Melanotic Mutants in Drosophila: Pathways and Phenotypes

Abstract

Mutations in >30 genes that regulate different pathways and developmental processes are reported to cause a melanotic phenotype in larvae. The observed melanotic masses were generally linked to the hemocyte-mediated immune response. To investigate whether all black masses are associated with the cellular immune response, we characterized melanotic masses from mutants in 14 genes. We found that the melanotic masses can be subdivided into melanotic nodules engaging the hemocyte-mediated encapsulation and into melanizations that are not encapsulated by hemocytes. With rare exception, the encapsulation is carried out by lamellocytes. Encapsulated nodules are found in the hemocoel or in association with the lymph gland, while melanizations are located in the gut, salivary gland, and tracheae. In cactus mutants we found an additional kind of melanized mass containing various tissues. The development of these tissue agglomerates is dependent on the function of the dorsal gene. Our results show that the phenotype of each mutant not only reflects its connection to a particular genetic pathway but also points to the tissue-specific role of the individual gene.

HALF a century ago melanotic tumors in Drosophila larvae and adults were viewed as the equivalent of cancer and as events of controlled histological differentiation that could be manipulated genetically. The participation of blood cells in the formation of some melanotic tumors was reported at about the same time (Oftedal 1953; Barigozzi 1958; Rizki 1960). Black melanotic spots are found in a number of different mutants and have been called, interchangeably, melanotic tumors or pseudotumors. These “tumors” are usually not invasive and involve tumorous overgrowth only in some instances. Therefore we use the term “melanotic masses” to describe the phenotype generally and “melanotic nodules” and “melanizations” to describe more specific phenotypes.

Well-documented studies of larval melanotic masses have shown the involvement of the immune response mediated by hemocytes (Rizki and Rizki 1983; Harrison et al. 1995; Rodriguez et al. 1996; Qiu et al. 1998; Nappi et al. 2005). This cellular immune response includes cell aggregation, phagocytosis, encapsulation of foreign material, and induction of the melanization cascade (for reviews see De Gregorio et al. 2002a; Fossett et al. 2003; Meister and Lagueux 2003; Nappi and Christensen 2005). Phagocytosis is carried out by the most abundant class of hemocytes, the plasmatocytes. Lamellocytes, functioning in encapsulation, are rare in healthy larvae and increase substantially during metamorphosis and after infection. Melanization/melanogenesis is facilitated by a distinct type of blood cells, the crystal cells. Crystal cells express the enzyme phenoloxidase (Pro-phenoloxidase A1), responsible for the initiation of the melanogenesis cascade, at rate-limiting levels (Lebestky et al. 2000; Duvic et al. 2002; Sugumaran 2002; Nappi and Christensen 2005). The activation of Pro-phenoloxidase is partially controlled by the serine protease inhibitor serpin 27A (Spn27A). Spn27A mutant larvae show a melanotic phenotype and excessive melanization in response to immune challenge (Nappi et al. 2005). This phenotype is linked to the activation of the Toll pathway (Tingvall et al. 2001; De Gregorio et al. 2002b; Ligoxygakis et al. 2002a,b; Nappi et al. 2005).

The Toll pathway is responsible for the formation of melanotic masses in Drosophila (Lemaitre et al. 1995). The pathway is named after the transmembrane receptor Toll that, when activated, controls the nuclear targeting of the Drosophila NF-κB/Rel proteins Dorsal and Dif. The pathway is a major effector of innate immunity and is also involved in hematopoiesis. Constitutive activation of the pathway in Toll gain-of-function or cactus loss-of-function mutants leads to overproliferation of hemocytes, in particular lamellocytes, resulting in the formation of melanotic nodules (Lemaitre et al. 1995; Qiu et al. 1998; Lavine and Strand 2002). Jak/STAT is another signaling pathway functioning in innate immunity and hematopoiesis. Constitutive activation of this pathway in Hop^Tum mutants leads to phenotypes similar to those observed in constitutive activation of Toll, including overproliferation of hemocytes and melanotic nodules (Harrison et al. 1995; Lagueux et al. 2000; Luo et al. 2002; Evans et al. 2003; Meister 2004).

The Drosophila immune deficiency (IMD)/Relish pathway also functions in the immune response and has a connection to melanotic nodule formation. After immune challenge of larvae with constitutive expression of the peptidoglycan-recognition protein-LE, functioning upstream of the IMD pathway, melanotic masses were observed in the cuticle and hemolymph (Takehana et al. 2002).

