Gene name - Suppressor of variegation 3-9
Cytological map position - 88E6-8
Function - enzyme - methylates histone H3
Symbol - Su(var)3-9
FlyBase ID: FBgn0003600
Genetic map position - 3-56.4
Classification - GTP-binding elongation factor, chromo domain, nucleotide triphosphate hydrolase
Cellular location - nuclear
|Recent literature||Maksimov, D. A., Laktionov, P. P., Posukh, O. V., Belyakin, S. N. and Koryakov, D. E. (2017). Genome-wide analysis of SU(VAR)3-9 distribution in chromosomes of Drosophila melanogaster. Chromosoma [Epub ahead of print]. PubMed ID: 28975408
Histone modifications represent one of the key factors contributing to proper genome regulation. One of histone modifications involved in gene silencing is methylation of H3K9 residue. Present in the chromosomes across different eukaryotes, this epigenetic mark is controlled by SU(VAR)3-9 and its orthologs. Despite SU(VAR)3-9 was discovered over two decades ago, little is known about the details of its chromosomal distribution pattern. To fill in this gap, DamID-seq approach was used and high-resolution genome-wide profiles were obtained for SU(VAR)3-9 in two somatic (salivary glands and brain ganglia) and two germline (ovarian nurse cells and testes) tissues of Drosophila melanogaster. Analysis of tissue and developmental expression of SU(VAR)3-9-bound genes indicates that in the somatic tissues tested, as well as in the ovarian nurse cells, SU(VAR)3-9 tends to associate with transcriptionally silent genes. In contrast, in the testes, SU(VAR)3-9 shows preferential association with testis-specific genes, and its binding appears dynamic during spermatogenesis. In somatic cells, the mere presence/absence of SU(VAR)3-9 binding correlates with lower/higher expression. No such correlation is found in the male germline. Interestingly, transcription units in piRNA clusters (particularly flanks thereof) are frequently targeted by SU(VAR)3-9, and Su(var)3-9 mutation affects the expression of select piRNA species. This analyses suggest a context-dependent role of SU(VAR)3-9. In euchromatin, SU(VAR)3-9 may serve to fine-tune the expression of individual genes, whereas in heterochromatin, chromosome 4, and piRNA clusters, it may act more broadly over large chromatin domains.
|Masuko, K., Fuse, N., Komaba, K., Katsuyama, T., Nakajima, R., Furuhashi, H. and Kurata, S. (2018). winged eye induces transdetermination of Drosophila imaginal disc by acting in concert with a histone methyltransferase, Su(var)3-9. Cell Rep 22(1): 206-217. PubMed ID: 29298422
Drosophila imaginal disc cells exhibit a remarkable ability to convert cell fates in response to various perturbations, a phenomenon called transdetermination (TD). winged eye (wge) has been identified as a factor that induces eye-to-wing TD upon overexpression in eye imaginal discs, but the molecular mechanisms underlying TD have remained largely unclear. This study found that wge induces various histone modifications and enhances the methylation of Lys9 on histone H3 (H3K9), a feature of heterochromatin. A histone methyltransferase, Su(var)3-9, is required for wge-mediated H3K9 methylation and eye-to-wing TD. Su(var)3-9 is also required for classical wound-induced TD but not for normal development, suggesting its involvement in several types of imaginal disc TDs. Transcriptome analysis revealed that wge represses eye identity genes independently of Su(var)3-9 and activates TD-related genes by acting together with Su(var)3-9. These findings provide new insights into diverse types of chromatin regulation at progressive steps of cell-fate conversions.
Modifier mutations of position-effect variegation (PEV) represent a useful tool for a genetic and molecular dissection of genes connected with chromatin regulation in Drosophila. Su(var)3-9 mutations show a strong suppressor effect even in the presence of PEV enhancer mutations, indicating a central role of this gene in the regulation of PEV (Tschiersch, 1994). Human (SUV39H1) and mouse (Suv39h1) homologs (Aagaard, 1999) of Drosophila Su(var)3-9, encode novel enzymes which selectively methylate histone H3 at lysine 9 (Rea, 2000). Mammalian SUV39H1 proteins associate with HP1 (Aagaard, 1999), and the SUV39H1-dependent methylation mark in the H3 N terminus generates a heterochromatic affinity for the SUV39H1-HP1 complex (Lachner, 2001; Bannister, 2001). These data define the SUV39H1-HP1 methylation system as an important regulatory mechanism for the induction and propagation of heterochromatic subdomains in mammalian chromatin. Moreover, over-expression of the SUV39H1 HMTase in HeLa cell lines redistributes endogenous HP1 proteins and results in growth retardation (Melcher, 2000; Firestein, 2000), indicating that SUV39H1-mediated modulation of heterochromatin can impair cell cycle progression (Czvitkovich, 2001 and references therein).
