A distinct PP2A subunit regulates local protein phosphorylation at the axon initial segment

Abstract

Protein phosphorylation plays a crucial role in regulating the cytoskeletal and membrane proteins at the axon initial segment (AIS). However, our knowledge of AIS-specific kinases and phosphatases is very limited. Here, we report the identification of a protein phosphatase 2A (PP2A) B55 regulatory subunit enriched at the AIS in mice: Ppp2r2c. Our results demonstrate that PP2A-B55 subunits exhibit substantial heterogeneity in their subcellular localization and function. Notably, the Ppp2r2c subunit is selectively concentrated at the AIS, and this enrichment is driven by its unique structure. Utilizing a microelectrode array system (MEA), we show that Ppp2r2c modulates neuronal activity during in vitro development. With phosphoproteomics, we further reveal that the potassium channel Kv1.2 is one of the downstream targets that link Ppp2r2c activity to neuronal excitability. Together, these data provide a critical entry point for understanding the mechanisms of PP2A-mediated local phospho-regulation at the AIS.

The existence of phosphatases specialized in regulating the axon initial segment (AIS) was unclear. Here, the authors show that a PP2A-B55 subunit is selectively enriched at the AIS and contributes to its local phospho-regulation.

Introduction

The axon initial segment (AIS) is a distinguishing feature of the neuron that governs action potential generation and neuronal polarity. It has long been recognized as a hotspot for protein phosphorylation, which regulates dynamic processes such as protein interaction, membrane trafficking, receptor and ion channel activity, and plasticity^1–3. For instance, casein kinase 2 (CK2) phosphorylates the Ankyrin-G (AnkG)-binding motif of voltage-gated sodium channels (NaVs) and Kv7-family potassium channels (KCNQ2/3), which strongly promotes their affinity for the AnkG scaffold^4–6. The electrophysiological property of NaVs is also tightly regulated by protein kinase A (PKA) and glycogen synthase kinase 3 (GSK3)^7–10. Cyclin-dependent kinases (CDKs) phosphorylate the Kvβ2 auxiliary subunit and regulate the axonal membrane targeting of the Kv1 channel¹¹. Further, AIS-enriched L1-family adhesion molecules, such as neurofascin-186 and neuronal cell adhesion molecule (NrCAM), require dephosphorylation at specific tyrosine residues for AnkG interaction¹². In addition, several phosphatases, such as calcineurin and protein phosphatase 2A (PP2A), are known to play a role in various forms of AIS plasticity^13,14. However, in contrast to the notion that phosphorylation is a central mechanism for AIS regulation, our knowledge of the AIS-enriched kinases and phosphatases is surprisingly lacking. Indeed, CK2 and GSK3^5,15 are perhaps the only two kinases known to be concentrated at the AIS, and no phosphatase with clear AIS enrichment has been identified to date. Without knowing these AIS-specific modulators, researchers face a significant challenge in understanding how protein phosphorylation is locally regulated at the AIS in physiology and pathology.

Recent advancements in exploratory proteomics identified an evolving list of protein constituents at the AIS, including kinases and phosphatases that were not previously known to associate with the domain^16–20. In recent work, endogenous proximity proteomics was employed to map the proteomic interactions of autism spectrum disorder (ASD) risk proteins, including the AIS-enriched AnkG, NaV1.2, and NaV1.6¹⁸. Within this dataset, we identified a PP2A regulatory subunit Ppp2r2c (PP2A-B55γ). This suggested a previously unrecognized link between specific PP2A-B55-family subunits and the AIS, providing the field with a new premise to ask questions about their functions in neuronal signaling.

The PP2A phosphatases are trimeric assemblies from 2 scaffolding subunits, 2 catalytic subunits, and 15 distinct regulatory subunits, leading to a remarkable variety of nearly 100 heteromeric combinations^21–23. The regulatory subunits play a critical role in determining the functional specificity of PP2A holoenzymes. Yet, their heterogeneity in subcellular localization, specificity in substrates, and uniqueness in function remain largely unknown. The AIS is densely packed with ion channels that play crucial roles in action potential generation, waveform regulation, and neuronal excitability modulation^24–30. They represent prime candidates for local phospho-regulation to facilitate neuronal signaling. In humans, mutations of the PP2A-B55-family genes are linked to brain conditions such as ASD, intellectual disability, bipolar disorder (BD), and spinocerebellar ataxia^22,31. It is worth noting that these conditions exhibit remarkable overlap with brain disorders associated with AIS and ion channel dysfunction^24,32–34, suggesting an intriguing phenotypic convergence.

In this study, we employed CRISPR-mediated gene editing, predictive protein structure modeling, phosphoproteomics, and electrophysiological approaches to examine how Ppp2r2c functionally regulates protein phosphorylation at the AIS in mice. Our data shows that Ppp2r2c is selectively enriched at the AIS due to its unique structure. It functionally interacts with the Kv1.2 channel to regulate its expression at the AIS and modulates neuronal activity during development. Together, our work provides critical insights into the localization, function, and mechanism of Ppp2r2c at the AIS, and offers a new perspective for interpreting its role in the pathophysiology of brain disorders.

Results

PP2A-B55-family subunits exhibit distinct subcellular localizations

Recently, Ppp2r2c was identified in the endogenous proximity proteome of AnkG and NaV1.2¹⁸, prompting us to investigate the roles of PP2A-B55 subunits in regulating AIS protein phosphorylation. We first employed homology-independent universal genome engineering (HiUGE) to label these subunits with a “spaghetti monster” fluorescent protein (smFP) for mapping their subcellular localization^35,36 (Fig. 1A). The results revealed a striking enrichment of Ppp2r2c at the AIS both in vitro and in vivo (Fig. 1B, C), consistent with the proximity proteomics finding. To validate Ppp2r2c’s AIS localization, we also used adeno-associated viruses (AAV) to express cDNA encoding V5-tagged Ppp2r2c (canonical isoform) and observed a similar profile (Figs. 1D and S1A, B). In addition, immunofluorescence labeling with a validated monoclonal antibody against Ppp2r2c further substantiated this finding (Figs. 1E, S1C, S2, and S4E). With stimulated emission depletion microscopy (STED), we found that Ppp2r2c’s localization at the AIS was in register with the periodicity of the βIV-spectrin cytoskeleton (Fig. S1D, E). We further showed that CRISPR-depletion of AnkG completely abolished Ppp2r2c’s AIS enrichment, suggesting that the localization depends on the AnkG scaffold (Fig. S1F, G). Moreover, we found that another PP2A-B55 subunit, Ppp2r2b, also showed prominent enrichment at the AIS, both endogenously tagged and recombinantly expressed (Fig. 1F, G, the cDNA of the canonical isoform was used). In contrast, the expression of Ppp2r2a was largely diffuse throughout the cytosol (Fig. 1H, I). Further, we found that the recombinantly expressed Bβ2-isoform of Ppp2r2b showed mitochondrial localization, consistent with previous reports^37–39 (Fig. S1H). Together, these data demonstrate that PP2A-B55 subunits exhibit divergent subcellular localization, with the canonical isoforms of Ppp2r2b/c showing previously unrecognized enrichment at the AIS (Fig. 1J, K).

