Introduction

TDP-43 (Transactive response DNA binding protein 43 kDa) is a ubiquitously expressed protein that is required for the appropriate regulation, processing, and stability of thousands of transcripts1,2,3. In adult neurons, loss of TDP-43 nuclear function leads to neurodegeneration, suggesting that TDP-43 is critically required for the appropriate splicing and expression of genes essential for neuronal function and health1,4. Importantly, TDP-43 abnormalities such as TDP-43 cytoplasmic mislocalization, nuclear exclusion, aggregation, and downregulation are major hallmarks of various neurodegenerative diseases, including Alzheimer’s disease (AD) and AD-related dementias such as Amyotrophic Lateral Sclerosis (ALS) and ALS/Frontotemporal Dementia (FTD)3,5,6,7. These pathologies are accompanied by the loss of TDP-43 molecular function, which causes the incorporation of cryptic exons in TDP-43 target mRNAs, resulting in gene dysregulation and the synthesis of non-functional proteins6,8,9,10,11. Thus, TDP-43 dysfunction is thought to be a driver of neurodegeneration and loss across multiple neurodegenerative diseases. Several pathways are known to contribute to TDP-43 abnormalities in disease, including the erosion of the nuclear pore complex (NPC) and disrupted nucleocytoplasmic transport (NCT)7,12,13. However, the mechanisms by which these cellular changes in disease are initiated are still not well understood.

Glycerophosphodiester phosphodiesterase 2 (GDE2) is one of three six-transmembrane proteins that cleave the glycosylphosphatidylinositol (GPI) anchor that tethers some proteins to the outer leaflet of the plasma membrane14,15,16. GDE2 is expressed in neurons, a subset of terminally differentiated oligodendrocytes, and vascular endothelial cells, while GDE3 is expressed in oligodendrocyte precursor cells and astrocytes17,18,19,20,21. GDE6 is expressed in neural progenitors but has not been detected in mammals22,23. In the developing nervous system, neuronally expressed GDE2 regulates the differentiation of subsets of late-born spinal and cortical neurons and contributes to oligodendrocyte differentiation via the release of maturation signals18,24,25. GDE2 expression persists in the adult nervous system, where it has separate roles in regulating pathways that are important for neuronal function and survival26,27,28,29. Prominent among these roles is that GDE2 is essential for maintaining NPC integrity, appropriate NCT, and ensuring TDP-43 nuclear localization and function29. Mechanistic studies in mouse models and neurons derived from human induced pluripotent stem cells (iPSCs) determined that GDE2 is required to prevent the sustained activation of canonical Wnt signaling in neurons, which is sufficient to erode the NPC, impair NCT, and disrupt TDP-43 nuclear expression29. Thus, GDE2 is a physiological regulator of neuronal canonical Wnt activation that is important for preserving NPC, NCT, and TDP-43 nuclear expression and function. Further evidence suggests that the GDE2/Wnt regulatory pathway is relevant to TDP43 pathologies in disease. GDE2 aberrantly accumulates in intracellular compartments in the neurons of postmortem brains of patients with AD, ALS, and ALS/FTD27,28; strikingly, neurons with GDE2 accumulations correlate precisely with TDP-43 pathologies29. Further, iPSC-derived neurons from patients with ALS show a reduction in GDE2 protein and increased canonical Wnt activation, which, when dampened by pharmacological agents, partly rescues TDP-43 nuclear function29. Deeper insight into the mechanisms by which GDE2 regulates canonical Wnt signaling in mature neurons could help clarify initiating mechanisms that underlie NPC, NCT, and TDP-43 abnormalities in disease.

GDE2 acts at the cell surface to cleave the GPI-anchor of its substrates and release them from the plasma membrane14,15,16. Accordingly, we hypothesize that GDE2 regulates the cell surface expression and activity of GPI-anchored proteins that activate or potentiate Wnt signal transduction to regulate NPC integrity, NCT, and TDP-43 localization. Glypican (GPC)2, GPC4, and GPC6 are established substrates of GDE216. They belong to the Glypican family of GPI-anchored heparan sulfate proteoglycans that undergo extensive post-translational modifications, including the attachment of glycosaminoglycan (GAG) chains and endoproteolytic cleavage by furin-like convertases30. Glypicans are evolutionarily conserved from Drosophila to humans. In mice and humans, there are six Glypican family members that are divided into two subfamilies: GPC1/2/4/6, which correspond to Drosophila Dally-like protein (Dlp), and GPC3/5, which are homologues of Drosophila Dally30. Between the two subfamilies, GDE2 can cleave GPC1/2/4/6 but not GPC3/516. Dlp has been shown to potentiate Wnt activation at suboptimal concentrations and can bind to the lipid moieties of Wnt ligands, facilitate their transport across cells, and accordingly regulate their accessibility and binding to Wnt receptors30,31,32. Thus, Glypicans related to Dlp are important regulators of canonical Wnt activation. Interestingly, GPC1, GPC4, and GPC6 show impaired release in the spinal cord of the SODG93A mouse model of familial ALS, which exhibits impaired GDE2 activity26. Moreover, mutations in GPC6 are associated with AD and ALS in genome-wide association studies (GWAS), and GPC6 exhibits increased expression and altered localization in spinal cord tissues from patients with ALS with TDP-43 proteinopathies33,34,35,36,37. These observations suggest that the Dlp-class Glypicans are good candidates for substrates targeted by GDE2 to regulate Wnt-mediated effects on NPC, NCT, and TDP-43 localization in health and disease.

Here, we sought to determine the physiologically relevant Glypicans that are regulated by GDE2 to modulate canonical Wnt signaling in adult cortical neurons. We show that GDE2 regulates GPC6 surface expression in cortical neurons to control Wnt activation and that increased levels of GPC6 are sufficient to promote sustained canonical Wnt signaling, causing abnormalities in the NPC, NCT, and disrupting the nuclear localization of TDP-43. Thus, the GDE2/GPC6 regulation of Wnt activation is a physiological pathway that, when impaired, may contribute to NPC, NCT, and TDP-43 pathologies in disease.

