The human leukemic oncogene MLL-AF4 promotes hyperplastic growth of hematopoietic tissues in Drosophila larvae

Summary MLL-rearranged (MLL-r) leukemias are among the leukemic subtypes with poorest survival, and treatment options have barely improved over the last decades. Despite increasing molecular understanding of the mechanisms behind these hematopoietic malignancies, this knowledge has had poor translation into the clinic. Here, we report a Drosophila melanogaster model system to explore the pathways affected in MLL-r leukemia. We show that expression of the human leukemic oncogene MLL-AF4 in the Drosophila hematopoietic system resulted in increased levels of circulating hemocytes and an enlargement of the larval hematopoietic organ, the lymph gland. Strikingly, depletion of Drosophila orthologs of known interactors of MLL-AF4, such as DOT1L, rescued the leukemic phenotype. In agreement, treatment with small-molecule inhibitors of DOT1L also prevented the MLL-AF4-induced leukemia-like phenotype. Taken together, this model provides an in vivo system to unravel the genetic interactors involved in leukemogenesis and offers a system for improved biological understanding of MLL-r leukemia.

7][28] The MLL-AF4 fusion protein is thought to drive a feedforward loop resulting in high activity of the histone H3 lysine 79 methyltransferase DOT1L, culminating in aberrant transcription of target genes. 23MLL-AF4 target genes include HOXA9, which has been shown to be dependent on its cofactors MEIS1 and PBX3 to induce leukemic transformation. 29MEIS1 and PBX3 are themselves also well-known MLL fusion protein targets. 30,31Even in the presence of this molecular understanding of the impact of MLL fusion proteins, therapies developed toward these targets have not yet successfully entered clinical practice.In this context, additional genetically tractable animal models that allow rapid and unbiased in vivo genetic analysis and biological understanding could complement studies in existing mouse models.
Drosophila has a long tradition as a system to understand gene function and to model disease, including cancer and leukemia. 32rosophila hematopoiesis has several key similarities to mammalian hematopoiesis as it is governed by conserved transcriptional programs that result in generation of hemocytes with functional similarity to human blood cells. 33Fly immune cells reside in specific compartments where they interact with the microenvironment. 34Briefly, hematopoiesis in Drosophila occurs in two waves, first in the embryo and later in the larval hematopoietic organ, the lymph gland. 35,36The lymph gland consists of two primary (anterior) and several secondary (posterior) lobes.The primary lobe can be divided into the progenitor-containing medullary zone, the cortical zone containing differentiated hemocytes, and the posterior signaling center (PSC), which is involved in regulating the balance between progenitors and differentiated hemocytes. 37raditionally, hematopoietic differentiation from prohemocytes has been known to result in three main types of hemocytes: plasmatocytes, crystal cells, and lamellocytes.The macrophage-like plasmatocytes make up the majority of hemocytes (around 95% of the hemocyte population) with a smaller fraction of crystal cells (around 5%) involved in melanization and wound healing. 38Finally, large, actin-rich lamellocytes develop upon specific insults, such as the presence of parasitic wasp eggs, and are not normally observed in healthy larvae. 39,40Recently, novel single-cell sequencing data have unraveled a more diverse range of hematopoietic cell subtypes, including intermediate stages. 41,42urthermore, hemocyte development is not limited to the linear development from prohemocytes as a common precursor.There are also mechanisms of transdifferentiation, where plasmatocytes can develop into lamellocytes, underpinning the plasticity and intricacy of the Drosophila hematopoietic system. 43otably, the MLL protein was first identified as Trithorax (Trx) in Drosophila melanogaster, 44 and orthologs of all the main MLL fusion partners are found in Drosophila.Human full-length MLL was previously found to partly rescue a trx mutant.Interestingly, expression of human MLL-AF4 or MLL-AF9 in Drosophila larval brain affected cell-cycle progression, whereas ubiquitous expression was lethal. 45Furthermore, expression of human MLL-AF10 was found to deregulate the activity of Polycomb group-responsive elements in a Drosophila adult eye reporter system, 46 indicating that these human leukemic fusion proteins are active also in Drosophila.
In this study, we exploited the conservation of the MLL complex and hematopoiesis in D. melanogaster to gain insight into the hematopoietic malignancies that occur upon chromosomal MLL rearrangements.We show that expression of human MLL-AF4 in the Drosophila hematopoietic system leads to increased levels of differentiated hemocytes, resulting in enlarged lymph glands and enhanced numbers of circulating hemocytes.These phenotypes were dependent on the main complex partners that have also been identified in human leukemia and could be rescued by genetic or chemical targeting of DOT1L.