Fig. 1. The heterogeneous subcellular localization of PP2A-B55 subunits.

A Schematic illustration of the HiUGE-mediated endogenous protein labeling strategy to identify new AIS-enriched proteins based on exploratory proteomics datasets. B, C Representative images of HiUGE-mediated smFP-V5 labeling of Ppp2r2c showing AIS-enriched localization both in vitro and in vivo. D Recombinant expression of Ppp2r2c showing enriched localization at the AIS. E Monoclonal antibody labeling of Ppp2r2c showing enriched localization at the AIS. F, G Endogenous labeling and recombinant expression of Ppp2r2b also showed enriched localization at the AIS. The canonical splice isoform was used for recombinant expression. H, I Endogenous labeling and recombinant expression of Ppp2r2a showed a diffuse localization throughout the neuron. Scale bars are indicated in each panel. Arrowheads indicate the AIS. J, K Fluorescence line profiles and polarity index quantification results of HiUGE labeling and AAV-mediated recombinant expression experiments. *: p < 0.05, ****: p < 0.0001, one-way ANOVA with Dunnett’s test, three (J) or four (K) independent experiments were performed, and the numbers of neurons analyzed for each group are shown on the graph. Dashed lines represent polarity index = 1 (no enrichment). Plots are mean ± SEM, a.u. arbitrary units. Source data are provided as a Source Data file. Figure 1A: created in BioRender. Gao, Y. (2025) https://BioRender.com/e34x072.

A unique “notched” β-propeller structure targets Ppp2r2c to the AIS

Why do some PP2A-B55 subunits enrich at the AIS, while others do not? To investigate the mechanisms behind the distinct localizations of these subunits, we first aligned their primary sequences. The most notable region was the N-terminus (N-term), which displayed a similarity between the AIS-enriched Ppp2r2b/c subunits that contrasted with the non-AIS-enriched subunits (Fig. 2A). We then aligned their predicted folding structures (retrieved from AlphaFold DB⁴⁰) and discovered a unique feature shared by Ppp2r2b/c that differentiated them from Ppp2r2a. Specifically, while these subunits all assumed 7-bladed β-propeller structures⁴¹, the N-term of Ppp2r2a was interlocked with the C-term to complete one of the folded “blades”, whereas the N-term of AIS-enriched Ppp2r2b/c were non-folded and extended (Fig. 2B). This rendered the β-propeller structure imperfect with a “notch” in Ppp2r2b/c (Fig. 2C). As the N-term sequences of PP2A-B55 subunits are known to influence their subcellular localization⁴², we hypothesized that there may be two putative AIS clustering mechanisms based on these structural predictions. Mechanism 1: the non-folded N-term of Ppp2r2c may serve as an extended motif that signals AIS targeting and enrichment. Mechanism 2: the interlocked N-term of Ppp2r2a obstructs a critical site for AIS targeting, while the open β-propeller of Ppp2r2c may enable unique interactions that lead to its enrichment at the AIS.

Fig. 2. A “notched” β-propeller structure is required for the AIS enrichment of Ppp2r2c.

A The sequence alignment of PP2A-B55 subunits revealed diverging N-term sequences and largely conserved C-term sequences. B The AlphaFold structural prediction shows that the N-terms of AIS-enriched Ppp2r2b and Ppp2r2c are non-folded and extended, whereas the N-term of Ppp2r2a is interlocked with the C-term in a β-strand interaction. C The N-term of Ppp2r2a completes one “blade” of the β-propeller by interlocking with the C-term, whereas the non-folded N-term of Ppp2r2c leaves an open “notch”. D Mutation experiments of Ppp2r2c showed that the AIS enrichment was retained in the N-term truncation (∆N22, polarity index = 4), but ablated in the C-term truncation (∆C14) and the N-term sequence swap with Ppp2r2a. *: p < 0.05, ****: p < 0.0001, one-way ANOVA with Dunnett’s test, four independent experiments were performed, and the numbers of neurons analyzed for each group are shown on the graph. E Mutation experiments of Ppp2r2a showed that the AIS enrichment was neomorphically gained in the N-term truncation (∆N26) and the N-term sequence swap with Ppp2r2c. *: p < 0.05, **: p < 0.01,***: p < 0.001, one-way ANOVA with Tukey’s test, four independent experiments were performed, and the numbers of neurons analyzed for each group are shown on the graph. Dashed lines represent polarity index = 1 (no enrichment). Plots are mean ± SEM. Scale bars are indicated in each panel. Arrowheads indicate the AIS. Diagrams in (D, E) depict the organization of β-strands within each propeller blade. Source data are provided as a Source Data file. Figure 2D, E: created in BioRender. Gao, Y. (2025) https://BioRender.com/e34x072.

To test these possibilities, we created truncational mutants to remove the N-term sequences from Ppp2r2a and Ppp2r2c. If the extended N-term sequence serves as the “GO” signal for Ppp2r2c (Mechanism 1), removing it (Ppp2r2c-ΔN22) would ablate AIS enrichment. Alternatively, if the folded N-term sequence acts as the “NO-GO” signal for Ppp2r2a (Mechanism 2), removing it (Ppp2r2a-ΔN26) would result in neomorphic AIS enrichment. Immunocytochemistry data indicated that Mechanism 2 was the predominant driver for AIS enrichment (Figs. 2D, E and S3). This suggests that the N-term of Ppp2r2a blocks a critical domain responsible for AIS enrichment. Based on the predicted structure, this domain is the “notched” blade formed by three β-strands at the C-term. We confirmed this by truncating one of the β-strands from the C-term of Ppp2r2c (ΔC14), which completely ablated its AIS enrichment (Fig. 2D and S3C, D). To further validate these findings, we generated N-term-swapped “hybrids” between these subunits, and their crossed-over localization pattern further substantiated that the “notched” β-propeller drives Ppp2r2c’s AIS enrichment (Fig. 2D, E). Finally, we noted statistical differences in the polarity indices amongst some of the AIS-enriched mutants, such as Ppp2r2c vs. Ppp2r2c-ΔN22, and Ppp2r2a-ΔN26 vs. Ppp2r2a-2c-N22-hybrid. This suggests that even though the “notched” β-propeller is the main determinant for AIS enrichment, the N-term sequence of Ppp2r2c also contributes to additional targeting specificity.

