Differential protein-protein interactions underlie signaling mediated by the TCR and a 4-1BB domain-containing CAR.

Cells rely on activity-dependent protein-protein interactions to convey biological signals. For chimeric antigen receptor (CAR) T cells containing a 4-1BB costimulatory domain, receptor engagement is thought to stimulate the formation of protein complexes similar to those stimulated by T cell receptor (TCR)-mediated signaling, but the number and type of protein interaction-mediating binding domains differ between CARs and TCRs. Here, we performed coimmunoprecipitation mass spectrometry analysis of a second-generation, CD19-directed 4-1BB:ζ CAR (referred to as bbζCAR) and identified 128 proteins that increased their coassociation after target engagement. We compared activity-induced TCR and CAR signalosomes by quantitative multiplex coimmunoprecipitation and showed that bbζCAR engagement led to the activation of two modules of protein interactions, one similar to TCR signaling that was more weakly engaged by bbζCAR as compared with the TCR and one composed of TRAF signaling complexes that was not engaged by the TCR. Batch-to-batch and interindividual variations in production of the cytokine IL-2 correlated with differences in the magnitude of protein network activation. Future CAR T cell manufacturing protocols could measure, and eventually control, biological variation by monitoring these signalosome activation markers.

SRC family kinase inhibition rescues molecular and behavioral phenotypes, but not protein interaction network dynamics, in a mouse model of Fragile X syndrome.

Glutamatergic synapses encode information from extracellular inputs using dynamic protein interaction networks (PINs) that undergo widespread reorganization following synaptic activity, allowing cells to distinguish between signaling inputs and generate coordinated cellular responses. Here, we investigate how Fragile X Messenger Ribonucleoprotein (FMRP) deficiency disrupts signal transduction through a glutamatergic synapse PIN downstream of NMDA receptor or metabotropic glutamate receptor (mGluR) stimulation. In cultured cortical neurons or acute cortical slices from P7, P17 and P60 FMR1-/y mice, the unstimulated protein interaction network state resembled that of wildtype littermates stimulated with mGluR agonists, demonstrating resting state pre-activation of mGluR signaling networks. In contrast, interactions downstream of NMDAR stimulation were similar to WT. We identified the Src family kinase (SFK) Fyn as a network hub, because many interactions involving Fyn were pre-activated in FMR1-/y animals. We tested whether targeting SFKs in FMR1-/y mice could modify disease phenotypes, and found that Saracatinib (SCB), an SFK inhibitor, normalized elevated basal protein synthesis, novel object recognition memory and social behavior in FMR1-/y mice. However, SCB treatment did not normalize the PIN to a wild-type-like state in vitro or in vivo, but rather induced extensive changes to protein complexes containing Shank3, NMDARs and Fyn. We conclude that targeting abnormal nodes of a PIN can identify potential disease-modifying drugs, but behavioral rescue does not correlate with PIN normalization.

Protein interaction network analysis of mTOR signaling reveals modular organization.

The mammalian target of rapamycin (mTOR) is a serine-threonine kinase that acts as a central mediator of translation and plays important roles in cell growth, synaptic plasticity, cancer, and a wide range of developmental disorders. The signaling cascade linking lipid kinases (phosphoinositide 3-kinases), protein kinases (AKT), and translation initiation complexes (EIFs) to mTOR has been extensively modeled, but does not fully describe mTOR system behavior. Here, we use quantitative multiplex coimmunoprecipitation to monitor a protein interaction network (PIN) composed of 300+ binary interactions among mTOR-related proteins. Using a simple model system of serum-deprived or fresh-media-fed mouse 3T3 fibroblasts, we observed extensive PIN remodeling involving 27+ individual protein interactions after 1 h, despite phosphorylation changes observed after only 5 min. Using small molecule inhibitors of phosphoinositide 3-kinase, AKT, mTOR, MEK and ERK, we define subsets of the PIN, termed "modules", that respond differently to each inhibitor. Using primary fibroblasts from individuals with overgrowth disorders caused by pathogenic PIK3CA or MTOR variants, we find that hyperactivation of mTOR pathway components is reflected in a hyperactive PIN. Our data define a "modular" organization of the mTOR PIN in which coordinated groups of interactions respond to the activation or inhibition of distinct nodes, and demonstrate that kinase inhibitors affect the modular network architecture in a complex manner, inconsistent with simple linear models of signal transduction.

