Drosophila Hox genes induce melanised pseudo-tumours when misexpressed in hemocytes

Background: Hox genes are key early determinants of cell identity along the anterior-posterior body axis across bilaterians. Recently, several late non-homeotic functions of Hox genes have emerged in a variety of processes involved in organogenesis in several organisms, including mammals. Being crucial factors in determining cell identity and organogenesis, the misregulation of Hox genes is likely to be associated with defects in these processes. Several studies have reported the misexpression of Hox genes in a variety of malignancies including acute myeloid leukaemia. Methods: The Hox genes Dfd, Ubx, abd-A and Abd-B were overexpressed via the UAS-Gal4 system using Cg-Gal4, Lsp2-Gal4, He-Gal4 and HmlD3-Gal4 as specic drivers. Genetic interaction was tested by bringing overexpression lines in heterozygous mutant backgrounds of Polycomb and trithorax group factors. Larvae were visually scored for melanised bodies. Hemocytes were quantied by dissecting larvae for lymph in 4mm wells and staining nuclei with DAPI and tested for differentiation by staining them with anti-myospheroid and for proliferation with anti-PH3. Pupal lethality was carried out by letting pupae eclose and scoring those that failed after the time point. Results: Expression of Dfd, Ubx and abd-A, but not Abd-B in the hematopoietic compartment of Drosophila led to the appearance of circulating melanised bodies, and increase in cell numbers, cell-autonomous proliferation and differentiation of hemocytes. Pupal lethality and the melanised pseudo-tumor phenotype were suppressed by the mutations in Psc 1 and esc 2 background while polycomb group member mutations Pc 1 and Su(z)12 3 and trithorax group member mutation TrlR 85 increased the phenotype. Conclusions: Dfd, Ubx and abd-A are leukemogenic. Mutations in Polycomb and trithorax group members, which are responsible for maintaining the expression state of the Hox genes, modulate the leukemogenic phynotype. exclusive driver, however, Ubx and abd-A gave a signicant increase in hemocyte number, despite them not manifesting melanised spots. Our results show that melanised spots (or pseudo-tumors), which have been reported as the hallmarks of a “leukemia-like” phenotype in Drosophila, may not reect an actual increase in hemocytes. Additionally, many studies have used the strong driver Cg-Gal4, which drives expression in fatbody as well as the blood cells. As our results show that perturbations in the fatbody may indeed lead to an increase in circulating hemocytes. It may also be that the number of circulating cells when we expressed the hox genes under cg-Gal4 may be due to circulating cells being trapped in the melanised peseudo tumors. has shown that abd-A, one of the three Hox genes of the bithorax complex of Drosophila, acts as a growth promoter in Histoblast Nest Cells 74 . We show this to be a normal additional function of abd-A which involved adult cuticle formation during pupation. In the present study, we tested four Hox genes of Drosophila by ectopically expressing them in the blood cells and show that they are capable of inducing melanised bodies in circulation. These melanotic spots appear only when expressed in the lymph gland and circulating blood cells. The ectopic expression of the Hox genes also triggers cell proliferation. The cells appear to divide in a cell-autonomous manner, which is reected in the detection of PH3+ cells in circulation. The presence of myospheroid positive elongated cells, seen in circulation, also suggests that Hox overexpression leads to the differentiation of the circulating blood cells into lamellocytes. Overexpression of abd-A, shows a relatively stronger phenotype while Abd-B overexpression does not. It supports an earlier nding in which we observed a non-homeotic growth promoter role of abd-A but not of Abd-B during the formation of the adult cuticle during pupation 74 . Our results indicate that Hox genes are causal in leukaemia, reinforcing previous studies in vertebrate model systems, and extending these ndings to Drosophila. This also opens the possibility that Hox gene induced leukaemias, especially those of the myeloid lineage, can be studied and modelled in Drosophila.


5.
Our study reports the causal link of Hox genes to the process of cancer conditions.
