BIOLOGICAL OVERVIEW

Ecdysteroid pulses trigger the major developmental transitions during the Drosophila life cycle. These hormonal responses are thought to be mediated by the ecdysteroid receptor (EcR) and its heterodimeric partner Ultraspiracle (Usp). Evidence is provided for a second ecdysteroid signaling pathway mediated by Hormone receptor-like in 38 (Hr38), the Drosophila ortholog of the mammalian NGFI-B subfamily of orphan nuclear receptors. Hr38 also heterodimerizes with Usp, and this complex responds to a distinct class of ecdysteroids in a manner that is independent of EcR. This response is unusual in that it does not involve direct binding of ecdysteroids to either Hr38 or Usp. X-ray crystallographic analysis of Hr38 reveals the absence of both a classic ligand binding pocket and coactivator binding site, features that seem to be common to all NGFI-B subfamily members. Taken together, these data reveal the existence of a separate structural class of nuclear receptors that is conserved from fly to humans (Baker, 2003).

Of the 18 Drosophila genes that encode canonical nuclear receptors, only EcR, in conjunction with Usp, has been shown to be ligand responsive. Similarly, although a wide spectrum of ecdysteroids with biological activity have been identified in the insect hemolymph, only a small subset of these can activate EcR/Usp at physiologic concentration. These results have suggested that other signaling pathways, perhaps mediated by one or more orphan nuclear receptors, may participate in ecdysteroid responses. To that end, the possibility was investigated that Hr38 (NR4A4) may be an ecdysteroid-responsive factor based on the following observations. (1) Hr38, like the ecdysteroid-responsive EcR, is the only other Drosophila nuclear receptor known to heterodimerize with Usp, the ortholog of the vertebrate retinoid X receptor (RXR). (2) Hr38 is the Drosophila ortholog of the mammalian NGFI-B subfamily of orphan nuclear receptors, that includes NGFI-B (NR4A1), Nurr1 (NR4A2), and NOR1 (NR4A3) (Philips, 1997; Wilson, 1991; Cheng, 1997; Paulsen, 1995; Zetterstrom, 1996; Perlmann, 1995; Forman, 1995). Like other orphan receptors that heterodimerize with RXR, the NGFI-B/RXR and Nurr1/RXR heterodimers are ligand responsive, suggesting that the Hr38/Usp heterodimer may also be ligand activated (Perlmann, 1995). (3) Like EcR and usp, Hr38 is broadly expressed during the third instar, prepupal, and pupal stages, suggesting that its temporal specificity may also be conferred by a hormone (Fisk, 1995; Kozlova, 1998). (4) Both Hr38 and usp mutant flies have abnormalities in cuticle formation that are not seen in EcR mutants, thereby uncoupling the action of the two receptor heterodimer complexes and suggesting they may govern distinct ecdysteroid signaling pathways (Kozlova, 1998; Hall, 1998). Taken together, these findings raise the possibility that Hr38 participates in an ecdysteroid response pathway that is different from the one transduced by the EcR/Usp heterodimer (Baker, 2003).

Evidence is provided for the existence of an ecdysteroid signaling pathway mediated by the orphan nuclear receptor Hr38. The existence of this pathway is supported by four independent experimental findings. (1) Transactivation assays in insect cells demonstrated that a distinct group of endogenous ecdysteroids, several with no previously known function, can potentiate Hr38-dependent transcription when heterodimerized with a preactivated partner (i.e., rexinoid bound RXR or VP16-Usp). (2) Organ explants from transgenic flies bearing a Hr38-specific reporter gene were shown to be similarly responsive to ecdysteroids, indicating that this pathway can function in vivo. Importantly, the specificity of the Hr38 ecdysteroid activators and the use of RNAi methodology have excluded the involvement of EcR in mediating this response. (3) Neither the ecdysteroid agonists nor any of the known nuclear receptor coactivators are capable of binding directly to Hr38. (4) X-ray crystallographic structure analysis of the Hr38 ligand binding domain shows that Hr38 lacks the classic binding sites for either a ligand or a conventional coactivator, features that are hallmarks of all other known inducible nuclear receptors. These findings provide compelling evidence for an atypical nuclear receptor transcriptional signaling pathway that mediates ecdysteroid responses in insects (Baker, 2003).

