SIDL Interacts with the Dendritic Targeting Motif of Shal (Kv4) K⁺ Channels in Drosophila

Yeast-Two-Hybrid (Y2H) Assays for Mapping Interaction Sites

The C-terminus of Shal1 (Shal1C; amino acids 411-571) was used as “bait” in a Y2H screen of pGADT7 “prey” constructs containing cDNAs reverse transcribed from RNA isolated from Drosophila embryos (aged 0-24 hrs; Clontech Laboraties Inc., Palo Alto, CA), as previously described (Diao et al., 2009).

For direct Y2H assays testing the specificity of the SIDL-Shal interaction, the C-termini of Shaker and Shaw were amplified from cDNA clones of ShA1 and Shaw2 by PCR, the C-terminus of Shab-RB was amplified by RT-PCR, the N-terminus of Shal and C-terminus of mouse K_v4.2 (mK_v4.2C) were amplified from cDNA clones by PCR; amplified products were subcloned into the pGBKT7 bait vector. For mapping the SIDL-Shal interaction, deletion constructs of SIDL and Shal were amplified and cloned into pGADT7 and pGBKT7 vectors, respectively. Shal2C(LL→AA) and Shal2C(LL→VV) constructs were generated using primers that incorporated the corresponding leucines substitutions. The LL-motif construct was obtained by extending two overlapping primers. All constructs were confirmed by DNA sequencing, co-transformed into AH109 yeast, and assayed for interaction by the expression of reporter genes (HIS3, Ade2, and MEL1).

GST-Pulldown Assays

Shal1C, Shal2C and Shal2CΔ9 were amplified from Shal cDNA clones using appropriately engineered primers. The C-terminus of SIDL (SIDLC) was amplified from the clone identified in our Y2H screen; PCR amplification rounds were designed to fuse the epitope c-myc to the N-terminus of SIDL. Shal and myc-SIDLC products were then subcloned into pGEX-2TK and transformed into BL21 cells to express GST-fusion proteins. GST and GST-fusion proteins were purified using glutathione-agarose beads (Sigma Aldrich, St. Louis, MO). myc-SIDLC protein was cut with biotinylated thrombin (Novagen Inc, San Diego, CA) from GST-myc-SIDLC, which contains an engineered thrombin site. Thrombin was removed using streptavidin-agarose beads. GST proteins were then incubated with purified myc-SIDLC or Drosophila embryo membrane extracts. Glutathione-agarose beads were washed, boiled in 4×SDS loading buffer, and run by standard SDS-PAGE techniques. Proteins were transferred to nitrocellulose and immunoblots were probed with designated antibodies.

In Situ Hybridization

Wild type (w¹¹¹⁸) embryos aged 12 hours (at 25°C) were collected, washed with water, dechorionated with bleach, and fixed with 4% paraformaldehyde in PBS. After rehydration, embryos were treated with protease K and then hybridized with probe at 55°C. Single-stranded digioxygenin-labeled (DIG-11-UTP, DIG DNA Labeling and Detection Kit, Roche, Basel, Switzerland) probes were amplified by PCR. The linearized SIDLC clone from the Y2H screen was used as template. The 3′activation domain primer was used to amplify the SIDLC probe. After hybridization and wash, embryos were incubated with anti-DIG antibody and developed by NBT/BCIP.

Antibodies

The anti-Shal antibody was generated by our lab, using the peptide sequence CGIELDDNYRD, as previously described (Diao et al., 2009). Monoclonal anti-syntaxin (8C3), anti-myc (9E10), anti-elav (Elav-9F8A9), and anti-β-galactosidase (40-1a) were obtained from the Developmental Studies Hybridoma Bank (DSHB, University of Iowa). Anti-GST was obtained from Invitrogen Molecular Probes (Eugene, OR).

