NMDA Receptor Dysregulation by Defective Depalmitoylation in the Infantile Neuronal Ceroid Lipofuscinosis Mouse Model

Protein palmitoylation and depalmitoylation alter protein function. This post-translational modification is critical for synaptic transmission and plasticity. Mutation of the depalmitoylating enzyme palmitoyl-protein thioesterase 1 (PPT1) causes infantile neuronal ceroid lipofuscinosis (CLN1), a pediatric neurodegenerative disease. However, the role of protein depalmitoylation in synaptic maturation is unknown. Therefore, we studied synapse development in Ppt1-/- mouse visual cortex. We demonstrate the stagnation of the developmental N-methyl-D-aspartate receptor (NMDAR) subunit switch from GluN2B to GluN2A in Ppt1-/- mice. Correspondingly, GluN2A-mediated synaptic currents are diminished and Ppt1-/- dendritic spines maintain immature morphology in vivo. Further, GluN2B is hyperpalmitoylated in Ppt1-/- neurons and associated with extrasynaptic, diffuse calcium influxes and enhanced vulnerability to NMDA-induced excitotoxicity. Remarkably, Ppt1-/- neurons treated with palmitoylation inhibitors demonstrate normalized levels of palmitoylated GluN2B and Fyn kinase, reversing susceptibility to excitotoxic insult. Thus, depalmitoylation of GluN2B by PPT1 plays a critical role in postsynapse maturation and pathophysiology of neurodegenerative disease.


Introduction 29
The neuronal ceroid lipofuscinoses (NCLs) are a class of individually rare, primarily autosomal recessive, 30 neurodegenerative diseases occurring in an estimated 2 to 4 of 100,000 live births (Nita et al., 2016). Collectively, 31 NCLs represent the most prevalent class of hereditary pediatric neurodegenerative disease (Haltia, 2006). The 32 NCLs are characterized by progressive neurodegeneration, blindness, cognitive and motor deterioration, seizures, 33 and how depalmitoylation regulates the turnover of these proteins, let alone during the GluN2B to GluN2A subunit 81 switch, is unclear. 82 In the current study, we investigated the cellular and synaptic effects of PPT1-deficiency using the Ppt1 -/-83 mouse model of CLN1 disease. We focused on the visual system in Ppt1 -/animals for two reasons. First, cortical 84 blindness is a characteristic feature of CLN1 disease. Second, the rodent visual system is a well-studied model of 85 cortical development and synaptic plasticity/maturation and it therefore serves as an optimal experimental model 86 to examine the role of PPT1-mediated depalmitoylation during development. We found that lipofuscin 87 accumulated very early in the Ppt1 -/visual cortex, shortly after eye-opening at postnatal day (P) 14, a timing 88 earlier than previously documented (Gupta et al., 2001). Using biochemistry and electrophysiology, we found 89 impeded developmental NMDAR subunit switch from GluN2B to GluN2A in Ppt1 -/mice compared to wild-type 90 (WT). This NMDAR disruption is associated with disrupted dendritic spine morphology in vivo. To gain further 91 mechanistic insight into neurodegeneration in CLN1, we used cultured cortical neurons and found that Ppt1 -/-92 neurons recapitulate the disrupted GluN2B to GluN2A switch, leading to excessive extrasynaptic calcium 93 transients and enhanced vulnerability to NMDA-mediated excitotoxicity. We directly examined protein 94 palmitoylation state and found hyperpalmitoylation of GluN2B as well as Fyn kinase, which facilitates GluN2B 95 surface retention, in Ppt1 -/neurons. Finally, we demonstrate that chronic treatment of Ppt1 -/neurons with 96 palmitoylation inhibitors normalized GluN2B and Fyn kinase hyperpalmitoylation and rescued the enhanced 97 susceptibility to excitotoxicity. Our results indicate that PPT1 plays a critical role in the developmental GluN2B 98 to GluN2A subunit switch and synaptic maturation. Further, our results indicate that these dysregulated 99 mechanisms contribute to CLN1 pathophysiology and may be shared features of common adult-onset 100 neurodegenerative diseases. 101 6 P11 to P60, and measured levels of GluN2B and GluN2A subunits in whole lysates and synaptosomes of WT and 142 Ppt1 -/visual cortices. Whereas GluN2B levels were comparable between WT and Ppt1 -/at all ages, GluN2A 143 levels in synaptosomes were significantly lower in Ppt1 -/than WT (Figure 2A) . This decrease was present at  144  time points during, and just following, the critical period in visual cortical development (P33, P42, and P60).  145 When analyzed as a ratio of GluN2A/GluN2B, a robust and persistent decrease is observed in Ppt1 -/visual cortex 146 ( Figure 2B). GluN1 levels were unchanged between WT and Ppt1 -/in synaptosomes (Figure 2C) GluN2B and GluN2A, while SAP102 levels remained unchanged, we observed a decrease in PSD-95 levels at 156 P33-P60, the same developmental time points where GluN2A expression was reduced ( Figure 2D). Together, 157 these results suggest reduced incorporation and scaffolding of GluN2A-containing NMDARs in Ppt1 -/synapses, 158 indicating immature or dysfunctional synaptic composition. 