All animal procedures were performed in accordance with the guidelines of the University of Illinois at Chicago Institutional Animal Care and Use Committee. Ppt1^+/− (heterozygous) mice were maintained on 12-h light/dark cycle with food and water ad libitum and bred to generate Ppt1^−/− (CLN1), Ppt1^+/−, and Ppt1^+/+ (wild-type, WT) animals (Gupta et al., 2001). Animals of both sexes were allocated to groups based on their genotype (i.e., WT or Ppt1^−/−), without specific regard to equally balancing the number of males or females across groups. Therefore, males and females were used in roughly equal proportion. Imaging data were acquired randomly (no criteria were used for selecting cells, view fields, etc. except where anatomically necessary). All data were acquired and maintained without descriptive naming/labeling to ease randomization. Severe seizure, defined as an animal suffering from either prolonged (3–5 min) epileptic episodes or a failure to recover normal locomotor activity from such episodes, was considered as a humane endpoint criterion and prompted euthanasia but was not systematically quantified.

WT and Ppt1^−/− brains were collected as previously (Koster et al., 2019, 2023). For biochemical analysis of total protein content, isolated visual cortices from Ppt1^−/− and WT animals were homogenized in ice-cold synaptosome buffer (320 mM sucrose, 1 mM EDTA, 4 mM HEPES, pH 7.4) containing 1× protease inhibitor cocktail (Roche), 1× phosphatase inhibitor cocktail (Roche), and 1 mM PMSF using 30 strokes in a Dounce homogenizer. Aliquots of lysates were stored at −80°C, and the remaining sample was used for synaptosome preparation, which was performed as follows: whole lysates were centrifuged at 1000 × g to remove cellular debris, after which supernatant was centrifuged at 12,000 × g for 15 min to generate pellet P2. The P2 pellet was resuspended in synaptosome buffer and centrifuged at 18,000 × g for 15 min to produce the synaptosomal membrane fraction, LP1, referred to hereafter as synaptosomes.

For immunoblotting, the protein concentration of each sample was determined using BCA protein assay (Pierce). Samples were then brought to 20 μg total protein in 2× Laemmli buffer containing 10% β-mercaptoethanol (Bio-Rad), heated at 70°C for 10 min, and loaded into 4–20% precast gels (Bio-Rad) for electrophoresis (130 V, 1.5–2 h). Proteins were wet-transferred to PVDF membranes (Immobillon-P, Millipore), blocked in TBS, pH 7.4, containing 5% non-fat milk and 0.1% Tween-20 (TBS-T + 5% milk). Membranes were incubated in primary antibody solutions containing 2% BSA in TBS-T for 2 h at room temperature (RT) or overnight at 4°C. Primary antibodies were used according to Table 4. Membranes were then incubated with appropriate secondary, HRP-conjugated antibodies (Jackson ImmunoResearch) at either 1:1,000 or 1:5,000 for 1 h at RT before washing three times with TBS-T. Visualization and quantification were performed using the Pierce SuperSignal ECL substrate and Odyssey-FC chemiluminescent imaging station (LI-COR). Signal density for each synaptic protein was measured using the LI-COR software, Image Studio Lite (version 5.2), and normalized to the signal density for β-actin loading control for each lane. Two technical replicates for each experiment were averaged together to get one n.

Wild-type and Ppt1^−/− mice were anesthetized using isoflurane and transcardially perfused with ice-cold PBS (pH 7.4, ~30 mL/mouse), followed by 4% paraformaldehyde (PFA) in PBS (~15 mL/mouse). Brains were removed and post-fixed overnight at 4°C in 4% PFA and transferred to PBS, pH 7.4, containing 30% sucrose solution for 48 h prior to sectioning at either 50 or 100 μm in cold PBS using a Vibratome 1,000 (Technical Products International, St. Louis, MO). Serial sections were stored free floating in cryoprotectant solution (30% glycerol, 30% ethylene glycol in PBS) at −20°C until analysis of AFSM or immunohistochemistry was performed.

For AFSM analysis (as in Koster et al., 2019), 3–4 mid-sagittal sections were mounted on Superfrost Plus microscope slides (VWR) using Vectamount mounting media containing DAPI (Vector Laboratories). Images were acquired for at least two sections from each animal using a Zeiss LSM710 confocal laser scanning microscope at 40× magnification. DAPI signal was visualized using excitation at 405 nm, while AFSM was detected using 561 nm excitation. Although AFSM signal is detectable without any deliberate staining across a wide spectral range, we chose to quantify it in the red fluorescence range to allow for consistency across experiments where the green (488 nm) channel was occupied by immunostaining for another protein of interest (e.g., Iba1). All sections were imaged using identical capture conditions. Quantification of AFSM was performed by generating a binary mask of AFSM-positive pixels (i.e., fluorescent signal in the 561 nm channel) in FIJI. An identical threshold was applied to each image. Percent area occupied by AFSM puncta that satisfied the threshold was then calculated using the “analyze particles” tool in FIJI. This analysis was performed for 2–4 sections (a total of 10–20 images, as imaging an entire cortical column is typically five interlaced images) from each animal and averaged together to give a single value, representative of the total area occupied by AFSM in the cortical column imaged. Animal numbers are reported in the relevant figure legends.