Su(var)3-9 possesses a chromo domain, evidencing its evolutionary affinities with genes of the Polycomb-group and trithorax-group. The majority of Pc-G and trx-G gene products differ from classical transcription factors because they can control gene activity in an apparently sequence-independent manner, suggesting a role in modulating gene activity by altering the chromatin structure. For example, the Pc-G protein Ezh2 is associated with a complex containing histone deacetylase (HDAC) activity and several trx-G proteins interact with components of chromatin remodeling machines. Some Pc-G and trx-G gene products contain evolutionarily conserved protein motifs, e.g. the chromo or the SET domain, that are also present in another group of bona fide chromatin regulators -- the modifiers of position-effect-variegation (PEV). A few Pc-G and trx-G members can indeed be classified as PEV modifying genes. The PC protein shares the 60 amino acid chromo domain with the heterochromatin protein HP1, one of the best characterized modifier genes in Drosophila and mammals (Czvitkovich, 2001 and references therein).
Because HP1 also interacts with a variety of transcriptional co-repressors, it could function as an adaptor molecule to restrict gene expression programs by inducing a heterochromatic positioning of target genes. The suppressor of position-effect-variegation, SU(VAR)3-9 (Tschiersch, 1994), the Pc-G protein Enhancer of zeste and the trx-G protein Trithorax contain the 130 amino acid SET domain, which has recently been linked with histone methyltransferase (HMTase) activity (Rea, 2000). Moreover, the SET domain is a target for phosphorylation-dependent signaling pathways through the anti-phosphatase Sbf1, which can trigger the oncogenic potential of the mammalian TRX homolog, Mll/HRX. These observations suggest that some PEV modifier genes could, at least in part, induce Pc-G- or trxG-like phenotypes by modulating the regional organization of chromatin domains (Czvitkovich, 2001 and references therein).
Su(var)3-9 is a dominant dose-dependent modifier, and extra gene copies significantly enhance silencing of different PEV marker genes (Tschiersch, 1994). Studies by Aagaard (1999) demonstrate the functional homology between human and fly Su(var)3-9 genes. Using the 'triplo-enhancer effect' of Su(var3-9) as an experimental assay, transgenic fly lines were established that carry the human SUV39H1 or a (myc)3-tagged variant cDNA under the control of the heat shock promoter hsp70. From a total of 14 transformed fly lines, nine insertions in the second and third chromosome were selected, and basal activity of the transgene was confirmed by expression analysis. As controls, transgenic flies carrying Su(var)3-9 cDNAs or a genomic fragment comprising the Su(var)3-9 locus (Tschiersch, 1994) were used. All transgenic lines were crossed into the In(1)wm4h indicator strain, which contains an inversion placing the white marker gene adjacent to pericentric X heterochromatin. In this strain, heterochromatin-mediated, variegated white gene expression can be easily detected as red (active transcriptional state) or white (repressed transcriptional state) patches in the Drosophila eye (Aagaard, 1999).
Visual inspection of progeny derived after crossing In(1)wm4h transgenic females into In(1)wm4h males indicates that all Su(var)3-9 and SUV39H1 transgenes induce a significant increase in the proportion of unpigmented areas in the eyes, therefore demonstrating repression of wm4h gene activity. This 'triplo-enhancer effect' is largely independent of heat shock treatment and correlates with basal transcription of the preselected transgenes. In contrast, ongoing studies reveal that partial rescue of the 'haplo-suppressor effect' of Su(var)3-9 heterozygotes by human SUV39H1 requires the correct developmental expression of the transgene from very early embryogenesis (Aagaard, 1999).