The insect hemolymph carries a wide range of endogenous ecdysteroids, some of which are only present at specific stages during development. These may be supplemented by phytoecdysteroids that can enter the animal through its diet. Until recently, it was thought that the vast majority of these compounds were unable to elicit a biological response. Mounting evidence, however, indicates that alternate transcriptional pathways exist that are driven by ecdysteroids other than 20E. Coordinate changes in ecdysteroid-regulated gene expression occur at several stages in the Drosophila life cycle at times when the 20E titer is known to be low. In addition, the let-7 and miR-125 small temporal RNAs are induced at puparium formation in precise synchrony with the E74A 20E-inducible gene, but in a manner that is independent of either 20E or EcR (Bashirullah, 2003). Of particular relevance to Hr38 functions, α-ecdysone has been shown to drive neuroblast proliferation during early pupal development in the hornworm Manduca sexta, providing in vivo evidence that this hormone is responsible for a specific response in insects. Similarly, 3-dehydro-20E has been shown to have a potency indistinguishable from 20E in Manduca and is also observed to have high activity in Drosophila larval fat body, while it has been noted that makisterone A and not 20E is the major ecdysteroid present during the last larval instar of the honeybee. Given the reported activity of these ecdysteroids, it seems reasonable to expect that at least one of the pathways governing these responses is mediated by the Hr38 pathway described here. Further support for the hypothesis that a Hr38/Usp heterodimer may play an essential role in ecdysteroid signaling comes from the observation that Hr38 and Usp are each required for ecdysteroid-induced cuticle formation during Drosophila development. A key to the future characterization of this developmental pathway will be the use of the Hr38/Usp heterodimer and ecdysteroid agonists as tools to identify downstream target genes, which at present remain unknown (Baker, 2003 and references therein).

An interesting feature of the Hr38 response is the broad specificity and increased sensitivity that a number of ecdysteroids have for Hr38 compared to the previously described signaling pathway mediated by EcR. Indeed, even the response to 20E appears to be an order of magnitude more potent for Hr38 than for EcR. Thus, discovery of the Hr38 response pathway may not only provide a mechanism of action for other ecdysteroids in insects, but may also provide a means of augmenting the ecdysteroid-mediated functions of EcR at specific stages in the life cycle (Baker, 2003).

Another striking feature of the Hr38 response is that it requires coactivation of its heterodimer partner to become competent for transcriptional activation via ecdysteroids. The finding that VP16-Usp is able to substitute for ligand-activated RXR in transfection assays is intriguing and suggests that in vivo, wild-type Usp is capable of activation by ligand or some other coactivation mechanism. The existence of a ligand for Usp is supported by X-ray crystal data on the Usp ligand binding domain showing the presence of a large hydrophobic pocket that can be occupied by lipophilic ligands (Billas, 2001; Clayton, 2001). The observation that the Hr38 ecdysteroid response can occur in larval organs that contain wild-type Usp supports this hypothesis. Identification of the Usp ligand and/or coactivator represents a critical next step toward defining the mechanism of Hr38 action (Baker, 2003).

Hr38's distinct ecdysteroid-regulated activity points to a role that is substantially different from that of EcR, both in terms of ligand specificity and mechanism of action. Although both receptors require heterodimerization with Usp to be ecdysteroid responsive, only the EcR response appears to require conventional binding of the ecdysteroid agonist. Furthermore, the role of Usp in the EcR heterodimer is that of a silent partner (i.e., the transcriptional activity of Usp is dispensable for the ecdysteroid response). In contrast, the Hr38 pathway requires transcriptional activation of both itself and its heterodimeric partner. Surprisingly, however, this response occurs in the absence of ecdysteroid binding directly to receptor, implying the existence of a nonclassical mechanism of action (Baker, 2003).

The structure of the Hr38 ligand binding domain offers an intriguing framework from which several clues about the mechanism of Hr38 action can begin to be elucidated. Although the possibility that a ligand could bind to Hr38 by an induced-fit mechanism or to an allosteric site cannot be formally ruled out, both possibilities are considered unlikely. The tight spatial constraints forced upon the protein by the four phenylalanines within the conventional ligand binding pocket almost completely exclude the induced-fit possibility. Likewise, the inability to demonstrate any type of specific ligand binding to the protein under a variety of conditions (e.g., in the presence or absence of activated heterodimer partner) using a number of assays argues against the existence of a second binding site on the protein. An equally important finding is the loss of the charge clamp, which fundamentally excludes the Hr38 ligand binding domain from interacting with the p160 family of coactivators in a conventional fashion. This finding is consistent with the inability to observe any interactions with these coactivators in either cell-based or biochemical assays. Taken together, these results provide strong evidence that the ecdysteroid response by the Hr38/Usp heterodimer occurs through a mechanism that is different from the well-documented, direct binding paradigm that has been exhibited for numerous other RXR heterodimers. Therefore, the signaling pathway between ecdysteroid and Hr38-mediated transcription must be transduced in an atypical fashion. This mechanism, however, still appears to require the AF-2 domain of Hr38. Although it is not clear how the AF-2 contributes to receptor transactivation, the data support a model in which ecdysteroids may indirectly activate Hr38, perhaps by recruiting a specific cofactor to the Hr38/Usp heterodimer. In this model, it is tempting to speculate that ecdysteroids may activate the cofactor through a direct interaction or through a second message pathway. Regardless, the requirement for a Hr38 cofactor is implicit in these findings and its future characterization will be important to fully understand the mechanism of this new signal transduction pathway (Baker, 2003).

