BIOLOGICAL OVERVIEW

Relish was identified in a screen for genes involved in the Drosophila immune system by using PCR differential display to identify genes induced in infected flies. Unlike the single domain proteins Dorsal and Cactus (respectively the classic NF-kappaB- and IkappaB-related proteins of Drosophila), Relish contains both a Rel homology domain (hence the name Relish) and an IkappaB-like domain with six ankyrin repeats. Thus Relish is a dual domain protein. In this respect Relish is similar to the compound mammalian NF-kappaB precursors p100 and p105, although no obvious similarity is seen outside the two conserved domains (Dushay, 1996).

Interesting features have been noted in the regions outside the conserved domains. Like RelB, Relish has an unusually long region N-terminal of the Rel homology domain. Furthermore, just downstream of the Rel homology domain there is a serine-rich stretch, corresponding to the position where p100 and p105 have a glycine-rich region that serves as a processing signal for the generation Rel domain containing p50. The serine-rich sequence in Relish may serve a similar function. Another serine-rich region is found in the N-terminal region of Relish. Finally, there are several potential target sites for phosphorylation by casein kinase II, including four in the spacer between the Rel and ankyrin domains, and five either in or near the PEST region. Casein kinase II has been implicated in the constitutive phosphorylation of the PEST region in IkappaB, and the signal-induced phosphorylation of the same protein is mediated by a kinase with similar target sites (Dushay, 1996).

There is experimental evidence for at least three independent signal transduction pathways in the activation of the immune response in Drosophila. One pathway, defined by the Toll gene, is involved in the defense against fungi. A constitutively active form of Toll can mediate superinduction of the Cecropin gene promoter in the transfected hemocyte line mbn-2. Toll is preferentially involved in the induction of antifungal components such as Drosomycin and metchnikowin. The Toll pathway also requires other dorsal group genes such as tube, pelle, cactus, and spatzle (spz), but not dorsal itself. A second pathway directs an antibacterial response. It is defined by the immune deficiency (imd) mutation, which affects the expression of diptericin and other bactericidal factors, but not of drosomycin. In agreement with this model, factors that have both antibacterial and antifungal activity, such as the cecropins and metchnikowin, appear to respond to both pathways. In addition to these two pathways, recent experiments with the Toll-like gene, 18-wheeler, indicate that a third pathway may exist. Mutations in 18-wheeler affect the induction of Attacin, but not Diptericin or Drosomycin (Hedengren, 1999 and references therein).

Although one or more Rel factors are likely to mediate signals from these pathways in the immune response, the specific role of the three known Rel factors in Drosophila is unclear. Dorsal is not required, because dorsal mutants have no phenotypic effect on the induction of the antimicrobial genes. However, recent data show that Dif is specifically involved in the induction of drosomycin and that Dorsal may play a redundant role in the same pathway. Using P element-mediated mutagenesis, a series of deletions in the Relish gene have been created. These mutations have a profound effect on the induction of the entire set of antimicrobial genes, and it is concluded that Relish is a key factor that controls the antibacterial as well as the antifungal defense in Drosophila. The most straightforward interpretation of the broad effect of the Relish mutants on the antibacterial and antifungal responses is that Relish is involved in all three immunity pathways of Drosophila (Hedengren, 1999)

Relish is a key factor in the induction of the humoral immune response in Drosophila, including antibacterial as well as antifungal factors. In striking contrast, there is a complete lack of effect of Relish on cellular immune reactions and hematopoiesis. These observations indicate that cellular and humoral immune responses are controlled by distinct and independent systems in Drosophila. Hematopoiesis, and perhaps the activation of cellular reactions, may instead be controlled by another Rel factor, Dorsal. Overexpression of this factor has been shown to induce lamellocyte differentiation and cause the formation of melanotic capsules, and similar effects have been observed with mutations that lead to the activation of the Toll pathway (Hedengren, 1999 and references therein).

It is interesting to compare the roles of different Rel factors in the cellular and humoral immune responses of Drosophila with the situation in mammals. A parallel can be drawn between the phenotype of the Relish mutants and the effects seen in knockout mice that carry a disruption of the p105 gene. Such mice have a normal complement of lymphocytes, but they are deficient in both specific and nonspecific responses to infection (Sha, 1995). In contrast, disruption of the RelA and RelB genes in mice gives more complex developmental phenotypes, including effects on hematopoiesis. Thus, although the picture is less clear in mice than in Drosophila, it appears to be a conserved feature that different Rel factors regulate immune responses and hematopoiesis (Hedengren, 1999 and references therein).

