Heterotrimeric Go protein links Wnt-Frizzled signaling with ankyrins to regulate the neuronal microtubule cytoskeleton

Drosophila neuromuscular junctions (NMJs) represent a powerful model system with which to study glutamatergic synapse formation and remodeling. Several proteins have been implicated in these processes, including components of canonical Wingless (Drosophila Wnt1) signaling and the giant isoforms of the membrane-cytoskeleton linker Ankyrin 2, but possible interconnections and cooperation between these proteins were unknown. Here, we demonstrate that the heterotrimeric G protein Go functions as a transducer of Wingless-Frizzled 2 signaling in the synapse. We identify Ankyrin 2 as a target of Go signaling required for NMJ formation. Moreover, the Go-ankyrin interaction is conserved in the mammalian neurite outgrowth pathway. Without ankyrins, a major switch in the Go-induced neuronal cytoskeleton program is observed, from microtubule-dependent neurite outgrowth to actin-dependent lamellopodial induction. These findings describe a novel mechanism regulating the microtubule cytoskeleton in the nervous system. Our work in Drosophila and mammalian cells suggests that this mechanism might be generally applicable in nervous system development and function.

3 were raised using the recombinant protein purified from bacteria (Kopein and Katanaev, 2009); the antiserum was used at 1:100 for immunostaining. Specificity of this antiserum was confirmed by western blots on Drosophila head extracts, as well as by immunostaining of wing imaginal discs. Secondary antibodies were HRP labeled for western blots (1:4000) or Cy3-and Cy5labeled in immunostaining (1:400 in PBT, 2h incubation at room temperature). The preparations were mounted in Vectashield (Vector Labs), dorsal side up.

Microscopy and analysis of NMJs
The well-characterized NMJs of muscle 6/7 in segment 2-4 were analyzed in all experiments.
Maximally, two segments per animal (e.g. segment A3 and A4, both in the same hemisphere, or both segments A3 in the two hemispheres) were analyzed. NMJs were imaged with a confocal microscope (Zeiss LSM 510 or Zeiss LSM710). For statistical analysis, one optical slice with a thickness of 2.3 µm was taken with a 20x or a 25x objective in the optical plane of each NMJ and the boutons were measured manually with the help of the program AxioVision 4.7 (Zeiss). A bouton was identified by the CD8-GFP-Sh, anti-Dlg and/or anti-HRP staining as a circular or slightly oval structure with clear borders, connected by neurites to the neighboring bouton; all these methods resulted in identical bouton quantifications. Type 1b boutons were distinguished from type Is by more intense anti-Dlg staining and their larger size (Packard et al., 2002). Bouton number values are depicted as percentage of the respective control. The length of the NMJ was measured from the first to the last bouton along the synaptic cleft and all side branches on the muscle surface with more than three boutons were measured and added to the total length of the NMJ. The lengths of the NMJ slightly varied depending on the phenotype, in agreement with (Mathew et al., 2005) (Fig. S2E).
To confirm that the OK371-Gal4/UAS-RNAi system was efficient to downregulate Gαo, Wg and Fz2, respective immunostainings of wild-type and RNAi-expressing NMJs were Development | Supplementary Material 4 performed in parallel. NMJs were imaged with a LSM710 (Zeiss) confocal microscope using identical settings for all images. Quantification of the fluorescence was performed with ImageJ (NIH). The presynaptic cell was outlined with the freehand selection tool following the borders of the staining and the mean value of the fluorescence of this area was measured with the measure tool. A noticeable downregulation in the levels of the respective proteins was achieved ( Fig.   S1I,M,O), and quantification revealed a ~50% decrease in anti-Gαo/Wg/Fz2 staining in the NMJ (Fig. S1J,N,P). However, this is likely to be a gross underestimation of the efficiency of the RNAi-mediated downregulation: using a pan-neuronal driver (elav-Gal4), we find a comparable decrease in anti-Gαo immunostaining (Fig. S1K), but in western blots on whole-head extracts of the control versus the RNAi-Gαo constructs, a dramatic decrease in Gαo levels could be seen ( Fig. S1L). Mouse anti-tubulin (Sigma, 1:2500) staining served as loading control.