Expression of the human oncogene MLL-AF4 induces a leukemia-like phenotype in flies
Due to the conserved function of the equivalent of the human MLL complex in Drosophila melanogaster, the trithorax complex, we hypothesized that expressing the human oncogene MLL-AF4 in the hematopoietic system of D. melanogaster could induce a leukemic phenotype.We expressed either human full-length MLL or the human fusion protein MLL-AF4 in the larval hematopoietic system using the pxn-Gal4 driver and co-expressing GFP (pxn-Gal4>UAS-GFP).The pxn-Gal4 driver was chosen because peroxidasin is expressed in the cortical zone of the lymph gland from the second instar larval stage and onwards in hemocytes that mature into plasmatocytes and crystal cells. 36,47e assessed the overall effects of MLL-AF4 expression on the lymph gland by measuring lymph gland size changes.Furthermore, we evaluated the effects of MLL-AF4 expression on the cortical zone by quantification of the GFP+ volume fraction of the lymph gland, to identify whether this compartment changed in size relative to the overall lymph gland.Interestingly, while the lymph glands in larvae expressing fulllength, normal MLL in the cortical zone were similar to wild-type (WT) lymph glands, expression of the MLL-AF4 oncogene induced significant lymph gland enlargement and an increase in GFP+ hemocytes in the lymph gland compared to WT larvae (Figures 1A-1G).Furthermore, lymph glands expressing MLL-AF4 in the cortical zone displayed increased levels of GFP volume per total lymph gland volume (Figure 1H).Flow cytometry of dissociated lymph glands also showed an increase in the total number of GFP+ hemocytes per lymph gland as well as an increase in the percentage of GFP+ hemocytes per lymph gland (Figures 1I and 1J), confirming that these lymph glands are hyperplastic.Lymph glands expressing either the MLL N-terminal or the AF4 C-terminal part of the MLL-AF4 fusion protein in the cortical zone were similar to controls (Figures S1A-S1F).We verified by qRT-PCR that the MLL N-terminal or the AF4 C-terminal portions were expressed to similar levels as the MLL-AF4 fusion protein (Figures S1G-S1I).We also found that simultaneous expression of two independent upstream activating sequence (UAS)-driven MLL-AF4 transgenes induced even larger lymph glands relative to a single MLL-AF4 transgene, bolstering the causal effect of MLL-AF4 expression on the hyperplasia phenotype (Figures S1J-S1L).However, larvae expressing MLL-AF4 were still able to develop into adult flies similar to the controls (Figure S1M).To investigate whether the increase in lymph gland size was a result of increased proliferation, we stained lymph glands from L2, feeding L3 and wandering L3 stages for the proliferative marker phosphorylated histone 3 serine 10 (pH3Ser10).Indeed we found that the lymph gland overgrowth could be explained by increased proliferation in MLL-AF4-expressing lymph glands at the L2 larval stage, whereas there was no difference in proliferation in lymph glands from feeding and wandering L3 larvae (Figures 1K-1Q).
Given that pxn-Gal4 mainly drives MLL-AF4 expression in differentiating hemocytes, 36 we determined whether its expression affected hemocyte maturation.The observed increase of GFP+ hemocytes was identified as plasmatocytes as shown by staining the primary lobes for NimC1/P1 48 (Figures S2A-S2D).Furthermore, crystal cell levels were increased upon MLL-AF4 expression as evaluated by staining for Hindsight (Hnt) 49 (Figures S2E-S2H).Finally, lamellocytes were occasionally observed in MLL-AF4-expressing lymph glands (Figures S2I-S2K).We also assessed the effect of cortical zone expression of MLL or MLL-AF4 on the PSC, which can be identified by antp-positive cells.The number of antp-positive cells was reduced upon expression of either MLL or MLL-AF4 (Figure S2L), suggesting that this was independent of the lymph gland hyperplasia induced by MLL-AF4 expression specifically.
To further understand how MLL-AF4 affects the larval hematopoietic system, we investigated the effect of expressing MLL-AF4 in circulating hemocytes using the hml-Gal4 driver and co-expressing GFP (hml-Gal4>UAS-GFP). 50Whereas wL3 larvae expressing full-length MLL had normal levels of circulating hemocytes, MLL-AF4 expression resulted in a significant increase in circulating hemocyte levels (Figures 2A-2D).In L2 larvae the number of circulating hemocytes was similar to that of controls, indicating that the expansion occurs after this stage (Figures S3A-S3D).We also noticed that in larvae expressing MLL-AF4, lamellocytes were present in the hemolymph, again indicative of aberrant hemocyte differentiation (Figures 2A-2C and 2E).To test whether MLL-AF4-expressing hemocytes were still able to perform their phagocytic function, hemocytes were exposed to E. coli beads that fluoresce only at lower pH, indicative of internalization by phagocytosis.Larvae with hemocytes expressing either MLL or MLL-AF4 had similar levels of phagocytic cells compared to the control (Figure S3E), and these cells were performing phagocytosis at similar levels (Figure S3F).Taken together, these results show that MLL-AF4 induces hyperplasia of hematopoietic tissues in Drosophila primarily through an increase in plasmatocytes.In addition, MLL-AF4 expression affects hemocyte fate since it promotes the development of lamellocytes.