The relative importance of the humoral factors in Drosophila immunity is dramatically illustrated by the phenotype of the Relish mutants. A single bacterial cell is sufficient to kill these mutants and the resistance to fungal infections is also impaired, even though the cellular immune defense appears to be intact and parasites are efficiently eliminated. This extreme sensitivity to a bacterial infection is surprising, since phagocytosis of bacteria appears to be normal. One possible explanation is that the antibacterial effector functions of the phagocytic cells are impaired in the Relish mutants. Alternatively, and perhaps more likely, enterobacteria such as Enterobacter cloacae may be relatively resistant to the killing mechanisms of the hemocytes. Indeed, injected cells of Enterobacter cloacae or Escherichia coli can survive for a long time in wild-type Drosophila, even though bacteria injected later are eliminated efficiently by the induced immune response. The surviving bacteria probably reside within phagocytic cells, where they are protected from the humoral factors. Thus, the humoral immune response may be the only efficient defense against bacteria such as Enterobacter cloacae or Escherichia coli (Hedengren, 1999 and references therein).

In spite of the dramatic effects on the immune response, homozygous Relish mutants are fertile and give rise to normal offspring. This lack of Relish developmental effects is surprising, considering the fact that early embryos express a specific, maternally contributed form of Relish that lacks the N-terminal domain (Dushay, 1996). The existence of a specific embryonic form suggests that Relish, like dorsal, could play a role in embryogenesis. However, the function of Relish in the early embryo, if any, must be redundant or else it is involved in a more subtle process that is not required for survival (Hedengren, 1999).

Two alternative models are suggested for the role of Relish in the three genetically defined induction pathways of immune responses in Drosophila. The most straightforward interpretation of the broad effect of the Relish mutants on the antibacterial and antifungal responses is that Relish is involved in all three pathways. The case is most clear for the imd pathway, which is of particular importance for the induction of Diptericin and other genes of the antibacterial defense. Since Diptericin induction has an absolute requirement for a functional Relish gene, it has been concluded that Relish is likely to be part of this pathway. Furthermore, Relish could also be involved in the Toll and 18-wheeler pathways, since the inducibility of both Drosomycin and Attacin is significantly reduced in Relish mutants. However, additional factors have to be invoked in order to explain how different pathways can specifically induce different genes. For the Toll pathway, this specificity could be provided by Dif and/or dorsal since these factors have been shown to participate in the Toll-dependent induction of Drosomycin, perhaps in the form of a Relish-Dif (or Relish-Dorsal) heterodimer. In this case, a homo- or a hetero-dimer of Dif and/or Dorsal could partially substitute for loss or Relish, explaining the residual expression of Drosomycin in the Relish mutants. Data supporting this model were very recently obtained by Han (1999). A synergistic effect on Drosomycin expression was found when Relish was overexpressed together with Dif. This indicates that the two factors may indeed cooperate, perhaps as a heterodimer, in the Toll pathway. For the other pathways, no candidates for such specificity factors have been found. The induction of Drosomycin and Attacin is not completely abolished in the Relish mutants, and for this reason at least one Relish-independent pathway must be included in the model (Hedengren, 1999 and references therein).

A second model, which is also consistent with these findings, is suggested by the fact that there is a residual inducibility of Drosomycin in Toll mutants and of Attacin in 18-wheeler mutants. According to this model, Relish and Toll (the latter probably via Dif) act independently to induce Drosomycin, and similarly Relish and 18-wheeler form independent pathways for the induction of Attacin (Hedengren, 1999 and references therein).

It is interesting to observe the complexity of the different signal transduction pathways that converge to activate the immune responses in Drosophila. At least two of these pathways depend on receptors of the Toll family. Several other members of this family can be identified in the database of the Drosophila genome project, and it is possible that yet other signal pathways will be identified as the role of these receptors becomes clear. The availability of Relish and other mutants will help in understanding these different pathways and their role in immunity, not only in Drosophila but probably also in humans (Hedengren, 1999).

