mistarget the regulating complexes, hijacking their activities [7]. As a result, MLL-rearranged acute leukemia typically presents with a highly abnormal epigenome, reflected by aberrant DNA methylation patterns [8][9][10] and histone modification signatures [11], which alter the epigenetic and transcriptomic landscape of the cell. Consequently, several epigenetic drug classes, including DOT1L histone methyltransferase, BET protein, and histone deacetylases (HDAC) inhibitors, have shown promising results in MLL-rearranged ALL animal models [12][13][14][15][16], providing preclinical rationales for their implementation in current and future clinical trials [17].
However, despite their known cytotoxicity against MLL-rearranged ALL cells in vitro [7,10,17,18], preclinical in vivo activity studies of another pivotal class of epigenetic drugs, that is, the DNA methyltransferase inhibitors (DNMTi), such as decitabine and 5-azacytidine, are limited. Therefore we assessed the in vivo anti-leukemic potential of low and clinically relevant dosages of decitabine for a prolonged timespan in a MLL-rearranged ALL xenograft mouse model. Furthermore, using high-throughput combinatorial drug library screens, we explored whether prolonged low-dose decitabine would epigenetically prime and chemo-sensitize MLL-rearranged ALL cells toward standard chemotherapy, as well as toward an array of other, mostly FDAapproved compounds. were humanely culled, and systemic leukemic burden was determined using multicolor flow cytometry, as described before [15]. Statistical significance was determined by log-rank testing.

Cell culture
The

High-throughput drug screening
The MLL-rearranged ALL cell lines SEM was pretreated for 14 days with  Table S1.
The decitabine or vehicle pre-treated cell lines were seeded in 384well plates at 10 000 cells/well and treated with 10, 100, or 1000 nM of the compounds using the Sciclone ALH 3000 liquid handling robot (Perkin Elmer). Control samples were treated with DMSO (maximum concentration 0.5% v/v). The cell viability was assessed by a 4-day thiazolyl blue tetrazolium bromide (MTT; Sigma) assay as previously described [19]. The cell viability was normalized to the DMSO controls.
This normalized cell viability of the three concentrations of each compound was used to calculate the area under the curve (AUC) for the compound using GraphPrism. The top hits were defined as drugs with a reduction of more than 30% AUC in the decitabine pretreated SEM cells compared to vehicle-treated cells. Drug synergy between decitabine and the combined compounds was determined using BLISS independence model calculations [20], with the equation: showed an excess over BLISS of >10%, while antagonism was defined if the E combi showed an excess over BLISS of <−10%.

Western blotting
Cell pellets of the cell lines pretreated with decitabine or vehicle were collected at several time points and lysed with RIPA buffer supple-

Decitabine monotherapy mildly attenuates leukemia progression in MLL-rearranged ALL xenografts
Xenograft mouse models still represent the standard for in vivo anti-leukemic drug efficacy testing in MLL-rearranged ALL, as to date bona fide genetic mouse models have not yet been reliably established for this type of leukemia. In order to generate xenografts, we used a previously described reporter cell line, SEM-SLIEW, which is derived from the MLL-AF4-positive B-cell precursor ALL cell line SEM. SEM-SLIEW was modified to express the luciferase reporter gene, allowing for longitudinal in vivo disease monitoring by bioluminescence [15].
Decitabine dose response curves showed comparable sensitivity of SEM-SLIEW and its parental cell line SEM to the drug in vitro ( Figure   S1). Xenografts were established by injecting 10 5 SEM-SLIEW cells intrafemurally into the bone marrow of immunodeficient NSG mice, creating an orthotopic model. Successful engraftment was confirmed by bioluminescence post-transplantation, and mice were divided into a control (n = 13) and treatment group (n = 13). The treatment group was intraperitoneally injected with a low dose of decitabine (0.1 mg/kg), three times a week. The control group was injected with the corresponding vehicle (10% DMSO in saline) ( Figure 1A). One of the therapeutic limitations of decitabine in vivo is its short physiological half-life due to metabolization by liver cytidine deaminases [21]. As previous animal studies have shown that co-administration of tetrahydourine (THU), a cytidine deaminase inhibitor, elevates decitabine plasma levels 10-fold, while revealing no anti-leukemic efficacy in monotherapy, we co-injected decitabine with THU (4 mg/kg in saline) [22][23][24][25][26][27].  Previous reports hinted on the use of decitabine as a chemosensitizer in a variety of cancer types [28][29][30]. Hence, to elucidate whether decitabine would exert chemo-sensitizing effects in MLLrearranged ALL, we next performed high-throughput combinatorial drug screens.