Protein interaction network analysis of mTOR signaling reveals modular organization.

The mammalian target of rapamycin (mTOR) is a serine-threonine kinase that acts as a central mediator of translation, and plays important roles in cell growth, synaptic plasticity, cancer, and a wide range of developmental disorders. The signaling cascade linking lipid kinases (PI3Ks), protein kinases (AKT) and translation initiation complexes (EIFs) to mTOR has been extensively modeled, but does not fully describe mTOR system behavior. Here, we use quantitative multiplex co-immunoprecipitation to monitor a protein interaction network (PIN) composed of 300+ binary interactions among mTOR-related proteins. Using a simple model system of serum deprived or fresh-media-fed mouse 3T3 fibroblasts, we observed extensive PIN remodeling involving 27+ individual protein interactions after one hour, despite phosphorylation changes observed after only five minutes. Using small molecule inhibitors of PI3K, AKT, mTOR, MEK and ERK, we define subsets of the PIN, termed 'modules', that respond differently to each inhibitor. Using primary fibroblasts from individuals with overgrowth disorders caused by pathogenic PIK3CA or MTOR variants, we find that hyperactivation of mTOR pathway components is reflected in a hyperactive PIN. Our data define a "modular" organization of the mTOR PIN in which coordinated groups of interactions respond to activation or inhibition of distinct nodes, and demonstrate that kinase inhibitors affect the modular network architecture in a complex manner, inconsistent with simple linear models of signal transduction.

Activity dependent dissociation of the Homer1 interactome.

Neurons encode information by rapidly modifying synaptic protein complexes, which changes the strength of specific synaptic connections. Homer1 is abundantly expressed at glutamatergic synapses, and is known to alter its binding to metabotropic glutamate receptor 5 (mGlu5) in response to synaptic activity. However, Homer participates in many additional known interactions whose activity-dependence is unclear. Here, we used co-immunoprecipitation and label-free quantitative mass spectrometry to characterize activity-dependent interactions in the cerebral cortex of wildtype and Homer1 knockout mice. We identified a small, high-confidence protein network consisting of mGlu5, Shank2 and 3, and Homer1-3, of which only mGlu5 and Shank3 were significantly reduced following neuronal depolarization. We identified several other proteins that reduced their co-association in an activity-dependent manner, likely mediated by Shank proteins. We conclude that Homer1 dissociates from mGlu5 and Shank3 following depolarization, but our data suggest that direct Homer1 interactions in the cortex may be more limited than expected.

Non-synaptic Cell-Autonomous Mechanisms Underlie Neuronal Hyperactivity in a Genetic Model of PIK3CA-Driven Intractable Epilepsy.

Patients harboring mutations in the PI3K-AKT-MTOR pathway-encoding genes often develop a spectrum of neurodevelopmental disorders including epilepsy. A significant proportion remains unresponsive to conventional anti-seizure medications. Understanding mutation-specific pathophysiology is thus critical for molecularly targeted therapies. We previously determined that mouse models expressing a patient-related activating mutation in PIK3CA, encoding the p110α catalytic subunit of phosphoinositide-3-kinase (PI3K), are epileptic and acutely treatable by PI3K inhibition, irrespective of dysmorphology. Here we report the physiological mechanisms underlying this dysregulated neuronal excitability. In vivo, we demonstrate epileptiform events in the Pik3ca mutant hippocampus. By ex vivo analyses, we show that Pik3ca-driven hyperactivation of hippocampal pyramidal neurons is mediated by changes in multiple non-synaptic, cell-intrinsic properties. Finally, we report that acute inhibition of PI3K or AKT, but not MTOR activity, suppresses the intrinsic hyperactivity of the mutant neurons. These acute mechanisms are distinct from those causing neuronal hyperactivity in other AKT-MTOR epileptic models and define parameters to facilitate the development of new molecularly rational therapeutic interventions for intractable epilepsy.

Synaptic protein interaction networks encode experience by assuming stimulus-specific and brain-region-specific states.