To quantify the degree of PEV enhancement, In(1)wm4h; transgenic males were crossed into the 'sensitized' In(1)wm4h; Su(var)2-1 indicator strain, which allows a more accurate measurement of red-eye pigments as in the In(1)wm4h strain. Eye pigments were extracted from male progeny and pigment absorbance at 480 nm was determined. The results of these quantitations show that one extra gene copy of genomic Su(var)3-9 induces a pronounced (8- to 14-fold) reduction in the concentration of red-eye pigments, which was reflected by 8- to 28-fold reduced levels in the three lines carrying Su(var)3-9 cDNAs. Importantly, the nine lines with SUV39H1 cDNAs also displayed a 2- to 7-fold reduction in red eye pigmentation. Despite some variation among the transgenic lines and although a significant fraction of transgenic flies carrying Su(var)3-9 and SUV39H1 cDNAs display paternal effects, these results demonstrate functional homology between human and fly Su(var)3-9 genes, and indicate that SUV39H1 is capable of repressing gene activity in the vicinity of heterochromatin (Aagaard, 1999).
To examine its role in heterochromatin-mediated gene repression in more detail, the distribution of the transgenic human (myc)3-SUV39H1 protein at Drosophila polytene chromosomes was analyzed. Third instar larvae of transgenic line mA were heat shocked for 2h to increase expression of the hsp70-driven transgene, and localization of (myc)3-SUV39H1 protein was detected with monoclonal alpha-myc (9E10) antibodies at polytene chromosomes that were prepared 30 and 60 min after heat shock (Aagaard, 1999).
Immunolocalization of (myc)3-SUV39H1 30 min after heat shock reveals a pronounced staining of the chromocenter and of the fourth chromosome, indicating a preferred association with heterochromatin. At a later timepoint (60 min), staining at the chromocenter appears enhanced and several discrete euchromatic sites are visualized. A similar distribution of transgenic (myc)3-SUV39H1 has also been observed in other lines and is even reflected by ectopic SU(VAR)3-9 proteins that contain green fluorescent protein as a tag. These data demonstrate preferred binding of SU(VAR)3-9 related proteins to the chromocenter and thus support their direct involvement in regulating heterochromatin-mediated gene repression of pericentric marker genes (Aagaard, 1999).
Suv39h1/SUV39H1 and a mammalian homolog of Drosophila HP1, M31, were found to colocalize at heterochromatic foci in mouse interphase nuclei. The significant co-localization with M31 suggests that Suv39h1/SUV39H1 and M31 may be components of a heterochromatic protein complex in vivo. To address this notion directly, co-immunoprecipitations (co-IPs) were performed with nuclear extracts from murine Cop8 cells, human HeLa cells and HeLa-B3 cells that 'stably' overexpress (myc)3-SUV39H1. The HeLa-B3 cells were chosen, because higher amounts of protein can be immunoprecipitated with monoclonal alpha-myc (9E10) antibodies. To detect possible complex formation between endogenous Suv39h1/SUV39H1 and M31, co-IPs were also performed with monoclonal alpha-M31 antibodies. Since the sizes of SUV39H1 (48 kDa), (myc)3-SUV39H1 (55 kDa) and M31 (25 kDa) largely co-migrate with either the heavy or the light chain of immunoglobulins, alpha-myc and alpha-M31 antibodies were covalently coupled to protein G-Sepharose beads. Following IP with these antibody beads, immunoprecipitates were separated by SDS-PAGE, transferred to nitrocellulose membranes and probed with alpha-myc, alpha-Suv39h1, alpha-M31 and human auto centromeric antibodies (hACA) antibodies (Aagaard, 1999).
The results of these co-IPs show that the alpha-myc beads specifically immunoprecipitate (myc)3-SUV39H1 from nuclear extracts of HeLa-B3 cells but, as expected, not from HeLa or Cop8 cells. Interestingly, M31 is present in the precipitated material, indicating complex formation with ectopically expressed (myc)3-SUV39H1. In contrast, CENP-A (19 kDa), which is a crucial hACA epitope of the inner centromeric region, does not co-immunoprecipitate with (myc)3-SUV39H1. Using the alpha-M31 beads in the converse co-IPs, similar amounts of endogenous M31 are enriched from nuclear extracts of all three cell lines. Importantly, endogenous SUV39H1 or Suv39h1 is co-immunoprecipitated from HeLa or Cop8 nuclear extracts. In addition, (myc)3-SUV39H1 appears over-represented in co-IPs from HeLa-B3 nuclear extracts, suggesting that the ectopic protein can efficiently compete with the lower abundant endogenous SUV39H1 for putative M31 interaction surfaces (Aagaard, 1999).