The activation of other pathways like the Ras/MAPK in hemocytes by the expression of transgenes leads to hemocyte proliferation and formation of melanotic masses (Asha et al. 2003; Zettervall et al. 2004). There are also several loss-of-function mutations leading to similar phenotypes. These include mutations in yantar, functioning in hemocyte differentiation (Sinenko et al. 2004; Sinenko and Mathey-Prevot 2004), mutations in the ribosomal gene rpS6 that lead to hemocyte malignancies and aberrant immune system behavior (Watson et al. 1992; Stewart and Denell 1993), black-pearl (blp), a distant homolog of DnaJ (Becker et al. 2001), and the malignant blood neoplasm gene l(3)mbn (Konrad et al. 1994). Additional mutants resulting in melanotic phenotypes were identified in genetic screens and many were linked to immunity (Watson et al. 1991, 1994; Rodriguez et al. 1996; Zettervall et al. 2004).

A large number of “tumorogenic” mutants were identified before the era of whole-genome sequencing and modern methods of gene mapping. The observed phenotypes lead to gene names like tumorous (tu-s), malignant blood neoplasm (mbn-s), and aberrant immune response (air-s; reviewed in Gateff 1977; Watson et al. 1991, 1994; Lindsley and Zimm 1992). More than 30 genome annotated genes have been reported to cause melanotic phenotypes in larvae. The tumors in ∼20 of these genes were not histologically characterized and the involvement of blood cells was not established. Rather, it is commonly assumed that all melanotic masses are caused by abnormalities in the immune response.

Here we report a morphological and immuno-histochemical analysis of melanotic nodules produced by mutations in 14 genes, representing an array of pathways and cellular functions and characterized to various degrees. Our experiments are aimed at understanding if all melanotic phenotypes are linked to the cellular immune response conventionally associated with encapsulation by hemocytes. We found that nodules in half of the mutants are surrounded by blood cells, in particular, lamellocytes. Most of these mutations are in genes previously reported to function in hematopoiesis and in the immune response. Three genes, zfrp8, Su(var)205, and spag, have melanotic nodules surrounded by hemocytes, but the function of these genes in hematopoiesis or immunity is not established.

Interestingly, more than half of the mutants produce melanizations in the gut, trachea, cuticle, or salivary glands, and no hemocytes were detected surrounding these nodules. This suggests that the mechanism for their formation does not involve most aspects of the cellular immune response, except possibly melanization.

Our results show that the histological location of black masses is a reliable primary indication if encapsulation by hemocytes is involved. While most mutants in a given pathway showed similar phenotypes, we also observed tissue-specific effects particular to individual pathway components. This is seen, for instance, in hypomorphic alleles of cact, which produce melanotic aggregates not seen in Toll gain-of-function alleles. Our results show that cact may control the activity of Rel proteins in some tissues independently of Toll.

MATERIALS AND METHODS

Fly strains:

Flies were raised on standard cornmeal/molasses food at 21°. Fly stocks bearing the P-element insertions spag^k12101, dcp-1^k05606, dcp-1⁰²¹³², Ark^k11502, dpld^k08815a, skpA^G0037, skpA^G0389, and pr-set7^P{hsneo}1, the mutations Toll³, cact¹, l(3)mbn^E1, and hop^Tum, corresponding deficiency alleles, and the armadillo (arm-GAL4) driver were obtained from the Bloomington Stock Center. The hemocyte–fat-body-specific driver collagen (cg-GAL4), dronc², dronc⁵¹, Su(var)205⁰², and Su(var)205⁰⁴ were kindly provided by C. Dearolf (Massachusetts General Hospital, Boston), A. Bergmann (University of Texas, Houston), J. Abrams (University of Texas Southwestern Medical Center, Dallas), and L. Wallrath (University of Iowa, Iowa City, IA), respectively. Flies carrying UAS-EGFP-dl, UAS-ECFP-Dif, and UAS-p-Relish were obtained from T. Ip (University of Massachusetts, Worcester, MA).

GFP-marked balancers were used to identify mutant larvae. Homozygous mutant larvae were collected on egg-laying plates and incubated with additional yeast on standard cornmeal/molasses food at 23° until the individuals reached the third instar larval stage and developed melanotic nodules.