A core network of widely expressed proteins within the glutamatergic post-synapse mediates activity-dependent synaptic plasticity throughout the brain, but the specific proteomic composition of synapses differs between brain regions. Here, we address the question, how does proteomic composition affect activity-dependent protein-protein interaction networks (PINs) downstream of synaptic activity? Using quantitative multiplex co-immunoprecipitation, we compare the PIN response of in vivo or ex vivo neurons derived from different brain regions to activation by different agonists or different forms of eyeblink conditioning. We report that PINs discriminate between incoming stimuli using differential kinetics of overlapping and non-overlapping PIN parameters. Further, these "molecular logic rules" differ by brain region. We conclude that although the PIN of the glutamatergic post-synapse is expressed widely throughout the brain, its activity-dependent dynamics show remarkable stimulus-specific and brain-region-specific diversity. This diversity may help explain the challenges in developing molecule-specific drug therapies for neurological disorders.

A Repurposed Drug Screen Identifies Compounds That Inhibit the Binding of the COVID-19 Spike Protein to ACE2.

Repurposed drugs that block the interaction between the SARS-CoV-2 spike protein and its receptor ACE2 could offer a rapid route to novel COVID-19 treatments or prophylactics. Here, we screened 2,701 compounds from a commercial library of drugs approved by international regulatory agencies for their ability to inhibit the binding of recombinant, trimeric SARS-CoV-2 spike protein to recombinant human ACE2. We identified 56 compounds that inhibited binding in a concentration-dependent manner, measured the IC50 of binding inhibition, and computationally modeled the docking of the best inhibitors to the Spike-ACE2 binding interface. The best candidates were Thiostrepton, Oxytocin, Nilotinib, and Hydroxycamptothecin with IC50's in the 4-9 µM range. These results highlight an effective screening approach to identify compounds capable of disrupting the Spike-ACE2 interaction, as well as identify several potential inhibitors of the Spike-ACE2 interaction.

Remodeling of the Homer-Shank interactome mediates homeostatic plasticity.

Neurons maintain stable levels of excitability using homeostatic synaptic scaling, which adjusts the strength of a neuron's postsynaptic inputs to compensate for extended changes in overall activity. Here, we investigated whether prolonged changes in activity affect network-level protein interactions at the synapse. We assessed a glutamatergic synapse protein interaction network (PIN) composed of 380 binary associations among 21 protein members in mouse neurons. Manipulating the activation of cultured mouse cortical neurons induced widespread bidirectional PIN alterations that reflected rapid rearrangements of glutamate receptor associations involving synaptic scaffold remodeling. Sensory deprivation of the barrel cortex in live mice (by whisker trimming) caused specific PIN rearrangements, including changes in the association between the glutamate receptor mGluR5 and the kinase Fyn. These observations are consistent with emerging models of experience-dependent plasticity involving multiple types of homeostatic responses. However, mice lacking Homer1 or Shank3B did not undergo normal PIN rearrangements, suggesting that the proteins encoded by these autism spectrum disorder-linked genes serve as structural hubs for synaptic homeostasis. Our approach demonstrates how changes in the protein content of synapses during homeostatic plasticity translate into functional PIN alterations that mediate changes in neuron excitability.

Altered network and rescue of human neurons derived from individuals with early-onset genetic epilepsy.

Early-onset epileptic encephalopathies are severe disorders often associated with specific genetic mutations. In this context, the CDKL5 deficiency disorder (CDD) is a neurodevelopmental condition characterized by early-onset seizures, intellectual delay, and motor dysfunction. Although crucial for proper brain development, the precise targets of CDKL5 and its relation to patients' symptoms are still unknown. Here, induced pluripotent stem cells derived from individuals deficient in CDKL5 protein were used to generate neural cells. Proteomic and phosphoproteomic approaches revealed disruption of several pathways, including microtubule-based processes and cytoskeleton organization. While CDD-derived neural progenitor cells have proliferation defects, neurons showed morphological alterations and compromised glutamatergic synaptogenesis. Moreover, the electrical activity of CDD cortical neurons revealed hyperexcitability during development, leading to an overly synchronized network. Many parameters of this hyperactive network were rescued by lead compounds selected from a human high-throughput drug screening platform. Our results enlighten cellular, molecular, and neural network mechanisms of genetic epilepsy that could ultimately promote novel therapeutic opportunities for patients.

A repurposed drug screen identifies compounds that inhibit the binding of the COVID-19 spike protein to ACE2.