Results

Gde2 and Gpc6 expression and regulation in cortical neurons

To identify candidate Glypicans that overlap with GDE2 expression in excitatory cortical neurons, we took advantage of publicly available single-cell RNA-sequencing (scRNAseq) datasets of the adult mouse cortex38, focusing on excitatory neurons that show Gde2 expression. Consistent with previous studies, Gde2 is expressed throughout the adult cortex, and is enriched in excitatory neurons in deep layers of the cortex corresponding to Layers 5 and 621 (Fig. 1a). Of the 6 Glypican family members, Gpc6 shows the highest mRNA expression in Gde2-expressing cortical excitatory neurons (Fig. 1a). Gpc6 is expressed in subsets of neurons across all cortical layers, with robust expression in Layer 5 and Layer 6 excitatory pyramidal tract (PT) and intratelencephalic (IT) neurons. Gpc1, 4, and 5 are expressed at lower levels than Gpc6 in subsets of deep-layer neuronal layers, with Gpc5 also showing expression in Layers 2/3. In contrast, Gpc2 and Gpc3 are not co-expressed with Gde2 in cortical neurons.

Fig. 1
figure 1

Gpc6 is co-expressed with Gde2 in the mouse cortex and regulated by Gde2 expression. (a) Heat map showing the expression of Gde2 (Gdpd5) in excitatory neurons in the mouse cortex (left) and the expression of Glypican family proteins and Wnt signaling components in Gde2-expressing cortical neurons (right). Each subpanel is scaled differently to highlight the range of expression levels. Data from the Allen Brain Atlas. L: layer. (b,c). Representative images of RNAscope probing for Gpc6 transcripts in coronal sections of the adult mouse brain. The boxed area of the motor cortex in (b) is shown in magnified view in (c). Scale bar: (b) = 500 μm; (c) = 100 μm. (d,e). Representative western blot images of surface biotinylation of DIV21 WT and Gde2KO primary cortical neurons (d) and graph quantifying the surface levels of furin-converted GPC6 normalized to the membrane protein, Na-K ATPase (e). WT vs. Gde2KO: **p = 0.0015. n = 4 WT, 5 Gde2KO. Welch’s t-test. All graphs: mean ± sem.

Full size image

We focused our analysis on GPC6 because of the level and extent of its co-expression with Gde2 in cortical neurons. RNAscope experiments on sections of 4-month-old WT mice confirm that Gpc6 is expressed throughout the brain and is distributed in neurons across all cortical layers (Fig. 1b,c). These observations, combined with the knowledge that GPC6 is an established physiological substrate of GDE2, suggest GPC6 as a potential candidate for mediating GDE2-dependent regulation of canonical Wnt signaling in excitatory neurons. GDE2 acts at the cell surface to cleave the GPI-anchor of its substrates and release them into the extracellular space. Accordingly, if GPC6 is a substrate of GDE2 in excitatory cortical neurons, loss of GDE2 should result in increased GPC6 surface expression and, conceivably, an increase in total levels of GPC6 protein. To test this idea, we performed surface biotinylation to examine the expression of GPC6 and GDE2 in cortical neurons prepared from WT and Gde2KO postnatal day (P) 0 pups that were cultured for 21 days in vitro (DIV). Western blot analysis showed that there is an increase in surface GPC6 expression and in the total amounts of GPC6 protein in Gde2KO neurons compared with WT (Fig. 1d,e). No changes in Gpc6 transcript amounts were found between WT and Gde2KO cortices using published RNAseq datasets (Fig. S1a), suggesting that GDE2 regulates the cell surface expression of GPC6 post-transcriptionally. Taken together, these observations are in line with the model that GDE2 regulates GPC6 surface expression to suppress sustained canonical Wnt signaling in excitatory cortical neurons in the adult brain.

GPC6 potentiates Wnt7b mediated signaling

Wnt ligands comprise a large family of proteins, and to date, 19 Wnt ligands have been identified in mice39,40,41. We analyzed the expression of Wnt ligands in the same scRNAseq data described above and found that Wnt4, Wnt9a, and Wnt7b are co-expressed with Gde2 and Gpc6 in excitatory cortical neurons (Fig. 1a and Fig. S1b). Wnt4 has been reported as a non-canonical Wnt ligand42, while Wnt9b is implicated in both canonical and non-canonical Wnt signaling pathways43. Wnt7b typically mediates canonical Wnt signaling44,45. Because GDE2 is known to regulate canonical Wnt signaling in cortical neurons18,29, we examined the relationship between GPC6 and Wnt7b-dependent canonical Wnt signaling. To examine if GPC6 is capable of potentiating Wnt7b-mediated signaling, we utilized SH-SY5Y cells, a neural-like immortalized cell line responsive to Wnt activation. To facilitate the detection of GPC6, we generated an N-terminal HA-tagged GPC6 (HA-GPC6). Notably, addition of the HA tag did not interfere with GPC6 processing by furin convertases (Fig. 2a), GPC6 trafficking to the cell surface (Fig. S2a-b), or GPC6 release by GDE246. Transfection of SH-SY5Y cells with Wnt7b plasmids elicited a dose-dependent increase in non-phosphorylated (p)-β-catenin levels (Fig. S2c), allowing us to examine the role of GPC6 in modulating Wnt7b signaling transduction. First, we transfected SH-SY5Y cells with a low amount of plasmid expressing Wnt7b or GFP where Wnt7b expression did not alter basal levels of non-phosphorylated (p)-β-catenin amounts, consistent with suboptimal Wnt pathway activation (Fig. 2a-b and Fig. S2c, grey bars). Under the same conditions, co-transfection of Wnt7b with GPC6 led to a significant increase in the amounts of non-p-β-catenin; however, no synergy was observed when GPC6 was co-transfected with a control plasmid expressing GFP (Fig. 2a-b, green bars). These observations support the notion that increased surface expression of GPC6 is sufficient to potentiate canonical Wnt7b-mediated signaling. Interestingly, total levels of GPC6 were reduced in the presence of Wnt7b (Fig. 2a), consistent with previous studies that glypicans are internalized with the Wnt-receptor complex47,48, which we speculate is targeted for degradation.

