SEARCH     advanced search my account e-alerts subscribe register
Journal home
Advance online publication
Current issue
Press releases
For authors
For referees
Contact editorial office
About the journal
For librarians
Contact NPG
Customer services
Site features
NPG Subject areas
Access material from all our publications in your subject area:
Biotechnology Biotechnology
Cancer Cancer
Chemistry Chemistry
Dentistry Dentistry
Development Development
Drug Discovery Drug Discovery
Earth Sciences Earth Sciences
Evolution & Ecology Evolution & Ecology
Genetics Genetics
Immunology Immunology
Materials Materials Science
Medical Research Medical Research
Microbiology Microbiology
Molecular Cell Biology Molecular Cell Biology
Neuroscience Neuroscience
Pharmacology Pharmacology
Physics Physics
Browse all publications
22 November 1999, Volume 18, Number 49, Pages 6875-6887
Table of contents    Previous  Article  Next   [PDF]
Control of development and immunity by Rel transcription factors in Drosophila
Shubha Govind

Department of Biology, City College and The Graduate Center of The City University of New York, 138th Street and Convent Avenue, New York, NY 10031, USA

Correspondence to: Shubha Govind, Current address: Novartis Institute for Biomedical Research, Functional Genomics, 133/2028A, 556 Morris Avenue, Summit, New Jersey, NJ 17901-1398, USA


The Drosophila Rel/NF-kappaB transcription factors - Dorsal, Dif, and Relish - control several biological processes, including embryonic pattern formation, muscle development, immunity, and hematopoiesis. Molecular-genetic analysis of 12 mutations that cause embryonic dorsal/ventral patterning defects has defined the steps that control the formation of this axis. Regulated activation of the Toll receptor leads to the establishment of a gradient of nuclear Dorsal protein, which in turn governs the subdivision of the axis and specification of ventral, lateral and dorsal fates. Phenotypic analysis of dorsal-ventral embryonic mutants and the characterization of the two other fly Rel proteins, Dif and Relish, have shown that the intracellular portion of the Toll to Cactus pathway also controls the innate immune response in Drosophila. Innate immunity and hematopoiesis are regulated by analogous Rel/NF-kappaB-family pathways in mammals. The elucidation of the complex regulation and diverse functions of Drosophila Rel proteins underscores the relevance of basic studies in Drosophila.


Dorsal-ventral polarity; immunity; hematopoiesis; muscle; Drosophila; Dorsal


Studies on model organisms, such as Drosophila melanogaster, have provided critical insights into regulatory processes in mammals because molecular mechanisms controlling specific biological processes are frequently conserved. Due to the availability of a vast array of molecular, genetic and biochemical tools, and the comparatively simple biology of Drosophila, it is possible to address in vivo many basic mechanistic questions that are still relatively difficult to approach in vertebrates. Our understanding of vertebrate biology thus continues to be influenced by fundamental findings in the Drosophila field. Herein, I review the structure and regulation of the Drosophila Rel transcription factors and their functions in different biological processes.

Three Rel/NF-kappaB transcription factors have been identified in Drosophila: Dorsal, Dif and Relish (Figure 1). Like their vertebrate counterparts, these fly proteins contain an approximately 300-amino acid long, multifunctional and conserved Rel homology (RH) domain. The extent of sequence similarity among the Drosophila Rel proteins is between 45 - 50% (Dushay et al., 1996; Ip et al., 1993; Steward, 1987). The RH domain mediates dimerization, DNA binding, and interactions with IkappaB inhibitor proteins. The Drosophila IkappaB inhibitor, Cactus, has six ankyrin repeats and a PEST domain, and it serves to retain Dorsal and Dif in the cytoplasm (Geisler et al., 1992; Kidd, 1992). Relish is a compound Rel protein, which contains ankyrin repeats in its C-terminal portion. Although similar in organization to the mammalian Rel/NF-kappaB p100 and p105 proteins, the Relish RH domain and ankyrin repeats are structurally quite different from all other known Rel or IkappaB proteins (for phylogenetic relationships among Rel/NF-kappaB and IkappaB superfamily proteins, see Huguet et al., 1997). The vertebrate Rel factors include the NF-kappaB subunits p50 and RelA, p52, RelB, c-Rel and the viral oncoprotein v-Rel (Chen and Ghosh, 1999; Gilmore, 1999; both this issue). Inducible by a broad range of stimuli such as cytokines, phorbol esters, viruses, and bacterial products (Pahl, 1999, this issue), vertebrate Rel and IkappaB proteins regulate many biological processes, including immune cell development and activation, development of the fetal liver, and glial and neuronal cell functions (Gerondakis et al., 1999, this issue; O'Neil and Kaltschmidt, 1997).

The Drosophila Rel factors also control several biological functions (Figure 2), including the specification of embryonic dorsal-ventral (d/v) polarity (Figure 2a; Belvin and Anderson, 1996; Drier and Steward, 1997; Rusch and Levine, 1996), humoral immunity (Figure 2b; Engström, 1999; Hoffmann et al., 1999; Hoffmann and Reichhart, 1997), hematopoiesis (Figure 2c; Qiu et al., 1998), and muscle development (Halfon and Keshishian, 1998). These biological processes exploit the common intracellular Tollright arrowRel protein `cascade', with homologous components in vertebrates (Figure 2). Toll, the founding member of the family of Toll-like receptors (TLRs; Rock et al., 1998) is a multifunctional transmembrane receptor that transduces an extracellular signal to cytoplasmic proteins Tube and Pelle (Hashimoto et al., 1988; Stein et al., 1991). Tube is a novel protein (Letsou et al., 1991), with no known vertebrate homolog. Pelle functions as a protein kinase (Shelton and Wasserman, 1993), whose mammalian counterpart is called IRAK (Interleukin Receptor Associated Kinase; Cao et al., 1996). The Tollright arrowTuberight arrowPelle signal results in the degradation of Cactus and activation of Dorsal/Dif, which can then enter the nucleus. Nuclear Rel proteins bind to kappaB sites within target-gene promoters and regulate their transcription. Four of the five components, Toll, Pelle, Cactus and Dorsal/Dif, have mammalian counterparts, which share molecular and biochemical features with the Drosophila proteins (Figure 2d). Of all the Rel factors, the regulation and function of Dorsal and Cactus in the specification of embryonic d/v polarity is best understood. Because Dorsal regulation and function continues to serve as a paradigm for analysis of other Rel pathways in Drosophila, it is considered in detail in the next section.

Rel factors in embryonic dorsal/ventral patterning

The d/v axis of the Drosophila embryo is established in two successive phases. In the early phase, the oocyte nucleus assumes an asymmetric anterior-dorsal position. This asymmetric location of the nucleus provides a dorsalizing signal to the adjacent follicle cells. Signaling proteins, Gurken/TGF-alpha and Torpedo/Drosophila EGF-receptor control this process. In the absence of the signal or the receptor, dorsal follicle cells differentiate as follicle cells on the ventral side, and the egg shell, secreted from the follicle, is ventralized (Ray and Schüpbach, 1996). Further description of these factors and their regulation is beyond the scope of this review.

Twelve genes participate during the second, late phase of d/v patterning (Figure 3). Ventral follicle cells of the egg chamber provide a localized factor, that is transmitted to the embryo. Shortly after fertilization, this factor somehow participates in the localized processing of Spätzle, within the fluid-filled extracellular compartment of a precellular embryo. Processed Spätzle is a ligand that activates Toll ventrally, even though Toll is present on the plasma membrane uniformly along the d/v axis (Hashimoto et al., 1991). The Toll signal is transduced across the cytoplasm via Tube and Pelle, and causes the graded nuclear localization of evenly-distributed cytoplasmic Dorsal within the embryo. While nuclei at ventral positions receive the highest concentration of Dorsal, nuclei at lateral positions have low, but detectable, levels of Dorsal. In contrast, nuclei located dorsally remain devoid of Dorsal (Steward et al., 1988). This second phase of d/v embryonic patterning can be divided into three parts (Figure 3): (1) action of Pipe, Windbeutel and Nudel in the somatic follicle cells of the ovary; (2) activation of the proteolytic cascade (Gastrulation Defective, Snake, Easter) and Spätzle in the extra-embryonic perivitelline space; and (3) the intracellular Tollright arrowTuberight arrowPelleright arrowCactusright arrowDorsal signal transduction cascade.

The 12 `maternal effect' genes (11 `dorsal group' genes and cactus; Figure 3) that participate in the second phase of d/v axis specification were discovered in genetic screens for female sterility. Homozygous females, mutant in any of these genes, produce embryos with d/v pattern defects. Females carrying null mutations in any one of the 11 `dorsal group' genes give rise to dorsalized embryos, in which all cells along the d/v axis assume a dorsal fate (Anderson and Nüsslein-Volhard, 1986). In these embryos (except those from dorsal females), Dorsal remains in the cytoplasm of the embryo (Roth et al., 1989; Rushlow et al., 1989; Steward, 1989). Loss of function of the 12th gene, cactus, results in ventralized embryos, as ventral fates are conferred on cells in lateral and dorsal positions (Roth et al., 1991). In Cactus-deficient embryos, the gradient of Dorsal is extended dorsally, resulting in ventralized embryos (Geisler et al., 1992; Kidd, 1992; Roth et al., 1991). Thus, the nuclear concentration of Dorsal directly specifies ventral, lateral or dorsal fates.