Immunostaining Embryonic Cell Cultures

Embryos aged 5-6 hours (25°C; stage 9-10) were dissociated into culture media (Schneider’s Drosophila media supplemented with 18% fetal bovine serum, 100 units/mL penicillin, 100 ug/mL streptomysin) using a glass micropipet, as previously described (Tsunoda and Salkoff, 1995a). Cell cultures grown for 2 to 7 days were immunostained, as previously described (Diao et al., 2009). Briefly, cultures were fixed (4% formaldehyde), incubated with primary antibody, washed (0.1% sapponin in PBS), incubated with rhodamine or FITC-conjugated secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA; 2 hours, room temperature), washed again, and mounted in 90% glycerol with p-phenylenediamine (Sigma Aldrich, St. Louis, MO).

Transgenic UAS-myc-SIDL, UAS-GFP-Shal, UAS-GFP-ShalΔLL-motif, and UAS-UAS-GFP-Shal(LL→AA) Lines

The Gateway System was used to generate pUAST-6×myc-SIDL (referred to as UAS-myc-SIDL), pUAST-GFP-Shal2 (referred to as UAS-GFP-Shal), and pUAST-GFP-Shal2ΔLL-motif (referred to as UAS-GFP-ShalΔLL-motif) clones. SIDL was amplified by PCR from an available cDNA clone (Drosophila Genomics Resource Center, DGRC). Shal2 and Shal2ΔLL-motif were amplified by PCR from a cDNA clone of Shal2 using appropriate primers. SIDL, Shal2 and Shal2ΔLL-motif were then subcloned into the “entry vector”, pENTR1A (Invitrogen, Carlsbad, CA), such that sequences were flanked with attL1 and attL2 recombination sites. SIDL, Shal2 and Shal2ΔLL-motif were then recombined in vitro using lambda integrase into “destination” vectors (Drosophila Gateway Vector collection, DGRC), containing the attR1 and attR2 recombination sites. pUAST Drosophila transformation destination vectors have been designed to fuse 6 sequential myc epitopes (6xmyc) (pTMW) or GFP (pTGW) to the N-termini of the corresponding sequences, under the control of the upstream-activating-sequence (UAS). For UAS-GFP-Shal(LL→AA), the resulting Shal2 recombined into pTGW vector was altered so that nucleotides (1441-1446) corresponding to the two conserved leucines in the LL-motif were changed to encode two alanines; this site-directed change was performed by GenScript (Piscataway, NJ). All transgenic lines were generated by microinjection of constructs into w¹¹¹⁸ embryos and standard P-element mediated germline transformation.

Fly Brain Imaging

Adult brains were dissected in 0.1M Phosphate buffer, fixed for 20 minutes in 4% paraformaldehyde and washed in 0.1M phosphate buffer for 30 minutes. Images were captured using Zeiss LSM 510 confocal microscope and processed using Adobe Photoshop. Images are of a single frontal section approximately 60um from the antennal lobe, into the brain. For quantification in Figure 5, GFP intensity was calculated by first measuring the mean signal intensity along a line drawn over the medial antennocerebral tract (mACT) using ImageJ software (Rasband, W.S., ImageJ, U. S. National Institutes of Health; http://rsb.info.nih.gov/ij/, 1997-2009.). Background fluorescence was then measured by placing the same line over the neighboring neuropil for each antennal lobe. Mean GFP signal intensity was adjusted by subtracting the background, and then expressed as a percentage above background fluorescence.

Somato-Dendritic Localization of Shal K⁺ Channels is Conserved in Drosophila

To determine if the somato-dendritic localization of Shal K⁺ channels seen in mammalian systems is also conserved in Drosophila, we used the tissue-specific UAS-GAL4 expression system (Brand and Perrimon, 1993) to examine the localization of GFP-tagged Shal channels in an identified cell type. We chose to use the projection neurons (PNs) of the antennal lobe of the adult Drosophila brain because axonal and dendritic processes are easily discernable in vivo. PN cell bodies are localized in three major clusters (anterodorsal, lateral, and ventral) around the periphery of the antennal lobe. Like many insect neurons, PNs are unipolar and this single process extends into the antennal lobe where it forms dendritic arbors that receive input from olfactory receptor neurons, as well as an axonal projection that extends in the opposite direction to higher centers in the brain (Stocker et al., 1997; Stocker et al., 1990; Wong et al., 2002). We used the GAL4-GH146 transgene to drive expression of UAS-transgenes specifically in PNs (Stocker et al., 1997; Wong et al., 2002). For orientation, we first used GAL4-GH146 to drive expression of three different UAS-transgenes previously shown to display expression in dendrites only (Dscam[17.1]-GFP; Wang et al., 2004), axons and dendrites (Dscam[17.2]-GFP;(Wang et al., 2004), or nuclei (ms2::GFP-nls; Ashraf et al., 2006). Adult brains were dissected, fixed, and subjected to confocal microscopy to examine GFP localization in PNs. As expected, Dscam [17.1]-GFP signal was observed exclusively in the dendrite enriched glomeruli within the antennal lobe, while Dscam [17.2]-GFP was localized to axons in the antennocerebral tract as well as dendrites (Figure 1A). The ms2::GFP-nls nuclear signal illustrated the position of PN cell bodies (Figure 1A).

