Abstract
Drosophila larval hematopoietic organs produce circulating hemocytes that ensure the cellular host defense by recognizing and neutralizing non-self or noxious objects through phagocytosis or encapsulation and melanization. Hematopoietic lineage specification as well as blood cell proliferation and differentiation are tightly controlled. Mutations in genes that regulate lymph gland cell proliferation and hemocyte numbers in the body cavity cause hematopoietic organ overgrowth and hemocyte overproliferation. Occasionally, mutant hemocytes invade self-tissues, behaving like neoplastic malignant cells. Two alleles of the Polycomb group (PcG) gene multi sex combs (mxc) were previously isolated as such lethal malignant blood neoplasm mutations. PcG genes regulate Hox gene expression in vertebrates and invertebrates and participate in mammalian hematopoiesis control. Hence we investigated the need for mxc in Drosophila hematopoietic organs and circulating hemocytes. We show that mxc-induced hematopoietic hyperplasia is cell autonomous and that mxc mainly controls plasmatocyte lineage proliferation and differentiation in lymph glands and circulating hemocytes. Loss of the Toll pathway, which plays a similar role in hematopoiesis, counteracted mxc hemocyte proliferation but not mxc hemocyte differentiation. Several PcG genes tested in trans had no effects on mxc hematopoietic phenotypes, whereas the trithorax group gene brahma is important for normal and mutant hematopoiesis control. We propose that mxc provides one of the regulatory inputs in larval hematopoiesis that control normal rates of plasmatocyte and crystal lineage proliferation as well as normal rates and timing of hemocyte differentiation.
DROSOPHILA larval circulating blood cells carry out the cellular host defense response through phagocytosis or encapsulation and melanization. Most wild-type circulating hemocytes belong to a monocyte-like phagocytic cell lineage (reviewed in Dearolf 1998; Fossett and Schulz 2001). They are called plasmatocytes, podocytes, or macrophages according to their shape, their adhesion properties, and the stage of development (Gateff 1978; Rizki 1978; Lanotet al. 2001; Meister and Govind 2002). Phagocytic cells eliminate small foreign objects and, in pupae, the lysing larval tissues. Less than 1% of circulating hemocytes are flat lamellocytes that encapsulate larger “non-self” objects (Rizki 1978; Rizki and Rizki 1980). Crystal cells represent a second hematopoietic cell lineage (Dearolf 1998; Fossett and Schulz 2001), which is required for melanization of lamellocytes after encapsulation (Rizki 1978). Specification of hemocyte progenitor cells and regulation of lineage commitment and differentiation depend on a series of evolutionarily conserved transcription factors (Lebestkyet al. 2000; reviewed in Fossett and Schulz 2001). Drosophila blood cells are produced during two successive waves of hematopoiesis in embryos and larvae. All larval hemocytes are produced by the hematopoietic organs or lymph glands. This structure differentiates during late embryogenesis as two lobes along the anterior part of the dorsal vessel (Rugendorffet al. 1994) and has in third instar larvae four to seven lobe pairs (Gateff 1978; Rizki 1978). Large anterior lobes contain most types of circulating hemocytes, whereas smaller posterior ones contain undifferentiated blast cells that normally give rise to macrophages at pupariation (Lanotet al. 2001). Lymph gland overgrowth, hemocyte proliferation, and hemocyte differentiation are inducible by immune challenge or by external aggression such as infestation by a parasitoid wasp, but they are also observed in certain mutant contexts (Watsonet al. 1991; Rizki and Rizki 1992; reviewed in Dearolf 1998; Lanotet al. 2001). The corresponding larval phenotypes include hypertrophied hematopoietic organs, increased numbers of circulating hemocytes, and abnormal differentiation of lamellocytes that represent up to 50% of the cells. Melanotic masses often occur. They are formed by lamellocyte-covered capsules that contain melanized self-tissue (pseudotumors; Sparrow 1978; Watsonet al. 1991; Dearolf 1998).
Such abnormal immune response phenotypes are caused by a number of mutations (Watsonet al. 1991; Toroket al. 1993; Dearolf 1998), making these loci potential candidates for genes directly regulating hematopoiesis. Two lethal (1) malignant blood neoplasm [l(1)mbn] alleles were isolated as causing overproliferation of lymph glands and circulating hemocytes (Gateff 1978; Shrestha and Gateff 1982; Gateff and Mechler 1989). Allelism was later established between l(1)mbn and the Polycomb group (PcG) gene multi sex combs (mxc; Santamaria and Randsholt 1995; Sagetet al. 1998). PcG genes form a conserved group that collectively maintain expression patterns of important selector genes in vertebrates and invertebrates (reviewed in Pirrotta 1998; van Lohuizen 1999; Gebuhret al. 2000; Brock and van Lohuizen 2001; Simon and Tamkun 2002). First isolated in Drosophila as negative trans-regulators of the Hox genes, PcG genes have been shown to act in conjunction with trithorax group (trxG) genes to maintain transcriptional regulation and provide a cellular memory mechanism throughout development, probably by changes in chromatin structure. Several mammalian PcG and trxG members are involved in hematopoiesis control (reviewed in van Lohuizen 1999; Gebuhret al. 2000; Takihara and Hara 2000; Raaphorstet al. 2001). PcG genes show stage-specific expression differences in human bone marrow cells (Lessardet al. 1998). The mouse PcG gene embryonic ectoderm development (eed) negatively regulates myeloid and lymphoid progenitor cell proliferation in bone marrow (Lessardet al. 1999), whereas targeted disruption of murine PcG genes Bmi-1, Mel 18, Rae23/Mph1, and M33 all lead to loss or hypoproliferation of various hematopoietic tissues. The human trithorax homolog MLL is often affected in translocations associated with acute myeloid or lymphoblastic leukemias (reviewed in van Lohuizen 1999; Muller and Leutz 2001), and MLL-/+ mice present severe hematopoietic abnormalities (Yuet al. 1995). Furthermore, mammalian SWI/SNF chromatin remodeling proteins, which are homologs of the Drosophila trxG protein Brahma (BRM), have been implicated as important cofactors in the regulation of myeloid and erythroid genes (reviewed in Gebuhret al. 2000; Muller and Leutz 2001). Little is known about possible roles of PcG and trxG genes in Drosophila hematopoiesis. The domino (dom) gene encodes SWI/SNF family DNA-dependent ATPases that interact with PcG products in negative homeotic gene regulation, and dom mutations induce hematopoietic disorders. Lymph glands show proliferation defects and the rare dom hemocytes that differentiate cannot cross the lymph gland basement membrane (Braun et al. 1997, 1998; Ruhfet al. 2001). mxc is the only Drosophila gene known to cause both abnormal hematopoiesis and homeotic transformations due to HOM/Hox gene gain of function (Sagetet al. 1998). In view of this, we decided to analyze how regulation of larval hematopoiesis and of circulating hemocyte density were affected by mxc mutations, alone or in other mutant contexts that also control these processes.