Fig. 2
figure 2

GPC6 potentiates Wnt7b signaling transduction. (a,b) Representative western blots (a) of SH-SY5Y cells transfected with plasmids expressing HA-GPC6 and 250ng Wnt7b (see Fig. S2c) and graph (b) quantifying non-p-β-catenin levels normalized to Tubulin. Arrow marks full-length HA-GPC6; Asterisk marks N-terminal HA-GPC6 protein after cleavage by furin convertase. GFP-GFP vs. GFP-GPC6: ns = 0.9449; Wnt7b-GFP vs. Wnt7b-GPC6: *p = 0.0253; GFP-GFP vs. Wnt7b-GFP: ns = 0.7818. RM two-way ANOVA. n = 3, mean ± s.e.m. (c). Graph quantifying GPC6 mRNA levels treated with siRNAs. GFP-scramble vs. GFP-siGPC6: ***p = 0.0003; Wnt7b-scramble vs. Wnt7b-siGPC6: ***p = 0.0003. RM two-way ANOVA with Sidak’s multiple comparison test. n = 4, mean ± s.e.m. (d). Representative western blots of SH-SY5Y cells transfected with 500ng of plasmids expressing Wnt7b (see Fig. S2c) and treated with siRNAs. e. Graph quantifying non-p-β-catenin protein levels normalized to Tubulin. GFP-scramble vs. GFP-siGPC6: ns = 0.8438; GFP-scramble vs. Wnt7b-scramble: *p = 0.0372; Wnt7b-scramble vs. Wnt7b-siGPC6: *p = 0.0338; GFP-siGPC6 vs. Wnt7b-siGPC6: ns = 0.6939. RM two-way ANOVA. n = 4, mean ± s.e.m. (f,j). Topflash luciferase assay in STF cells quantifying Firefly luciferase activity, reporting Wnt transcriptional activity, normalized to Renilla expressed under a universal promoter. (f) GFP vs. Wnt7b with Fzd1: ***p = 0.0003; Fzd3: ns p = 0.9898; Fzd4: *p = 0.0207; Fzd7: **p = 0.0086; Fzd8: ****p < 0.0001. Ordinary two-way ANOVA with Sidak’s multiple comparison test, with a single pooled variance. n = 5, mean ± s.e.m. (g) Fzd1: GFP-GFP vs. GFP-GPC6 **p = 0.0034; GFP-GFP vs. GFP-Wnt7b ****p < 0.0001; Wnt7b-GFP vs. Wnt7b-GPC6 **p = 0.0019; GFP-GPC6 vs. Wnt7b-GPC6: ****p < 0.0001. (h) Fzd4: GFP-GFP vs. GFP-GPC6 ns p = 0.1370; GFP-GFP vs. GFP-Wnt7b **p = 0.0076; Wnt7b-GFP vs. Wnt7b-GPC6 *p = 0.0358; GFP-GPC6 vs. Wnt7b-GPC6: **p = 0.0034. i: Fzd7: GFP-GFP vs. GFP-GPC6 ***p = 0.0008; GFP-GFP vs. GFP-Wnt7b ****p < 0.0001; Wnt7b-GFP vs. Wnt7b-GPC6 **p = 0.0020; GFP-GPC6 vs. Wnt7b-GPC6: ***p = 0.0002. j: Fzd8: GFP-GFP vs. GFP-GPC6 ns p = 0.9327; GFP-GFP vs. GFP-Wnt7b ***p = 0.0004; Wnt7b-GFP vs. Wnt7b-GPC6 ns p = 0.0644; GFP-GPC6 vs. Wnt7b-GPC6: ***p = 0.0002. RM-two-way ANOVA. n = 5, mean ± s.e.m.

Full size image

SH-SY5Y cells express endogenous GPC646, which likely contributes to basal levels of Wnt activation determined by non-p-β-catenin expression (Fig. 2a and Fig. S2c). To test whether GPC6 is required to facilitate Wnt7b-dependent Wnt stimulation, we transfected SH-SY5Y cells with a high level of Wnt7b (Fig. S2c) and asked whether GPC6 knockdown dampens Wnt activation. Treatment of SH-SY5Y cells transfected with plasmids expressing GFP or Wnt7b with siRNAs targeting GPC6 led to an approximately 50% reduction of endogenous GPC6 mRNA expression when compared to control siRNAs (Fig. 2c). Notably, GPC6 knockdown ablated the Wnt7b-dependent increase in non-p-β-catenin that was observed in control siRNA treated conditions (Fig. 2d,e), indicating that endogenous GPC6 is required for Wnt7b-dependent canonical Wnt activation. Taken together, these observations suggest that GPC6 is necessary and sufficient to facilitate Wnt7b signal transduction.

We next examined the surface components that mediate Wnt7b signaling in concert with GPC6. Notably, β-catenin also plays major roles at adherens junctions and links cadherin cell adhesion molecules to the actin cytoskeleton49. To focus our analysis on nuclear β-catenin relevant to Wnt7b-dependent activation of transcription, we utilized a heterologous system using the Wnt reporter cell line, SuperTopflash (STF)44. STF is a stable HEK293T cell line expressing a firefly luciferase reporter gene under the regulation of 7 TCF/LEF binding sites, which is activated when non-p-β-catenin binds to TCF/LEF transcription factors in response to canonical Wnt activation39,44,50. Analysis of the sc-RNAseq data described in Fig. 1 showed that the Wnt receptors Fzd1, Fzd3, Fzd4, Fzd7, and Fzd8, and the Wnt co-receptor Lrp6 are co-expressed with Gde2 and Gpc6 in deep-layer cortical neurons (Fig. 1a and Fig. S1c). STF cells do not express Lrp6; accordingly, we co-expressed Wnt7b, Lrp6, and each identified Fzd receptor in STF cells and assayed cells for luciferase activity (Fig. 2f). We found that Wnt7b activates canonical Wnt signaling when co-transfected with Lrp6 and Fzd1, Fzd4, Fzd7, or Fzd8 (Fig. 2f), but not Fzd3. This is consistent with previous reports that Fzd3 mainly mediates non-canonical Wnt signaling44. Next, we tested whether GPC6 potentiates Wnt7b activation when co-expressed with Fzd1, Fzd4, Fzd7, and Fzd8 in this paradigm. We found that GPC6 is capable of potentiating Wnt7b activation when co-expressed with Fzd1, Fzd4, and Fzd7 but not Fzd8 (Fig. 2g-j). Moreover, we observed a modest increase in Wnt activation when GPC6 is overexpressed with Fzd1 and Fzd7 in the absence of exogenous Wnt7b expression, suggesting that GPC6 can potentiate endogenous Wnt signaling in STF cells (Fig. 2g and i). Taken together, these in vitro observations suggest that GPC6 can potentiate canonical Wnt7b/Lrp6 activation via a subset of Fzd receptors that include Fzd1, Fzd4, and Fzd7, all of which are co-expressed with GDE2 in deep-layer cortical neurons.