As proteins that regulate the nuclear localization of Dorsal in this second phase are required during oogenesis or very early in embryonic development, their genes are active in either the somatic follicle cells or the germline nurse cells of the developing egg chamber. Thus, the genotype of the mother, and not the father, determines the function of these genes. Therefore, these genes are termed `maternal-effect' genes. After the gradient of nuclear Dorsal is established, `zygotic' gene expression is initiated. Differences in the absolute concentrations of nuclear Dorsal in ventral, lateral and dorsal positions result in distinct sets of zygotic genes being expressed along the d/v axis. This leads to specification of cell identities along the d/v axis. Further regulation within each region then leads to regional differentiation. Thus, formation of the Dorsal gradient represents a critical regulatory link between global positional information and pattern formation.

Somatically active genes and the proteolytic cascade

The first important step that generates internal embryonic d/v asymmetry is the ventral activation of Spätzle. Spätzle is secreted uniformly from the embryo into the perivitelline fluid as an inactive precursor (Morisato and Anderson, 1994; Schneider et al., 1994; Stein and Nüsslein-Volhard, 1992). Upon proteolytic processing, a C-terminal proteolytic fragment of Spätzle serves as the ligand for the Toll receptor, and this engagement initiates the signal transduction cascade that ultimately results in the activation of Dorsal.

Processing of Spätzle is controlled by the action of gastrulation defective, snake and easter, each of which encodes zymogens of trypsin-like serine proteases (Chasan and Anderson, 1989; DeLotto and Spierer, 1986; Konrad et al., 1998). These zymogens are activated by proteolytic cleavage and are thought to act sequentially in a proteolytic cascade similar to that of the vertebrate coagulation cascade (LeMosy et al., 1999). Genetic epistasis experiments have placed Gastrulation Defective upstream of Snake, and Snake upstream of Easter (Chasan et al., 1992; Smith and DeLotto, 1994). Like Spätzle, Snake and Easter activities are freely diffusible within the perivitelline fluids, whereas Gastrulation Defective activity is not diffusible (Stein and Nüsslein-Volhard, 1992). Based on genetic experiments, Gastrulation Defective is believed to establish a localized complex of Nudel, Easter and Snake, which leads to the activation of Spätzle (Konrad et al., 1998; LeMosy et al., 1998).

How is the processing of Spätzle spatially controlled? The answer to this question, at least in part, lies in the action of the nudel, pipe and windbeutel, genes that are transcribed and translated during oogenesis in the follicle cells, hours before Toll activation. nudel encodes a modular protein with a central serine protease domain and with sequences resembling extracellular matrix proteins (Hong and Hashimoto, 1995). During oogenesis, Nudel is secreted as a zymogen. However, processing at the zymogen cleavage site in Nudel occurs during early embryogenesis. This cleavage is dependent on the protease function of Nudel itself, and it does not require the functions of other dorsal group genes. The protease domain of Nudel is uniformly associated with the plasma membrane of the early embryo (LeMosy et al., 1998). windbeutel encodes a protein that has sequence similarity to the vertebrate endoplasmic reticulum protein ERp29, and may be required for the biogenesis of a secreted factor involved in the establishment of d/v polarity (Konsolaki and Schüpbach, 1998). windbeutel and nudel transcripts are expressed evenly in all follicle cells, even though windbeutel function is required only in the ventral follicle cells (Nilson and Schüpbach, 1998).

Like windbeutel, the genetic requirement for pipe is also restricted to the ventral follicle cells (Nilson and Schüpbach, 1998), and consistent with this requirement, expression of pipe is spatially restricted to ventral follicle cells (Sen et al., 1998). Normally, pipe is repressed in the dorsal follicle by the Gurken/Torpedo signal provided by the oocyte. But when pipe is ectopically expressed in these dorsal follicle cells in a pipe-deficient background, the embryonic d/v axis is inverted with respect to the eggshell. This observation suggests that Pipe defines the orientation of the embryonic d/v axis (Sen et al., 1998). pipe encodes an enzyme similar to heparan sulfate 2-O-sulfotransferase, an enzyme that modifies glycosaminoglycan side chains of proteoglycans (Sen et al., 1998). Like Windbeutel, Pipe is expected to be active in the Golgi of the ventral follicle cells, and it is possible that Windbeutel and Pipe are both required for the modification of a proteoglycan, which is secreted from the ventral follicle cells. Such a molecule has not yet been identified, but its action, limited ventrally, is predicted to spatially restrict the activation of the protease cascade and processing of the Spätzle precursor (Anderson, 1998; LeMosy et al., 1998; Roth, 1998; Sen et al., 1998).

Intracellular cytoplasmic signaling

The intracellular portion of the maternal d/v cascade consists of the remaining four dorsal group genes, Toll, tube, pelle and dorsal, as well as cactus (Figure 4). The order in which these proteins act in the embryo has been determined from genetic experiments (Galindo et al., 1995; Grobetahans et al., 1994; Hecht and Anderson, 1993; Roth et al., 1991; Figure 3). The Toll receptor acts upstream of Tube and Pelle. The signal from Tube and Pelle, probably indirectly, leads to the phosphorylation of Cactus and Dorsal (Belvin et al., 1995; Drier et al., 1999), which causes a cytoplasmic gradient of Cactus degradation on the ventral side and the consequent nuclear localization of Dorsal (Bergmann et al., 1996; Reach et al., 1996).

The Drosophila Toll protein is a large (1097 amino acids), trans-membrane receptor, with multiple leucine-rich repeats in the extracellular domain and a highly-conserved intracellular Toll-homology (TH) domain (Hashimoto et al., 1988; Rock et al., 1998; Schneider et al., 1991). The TH domain is shared by a family of interleukin-1 receptors, mammalian MyD88 factors and plant disease-resistance proteins (Hoffmann et al., 1999; Rock et al., 1998). In humans, at least five Toll-like receptors (TLRs) have been identified (Medzhitov et al., 1997; Rock et al., 1998), and there may be other Toll-family proteins in Drosophila as well (Hoffmann et al., 1999).

The structural features of proteins downstream of Toll are also known. Tube (462 amino acids) contains an N-terminal `death domain' and C-terminal `Tube repeats' (Edwards et al., 1997; Letsou et al., 1991). Pelle (501 amino acids) has a `death domain' towards its N terminus, and a catalytic domain in its C-terminal end (Edwards et al., 1997; Shelton and Wasserman, 1993). Previously identified in proteins regulating apoptosis, `death domains' are thought to play a general role in protein-protein interactions in this complex (Edwards et al., 1997). In addition to its N-terminal RH domain, Dorsal has a C-terminal transactivation domain (Figure 5).

In the cytoplasm, the Dorsal dimer exists in a complex with Cactus (Govind et al., 1992; Isoda and Nüsslein-Volhard, 1994; Kidd, 1992; Whalen and Steward, 1993). Dorsal interacts with Tube and Pelle (Edwards et al., 1997; Yang and Steward, 1997); Tube and Pelle also interact with one another (Galindo et al., 1995; Grobetahans et al., 1994). However, an interaction of Cactus and Tube, or of Cactus and Pelle, has not been demonstrated.

There is strong evidence that active Toll recruits both Tube and Pelle to the membrane. Tube is normally localized to the inner surface of the embryo (Galindo et al., 1995). However, Tube appears to become more concentrated along the ventral midline where Toll is activated (Towb et al., 1998). Even though no such asymmetry was found for Pelle distribution in wild-type embryos, when a mutant hyperactive Toll protein was ectopically expressed at the anterior pole, high levels of both Tube and Pelle were found in the anterior sub-cortical cytoplasm (Towb et al., 1998). If Tube or Pelle are experimentally forced to the membrane, the need for Toll activation is alleviated (Galindo et al., 1995; Grobetahans et al., 1994; Towb et al., 1998). An actin-binding protein, Filamin, which interacts with both Toll and Tube, is thought to link Toll to the Tube/Pelle/Dorsal/Cactus complex (Edwards et al., 1997). These studies suggest that assembly of a multi-protein complex and its proximity to Toll ensures efficient intracellular signaling (Edwards et al., 1997; Yang and Steward, 1997).

The Tollright arrowTuberight arrowPelle signal induces the degradation of Cactus, ventrally and laterally, releasing Dorsal for nuclear import (Belvin et al., 1995; Roth et al., 1991). Cycloheximide treatment of permeabilized embryos (preventing de novo synthesis of Cactus) allowed the visualization of a gradient of Cactus degradation (Bergmann et al., 1996). Cactus is stabilized by its interaction with Dorsal, and the level of Cactus in the embryo correlates with that of Dorsal (Bergmann et al., 1996; Govind et al, 1993; Kidd, 1992; Whalen and Steward, 1993). The stability of Cactus is regulated by two independent mechanisms, each with a distinct sequence requirement. As with the vertebrate IkappaBalpha protein, signal-independent degradation of Cactus occurs through a C-terminal PEST sequence; in the absence of Dorsal, a Cactus mutant lacking the PEST sequence is stable, but can still be degraded in response to the ventral signal. In contrast, mutations towards the N terminus of Cactus, including N-terminal deletions or modifications of two pairs of serine residues, make Cactus refractile to the Toll signal (Belvin et al., 1995; Bergmann et al., 1996; Reach et al., 1996; Figure 5). While the Pelle kinase is important for signaling Cactus degradation, it is unlikely that the N-terminal serines of Cactus are its direct targets. That is, Pelle cannot phosphorylate Cactus in vitro, and the mammalian IkappaB kinases are not similar to Pelle (Karin, 1999, this issue). As with Cactus, mammalian IkappaB degradation occurs via phosphorylation of serine residues in its N terminus, and the IkappaB kinases that serve this function have been identified (Karin, 1999, this issue). IkappaB kinase-like complexes performing analogous modifications of Cactus will no doubt be found in Drosophila. Moreover, there is evidence that a ubiquitin-mediated proteasome pathway is responsible for signal-dependent Cactus degradation and nuclear import of Dorsal (Spencer et al., 1999). Signal-independent phosphorylation of three serines within the Cactus PEST domain by the Drosophila casein kinase II controls Cactus activity and levels, but not its interaction with Dorsal (Liu et al., 1997).