Low magnification (4×) images were acquired from 2 to 3 matched sections for all animals used for AFSM analysis using a Zeiss Axio Imager M1. Images were analyzed in Fiji by drawing an ROI perpendicular to the cortical surface that extended down to the start of the subcortical white matter, i.e., at the lowest bounding edge of cortical L6, where DAPI staining sharply disappears. ROI length was measured for three mid-sagittal sections from each animal corresponding to V1. These values were averaged to give the cortical thickness in mm.

For immunohistochemistry, 3–4 medial sections were first incubated in TBS for 10 min before undergoing permeabilization (TBS + 0.5% Triton X-100) for 30 min at RT. Next, samples underwent antigen retrieval by heating in tris-EDTA (pH 9.0) at 95°C for 30 min before being equilibrated to room temperature in Tris-EDTA solution for 40 min. Tissue was then blocked (TBS + 0.1% Triton X-100, 4% BSA, and 5% normal goat serum) for 2 h at RT before being incubated in rabbit anti-Iba1 (1:1,000, in TBS + 0.1% Triton X-100 + 2% BSA) for 48 h. After washing four times, 10 min each, tissue was incubated in Alexa Fluor 488 goat anti-rabbit (1:1,000; Thermo Fisher) overnight at RT. The tissue was then washed (4×, 10 min) and mounted as above.

Microglial images were acquired with a Zeiss LSM710 confocal microscope either at 10× (low magnification images) or 63× for Sholl analysis. 63× images were acquired as 30–60 plane Z-stacks (Z-interval = 2 μm) in random fashion (the only criterion is that at least one full microglia had to be centered in the stack) at the border of layers 2/3 and 4 in the visual cortex. 3–4 images were taken from two tissue sections for each animal. Images were analyzed using the Sholl analysis tool in Fiji by lab members blinded to the condition. Briefly, the image was collapsed into a maximum-intensity projection to ensure all microglia processes were captured, and individual microglia were outlined with a freehand ROI. The surrounding area was removed (“clear outside” function in Fiji), and the image was then thresholded to generate a mask of all microglia processes, and an ROI was created at the center of the cell soma. The mask containing one individual microglia was then skeletonized using the “skeletonize” plugin in Fiji. Sholl analysis was performed according to these parameters using the Fiji Sholl tool: start radius (from the center of cell soma ROI) = 5 μm, step size = 2 μm, end radius = 70 μm. The number of intersections at each 2 μm step was averaged for all microglia from each animal and counted as one n.

To analyze the number of microglia across age, genotype, and condition (LR vs. DR), the number of Iba1-positive cell bodies was counted manually in a 0.135 × 0.135 × 0.05 mm (width × length × depth) volume using the imaging parameters described above. Cell bodies straddling the top and right borders of the image were not counted, while those on the left and bottom borders were. The number of microglia in this view field was then extrapolated to estimate the number of microglia in a 1 mm³ volume.

Mice between P20 and P40 were anesthetized via isoflurane inhalation (4% induction, 1–1.5% maintenance) and placed in a stereotaxic frame (RWD Life Science Inc.). After expressing a small amount of vitreous from the eye, intravitreous injections of 2 μL fluorophore-conjugated CTB (1 mg/mL; 488 nm conjugate in the left eye, 555 nm conjugate in the right eye) were performed for each mouse. Following the injection, antibiotic ointment was applied to each eye, and the animals were allowed to recover in their home cage for 24 h before undergoing transcardial perfusion as detailed above. Sectioned brains (100 μm sequential coronal sections encompassing the entire dLGN) were then assessed for the quality of the injections by a blinded researcher, and only those brains with well-traced retinogeniculate projections were immunostained for Iba1. Matched sections were then imaged under high magnification at the border of converging retinal projections from both eyes in each dLGN using an LSM710 confocal microscope and compared for microglial morphology (Sholl analysis) as described above.

WT and Ppt1^−/− occipital cortices from animals aged to P42 were used for palmitoyl-proteomic analysis. Lysates and synaptosomes were collected as described above. The palmitoyl-proteomic protocol was then carried out according to Wan et al. (2007) with slight modifications. Before beginning the assay, a BCA assay was performed to start with equal (600 μg) protein content for each sample. Blocking (N-ethylmaleimide; NEM), hydroxylamine, and biotinylation (HPDP-biotin) steps were all performed as recommended in the protocol. The elution protocol was also followed (Wan et al., 2007), with the exception that instead of streptavidin resin, magnetic streptavidin-coated beads (Dynabeads™, Thermo Fisher) were used (100 μL beads/sample). The final eluent was frozen at −80°C and prepared for mass spectrometry (see below). Due to the small starting material (occipital cortex only), the whole procedure was scaled down into 2 mL tubes, including chloroform-methanol precipitations, which were carried out with the following volumes: 150 μL sample, 600 μL methanol, 150 μL chloroform, and 450 μL water. This limited protein loss was evident in trial runs using 15 mL conical tubes.