We then generated a GFP-tagged Shal2 construct under the control of UAS (UAS-GFP-Shal). As a control for cell morphology, we used an available transgenic line expressing UAS-DsRed. We then made transgenic lines in which GAL4-GH146 was used to drive simultaneous expression of both UAS-GFP-Shal and UAS-DsRed in the same brains. We found that GFP-Shal channels were indeed localized to dendrites and cell bodies, but not at all to axons; figure 1A shows collapsed confocal planes in order to display cell bodies, the antennal lobe (dendrites), and axonal tracts in the same image. Ds-Red in the same PNs displayed expression throughout the cell bodies, axons, and dendrites, suggesting that the localization pattern of GFP-Shal was specific, and not due to any structural abnormality. Note that collapsed confocal images also amplified background, especially in the mushroom bodies and can be seen in both GFP and DsRed images. The major axonal tract of PNs, the inner antennocerebral tract (iACT) is indicated (arrows) for clarity (Figure 1A). Figure 1B shows a higher magnification overlay of GFP-Shal and DsRed in a cluster of PN cell bodies and their proximal dendritic tracts extending into the antennal lobe. These results suggest that somato-dendritic localization of GFP-Shal is conserved in Drosophila, making Drosophila, and specifically the PNs, a viable model for investigating mechanisms underlying dendritic targeting of Shal channels.

Identification of SIDL as a Novel Interactor of the Highly Conserved Di-Leucine Motif of Shal K⁺ Channels

The Shal LL-motif is likely to bind component(s) involved in dendritic localization of Shal channels. Such an interactor may regulate the release of Shal channel subunits from the ER/trans-Golgi network, or function in the targeting, transport, retention, or recycling of Shal channels in dendrites. To take an unbiased approach to identifying protein interactors of the LL-motif, we performed a yeast-two-hybrid (Y2H) screen for candidate interactors. We used the entire 161 residue C-terminal tail of the Drosophila Shal1 isoform (Shal1C), containing the LL-motif, as “bait” and screened for interacting “prey” from a cDNA library (Diao et al., 2009). One of the interactors identified was the C-terminal 382 residues of a protein we now call SIDL (for Shal Interactor of Di-Leucine Motif). To examine the specificity of the SIDL-Shal1 interaction, we used direct Y2H assays to test whether the C-terminus of SIDL (SIDLC) was also able to interact with the C-termini of three other Drosophila voltage-gated K⁺ channels, Shaker (K_v1), Shab (K_v2), and Shaw (K_v3), and/or the C-terminus of the mouse Shal K⁺ channel, mK_v4.2. We found that SIDLC interacted with Drosophila Shal1 and mK_v4.2, but not the C-termini of the other Drosophila K⁺ channels tested (Figure 2A). These results suggest that SIDL interacts specifically, even across species, with Shal K⁺ channels.

