Genetic Analysis of Slipper/Mixed Lineage Kinase Reveals Requirements in Multiple Jun-N-Terminal Kinase-Dependent Morphogenetic Events During Drosophila Development
- Stephanie Polaski,
- Lisa Whitney,
- Barbara White Barker and
- Beth Stronach 1
- 1Corresponding author: Department of Biological Sciences, University of Pittsburgh, 202 Life Sciences Annex, 4249 Fifth Ave., Pittsburgh, PA 15260. E-mail: stronach{at}pitt.edu
Abstract
Mixed lineage kinases (MLKs) function as Jun-N-terminal kinase (JNK) kinase kinases to transduce extracellular signals during development and homeostasis in adults. slipper (slpr), which encodes the Drosophila homolog of mammalian MLKs, has previously been implicated in activation of the JNK pathway during embryonic dorsal epidermal closure. To further define the specific functions of SLPR, we analyzed the phenotypic consequences of slpr loss and gain of function throughout development, using a semiviable maternal-effect allele and wild-type or dominant-negative transgenes. From these analyses we confirm that failure of dorsal closure is the null phenotype in slpr germline clones. In addition, there is a functional maternal contribution, which can suffice for embryogenesis in the zygotic null mutant, but rarely suffices for pupal metamorphosis, revealing later functions for slpr as the maternal contribution is depleted. Zygotic null mutants that eclose as adults display an array of morphological defects, many of which are shared by hep mutant animals, deficient in the JNK kinase (JNKK/MKK7) substrate for SLPR, suggesting that the defects observed in slpr mutants primarily reflect loss of hep-dependent JNK activation. Consistent with this, the maternal slpr contribution is sensitive to the dosage of positive and negative JNK pathway regulators, which attenuate or potentiate SLPR-dependent signaling in development. Although SLPR and TAK1, another JNKKK family member, are differentially used in dorsal closure and TNF/Eiger-stimulated apoptosis, respectively, a Tak1 mutant shows dominant genetic interactions with slpr, suggesting potential redundant or combinatorial functions. Finally, we demonstrate that SLPR overexpression can induce ectopic JNK signaling and that the SLPR protein is enriched at the epithelial cell cortex.
CELLULAR protein kinases are key mediators of information transfer within eukaryotic cells in response to a variety of signals. The Jun N-terminal kinase (JNK) pathway, for example, consists of a hierarchy of sequential kinase reactions, culminating in activation of transcription factors capable of modulating gene expression and cell behavior (Davis 2000). JNK signaling can be activated in response to many different stimuli, including polypeptide signals like cytokines, environmental stress signals like heat and osmotic shock, bioactive lipid mediators like ceramide, and others. The resulting output from such signals includes cell proliferation, cell-shape change, and cell death. Consequently, JNK-signaling activity is widespread in organisms and modulates the balance between cell growth and death during development, inflammation, and cancer (Manning and Davis 2003). Major questions remain to be solved regarding the organization and regulation of this essential signaling system. For instance, how is the JNK pathway activated in response to different inputs and then how are different outputs achieved?
Previously, we identified Drosophila slipper (slpr), encoding a mixed lineage kinase (MLK), as an upstream kinase stimulating the JNK pathway controlling the morphogenetic process of dorsal closure. Dorsal closure occurs during midembryogenesis with the movement of the dorsal ectoderm on each side of the embryo toward the dorsal midline to seal the embryo in a seamless, protective epidermis (Kiehart et al. 2000; Harden 2002; Jacinto et al. 2002). Molecular markers reporting JNK-signaling activity are present at the leading edge of the advancing epithelium and mutations removing the function of JNK pathway components result in failure of closure (Glise et al. 1995; Riesgo-Escovar et al. 1996; Sluss et al. 1996; Hou et al. 1997; Kockel et al. 1997; Stronach and Perrimon 2002). The phenotype of such mutants, including slpr, is visualized as a large, dorsal anterior hole in the secreted larval cuticle. Loss of JNK signaling disrupts closure by disrupting both gene expression and the organization of the cytoskeleton (Jasper et al. 2001; Kaltschmidt et al. 2002; Xia and Karin 2004). slpr mutants fail to develop properly because of a failure to activate JNK signaling (Stronach and Perrimon 2002).
SLPR is the Drosophila homolog of mammalian MLKs. MLKs belong to the MAPKKK family capable of activating the JNK and p38 MAPKs through canonical intermediary JNKK/MKK kinases (Gallo and Johnson 2002). Specifically, the MKK7 homolog in flies, encoded by the hemipterous (hep) gene, is proposed to be the substrate for SLPR (Sathyanarayana et al. 2003). However, unconventional MLK substrates may include transcription factors and regulators of endocytosis (Akbarzadeh et al. 2002; Marcora et al. 2003; Cha et al. 2004). On the basis of sequence homology, seven different mammalian MLKs have been identified, clustering into three subfamilies: the core MLKs (MLK1–4), the dual leucine zipper (LZ) kinases, and the zipper sterile-α-motif kinase (see Gallo and Johnson 2002 for a recent review). All family members activate the JNK pathway when overexpressed in cultured cells and it is not yet clear what endogenous activators or substrates they have, or to what extent they play differential roles in vivo (Rana et al. 1996; Tibbles et al. 1996; Hirai et al. 1997; Merritt et al. 1999).
Currently, MLK3 is the only MLK family member to be characterized genetically in mammals. While MLK3 gene knockout mice are viable, they display a mild dorsal midline discontinuity in the epidermis (Brancho et al. 2005), suggesting that MLK3 plays a role, similar to that of SLPR, in epithelial tissue closure. Moreover, among many activating signals tested, murine MLK3 knockout cells in culture were mildly impaired only in their ability to transduce signals from tumor necrosis factor α (TNFα) to JNK (Brancho et al. 2005). It remains to be determined whether TNF is the main physiological agonist for MLK3 in vivo and whether other MLKs are partially redundant with MLK3 in animal development.
Structurally, MLKs, including SLPR, contain an N-terminal SH3 domain, a kinase domain, a LZ region, and a Cdc42/Rac interacting binding (CRIB) domain that serves as a site of interaction with GTPases of the Rho/Rac family (Burbelo et al. 1995; Teramoto et al. 1996; Bock et al. 2000). Full activation of MLK proteins is a multistep process, which involves relief of inhibition, dimerization, and autophosphorylation (Leung and Lassam 1998; Bock et al. 2000; Vacratsis and Gallo 2000; Zhang and Gallo 2001). SLPR is most closely related to the human MLK core subfamily with up to 56% aa identity in the N-terminal half of the protein. The high degree of sequence conservation of SLPR makes it an excellent representative MLK family member for genetic analysis to shed light on the function of the MLK family in vertebrates.
