Sushi domain-containing protein 4 controls synaptic plasticity and motor learning

Figures and figure supplements Sushi domain-containing protein 4 controls synaptic plasticity and motor learning Ine ´s Gonza ´lez-Calvo et al Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 1 of 26 RESEARCH ARTICLE A CCP TM NTCT CCPCCP CCP C Accelerating Rotarod 123123123123123 0 100 150 200 250 Time to fall (sec.) DAY1 DAY2 DAY3 DAY4 DAY5 WT (n=11) KO (n=7) Extracellular Intracellular p=0.0079 50 Sushi domain-containing protein 4 Susd4 131.7 kb 1 UTR Deletion 5’ 3’ 8765432 UTR E F WT KO CTX BS CB CTX BS CB G 0 10 20 30 40 Cerebellum Brainstem P0 P3 P7 P14 P21 3mo Susd4 expression (relative to Rpl13a) B AAV2-hSYN-DIO-HA-SUSD4-2A-eGFP SUSD4 GluD2 SUSD4 GFP GluD2 SUSD4 CTL GluD2 CTL GFP GluD2 CTL AAV2-hSYN-DIO-eGFP D SUSD4 GFP SUSD4 GFP SUSD4 GFP Figure 1. SUSD4 is necessary for motor coordination adaptation and learning. (A) Diagram of the protein SUSD4 showing its domain organization with four extracellular Complement Control Protein (CCP) domains, one transmembrane (TM) domain and a cytoplasmic domain (C T ). (B) Quantitative RTPCR shows an increase in Susd4 mRNA expression (relative to the housekeeping gene Rpl13a) during postnatal development in the cerebellum and in the brainstem. Extracts were prepared from tissue samples of mice aged from 0 to 21 days (P0-21) and 3 months (3mo). Mean ±s.e.m. (n = 3 independent experiments). (C) HA-tagged SUSD4 is found in dendrites (left panel, single plane) and in some of the distal dendritic spines (right panel, arrowheads, projection of a 1.95 mm z-stack) in adult cerebellar Purkinje cells. Anti-HA and anti-GFP immunolabeling was performed on parasagittal cerebellar sections obtained from adult L7Cre mice after stereotaxic injection of AAV particles driving the expression of HA-SUSD4 and soluble GFP. Figure 1 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 2 of 26 Research article Neuroscience Figure 1 continued Scale bars: 10 mm (left panel) and 2 mm (right panel). (D) Purkinje cells from primary mixed cerebellar cultures of L7Cre mice were transduced at 3 days in vitro (DIV3) with an HA-tagged SUSD4 expressing virus (AAV2-hSYN-DIO-HA-SUSD4-2A-eGFP) or with a control virus expressing GFP alone (AAV2- hSYN-DIO-eGFP), and immunostained in non-permeabilizing conditions at DIV17 for HA to localize surface SUSD4 (anti-HA, red), and in permeabilizing conditions to detect the green fluorescent protein (anti-GFP, green) and the endogenous GluD2 subunit (anti-GRID2, blue). Scale bar 5 mm. (E) Genomic structure of the Susd4 gene. White boxes represent exons. Exon 1 is deleted in the Susd4 loss-of-function mouse model. See also Figure 1— figure supplement 2. (F) Motor coordination and learning is deficient in adult male Susd4 -/- (knockout [KO]) mice compared to age-matched Susd4 +/+ (wild-type [WT]) littermates. Each mouse was tested three times per day during 5 consecutive days on an accelerating rotarod (4-40 r.p.m. in 10 min) and the time spent on the rotarod was measured. Mean ±s.e.m. (WT n = 11 and KO n = 7 mice, two-way ANOVA with repeated measures, interaction (time and genotype): **p=0.0079, F(14, 224)=2.22; time: ****p<0.0001, F(14, 224)=3.469; genotype: p=0.0553, F(1, 16)=4.272). (G) In situ hybridization experiments were performed on brain sections from 1-month-old Susd4 WT and Susd4 KO mice to detect Susd4 mRNA using a probe encompassing exons 2–5 (see also Figure 1—figure supplement 2). Susd4 expression was found in many regions of the brain in Susd4 WT mice (see also Figure 1— figure supplement 1) including the cerebral cortex (CTX), the cerebellum (CB), and the brainstem (BS). No labeling was found in the brain of Susd4KO mice. Scale bar 500 mm. Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 3 of 26 Research article Neuroscience A B P0 P7 CTX IO CB DG CA3 CTX IO CB DG CA3 CTX C IO CB PCs P30 Figure 1—figure supplement 1. Susd4 mRNA expression in the developing mouse brain. (A)Susd4 mRNA expression was visualized in the brain of wild-type mice by in situ hybridization. Coronal (left) and sagittal (right) Figure 1—figure supplement 1 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 4 of 26 Research article Neuroscience Figure 1—figure supplement 1 continued sections are presented at postnatal day 0 (P0), (B) postnatal day 7 (P7), and (C) postnatal day 30 (P30). Susd4 expression was found in many regions including the cerebral cortex (CTX), the dentate gyrus (DG) and CA3 regions in the hippocampus (coronal section, left), the cerebellum (CB), in particular Purkinje cells (PCs), and the inferior olive (IO; sagittal section, right). Scale bars 250 and 500 mm (inset C). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 5 of 26 Research article Neuroscience WT Genotyping TaqMan 1UTR SP 5’ 3’ UTR UTR 1 8765432 Susd4 131697 bpA BWT KO 800bp 300bp CTX CB BS CTX CB BS Wild-type allele Knockout allele Selection Cassette KO Genotyping Figure 1—figure supplement 2. Characterization of Susd4 knockout (KO) mice. (A) Structure of the Susd4 gene and strategy for the generation of the KO mouse. The gene coding for the Susd4 mRNA contains eight exons. The wild-type (WT) allele is presented indicating the localization of the primers used for genotyping and of the probes used for TaqMan RT-qPCR. In the KO allele, the 5’UTR and first exon are entirely deleted and replaced by the selection cassette. (B)Susd4 expression was assessed by RT-PCR using primers encompassing exons 6-8 in extracts from cortex (CTX), cerebellum (CB), and brainstem (BS) in Susd4 WT and Susd4 KO mice. Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 6 of 26 Research article Neuroscience FORELIMB HINDLIMB WT KO Print Separation Stride Stance WT KO 0 2 4 6 8 WT KO 0 2 4 6 8 WT KO 0 1 2 3 4 5 WT KO 0 1 2 3 4 5 WT KO 0.0 0.5 1.0 1.5 2.0 P=0.0148 Forelimb stride (cm) Hindlimb stride (cm) Forelimb stance (cm)Hindlimb stance (cm) Print separation (cm) Footprint Test Figure 1—figure supplement 3. Footprint analysis in Susd4 knockout (KO) mice. Footprint patterns of P30 Susd4WT and Susd4 KO mice were quantitatively analyzed by measuring stride length for the fore paws (magenta) and hind paws (cyan), stance length for the forelimbs and hindlimbs, and print separation. Mean ±s.e.m. (WT n = 9 and KO n = 10 mice; unpaired Student’s t-test; forelimb stance: p=0.3059; forelimb stride: p=0.5882; hindlimb stance: p=0.4533; hindlimb stride: p=0.3580; print separation: *p=0.0148). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 7 of 26 Research article Neuroscience WT KOWT KO Area of cerebellum (mm2) Extent of Molecular Layer (µm) WT KO WT KO 0 25 50 75 100 125 150 175 0 2 4 6 8 10 CABP A B HOECHST WT KO Figure 1—figure supplement 4. Normal cerebellar cytoarchitecture in Susd4 knockout (KO) mice. (A) Parasagittal sections of P30 Susd4 wild-type (WT and Susd4 KO cerebella were stained with Hoechst and used for quantitative analysis of the mean area of the cerebellum). Mean ±s.e.m. (n = 3 WT mice and n = 3 KO mice). Scale bar 500 mm. (B) Calbindin protein (CABP) immunostaining was used for quantitative analysis of the mean height of the molecular layer. Mean ±s.e.m. (WT n = 5 and KO n = 6 mice). Scale bar 30 mm. Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 8 of 26 Research article Neuroscience AB C WT KO 100 μV 100 ms 100 μV 100 ms 0.0 0.5 1.0 1.5 2.0 Coefficient of Variation (CV) 0 10 20 30 40 Mean firing rate (spikes per second) 0.0 0.2 0.4 0.6 0.8 1.0 WT KO WT KO WT KO CV2 Figure 1—figure supplement 5. High-density microelectrode array (MEA) analysis of Purkinje cell spiking in acute cerebellar slices from Susd4 knockout (KO) compared to Susd4 wild type (WT). (A) Image of a cerebellar acute slice from a WT mouse overlapped with the image of the color map of the MEA recording. Each pixel represents one channel, where the active units are in red. The black square highlights one of the channels. (B) Representative traces of electrical activity recorded in one channel from Susd4 WT and KO mice. Each tick points out one action potential that has been detected and sorted by the Brainwave software. (C) Histograms of the mean firing rate, coefficient of variation (CV) of interspike intervals, and CV2. Mean ±s.e.m. (WT n = 5 and KO n = 4 mice; Mann-Whitney test; mean firing rate: p=0.2857; CV: p=0.4127; CV2: p=0.5373). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 9 of 26 Research article Neuroscience GFPCABP A Number of STEPS Percent of cells 0 20 40 60 80 100 1 >1 26/9 26/7 WT KO BCF synapse elimination Susd4 WT Susd4 KO CF EPSCs kinetics 0 10 20 30 40 50 0 1 2 3 4 5 Decay time (ms) Rise time (ms) KOWT WT KO C D 0 1 2 3 4 5 0 1 2 3 Time to peak (ms) Rise Time 10% to 90% (ms) WT KO CF mEPSCs kinetics WT KO Area CF (mV*ms) Area hyperpolarization (mV*ms) 0 20000 40000 60000 80000 100000 Sweep # -150000 -100000 -50000 0 Sweep # CF-EPSP area (mV*ms) CF-EPSP hyperpolarization area (mV*ms) 300287274 261 248235222209196183170157144131118105927966534027141 300287 274 261248 235 222209 196183 170 157 144131118105927966534027141 WT (16/11) KO (14/10) WT (16/11) KO (14/10) 0 2 4 6 8 10 Number of Spikelets WT (16/11) KO (14/10) Long Term Depression induction (Current Clamp)E 20mV −60mV Area CF Area hyperpolarization 100ms WT KO * ** * * 20ms 20mV CF STIM Figure 3—figure supplement 1. Characteristics of the climbing fiber (CF)/Purkinje cell (PC) synapse. (A) CFs were visualized in Susd4 wild-type (WT) and knockout (KO) mice crossed with Htr5b-eGFP reporter mice expressing the green fluorescent protein (GFP) specifically in inferior olivary neurons. Anti-GFP and anti-CABP (to visualize PCs) immunolabeling was performed on parasagittal sections of P30 mice and showed no qualitative differences Figure 3—figure supplement 1 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 16 of 26 Research article Neuroscience Figure 3—figure supplement 1 continued in the absence of Susd4 expression. Scale bar 10 mm. (B) Patch-clamp recordings of PCs showed a similar percentage of mono- (one climbing fiber) and multi-innervation (>1 CFs) of PCs in P30 Susd4 KO and WT mice, as measured by the number of steps elicited by electrical stimulation of the CFs (WT n = 26 cells from nine mice and KO n = 26 cells from seven mice; chi-square test, p=0.5520). (C) No change in the rise and decay times of CF/PC excitatory postsynaptic currents (EPSCs) was induced by Susd4 deletion. Mean ±s.e.m. (WT n = 26 cells from nine mice and KO n = 26 cells from seven mice; rise time: unpaired Student’s t-test, p=0.3750; decay time: Mann-Whitney test, p=0.7133). (D) No change in the time to peak and in the rise time of CF/PC synapse-delayed EPSC quanta was induced by Susd4 