Ppp2r2c modulates neuronal activity during in vitro development

Since the AIS is a crucial domain that controls neuronal firing, we next asked the question of how these subunits selectively modulate neuronal activity. Using a microelectrode array (MEA) system, we compared the effects of AIS-enriched Ppp2r2c versus non-enriched Ppp2r2a on spontaneous neuronal activity. We first tested the effects of Ppp2r2a/c deficiency using a CRISPR-mediated depletion strategy as previously described^16,17. On days-in-vitro 1 (DIV1), primary neurons were transduced with AAVs expressing three gRNAs (3 × gRNAs) targeting either Ppp2r2a or Ppp2r2c to induce insertion-deletion-based expression disruption (Ppp2r2a- and Ppp2r2c-deplete, Fig. S4A, B, E). A non-targeting vector served as the control. Neuronal activity was subsequently measured by the MEA on DIV08, 11, and 14 (Fig. 3A). We found that the depletion of both Ppp2r2a and Ppp2r2c resulted in significant suppression of activity at DIV11 and 14, as measured by reduced mean firing rate, burst frequency, and network burst frequency (Fig. 3B, C). Cell viability, as measured by impedance in MEA experiments, remained unaffected (Fig. 3C). We also did not observe any overt phenotype in general neuronal maturation in vitro (Fig. S5). These findings suggest that both Ppp2r2a and Ppp2r2c are involved in regulating baseline neuronal activity without affecting overall cell condition.

Fig. 3. PP2A-B55 subunits differentially modulate neuronal activity during in vitro development.

A Schematic illustration of the MEA experiment under 3 × gRNA CRISPR-depletion treatments (Ppp2r2a- and Ppp2r2c-deplete). A non-targeting vector was used as the control. B Representative raster plot at DIV11. C Quantification of neural metrics data showing the effect of Ppp2r2a- and Ppp2r2c-deplete treatments on neuronal activity. ****: p < 0.0001; ns non-significant, two-way ANOVA with Tukey’s test, n = 80 wells per group from three independent experiments. D Schematic illustration of MEA experiment under Ppp2r2a and Ppp2r2c overexpression (OE) treatments. GFP-OE was used as the control. E Representative raster plot at DIV11. F Quantification of neural metrics data showing the effect of Ppp2r2a- and Ppp2r2c-OE treatments on neuronal activity. **: p < 0.01, ***: p < 0.001, ****: p < 0.0001; ns non-significant, two-way ANOVA with Tukey’s test, n = 96 wells per group from three independent experiments. Plots are mean ± SEM. Source data are provided as a Source Data file. Figure 3A, D: created in BioRender. Gao, Y. (2025) https://BioRender.com/e34x072.

We then tested the effects of AAV-mediated overexpression (OE) of Ppp2r2a and Ppp2r2c (Fig. S4C, D, F). AAVs expressing the canonical splice isoforms of Ppp2r2a or Ppp2r2c were added to neuronal cultures at DIV07. GFP-OE served as the control. Neuronal activity was measured on DIV08, 11, and 14 (Fig. 3D). Ppp2r2c-OE led to a remarkable potentiation of neuronal activity at DIV11 and DIV14. In contrast, Ppp2r2a-OE showed no significant effect (Fig. 3E, F). Again, cell viability remained unaffected (Fig. 3F). Collectively, our data suggest that while both subunits play a role in maintaining baseline neuronal firing, the distinct enrichment of Ppp2r2c at the AIS selectively potentiates neuronal activity under the OE condition.

To further validate the effects of Ppp2r2c on neuronal activity, we performed a rescue experiment by expressing a CRISPR-resistant Ppp2r2c vector, which restored the neuronal activity deficits associated with Ppp2r2c-deplete (Fig. S6A). We also conducted a Ppp2r2c-OE experiment on a shorter timescale and observed a responsive phenotype (Fig. S6B). We further tested the effects of the mutant Ppp2r2a-ΔN26, which, as previously shown, gained neomorphic enrichment at the AIS (Fig. 2E). Unlike the wild-type, the mutant Ppp2r2a-ΔN26-OE significantly potentiated neuronal activity, mimicking the effects of the AIS-enriched Ppp2r2c (Fig. S6C). Together, these additional data further support the model that the AIS localization of the PP2A-B55 subunits is an essential determinant of their modulatory effects on neuronal activity.

Ppp2r2c regulates AIS protein phosphorylation

What does Ppp2r2c do at the AIS to modulate neuronal activity? To answer this question, we employed phosphoproteomics to identify phosphopeptides that were altered by Ppp2r2a-OE or Ppp2r2c-OE, as they showed distinct electrophysiological phenotypes on the MEA. LC-MS/MS revealed numerous phosphopeptides that were significantly downregulated in comparison to the GFP-OE control, suggesting candidate sites that were dephosphorylated under the Ppp2r2a-OE or Ppp2r2c-OE conditions (Fig. 4A, B and Supplementary Data 1). The Venn diagram showed that these sites were largely non-overlapping, consistent with our central hypothesis on the functional heterogeneity of the PP2A regulatory subunits (Fig. 4C). This was further evident in the gene ontology (GO) analysis, which showed “postsynaptic cytoskeleton” and “presynaptic cytosol” as the top terms for the diffuse Ppp2r2a-OE, compared to “paranodal junction” and “axon initial segment” for the AIS-enriched Ppp2r2c-OE (Fig. 4D, note: we recognize the ambiguity of the GO term “paranodal junction” with proteins found at the juxtaparanode and the AIS).

Fig. 4. Ppp2r2c regulates protein phosphorylation at the AIS.