Eluents from the ABE assay were dried for approximately 30 min and digested using the S-trap Micro Spin Column Digestion Protocol (Protifi, Huntington, NY) with minor changes. Briefly, 30 μL of 10% sodium dodecyl sulfate (SDS) and 100 mM triethylammonium bicarbonate (TEAB) with Pierce protease inhibitor cocktail (Thermo Fisher Scientific, Waltham, MA) and phosphatase inhibitors (10 mM sodium pyrophosphate, 1 mM PMSF, 1 mM sodium orthovanidate, and 1 mM β-glycerolphosphate). Proteins were reduced with a final concentration of 20 mM dithiothreitol (DDT) at 95°C for 10 min, followed by alkylation in the dark at room temperature with 40 mM of iodoacetamide. Next, phosphoric acid was added for a final concentration of 1.2%. Samples were briefly vortexed to mix before 300 μL of S-trap binding buffer (90% MeOH, 100 mM TEAB) was added. Samples were vortexed again prior to loading onto the S-Trap Micro Spin Columns. After four washes with 150 μL of S-trap binding buffer with centrifugation at 1,000×g, 40 μL of 50 mM TEAB containing 0.75 μg of trypsin was added and incubated overnight at 37°C.

Peptides were eluted with 40 μL of each of the following solutions: 50 mM TEAB, 0.2% formic acid (FA), 50% acetonitrile (ACN), and 0.1% FA. The spin column was spun at 4,000×g after adding each solution. Pooled eluents were dried down prior to resuspension in 100 μL of 3% ACN and 0.1% FA.

Raw files were searched with Proteome Discoverer 2.3 (Thermo Fisher Scientific, Waltham, MA) using the Sequest HT search engine against the UniProt Mus musculus database (22,286 gene sequences; downloaded on 27 April 2017). Mass error tolerance was set to 10 ppm for precursors, cleaved by trypsin, allowing a maximum of two missed cleavages, with sequence lengths between 6 and 144 amino acids. Fragment masses were searched with a tolerance of ±0.02 Da. Dynamic modifications included oxidation (M), deamidation (N, R, Q), and acetylation (N-terminus). Carbamidomethylation was set as a static modification (C). Both peptides and PSMs were set to a target false discovery rate (FDR) ≤0.01 for matches with high confidence. Label-free quantification (LFQ) was performed using precursor ion intensity. Samples were normalized using the average intensity of all peptides. The top five most abundant peptides were used for protein abundance calculation. t-test was used to determine the p-values between the two conditions.

Datasets filtered for two unique peptides were further narrowed by filtering for proteins showing an increase in their abundance ratio (Ppt1^−/−/WT) of >1.2-fold. These filtered gene lists for lysates and synaptosomes were input separately into the SynGO online tool (Koopmans et al., 2019). The genes encompassed in the top significant biological process (BP) SynGO term for lysates and synaptosomes were then input into STRING, the online protein–protein interaction database.¹ A K-means cluster analysis was performed to detect clusters of functionally related proteins and for clarity of visualization.

The APEGS assay was performed as described (Kanadome et al., 2019), following the guidelines for tissue samples. Homogenate buffer as described in the APEGS protocol (20 mM Tris–HCl, 2 mM EDTA, 0.32 M sucrose, pH 8.0) was used to homogenize WT and Ppt1^−/− occipital cortices. Lysates and synaptosomes were then brought to 300 μg total protein in a final volume of 0.5 mL buffer B (PBS containing 4% SDS, 5 mM EDTA, 8.9 M urea, and protease inhibitors). Proteins were then reduced by the addition of 25 mM Bond-Breaker™ TCEP (0.5 M stock solution, Thermo Fisher) and incubation at RT for 1 h. To block free thiols, freshly prepared N-ethylmaleimide (NEM) in 100% ethanol was added to lysates (to 50 mM), and the mixture was rotated end-over-end for 3 h at RT. Following 2× chloroform-methanol precipitation, samples were divided into +hydroxylamine (HA) and −HA groups, which were exposed to three volumes of HA-containing buffer (1 M HA to expose palmitoylated cysteine residues) or tris-buffer control (−HA), respectively, for 1 h at 37°C. Following chloroform-methanol precipitation, the samples were solubilized and exposed to 10 mM TCEP and 20 mM mPEG-5 k (Laysan Bio Inc., see Table 4) for 1 h at RT with shaking (thereby replacing palmitic acid with mPEG-5K on exposed cysteine residues). Following the final chloroform-methanol precipitation, samples were solubilized in 70 μL of PBS containing 1% SDS, and protein concentration was measured by BCA assay (Pierce). Samples were then brought to 20 μg protein in Laemmli buffer with 2% β-mercaptoethanol for immunoblot analyses as above. Quantification of palmitoylated vs. non-palmitoylated protein was carried out for standard immunoblot analysis, with the additional consideration that signal from palmitoylated bands demonstrating the APEGS-dependent molecular weight shift was divided by the signal from the non-palmitoylated band, the location of which was verified by matching to the −HA control sample. This ratio was divided by β-actin control from the same lane for normalization.