Electrophysiology and muscle contraction
ChR2-mediated stimulation of synaptic potentials was performed as described (Schroll et al., 2006;Hornstein et al., 2009). Larvae expressing ChR2 in motoneurons using the driver OK371-Gal4 were grown on standard corn food supplemented with 1 mM all-trans-retinal (Sigma) at 25°C in the dark. Wandering third instar larvae were dissected in cold Ca 2+ -free HL-3 saline (Zhang and Stewart, 2010) and washed three times in cold HL-3 supplemented with 1.5 mM CaCl 2 before performing the measurements in same buffer. Intracellular potentials were recorded in body wall muscles 6/7 using a pipette with a resistance of 15-30 MΩ when filled with 1 M KCl. To evoke single action potentials, animals were stimulated by a 20 ms light pulse of 470 nm using a high-power LED placed 10 cm from the larvae (light pulse triggered at 1.2V, Thorlabs) controlled by Chart Master software (HEKA). Electrophysiological signals were pre-amplified (without filtering) using the LPF-8 signal conditioner (Warner Instruments) and the 50 Hz noise was reduced using the HumBug (Quest scientific) noise reducer. Finally, analog signals were Development | Supplementary Material measured using a KS-700 amplifier (World Precision Instruments) then digitized using LIH8+8 (HEKA). Data were acquired using the Chart Master software at a sampling frequency of 20 kHz.
The data were low-pass filtered at 2 kHz before analysis of EJPs and mEJPs with the Mini analysis program (Synaptosoft). The threshold for detection of peaks was set to 0.3 mV. For analysis only muscles with a resting potential more negative than -46 mV were used. Note that the optogenetic measurements used here and the traditional electrophysiological recordings produce identical EJP amplitudes when measured side-by-side (Pulver et al., 2011) (note also that, in this particular work, utilizing the same OK371-Gal4 driver as used by us, the EJP amplitude in the wild-type is measured as ~12.5 mV, very similar to our measurement of 11 mV; see Fig. 1H). Furthermore, when performing our own optogenetic measurements, we sometimes (rarely) observed non-stimulated, spontaneous action potentials which were of the same amplitude as the light-induced ones (such an example is shown on Fig. S1T).
For the locomotion test, third instar larvae were placed on a 1% agarose plate and allowed to adjust for 1 min. The number of whole body contractions per minute was counted.

Yeast two-hybrid screen
Isoform II of Drosophila Gαo was used as the bait, and the cDNA library from Drosophila head was used as the prey in the screening custom-performed by Hybrigenics (Paris, France). Fifty four million clones were screened and analyzed as described (Kopein and Katanaev, 2009). The three clones of Ank2 representing partial open reading frames each had the high confidence interaction score ]biological significance score (Formstecher et al., 2005)] (B, E-value < 1e-5).
Transformed cells were grown at 37°C until OD 600 = 0.7, cooled to 17°C before induction with 0.1 mM IPTG and subsequent growth overnight at 17°C, and then harvested by 15 min centrifugation at 4000 g. The pellet was resuspended in column buffer 200 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mg/ml lysozyme, 1x Complete protease inhibitor cocktail (Roche)] and incubated on ice for 30-60 min before sonification lysis, followed by centrifugation for 30 min at 16,000 g to remove cell debris. The protein was bound to amylose resin (New England BioLabs) by incubation at 4°C for at least 1 h. The amylose beads were washed three times for 10-15 min with column buffer and the protein was eluted with 10 mM maltose in column buffer. Control MBP was prepared using the pMAL-c2x plasmid in parallel.

Pull-down assays
Thirty micrograms of His 6 -Gαo were incubated with a twofold molar excess of MBP or MBP-Ank2_12 in HKB* buffer (100 mM KCl, 50 mM HEPES-KOH, 10 mM NaCl, 5 mM MgCl 2 , 2 mM EGTA, 1 mM DTT, 5% glycerol, 0.5% NP40, 0.1% Tween) at 17°C for 1.5 h, prior to addition to 100 µl 50% amylose resin slurry (New England BioLabs) pre-equilibrated with HKB* for an additional 1.5 h incubation at 17°C. The resin was washed four times with 1.5 ml HKB* for 15 min. The retained proteins were eluted by a 15 min incubation with 50 µl 10 mM maltose in HKB*. The proteins were resolved on 10% SDS-PAGE, electrotransferred to nitrocellulose membranes (Whatman) and detected by immunoblotting using rabbit anti-Gαo/i at 1:1000 (Merck). Equal loading was ensured by immunoblotting using rabbit anti-MBP antibodies at 1:4000 (New England Biolabs). For experiments with Gβγ, His6-Gαo was pre-incubated with the βγ dimer purified from porcine brains (Koval et al., 2010) for 45 min at room temperature before addition of equimolar amounts of MBP-Ank2_12 or MBP and incubation for an additional 1.5 h at 4°C. The pre-equilibrated resin was added to the proteins, incubated for 1.5 h and washed as described above. Proteins were eluted by addition of 5x sample buffer and boiling.