Immunostaining:

For antibody staining, third instar larvae were dissected in phosphate-buffered saline (PBS) and tissues containing melanotic nodules were immediately transferred to a glass slide. Excess liquid was removed with tissue paper and the samples were air dried for 30 sec and then fixed at 80° for 1 min. The slides were kept at −70° for several days. For each mutant samples from 10 to 20 larvae were analyzed. The samples were additionally fixed for 40 min in 4% paraformaldehyde in PBST (PBS, 0.1% Tween 20) and washed several times in PBST. Antibody staining was performed essentially as described in Whalen and Steward (1993). Anti-Hemese monoclonal antibody (H2) and antibodies specific for lamellocytes (L1) were obtained from I. Ando (Biological Research Center, Szeged, Hungary) and used at 1:500 dilution. Secondary Cy-3-conjugated anti-mouse (Jackson ImmunoResearch Laboratories, West Grove, PA) was used at 1:500. F-actin was visualized by incubation with Alexa488 phalloidin (Molecular Probes, Eugene, OR) at 1:80 for 2 hr. DNA was stained using Hoechst 33258 (Molecular Probes). Samples were mounted in Vectashield (Vector Laboratories, Burlingame, CA) and examined with a Zeiss Axioplan-2 microscope. Hematopoietic nodule images were captured using a Leica DM IRBE laser scanning confocal microscope. The images were analyzed with Image Pro Plus and Leica Microsystems software and further processed using Adobe PhotoShop.

RESULTS

Analysis of melanotic masses:

We decided to investigate if all melanotic masses indeed involve the cellular immune response, particularly the encapsulation of the mass by hemocytes. We also wanted to establish if mutations in genes functioning in the same process or in a given pathway result in the formation of melanotic nodules with similar characteristics. To this end, we analyzed whenever possible at least two mutant alleles in 14 genes, all reported to cause melanotic phenotypes. For this study we selected mutations in Toll and cact affecting the NF-κB pathway: dronc, Dcp-1, and Ark involved in apoptosis; Pr-Set7, Su(var)205, and skpA functioning in chromatin organization; hop activating the Jak/STAT pathway; and a number of genes whose functions have not been studied in detail, l(3)mbn, mxc, spag, zfrp8, and dpld (see Table 1).

TABLE 1.

Mutations causing different types of melanotic masses

		Type of melanotic phenotype
Mutation	Gene name, function, references	Ha	Gutb	Otherc	Uniqued
skpAG0037	skpA, chromatin condensation, cell cycle, Peter et al. (2002); Murphy (2003)	−	−	+	−
skpAG0389		−	+	−	−
pr-Set720	pr-set7, histone H4 methylation, chromatin structure, cell cycle, Karachentsev et al. (2005)	+	−	+	−
pr-set7P{hsneo}1/Df(3R)red3l		−	−/+	−	−
Su(var)20502/Df(2L)TE29Aa-11	Supressor of variegation, chromatin structure cell cycle, Bushey and Locke (2004); Norwood et al. (2004)	+	−	−	−
Su(var)20502/Su(var)20504		+	−	−	−
Su(var)20504/Df(2L)TE29Aa-11		+	−	−	−
dronc2/dronc51	dronc or Nc, Nedd2-like caspase, programmed cell death, Laundrie et al. (2003); Chew et al. (2004); Xu et al. (2005)	+	+	+	−
dronc2/Df(3L)AC1		−	+	+
Dcp-1k05606	Death caspase-1, programmed cell death, also affect pita gene, Laundrie et al. (2003)	−	+	−	−
Dcp-102132		−	+	−	−
Dcp-1k05606/Df(2R)or-BR6		−	+	−	−
Arkk11502	Apaf-1-related killer, caspase activator, programmed cell death, Rodriguez et al. (1999, 2002); Zhou et al. (1999); Zimmermann et al. (2002)	−	+	+	−
Arkk11502/Df(2R)BSC49/		−	+	−	−
Df(2R)P803-D1/Arkk11502		−	+	−	−
dpldk08815a	dappled, homolog of C. elegans lin-41, Rodriguez et al. (1996)	−	+	−	−
dpldk08815a/Df(2R)ST1		−	+	−	−
mxc22a-6/Y	multi sex comb, tumor suppressor, larval hematopoiesis, Santamaria and Randsholt (1995); Saget et al. (1998); Remillieux-Leschelle et al. (2002)	−	+	−	−
hopTum/+	Hopscotch, Janus kinase, Jak-STAT signaling, Harrison et al. (1995); Luo et al. (2002)	+	−	−	−
l(3)mbnE1	malignant blood neoplasm, Konrad et al. (1994)	+	−	−	−
l(3)mbnE1/+		+	−	−	−
cact A2	cactus, IκB homolog, Rel pathway, Roth et al. (1991); Belvin et al. (1995); Qiu et al. (1998)	+	+	−	+
cactA2/cactE10R01		+	+	−	+
cactG8		+	+	−	+
cact1		+	+	−	+
Toll10B/+	Toll, transmembrane receptor, Rel pathway, Gerttula et al. (1988)	+	−	−	−
Toll3/+		+	−	−	−
zfrp8M-1-1	zinc finger RP8, homolog of mammalian PDCD2 (programmed cell death 2), Minakhina et al. (2003); S. Minakhina (unpublished results)	+	−	−	−
zfrp8M-1-1/Df(2R)SM183		+	−	−	−
spagk12101	spaghetti, TPR-protein intaracting with Hsp90, Marhold (2000)	+	−	−	−