Repurposed drugs that block the interaction between the SARS-CoV-2 spike protein and its receptor ACE2 could offer a rapid route to novel COVID-19 treatments or prophylactics. Here, we screened 2701 compounds from a commercial library of drugs approved by international regulatory agencies for their ability to inhibit the binding of recombinant, trimeric SARS-CoV-2 spike protein to recombinant human ACE2. We identified 56 compounds that inhibited binding by <90%, measured the EC 50 of binding inhibition, and computationally modeled the docking of the best inhibitors to both Spike and ACE2. These results highlight an effective screening approach to identify compounds capable of disrupting the Spike-ACE2 interaction as well as identifying several potential inhibitors that could serve as templates for future drug discovery efforts.

Reintroduction of the archaic variant of NOVA1 in cortical organoids alters neurodevelopment.

The evolutionarily conserved splicing regulator neuro-oncological ventral antigen 1 (NOVA1) plays a key role in neural development and function. NOVA1 also includes a protein-coding difference between the modern human genome and Neanderthal and Denisovan genomes. To investigate the functional importance of an amino acid change in humans, we reintroduced the archaic allele into human induced pluripotent cells using genome editing and then followed their neural development through cortical organoids. This modification promoted slower development and higher surface complexity in cortical organoids with the archaic version of NOVA1 Moreover, levels of synaptic markers and synaptic protein coassociations correlated with altered electrophysiological properties in organoids expressing the archaic variant. Our results suggest that the human-specific substitution in NOVA1, which is exclusive to modern humans since divergence from Neanderthals, may have had functional consequences for our species' evolution.

Mutant Huntingtin Is Cleared from the Brain via Active Mechanisms in Huntington Disease.

Huntington disease (HD) is a neurodegenerative disease caused by a CAG trinucleotide repeat expansion in the huntingtin (HTT) gene. Therapeutics that lower HTT have shown preclinical promise and are being evaluated in clinical trials. However, clinical assessment of brain HTT lowering presents challenges. We have reported that mutant HTT (mHTT) in the CSF of HD patients correlates with clinical measures, including disease burden as well as motor and cognitive performance. We have also shown that lowering HTT in the brains of HD mice results in correlative reduction of mHTT in the CSF, prompting the use of this measure as an exploratory marker of target engagement in clinical trials. In this study, we investigate the mechanisms of mHTT clearance from the brain in adult mice of both sexes to elucidate the significance of therapy-induced CSF mHTT changes. We demonstrate that, although neurodegeneration increases CSF mHTT concentrations, mHTT is also present in the CSF of mice in the absence of neurodegeneration. Importantly, we show that secretion of mHTT from cells in the CNS followed by glymphatic clearance from the extracellular space contributes to mHTT in the CSF. Furthermore, we observe secretion of wild type HTT from healthy control neurons, suggesting that HTT secretion is a normal process occurring in the absence of pathogenesis. Overall, our data support both passive release and active clearance of mHTT into CSF, suggesting that its treatment-induced changes may represent a combination of target engagement and preservation of neurons.SIGNIFICANCE STATEMENT: Changes in CSF mutant huntingtin (mHTT) are being used as an exploratory endpoint in HTT lowering clinical trials for the treatment of Huntington disease (HD). Recently, it was demonstrated that intrathecal administration of a HTT lowering agent leads to dose-dependent reduction of CSF mHTT in HD patients. However, little is known about how HTT, an intracellular protein, reaches the extracellular space and ultimately the CSF. Our findings that HTT enters CSF by both passive release and active secretion followed by glymphatic clearance may have significant implications for interpretation of treatment-induced changes of CSF mHTT in clinical trials for HD.

Plasma From Recovered COVID-19 Patients Inhibits Spike Protein Binding to ACE2 in a Microsphere-Based Inhibition Assay.

We present a microsphere-based flow cytometry assay that quantifies the ability of plasma to inhibit the binding of spike protein to angiotensin-converting enzyme 2. Plasma from 22 patients who had recovered from mild coronavirus disease 2019 (COVID-19) and expressed anti-spike protein trimer immunoglobulin G inhibited angiotensin-converting enzyme 2-spike protein binding to a greater degree than controls. The degree of inhibition was correlated with anti-spike protein immunoglobulin G levels, neutralizing titers in a pseudotyped lentiviral assay, and the presence of fever during illness. This inhibition assay may be broadly useful to quantify the functional antibody response of patients recovered from COVID-19 or vaccine recipients in a cell-free assay system.

