ABSTRACT
Filamins are highly conserved actin-crosslinking proteins that regulate organization of the actin cytoskeleton. As key components of versatile signaling scaffolding complexes, filamins are implicated in developmental anomalies and cancer. Multiple isoforms of filamins exist, raising the possibility of distinct functions of each isoform during development and in diseases. Here, we provide an initial characterization of jitterbug (jbug), which encodes one of the two filamin-type genes in Drosophila. We generate Jbug antiserum that recognizes all of the spliced forms, which reveals differential expression of different Jbug isoforms during development with a significant maternal contribution of Jbug protein. To reveal the function of Jbug isoforms, we create new genetic tools, including a null allele that deletes all isoforms, hypomorphic alleles that affect only a subset, and UAS lines for expression of the major isoforms. Using these tools, we demonstrate that Jbug is required for viability and that specific isoforms of Jbug are required in the formation of actin-rich protrusions such as thoracic bristles in adults and ventral denticles in the embryo. We also provide evidence for trans-splicing in the jbug locus.
Introduction
Actin-crosslinking proteins organize actin filaments into higher-order structures such as orthogonal actin arrays and parallel actin bundles. Filamins are highly conserved large cytoplasmic proteins that crosslink actin filaments into dynamic three-dimensional structures (Nakamura et al. 2011; Razinia et al. 2012). Filamins consist of actin-binding Calponin-homology (CH) domains at their N-terminus, followed by a long C-terminal rod-like domain of filamin-type immunoglobulin (IG-FLMN) repeats. Dimerization of filamins through the last C-terminal IG-FLMN repeat allows the formation of a flexible V-shaped structure that holds two actin filaments at large angles to create either a loose three-dimensional actin network (Nakamura et al. 2011; Razinia et al. 2012) or parallel bundles of actin filaments (Sokol and Cooley 1999; Gay et al. 2011). In addition to binding to actin, filamins interact with transmembrane receptors, adhesion molecules and even transcription factors, and are thus involved in multiple cell functions including motility, maintenance of cell shape, and differentiation (Nakamura et al. 2011; Razinia et al. 2012). Mutations in filamins are associated with a wide range of congenital anomalies and have been shown to both promote and inhibit metastasis and cancer growth (Razinia et al. 2012; Savoy and Ghosh 2013; Shao et al. 2016; Sasaki et al. 2019), emphasizing the multiple critical roles of filamins in development and disease.
Studies in Drosophila have helped refine our understanding of the roles of filamins in vivo. Drosophila has two orthologs of human filamin A: Cheerio (Cher) and Jitterbug (Jbug). Cher shares the organization of the protein and 46% identity and 61% similarity in amino acid sequence with human filamin A. Jbug, with 23% identity and 36% similarity in amino acid sequence with human filamin A, has some distinct features, including an actin-binding domain consisting of three, instead of two, CH domains. Several studies have revealed roles for Cher in organizing the F-actin cytoskeleton in multiple developmental contexts and diseases, including formation of ovarian germline ring canals (Li et al. 1999; Sokol and Cooley 1999; Huelsmann et al. 2016), follicle cell migration (Sokol and Cooley 2003), tissue morphogenesis during cellularization (Krueger et al. 2019), and Drosophila tumorigenesis (Külshammer and Uhlirova 2013). Compared to the extensive studies on Cher, our understanding of the roles of Jbug is more limited. Originally identified as a gene that causes bang-sensitive seizure – the phenotype that gives the gene its jitterbug name (Song and Tanouye 2006), jbug has been shown to be required in photoreceptor cells for axon targeting (Oliva et al. 2015) and in tendon cells to maintain their shape at the muscle-tendon junction (Olguín et al. 2011; Manieu et al. 2018).
Multiple isoforms of filamin proteins exist in both Drosophila and mammals (Sokol and Cooley 1999; Browne et al. 2000; Van der Flier and Sonnenberg 2001; Gorlin et al. 1990; Wang et al. 2007). Originally, only two Cher isoforms had been reported: a large isoform (FLN240) that contains both actin-binding CH domains and IG-FLMN repeats and a smaller one (FLN90) that contains only IG-FLMN repeats (Sokol and Cooley 1999). Recent annotations indicate that cher encodes six longer isoforms (>240 kDa), which include two actin-binding CH domains and 18-22 IG-FLMN repeats, and four shorter isoforms (90-100 kDa), which include only eight IG-FLMN repeats without any CH domains (Flybase; www.flybase.org; Thurmond et al. 2019). Importantly, a recent study in Drosophila larval neuromuscular junctions revealed a role for the shorter Cher isoform (FLN90) in synapse formation (Lee and Schwarz 2016), suggesting distinct roles for different isoforms. Like cher, jbug also encodes multiple isoforms with a range of protein sizes. The roles for jbug, however, have been revealed mostly using RNA interference (RNAi) constructs that target most of the existing splice forms, preventing the characterization of isoform-specific roles for Jbug.
Here, we report on the creation of several new genetic tools for parsing out the roles of the different Jbug isoforms, including a null allele that deletes all isoforms and hypomorphic alleles that affect only a few. We also create UAS constructs for expressing some of the major Jbug isoforms. We demonstrate that different isoforms are differentially expressed during development and localize to different domains in embryonic epithelia. We further show an essential role for Jbug in organismal viability and epithelial morphogenesis. We also show isoform-specific roles of Jbug in formation of epidermal denticles during embryogenesis and thoracic bristles in adults. We also report on the rescue of jbug function through trans-splicing.