A Schematic illustration of the phosphoproteomics experiments to assess the effects of PP2A-B55 subunits on protein phosphorylation. B Volcano plots showing the downregulated phosphopeptides under the Ppp2r2a-OE and Ppp2r2c-OE conditions against the GFP-OE control, indicating the candidate dephosphorylated sites. C Venn diagram showing the overlap between the downregulated phosphosites of Ppp2r2a-OE versus Ppp2r2c-OE. D Top Cellular Component Gene Ontology (GO) terms for the downregulated phosphopeptide genes of Ppp2r2a-OE versus Ppp2r2c-OE (the top two terms were highlighted with red asterisks). E Western blot image showing that V5-tagged Ppp2r2c (arrow) co-immunoprecipitated with the GFP-tagged Kv1.2 (full-length, FL, arrowhead). The experiment was independently repeated three times with similar results. F Representative images of the Kv1.2 immunoreactivity under the control (GFP-OE) and Ppp2r2c-OE conditions. Scale bars are indicated in each panel. G Fluorescence line profiles of the Kv1.2 immunoreactivity along the axon. Plots are mean ± SEM, a.u. arbitrary units. H Quantification of the Kv1.2 immunoreactivity intensity at the AIS. ****: p < 0.0001, two-tailed Welch’s t-test, three independent experiments were performed, and the numbers of neurons analyzed for each group are shown on the graph. Plots are mean ± SEM. I A proposed model illustrating the functional interaction between Ppp2r2c and Kv1.2. Source data are provided as a Source Data file. Figure 4A, I: created in BioRender. Gao, Y. (2025) https://BioRender.com/e34x072.

Within the dataset, we identified that the phosphosite pS440/441 of a Kv1 potassium channel subunit, Kv1.2 (Kcna2), was significantly reduced in the Ppp2r2c-OE group. This locus is crucial for the membrane trafficking and functional expression of the AIS-enriched Kv1 channel, which conducts major outward currents at the AIS and regulates neuronal excitability^25–29. We thus followed up on this proteomic candidate. Co-immunoprecipitation experiments showed that Ppp2r2c interacted with Kv1.2 at its intracellular tails (Figs. 4E and S7A, B), suggesting that Ppp2r2c may selectively engage Kv1.2 at the AIS. Consistent with the hypothesis that dephosphorylation of pS440/441 may disrupt Kv1.2 expression, we found that the Kv1.2 immunoreactivity intensity at the AIS was significantly reduced under the Ppp2r2c-OE condition (Fig. 4F–H). Further, since phosphorylation is known to regulate the membrane trafficking and plasticity of NaV channels¹³, we measured their expression at the AIS as well. Interestingly, we found that NaV immunoreactivity intensity at the AIS remained unaffected by Ppp2r2c-OE treatment, suggesting that Ppp2r2c-OE does not significantly impact NaV abundance under baseline conditions (Fig. S8). This indicates a distinct shift in the AIS ion channel composition that favors excitation under Ppp2r2c-OE, and offers a possible mechanistic explanation of the phenotypes seen in the MEA experiments (Fig. 4I).

To further validate the effects of Ppp2r2c on Kv1.2, we conducted a Ppp2r2c-OE experiment on a shorter timescale and observed a responsive phenotype of Kv1.2 down-regulation (Fig. S9A–C). We also observed an up-regulation of Kv1.2 under Ppp2r2c-deplete treatment, as expected (Fig. S9D–F). We further tested the effects of the AIS-enriched Ppp2r2a-2c-N22-hybrid mutant and found that it phenocopied Ppp2r2c-OE in down-regulating Kv1.2 (Fig. S9G–I); this once again suggests that the AIS localization is an essential determinant of the subunits’ function. Together, these molecular and functional data provide a critical entry point to expand the mechanisms of local phospho-regulation at the AIS and their impact on neuronal signaling.

Discussion

Protein phosphorylation at the AIS is a rapidly evolving field. Defining the identity and function of the kinases and phosphatases involved in this process is vital for understanding AIS physiology and pathology. Following the initial identification of Ppp2r2c in the AIS proteomes¹⁸, we present here a mechanistic study to examine its unique localization and function at the AIS.

First, we revealed the distinct subcellular localizations of PP2A-B55 subunits in neurons. Canonical isoforms of Ppp2r2c (and Ppp2r2b) were enriched at the AIS, whereas Ppp2r2a was largely diffuse. The different localizations determine the molecular landscape surrounding these subunits, providing unique contexts for their cellular function. In future experiments, it will be important to determine how splice variations contribute to additional complexity in the subcellular localization of these subunits. Additionally, it will be equally important to determine if other mechanisms, such as post-translational modifications (PTMs), may also contribute to localization regulation. For example, a recent study showed that phosphorylation of Ppp2r2c by p21-activated kinase 6 (PAK6) modulates its subcellular localization and association with leucine-rich repeat kinase 2 (LRRK2)⁴³. It will be interesting to test whether phosphorylation at this site, which is near the “notched” β-propeller blade in Ppp2r2c, may affect its AIS enrichment.

WD40-repeats and β-propellers may function as scaffolds for protein complex assembly and subcellular targeting^42,44,45. Our model suggests that the “notched” β-propeller facilitates the enrichment of Ppp2r2c at the AIS. In contrast, the interlocked β-propeller of Ppp2r2a obstructs the essential domain for AIS targeting, although we suspect this inhibition to exist within dynamic equilibrium and may be influenced by complex PTMs and chaperone interactions. Interestingly, recent studies identified many other β-propeller-containing proteins within the AIS proteomes, such as Bcas3, Gnb5, Kctd3, Stxbp5, Stxbp5l, Wdr7, Wdr20, Wdr44, and Wdr47^17,18. Future work is required to determine whether interactions associated with β-propellers represent a generalizable mechanism for AIS affinity. Further, we expect the interactions at the “notch” domain to be heterogeneous and multivalent, likely involving the “β-strand addition” mechanisms⁴⁶. We observed that Kv1.2 was distally shifted in reference to Ppp2r2c (Fig. S7C), and the accumulation of Ppp2r2c at the AIS started very early during in vitro maturation (Fig. S1C). Given the late onset of Kv1.2 expression at the AIS⁴⁷, these data suggest that the interaction with Kv1.2 is not the primary AIS clustering mechanism of Ppp2r2c. Future experiments are needed to determine how the “notch” domain induces complex interactions with other AIS proteins and how they ultimately contribute to the AIS targeting of Ppp2r2c.

We show that Ppp2r2c modulates neuronal activity during in vitro development, which aligns with its prominent placement at the AIS. This may explain how Ppp2r2c dysfunction may contribute to excitatory/inhibitory (E/I) imbalance, a prevailing hypothesis underlying neurodevelopmental disorders^48,49. While the depletion of both Ppp2r2a and Ppp2r2c reduces baseline firing, only Ppp2r2c-OE potentiates neuronal network activity. This corresponds with the observation that both Ppp2r2a and Ppp2r2c are present at the AIS, but only the latter is selectively enriched there. Further, in sharp contrast to its AIS enrichment, the presence of the recombinantly expressed Ppp2r2c canonical isoform is minute at the dendritic excitatory and inhibitory synapses (Fig. S1A, B), consistent with previous studies indicating the absence of Ppp2r2c in synaptically-enriched proteomes^18,50,51. Therefore, we expect the effect of Ppp2r2c on neuronal signaling to be primarily driven by the AIS, whereas for the diffuse Ppp2r2a, it may be a summation of dispersed cytosolic effects, many of which may counteract each other. Future experiments are needed to discern the subunits’ unique functions in different cell types, such as the excitatory versus the inhibitory neurons, both of which express Ppp2r2c at their AIS (Fig. S2). It will also be important in future work to determine the presence and function of these subunits at the node of Ranvier and nearby axonal microdomains in vivo.