Two signal transduction pathways are well characterized as controlling hemocyte proliferation and steady-state numbers of hemocytes in Drosophila larvae. The conserved Toll/cactus/Rel-NF-κB signaling pathway is one (Qiuet al. 1998; reviewed in Dearolf 1998; Mathey-Prevot and Perrimon 1998; Meister and Govind 2002). Toll pathway activation leads to nuclear translocation of Rel/NF-κB transcription factors that regulate hemocyte division and differentiation. Enhanced Toll signaling induces hematopoietic organ hyperplasia, increases in circulating hemocytes, abnormal lamellocyte differentiation, and pseudotumors, whereas larvae with reduced Toll signaling have fewer hemocytes. Toll signaling and Rel/NF-κB proteins also control Drosophila humoral host defense (Govind 1999), and homologous vertebrate Toll/cactus-I-κB/Rel pathways ensure similar functions (Qiuet al. 1998; Grossmannet al. 1999). Hematopoiesis control likewise involves the Drosophila JAK/STAT signal transduction pathway (reviewed in Dearolf 1998; Mathey-Prevot and Perrimon 1998; Luo and Dearolf 2001). Constitutive activation of the JAK nonreceptor tyrosine kinase encoded by hopscotch (hop) causes hematopoietic overproliferation, overproduction of lamellocytes, and pseudotumors (Hanratty and Ryerse 1981; Silvers and Hanratty 1984; Hanratty and Dearolf 1993; Harrisonet al. 1995; Luo et al. 1995, 1997; Lanotet al. 2001). Gain-of-function hop proteins hyperactivate the signal transducer and activator of transcription protein encoded by D-stat, and haplo-insufficiency for D-stat partially suppresses hematopoietic gain-of-function phenotypes of hop (Houet al. 1996; Yanet al. 1996; Luoet al. 1997; reviewed in Dearolf 1998; Luo and Dearolf 2001). Hematopoiesis control by the JAK/STAT pathway is also evolutionarily conserved since alterations of mammalian JAK/STAT affect hematopoietic cell proliferation, differentiation, and apoptosis (Lacroniqueet al. 1997; Nosakaet al. 1999; reviewed in Dearolf 1998; Luo and Dearolf 2001).
Here, we describe hematopoietic phenotypes of several increasingly severe mxc alleles and confirm that mxc directly controls hematopoiesis. Development of plasmatocyte and crystal cell lineages is affected by loss of mxc. We compared hematopoietic defects due to mutations of mxc, of the Toll, or of the JAK/STAT pathway and analyzed epistatic relations between mxc and these mutants. Loss of Toll signal is epistatic to loss of mxc. Diminished D-stat activity had no effect on mxc hematopoietic phenotypes whereas it partially rescued lamellocyte differentiation of Toll gain of function and both hemocyte overproliferation and differentiation induced by constitutive JAK activation. Any genetic combination of these proliferation activator contexts caused extreme lymph gland overgrowth together with reductions in circulating hemocyte numbers. Finally, we found that proliferation control by mxc is less dependent on other PcG genes than is segmental identity control, whereas the trithorax group gene brahma is important for normal and mutant hematopoiesis. We propose that mxc provides a hematopoiesis regulatory input that controls normal plasmatocyte and crystal cell lineage development as well as normal rates and timing of hemocyte differentiation.
MATERIALS AND METHODS
Fly strains and culture: Flies were grown on standard culture medium at 25°, unless otherwise stated. mxc alleles have been described (Santamaria and Randsholt 1995; Docquieret al. 1996). Dorothy (Dot) encodes an ecdysteroid UDP-glucosyl/UDP glucuronosyl transferase that is expressed in pericardial and lymph gland cells; the Dot-LacZ strain was a gift from D. Kimbrell (Rodriguezet al. 1996). Reporter line l(2)113/28 (from M. Meister) expressing β-galactosidase in lamellocytes is described by Braun et al. (1997). The ubiquitously β-galactosidase-expressing strain hsp83-LacZ contains several P[hsp83-LacZ] inserts. The β-galactosidase-free strain βgalnl was provided by A. Shearn (Woodhouseet al. 1998). hopM38/FM7 act-GFP flies were from M. Lagueux. Other Toll and JAK/STAT mutants were gifts from C. Dearolf or B. Lemaitre. Mutant domino (dom) phenotypes are described by Ruhf et al. (2001); dom strains were provided by M. Meister. Balancer chromosomes and all other mutants, including Black cells (Bc), tube (tub), Toll (Tl), cactus (cact), hopscotch (hop), D-stat, trxG, and PcG genes are described in FlyBase (1999).
Hemocyte counts: Wandering third instar larvae were bled in 5 μl of Drosophila Ringer on a hemocytometer. Mean hemocyte numbers per milliliter of hemolymph were estimated by counting the number of cells in a given surface, under a dissection microscope. Each experiment involved an internal, nonmutant control category provided by sibling larvae issued from the same cross. Control larvae were collected from the same vials as experimental larvae and counted under similar conditions (see table legends for controls in each experiment). Hemocytes from at least 10 larvae were counted per genotype and most experiments were repeated. Genotypes were recognized by mouth-hook color, body shape, or the presence of actin- or histone3-driven GFP. mxc mutant chromosomes carry y1 and are maintained over Binsn. Phagocytic cells were recognized by their ability to absorb particles of India ink injected 2 hr previously into the larva (Lanotet al. 2001). Lamellocytes were identified by β-galactosidase expression, using the lamellocytes LacZ-reporter l(2)113/28 (Braunet al. 1997). Crystal cells were counted per larva; they were visualized by heating larvae for 10 min at 70° in a water bath. Blackened cells were counted under a dissection microscope. Hemocyte numbers were compared by Student’s t-tests. Blood cell type distributions and numbers of dividing cells were compared using chi-square tests.
Lymph gland transplantations: y1 mxc/Binsn females were crossed to homozygous hsp83-LacZ males at 23°. Lymph glands from wandering third instar progeny, [yellow] for experimental and male [yellow+] for control, were dissected in Drosophila Ringer and injected into the abdomen of 2- to 5-day-old βgalnl / βgalnl females, using a Drummond nanobject automatic injector (Drummond Scientific, Broomall, PA). Injected females were grown overnight at 20°, followed by 5 days at 23° for proliferation tests and 3 weeks for survival tests. Proliferation test females were then dissected and X-Gal stained without prior fixation for 1-3 hr at 37°. Stained tissues were fixed for 10 min in 3.7% formaldehyde in 1× PBS, rinsed twice in 1× PBS, and mounted in PBS:glycerol.
Genetic interactions: Tl10b effects on mxcG43 were analyzed in male progeny of y1 mxcG43/Binsn females crossed to Tl10b/His3-GFP males. Loss of Toll function in mxcG43 larvae was examined in male progeny of y1 mxcG43/Binsn;TlRXA/TM6c females and Binsn/Y;Tlr632/TM6c males. Other Toll pathway effects on mxcG43 mutants were examined in male larvae from strains y1 mxcG43/Binsn;cactA2/CyO act-GFP and y1 mxcG43/Binsn;tub238/TM6c. Interactions between D-stat and hop were examined in progeny of y w hopTum-l/FM7 females and D-stat6346/His3-GFP males. For other effects of D-stat, y1 mxcG43/Binsn females were mated to D-stat6346/His3-GFP males, y1 mxcG43/Binsn D-stat6346/ TM6c females were crossed to Binsn/Y;D-statHJ/TM6c males, and D-stat6346/TM6c females were crossed to Tl10b/His3-GFP males. Joint Toll signal and JAK gain-of-function effects were evaluated among male progeny from strain y w hopTum-l/Binsn;cactA2/CyO act-GFP or from crosses between y w hopTum-l/FM7 females and Tl10b/His3-GFP males. To examine effects of mxcG43 and hopTum-l together, an X chromosome carrying both mutations was obtained by recombination. PcG- and trxG-mxc interactions were tested among male progeny of y1 mxcG43/ Binsn females crossed to esc4/His3-GFP, PscArp1/His3-GFP, PclXM3/ His3-GFP, PcK/His3-GFP, ScmD1/His3-GFP, brm2/His3-GFP, brm2 trxE2/His3-GFP, Df(3R)trx1/His3-GFP, or mor1/His3-GFP males.
X-Gal staining: Wandering larvae were dissected in 1× PBS; fixed in 1× PBS, 3.7% formaldehyde for 10 min at room temperature; washed in 1× PBS; and stained 4-16 hr as described by Docquier et al. (1996). Hemocytes were smeared onto polylysine-coated glass slides, air dried for 5 min, fixed for 30 sec in 3.7% formaldehyde in 1× PBS, washed twice in 1× PBS, and then stained as other imaginal tissues.