Materials and Methods
Fly stocks and husbandry
Fly lines used in our experiments were: jbug20, jbug30, jbug133, UAS-Jbug-RC, UAS-Jbug-RF (this work); UAS-Jbug-RL (Olguín et al. 2011); Oregon R, fkh-Gal4 (Henderson and Andrew 2000), 69B-Gal4 (Rorth 1996); da-Gal4 (Wodarz et al. 1995); matα-Gal4 (Häcker and Perrimon 1998), Df(2R)59AB, Df(2R)Exel6079, TRiP.JF01116 (Bloomington Stock Center); GD8664, GD13033 (Vienna Drosophila RNAi Center). All crosses were performed at 25°C.
Generation of jbug mutant alleles
jbug mutants were generated by homologous recombination (Gong and Golic 2003). For jbug20 null mutants, genomic fragments upstream and downstream of the jbug ORF were amplified by PCR using the following primers (restriction sites in bold and linkers in italic):
in-fusion_L-22737893, 5’-CTAGTCTAGGGCGCGCCGATGAGTTGTGGCTTGAGCA-3’
in-fusion_R-22742442, 5’-TAGGGGATCACGTACGGCCAAACCAATGCTGAAGAT-3’
NotI_L-22717574, 5’-ATGCGGCCGCTAGAGGGGAGAGCAAGTGGA-3’
Acc65I_R-22721186, 5’-ATAGGTACCGTGTGAGCTTCGGGATCAAT-3’
For jbug30 and jbug133 hypomorphic alleles, genomic fragments upstream and downstream of an exon common to all jbug splice forms were amplified using the following primers (restriction sites in bold):
AscI_L-22727168, 5’-ATGGCGCGCCAAAGCCGTTGAAGAAAGCAA-3’
BsiWI_R-22731840, 5’-ATACGTACGCTCGATGCACTTTGTCTCCA-3’
NotI_L-22717574, 5’-ATGCGGCCGCTAGAGGGGAGAGCAAGTGGA-3’
Acc65I_R-22721186, 5’-ATAGGTACCGTGTGAGCTTCGGGATCAAT-3’
PCR fragments were cloned into pW25, which carries white+, the recognition site for I-SceI endonuclease, and FRT sites. The constructs were injected into w1118 embryos by Rainbow Transgenic Flies, Inc. Transformants were crossed to flies carrying hs-I-SceI and hs-Flp and progeny were heat shocked (37°C) for 1 hour 48–72 hours after egg laying (AEL). Recombination and insertion were confirmed by genomic PCR and reverse-transcriptase PCR (RT-PCR) followed by sequencing.
The following primers were used for RT-PCR for jbug20 shown in Supplemental Figure 1 (restriction sites in bold and linkers in italic):
jbug 5-1,5’-GGAGAACAAGTACCGGGTGA-3’
jbug 5-2, 5’-GGATCCTCGAGCATATGTCCTTCCACGTTCACCTCGA-3’ (same as Jbug-Ab-3’ for making Jbug antibody)
jbug 3-1, 5’-ATGCGGCCGCTAGAGGGGAGAGCAAGTGGA-3’ (same as NotI_L-22717574 used for amplifying a homology arm downstream of the jbug ORF)
jbug 3-2, 5’-CGCGCGGCAGCCATATGATGTCCTCACCCGGCCTAAC-3’ (same as Jbug-Ab-5’ for making Jbug antibody)
Generation of UAS-Jbug-RC and-RF transgenic flies
An ORF for jbug-RC (1230 bp) was PCR-amplified using the existing cDNA RE40504 as a template. A full cDNA for jbug-RF (9430 bp) was created by In-Fusion cloning of three fragments. The 5’ fragment was prepared by enzyme digestion of RE40504 (cut with ApaI and BamHI). The middle fragment (1189 bp) was amplified by reverse transcriptase-PCR (RT-PCR) and the ApaI and BglII sites were added to the 5’ and the 3’ end of the fragment, respectively. The 3’ fragment was amplified by PCR using GH03118 from the BglII site to the end of the ORF and the BamHI site was added to the 3’ end. The ligated full cDNA was confirmed by sequencing. jbug-RC and jbug-RF ORFs were PCR-amplified and cloned into the pUAST vector (Brand and Perrimon, 1993) to create UAS-Jbug-RC and UAS-Jbug-RF. The DNA constructs were injected into w1118 embryos by Rainbow Transgenics, Inc. (UAS-Jbug-RC) and by GenetiVision (UAS-Jbug-RF).
Generation of Jbug antibody
A PCR fragment of 1212 bp that is common to all jbug splice forms was cloned into the pET15b vector. The fragment corresponds to almost the entire open reading frame for the Jbug-PC isoform except for the last six amino acids and contains four IG-FLMN repeats. Primers used for In-Fusion cloning were (restriction sites in bold and linkers in italic):
Jbug-Ab-5’, 5’-CGCGCGGCAGCCATATGATGTCCTCACCCGGCCTAAC-3’
Jbug-Ab-3’, 5’-GGATCCTCGAGCATATGTCCTTCCACGTTCACCTCGA-3’
The construct was transformed into BL21-DE3 cells for protein induction with IPTG. Recombinant protein was purified using Ni-NTA agarose (Qiagen). Rabbit polyclonal antibodies were generated by Covance, Inc. and used at a dilution of 1:500 for immunohistochemistry and 1:5,000 for western blot.