Plasma from recovered COVID19 subjects inhibits spike protein binding to ACE2 in a microsphere-based inhibition assay.

High throughput serological tests that can establish the presence and functional activity of anti-SARS-COV2 antibodies are urgently needed. Here we present microsphere-based Flow Cytometry assays that quantify both anti-spike IgGs in plasma, and the ability of plasma to inhibit the binding of spike protein to angiotensin converting enzyme 2 (ACE2). First, we detected anti-trimer IgGs in 22/24 and anti-RBD IgGs in 21/24 COVID+ subjects at a median of 36 (range 14-73) days following documented SARS-CoV-2 RNA (+) secretions. Next, we find that plasma from all 22/24 subjects with anti-trimer IgGs inhibited ACE2-trimer binding to a greater degree than controls, and that the degree of inhibition correlated with anti-trimer IgG levels. Depletion of trimer-reactive Igs from plasma reduced ACE2-trimer inhibitory capacity to a greater degree than depletion of RBD-reactive Igs, suggesting that inhibitory antibodies act by binding both within and outside of the RBD. Amongst the 24 subjects, presence of fever was associated with higher levels of anti-trimer IgG and inhibition of binding to human ACE2. This inhibition assay may be broadly useful to quantify the functional antibody response of recovered COVID19 patients or vaccine recipients in a cell-free assay system.

Resolving the Synaptic versus Developmental Dichotomy of Autism Risk Genes.

Genes that are mutated in Autism Spectrum Disorders (ASD) can be classified broadly as either synaptic or developmental. But what if this is a false distinction? A recent spate of publications has provided evidence for developmental mechanisms that rely on neural activity for proper cortical development. Conversely, a growing body of evidence indicates a role for developmental mechanisms, particularly chromatin remodeling, during learning or in response to neural activity. Here, we review these recent publications and propose a model in which genes that confer ASD risk operate in signal transduction networks critical for both cortical development and synaptic homeostasis.

Quantification of Protein Interaction Network Dynamics using Multiplexed Co-Immunoprecipitation.

Dynamic protein-protein interactions control cellular behavior, from motility to DNA replication to signal transduction. However, monitoring dynamic interactions among multiple proteins in a protein interaction network is technically difficult. Here, we present a protocol for Quantitative Multiplex Immunoprecipitation (QMI), which allows quantitative assessment of fold changes in protein interactions based on relative fluorescence measurements of Proteins in Shared Complexes detected by Exposed Surface epitopes (PiSCES). In QMI, protein complexes from cell lysates are immunoprecipitated onto microspheres, and then probed with a labeled antibody for a different protein in order to quantify the abundance of PiSCES. Immunoprecipitation antibodies are conjugated to different MagBead spectral regions, which allows a flow cytometer to differentiate multiple parallel immunoprecipitations and simultaneously quantify the amount of probe antibody associated with each. QMI does not require genetic tagging and can be performed using minimal biomaterial compared to other immunoprecipitation methods. QMI can be adapted for any defined group of interacting proteins, and has thus far been used to characterize signaling networks in T cells and neuronal glutamate synapses. Results have led to new hypothesis generation with potential diagnostic and therapeutic applications. This protocol includes instructions to perform QMI, from the initial antibody panel selection through to running assays and analyzing data. The initial assembly of a QMI assay involves screening antibodies to generate a panel, and empirically determining an appropriate lysis buffer. The subsequent reagent preparation includes covalently coupling immunoprecipitation antibodies to MagBeads, and biotinylating probe antibodies so they can be labeled by a streptavidin-conjugated fluorophore. To run the assay, lysate is mixed with MagBeads overnight, and then beads are divided and incubated with different probe antibodies, and then a fluorophore label, and read by flow cytometry. Two statistical tests are performed to identify PiSCES that differ significantly between experimental conditions, and results are visualized using heatmaps or node-edge diagrams.

Activity-dependent changes in synaptic protein complex composition are consistent in different detergents despite differential solubility.