. These ndings open a new possibility of addressing the function of Hox genes in leukemogenic hematopoiesis the y as model system.
Life comes in a variety of body forms. Despite this variety, there is similarity at the genetic and molecular level in the developmental mechanisms that lead to this variety across species. For example, in spite of the evolutionary distance between vertebrates and Drosophila, many organ and tissue types show a degree of homology with each other and many key developmental pathways governing their development and function are conserved.
The hematopoietic system is no exception. Hemocytes of Drosophila resemble the myeloid lineage of blood cells 1  This conservation includes the molecular pathways involved in the development of these cell types. For example, Drosophila serpent 6 is related to GATA 1, 2 and 3 of vertebrates. GATA-2 is responsible for blood progenitor proliferation and survival 7,8 while GATA-1 is required for progenitor differentiation into erythrocytes, megakaryocytes and eosinophils [9][10][11] . Similar to GATA-2, srp is required for progenitor maintenance and proliferation. Loss of function in srp leads to a reduced number of progenitors and a loss of all hemocytes. It is also required in plasmatocyte differentiation, similar to GATA-1 12 . Additionally, Drosophila u-shaped is related to the Friend of GATA (FOG) family. FOG-1 and GATA-1 together are required for erythrocyte and megakaryocyte differentiation 13,14 . FOG-1 interacts with GATA-1 to represses eosinophil differentiation and must be downregulated for eosinophil differentiation 15 . Similarly, ush is expressed in hemocyte precursors and plasmatocytes, and must be downregulated for crystal cell development 16 .
Signalling pathways involved in regulating hematopoiesis, as would be anticipated, are conserved between vertebrates and Drosophila. For example, Jagged-1, the vertebrate homolog of Serrate and a ligand of Notch, is produced by the stromal cells of the bone marrow, to regulate Hematopoietic Stem Cell (HSC) proliferation and survival 17 . Ser performs a similar role, being released by cells of the Posterior Signalling Centre (PSC) 18 via cytonemes 19,20 , a set of regulatory cells at the posterior end of the Lymph Gland (LG). Vertebrate JAK2 is required for erythropoiesis 21 , while STAT5 is required for proper progenitor and myeloid cell function 22 . The Drosophila JAK/STAT pathway is required within the LG for the maintenance of prohemocytes, among other things 23 . Transformations in JAK2 can lead to leukemogenesis in vertebrates 24 , similar to how gain of function hop mutants behave 25 . The Toll pathway is also conserved, playing a major role in innate immunity in both vertebrates and ies 26 .
One aspect of vertebrate hematopoiesis that has not been mirrored in Drosophila is the role of Hox genes.
Hox genes are well known for their conserved role in body axis formation across all bilaterians 27 , but also play roles in vertebrate hematopoiesis 28 , autophagy 29 , as well as cell proliferation, differentiation, migration and apoptosis 30 . Hox genes are transcribed in HSCs as well as lineage progenitors, and are suppressed in differentiated blood cells [31][32][33][34][35] Overexpression models show blockages in certain stages of development, expansion of HSCs, the circulation of blast cells, etc. 36-41 . For example, Hoxa7 and Hoxa9 have been shown to have a role in the development of hematopoietic progenitors of different lineages in mice. On the other hand, in Drosophila, other than Antp, which is implicated in setting up the location of the LG 42 , as well as later marking the PSC 43 , Hox genes have not been reported to play any role.
The expression of genes of the Hox cluster during, and after development is regulated by two chromatin remodelers, Polycomb and trithorax group (PcG and trxG) of proteins, which were discovered as transcriptional repressors (PcG) and activators (trxG) of Hox genes in Drosophila 44 . Later, these proteins were shown to regulate many biological processes such as cell fate and lineage, cellular memory, stem cell function, and tissue homeostasis in cell lines and mouse models [45][46][47][48] . The deregulation of Hox genes via Polycomb or trithorax proteins can lead to leukemogenesis by misregulation of hematopoiesis.