Gross deletion analysis of Cactus has mapped the dorsal-binding region of Cactus to the ankyrin repeats (Kidd, 1992), suggesting that the tertiary structure of these repeats is essential for Dorsal binding. However, a detailed basis for loss of Dorsal interaction in the large number of cactus alleles is still not available. This information will be useful as Cactus can also interact with Dif (Lehming et al., 1995; Tatei and Levine, 1995). By comparison to the crystal structures of the IkappaBalpha/NF-kappaB complex (Huxford et al., 1998; Jacobs and Harrison, 1998), molecular characterization of cactus alleles will provide an explanation for Rel-Cactus specificity in various biological processes.

Like NF-kappaB p50, the Dorsal RH domain has two distinct functional domains, termed here RHD1 and RHD2 (Figure 5, Ghosh et al., 1995; Govind et al., 1996; Muller et al., 1995). As described above, many functions have been ascribed to the Dorsal RH domain, including dimerization, DNA binding, Cactus interaction, Tube interaction, Pelle interaction, nuclear localization and signal-dependent phosphorylation (Drier et al., 1999; Govind et al., 1996; Ip et al., 1991; Isoda et al., 1992). The Dorsal nuclear localization signal (NLS) is not itself required for interaction with Cactus (Govind et al., 1996), even though it is believed that, as with vertebrate Rel and IkappaB proteins, Cactus interaction masks the Dorsal NLS. A serine at position 234 of Dorsal contained within RHD2 is critical for interaction with Cactus (Lehming et al., 1995). In the embryo, a mutant Dorsal, harboring an S234P mutation, destabilizes Cactus and undergoes nuclear localization along the entire d/v axis (Drier and Steward, personal, communication).

Cactus has been considered to be the primary downstream target of the Toll signal, even though it has been known for some time that Dorsal is also phosphorylated and that its phosphorylation status is affected by Toll signaling (Gillespie and Wasserman, 1994; Whalen and Steward, 1993). Drier et al. (1999) have recently shown that Dorsal is phosphorylated by the Toll signal independently of Cactus degradation. This phosphorylation is essential for the efficient nuclear localization of Dorsal on the ventral side of the embryo. Mutations of six evolutionarily conserved serines within the RH domain (Figure 5) render Dorsal non-functional and constitutively cytoplasmic. Of these six serines, S317 appears to be a critical target for signal-dependent phosphorylation of Dorsal (Drier et al., 1999). Thus, gradient formation is not simply a consequence of Cactus degradation. Cactus degradation laterally and ventrally results in the nuclear import of unmodified Dorsal to the moderate levels that are normally observed in lateral nuclei. Phosphorylation of Dorsal by the Toll signal is essential for the enhanced localization of Dorsal in the ventral-most nuclei. These observations are consistent with genetic experiments, which suggested that both Cactus and Dorsal are modified by the Toll signal (Govind et al., 1993). It is likely that distinct kinases phosphorylate Dorsal and Cactus, and thus convert the putative gradient of Toll activation to discrete step-functions of ventral versus lateral levels of nuclear Dorsal.

The role of Dorsal in establishing dorsal/ventral polarity

A direct consequence of the gradient of nuclear Dorsal protein is the subdivision of the embryonic axis into three non-overlapping territories. Dorsal regulates a number of zygotic genes in a concentration-dependent manner. As a result, three embryonic tissues - the mesoderm, the adjacent neuroectoderm and the dorsal ectoderm - are specified. High nuclear levels of Dorsal directly active twist and snail in the ventral-most region that gives rise to the mesoderm. In the presumptive neuroectoderm where nuclear Dorsal levels are lower, genes such as short gastrulation and rhomboid are activated in broad stripes. In both the ventral and lateral nuclei, high and low levels of Dorsal repress genes such as zerknüllt (zen), decapentaplegic (dpp), tolloid and twisted gastrulation. These genes are therefore expressed in the dorsal nuclei of the blastoderm, which is the future dorsal ectoderm of the embryo (Figure 6). Promoters of each of these three classes of zygotic genes have sites that bind Dorsal with different affinities, and in combination with other transcription factors, they exhibit an on/off threshold response to different concentrations of Dorsal (Rusch and Levine, 1996).

Deletion analysis of the twist and snail promoters with the lacZ reporter constructs in transgenic flies led to the identification of multiple low-affinity binding sites for Dorsal (Ip et al., 1992b; Jiang et al., 1991; Pan et al., 1991; Thisse et al., 1987, 1991). In addition to Dorsal, helix - loop - helix (HLH) proteins are also important for the establishment of the mesoderm. twist encodes a transcription factor that contains a basic helix - loop - helix motif. Embryos that are double heterozygous for dorsal and twist (dl+/+twi) show abnormal patterns of twist and snail expression, and display disruptions in mesoderm differentiation (Kosman et al., 1991). Once twist transcription is turned on in the presumptive mesoderm, it upregulates its own expression (Ip et al., 1992b). Twist also binds to the snail promoter, and the synergistic interaction between Dorsal and Twist combines the delimit the `on/off' boundary of snail at the mesoderm/neuroectoderm junction (Kosman et al., 1991; Ip et al., 1992b; Leptin, 1991; Shirokawa and Courey, 1997).

In addition to HLH proteins, other transcription factors are also important in the activation of ventrally-active promoters. For example, the Drosophila CREB-binding protein (dCBP) is required for the Dorsal-dependent activation of twist; in dCBP mutants, twist expression does not occur, and, similar to twist embryos, dCBP embryos have a dorsalized phenotype (Akimaru et al., 1997). There is also genetic and biochemical evidence for Dorsal interaction with proteins of the TFIID complex (TAFII110 and TAFII60) in the activation of snail and twist promoters (Zhou et al., 1998).

In the neuroectodermal regions, low levels of Dorsal activate the expression of many target genes. short gastrulation and rhomboid are expressed in broad stripes, whereas lethal of scute, m7 and m8 genes of Enhancer of Split complex and single minded are expressed in narrower stripes (reviewed in Rusch and Levine, 1996). The best studied of these Dorsal target genes is rhomboid. The rhomboid promoter has high-affinity Dorsal-binding sites that are adjacent to the E-box sequences, where HLH proteins such as Daughterless and T4 (Scute) bind (Ip et al., 1992a). Mutations in either the Dorsal-binding site or the E-box sequence result in significant reductions in rhomboid promoter activity (Ip et al., 1992a). Thus, as with the twist and snail promoters, cooperative interactions between Dorsal and HLH proteins promote rhomboid gene activation. These interactions maximize the occupancy of Dorsal-binding sites, allowing efficient transcription of neuroectodermal genes in regions where Dorsal concentration is low (Gonzales-Crespo and Levine, 1993; Jiang and Levine, 1993). The rhomboid promoter also contains Snail-binding sites. snail encodes a zinc-finger protein, that represses the expression of rhomboid in the presumptive mesoderm (Boulay et al., 1987; Ip et al., 1992a), limiting its expression laterally.

Dorsal also acts as a repressor of a number of target genes. zen, dpp and tolloid are repressed ventrally and laterally by high and low concentrations of Dorsal. Therefore, their expression is restricted to the dorsal ectoderm (Figure 6). The zen, dpp and tolloid promoters have `ventral repression elements' or VREs, with high-affinity Dorsal-binding sites, which are efficiently occupied by the low lateral and high ventral levels of nuclear Dorsal (Huang et al., 1993; Ip et al., 1991; Kirov et al., 1994). Transcriptional repression by Dorsal is mediated by co-repressors such as DSP1 or Groucho, which bind to the zen and dpp promoters in close proximity to Dorsal (Dubnikoff et al., 1997; Huang et al., 1995; Lehming et al., 1994). Groucho does not bind to DNA directly, but is required for the repression of many genes by interacting directly with a number of DNA-bound transcriptional repressors (Fisher and Caudy, 1998). In embryos lacking maternally-expressed Groucho, zen and dpp expression is observed all along the d/v axis (Dubnikoff et al., 1997).

Even though Dorsal can act as a repressor, it appears to intrinsically be a transcriptional activator. For example, when an `activator' site in the twist promoter was placed in the context of the zen promoter, repression was achieved. However, when a Dorsal-binding site from zen was placed within the twist promoter, or in the context of a minimal promoter, transcriptional activation was observed (Jiang et al., 1992; Pan and Courey, 1992). Moreover, Dorsal activates sequences from the zen promoter in yeast (Kamens and Brent, 1991; Lehming et al., 1994), which presumably lack Drosophila co-activators and repressors. Thus, Dorsal by itself functions as an activator, and mediates repression by recruiting co-repressors to adjacent negative response elements.