loss-of-function. Mean ±s.e.m. (WT n = 10 cells from six mice and KO n = 8 cells from three mice; time to peak: unpaired Student’s t-test, p=0.1692; rise time 10-90%: Mann-Whitney test, p=0.0639). (E) CF/PC transmission was monitored during long-term depression (LTD) induction in current-clamp mode. Representative traces of the CF-EPSP are shown. The quantification of the number of spikelets (spikes induced by repeated CF stimulation [Mathy et al., 2009;Titley et al., 2019]) did not reveal any difference between WT and KO PCs (unpaired Student’s t-test, p=0.5791). Bottom: Measurements of the CF-EPSP area and hyperpolarization area during the LTD induction protocol. Mean ±s.e.m. (WT n = 16 cells from 11 mice and KO n = 14 cells from 10 mice). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 17 of 26 Research article Neuroscience WT KO GluA2 ßACTIN +Leupeptin +MG132 CTL K-Glu CTL K-Glu CTL K-Glu CTL K-Glu 1.0 0.8 0.6 0.4 0.2 0.0 300020001000 Mean intensity of GluA2 puncta WT (n=5) KO (n=5) Cumulative probability 40000 Number of GluA2 puncta per VGLUT2 bouton GluA2 VGLUT2 WT KO WT KO 0.0 0.1 0.2 0.3 0.4 WT KO Fraction of total number p<0.0001 01 2 3≥4 D A C GluA2 clusters around climbing fiber boutons WT KOWT KO P=0.0107 0.0 0.5 1.0 1.5 2.0 Total GluA2 protein levels after cLTD induction (relative to control) P=0.3755 P=0.3176 P=0.2338 B WT KO 0.0 0.5 1.0 1.5 2.0 Surface GLUA2 protein levels after cLTD induction (relative to control) P=0.0212 P=0.0538 WT KO CTL K-Glu CTL K-Glu GluA2 GluA2 ßACTIN IP: INPUT: Cultured Purkinje Cells infected with AAV2-hSYN-DIO-HA-SUSD4-2A-eGFP 5 µm SUSD4 SUSD4 + GluA2 SUSD4 + GluA2 + GFP GluA2 SUSD4 Binary Images GluA2 Colocalization GFP + Colocalization Figure 4. Loss of SUSD4 leads to misregulation of the AMPA receptor subunit GluA2. (A) The number of GluA2 clusters (anti-GluA2 immunolabeling, green) per climbing fiber presynaptic bouton (anti-VGLUT2 immunolabeling, red) and their intensity were quantified in cerebellar sections of juvenile Susd4 knockout (KO) mice and Susd4 wild-type (WT) littermates. Cumulative plot for the mean GluA2 intensity per VGLUT2 bouton shows no significant change between WT and KO. The distribution of the VGLUT2 boutons according to the number of associated GluA2 clusters is significantly different between WT and KO. Mean ±s.e.m. (WT n = 5 and KO n = 5 mice; intensity: Kolmogorov-Smirnov test, p=0.5009; distribution: chi-square contingency test, ****p<0.0001). Scale bars 30 mm (top) and 15 mm (bottom). (B) Activity-dependent changes in surface localization of GluA2 was studied in cerebellar acute slices from Susd4 KO mice and control Susd4 WT littermates using a chemical LTD protocol (cLTD; K-Glu: K + 50 mM and glutamate 10 mM for 5 min followed by 30 min of recovery). Surface biotinylation of GluA2 subunits was followed by affinity purification of biotinylated GluA2 subunits and anti-GluA2 immunoblot analysis. The fraction of biotinylated GluA2 was obtained by measuring the levels of biotinylated GluA2 in affinity-purified samples and total GluA2 normalized to bACTIN in input samples for each condition. The ratios between the fraction of biotinylated GluA2 after cLTD and control conditions are represented. Mean ±s.e.m. (n = 8 independent experiments; two-tailed Student’s one sample t-test was performed on the ratios with a null hypothesis of 1, P WT = 0.0212 and P KO = 0.0538). (C) Activity-dependent degradation of GluA2 was assessed in cerebellar acute slices from Susd4 KO and WT mice after induction of chemical LTD (cLTD; K-Glu: K + 50 mM and