SLPR/MLKs are related to a larger family of JNKKK proteins, six altogether in Drosophila, with the potential to activate JNK (Stronach 2005 and references therein). To date, biochemical and cell culture assays, as well as phenotypic data from mutant animals, provide clear evidence that at least two members, SLPR and TAK1, do activate the common substrate HEP/MKK7, but appear to do so in response to distinct signals (Chen et al. 2002; Sathyanarayana et al. 2003; Silverman et al. 2003). To investigate further the selective role of SLPR throughout development and in homeostasis, we took three approaches. First, a genetic and developmental analysis with slpr alleles uncovered numerous morphogenetic roles for SLPR, which also require JNK signaling, implicating SLPR in JNK activation in multiple tissues and throughout development. Second, an antibody was generated to characterize endogenous SLPR protein in normal and mutant animals. A missense mutation in the kinase domain results in stable expression of a dominant interfering protein while two nonsense alleles result in significant reduction or loss of protein expression. Third, the requirement for SLPR in TNF-dependent apoptosis was tested directly, and SLPR was found to be dispensable for cell death induced by TNF/Eiger overexpression.
MATERIALS AND METHODS
Drosophila stocks:
Unless otherwise noted, all stocks were obtained from the Bloomington Stock Center and are described in FlyBase (Drysdale and Crosby 2005).
Mutants obtained were:
slpr921 and slpr3P5 (Stronach and Perrimon 2002);
Rac1J10, Rac2Δ, MtlΔ (Hakeda-Suzuki et al. 2002);
Rho172O (Strutt et al. 1997);
msn102 (Treisman et al. 1997);
msnj1E2 (Spradling et al. 1999);
hep1 (Glise et al. 1995);
hep699 [a pupal lethal mutation obtained in a mutagenesis screen for embryonic phenotypes from germline clones (Chou and Perrimon 1996)];
bsk1 (Sluss et al. 1996);
junIA109, kay1 (Riesgo-Escovar and Hafen 1997a);
pucE69 (puc-LacZ) (Ring and Martinez Arias 1993);
raw1, raw02418 (Byars et al. 1999);
Tak12527, identified in an imd enhancer screen, introduces a premature stop codon (W178stop) into the activation loop of the kinase domain of Tak1 and behaves as an amorph (D. Schneider, personal communication).
Gal4 lines obtained were:
arm-GAL4 (P{w[+mW.hs] = GAL4-arm.S}11);
Act5C-GAL4 (P{w[+mC] = Act5C-GAL4}17bFO1);
pnr-GAL4 (P{w[+mW.hs] = GawB}pnr[MD237]);
GMR-GAL4 (P{w[+mC] = GAL4-ninaE.GMR}12).
Transgenes:
p{w+ slpr-gen}, a genomic rescue construct for slpr, has been described (Stronach and Perrimon 2002). UAS-slprWT was pieced together from two EST clones. The full-length GH26507 EST contains a point mutation in kinase subdomain IX, replacing invariant amino acid D314 with Y, and is thus nonfunctional in genetic rescue assays. The AT02557 EST is a partial cDNA, lacking 5′ coding sequences, including the translation start codon, but it does contain a wild-type kinase domain and all C-terminal sequences. Thus a wild-type coding region was created as follows. From the GH26507 EST, an ∼3740-bp XmnI restriction fragment was cloned into the SmaI site of pBluescriptKS to create pBSX3. Thus, hybrid SmaI/XmnI sites define the endpoints of the recombinant clone, excluding ∼690 nt of 5′UTR and ∼700 nt of 3′UTR sequences, but including ∼76 nt of 5′UTR and 217 nt of 3′UTR flanking the complete coding region. pBSX3 was completely digested with SbfI and partially digested with SalI to yield a 5621-bp fragment, including the vector backbone and 5′ sequences of slpr. To this backbone, a 1080-bp replacement SbfI–SalI wild-type coding fragment isolated from the AT02557 EST was ligated to generate pBSX3WT. The recombinant clone was then fully sequenced to ensure wild-type coding capacity. Finally, the insert was transferred into the transformation vector pUASp as a KpnI–NotI fragment. UAS-slprKD is represented by two different types of transgenes, which behave similarly in all assays that we have used. One is derived from the XmnI fragment of GH26507 as described above (from pBSX3), containing the D314Y mutation. The second construct has the same D314Y mutation but, in addition, was engineered by site-directed mutagenesis to replace the lysine (AAG) in the VAIK phospho-transfer motif of kinase subdomain II with methionine (ATG) (Hanks et al. 1988). Transgenic lines were generated according to standard procedures (Spradling 1986).
regg1 is essentially UAS-egr and UAS-egr-IR is an inverted repeat construct to knockdown egr gene product in vivo (Igaki et al. 2002). UAS-takKD contains a mutation K46R and has been described previously (Mihaly et al. 2001).
Isolation of slprBS06 allele:
Mutagenesis was performed by feeding w− males 25 mm ethyl methanesulfonate (Grigliatti 1998). Mutagenized males were crossed to y w f C(1)DX/Y; p{w+, slpr-gen} females. The attached-X in females causes segregation of the paternal X to sons. Approximately 1000 F1 w */Y; p{w+, slpr-gen}/+ males with w+ eyes were backcrossed to y w f C(1)DX/Y females without the slpr transgene. F2 progeny were scored to recover vials that produced only red-eyed males, indicative of survival dependent on the presence of the slpr transgene. The mutant X chromosome was then recovered and maintained in females using the FM7 balancer. The molecular nature of the allele was determined by PCR amplification of coding sequences from genomic DNA and sequencing of the PCR products. Multiple independent amplifications were performed as well as sequencing of both DNA strands. One nucleotide change was identified within the coding sequence of slprBS06 (see results).
Clonal analysis:
FRT101 marked with mini-w+ was recombined onto the w, slprBS06 chromosome. Recombinant lines were established and observed for reduced frequency of non-FM7 males. Lines were kept if the rate of mutant male eclosion was restored by the presence of a slpr-rescuing genomic transgene. To generate germline clones, slprBS06 FRT101 recombinant females balanced over FM7 were crossed to w ovoD1 FRT101/Y; hs-flp38 males. Vials with third instar larvae were heat-shocked for 1 hr at 37° 2 days in a row. Non-FM7 mosaic females were collected and mated to either +/Y or slprBS06/Y males and allowed to lay eggs. For follicle cell clones, slpr FRT101/ Ubiq-GFP, FRT101;e22c-GAL4, UAS-flp/+ females were mated to wild-type males. After 2–6 days, eggs were harvested, washed, and mounted for observation of the eggshells by dark-field microscopy. In addition, ovaries from females were dissected, fixed, and processed for immunofluorescence with an anti-SLPR antibody or simply stained with rhodamine–phalloidin to highlight actin. For disc clones, slpr FRT101/FM7iGFP females were crossed to hs-flp, hs-nuclearGFP, FRT101/Y males. In separate experiments, 1 hr of heat shock at 37° was applied to either late first or second instar larvae. Just before dissecting third instar larval discs, larvae were subjected to another 1-hr, 37° heat shock to induce acute expression of GFP. After 1–2 hr recovery, larvae were dissected, and discs were fixed in PBS with 4% formaldehyde for 20 min. Discs were subsequently processed for immunofluorescence with the anti-SLPR antibody using standard procedures.