Rel factors in humoral immunity

In Drosophila, a septic wound induces a systemic response, which consists of a rapid induction of a battery of anti-bacterial peptides that includes Attacin, Diptericin, Drosocin, and insect Defensin. These peptides have a broad spectrum of activity against Gram-positive and Gram-negative bacteria. Metchnikowin and Cecropins have both anti-bacterial and anti-fungal properties, whereas Drosomycin acts specifically against fungi (Ekengren and Hultmark, 1999; Levashina et al., 1998; for reviews see Bulet et al., 1999; Engström, 1999; Hoffmann et al., 1999). The induction of anti-microbial peptides is not completely non-specific. Infection by Gram-negative bacteria elicits the upregulation of various anti-bacterial peptide genes, whereas the strongest effect of fungal infection is on drosomycin induction. In contrast, Metchnikowin is induced by all microbes tested (Lemaitre et al., 1997).

Anti-microbial peptides are synthesized primarily in the larval or adult fat body, and are secreted into the hemolymph. Regulated at the transcriptional level, high RNA levels can be detected within 1 - 2 h after infection, peaking by 4 h. The involvement of Rel factors in this acute phase gene induction was first suggested by the presence of kappaB motifs in the cecropin and diptericin promoters (Hultmark, 1993; Meister et al., 1994; Reichhart et al., 1992; Sun et al., 1991; Tryselius et al., 1992; also see citations in Engström, 1999 and Hoffmann and Reichhart, 1997, for promoter organization and requirements for induction). These observations led to the search for the expression of trans-acting Rel-like kappaB-binding proteins in the fat body as well as promoter studies to test their function (Engström et al., 1993; Kappler et al., 1993).

From such studies, two additional Rel transcription factors, Dif and Relish, were identified. The Drosophila Rel family member dif was cloned by sequence similarity to dorsal, and it was shown that Dif is not only expressed in the larval fat body, but also that its expression is induced by immune challenge (Ip et al., 1993). At the same time, dorsal expression and induction in the fat body was reported (Lemaitre et al., 1995; Reichhart et al., 1993). The relish cDNA was identified in a differential display screen that searched for induced mRNAs from bacterially-infected flies. relish RNA is expressed maternally as well as in later stages (Dushay et al., 1996). Dif and Dorsal are normally localized to the cytoplasm, but are imported into the nucleus upon bacterial challenge. This nuclear localization of Dorsal, affected by immune challenge, is regulated by Toll, tube, pelle and cactus (Lemaitre et al., 1995). The nuclear localization of Dif depends at least in part on the activity of 18-wheeler, a gene encoding a Toll family receptor (Williams et al., 1997). In protein interaction experiments in yeast as well as in in vitro binding assays, Cactus was found to bind not only to Dorsal, but also to Dif (Lehming et al., 1995; Tatei and Levine, 1995). In vitro experiments also show that Dorsal and Dif can form heterodimers (Gross et al., 1996).

The existence of d/v pathway mutants and availability of anti-microbial gene probes facilitated the dissection of the pathway leading to anti-microbial gene induction. In a series of Northern blots, various d/v pathway mutants were subjected to bacterial and fungal infection, and the induction of anti-bacterial genes (cecropin A, diptericin, drosocin, attacin and defensin) and anti-fungal genes (drosomycin and cecropin A) was measured (Lemaitre et al., 1996). This analysis showed that the wild-type functions of spätzle, Toll, tube, pelle and cactus are required for the full induction of drosomycin, and, to a lesser extent, also for cecropin A, attacin and defensin (Figure 7). In constitutively-active TollD/+ and loss-of-function cactus backgrounds, the drosomycin gene is constitutively expressed. The induction of diptericin and drosocin was largely unaffected by these mutations and remained fully inducible. Mutations in gastrulation defective, snake, easter or dorsal did not modify any aspect of the anti-microbial induction tested (Lemaitre et al., 1996).

Consistent with their inability to mount a robust immune induction of the anti-fungal peptide, Toll-deficient mutant adults are immune-compromised and quickly succumb to Aspergillus fumigatus infection, but not to infection by E. coli. (Lemaitre et al., 1996). However, flies mutant in another gene called immune deficiency (imd; Lemaitre et al., 1995), which are impaired in anti-bacterial gene induction (mainly diptericin and drosocin, but also cecropin, attacin and defensin), show normal survival to A. fumigatus infection, but survive poorly when exposed to E. coli. These results provided the simple model that both the Toll and the imd pathways control the induction of anti-bacterial genes, whereas the induction of anti-fungal genes is governed mainly by the Toll pathway (Lemaitre et al., 1996).

It was important to arrive at this scenario of anti-microbial gene induction, as it provided a working model for further genetic dissection. With the availability of more regulatory mutants (e.g. 18-wheeler, Dif, Relish, necrotic) and additional anti-microbial targets (metchnikowin; Levashina et al., 1998), it is now possible to extend and refine this picture (Figure 7). Indeed, Eldon and colleagues (Williams et al., 1997) found that the induction of attacin is almost completely abolished in 18-wheeler mutants, whereas induction of cecropin and diptericin is affected to a lesser degree. In addition, mutant 18-wheeler larvae are more susceptible to bacterial infections, than are their wild-type counterparts (Williams et al., 1997).

The analysis of anti-microbial gene induction in dorsal mutants showed that it is dispensable in the normal regulation of all anti-microbial genes tested, even though Dorsal is expressed in the fat body cells and its nuclear localization is controlled by immune challenge (Lemaitre et al., 1995, 1996). This result questioned the validity of observations made from in vitro DNA-binding assays, as well as from transfection experiments, wherein all three Rel proteins were found to trans-activate reporter genes in a sequence-specific manner, under the control of kappaB-related motifs, (Dushay et al., 1996; Ip et al., 1993; Reichhart et al., 1993; reviewed in Engström, 1999). To assess the relative contributions of Dif and Relish in anti-microbial gene activation, dif and relish mutants have been recently obtained. Analyses of dif and dif/dorsal double mutant adults (Ferrandon et al., 1999; Hedengren et al., 1999; Meng et al., 1999) and dif/dorsal clones made in the larval fat body (Manfruelli et al., 1999) reveals that the primary function of Dif is to regulate anti-fungal gene expression mediated by the Toll-Cactus pathway. In larvae, Dif and Dorsal exhibit functional redundancy, and either of these proteins can activate drosomycin in fat body cells that lack both of these proteins (Manfruelli et al., 1999). This functional redundancy explains why dorsal mutants exhibit normal anti-microbial gene induction (Lemaitre et al., 1995). Disruption of relish leads to the elimination of diptericin and cecropin induction as well as drastic reduction of attacin and drosomycin induction. Consistent with these observations, relish mutant flies show increased susceptibility to both bacterial and fungal infections (Hedengren et al., 1999).

As the specific contributions of various Rel proteins in the intracellular aspect in anti-microbial gene induction are becoming clearer, so are events in the extracellular portion of the pathway. While the proteases that regulate processing of Spätzle have not yet been identified, a role for serine protease inhibitors called Serpins has recently been discovered in necrotic (nec) mutants of Drosophila. Mutant nec adults have melanized spots on body and leg joints. The nec gene encodes three putative Serpins (Levashina et al., 1999). Analysis of anti-microbial peptide gene induction in nec animals revealed that anti-bacterial gene expression remains unimpaired, whereas drosomycin expression is constitutively activated. This constitutive drosomycin activation is suppressed by removal of spätzle or Toll function, placing Serpin function upstream of Spätzle activation (Levashina et al., 1999; Figure 7).

Immune challenge not only activates the transcription of anti-microbial genes, but also of genes in the spätzle-to-dorsal (except tube) cascade. In this respect, transcriptional upregulation of cactus is especially noteworthy, as it follows acute phase induction kinetics, much like that of drosomycin and represents an autoregulatory step in the induction process (Nicolas et al., 1998). Thus, cactus upregulation is dependent on the Spätzleright arrowTollright arrowCactus pathway; cactus transcripts are induced in genetic backgrounds where the pathway is constitutively active (TollD/+ or cactus/cactus), whereas its immune induction is impaired in spätzle, Toll, tube or pelle mutants. In addition to the transcriptional activation of the cactus locus, there is also post-translational regulation of Cactus degradation that is controlled by the Toll signal. Immune activation of the Tollright arrowCactus pathway in the fat body leads to the rapid degradation of pre-existing Cactus isoforms and the synthesis of new proteins. This autoregulation of cactus induction and stability parallels regulation of mammalian IkappaBalpha and serves to illustrate another conserved aspect of mammalian and insect Rel signaling pathways (Nicolas et al., 1998).

