Targeting the SUMO Pathway Primes All-trans Retinoic Acid–Induced Differentiation of Nonpromyelocytic Acute Myeloid Leukemias

Acute myeloid leukemias (AML) are a heterogeneous group of severe hematologic malignancies. They arise through the acquisition of oncogenic mutations by hematopoietic stem or progenitor cells. Instead of differentiating into normal blood constituents, leukemic cells are blocked at intermediate differentiation stages and proliferate and infiltrate the bone marrow, leading to the disease (1). Except for the acute promyelocytic leukemia (APL) subtype, the standard treatment of AMLs has changed little over the past 40 years. It generally consists of intensive chemotherapy composed of one anthracycline (daunorubicin or idarubicin) and the nucleoside analogue cytarabine (Ara-C). However, relapses are frequent (40%–70% of patients depending on prognosis factors; ref. 2) and the overall survival very low, making novel treatments urgently needed.

Differentiation therapies have appeared as a powerful strategy for AML treatment. They rely on the idea that restoration of differentiation is associated with cell division arrest, followed by death due to the naturally limited lifespan of differentiated cells. This approach has proved particularly efficient at curing APL (3), an AML subtype characterized by the expression of oncogenic fusion proteins engaging the retinoic acid receptor α (RARα), which belongs to the nuclear receptor family. APL therapy is based on pharmacologic doses of its natural ligand, all-trans-retinoic acid (ATRA), used in combination with arsenic trioxide (4). ATRA leads to the degradation of the oncogenic RARα fusion protein and activates wild-type RARα. This activates a specific transcriptional program, which drives differentiation, cell-cycle arrest, and death of the leukemic cells (5, 6). ATRA also induces to various degrees the in vitro differentiation of certain non-APL AML cell lines and primary cells from a substantial number of non-APL AMLs (6). This includes AMLs mutated in NPM1 (7), IDH1/IDH2 (8), or FLT3/ITD genes (9) or overexpressing the transcription factor EVI-1 (10). However, the clinical trials conducted so far have failed to prove a significant efficacy of ATRA on patients with non-APL AML, including in combination with other drugs (11). This lack of effect has been attributed to the inability of ATRA to induce the expression of critical genes involved in differentiation, cell-cycle arrest, or apoptosis (6, 11). Interestingly, targeting epigenetic enzymes has recently appeared promising to restore the ability of ATRA to activate RARα target genes and potentiate ATRA-induced differentiation of non-APL AMLs (6, 11). For instance, inhibition of histone deacetylases (HDAC) with valproic acid was shown to favor ATRA-induced differentiation of non-APL AML cells (12). However, the clinical trials conducted so far showed only limited effects in a subset of the patients treated with the combination of ATRA and valproic acid (13–15). More recently, inhibition of the histone demethylase LSD1/KDM1A was also shown to strongly sensitize AML cells to ATRA in preclinical models via transcriptional reprogramming (16). Phase I/II trials are ongoing to determine the clinical efficacy of the combination between ATRA and LSD1 inhibitors.

SUMO is a group of three (SUMO-1–3) ubiquitin-related polypeptidic posttranslational modifiers covalently and reversibly conjugated to numerous intracellular proteins to regulate their function and fate (17). Its conjugation involves a unique E1 SUMO-activating enzyme, a unique E2-conjugating enzyme (Ubc9), and several E3 SUMO ligases (18). Deconjugation is ensured by deSUMOylases, in particular of the SENP family. Increasing evidence links deregulation of the SUMO pathway to cancer (19, 20), including in hematologic malignancies such as lymphomas (21) and multiple myeloma (22). In the case of AMLs, the SUMO pathway is essential for efficient differentiation therapy of APLs by an ATRA + arsenic trioxide combination treatment. Arsenic trioxide induces the rapid SUMOylation of the PML/RARα oncoprotein, which initiates its elimination by the ubiquitin/proteasome system (23–25). In addition, we have shown that inhibition of the SUMO pathway by genotoxics-induced reactive oxygen species (ROS) is essential for fast and efficient cell death of chemosensitive non-APL AML cells subjected to anthracyclines or Ara-C treatment (26).