Using phosphoproteomics, co-immunoprecipitation, and immunofluorescence microscopy, we identified Kv1.2 as one of the many putative downstream effectors linking Ppp2r2c to neuronal excitability. This was further validated by Western blot analyses of independent brain lysate samples, which revealed a remarkable loss of Kv1.2-pS440/441, and a smaller, yet still significant, reduction of total Kv1.2 (Fig. S7D, E). This suggests that the disruption of Kv1.2 expression is likely a consequence, rather than the cause, of the loss of detected pS440/441 phosphorylation. The shift of AIS ion channel composition as a result of Kv1.2 loss provides a possible mechanistic explanation for the electrophysiological phenotypes seen in Ppp2r2c-OE. Future studies are needed to determine the roles of the other phosphoproteomic candidates, such as the hyperpolarization-activated cyclic nucleotide-gated (HCN) channels, the BK potassium channel Kcnma1, the scaffolding protein Tanc2, and the microtubule regulator Stmn1. Further, a previous study using heterologous cell lines showed that Ppp2r2c may interact and modulate the KCNQ2 channel³¹, indicating another target worthy of further investigation in neurons. In addition, it has been shown that the inhibition of PP2A with a low dose of okadaic acid disrupts NMDA-induced AIS plasticity¹³. It will be interesting to determine how the local dynamics of PP2A at the AIS, likely driven by the Ppp2r2b/c-containing holoenzymes, may contribute to AIS plasticity and activity homeostasis using high-density MEA or live imaging^52–54. Overall, even though Ppp2r2c is locally enriched, the effect of its associated holoenzyme is still expected to be broad-spectrum at the AIS. It will be an important future direction to explore other direct, indirect, and compensatory mechanisms that may contribute to Ppp2r2c’s effects on AIS function. Finally, we do not rule out potential structural roles of Ppp2r2c beyond regulating PP2A enzymatic activity, as is often observed with other signaling molecules in neurons^55,56.

Ppp2r2c is implicated in ASD, intellectual disability, and BD^31,57,58, showing an intriguing convergence with brain disorders associated with AIS and ion channel dysfunction^24,32–34. The identification of Ppp2r2c at the AIS offers a new perspective for interpreting how it may drive phenotypes. Currently, not much is known about how the dysregulation of PP2A-B55 subunits contributes to behavioral abnormalities in rodent models. Data from the International Mouse Phenotyping Consortium (IMPC, www.mousephenotype.org) show that Ppp2r2a-KO causes preweaning lethality, Ppp2r2b-KO results in hyperactivity, and Ppp2r2c-KO shows no overt deficits^59,60. There is an excellent opportunity to explore the effects of these global and conditional KOs using refined behavioral paradigms that evaluate learning, memory, and social interaction in the context of neurodevelopmental disorders.

Despite the profound impact of protein phosphorylation on AIS function and neuronal signaling, we are only beginning to understand how unique kinases and phosphatases regulate this process. Our work reveals a previously unrecognized AIS specificity for PP2A-B55 subunits, and spurs future advancements in understanding AIS function through proteomic discoveries.

Methods

Animals

H11-Cas9 mice (Jackson Laboratory #28239) and C57BL/6J mice (Jackson Laboratory #000664) were used. Mice were group housed with a temperature range of 68–72 °F, humidity range of 30–70%, and light cycle of 7:00 am–7:00 pm. Both males and females were used. All procedures were performed following protocols approved by the Baylor College of Medicine Institutional Animal Care and Use Committee (IACUC #AN-9158 and #AN-4634), in accordance with US National Institutes of Health guidelines.

DNA vector preparations

HiUGE plasmids were constructed following the previously described methods^18,35. Briefly, a DNA insert sequence for knock-in was flanked by sequences that were recognized by a synthetic donor-specific gRNA (DS-gRNA), inert to the host genome. For endogenous protein labeling, spaghetti monster fluorescent protein (smFP-V5)³⁶ was used as the insert. The gene-specific gRNAs (GS-gRNAs) targeting the C-terms were designed using CRISPOR⁶¹, and a pair of 23–24 mer oligonucleotides were annealed and ligated into the SapI site of the GS-gRNA expression cassette behind a U6 promoter. The following mouse genomic sequences were targeted: Ppp2r2a: CAAGACAAAGTGAATTAGGCTGG, Ppp2r2b: CCAAGACAAGGTTAACTAGAAGG, Ppp2r2c: CAACTCTGACATGCACTAGCTGG.

For CRISPR-mediated expression depletion, a 3 × gRNA AAV vector was used as previously described^16,17. Briefly, three gRNAs targeting specific exon sequences were cloned into the 3 × gRNA backbone using NEBuilder assembly (New England Biolabs). The following mouse genomic sequences were targeted: Ppp2r2a (GCGAATTAATCTCTGGCATCTGG, GATGTGATAAGTGTGGGCGTTGG, CAATATGGAAGAGCTCACGGAGG), Ppp2r2c (AGTGTGGTTGAACTCGACAGTGG, GGTAGTCAAACTCCGGCTCGTGG, AGATTACCGAACGAGACAAGAGG), Ank3 (GCTTTATGGTGGACGCGAGAGGG, GCCAGTAGGCTGGTAGAAATGGG, GCGTGTCCAATGGGTACAAGGGG).

For recombinant expression, protein sequences of the canonical splice isoforms were retrieved from UniProt. V5-epitope or GFP-tagged cDNA sequences were codon-optimized and synthesized by Twist Bioscience (South San Francisco, CA, USA), then cloned into an AAV vector downstream of an Ef1α promoter.