Furthermore, PcG members EZH2, a human homolog of Drosophila E(z) protein, EED (Esc in Drosophila), SUZ(12) (Drosophila Su(z)12) and BMI-1(homolog of Drosophila Psc) have been shown to have a role in different cancers in knock out studies carried out in cell lines as well as mouse model [49][50][51][52] . Mixed Lineage Leukemia (MLL), a human homolog of Drosophila Trithorax (Trx) protein, regulates Hoxa expression in HSCs. MLL is a frequent fusion protein partner in acute leukemia 53 . Evidence for the role of of PcG and trxG genes in regulating HSC development in Drosophila remains largely to be explored 54,55 .
In this study, we show that overexpression of the Hox genes, Dfd, Ubx and abd-A in blood cells not only leads to melanised pesudo-tumors, but also to a signi cant increase in blood cell number and the induction of lamellocyte differentiation. Further, we present genetic evidence to show the role of PcG members, Psc and Esc, in the melanised pseudo-tumor formation induced by Hox genes. These ndings will be helpful in understating the biological events associated with leukaemia in humans, which may open new possibilities of markers and therapy.

Methods
Fly strains and culture: Flies were cultured in standard cornmeal and sucrose agar. The wild-type ies used in this study were Canton-S. Flies were maintained at 25 o C. For all experiments, ies were allowed to lay eggs for 6 hours before being transferred to a fresh vial. Larvae were screened and used for immunohistochemistry at 96-102 hours post egg laying, before the onset of metamorphosis. Supplementary Table S1and S2 list the y stocks used in this study.
Larval screening for percent penetrance and severity of the phenotype: For the over-expression of different Hox genes, the UAS-Gal4 binary system was used. To assess the effect of PcG and trxG members had in modifying the phenotype, heterozygous mutant lines were recombined with the Cg-Gal4 driver (Supplementary Table S2 for all recombined stocks made in the lab). Con rmation of recombination was based on expression of w+ linked with the Cg-Gal4 transgene and lethality when backcrossed with the mutant line. Recombined mutants with Cg-Gal4 were maintained over the CyO-GFP balancer for GFP screening. Third chromosome mutants were crossed with homozygous Cg-Gal4 lines and maintained over TM6B for screening via the Tubby phenotype. Experimental crosses were set between recombined strains (Cg-Gal4 with mutant) and UAS-abd-A at a density of 12 females and 6 males for each cross. Egg lay was allowed for 6 hours and progeny were collected after 96 hours post egg lay, at the L3F stage. Screening was done using a stereomicroscope. Penetrance was calculated by calculating the percentage of melanotic pseudo-tumor manifesting larvae. Severity of the phenotype was assessed visually. One-way ANOVA (Dunnett's multiple comparisons) was performed to test the signi cance.
Pupal lethality count: To assess the pupal lethality, larvae were allowed to develop into pupae and were observed beyond 10 days post egg-lay. Eclosed progenies were considered as survivors. Dead pupae were counted manually. For heterozygous mutant experimental pupae, larvae were rst screened to con rm the presence of the Cg-Gal4 driven expression of UAS-abd-A and the presence of the mutation before being transferred to fresh vials. Second chromosome mutants were con rmed by selecting non-GFP larvae while third chromosome mutations were Tb + Immunostaining and cell quanti cation: For staining proliferative cells we made use the M -phase marker, Anti-PhosphoHistone 3 at serine 10, from Upstate (cat# 07-212, 1ng/μL). For con rming the presence of lamellocytes, we used anti-myospheroid (DSHB #CF.6G11, 27pg/μL). Blood cells were prepared using an established protocol 56 . Blood cells numbers were quanti ed using a modi ed version of the protocol by Petraki, Alexander, & Bruckner, 2015 57 . Larvae were dissected in 4mm wells, their hemolymph allowed to settle down, before being xed with 1% formaldehyde and stained with DAPI. Each well was scanned using an Olympus IX83 at 20X, with 32 images stitched. Cells were quanti ed using CellPro ler by counting individual nuclei. Signi cance was tested using an unpaired t-test with Welch's correction between control and overexpression genotypes.