Knockdown of MLL-AF4 complex partners rescues the leukemic phenotype
Next, we tested whether the observed leukemia-like phenotypes depend on fly orthologs of the known complex partners of the human MLL-AF4 fusion protein (see Figure 3A for a schematic overview).To this end, we used RNAi-mediated knockdown of these orthologs.For MLL interaction partners we depleted Mnn1 (the Drosophila ortholog of MEN1) or Jasper (the Drosophila orthologue of LEDGF/PSIP1), since See also Figure S3.
the human orthologs are required for MLL to interact with chromatin (Figure 3A).For interaction partners of human AF4, we targeted Su(Tpl) (the Drosophila ortholog of ELL) and ear (the Drosophila ortholog of ENL/MLLT1 and AF9/MLLT3), which are part of the transcriptional super elongation complex (SEC) 51 (Figure 3A).Knockdown efficiency was evaluated by qRT-PCR in hemocytes for several available RNAi lines against each of these targets (Table S1; Figures S4A-S4D), and a subset of lines producing efficient knockdown was selected for phenotypic analysis.Pxn-driven depletion of most of these gene products in the cortical zone of otherwise WT lymph glands did not result in obvious changes in lymph gland size or morphology (Figures S4E-S4L).The exception was depletion of Mnn1, where lymph glands presented with larger secondary lobes than those in the WT, and the overall lymph gland size was smaller than that of WT (Figures S4G and S4L).Interestingly, in lymph glands expressing MLL-AF4 in the cortical zone, concurrent depletion of Mnn1, Jasper, ear, or Su(Tpl) resulted in a reduction in overall lymph gland size (Figures 3B-3L).Taken together, this shows that the observed leukemia-like phenotype of MLL-AF4 depends on the same conserved interaction partners as in human leukemia.The leukemic phenotype of the Drosophila MLL-AF4 model depends on the DOT1L fly homolog gpp In MLL-r leukemia, the histone methyltransferase DOT1L has been shown to be associated with the MLL fusion protein complex, and DOT1L is required for leukemogenesis in different cellular and organismal model systems. 52,535][56] To test whether DOT1L is required for the MLL-AF4 phenotype in Drosophila, we first depleted the Drosophila DOT1L homolog gpp 57 in lymph glands expressing MLL-AF4 in the cortical zone.Interestingly, gpp knockdown using two different RNAi lines resulted in complete rescue of the MLL-AF4-driven increase in lymph gland size (Figures 4A-4D and 4G).A similar rescue of the leukemic phenotype was also observed when MLL-AF4 was expressed in a heterozygous genetic background with a mutant allele of gpp (gpp E-60 or gpp-03342) (Figures 4E-4G), in which gpp expression is reduced by half (Figure S5H).In contrast, in WT larvae, knockdown of gpp or introduction of gpp mutant alleles did not result in a significant reduction of lymph gland size (Figures S5A-S5F).Reduction of gpp expression in hemocytes by RNAi or mutant alleles was verified by qRT-PCR (Figures S5G and  S5H).Altogether, these data show that Drosophila DOT1L is essential for the phenotype induced by human MLL-AF4 in this in vivo model system.