Western blot
Protein extracts from embryos 0-2 and 2-16 hours AEL, 1st, 2nd, 3rd instar larvae, pupae (2 days after puparium formation (APF), and 2-day old male and female adults were used. For embryos homozygous for jbug20, Df(2R)59AB and Df(2R)Exel6079 and for 1st instar larvae homozygous for jbug20 and transheterozygous for jbug30/Df(2R)Exel6079, a balancer chromosome that contains the twi-Gal4 UAS-GFP construct was used to distinguish desired genotypes from heterozygous siblings. Rabbit a-Jbug (this work; 1:5,000), mouse a-Actin (RRID:AB_11004139; Invitrogen; 1:2,000), and HRP-labelled secondary antibodies (RRID:AB_430833; RRID:AB_430834; Promega; 1:2,000) were used. Protein bands were detected using Pierce ECL Western Blotting Substrate.
Scanning electron microscopy
Wild type (Oregon R) and jbug30 homozygous male flies were coated with gold palladium and examined and photographed in a LEO/Zeiss Field-emission SEM.
Antibody staining and confocal microscopy
Embryos were collected on grape juice-agar plates and processed for immunofluorescence using standard procedures. Briefly, embryos were dechorionated in 50% bleach, fixed in 1:1 heptane:formaldehyde for 40 min and devitellinized with 80% EtOH, then stained with primary and secondary antibodies in PBSTB (1X PBS, 0.1% Triton X-100, 0.2% BSA). For Jbug and phalloidin staining, embryos were hand-devitellinized. Antibodies used include: mouse a-Crb (RRID: AB_528181; DSHB 1:10), rabbit a-CrebA (RRID: AB_10805295; 1:5,000), guinea pig a-Sage (Fox et al. 2013), chicken a-GFP (RRID:AB_2534023; Invitrogen; 1:500), rat a-E-Cad (RRID: AB_528120; DSHB; 1:50), rabbit a-Jbug (this work; 1:500), rabbit α-ß-Gal (RRID: AB_221539; Invitrogen; 1:500). Alexa fluor 488, 568, 647-labelled secondary antibodies were used at 1:500 (RRID: AB_142924; RRID: AB_141778; RRID: AB_2535766; RRID: AB_143157; RRID: AB_2535812; Molecular Probes). Alexa fluor-568-labeled phalloidin was used for F-actin labelling (RRID: AB_2632953; Molecular Probes; 1:250). All images were taken with a Leica SP8 confocal microscope using either a 63x, 1.4 NA or a 40x, 1.3 NA oil objective and the LAS X software (Leica).
Quantification of denticle precursors in the ventral epidermis
Embryos stained for phalloidin and E-Cad were imaged with a Leica SP8 confocal microscope. A maximum intensity projection was generated using two focal planes (0.3 μm apart) in the region of adherens junctions. Denticle belts in abdominal segments 4-6 in stage 16 embryos were selected for quantification. Six columns of cells from anterior to posterior of each belt were analyzed. In each column, denticle precursors of four cells closest to the ventral midline were counted. For each genotype, five to eight embryos were quantified. Unpaired Student’s t-test with Welch’s correction was used for statistical analysis.
Results
Different splice forms of jbug are differentially expressed during development
Ten annotated splice forms of jbug, which encode six different protein isoforms, are reported in Flybase (www.flybase.org; Thurmond et al. 2019). All isoforms contain multiple repeats of filamin-type immunoglobulin (IG-FMLN) domains (Figure 1, A and B), which form a rod-like structure in filamin. Two Jbug isoforms consist exclusively of these repeats; Jbug-PC/G/H/J/K (each splice form encodes the same 42.4 kDa protein; hereafter refered to as Jbug-PC) has four IG-FMLN repeats and Jbug-PI (185.7 kDa) has sixteen (Figure 1B). On the other hand, Jbug-PL (175.6 kDa) and Jbug-PF/M/N, the three longest Jbug isoforms (>300 kDa), contain three tandem calponin-homology (CH) actin-binding domains in their N-termini as well as multiple IG-FMNLs in their C-termini (seven and nineteen IG-FMLNs for Jbug-PL and Jbug-PF/M/N, respectively) (Figure 1B).
High-throughput expression data available in Flybase (www.flybase.org; Thurmond et al. 2019) showed high expression of jbug splice forms throughout all developmental stages, except during 0-18 hour embryogenesis when there is very little expression of jbug-RF, RM, RN and jbug-RL (Figure 1C; FB2020_04). To analyze protein expression levels of Jbug during Drosophila development, we generated Jbug antiserum that recognizes a protein domain encoded by all ten jbug splice forms (Figure 1A). This domain includes the four IG-FLMN repeats that are present in all Jbug isoforms (Figure 1B). Our Western blot analysis using wild-type protein extracts revealed that Jbug proteins are detected at high levels throughout all developmental stages (Figure 1D). Strong Jbug signals in protein extracts from 0-2 hour embryos (Figure 1D) also suggest a maternally provided pool of Jbug.
The Jbug antiserum detected bands in the size ranges that are predicted from the major splice forms (42.4 kDa, 175.6 kDa, 185.7 kDa, >300 kDa). Importantly, the different isoforms were differentially expressed during development (Figure 1D). Jbug-PC (42.4 KDa) was abundant only in pupae and Jbug-PL (175.6 kDa) was most abundant in adults (Figure 1D). Bands corresponding in size to Jbug-PI (185.7 kDa) and Jbug-PF/M/N (>300 kDa) were detected at all stages with slight differences in protein levels (Figure 1D). Jbug-PF/M/N isoforms differ only slightly from one another in regions outside the CH and IG-FLMN domains. Due to their huge sizes and a very slight difference in molecular weight among Jbug-PF/M/N, we were not able to distinguish exactly which of the three isoforms is expressed at any specific stage. The large size of these isoforms also diminishes membrane transfer, leading to relatively weak signals. Notably, several intermediate-sized bands that did not match the predicted size of annotated isoforms were also detected at different levels throughout development (Figure 1D). The existence of these bands suggests either that more Jbug isoforms exist than have been annotated or that proteolyzed and/or phosphorylated fragments are being detected (also see Figure 2F; most of these extra bands are absent in jbug null mutant larval extracts). Indeed, roles for proteolytic forms of Filamin A (Bedolla et al. 2009; Savoy and Ghosh 2013; Yue et al. 2013) as well as phosphorylated Filamin A (Ohta and Hartwig 1996; Tirupula et al. 2015) have been reported from studies of the mammalian proteins. Taken together, our data indicate that different splice forms of jbug are differentially expressed during Drosophila development.