At the post-synaptic density (PSD), large protein complexes dynamically form and dissociate in response to synaptic activity, comprising the biophysical basis for learning and memory. The use of detergents to isolate the PSD and release its membrane-associated proteins complicates studies of these activity-dependent protein interaction networks, because detergents can simultaneously disrupt the very interactions under study. Despite widespread recognition that different detergents yield different experimental results, the effect of detergent on activity-dependent synaptic protein complexes has not been rigorously examined. Here, we characterize the effect of three detergents commonly used to study synaptic proteins on activity-dependent protein interactions. We first demonstrate that SynGAP-containing interactions are more abundant in 1% Deoxycholate (DOC), while Shank-, Homer- and mGluR5-containing interactions are more abundant in 1% NP-40 or Triton. All interactions were detected preferentially in high molecular weight complexes generated by size exclusion chromatography, although the detergent-specific abundance of proteins in high molecular weight fractions did not correlate with the abundance of detected interactions. Activity-dependent changes in protein complexes were consistent across detergent types, suggesting that detergents do not isolate distinct protein pools with unique behaviors. However, detection of activity-dependent changes is more or less feasible in different detergents due to baseline solubility. Collectively, our results demonstrate that detergents affect the solubility of individual proteins, but activity-dependent changes in protein interactions, when detectable, are consistent across detergent types.

The early proximal αß TCR signalosome specifies thymic selection outcome through a quantitative protein interaction network.

During αß T cell development, T cell antigen receptor (TCR) engagement transduces biochemical signals through a protein-protein interaction (PPI) network that dictates dichotomous cell fate decisions. It remains unclear how signal specificity is communicated, instructing either positive selection to advance cell differentiation or death by negative selection. Early signal discrimination might occur by PPI signatures differing qualitatively (customized, unique PPI combinations for each signal), quantitatively (graded amounts of a single PPI series), or kinetically (speed of PPI pathway progression). Using a novel PPI network analysis, we found that early TCR-proximal signals distinguishing positive from negative selection appeared to be primarily quantitative in nature. Furthermore, the signal intensity of this PPI network was used to find an antigen dose that caused a classic negative selection ligand to induce positive selection of conventional αß T cells, suggesting that the quantity of TCR triggering was sufficient to program selection outcome. Because previous work had suggested that positive selection might involve a qualitatively unique signal through CD3δ, we reexamined the block in positive selection observed in CD3δ0 mice. We found that CD3δ0 thymocytes were inhibited but capable of signaling positive selection, generating low numbers of MHC-dependent αß T cells that expressed diverse TCR repertoires and participated in immune responses against infection. We conclude that the major role for CD3δ in positive selection is to quantitatively boost the signal for maximal generation of αß T cells. Together, these data indicate that a quantitative network signaling mechanism through the early proximal TCR signalosome determines thymic selection outcome.

Clustering the autisms using glutamate synapse protein interaction networks from cortical and hippocampal tissue of seven mouse models.

Background: Autism spectrum disorders (ASDs) are a heterogeneous group of behaviorally defined disorders and are associated with hundreds of rare genetic mutations and several environmental risk factors. Mouse models of specific risk factors have been successful in identifying molecular mechanisms associated with a given factor. However, comparisons among different models to elucidate underlying common pathways or to define clusters of biologically relevant disease subtypes have been complicated by different methodological approaches or different brain regions examined by the labs that developed each model. Here, we use a novel proteomic technique, quantitative multiplex co-immunoprecipitation or QMI, to make a series of identical measurements of a synaptic protein interaction network in seven different animal models. We aim to identify molecular disruptions that are common to multiple models. Methods: QMI was performed on 92 hippocampal and cortical samples taken from seven mouse models of ASD: Shank3B, Shank3Δex4-9, Ube3a2xTG, TSC2, FMR1, and CNTNAP2 mutants, as well as E12.5 VPA (maternal valproic acid injection on day 12.5 post-conception). The QMI panel targeted a network of 16 interacting, ASD-linked, synaptic proteins, probing 240 potential co-associations. A custom non-parametric statistical test was used to call significant differences between ASD models and littermate controls, and Hierarchical Clustering by Principal Components was used to cluster the models using mean log2 fold change values. Results: Each model displayed a unique set of disrupted interactions, but some interactions were disrupted in multiple models. These tended to be interactions that are known to change with synaptic activity. Clustering revealed potential relationships among models and suggested deficits in AKT signaling in Ube3a2xTG mice, which were confirmed by phospho-western blots. Conclusions: These data highlight the great heterogeneity among models, but suggest that high-dimensional measures of a synaptic protein network may allow differentiation of subtypes of ASD with shared molecular pathology.