Immunohistochemistry: Wandering larvae of adequate genotypes were dissected in 1× PBS; tissues were fixed in 1× PBS, 3.7% formaldehyde for 15 min at room temperature and then immunostained as described by LaJeunesse and Shearn (1996). Circulating hemocytes were smeared on polylysine-coated slides, air dried and fixed for 5 min in 3.7% formaldehyde, and then immunostained as other imaginal tissues. Antiphosphohistone H3 antibody (Upstate Biotechnology, Lake Placid, NY) was used at 1:10,000 without prior incubation. Staining was revealed after incubation with secondary biotinylated anti-rabbit antibody (Roche, Indianapolis; 1:400) using the Vectastain ABC (Elite) kit. Apoptosis in imaginal tissues was visualized using TUNEL (Whiteet al. 1994) with modifications. Dissected imaginal tissues were fixed for 15 min in 1× PBS, 3.7% formaldehyde and for additional 15 min in 1× PBS, 3.7% formaldehyde, 0.6% Triton X-100 and then washed three times in PTW (1× PBS, 1% Tween 20) and three times in 1× PBS. Tissues were incubated 30 min in H2O2:methanol (3:1000); washed three times in 1× PBS; incubated 4 min with proteinase K in 1× PBS; washed in PTW glycine (2 mg/ml); fixed again for 15 min in 3.7% formaldehyde, 0.2% gluteraldehyde, 1× PBS; washed in PTW; and incubated for 90 min with terminal transferase and biotinylated dUTP. Label was revealed using DAB and the Vectastain ABC (Elite) kit. Tissues were mounted in PBS:glycerol.
RESULTS
mxc mutations cause lymph gland overgrowth and hemocyte overproliferation: Two early pupal lethal mxc/l(1)mbn mutants were isolated as showing severe hematopoietic neoplasia (Gateff 1978; Shrestha and Gateff 1982; Gateff and Mechler 1989). To determine whether such defects were particular for these alleles, we examined lymph glands of four hypomorphic mxc mutants with increasingly severe homeotic and developmental phenotypes that range from viable to larval lethal. Their characteristics are summarized in Table 1. mxc mutant larvae develop pseudotumors mainly when raised under crowded conditions. In that case, <10% of mxcG43 larvae exhibit pseudotumors while up to 25% older mxcmbn1 larvae and most older mxc16a-1 larvae do (Sparrow 1978; Sagetet al. 1998). Effects of mxc on lymph glands were observed in X-Gal-stained mxc/Y;Dot-LacZ/+ late third instar larvae, compared to Binsn/Y;Dot-LacZ/+ siblings (Rodriguezet al. 1996). All four mutants exhibit overgrown lymph glands (Table 1; Figure 1). In mxcG46/Y, only some second and third lymph gland lobes showed hypertrophy, whereas more severe alleles induced stronger overgrowth in more posterior lobe pairs. To determine whether overgrowth was associated with increased cell divisions, we stained lymph glands with anti-phosphohistone H3 antibody, which labels cells undergoing mitosis. Lymph gland lobes from all mxc mutants exhibit increased mitotic activity as compared to wild type (Figure 2).
Lymph glands and hemocytes in wild-type and mxc mutants
—Overgrown hematopoietic organs of mxc larvae. Lymph glands from wild-type and mxc late third instar larvae carrying a Dot-LacZ enhancer trap were stained with X-Gal. Anterior lobes are to the left. (A) Wild type; (B) mxcG46; (C) mxcG43. Note that posterior lobes show stronger hypertrophy in C (arrowhead) than in B.
We examined numbers and relative proportions of circulating hemocytes in mxc larvae (Table 1; Figure 3). Numbers and ratios of plasmatocytes or macrophages, lamellocytes, and crystal cells in wild type have been described (Rizki 1978; Lanotet al. 2001; see Introduction). All control larvae showed mean hemocyte concentrations per milliliter that were within the range of previously published values for animals of comparable age (Gateff 1978; Brehelin 1982; Silvers and Hanratty 1984; Luo et al. 1995, 1997; Table 1). The same was true for the number of crystal cells per control larva (Table 1; Lanotet al. 2001; Sorrentinoet al. 2002). mxcG43, mxcmbn1, and mxc16a-1 larvae exhibited three- to fivefold increases in circulating hemocytes (Table 1). mxc blood cell types appeared similar to wild type with one possible exception: mutant larvae exhibited a significant increase (4-5% compared to <0.1% in controls) in a cell type with large pseudopod-like extensions. The cells appeared pear-shaped or spindle-shaped and were different from round plasmatocytes, even with cytoplasmic filaments, or from flattened macrophages (Figure 3). Like phagocytic plasmatocytes, these cells absorbed particles of India ink injected into the larva (Lanotet al. 2001; Figure 3); they presented numerous cytoplasmic organelles and vacuoles and resembled mxcmbn1 podocytes (Gateff 1978; Shrestha and Gateff 1982). Accordingly, we considered them as phagocytes, possibly with altered adhesion capacities, and call them podocytes hereafter. Mutant larvae contained 2-3% lamellocytes, whereas such cells often represented <0.1% in wild type (Table 1). Lamellocytes were already differentiated within mxc mutant lymph glands, as revealed by the lamellocyte-specific enhancer trap line l(2)113/28 (Figure 3). Staining of hemocytes from wandering larvae with anti-phosphohistone H3 antibody revealed increased mitoses in mxc mutants (Table 2; Figure 2). Among wild-type hemocytes 0.64% showed mitotic figures in good agreement with previous data (Rizki 1978; Qiuet al. 1998), whereas mitoses were up to four times as frequent in mxc mutants. Hence increased proliferation is present in lymph glands and in circulating blood cells. We compared crystal cell numbers per mutant larva with two controls: the y1 ac1 z1 strain in which mxcG46 and mxcG43 were induced (Santamaria and Randsholt 1995) and the balancer chromosome Binsn. Crystal cells, visualized by heat treatment, were fewer in all mxc contexts than in wild-type controls (Table 1; Figure 3). Similar results were observed when black cells were compared in Binsn/Y;Bc/+ and mxc/Y;Bc/+ animals (not shown). The reduction was statistically significant for mxcmbn1 and mxc16a-1 larvae when compared to both controls (Table 1). mxcG46 and mxcG43 crystal cell numbers were statistically different only from the Binsn/Y control.
—Cell division in lymph glands and circulating hemocytes. Cell division was visualized with anti-phosphohistone H3 antibody. (A) Lymph gland chain from Binsn/Y larva. Dividing cells are detected in posterior lobes (arrow). Bar, 100 μm. (B) First and second lymph gland lobes from mxcG43/Y larva. Numerous dividing cells are present in both. Bar, 100 μm. (C) Mitotic figure (arrow) in circulating hemocyte from mxcG43/Y larva. Bar, 10 μm.
—Circulating blood cell types in mxc larvae. Third instar larvae were dissected in Drosophila Ringer solution and hemocytes were visualized with a Leitz DMRD photomicroscope. (A) Plasmatocytes and spindle-shaped podocytes from mxcG43/Y larva absorb particles of India ink injected 2 hr previously; (B) macrophage; (C) the l(2)113/28 reporter line drives nuclear β-galactosidase expression in lamellocytes (la), here from mxcG43/Y larva. Bars in A-C, 10 μm. (D) X-Gal staining of mxcG43/Y;l(2)113/28/+ reveals differentiated lamellocytes inside the lymph glands (arrowheads). Bar, 100 μm. (E and F) Crystal cells, visualized by heat treatment, are more frequent in control Binsn/Y (E) than in mxc16a-1/Y (F) larvae.