Concentrated AAV vectors were prepared following previously described methods^16,62 with modifications. Briefly, HEK293T cells (Takara, AAVpro #632273) grown on two 15 cm dishes were transfected with 15 μg AAV vector, 30 μg helper plasmid, and 15 μg serotype plasmid (pUCmini-iCAP-PHP.eB, a gift from Viviana Gradinaru⁶³) using PEI-Max (Polysciences #24765-100). Three days after transfection, AAV particles were precipitated using PEG8000 (Santa Cruz #sc-281693) from the media at 4 °C overnight. The pellet was washed with sterile glutamine-free DMEM, resuspended in sterile Dulbecco′s PBS (DPBS, 1:100 of original volume), and stored at −80 °C. Alternatively, the preparation was further purified by repeated DPBS washes with Amicon ultrafiltration (100 kDa MWCO, Sigma #UFC8100) for in vivo use. The titers of the concentrated AAVs were ~10¹⁰ GC/μL. Small-scale AAVs were prepared as previously described³⁵. Briefly, HEK293T cells grown on 24-well plates were triple-transfected with 0.2 μg AAV vector, 0.4 μg helper plasmid, and 0.2 μg serotype plasmid. Three days after transfection, the AAV-containing medium was filtered through Costar Spin-X columns (Sigma #8162) and applied directly to neuronal cultures.

HiUGE-mediated endogenous protein labeling

HiUGE-mediated endogenous protein labeling was performed as previously described^18,35. Briefly, neonatal H11-Cas9 mice were euthanized by decapitation, and forebrain tissues (containing cortices and hippocampi) were rapidly isolated and dissociated with papain (Worthington #LS003120). Cells were seeded at a density of ∼250,000 cells per cm² on poly-L-lysine (Sigma # P2636) coated coverslips and maintained in the Neurobasal Plus culture system (ThermoFisher #A3653401). On DIV1, 100 μL small-scale AAVs encapsulating HiUGE components (GS-gRNA and smFP-V5 donor) were added to cells. Cells were fixed at DIV11-14 for immunodetection of smFP-V5-tagged proteins. For in vivo labeling, purified AAVs were intracranially injected into neonatal H11-Cas9 pups between P0-2 (2 μL per hemisphere). Mice were deeply anesthetized with isoflurane and euthanized by decapitation 2–3 weeks following the injection for brain collection.

Immunocytochemistry and immunohistochemistry

Primary neurons grown on coverslips (DIV11-14) were fixed with 4% PFA at 4 °C for 3–5 min. Alternatively, for Ppp2r2c-mAb staining, neurons were fixed with a fixative consisting of 3% glyoxal, 0.8% glacial acetic acid, and 20% methanol, pH 5.0, at 4 °C for 10 min^64,65. Cells were blocked with a blocking buffer (5% goat or donkey serum in PBS with 0.3% Triton-X-100) at RT for 1 h. Coverslips were incubated with primary antibody diluted in blocking buffer overnight at 4 °C or for 1 h at RT. After three washes in PBS with 0.3% Triton-X-100 (PBST), the coverslips were incubated with species-matched Alexa Fluor-conjugated secondary antibodies (ThermoFisher or Jackson ImmunoResearch, 1:1000) diluted in blocking buffer for 30 min at RT. After three washes in PBST, cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI), mounted with FluorSave reagent (Millipore Sigma #345789), and imaged on Zeiss microscopes. For immunohistochemistry of HiUGE labeling, approximately 3-week-old mice were euthanized by isoflurane overdose followed by decapitation. Brains were dissected out and frozen on crushed dry ice. Coronal sections were obtained at 20 µm thickness using a cryostat (ThermoFisher #Cryostar NX70), attached to slides, and fixed with 4% PFA at 4 °C for 5 min. Sections were blocked with blocking buffer and incubated with primary antibody overnight at 4 °C or 1 h at RT. After three washes in PBST, sections were incubated with species-matched Alexa Fluor-conjugated secondary antibodies for 30 min at RT. After three washes in PBST, sections were counterstained with DAPI, cover-slipped with FluorSave reagent (Millipore Sigma #345789), and cortical areas were imaged on Zeiss microscopes. Alternatively, for Ppp2r2c-mAb staining, 1-month-old C57BL/6J mice were euthanized by isoflurane overdose followed by transcardial perfusion with 20 mL ice-cold PBS and 20 mL glyoxal-based fixative (3% glyoxal + 0.8% acetic acid + 20% methanol). The brains were dissected out and post-fixed in the same solution at 4 °C overnight. Tissues were sequentially dehydrated in 20% sucrose followed by 30% sucrose at 4 °C overnight. Brains were embedded in Tissue-Tek OCT (Sakura Finetek #4583) blocks. Serial coronal sections were obtained at 20 μm intervals and mounted onto Superfrost Plus slides (Fisher Scientific #1255015). Sections were washed three times with PBS at RT to remove OCT, followed by permeabilization with PBST at RT for 5 min. Tissues were then blocked in a blocking buffer containing 10% goat serum in PBST for 1 h at RT before incubation with primary antibodies at 4 °C overnight. After three washes in PBST, the slides were incubated with species-matched Alexa Fluor-conjugated secondary antibodies diluted in blocking buffer at RT for 1 h (ThermoFisher, 1:500) and washed with PBST again three times. Slides were then mounted with VECTASHIELD HardSet Antifade Mounting Medium (Vector Laboratories #H-1400-10), and cortical areas were imaged. Microscopic images were obtained using a Zeiss Apotome.2 Microscope. Images were pseudo-colored and processed using Zeiss ZEN lite or FIJI⁶⁶. The AIS enrichment analyses and the fluorescence-based quantifications were performed by observers blinded to the experimental conditions in Zeiss ZEN lite or FIJI⁶⁶. For the quantification of fluorescence signal at the AIS, a universal background correction was first applied, and the signal intensity was recorded either within a region of interest representing the AIS or as a line profile along the AIS. For intensity quantifications, the values were first normalized to the means within each independent experiment, then normalized to the means of the control group across independent experiments. For fluorescence line profiles, the values were first averaged within each independent experiment, then normalized to the highest intensity in a single channel. We also analyzed the polarity index following a previously described procedure¹⁷, using the formula AIS/Dendrite ratio = AIS correct mean intensity (CMF)/Dendrite CMF. The following primary antibodies were used: mouse anti-V5-epitope (ThermoFisher #R96025, 1:500), rabbit anti-V5-epitope (Cell Signaling #13202S, 1:1000), mouse anti-HA-epitope (Cell Signaling #2367S, 1:1000), rabbit anti-HA-epitope (Cell Signaling #3724S, 1:1000), mouse anti-Ppp2r2c (Santa Cruz #sc-100417, 1:250), mouse anti-AnkG (Santa Cruz #sc-12719, 1:250), mouse anti-AnkG (Rasband Lab N106/36, 1:2000), guinea pig anti-AnkG (Synaptic Systems #386004, 1:500 for immunohistochemistry and 1:2000 for immunocytochemistry), rabbit anti-βIV-spectrin (Rasband Lab C9831, 1:2000), chicken anti-Map2 (Biosensis #C-1382-50, 1:1000), mouse anti-PSD95 (Santa Cruz #sc-32290, 1:250), mouse anti-Gephyrin (Santa Cruz #sc-25311, 1:250), mouse anti-Kv1.2 (AntibodiesInc #75-008, 1:1000), mouse anti-Pan-NaV channel (Rasband Lab K58/35, 1:1000), rabbit anti-TOM20 (Cell Signaling #42406T, 1:500), rabbit anti-CamKII (Proteintech #20666-1-AP, 1:500), rabbit anti-Parvalbumin (Proteintech #29312-1-AP, 1:500), mouse anti-LAMP1 (Santa Cruz #sc-20011, 1:100), rabbit anti-EEA1 (Cell signaling #3288T, 1:500).

