Based on the consensus protein sequence for a SOCS box derived by Hilton and colleagues 30, a tBLASTn search of the Berkeley Drosophila Genome Project (BDGP) database 31 was conducted to examine all possible reading frames. Three putative loci containing both a SOCS box and an SH2 domain were identified using this strategy. All three match the arrangement of mammalian SOCS genes in that the SOCS box is at the carboxyl terminus with the SH2 domain directly preceding it. Each of these putative homologues also overlaps with a predicted gene from the BDGP. We named these three genes Socs16D (overlapping with CG8146), Socs36E (overlapping with CG15154), and Socs44A (CG2160), based upon their cytological location. Comparison of these three fly SOCS genes with vertebrate SOCS reveals that Socs36E is most similar to the mouse SOCS5, while Socs16D and Socs44A are less similar to specific mouse SOCS (see Fig. 1). While the amino termini are quite different, SOCS5 and Socs36E are 62% identical (71% similar) at the carboxy terminus from the region just before the SH2 domain to the end of the SOCS domain (region shown in Fig. 1A). Within that same C-terminal region, Socs44A is most similar to SOCS6 and SOCS7 (46% and 39% similar, respectively). Socs16D also has highest similarity to SOCS6 and SOCS7 (47% to each) over the same carboxyl region. These similarities suggest that the ancestral versions of Socs36E and a common predecessor of Socs16D and Socs44A existed as two separate SOCS genes at the time of divergence of mammals and dipterans (Fig. 1B).

In mammals, regulation of JAK signaling through SOCS proteins is based on a simple negative feedback mechanism. Specifically, the activity of the JAK pathway stimulates the expression of SOCS genes, because activated STATs bind to enhancers for the SOCS genes and induce transcription. Socs36E is similarly regulated during embryogenesis by Drosophila JAK signaling 29. Socs36E is expressed dynamically, in a striped pattern that later becomes restricted predominantly to the tracheal pits [2829, and Fig. 3], very similar to upd, the gene encoding the embryonic ligand for the JAK pathway 32. Indeed, activation of JAK signaling is both necessary and sufficient for Socs36E expression in embryogenesis 29. Furthermore, the expression of Socs36E during oogenesis matches the known activation of JAK signaling. The expression of upd in the ovaries is restricted to the two polar follicle cells at either end of the egg chambers of the vitellarium [26 and Fig. 2C]. Socs36E is expressed in a larger number of follicle cells centered at the two poles of the egg chamber (Fig. 2D). Given that secreted Upd protein is produced in the polar follicle cells and activates JAK signaling in neighboring cells 33, this suggests that Socs36E expression is controlled by JAK activity in oogenesis, as well as embryogenesis.

The Socs44A gene that was predicted based on protein homology is identical to hypothetical gene CG2160. A single cDNA corresponding to the locus (LP02169) was isolated by the BDGP, has been completely sequenced and encodes the expected SH2 and SOCS domains at the carboxyl terminus (gb AF435923). To determine whether Socs44A is similarly regulated by JAK pathway activity, in situ hybridization to embryos and ovaries was performed. No specific expression of Socs44A was detected until very late in embryogenesis. The only striking staining pattern observed was in the trachea of late embryos (Fig. 2A). Non-specific tracheal staining is sometimes seen with probes to late embryos, however this pattern was never observed when sense probe was used (Fig. 2B). Unfortunately, embryos homozygous for any available deletions that remove Socs44A die prior to formation of trachea, therefore we cannot conclusively determine whether the late tracheal staining reflects RNA expression. Nonetheless, because the JAK pathway is activated in a segmentally repeated pattern during embryogenesis, the lack of Socs44A expression suggests that it is not responsive to JAK signaling. Consistent with this conclusion, expression of Socs44A in the ovary is restricted to only germline expression late in oogenesis, with no detectable RNA in the follicular epithelium (Fig. 2E).

To directly test whether Socs44A expression is regulated by JAK pathway activity, in situ hybridization to Socs44A RNA was performed in embryos that lack JAK pathway activity. The product of the hop gene is required in early embryogenesis and must be provided maternally for proper segmentation of the embryo. The dominant female sterile (DFS) technique was used to generate females that fail to produce hop in the germline 34. In situ hybridization of hop germline clone embryos using Socs36E as probe demonstrates a strong reduction in Socs36E expression in the mutant embryos as compared with wild-type (Fig. 3). Similar results have been reported in embryos lacking upd activity 29. These data demonstrate that hop is required to stimulate the normal segmentally-repeated Socs36E expression in the embryo. However, expression of Socs44A does not appear to be affected by maternal loss of hop. Although the trachea are malformed and dramatically reduced in embryos lacking JAK pathway activity [3536 and Fig. 3G,3H], the remaining segments of trachea continue to express Socs44A at apparently normal levels (Fig. 3E,3F). Thus the failure of endogenous Socs44A to be expressed in the normal pattern of JAK pathway activation and of Socs44A expression to be eliminated by loss of JAK activity indicate that Socs44A expression is not stimulated by the pathway.