Visualisation of lymph glands in larvae over expressing Hox genes and Quanti cation of relative GFP levels: Larave were grown as described above. Virgin Hml-Gal4, UAS-GFP, ies were used to drive the expression of the individual Hox genes. Cg-Gal4, Hml-Gal4, He-Gal4 and Lsp2-Gal4 lines were crossed with mcd8-GFP lines Larvae were harvested and visualised under a Zeiss Axiozoom.V16 for GFP. For comparison between He-Gal4, Hml-Gal4 and cg-Gal4, whole larval maximum intensities of He-Gal4 and Hml-Gal4 were compared with regions devoid of the fatbody in cg-Gal4. For comparison between cg-Gal4 and Lsp2-Gal4, hole larval maximum intensities were compared.

Results
Tumor phenotype correlates with the tissue speci city and strength of the driver: In Drosophila, the collagen-Gal4 (Cg-Gal4) driver induces the strong expression UAS tagged genes in the fatbody as well as in the hematopoietic system 58 . The different UAS Hox genes lines, Dfd, Ubx, abd-A and Abd-B, when brought under the Cg-Gal4 driver, induced melanised pseudo-tumors in larvae. This phenotype manifested in 26% of Cg-Gal4>UAS Dfd larvae, 60% of Cg>Ubx larvae, 82% of Cg-Gal4>UAS abd-A larvae and 4% of Cg-Gal4>UAS-Abd-B larvae (Figure 1, 2A, 2B and Supplementary Table S3).
We then used the Hemese-Gal4 (He-Gal4) driver, which expresses throughout the lymph gland as well as in circulating hemocytes, and the HemolectinD3-Gal4 (HmlD3-Gal4) driver, which expresses in the cortical region of the lymph gland as well as in mature circulating hemocytes. While melanised pseudo tumours were observed in these genotypes, they appeared smaller and the penetrance of the phenotype was very low, manifesting in 3% of He-Gal4>UAS Dfd, 6% in HmlD3-Gal4>UAS Dfd, 9% in He-Gal4>UAS Ubx, 2% in HmlD3-Gal4>UAS Ubx, 8% in He-Gal4>UAS abd-A, 4% in HmlD3-Gal4>UAS abd-A, 3% in He-Gal4>UAS Abd-B and 2% in HmlD3-Gal4>UAS Abd-B (Figure 1, 2A, 2B and Supplementary Table S3). He-Gal4 induces expression throughout the lymph gland and in sessile cell pockets which reside underneath the larval cuticle. Thus, it expresses in all areas involved in hematopoiesis 59 . Over-expression of Hox genes with the He-Gal4 driver always showed a higher penetrance of the phenotype when compared to HmlD3-Gal4. Lamellocytes are responsible for the encapsulation mechanism in combating an immune challenge, and they do not express Hemolectin. The low penetrance of the phenotype in HmlD3-Gal4 could be due to a lack of expression in lamellocytes 60 . Also, Hemolectin does not express in the medullary zone of the lymph gland, where cell proliferation and differentiation takes place 61 . It shows the phenotype is associated with active proliferation and differentiation of hemocytes of developing larvae. To test that the phenotype was not due to expression of the Hox genes in the fatbody (as Cg-Gal4 expresses in both blood cells as well as the fatbody) we over-expressed these genes using the fatbody speci c driver Lsp2-Gal4. Lsp2-Gal4 functions in L3 larval fat bodies 62 . No melanised spots were observed in such larvae, indicating that the pseudo-tumor phenotype is not induced by the misexpression of of Hox gene in the fatbody. To test whether the relative strength of the Gal4s, we overexpressed mcd8-GFP under Cg-Gal4, Hml-Gal4, He-Gal4 and Lsp2-Gal4. Hml-Gal4 was signi cantly weaker than He-Gal4 and cg-Gal4. He-Gal4 and cg-Gal4 appear to drive expression at similar levels. However, as we compared whole larvae of He-GFP expressing larvae to regions devoid of the fatboy in cg-Gal4 larvae, this similarity may be artefactual ( Figure Table S3). Taken together, this implied that the melanised pseudo-tumour phenotype we observe is of hemocyte origin.