Null and hypomorphic jbug alleles have been generated
To investigate roles for jbug in development, we generated mutant alleles for jbug using homologous recombination (Gong and Golic 2003). Two donor DNA constructs were used to target different genomic regions. One construct targets the genomic region encompassing the translation start sites for all splice forms and extends through the exons shared by all splice forms; the other targets only the region containing the exons shared by all splice forms (Figure 2, A and B). With the construct that targets the larger region, we obtained a null mutant, jbug20 (Figure 2A). PCR analysis confirmed that the entire targeted region was replaced with the mini-white+ (mini-w+) gene (data not shown). Reverse transcriptase-PCR (RT-PCR) using mRNA extracted from homozygous jbug20 embryos further confirmed that no jbug transcripts are produced in jbug20 embryos (Supplemental Figure 1, A-C). Both jbug20 homozygous animals and transheterozygotes for jbug20 over either of the two deficiency lines that delete all or part of the jbug locus (59AB and Exel6079; Figure 2A) died during the first instar larval stage (Table 1). Western blot analysis using Jbug antiserum showed that whereas a significant amount of the Jbug proteins were detected in jbug20 embryos (Figure 2E), almost all protein bands were absent by the first instar larval stage, with the exception of two low molecular weight bands that are likely to be from nonspecific cross-reactivity (Figure 2F). Similarly, all Jbug isoforms were detected both in 59AB and Exel6079 homozygous embryos (Figure 2I). These data suggest that Jbug proteins from maternally provided jbug mRNA or protein persist through embryogenesis. Since only very low level of jbug mRNA is detected in 0-2 hour wild-type embryos, based on RNA-Seq (Flybase; www.flybase.org) and in situ hybridization (Fly-FISH; http://fly-fish.ccbr.utoronto.ca), it is more likely that the Jbug protein, and not the mRNA, is maternally provided.
With the DNA construct that aimed to replace the genomic region common for all splice forms of jbug, we unexpectedly obtained two hypomorphic alleles, jbug30 and jbug133, that affect only a subset of splice forms (Figure 2, B-D). Incomplete recombination occurred in these two alleles, resulting in either a part (for jbug30) or the entire DNA construct (for jbug133) being inserted into the jbug locus (Figure 2B). In jbug30, one homology arm in the donor DNA recombined correctly, but the other homology arm was inserted into the genome along with the mini-w+ gene, rather than being recombined (Figure 2C). In jbug133, homologous recombination failed to occur; instead, the entire donor construct was inserted into the 3’ UTR of the splice forms that have longer C-termini (jbug-RI and jbug-RF/M/N) (Figure 2D). These recombination events disrupt jbug-RI and jbug-RF/M/N in both alleles. Western blot analysis revealed that the Jbug-PF/M/N isoforms (>300 kDa) were either absent or decreased to undetectable levels in jbug30 mutants at all stages (Figure 2, E-H and J). Jbug-PI (185.7 kDa) was also absent in jbug30, and instead, a smaller protein (~90 kDa) was detected in jbug30 mutant animals at all stages (Figure 2, E-H and J). This additional band was not detected in wild type or jbug133 protein extracts at any stages, suggesting that insertion of the DNA construct generates a truncated form of Jbug-PI in jbug30. In jbug133, both Jbug-PF/M/N and Jbug-PI protein bands were slightly decreased in intensity (Figure 2, E-H), suggesting that insertion of the construct in the 3’ UTR region either affects mRNA stability or inhibits translation of these isoforms. Consistent with the molecular data, we did not detect any significant changes in the level of Jbug-PC (42.4 kDa) and Jbug-PL (175.6 kDa) in jbug30 and jbug133 protein extracts (Figure 2, E-H). Taken together, we generated new alleles for jbug, including a null mutant and two hypomorphic alleles that disrupt a subset of Jbug isoforms, specifically Jbug-PI and Jbug-PF/M/N.
jbug hypomorphs show semi-lethality and defects in bristle development on the thorax
Many homozygous flies in jbug30 and jbug133 died during pupation, leaving a number of dead pupae that never eclosed. In both alleles, homozygous adult flies that survived were fertile and homozygous lines could be maintained. Interestingly, adult flies homozygous for jbug30 or jbug133 produced short, bent and rough-ended bristles on the thorax (Figure 3A), suggesting a role for Jbug in bristle development, perhaps in crosslinking actin bundles that form the bristles. Consistent with our findings that levels of Jbug-PF/M/N and Jbug-PI were more significantly reduced in jbug30 than in jbug133 (Figure 2, E-H), jbug30 flies showed a higher expressivity of bristle defects than jbug133. Whereas almost every thoracic bristle had defects in jbug30 homozygous adults, only a few bristles showed distinguishable defects in jbug133 homozygotes (Figure 3A).