Cell division in circulating hemocytes
Together the data show that all four mxc mutants affect lymph gland and circulating hemocyte proliferation and differentiation. Loss-of-function alterations of mxc result in abnormal numbers of circulating larval hemocytes of both the plasmatocyte and the crystal cell hematopoietic lineages.
mxc hematopoietic tissue is intrinsically overproliferating and invasive: mxc hematopoietic phenotypes could reflect a defense response to apoptosis, which occurs in mxc imaginal discs (reviewed in Dearolf 1998; Sagetet al. 1998). Alternatively, mxc could directly control prohemocyte and hemocyte proliferation and differentiation since l(1)mbn lymph gland cells can divide after transplantation into larvae or adults: they invade healthy tissues and can, according to Gateff and Mechler (1989), kill the host. To choose between these alternatives, we first compared the number of apoptotic cells in wild-type and mxc mutant lymph glands. TUNEL labeling of apoptotic cells revealed hardly any increase in cell death in lymph glands from mxc mutant animals (Figure 4). Hence mxc lymph gland overproliferation is likely not induced to compensate for cell death in the hematopoietic organs.
We performed a series of transplantation experiments, since transplantation of Drosophila cells with mutations in tumor suppressor genes into wild-type adult hosts can reveal their capacity for autonomous, uncontrolled proliferation (Hanratty and Ryerse 1981; Woodhouseet al. 1998). mxcG43, mxcmbn1, or mxc16a-1 lymph gland tissues that were all hsp83-LacZ/+ were injected into βgalnl/βgalnl females. We found no differences in viability between females transplanted with mxc or with wild-type lymph gland tissue, even 3 weeks after injection. To test for growth, a number of transplanted females were observed 24 hr after injection. This provided sizes of the original transplants. The rest were allowed to grow for 5 days. β-Galactosidase-expressing cells were revealed by X-Gal staining. Results are summarized in Table 3. From 50 to 100% of the transplants were detected after 5 days of growth. Furthermore, 60-80% of mxc lymph gland transplants multiplied in the hosts, whereas none of the 46 control transplants did so. Growing transplanted cells either remained localized in the abdomen of the host or invaded from part to all of the body cavity (Figure 5). Hence overproliferation of mxc lymph gland tissue is an autonomous and intrinsic characteristic of these cells. We conclude that mxc wild-type product is directly involved in control of lymph gland and hemocyte proliferation in larvae.
—Cell death in wild-type and mxc mutant lymph glands. Dark-field photograph of apoptotic cells, revealed by Tunel. (A) First lymph gland lobe pair and second lymph gland lobe from Binsn/Y wandering larva. Rare apoptotic cells are detected (arrowheads). (B) Detail of hypertrophic posterior lymph gland lobes from mxcG43/Y larva. Arrowheads show apoptotic cells. Bars, 100 μm.
Transplantation of mxc lymph glands into wild-type hosts
mxc and Toll signaling in hematopoiesis: The Toll pathway controls insect cellular defense responses, and Toll gain of function (g.o.f.) causes hematopoietic overgrowth (Qiuet al. 1998). We compared hematopoietic phenotypes of mxc- and of Toll signal g.o.f., due to either constitutive receptor activation by Tl10b or loss of the cytoplasmic inhibitor encoded by cactus in cactA2/cactA2 hypomorphs (Table 4). Qiu et al. (1998) reported a low incidence of pseudotumors along with a 2-fold hemocyte increase in Tl10b/+ larvae and a >10-fold one in the strong cactS1/cactS1 mutant. We found that Tl10b/+ larvae exhibited a 3- to 5-fold increase in hemocyte numbers and that 99% presented pseudotumors; hypomorphic cactA2/cactA2 larvae showed a 3-fold hemocyte increase (Table 4). These values are in the same range as mxc-induced overproliferation. In contrast to even the most severe mxc mutants (Table 1), which exhibit only 3-5% lamellocytes, both gain-of-Toll signal contexts were associated with higher lamellocyte ratios of, respectively, 13.8 and 17.0% (similar to previous Tl10b/+ data; Lanotet al. 2001). Spindle-shaped podocytes were also present, but in lesser amounts (4.0 and 1.1% of circulating cells, respectively).
To address how mxc related to Toll signaling in hemocyte control, we constructed mxcG43/Toll pathway double mutants. Loss of the pathway was obtained in either heteroallelic TlRXA/Tlr632 larvae or tub238/tub238 animals where cytoplasmic transduction of Toll signal is blocked (Govind 1999). Results are summarized in Table 4. In agreement with Qiu et al. (1998), we found that diminished Toll signaling reduces hemocyte numbers (68 and 77% of sibling Tl/+ or tub238/+ controls, respectively). Furthermore, both Toll signal loss-of-function (l.o.f.) contexts counteracted the effects of mxcG43 l.o.f. on hemocyte production. Mean hemocyte numbers of mxcG43/Y;TlRXA/Tlr632 or mxcG43/Y;tub238/tub238 larvae were not statistically different from the corresponding Binsn/Y;Tl/+ or Binsn/Y;tub238/+ internal controls, whereas mxcG43/Y;Tl/+ and mxcG43/Y;tub238/+ sibling larvae always had more circulating hemocytes (P < 0.01). Podocytes and lamellocytes, on the other hand, were still more numerous in mxcG43/Y;TlRXA/Tlr632 and mxcG43/Y;tub238/tub238 larvae compared to Binsn/Y;Tl/+ or Binsn/Y;tub238/+ controls. Hence for hemocyte density, loss of Toll signal is epistatic to the loss of mxc. We conclude that loss of mxc can increase body cavity hemocyte numbers only when the Toll pathway is functional.
—mxc lymph gland cells proliferate after transplantation into wild-type hosts. Lymph glands were dissected from third instar Binsn/Y;hsp83-LacZ/+ or mxc/Y;hsp83-LacZ/+ larvae and transplanted into the abdomen of βgalnl/βgalnl females. Hosts were dissected 5 days after injection and stained with X-Gal. (A) Transplanted wild-type cells show localized β-galactosidase expression (arrow); (B) transplanted mxcG43/Y cells; (C) transplanted mxcmbn1/Y cells. The dissected abdomen in C shows that β-galactosidase-expressing cells have partially invaded the body cavity.
mxc and JAK/STAT control of hematopoiesis: Gain of function of the JAK kinase encoded by hopscotch can be lethal and can induce pseudotumors, strong lymph gland and hemocyte overgrowth, abnormal differentiation of lamellocytes, and reduced crystal cell numbers (Hanratty and Ryerse 1981; Hanratty and Dearolf 1993; Harrisonet al. 1995; Luo et al. 1995, 1997; Lanotet al. 2001). Some of these effects are mediated by JAK overactivation of D-stat product and can be rescued by loss of D-stat (Houet al. 1996; Yanet al. 1996; Luoet al. 1997). As for the Toll pathway, we compared the effects of mxc mutations on hematopoiesis to those of loss or gain of function of the JAK/STAT pathway. Loss of hop reduces cell proliferation in larval tissues such as imaginal discs and brain (Perrimon and Mahowald 1986), but no data are, to our knowledge, available concerning hemocyte numbers and ratios in hop amorphic animals that are larval/pupal lethals. We examined hemocyte production in amorphic hopVA275 and hopM38 males raised at 25°. Interestingly, both hop null alleles exhibit hemocyte numbers and cell type distributions that were no different from control Binsn/Y or FM7/Y siblings (Table 5). This result suggests that, contrary to loss of Toll signaling, loss of hop/JAK might have no effect on the number of circulating larval hemocytes.