SUMO is increasingly viewed as an epigenetic mark highly enriched at gene promoters (27–29). It regulates gene expression via modification of numerous transcription factors/coregulators, histone-modifying enzymes, RNA polymerases, and even histones (30, 31). Although sometimes associated with transcriptional activation (32), SUMOylation at gene promoters is mostly known to limit or repress transcription (27, 31, 33–36). In particular, SUMOylation facilitates the recruitment of SUMO-interacting motifs (SIM) containing corepressors on promoters (30, 37, 38). We show here that SUMOylation takes part in the epigenetic silencing of ATRA-responsive genes in non-APL AMLs. Its inhibition activates the prodifferentiating and antileukemic effect of ATRA in these cancers, which opens new perspectives in the treatment of this poor-prognosis cancer.

U937, HL60, and THP1 cell lines were authenticated by the ATCC using short tandem repeat analysis. MOLM14 cells were obtained directly from the ATCC. All cells were regularly tested negative for Mycoplasma. They were cultured in RPMI containing 10% fetal bovine serum (FBS) and streptomycin/penicillin at 37°C in the presence of 5% CO₂. After thawing, cells were passaged at 0.3.10⁶/mL every 2 to 3 days for no more than 10 passages. U937 cells resistant to Ara-C were generated by culturing the cells for 2 months in the presence of increasing Ara-C concentrations (up to 0.1 μmol/L). Patient bone marrow aspirates were collected after obtaining written informed consent from patients under the frame of the Declaration of Helsinki and after approval by the Institutional Review Board (Ethical Committee “Sud Méditerranée 1,” ref 2013-A00260-45, HemoDiag collection). Fresh leukocytes were purified using density-based centrifugation using Histopaque 1077 from Sigma-Aldrich and resuspended at a concentration of 10⁶/mL in IMDM (Sigma-Aldrich) complemented with 1.5 mg/mL bovine serum albumin, 4.4 μg/mL insulin (Sigma-Aldrich), 60 μg/mL transferrin (Sigma-Aldrich), 5% streptomycin + penicillin, 5% FBS, 5 μM β-mercaptoethanol, 1 mmol/L pyruvate, 1x MEM nonessential amino acids (Life Technologies), 10 ng/mL IL3 (PeproTech), 40 ng/mL SCF (PeproTech), and 10 ng/mL TPO (PeproTech).

ATRA was from Sigma-Aldrich. It was resuspended at a 100-mmol/L concentration in DMSO and stored at −20°C for a maximum of 2 weeks. Anacardic acid (AA) was from Santa Cruz Biotechnology and 2-D08 from Merck Millipore. Anti–SUMO-1 (21C7) and SUMO-2 (8A2) hybridomas were from the Developmental Studies Hybridoma Bank. The anti-H3K4Me₃ antiserum was from Abcam and the Ubc9 antibody from Santa Cruz Biotechnology.

Cells were washed in PBS containing 2% FBS and incubated at 4°C for 30 minutes in the presence of the following fluorophore-conjugated antibodies: CD45-Pacific Blue (A74763; Beckman Coulter), CD14-PE (130-091-242; Miltenyi Biotec), CD15-PE-Vio770 (130-100-425; Miltenyi Biotec), and CD11b-APC (130-109-286; Miltenyi Biotec). Matched isotype controls were used for each treatment condition. After washing, cells were analyzed using the LSR Fortessa flow cytometer (Becton Dickinson) and FACSDiva software. Data were analyzed using FlowJo software (version 10). For patient samples, median fluorescent intensities (MFI) for each differentiation marker were measured on leukemic cells previously selected using CD45/SSC gating (39). MFIs from isotype controls were subtracted from each treatment condition.

Cell lines or patient samples were cytospun on microscope slides (1,500 rpm for 5 minutes), dried for 5 minutes, and stained, first with May–Grunwald (5 minutes) and then, Giemsa (1/10 dilution, 15 minutes) stain (MGG) staining. Microscopic examinations were performed using the Axio Imager Z2 microscope (Zeiss).