Tumor phenotype is co-related with lethality at the pupal stage: We also noticed a signi cant level of pupal lethality when Hox genes were misexpressed in these conditions. Pupal lethality with the Cg-Gal4 driver was highest when it drives UAS-abd-A (99 %). Cg-Gal4>UAS Dfd (53%) and Cg>Ubx (24%) also show an increased lethality at pupal stage. It was negligible in Cg-Gal4>UAS-Abd-B (2%). We observed lethality when the same genes were over expressed in the fatbody with Lsp2-Gal4. However, Lsp2-Gal4 driven Hox expression induced lethality was lower compared to Cg-Gal4 driven Hox expression induced lethality. But it must be noted that it was greater than that induced by the blood speci c drivers used by us. Pupal lethality with Lsp2-Gal4 driver was observed 9% in Lsp2-Gal4>UAS Dfd, 26% in Lsp2-Gal4>UAS Ubx and 31% in Lsp2-Gal4>UAS abd-A. It has previously been shown that aberrant blood cells can induce pupal lethality 63 . However, while we did observe some pupal lethality when the Hox genes were expressed under He and Hml, the lethality was most prominent in when the Cg-Gal4 or Lsp2-Gal4 driver was used ( Figure 2B, Supplementary Table S4) which supports the earlier report suggesting that Hox genes are repressors of autophagy in the fatbody 29 . Thus, while we do observe insigni cant lethality with blood speci c drivers since the expression of Hox genes in the fatbody does indeed induce lethality, the greater lethality when Cg-Gal4 is used may be due to the concomitant expression induced in the fatbody as well as blood cells.
Hox genes over-expression induces hemocyte proliferation and differentiation: Change in number cells and types of cells become important considering the phenotype observed upon misexpression of Hox genes. We quanti ed the number of blood cells in our overexpression lines using a modi ed version of established methods 56,57 . When expressed by blood speci c driver, Dfd, Ubx and abd-A led to a signi cant increase in the number of circulating hemocytes ( Figure 3A and 3B, Supplementary Table S5-8). Interestingly, while the penetrance of melanised spots was lower, blood speci c drivers showed a larger number of blood cells ( Figure 3B). Under the control of, Lsp2, the fatbody exclusive driver, however, Ubx and abd-A gave a signi cant increase in hemocyte number, despite them not manifesting melanised spots. Our results show that melanised spots (or pseudo-tumors), which have been reported as the hallmarks of a "leukemia-like" phenotype in Drosophila, may not re ect an actual increase in hemocytes. Additionally, many studies have used the strong driver Cg-Gal4, which drives expression in fatbody as well as the blood cells. As our results show that perturbations in the fatbody may indeed lead to an increase in circulating hemocytes. It may also be that the number of circulating cells when we expressed the hox genes under cg-Gal4 may be due to circulating cells being trapped in the melanised peseudo tumors.
Previous studies have shown that cells of the LG do not enter into circulation until the onset of metamorphosis. However, Hml and Cg express in the cortical region of the LG, and He expresses throughout. Thus, the question arose as to whether the increase in cell number was due to an increase in cell proliferation at the LG or were circulating cells proliferating in a cell-autonomous manner. Hence, we checked for the presence of the mitotic cell marker PH3. We observed cells positive for PH3, when Hox genes were expressed in the blood cells, and not when expressed exclusively in the fatbody ( Figure 3A, Supplementary Figure 1A-D). Unlike previous reports, we did not nd proliferative cells in our control experiments 64 . This may be due to a loss of cells in our preparations or more robust immunostaining on our part. Thus, while we cannot rule out the possibility that LG cells contribute to this increase, at least a fraction of the increase takes place due to the cell autonomous division of Hox overexpressing cells. As cells of the LG could potentially prematurely be released into circulation on account of the Hox gene over expression, we checked for the integrity of the LG by overexpressing UAS-Dfd, UAS Ubx and UAS-abd-A in an Hml-Gal4, UAS-GFP background.