The principles of Hr38 action may be of help in characterizing its mammalian orthologs, the NGFI-B family of receptors. Like Hr38, these orphan receptors can function as monomers or RXR heterodimers and be activated by RXR ligands (Giguere, 1999). However, little is known about the agonist or cofactor specificity of these proteins or the mechanistic details of how they promote transactivation. The analysis carried out in this study shows that the overall conservation between Hr38 and the three mammalian NGFI-B family members is well conserved in the putative ligand binding pocket. Indeed, as shown for Hr38, modeling of the 3D structure of the NGFI-B receptors predicts the absence of both a ligand binding pocket and a coactivator binding site, suggesting that a common mechanism of action may exist for governing these receptors in mammals. Given these similarities between Hr38 and its vertebrate counterparts, it should not be surprising that, like many other insect signaling pathways, there is a lot to learn from the fly (Baker, 2003).

GENE STRUCTURE

Hr38 is a member of the steroid receptor superfamily in Drosophila homologous to the vertebrate NGFI-B-type orphan receptors. In addition to binding to specific response elements as a monomer, Hr38 interacts with the USP component of the ecdysone receptor complex in vitro, in yeast and in a cell line, suggesting that Hr38 might modulate ecdysone-triggered signals in the fly. The molecular structure and expression of the Hr38 gene has been characterized and an in vivo analysis of its function(s) in development has been initiated. The Hr38 transcription unit spans more than 40 kb in length, includes four introns, and produces at least four mRNA isoforms differentially expressed in development; two of these are greatly enriched in the pupal stage and encode nested polypeptides (Kozlova, 1998).

A genomic walk of approximately 50 kb was performed in the chromosomal 38E region where Dhr38 resides. A second Hr38 cDNA clone, cTK61, which overlaps with cTK11 (Sutherland, 1995), was partially sequenced, and both clones were mapped with respect to genomic DNA from the region using hybridization and PCR scanning of the genomic DNA with cDNA sequencing primers to identify potential introns. Genomic sequences in 38E region corresponding to part of intron 2 of Hr38 and further upstream are available (BDGP) and were supplemented with partial sequencing of genomic subclones. Genomic and cDNA sequences were compared. Taken together these experiments show that the Hr38 gene extends over at least 40 kb and includes at least 5 exons. The first intron is more than 20 kb in length, whereas the second through fourth introns are short (approximately 600 bp, 117 bp, and 184 bp, respectively). The first and second exons are unique to cTK61, and exon four is shared in its entirety by the two cDNAs; exon three is complete in cTK61 and incomplete in cTK11, and conversely exon five is shorter in cTK61 relative to cTK11. As previously reported (Sutherland, 1995) the cTK11 clone contains an open reading frame capable of encoding the Hr38 protein with a calculated molecular mass of 61 KDa, and more than 3 kb of 3' untranslated region (3' UTR). Using BDGP genomic and the cDNA sequences, a composite sequence of cTK61 was generated. Computer analysis of this sequence showed a long open reading frame of 1071 amino acids beginning at nucleotide position 1121 with double methionine codons; the third methionine of this open reading frame is also unique to cTK61, whereas the fourth corresponds to the initiation codon of cTK11. Translation beginning with the first methionine codon of cTK61 would produce a polypeptide of 1071 residues and a calculated molecular mass of 117 KDa, which includes a transcription activation domain 522 amino acids longer than in the polypeptide encoded by cTK11; the DNA and ligand binding domains of the two polypeptides appear identical. A polypeptide of ca. 120 KDa is indeed detected immunochemically upon overexpression of the hsTK61 transgene. Expression studies using RNA blots and RT-PCR, together with the fact that both cDNA clones are terminated with a poly(A) stretch, suggest that these clones correspond to distinct RNA isoforms and that the gene contains alternative promoters and polyadenylation sites. This conclusion is reinforced by the findings of Fisk (1995) who described another, shorter cDNA clone of Hr38, pLF16 (Kozlova, 1998).