Activity of the JAK pathway is both necessary and sufficient for the expression of Socs36E. The ectopic activation of the JAK pathway by misexpression of upd results in expression of Socs36E in the same pattern [29 and data not shown]. In contrast, similar misexpression of UAS-upd with the paired-GAL4 driver failed to stimulate any detectable expression of Socs44A in the embryo (not shown). We conclude that Socs44A expression is not responsive to JAK pathway activity, therefore cannot function via a traditional auto-regulatory feedback loop.

The lack of transcriptional regulation by JAK signaling does not preclude a role for Socs44A in the control of JAK activity. To test whether it can attenuate JAK signaling, Socs44A was misexpressed using the GAL4/UAS system. Similar experiments performed with Socs36E have demonstrated that expression in the developing wing reproducibly results in the production of ectopic wing vein near the posterior crossvein [Fig. 4C and 28]. This phenotype is quite similar to that noted for viable mutants of hop or Stat92E [Fig. 4B and 37], suggesting that Socs36E misexpression may cause a reduction in JAK signaling in the wing. But unlike observed JAK mutations, the anterior crossvein was also completely missing from Socs36E misexpression wings, perhaps suggesting an additional role for Socs36E that is independent of the JAK pathway. Callus and Mathey-Prevot 28 demonstrated that the additional influence on wing venation may be due to the suppression of the EGFR pathway.

Using the engrailed-GAL driver, GAL-e16E, expression of Socs44A in the posterior compartment of the wing caused mild venation defects similar, but not identical, to Socs36E (Fig. 4D). Expression of Socs44A caused production of ectopic wing vein near the posterior crossvein, but unlike Socs36E, the ectopic vein was seen predominantly posterior to L5, not between L4 and L5. Furthermore, the anterior crossvein was not reduced or eliminated by Socs44A expression, but a substantial arching of L3 was noticed. Both the ectopic vein and arching of L3 were enhanced in animals heterozygous for a null allele of hop (Fig. 4E), indicating that the phenotype is sensitive to a reduction in JAK pathway activity. Misexpression of hop activates JAK signaling and causes reduction of wing venation in the posterior of the wing, somewhat the opposite of Socs44A misexpression (Fig. 4F). The simultaneous misexpression of hop and Socs44A results in a phenotype similar to expression of Socs44A alone (Fig. 4G). Therefore, the activity of Socs44A is capable of negating the influence of ectopic JAK activity in the wing.

Loss of JAK function in embryos is lethal, but various combinations of weak alleles of hop show some viability (Table 1). If Socs44A were negatively regulating the JAK pathway, misexpression of Socs44A in a hop mutant background would be expected to further reduce viability. The ability of Socs44A misexpression to enhance the lethality of weak heteroallelic combinations of hop was tested. For all alleles examined, expression of Socs44A in the engrailed pattern caused complete lethality. For the weakest hop allelic combination, hop^msv/hop^M75, misexpression of Socs44A caused viability to drop from 62% to 0% (Table 1). These data are consistent with the hypothesis that ectopic Socs44A acts to further reduce pathway activity in these JAK activity depleted animals, causing lethality.

While the above data indicate that ectopic Socs44A is capable of downregulating JAK activity, they do not address whether Socs44A has an endogenous role in JAK pathway regulation. To determine if endogenous Socs44A downregulates JAK activity, we assayed the effect of a Socs44A deficiency on hop mutant phenotypes. The hop^M38/msv heteroallelic mutant exhibits wing vein material at the posterior crossvein (Fig 4B) that is 98% penetrant. Removal of a single copy of Socs44A using either of two deficiencies in the region reduced the penetrance of the hop phenotype by as much as 52% (Table 2). An overlapping deficiency that did not remove the Socs44A locus had little effect on penetrance of the phenotype. These results suggest that regulation of JAK activity in the wing is a normal endogenous function of Socs44A.