For primary cortical neuron cultures, embryos at embryonic day 15.5 from timed-pregnant Ppt1^−/+ dams were removed, decapitated under anesthesia, and cortices resected. All dissection steps were performed in ice-cold HBSS, pH 7.4. Following cortical resection, tissue from each embryo was individually collected in a separate microtube, genotyped, and digested in HBSS containing 20 U/mL papain and DNAse at 37°C (20 min total; tubes flicked at 10 min) before sequential trituration with 1 mL (~15 strokes) and 200 μL (~10 strokes) pipettes, generating a single-cell suspension. For live-cell/immunohistochemical experiments, cells were counted and then plated at 150,000–180,000 cells/well in 24-well plates containing poly-d-lysine/laminin-coated coverslips. Plated cells were incubated at 37°C in plating medium (neurobasal medium containing B27 supplement, l-glutamine, and glutamate) for 3–5 DIV, before replacing half medium every 3 days with feeding medium (plating medium without glutamate). For synaptic scaling experiments, neurons were treated with either bicuculline (20 mM, solubilized in DMSO, Tocris) or TTX citrate (1 mM, solubilized in sterile water, Tocris) for 48 h where indicated.

Analysis of NFAT nuclear translocation following culture-wide depolarization was performed as in Murphy et al. (2014) with minor modification. Neurons were transfected at DIV 12 with a 1:1 mixture of CAG-mCherry and mouse GFP-NFATc3 (pCMV-SGFP2-mNFATc3.dna, courteously provided by Dr. Dell’Acqua), 1 mg DNA per coverslip using Lipofectamine^® 2000 (Thermo Fisher) as above. A subset of neurons was treated with TTX (1 μM) for 48 h leading up to the assay to induce synaptic scaling. Solutions were composed as follows: Tyrode + TTX (in mM: 135 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl2, 25 HEPES, and 10 glucose, pH 7.4 and 1 mM TTX), depolarization solution was isotonic, but with 50 mM KCl (85 mM NaCl) and without TTX, and recovery solution was standard Tyrode without TTX. Cells were incubated at 37°C in between steps. For every coverslip of depolarized (KCl) neurons, a control with no depolarization (sham, Tyrode solution + TTX) and depolarization with the selective GluA1-containing AMPAR blocker NASPM (KCl + 10 μM NASPM) added were performed in parallel. Following fixation at the end of the assay, cells were immunostained with anti-GFP 488 antibody (Thermo Fisher) following the protocol above to amplify the GFP-NFAT signal. GFP-NFAT nucleus/soma ratio was analyzed by manually tracing the nucleus, based on the DAPI staining signal, and soma as independent ROIs and dividing the integrated fluorescence value (measured by multi-measure tool in Fiji) for the nucleus by that for the soma.

Animals were treated with FK506 at a dose of 3 mg/kg/day (Muthuraman and Sood, 2010), similar to a previous study (Yoshiyama et al., 2007). Specifically, FK506 (Selleck Chemicals) was solubilized in DMSO at a concentration of 9 mg/mL to make stock solution aliquots. The FK506 stock was solubilized into sucrose-supplemented (2%) drinking water in mice at a dilution of 1:400 (bringing the final DMSO concentration to 0.25%) and changed every other day. Considering animals drink ~4 mL/day, they are expected to consume 0.09 mg/day via this passive administration route, or the equivalent of 3 mg/kg at an average weight of 30 g. Analysis of AFSM and gliosis in FK506-treated animals was performed as mentioned above.

Each animal performed three trials on a fixed speed rotarod, and the latency to fall was timed for each animal manually by an observer. The average latency to fall was calculated for each subject.

A limitation of our previous studies demonstrating that loss of Ppt1 function impairs the function and plasticity of synaptic receptors (Koster et al., 2019, 2023) is that it remains largely unclear how these early-disease changes contribute to classical disease features, like AFSM accumulation or neuroinflammation. To uncover dysregulated molecular pathways in CLN1 that might provide a link between these processes, we performed a palmitoyl-proteomic screen using the acyl-biotin exchange method (Drisdel and Green, 2004; Roth et al., 2006; Wan et al., 2007) on visual cortical lysates and synaptosomes from young (postnatal day (P) 42) Ppt1^−/− mice. We chose mice at this timepoint to capture early changes to synaptic protein palmitoylation, prior to the onset of severe disease.

This analysis detected 512 palmitoylated proteins in lysates and 596 in synaptosomes from WT and Ppt1^−/− visual cortices (Figures 1A,B). In lysates, we did not detect any significant changes in the abundance ratio (Ppt1^−/−/WT) for any proteins (Figure 1A). In synaptosomes, we detected significantly increased palmitoylation levels of only two proteins, acid ceramidase and cathepsin D, demonstrating that these proteins are consistently overrepresented in the CLN1 brain across several studies (Figure 1B; Chandra et al., 2015; Sleat et al., 2017; Atiskova et al., 2019; Gorenberg et al., 2022). Despite only a few proteins showing statistically significant excessive palmitoylation, we found that most of the identified proteins showed modest over-palmitoylation in the Ppt1^−/− brain, particularly in synaptosomes (Figures 1A,B). Specifically, in Ppt1^−/− lysates, 144 proteins (28.1%) show a > 1.2-fold increase in the raw abundance ratio compared to WT, while only 13 (2.5%) show a < 0.80-fold change (Figure 1C). This trend is even more robust in Ppt1^−/− synaptosomes, with 380 proteins (63.8%) demonstrating a > 1.2-fold change increase, and only 31 (5.2%) showing a < 0.80-fold change reduction (Figure 1D). These data indicate a bulk increase in the palmitoylation level of synaptic proteins in young mice with loss of Ppt1 function, consistent with Ppt1 regulating a substantial proportion of the synaptic palmitome (Gorenberg et al., 2022).