The above data demonstrate complex formation between SUV39H1 and M31, and provide the first evidence for the existence of a mammalian SU(VAR) protein complex. To characterize the approximate size of this complex, HeLa nuclear extracts, the same as used for the co-IPs, were sedimented by velocity centrifugation in a 10%-40% sucrose gradient. Twenty fractions were collected and subsequently analysed by Western blotting with the hACA, alpha-Suv39h1 and alpha-M31 antibodies. Whereas CENP-A is distributed over a broad range, other hACA epitopes (CENP-B) are enriched in the lower molecular mass fractions. In contrast, although a minor portion was detected towards the top of the gradient, the majority of SUV39H1 protein is found in fractions 8-11, which overlap with the M31 peak (fractions 7-9) (Aagaard, 1999).
M31 (also called HP1beta) represents one of several mammalian HP1 isoforms. The same protein blots were probed with antibodies that are specific for HP1alpha or M32 (also called HP1gamma). However, both M32 and HP1alpha peak in fractions 2-3, with the euchromatic M32 protein being restricted to the low molecular mass range, whereas the heterochromatic HP1alpha protein extends into higher fractions. These results indicate distinct sedimentation profiles for the three different mammalian HP1-related proteins and, together with the co-IPs shown above, provide supporting evidence that M31 is the most likely partner for endogenous SUV39H1 to be present in a multimeric mammalian SU(VAR) protein complex, which sediments at ~20S (Aagaard, 1999).
This functional analysis of human (SUV39H1) and mouse (Suv39h1) homologs of the Drosophila PEV modifier Su(var)3-9 characterizes SUV39H1 as the first mammalian Su(var) gene to be shown to modulate chromatin-dependent gene activity. Suv39h1/SUV39H1 are chromosomal proteins that are enriched at heterochromatic foci in interphase and which accumulate at centromeres of metaphase chromosomes. Moreover, Suv39h1/SUV39H1 associate with M31, providing direct evidence for the existence of a mammalian SU(VAR) protein complex. These data define Suv39h1/SUV39H1 as novel heterochromatic components and implicate these proteins in both epigenetic gene control and the structural organization of mammalian higher order chromatin (Aagaard, 1999).
The preferred affinity of endogenous Suv39h1/SUV39H1 for heterochromatic regions and of ectopic (myc)3-SUV39H1 for the polytenic chromocenter in Drosophila suggests a direct role in the organization of repressive chromatin domains and the regulation of heterochromatin-dependent gene silencing. For example, variegation and the clonal nature of gene repression have been explained by the variable and co-operative extension of heterochromatin from the chromocenter along the chromosome ('cis-silencing'). However, variegation at centromere-distal positions, like repeat-induced silencing, or even 'trans-inactivation' across homologous chromosomes is also modulated by Su(var) gene dosage. Furthermore, centromeric heterochromatin appears to be able to selectively recruit repressed genes into transcriptionally inactive subnuclear compartments. Thus, a more general model has been proposed, in which the nucleation of repressive chromatin domains is largely dictated by the pairing or looping potential of target sequences. Repeat-driven looping or pairing may induce an altered structure which is then stabilized and expanded in response to the local concentration of heterochromatin-specific proteins. SU(VAR)3-9-related proteins represent excellent candidates to match most of these required functions and, since Suv39h1/SUV39H1 are components of mitotic chromatin, they could also propagate distinct transcriptional states during cell divisions (Aagaard, 1999 and references therein).
The subnuclear distribution and chromatin association of endogenous Suv39h1/SUV39H1 proteins indicates significant co-localization with heterochromatin-specific M31 during interphase and partial overlap with epitopes recognized by human anti-centromeric autoantibodies (hACA) during metaphase. Interphase heterochromatin and mitotic chromatin most probably differ in their condensation levels, and the mitotic restructuring of chromosomes has been proposed to induce dynamic redistributions for several chromatin regulators. In this respect, the localization of Suv39h1 protein during interphase is spatially separated from hACA epitopes, which appear to cluster in the vicinity of heterochromatic foci. However, SUV39H1 specifically accumulates at centromeric positions of human metaphase chromosomes, but does not decorate pericentromeric heterochromatin. Thus, Suv39h1/SUV39H1 resembles dynamic chromosomal proteins that display highest affinities for non-centromeric, heterochromatic foci during interphase and centromeric heterochromatin at metaphase (Aagaard, 1999).