To analyze the bristle defects in more detail, we performed scanning electron microscopy (SEM) in wild type and jbug30 homozygous mutant adults. Whereas wild-type animals showed nicely elongated and tapered bristles that suggested evenly aligned actin bundles (Figure 3B, a and a’), jbug30 mutant flies had bristles that were short, curved (Figure 3B, c), bent (Figure 3B, d) and rough-ended (Figure 3B, e). Even when the bristles looked relatively normal, they were often shorter than wild type and mildly twisted (Figure 3B; compare b and b’ to f and f’).
Transheterozygotes of jbug30 or jbug133 over the jbug20 null allele or the 59AB deficiency line that deletes the entire jbug gene (Figure 2, C and D) showed both semi-lethality and the same bristle defects as observed in homozygous jbug30 and jbug133 adults (Table 1). These data suggest that both phenotypes are due to loss of jbug function. Consistent with our data that jbug30 is a more severe hypomorphic allele than jbug133 (Figure 2, E-H; Figure 3A), fewer transheterozygous jbug30/jbug20 or jbug30/59AB flies survived to the adult stage compared to transheterozygous jbug133/jbug20 or jbug133/59AB flies (Table 1). Taken together, these data suggest that Jbug-PF/M/N and/or PI are important for viability and for actin bundle arrangement within bristles.
We also performed complementation tests using the deficiency line Exel6079, which deletes through the genomic region encoding the N-terminal exons of transcripts of jbug-RL and jbug-RF/M/N, but leaves the coding region for jbug-RC and jbug-RI intact (Figure 2, C and D). Interestingly, transheterozygous jbug30/Exel6079 (and jbug133/Exel6079) flies showed neither semi-lethality nor bristle defects (Table 1). This finding was unexpected because Jbug-PF/M/N should be absent in transheterozygous jbug30/Exel6079 flies; jbug30 and Exel6079 disrupt the 3’ and the 5’ ends of jbug-RF/M/N, respectively. To test which isoforms are present in jbug30/Exel6079 flies, we performed Western blot using extracts from jbug30/Exel6079 transheterozygous 1 st instar larvae. Surprisingly, strong Jbug-PF/M/N bands were detected in jbug30/Exel6079 (Figure 2J). The rescue of jbug30 and Jbug133 phenotypes by Exel6079 and the presence of protein isoforms that should be missing in either Exel6079 or in the jbug hypomorphic alleles, suggest some form of trans-splicing, wherein the 5’ exons encoded on the jbug hypomorphic chromosomes are spliced to 3’ exons encoded on the Exel6079 chromosome.
Expression of Jbug-RF rescues the bristle defects and semi-lethality of jbug hypomorphs
To begin to parse out the functions of the different Jbug isoforms, we created two new UAS-lines, one encoding Jbug-PC (42 kDa protein) and one encoding Jbug-PF (>300 kDa protein). We also obtained a UAS construct for Jbug-PL (175.6 kDa protein) expression from the Mlodzik lab (Olguín et al. 2011). We first tested whether expression of any specific jbug splice form(s) could rescue the lethality of jbug20 null mutants. Each of the three constructs was expressed in jbug20 mutant background using daughterless-Gal4 (da-Gal4; Wodarz et al., 1995), which is expressed in all cells, or 69B-Gal4 (Rorth 1996), which is expressed in ectodermal cells. None of the UAS-constructs rescued the lethality of jbug20; indeed, no larvae survived past the 1st instar larval stage. These data suggest that multiple Jbug isoforms are required for viability.
We next tested whether expression of specific Jbug isoforms could rescue the bristle defects in jbug30, the severe hypomorphic allele. Consistent with the data that Jbug-PC is not affected in jbug30 mutants, overexpression of jbug-RC with da-Gal4 or with 69B-Gal4 in an otherwise wildtype background did not result in any overt defects in bristles (Figure 3C) nor did it rescue the bristle defects of jbug30 homozygotes (Figure 3E). On the other hand, overexpression of jbug-RF in otherwise wild-type animals resulted in shorter and thicker bristles (Figure 3D), suggesting that proper levels of Jbug-PF/M/N are critical for normal bristle development. Importantly, expression of jbug-RF in jbug30 mutants using da-Gal4 or 69B-Gal4 nearly completely rescued the bristle phenotypes (Figure 3F). Moreover, jbug-RF expression in jbug30 mutants also rescued semi-lethality. 86% (27/32) of pupae emerged as adult flies when jbug-RF was overexpressed in jbug30 mutants whereas only 26% (9/35) of jbug30 homozygous pupae emerged as adult flies. These data suggest that the bristle defects and semi-lethality observed in jbug30 are due to loss of the longest (>300 kDa) Jbug isoforms.
Jbug hypomorphs show delay in actin-based prehair formation and mild defects in wing hair orientation
Besides thoracic bristle defects, jbug30 and jbug133 homozygous adults also showed a mild swirling pattern of wing hairs (Supplemental Figure 2, B and C). Phalloidin staining of actin-rich prehairs at 32 hours after puparium formation (APF) revealed that hair formation is delayed in jbug30 pupal wings (Supplemental Figure 2, D and E). A swirling pattern of wing hairs and delayed prehair formation are often observed in mutants for planar cell polarity (PCP) genes (Adler 2012). jbug30 mutant eyes, however, did not have any defects in ommatidial rotation characteristic of loss of key PCP genes (Supplemental Figure 2, F and G), suggesting that Jbug is not involved in ommatidial rotation. Overall, these data suggest that proper levels for Jbug-PF/M/N and/or Jbug-PI are required for the timely maturation of the actin bundles that form the prehairs.