Hemocytes in Toll pathway and mxc mutants
Hemocyte phenotypes induced by loss of hop and loss of D-stat
Hou et al. (1996), Yan et al. (1996), and Luo et al. (1997) have reported partial suppression of g.o.f. hop-associated lethality and pseudotumors by mutations of D-stat, mainly at 29°. Furthermore, Luo et al. (1997) reported inhibition of hopTum-l/Y-associated lamellocyte differentiation by loss of a D-stat copy. To establish whether similar effects are observed at 25°, we examined plasmatocyte and lamellocyte production in a hopTum-l/Y;D-stat6346/+ context at this temperature. In agreement with the authors cited above, we found that hopTum-l/Y induced strong hemocyte overproduction and that 25% of these were lamellocytes; pseudotumors were less frequent in female hopTum-l/+;D-stat6346/+ larvae compared to their hopTum-l/+;His3-GFP/+ siblings at this semirestrictive temperature (10 and 24%, respectively). Interestingly, larvae raised at 25° revealed a suppressive effect of loss of D-stat on hopTum-l/Y hemocyte numbers since hopTum-l/Y;D-stat6346/+ males had significantly fewer circulating blood cells than did their hopTum-l/Y;His3-GFP/+ siblings (Table 5). Hence D-stat product could be required for both hemocyte proliferation and hemocyte differentiation in response to hopTum-l-encoded product. D-stat6346 null mutants are larval/pupal lethals and exhibit small imaginal discs, indicating that D-stat plays a role in control of imaginal cell proliferation (Houet al. 1996). Our data suggest that this role extends to hemocyte proliferation.
We first looked for mxc and JAK/STAT pathway interactions in mxcG43/Y;D-stat6346/+ larvae. Total circulating hemocyte numbers as well as lamellocyte numbers were similar in mxcG43/Y;TM6c/+ and mxcG43/Y;D-stat6346/+ larvae (Table 6). Similar results were obtained for mxcG43/Y;D-statHJ/D-stat6346 larvae compared to mxcG43/Y;D-stat/+ animals (data not shown). Hence mxc effects on blood cell numbers are likely not modified by loss of D-stat. We wondered whether the same held true for Tl10b and assessed this by looking for effects of D-stat6346 in trans with Tl10b. Heterozygosity for D-stat6346 reduced lamellocyte differentiation in Tl10b/D-stat6346 compared to Tl10b/His3-GFP larvae (Table 6). Therefore the strong induction of lamellocyte differentiation caused by gain of Toll requires D-stat.
To conclude, our data indicate that hemocyte production and lamellocyte differentiation in various genetic and experimental contexts show different sensitivities to changes in D-stat dosage. D-stat quantity is apparently limiting for the strong hemocyte production induced by constitutive hop/JAK activation at 25°, whereas the lesser blood cell productions of mxcG43 or Tl10b are not affected by D-stat hemizygosity. Furthermore, as previously reported for hopTum-l lamellocyte production (Luoet al. 1997), we find that the strong lamellocyte differentiation in Tl10b also depends on D-stat product, whereas the lesser lamellocyte production of mxcG43 is unaffected in both a D-stat6346/+ and a D-statHJ/D-stat6346 context.
Lymph gland proliferation and circulating hemocytes in double inductive conditions: We examined the effects on hemocyte production of several double mutant contexts that each alone increase hemocyte numbers. Interestingly, mxcG43/Y;Tl10b/+ and mxcG43/Y;cactA2/cactA2 larvae had fewer hemocytes than did Tl10b/+, mxcG43/Y, or cactA2/cactA2 animals (Table 7; P < 0.01). Furthermore, joint activation of the Toll and the JAK/STAT pathways had comparable effects in hopTum-l/Y;Tl10b/+ and hopTum-l/Y;cactA2/cactA2 larvae (not shown). mxcG43 hopTum-l/Y larvae had 3.1 × 106 (SD ± 1.3 × 106) hemocytes/ml, which represents an increase compared to the internal wild-type control (0.7 ± 0.4 × 106 hemocytes/ml), but is considerably less than the added effects of mxcG43 and hopTum-l alone. All such double mutant animals contained abnormally large hemocytes. To understand this phenomenon, we examined lymph glands from mxcG43 hopTum-l/Y and mxcG43/Y;cactA2/cactA2 animals. Such hematopoietic organs were very fragile and difficult to dissect. When the lymph glands could be isolated, they showed severe overgrowth and contained high numbers of differentiated blood cells (mxcG43/Y;cactA2/cactA2 in Figure 6). This was also associated with intense mitotic activity, as revealed by staining with anti-phosphohistone H3 antibody (Figure 6). TUNEL label revealed no increase in apoptotic cell death in these lymph glands (mxcG43/Y; cactA2/cactA2 example in Figure 6). Lamellocytes differentiated within the hematopoietic organs; in some larvae, the glands were covered by lamellocytes, as shown by the lamellocyte-specific reporter l(2)113/28, and progressively melanized (Figure 6). Together these data suggest that differentiated blood cells in the dramatically overgrown lymph gland lobes are not released into the hemolymph. The lymph glands could end by being recognized as non-self and encapsulated by the cellular host defense system of the animal.
Hemocytes in mcx and D-stat/+ or D-stat/Tl10b trans-heterozygotes
Hemocytes in mxc and Toll signal g.o.f. double mutants
mxc, PcG, and trxG genes and larval hemocyte control: During Drosophila and vertebrate development, trxG and PcG genes control the expression of many targets, including the Hox genes. Furthermore, mammalian hematopoiesis is a target of trxG and PcG regulation (reviewed in Takihara and Hara 2000; Muller and Leutz 2001; Raaphorstet al. 2001). Hence, we asked whether Drosophila larval hematopoiesis depended on PcG and trxG gene products to the same extent as homeotic genes do for identity specification along the anterior/posterior body axis. Indeed, homeotic transformations of single PcG mutants are synergistically enhanced by adding a second PcG mutation in trans (Jürgens 1985), whereas trxG mutations in trans suppress PcG mutant phenotypes (Kennison and Tamkun 1988). Hemocytes were counted for mxcG43/Y in trans with alleles of Sex comb on Midleg (Scm), Polycomb-like (Pcl), Polycomb (Pc), and Posterior sex combs (Psc), which strongly enhance adult homeotic transformations of the viable mxcM1 allele (Sagetet al. 1998). We also tested loss of extra sex combs (esc). None of the five mutations affected control or mxcG43 hemocyte numbers or ratios (data not shown). This suggests that Drosophila PcG genes are required differently in larval hemocyte production and in HOM/Hox gene regulation. Indeed, mxc, Pcl, Pc, Scm, and Psc exert common negative control on the latter process, whereas mxc alone seems critically required for the former.
—Lymph glands in mxc;cact and mxc;hopTum-l double mutants. (A) Whole view of dramatically overgrown lymph gland chain from mxcG43/Y;cactA2/cactA2 larva, stained with anti-phosphohistone H3 antibody. (B) Enlarged view of detail of lymph gland in A reveals anti-phosphohistone H3 antibody label in dividing cells. (C) Detail of Tunel-labeled mxcG43 hopTum-l/Y lymph gland. Apoptotic cells are indicated by arrowheads. (D) mxcG43hopTum-l lymph glands contain differentiated hemocytes; lymph glands were squashed with a coverslip after dissection in 1× PBS. (E) Lamellocytes differentiate inside overgrown mxcG43hopTum-l/Y lymph glands, revealed by β-galactosidase expression of the l(2)113/28 reporter line. (F) X-Galstained mxcG43hopTum-l/Y;l(2)113/28 /+ lymph gland lobe, covered by lamellocytes and partially melanized (arrowhead); same scale as E. Bars in A and E, 150 μm; bars in B, C, and D, 40 μm.