This Suv39h1/SUV39H1 staining pattern is clearly distinct from the interphase distribution of several mammalian Pc-G proteins. In addition, only a minor fraction of M33 (PC homolog) and BMI1 (PSC homolog) remains associated with mitotic chromatin, whereas EZH [E(Z) homologs] proteins do not appear to localize at human metaphase chromosomes. Direct examination of possible interactions with Pc-G proteins indicate no physical in vivo association between SUV39H1 and M33 or EZH2. These distinct staining and interaction patterns are in agreement with the described differences between Pc-G and Su(var) gene function, despite several common sequence motifs, including chromo and SET domains, that are shared by some PEV modifiers and chromosomal regulators of HOM-C (Aagaard, 1999).
The high-affinity association with centromeric positions on metaphase chromosomes implicates a direct role for SUV39H1/Suv39h1 in mammalian centromere activity. This interpretation is supported by the functional analysis of clr4 mutations that result in perturbed chromosome segregation and disrupt localization of the centromere component SWI6 (Ekwall, 1996), which represents the HP1 homolog in S.pombe. However, human SUV39H1 has so far failed to rescue clr4-dependent centromeric gene silencing in S.pombe. However, overexpression of (myc)3-SUV39H1 in HeLa cells appears to perturb chromosome segregation. Since SUV39H1 is specifically localized at the outer region of the centromere, and because clr4-dependent segregation defects are synergistically enhanced by beta-tubulin mutations (Ekwall, 1996), deregulated SUV39H1 function could probably interfere with kinetochore assembly (Aagaard, 1999).
In mammals, several centromere-specific proteins (CENPs) have been identified, of which CENP-A appears to be a crucial component of active centromeres. CENP-A resembles a histone H3-variant that is cell cycle regulated and has been implicated to target assembly of (CENP-A/H4)2 tetramers to centromeric heterochromatin, specifically during late replication. Despite the apparent similarities in centromeric localization and partly overlapping sedimentation profiles, no physical association between SUV39H1 and CENP-A, or other hACA epitopes (CENP-B and CENP-C) was detected. Instead, SUV39H1 is present in a complex with M31. According to current models, centromere function is likely to be co-regulated at multiple levels: whereas CENP-A containing tetramers may induce an altered nucleosomal array, higher order chromatin appears to be required to 'imprint' active centromeres. Based on the data, it is proposed that Suv39h1/SUV39H1 are involved in the organization of such a higher order chromatin structure at mammalian centromeres (Aagaard, 1999).
In interphase, SUV39H1/Suv39h1 significantly co-localize and co-immunoprecipitate with M31. In addition, ectopic (myc)3-SUV39H1 also associates with M31 in vivo. However, binding between in vitro co-translated SUV39H1 and M31 or retention of endogenous M31 on affinity columns that contain bacterially expressed GST-Suv39h1 has not been detected, suggesting that possible direct interactions are dependent on post-translational modifications. The sedimentation profiles of SUV39H1 and mammalian HP1 isoforms are most consistent with SUV39H1 and M31 being present in a common, multimeric complex of ~20S. In contrast, the also heterochromatic HP1alpha or the euchromatic M32 are restricted to lower molecular mass fractions. These results underscore the specificity of the SUV39H1-M31 complex and are in agreement with described differences in interacting partners for M31 and HP1alpha that have been identified through yeast two-hybrid screens (Aagaard, 1999).