The MLL-AF4 leukemia phenotype can be rescued by DOT1L inhibitors and MLL-Menin interaction inhibitors
To understand if this model offers a reliable and rapid platform for identifying pharmacological inhibitors for human MLL-r leukemia, we next addressed whether drugs in preclinical and early-stage clinical studies are efficacious in the Drosophila MLL-AF4 leukemia model.We selected inhibitors targeting different downstream mechanisms of MLL-AF4.Specifically, we tested the inhibitor of MLL-menin interaction MI-463, the bromodomain and extraterminal motif inhibitor i-BET, and the DOT1L inhibitor SGC0946.Drugs of interest were administered to developing larvae by mixing them into the standard fly food.Drug effect on the hematopoietic system was assessed in wandering 3 rd instar larvae.Interestingly, initial experiments assessing drug effects by visualizing the hematopoietic system through the cuticle of intact larvae revealed that treatment with the DOT1L inhibitor seemingly reduced lymph gland size (Figures S6A-S6L), whereas MI-463 had little effect and i-BET induced larval death (Figures S6M-S6T).Therefore, we decided to investigate the DOT1L inhibitor effect further.Larvae were treated as before, and lymph glands were dissected out to assess drug effects on lymph gland size and morphology.Treatment with 20 mM SGC0946 was shown to fully rescue the lymph gland hyperplasia phenotype observed in the leukemia model (Figures 5A-5F).Notably, a subset of SGC0946-treated MLL-AF4-expressing lymph glands were even smaller than WT lymph glands, indicating that MLL-AF4-expressing hemocytes are particularly sensitive to inhibition of gpp/DOT1L.Although DOT1L inhibitors showed promising results in initial studies, they have failed in clinical trials.To address whether the Drosophila MLL-AF4 leukemia model would respond to a more relevant novel therapy, we treated the larvae with the MLL-Menin interaction inhibitor VTP50469 that has showed potential in preclinical studies. 58Treatment with 20 mM showed no significant reduction in lymph gland size, whereas treatment with 200 mM, which was the limit concentration of solubility in DMSO, resulted in significantly smaller lymph glands in MLL-AF4-expressing larvae (Figures 5F-5J).Together, these data provide proof of principle that the MLL-AF4 Drosophila model is sensitive for pharmacological inhibition of clinically relevant targets that can be used to explore leukemogenesis.Hth and exd, the Drosophila homologs of MEIS1 and the PBX family, are crucial for the MLL-AF4 leukemic phenotype MLL-r leukemia has previously been shown to depend on MEIS1 and co-depend on redundant contributions of PBX proteins in mammalian leukemia models. 30,59Furthermore, depletion of either of the Drosophila orthologs of MEIS1 or the PXB family, hth or exd, respectively, had previously been shown to rescue a HOXA9-NUP98 leukemia model in the Drosophila eye. 60To elucidate part of the biological mechanism leading to lymph gland hyperplasia in our MLL-AF4 leukemia model, we next investigated the role of hth and exd.Downregulation of either hth or exd rescued the MLL-AF4 lymph gland overgrowth (Figures 6A-6E), whereas it had no effect on WT lymph glands (Figures S7A-S7D).Furthermore, using immunostaining for hth, we could observe increased levels of hth-positive cells in the cortical zone of the lymph gland, where hth expression was highly upregulated in MLL-AF4-expressing hemocytes (Figures 6F-6I).The observed increase of hth-positive cells was not only a consequence of a larger cortical zone in MLL-AF4-expressing lymph glands, as shown by a higher fraction of hth expression also within the cortical zone (Figure 6J).Depletion of gpp reduced the hth signal back to basal levels, suggesting that the hth/exd axis is downstream of gpp recruitment and transcriptional activation.the associated comprehensive toolbox for genetic manipulation and unbiased screening will allow for novel discoveries and sought-after therapeutic options in vivo.
In mouse models, DOT1L has been studied as a potential therapeutic target in MLL-r leukemia, using inhibitors that either inhibit its catalytic activity or disrupt its interaction with MLL fusion partners. 61Multiple mechanisms have been proposed to explain how the methyltransferase DOT1L contributes to leukemogenesis.Traditionally, it is believed that the fusion partner homologs AF9 and ENL are responsible for recruitment of DOT1L through the multi-subunit complex EAP (ENL-associated proteins), but DOT1L has also been shown to be directly associated with AF10. 62Furthermore, although MLL-AF4 does not appear to be directly associated with DOT1L, it can stimulate DOT1L activity indirectly through the recruitment of AF9. 63,64Pharmacological inhibition of DOT1L with small molecules selectively kills MLL-r leukemia cells in vitro and in vivo. 65,66In line with this, we found that a subset of MLL-AF4-expressing lymph glands from larvae raised on SGC0946-containing food were even smaller than WT lymph glands, indicating that MLL-AF4-expressing hemocytes are particularly sensitive to pharmacological gpp/DOT1L inhibition.Although SGC0946 effectively inhibits DOT1L activity, the failure of DOT1L inhibitors in clinical trials is concerning and may be attributed to the existence of compensatory mechanisms or alternative pathways that contribute to leukemogenesis.This fly model of leukemia could also be used as a convenient in vivo screening tool for identifying such mechanisms and pathways.
We also found genetic depletion of gpp/DOT1L to rescue the leukemia-like phenotype of our MLL-r leukemia model.Interestingly, the gpp allele E-60 was identified as an enhancer of the phenotype caused by expression of the human leukemic oncogene NUP98-HOXA9 in the Drosophila eye. 60In contrast, we found this allele to suppress the phenotype in our larval model of MLL-r leukemia, similar to the gpp allele 03342, which previously failed to complement lethal gpp alleles. 578][69] As these oncogenes are known to activate HOX genes, DOT1L is probably promoting aberrant transcription of the HOX gene cluster in these malignancies.Interestingly, depletion of gpp in MLL-AF4-expressing lymph glands reduced the size to less than that in WT larvae, suggesting that they are unable to maintain their transcriptional regulation.These results are in concordance with our findings from the pharmacological inhibition of DOT1L where such sensitivity also was observed.When it comes to NUP98-HOXA9, the lack of gpp/DOT1L dependency might be due to the fusion oncoprotein containing the downstream target itself, hence bypassing the regulative effect of DOT1L on HOX genes.In this case, based on the Drosophila model, 60 gpp/DOT1L appears to rather oppose NUP98-HOXA9 function, underscoring the need to critically evaluate the use of DOT1L inhibitors across cytogenetically different subtypes of leukemia.
Inhibitors of the BET family of proteins such as i-BET have shown promising results in preclinical models and in early clinical trials on hematologic malignancies, 70 although clinical implementation of most of these inhibitors has been hampered by toxicity. 71Indeed, larvae feeding on food containing i-BET died at early larval stages irrespective of the presence of MLL-AF4, indicating general toxicity at concentrations above 5 mM.Therefore, we believe that our Drosophila model can be used to quickly discard candidate compounds or derivatives that are too toxic for further clinical development, which may help to reduce attrition commonly associated with drug development. 724][75] In agreement, genetic depletion of the Drosophila orthologs Mnn1 and Jasper suppressed the lymph gland enlargement in this MLL-AF4 fly model.Lymph glands where Mnn1 was depleted in the presence of MLL-AF4 had a tendency to be smaller than WT on average, suggesting strong inhibition of the lymph gland growth in the absence of Mnn1.In contrast, the small molecule MI-463, which inhibits MEN1-MLL interaction, only produced a modest effect on the MLL-AF4-induced lymph gland hyperplasia.However, the novel MLL-Menin interaction inhibitor VTP50469 rescued the phenotype at high concentrations.VTP50469 has demonstrated significant promise as a therapeutic agent. 58Recently, additional novel Menin inhibitors have shown promising effects in clinical trials, underpinning the role of Menin as therapeutic target in MLL-r leukemia. 76he need to use high concentrations of VTP50469 in our model system could be due to limited bioavailability in the larvae, or that this compound is less efficient at disrupting the interaction of human MLL with Drosophila Mnn1.
Interestingly, we only observed a transient proliferative effect of MLL-AF4 in L2 larvae, whereas this effect was no longer present in wL3 larvae.It is possible that other cooperative factors are needed for a stronger phenotype and sustained proliferation, such as the reciprocal fusion protein AF4-MLL.Several model systems have failed to properly represent MLL-AF4-rearranged leukemia, and it has been proposed that this is due to the lack of the reciprocal fusion protein.Most patients with translocations of the MLL gene have also the reciprocal fusion protein present, 27 and the reciprocal fusion protein has been shown to be necessary to establish a leukemic mouse model. 26A human infant model of MLL-r iALL, using CRISPR/Cas9 editing of human fetal cells to generate cells that express both MLL-AF4 and AF4-MLL, has also elucidated the importance of AF4-MLL. 77A future approach would be to co-express AF4-MLL in our fly leukemia model, preferably for a controlled duration since AF4-MLL may be lost during clonal evolution 27 and may be most important at early disease development. 28otably, in our model, depletion of hth or exd, the Drosophila orthologs of MEIS1 and PBX3, rescued the phenotype.These findings are in line with the role of MEIS1 and PBX3 in leukemic transformation in MLL-r leukemia previously described. 30,59In Drosophila, hth and exd have been shown to be associated with Hox proteins to regulate target developmental genes, a similar role of MEIS1 and PBX family. 78In the context of hematopoiesis, hth has been described in the regulation of the PSC, which is important for normal hematopoiesis.Specifically, hth mutant larvae fail to develop lymph glands and hth overexpression causes disappearance of the PSC. 79,80In agreement with this, we found that MLL-AF4 expression in the cortical zone lead to upregulation of hth protein levels and a decrease in the number of cells in the PSC.In contrast, the expression of NUP98-HOXA9 in the cortical zone induced an enlargement of the PSC, 81 suggesting differences in the mechanisms of how these human oncogenes induce lymph gland hyperplasia.The fact that we do not see lymph gland development failure with hth knockdown is most likely because we restrict the expression pattern to mature hemocytes in our model, compared to the previously described hth null mutants.Since hth levels and lymph gland size could be rescued by gpp depletion, we suggest that hth and exd may function downstream of gpp recruitment and transcriptional activation.Future studies, such as by single-cell RNA sequencing, should investigate the changes in gene expression in this fly leukemia model dependent on gpp and hth/exd function, which may also explain the observed differentiation into lamellocytes.Interestingly, lamellocyte fate is often regarded as a suppressed cell fate and MLL-AF4 expression may relieve this suppression. 82lthough previous MLL-r leukemia model systems have been reported in D. melanogaster, they have been limited to the observation of lethality, cell-cycle defects, 45 or eye phenotypes in a reporter assay of Polycomb group-responsive element activity. 46These studies showed that expression of human MLL-AF4, MLL-AF9, or MLL-AF10 affected cell-cycle progression, normal development, or transcriptional regulation.In contrast, our viable in vivo model allows for a more detailed investigation of the hematopoietic irregularities that are induced upon MLL-AF4 expression.We observed changes in levels of differentiated hemocytes, specifically increased levels of plasmatocytes and crystal cells in the lymph gland as well as the presence of lamellocytes.These results are qualitatively more similar to those of the larval leukemia model based on NUP98-HOXA9 expression. 814][85][86][87][88] In the context of BCR-ABL, a fly model of BCR-ABL expression in the fly eye has been put forward as a screening platform for identification of novel or improved compounds. 85In our study we found that the larval aberrant hematopoietic phenotype induced by human MLL-AF4 expression can be suppressed by DOT1L or Menin-MLL interaction inhibitors, indicating that our model could be used to test derivatives of current inhibitors or novel compounds.The phenotype is amenable to imaged-based screening, similar to what has previously been performed for a Drosophila Ras-driven tumor model, 89 as the enlarged GFPpositive lymph gland can be visualized through the semi-transparent larval cuticle.
Collectively, we present a genetically tractable MLL-r leukemia model that is sensitive to pharmacological and genetic perturbations that can be exploited to further investigate the role of MLL-AF4 in leukemogenesis.