In response to infection, both Dorsal and Dif translocate from the cytoplasm to the nucleus of the larval fat body cells (Ip et al., 1993; Lemaitre et al., 1995). Although the Toll/Cactus pathway is required for Dorsal relocalization, Toll and tube are not required for the relocalization for Dif (Wu and Anderson, 1998), even though Dif is constitutively nuclear in TollD/+ as well as in the cactus-deficient fat body (Ip et al., 1993; Wu and Anderson, 1998). To analyse the roles of Dorsal-Cactus and Dif-Cactus complexes in Toll signaling in the fat body and to identify new factors that contribute to anti-microbial gene induction, Wu and Anderson (1998) performed a genetic screen. Using a diptericin-lacZ promoter transgenic fly strain (Reichhart et al., 1992), they recovered more than 40 different mutations on the third chromosome, which are consistently compromised in their ability to induce the diptericin promoter, at both the transgenic and the endogenous loci. Using the nuclear translocation of Dif in the mutant fat body as a second assay, they were able to place these mutants into two groups. One group (class I), including three loci (ird 4, ird 6 and ird 8; ird for immune response deficient), affects the nuclear localization of Dif before and after infection. The second group (class II) includes three other loci (ird 5, ird 9 and ird 10) and in this group Dif is nuclear both before and after infection. It is likely that the class I mutants block diptericin induction by mechanisms other than directly blocking the nuclear import of Dif, as Dif cannot bind to or activate the diptericin promoter (Gross et al., 1996; Ip et al., 1993; Peterson et al., 1995) and diptericin activation is normal in cells devoid of Dif (Manfruelli et al., 1999; Meng et al., 1999). Nevertheless, the value of this screen lies in the isolation of many new genes involved in diptericin activation. Given that there is some overlap in diptericin regulation with that of other anti-bacterial genes (drosocin, cecropin etc., Lemaitre et al., 1996), some of these ird genes are likely to contribute to the co-regulation of anti-bacterial peptide genes, possibly via Rel/Cactus pathways.

As we survey the evidence for Rel factor functions in the Drosophila fat body, it is clear that Rel pathways use Toll and 18-Wheeler receptors to sense extracellular signals generated after pathogen attack, and that Rel proteins Dorsal, Dif and Relish control the rapid upregulation of anti-microbial peptide genes. Thus, the parallels between the Drosophila and mammalian systems are striking, specially in light of reports that human Toll-like receptors function upstream of NF-kappaB activation and cytokine gene expression (Medzhitov et al., 1997; Yang et al., 1998). Protein motifs similar to the ones found in these conserved signaling pathways (leucine-rich repeats, Toll homology domains, serine-threonine kinases), are also found in proteins involved in plant disease resistance (Rock et al., 1998; Wilson et al., 1997; Yang et al., 1997). Effector anti-microbial peptides, employed by the hosts to kill pathogens are also similar (Hoffmann et al., 1999). The conservation of these ancestral modules in innate immunity in plants and animals suggests a common and ancient origin, and raises interesting questions about the co-evolution of hosts and their pathogens.

Rel factors in hematopoiesis

Genetic analysis of various alleles of the d/v component mutants suggested that at least some of these genes, such as Toll, tube, pelle and cactus, have important zygotic functions during post-embryonic stages of the fly life cycle. This indication came from observations that null mutations in Toll, tube and pelle compromise the viability of a majority of homozygous mutants (Gerttula et al., 1988; Hecht and Anderson, 1993), whereas null mutations in cactus lead to lethality of all animals prior to their reaching adulthood (Roth et al., 1991). Hyperactivation of the Rel pathway resulting from loss of Cactus function, from constitutive Toll or the Toll-like 18-Wheeler receptors, or from over-expression of Dorsal (or even mouse NF-kappaB p50) leads to a high incidence of melanotic tumors (Eldon et al., 1999; Gerttula et al., 1988; Govind, 1996; Roth et al., 1991). In such mutants, activated hemocytes called lamellocytes form aggregates around self-tissue. These aggregates are melanized to form melanotic capsules. Similar capsules are also formed around large invading parasites, and this cellular immune encapsulation of the parasite serves to limit its growth and development (Rizki and Rizki, 1984).

An explanation for both the lethality and melanotic capsules of Toll and cactus mutants was recently provided by Qiu et al. (1998). Using strong alleles of cactus, they first showed a direct correlation between the penetrance of melanotic capsules and the strength of the cactus lethal phenotype. Using the UAS-cactus/GAL4 system, they then rescued these phenotypes simultaneously by selective expression of wild-type Cactus protein in the larval lymph gland. However, cactus phenotypes were not rescued when Cactus protein was provided in other organs including the fat body, where Cactus and other pathway components control the anti-microbial gene expression.

Lethality and melanotic tumors of cactA2, a strong cactus hypomorph mutant, were also rescued by mutations in Toll, tube and pelle, but not by mutations in gastrulation defective, snake, easter, spätzle or even dorsal. Consistent with these observations, Qiu et al. (1998) found that hemolymph of TollD/+ and cactus/cactus larvae has an overabundance of hemocytes (with an increase in the proportion of blood cells in division), whereas hemolymph derived from loss-of-function Toll, tube and pelle mutant shows a deficit in hemocyte number. These observations support a model in which Toll, Tube, Pelle and Cactus play a role in controlling the steady-state hemocyte density (Qiu et al., 1998).

Genetic studies in mice have also shown that NF-kappaB/IkappaB proteins play a central role in mammalian hematopoiesis. The cactus phenotype of flies is strikingly similar to the phenotype observed in mice deficient in IkappaBalpha. IkappaBalpha knockout mice have a significantly elevated number of granulocyte (neutrophil) precursors and show neonatal lethality (Beg et al., 1995). Moreover, mice lacking RelB show an increase in the number of erythroid precursors in the spleen, but a lower number in the bone marrow (Weih et al., 1995). The hematopoietic defects observed in cactus mutants are also reminiscent of lymphoid malignancies associated with mutations in vertebrate Rel proteins (Rayet and Gélinas, 1999, this issue). These parallels suggest that the mammalian and Drosophila NF-kappaB/IkappaB pathways control hematopoiesis and immune functions by a shared mechanism.

Rel factors in muscle development

The abdominal hemisegments of the Drosophila larva have 30 somatic muscles fibers, which develop in a highly stereotypic pattern from 30 muscle founder cells in the embryo. By extending growth cone-like filipodia, muscle fibers search and contact the specialized epidermal muscle attachment cells (Keshishian et al., 1996). The requirement for Toll signaling for proper muscle development was first noticed as a dominant disruption of muscle patterning in Toll-deficient embryos: loss of one or both copies of Toll leads to extensive errors in muscle pattern, including deletion and duplication of muscle fibers as well as errors in muscle insertion in all 30 fibers.

The role of Toll in muscle development is non-autonomous. In a reverse mosaic experiment, wherein wild-type Toll protein was selectively expressed in a subset of tissues that includes the epidermis and the epidermal muscle attachment cells, the muscle development defects were rescued. This rescue was not achieved by Toll expression in mesodermally-derived tissues, even though Toll protein is normally expressed in some of the 30 somatic muscles of wild-type stage 12 - 17 embryos (Halfon et al., 1995; Halfon and Keshishian, 1998). Some, but not all, other genes in the d/v pathway are also involved in muscle development. The requirement for other d/v pathway components in muscle development was tested by examining the muscle phenotypes of mutant easter, tube and pelle embryos. While easter mutants showed normal muscle development, spätzle, tube and pelle mutants showed a phenotype identical to that of Toll embryos (Halfon and Keshishian, 1998). These results suggest that the Spätzleright arrowTollright arrowTuberight arrowPelle proteins are required epidermally to bring about normal muscle development in Drosophila.

The possible involvement of Dorsal and Cactus in muscle development and function also comes from the analysis of muscle phenotypes in dorsal larvae (Cantera et al., 1999a). Loss of dorsal function leads to larval muscle misinsertions as well as duplications; the most striking phenotype associated with muscle five is similar to that described for spätzle, tube and pelle (Halfon et al., 1995; Halfon and Keshishian, 1998). Given that muscle development is essentially complete during embryogenesis, it is likely that the phenotype described for dorsal larvae originates in the embryo. Taken together, these studies suggest that, at least in muscle five, the Tollright arrowTuberight arrowPelle signal targets the Dorsal/Cactus complex. However, there are a number of issues about this model that remain unresolved. For example, it is not clear whether Dorsal (and Cactus) is expressed in the embryonic epidermis or if this expression is sufficient to rescue the observed muscle five defect in dorsal mutants. Analysis of muscle phenotypes in cactus and dif mutants in the embryo and the larva (as well as double mutants of cactus and Toll, tube, pelle, dorsal or dif) should further define the contributions of these genes to the muscle pathway. In addition, temperature shift studies with appropriate temperature-sensitive alleles should also help define the timing of the events in question.

Rel factors in other functions

Drosophila Rel factors govern additional biological processes. For example, mutant Toll embryos and dorsal larvae show changes in the number of motorneurons in the central nervous system and abnormal neuromuscular junctions (Cantera et al., 1999b; Halfon et al., 1995; Rose et al., 1997), suggesting that Toll and Dorsal regulate synaptogenesis. A role for Dif and Cactus in the larval central nervous system is suggested by their intriguing co-localization pattern (Cantera et al., 1999b). Dif and Cactus are both present at low levels in most cells of the brain and the nerve cord, but are present at high levels in mushroom bodies, the major associative brain center, involved in learning and memory in flies. Moreover, both proteins are also co-localized in a small subset of the cerebral neuroendocrine system, thought to control the release of neuropeptides like ecdysone and juvenile hormone into circulation. These results raise interesting questions regarding the role of Rel pathways in central nervous system function, and analysis of mutants will further our understanding of their roles in these processes. Finally, Toll, tube or pelle mutants form abnormally short pupae, whereas those that lack cactus make abnormally long and thin pupae (Letsou et al., 1991). These observations suggest that the Tollright arrowCactus pathway contributes to the development of the overall body shape, even though it is not understood how pupal shape is controlled or where in the body these genes function.