The interaction of SIDLC with mK_v4.2C suggested that SIDL interacts with a conserved binding site on Shal channels. Using direct Y2H assays, we tested whether the binding site is the highly conserved LL-motif required for dendritic targeting. We first tested whether SIDLC was able to bind the shorter Shal2C isoform, which also contains the LL-motif. SIDLC also showed strong interaction with Shal2C. We then made deletion constructs of Shal2C as illustrated in Figure 2A. We found that deletions affecting the 16 amino acid LL-motif resulted in a loss of interaction between SIDLC and Shal. This was first seen in larger deletions of up to 40 residues of the C-terminal end of Shal2C, and then in progressively smaller deletions. A deletion of the entire LL-motif (Shal2CΔ16), or the last 9 residues of the LL-motif (Shal2CΔ9), including the two conserved leucines, resulted in a complete loss of interaction (Figure 2A). Substitution of the two leucines to alanines (Shal2C(AA)) also significantly reduced interaction with SIDLC, while substitution to more similar valine residues (Shal2C(VV)) did not. Finally, we tested whether the LL-motif alone was sufficient to bind SIDLC. We found that SIDLC displayed equally strong interaction with the 16-residue LL-motif as with the entire Shal2C (Figure 2A). Thus, our results show that the conserved LL-motif on the C-terminus of Shal is both required and sufficient for binding SIDL.

To validate the SIDL-Shal interaction using another independent approach, we used the GST-pull down assay. We generated constructs to express GST fused to Shal1C (GST-Shal1C), Shal2C (GST-Shal2C), and Shal2CΔ9 (GST-Shal2CΔ9). GST-fusion proteins were expressed and incubated with the C-terminal 382 amino acids of SIDL tagged with myc (myc-SIDLC). We found that GST-Shal1C and GST-Shal2C indeed pulled-down myc-SIDLC, while GST alone and GST-Shal2CΔ9 did not (Figure 2B). To test the GST pull-down assay in the opposite configuration, we fused SIDLC to GST (GST-SIDLC) and incubated GST and GST-SIDLC with embryonic membrane extracts this time. We found that GST-SIDLC indeed pulled-down native Shal protein, while GST alone did not (Figure 2C).

SIDL is Primarily Expressed in the Nervous System

SIDL, which has not previously been described, contains 1146 amino acids, with no consensus protein binding or enzymatic domains. Sequence analysis using multiple transmembrane prediction programs resulted in conflicting predictions. Some programs predicted SIDL to be a soluble protein while others predicted an integral membrane protein. Two of the Kyte-Doolittle based prediction programs (waveTM, Liakopoulos et al., 2001), available at http://athina.biol.uoa.gr/bioinformatics/waveTM/, and TopPred2, (Claros and von Heijne, 1994), available at http://bioweb.pasteur.fr/seqanal/interfaces/toppred.html) predicted SIDL to contain one to two transmembrane domains at 787-807 (amino acids) and possibly also 1079-1099 (amino acids; see Figure 3A).

To examine the expression of SIDL at different developmental stages, we performed RT-PCR for SIDL from embryos, at 5, 12, and 17 hours after egg laying (AEL), larvae in all three instar stages, pupae at 1 day and 4 days post-pupariation, and adult fly heads and bodies. We found that SIDL is expressed at all of these developmental stages (Figure 3B). To examine whether SIDL is expressed in the nervous system, we examined the expression of SIDL in the developing embryo, where Shal channels have been shown to underlie the transient A-type K⁺ current present in virtually all neurons (Tsunoda and Salkoff, 1995a). We performed RNA in situ hybridizations in wild-type embryos (12 hours AEL) using a single-stranded digoxygenin-labeled probe for SIDL. We found that SIDL displayed expression primarily in the nervous system, similar to the nervous system specific elav protein (Figure 3C).

SIDL Co-Localizes with Shal K⁺ Channels in Neurons

To determine if SIDL co-localizes with Shal channels at distinct subcellular sites, we examined neurons in primary cultures from transgenic lines. We first used the neuron-specific elav-GAL4 transgene to drive expression of UAS-GFP-Shal alone. We found that GFP-Shal protein was localized in neuronal cell bodies and in distinct puncta along neuronal processes (Figure 3D, 3F; Diao et al., 2009). As a control, we examined a transgenic line expressing the membrane targeted CD8-GFP fusion protein under the control of the same GAL4 neuronal driver. In contrast to GFP-Shal, CD8-GFP was localized uniformly throughout neuronal processes (Figure 3D), suggesting that GFP-Shal puncta in neuronal processes are indeed distinct sites of localization. To confirm that GFP-Shal puncta are indeed localized in dendrites, we prepared primary cultures from embryos expressing GFP-Shal and the fusion protein, nod-ß-Gal, which localizes only in dendrites (Andersen et al., 2005; Clark et al., 1997; Lee et al., 2000; Rolls et al., 2007; Yamamoto et al., 2006). Indeed, GFP-Shal positive processes were also positive for nod-ß-Gal (Figure 3E).