159 To examine whether the reduction in GluN2A is due to selective exclusion from the postsynaptic site or 160 alterations in the total protein amount, we also measured NMDAR subunit levels in whole lysates. These findings 161 closely match our findings in synaptosomes. Namely, GluN2A levels showed reductions in Ppt1 -/lysates 162 beginning at the same time point (P33), while GluN2B levels were stable (Supplementary Figure 1A). The 163 GluN2A/2B ratios in Ppt1 -/whole lysates were also lower than those in WT lysates and the reduction was 164 comparable to that observed in synaptosomes. (Supplementary Figure 1B). GluN1 levels, however, were 165 unaltered between genotypes (Supplementary Figure 1C). Collectively, these results indicate a selective 166 decrease in the total amount of mature synaptic components in Ppt1 -/brains and suggest that synaptosomal 167 reductions in GluN2A and PSD-95 may result from altered transcription or translation. 168 NMDAR-mediated EPSCs are altered in Ppt1 -/visual cortex 169 Next, we sought to correlate our biochemical findings electrophysiological changes in NMDAR 170 functionality (Figure 2 NMDA-EPSCs were pharmacologically isolated (see Methods section) and were recorded in whole cell patch 176 mode clamped at +50mV ( Figure 3A). As GluN2A-and GluN2B-containing NMDARs exhibit differential 177 receptor kinetics, with GluN2A displaying fast (~50ms) and GluN2B displaying slow decay kinetics (~300ms), 178 their relative contribution is reliably interpolated by fitting the EPSC decay phase with a double exponential 179 function (Stocca and Vicini, 1998;Vicini et al., 1998). From the fitting, we measured the following parameters: 180 the amplitude (A) of the fast (Af) and slow (As) components; the ratio Af/ Af+As; the decay time constants ( ) of 181 the fast ( f) and the slow components ( s), and the weighted decay ( w). Ppt1 -/mice showed significant decreases 182 in the Af and ratio of Af/Af+As as compared to WT, while As showed no significant change ( Figure 3B). Further, 183 7 the f significantly decreased in Ppt1 -/mice vs. WT, while w and s were comparable ( Figure 3C) We hypothesized that dendritic spine morphology is immature or disrupted in Ppt1 -/neurons, particularly 199 given that GluN2A subunit incorporation is disrupted in vivo. Thus, we used in utero electroporation to sparsely 200 label layer II/III cortical neurons in visual cortex using a GFP construct (Matsuda and Cepko, 2004). GFP-201 expressing cells from WT and Ppt1 -/animals were imaged for detailed analysis of dendritic spine morphology 202 (spine length, spine volume, and spine head volume) at P33, a time point when dendritic spine morphology is 203 typically considered mature and GluN2A is reduced at Ppt1 -/synapses. 204 Electroporated, GFP-expressing cells (procedure schematized in Figure 4A) from WT and Ppt1 -/visual 205 cortex ( Figure 4B) were analyzed in Imaris (Bitplane) for dendritic spine characteristics. While WT neurons 206 exhibited mushroom-type spine morphology with high-volume spine heads ( Figure 4C, arrows), Ppt1 -/neurons 207 showed longer, filipodial protrusions or stubby spines (Figure 4C, arrowheads). Quantification of spine length 208 and spine volume demonstrated that Ppt1 -/spines were longer and less voluminous compared to WT ( Figure 4D-209 E). Further, the volume of dendritic spine heads was reduced in Ppt1 -/neurons ( Figure 4E, inset). These data 210 indicate that dendritic spine morphology is disrupted in the developing CLN1 visual cortex, correspond with the 211 finding that NMDAR composition is immature at P33, and suggest a reduced ability to compartmentalize calcium 212 and other localized biochemical signals in CLN1.

213
NMDAR subunit composition and dendritic spine morphology are also immature in Ppt1 -/primary cortical 214 neurons 215 The GluN2B to GluN2A switch and maturation of dendritic spine characteristics in WT primary neurons 216 has been previously demonstrated (Williams et al., 1993;Zhong et al., 1994;Papa et al., 1995). We established 217 that the developmental switch from GluN2B-to GluN2A-containing NMDARs and dendritic spine morphology 218 are impaired in the Ppt1 -/mouse brain. To understand these mechanisms more comprehensively and examine 219 protein palmitoylation more directly, we used dissociated neuronal cultures. First, we analyzed these 220 developmental events in WT and Ppt1 -/primary cortical neurons to determine whether the biochemical and 221 structural features of disease are recapitulated in vitro.