Limitations of the study
Despite the valuable insights that can be gained from our fly leukemia model, we acknowledge the limitations of our system.There are several limitations when it comes to using this for drug screening in terms of scaling and pharmacokinetics.Based on an imaging approach, it is likely not realistic to use this platform for thousands of compounds.Moreover, actual drug concentrations in the larvae are not easily measured.It is possible to implement feeding assays to control and assess dosage of the drugs in each individual larva, but this will not be feasible for largescale drug screens.Furthermore, we have not evaluated the potential degradation of drugs when mixed in the fly food.This could potentially lead to false negatives, and one should assess the pharmacological properties of the drug before exposing it to fly food.The drug platform serves better for proof-of-concept studies in the context of MLL-r leukemic pathways.
Both the DOT1L inhibitor SGC0946 and the MLL-Menin interaction inhibitor VTP50496 rescued the MLL-AF4-induced lymph gland enlargement.One significant limitation is that our model system could not accurately predict which of the two drugs would exhibit favorable outcomes in clinical trials.The failure of DOT1L inhibitors in clinical trials minimizes its role as a therapeutic candidate, whereas the potential of MLL-Menin interaction and targeted Menin inhibitors as agents is highly promising. 58,76Findings observed in our model system may not necessarily translate directly to the clinical setting due to complex biological variations between Drosophila and humans.
Although we observe hematopoietic aberrations in Drosophila upon the expression of MLL-AF4, our model differs from human leukemia in some important aspects.Whereas accumulation of immature blasts is observed in leukemia resulting from MLL-r leukemia in humans, we do not observe an accumulation of immature hemocytes in our model.Based on the available assays and markers, the increase in hemocyte numbers in circulation and in the cortical zone of the lymph gland appears to represent terminally differentiated and functional hemocytes.
This model only includes MLL-AF4 and not the reciprocal translocation AF4-MLL.AF4-MLL may be essential to sustain the initial increase in proliferation (pSer10His3 staining) that we observe in L2 lymph glands, similar to a suggested role in human MLL-r leukemia. 28There might be leukemic transformations that are not properly modeled due to the lack of this and other important genetic factors.However, despite its limitation as a drug candidate screening system, our model system holds great potential for understanding MLL biology in the context of hematopoiesis and leukemia.By interrogating the underlying molecular mechanisms and drug responses, our model system can aid in the discovery of new targets and guide the development of more effective treatments for leukemia.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Helene Knaevelsrud (helene.knavelsrud@medisin.uio.no).