In mammals, there are multiple points of cross-talk between the JAK and EGFR/MAPK signaling pathways 3383940. EGFR signaling plays a prominent role in many developmental processes in Drosophila, including wing venation 4142. As mentioned above, expression of Socs36E has been reported to suppress EGFR signaling in the wings 28. To determine the relationship of Socs44A to EGFR/MAPK signaling, wing phenotypes due to misexpression of Socs44A were examined in the background of heterozygous mutations for components of the EGFR signaling pathway. Engrailed-GAL4 driven misexpression phenotypes of Socs44A were suppressed in the background of heterozygous mutations for Ras85D, Son of sevenless (Sos), and Egfr (Fig. 5A,5B,5C,5D). Consistent with these observations, reduction in the dosage of the EGFR negative regulator argos enhanced the Socs44A misexpression phenotype (Fig. 5E). In contrast, concurrent misexpression of Socs44A and argos had antagonistic effects. Misexpression of two copies of an argos transgene under the engrailed-GAL4 driver resulted in wings lacking the 4^th lateral vein (L4) as well as both cross-veins (Fig. 5H). Concurrent misexpression of a single copy of the Socs44A transgene in this background was able to rescue this phenotype, restoring the posterior crossvein and both the most proximal and distal portions of L4 (Fig. 5I). The resulting wing phenotype mimicked that seen when only a single copy of argos was used in the misexpression assay (Fig. 5J) or what is seen in heteroallelic Egfr mutants (Fig. 5G). Finally, concurrent misexpression of a single copy of the argos and Socs44A transgenes produced a nearly wildtype wing (Fig. 5K). These data indicate that Socs44A expression is able to suppress argos misexpression phenotypes in a dose-dependent manner. It should be noted that concurrent misexpression of UAS-GFP did not affect the UAS-argos phenotype (not shown), indicating that the suppression by UAS-Socs44A was not merely a consequence of titrating GAL4.

Although these misexpression data indicate that Socs44A can enhance EGFR signaling, they do not necessarily demonstrate that this is a normal function of Socs44A. To address whether this is an endogenous function of Socs44A, we assayed the influence of a deficiency that removes Socs44A in the argos misexpression background. Engrailed-GAL4 misexpression of argos produces a range of phenotypic classes in which parts or all of L4 and/or the posterior cross-vein are missing (Fig. 6A). Addition of a single copy of a deficiency that removes Socs44A shifted the distribution of phenotypes to the more severe classes (Fig. 6B). In contrast, addition of an overlapping deficiency that does not include the Socs44A locus did not show such a shift. While it cannot be unambiguously stated that this effect is due to loss of Socs44A specifically, these results are consistent with the misexpression analyses and suggest that Socs44A normally plays a role in enhancing EGFR signaling in the Drosophila wing.

Evidence presented here and elsewhere indicates that Socs36E and Socs44A can downregulate JAK signaling in the wing 28. However, the ability of specific mammalian SOCS to regulate JAK activity has been observed to differ, depending upon the tissue examined 43. To determine whether there is a similar context specificity for the Drosophila SOCS, regulation was examined in another tissue in which JAK and EGFR functions have been well characterized. Both pathways are required for proper patterning of the follicular epithelium surrounding developing egg chambers during oogenesis 263344454647. One of the distinct cell populations requiring these pathways is the posterior terminal follicle cells 33. These cells are molecularly identified by the expression of the ETS domain transcription factor, pointed 474849. In clones of cells that lack hop activity (Fig. 7C,7D,7E) or egfr activity (not shown), there is a loss of pnt-lacZ expression, indicating failure to specify the posterior terminal follicle cells.

To test whether Socs36E and Socs44A can downregulate JAK or EGFR activity during oogenesis, clones of cells misexpressing these genes in developing egg chambers were examined. In clones misexpressing Socs36E at high levels in posterior cells of the developing egg chamber, there was a dramatic loss of the pnt-LacZ marker (Fig. 7F,7G,7H). This loss was restricted to only those cells that misexpressed Socs36E and did not influence neighboring cells. These results indicate that JAK and/or EGFR signaling was attenuated by Socs36E activity. In contrast, for cells in which Socs44A was misexpressed in a similar fashion, there was no reduction of pnt-LacZ expression (Fig. 7I,7J,7K). We conclude that Socs44A is unable to attenuate JAK activity in the follicle cells. This ability of Socs44A to regulate JAK signaling in the wing, but not in the ovary, indicates that SOCS activity in invertebrates can also be context specific. Furthermore, the differential ability of the fly SOCS to attenuate JAK and EGFR signaling in the ovary demonstrates distinct functions for these two proteins.