Antibodies:
A fragment containing the N-terminal 106 aa of SLPR including the SH3 domain was PCR amplified and cloned into the pET16b vector (Novagen) to create HIS-tagged SLPR-SH3. The tagged protein was expressed in Escherichia coli and purified by affinity chromatography using the Ni-NTA spin column according the manufacturer (QIAGEN, Chatsworth, CA). Purified protein was used to immunize rabbits (Covance). Affinity-purified rabbit anti-SLPR serum was used at 1:600 for Western immunoblot with BSA as a blocking agent and 1:400 for immunofluorescence. Mouse anti-β-gal (Promega, Madison, WI) was used at 1:1000. All fluorophore or enzyme-conjugated secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA).
Image acquisition:
Images of adult flies were captured using a Leica DFC300F camera mounted on a Leica MZ16 stereomicroscope. Pictures of embryos and larval discs were obtained by laser confocal scanning (Biorad Radiance 2000) on a Nikon E800 compound microscope. Cuticles and eggshells were observed with dark-field optics on an E800 microscope and photographed with an attached SPOT camera. All images were assembled with Adobe Photoshop.
RESULTS
Molecular characterization of slpr alleles:
Previous characterization of two zygotic slpr mutants, slpr921 and slpr3P5 (Figure 1), revealed that slpr is required for embryonic dorsal closure (Stronach and Perrimon 2002). slpr921 zygotic mutants are embryonic lethal with a severe dorsal closure phenotype (Figure 2, cross 1). Germline clones of this allele, producing maternal and zygotic mutant embryos, did not enhance the cuticle phenotype further, as might be expected if slpr participates in other aspects of embryonic development. slpr3P5 is a weaker allele as evidenced by an increased larval hatch rate of the zygotic mutant relative to slpr921 (Figure 2, cross 2). In contrast to slpr921, germline clones of slpr3P5 shifted the lethal phase and cuticle phenotype to resemble the severe defects observed with the slpr921 allele (not shown), indicating that slpr may have a maternal contribution. This is consistent with expression data from in situ hybridization experiments showing that slpr transcripts are present in early embryos prior to the time that zygotic transcription starts (Sathyanarayana et al. 2003). Sequence analysis indicated that both alleles encode point mutations in the catalytic kinase domain, with slpr3P5 predicted to encode a truncated protein (Figure 1). The nature of these alleles suggests that impairment of the kinase function of SLPR results in an embryonic dorsal open phenotype.
Schematic of the SLPR protein indicating conserved protein domains and the molecular nature of the mutant alleles. As shown previously, slpr921 encodes a point mutation in the kinase domain and slpr3P5 encodes a stop codon near the end of the kinase domain. Here, we show that slprBS06 encodes a stop codon truncating the protein after amino acid 47, which likely represents a null allele (see text). Immune antisera were generated to the region of the SLPR protein indicated by the horizontal line. SH3, Src homology 3.
Embryonic lethality and dorsal open phenotype associated with variable reduction in maternal vs. zygotic slpr function. Parental crosses (1–11) are indicated along the x-axis with maternal genotype first and paternal genotype second. Embryonic lethality was measured as the number of unhatched larvae/total number of larvae and expressed as a percentage (shaded bars). The total number of larvae scored for each cross is given above the shaded bars. Among the unhatched larvae, cuticles were recovered and scored for the dorsal open phenotype (solid bars), indicated by a large dorsal hole in the cuticle regardless of whether the anterior head structures were normal or everted. The data indicate that slprBS06/Y zygotes (3) survived embryogenesis much better than slpr921/Y (1) or slpr3P5/Y (2) zygotes; but further reducing the maternal slpr contribution to zygotes derived from null slprBS06 homozygous females dramatically increased the lethality and dorsal open phenotype (4 and 5), unless rescued by the wild-type paternal chromosome zygotically (4). Because of a functional maternal contribution of slpr gene product, zygotes expressing dominant-negative slpr (slprKD) or Tak1 (takKD) had minimal effect on embryonic lethality and dorsal closure (7 and 8); yet, sensitizing the maternal background with slprBS06/+ heterozygous mothers significantly increased the potency of dominant-negative kinase expression zygotically (10 and 11). Notably, slprKD was much more effective at blocking dorsal closure in this background (10) than takKD (11), while armG4 alone had a negligible effect (9). Reduction in Tak1 function either by expression of a dominant-negative transgene (11) or in a double-mutant combination (6) does exacerbate the embryonic lethality and dorsal defects associated with reduced slpr function, suggesting that there may be a limited role for TAK1 in the embryo when JNK signaling is compromised. The severe effect of slpr921 or slpr3P5 zygotically (1 and 2) is likely due to the antimorphic nature of the alleles encoding proteins that dominantly interfere with residual slpr maternal product in the zygote.
By screening for additional slpr alleles that survive in the presence of a rescuing transgene, rather than for mutations with a dorsal closure phenotype per se, we sought to recover alleles that would facilitate further developmental analysis and insight into the functions of other regions of the SLPR protein. Among ∼1000 F1 crosses (see materials and methods), a new allele of slpr has been recovered and is designated slprBS06. slprBS06 is a semiviable, maternal-effect mutation. Sequence analysis shows that BS06 encodes a transition mutation, changing the UGG tryptophan48 codon to a UAG amber codon, causing a premature stop after amino acid 47, just before the SH3 domain (Figure 1). The molecular lesion, coupled with rescue by a slpr transgene, confirms that BS06 is a bona fide slpr allele. The predicted mutant protein would not retain any conserved domains and is likely to represent a null allele of slpr.
slprBS06/Y zygotic mutants have a low rate of embryonic lethality (Figure 2, cross 3) and a variable lethal phase. A proportion die during the pupal stage, but some adult males eclose at ∼10% of the expected frequency (Table 1); for example, on average one slprBS06/Y male ecloses for every 10 slpr+/Y brothers. The eclosion frequency is normalized by the presence of the genomic rescuing transgene. Although the emergence and fertility of slprBS06 males is poor, some males are capable of mating to produce homozygous slprBS06 females or heteroallelic females. The results of complementation in females are shown in Table 1. slprBS06 poorly complements the lethality associated with slpr3P5 and slpr921 and rare escapers eclose as adults. slpr mutant females show a maternal effect on their progeny; slprBS06 maternal and zygotic mutant embryos laid by either homozygous slprBS06 or heteroallelic females or by making slprBS06 germline clones using the “FLP-ovoD1” dominant female sterile technique (Chou and Perrimon 1996) exhibit a severe dorsal closure phenotype (Figure 2, cross 4 and cross 5), providing additional genetic evidence that BS06 is a slpr allele and that failure of dorsal closure is the null phenotype.