LGs remained intact 96hrs post egglay ( Figure 6) While imaging the blood cells, we noticed that there were larger, attened cells in circulation, reminiscent of lamellocytes. To test whether they were bona de lamellocytes, we stained the hemocytes for the lamellocyte marker myospheroid ( Figure 4A, Supplementary Figure 1A-D). Control larvae infrequently showed the presence of lamellocytes. In our overexpression lines, however, we noticed that a signi cant number of cells were lamellocytes mys+. Some plasmatocytes also stained positive for mys. None of the plasmatocytes in the control ies or those overexpressing Abd-B were positive for mys. Previous reports have sugessteded that circulating plasmatocytes may differentiate into lamellocytes 65,66 . Thus, it may be that these circulating mys+ plasmatocyte like cells are differentiating into lamellocytes. However, Lsp2-Gal4>UAS Ubx also had a signi cant number of lamellocytes. This is in keeping with reports that signals from the fat body can drive lamellocyte differentiation 67,68 . Thus, we speculate that these cells, upon Hox overexpression, are pushed toward the lamellocyte fate ( Figure 4A, 4B).

Effect of PcG and trxG genes
PcG members are known to function primarily through two distinct complexes, PRC1 (consisting of Pc, Psc, Su(z)2 and Sce) and PRC2 (consisting of E(z), Su(z)12, Esc and Caf 1-55) 69 . Members of the PcG and trxG have been shown to have a role in hematological malignancies in different clinicopathological data in leukemic patients and mice models 70,71 . To determine their role in melanized pseudo-tumor formation in ies, we over-expressed abd-A using Cg-Gal4 in the background of different PcG and trxG mutants. We selected Psc 1 , Pc 1 , Su(z)2, Su(z)12, E(z) and esc2 from the PcG and brm 2 , Trl from the trxG. Melanotic pseudo-tumor phenotype was used in our study to assay the effect of the mutants as it is convenient and robust. All experiments were performed in biological triplicates. The PcG mutants Pc 1 , Su(z)12 3 , and trxG member brm 2 showed an increase in melanotic body formation (Figure 5A, 5B and   Supplementary Table S9), and enhanced the phenotype upto 100 per cent. Pc 1 and Su(z)12 3 not only enhanced the penetrance (percentage phenotype showing larvae) but showed an increase in severity (scored as number and size of the black spots) compared to abd-A over-expressed in absence of mutants ( Figure 5A). Pc and Su(z)12, both are the core proteins of PRC1 complex and play a role in negative regulation of their target genes. Our results indicate these proteins might regulate melanotic body formation. Surprisingly, E(z) does not show any signi cant effect on penetrance. On the other hand, esc 2 (PRC2 member) and Psc 1 (PRC1 member) showed a signi cant decrease in penetrance 15% and 17% respectively ( Figure 5B, Supplementary Table S9). The severity of the phenotype is also reduced in both the mutant background. These results indicate that genes involved in melanotic pseudo-tumor causing phenotype might be the target of the Esc and Psc proteins. Although it has been shown that Esc-E(z) complex is a thousand times effective to E(z) alone 72,73 , our results suggest that Esc regulates its targets independent of E(z) activity or, for that matter, any other member of the PRC2 complex in the observed phenotype. Similarly, Psc mutation rescued the phenotype. We tested whether bringing our overexpression in the PcG and trxG backgrounds affected the number of PH3 positive nuclei. We did not observe a signi cant change ( Figure 5D, Supplementary table 12). As the average number of PH3+ nuclei in cg-Gal4>UAS abd-A larvae was 0.08% of the average of total hemocytes, it may be that the total number of dividing nuclei are too few to signi cantly differ.