For stimulated emission depletion (STED) super-resolution imaging, samples were blocked with a blocking buffer containing 10% goat serum, followed by overnight incubation with primary antibodies at RT (mouse anti-V5-epitope, ThermoFisher #R960CUS, 1:100; and rabbit anti-βIV-spectrin, Rasband Lab C9831 at 1:200). Secondary antibodies were incubated for 1 h at RT (Abberior #STRED-1001 and #STORANGE-1002, 1:300). After washes, the coverslips were mounted with a liquid antifade mounting solution (Abberior #MM-2009). Images were acquired with a STEDyCON system (Abberior) on a Nikon Eclipse Ti-2 microscope. Images were deconvolved using Huygens Essential software.

Western blot

For Western blot, protein samples were heat-denatured in Laemmli buffer and subjected to SDS-PAGE. The Kv1.2 blots from brain lysates were an exception, where the samples were not exposed to heat to avoid aggregation. After transferring to nitrocellulose membranes, the blot was incubated with blocking buffer (Rockland #MB-070, or 5% BSA in TBS) for 30 min and then with primary antibodies overnight at 4 °C or 1 h at RT. Species-matched fluorescent secondary antibodies (LI-COR or CST, 1:10,000) were incubated with the blot for 30 min at RT, and the immunosignal was detected using an Odyssey FC imager (LI-COR). Alternatively, HRP-conjugated secondary antibodies (CST, 1:1000) were incubated with the blot for 30 min at RT, and the immunosignal was developed using Femto Substrate (ThermoFisher #34094). For quantifications, we first computed the relative intensity ratios between the protein of interest and the housekeeping reference. These ratios were then normalized to the means of the control group for plotting. The following primary antibodies were used: rabbit anti-Ppp2r2a (Proteintech #16569-1-AP, 1:1000), mouse anti-Ppp2r2c (Santa Cruz #sc-100417, 1:250), rabbit anti-V5-epitope (Cell Signaling #13202S, 1:1000), mouse anti-GFP (Proteintech #66002-1-Ig, 1:1000), rabbit anti-Kv1.2-pS440/441 (a gift from Dr. James Trimmer, 1:1000), mouse anti-Kv1.2 (AntibodiesInc #75-008, 1:1000), mouse anti-PSD95 (Santa Cruz #sc-32290, 1:250), rabbit anti-GAPDH (Proteintech #10494-1-AP, 1:1000; or Cell Signaling #2118, 1:5000), mouse anti-GAPDH (Proteintech #60004-1-Ig, 1:1000), rabbit anti-β-Actin (GeneTex #GTX637675, 1:5000).

Immunoprecipitation

V5-epitope and GFP-tagged cDNA constructs were co-transfected into HEK293T cells using PEI-Max (Polysciences #24765-100). Three days following transfection, cells were lysed using an NP40-based lysis buffer for Nanobody Trap experiments following the manufacturer’s protocol (Chromotek #gtma). The bound proteins were eluted by boiling in 2× Laemmli buffer for 5 min, and subjected to Western blot analysis (immunoprecipitated IP fraction). The lysates were also subjected to Western blot (input fraction).

Protein structure prediction

Protein structures were predicted using AlphaFold^40,67,68. Specifically, predicted structures of PP2A-B55 subunits (Ppp2r2a: AF-Q6P1F6-F1, Ppp2r2b: AF-Q6ZWR4-F1, Ppp2r2c: AF-Q8BG02-F1) were retrieved from AlphaFold DB⁴⁰. Molecular graphics and analyses were performed in UCSF ChimeraX⁶⁹.

Microelectrode array (MEA)

MEA recordings were performed as previously described¹⁸ with slight modifications. Briefly, 96-well CytoView MEA plates (Axion Biosystems #M768-tMEA-96W-5) were coated with 1 mg/mL poly-L-lysine (Sigma #P2636) in borate buffer (pH 8.5) overnight. Neonatal H11-Cas9 mice were euthanized by decapitation, and forebrain tissue was rapidly isolated and dissociated with papain (Worthington #LS003120). Cells were seeded as droplets onto the MEA plates (~150,000 cells per well) and maintained in the Neurobasal Plus culture system (ThermoFisher #A3653401). Unless otherwise stated, AAVs were added on DIV1 for CRISPR-mediated depletion or on DIV 7 for overexpression. Recordings of spontaneous neuronal activities were conducted using a Maestro Pro MEA system on DIV 8, 11, and 14, then analyzed using AxIS Navigator (Axion Biosystems). Electrode bursts were defined as a minimum of 10 spikes, separated by an inter-spike interval (ISI) of no more than 100 msec. Network bursts were defined as a minimum of 100 spikes, separated by an ISI of no more than 100 msec with at least 35% participating electrodes.