Complementation, survival, and phenotypic defects of slpr alleles
Threshold levels of JNK pathway activity in slprBS06 zygotic mutants:
slprBS06 zygotic mutants are phenotypically weaker, with respect to lethal phase, than the two embryonic lethal alleles (921 and 3P5) that had been characterized previously (Figure 2). To explain how the null slprBS06 allele could behave as a weaker allele, we postulate that zygotic slprBS06 mutants survive embryogenesis due to the presence of functional maternal product in the embryo, whereas the other alleles, slpr3P5 and slpr921, encode antimorphic proteins, which reduce the activity of the remaining maternal contribution in the zygote, resulting in severe defects in dorsal closure and embryonic lethality. Thus, having residual antimorphic protein is worse than no protein at all. This hypothesis makes several predictions. First, reducing the maternal dosage of slpr coupled with expression of dominant-negative transgenic protein should increase the percentage of embryonic lethality and dorsal open phenotype. Second, mutant SLPR protein should be present in the slpr921 and slpr3P5 mutants, not in the slprBS06 mutant. Third, the maternal protein in slprBS06 zygotic mutants should be sensitive to levels of other JNK pathway regulators.
To address the first prediction, we compared expression of the UAS-slprKD transgene in an otherwise wild-type background to the expression of the transgene in embryos from heterozygous slprBS06/+ mothers, which would contribute 50% of the normal dose of slpr gene product. Indeed, we find that expression of slprKD throughout the wild-type embryo using armadillo-GAL4 (armG4) results in very mild embryonic lethality and a low percentage of dorsal open embryos (Figure 2, cross 7). slprBS06 zygotic mutant embryos, carrying the Gal4 driver alone, do not display a dorsal open phenotype (Figure 2, cross 9). However, further reduction in the maternal function coupling the slprBS06 allele, with ubiquitous zygotic expression of slprKD, significantly increases the percentage of embryonic lethality and severity of cuticle defects (Figure 2, cross 10), consistent with the prediction. Moreover, the slpr921,3P5 alleles are likely to encode antimorphic proteins like slprKD, whose presence in the mother and zygote poison functional maternal protein, leading to a more severe phenotype than that produced by slprBS06 alone. Their antimorphic character is likely explained by the nature of the alleles that are predicted to encode proteins with residual protein-binding interaction motifs.
To assess whether SLPR protein is stably made from the mutant alleles, we generated an antibody to the SLPR SH3 domain and used this reagent to visualize SLPR immunoreactivity in tissues with slpr mutant clones (Figure 3A). As expected for a null allele, we observed loss of protein immunoreactivity in slprBS06 mutant clones in the wing imaginal disc compared with neighboring heterozygous tissue, marked by the presence of GFP (Figure 3A, top). Western immunoblotting (Figure 3B) confirms that there is no full-length SLPR protein present in lysates made from slprBS06 mutant adult flies, while protein is clearly detected in lysates from heterozygotes or mutant males carrying a genomic rescue transgene. In contrast, clones of slpr921 mutant tissue show normal SLPR protein staining relative to nonmutant tissue (Figure 3A, middle), demonstrating that the mutant protein is stably expressed. slpr3P5 mutant clones show significant reduction in protein immunoreactivity (Figure 3A, bottom), suggesting that the truncated mutant protein is poorly expressed, unstable, or less antigenic, relative to wild type. However, some slpr3P5 mutant protein must be produced to explain the increased zygotic embryonic lethality relative to the null allele, which indicates that slpr3P5 has some dominant “interfering” capability. Thus, the absence of protein in slprBS06 mutant tissue supports the idea that BS06 is a null allele and provides an explanation for why slprBS06 zygotic mutants may survive better than their counterparts who inherit residual mutant protein, which interferes with the reduced pool of maternal SLPR activity. It is important to note, however, that all alleles are fully rescued by a genomic transgene indicating not only that there are no second-site lethal mutations in the background, but also that the balance of SLPR activity can be restored ectopically by providing an additional wild-type allele.
SLPR protein expression in slpr mutants assessed by immunofluorescence analysis in tissue clones and by Western immunoblot. (A) Clonal analysis of slpr alleles in larval wing imaginal discs. Homozygous mutant tissue is indicated by lack of GFP fluorescence (left column, and green in right column). SLPR immunoreactivity is shown (middle column, and red in right column). Yellow outlines a clone of mutant tissue for each allele. Note that there is a loss of SLPR staining in slprBS06 mutant clones (top), consistent with slprBS06 encoding a null allele. In contrast, SLPR protein encoded by slpr921 is clearly detected in the clone (middle), indicating that the mutant protein is stably expressed. Antibody staining is significantly diminished in slpr3P5 mutant tissue (bottom). These results confirm the specificity of the antibody in vivo. (B) Western immunoblot of adult protein extracts from the genotypes indicated, using SLPR antisera. Full-length SLPR migrates near 150 kDa. The ∼80-kDa band is nonspecific and serves as an internal loading control.
If slprBS06 zygotic mutants depend on wild-type maternal slpr contribution, then modulating the dosage of gene products that attenuate or potentiate the maternal SLPR function might worsen or improve slprBS06 zygotic phenotypes. Thus, although slprBS06 is a null allele, it may provide a sensitized background when zygotically mutant because the dwindling supply of maternal gene product places the mutant animals at a threshold for viability, exemplified by the rate of slprBS06/Y eclosion at 10% of expected. Reducing the dosage of several positively acting components of the embryonic JNK pathway, Rac, bsk, and kay did worsen (or enhance) the rate of slprBS06/Y lethality (Figure 4), while Rho1, which is thought to play role in dorsal closure largely independent of the JNK pathway, did not (Lu and Settleman 1999; Magie et al. 1999). However, reduction of jun did not have a significant effect on slprBS06 lethality, although only the junIA109 allele was tested. That only one of three alleles of msn tested showed a dominant genetic interaction with slprBS06 underscores potential allele-specific differences among the interactions with slpr. Alternatively, reduction of JUN function may not, like the other transducers, be limiting for development in a slprBS06 zygotic mutant background.
Dominant effects of JNK pathway regulators on slprBS06/Y male survival. slprBS06/FM7 females were crossed to w or yw males to determine the average rate of slprBS06/Y adult male eclosion relative to FM7/Y eclosion, which is expressed as a percentage (shaded bars). In these controls, slpr mutant males emerge at an average frequency of ∼10% of expected, relative to the balancer class. The total number of males scored, combined from each class, is indicated for each mutant tested. To test whether loss of JNK pathway regulators dominantly modifies the rate of slprBS06 male eclosion, we crossed slprBS06/FM7 females to heterozygous males from the mutant stocks indicated below the x-axis. Numbers of adult male progeny were counted and compared as above (excluding the autosomal balancer classes), e.g., slprBS06/Y;mutant/+ vs. FM7/Y;mutant/+ (solid bars). Reduced dosage of many positively acting JNK pathway components (e.g., bsk, kay) decreased the rate of slpr mutant male recovery, while halving the dosage of negative regulators of JNK signaling (e.g., raw, puc) significantly improved the rate of slprBS06 male eclosion to near the expected frequency.