GPC6 overexpression in the cortex phenocopies Gde2KO

Previous studies show that GDE2 prevents the sustained activation of canonical Wnt signaling in neurons to ensure NPC integrity, appropriate NCT, and TDP-43 nuclear expression29. We hypothesize that increased surface GPC6 expression in neurons elicited by GDE2 loss causes aberrant Wnt activation that drives NPC, NCT, and TDP-43 abnormalities. To test this hypothesis, we utilized a transgenic Wnt reporter mouse line that contains a H2B-MYC-GFP fusion reporter under the control of TCF/LEF1 binding sites integrated into the Rosa26 genetic locus (WT;WNT-GFP-MYC)45, to examine whether GPC6 overexpression can activate neuronal Wnt signaling in vivo. In these animals, Wnt activation can be visualized at single-cell resolution by MYC or GFP expression. We generated adeno-associated virus (AAV) expressing HA-GPC6 (AAV HA-GPC6) or control virus (AAV HA-GFP-MYC or AAV HA-GFP). We injected WT;WNT-GFP-MYC mice at P28 with AAV HA-GPC6, AAV HA-GFP, or AAV-HA-GFP-MYC by the retro-orbital route and examined animals 10 days later for Wnt activation by immunostaining cortical neurons for MYC expression (see Methods). We found that cortical neurons expressing HA-GPC6 showed an increase in nuclear MYC expression compared to neurons expressing HA-GFP, suggesting that GPC6 overexpression induces canonical Wnt signaling in neurons in vivo (Fig. 3a-b).

Fig. 3
figure 3

GPC6 overexpression activates Wnt signaling and drives NPC, NCT, and TDP-43 abnormalities. (a,c–n) Representative images of immunostained cortical sections of AAV-transduced WT;WNT-GFP-MYC animals 10 days post-injection. Dashed lines highlight the nuclei of cells transduced with AAV HA-GFP or AAV HA-GPC6 in (a) and (k–n); dashed lines outline the cell body of transduced cells in (c–f); arrows highlight transduced cells in (g–j). Scale bar: (a,c-n) 10 μm. (b) Graph quantifying nuclear MYC immunofluorescence intensity in neurons expressing HA-GFP or HA-GPC6. ****p < 0.0001, Welch’s t-test, N = 3 animals, n = 49 HA-GFP, 47 HA-GPC6-expressing cells. (o–r) Graphs quantifying neurons expression HA-GFP-MYC or HA-GPC6 comparing (o) Nup98 C/Nuclear envelope (NE) ratio (****p < 0.0001, Welch’s t-test, N = 3 animals, n = 78 HA-GFP-MYC, 77 HA-GPC6-expressing cells.), (p) Ran N/C ratio (****p < 0.0001,Welch’s t-test, N = 3 animals, n = 79 HA-GFP-MYC, 76 HA-GPC6-expressing cells.), (q) TDP-43 N/C ratio (****p < 0.0001, Welch’s t-test, N = 3 animals, n = 67 HA-GFP-MYC, 62 HA-GPC6-expressing cells.), (r) TDP-43 nuclear intensity (****p < 0.0001, Welch’s t-test, N = 3 animals, n = 67 HA-GFP-MYC, 62 HA-GPC6-expressing cells.). A.U.: arbitrary unit. All graphs: mean ± sem.

Full size image

Ablation of GDE2 results in sustained canonical Wnt signaling in cortical neurons that disrupts NPC, NCT, and TDP-43 localization29. Specifically, Nucleoporin (Nup) 98, a structural component of the NPC that is required for maintaining the nuclear barrier and selective protein import and export, abnormally accumulates in the cytoplasm51. Moreover, the Ras-related GTPase (Ran) that controls the nuclear import and export of proteins through a gradient of nuclear-enriched Ran-GTP compared with cytoplasmic Ran-GDP is disrupted, leading to a reduction in Ran nuclear/cytoplasmic (N/C) ratio52. Nuclear import of TDP-43 utilizes the Ran gradient12,53, and consistent with disrupted NCT, TDP-43 N/C ratios and TDP-43 nuclear intensity are reduced upon GDE2 ablation and ectopic Wnt activation29. Strikingly, HA-GPC6-expressing neurons recapitulated these changes, in that they showed an increase in Nup98 cytoplasmic accumulation (Fig. 3c-f, o, Fig. S3a, c) and a reduction in the Ran N/C ratio (Fig. 3g-j, p, Fig. S3b, d) compared to neurons expressing HA-GFP-MYC. Further, neurons expressing HA-GPC6 showed a decrease in TDP-43 nuclear intensities and N/C ratio relative to neurons expressing control virus (Fig. 3k-n, q-r, Fig. S4). Therefore, GPC6 overexpression in adult neurons phenocopies Gde2KO and ectopic Wnt activation by its ability to increase canonical Wnt signaling, disrupt NPC and NCT protein distribution, and promote TDP-43 cytoplasmic mislocalization and nuclear reduction.

GPC6 genetic reduction rescues TDP-43 mislocalization in Gde2KO animals

Our previous observations demonstrate that increased GPC6 surface expression in cortical neurons is sufficient to activate canonical Wnt signaling and trigger NPC/NCT disruptions and TDP-43 mislocalization. We reasoned that if GDE2 is a physiological regulator of GPC6 surface expression that is important for this pathway, then reducing the levels of endogenous GPC6 should rescue these changes in Gde2KO animals. To test this hypothesis, we genetically reduced GPC6 in Gde2KO and Gde2KO;WNT-GFP-MYC mice by generating Gde2KO;Gpc6+/- (Gde2KO;Gpc6Het) and Gde2KO;Gpc6Het;WNT-GFP-MYC animals. We did not completely ablate GPC6 in these animals because Gpc6KO mice die immediately after birth54. Animals were aged to 4 months before analysis by Western blot and immunohistochemistry. Western blots of cortical lysates from Gde2KO;Gpc6WT and Gde2KO;Gpc6Het animals showed that this genetic strategy reduced GPC6 protein amounts in Gde2KO animals by approximately 30% (Fig. 4a-b). To determine whether GPC6 reduction can rescue canonical Wnt activation in cortical neurons of Gde2KO animals, we examined neuronal MYC expression in 4-month-old WT;WNT-GFP-MYC, Gde2KO:WNT-GFP-MYC, and Gde2KO;Gpc6Het;WNT-GFP-MYC animals. Compared to WT;WNT-GFP-MYC animals, Gde2KO;WNT-GFP-MYC mice showed a marked increase in the proportion of nuclear MYC + neurons, consistent with aberrant neuronal Wnt activation upon GDE2 ablation (Fig. 4c, c’, d). In contrast, genetic reduction of Gpc6 in Gde2KO;Gpc6Het;WNT-GFP-MYC animals reduced the proportion of nuclear MYC + neurons normally seen in Gde2KO:WNT-GFP-MYC mice to WT:WNT-GFP-MYC levels (Fig. 4c, c’, d). No changes were observed in non-neuronal MYC + cells between the three genotypes (Fig. 4c, c’, e), which is consistent with previous studies showing that GDE2 loss does not perturb canonical Wnt signaling in non-neuronal cells. Thus, these observations are consistent with the model that GDE2 regulates the surface expression of GPC6 to regulate canonical Wnt signaling in neurons.