Chemo-sensitizing effect of decitabine toward conventional chemotherapeutics in MLL-rearranged ALL
The chemo-sensitizing capability of decitabine was assessed by performing a combinatorial screen of decitabine with prednisone, asparaginase, cytarabine, daunorubicin, or vincristine, which represent cornerstone drugs in current MLL-rearranged infant ALL treatment [2,3].
Prior to synergy testing, the MLL-rearranged ALL cell lines SEM, ALLPO, and KOPN8 were first pretreated with a low dose of 5 nM decitabine or corresponding vehicle (controls). Since high concentrations of decitabine cause DNA damage by the formation of DNA double strand breaks [31,32], we used a low-dose decitabine similar to others [33] to solely evaluate the demethylating effect of decitabine.
The low dose of decitabine is clinically relevant, since in pediatric patients with acute myeloid leukemia, a dosage of 20 mg/m 2 decitabine is safely achievable [34,35], leading to overall maximal plasma levels of 100 ng/mL or 0.4 μM, which will decrease substantially within 1 h. As DNMT inhibitors typically require several cell divisions to fully exert their demethylating activity, decitabine pretreatment was performed using a prolonged period of exposure of 14 days [36]. Due to the short half-life of decitabine, the drug was refreshed every other day, and passaging of the cells was performed every 4 days for optimal cell growth conditions (Figure 2A).
The demethylating effect of decitabine during 14 days of pretreatment was assessed by monitoring DNMT1 protein expression.
Decitabine represents a deoxycytidine analogue that, like normal deoxycytidines, becomes incorporated into the DNA during replication. Once incorporated, decitabine covalently binds and traps DNA methyltransferases (DNMTs), thereby depleting subsequent daughter cells from functional DNMTs in consecutive cell cycles [37]. Depletion of DNMT1 is commonly used as a reliable read-out for DNA demethylation, and we confirmed that expression of DNMT1 was completely lost after the 14-day pretreatment with low-dose decitabine in all cell lines tested ( Figure 2B). In KOPN8 reduction of DNMT1 expression is seen after 4 days and further reduced in the later timepoints until a total loss after 14 days. In SEM and ALLPO inhibition of DNMT1 expression is evident after 4 days and completely lost after 8 days. In these cell lines, the band for DNMT1 reappears after 12 days, probably due After the 14-day period of decitabine pretreatment, the leukemic cells were subjected to synergy testing, using 5 nM decitabine in combination with prednisone, asparaginase, cytarabine, daunorubicin, F I G U R E 2 Setup of combinatorial synergy screening and effect of decitabine monotherapy on cell lines. Schematic setup in vitro synergy testing of decitabine: the MLL-rearranged ALL cell lines SEM, ALLPO, and KOPN8 were pre-treated for 14 days with 5 nM decitabine or vehicle and subsequently exposed to synergy testing (A), the effect on DNMT1 protein expression was determined using western blotting with β-ACTIN as a loading control (B). Mean percentage of cell viability effects of 14 days of 5 nM decitabine (DEC) pretreatment or vehicle control in MLL-rearranged ALL cell lines SEM, ALLPO, and KOPN8 with standard deviation. Viability was determined using MTT assays. Data has been corrected for the effect of vehicle (C) or vincristine, which are currently used in the treatment of MLLrearranged infant ALL. The in vitro efficacy of each drug combination was assessed by 4-day dose-response curves (MTT assays), normalized to the effects of decitabine as a single agent and analyzed for synergistic, additive, or antagonistic effects by means of the Bliss Independence model [20]. For all three tested cell lines, a mild chemosensitizing effect was observed toward asparaginase in the decitabine treated cells (Figure 3). The combination of decitabine and 15.6 nM cytarabine appeared to have a synergistic effect in ALLPO, while showing an antagonistic effect for the combination of decitabine and 250 nM of cytarabine. Enhanced sensitivity toward all the conventional drugs was evident in KOPN8, although this is not considered as synergy by the Bliss independence model due to the increased effect of decitabine alone reducing cell viability by 70% ( Figure 2C). Taken together we observe decitabine chemo-sensitizes toward the conventional drugs prednisolone, vincristine, daunorubicin, asparaginase, and cytarabine for inhibition of cell survival of MLL rearranged ALL in a limited range of concentrations. Additionally, the observed synergy was not consistent in all of the cell lines tested. These results indicate prolonged exposure to low-dose decitabine is sufficient to completely deplete MLL-rearranged ALL cells from DNTM1, however only mildly sensitizes MLL-rearranged ALL cells toward conventional chemotherapeutics.