We next examined whether SIDL co-localizes with GFP-Shal in these dendritic puncta. We generated transgenic lines expressing both UAS-GFP-Shal and UAS-6xmyc-tagged SIDL (UAS-myc-SIDL) in neurons, using the elav-Gal4 driver. We found that GFP-Shal and myc-SIDL co-localized both in neuronal cell bodies and in puncta along neuronal processes (Figure 3F). Nearly all GFP-Shal positive puncta were also positive for myc-SIDL; there were a minority of myc-SIDL positive puncta that did not appear to contain GFP-Shal. One possibility is that SIDL may function in clustering Shal channels at these sites. The co-localization of GFP-Shal and myc-SIDL in dendritic puncta further implicated a role for SIDL in Shal K⁺ channel regulation.

Disruption of the Shal-SIDL Interaction Results in Axonal Localization of Shal Channels in a Subset of Projection Neurons in the Adult Brain

To test whether the Shal-SIDL interaction is required for somato-dendritic localization of Shal K⁺ channels, we replicated mutations in Shal that resulted in a loss of the Shal-SIDL interaction in Y2H assays (Figure 2) and created transgenic lines expressing these Shal variants tagged with GFP. The two transgenic lines we generated included one expressing UAS-GFP-Shal with a deletion of the 16-residue LL-motif (UAS-GFP-ShalΔLL-motif), and another with the two central leucines of the LL-motif changed to alanines (UAS-GFP-Shal(LL→AA)). We then used GAL4-GH146 to drive expression of UAS-GFP-ShalΔLL-motif and UAS-GFP-Shal(LL→AA) in PNs. In case any observed effects might be due to gross morphological changes in the PNs, we simultaneously expressed UAS-DsRed in the same fly brains with each of the UAS-GFP-Shal variants. Whole brains were dissected, fixed and imaged by confocal microscopy. Analysis of confocal planes appropriate to examine PN clusters and dendrites in the antennal lobe, showed that GFP-Shal was localized in both cell bodies and dendrites, as described earlier. Figure 4 shows confocal images collapsed to display PN cell body clusters, the antennal lobe with PN dendrites, and the major iACT PN axonal tract. Interestingly, however, GFP-ShalΔLL-motif and GFP-Shal(LL→AA) channels displayed a distinctly punctate localization pattern in both cell bodies and in the antennal lobe (Figure 4). It is unlikely that this was due to a defect in PN morphology since DsRed was present diffusely throughout PNs in GFP-Shal, GFP-ShalΔLL-motif, and GFP-Shal(LL→AA) expressing transformants (Figure 4). The clustering of GFP-ShalΔLL-motif and GFP-Shal(LL→AA) channels may indicate a role for the Shal-SIDL interaction in maintaining channel distribution.

We were surprised that GFP-ShalΔLL-motif and GFP-Shal(LL→AA) channels were excluded from axons in the iACT. To investigate whether these GFP-Shal mutants were excluded from all PN axons, we examined single confocal planes through the brain that captured PN axons projecting in the middle ACT (mACT). PNs extend axons to higher centers in the brain in three antennocerebral tracts (ACTs): inner (iACT), middle (mACT), and outer (oACT). The iACT axons project from cell bodies that reside in anterodorsal and lateral PN clusters to the calyx of the mushroom body and the lateral horn (LH) in the protocerbrum (Figures 1,4,5; see arrows), while mACT axons project from cell bodies that reside in the ventral PN cluster directly to the LH (Figure 5, see arrowheads; Okada et al., 2009; Stocker, 1994; Stocker et al., 1990). oACT axons project from anterodorsal and lateral cluster PNs to the LH and calyx by way of an outer tract. We did not observe expression in the oACT with the GH146-Gal4 driver, most likely because the GH146-Gal4 only expresses Gal4 in a single PN that projects via the oACT (Wong et al., 2002) and it was not detectable in our hands. GH146-Gal4 clearly promoted expression in PNs with axons in the iACT, as shown in Figures 1 and 4, and the mACT, as shown in Figure 5. GFP-Shal, however, was never seen to extend into either tract (Figures 1,5), consistent with an exclusively somato-dendritic localization pattern. Interestingly, in confocal planes used to image the mACT, we also observed GFP-Shal in the very proximal segment of the iACT axons just where they bundle to leave the antennal lobe (Figure 5, see arrows). One possibility is that Shal channels are excluded from axons by a diffusion barrier present in this proximal segment, similar to mechanisms proposed in mammalian neurons (Winckler et al., 1999).