222
Cortical neurons cultured for 7, 10, or 18 days in vitro (DIV 7, 10, or 18) were harvested and lysates 223 subjected to immunoblot analysis for markers of immature (GluN2B) or mature (GluN2A, PSD-95) excitatory 224 8 synapses. Expression of GluN2B clearly preceded that of mature synaptic markers, peaking in both WT and Ppt1 -225 /neurons at DIV10 and decreasing thereafter ( Figure 5A). In contrast, levels of both GluN2A and PSD-95 226 remained low until DIV18, at which point expression was robust ( Figure 5B, C). Importantly, GluN2A, PSD-95, 227 and GluN2A/GluN2B ratio levels were reduced in Ppt1 -/neurons compared to WT at DIV18, indicating that the 228 biochemical phenotype is recapitulated to an extent in vitro (Figure 5B-D). 229 To analyze dendritic spine morphology, primary cortical neurons from fetal WT and Ppt1 -/mice were 230 transfected with a GFP construct (Matsuda and Cepko, 2004) and were cultured until DIV 15 or 20 when live cell 231 imaging was performed ( Figure 6A). We measured dendritic spine length and volume in transfected cells using 232 the Imaris software (Bitplane). At both DIV 15 and 20, we observed a significant shift in the dendritic spine length 233 and volume (Figure 6B and C). Ppt1 -/neurons demonstrated a significantly higher percentage of long, thin 234 protrusions (filipodia-type; Figure 6B), and a significant reduction in the percentage of mature, mushroom-type 235 dendritic spines with volumes greater than 0.2 m 3 ( Figure 6C). When averaged, the data demonstrate a robust 236 increase in mean spine length and a decrease in spine volume in Ppt1 -/neurons compared to WT controls at both 237 time points (Figure 6B and C, right). Together, these data demonstrate that Ppt1 -/neurons in culture give rise 238 to morphologically immature dendritic spines and corroborate our in vivo findings. individual synaptic site demonstrates that local fluorescence increases in WT cells are confined to a short distance 258 from the peak F/F0 at synaptic sites ( Figure 7B and C, left), while those of Ppt1 -/neurons diffuse longer 259 distances within the dendrite (Figure 7B and C, right). To quantitatively compare these properties, we performed 260 measurements of area under the curve (AUC) and calcium diffusion distance (see shaded region in Figure 7C) 261 for each synaptic site from WT and Ppt1 -/neurons. These analyses revealed a robust increase in both the AUC 262 ( Figure 7D) and the calcium diffusion distance ( Figure 7E) in Ppt1 -/neurons compared to WT. Furthermore, 263 performing correlation analysis of calcium events across time (see Methods) within a given neuron demonstrates 264 that calcium influxes are more synchronous (increased correlation coefficient) in Ppt1 -/neurons compared to WT 265 ( Figure 7F). This result may involve the mechanisms underlying synaptic cluster plasticity, including synaptic 266 9 integration via translational activation (SITA) influenced by excessive Ca 2+ entry in Ppt1 -/neurons (Govindarajan 267 et al., 2006). Together, these data indicate that calcium entry and dispersion are enhanced at Ppt1 -/synapses in 268 vitro. 269 These data are in line with our biochemical and electrophysiological findings and suggest that  containing NMDARs mediate the observed calcium signals. To further test this possibility, we next treated WT 271 and Ppt1 -/neurons with Ro 25-6981 (1 M, added in imaging medium following 2.5min imaging at baseline), a 272 GluN2B-containing NMDAR specific antagonist, and performed calcium imaging. Ro 25-6981 had virtually no 273 effect on calcium signals recorded from WT cells (Figure 7G-I, see Video 3). In contrast, Ppt1 -/neurons treated 274 with Ro 25-6981 showed a reduction in dendritic calcium influxes within shafts, while few residual, 275 compartmentalized transients persisted ( Figure 7G-I, see Video 4). Quantitatively, both AUC ( Figure 7H) and 276 calcium diffusion ( Figure 7I) distance were rescued to WT levels following Ro 25-6981 treatment of Ppt1 -/-277 neurons. Together, these data suggest that Ppt1 -/neurons have extrasynaptic calcium signaling compared to WT 278 that is sensitive to GluN2B-NMDAR blockade. 4.5%; **p= 0.0043; Figure 8B). At 300 M NMDA treatment this effect plateaued, as cell viability between WT 293 and Ppt1 -/neurons was comparable ( Figure 8B). 294 Palmitoylation inhibitors rescue enhanced vulnerability to NMDA-mediated excitotoxicity in Ppt1 -/-295 cultured neurons 296 We next asked whether this enhanced vulnerability to excitotoxicity is due to hyperpalmitoylation of 297 neuronal substrates, and if it can be corrected by balancing the level of synaptic protein 298 palmitoylation/depalmitoylation. First, we found that 77% of cultured Ppt1 -/neurons accumulate ALs 299 spontaneously at DIV18-20 ( Figure 9A and B). Immunostaining for a marker of lysosomal compartments, 300 lysosomal-associated membrane protein-2 (LAMP-2), corroborated these findings, revealing that the observed 301 AL accumulates within lysosomes of Ppt1 -/but not WT neurons (vehicle treatment in Figure 9B). Further, 302 lysosomes appeared swollen in vehicle-treated Ppt1 -/neurons (see arrows in Figure 9B). Treatment with the 303 palmitoylation inhibitors, 2-bromopalmitate (2-BP, 1 M, 7-day treatment) and cerulenin (1 M, 7-day treatment) 304 reduced the percentage of AL-positive neurons ( Figure 9C) and the area occupied with ALs per neuron ( Figure  305 9D). Further, the mean lysosomal size also normalized in Ppt1 -/neurons when these cells were treated with 2-BP 306 or cerulenin ( Figure 9E). 307