For proliferation assays on cell lines, cells were seeded at a concentration of 3 × 10⁵/mL, and viable cells were counted at regular intervals using the Trypan blue exclusion method with an EVE automatic cell counter. Cell-cycle distribution was analyzed by propidium iodine (PI) staining. Cells were washed once with PBS and fixed with cold 70% ethanol for 10 minutes and washed once with PBS. RNAase A (100 μg/mL; Sigma-Aldrich) was then added for 10 minutes at room temperature. Cells were washed with PBS and stained with 50 μg/mL PI (Sigma-Aldrich) for 10 minutes at room temperature. Cells were then washed once with PBS and analyzed by flow cytometry. For patient cells, equal numbers of CountBright absolute counting beads (C36950; Life Technologies) were added to each sample. Viable cells were selected using the CD45/SSC gating, and their number was normalized to the number of beads counted in the same sample.

Retroviral constructs expressing either Ubc9, SENP2, or SENP5 were constructed by inserting human cDNA using the Gateway cloning technology (Thermo Fisher Scientific) into the pMIG retroviral vector (40), which also coexpresses EGFP from the same polycistronic mRNA. Retroviruses were produced by cotransfection of these constructs with gag-pol and VSV-G expression vectors into HEK293T cells using Lipofectamine 2000 (Invitrogen). Viral supernatants were collected 48 hours later, 0.45 μm-filtered, and directly used to infect AML cell lines. Only EGFP-positive cells were considered in flow cytometry analyses. Where indicated, the EGFP-positive cells were sorted using the FACSAria cell sorter (Becton Dickinson).

Total mRNA was purified using the GenElute Mammalian Total RNA kit (Sigma-Aldrich). After DNase I treatment, 1 μg of total RNA was used for cDNA synthesis using the Maxima First Strand cDNA kit (Thermo Fisher Scientific). qPCR assays were conducted using Taq platinum (Invitrogen) and the LightCycler 480 device (Roche) with specific DNA primers (see Table 1). Data were normalized to the housekeeping TBP mRNA levels.

A total of 30.10⁶ cells were cross-linked with 1% paraformaldehyde for 8 minutes. Paraformaldehyde was then neutralized with 125 mmol/L glycine for 10 minutes. Cross-linked cells were washed with cold PBS, resuspended in a cell lysis buffer (5 mmol/L PIPES pH 7.5, 85 mmol/L KCl, 0.5% NP40, 20 mmol/L N-ethyl maleimide, 1 μg/mL of aprotinin, pepstatin, leupeptin, 1 mmol/L AEBSF) and incubated at 4°C for 10 minutes with rotation. Nuclei were centrifuged (5,000 rpm for 10 minutes at 4°C) and resuspended in a nucleus lysis buffer (50 mmol/L Tris-HCl pH 7.5, 1% SDS, 10 mmol/L EDTA, 20 mmol/L N-ethyl maleimide, 1 μg/mL of aprotinin, pepstatin, leupeptin, 1 mmol/L AEBSF) and incubated at 4°C for 2.5 hours. Lysates were then sonicated for 30 cycles of 30 seconds each at 4°C using the Bioruptor Pico (Diagenode) under standard conditions. After sonication, samples were centrifuged (13,000 rpm for 10 minutes at 4°C), and the supernatants were diluted 10-fold in the immunoprecipitation buffer (1.1% Triton X100, 50 mmol/L Tris-HCl pH 7.5, 167 mmol/L NaCl, 5 mmol/L N-ethyl maleimide, 1 mmol/L EDTA, 0.01% SDS, 1 μg/mL of aprotinin, pepstatin, leupeptin, 1 mmol/L AEBSF) with 2 μg of antibodies and Dynabeads Protein G (Thermo Fisher Scientific). Immunoprecipitations were performed overnight at 4°C. Beads were then washed in low-salt buffer (50 mmol/L Tris-HCl pH 7.5, 150 mmol/L NaCl, 1% Triton X100, 0.1% SDS, 1 mmol/L EDTA), high-salt buffer (50 mmol/L Tris-HCl pH 7.5, 500 mmol/L NaCl, 1% Triton X100, 0.1% SDS, 1 mmol/L EDTA), LiCl salt (20 mmol/L Tris-HCl pH 7.5, 250 mmol/L LiCl, 1% NP40, 1% deoxycholic acid, 1 mmol/L EDTA), and TE buffer (10 mmol/L Tris-HCl pH 7.5, 0.2% Tween20, 1 mmol/L EDTA). Elution was done in 200 μL of 100 mmol/L NaHCO₃, 1% SDS. Chromatin cross-linking was reversed by overnight incubation at 65°C with 280 mmol/L NaCl followed and 2 hours at 45°C with 35 mmol/L Tris-HCl pH6.8, 9 mmol/L EDTA, 88 μg/mL RNAse, and 88 μg/mL proteinase K. Immunoprecipitated DNAs were purified using the NucleoSpin Gel and PCR Clean-up Kit (Macherey-Nagel). Immunoprecipitated DNA and DNA inputs purified from samples before immunoprecipitation were subjected to PCR analysis using the Roche LightCycler 480 device using appropriate primers (see Table 1).