In the 20 years since the genetic identification of the dorsal gene (Nüsslein-Volhard, 1979), we have come a long way in understanding its role in pattern formation. Now that all the `dorsal group' genes and cactus have been cloned, a more complete picture of the events before gradient formation has emerged. Analysis of the three somatically-active `dorsal group' genes provides the view that macromolecular complexes are assembled in close proximity to Toll on the plasma membrane of the embryo. The importance of carbohydrate modification in d/v pattern formation, while unexpected, fits well with a role for proteoglycans and extracellular matrix in localizing molecular information. Studies on the intracellular d/v pathway have also revealed that signaling occurs in multi-protein complexes, and it relies on specific protein-protein interactions. With many of the Dorsal target promoters characterized, there is also a clearer idea of how the Dorsal gradient controls the expression of the three classes of zygotic gene, which leads to the subdivision of the d/v axis.

A second area of progress has been the demonstration that Rel factors control other important biological functions in Drosophila. The identification of Dif and Relish, and analysis of their regulation and function in humoral immunity has been especially exciting. The identification of human Toll-like receptors has extended the functional homology of the Toll/IkappaB/NF-kappaB pathway in Drosophila and human immunity. Given that protein modules found in the immunity pathways have been conserved during a billion years of evolution, it is hardly surprising that these proteins are exploited for controlling diverse and unrelated functions. The identification of additional Toll-related proteins in Drosophila and their functional analysis will be important in learning more about the circuitry and cross-talk between various Toll and Rel proteins. The completion of the sequencing of the Drosophila genome will reveal whether additional Rel/IkappaB pathway proteins are present. As reports of Rel family proteins from other invertebrates become available (Gambif1, from the human malaria vector Anopheles gambiae, Barillas-Mury et al., 1996; SpNF-kappaB from sea urchins, Pancer et al., 1999), it will be possible to obtain a phylogenetic perspective on Rel-factor structure and function. With cross-fertilization of ideas between invertebrate and vertebrate systems, findings over the coming years should be just as exciting and fruitful as they have in the last few years.


I thank Drs D Hultmark, E Drier and R Steward for communicating results prior to publication and to my colleagues for their insightful comments on sections of this manuscript. Research in my laboratory is supported by funds from the American Heart Association, American Cancer Society and PSC-CUNY.


Akimura H, Hou DX and Ishii S. (1997). Nature Genet. 17, 211-214. MEDLINE

Anderson KV. (1998). Cell 95, 439-442. MEDLINE

Anderson KV and Nüsslein-Volhard C. (1986)). In: Gall JG (ed). Gametogenesis and the Early Embryo. Liss: New York, pp. 177-194.

Barillas-Mury C, Charlesworth A, Gross I, Richman A, Hoffmann JA and Kafatos FC. (1996). EMBO J. 15, 4691-4701. MEDLINE

Beg AA, Sha WC, Bronson RT and Baltimore D. (1995). Genes Dev. 9, 2736-2746. MEDLINE

Belvin MP and Anderson KV. (1996). Annu. Rev. Cell. Dev. Biol. 12, 393-416. MEDLINE

Belvin MP, Jin Y and Anderson KV. (1995). Genes Dev. 9, 783-793. MEDLINE

Bergmann A, Stein D, Geisler R, Hagenmaier S, Schmid B, Fernandez N, Schnell B and Nüsslein-Volhard C. (1996). Mech. Dev. 60, 109-123. Article MEDLINE

Boulay JL, Dennefeld C and Alberga A. (1987). Nature 330, 395-398. MEDLINE

Bulet P, Hetru C, Dimarcq J-L and Hoffmann D. (1999). Dev. Comp. Imm. 23, 329-344.

Cantera R, Kozlova T, Barillas-Mury C and Kafatos FC. (1999a). Mol. Cell. Neurosci. 13, 131-141.

Cantera R, Roos E and Engström Y. (1999b). J. Neurobiol. 38, 16-26.

Cao Z, Henzel WJ and Gao X. (1996). Science 271, 1128-1131. MEDLINE

Chasan R and Anderson KV. (1989). Cell 56, 391-400. MEDLINE

Chasan R, Jin Y and Anderson KV. (1992). Development 115, 607-616. MEDLINE

Chen FE and Ghosh G. (1999). Oncogene 18, 6845-6852. MEDLINE

DeLotto R and Speirer P. (1986). Nature 323, 688-692. MEDLINE

Drier E, Huang LH and Steward R. (1999). Genes Dev. 13, 556-568. MEDLINE

Drier E and Steward R. (1997). Sem. Cancer Biol. 8, 83-92.

Dubnikoff T, Valentine SA, Chen G, Shi T, Lengyel JA, Paroush Z and Courey AJ. (1997). Genes Dev. 11, 2952-2957. MEDLINE

Dushay MS, Äsling B and Hultmark D. (1996). Proc. Natl. Acad. Sci. USA 93, 10343-10347. Article MEDLINE

Edwards DN, Towb P and Wasserman SA. (1997). Development 124, 3855-3864. MEDLINE

Ekengren S and Hultmark D. (1999). Insect Biochem. Molec. Biol. (in press).

Eldon E, Reichhart J-M and Williams M. (1999). A. Conf. Dros. Res. 40, 393B.

Engström Y. (1999). Dev. Comp. Imm. 23, 345-358.

Engström Y, Kadalayil L, Sun SC, Samakovlis C, Hultmark D and Faye I. (1993). J. Mol. Biol. 232, 327-333. Article MEDLINE

Ferrandon D, Jung A, Rutschmann S, Reichhart J-M and Hoffmann JA. (1999). A Conf. Dros. Res. 40, 422A.

Fisher AL and Caudy M. (1998). Genes Dev. 12, 1931-1940. MEDLINE

Francois V, Solloway M, O'Neill JW, Emery J and Bier E. (1994). Genes Dev. 8, 2602-2616. MEDLINE

Galindo RL, Edwards DN, Gillespie SKH and Wasserman SA. (1995). Development 121, 2209-2218. MEDLINE

Geisler R, Bergmann A, Hiromi Y and Nüsslein-Volhard C. (1992). Cell 71, 613-621. MEDLINE

Gerondakis S, Grossman M, Nakamura Y, Pohl T and Grumont R. (1999). Oncogene 18, 6888-6895. MEDLINE

Gerttula S, Jin Y and Anderson KV. (1988). Genetics 119, 123-133. MEDLINE

Ghosh G, Van Duyne G, Ghosh S and Sigler P. (1995). Nature 373, 303-310. MEDLINE

Gillespie SKH and Wasserman SA. (1994). Mol. Cell. Biol. 14, 3559-3568. MEDLINE

Gilmore TD. (1999). Oncogene 18, 6925-6937. MEDLINE

Gonzales-Crespo S and Levine M. (1993). Genes Dev. 7, 1703-1713. MEDLINE

Govind S. (1996). Biochem. Soc. Trans. 24, 39-44. MEDLINE

Govind S, Brennan L and Steward R. (1993). Development 117, 135-148. MEDLINE

Govind S, Drier EA, Huang LH and Steward S. (1996). Mol. Cell. Biol. 16, 1103-1114. MEDLINE

Govind S, Whalen AM and Steward R. (1992). Proc. Natl. Acad. Sci. USA 89, 7861-7865. MEDLINE

Grobetahans J, Bergmann A, Haffter P and Nüsslein-Volhard C. (1994). Nature 372, 563-566. MEDLINE

Gross I, Georgel P, Kappler C, Reichhart J-M and Hoffmann JA. (1996). Nucleic Acids Res. 89, 7861-7865.

Gross I, Georgel P, Oertel-Buchheit P, Scnarr M and Reichhart J-M. (1999). Gene 228, 233-242. MEDLINE

Halfon M, Hashimoto C and Keshishian H. (1995). Dev. Biol. 196, 151-167.

Halfon M and Keshishian H. (1998). Dev. Biol. 199, 164-174. MEDLINE

Hashimoto C, Gerttula S and Anderson KV. (1991). Development 111, 1021-1028. MEDLINE

Hashimoto C, Hudson KL and Anderson KV. (1988). Cell 52, 269-279. MEDLINE

Hecht PM and Anderson KV. (1993). Genetics 135, 405-417. MEDLINE

Hedengren M, Asling B, Dushay MS, Ando I, Ekengren S, Wihlborg M and Hultmark D. (1999). Molecular Cell In Press.

Hoffmann JA, Kafatos FC, Janeway Jr CA and Ezekowitz RAB. (1999). Science 284, 1313-1318. Article MEDLINE

Hoffmann JA and Reichhart J-M. (1997). Trends Cell Biol. 7, 309-316.