Fig. 4
figure 4

Gpc6 genetic reduction restores NCT and TDP-43 nuclear localization in Gde2KO animals. (a,b) Representative western blot images of GPC6 in Gpc6WT and Gpc6Het cortical extracts (a) and graph (b) quantifying GPC6 levels normalized to total protein. Arrow points to the band for furin cleaved GPC6. Gpc6WT vs. Gpc6Het: *p = 0.0215, n = 4, Unpaired t-test. (c,d). Representative images of immunostained cortical sections of 4-month-old WT;WNT-GFP-MYC, Gde2KO:WNT-GFP-MYC, and Gde2KO;Gpc6Het;WNT-GFP-MYC animals (c,c’) and graphs quantifying the proportion of MYC + neurons (d) and number of MYC + non-neuronal cells per field of view (e) in the cortex. Images in (c’) are magnified from (c). Scale bars: c = 50 μm, (c’) = 20 μm. (d) WT;WNT-GFP-MYC vs. Gde2KO:WNT-GFP-MYC: **p = 0.0044; Gde2KO:WNT-GFP-MYC vs. Gde2KO;Gpc6Het;WNT-GFP-MYC: **p = 0.0097; WT;WNT-GFP-MYC vs. Gde2KO;Gpc6Het;WNT-GFP-MYC: ns p = 0.7573. N = 4–5 animals, n ~ 600 neurons per animal. Ordinary one-way ANOVA with Tukey’s multiple comparisons test. (e) WT;WNT-GFP-MYC vs. Gde2KO:WNT-GFP-MYC: ns p = 0.2294; Gde2KO:WNT-GFP-MYC vs. Gde2KO;Gpc6Het;WNT-GFP-MYC: ns p = 0.9888; WT;WNT-GFP-MYC vs. Gde2KO;Gpc6Het;WNT-GFP-MYC: ns p = 0.2794. N = 4–5 animals. (f, h) Representative images of immunostained cortical sections of 4-month-old WT, Gde2KO, and Gde2KO;Gpc6Het. Scale bar = 50 μm. Arrows highlight cells in Gde2KO exhibiting nuclear exclusion of Ran (f) and cytoplasmic mislocalization of TDP-43 (h). (g,i). Graph quantifying Ran N/C ratios (g. WT vs. Gde2KO: ***p = 0.0007; Gde2KO vs. Gde2KO;Gpc6Het: **p = 0.0039; WT vs. Gde2KO;Gpc6Het: ns p = 0.6670. Ordinary one-way ANOVA with Tukey’s multiple comparisons test) and TDP-43 N/C ratios (i. WT vs. Gde2KO: **p = 0.0070; Gde2KO vs. Gde2KO;Gpc6Het: *p = 0.0314; WT vs. Gde2KO;Gpc6Het: ns p = 0.9577. Brown-Forsythe and Welch ANOVA tests). N = 6 animals, ≥ 120 cells per animal. All graphs: mean ± sem.

Full size image

To determine if the dampening of Wnt activation by the genetic reduction of Gpc6 is sufficient to rescue NCT and TDP-43 abnormalities in Gde2KO animals, we examined Ran and TDP-43 distribution in cortical neurons of 4-month-old WT, Gde2KO, and Gde2KO;Gpc6Het animals by immunohistochemistry. Gde2KO cortical neurons exhibit a reduction in the N/C ratio of Ran, consistent with the disruption of the Ran gradient important for NCT (Fig. 4f-g). Notably, Gde2KO;Gpc6Het mice showed an equivalent Ran N/C ratio to WT animals, suggesting that NCT trafficking is restored upon GPC6 reduction (Fig. 4f-g). In line with this observation, the reduction in TDP-43 N/C ratio observed in Gde2KOs, indicative of impaired TDP-43 localization, is rescued to WT levels in Gde2KO;Gpc6Het mice (Fig. 4h-i). In summary, these observations provide evidence that GDE2 regulates GPC6 expression to modulate neuronal Wnt signaling, which ensures the integrity of NCT and TDP-43 nuclear localization.

Discussion

Previous studies have determined that GDE2 encodes a physiological pathway that prevents sustained canonical Wnt signaling in neurons to ensure NPC integrity, appropriate NCT, and nuclear TDP-43 expression and function29. We show here that GDE2 inhibits persistent canonical Wnt signal transduction in neurons by regulating the surface expression of GPC6. Excessive GPC6 expression in vivo potentiates Wnt activation, resulting in the disruption of the NPC, alterations in Ran-dependent NCT, and TDP-43 mislocalization, while genetic reduction of GPC6 rescues these cellular changes in Gde2KO animals. These observations highlight GPC6 as an important activator of canonical Wnt signaling in neurons, whose regulation by GDE2 is critical for preserving the NPC, NCT, and appropriate localization of TDP-43. These observations provide insight into fundamental, physiological pathways that are necessary for postmitotic neuronal function and viability. GDE2 dysfunction, GDE2-dependent Wnt activation, and GPC6 have been independently implicated in neurodegenerative diseases with TDP-43 abnormalities26,27,28,29,37. Our observations suggest the compelling possibility that GDE2, GPC6, and canonical Wnt pathway regulation integrate to form a physiologically important signaling axis that, when disrupted, contributes to NPC, NCT, and TDP-43 pathologies in disease.