10
To examine the efficacy of these compounds in preventing NMDA-mediated toxicity, we pretreated a 308 subset of neurons with the same palmitoylation inhibitors, 2-BP (1 M, DIV12-18) and cerulenin (1 M, DIV12-309 18) prior to treatment with NMDA and glycine. Notably, pretreatment with both 2-BP and cerulenin improved 310 cell viability of Ppt1 -/neurons to that of WT following excitotoxicity induction ( Figure 8F). These results 311 indicate Ppt1 -/neurons are more vulnerable to excitotoxicity and are consistent with our calcium imaging data 312 that demonstrated the predominance of extrasynaptic, GluN2B-mediated NMDAR activity. 313 Palmitoylation inhibitors rescue GluN2B and Fyn kinase hyperpalmitoylation in Ppt1 -/neurons 314 Finally, we directly examined the palmitoylation state of neuronal proteins to determine the mechanisms by 315 which hyperpalmitoylation of neuronal substrates may lead to NMDA-mediated excitotoxicity and asked whether 316 palmitoylation inhibitors correct these abnormalities. In particular, we focused on GluN2B palmitoylation, given 317 our evidence implicating GluN2B in the synaptic dysfunction present in Ppt1 -/neurons. We employed a modified 318 acyl-biotin exchange procedure (Drisdel and Green, 2004), termed the APEGS assay (acyl-PEGyl exchange gel-319 shift) (Yokoi et al., 2016). The APEGS assay effectively tags the palmitoylation sites of neuronal substrates with 320 a 5kDa polyethylene glycol (PEG) polymer, causing a molecular weight-dependent gel shift in immunoblot 321 analyses. Thus, we quantitatively analyzed the palmitoylated fraction of synaptic proteins and palmitoylated 322 signaling molecules that may influence NMDAR function. 323 We subjected DIV18 WT and Ppt1 -/primary cortical neuron lysates to the APEGS assay to determine the 324 palmitoylation state of GluN2B. Indeed, GluN2B palmitoylation was increased in Ppt1 -/neurons compared to 325 WT at DIV18 ( Figure 10A). This suggests enhanced surface retention of GluN2B-containing NMDARs in Ppt1 - (1 m, DIV12-18) as in Figure 9 normalized levels of GluN2B palmitoylation to those of WT ( Figure 10B), 329 implying enhanced turnover or reduced surface retention of GluN2B. We also examined another palmitoylated 330 protein, Fyn kinase, as a candidate for facilitating the enhanced surface retention of GluN2B-containing NMDARs. 331 Fyn is a prominent member of the Src kinase family known to directly interact with and affect the synaptic Ppt1 -/neurons compared to WT ( Figure 10C). Importantly, the palmitoylation of Fyn significantly decreased in 334 both WT and Ppt1 -/neurons following treatment with 2-BP and cerulenin ( Figure 10D). Together, these data 335 point to two potentially overlapping mechanisms by which palmitoylation inhibitors reduce the stabilization or 336 retention of GluN2B at the synaptic compartment, thereby reducing cellular calcium load in Ppt1 -/neurons and 337 mitigating the enhanced susceptibility to excitotoxicity. Further, these data implicate the palmitoylation of Fyn 338 kinase, a molecule that is being targeted for the treatment of Alzheimer's disease ( Since the development of this, and alternative models of CLN1, much progress has been made in understanding 343 the temporal, regional, and cell-type specific effects of lipofuscin accumulation and neuronal degeneration, 344 particularly GluN2B and Fyn kinase, which facilitates retention of GluN2B-containing receptors on the cell surface. This 363 dysregulation likely deviates the GluN2A/2B composition and spine morphology toward immaturity, causing 364 enhanced vulnerability to excitotoxic insult. Importantly, chronic palmitoylation inhibitor treatment alleviates 365 Ppt1 -/--induced dysfunction. Together, these data implicate dysregulated GluN2 subunit switch as a major 366 pathogenic mechanism in CLN1. 367 GluN2B Here, we have identified the novel mechanism whereby impaired GluN2B to GluN2A subunit switch contributes 377 to the core pathophysiology of a pediatric neurodegenerative disease. 378 We have demonstrated a novel role for PPT1 in the regulation of palmitoylated postsynaptic proteins. In 379 particular, lack of PPT1 disrupts the NMDA receptor GluN2B to GluN2A subunit switch characteristic of 380 excitatory synaptic maturation. Initially, we predicted synaptic markers, particularly PSD-95, would be 381  The current work supports a role for PPT1 in regulating postsynaptic proteins. However, whether the pre-413 or postsynaptic site is affected foremost in Ppt1 -/mouse visual cortex during development is still unknown. 