Jbug proteins localize in a dynamic pattern in the embryonic epithelium
We next analyzed the subcellular localization of Jbug proteins during embryogenesis in wild type and jbug mutant alleles. During early development, Jbug was detected as small punctate structures in the cytoplasm of all cells (Figure 4A). During stages 11-13, Jbug was detected at a high level both near the apical surface and at and near the adherens junctions (AJs) in all epidermal cells (Figure 4B). AJ Jbug signals became weaker over time, and by stage 15, only mesh-like signals on the apical surface remained (Figure 4C). In jbug20 mutant embryos, AJ signals were absent but strong signals on the apical surface were still detected (Figure 4D). At stage 15 and later, apical surface signals were slightly decreased compared to wild type levels (Figure 4E). Since the jbug30 and jbug133 homozygotes that survived to adult stages were fertile, we analyzed Jbug signals in jbug30 or jbug133 embryos from homozygous mothers. Interestingly, in jbug30 embryos from jbug30 homozygous mothers, Jbug signals were both very weak and dispersed in epidermal cells throughout the stages, and largely absent in AJs (Figure 4, F and G), suggesting that Jbug-PF/M/N and Jbug-PI might be major isoforms that show strong embryonic expression and AJ localization. Expression of Jbug-RF in the jbug30 mutant background restored strong signals at AJs (Figure 4G) and on the apical surface (Figure 4K) at stage 13 and 15, respectively. Consistent with the minor changes in Jbug protein levels in jbug133, Jbug staining was mostly normal, albeit slightly reduced, in the epidermal cells in jbug133 embryos from homozygous jbug133 mothers (Figure 4, H and I).
BDGP Expression Data showed that, in addition to the strong ectodermal expression at earlier stages, jbug is also upregulated in epithelial tubular organs, including the trachea and the salivary gland (https://insitu.fruitfly.org/cgi-bin/ex/insitu.pl). Our previous work also showed that tracheal expression of jbug is upregulated by Trachealess, a major transcription factor in the trachea (Chung et al. 2011). Consistent with mRNA expression, antibody staining in wild-type embryos revealed accumulation of Jbug protein in the trachea and the salivary gland (Figure 5, A and E). Interestingly, Jbug was abundant in both the cytoplasm and in a distinct apical domain of both tracheal and salivary gland cells, which appeared to be just under the apical surface(Figure 5, A and E), as opposed to following the AJ pattern of E-Cad. This apical staining was prominent up to stage 14 in the salivary gland; at later stages, Jbug signals were observed mostly in the cytoplasm and detected in the apical domain only as a thin layer along the lumen (compare Figure 5E to Figure 6A).
Jbug signals in tubular epithelial organs were significantly decreased in jbug20 embryos, even with regards to surrounding tissues (Figure 5, B and F), suggesting that zygotic jbug expression is required for upregulation of Jbug in these tissues. In jbug20, the apical staining in the trachea was absent and in the salivary gland was more punctate. In jbug30, apical staining of Jbug in the trachea and salivary glands was also absent; instead, uniform Jbug signals were detected in the cytoplasm (Figure 5, C and G). Consistent with slight changes in Jbug protein levels in jbug133, Jbug signals were mostly normal, albeit slightly reduced, in the trachea and salivary glands in jbug133 (Figure 5, D and H; data not shown). These data suggest that Jbug-PF/M/N (>300 kDa) and/or Jbug-PI (185 kDa) are enriched in the apical domain in tubular epithelial organs in the embryo.
Jbug function in embryonic epithelial tubes
Despite the absence or mislocalization of Jbug in the apical domain of trachea and salivary glands, both organs developed normally without showing overt morphological defects in jbug20 and jbug30 embryos (Figure 5, B, C, F and G). To ask if maternally supplied jbug is sufficient for embryonic development, we maternally knocked down jbug using the mata-Gal4 driver (Häcker and Perrimon 1998) to express a jbug RNAi construct that targets all jbug splice forms. Maternal knockdown of jbug did not result in overt defects in overall embryo morphology (data not shown). However, knockdown of jbug both maternally and zygotically using mata-Gal4 and fork head-Gal4 drivers (fkh-Gal4; a driver that drives expression in several embryonic tissues including the salivary gland; Henderson and Andrew 2000) resulted in severe morphological defects in the embryo, including deeper segmental grooves and a wider salivary gland lumen (Figure 5J). These data suggest that both maternal or zygotic pools of jbug are required for tubular epithelial organ morphogenesis.
To test the effect of Jbug overexpression in tubular epithelial tissues, we overexpressed different Jbug isoforms (Jbug-PC, PL and PF) in the salivary gland using fkh-Gal4. Interestingly, overexpression of each isoform caused an expanded and irregular apical domain (Figure 6, B-F). Moreover, the apical protein Crb, which is normally enriched in a sub-apical domain just apical to the AJs in many epithelial cells including salivary gland cells (Fig 6A, inset in Crb column; Wodarz et al. 1995; Tepass 1996; Chung and Andrew 2014), was mislocalized and dispersed along the entire apical surface in Jbug-overexpressing salivary glands (Figure 6, B-F). These data indicate that overexpression of multiple Jbug isoforms can disrupt the sub-apical domain enrichment of Crb. Interestingly, unlike other isoforms that are primarily localized in the cytoplasm when overexpressed (Figure 6, D-F), both untagged and C-terminal GFP-tagged Jbug-PC were detected in the cytoplasm and in nuclei (Figure 6, B and C). These data suggest that this smallest isoform – which is primarily expressed in pupae (Figure 1D) – might function in the nucleus as well. Taken together, these data suggest that appropriate levels of Jbug isoforms are important for apical organization of tubular epithelial cells during organ development.