When we examined the localization of GFP-ShalΔLL-motif and GFP-Shal(LL→AA) channels in PNs, we found that although these mutant channels were not present in the distal iACT (Figure 4), they were present in the proximal segment of the iACT, similar to GFP-Shal (Figure 5). In contrast, GFP-ShalΔLL-motif and GFP-Shal(LL→AA) channels were consistently found throughout the mACT (Figure 5, see arrowheads). Thus, mislocalization to axons appeared to occur only in those PNs with axons that project into the mACT, but not into the iACT. Quantification of GFP signal intensity along the mACT, expressed as percent above background, showed that GFP-ShalΔLL-motif and GFP-Shal(LL→AA) displayed 78.9% and 68.4% signal in mACT over background, respectively, compared with only 3.2% over background for GFP-Shal (Figure 5C). Our studies show that disruption of the LL-motif, and therefore the Shal-SIDL interaction, result in axonal targeting in the subset of PN neurons, those that reside in the ventral cluster, and project axons in the mACT tract. These results suggest then that there are likely to be multiple mechanisms for targeting Shal K⁺ channels in different neurons. The role of other potential Shal interactors, such as SKIP3 (Diao et al., 2009) and a putative KChIP homolog, in trafficking will need to be explored in the future as well.

Mislocalization of GFP-ShalΔLL-motif Channels in Sensory Neurons

To test whether the localization patterns of GFP-Shal variants showed any differences in another system, we used the ppk-Gal4 transgene to drive expression of GFP-Shal variants in identified neurons in the peripheral nervous system (PNS). In the Drosophila PNS, the dendritic arborization (da) neurons are present in four clusters (dorsal (d), lateral (l), ventral’ (v’), and ventral (v)) and project multiple dendrites along the epidermis. The ppk-Gal4 transgene drives expression in class IV da neurons (Grueber et al., 2003; Grueber et al., 2007). Class IV da neurons display the most extensive array of dendritic arborization of the da neurons (Grueber et al, 2002), and three class IV da neurons, named ddaC, v’ada, vdaB, have been identified in each abdominal hemisegment (A2-A6), present in the dorsal, ventral’, and ventral clusters, respectively. da neurons are easily examined in the late embryo through larval stages, where their cell bodies and dendritic projections are found just beneath the surface of a rather transparent cuticle. In contrast, da axons bundle and project ventrally to the ventral nerve cord (VNC) of the CNS, which is also easily imaged in larvae. We examined the localization of mCD8-GFP and GFP-Shal in dendrites and axons of class IV da neurons in live 2^nd instar larvae. mCD8-GFP was found throughout dendrites and in axons projecting to the VNC, as expected (Figure 6). Surprisingly, GFP-Shal was localized primarily to cell bodies in these neurons, with little to no signal in dendrites (Figure 6). It is unclear whether this is representative of endogenous Shal localization in these neurons. When we examined axons extending to the VNC, no GFP-Shal signal was evident here (Figure 6), as expected. GFP-ShalΔLL-motif channels, however, were present in both cell bodies and axons of class IV da neurons (Figure 6). GFP-ShalΔLL-motif signal was consistently observed in a punctuate pattern in axons extending from the cell bodies (Figure 6, top-right, see arrows) and clearly in the VNC (Figure 6, bottom-right). These results suggest that exclusion of Shal K⁺ channels from axons in class IV da neurons is also dependent on the LL-motif, and perhaps its interaction with SIDL.