414 Indeed, we detect a significant reduction in GluN2A and PSD-95 beginning at P33, but not before, while 415 lipofuscin deposition begins as early as P14. However, our electrophysiological, biochemical, and in vitro calcium 416 imaging data suggest that this reduction is manifested postsynaptically. Furthermore, we show direct evidence for 417 hyperpalmitoylation of two postsynaptic proteins, the GluN2B subunit and Fyn. One direction of our future work 418 is to examine in more detail the timing, subcellular specificity, and trafficking of both pre-and post-synaptic 419 proteins in Ppt1 -/mouse. 420 Excitotoxicity these previous studies are consistent with our observations that Ppt1 -/neurons are biased toward extrasynaptic 436 calcium transients (Figure 7 and Video 2) and that they are more susceptible to excitotoxicity (Figure 8). 437 Furthermore, the most significant outcome of this study is that palmitoylation inhibitors mitigated the pro-438 apoptotic predisposition of Ppt1 -/neurons in vitro (Figure 9). 439 The incorporation of GluN2A For collection of brain for biochemistry (immunoblot), Ppt1 -/and WT animals were decapitated following 500 isoflurane anesthesia, then the brain was removed, and washed in ice cold PBS. The occipital cortex (visual 501 cortex), hippocampus, and remaining cortex were separately collected on ice. Isolated visual cortices from Ppt1 -502 /and WT animals were homogenized in ice-cold synaptosome buffer (320mM sucrose, 1mM EDTA, 4mM 503 HEPES, pH7.4 containing 1x protease inhibitor cocktail (Roche), 1x phosphatase inhibitor cocktail (Roche) and 504 1mM PMSF) using 30 strokes in a Dounce homogenizer. Aliquots for whole lysate (WL) were stored and the 505 remaining sample was used for synaptosome preparation, performed as previously with slight modification. In 506 brief, WLs were centrifuged at 1,000 x g to remove cellular debris, supernatant was then centrifuged at 12,000 x 507 g for 15min to generate pellet P2. P2 was resuspended in synaptosome buffer and spun at 18,000 x g for 15min 508 to produce synaptosomal membrane fraction, LP1, which was used for downstream biochemical analyses 509 (synaptosomes). For immunoblot, protein concentration of each sample was determined using BCA protein assay 510 (Pierce). Samples were then measured to 20 g total protein in 2x Laemmli buffer containing 10% -511 mercaptoethanol (Bio-rad), boiled at 70°C for 10min and loaded into 10% tris-glycine hand cast gels (Bio-rad), 512 or 4-20% precast gels (Bio-rad) for electrophoresis (110V, 1.5-2h). Proteins were wet-transferred to PVDF 513 membranes (Immobilon-P, Millipore), blocked in TBS, pH7.4 containing 5% non-fat milk and 0.1% Tween-20 514 (TBS-T+5% milk). Membranes were incubated in primary antibody solutions containing 2% BSA in TBS-T for 515 2h at RT or overnight at 4°C. Primary antibodies were used as follows: GluN2A (Cat: NB300-105, 1:1,000, Novus 516 Biologicals), GluN2B (Cat: 75/097, 1:1,000, Neuromab), GluN1 (Cat: 75/272, 1:1000, Neuromab), PSD-95 (Cat:  517 K28/74, 1:2,000, Neuromab), SAP102 (Cat: N19/2, 1:2,000, Neuromab), Fyn kinase (Cat: 4023, 1:1,000, Cell 518 signaling) and -actin-HRP (Cat: MA5-15739-HRP, 1:2,000, ThermoFisher). Membranes were then incubated 519 with appropriate secondary, HRP-conjugated antibodies (Jackson ImmunoResearch) at either 1:5,000, 1:10,000, 520 or 1:30,000 (PSD-95 only) for 1h at RT. Visualization and quantification was performed using Pierce SuperSignal 521 ECL substrate and Odyssey-FC chemiluminescent imaging station (LI-COR). Signal density for each synaptic 522 protein was measured using the LI-COR software, Image Studio Lite (version 5.2) and was normalized to the 523 signal density for -actin loading control for each lane. A total of four independent experiments was performed 524 for both WL and LP1 analyses, with a minimum of two technical replicates for each experiment averaged together. 525 Histology and autofluorescent lipopigment quantification 526 Ppt1 -/and WT mice were anesthetized using isoflurane and transcardially perfused with ice cold PBS (pH 527 7.4, ~30ml/mouse) followed by 4% paraformaldehyde (PFA) in PBS (~15ml/mouse). Brains were removed and 528 post-fixed for 48h at 4°C in 4% PFA and transferred to PBS, pH7.4 containing 0.01% sodium azide for storage if 529 necessary. Brains from Ppt1 -/and WT animals were incubated in 30% sucrose solution for 48h prior to sectioning 530 16 using Vibratome 1000 in cold PBS. For imaging and quantification of AL, sagittal sections were cut at 100 m. 531 Every third section was mounted on Superfrost Plus microscope slides (VWR) using Vectamount mounting media 532 containing DAPI (Vector Laboratories, cat: H-5000). Interlaced/overlapping images of visual cortex area V1 from 533 the cortical surface to subcortical white matter (or subiculum), which was localized using Paxino's mouse atlas 534 (sagittal), were collected for 2-4 sections from each animal using a Zeiss LSM710 confocal laser scanning 535 microscope at 40x magnification (excitation at 405nm to visualize DAPI and 561nm to visualize AL). All sections 536 were imaged using identical capture conditions. Quantification of AL was performed by thresholding images in 537 FIJI (NIH), generating a binary mask of AL-positive pixels (satisfied threshold) vs. background. The identical 538 threshold was applied to each image (from cortical surface to subcortical white matter and across animals). 