cDNA clone length - 5206 bp

Bases in 5' UTR - 162

Exons - 5 for Hr38-P1 and 3 for Hr38-P2

Bases in 3' UTR - 3388

PROTEIN STRUCTURE

Amino Acids - 1073 and 556

Structural Domains

The recombinant lambda clone 4-2 containing the genomic region of the Drosophila hormone receptor 38 (DHR 38) gene, homologous to mammalian neuronal growth factor I-B (NGFI-B), was isolated by radioactive labelled oligonucleotide hybridization. The nucleotide sequence of the genomic clone reveals three exons that encode the functional domains of the protein. The N-terminal exon1 has no homology at the amino acid level with NGFI-B, the mammalian homolog. A glutamine-rich region, probably involved in transcriptional activation, is observed at the C-terminal part of this exon. A similar motif is also present upstream in another reading frame of the same strand. Both motifs are preceded by a repetitive nonanucleotide sequence containing an AluI site, resembling a duplicated human Alu-sequence. A monomeric version of this sequence, coding similarly for an oligoglutamine peptide, occurs at a surprisingly high frequency in other regulatory genes in Drosophila. In contrast to mammalian Alu sequences, this sequence is found almost exclusively in the coding regions of Drosophila genes, but not in the non-coding parts of the genes. The DNA-binding domain with two zinc-fingers, and at least part of the ligand-binding peptide, is coded by the largest middle exon2 in Hr38 and exhibits up to 100% homology in short peptide motifs to its mammalian counterpart, where these domains are split into exons 3, 4, 5, and 6. However, the length, information content, stop codon, and splice site are conserved in the last exons in both fly and rat. In situ hybridization to 0-24 h wholemount embryos shows strong expression of Hr38 in neurogenic regions and in the intestinal tract during embryogenesis, suggesting a spatial and temporal control of transcription, partially analogous to the central nervous system-specific expression of NGFI-B in mammals (Komonyi, 1996).

Activity studies indicate that Hr38 may function as an ecdysteroid receptor and, like other nuclear receptors, recruit coactivators upon ligand binding. However, despite several attempts using a variety of techniques, including direct radioligand binding and cofactor recruitment assays, it was not possible to demonstrate that Hr38, RXR, or Usp (alone or as a heterodimer) directly binds any of the potent ecdysteroid agonists or their metabolites. In addition, no interactions were detected between Hr38 and any of the known nuclear receptor cofactors, including SRC-1, GRIP1, ACTR, or NCoR. Likewise, mammalian NGFI-B family members have also been shown to lack interactions with known cofactors (Castro, 1999; Wansa, 2002; Maira, 2003), which has been suggested to be due to the lack of a coactivator interaction surface on the receptor (Wansa, 2002). Taken together, these findings suggest the existence of an atypical signal transduction mechanism that governs Hr38 transactivation and that apparently does not require direct binding of ligand or known coactivators. To begin to explore this mechanism in more detail, the X-ray crystal structure of the Drosophila Hr38 ligand binding domain was solved. The overall architecture of the Hr38 ligand binding domain is similar to other members of the nuclear receptor family. The structure consists of 11 α helices with a small three-stranded β sheet arranged in a three-layer helical sandwich. Helices H4, H5, H8, and H9 are packed between helices H1 and H3 on one side, with H7 and H10 on the other side. Hr38 does not contain helix H2 but has an additional short helical segment between helices H9 and H10. The AF-2 helix of Hr38 (residues 1060-1069) was found in the 'active' conformation, characteristic of agonist bound nuclear receptors (Baker, 2003).

In many nuclear receptor ligand binding domain crystal structures, the protein exists as a homo- or heterodimer with an extensive dimer interface along helix H10. In the Hr38 structure, the protein appears as a monomer despite the fact that two molecules are found in the asymmetric unit. Hr38 is known to bind to specific response elements as a monomer as well as a heterodimer with Usp. No significant crystal contacts were observed in the structure that would suggest a biologically relevant homodimer interface. This finding is consistent with the results of size exclusion chromatography using the crystallography construct. Although the structural basis of Hr38's ability to heterodimerize with Usp must await the structure of the Hr38/Usp complex, Hr38 does contain the consensus heterodimerization motif of φ ψKψψ ψKψψ Σ ψRψψ in the first half of helix H10, where φ = the hydrophobic aromatic residues Phe, Trp, or Tyr; ψ = a hydrophobic aliphatic residue with preference for Met, Leu, Val, Pro, Ala, or Ile; and Σ = the acidic residue Asp or Glu. This motif corresponds to F SRLL GKLP E LRSL in Hr38 (residues 1027-1040) (Baker, 2003).