Halving the dose of two known negative regulators of JNK signaling in the embryo, puc and raw (Martin-Blanco et al. 1998; Byars et al. 1999), significantly improves (or suppresses) the lethality of slprBS06/Y zygotic mutant males to near the expected frequency of survival. Another negatively acting component of the JNK pathway during dorsal closure, the ETS domain transcriptional repressor aop (Riesgo-Escovar and Hafen 1997b), had a moderate effect in boosting slprBS06 male eclosion. Taken together, we conclude that the maternal supply of SLPR gene product is functional for dorsal closure in the absence of zygotic activity and can be potentiated to promote adult survival by twofold reduction in components that downregulate JNK-signaling pathway activity.
Zygotic phenotypes of the slpr maternal-effect allele:
Adult slprBS06 mutant males survive at a frequency averaging 10% and display an array of visible morphological defects. The defects are incompletely penetrant but include crumpled wings (42%) (Figure 5, A and A′), genital rotation, and eversion defects (37%) (Figure 5, D, D′, and D″), dorsal abdominal clefts or segmental fusions (30%) (Figure 5, B and B′), missing or deformed maxillary palps (35%) (Figure 5, E and E′), and rarely, cleft nota (2%) (Figure 5, C and C′). The underlying cause of these defects at the tissue and molecular level is still unknown but several of the phenotypes have been previously linked to JNK signaling. First, recent work demonstrates a link between puckered (JNK phosphatase) levels and proper male terminalia development (Macias et al. 2004; McEwen and Peifer 2005). Second, thorax closure is mediated in part by JNK activity such that reduced signaling can result in cleft nota (Agnes et al. 1999; Zeitlinger and Bohmann 1999).
Zygotic phenotypes observed in slprBS06 mutant males. Comparison of the morphology of wings (A), abdomen (B), nota (C), terminalia (D), and palps (E) between slprBS06/Y mutants plus (A–E) or minus (A′–E′) a genomic rescue transgene. (A and A′) In comparison to the normal flat expanded morphology of the wing, mutant wings appear crumpled, as if the wing failed to inflate and expand properly. (B and B′) The abdominal dorsal midline is disfigured in the mutant with segmental gaps and misalignments very rarely seen in wild-type adult males. (C and C′) This example shows morphological defects of the left heminota in the mutant, resulting in a shallow, anterior-biased cleft, compared to wild type. (D, D′, and D″) Male terminalia are shown with wild-type orientation of the penis–anus axis indicated by the red arrowhead in D. Two different defects are observed in slpr mutants; a failure to evert the terminalia altogether (D′) or a misorientation of the penis–anus axis of the terminalia (D″), where the arrowhead indicates a 90° misrotation in this example. (E and E′) Compared to the wild-type pair of maxillary palp appendages (asterisks in E), slpr mutant palps are either missing (arrow in E′) or deformed (arrowhead in E′).
Many of these defects are also observed with varying penetrance in females with compromised SLPR function, with the maxillary palp and terminalia defects being most prevalent (Table 1 and Figure 6, C, F, and G). In females with terminalia defects, abdomens appear truncated at tergite 7 or 8, with reduction of the gonopod and loss of (or failure to evert) the proctodeal structures forming the analia (Figure 6G; compare to heterozygous hep1/FM7, Figure 6D, for wild-type morphology). Females with truncated terminalia are essentially sterile because they are unable to mate or lay eggs; however, other females, which do not show terminal defects, are capable of mating. Notably, eggs laid by slprBS06 homozygous or heteroallelic mutant females show dorsal appendage defects of the eggshell, reflecting a requirement for slpr in the somatic follicle cells during oogenesis (Figure 6, H and J). Similarly, thin, shortened appendages were also observed among eggs collected from females with slpr921 mutant follicle cell clones (Figure 6I). Dorsal appendage formation has also been previously shown to require JNK signaling (Suzanne et al. 2001).
Morphogenetic defects associated with loss of slpr or hep signaling in adult females or in somatically derived eggshell appendages. (A–C) Adult heads. Similar to the phenotype in slprBS06 mutant males, unilateral loss of a maxillary palp is observed in a heteroallelic combination of hep mutations (B) compared to hep heterozygotes (A). Expression of dominant-negative slpr using the Act5C-GAL4 driver also phenocopies the palp defect (C). In (A–C), asterisks mark normal palps and arrows point to the region missing a palp. (D–G) Lateral view of terminalia of the adult female. (D) hep heterozygotes. Note wild-type gonopod and proctodeal structures (red asterisks) and the position of abdominal tergite 7 (white asterisk). (E–G) Loss of JNK signaling correlates with truncation of structures posterior to tergite 7. Terminalia defects in females mutant for hep (E) and slpr (G) or in females expressing dominant-negative slpr (F). These females are unable to defecate or lay eggs. Some slpr mutant females develop normal terminalia and are capable of mating and laying eggs. (H–J) Dark-field images of eggshells; dorsal view with anterior to the left, except J is the lateral view. (H) Eggs laid by wild-type mothers have a pair of paddle-shaped, extended dorsal appendages (arrowhead). (J) Eggs laid by slprBS06 homozygous females have defective, stunted dorsal appendages. (I) Thin, shortened appendages are also observed among eggs in which slpr921 mutant follicle-cell clones have been generated.
As shown in Table 1 and in Figure 6, zygotic expression of an engineered kinase-dead form of SLPR (UAS-slprKD) using the actin5C-Gal4 driver phenocopies the palp and terminalia defects observed in slpr mutant females, indicating dominant-negative activity of the transgenic construct and reinforcing the observed requirement for SLPR function in the development of those tissues (Table 1 and Figure 6, C and F). Moreover, these same defects can be observed in hep (JNKK/MKK7) mutant adults (Figure 6, B and E), compared to hep1/FM7 heterozygotes resembling wild type (Figure 6, A and D), although they had not been previously described. Importantly, these data indicate that the morphogenetic defects observed in slprBS06 zygotic mutant animals are due to loss of SLPR-mediated activation of HEP and JNK signaling. Together, the collection of malformed tissues observed at various developmental stages in slpr mutants suggests a common defect in tissue closure, elongation, or eversion.
Wild-type SLPR overexpression activates JNK signaling:
Loss of slpr function results in failure to activate the JNK pathway to mediate various morphogenetic events. To test whether SLPR can ectopically activate JNK signaling in vivo, a cDNA-based UAS-slprWT transgene was generated. To confirm the activity of the transgene, we tested its ability to rescue the embryonic dorsal closure defects of slpr921 zygotic mutants by ubiquitous expression using the armG4 driver. slpr921 zygotic mutants are embryonic lethal, dying at the expected 25% Mendelian frequency, so we first assessed rescue by monitoring embryonic survival rather than survival to adulthood. Indeed, we found that slpr921 embryonic lethality is reduced from ∼23 to 9% with transgene expression, and among the unhatched larvae, the severe dorsal open cuticle phenotype is reduced from >90% without the transgene to 2% with the rescuing transgene (not shown). Moreover, expression of the UAS-slprWT transgene at 18° rescues slpr921, slprBS06, and slpr3P5 mutant males to adulthood, demonstrating the ability of ubiquitous slpr cDNA expression to fully complement.