GTP-binding assay
Poorly hydrolysable fluorescent GTP analog Eu-GTP (PerkinElmer) was used in the GTPbinding assay with purified His 6 -tagged Drosophila Gαo and Gαo[G203T] as described (Koval et al., 2010). The indicated (Fig. S2C) concentrations of the proteins were incubated for 2 hours in the presence of 5 nM GTP-Eu in 1xHKB buffer (10 mM HEPES-NaOH, 135 mM KCl, 10 mM NaCl, 2 mM EGTA, pH 7.5) supplemented with 5 mM MgCl 2 . The reaction mixtures were subsequently transferred to AcroWell BioTrace NT 96-well plates (Pall), filtered using a vacuum manifold, and the membranes were washed twice with ice-cold washing buffer (20 mM Tris-HCl, 0.1 mM MgCl 2 , pH 8.0). Fluorescence of the label retained on the membranes was measured immediately in a Victor 3 multilabel counter (PerkinElmer) in TRF mode. All experimental points were measured in duplicate. Curve fitting was performed in Prism 5 software (GraphPad).

Development | Supplementary Material
Mouse neuroblastoma N2a cells were cultured in MEM supplemented with 10% FCS, Lglutamine and penicillin/streptomycin (all from Gibco, Life Technologies). Vector transfections were carried out with X-tremeGENE 9 (Roche) according to the manufacturer's instructions.

Neurite outgrowth assay
N2a cells were co-transfected for 24 h with pEGFP-C1 (Clontech) and pcDNA3.1+ (Invitrogen) or a plasmid encoding human Gαo (Missouri S&T cDNA Resource Center). Cells were trypsinized and seeded on poly-L-lysine-coated coverslips for an additional 24 h to allow neurite formation. Alternatively, cells were transfected with EGFP-tagged ankyrin-B or ankyrin-G vectors (Ayalon et al., 2008) or co-transfected with the Gαo plasmid and prepared as above.
Transient ankyrin double knockdowns were obtained by co-transfection of the shRNA-stably transfected N2a cell lines with the shRNA vectors shluc, shankB or shankG in addition to the plasmids described above. Ankyrin-independent neurite outgrowth was analyzed using an EGFPtagged MARK2 (PAR1b) plasmid (Nishimura et al., 2012) under the transient double ankyrin knockdown conditions described above. Additionally, an mRFP-tagged MARK2 plasmid was generated by subcloning the BglII-KpnI fragment including the MARK2 sequence into the same sites of the pmRFP-C1 vector. Then, N2a cells were co-transfected with the mRFP-MARK2 and AnkB-GFP plasmids and prepared as above. AnkB-GFP mean fluorescence intensities at the rear end of neurites (10-20 µm 2 ) and at whole neurites were scored from 40-80 neurites per condition using ImageJ, and the ratio values were used to determine AnkB-GFP accumulation at neurite tips in cells co-transfected with control pcDNA3.1+, Gαo or mRFP-MARK2 plasmids. For Nocodazole treatment, transfected N2a cells were allowed to adhere on coverslips for 6 h before incubation for an additional 18 h with Nocodazole (Sigma-Aldrich) in normal medium at the concentrations indicated in the corresponding figures. Cells were finally fixed with paraformaldehyde, stained with phalloidin-Rhodamine (Molecular Probes, Life Technologies) and DAPI (Sigma-Aldrich) or anti-Gαo antibody and mounted for microscopy analysis. Samples were recorded with an α-Plan-Apochromat 63x/1.4 or a Plan-Neofluar 20x/0.50 objective on an