Animals were used in accordance to a protocol reviewed and approved by the Institutional Animal Care and Use Committee of Région Midi-Pyrénées (France). Tumors were generated by injecting subcutaneously 2 × 10⁶ U937 cells (contained in 100 μL of PBS) on both flanks of NOD-Scid-IL2gRnull (NSG) mice (adult males and females; 25 g each; Charles River Laboratories). When tumors reached 100 mm³, mice received peritumoral injections of ATRA (2.5 mg/kg/day) or 2-D08 (10 mg/kg/day), or both every 2 days. Injections of vehicle (DMSO) were used as controls. Tumor sizes were measured with a caliper and volumes calculated using the formula: |$\nu = {\frac{\pi }{{6\ \times A \times {B^2}}}}}$|⁠, where A is the larger diameter and B is the smaller diameter.

Statistical analyses were performed using Prism 5 software. The two-tailed paired Student t test, the Wilcoxon matched-pairs signed rank test, and the Mann–Whitney test were used for analysis of experiments with cell lines, patient samples, and xenografts, respectively. Differences were considered as significant for P values of <0.05. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. ns, nonsignificant.

To assess the role of SUMOylation in ATRA-induced differentiation of non-APL AMLs, we first resorted to the reference and well-characterized U937 and HL-60 non-APL AML cell lines. They belong to the M4 and M2 subtypes, respectively [of eight AML subtypes of the French–American–British (FAB) classification] and can differentiate in vitro in the presence of ATRA, albeit at low efficiency. We used two commercially available pharmacologic inhibitors of the SUMO pathway: 2-D08, an inhibitor of the E2 SUMO-conjugating enzyme Ubc9 (41, 42), and AA, an inhibitor of the SUMO-activating enzyme Uba2/Aos1 (43). Both inhibitors were used at concentrations leading to moderate hypoSUMOylation of cell proteins (Fig. 1A). This partial inhibition of the SUMO pathway resulted in a significant increase in ATRA-induced differentiation of both U937 (Fig. 1B) and HL60 (Fig. 1C) cells, as assessed by enhanced expression of CD15 or CD11b differentiation markers, respectively. Combining ATRA with 2-D08 also increased the differentiation of THP1 (Fig. 1D), MOLM14 (Fig. 1E), and Ara-C–resistant U937 cells developed in our laboratory (Fig. 1F). The combination of ATRA and SUMOylation inhibitors increased the number of U937 (Fig. 2A), HL-60 (Fig. 2B), and THP1 (Supplementary Fig. S1) cells showing morphologic changes typical of terminal myeloid differentiation, such as nuclear lobulation and appearance of numerous cytosolic granules. Altogether, these data suggest that the SUMO pathway limits the differentiating effects of ATRA on various non-APL AML cell lines and that its inhibition could favor their ATRA-induced differentiation, including in the case of resistance to chemotherapeutics used in AML treatment.