However, although SUV39H1 and M31 share part of their peak fractions, they do not entirely co-sediment, raising the possibility that both proteins may also participate in more promiscuous interactions and in the formation of additional complexes. A variety of heterogeneous partners implicated in transcriptional regulation, replication and subnuclear architecture have been described for HP1-related proteins. Although these interactions would be consistent with the proposed molecular nature of HP1 as an 'adaptor protein', direct biochemical interactions in vivo have been difficult to define. Recently, physical association between Drosophila SU(VAR)3-7 and HP1 has been reported. Since Su(var)3-9, Su(var)3-7 and Su(var)2-5 (HP1) are all dose-dependent modifiers of PEV, this finding predicts the putative mammalian SU(VAR)3-7 homolog(s) as another likely candidate to be present in the SUV39H1-M31 protein complex. In summary, this analysis of mammalian SU(VAR)3-9-related proteins characterizes SUV39H1/Suv39h1 as novel heterochromatic components and provides an entry point to dissect the structural principles that underlie the formation and function of mammalian higher order chromatin (Aagaard, 1999 and references therein).
The Drosophila suppressor of position-effect variegation Su(var)3-9 encodes a heterochromatin-associated protein that is evolutionarily conserved. In contrast to its yeast and mammalian orthologs, the Drosophila Su(var)3-9 gene is fused with the locus encoding the gamma subunit of translation initiation factor eIF2. Synthesis of the two unrelated proteins is resolved by alternative splicing. A similar dicistronic Su(var)3-9/eIF-2gamma transcription unit was found in Clytus arietis, Leptinotarsa decemlineata, and Scoliopterix libatrix, representing two different orders of holometabolic insects (Coleoptera and Lepidoptera). In all these species the N terminus of the eIF-2gamma, which is encoded by the first two exons, is fused to SU(VAR)3-9. In contrast to Drosophila melanogaster, RT-PCR analysis in the two coleopteran and the lepidopteran species demonstrate the usage of a nonconserved splice donor site located within the 3' end of the SU(VAR)3-9 ORF, resulting in removal of the Su(var)3-9-specific stop codon from the mRNA and complete in-frame fusion of the SU(VAR)3-9 and eIF-2gamma ORFs. In the centipede Lithobius forficatus eIF-2gamma and Su(var)3-9 are unconnected. Conservation of the dicistronic Su(var)39/eIF-2gamma transcription unit in the studied insects indicates its origin before radiation of holometabolic insects and represents a useful tool for molecular phylogenetic analysis in arthropods (Krauss, 2000).
By molecular analysis, Su(var)3-9 can be correlated with a 2.4 kb transcript which encodes a putative protein of 635 amino acids containing a chromo domain and a region of homology to Enhancer of zeste and Trithorax, two antagonistic regulators of the Antennapedia and Bithorax gene complexes, as well as to the human protein ALL-1/Hrx, which is implicated in acute leukemias. This region of homology is found in all four proteins at the C-terminus. The homology of Su(var)3-9 to both negative (Polycomb and Enhancer of zeste) and positive (trithorax) regulators of the Antennapedia and Bithorax complexes also suggests similarities in the molecular processes connected with stable transmission of a determined state and the clonal propagation of heterochromatinization (Tschiersch, 1994).
Sequence comparisons of the 412 amino acid human SUV39H1 and murine Suv39h1 proteins indicate overall identities of 95%. Both the human and mouse homologs lack 155 N-terminal amino acids of SU(VAR)3-9 (635 amino acids) (Tschiersch, 1994). Interestingly, the 490 amino acid CLR4 protein also does not contain the fly-specific N-terminal extension (Aagaard, 1999).
Overall, cross-species amino acid identities reach 42% between the fly and the two mammalian proteins, and 38% between CLR4 and SUV39H1/Suv39h1. Alignment of all four proteins reveals three regions of sequence identity. Most highly related is the 130 amino acid SET domain core (36% identity), which is followed at the very C-terminal tail by three conserved cysteine residues. N-terminal to the SET domain is a 110 amino acid domain (27% identity) that contains several conserved cysteine residues. Cysteine-rich regions appear specifically associated with the SET domains of most, but not all SET domain proteins. Despite the lack of homology to other well-defined cysteine stretches, such as RING (C3HC4)- and PHD (C4HC3)-fingers or LIM domains (C2HC5), this cysteine-rich cluster -- and probably also the three C-terminal cysteine residues -- may participate in facilitating molecular interactions. The third conspicuous sequence motif is the 40 amino acid chromo domain (20% identity) which is located close to the N-termini of SUV39H1/Suv39h1. Finally, although not present in S.pombe CLR4, both mammalian N-termini share a 45 amino acid region (29% identity) with SU(VAR)3-9 (Aagaard, 1999).
date revised: 4 April 2002
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.