^aMelanotic nodules found in the hemocoel or in association with Hemese-positive lymph glands.

^bGut melanization not accompanied by tissue overgrowth and hemocyte encapsulation.

^cMelanizations found in other tissues, such as the trachea and salivary glands.

^dUnique melanotic mass-type-containing variable tissues.

All melanotic masses were dissected from third instar mutant larvae. The nodules are often located in the hemocoel (body cavity) but in some mutants black masses are found in other tissues. To investigate whether all the melanotic masses were associated with blood cells, we dissected the melanized tissues and determined whether the hemocyte-specific marker Hemese was expressed in or around the nodules (Figure 1 and Figure 2; Kurucz et al. 2003). To reveal the cell shape and the size of nuclei, we also stained the tissues for filamentous actin and DNA.

Figure 1.—

Melanizations and melanotic nodules. Melanotic nodules isolated from the hemocoel of dronc²/dronc⁵¹ (A and B), cact ^A2 (C and D), and spag^k12101(E and F). Trachea isolated from Pr-Set7²⁰ (G and H; note melanization within the circles), Su(var)205⁰²/Su(var)205⁰⁴ (I and J; darkening tissue and melanized spot within the circles), nodules in the intestine observed in dronc²/dronc⁵¹ (K and L), mxc^22a-6 (M and N), Ark^k11502 (O and P), and dcp-1^k05606 (Q and R). Melanization of salivary gland cells in Ark^k11502 (S and T). In B, D, F, H, J, L, N, P, R, and T, tissues were stained with anti-Hemese (red), phalloidin for F-actin (green), and Hoechst33258 for DNA (blue). Bar, 40 μm.

Figure 2.—

Melanotic nodules. Confocal crossections taken close to the surface of melanized nodules derived from (A) Hop^Tum/+, (B) l(3)mbn^E1/+, (C) Su(var)205⁰²/Su(var)205⁰⁴, (D) zfrp8^M-1-1, (E) Toll^10B/+, and (F, I, and J) cact ^A2. Muscle and tube-like structures are indicated with arrows and polyploid nuclei with arrowheads in I and J. (G and H) cact ^A2 and dl². Tissues were stained with anti-Hemese (red), phalloidin for F-actin (green), and Hoechst33258 for DNA (blue). In A–H, Hemese staining alone is shown in the top panels. The melanized foci usually do not stain and are indicated by stars. Bar, 40 μm.

Surprisingly, only half of the mutants showed the classical melanotic phenotype: nodules that are located in the hemocoel and contain hemocytes, Hemese-positive cells (Figure 1, A–F, and Figure 2, Table 1). The other mutations, previously reported to have a melanotic phenotype, generally have melanizations found in nonhematopoietic tissues such as gut and salivary glands or melanization of trachea without involvement of hemocytes (see Figure 1, G–T; Table 1).

Melanizations in nonhematopoietic tissues:

Melanizations in different regions of the larval gut were observed in mxc, dpld, Dcp-1, and Ark and less frequently in dronc and cact larvae (see Table 1). The location and frequency of the melanizations varies in different mutants. Ark^k11502 homozygous mutant larvae develop multiple black spots along the gut and sometimes in the salivary glands, while other allelic combinations of Ark show only gut melanizations. Most hemizygous and a large number of heterozygous mxc^22a-6 larvae develop melanizations in the gut walls. This was also observed by Saget et al. (1998) who reported strong melanization associated with the hindgut ring in several mxc alleles. A total of 30–50% of dpld larvae show gut melanization, while the rest die at earlier stages without a detectable melanotic phenotype. Gut abnormalities as well as melanotic phenotypes were previously described for other alleles of dpld (Rodriguez et al. 1996). Several Dcp-1 mutant alleles also show consistent gut melanization. Gut melanizations are found in 20–30% of cact and dronc larvae along with the blood-associated melanotic nodules that we discuss below.