Wnt signaling in mammals is highly complex, with 19 identified Wnt ligands, 2 LRP receptors, and 10 Fzd co-receptors39. We utilized a scRNA-seq dataset of the adult mouse cortex38 to identify Wnt components that mediate the regulation of Wnt signaling by GPC6, and focused on the brain-specific Wnt ligand, Wnt7b, Wnt receptor LRP6, and a subset of Fzd co-receptors that are co-expressed with GDE2 and GPC6 in the adult cortex. Using a cell line that reports canonical Wnt activity, we found that GPC6 potentiates Wnt activation elicited by Wnt7b when co-expressed with Fzd1, Fzd4, and Fzd7, but not Fzd8. GPC6 belongs to the Dlp-class of Glypicans30. Mechanistically, Dlp-class Glypicans shield the lipid moiety of the Wnt ligand and facilitate their transfer to Fzd receptors, thus acting as a reservoir of signaling-competent Wnt ligand at the cell surface31,32. Our study suggests that there is some specificity in the mechanism of GPC6-dependent transfer of Wnt ligands to Fzd co-receptors. However, the basis of this specificity is unclear. Future studies that interrogate the structural interactions between GPC6, Wnt ligands, and Fzds may clarify this issue. Our analysis of GDE2 and GPC6 regulation of Wnt signaling focuses primarily on deep-layer cortical neurons. Notably, GPC1 and GPC4 are also co-expressed with GDE2 in deep-layer neurons, albeit at much lower levels than GPC6. It will be interesting to examine the contributions of GPC1 and GPC4 to GDE2-dependent Wnt activation and determine whether these pathways, in addition to GPC6, extend to other neuronal subtypes in the brain and contribute to NPC, NCT and TDP-43 abnormalities. Further molecular insight can be gleaned by determining the in vivo contributions of Fzd1, Fzd4, and Fzd7 to GDE2/Glypican-mediated Wnt7b signal transduction in different cellular contexts.

Studies of WNT-GFP-MYC mice here and elsewhere suggest that cortical neurons exhibit low and transient activation of canonical Wnt signaling in vivo39,55,56. A key feature of Dlp-class Glypicans that facilitates ligand transfer to Wnt receptors at the cell surface is the GPI-anchor, which enables their rapid diffusion through the plane of the membrane30. We propose that GDE2 GPI-anchor cleavage of GPC6 provides an effective and rapid means to slow/prevent ligand transfer to ensure low levels of Wnt signaling in cortical neurons. This is critical to prevent a buildup of GPC6 surface expression that would increase and sustain Wnt activation, which would negatively impact neuronal health and viability due to downstream effects on the NPC, NCT, and TDP-43 nuclear localization and function. Importantly, complete blockade of Wnt signaling is detrimental, as transient activation of Wnt signaling in neurons is required for neuronal functions such as dendritic plasticity55,56. Accordingly, we propose that periodic downregulation of GDE2 activity is essential to preserve the dynamics of neuronal Wnt activation. Studies of GDE2 developmental functions reveal that thiol-redox mechanisms regulate GDE2 trafficking to the cell surface and surface activity57,58. This suggests the compelling possibility that similar pathways might act physiologically to regulate GDE2 surface activity in adult neurons; however, this requires further investigation.

To date, GDE2 cleavage of its GPI-anchored substrates is known to occur in cis. However, the function of the substrate in question can have context-dependent cell-autonomous or non-cell-autonomous effects. For example, during development, GDE2 cleaves and inactivates the GPI-anchored metalloprotease RECK, which leads to the ADAM10-dependent cleavage of the Notch ligand Dll-116. This leads to the non-cell autonomous downregulation of Notch signaling and the initiation of neuronal differentiation16. In contrast, GDE2-dependent inactivation of RECK in mature neurons leads to the cell-autonomous regulation of APP by ADAM1027. In this study, we show that the loss of GDE2 regulation of GPC6 surface expression leads to the cell-autonomous potentiation of canonical Wnt signaling in cortical neurons that drives NPC, NCT and TDP-43 abnormalities. It is possible that this action might also have non-cell-autonomous consequences. For example, in the developing Drosophila wing disc, released soluble Dlp can shield the lipid moieties of Wnt ligands to enable long-distance signaling and contribute to target gene expression and timely growth31. Whether the reduction of released soluble forms of GPC6 consequent to GDE2 disruption affects neuronal function or viability remains to be tested.

Studies in animal models of disease, human postmortem brain, cerebrospinal fluid, and iPSC-derived neurons (iSNs) suggest that GDE2 dysfunction and abnormal activation of Wnt signaling contribute to TDP-43 mislocalization and impaired nuclear function in ALS and ALS/FTD28,29. Our discovery that GDE2 regulation of GPC6 is a physiological pathway that prevents sustained Wnt activation in neurons suggests potential contributions of GPC6 to diseases with TDP-43 proteinopathies in the context of GDE2 dysfunction/Wnt activation. Interestingly, GPC6 and its family member GPC4 are associated with AD and ADRDs in GWAS and GPC6 shows increased expression and mislocalization in spinal cord tissues of patients with ALS33,34,35,36,37. Further, GPC6 and GPC4 release is reduced in the SOD1G93A model of ALS26, and work in Drosophila suggests Dlp is a target of TDP-43 proteinopathy, with reduced expression at the neuromuscular junction (NMJ)37. These observations, combined with our study, suggest that GPC6 and potentially other GPC family members, are likely to have complex and varied contributions to neurodegenerative disease onset, progression, and pathology.

Methods

Animals

Mice were bred and maintained in accordance with approved Johns Hopkins University Institutional Animal Care and Use Committee (JHU IACUC) protocols. Gde2KO24, WT; Wnt-GFP-MYC (Rosa26 Tcf/Lef H2B-EGFP-6xMYC)45, and Gpc6+/-54 mice were bred, maintained, and genotyped as described previously. All experimental protocols were approved by JHU IACUC. All experiments were carried out in accordance with relevant guidelines and regulations, including JHU IACUC and the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

For each experiment, the groups being compared, the respective experimental units, and the sample sizes (number of cells and animals analyzed) are listed in the figure legends. No statistical power analysis was used to predetermine sample size. Sample sizes were determined to be similar or to exceed those previously reported in the literature. There were no exclusions of animals in this study. No randomization sequence was generated or used to allocate animals to control or treatment groups. However, animals were randomly picked for injections, and AAV injections were performed alternating between control and AAV GPC6 to minimize confounding effects from the order of treatments. Animals with different treatments or genotypes were maintained in mixed housing, separated by sex, until the age/timepoint listed in the figure legends for each experiment. Both males and females were used in all experiments, and the genotypes for each experiment are listed in the figures and figure legends. The statistical methods used are included in the figure legends. Datasets were tested for assumptions in their distribution, and corrections were made in statistical testing if the assumptions were not met.