Effect of PcG mutants on the melanized pseudo-tumor related pupal lethality:
To test the effect of mutations on pupal lethality, L3F larvae from each combination, which manifested melanised pseudo-tumours, were transferred to fresh vials and allowed to pupate and eclose. Larvae from overexpressed abd-A (driven by Cg-Gal4) with melanotic body showed up to 99% lethality at the pupal stage. Further, we checked pupal lethality in mutant background. Since all mutants are maintained over balancers (Table S2), we selected overexpressed progenies without balancer to con rm mutant in the same progeny and transferred them in new food vials. Pupal lethality in Su(z)12 3 , Pc 1 and Su(z)2 1.a1 was always 100% while we could get a few survivors from Cg-Gal4>UAS abd-A ( Figure 3C, Table S10). A decline in lethality was seen in Psc 1 ,esc 2 brm 2 and Trl R85 . The survivors from Psc 1 and esc 2 were quite healthy as compared to the survivors of Cg-Gal4>UAS abd-A. This reduction in lethality indicates that Esc and Psc proteins are strongly suppressing the melanotic pseudo-tumour phenotype and its consequences on development. Although brm 2 showed an increase in penetrance it decreases pupal lethality 89% compare to abd-A alone. Trl R85 showed a decrease in pupal lethality (79%).

Conclusion
Homeotic genes or Hox genes determine the cell identity across the anterior-posterior body axis early during development, a function that is conserved across all bilaterians. The later functions played by these genes, however, are less well studied. A number of reports indicate that they play a variety of nonhomeotic functions later in development. Our lab has shown that abd-A, one of the three Hox genes of the bithorax complex of Drosophila, acts as a growth promoter in Histoblast Nest Cells 74 . We show this to be a normal additional function of abd-A which involved adult cuticle formation during pupation. In the present study, we tested four Hox genes of Drosophila by ectopically expressing them in the blood cells and show that they are capable of inducing melanised bodies in circulation. These melanotic spots appear only when expressed in the lymph gland and circulating blood cells. The ectopic expression of the Hox genes also triggers cell proliferation. The cells appear to divide in a cell-autonomous manner, which is re ected in the detection of PH3+ cells in circulation. The presence of myospheroid positive elongated cells, seen in circulation, also suggests that Hox overexpression leads to the differentiation of the circulating blood cells into lamellocytes. Overexpression of abd-A, shows a relatively stronger phenotype while Abd-B overexpression does not. It supports an earlier nding in which we observed a non-homeotic growth promoter role of abd-A but not of Abd-B during the formation of the adult cuticle during pupation 74 .
Our results indicate that Hox genes are causal in leukaemia, reinforcing previous studies in vertebrate model systems, and extending these ndings to Drosophila. This also opens the possibility that Hox gene induced leukaemias, especially those of the myeloid lineage, can be studied and modelled in Drosophila.
Till date, the only known Hox gene to participate in Drosophila hematopoiesis is Antp, which marks in the PSC, and provides spatial signals for the development of the LG. In vertebrates, Hox genes have been shown to express within progenitor cells and are rapidly switched off during cell maturation. As our overexpression lines perturb both cell number and differentiation, it is possible that multiple Drosophila Hox genes are involved in netuning the precise programme of Drosophila blood cell development as well.
The fact that the cells appear to be phenotypically con ned to plasmatocytes and lamellocytes implies that expression of these genes works in tandem with, and above the speci c programme of the cell types. It would be interesting to know which genes are being modulated in our overexpression lines, by pro ling their transcription states as well as the binding sites of the individual Hox proteins. In the absence of this information, we speculate that Hox gene overexpression leads to the aberrant transcription of genes. It is known from previous studies that Hox dysregulation in leukaemia is usually concomitant with gain or loss of function mutations in upstream regulators, most commonly in Mixed Lineage Leukemia-1 (MLL-1) fusion proteins 75,76 , or loss of function Enhancer of Zeste Homolog 2 (EZH2) mutations 77 . It has been reported that EZH2 mutations have the lowest number of co-operating mutations to induce leukemogenesis 78 .