Melanization of the trachea was observed in two mutants, Pr-Set7²⁰, a null mutation in the histone H4-K20 mono-methytransferase, and in a strong, P-element-induced mutation in the skpA (skpA^G0037) gene encoding a component of the SCF ubiquitin ligases involved in centrosome duplication (Murphy 2003; Karachentsev et al. 2005). These two genes show the largest variations of phenotypes of all the mutants studied here. Some homozygous Pr- Set7²⁰ larvae have melanotic nodules in the hemocoel and cuticle along with melanization of the trachea. In Pr-Set7^P{hsneo}1/Df(3R)red3l, melanizations are rare and found only in the gut and cuticle. A similar phenotype is observed in skpA^G0389 larvae, suggesting that the histological appearance of the melanotic phenotype in both Pr-Set7 and skpA probably depends on the strength of the allele.

Notably, all melanizations found in the gut wall were surrounded by relatively normal-looking gut tissue that does not express the Hemese marker. Moreover, Hemese is not detected in or around pigmented trachea and melanized salivary gland tissue (Figure 1, G–J, S, and T). Thus the melanization in these organs can be induced without the activation of the cellular immune response, or, more precisely, without the encapsulation response.

Melanotic nodules associated with hemocytes:

In about half of our mutants all melanotic masses were found in the hemocoel or in association with the lymph glands and they all show high levels of Hemese expression (Figure 1, B, D, and F; Figure 2). In this study we call this phenotype “melanotic nodules.”

Criteria for classifying melanotic nodules were suggested by Watson et al. (1991, 1994). On the basis of the dissection of nodules and the morphology of lymph glands, Watson and colleagues divided mutants with melanotic phenotypes into two classes. The class 1 mutants produce melanotic nodules as a “response of an apparently normal immune system to the presence of abnormal target tissues” such as brain or imaginal disc tissue. This process is often called an “autoimmune response” (Watson et al. 1991; Nappi et al. 2004). Class 2 represents the “true blood cell tumor” mutants showing overgrowth of hematopoietic organs, leading to blood cell aggregations that form melanotic nodules. In both classes, the cellular immune system is believed to be activated and to trigger encapsulation and melanization.

Our melanotic nodule group consists partly of mutants with well-defined tumor phenotypes—hop, l(3)mbn, cact, Toll—and mutants in which the melanotic phenotype has not been described in detail before—Su(var)205, spag, and zfrp8 (alleles and phenotypes are listed in Table 1).

P-induced lethal mutations in spag (spag^k12101) cause multiple nodules surrounded by hemocytes (Figure 1B). The lymph glands in these mutant larvae are about two times larger than wild-type glands. The nodules are located symmetrically along the body wall and often replace imaginal discs, suggesting that the formation of melanotic nodules may be provoked by an immune response to the self-tissue. The enlargement of lymph glands in this case may happen as part of this autoimmune reaction. Therefore, this is the only mutant that we analyzed that could be included in the Watson class 1 genes. The other molecularly characterized class 1 gene, krz, was described by Roman et al. (2000). Krz is expressed in the nervous system and fat body and mutants have melanotic nodules in the fat bodies that are surrounded by hemocytes.

All other mutations with melanotic nodules tested in our study cause overproliferation of hemocytes and belong to class 2, following the Watson classification. RpS6/air8 and hop are original members of this class (Watson et al. 1991, 1992, 1994). How mutations in RpS6/air8 lead to aberrant immune response and to hemocyte overproliferation is not known. It is, however, known that gain-of-function mutations in hop (hop^Tum) and also expression of an activated hop transgene result in the constitutive activation of the Jak/STAT pathway, resulting in hemocyte proliferation, lamellocyte differentiation, and constitutive immune response (Luo et al. 2002; Asha et al. 2003; Agaisse and Perrimon 2004; Zettervall et al. 2004). Similar phenotypes are produced by overactivation of the Nf-κB/Rel pathway in the gain-of-function Toll^10B mutant and the loss-of-function cact^A2 mutant (Qiu et al. 1998). Our results confirm the hematopoietic origin of all nodules found in hop^Tum and Toll (Toll^10B and Toll³) mutants, where Hemese-positive cells aggregate to form masses with one or multiple melanized foci. The shapes of cells in these masses are deformed and F-actin is increased (Figure 2, A and E).