We then wondered if the repressive action of SUMOylation on ATRA-induced differentiation could be linked to its well-characterized ability to repress/limit gene expression (37). Inhibition of SUMOylation with 2-D08 was sufficient on its own to increase the expression of various ATRA-responsive genes associated with myeloid differentiation, such as RARA, CEBPA, TNFSF10, ITGAX, ITGAM, and IL1B (Fig. 3A). This suggests that inhibition of SUMOylation might prime differentiation by increasing the basal expression of master genes involved in myeloid differentiation. In addition to basal expression, 2-D08 also increased the ATRA-induced expression of most of those genes (Fig. 3A). To determine if SUMOylation controls the expression of these genes at the level of chromatin, we assayed the active transcription-associated histone mark H3K4me₃ on their promoters. An increase of this mark correlated with the level of expression of RARA, ITGAX, CEBPA, and TNFSF10 mRNAs (Fig. 3B), suggesting that SUMOylation represses ATRA-responsive gene induction at the level of chromatin.

Differentiated cells stop proliferating and have a shorter lifespan than undifferentiated cells. Consistently, ATRA and 2-D08 synergized to block the proliferation of both U937 cells (Fig. 4A) and their Ara-C–resistant variants in vitro (Fig. 4B). This correlated with an accumulation of cells in G₀–G₁ (Fig. 4C) and a strong activation of the CDKN1A gene encoding the CDK inhibitor p21CIP1 (Fig. 4D). This suggests that inhibition of SUMOylation increases the antiproliferative effects of ATRA in vitro by inducing cell-cycle arrest. To determine if this would also be the case in vivo, we treated immunodeficient mice subcutaneously xenografted with U937 cells with ATRA, 2-D08, or ATRA + 2-D08 after engraftment. Only the combination of ATRA and 2-D08 induced a significant reduction in tumor growth, whereas ATRA and 2-D08 alone showed slight, if any, effects (Fig. 4E and F). Thus, the combination of ATRA with an inhibitor of SUMOylation can not only promote non-APL AML cell differentiation but also exert antiproliferative effects both in vitro and in vivo.

To rule out possible off-target effects of 2-D08 and AA, we resorted to the genetic manipulation of the SUMO pathway to confirm the role of SUMOylation on ATRA-induced differentiation in non-APL AMLs. In a first step, we overexpressed either the SENP-2 or the SENP-5 desumoylase in U937 cells (Supplementary Fig. S2A). This led to a significant increase in their ATRA-induced differentiation, as assayed by the expression of CD11b (Fig. 5A). This also correlated with a reduction in their proliferation (Fig. 5B). Overexpression of SENP-2 also increased ATRA-induced differentiation of HL-60 cells (Fig. 5C). Interestingly, it strongly decreased SUMOylation, in particular by SUMO-2, of chromatin-bound proteins, as analyzed by chromatin immunoprecipitation (ChIP)–qPCR on the promoter of the RARA gene (Supplementary Fig. S2B). This was associated with stronger expression of RARA as well as other ATRA-responsive genes (CEBPA, ITGAM, and IL1B; Fig. 5D). Although SENP-2–expressing HL60 cells did not appear more differentiated than control cells in the absence of ATRA, they showed higher basal expression of these genes. This further supported the idea that inhibition of SUMOylation could prime AML cells for differentiation. In a second step, we increased SUMOylation in THP1 cells via overexpression of the SUMO E2-conjugating enzyme Ubc9 (Fig. 6A). This led to a massive decrease in ATRA-induced differentiation, as assayed by CD14 expression (Fig. 6B), and morphologic changes (appearance of pseudopods, cytosolic vesicles, and granules; Fig. 6C). Moreover, Ubc9 overexpression strongly decreased basal expression and/or induction of genes involved in myeloid differentiation by ATRA (Fig. 6D). Altogether, this confirmed the repressive role of the SUMO pathway on ATRA-induced differentiation of non-APL AMLs.