Materials availability
This study did not generate new unique reagents.All fly lines used in this manuscript are either available from stock centers or will be made available upon request.

Data and code availability
d Microscopy, flow cytometry and qPCR data reported in this paper will be shared by the lead contact upon request.d ImageJ macros and NIS Elements software setting used for quantification will be shared by the lead contact upon request.d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Fly stocks
Drosophila melanogaster fly stocks were mainly obtained from Bloomington Stock Center or VDRC.UAS-MLL-AF4 ML5 (on 2nd), UAS-MLL-AF4 ML6 (on 3rd), UAS-MLL FL on 2 nd , UAS-MLL FL on 3 rd , UAS-MLL N-term (on 3rd), UAS-MLL N-term (on X) and UAS-AF4 C-term (on X) were reported elsewhere. 45pxn-Gal4, UAS-GFP was also previously reported. 91A full list of the fly lines used in this study can be found in Table S1.All crosses were performed on standard potato mash fly food (27.3 g/liter dry yeast, 32.7 g/liter dried potato powder, 60 g/liter sucrose, 0.73% agar, 0.2% 4-hydroxybenzoic acid methyl ester (nipagin)/ethanol and 0.45% propionic acid) with additional yeast paste and incubated at 25 C for the first 72 hrs followed by 48-72 hrs at 29 C to boost Gal4-driven expression.

METHOD DETAILS GFP imaging of entire larvae
Larvae were heat fixed to visualize the lymph gland in intact animals.Wandering L3 larvae were collected, washed once in PBS with 0.2% Triton-X100 (Sigma Aldrich, Cat#T9485) (PBT 0.2%) and once in PBS, followed by heat fixation in a drop of glycerol on a microscope slide

qRT-PCR
qRT-PCR was used to determine RNA levels of genes of interest, as well as to validate knockdown efficiency of RNAi fly lines.For qPCR on circulating hemocytes, the hemocytes were extracted by a protocol adapted from Tattikota and Perrimon. 92For each biological replicate, 15-20 larvae were collected in a 2 mL Eppendorf tube containing 500 ml glass beads and 200 ml PBS.The tube was vortexed for 2 min at 2000 rpm.
Only larvae where the heartbeat was detectable were included in the following.Each larva was dried and placed in a 10ml drop of ice-cold PBS before peeling the cuticle open allowing the hemocytes to enter the media by diffusion.The next larva was added to the same drop and the procedure repeated until all larvae were processed before the hemolymph was transferred to a new tube.RNA was extracted using TRIzol (Life Technologies, Cat#15596026, Lot#16200701) and the Zymo Research Direct-zolä RNA MicroPrep kit (Cat# R2062, Lot# ZRC201718) and RNA concentration measured by NanoDrop.RNA was converted to cDNA using the iScriptä cDNA Synthesis Kit (BIO-RAD Cat#1708891) on the Applied Biosystems SimpliAmp PCR machine.qRT-PCR was performed with Fast SYBRGreen Master Mix (Applied Biosystems, Cat#4385614) and run on Applied Biosystems StepOnePlus Real Time PCR system with StepOne v2.3 software.Efficiency of primer sets was tested before running expression level analysis.qPCR primers were selected from http://www.flyrnai.org/flyprimerbank. 93All primers used are listed in Table S2.The reaction was quenched with 300 mL of PBS with 2% FBS, before the sample was filtrated in FACS tubes with 35 mm filter cap (Falconâ, Cat# 352235). 1 mL of 1,677 mg/mL Propidium Iodide (Sigma Aldrich, Cat# P4170) was added right before analysis to identify live and dead cells.Samples were run on LSRII UV laser flow cytometer and the analysis was performed with FACSDiva and FlowJo software.

Phagocytosis assay of circulating hemocytes
To assess if the circulating hemocytes expressing MLL-AF4 were functional, we measured phagocytic activity with flow cytometry.Hemolymph from 15 wandering third instar larvae was collected by bleeding in 100 mL ice-cold ESF921 (Expression Systems, Cat# 500302) and kept on ice after collection.Cells were counted using a Bu ¨rker chamber and equal number of cells were mixed with 1, 2.5 and 5 mL pHrodoä Red E. coli BioParticlesä (Invitrogen, Cat# P35361, Lot# 1387791) for 20 minutes at room temperature.300 mL PBS with 2% FBS was added before the sample was filtrated in FACS tubes with 35 mm filter cap (Falconâ, Cat# 352235).Samples were run on LSRII UV laser flow cytometer and the analysis was performed with FACSDiva and FlowJo software.