The phenotype of melanotic masses found in cact larvae stands apart. About half of the melanotic masses are Hemese positive and are similar to the nodules seen in Toll mutants, while others have additional features that will be discussed below.

Melanotic nodules in l(3)mbn, encoding a plasma membrane protein, result from excessive proliferation of hemocytes and abnormal differentiation of these cells into giant plasmatocytes and lamellocytes. (Konrad et al. 1994; Kurucz et al. 2003). Our analysis shows that all l(3)mbn^E1 melanotic nodules are found in the hemocoel and are covered by large, flat F-actin/Hemese-positive cells, stretched over the surface of the nodules (Figure 2B). Often hemocytes on the surface of the melanized nodules stain more strongly with anti-Hemese antibody than other hemocytes (Figure 2). Su(var)205 nodules look similar to those observed in l(3)mbn but they are associated with the lymph gland lobes. Homozygous or hemizygous mutants in zfrp8 (Minakhina et al. 2003; S. Minakhina, unpublished results), a homolog of mammalian PDCD2 (programmed cell death-2), also cause melanotic nodules in the lymph glands. Because nodules caused by mutations in these two genes are always found in association with the lymph gland, the genes may have an essential function in hematopoiesis.

Encapsulation of melanotic nodules:

All melanotic masses found in the hemocoel or in association with the lymph gland contain hemocytes. The structure of the nodules, size, shape, level of melanization, and the amount of surrounding tissue vary not only between different mutants, but also within one larva. To further characterize the tissue surrounding the melanotic nodule and to investigate whether all the nodules result from encapsulation by lamellocytes, we used, in addition to the hemocyte-specific antibody (Hemese; Figure 2), the L1 antibody, specific for lamellocytes (Asha et al. 2003; Kurucz et al. 2003). Crystal cells are difficult to detect, because they are unstable and are destroyed once they have caused the initiation of the melanization cascade by releasing pro-phenoloxydase (Lanot et al. 2001; Evans and Banerjee 2003).

The cells in immediate contact with the melanized tissue in spag, mbn, Hop, Toll, and Su(var)205 mutant larvae are L1 positive, indicating that they are lamellocytes (Figure 3, A–E). This suggests that the encapsulation by lamellocytes and the proposed function of crystal cells are sufficient for the formation of melanotic nodules.

Figure 3.—

Participation of lamellocytes in melanotic nodule encapsulation. Confocal crossection through the central part of melanotic nodules derived from (A) l(3)mbn^E1/+, (B) spag^k12101, (C) hop^Tum/+, (D) Toll³/+, (E) Su(var)205⁰²/Df(2L)TE29Aa-11, (F and G) zfrp8^M-1-1/Df(2R)SM183, and (H, I, and J) cact¹ and cact^G8. Tissues were stained with L1-lamellocyte-specific antibodies (red), phalloidin for F-actin (green), and Hoechst33258 for DNA (blue). L1 staining alone is shown in the top panels; melanized tissues (indicated with stars) generally do not stain (A, B, E, G, and I), but sometimes accumulate the L1 marker (D and F) . Bar, 40 μm.

In cact and zfrp8 mutants, there are melanotic nodules that are surrounded by lamellocytes (Figure 3, F and I) as well as nodules that are surrounded by other cells (Figure 3, G and J). In cact mutants, three types of nodules are found (see below). The zfrp8 nodules are located in the lymph gland, a tissue that contains hemocytes at different stages of differentiation (Lanot et al. 2001; Jung et al. 2005). In this case, melanized tissues can be surrounded either by differentiated lamellocytes (Figure 3F) or by small, L1-negative prohemocytes (Figure 3G).

Melanotic masses in cactus:

The majority of the melanotic nodules in all cactus alleles are located in the hemocoel. Only half of the cact^A2 melanotic nodules have the same composition as the nodules found in Toll gain-of-function mutants (Figure 3, D and H). This is surprising because Toll functions upstream from cact in the Rel pathway and in both Toll^10B and cact^A2 the NF-κB/Rel proteins Dorsal and Dif are constitutively nuclear.