539 Percent area occupied by AL puncta that satisfied the threshold was then calculated using the "analyze particles" 540 tool in FIJI. This analysis was performed for 2-4 sections (total of ~10-20 images, as imaging an entire cortical 541 column is typically 5 interlaced images) from each animal and averaged together to give a single value, 542 representative of the total area occupied by AL in the cortical column imaged. Three to six animals per group 543 were analyzed this way and averaged to give the mean area occupied by AL at each time point, for both genotypes 544 (n=4-6 animals/group). 545 Electrophysiology 546 WT and Ppt1 -/animals at P42 were deeply anesthetized using isoflurane drop method and decapitated. 547 Brains were resected in semi-frozen oxygenated (95% O2 and 5% CO2) artificial cerebrospinal fluid (aCSF, in 548 mM: NaCl 85, sucrose 75, KCl 2.5, CaCl2 0.5, MgCl2 4, NaHCO3 24, NaH2PO4 1.25, D-glucose 25, pH 7.3), and 549 350µm sections containing visual cortex area V1 were sectioned using a Leica VT1200 S vibratome in semi-550 frozen aCSF. After recovery (1h) in aCSF at 30°C, sections were transferred to the recording chamber, perfused 551 at 2ml/min with aCSF at 30°C. Following localization of visual cortex area V1 using Paxinos mouse brain atlas, 552 a stimulating electrode was placed in layer IV, and pyramidal neurons from layer II/III were blindly patched 553 (patch solution in mM: CsOH monohydrate 130, D-Gluconic acid 130, EGTA 0.2, MgCl2 1, CsCl 6, Hepes 10, 554 Na2-ATP 2.5, Na-GTP 0.5, Phoshocreatine 5, QX-314 3; pH 7.3, osmolarity 305 mOsm) and recorded in voltage 555 clamp mode at +50mV (VH) to remove Mg 2+ block from NMDARs. NMDA-EPSCs were pharmacologically 556 isolated via addition of CNQX (10μM), (+)-Bicuculline (60 μM) and SCH 50911 to block AMPA, GABAA and 557 GABAB receptors, respectively. Stimulation intensity was titrated to give a saturating postsynaptic response, and 558 EPSCs were then recorded, averaging 5-10 sweeps. The decay phase of the averaged NMDAR-EPSCs were then 559 fitted to a double exponential (Carmignoto and Vicini, 1992). We calculated for each cell: the amplitude of the 560 fast (Af) component (GluN2A-mediated), the amplitude of the slow (As) component (GluN2B-mediated), the 561 contribution of the fast component Af/ Af+As to the overall decay phase, the fast ( f), the slow ( s) and the 562 weighted ( w) in WT and Ppt1 -/mice following this formula: w= fx(Af/Af+As) + sx(As/Af+As) (n=10/4 563 (cells/animals), WT; n=12/5 PPT-KO). 564 In utero electroporation 565 In utero electroporation was performed as previous (Yoshii et al., 2011(Yoshii et al., , 2013. Timed-pregnant dams at E16.5 566 were deeply anesthetized via isoflurane (3% induction, 1-1.5% for maintenance of anesthesia during surgery) and 567 laparotomized. The uterus was then externalized and up to ~1μl of solution containing GFP construct (2μg/μl) 568 and fast green dye was delivered into the left lateral ventricle through the uterine wall using a micropipette. Using 569 an ECM 830 Square Wave electroporator (Harvard Apparatus, Holliston MA), brains were electroporated with 5 570 pulses of 28V for 50msec at intervals of 950msec at such an angle to transfect neurons in visual cortex. After 571 recovery, pregnancies were monitored and pups were delivered and nursed normally. Electroporated pups were 572 genotyped, raised to P33, and sacrificed via transcardial perfusion as described above. Electroporated brains from 573 WT and Ppt1 -/mice (procedure schematized in Figure 4A) were sectioned and sequentially mounted. Transfected 574 neurons in visual cortex ( Figure 4B) were imaged to capture all apical neurites and 3D reconstructed images were 575 analyzed in Imaris (Bitplane) for dendritic spine characteristics known to be associated with synaptic maturity 576 (spine length, spine volume, and spine head volume). At least two z-stack images (typically >100 z-planes/image) 577 were stitched together to capture the prominent apical neurites and extensions into the cortical surface for each 578 cell. Each stitched image, equivalent to one cell, was considered one n. 579 Primary cortical neuron culture 580 For primary cortical neuron cultures, embryos from timed-pregnant, Ppt1 -/+ dams were removed, 581 decapitated, and cortices resected at embryonic day (E) 15.5. All dissection steps were performed in ice cold 582 HBSS, pH7.4. Following cortical resection, tissue from each individually-genotyped embryo were digested in 583 HBSS containing 20U/ml papain and DNAse (20min total, tubes flicked at 10min) before sequential trituration 584 with 1ml (~15 strokes) and 200 l (~10 strokes) pipettes, generating a single-cell suspension. For live-cell imaging 585 experiments, cells were counted then plated at 150,000-180,000 cells/well in 24-well plates containing poly-D-586 lysine/laminin-coated coverslips. For biochemical experiments, i.e. immunoblot, APEGS assay in vitro, cells 587 were plated on poly-D-lysine/laminin-coated 6-well plates at 1,000,000 cells/well. Cells were plated and stored 588 in plating medium (Neurobasal medium containing B27 supplement, L-glutamine and glutamate) for 3-5 DIV, 589 before replacing half medium every 3 days with feeding medium (plating medium without glutamate). Cultures 590 used in chronic palmitoylation inhibitor treatment were exposed to either DMSO (vehicle), 2-BP (1 m, Sigma, 591 cat: 238422) or cerulenin (1 m, Cayman Chemicals, cat: 10005647) every 48 hours between DIV 11 and 18. 