Drug treatment
SGC0946 (Selleckchem, Cat# S7079 and MedChemtronica Cat# HY-15650/CS-3531, Lot#58947), VTP50469 (MedChemExpress, Cat# HY-114162, Lot# CS-0077527), i-BET (Merck Millipore, Cat# 401010) and MI-463 (Selleckchem, Cat# S7816) were dissolved in DMSO as a stock solutions and added to molten ($40 C) standard fly food mixed with blue food colorant (Panduro, Cat# 1018108-104) at final concentrations of the drug and 0.1% DMSO.DMSO concentration was kept equal across all experimental vials.The exception was for the preliminary drug experiments in Figure S6, where DMSO concentration varied and DMSO only controls were provided for each drug concentration.5 mL of food was added to each vial and allowed to solidify.Crosses were initially performed on normal fly food with yeast paste before they were flipped onto vials with drugs 24 hours after crossing to ensure sufficient fertilization and egg laying.Incubation times and temperature conditions were as in regular crosses described above.The presence of blue color in the larval gut was used as an indication to ensure exposure to the drug by ingestion of the drug-containing food.

QUANTIFICATION AND STATISTICAL ANALYSIS Image processing and quantification
Within each set of experiments, images were captured with identical settings below pixel value saturation and post-processed identically.
Intensity enhancements for visualization purposes on confocal microscopy images were performed using Adobe Photoshop, ImageJ or NIS Elements, and all intensity alterations were done equally for all genotypes to ensure comparativeness within each experiment.Confocal images shown in figures are maximum intensity projections made using the NIS Elements software unless otherwise stated.For confocal imaging performed on the Nikon Yokogawa CSU-W1 microscope, lymph gland volume and cell type volume fractions were calculated using the NIS Elements software.Quantifications for circulating hemocytes, lamellocytes, NimC1/P1+ cells, Hnt+ cells, antp+ cells and hth+ area were also performed with NIS Elements.NimC1/P1 positive cells were counted in 1 Z-plane visualizing the surface plane only due to improper antibody penetration of lymph gland tissue.Phospho-H3 positive cell quantifications were done using ImageJ and is reported per lymph gland area to account for size differences.

Experimental design and statistics
Samples were not masked or randomized during data collection or analysis.Statistical analysis was performed using GraphPad Prism 5.The data was assumed to be normally distributed.For multiple comparisons, one-way ANOVA with Bonferroni post-testing was used.For single comparisons, unpaired two-sided Student's t-test was performed.For each panel, the employed statistical test is stated in the corresponding figure legend.

Figure 1 .
Figure 1.Expression of the human oncogene MLL-AF4 in the hematopoietic system of D. melanogaster induces a leukemia-like phenotype in the lymph gland (A-C) Representative images of wandering third-instar larvae expressing full-length human MLL or human MLL-AF4 driven by pxn-Gal4 and imaged by widefield fluorescence microscopy.The hematopoietic system is marked by GFP expression driven by pxn-Gal4.Arrowheads indicate enlarged lymph glands.(D-F) Immunofluorescence confocal images of lymph glands that are WT or expressing MLL or MLL-AF4 driven by pxn-Gal4.The cortical zone is marked by GFP, and the posterior signaling center (PSC) is detected by immunostaining against antp (in red).Images are maximum intensity projection of z stacks.Scale bars 100 mm.(G) Quantification of lymph gland area (mm 2 ) from immunofluorescence confocal images.(H) Quantification of GFP-positive lymph gland volume presented as ratio of total lymph gland volume based on immunofluorescence confocal stacks.G, H: Each data point represents one animal.Boxplots show the mean G the 95 th percentile as error bars.Statistical significance was determined by one-way ANOVA with Bonferroni post-test to assess significant differences from WT. ns: not significant, ***: p < 0.001.(I) Quantification of GFP-positive hemocytes per lymph gland by flow cytometry.(J) Quantification of percentage of GFP-positive cells in larval lymph glands measured by flow cytometry.I, J: Data points are mean values from around 15 lymph glands per genotype from 6 independent experiments.Boxplots show the mean G the 95 th percentile as error bars.Statistical significance was determined by two-sided Student's t test to assess significant differences from WT. **: p < 0.01, ***: p < 0.001.(K-P) Immunofluorescence confocal images of lymph glands from larvae that are WT or expressing MLL-AF4 driven by pxn-Gal4, in 2 nd instar larva (L2), feeding 3 rd instar larva (fL3) and wandering 3 rd instar larva (wL3), respectively.Cortical zone is marked by GFP expression and phospho-histone 3 Ser10 (pH3Ser10)-positive cells are shown in grayscale.Images are maximum intensity projection of z stacks.Scale bars 100 mm.(Q) Quantification of phospho-histone 3 Ser10-positive cells per lymph gland area for L2, fL3, and wL3 larval stages.All values are normalized to WT mean of wL3 larval stage.Values are shown as mean and error bars show 95 th percentile.Two-sided Student's t test was performed to assess significant differences between WT and MLL-AF4 for each larval stage.ns: not significant, ***: p < 0.001.Genotypes: A, D, G-J, K-M, Q: pxn-Gal4, UAS-GFP/+ (WT) B, E, G-J: pxn-Gal4, UAS-GFP/UAS-MLL C, F, G-J, N-P, Q: pxn-Gal4, UAS-GFP/UAS-MLL-AF4. See also Figures S1 and S2.