The other half of the nodules consist of tissue agglomerates that contain one or more melanized foci and also tube-like structures, possibly tracheae, visceral muscles, and polyploid cells (arrows and arrowheads in Figure 2, I and J; Figure 3J). Only some of the agglomerates containing the melanotic foci are surrounded by lamellocytes (compare Figure 3, I and J). Similar masses are found at lower frequency in the homozygous cact ^G8 and cact ¹ larvae.

The differentiation of cells within cact melanotic masses into various tissues may result from uncontrolled Rel-protein activation. To check if dorsal (dl) is responsible for the tissue transformation observed in cact mutants, we analyzed the nodules derived from cact^A2, dl² double mutants. Multiple cact^A2, dl² melanotic nodules show clear hematopoietic origin (Figure 2, G and H). They are located in the hemocoel and lymph glands and consist of large, flat, F-actin-rich lamellocytes or both lamellocyte and plasmatocyte-like cells. Melanotic agglomerates containing different tissue types are not observed. The fact that in the double mutant the melanotic nodule phenotype is still detected, but dl loss of function suppresses the formation of the complex agglomerates, may be due to the redundant function of Dorsal and Dif in hematopoiesis and immunity and is consistent with cact and dl having additional functions in development.

General overexpression of dorsal is lethal:

We reasoned that upregulating the three Rel proteins in different tissues and observing the phenotypes produced by this overexpression would help us understand the cact loss-of-function phenotype. To this end, we expressed the Drosophila Rel proteins coupled to fluorescent peptides, EGFP-Dorsal, ECFP-Dif, and EYFP-Relish (Bettencourt et al. 2004) under the control of the arm-GAL4 and cg-GAL4 driver. The fluorescence observed in larvae confirmed that arm-GAL4 induces virtually ubiquitous but low expression of transgenes, while cg-GAL4 drives a high level of expression specifically in hemocytes and fat body (Asha et al. 2003).

The expression of the three Rel transgenes did not cause melanotic nodule formation but had specific effects on larval viability. The expression of EYFP-Relish does not reduce larval and adult viability regardless of driver. ECFP-Dif larvae have reduced viability (∼25% viability) when driven by arm-GAL, but cg-GAL4; ECFP-Dif animals are viable. Activation of EGFP-dl expression in immune tissues (CG-GAL4) does not affect viability; however, EGFP-dl expression driven by arm-GAL4 causes 100% lethality at the first and second instar larval stages. This phenotype resembles the phenotype of cact null alleles, which are also lethal at the first and second instar larval stages (Roth et al. 1991; Qiu et al. 1998) and suggests that the essential function of Cactus during postembryonic development is to inhibit Dorsal activity in somatic tissues and to regulate Dorsal and Dif function in the immune response.

DISCUSSION

We could subdivide melanotic masses into two categories on the basis of staining with the hemocyte-specific marker Hemese: (1) the melanotic nodules in the hemocoel or associated with the lymph glands evidently surrounded by blood cells and (2) Hemese-negative melanizations located in the gut, salivary glands, and tracheae. Absence of tissue overgrowth and lack of hemocytes surrounding the melanized foci indicate that they may result from cell death, possibly necrosis. It is not clear if activation of the melanization of these masses involves crystal cells.

Mutants in genes functioning in apoptosis generally show only melanizations. Mutants in Ark, dcp-1, and dronc develop melanizations located in the gut (see Table1). In Ark and dcp-1 mutants, no evidence of additional hyperplasia of the lymph glands or overproliferation of blood cells was observed (see also Song et al. 1997), but lymph-gland-associated nodules are sometimes seen in dronc larvae, and in some Ark alleles melanized foci are detected in the salivary glands. These variations may reflect the tissue-specific functions of the apoptosis genes or may be due to genetic background.

Mutations in Pr- Set7, skpA, and Su(var)205 show the highest divergence of phenotype. While Su(var)205 has a clear hematopoietic phenotype, both Pr-Set7 and skpA larvae show melanizations in other tissues, and the location of the melanized masses varies in different alleles. This pleiotropy may be explained by the fact that all three genes are involved in chromatin and chromosome structure and therefore may interfere with basic mechanisms of cell maintenance (Murphy 2003; Bushey and Locke 2004; Karachentsev et al. 2005).