AAV vector construction and injections

All AAV plasmid backbones were based on AAV-GFP.Cre59 (Addgene, 49056). AAV-GFP was constructed by deleting the coding sequence of Cre. For AAV-HA.GFP.MYC and AAV-HA.GPC6, the coding sequences were synthesized from pRK-HA.GFP.MYC (Addgene, 137763) and pDisplay HA.GPC616, respectively, and subcloned into the AAV backbone by replacing GFP/Cre. AAV-HA.GFP was constructed by deleting the MYC tag. The AAV vectors were packaged by Janelia Viral Tools using the PHP.eB capsid60,61.

AAV vectors were delivered to mice retro-orbitally at p28 at 1011 viral genome per animal as previously described60. Prior to retroorbital injections, animals weighed around 15 g and were anesthetized by intraperitoneal injection of 0.01 ml/g Avertin (1.3% 2,2,2-Tribromoethanol (T48402-25G) and 0.7% 2-methyl-2-butanol (Sigma 240486) in Phosphate Buffered Saline (PBS)). Animals were monitored post-injection. Prior to tissue harvesting at the time points indicated, animals were euthanized by intraperitoneal injection of 0.02 ml/g Avertin (see also Immunohistochemistry).

Immunohistochemistry (IHC) and image acquisition

4-month-old animals weighed around 25 g prior to tissue harvesting, and there were no gross weight differences between genotypes. Mice were anesthetized with 0.02 ml/g Avertin solution (1.3% 2,2,2-Tribromoethanol (T48402-25G) and 0.7% 2-methyl-2-butanol (Sigma 240486) in Phosphate Buffered Saline (PBS)) by intraperitoneal injection before transcardial perfusion with 0.1 M Phosphate Buffer (PB) and 4% Paraformaldehyde (PFA) in 0.1 M PB. Brains were dissected, post-fixed in 4% PFA for 18–20 h, washed with PBS, and prepared for embedding in cryomolds or paraffin blocks as previously described26.

Cryo-embedded brains were sectioned on a cryostat (Leica CM3050S) at 30–40 μm. IHC-immunofluorescence (IF) was performed on free-floating sections. Paraffin blocks were cut into 4 μm sections on a Rotary microtome (Leica RM2235) and collected on slides. Paraffin sections were deparaffinized with Xylenes and rehydrated in an ethanol series immediately prior to staining. Immunostaining was performed as previously described29. Briefly, sections were washed in PBS and permeabilized in 0.3% Triton-X-100 in PBS (PBST). For paraffin sections, antigen retrieval was performed with 0.1 M sodium citrate buffer (pH 6) for 20 min in a 95 °C water bath. Sections were then incubated for at least 1 h with blocking solution (5% Normal Donkey Serum (NDS) or 5% Bovine Serum Albumin (BSA) (Sigma, A9647-100G) in PBS and 0.3% Triton-X-100 in PBS). Sections were incubated overnight with primary antibodies at 4 °C and washed with PBS the next day. Sections were then incubated with the appropriate fluorescently conjugated secondary antibodies (Jackson Immunoresearch) for 1–2 h at room temperature (RT). Nuclei were stained with Hoechst 33,342 (1:500, Thermo Fisher) in PBS for 15 min at RT. Sections were mounted on slides with Prolong Gold mounting media (ThermoFisher, P36931) and coverslipped before imaging. See Table 1 for antibody information.

For immunostaining of WNT-GFP-MYC animals injected with AAV, transduced cells were identified by staining for HA with Alexa Fluor 488-conjugated secondary antibodies. MYC expression was used as the Wnt reporter readout and visualized by immunostaining with Cy3-conjugated secondary antibodies (ex/em = 554/566 nm). This allowed us to separate MYC expression from any endogenous GFP signal (ex/em = 488/513 nm) that could confound the analysis. Images were acquired with a Zeiss LSM 700 microscope. All channels were acquired on separate tracks to avoid any signal bleed through. The same settings were used for all images acquired within the same experiment. For image presentation, the same adjustments in brightness and contrasts were applied to images within the same experiment.

RNAscope

RNAscope coupled with IHC was carried out using the RNAscope Multiplex Fluorescent v2 Assay with the Co-Detection kit (ACDbio, 323110) according to the manufacturer’s protocol. Gpc6 mRNA was virtualized using the probe Mm-Gpc6-C2 (ACDBio, Cat No. 442841-C2) and Opal-570 fluorophore (Akoya Biosciences, FP1488001KT, 1:500), and costained with NeuN (synaptic systems, 266004,1:500). Slides were mounted with ProLong Gold mounting media with DAPI (ThermoFisher, P36931), coverslipped, and imaged on a Zeiss LSM800 confocal microscope.

Quantitative image analysis

All images were quantified using Fiji (ImageJ, imagej.org, version 1.54p) as previously descirbed29. For mouse tissues, 2–4 sections per mouse were analyzed depending on the experiment. For fluorescent stains, threshold-based cell counting macros created in Fiji were used to quantify NeuN-positive cells and cells double positive for both NeuN and MYC. Neurons were considered positive for MYC if their nuclear intensity was more than 2x standard deviations above the mean background intensity.

To quantify nuclear/cytoplasmic (N/C) or C/nuclear envelope (NE) ratios, DAPI was used to delineate the nucleus from the cytoplasm in neurons. NeuN signal outside of DAPI was used to define the cytoplasm. The nuclear and cytoplasmic intensities were measured as the average fluorescence intensities from three regions of interest (ROIs) of the same size in the nucleus and in the cytoplasm in the channel of interest. The NE intensity was measured by drawing a freehand line of 2-pixel width around the nucleus. Five background measurements were taken per image, the average of which was used to correct for differences in background intensity. The final N/C ratio was calculated by dividing the average background-corrected nuclear intensity by the average background-corrected cytoplasmic intensity for each cell. The final C/NE ratio was calculated by dividing the average background-corrected cytoplasmic intensity by the average background-corrected NE intensity for each cell. Nuclear intensities were calculated using a mask of the area within DAPI to measure mean intensity in the relevant channel of interest.