Phylogenetically, SOCS fall into three general clades. The first includes the best studied vertebrate SOCS, CIS and SOCS1-3. Interestingly, there are no representatives of this group found in the fly genome. Vertebrate SOCS of the remaining two clades have yet to be fully characterized with regard to their physiological roles, as well as mechanistic roles in JAK/STAT signaling. Socs36E is most similar to the vertebrate SOCS of the second clade, containing SOCS4 and SOCS5. It shares similarity not only in the SH2 and SOCS domain, but also in the region upstream of the SH2 domain. Mutational analysis has shown that SOCS5 inhibits IL-6 50, whereas nothing is known about the activity of SOCS4. Socs44A falls into the third clade occupied by vertebrate SOCS6 and SOCS7, as well as the only C. elegans homologue. SOCS6 has been shown to downregulate the insulin receptor 5152. Very little is known about SOCS7, other than its ability to interact with Nck, Ash, and PLCγ 53. Because of the relative lack of information about these latter two clades, study of the Drosophila SOCS may identify general properties of these homologues that span each clade.

Although mammalian genomes encode large families of specific JAK pathway components, Drosophila has only one characterized receptor, domeless, one Janus kinase, hop, and a single STAT, stat92E. Despite the simplicity of the transduction machinery for the JAK pathway, there are three SOCS genes in flies. Furthermore, there is only one Drosophila homologue of the PIAS negative regulatory family, zimp, and it is also capable of inhibiting JAK pathway activity 5455. In an organism with few functionally redundant genes, why are there three Drosophila SOCS? Two possible explanations for the apparent abundance of SOCS are that the different Drosophila SOCS may be expressed differently or they may differently regulate signaling through pathways other than JAK. Indeed, we presented evidence for both of these distinctions for Socs36E and Socs44A.

It has been demonstrated that, like the classical vertebrate SOCS genes, Socs36E is transcriptionally responsive to JAK pathway activity [29 and this work]. In both embryos and ovaries, the expression of Socs36E mirrors the known pattern of JAK activation and, indeed, altered JAK activation in the embryo elicits a transcriptional alteration in Socs36E. Unlike Socs36E, the expression of Socs44A did not match that of JAK induction. In the embryo, detectable Socs44A expression was absent until late stages of embryogenesis, when it was restricted to the developing trachea. JAK activation does occur in the tracheal pits and has been implicated in tracheal morphogenesis 3536, but Socs44A expression was lacking in the other tissues of the early embryo where JAK activation has been described. More telling was the finding that neither reduction nor expansion of JAK activation in the embryo had any effect on Socs44A expression. This disparity between Socs44A and Socs36E support the hypothesis that these genes are not redundant.

Despite the difference in expression of the two SOCS genes, both are able to downregulate JAK activity in some tissues. Misexpression of Socs36E is able to suppress JAK activity in the developing adult (imaginal) wing and thorax 28. Similarly, misexpression of Socs44A reduced JAK activity in the imaginal wing, as illustrated by the enhancement of that phenotype by reduction of endogenous hop. Furthermore, misexpression of Socs44A rescued wing vein loss resulting from misexpression of hop. Perhaps most importantly, introduction of deficiencies that remove Socs44A rescued a hop wing vein phenotype. Taken together, these data strongly suggest that Socs44A downregulates JAK pathway activity during normal wing development. However, misexpression of Socs44A had no effect on expression of a marker for JAK pathway activity during oogenesis. This indicates that there is context specificity to SOCS action in Drosophila, a phenomenon that has been observed in the study of mammalian SOCS 43. In contrast, misexpression of Socs36E was able to downregulate expression of the pnt-lacZ marker in follicle cells, although it cannot be distinguished whether this is due to reduction of signaling through JAK or EGFR. However, because Socs36E is expressed in the pattern of JAK activation in follicle cells, it is likely that it has a function in regulating JAK signaling in the ovary.