Hong CC and Hashimoto C. (1995). Cell 82, 785-794. MEDLINE

Huang J-D, Dubnicoff T, Liaw G-J, Bai Y, Valentine S, Shirokawa JM, Lengyel J and Courey AJ. (1995). Genes Dev. 9, 3177-3189. MEDLINE

Huang J-D, Schwyter DH, Shirokawa JM and Courey AJ. (1993). Genes Dev. 7, 694-704. MEDLINE

Huguet C, Crepieux P and Laudet V. (1997). Oncogene 15, 2965-2974. MEDLINE

Hultmark D. (1993). Trends Genet. 9, 178-183. MEDLINE

Huxford T, Huang D-B, Malek S and Ghosh G. (1998). Cell 95, 759-770. MEDLINE

Ip YT, Kraut M, Levine M and Rushlow CA. (1991). Cell 64, 439-446. MEDLINE

Ip YT, Park RE, Kosman D, Bier E and Levine M. (1992a). Genes Dev. 6, 1728-1739. MEDLINE

Ip YT, Park RE, Kosman D, Yazdanbakhsh K and Levine M. (1992b). Genes Dev. 6, 1518-1530.

Ip YT, Reach M, Engström Y, Kadalayil L, Cai H, Gonzalez-Crespo S, Tatei K and Levine M. (1993). Cell 75, 753-763. MEDLINE

Isoda K and Nüsslein-Volhard C. (1994). Proc. Natl. Acad. Sci. USA 91, 5350-5354. MEDLINE

Isoda K, Roth S and Nüsslein-Volhard C. (1992). Genes Dev. 6, 619-630. MEDLINE

Jacobs MD and Harrison SC. (1998). Cell 95, 740-758.

Jiang J, Kosman D, Ip YT and Levine M. (1991). Genes Dev. 5, 1881-1891. MEDLINE

Jiang J and Levine M. (1993). Cell 72, 741-752. MEDLINE

Jiang J, Rushlow CA, Zhou Q, Small S and Levine M. (1992). EMBO J. 11, 3147-3154. MEDLINE

Kamens J and Brent R. (1991). New Biol. 10, 1005-1013.

Kappler C, Meister M, Laguex M, Gateff E, Hoffmann JA and Reichhart J-M. (1993). EMBO J. 12, 1561-1568. MEDLINE

Karin M. (1999). Oncogene 18, 6867-6874. MEDLINE

Keshishian H, Broadie K, Chiba A and Bate M. (1996). Annu. Rev. Neurosci. 19, 545-575. MEDLINE

Kidd S. (1992). Cell 71, 623-635. MEDLINE

Kirov N, Childs S, O'Connor M and Rushlow C. (1994). Molec. Cell Biol. 14, 713-722. MEDLINE

Konrad KD, Goralski TJ, Mahowald AP and Marsh JL. (1998). Proc. Natl. Acad. Sci. USA 95, 6819-6824. Article MEDLINE

Konsolaki M and Schüpbach T. (1998). Genes Dev. 12, 120-131. MEDLINE

Kosman D, Ip YT, Levine M and Arora K. (1991). Science 254, 118-122. MEDLINE

Lehming N, McGuire S, Brickman JM and Ptashne M. (1995). Proc. Natl. Acad. Sci. USA 92, 10242-10246. MEDLINE

Lehming N, Thanos D, Brickman JM, Ma J, Maniatis T and Ptashne M. (1994). Nature 371, 175-179. MEDLINE

Lemaitre B, Meister M, Govind S, Georgel P, Steward R, Reichhart J-M and Hoffmann JA. (1995). EMBO J. 14, 536-545. MEDLINE

Lemaitre B, Nicholas E, Michaut L, Reichhart J-M and Hoffmann JA. (1996). Cell 86, 973-983. MEDLINE

Lemaitre B, Reichhart JM and Hoffmann JA. (1997). Proc. Natl. Acad. Sci. USA 94, 14614-14619. Article MEDLINE

LeMosy EK, Hashimoto C and Hong CC. (1999). Trends Cell Biol. 9, 102-107. MEDLINE

LeMosy EK, Kemler D and Hashimoto C. (1998). Development 125, 4045-4053. MEDLINE

Leptin M. (1991). Genes Dev. 5, 1568-1576. MEDLINE

Letsou A, Alexander S, Orth K and Wasserman SA. (1991). Proc. Natl. Acad. Sci. USA 88, 810-814. MEDLINE

Levashina EA, Ohresser S, Lemaitre B and Imler J-L. (1998). J. Mol. Biol. 278, 515-527. Article MEDLINE

Levashina EA, Langley E, Green C, Gubb C, Ashburner M, Hoffmann JA and Reichhart JM. (1999). Science 285, 1917-1919. Article MEDLINE

Liu Z-P, Galindo RL and Wasserman SA. (1997). Genes Dev. 11, 3413-3422. MEDLINE

Manfruelli P, Reichhart J-M, Steward R, Hoffmann JA and Lemaitre B. (1999). EMBO J. 18, 3380-3391. Article MEDLINE

Mason ED, Konrad KD, Webb CD, Marsh JL. (1994). Genes Dev. 8, 1489-1501. MEDLINE

Medzhitov R, Preston-Hurlburt P and Janeway CA. (1997). Nature 388, 394-397. Article MEDLINE

Meister M, Braun A, Kappler C, Reichhart JM and Hoffmann JA. (1994). EMBO J. 13, 5958-5966. MEDLINE

Meng X, Khanuja BS and Ip YT. (1999). Genes Dev. 13, 792-797. MEDLINE

Morisato D and Anderson KV. (1994). Cell 76, 677-688. MEDLINE

Muller CW, Rey FA, Sodeoka M, Verdine GL and Harrison SC. (1995). Nature 373, 311-317. MEDLINE

Nicolas E, Reichhart J-M, Hoffmann JA and Lemaitre B. (1998). J. Biol. Chem. 273, 10463-10469. Article MEDLINE

Nilson L and Schüpbach T. (1998). Cell 93, 253-262. MEDLINE

Nüsslein-Volhard C. (1979)). In: Determinants of Spatial Organization, Subtelny S and Koenigsberg IR (eds). Academic Press, New York., pp. 185-211.

O'Neil LAJ and Kaltschmidt C. (1997). Trends Neurosci. 20, 252-258. Article MEDLINE

Pahl H. (1999). Oncogene 18, 6853-6866. MEDLINE

Pan DJ and Courey AJ. (1992). EMBO J. 11, 1837-1842. MEDLINE

Pan DJ, Huang JD and Courey AJ. (1991). Genes Dev. 5, 1892-1901. MEDLINE

Pancer Z, Rast JP and Davidson EH. (1999). Immunogenetics 49, 773-786. MEDLINE

Peterson U-M, Bjorklund G, Ip YT and Engström Y. (1995). EMBO J. 14, 3146-3158. MEDLINE

Qiu P, Pan PC and Govind S. (1998). Development 125, 1909-1920. MEDLINE

Ray R and Schüpbach T. (1996). Genes Dev. 10, 1711-1723. MEDLINE

Rayet B and Gélinas C. (1999). Oncogene. 18, 6938-6947. MEDLINE

Reach M, Galindo RL, Towb P, Allen JL, Karin M and Wasserman SA. (1996). Dev. Biol. 180, 353-364. Article MEDLINE

Reichhart J-M, Georgel P, Meister M, Lemaitre B, Kappler C and Hoffmann JA. (1993). CR Acad Sci. 316, 1218-1224.

Reichhart J-M, Meister M, Dimarcq JL, Hoffmann D, Ruiz C, Richards G and Hoffmann JA. (1992). EMBO J. 11, 1469-1477. MEDLINE

Rizki TM and Rizki RM. (1984)). In: King RC (ed). Insect Ultrastructure. vol. 2 Plenum, New York., pp. 579-604.

Rock FL, Hardiman G, Timans JC, Kastelein RA and Bazan JF. (1998). Proc. Natl. Acad. Sci. USA 95, 588-593. Article MEDLINE

Rose D, Zhu X, Kose H, Hoang B, Cho J and Chiba A. (1997). Development 124, 1561-1571. MEDLINE

Roth S. (1998). Curr. Biol. 8, R906-R910. MEDLINE

Roth S, Hiromi Y, Godt D and Nüsslein-Volhard C. (1991). Development 112, 371-388. MEDLINE

Roth S, Stein D and Nüsslein-Volhard C. (1989). Cell 59, 1189-1202. MEDLINE

Rusch J and Levine M. (1996). Curr. Opin. Genet. Devel. 6, 416-423.