The true melanotic nodule phenotype manifested by encapsulation of melanotic foci by lamellocytes was present in mutants with established defects in hematopoiesis and immunity, such as hop and Toll. It was also observed in mutants with no reported phenotypes in blood development, such as Su(var)205 and zfrp8. The finding of melanotic nodules in association with the lymph glands suggests that zfrp8 and Su(var)205 have an important function in blood development. Interestingly, nodules found in the lymph gland are often surrounded by lamellocytes, of which there are only a few in wild-type glands. Unlike in Su(var)205, in zfrp8 larvae melanotic capsules also can be formed by L1-negative blood cells, possibly prohemocytes. This lamellocyte-type function of the prohemocytes may indicate that these cells have the potential to differentiate into lamellocytes.

We assigned one new mutation, spag, to class 1 of Watson et al. (1991, 1994), in which melanotic nodules are formed as an auto-immune response to abnormalities in the imaginal discs. But we found it generally difficult to distinguish between class 1 and class 2 “melanotic tumors” as described by Watson et al. All melanotic nodules are associated with overproliferation of hemocytes. This overgrowth and the ensuing clumps of cells may be recognized by the immune system as “abnormal” and provoke an auto-immune response. Most melanotic nodules are surrounded by lamellocytes, similarly to the encapsulation of foreign objects.

Mutations in genes in a given pathway do not necessarily result in identical phenotypes. This is clearly illustrated by the analysis of Toll and cact mutants. Cact encodes the Drosophila IκB protein, that controls the activity of the Rel/Nf-κB transcription factors Dorsal and Dif. Activation of the transmembrane receptor Toll causes degradation of the IκB protein Cactus and translocation of activated Rel proteins into nuclei. Hence loss of function of cact and the gain of function of Toll are expected to cause similar phenotypes resulting from increased nuclear Rel activity. Indeed, such mutants exhibit comparable defects in innate immunity and hematopoiesis, including overproduction of hemocytes and multiple melanotic nodules (Whalen and Steward 1993; Belvin et al. 1995; Belvin and Anderson 1996; Govind 1996; Qiu et al. 1998; this work).

In addition, the partial loss of function of cact leads to formation of melanotic masses of complex composition. They often include patches of cells reminiscent of somatic tissues, such as muscles, tracheae, and polyploid cells, typical for fat body and gut. The formation of these tissue aggregates is suppressed in cact, dl double-homozygous mutants, suggesting that the transformation results from uncontrolled activation of the NF-κB/Rel protein Dorsal.

Unlike other Rel proteins, Dorsal has functions in addition to the immune response. Maternally deposited Dorsal is the ventral morphogen and directly activates and represses a large set of genes, thereby controlling the formation of the different germ layers. Later in development Dorsal and its inhibitor Cactus function in the maintenance of innervation of somatic muscles (Cantera et al. 1999; Beramendi et al. 2005). It is likely that in late embryos and larvae, in the absence of Cactus, Dorsal protein is constitutively active and this leads to abnormal gene expression and tissue differentiation. It has been shown before that the ectopic activation of one of the Dorsal targets, twist, can activate the muscle-specific marker tinmann and cause tissue transformations, including muscle differentiation (Furlong et al. 2001; Furlong 2004).

Why does activation of the Rel pathway in Toll mutants not cause tissue aggregates, as seen in cact? The answer may lie in the expression patterns and in the different requirements of Toll and cact. It is known that the expression pattern of Toll does not completely overlap with that of dl and cact. Toll expression is reportedly absent in postembryonic muscles and lymph glands (Gerttula et al. 1988; Hashimoto et al. 1991; Nose et al. 1992; Halfon et al. 1995; Kambris et al. 2002; Wang et al. 2005). dl is expressed in immune tissues and at neuromuscular junctions, gut, salivary glands, and trachea (Cantera et al. 1999; Beramendi et al. 2005). It is likely that the function of Dorsal is regulated in all these tissues by its cytoplasmic inhibitor, Cactus, under the control of a different transmembrane receptor, possibly one of the Toll homologs.

Some of the larval tissues appear particularly sensitive to the uncontrolled activation of Dorsal. This is manifested by the cact tissue transformation phenotype and early larval lethality of cact null mutants (Roth et al. 1991). Lethality at the same period is also observed when a dl transgene is generally expressed.

In conclusion, we propose to subdivide melanotic phenotypes into three classes, depending on the detection of hemocytes. True melanotic nodules are generally found in hematopoietic and immunity mutants. These nodules usually present lamellocyte-mediated encapsulation. Mutations in several genes form melanizations only in nonhematopoietic tissues without encapsulation. Melanotic masses consisting of tissue agglomerates reminiscent of vertebrate tumors are seen in cact mutants and may be caused by the uncontrolled activity of the NF–κB protein Dorsal.