Implication of trxG products in hematopoiesis was tested in trans-heterozygous mxc;trxG genetic contexts, using trithorax (trx), moira (mor), and brahma (brm) mutations that counteract PcG homeotic phenotypes in trans (FlyBase 1999). trx and mor mutations had no effect on wild-type or mxc hemocyte numbers and ratios. Interestingly, whereas lymph glands of brm2/+ were no different from wild type (not shown), loss of brm reduced circulating hemocytes in brm2/+ and in y1 mxcG43/Y;brm2/+ larvae (Table 8). Similar results were observed for brm2 trxE2/+ and y1 mxcG43/Y;brm2 trxE2/+ larvae, indicating that brm alone affects circulating hemocyte numbers. The trxG protein BRM is similar to yeast SWI/SNF chromatin-remodeling proteins. The Drosophila gene domino also encodes SWI/SNF chromatin proteins that are, like their mammalian homologs, involved in control of cell proliferation (reviewed in Gebuhret al. 2000; Ruhfet al. 2001). To further characterize the role of brm in hemocyte control, we examined hemocyte numbers in Tl10b/+ and hopTum-l/Y larvae carrying a brm2 allele (Table 8). The brm2/+ context had no effect on Tl10b/+ phenotype, whereas hopTum-l/Y;brm2/+ larvae contained significantly fewer hemocytes than did their hopTum-l/Y;His3-GFP/+ siblings (P < 0.01). Lamellocyte ratios remained unchanged (respectively, 28.3 and 28.4%); hence brm2 does not block hopTum-l-induced lamellocyte differentiation. Together, these data show, first, that two Drosophila SWI/SNF proteins, BRM and DOM, participate in hematopoiesis control and, second, that upregulation of hemocyte production by Toll is less sensitive to the level of brm product than upregulation either by activation of JAK or by loss of mxc.
DISCUSSION
mxc controls blood cell proliferation and differentiation: Here, we confirm that wild-type mxc directly regulates hemocyte proliferation and differentiation in Drosophila larvae. Gateff and Mechler (1989) reported modification of these processes by mxc l(1)mbn alleles, which were screened as causing hyperplasia and neoplasia. We examined hematopoietic phenotypes of three mutants that were isolated on the basis of other criteria (Santamaria and Randsholt 1995; Docquieret al. 1996). Hematopoiesis was not analyzed in strong mxc hypomorphic or amorphic embryos, since the maternal mxc component allows normal development until the second instar. Nor did we study mxc/+ females alone or in trans with other mutations, since no haplo-insufficiency has ever been observed for any set of mxc phenotypes. We show that overproliferation and premature lamellocyte differentiation take place in the hyperplastic lymph gland lobes and that increased proliferation is also seen in circulating cells. Lymph gland overgrowth is progressively stronger from mxcG46 to mxc16a-1 (Table 1) and affects more and more posterior lobes, in good agreement with Shrestha and Gateff (1982) who described the whole hematopoietic organ of mxcmbn1 as hyperplastic. Lanot et al. (2001) proposed that posterior lymph gland lobes contain hematopoietic blast cells, similar to bone marrow cells, which can be solicited to differentiate into given blood cell lines. Our data support the notion that anterior lobe prohemocytes are more readily solicited to divide and differentiate than are posterior ones.
Hemocytes of mxc, Tl, hopTum-l, and brm trans-heterozygotes
Interestingly, the stronger mxc alleles showed lower numbers of circulating crystal cells that are about 10 times less frequent in mxcmbn1 and mxc16a-1 larvae compared to wild type. It has been suggested that crystal cell numbers might also be reduced in hopTum-l/Y animals at non-permissive temperatures (Lanotet al. 2001). This could in both cases reflect recruitment of crystal cells in the melanotic capsules of mutant larvae. Still, such capsules were far from always detected in mxc larvae, since only up to 25% mxcmbn1/Y animals developed pseudotumors. Hence mxc might affect crystal cell development. During hematopoiesis, plasmatocytes and crystal cells develop from a common pool of cells expressing the GATA protein Serpent (reviewed in Fossett and Schulz 2001; Meister and Govind 2002). The plasmatocyte cell lineage is specified by the glial cells missing (GCM) conserved transcription factor; crystal cell lineage development depends on expression of the Acute Myeloid Leukemia 1 (AML-1)-like transcription factor Lozenge and is repressed by the friend of GATA protein U-shaped (USH; Lebestkyet al. 2000; Fossettet al. 2001; Fossett and Schulz 2001). As some crystal cells are still present in the most severe mxc mutant, we propose that procrystal cell proliferation or differentiation into crystal cells, rather than crystal cell lineage specification per se, might be affected. Future investigation will determine whether modifications of mxc product levels also affect crystal cell precursors in the lymph glands and, if so, through which targets.
Phagocytic cell types in mxc mutants: Circulating hemocytes in mxc larvae consistently exhibited several percent of phagocytic cells, which we called podocytes (Figure 2), like the phagocytic cells described by Shrestha and Gateff (1982) as a prevailing cell type in mxcmbn1 larvae. Like lamellocytes, such podocytes are rarely seen in wild-type larvae (<0.1% of circulating cells). We examined wild-type hemocytes from mid-third instar until 20 hr after pupariation without finding large amounts of similar cells. Changes in phagocytic cells from rounded plasmatocytes to flattened macrophages with membranous extensions and changed adhesion capacities normally occur at the end of the third instar and can be induced before by increasing ecdysone titer (Lanotet al. 2001). As suggested by Rizki (1978), podocytes might represent an intermediary form between these phagocytic cell types, which would differentiate in greater numbers in mxc mutants before pupariation. Their spindle shape suggests changed adhesion features and raises the question whether these cells could play an invasive role in mxc mutants.
mxc invasiveness and neoplasia: mxc hematopoietic tissue from different alleles divides autonomously when transplanted into a wild-type host. Proliferative capacities of mxc lymph glands appear higher than those of larval brain or imaginal disc cells mutated for tumor suppressor genes. Indeed, growth of lethal giant larvae, brain tumor, or discs large transplants required 9-12 days at 25° (Woodhouseet al. 1998), whereas mxc transplants increased after 5 days at 23°. hopTum-l lymph gland cells, grown at the restrictive temperature of 29° before and after transplantation, also increase after 3-5 days in wild-type females (Hanratty and Ryerse 1981). This might reflect differences in proliferation control mechanisms between the loosely structured lymph gland tissue where cell division is inducible and the highly structured larval disc and brain tissue (Harrisonet al. 1995). Such differences in growth control might explain why mxc and other hematopoiesis mutants exhibit blood cell overproliferation together with reduced imaginal discs and brains (Watsonet al. 1991; Toroket al. 1993; Sagetet al. 1998).
Contrary to the data reported by Gateff and Mechler (1989), we found no lessened viability in females transplanted with mxc lymph glands, even 3 weeks after transplantation. This result was unexpected for the mxcmbn1 and mxc16a-1 pupal lethal alleles and less so for mxcG43, which shows partial viability at 23°. As mxc mutations are temperature sensitive (Sagetet al. 1998), 23° may not induce sufficient proliferation to endanger the transplanted hosts. Indeed, at 29° hopTum-l is lethal and transplanted hopTum-l lymph gland tissue kills a wild-type host, whereas hopTum-l animals are viable at 25°. We propose that loss of mxc function leads to hematopoietic neoplasia, invasiveness, and altered blood cell composition but not necessarily to a lethal malignant transformation.
Hematopoietic defects and other mxc phenotypes: Severity of hematopoietic defects was correlated with that of other mxc phenotypes. Previous genetic studies showed mxcG46 as a viable and mxcG43 as a medium severe allele, whereas mxcmbn1 is a very strong and mxc16a-1 an even stronger hypomorph. mxc adult males exhibit ANT-C and BX-C gene gain-of-function-like homeotic transformations, with increasing penetrance and expressivity from mxcM1 via mxcG46 to mxcG43 (Santamaria and Randsholt 1995). mxcmbnSO, mxcmbn1, and mxc16a-1 larvae die before the pharate stage with ectopic homeotic gene expression in many discs (Sagetet al. 1998), hence strong HOM gene deregulation is also present in these mutants. Severity of germline proliferation defects increases similarly from partial mxcM1 sterility to reduced mxcG43 gonads in males and to almost no detectable gonad development in mxcmbn larvae (Docquieret al. 1996). mxc animals show increasing developmental delays from mxcM1 to mxc16a-1 (the strongest allele that reaches the third larval instar), whereas imaginal discs and brains of some mxcmbn1 and all mxc16a-1 larvae are reduced in size and amorphic mxc alleles are lethal in clones (Sagetet al. 1998). Mitotic figures or metaphase chromosome morphology are not visibly affected by loss of mxc (Sagetet al. 1998), hence mxc+ could have a more subtle effect on cell division rates, possibly through the cell cycle. Interestingly, recent data suggest several links between PcG function and regulation of the cell cycle (reviewed in Brock and van Lohuizen 2001).