GENE STRUCTURE

All three Relish transcripts contain both Rel and ankyrin domains, since hybridization of Northern blots with probes from both domains show identical results. To further characterize the maternal transcript, five cDNA clones were isolated from an ovarian library. These clones are identical to those from the adult and mbn-2 cell libraries except for various degrees of truncation at the 5' end. The longest clone has an insert of 2.7 kb and thus must be near full-length. 5'-rapid amplification of cDNA end products from 0- to 2-h-old embryos show the same sequence and terminate near the 5' end of this clone. It is concluded that the maternal transcript differs from the other two Relish transcripts at the 5' end only, and that if there is a unique 5' exon it must be very short. The maternal transcript is too short to encode the N-terminal part of the open reading frame, and has an alternative translation start site (Dushay, 1996).

Relish maps near the end of the 85C region on the right arm of the third chromosome. This is a well-characterized region of the genome, covered by contiguous genomic walks around the oskar and pumilio genes, as well as by genomic clones around neuralized. Cosmid clones were obtained from the relevant interval and they were probed with the Relish cDNA clone 5.3 (Dushay, 1996). In this way, Relish could be mapped to a cosmid, cos 1698, between pumilio and neuralized. 10.8 kb around the Relish gene were sequenced and several cDNA clones corresponding to the transcripts of this region were isolated. The genomic organization in this region turns out to be complex, with overlapping transcripts from at least four different genes. Twenty cDNA clones from the Relish gene have been described (Dushay, 1996). They are all collinear, and there is no evidence for alternative splicing of the Relish gene or for the use of alternative polyadenylation sites. Nevertheless, at least three different Relish transcripts of 2.7, 3.1, and 3.4 kb can be detected on Northern blots (Dushay, 1996). The 2.7 kb transcript is expressed in females and in early embryos, presumably as a maternally contributed transcript. This conclusion is supported by the observation that Relish is expressed in nurse cells in the ovary, as detected by in situ hybridization (Hedengren, 1999).

The 3.1 kb transcript is very strongly induced in infected flies, whereas the 3.4 kb transcript is constitutively present. 5' RACE has been used to sequence and accurately map the 5' ends of these transcripts, as well as of a fourth, 3.5 kb transcript of low abundance. The latter transcript can also occasionally be detected as a faint band on Northern blots. These data show that the different Relish transcripts originate from one minor and three major transcription start sites that are arranged in tandem. The entire Relish transcription unit is situated inside an intron of the previously described Nmdmc gene (Price, 1993), which encodes a putative NAD-dependent methylenetetrahydrofolate dehydrogenase-methenyltetrahydrofolate cyclohydrolase (Hedengren, 1999).

This is a bifunctional enzyme involved in tetrahydrofolate metabolism, with no obvious relationship to immune responses. From the Nmdmc gene, four different cDNA clones have been isolated, all of them representing the same splice form (transcript A) with a first exon 2.4 kb upstream of Relish, spliced to two exons downstream of the Relish gene. The first exon of this transcript differs from that of the previously published Nmdmc transcript (transcript B; Price, 1993) for which the first exon could be mapped to a position 1.2 kb upstream of Relish. Thus, the Nmdmc gene has two alternative splice forms, both of them spanning the region that contains the four Relish transcripts. The protein products encoded by the two transcripts are essentially identical. In transcript B, the first exon is noncoding, whereas transcript A encodes an additional six N-terminal amino acid residues (Hedengren, 1999).

PROTEIN STRUCTURE

Amino Acids - 817

Structural Domains

Relish contains both a Rel homology domain and an IkappaB-like domain with six ankyrin repeats. A phylogenetic reconstruction of the possible relationship between the different Rel homology domains shows that Relish may have branched off from other Rel proteins at a very early stage. The sequences of the ankyrin repeats are also quite different from those of other IkappaB-like proteins, and they are about equally close to the ankyrin repeats in the Notch, ankyrin and IkappaB families. As in other Rel proteins, a putative nuclear localization sequence is found at the C-terminal end of the Rel homology domain, and the ankyrin repeats are followed by an acidic, PEST-like sequence. PEST sequences are rich in proline, glutamic acid, serine and threonine residues, and they have been implicated in protein turnover (Dushay, 1996).

date revised: 5 November 2000

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.