Cell culture

Mouse primary cortical neuronal culture

P0 or P1 pups were euthanized by cryoanesthesia followed by decapitation. Mouse primary cortical cultures were prepared from the extracted mouse cortices and plated on poly-l-lysine-coated plates or acid-washed coverslips as previously described27 with minor modifications. Cells were maintained in Neurobasal medium supplemented with 2% B27, 1% L-glutamine, and 1% Pen/Strep. 5µM cytosine arabinoside was added on DIV2 to inhibit glial growth and removed on DIV3. From DIV4, cultures were fed every 3 days and maintained at 37 °C until harvest at DIV21.

Immortalized cell lines

SH-SY5Y (CRL-2266) cells were obtained from American Type Culture Collection (ATCC) and cultured according to the manufacturer’s protocol. SuperTopflash (STF) cells were a gift from Dr. Jeremy Nathans (Johns Hopkins University). Plasmids were transfected with FuGENE HD (Promega E2311) for STF cells and Lipofectamine 2000 (Invitrogen 11668-027) for SH-SY5Y cells accordingly to the manufacturer’s protocol. The plasmid expressing HA-GPC6 was previously described16. Plasmids expressing Wnt7b, Lrp6, Fzd1, Fzd3, Fzd4, Fzd7, and Fzd8 are kind gifts from Dr. Jeremy Nathans. siRNAs were purchased from Santa Cruz Biotechnology to knock down hGPC6 (sc-75152) compared to non-targeting control (sc-37007).

Cell surface biotinylation and fractionation

Cell surface biotinylation and fractionation were performed according to a previously described protocol62 with minor modifications. Briefly, culture plates were placed on ice for 15 min prior to washing to prevent cold-shock-induced lysis. Cells were rinsed once with ice-cold PBSCM (PBS-calcium-magnesium: 1× PBS, 1mM MgCl2, 0.1mM CaCl2 (pH 8.0)), and then incubated with Sulfo-NHS-SS-biotin (1.0 mg/mL, Thermo Scientific 21331) for 30 min at 4 °C. Cells were then washed with PBSCM and incubated in 20 mM glycine in PBSCM twice for 5 min to quench unreacted biotinylation reagent. Neurons were lysed in radioimmunoprecipitation assay (RIPA) buffer containing 1x protease inhibitor cocktail (Sigma, P8340) and spun down at 21,000 g for 15 min at 4 °C. Protein levels were measured using a BCA Protein Assay kit (ThermoFisher, 23225). Equal amounts of proteins were incubated overnight with NeutrAvidin agarose beads (ThermoFisher, 29201) at 4 °C on a rotator. The beads were then washed four times with RIPA buffer containing protease inhibitor, and the captured biotinylated fraction was eluted by incubating the beads in 2x Laemmli buffer at 95 °C for 10 min. The total and surface fractions were then analyzed using immunoblotting.

Immunoblotting

4-month-old animals weighed around 25 g prior to tissue harvesting, and there were no gross differences between genotypes. Mice were anesthetized with 0.02 ml/g Avertin solution (1.3% 2,2,2-Tribromoethanol (T48402-25G) and 0.7% 2-methyl-2-butanol (Sigma 240486) in Phosphate Buffered Saline (PBS)) by intraperitoneal injection before decapitation and tissue harvesting. Mouse cortical tissues were sonicated in RIPA buffer containing 1x protease inhibitor cocktail (Sigma, P8340) and spun down at 21,000 g for 20 min at 4 °C. Protein levels were standardized across all samples using a BCA Protein Assay kit (ThermoFisher, 23,225) and 4x Laemmli buffer was added to samples to a final concentration of 1x. Cultured cells were lysed directly in 1x Laemmli buffer, sonicated, and spun down at 21,000 g for 10 min at RT.

Immunoblotting was performed as previously described29. Samples were boiled and run on 7.5% or 10% polyacrylamide gels in tris/glycine buffer before transferring to polyvinylidene difluoride (PVDF) membranes. PVDF membranes were blocked with 5% milk in tris-buffered saline containing 0.3% Tween-20 (TBST) or Everyblot blocking buffer (Bio-Rad laboratories,12010020) for 1–2 h at RT before applying primary antibodies overnight at 4 °C. After washing with TBST, the membranes were incubated with the appropriate horseradish peroxidase (HRP) or fluorescent protein-conjugated secondary antibodies for 1 h at RT. Membranes were washed again with TBST, developed using enhanced chemiluminescence substrate when appropriate (Kindle Biosciences, R1004), and imaged using a ChemiDoc Imager (Bio-Rad). Blots were analyzed using Fiji (ImageJ, imagej.org, version 1.54p). See Table 1 for antibody information.

TOPFlash luciferase assay

Renilla luciferase control construct (pRL-TK) was purchased from Promega (E2241). Dual-luciferase reporter assay was performed according to the manufacturer’s instructions (Promega, E1910). The readings of Firefly and Renilla luminescence were recorded using a Turner BioSystems Luminometer (TD-20/20). Relative luciferase activity was calculated by normalizing Firefly/Renilla values.

Analysis of publicly available ScRNAseq and RNAseq datasets

scRNAseq data of the adult mouse cortex and hippocampus from the Allen Brain Atlas38 were retrieved using CellxGene Census63 using the accession ID d7291f04-fbbb-4d65-990a-f01fa44e915b and visualized using the Scanpy64 and anndata65 packages in Python. No custom code was used in preparation of this manuscript. Bulk RNAseq datasets of 4-month-old WT and Gde2KO cortices were previously published29 and available at GEO using the accession ID GSE246462. Sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality using Trimmomatic v.0.36. The trimmed reads were mapped to the Mus musculus reference genome GRCm38 available on ENSEMBL using the STAR aligner v.2.5.2b to generate BAM files. Unique gene hit counts were calculated using feature Counts from the Subread package v.1.5.2. Only unique reads that fell within exon regions were counted. The gene hit counts table was normalized using transcript per million to examine the expression of Gpc6 (ENSMUSG00000058571).

Statistical analyses

Data were analyzed and plotted using GraphPad Prism (graphpad.com, version 10). Statistical significance for pairwise comparisons was derived using a two-tailed Student’s t test with relevant corrections. For multiple comparisons, we used ANOVA with corrections for multiple comparisons. All values are reported as mean ± s.e.m. In figures, asterisks denote statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Specific statistical information for each experiment is included in the figure legends.

Table 1 Antibody Information.
Full size table