LC-MS/MS-based phosphoproteomic profiling

Sample processing and LC-MS/MS detection

Neonatal mice received intracranial injections of GFP-OE, Ppp2r2a-OE, or Ppp2r2c-OE AAV (2 μL per hemisphere). Forebrain tissue was collected on P21. Sample preparation for deep-scale phosphoproteomic profiling was performed as described previously⁷⁰, with minor modifications. Briefly, samples were cryo-pulverized and lysed with 8 M Urea lysis buffer, reduced/alkylated with dithiothreitol/iodoacetamide, and digested using LysC and trypsin enzyme. The peptides were desalted using Sep-Pak Vav 1cc C18 cartridges (Water WAT054955) and dried in a speed vac. The peptides (90 µg per sample) were labeled with TMT10plex Label Reagent Set (Thermo Scientific 90110) according to the manufacturer’s protocol. The TMT channel assignment was as follows: three GFP-OE control tissues- 126, 127 N, 127 C, three Ppp2r2A-OE tissues- 128 C, 129 N, 129 C, and three Ppp2r2C-OE tissues- 130 N, 130 C, 131. The labeled peptides were mixed and dried with a speed vacuum concentrator. Offline fractionation of TMT-labeled peptides was done using an Agilent 300Extend-C18 column (4.6 mm × 250 mm, 5 µm) on an Agilent 1260 Infinity II system at 1 ml/min for 96 min. The 96 fractions were concatenated into 24 peptide pools and a flow-through pool and acidified with s final concentration of 0.1% formic acid (FA). 5% of the peptide samples were used for proteome profiling, and 95% were used for phospho-proteome profiling. For the phospho-proteome, the 24 peptide pools were further concatenated to make 12 pools. The phospho-peptide enrichment was carried out using the High Select™ Phosphopeptide Enrichment Kit (Thermo Scientific A52284) according to the manufacturer’s protocol. The peptides were separated on an online nanoflow Easy-nLC-1200 system (Thermo Fisher Scientific) coupled to Orbitrap Exploris480 mass spectrometer (Thermo Fisher Scientific). 250 ng of each fraction for proteome was loaded on MS, while 25% of each fraction for phosphoproteome was loaded onto the MS. The peptide loading was done on a pre-column (2 cm × 100 µm I.D.) and separated on an in-line 20 cm × 75 µm I.D. column (Reprosil-Pur Basic C18aq, Dr. Maisch GmbH, Germany) equilibrated in 0.1% FA. Peptide separation was done at a flow rate of 200 nl/min over a 110-min run time. The MS data acquisition was carried out in TMT MS2 mode. The MS1 was done in Orbitrap (120000 resolution, scan range 375–1500 m/z, 50 ms Injection time), followed by MS2 in Orbitrap at 30000 resolution (HCD 38%) with TurboTMT algorithm. Raw instrument files were converted to mzML using MSConvert⁷¹. Database search for total protein profiling used mouse GENCODE (release 32). For phosphoproteomics, UniProt reviewed (SwissProt) (UP000000589) sequences downloaded on November 11, 2024.

Protein profiling

Precursor ion AUCs and reporter ion intensities were extracted with MASIC⁷². Butterworth smoothing method was used with a sampling frequency of 0.25 and an SIC tolerance of 10 ppm. Reporter ion tolerance was set to 0.003 Da with reporter ion abundance correction enabled. MSFragger version 3.8⁷³ was used for database search with deisotoping⁷⁴ and mass calibration plus parameter optimization⁷⁵. A strict trypsin digest rule was used to generate peptides between 7 and 50 amino acids between 350 and 10,000 Dalton, and clipped N-terminus methionine cleavage was enabled. Charges 2–6 were considered. Initial precursor mass error was set to 20 ppm, and fragment mass tolerance was set to 0.02 Dalton. Isotope error values were set to −1/0/1/2. Variable modifications were set to oxidation (15.9949) on methionine, peptide N terminal TMT10 reagent (229.1629) peptide N terminal pyroglutamic acid formation on Q (−17.02650) or E (−18.01060), and protein N-terminal acetylation (42.010565). Fixed modifications include TMT10 reagent (229.1623) lysine and carbamidomethylation (57.02146) of cystine (57.02146). Peptide validation was performed using a semi-supervised learning procedure in Percolator⁷⁶ as implemented in MokaPot⁷⁷ and filtered to 1% PSM FDR. Finally, peptides rolled up to the gene level with gpGrouper⁷⁸.

Phosphosite profiling

Raw data were processed with FragPipe version 22.0, driven with the Philosopher Pipeline⁷⁹. MSFragger version 4.1⁷³ was used for database search with deisotoping⁷⁴ and mass calibration plus parameter optimization⁷⁵. A strict trypsin digest rule was used to generate peptides between 7 and 50 amino acids between 200 and 5000 Dalton. Clipped N-terminus methionine cleavage was enabled, and charges 2–6 were considered. Initial precursor and fragment mass tolerances were set to 20 ppm. The isotope error value allowance was set to 0/1/2/3. Variable modifications are oxidation (15.9949) on methionine, phosphorylation (79.9663) of STY, and TMT10 reagent (229.1623) on serine. Fixed modifications include TMT10 reagent (229.1623) on N-term peptide and lysine, and carbamidomethylation of cystine (57.02146). Clear MS2 m/z range 125.5–131.5 to account for the TMT label. Peptide validation was performed with Percolator⁷⁶ after addition of additional features calculated by MSBooster⁸⁰. Site localization probabilities were calculated with PTMProphet⁸¹, and filtered to 1% protein FDR using a sequential picked FDR strategy⁸². MS1 peak chromatograms and MS2 reporter ions were extracted with IonQuant v 1.10.27 with default parameters for high-resolution OT-OT TMT data⁸³. Final site-level expression tables were generated with TMT integrator using the top three peptides with a min peptide probability of 0.5 and a min site probability of 0.75, and median normalization⁸⁴.

Statistics and gene ontology analyses

The protein-level and phosphopeptide-level quantification data were analyzed using PolySTest⁸⁵. Abundances were considered significantly altered if they met false discovery rate FDR < 0.05 (PolySTest) and abs(log2-FC) > 0.585 (1.5 fold change) compared to controls. Gene ontology (GO) analyses were performed using ShinyGO⁸⁶ against a custom statistical domain of all identified brain proteins from cumulative mouse brain proteomic studies with recent updates (9884 unique proteins)¹⁸. The pathway size boundary was set between 10 and 500 to exclude ambiguous terms for querying the Cellular Component pathway database. Venn diagrams were made with BioVenn⁸⁷.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.G. conceived the study. Y.G., A.P.A., S.K., A.J.M., X.D., W.Z., and A.B.S. performed experiments and analyses. Y.G., M.N.R., and A.M. supervised the collaboration. Y.G. wrote the paper with input from all authors.

Peer review

Peer review information

Nature Communications thanks Amélie Freal and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Requests for data, resources, and reagents in this study should be directed to and will be fulfilled by the Corresponding Author, Dr. Yudong Gao. HiUGE-related plasmids are available from Addgene (#200386–200388). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE⁸⁸ partner repository with the dataset identifier PXD062000. Source data are provided with this paper.

Code availability

The study does not include custom code or mathematical algorithm that is deemed central to the conclusions.

Competing interests

Y.G. has a patent related to the HiUGE technology (U.S. Patent No. 12,325,855). The intellectual property was licensed to CasTag Biosciences. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-66120-0.