Of significant interest was the finding that the apo-Hr38 structure does not contain a well-defined ligand binding site. The side chains of four phenylalanines (F881, F922, F939, and F954) point into the interior of the pocket and essentially fill the entire space. Calculation of the volume of the cavity within the putative binding pocket revealed that the largest contiguous pocket is only ∼30 ų, which is too small to allow binding of any organic small molecule. Three of the four phenylalanines adopt the only available low energy conformations, while the fourth can only move within a single plane before encountering a significant steric barrier. Thus, it does not appear that any simple structural changes are available that might open up the binding pocket for access to a potential ligand (Baker, 2003).

Most nuclear receptors contain a conserved arginine within helix H5 that serves to anchor and correctly position ligands within the predominantly hydrophobic ligand binding pocket. While this arginine is conserved in Hr38 (R929), its side chain points out toward solvent rather than toward the interior of the protein. The β sheet packs tightly against the end of helix H5, excluding R929 from the potential binding pocket. The absence of a ligand binding pocket is consistent with the inability to detect ecdysteroid binding to Hr38. It is concluded from these data that ecdysteroid activation of the Hr38/Usp heterodimer must occur through an alternative mechanism that does not involve direct binding of the agonist to the receptor complex. The details of this mechanism are currently under study (Baker, 2003).

Hr38 has high sequence identity with the ligand binding domain of the three human NGFI-B subfamily members. A pairwise analysis of these sequences shows that Hr38 has the same level of conservation with each of the human NGFI-B family members as the human receptor subtypes have among themselves, suggesting that each of these four receptors evolved from a common ancestor. Further analysis of the Hr38 and mammalian sequences reveals that the four phenylalanines that fill the ligand pocket are conserved, as are almost all the amino acids that make up the core of the ligand binding domain. The remarkable conservation between these receptors suggests that the vertebrate NGFI-B subfamily members also do not have a conventional ligand binding pocket (Baker, 2003).

The coactivator binding site in other nuclear receptors consists of a hydrophobic groove formed by helices H3, H4, H5, and the AF-2 helix. In these receptors, the LXXLL motif of the coactivator is positioned within the groove by a charge-clamp interaction involving a highly conserved glutamic acid on AF-2 and a lysine on helix H3. Sequence alignments of the Hr38/NGFI-B family with other nuclear receptors indicate that the charged clamp residues are not conserved. The conserved glutamic acid on the AF-2 is an asparagine in Hr38 (N1065) and a lysine in the mammalian NGFI-B receptors. The conserved lysine on helix H3 is a glutamic acid in both Hr38 (E897) and its mammalian orthologs. Interestingly, in the Hr38 structure, the AF-2 helix is shifted by one turn relative to its position in other nuclear receptor ligand binding structures. The AF-2 helix is held in this position by a series of hydrophobic contacts with the main body of the ligand binding domain. The side chains of I1063 and M1066 are buried completely at the interface, with the M1066 side chain packing in a hydrophobic depression created by the side chains of L885, L918, L1045, and I1048. Due to the shift of the AF-2 helix in Hr38, A1061, rather than N1065, sits at the same position as the conserved glutamic acid of the charge clamp. These features result in the loss of the charge clamp in Hr38. In addition, the hydrophobic cleft that makes up the LXXLL motif binding site is blocked by a number of hydrophilic residues in Hr38. The side chains of L893 and N1065 point into the groove, partially blocking one of the leucine binding pockets and closing off one end of the groove. Thus, it is unlikely that the Hr38 AF-2 helix can stabilize the binding of coactivator proteins through the LXXLL coactivator motif. These observations may explain why none of the well-characterized coactivator proteins for all other nuclear receptors have been shown to interact with any members of the Hr38/NGFI-B family. Taken together with the finding that these orphan nuclear receptors lack a conventional ligand binding pocket, the structural analysis supports the notion that these receptors must use an alternate mechanism for effecting transactivation of gene expression (Baker, 2003).

date revised: 20 November 2003

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.