Another distinction we noted between the Drosophila SOCS was in their abilities to regulate signal transduction cascades in addition to JAK/STAT. Precedence for such additional roles for vertebrate SOCS include regulation of Tec, Vav, TCR, c-kit, and FAK mediated signaling 5657585960. It has been previously shown that Socs36E can suppress signaling not only through the JAK pathway, but also through the EGFR/MAPK pathway 28. Socs44A was also able to regulate EGFR/MAPK signaling, but acted in the opposite manner. Socs44A was able to rescue misexpression of the EGFR negative regulator argos in a dose-dependent manner. Furthermore, mutations in EGFR pathway components rescued Socs44A misexpression phenotypes. Importantly, a reduction of endogenous Socs44A activity enhanced the argos phenotype. Taken together, these data suggest that a normal function for Socs44A is to enhance the EGFR pathway. A potential mechanism for this genetic interaction can be found in a recent report describing physical interaction between SOCS3 and the p120 RasGAP 61. p120 RasGAP, a GTPase-Activating Protein, is an antagonist of MAPK signaling that is responsible for inactivating Ras. It does so by stimulating Ras GTP hydrolytic activity, leaving Ras in a GDP-bound, inactive configuration. Upon interaction with SOCS3, p120 RasGAP is unable to inactivate Ras, resulting in an upregulation of the EGFR/MAPK pathway. Perhaps Socs44A is acting in an analogous manner. Indeed, there are three candidate RasGAP genes in the fly genome. Biochemical analyses will be required to address this hypothesis.

Putative Drosophila SOCS genes were identified using a simple tBLASTn 2.0 query with a consensus sequence for the vertebrate SOCS domains 30 used to probe the complete genome contig sequences available from the BDGP. Identified homologies were compared with the predicted gene structures reported as "CG" sequences in the annotations of the genomic contigs. The translated sequences of the three putative SOCS gene genomic regions were scanned manually for possible alternative structures. The sequences surrounding the SOCS and SH2 domains were used to generate primers for the amplification of DNA corresponding to each putative gene. Amplification products were cloned and used to generate probes for the identification of cDNAs as described below. Phylogenetic comparison of SOCS proteins was performed using AlignX (VectorNTI 9.0), based on the ClustalW algorithm, to generate protein alignments and a neighbor-joining algorithm to create a phylogenetic tree.

A cDNA library constructed from RNA of 12–24 hr old embryos 62 was screened using 800bp of genomic DNA derived from the 3' end of the Socs36E coding region, including the SOCS box and SH2 domain. Two independent clones (Genbank accessions AF435838 and AF435839) were recovered, with the former being structurally similar to an EST from the BDGP (clot #7147). The BDGP also recovered two cDNA clones representing socs44A which have been designated as clot #8463. We have determined the complete sequence of the longer clone, LP02169 (Genbank AF435923).

In situ hybridizations to embryos were performed as previously described 32. Digoxigenin labeled probes for Socs36E and Socs44A were generated from the 5' ends of the respective cDNAs and did not include the coding region for the conserved SH2 and SOCS domains. Germline clone mutants for the hop^c111 null allele were generated using the ovo^D1 dominant female sterile technique 34. Embryos derived from mutant mothers were collected overnight and prepared for hybridizations as previously indicated. Embryos misexpressing upd in a specific pattern were generated by crossing females carrying a UAS-upd transgene with males heterozygous for paired-GAL4, which expresses GAL4 in the seven stripe pair-rule pattern of the paired gene. Progeny were collected and hybridized as above. Trachea in germline clone-derived hop^c111 embryos were visualized with the trh¹⁰⁵¹² enhancer trap 63 using anti-β-gal antibody (Cortex Biochemical, at 1:1000) as previously described 26.

To express Socs36E and Socs44A under control of GAL4, the full-length cDNAs described above were cloned into the pUAST vector 64. Germline transformations were performed 65 and transgenic lines established. For wing phenotypes, engrailed-GAL4 (e16E-GAL) was used to drive expression of the transgenes in the posterior compartment. Wings were dissected and mounted in Hoyer's medium 66 for photography.

Ovarian clones of the null allele, hop^c111, were generated by hsFLP mediated mitotic recombination as previously described 2633. Misexpression clones of Socs36E and Socs44A were generated using a GAL4 flip-out cassette 67. Genotypes of those animals were w [hsFLP]1; [Act5C>y>GAL4] [UAS-GFP.S65T]/ [UAS-socs36E]11.2 and w [hsFLP]1; [Act5C>y>GAL4] [UAS-GFP.S65T]/+; [UAS-socs44A]/ pnt-LacZ, respectively. For each, ovaries were fixed and stained with anti-β gal and anti-GFP as previously described 2633.

All in situ hybridization and wing images were acquired using a Spot Camera (Diagnostic Instruments) on a Nikon E800 microscope using differential interference contrast (DIC). A Leica TCS-SP laser scanning confocal microscope was used to capture all fluorescence micrographs. All images were then exported to Adobe Photoshop for manipulation and annotation.