To identify the characteristics of synaptic proteins that are overrepresented in the palmitoyl fraction of Ppt1^−/− visual cortices, we performed a synapse-specific gene ontology analysis using SynGO (Koopmans et al., 2019). For both datasets, SynGO analysis demonstrated a robust enrichment for synaptic proteins compared to a whole-brain proteomic background dataset (Figures 1E,F), emphasizing the role of palmitoylation at the synapse (Sanders et al., 2015). The top 10 enriched SynGO terms demonstrated substantial overlap between lysates and synaptosomes, which is expected given the enrichment for palmitoylated proteins in both populations. “Process in the synapse,” “process in the presynapse,” “structural constituent of the postsynapse,” “process in the postsynapse,” and “regulation of postsynaptic receptor levels” are represented in the top 10 enriched terms in both lysates and synaptosomes (Figure 1G and Supplementary Figure 1A).

To highlight specific pathways that are dysregulated by a lack of Ppt1, we performed a network analysis on all genes annotated with the top SynGO term “process in the synapse” using the online protein–protein interaction database, STRING (Snel et al., 2000; Szklarczyk et al., 2017). As expected from the SynGO analysis, the networks in lysates and synaptosomes demonstrated substantial overlap (Figure 1H and Supplementary Figure 1B). We took note of a particular cluster that appeared with minor variation in both lysates and synaptosomes, consisting of Akap5, the cAMP-dependent protein kinase subunits Prkacb, Prkar1b, and Prkar2b, the protein kinase C subunits Prkcb and Prkcg, and the calcineurin subunit Ppp3ca (Figure 1H, red circle).

The Akap5 gene encodes the A-kinase anchoring protein 5 (Akap5), a postsynaptic scaffolding protein that anchors protein kinase A, protein kinase C, and calcineurin to the postsynaptic density, where it indirectly interacts with AMPARs (Colledge et al., 2000; Robertson et al., 2009). Through these interactions, Akap5 can link synaptic activity to downstream signaling (Sanderson and Dell’Acqua, 2011), making it a top candidate for a potential link between aberrant synaptic plasticity and long-term pathogenic changes to the Ppt1^−/− brain. This notion is further supported by two related lines of evidence. First, that Akap5 regulates calcium-permeable (CP)-AMPAR incorporation during synaptic scaling (Diering et al., 2014; Sanderson et al., 2018), and second, our recent finding that synaptic upscaling is affected by loss of Ppt1 function, particularly in vitro (Koster et al., 2023). Therefore, we focused on whether Akap5 levels are dysregulated in the Ppt1^−/− brain.

We first measured Akap5 levels across cortical development (P11–P60) and did not detect changes in the total levels of Akap5 in lysates or synaptosomes of Ppt1^−/− visual cortices (Figures 2A,B). Furthermore, we did not detect increased Akap5 palmitoylation in visual cortical lysates at P42 using the acyl-PEGyl exchange gel shift (APEGS) assay (Figure 2C; Yokoi et al., 2016; Kanadome et al., 2019). However, performing the APEGS assay in visual cortical synaptosomes revealed that Akap5 is excessively palmitoylated at Ppt1^−/− synapses at this age (Figure 2D). Extending our previous in vitro observations (Koster et al., 2019), we also demonstrated increased GluN2B palmitoylation in Ppt1^−/− visual cortical synaptosomes (Supplementary Figure 2).

As mentioned, converging lines of evidence highlight the centrality of Akap5 and its associated signaling molecules in regulating synaptic scaling (Diering et al., 2014; Sanderson et al., 2018). Consistent with these accumulating data, we demonstrate here an over-palmitoylation of Akap5 in Ppt1^−/− brains and have shown previously that loss of Ppt1 leads to exaggerated synaptic upscaling of CP-AMPARs. However, the bulk of our prior synaptic upscaling experiments were performed in cortical neurons (Koster et al., 2023). Therefore, to test if this plasticity mechanism was also affected by loss of Ppt1 function in vivo, we employed a dark rearing (dark rearing/dark reared, DR) protocol in WT and Ppt1^−/− mice that reliably induces upscaling in the visual cortex (Goel et al., 2006, 2011; Goel and Lee, 2007).

Corroborating previous findings (Goel et al., 2006, 2011; Goel and Lee, 2007; Diering et al., 2014), levels of synaptosomal GluA1, but not GluA2 or Akap5, were increased in both DR-WT and DR-Ppt1^−/− visual cortices compared to their light reared (light rearing/light reared, LR) counterparts (Figures 3A,B). Notably, the increase of GluA1 levels in DR-Ppt1^−/− cortices exceeded those in DR-WT animals (Figures 3A,B), indicating synaptic upscaling of GluA1-containing AMPARs is exaggerated in vivo as well as in vitro (Koster et al., 2023).