We finally asked whether inhibiting SUMOylation could also favor the in vitro differentiation of primary AML cells using bone marrow aspirates from patients at diagnosis (see Supplementary Table S1). ATRA alone did not significantly induce primary AML cell differentiation, as assayed by CD15 expression. 2-D08 and AA alone showed a slight prodifferentiating trend, however, with no statistical significance. In contrast, combining either with ATRA significantly increased CD15 expression compared with cells treated with ATRA alone (Fig. 7A). Some patient cells were more sensitive to the differentiating effects of the ATRA + SUMOylation inhibitor combination than others, but they neither belonged to a unique FAB subtype nor shared cytogenetic/genetic abnormalities tested at diagnosis (Supplementary Table S1 and Supplementary Fig. S3). Inhibitors of SUMOylation also increased the number of cells showing morphologic changes typical of differentiation, such as nuclear lobulation, cytosol enlargement, or appearance of cytosolic granules on cells from the two patients tested (Fig. 7B and Supplementary Fig. S4). Interestingly, 2-D08 and AA also potentiated ATRA-induced differentiation of primary cells from 1 patient not responsive to induction chemotherapy (#16136), as well as from 2 (#16173 and #16188) out of 3 patients at relapse (Fig. 7C and Supplementary Table S1). Importantly, both inhibitors increased the antileukemic activity of ATRA on the primary AML cells taken both at diagnosis and relapse (Fig. 7D and Supplementary Fig. S5). Altogether, these data confirmed that SUMOylation inhibitors potentiate ATRA-induced differentiation of different non-APL AML subtypes and indicated novel therapeutic approach, including in the case of conventional chemotherapy failure.

Differentiation therapies using ATRA have transformed APL from a fatal to a highly curable disease. Unfortunately, their efficacy in other AML subtypes has been disappointing. Our work demonstrates that SUMOylation represses ATRA-induced differentiation of non-APL AMLs. Mechanistically, the SUMO pathway limits the expression of genes critical for myeloid differentiation (RARA and CEBPA), as well as genes required for cell-cycle arrest (CDKN1A) and retinoid-induced apoptosis (IL1B and TNFSF10). Inhibition of SUMOylation increases the presence of H3K4me3, a mark of active transcription, on their promoters and enhances both their basal and ATRA-induced expression. This suggests that targeting SUMOylation could prime AML cells for ATRA-induced differentiation, notably by augmenting the expression of critical regulators of myeloid differentiation. The ability of retinoids to induce cell-cycle arrest, in particular through transcriptional regulation of various cell-cycle regulators, is critical for their therapeutic effects (6, 44). In addition, other molecules inducing cell-cycle arrest, such as the HDAC inhibitor valproic acid (12, 45), also enhance ATRA-induced differentiation of non-APL AMLs. Thus, induction of cell-cycle arrest by the combination of ATRA with SUMO inhibitors might play an important role in prodifferentiating and antileukemic effects of these drugs.

More than 6,000 SUMOylated proteins have been identified so far (17, 46). Among them are many transcription factors and coregulators, some of which play key roles in myeloid differentiation. This is the case of CEBPα and CEBPϵ. Their SUMOylation was shown to repress and activate their transactivation capacities, respectively (47, 48). The ATRA receptor RARα also undergoes dynamic SUMOylation/deSUMOylation cycles essential for its ATRA-induced activation (49, 50). The SUMOylation of these transcription factors could participate in the silencing of ATRA-responsive genes that we have uncovered in non-APL AMLs. However, transcriptional repression of differentiation-associated genes by the SUMO pathway might not just result from SUMOylation of single transcription factors but from coordinated SUMOylation of multiple proteins bound to regulatory elements in the promoter/enhancers of these genes. Supporting the latter possibility, the concept of “group SUMOylation,” which has originally emerged from DNA repair regulation studies (51), implies that SUMO can exert its function wherever it is conjugated within a complex comprising several SUMOylatable proteins. SUMOylation of promoter-bound complexes are thought to favor, through SUMO/SIM interactions, the recruitment of transcriptional repressors (52) such as the N-Cor/HDAC (53) or the CoRest/LSD1 (54) complexes. Inhibition of SUMO conjugation would lead to a decrease in their recruitment on the chromatin and would in fine facilitate both basal and ATRA-induced expression of genes involved in myeloid differentiation.