Jbug is required for denticle formation in the epidermis
Jbug antibody staining in the ventral regions of stage 14 embryos revealed intense signals on the apical surface that presaged the actin-rich denticles that begin to form during stage 15 (Figure 7, B and C). This prompted us to ask if denticle formation was normal in jbug mutants. In wild type embryos, six rows of epidermal cells in the anterior portion of each segment normally produce one or more denticles, forming a trapezoidal belt (Bejsovec 2013; Figure 7, C and D; Figure 8A). Embryos homozygous for jbug20 did not show any overt morphological defects, likely due to strong maternal contribution (see Figure 1D and 2E), but they exhibited defects in denticle formation. Similarly, jbug30 embryos from a homozygous stock also exhibited defects in denticle formation, suggesting that Jbug-PF/M/N and/or Jbug-PI are important for actin bundle arrangement within denticles. Whereas the trapezoidal belts of actin-rich pre-denticles were relatively normal overall, there were both fewer and smaller denticle structures in jbug20 and jbug30 mutants compared to wild type (Figure 8, A-C). We counted the number of denticles in each cell from a grid of 24 cells (Figure 8H) for at least five individuals of each genotype. jbug20 embryos had fewer denticles per cell than wild type embryos in several rows of cells (C1, C3, C4 and C6; Figure 8, I and J). Irrespective of which row of cells (C1-C6) was examined, jbug30 embryos had, on average, fewer denticles per cell than wild type embryos (Figure 8, I and J). Expression of jbug-RF using the da-Gal4 driver in jbug30 mutants slightly increased the number of denticles in one row (C1), but not in other rows (C2-C6). Knocking down jbug-RL and jbug-RF/M/N by RNAi (GD13033; Figure 1A) using da-Gal4 also showed a decrease in the numbers of denticles per cell in rows C4 and C6 compared to wild type (Figure 8, F, I and J).
To test for a role of maternally provided Jbug in embryogenesis, we knocked down jbug by expressing each of the three RNAi lines that target different regions of jbug splice forms (Figure 1A) using a maternal driver mata-Gal4. Knocking down jbug using two strong RNAi lines (GD8664 and GD13033; Figure 1A) resulted in very few embryos, suggesting an essential role for Jbug in oogenesis. Knocking down jbug using a weak RNAi line JF01166 (Figures 1A) resulted in embryos with mild morphological defects such as slightly wavy epidermis and deeper grooves for some segmental grooves (Figure 8, G and L), suggesting that maternally provided Jbug proteins are required for normal development. These embryos also showed defects in denticle formation. As observed in jbug20 and jbug30 mutants or da-Gal4>jbug RNAi embryos, some cells failed to form denticles (Figure 8, G and I); others formed short denticles with weak phalloidin signals (Figure 8G). Quantification showed a decrease in the average number of denticles per cell in most columns (except for C2) of each denticle belt with mata>jbug RNAi (Figure 8J). Taken together, our data suggest that both maternally and zygotically provided Jbug proteins are required for denticle formation.
Discussion
Filamins play important roles in actin reorganization in many developmental processes and disease contexts. In this study we show that different isoforms of Jbug, the Drosophila filamin-type protein, are differentially expressed during development. Using new genetic tools that we generated, we reveal essential roles for different Jbug isoforms in viability, epithelial morphogenesis and formation of actin-rich structures.
Jbug plays a role in formation of actin-rich protrusions
Our results identify isoform-specific roles for Jbug in formation of two prominent actin-rich protrusions in Drosophila, bristles on the thorax and denticles in the ventral epidermis of the embryo (Figures 3 and 8). These defects are observed in the presence of Jbug-PL, which contains actin-binding domains and short IG-FLMN repeats, suggesting that this isoform is not sufficient to form actin-rich protrusions in Drosophila. Indeed, bristle defects are rescued by expression of Jbug-PF, one of the longest isoforms that contain long IG-FLMN repeats following the actin-binding domains (Fig. 3). These data suggest an important role for protein interactions through the long IG-FLMN repeats in forming actin-rich protrusions. Our interpretation is consistent with a role of filamins in acting as a molecular scaffold. More than 90 filamin-binding partners have been identified, including intracellular signaling molecules, receptors, ion channels, transcription factors, and cytoskeletal and adhesion proteins (Razinia et al., 2012).
Unlike defects in thoracic bristles, defects in denticle formation, although improved, were not fully rescued by overexpression of jbug-RF in jbug30 mutant embryos (Figure 8). These data suggest that either Jbug-PI, an isoform that contains long IG-FLMN repeats without actin-binding domains, might contribute to this process or that the expression system failed to provide enough of the jbug-RF splice form for rescue. Although Jbug is upregulated in the anterior side of each segment in the embryo, where denticle belts form, Jbug is not enriched in the denticles themselves (Figure 7). Interestingly, another Drosophila filamin – Cher – is enriched in denticles, colocalizing with actin protrusions (Dilks and DiNardo 2010). The different subcellular localization of Jbug and Cher in denticle forming cells suggests distinct roles for the two Drosophila filamins. Cher likely acts as a component of denticles that organizes actin filaments. A simple model for Jbug function is that the N-terminal actin-binding domains might act to stabilize actin filaments and that protein-protein interactions through long IG-FLMN repeats might recruit other key components for denticle formation to the apical plasma membrane where actin-rich protrusions arise. Indeed, a recent study has shown a role of the shorter Cher isoform (FLN90) that lacks actin-binding domains as a postsynaptic scaffold in Drosophila larval neuromuscular junctions (Lee and Schwarz 2016); FLN90 is required for synaptic localization of several key proteins, including type-A glutamate receptors and Ral GTPase (Lee and Schwarz 2016). Further studies will be needed to identify Jbug-interacting proteins and to reveal their roles in helping Jbug organize higher-order actin structures.