In contrast, slpr overexpression in an otherwise wild-type background can cause embryonic lethality, depending on the temperature and the transgenic line, which is likely due to ectopic activation of JNK signaling. Consistent with this, we observed cuticles with a puckered dorsal midline, reminiscent of puc mutant embryos in which JNK signaling is hyperactive (Figure 7, D and E). Indeed, programmed expression of UAS-slprWT in the dorsal ectoderm of a wild-type embryo with the strong pnr-GAL4 driver (Calleja et al. 1996) results in ectopic upregulation of puc-LacZ, a reporter of JNK pathway activity (Glise and Noselli 1997; Martin-Blanco et al. 1998), extending away from the leading edge but limited to a subset of the cells within the pnr domain (Figure 7, A and B). Expression of transgenic SLPR protein within the pnr domain is confirmed by immunofluorescence using SLPR-SH3 antisera (Figure 7C). For unknown reasons, endogenous SLPR protein is difficult to detect in the embryonic ectoderm. In contrast, transgenic slpr overexpression produces abundant, easily detectable protein. Notably, exogenous SLPR protein is enriched at the cell cortex of ectodermal epithelia (Figure 7C). Cortical localization is not dependent on kinase activity because catalytically inactive transgenic SLPR protein shows a similar distribution (not shown). In addition, enrichment of SLPR at the membrane does not appear to correlate with the activation state of the JNK pathway in the cell. For example, cells with JNK signaling below the threshold for induction of puc-LacZ expression still display cortical localization of transgenic SLPR protein (Figure 7C). The functional significance of this cortical localization, while intriguing, is still unknown. These expression data suggest that SLPR is limiting in JNK activation under normal developmental conditions. Under conditions of potent overexpression, SLPR is capable of only moderate JNK pathway upregulation, suggesting that additional components then become limiting; otherwise, puc-LacZ expression would be expected to encompass the entire pnr expression domain.
slpr overexpression induces ectopic JNK signaling in the embryo. (A–C) Immunofluorescent staining of stage 13/14 fixed embryos, dorsolateral view. puc-LacZ (encoded by pucE69) is a reporter of JNK activity at the leading edge of the dorsal ectoderm (A). Programmed overexpression of a wild-type slpr transgene using pnr-GAL4 (pnr-G4) results in ectopic upregulation of puc-LacZ reporter expression ventral to the leading edge (B) but within the pnr expression domain, indicated by immunofluorescent localization of transgenic SLPR protein (red staining in C). Note that SLPR protein is cortically localized in the ectodermal epithelia in comparison to the nuclear localization of β-gal expressed from the reporter (green staining in C). (D and E) Dark-field images of larval cuticles, dorsal view. Consistent with ectopic induction of JNK-signaling activity by SLPR, ubiquitous expression of wild-type slpr in the embryo results in a larval cuticle with a puckered dorsal midline (arrowhead, D), very similar to the cuticle phenotype of larvae mutant for the puckered phosphatase (E) in which JNK signaling is overactive.
The sensitivity of cells to SLPR-dependent JNK signaling is also revealed during the process of adult thorax morphogenesis (Figure 8). Some animals with weak ectopic upregulation of JNK activity, mediated by SLPR overexpression in the embryonic dorsal ectoderm, progress through development and eclose as adults with morphological defects in the thorax. These animals, pnr-GAL4 > UAS-slprWT, show a narrowing of the scutellum compared to wild type (Figure 8, A–C), contrasting with a cleft notum phenotype, which we observed upon expression of UAS-slprKD in the pnr domain (Figure 8D), indicative of JNK-pathway loss of function (Agnes et al. 1999; Zeitlinger and Bohmann 1999). Consistent with these phenotypes being due to altered JNK signaling, halving the dosage of puc phosphatase enhances the gain-of-function notum phenotype, essentially eliminating the scutellum altogether (Figure 8E), and suppresses the loss-of-function notum phenotype to near wild type (Figure 8F). Consequently, by transgenic SLPR expression and slprBS06 mutant analysis, we demonstrate that SLPR-mediated JNK signaling is responsible for shaping the adult thorax.
slpr expression in the pnr domain affects the morphology of the adult thorax. (A–F) Adult nota: dorsal views, anterior oriented to the top. Compared to the normal adult notum (A), overexpression of wild-type slpr (UASslprWT) narrows the scutellum (arrowhead, C). Notum defects are further enhanced by reducing the dosage of puc phosphatase (arrowhead, E). Dominant-negative slpr (UASslprKD) causes a cleft notum (arrowhead, D), which is dominantly suppressed by halving the dose of puc (arrowhead, F).
Specificity of SLPR and TAK1:
Previous genetic studies have suggested the prominent role of SLPR rather than TAK1 in regulating JNK signaling during dorsal closure, even though both are maternally contributed to the embryo and are known to share the common substrate, HEP/MKK7 (Mihaly et al. 2001; Vidal et al. 2001; Stronach and Perrimon 2002; Sathyanarayana et al. 2003). Tak1 null mutants are viable and show no maternal effect on dorsal closure (Vidal et al. 2001); nevertheless, the extant slpr mutants were not unequivocally null and could have dominantly interfered with TAK1 or other putative JNKKK family members functioning during dorsal closure. In this study, the null phenotype of slprBS06 germline clones allows us to conclude that SLPR is the primary activator of JNK signaling during closure and that endogenous TAK1 or another kinase cannot compensate to direct embryonic dorsal closure in the absence of maternal and zygotic slpr; thus the slpr and Tak1 null phenotypes are distinct. Notably, however, we observe that loss of Tak1 enhances the slprBS06 null phenotype in the double mutant (Figure 2, cross 6), such that no mutant male adults eclose and there is a slight increase in the relative embryonic lethality and dorsal closure defects of doubly mutant embryos. This genetic interaction suggests that additional redundant functions for slpr and Tak1 may exist.
In addition to the null phenotypes, expression of dominant-negative forms of the kinases in two different assays reveals their selective functions. While neither slprKD nor takKD is very effective at blocking dorsal closure in a wild-type embryo, expression of either is more potent in perturbing embryonic development in a sensitized background with reduced maternal SLPR function (Figure 2; compare cross 7 and cross 8 with cross 10 and cross 11). Yet, slprKD is significantly more effective at blocking dorsal closure than takKD under the same conditions. The measurable effect that takKD does have may reflect residual competition for hep, the common substrate between SLPR and TAK1, or may result from downregulation of basal levels of JNK signaling present in the embryo thought to modulate cell survival (McEwen and Peifer 2005).