592 Primary cortical neuron harvest and immunoblotting 593 Primary cortical neurons from E15.5 WT and Ppt1-/-embryos were cultured for 7, 10, or 18 DIV prior to harvest 594 for immunoblot or APEGS assay (only DIV18 used for APEGS). To harvest protein extracts, cells were washed 595 2x with ice-cold PBS before addition of lysis buffer containing 1% SDS and protease inhibitor cocktail, 596 500 l/well. Cells were incubated and swirled with lysis buffer for 5 minutes, scraped from the plate, triturated 597 briefly, and collected in 1.5ml tubes. Lysates were centrifuged at 20,000g for 15min to remove debris, and the 598 supernatant was collected for biochemical analysis. Immunoblotting analyses were performed as in section 2.2. 599 APEGS assay was carried out as described in section 2.7. 600 APEGS assay on primary cortical neuron lysates 601 The APEGS assay was performed as utilized in Yokoi, 2016 and recommended by Dr. M. Fukata (personal 602 communication, 06/2018). Briefly, cortical neuron lysates were brought to 150 g total protein in a final volume 603 of 0.5ml buffer A (PBS containing 4% SDS, 5mM EDTA, protease inhibitors, remaining sample used in aliquots 604 for "input"). Proteins were reduced by addition of 25mM Bond-Breaker™ TCEP (0.5M stock solution, 605 ThermoFisher) and incubation at 55°C for 1h. Next, to block free thiols, freshly prepared N-ethylmaleimide 606 (NEM) was added to lysates (to 50mM) and the mixture was rotated end-over-end for 3h at RT. Following 2x 607 chloroform-methanol precipitation (at which point, protein precipitates were often stored overnight at -20°C), 608 lysates were divided into +hydroxylamine (HA) and -HA groups for each sample, which were exposed to 3 609 volumes of HA-containing buffer (1M HA, to expose palmitolylated cysteine residues) or Tris-buffer control (-610 HA, see Figure 10), respectively, for 1h at 37°C. Following chloroform-methanol precipitation, lysates were 611 solubilized and exposed to 10mM TCEP and 20mM mPEG-5k (Laysan Bio Inc., cat# MPEG-MAL-5000-1g) for 612 1h at RT with shaking (thereby replacing palmitic acid with mPEG-5K on exposed cysteine residues). Following 613 18 the final chloroform-methanol precipitation, samples were solubilized in a small volume (60 l) of PBS containing 614 1% SDS and protein concentration was measured by BCA assay (Pierce). Samples were then brought to 10 g 615 protein in laemmli buffer with 2% -mercaptoethanol for immunoblot analyses as in section 2.2. Quantification 616 of palmitoylated vs. non-palmitoylated protein was carried out as in section 2.2, with the added consideration that 617 palmitoylated protein was taken as the sum of all (typically two-three distinct bands, see Figure 9) bands 618 demonstrating the APEGS-dependent molecular weight shift compared to the -HA control lane. Non-619 palmitoylated protein was quantified from the band size-matched to the -HA control sample. The ratio was taken 620 as the palmitoylated protein divided by non-palmitoylated protein, all divided by -actin control from the same 621 lane. 622 Transfection, dendritic spine and calcium imaging analyses 623 For analysis of dendritic spine morphology, WT and Ppt1 -/neurons were transfected between DIV6-8 624 with GFP using Lipofectamine® 2000 (ThermoFisher) according to manufacturer protocol. Briefly, GFP DNA 625 construct (~2 g/ l, added at ~1 g/well) was mixed with Lipofectamine-containing Neurobasal medium, 626 incubated for 30min to complex DNA-Lipofectamine, equilibrated to 37 C, and added to the cells 250 l/well for 627 1-1.5h. Following incubation, complete medium was returned to the cells. Neurons were then imaged at DIV15 628 and DIV20 for dendritic spine morphology using a Zeiss LSM 710 confocal microscope equipped with a heated 629 stage at 63x magnification. GFP-positive neurons were imaged at 0.2 m Z-plane interval (typically 25 Z-630 planes/image). Three to seven overlapping Z-stacks were stitched to visualize an entire neuron. Z-stack images 631 were collapsed into a single plane and dendritic spines were analyzed using semi-automated image processing 632 software, Imaris (Bitplane treated neurons, the same protocol was followed with the exception that calcium transients at an individual 703 synaptic site were split into "before application" and "after application" groups. 704 To analyze synaptic synchrony, F/F0 measurements for 20 randomly-chosen sites of synaptic activity 705 per neuron were correlated across the time dimension (500 frames of each video). A correlation matrix was 706 20 generated to determine the average correlation of each synaptic site with all other chosen sites. The average values 707 for each synaptic site, for 5 neurons/group are plotted in Figure 7. 