Upon synaptic calcium influx, like that from CP-AMPARs, activation of Akap5-associated calcineurin leads to dephosphorylation of the nuclear factor of activated T cells (NFAT) (Sanderson and Dell’Acqua, 2011; Li et al., 2012; Murphy et al., 2019). In turn, NFAT undergoes nuclear translocation and regulates transcription (Pan et al., 2012). Therefore, we hypothesized that an overload of palmitoylated Akap5 and excessive CP-AMPAR upregulation leads to the NFAT pathway being sensitized in upscaled Ppt1^−/− neurons.

To test this hypothesis, we performed a nuclear translocation assay as described in Murphy et al. (2014), following 48-h pretreatment of TTX to induce upscaling in WT and Ppt1^−/− neurons (Figure 4A). Following synaptic upscaling, WT and Ppt1^−/− neurons were depolarized (acute incubation in isotonic high K⁺ solution) and fixed for assessment of NFAT [specifically, the NFATc3 isoform (Murphy et al., 2014)] nuclear translocation (Figures 4B,C). While high K⁺-induced depolarization triggered the nuclear translocation of NFAT-GFP in both WT and Ppt1^−/− neurons, the nuclear/soma ratio was significantly higher in Ppt1^−/− cells, indicating a greater responsiveness to depolarization (Figures 4B,C). To test whether this effect resulted from an increased contribution of CP-AMPARs in upscaled Ppt1^−/− neurons, we performed the same assay and treated a subset of neurons with NASPM (10 μM), the CP-AMPAR selective antagonist (Koike et al., 1997), during the depolarization period. We found that NASPM reduced NFAT nuclear translocation to a greater degree in Ppt1^−/− cells compared to WT (Figures 4B,C). Together, these data suggest that calcium influx through CP-AMPARs more robustly drives downstream signaling through calcineurin and NFAT, which are both anchored by Akap5, in upscaled Ppt1^−/− neurons.

Although the panel of NFAT-target transcripts is incompletely characterized in neurons (Jayanthi et al., 2005; Luoma and Zirpel, 2008; Vashishta et al., 2009; Gómez-Sintes and Lucas, 2010), NFAT activation is associated with proinflammatory signaling in immune cell types, including microglia (Nagamoto-Combs and Combs, 2010; Minematsu et al., 2011; Pan et al., 2012). Neuroinflammation is a key correlate of disease severity in CLN1 and Ppt1^−/− mice (Kielar et al., 2007; Palmer et al., 2013; Macauley et al., 2014). Therefore, we next tested whether excessive synaptic upscaling of GluA1 via DR was associated with the severity of neuroinflammation in Ppt1^−/− mice.

We immunostained LR-WT, LR-Ppt1^−/−, DR-WT, and DR-Ppt1^−/− brains for the microglia marker, Iba1 (Figure 5A), and performed morphological (i.e., Sholl) analysis on individual cells in the visual cortex (Figures 5B,C), since a classical sign of inflammatory microglial activation is a loss of ramification and ameboid shape (Giulian, 1987; Gehrmann et al., 1995). At P33 and P42, there were no differences between LR-WT and LR-Ppt1^−/− cells in either the number of intersections as a function of distance from the cell soma (Figure 5D) or in terms of the total number of intersections (Figure 5E). By P60, however, there was a significant reduction in the number of intersections in LR-Ppt1^−/− mice compared to LR-WT that persisted to P78, suggesting microglial morphology analysis is a sensitive measure of the neuroinflammatory phenotype in CLN1 mice (Figures 5D,E).

Importantly, as early as P33 in the visual cortex, DR-Ppt1^−/− microglia exhibited a decrease in the total number of intersections compared to WT cells (Figure 5E). At P42, before any neurological deficit was detected in LR-Ppt1^−/− mice, DR-Ppt1^−/− mice exhibited significantly reduced microglial processes compared to LR-WT and DR-WT. Moreover, LR-Ppt1^−/− animals (Figure 5D,E) demonstrated an acceleration of the neuroinflammatory phenotype in DR animals. This effect was even more robust at P78 (Figure 5D,E). Similar to previous findings (Sadhukhan et al., 2021), we also detected an increase in the number of microglia in LR-Ppt1^−/− mice compared to WT counterparts at P78, though this effect was not exacerbated by DR (Figure 5F). These data demonstrate that DR worsens neuroinflammation in the visual cortex of Ppt1^−/− mice.

Gliosis and neurodegeneration follow a systematic pattern in CLN1, affecting the thalamus (especially the visual thalamus) before the cortex (Kielar et al., 2007). Therefore, we also examined the neuroinflammatory phenotype of microglia in the dorsal lateral geniculate nucleus (dLGN) at ~P33 before detectable changes occurred in the cortex (Supplementary Figures 3A–C). Microglia in this region are crucial for synaptic refinement during the development of the visual circuit (Shatz, 1990; Katz and Shatz, 1996), so we focused our analysis on microglia at the borders of ipsilateral and contralateral retinal projections reaching the dLGN. We found a subtle but significant change in microglia morphology that suggests an emerging neuroinflammatory phenotype in the dLGN of LR-Ppt1^−/− mice (Supplementary Figures 3A–C).