For pseudotumor formation, lymph gland overgrowth, and mitosis rate in circulating hemocytes, we found increasing severity from mxcG46, mxcG43, mxcmbn1 to mxc16a-1. The exception was the lower number of circulating hemocytes in mxc16a-1 compared to mxcmbn1 larvae (Table 1). Proliferation and survival of larval imaginal cells are more affected by mxc16a-1 than by mxcmbn1 (Sagetet al. 1998), hence this lower hemocyte density might reflect the final outcome of increased lymph gland and hemolymph cell division together with lesser survival of circulating hemocytes. Alternatively, mxc16a-1 might affect hemocyte identity in the lymph glands more than mxcmbn1, and fewer mxc16a-1 cells would consequently be released into the body cavity (see below).
mxc function in relation to the Toll pathway: Several mutant phenotypes suggested that mxc and the Toll pathway could regulate a common set of processes. First, both gain of Toll signaling and mxc mutations induce lymph gland and hemocyte overproliferation. Furthermore, mxc and cact mutants exhibit decreased adhesion among fat body cells, while overexpression of the target of Toll signal Dorsal/Rel and loss of mxc both induce salivary gland atrophy (Qiuet al. 1998; O. Saget, personal communication). Our genetic analysis shows that loss of Toll signal is epistatic over loss of mxc in circulating hemocyte production, indicating that increased hemocyte numbers in mxc mutants require the Toll pathway. But Toll signal g.o.f. and mxc phenotypes also show striking differences since lamellocyte differentiation is more strongly induced and pseudotumors are much more frequent in Toll g.o.f. larvae. A further difference is that crystal cell numbers are not affected in Toll g.o.f. mutants (Lanotet al. 2001). We interpret these differences as an indication that Toll signal and mxc represent different inputs, which both regulate wild-type hemocyte production. Loss of the Toll input could render lymph gland cells unable to respond to loss of mxc.
hop and D-stat in hematopoiesis control: Contrary to a study of larvae with lessened Toll signal (Qiuet al. 1998), we found no modification of circulating hemocyte numbers in wandering larvae with total loss of hop. Amorphic hop larvae survive until the late larval/pupal stage because of maternal hop product perdurance (Perrimon and Mahowald 1986). Hence wandering hop- larvae may not be sufficiently depleted of hop/JAK to show defective hemocyte production. Alternatively, this result could support the view (Luoet al. 1997) that in plasmatocytes hop/JAK regulates larval capacity to respond to proliferative and differentiative signals. Therefore, only the mutant, overactive forms of hop/JAK encoded by hopTum-l and hopT42 (Luoet al. 1997) would affect hemocyte production. Loss of hop might not change basic hemocyte production but could render the system less able to respond to an infection or immune challenge.
Our comparison between hopTum-l/Y and hopTum-l/Y; D-stat6346/+ males (as well as hopTum-l/+ and hopTum-l/+; D-stat6346/+ females) revealed that D-stat product is involved in hop g.o.f. plasmatocyte overproliferation. This possibility was previously suggested by Zeidler et al. (2000) and is upheld by the fact that D-STAT can bind to the D-raf promoter and activate its transcription (Kwonet al. 2000). Luo et al. (2002) recently reported that JAK g.o.f.-mediated hemocyte overproliferation and lamellocyte differentiation both require the D-Raf/D-MEK/mitogen-activated protein kinase pathway, linking again in a common regulatory network these two aspects of Drosophila hematopoiesis. If active in both, then D-STAT functions like its mammalian homolog STAT5, which regulates both proliferation and differentiation of hematopoietic cells (Nosakaet al. 1999; reviewed in Luo and Dearolf 2001). We observed a critical requirement for D-stat product on hemocyte numbers only in the hop g.o.f. context at 25°; indeed, loss of D-stat in D-statHJ/D-stat6346 larvae had no effect on hemocyte numbers of mxcG43/Y, and loss of one D-stat copy had no effect on plasmatocyte numbers of Tl10b/Y larvae. Gain of hop induced a >10-fold increase in circulating hemocytes whereas the gain-of-Toll or loss-of-mxc contexts that we examined induced but 3- to 4-fold increases. Therefore, this apparent difference in sensitivity to D-stat dosage could reflect a threshold situation, where only stronger hemocyte overproductions are visibly affected when D-stat is reduced.
Lamellocyte differentiation and D-stat: Mutation of D-stat partially suppresses pseudotumors and lamellocyte differentiation in hop g.o.f. (Houet al. 1996; Yanet al. 1996; Luoet al. 1997; this article). Lamellocyte differentiation induced in Tl10b/+ larvae also depends on D-stat dosage (Table 5). We were unable to establish whether Tl10b-induced lamellocyte differentiation depends on hop/JAK, since hopVA275/Y;Tl10b/+ larvae, in which JAK is absent, died before the third instar. As heterozygosity for D-stat suppresses +/Tl10b lamellocyte production, we propose that the Toll pathway is upstream of JAK/STAT signaling in this process. This would confirm the hypothesis of Mathey-Prevot and Perrimon (1998) who speculated that Toll might be upstream of JAK/STAT in hemocyte differentiation. Lagueux et al. (2000) have recently shown similar sequential effects of these two pathways on induction of a complement-like protein, which could have important roles in defense response to infection of Drosophila larvae. As argued above, the lack of effect of D-stat mutation in mxcG43 larvae could indicate that lamellocyte production in this mutant was too low to be sensitive to a reduction in D-stat product.
Lymph glands and circulating hemocytes in double stimulated hematopoiesis conditions: Double-mutant contexts, which each alone increase hemocyte production, yielded intriguing results of reduced hemocyte numbers compared to single mutants. This was true for combinations associating mxc with gain of function of Toll or of JAK, but also for double Toll signal and JAK g.o.f. contexts. The double-mutant animals contained abnormally large hemocytes with numerous inclusions or vacuoles and showed delayed development (24-48 hr in wandering larvae). This could reflect a lesser circulating hemocyte production, as in severe domino mutants (Braun et al. 1997, 1998; Ruhfet al. 2001). Alternatively, hemocytes could be attached to the imaginal discs and involved in phagocytosis, as in proliferation disruptor (prod) mutants that show delayed development and intense imaginal cell death (Toroket al. 1997). Lymph glands from mxcG43 hopTum-L/Y and mxcG43/Y;cactA2/cactA2 larvae showed dramatic overgrowth and extreme fragility, associated with intense mitotic activity and the presence of numerous differentiated hemocytes in the glands. The enlarged lymph glands were sometimes encapsulated by lamellocytes and melanized, recalling phenotypes of medium severe domino mutants that show massive overgrowth and blackening of the lymph glands together with rare, abnormally large circulating hemocytes (Ruhfet al. 2001). Recently, Luo et al. (2002) reported that hop g.o.f. animals without a functional D-raf pathway also exhibit overgrown lymph glands and dramatic reductions in circulating blood cell numbers and attributed this phenotype to a requirement of D-raf signaling for cell survival in the lymph glands, downstream of hop g.o.f.-induced proliferation. All these phenotypes illustrate that the final number of circulating larval hemocytes depends on several processes in the lymph glands, including control of prohemocyte division rates and of hemocyte differentiation (hop, D-stat, Toll, and mxc) but also hemocyte survival (dom, D-raf; Braun et al. 1997, 1998; Luoet al. 2002) and hemocyte capacity to cross the basement membrane (dom). A common explanation for all these data could be that strong deregulation of one or joint deregulation of any two of these processes could lead to production of cells whose modified characteristics (cell surfaces) hinder their passage into the body cavity or whose modified identities prevent survival. Overgrown mxcG43 hopTum-l/Y or mxcG43/Y;cactA2/cactA2 lymph glands showed numerous differentiated hemocytes and no enhanced cell death, suggesting that basement membrane passage rather than cell survival is affected in these animals. On the other hand, hemocyte identities would be so modified in strong dom1 mutant larvae that cell survival is impossible, hence the lack of circulating hemocytes and melanized lymph glands containing abnormal dying cells in these animals (Braunet al. 1997). This agrees with the fact that dom1 is not rescued by stimulated hemocyte production, as in dom1/dom1;Tl10b/+, hopTum-l/Y;dom1/dom1 and dom1/dom1;cactA2/cactA2 animals (Braunet al. 1998) or in mxcG43;dom1/dom1 and mxcmbn1;dom1/dom1 larvae (M. Meister, personal communication).