glutamate 10 mM for 5 min followed by 30 min of recovery). This degradation was absent when slices were incubated with 100 mg/mL leupeptin and with 50 mM MG132 (to inhibit lysosomal and proteasome degradation, respectively), or when slices were obtained from Susd4 KO mice. Band intensities of GluA2 were normalized to bACTIN. The ratios between levels with cLTD induction (K-Glu) and without cLTD induction (CTL) are represented. See also Figure 4—figure supplement 1. Mean ±s.e.m. (n = 8 independent experiments; two-tailed Student’s one sample t-test was performed on the ratios with a null hypothesis of 1, P WT = 0.0107, P WT+Leu/ MG132 = 0.3755, P KO = 0.3176 and P KO+Leu/MG132 = 0.2338). (D) Purkinje cells from primary cerebellar cultures of L7Cre mice were transduced at 3 days in Figure 4 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 18 of 26 Research article Neuroscience Figure 4 continued vitro (DIV3) with AAV particles driving the expression of HA-SUSD4 and soluble GFP (AAV2-hSYN-DIO-HA-SUSD4-2A-eGFP) and immunolabeled at DIV17 in non-permeabilizing conditions to localize surface SUSD4 (anti-HA, red) and surface GluA2 subunits (anti-GluA2, blue). Direct green fluorescent protein is shown (GFP, green). Right panels are binarized images of the anti-HA and anti-GluA2 immunolabelings and of the colocalization of these signals (maximum projection of a 1.8 mm z-stack). Scale bar 5 mm. Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 19 of 26 Research article Neuroscience 0.0 0.5 1.0 1.5 2.0 2.5 Total GluA2 protein level 0.0 0.5 1.0 1.5 Total GluD2 protein level GluA2 GluD2 WT KO WT KO A B GluD2 0.0 0.5 1.0 1.5 2.0 Total GluD2 protein levels after cLTD induction (relative to control) WT KOWT KO n.s. n.s. n.s. n.s. WT KO ßACTIN GluD2 +Leupeptin +MG132 CTL K-Glu CTL K-Glu CTL K-Glu CTL K-Glu C Surface GluA2 (Normalized) 0 10000 20000 30000 40000 50000 WT KO Figure 4—figure supplement 1. Basal surface GluA2 levels and total GluA2 and GluD2 levels in SUSD4 knockout (KO) mice. (A) Surface GluA2 levels (relative to normalized GluA2 input) in basal conditions were not different on average in acute cerebellar slices from Susd4 KO when compared to slices from wild-type (WT) mice. (B) and (C) Total protein levels (normalized to bACTIN) of GluA2 (B) and GluD2 (C) were not changed in acute cerebellar slices from Susd4 WT or KO mice in basal conditions. Mean ±s.e.m. (n = 8 independent experiments; unpaired Student’s t-test; GluA2: p=0.5424; GluD2: p=0.6821). (D) Cerebellar acute slices from control WT and Susd4 KO mice were incubated to induce chemical LTD (cLTD; K-Glu: K + 50 mM and glutamate 10 mM for 5 min followed by 30 min of recovery). Slices were incubated with 100 mg/mL leupeptin and with 50 mM MG132 (to inhibit lysosomal and proteasome degradation, respectively). Band intensities of GluD2 were normalized to bACTIN. The ratios between levels with cLTD induction (K-Glu) and without cLTD induction (CTL) are represented. Mean ±s.e.m. (n = 8 independent experiments; two-tailed Student’s one sample t-test was performed on the ratios with a null hypothesis of 1, P WT = 0.4973, P WT+Leu/MG132 = 0.1433, P KO = 0.3143, P KO+Leu/MG132 = 0.9538, n. s. = not significant). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 20 of 26 Research article Neuroscience SUSD4-GFP GluA2 A B INPUT anti-GFP IP GluA2 HA βACTIN PVRL3α76 52 76 150 kDa SEP-GluA2 HA-SUSD4 PVRL3α + + ++ + + + +- - - - Figure 4—figure supplement 2. Interaction and colocalization of HA-SUSD4 and the AMPA receptor subunit GluA2. (A) SUSD4 interaction with GluA2 was assessed by co-immunoprecipitation using HEK293 cells transfected with SEP-GluA2 together with PVRL3aas a control or HA-SUSD4. Affinity purification was performed with an anti- GFP antibody and extracts were probed for co-immunoprecipitation of GluA2 (with an anti-GluA2 antibody) and of HA-SUSD4 (anti-HA antibody) or PVRL3a(anti-PVRL3aantibody). bACTIN was used as a loading control. Representative of N = 3 independent experiments. (B) Mouse hippocampal neurons were transfected at 13 days in vitro (DIV13) with a GFP-tagged SUSD4 construct and immunostained at DIV17 for green fluorescent protein (GFP, green) to localize SUSD4 and for the endogenous GluA2 subunit (anti-GluA2, red). The arrowheads indicate the spines containing SUSD4 and GluA2. Scale bar 10 mm. Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 21 of 26 Research article Neuroscience DNAJA1ITCH ARRB1 PSMC5 G protein-coupled receptor binding NEDD4L ubi q uitin-lik e p rotein t r a n s f e r ase act i v i t y u bi q uiti n p rotein tr a n s f e r ase act i v i t y NEDD4 ubiquitin protein ligase activity WWP2WWP1 ubiq u i t i nlik e p rotein li g ase activit y pV <0.0005 pV 0.005-0.05 GFP + Synaptosomes GFP-SUSD4 + Synaptosomes IB: GFP 225 150 102 76 52 38 31 GFP + Synaptosomes GFP-SUSD4 + Synaptosomes Coomassie staining GFP-SUSD4GFP A C2 ITCH WWP1 NEDD4 HA 76 52 HA-SUSD4 HA-SUSD4∆CT GFP Synaptosome extract IP: HA B RMFQTKFKAHFPPRG PPRSSSSDPDFVVVDGVPVMLPTYDE...QSPPAYPG...MEEDP RMFQTKFKAHFPPRG PPRSSSSDPDFVVVDGVPVMLPTYDE...QSAPAAPG...MEEDP RMFQTKFKAHFPPRG PPRSSSSDPDFVVVDGVPVMAPTADE...QSPPAYPG...MEEDP RMFQTKFKAHFPPRG PPRSSSSDPDFVVVDGVPVMAPTADE...QSAPAAPG...MEEDP RMFQTKFKAHFPPRG PP LP_Y PP_Y C E PSD Map4 Microtubules Recycling endosome EZ EZ Early endosome Late endosome Lysosome Ub Lateral diffusion mRNA transport Translation mRNA granule CI-endocytosis Intracellular retention Endosomal transport Recycling Actin cytoskeleton Endosomal sorting Psmc5 Synaptic pool Extrasynaptic pool Intracellular pool Ubiquitination Nedd4l Nedd4 Wwp1 Wwp2 Itch Exocytosis Epb41l1 Epb41l2 Rph3a CM-endocytosis Ap2b1 Ap2a2 Arrb1 Stabilization Epb41l1 Epb41l2 Proteasome Autophagosome Degradation Pathways SUSD4 interactors AMPA Receptor Golgi / ER network Biosynthesis SUSD4 Mutants SUSD4 NTCT ∆PY ∆LY ∆PY/LY ∆CT ∆NT NT NTCT NTCT NTCT NTCT NTCT NTCT X X X X CCP 150 76 52 38 HA GluA2 IP: HA SEPGluA2 HA-SUSD4 HA-SUSD4∆PY HA-SUSD4∆LY HA-SUSD4∆PY/LY HA-SUSD4∆CT HA-SUSD4∆NT HA-SUSD4-NT HA-SUSD4 PVRL3a + + + + +- + + + NEDD4 102 D GluA2 IP (norm.)/ HA IP HA-SUSD4∆PY HA-SUSD4∆LY HA-SUSD4∆PY/LY HA-SUSD4∆CT HA-SUSD4∆NT HA-SUSD4-NT HA-SUSD4 SEPGluA2 + + + + + + + RMFQTKFKAHFPPRG PPRSSSSDPDFVVVDGVPVMLPTYDE...QSPPAYPG...MEEDP HA TM ... ... 0 1 2 3 4 5 6 Figure 5. SUSD4 binds NEDD4 ubiquitin ligases, known regulators of AMPA receptor turnover and degradation. (A) Mass spectrometry identification of SUSD4 interactors. Left: Affinity purification from cerebellar synaptosomes Figure 5 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 22 of 26 Research article Neuroscience Figure 5 continued was performed using either GFP-SUSD4 as a bait or GFP as a control. Proteins were then resolved using SDSPAGE followed by immunoblot for anti-GFP and coomassie staining of proteins. Right: Gene ontology (GO) enrichment analysis network (Molecular Function category) of the 28 candidate proteins (Cytoscape plugin ClueGO) identified in affinity purified samples (A) by liquid chromatography with tandem mass spectrometry (LC MS/MS). The ubiquitin ligase activity term is significantly enriched in