Figure 2 .
Figure 2. Expression of the human oncogene MLL-AF4 in the hematopoietic system of D. melanogaster induces a leukemia-like phenotype in circulating hemocytes (A-C) Immunofluorescence confocal images of circulating hemocytes from wL3 larvae that are WT or expressing MLL or MLL-AF4 driven by hml-Gal4.Images are maximum intensity projection of z stacks.Scale bars 100 mm.A 0 -C': GFP+ cells marked by the driver hml-Gal4, UAS-GFP.A 00 -C'': Phalloidin (F-actin) staining to visualize cytoskeleton and lamellocytes.Arrowheads indicate examples of differentiated lamellocytes.(D) Quantification of number of circulating hemocytes expressing GFP.Values are per larva across 4 larvae in 3 replicates.Boxplots show the mean G the 95 th percentile as error bars.(E) Quantification of lamellocytes in circulating hemocytes from A-C.Note that the x axis is split between 100 and 300 to enable visualization of all values.D,E: One-way ANOVA with Bonferroni post-test was performed to assess significant differences.ns: not significant, *: p < 0.05, **: p < 0.01.Genotypes: A, D, E: hml-Gal4, UAS-GFP/+ (WT) B, D, E: hml-Gal4, UAS-GFP/UAS-MLL C, D, E: hml-Gal4, UAS-GFP/UAS-MLL-AF4. See also Figure S3.

Figure 5 .Figure 6 .
Figure 5.The MLL-AF4 Drosophila model is sensitive to drug treatment targeting DOT1L and the MLL-Menin interaction (A-D) and (F-I) Lymph glands from wandering 3 rd instar larvae.Cortical zone is marked by GFP expression and PSC is detected by immunostaining for antp (in red).Images are maximum intensity projection of z stacks.Scale bars 100 mm.A, B: WT and MLL-AF4-expressing lymph glands treated with DMSO only.C, D: WT and MLL-AF4-expressing lymph glands treated with 20 mM SGC0946.F, G: WT and MLL-AF4-expressing lymph glands treated with DMSO only.H, I: WT and MLL-AF4-expressing lymph glands treated with 200 mM VTP50469.(E) Quantification of lymph gland area (mm 2 ) from immunofluorescence confocal images from drug treated larvae with either DMSO or the DOT1L inhibitor SGC0946.(J) Quantification of lymph gland area (mm 2 ) from immunofluorescence confocal images from drug treated larvae with either DMSO or the MLL-Menin interaction inhibitor VTP50469.E, J: Each data point represents one animal.Boxplots show the mean G the 95 th percentile as error bars.Statistical significance was determined by two-sided Student's t test to assess significant differences between drug-treated larvae and DMSO control for each genotype.ns: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001.Genotypes: A, C, E, F, H, J: pxn-Gal4, UAS-GFP/+ (WT) B, D, E, G, I, J: pxn-Gal4, UAS-GFP/UAS-MLL-AF4. See also Figure S6.

Flow
cytometry was used to determine lymph gland size and differentiation through levels of mature GFP-positive hemocytes in the lymph gland.Around 15 lymph glands were dissected out and trypsinized with 100 mL of 1x trypsin with EDTA without phenol red (Life Technologiesä, Cat#15400054) for 15 minutes at 25 C.Each sample was pipetted up and down 40 times to ensure proper tissue dissociation.

TABLE
d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS B Fly stocks d METHOD DETAILS B GFP imaging of entire larvae B Immunofluorescence confocal microscopy of lymph glands B Immunofluorescence of circulating hemocytes B qRT-PCR B Flow cytometry of live lymph gland cells B Phagocytosis assay of circulating hemocytes a heat block set at 70 C for approximately 10 seconds.A coverslip was added on top of the glycerol drop with the heat fixed larvae before imaging them with a GFP filter on Leica MZFLIII stereomicroscope and LAS v4.9 software.Post-imaging enhancement of GFP signal for sufficient visualization was performed in Adobe Photoshop.Santa Cruz, Cat# sc-26187).For staining with the Attila/L1 antibody, Triton-X100 was omitted in all washing, blocking and staining steps and replaced with PBS 1X to ensure proper staining.Lymph glands were next washed twice with PBT 0.2% and then incubated with fluorescent secondary antibodies for 2 hours in room temperature in the dark, followed by 10 min incubation with 1mg/mL Hoechst-33342 (Thermo Scientificä, Cat# 62249, Lot# PK1922313).The secondary antibodies used in this study were anti-mouse Alexa568 (1:1000, Molecular Probes, Cat# A10037), anti-rat DyLight649 (1:1000, Jackson Immunoresearch, Cat#712-495-153, Lot#93224) and anti-goat Alexa555 (1:500, Molecular Probes, Cat# A21432).After staining, lymph glands were washed twice in PBT 0.2%, detached from the carcass and mounted on microscope slides with 15 mL ProLongä Diamond Antifade Mountant (Invitrogenä, Cat#P36961).Typically, between 10 and 25 lymph glands were imaged and analyzed per genotype.Confocal imaging was performed on a Nikon Yokogawa CSU-W1 with a S Plan-FI LWD 20x air objective and a laser set of 405 nm, 488 nm and 561 nm.Lymph glands were imaged in a z-stack covering 25 mm with a z-step of 0.1 mm.
on For confocal imaging of circulating hemocytes, larvae were bled in ice-cold ESF921 media (Expression Systems, Cat# 500302) and the hemolymph from one larva was transferred to a corresponding well in a black 384-well glass-bottom plate (CellVis, Cat# P384-1.5H-N)The circulating hemocytes were fixed with 4% paraformaldehyde (Polysciences, Cat# 18814-20) and centrifuged for 5 min at 1000 rpm before incubation for 10 min at room temperature.Wells were then washed twice with PBT