mxc, PcG genes, and trxG genes in Drosophila hematopoiesis control: Several lines of evidence suggest that mammalian hematopoiesis depends on Hox genes. First, blood stem cells express a number of Hox genes; second, disruption of specific Hox genes in mice causes hematopoietic defects; finally, overexpression of individual Hox genes in hematopoietic cells can induce leukemogenesis (reviewed in Magliet al. 1997; Chiba 1998). Furthermore, murine PcG genes—M33, Bmi-1, Mel 18, and eed—that maintain Hox gene expression during development also participate in hematopoiesis regulation (reviewed in Takihara and Hara 2000; Raaphorstet al. 2001). An attractive, simple hypothesis would be that loss of mammalian PcG genes affects hematopoiesis through deregulation of Hox genes in hematopoietic cells, as PcG mutations modify anterior-posterior identity specification (reviewed in Pirrotta 1998; van Lohuizen 1999). In view of this, we asked whether other PcG mutations associated with mxc could affect hematopoiesis. We observed no hematopoietic phenotype interactions between mxc and Scm, Pcl, Pc, or Psc mutations, although these genes together regulate Drosophila HOM-Hox genes (Sagetet al. 1998) and include the M33 homolog Pc and the Bmi-1 and Mel 18 homolog Psc. Furthermore, no interaction was found between the eed homolog esc and mxc, although eed represses murine hematopoietic cell proliferation (Lessardet al. 1999) as mxc does in Drosophila larvae. Thus mxc hematopoiesis defects behave like mxc germline proliferation defects, which according to genetic analysis are also independent of other PcG gene products (Docquieret al. 1996). Some of these PcG genes may play a role in larval hematopoiesis, since we examined only trans-heterozygous mutant contexts. But in contrast to their role in segmental identity specification, no haplo-insufficiency for such an effect was detected in hematopoiesis. Hence Drosophila requirements for PcG gene products are different in A/P identity specification and in hematopoiesis. As mammalian hematopoiesis involves specification of many different cell lineages and differentiation of many tissues and cell types (reviewed in Orkin 2000), PcG genes may well have been recruited during evolution to control a number of these steps, which have no Drosophila equivalents. In agreement with this, a recent review (Raaphorstet al. 2001) underlined the role of PcG genes as regulators of mammalian lymphopoiesis.
Three trxG genes were tested for effects on hemocyte production. trx mutations had no effect on hemocyte numbers either, although overexpression of the human trithorax homolog MLL is associated with many acute myeloid or lymphoblastic leukemias (reviewed in van Lohuizen 1999) and MLL-/+ mice suffer severe hematopoietic defects (Yuet al. 1995). We found that overproduction of hemocytes in mxc larvae depends on the transcriptional activator BRM. BRM is homologous to yeast SWI2, a DNA-stimulated ATPase that is part of the large SWI/SNF protein complex that modifies target transcription by changes in chromatin structure (Tamkunet al. 1992; Dingwall 1995; Vignaliet al. 2000). In larvae, BRM activates HOM gene expression and is required for imaginal disc cell viability (Elfringet al. 1998). BRM has many targets, so our results could reflect a general requirement for BRM dosage on cell division. Still, brm2/+ animals show no developmental delay and loss of a brm+ copy had no effect on hematopoietic overproliferation induced by gain of Toll signal. Hence brm could well have a positive part in blood cell number or division control. brm could, as a trxG gene, be required at the same level as but antagonistic to mxc. Alternatively, since hemocyte phenotypes induced by loss of mxc and by activation of JAK are both partially suppressed by loss of brm, brm, mxc, and hop could all provide separate regulatory inputs, which together control hematopoietic cell divisions and cell density and cell survival in the larva.
moira encodes a Drosophila homolog of human and yeast chromatin-remodeling factors; mor and brm interact genetically and MOR interacts physically with BRM in the same large chromatin-remodeling protein complex in the embryo (Crosbyet al. 1999). Yet mor mutations in trans had no effect on hematopoiesis regulation. One possible explanation might be that composition of the SWI/SNF protein complex is different in embryos and in the lymph glands. Alternatively, as previously argued for the PcG, haploidy for mor (or for trx) may not reduce gene products enough to cause a mutant phenotype.
We found that three genes, mxc, brm, and dom, required for maintenance of HOM gene expression patterns are involved in control of Drosophila hematopoiesis. dom, like brm, encodes SWI2/SNF2 family DNA-dependent ATPases involved in gene expression control through modulation of chromatin structure (Ruhfet al. 2001). Interestingly, hypomorphic dom phenotypes indicate that dom, like mxc, negatively regulates lymph gland cell proliferation and maintains lymph gland cell identities (Braun et al. 1997, 1998; Ruhfet al. 2001), whereas brm activates proliferation. These effects recall dom and mxc repression and brm activation of HOM gene expression (Ruhfet al. 2001), suggesting the possibility that in the lymph glands, dom, mxc, and brm could participate in a common mechanism of proliferation and identity maintenance, which could involve modulation of chromatin structure.
mxc controls blood cell proliferation and differentiation but not lineage specification: We have shown that mxc functions as a cell autonomous regulator of cell divisions in the lymph glands, as well as in circulating blood cells, and that loss of mxc favors differentiation of plasmatocyte lineage-specific cells such as podocytes and lamellocytes. All blood cell types found in mxc mutants, even podocytes, are found in wild type. Crystal cells, although fewer, were always present in mxc mutants. We interpret this as meaning that loss of mxc does not change hemocyte lineage specification in the lymph glands. Rather, mxc controls steady-state hemocyte numbers in the body cavity of the larva. mxc and brm could both provide regulatory inputs in this process, together with domino, the JAK and Toll pathways, and other products, including the cell cycle regulated Pendulin protein, the Drosophila homolog of the mammalian S6 riboprotein, and the l(3)mbn-encoded plasma membrane protein (reviewed in Dearolf 1998). Together our data indicate that wild-type mxc product in larval hematopoiesis would maintain the normal rates of plasmatocyte proliferation and of crystal cell formation, as well as the normal timing of differentiation into self-recognizing macrophages. Under this hypothesis, mxc function in hematopoiesis would still be similar to PcG function in segmental identity specification in that both ensure that normal structures develop at the right time and place.
Acknowledgments
We thank B. Limbourg Bouchon, S. Govind, D. Kimbrell, M. Lagueux, B. Lemaitre, M. Meister, and A. Shearn for fly strains and M. Meister and O. Saget for sharing unpublished results. Thanks are due to our colleagues at the Centre de Génétique Moléculaire for stimulating discussions, to B. Lemaitre and M. Meister for critical reading of the manuscript, and to two unknown reviewers for interesting comments on former versions of this work. N.R.-L. was financed by the MENRT, the Fondation pour la Recherche Médicale, and the Association pour la Recherche contre le Cancer. This work was financed partly by a grant from the Association pour la Recherche contre le Cancer to P.S.
Footnotes
-
Communicating editor: K. V. Anderson
- Received June 4, 2001.
- Accepted August 15, 2002.
- Copyright © 2002 by the Genetics Society of America