particular due to the identification of several members of the NEDD4 family of HECT ubiquitin ligases. See also Table 1 (n = 3 independent experiments). (B) Immunoblot confirmation of SUSD4 interaction with NEDD4 ubiquitin ligases. Affinity purification from cerebellar synaptosomes was performed using full-length SUSD4 (HA-tagged, HA-SUSD4), a mutant lacking the C-terminal tail (HA-SUSD4DC T ), or GFP as a bait. Proteins were then resolved using SDS-PAGE followed by immunoblot for NEDD4, ITCH, WWP1, or HA-SUSD4 (anti-HA). HA-SUSD4 interacts with all three members of the NEDD4 family. This interaction is lost when the C-terminal tail of SUSD4 is deleted or when GFP is used instead of SUSD4 as a control. (C) Schematic representation of HA-tagged SUSD4 and different mutant constructs: SUSD4DC T (lacking the cytoplasmic tail), SUSD4DN T (lacking the extracellular domain), SUSD4N T (lacking the transmembrane and intracellular domains), SUSD4DPY (point mutation of the PPxY site), SUSD4DLY (point mutation of the LPxY), and SUSD4DPY/LY (double mutant at both PPxY and LPxY). (D) SUSD4 interaction with GluA2 and NEDD4 was assessed by co-immunoprecipitation using HEK293 cells transfected with SEP-GluA2 together with PVRL3aas a control or one of the HA-SUSD4 constructs represented in (C). Affinity purification was performed with an anti-HA antibody and extracts were probed for co-immunoprecipitation of GluA2 (with an anti-GluA2 antibody) and of the HECT ubiquitin ligase NEDD4 (anti-NEDD4 antibody). Co-immunoprecipitated GluA2 levels are normalized to input GluA2 and then represented as relative to the immunoprecipitated levels for each SUSD4 construct. N = 3 independent experiments. (E) Potential interactors of SUSD4 control several parameters of AMPA receptor turnover. Three different pools of AMPA receptors are found in dendrites and spines: synaptic, extrasynaptic, and intracellular. AMPA receptors are synthetized and delivered close to the synaptic spine to reach the synaptic surface. At the surface, AMPA receptors can move laterally (lateral diffusion) or vertically by endocytosis and exocytosis. Endocytosis can be mediated by clathrin (CM-endocytosis) or be clathrin-independent (CI- endocytosis). CM-endocytosis is often related to activity-dependent processes. After endocytosis, AMPA receptors can choose between two different pathways from the early endosomes, one for recycling and the other for degradation. Potential molecular partners of SUSD4 identified by our proteomic analysis could regulate AMPA receptor turnover at several levels of this cycle (in red). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 23 of 26 Research article Neuroscience Itch Nedd4l Wwp1 Nedd4 NEDD4 NEDD4L ITCH WWP1 C2 WW HECTWWWW A B C D E NTCT Figure 5—figure supplement 1. Expression of HECT ubiquitin ligases in adult mouse brain. (A) Schematic representation of four SUSD4 interactors: NEDD4, NEDD4L, ITCH, and WWP1. Legends: N T , N-terminus; HECT, Figure 5—figure supplement 1 continued on next page Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 24 of 26 Research article Neuroscience Figure 5—figure supplement 1 continued homologous to the E6-AP C-terminus domain; C T , C-terminus. (B) Pattern of expression of Nedd4 (RP_050712_03_C08), (C)Nedd4l (RP_040625_01_G10), (D)Itch (RP_050222_01_H06), and (E)Wwp1 (RP_050510_02_E12) mRNA in the adult mouse brain. From Allen Brain Atlas (http://www.brain-map.org). Gonza´lez-Calvo, Iyer, et al. eLife 2021;10:e65712. DOI: https://doi.org/10.7554/eLife.65712 25 of 26 Research article Neuroscience