SLPR does not mediate ectopic Eiger-induced apoptosis:
A second assay that discriminates between SLPR and TAK1 is described here. MLKs have been suggested as physiological mediators of TNF signaling on the basis of several observations. For example, TNF-α stimulation of Jurkat T cells activated MLK3 kinase activity and JNK phosphorylation, which could be blocked by addition of a small-molecule MLK inhibitor (Sathyanarayana et al. 2002). Also, primary fibroblasts derived from MLK3 knockout mice fail to support maximal JNK activation in response to TNF-α stimulation, although they respond normally to other stress signals that induce JNK activation (Brancho et al. 2005). The Drosophila TNF homolog Eiger is expressed dorsally in the early embryo (Stathopoulos et al. 2002; Kauppila et al. 2003), raising the possibility that it could stimulate the JNK pathway in the dorsal ectoderm, perhaps to trigger dorsal closure. Although this is a formal possibility, genetic evidence demonstrating that egr mutants are viable suggests the contrary (Igaki et al. 2002).
To test whether SLPR is required for transducing proapoptotic TNF signals in vivo, we overexpressed Eiger in the larval eye imaginal disc, which effectively ablates the adult eye due to JNK-dependent apoptosis (Figure 9A) (Igaki et al. 2002; Moreno et al. 2002). The phenotype is specific to Eiger overexpression because it is reversed by coexpression of an inverted-repeat Eiger transgene causing RNA interference in vivo (Figure 9B). Neither reducing the dosage of wild-type slpr genetically nor coexpression of a dominant-negative slpr transgene has any effect on Eiger-induced apoptosis in the eye (Figure 9, D and F). This is not due to lack of SLPR expression in the eye because SLPR protein is detected in the larval eye disc and we previously demonstrated the ability of slpr to suppress a rough-eye phenotype induced by overexpression of the wild-type Rac1 GTPase (Stronach and Perrimon 2002). In contrast to slpr, a hep mutation partially suppresses the Eiger phenotype (Figure 9C), suggesting that HEP is required in part to mediate Eiger signaling. Moreover, we found that, unlike SLPR, expression of dominant-negative TAK1 was able to block Eiger-induced apoptosis and restore, almost completely, the normal development of the adult eye (Figure 9E). Taken together, these data implicate TAK1 and not SLPR in TNF-induced, proapoptotic JNK signaling, at least in the context of the imaginal eye disc (see also Geuking et al. 2005). While we cannot rule out from these studies that SLPR may selectively mediate signals from endogenous Eiger, our observations in Drosophila are consistent with studies using murine MLK3−/− primary cells: that while MLK3 contributes to maximal JNK activation in response to TNF, MLK is not required for TNF-stimulated apoptosis (Brancho et al. 2005).
TNF/Eiger-induced apoptosis and loss of eye tissue is mediated by TAK1-, but not SLPR-, dependent JNK signaling. (A–F) Adult eyes: lateral views of the head. (A) Eiger overexpression in the developing eye disc under the control of GMR-GAL4 (GMR-G4) causes JNK-dependent apoptosis and ablation of the eye. regg1 is a GAL4-sensitive transgene insertion in the endogenous eiger locus. (B) The phenotype is reversed by in situ RNA interference of eiger (UAS-egr-IR). (C) Partial suppression by loss of hep (JNKK) indicates that the phenotype requires an intact JNK pathway; however, reducing slpr function by mutation (D) or expression from two copies of a dominant-negative transgene (UAS-slprKD) (F) has no effect on Eiger-induced eye ablation. In contrast, expression of dominant-negative TAK1 (UAS-takKD) rescues the loss of eye tissue almost completely (E).
DISCUSSION
Previous genetic studies have established a role for SLPR/MLK in JNK-pathway activation during embryonic tissue closure (Stronach and Perrimon 2002). Here, we have extended our initial findings about SLPR function in Drosophila by characterizing the phenotype of an allele affecting postembryonic development as well as protein products encoded by wild-type and mutant alleles. slprBS06 is a newly isolated null allele that encodes an early nonsense mutation and, consequently, no protein product is detected in mutant tissue clones or by Western immunoblot. Phenotypic comparison between the null allele and existing alleles confirms the role for SLPR in dorsal closure, clarifies that slpr has a maternal contribution and that the prior two alleles encode dominant-negative proteins, and uncovers additional roles for SLPR in metamorphosis of the adult.
The severe dorsal open phenotype of slprBS06 germline clones, maternally and zygotically mutant, indicates that dorsal closure is the earliest requirement for SLPR in embryogenesis and that a failure of dorsal closure is the null phenotype, consistent with the phenotype of the previously characterized 921 and 3P5 slpr alleles. In contrast, however, most slprBS06 zygotic mutants survive embryogenesis and adult mutant males are recovered at a low frequency. These males display several visible morphological phenotypes of variable penetrance, presumably as a consequence of the eventual depletion of functional maternal product. These observations indicate that the maternal slpr gene product is nearly sufficient for embryogenesis in the absence of zygotic product, but rarely provides enough function for metamorphosis, revealing additional roles for SLPR in postembryonic processes.
Defects observed in mutant adults implicate SLPR function for proper metamorphosis of the genital discs, dorsal abdomen and notum, maxillary palps, and wing. Females show somatic defects during oogenesis, demonstrating that SLPR is required for proper morphology of the chorionic dorsal appendages. Many of the defects, including those affecting the thorax, genitals, and dorsal appendages, have been documented previously to result from loss of JNK signaling (Agnes et al. 1999; Zeitlinger and Bohmann 1999; Martin-Blanco et al. 2000; Suzanne et al. 2001; Macias et al. 2004). Thus, the data reported here implicate SLPR as the upstream JNKKK family member required for JNK activation in these processes. This study also suggests that slpr function is mediated primarily, if not entirely, via HEP/MKK7 and the JNK pathway, as evidenced by the fact that hep mutants share all of the defects that we observed in slpr mutants and that reducing the dosage of two known negative regulators of JNK signaling, puc and raw, suppresses the slpr phenotypes. In light of these results, it will be informative to systematically test whether, in vivo, the mammalian MLK proteins activate alternative substrates and pathways as has been suggested from tissue culture studies (Chadee and Kyriakis 2004; Brancho et al. 2005).
Given that the slpr921 and slpr3P5 alleles are phenotypically more severe in zygotic mutants than the null allele, the encoded products have dominant-negative activity, which interferes with the functional pool of maternal slpr gene product. This is consistent with the molecular nature of the alleles, which predict that full-length (921) or partial (3P5) protein product would be expressed in the mutants. Indeed, clonal analysis and immunofluoresence staining confirm the expression of mutant protein in slpr921 animals. Protein levels in slpr3P5 mutant tissue appear reduced relative to wild type, but the encoded fragment retains the SH3 domain and most of the kinase domain, each of which, if folded properly, could engage in protein–protein interactions. Similarly, the catalytically inactive, full-length protein encoded by slpr921 retains several functional protein interaction motifs, which could account for the antimorphic properties of the protein.