Jbug functions in epithelial morphogenesis
Maternal knockdown of jbug using strong RNAi lines results in very few embryos, suggesting Jbug functions in oogenesis. When a weaker RNAi line was used to knock down maternal jbug, embryos with mostly normal morphology are generated. When jbug was knocked down both maternally and zygotically using the same RNAi line, enhanced morphological defects were observed (Figure 5), suggesting that both maternal and zygotic pools of jbug have roles in epithelial morphogenesis. Dynamic localization of Jbug to the apical surface during embryogenesis (Figures 4 and 5) further suggests roles for Jbug in remodeling the actin cytoskeleton and contributing to the mechanical stability of the plasma membrane and the cell cortex. Interestingly, in jbug30 hypomorphic allele, where only Jbug-PC (the smallest isoform that only contain short IG-FLMN repeats) and Jbug-PL (an intermediate isoform that contains actin-binding domains and short IG-FLMN repeats) are intact, no overt defects are observed during embryogenesis except for defects in denticle formation at later stages. As Jbug-PC is not expressed in the embryo (Figure 1), these data indicate that Jbug-PL is sufficient for epithelial morphogenesis during embryogenesis.
Several studies suggest roles for filamin A in tubular epithelial morphogenesis and diseases in tubular organs. During branching of breast cells, the levels of filamin A expression and the extent of the integrin-filamin interaction modulate collagen remodeling to optimize matrix stiffness to support tubulogenesis (Gehler et al. 2009). Filamin A is also an important regulator for Polycystin-2, a key molecule in autosomal polycystic kidney disease (Wang et al. 2015). Our data also suggests a potential role for Jbug in apical domain reorganization in tubular epithelial cells (Figure 6). Jbug shows strong apical enrichment in the embryonic trachea and the salivary gland (Figure 5) and Jbug overexpression causes an expanded and irregular apical membrane with dispersed Crb localization along the apical surface of salivary gland cells (Figure 6). These data support the idea that appropriate levels of Jbug are important for apical organization during tubular organ development. Loss of apical Jbug enrichment, however, does not cause overt defects in tubular organ formation in jbug null or hypomorphic alleles, at least during embryogenesis (Figure 5). In these alleles, abundant cytoplasmic Jbug signals still remain. Overexpressed Jbug also mainly localizes to the cytoplasm in salivary gland cells (Figure 6). These data suggest that apical enrichment may not be necessary for Jbug function in the formation of tubular epithelia. Knockdown of jbug both maternally and zygotically (in the salivary gland a few other tissues) results in severe morphological defects with an abnormal salivary gland formation (Figure 5), indicating that tube formation will be quite abnormal in the complete absence of jbug function.
Potential roles for the smallest isoform Jbug-PC in the nucleus
Jbug-PC only contains four IG-FLMN repeats and lacks actin-binding domains, suggesting that the major mode of action of this smallest Jbug isoform is not cytoskeletal organization. Compared to other Jbug isoforms that are continuously expressed throughout development, Jbug-PC is abundant only during pupation (Figure 1), suggesting transient and specific roles for Jbug-PC. Our study also shows that Jbug-PC localizes both to the nucleus and to the cytoplasm when overexpressed in the embryonic salivary gland (Figure 6). Similar nuclear localization has been observed in human Filamin A. Although full-length filamin A is predominantly cytoplasmic, a C-terminal fragment of Filamin A colocalizes with androgen receptor to the nucleus to act as a co-transcription factor (Loy et al. 2003). Therefore, a possible scenario is that Jbug-PC has a role in transcriptional regulation. It will be interesting to test which developmental processes require the function of Jbug-PC during pupation and to reveal its exact roles in those processes.
Trans-splicing occurs in the jbug locus
Trans-splicing, a gene regulatory mechanism that joins exons from two separate transcripts to produce chimeric mRNA, has been detected in most eukaryotes (Horiuchi and Aigaki 2006; Lasda and Blumenthal 2011). In Drosophila, trans-splicing has been demonstrated to be an essential process for two genes, longitudinals lacking (lola) and modifier of mdg4 (mod(mdg4)). In the lola locus, the 5’-end of lola transcripts contain five exons that splice into 20 variants of 3’ exons to generate 20 protein isoforms (Ohsako et al. 2003). In the mod(mdg4) locus, two mutant alleles of the gene, each located in one of the two homologous chromosomes, restore the wild type function of the gene (Mongelard et al. 2002). This phenomenon is very similar to the rescue that we have observed in jbug30/Exel6079 and jbug133/Exel6079 transheterozygotes (Table 1). The presence of Jbug-PF/M/N isoforms in transheterozygous animals for jbug hypomophs and Exel6079 (Figure 2) further supports the idea that trans-splicing occurs, where the 5’ exons of jbug-RF/M/N transripts on jbug30 or jbug133 are spliced to 3’ exons of jbug transcripts on the Exel6079 chromosome. Recent studies have revealed conserved sequences in the mod(mdg4) intron that promote trans-splicing in this gene (Gao et al. 2015; Tikhonov et al. 2018). Identification of the sequences that promote trans-splicing in the jbug locus will help better understand this surprising gene regulatory mechanism.
Acknowledgments
We thank the members of the Chung and the Andrew laboratories for comments and suggestions. We thank M. Mlodzik and the Bloomington stock center for fly stocks and the Developmental Studies Hybridoma Bank for antibodies. We thank Flybase for the gene information. This work is supported by start-up fund from Louisiana State University to S.C. and NIH RO1 DE013899 to D.A.
Footnotes
Discussion has been updated.