The regulation of synaptic calcium homeostasis, including calcium entry through CP-AMPARs, is crucial for neuronal health and prevention of epileptogenesis (McNamara et al., 2006). In addition, gliosis is not only a histopathological correlate of CLN1 disease progression but is associated with the onset of global symptoms, like seizures, in CLN1 mouse models (Zhang et al., 2022), where it may play a causative role (Vezzani et al., 2022). Having demonstrated an exaggerated synaptic upscaling phenotype, which alters neuronal calcium fluctuation (Koster et al., 2023), as well as exacerbated gliosis in DR-Ppt1^−/− animals, we next examined whether these pathophysiological changes influenced CLN1 disease severity.

Indeed, concomitant with excessive synaptic upscaling, the accumulation of AFSM increased in the visual cortex (Figures 6A,B). Furthermore, although dark rearing is known to slightly reduce visual cortical thickness in WT animals (Fifková, 1970; Takács et al., 1992), we observed a statistically significant reduction only in DR-Ppt1^−/− mice compared to WT animals raised under standard conditions (Figure 6C). Importantly, mortality occurred significantly earlier because of severe seizures (see Materials and Methods) in DR-Ppt1^−/− animals (Figure 6D). Thus, the lack of Ppt1-mediated depalmitoylation drives an exaggerated synaptic upscaling of GluA1 and enhanced neuroinflammation in vivo that corresponds to exacerbated CLN1 pathology. These results are consistent with a substantive connection between aberrant synaptic plasticity and CLN1 disease progression; however, further experiments are required to determine the causative agents (i.e., does excessive CP-AMPAR upregulation itself worsen disease).

Considering that the Akap5 complex is dysregulated in the Ppt1^−/− brain, NFAT activation is sensitized in upscaled Ppt1^−/− neurons, and in vivo induction of synaptic upscaling exacerbates disease symptoms, we reasoned that suppressing calcineurin activity should alleviate CLN1 progression. Therefore, we treated Ppt1^−/− animals from 1 to 4 months of age with the calcineurin inhibitor, FK506 (3 mg/kg dissolved in drinking water), which is an FDA-approved immunosuppressant under the trade names Tacrolimus or Prograf. Passive administration of FK506 in Ppt1^−/− mice demonstrated no effect on the degree of AFSM accumulation (Figures 7A,B). In addition, FK506 did not significantly suppress microglial activation as measured by the total number of intersections with a Sholl analysis (Figure 7C). However, further scrutiny of the data indicated that while inflammatory activation of microglia in both vehicle- and FK506-treated Ppt1^−/− mice was robust, FK506 treatment suppressed the transformation of microglia into a fully ameboid phenotype (Figure 7A). Accordingly, comparison of the Sholl profile (Figure 7D, left) revealed a greater number of microglia with ramifications at intermediate distances from the cell soma, which were largely absent in vehicle-treated mice (Figure 7D, right). Finally, as motor dysfunction is a primary symptom of CLN1 in humans and mice (Gupta et al., 2003; Dearborn et al., 2015; Nita et al., 2016), we next tested whether FK506 treatment improves motor symptoms in Ppt1^−/− mice using a Rotarod assay. Under these conditions, FK506 treatment appeared to slightly improve motor behavior compared to vehicle-treated Ppt1^−/− counterparts, though the effect did not quite reach statistical significance (p = 0.059, Student’s t-test vs. vehicle-treated Ppt1^−/− mice) (Figure 7E). Collectively, these results show that suppressing calcineurin activity via FK506 treatment may modestly slow neuroinflammatory progression in Ppt1^−/− mice.

While we maintain that over-palmitoylation of GluA1 likely contributes to the increased incorporation of CP-AMPARs at upscaled Ppt1^−/− synapses (Koster et al., 2023), we demonstrate here that Akap5 is also hyperpalmitoylated in Ppt1^−/− synaptosomes. Thus, the increased synaptic incorporation of GluA1 during upscaling may also arise from enhanced scaffolding by Akap5. Previous experiments implicate Akap5 in the regulation of CP-AMPARs during synaptic scaling (Sanderson et al., 2018). In addition, palmitoylation of Akap5 is required for the postsynaptic insertion of CP-AMPARs during LTP (Purkey et al., 2018); although this study also notes that palmitoylated Akap5 limits the synaptic incorporation of CP-AMPARs under basal conditions, suggesting that Akap5 palmitoylation differentially effects CP-AMPAR trafficking dynamics at a steady state compared to bouts of plasticity. Akap5 also requires depalmitoylation for its removal from the postsynaptic site and undergoes ubiquitination to downregulate GluA1-containing AMPARs during chemical LTD (Woolfrey et al., 2018; Cheng et al., 2020). Therefore, we anticipate that Akap5 normally undergoes depalmitoylation-dependent degradation (Cheng et al., 2020) and that this mechanism is diminished in Ppt1^−/− neurons. Consequently, we postulate that overly palmitoylated Akap5 harbors an enlarged perisynaptic pool of CP-AMPARs that is mobilized during synaptic upscaling in Ppt1^−/− neurons either at the extrasynapse or dendritic endosomes (He et al., 2009; Diering et al., 2014). Recent detailed analysis of the synaptic and perisynaptic localization of palmitoylated Akap5 agrees with such a notion (Chen et al., 2022).