Combinatorial high-throughput screening of decitabine with other drug classes
Next, we assessed whether a prolonged pretreatment of lowdose decitabine could sensitize MLL-rearranged ALL cells to other epigenetic-based or anticancer drugs. For this, SEM cells were exposed to a slightly higher concentration of 10 nM decitabine (or vehicle) The combination of the RTK inhibitor foretinib and decitabine decreased cell viability more potently than either drug alone and F I G U R E 3 Long-term low-dose decitabine treatment acts as a poor chemo-sensitizer in MLL-rearranged ALL cells. Chemo-sensitizing effect of decitabine pre-treatment (dec) on current chemotherapeutics. Decitabine pretreated cells were subsequently cultured with additional compounds in the presence or absence (ctrl) of the hypomethylating agent for 4 additional days. Synergy is determined using Bliss independence model. Percentage excess over Bliss (EoB) is indicated: ## EoB > 10%; ### EoB > 20%. Error bars represent the standard error of the mean (SEM). Graphs represent the average of n = 3 independent experiments (n = 2 for KOPN8) indicated moderate synergy according to the Bliss independence model in all three cell lines tested.
Taken together, we showed drug synergy by means of the Bliss Independence model appeared moderate and was evident at limited drug concentrations ( Figure 4B). Therefore, these data show that prolonged DNA demethylation by decitabine hardly sensitizes MLL-rearranged ALL cells to known epigenetic-based or anti-cancer drugs.
Additionally, we investigated whether prolonged low-dose decitabine pretreatment could chemo-sensitize MLL-rearranged ALL cells toward chemotherapeutics currently used in the treatment of this type of leukemia, as well as toward various other epigenetic or anti-neoplastic compounds. Earlier findings revealed that short-term exposure to high-dose decitabine could synergize with cytarabine to eradicate MLL-rearranged ALL cells in vitro [42], as well as with L-asparaginase to enhance cytotoxicity in the pediatric T-ALL [30].
Here, we demonstrate that prolonged exposure to low-dose decitabine occasionally sensitizes MLL-rearranged ALL cells to some of the current chemotherapeutics at certain concentrations in some of the cell lines tested. These observations were most notable for L-asparaginase and cytarabine, thereby confirming the results reported by others [42,30]. However, the in vitro chemo-sensitizing effects of decitabine are modest and therefore clinical relevance may be rather limited. Additionally, since synergy was observed for limited concentration ranges, reaching and maintaining these exact concentration ranges in patients would be challenging due the many factors influencing the pharmacokinetics of the drugs [43][44][45][46].
Furthermore, a recent study in relapsed pediatric ALL, all above 1 year of age, revealed that the combination of decitabine and HDAC inhibitor vorinostat in the current intensive chemotherapy protocol was determined not feasible due to the high incidence of infectious toxicities, despite encouraging response rates and pharmacodynamics [47]. The feasibility and efficacy of demethylating agents for the treatment of MLL-rearranged infant ALL will be further evaluated in currrent clinical trials.
In conclusion, our present study demonstrates that prolonged exposure to a clinically relevant low-dose of the DNA methyltransferase inhibitor decitabine significantly, but mildly delays leukemia progression in MLL-rearranged ALL xenograft mouse models. Moreover, long-term pretreatment with low-dose decitabine moderately sensitizes MLL-rearranged ALL cells toward conventional chemotherapeutics as well as toward known epigenetic-based compounds and antineoplastic agents, in vitro.
For a better understanding of the potential of demethylating agents in the treatment of MLL-rearranged ALL, agents with increased stability and bioavailability could be further evaluated. Ongoing clinical trials should shed more light on theefficacy of demethylating agents for the treatment of MLL-rearranged infant ALL.

ACKNOWLEDGMENTS
We thank the members of the Stam group of the Princess Maxima Center for Pediatric Oncology for their discussions and input.

STATEMENT ON ETHICS
Animal experiments were approved by the institutional Animal Ethical Committee of the Erasmus MC and performed in accordance to Dutch legislation.