What candidates are known to interact with the various regions of the SLPR protein to account for dominant-negative activity? By analogy with the mammalian MLK proteins, at least three recognized domains have potential protein-binding activity. The leucine zipper mediates homodimerization, which is requisite for autophosphorylation and substrate activation (Leung and Lassam 1998; Vacratsis and Gallo 2000). Mutant SLPR protein in slpr921 cells might trap wild-type protein in unproductive dimers. The CRIB domain binds to the activated form of the small GTPases Cdc42 and Rac1 (Burbelo et al. 1995; Bock et al. 2000), both implicated in dorsal closure. Titration of these GTPases by noncatalytic SLPR921 protein could also contribute to dominant interference of the wild-type SLPR protein. Finally, the N-terminal SH3 domain, retained in the proteins encoded by both slpr3P5 and slpr921, has the potential to engage in both intra- and intermolecular interactions. The SH3 domain of mammalian MLK3 can bind to a region between the LZ and CRIB domains through a critical proline residue (Zhang and Gallo 2001) that is conserved in Drosophila SLPR. The postulated intramolecular binding is thought to negatively regulate MLK activation by locking the protein in a closed conformation. This type of autoinhibition has been demonstrated for other modular kinases, such as Src tyrosine kinase (Boggon and Eck 2004). Also, the SH3 domain may serve as a docking site for upstream activating kinases of the Ste20 family (Kiefer et al. 1996), for which titration by an SH3-containing protein fragment could impair signal relay to the JNK pathway (Tibbles et al. 1996). Therefore, the modular domain organization of the SLPR protein with the potential for multiple regulatory protein interactions is likely to explain why residual mutant protein is more detrimental than complete loss of protein in the null mutant.
Why, then, does overexpression of an engineered kinase-dead SLPR transgene that is functionally equivalent to the protein encoded by slpr921 have such mild consequences in the embryo? Evidence suggests that it is due to a substantial functional maternal component, in addition to the zygotic contribution, because reducing the maternal pool in embryos derived from slpr−/+ heterozygous mothers exacerbates the effect of dominant-negative SLPR transgene expression. Further support for the function of the maternal gene product is demonstrated by the sensitivity of the maternal contribution to the dosage of additional positive and negative JNK pathway regulators, which we have monitored as the extent of recovery of slprBS06 mutant adult males.
Given that the maternal product is nearly sufficient for mutant males to survive to adulthood, it is curious that immunodetection of endogenous SLPR protein in the embryo has been difficult relative to the ease with which we can detect transgenic protein. This may suggest that the maternal pool of slpr gene product is largely mRNA rather than protein, that an active mechanism exists to maintain low levels of embryonic protein, or that protein complexes mask the ability to detect the epitope on the SLPR protein, either of which could be overcome by the abundant expression of exogenous transgenic protein. Although the mechanism is not clear, the genetic loss-of-function and overexpression data together indicate that certain cell types or developmental contexts are sensitive to the levels of SLPR protein in modulating JNK signaling. For example, while exogenous SLPR can induce JNK signaling in embryonic dorsal ectoderm cells, normally limited for JNK activity, not all cells are equally inducible, suggesting that there may be other limiting components or brakes that modulate the precise levels of JNK activity in cells. Inferring function from overexpression experiments in the absence of loss-of-function data can be misleading, however, because wild-type transgene expression can stimulate JNK signaling promiscuously, or at least where the endogenous protein appears not to be required. For example, transgenic expression of either SLPR or TAK1 can induce JNK signaling ectopically in the embryonic dorsal ectoderm under the control of pnr-GAL4 (this work; Mihaly et al. 2001), even though endogenous levels of TAK1 cannot provide enough JNK-signaling activity in slpr null embryos to rescue dorsal closure. Moreover, Tak1 mutants are viable, providing corroborating evidence that Tak1 is not required for dorsal closure (Vidal et al. 2001). In sum, JNK-signaling activity may be at a threshold level in most cells, easily overactivated by expression of many different upstream regulators, but whose selective use in physiological circumstances is revealed only through analysis of loss of function.
The combined gain- and loss- of-function analysis for SLPR described here supports two proposed mechanisms of signaling specificity among JNKKK proteins: first, that individual family members are used selectively in particular contexts, and second, that potential combinatorial or redundant functions may exist among members with common substrates (Chen et al. 2002). With respect to TNF/Eiger signaling, both loss-of-function analysis and dominant-negative constructs consistently implicate TAK1 rather than SLPR in this JNK-dependent response (this work; Igaki et al. 2002; Geuking et al. 2005). In addition, JNK-dependent developmental morphogenetic events, in particular dorsal closure in the embryo, selectively require SLPR. Yet, we have consistently observed that the slprBS06, Tak1 double mutant is more severe than either single mutant alone, suggesting that there are likely to be additional, perhaps redundant, functions of SLPR and TAK1 that are revealed only in the double mutant. These functions have yet to be investigated in detail.
At face value, genetic analysis has allowed the assignment of SLPR to mediate many JNK-dependent morphogenetic events and TAK1 to mediate JNK-dependent homeostatic responses, including apoptosis and immunity. However, it is still unclear whether the selective functions reflect differential transcriptional responses or whether cell and developmental context shapes what appear to be quite different cellular behaviors. In other words, although the developmental defects that arise as a consequence of loss of SLPR function may suggest a common failure in cell-shape change or cytoskeletal functions, similar to the defects that underlie the failure of embryonic dorsal closure in the mutants, that assumption may be too simplistic. It is a formal possibility that SLPR could regulate additional or alternative JNK-dependent cell responses in distinct contexts. For example, although SLPR appears not to mediate TNF-induced apoptosis in the Drosophila eye under conditions where Eiger is overexpressed, the male genital misrotation phenotype observed in slpr mutants may be linked to JNK-dependent developmental programmed cell death. Defective genital rotation is observed in certain viable alleles of hid, encoding a protein with proapoptotic function (Abbott and Lengyel 1991; Grether et al. 1995). The basis of the rotation defect may be due to an excess of genital disc cells, similar to the embryonic defects in head involution, the namesake phenotype of hid (Grether et al. 1995). Thus, both JNK signaling and HID function are required for proper genital rotation and, interestingly, there is precedent for hid being a transcriptional target of the JNK pathway downstream of Eiger (Moreno et al. 2002). Thus, it will be important to determine specifically whether SLPR mediates JNK-dependent HID expression or even apoptosis in imaginal discs or whether the requirement for SLPR in male genital rotation is unrelated to apoptosis. More generally, a full understanding of SLPR function will require systematic definition of the molecular and cellular mechanisms that underlie the morphological defects. If SLPR functions to regulate different outputs in different contexts, determining what regulates a selective response will be of considerable interest for future studies.
Acknowledgments
We are grateful to M. Miura, G. Campbell, A. Letsou, D. Bohmann, L. Kockel, S. Noselli, and A. Martinez-Arias for fly stocks. We also thank the lab of J. Hildebrand for technical assistance in affinity purification of the anti-SLPR antibody. We appreciate the comments of G. Campbell, B. McCartney, and V. Twombly on this manuscript and throughout the course of this work.
Footnotes
Communicating editor: A. J. Lopez
- Received January 31, 2006.
- Accepted July 24, 2006.
- Copyright © 2006 by the Genetics Society of America