Our data also indicate that the over-palmitoylation of Akap5 leads to a sensitization of the NFAT transcriptional program in Ppt1^−/− neurons. Not only does this provide a mechanistic link between synaptic alterations and gliosis in CLN1 (Jalanko et al., 2005; Kielar et al., 2007; Macauley et al., 2009, 2011, 2014), but, in addition, the calcineurin-NFAT pathway itself mediates synaptic scaling through the turnover of CP-AMPARs (Kim and Ziff, 2014). Therefore, mis-localization of Akap5 or impaired interactions with calcineurin due to loss of Ppt1 likely drives over-activation of NFAT-dependent transcription, causing a vicious cycle of exaggerated upscaling of CP-AMPARs in the Ppt1^−/− brain. In sum, several related mechanisms may collaborate to drive exaggeration of CP-AMPAR incorporation during synaptic scaling up in Ppt1^−/− neurons.

One motivation for employing the DR paradigm in Ppt1^−/− mice was brought on by previous work, which demonstrated that dark-rearing mouse models of other developmental disorders, like Rett and Angelman syndromes, mitigated disease symptoms (Yashiro et al., 2009; Durand et al., 2012). Accordingly, we originally anticipated an improvement in CLN1 pathophysiology in DR-Ppt1^−/− mice. However, our study demonstrates that DR-Ppt1^−/− animals have the opposite effect, exacerbating several markers of disease progression, including mortality. Why might this be the case?

Before the demonstration of its beneficial effects in neurodevelopmental disease models, DR was established as a model of synaptic upscaling in vivo (Goel et al., 2006, 2011; Goel and Lee, 2007). Here, we corroborate these data by demonstrating that DR of Ppt1^−/− mice exaggerates synaptic upscaling of GluA1, extending our previous findings (Koster et al., 2023) that used exogenous manipulations (e.g., TTX induction of upscaling). This is precisely the opposite effect conferred by loss of either Ube3a (Angelman syndrome) or MeCP2 (Rett syndrome), which suppresses upscaling (Blackman et al., 2012; Qiu et al., 2012; Pastuzyn and Shepherd, 2017). Therefore, we postulate that the opposing effects of DR on disease pathology in CLN1 versus Rett or Angelman syndrome mice arise from the differential impact that the proteins mutated in these conditions have on synaptic scaling. In other words, whereas DR alleviated symptoms in Rett and Angelman syndrome mice in part by overcoming the stagnation of synaptic upscaling, this same manipulation proved detrimental in the CLN1 model because it facilitated an exaggerated synaptic upscaling. Still, taking the broader context reveals that several proteins implicated in neurodevelopmental disorders regulate the common pathway of synaptic scaling and imply that normalizing synaptic scaling dynamics might alleviate disease symptoms.

The sensitization of calcineurin-NFAT activity in Ppt1^−/− neurons represents a new therapeutic target in CLN1 that has a history of being inhibited in other conditions requiring immunosuppression with FDA-approved drugs such as Tacrolimus (FK506) and cyclosporin. Drug repurposing is a major priority of the FDA (along with the incentives for orphan drug discovery) due to the economics and relative speed of getting potential therapeutics to patients relative to the traditional approach, which typically takes hundreds of millions of dollars and 10 or more years. Therefore, our demonstration here that FK506 has beneficial effects in CLN1 mice is a substantial first step toward its use in the clinic for patients suffering from CLN1. Although we observed quite a modest effect here, we offer the following caveat. Our aim was to perform a small pilot study of FK506 efficacy that employed several readouts, not just histology or behavior, but both, and as such, we chose a terminal time point for our cohort of 4 months with the hope of capturing such an effect. However, motor deficits in Ppt1^−/− mice emerge between 3 and 5 months, often reaching statistical significance by the 5-month mark (Macauley et al., 2009; Groh et al., 2021), while our own data show that neuroinflammation is robust by 3 months. Therefore, by attempting to capture the potential effects of FK506 across these two measures, we may have limited our power to detect either. Accordingly, our data warrant further, more detailed study of FK506 in CLN1.

Interestingly, Tacrolimus has demonstrated beneficial effects in multiple animal models of aging or neurodegeneration (Radhakrishnan et al., 2021; Fracassi et al., 2022; Sordo et al., 2022) and was the subject of an open-label clinical trial (NCT04263519). Moreover, human patients prescribed calcineurin inhibitors, including Tacrolimus, show reduced prevalence of developing Alzheimer’s disease (Silva et al., 2023). These studies suggest Tacrolimus or next-generation analogs may be a promising intervention in multiple neurodegenerative diseases. Future studies focused on testing the efficacy of FK506 or other calcineurin-directed therapeutics in a larger preclinical setting or in CLN1 patients will be of great interest.

In conclusion, we demonstrate here with an interdisciplinary approach how failed protein depalmitoylation disrupts synaptic plasticity via impaired proteostasis and triggers neuroinflammatory signaling that underlies a devastating disease. Importantly, these findings offer multiple novel targets for therapeutic intervention, at least one of which (calcineurin inhibitors) can benefit from the FDA and clinical priority to repurpose well-tolerated drugs and can therefore be more rapidly tested in suffering patients.