Rushlow CA, Han K, Manley JL and Levine M. (1989). Cell 59, 1165-1177. MEDLINE

Schneider DS, Hudson KL, Lin T-Y and Anderson KV. (1991). Genes Dev. 5, 797-807. Article MEDLINE

Schneider DS, Jin Y, Morisato D and Anderson KV. (1994). Development. 120, 1243-1250. MEDLINE

Sen J, Goltz JS, Stevens L and Stein D. (1998). Cell 95, 471-481. MEDLINE

Shelton CA and Wasserman SA. (1993). Cell 72, 515-525. MEDLINE

Shirokawa JM and Courey AJ. (1997). Mol. Cell. Biol. 17, 3345-3355. MEDLINE

Smith C and DeLotto R. (1994). Nature 368, 548-551. MEDLINE

Spencer E, Jiang J and Chen ZJ. (1999). Genes Dev. 13, 284-294. MEDLINE

Stein D and Nüsslein-Volhard C. (1992). Cell 68, 429-440. MEDLINE

Stein D, Roth S, Vogelsang E and Nüsslein-Volhard C. (1991). Cell 65, 725-735. MEDLINE

Steward R. (1987). Science 238, 692-694. MEDLINE

Steward R. (1989). Cell 59, 1179-1188. MEDLINE

Steward R and Govind S. (1993). Curr. Opin. Genet. Dev. 3, 556-561. MEDLINE

Steward R, Zusman SB, Huang LH and Schedl P. (1988). Cell 55, 487-495. MEDLINE

Sun S-C, Lindstrom I, Lee J-Y and Faye I. (1991). Eur. J. Biochem. 196, 247-254. MEDLINE

Tatei K and Levine M. (1995). Mol. Cell. Biol. 15, 3627-3634. MEDLINE

Thisse C, Perrin-Schmitt F, Stoetzel C and Thisse B. (1991). Cell 65, 1191-1201. MEDLINE

Thisse B, Stoetzel C, El Messal M and Perrin-Schmitt F. (1987). Genes Dev. 5, 1285-1298.

Towb P, Galindo RL and Wasserman SA. (1998). Development 125, 2443-2450. MEDLINE

Tryselius Y, Samakovlis C, Kimbrell DA and Hultmark D. (1992). Eur. J. Biochem. 204, 395-399. MEDLINE

Weih F, Carrasco D, Durham SK, Barton DS, Rizzo CA, Ryseck RP, Lira SR and Bravo R. (1995). Cell 80, 331-340. MEDLINE

Whalen AM and Steward R. (1993). J. Cell Biol. 123, 523-534. MEDLINE

Williams MJ, Rodriguez A, Kimbrell DA and Eldon ED. (1997). EMBO J. 16, 6120-6130. Article MEDLINE

Wilson I, Vogel J and Sommerville S. (1997). Curr. Biol. 7, R175-R178. MEDLINE

Wu LP and Anderson KV. (1998). Nature 392, 93-97. Article MEDLINE

Yang J and Steward R. (1997). Proc. Natl. Acad. Sci. USA 94, 14524-14529. MEDLINE

Yang R-B, Mark M, Gray A, Huang A, Xie MH, Zhang M, Goddard A, Wood WI, Gurnet AL and Godowski PJ. (1998). Nature 395, 284-288. Article MEDLINE

Yang Y, Shah J and Klessig DF. (1997). Genes Dev. 11, 1621-1639. MEDLINE

Zhou J, Zwicker J, Szymanski P, Levine M and Tjian R. (1998). Proc. Natl. Acad. Sci. USA 95, 13483-13488. MEDLINE


Figure 1 The Drosophila Rel factors. Rel proteins Dorsal, Dif (Dorsal-related Immunity Factor) and Relish and IkappaB Cactus. Dorsal B is derived by alternative splicing of the dorsal locus. Dorsal B is expressed in the larval fat body, but lacks a nuclear localization signal (Gross et al., 1999). A 300-amino acid Rel homology domain (RHD) is shown. This region is responsible for DNA binding, and protein-protein interactions. The C-terminal regions of Dorsal, Dorsal B and Dif contain transactivation domains and are different from one another. cactus encodes two, almost identical isoforms, differing only in the C termini. Relish and Cactus have several ankyrin repeats and a PEST sequence

Figure 2 Intracellular Drosophila Rel-IkappaB pathways. The ligand, Rel factor and effector genes vary among these pathways. Toll, Tube, Pelle and Cactus have been found to function in all processes identified so far. Toll has leucine-rich repeats in the extracellular domain and a conserved intracellular Toll-homology domain, just below the membrane. (a) Rel factors in embryonic d/v axis formation. Ventral activation of Toll by activated Spätzle triggers nuclear localization of Dorsal in ventral and lateral nuclei. The signal is mediated by Tube, Pelle and Cactus. (b) Rel factors in humoral immune induction. Fungal infection activates Spätzle. Signal transduction is through Toll, Tube, Pelle and the Dif-Cactus or Dorsal-Cactus complex. (c) Rel factors in hematopoiesis. Toll activation in hemocytes of the larval lymph gland is by an unknown ligand. Tube, Pelle and Cactus then regulate the function of an unidentified Rel protein. Target genes control hemocyte proliferation. Their identity is not known. (d) A simplified scheme of NF-kappaB activation in humans to show parallels with the Drosophila pathways. Activation of the human Toll-like receptor 4 (TLR4) initiates a cascade of cytoplasmic events. Although a Tube homolog has not been found in mammals, MyD88 may function as an adaptor between the TLR4 and the Pelle homolog IRAK. TLR4 activates NF-kappaB and triggers the production of inflammatory cytokines. Proteins for which Drosophila counterparts have not been identified are not shown here. For more details see other articles in this issue

Figure 3 Twelve `maternal-effect' genes involved in the specification of the dorsal-ventral axis. Genes are grouped based on the site of action of their products. The order in which windbeutel, pipe and nudel act relative to each other is not known. The sequence of action of the remaining nine genes was deduced from epistasis experiments and is as indicated. See text for details

Figure 4 Extracellular and intracellular d/v pathway in the ventral region of a precellular embryo. All components shown here are maternally encoded. Active Nudel protease is localized to the outside of the embryo (grey). Its interaction with proteins downstream in the pathway (e.g. Gastrulation Defective) is not known. A serine protease cascade of Gastrulation Defective, Snake and Easter results in the processing of the Spätzle precursor (pre-Spätzle) to active Spätzle. Pre-Spätzle is diffusible in the perivitelline fluid, that fills the space between the plasma and the vitelline membranes. This processing is believed to be confined ventrally. A gradient of active Spätzle is thought to cause a graded activation of Toll. Active Toll relays the signal to a multi-protein complex containing Tube (death domain, DD and tube repeats, TR) and Pelle (death domain and catalytic kinase domain, CD). Actin-binding protein Filamin, which interacts with both Toll and Tube, is proposed to facilitate the localization of the Tube-Pelle complex to the membrane. Pelle activation leads to the phosphorylation and degradation of Cactus, although Pelle probably does not directly phosphorylate Cactus. In addition, Dorsal phosphorylation by the Tube-Pelle signal augments its nuclear import to ventral nuclei

Figure 5 Schematic of the primary structures of Cactus and Dorsal. Serine residues 74, 78, 82, 83 in the N-terminal acidic domain of Cactus are important for signal-dependent phosphorylation (Bergmann et al., 1996; Reach et al., 1996). Serines at positions 463, 467 and 468 within the PEST domain are essential for signal-independent casein-kinase II phosphorylation (Liu et al., 1997). Based on the crystal structure of p50, the Dorsal RH domain (amino acids 46 - 340) consists of two subdomains (RHD1 and RHD2), and contains all the information for its selective nuclear import. The Dorsal NLS (aa 335-340) is not required for Cactus interaction. Serine residues (positions 70, 79, 103, 213, 312, 317) that are phosphorylated in Dorsal (Drier et al., 1999), and regions that are involved in Dorsal-Twist interaction and twist activation (Shirokawa and Courey, 1997), Dorsal-dCBP interaction (Akimaru et al., 1997), Dorsal dimerization (Govind et al., 1996), Dorsal-Cactus interaction (Govind et al., 1996; Lehming et al., 1995), and Dorsal binding with Tube and Pelle (Edwards et al., 1997; Yang and Steward, 1997) are shown. Also shown in the C-terminal transactivation domain are stretches of glutamine, alanine and asparagine residues

Figure 6 Schematic cross section of a cellular blastoderm with domains of zygotic gene expression. High levels of nuclear Dorsal in the presumptive mesoderm activate the expression of twist and snail. Lower levels of Dorsal in the neuroectoderm activate short gastrulation (Francois et al., 1994). This region gives rise to the central nervous system and the ventral epidermis. rhomboid is activated only a subset of the neuroectoderm. High and low levels of nuclear Dorsal repress zerknüllt, decapentapleigic tolloid and twisted gastrulation (see text and Mason et al., 1994). These genes are therefore expressed in the dorsal ectoderm in the absence of Dorsal. After gastrulation, the dorsal ectoderm gives rise to the dorsal-most amnioserosa (shown with dots) and the dorsal epidermis (grey). The expression of these zygotic target genes is essential for the subsequent subdivision of the d/v axis. (Modified from Steward and Govind, 1993)

Figure 7 Control of immune genes by Rel pathways. Many genes encoding anti-microbial peptides are activated by Rel pathways. The relative degrees to which these genes are controlled by Rel pathways are indicated. In adults, Dif (Dorsal or Dif in larvae) controls the activation of the anti-fungal gene drosomycin through the Spätzle-Cactus cassette. Activation of metchnikowin and cecropin is also partly under the control of this pathway. The embryonic protease cascade is dispensable for drosomycin activation. Protease inhibitors (Serpins) control the activation of drosomycin through Spätzle (Levashina et al., 1999). Relish is important for the activation of anti-bacterial and anti-fungal genes shown. 18-Wheeler function is critical for the activation of attacin. It is not known if Relish is regulated by the Toll or the 18-Wheeler receptors. Interactions between the Toll, 18-Wheeler and immune deficiency (not shown) pathways are not known

22 November 1999, Volume 18, Number 49, Pages 6875-6887
Table of contents    Previous  Article  Next    [PDF]