Our work indicates that combining ATRA with pharmacologic inhibitors of the SUMO pathway potentiates the prodifferentiating and antiproliferative/proapoptotic effects of ATRA on non-APL AMLs, including those resistant to the genotoxics currently used in the clinic. Considering the multiplicity of essential pathways controlled by SUMO, complete inhibition of SUMOylation would likely be too toxic in patients. However, similar to experiments with AA we conducted in a previous work (26), we observed no general toxicity of 2-D08 in mice. Both 2-D08 and AA are poorly efficient inhibitors of SUMOylation and induced only a slight decrease in SUMO conjugation at the doses we used. This probably explains their low toxicity. Finally, hemizygous mice expressing 50% of Ubc9 and showing slightly reduced SUMOylation activity are viable with no overt phenotype (55). This further suggests that limited reduction in cellular SUMOylation is not detrimental to essential cellular functions. Controlled inhibition of the SUMO pathway could thus, in combination with ATRA, target leukemic cells without overly affecting normal cells. In our in vivo experiments, the ATRA + 2-D08 combination only reduced AML growth in vivo but did not completely block it. Such limited effects could therefore have no clinical relevance. However, 2-D08, similar to AA, is characterized by a poor bioavailability linked to its hydrophobic nature, which could explain the limited effects observed in vivo. Novel, more soluble SUMOylation inhibitors, or improvement of the pharmacologic properties of the existing ones, are therefore necessary. ML-792 has been recently described as a novel inhibitor of SUMOylation (56). However, its pharmacologic properties and bioavailability have not been tested in vivo. Should they be better than those of 2-D08 and AA, this would permit an in-depth assessment of the therapeutic benefit of the ATRA + SUMOylation inhibitor association and, in fine, clinical use. In conclusion, our work suggests that targeting the SUMO pathway is a promising strategy to enhance the clinical efficacy of ATRA in non-APL AML and improve the treatment of this poor-prognosis cancer.

No potential conflicts of interest were disclosed.

Conception and design: H. Baik, M. Boulanger, J.E. Sarry, M. Piechaczyk, G. Bossis

Development of methodology: H. Baik, J. Kowalczyk, T. Salem, J.E. Sarry

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Boulanger, M. Hosseini, S. Zaghdoudi, Y. Hicheri, G. Cartron

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Baik, M. Boulanger, J. Kowalczyk, G. Bossis

Writing, review, and/or revision of the manuscript: Y. Hicheri, G. Cartron, G. Cartron, M. Piechaczyk, G. Bossis

Study supervision: G. Bossis

The authors are grateful to all members of the “Oncogenesis and Immunotherapy” group of IGMM for support and fruitful discussions and to Pierre Gâtel, Rosa Paolillo, and Frédérique Brockly for technical help. The authors thank Dr. Robert Hipskind for critical reading of the manuscript. Funding was provided by the CNRS, Ligue Nationale contre le Cancer (Programme Equipe Labellisée), FRM (Contract FDT20160435412), Cancéropole GSO (Programme Emergence), Région Languedoc-Roussillon (Contract Chercheur d’Avenir), INCA (ROSAML and METAML), Association Laurette Fugain (contract ALF-2017/02), the EpiGenMed Labex in the frame of the program “Investissement d’avenir” (ANR-10-LABX-12-01), and the ITMO Cancer in the frame of the Plan Cancer. The collection of clinical data and samples (HEMODIAG_2020) at the CHU Montpellier was supported by funding from the Région Languedoc-Roussillon and the SIRIC Montpellier Cancer.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.