Interestingly, we observe polycomb members Psc and Esc have a role in suppressing melanised pseudotumour formation. Both Psc and Esc mutants rescued the phenotype signi cantly which suggests that tumour suppressor genes may be their targets for repression. However, while there is evidence that Hox overexpression in vertebrate blood cells do induce lukemia, the modulation of the phenotype by PcG and trxG mutant backgrounds may be due to the differential regulation of immune genes in the overexpression background. As many Hox induced lukemaias occur in the background of PcG loss of function and trxG gain of function backgrounds, this may lead to the differential accessibility of the overexpressed transcription factor to immune genes, thus either enhancing or suppressing the phenotype BMI1, a mammalian counterpart of Psc, was discovered as a proto-oncogene. Overexpression of BMI1 in mouse models has been shown to induce both types of leukaemia (lymphoid and myeloid origin) 53,79,80 .
BMI1 cooperates with c-myc in the generation of lymphomas. MYC is a transcription factor that regulates many cellular processes, including proliferation and apoptosis. c-myc is an attractive target for chemotherapy as it has a role in multiple converging signalling cascades. In many cancers, MYC protein is overexpressed due to various processes like translocation, duplication or epigenetic misregulation 81-83 . BMI1 regulates MYC protein expression by inhibiting its downstream target ink4a-ARF, a tumour suppressor gene 84 . The role of EED, a mammalian homolog of Drosophila Esc, in leukaemia is not very well understood. Mutations in EED have been indicated to impair polycomb complex (PRC2) functionality and it is associated with myelodysplastic neoplasm 85,86 . While PcG and trxG genes are known to function in a complex, results in our lab indicate that they may have functions outside their canonical pathways, which would explain why genes within similar complexes elicit different effects on our overexpression backgrounds 87,88 .
Although Drosophila does not have ink4a homolog, it will be interesting to see the mechanism of regulation of melanotic tumor formation and over-proliferation of blood cells by Psc. Taken together, we speculate that Hox gene activation in hemocytes causes cell-autonomous proliferation and differentiation and induces leukaemia via aberrant transcription.
In summary, Drosophila the Hox genes Dfd, Ubx and particularly abd-A, when expressed in blood cells, are leukemogenic. This link of Hox genes to the pseudo-tumor phenotype supports the non-homeotic role of abd-A as a growth promoter later during development. The disease phenotype is modi ed by select PcG/trxG members. This reinforces previous studies in vertebrates that report the misregulation of Hox genes in several cancers and implicate the role of epigenetic factors in them. Hox induced leukaemia in Drosophila offers advantages of the y model to explore the biology of the process and develop novel potential markers and therapeutic options.

PH3
Phospho Histone 3 at Serine 10 LG -Acknowledgements Authors acknowledge Yacine Graba for the UAS lines used, N.R. Chakravarthi, C. Subbalakshmi, Aprotim Mazumder and P.S. Kesavan for access to and help with imaging facilities., Ramachandra for help y cultivation media. Authors are thankful to Indira Paddibhatla for help in useful discussions and familiarization with the Drosophila hematopoietic system.  Percentage of pupal lethality, indicated by larvae that fail to eclose. When Cg-Gal4 drives the genes Dfd, Ubx, abd-A and Abd-B do cause lethality, so does expressing them in the fatbody under Lsp2-Gal4. Driving these genes in the blood cells (He-Gal4 and HmlD3-Gal4) leads to a much lower penetrance of this phenotype.

Figure 3
Cell proliferation and quanti cation of hemocytes. A) Anti-PH3 staining for comparative study of cell proliferation with over-expression of abd-A gene driven by Cg-Gal4, He-Gal4, HmlD3-Gal4 and Lsp2-Gal4.