708 NMDA toxicity assays 709 To measure cell viability following exposure of WT and Ppt1 -/neurons to NMDA and glycine, neurons 710 were plated as above and grown to DIV18. For experiments presented in Figure 6, feeding medium was removed 711 from neurons, stored at 37°C, and replaced with B27-free Neurobasal medium with or without NMDA/glycine at 712 the following concentrations: 10/1 M, 100/10 M, or 300/30 M (ratio maintained at 10:1). Cells were incubated 713 for 2h at 37°C in treatment medium. Following incubation, treatment medium was removed and replaced with 714 the original feeding medium. Cells were then incubated an additional 22h before addition of PrestoBlue® cell 715 viability reagent (ThermoFisher). At 24h, fluorescence intensity of each well was measured using a Beckman 716 Coulter DTX 800 Multimode Detector. Cell viability for each treatment condition was calculated and expressed 717 as percentage of vehicle-treated control wells (no pretreatment, no NMDA application). Experiments in Figure 7  718 were performed similarly except that cultures were pretreated with either DMSO (vehicle), 2-BP (1 M, Sigma, 719 cat: 238422) or cerulenin (1 M, Cayman Chemicals, cat: 10005647) every 48 hours between DIV 11 and 18. 720 AL accumulation in vitro, palmitoylation inhibitor treatment, imaging and analysis 721 WT and Ppt1 -/neurons were cultured as above. To examine AL deposition, neurons were grown to 722 DIV18-20, fixed in 4% PFA for 10min at RT, and stored in PBS for up to 72 hours prior to immunocytochemistry. 723 To examine AL accumulation alone, cells were immunostained for the microtubule associated protein, MAP2 724 (Millipore Sigma, cat: AB5622) and mounted in DAPI-containing mounting medium. To assess AL localization, 725 DIV18-DIV20 neurons were immunostained for MAP2 and LAMP-2 (Abcam, cat: ab13524). Neurons were then 726 imaged at random using a Zeiss LSM 710 confocal microscope at 63x magnification. Z-stacks (0.4 m Z-plane 727 interval, 12-22 Z-planes/image) were taken at 512 x 512 pixel density. 7-10 neurons/group for three independent 728 experiments. 729 To semi-automatically analyze the percentage of AL-containing cells, the cytosolic area covered by AL 730 deposits, and the cytosolic area covered by lysosomes, images immunostained for MAP2 and LAMP-2 were 731 processed in FIJI. Each channel of the image: LAMP-2 (488nm), MAP2 (633nm), DAPI (405nm), AL (561nm) 732 was thresholded separately as to display only the lysosomes, cell soma, the nucleus, and AL deposits, respectively. 733 Thresholds were kept identical between images. Next, the areas of these compartments/deposits were measured 734 using the "analyze particles" tool restricted to an ROI tracing the cell soma. Lysosomes needed to have a 735 circularity of >0.5 to avoid counting small clusters of lysosomes as a single unit (Bandyopadhyay et al., 2014;736 Grossi et al., 2016). To measure AL deposits, the same approach was used with the additional constraint: AL 737 deposits were required to have a circularity >0.4 and comprise more than 8 adjacent pixels. Cytosolic area was 738 calculated by measuring MAP2 signal area and subtracting the area occupied by DAPI stain. 739 Immunocytochemistry 740 Coverslips were stained in runs so that all experimental and control groups were immunostained 741 simultaneously. Coverslips were washed 3x with TBS, permeabilized for 20min at RT with TBS containing 0.5% 742 Triton X-100, and blocked for 1h at RT in TBS containing 0.1% Triton X-100 and 5% BSA. Then, primary 743 antibody (MAP2 or LAMP-2) at 1:400 dilution was added to coverslips in TBS containing 0.1% Triton X-100 744 and 1% BSA and incubated for 2h at RT or overnight at 4 C. Following 4X washes with TBS containing 0.1% 745 21 Triton X-100, cells were incubated with 1:400 secondary, fluorophore-linked antibody (either Alexa Fluor 488, 746 cats: A-11034, A-11006; or Alexa Fluor 633, ThermoFisher, cat: A-21070) in TBS containing 0.1% Triton x-100 747 and 1% BSA. These steps are repeated for double immunostained cells. For LAMP-2/MAP2 double  748 immunostaining, saponin was used in place of Triton X-100 at the same concentrations. Coverslips are then 749 mounted on SuperFrost Plus slides in DAPI Vectamount medium. 750

751
The authors thank Dr. Froylan Calderon de Anda (Universitätsklinikum Hamburg-Eppendorf) for the GCaMP3 752 construct used to visualize calcium activity in Figure 6. This work is supported by startup funding awarded to 753 A.Y. by the University of Illinois at Chicago, Department of Anatomy and Cell Biology. repetitions/group) at each age using t-test and the significance was indicated as follows: *p<0.05, and **p<0.01. 269 Error bars represent s.e.m. 270 Representative confocal images of GFP-transfected dendritic segments from WT and Ppt1 -/neurons at P33. 285 Arrows mark mature, mushroom-type spines; arrowheads mark thin, filipodial spines or stubby, headless spines.