Organization of TNIK in dendritic spines

Abstract

TRAF2- and NCK-interacting kinase (TNIK) has been identified as an interactor of the psychiatric risk factor, Disrupted in Schizophrenia 1 (DISC1). As a step toward deciphering its function in the brain, we performed high-resolution light and electron microscopic immunocytochemistry. We demonstrate here that TNIK is expressed in neurons throughout the adult mouse brain. In striatum and cerebral cortex, TNIK concentrates in dendritic spines, especially in the vicinity of the lateral edge of the synapse. Thus, TNIK is highly enriched at a microdomain critical for glutamatergic signaling and implicated in the regulation of synaptic strength.

TRAF2- and NCK-interacting kinase (TNIK) is a serine/threonine kinase of the Ste20 family (Fu et al., 1999; Taira et al., 2004). In addition to its enzymatic activity, TNIK also contains scaffolding domains, raising the possibility that TNIK also acts as a scaffold that assembles molecular complexes for downstream signal transduction. TNIK was originally cloned from a human brain cDNA library, and its message is found at particularly high levels in the brain (Fu et al., 1999). Genetic association studies, as well as transcription profiling of blood and postmortem brain, support a role for TNIK as a risk factor for several psychiatric diseases, including bipolar disorder, attention deficit-hyperactivity disorder, and schizophrenia (Glatt et al., 2005; Matigian et al., 2007; Potkin et al., 2009; Shi et al., 2009; Ayalew et al., 2012; Elia et al., 2012). Moreover, TNIK binds to the well-known psychiatric risk factor DISC1 (Camargo et al., 2007; Wang et al., 2011; Coba et al., 2012).

Proteomic studies consistently find that TNIK protein is in the biochemically-defined postsynaptic density (PSD) (Jordan et al., 2004; Peng et al., 2004; Collins et al., 2006; Trinidad et al., 2008; Hussain et al., 2010; Wang et al., 2011), where it has been implicated in postsynaptic signaling (Hussain et al., 2010; Wang et al., 2011; Coba et al., 2012). Knockdown experiments in cultured primary neurons point to a role for TNIK in promoting surface expression of the AMPA-type glutamate receptor subunit GluR1 (Hussain et al., 2010; Wang et al., 2011). This effect may be related to the finding that TNIK can stabilize the levels of several other PSD proteins, including the scaffold protein PSD-95 and the transmembrane AMPA receptor regulatory protein γ-2, stargazin (Wang et al., 2011). TNIK is also linked to the NMDA receptor, via the A-kinase anchoring protein 9 (Yotiao), and activation of NMDA receptors and group I metabotropic glutamate receptors can modify TNIK phosphorylation (Coba et al., 2012).

To gain further insight into the neurobiology of TNIK, we have studied its localization in the brain of adult mouse, focusing on neocortex and striatum, two regions of particular interest in human neuropsychiatric disease.

MATERIALS AND METHODS

Animals

All procedures related to the care and treatment of animals were in accordance with institutional and NIH guidelines; all animals use protocols were reviewed and approved by the relevant Institutional Animal Care and Use Committee.

To generate TNIK knockout animal, a targeting vector containing 6.2 and 6.3 kb of genomic DNA 5′ and 3′ respectively of the targeted modification was constructed. A loxP site 5′ and a PGK promoter-neo^R-LoxP cassette 3′ of exon 7 was introduced into this construct, thereby flanking exon7 with LoxP sites (Fig. 1A). C57BL/6NTac ES cells were targeted using standard procedures, and appropriately targeted clones were identified by Southern blotting using genomic probes outside of the targeting vector on both the 5′ and 3′ sides. Appropriate incorporation of the LoxP site 5′ of exon 7 was confirmed by PCR.

Figure 1. Preparation of the knockout mice.

A: Gene targeting of TNIK locus and RT-PCR of whole brain mRNA from targeted mice. Genomic locus illustrating the region of exons 5–7. Targeting vector indicated showing homology arms, LoxP sites flanking exon 7 (▶), and neomycin (G418) resistance cassette used for selection. Neo^r cassette is flanked by Frt sites (●). After introduction into the germline, exon 7 was removed by Cre recombinase-mediated deletion. B: RT-PCR of whole brain RNA from animals of the indicated genotypes. PCR primers located in exons 6 and 8 indicated by arrows. Deletion of exon 7 results in a novel transcript in animals carrying the Δ7 allele that contains early termination codons as indicated. Identity of PCR products was verified by sequencing (data not shown). C: Immunofluorescence staining for TNIK in neocortex from WT (A) and TNIK^Δ7/TNIK^Δ7 (B) mouse brain.

Scale bar = 50 μm in C and D.

Targeted ES cells were used to establish this modification in the C57BL/6NTac mouse strain by standard procedures. Exon 7, which encodes 43 amino acids of the kinase domain, was removed by crossing mice carrying the targeted allele to a protamine-cre recombinase transgenic. In these C57BL/6NTac transgenics, Cre recombinase is expressed in the male germline, allowing animals bearing a fully recombined allele lacking exon 7 in all tissues to be derived. Subsequent breeding eliminated the transgene and established the recombined allele lacking exon 7, referred to as TNIK^Δ7, in the germline.

TNIK mRNA was detected in whole brain mRNA by RT-PCR using the following primer sets: Forward primer (exon 6) 5′-GGCCTGAGTCACCTGCACCAGC-3′; Reverse primer (exon 8) 5′-GGGCACCTTCTGCCATCTCA-3′ (Fig. 1B). TNIK^Δ7/TNIK^Δ7 homozygous mice showed a residual transcript in which exon 6 was spliced to exon 8, resulting in the introduction of 2 stop codons within the next 30 nucleotides after the novel 6/8 splice site.

Fully inbred C57BL/6NTac homozygous TNIK^Δ7/TNIK^Δ7 and wild type littermates were generated for this study by intercross of heterozygotes (+/TNIK^Δ7).

Antibody specificity

Table 1 provides a list of primary antibodies used in this study.

Table 1.

Primary Antibodies

Antigen	Immunogen	Source	Dilution
ChAT	Human placental choline acetyltransferase	Chemicon, AB144, RRID:AB_11212843	1:2,000
GABA	GABA conjugated to BSA	Millipore Corporation, clone 5A9, Cat. # MAB316, lot # 2080836	1:5,000
TNiK	Synthetic peptide from internal domain of human TNiK aa 809-933.	SIGMA Cat# HPA012297, RRID:AB_1858225	1:1,000
VGLUT1	C-terminal peptide rat VGLUT1: GATHSTVQPPRPPPPVRDY	Chemicon, cat. # AG208, RRID:AB_11213019	1:5,000

To identify TNIK we used a rabbit polyclonal antibody (Sigma-Aldrich Cat# HPA012297 RRID:AB_1858225) raised against human TNIK. To verify antibody specificity, we performed immunocytochemistry on brain sections from TNIK KO mice, run in parallel with sections from control mice. Staining with HPA012297 antibody was robust in tissue from WT mice, but extremely weak in tissue from TNIK^Δ7/TNIK^Δ7 mice, exhibiting a “background” pattern unrelated to that seen for WT brain sections (Fig. 1C, D). Three additional TNIK antibodies were also tested: MC-7403 (MBL International, Woburn, MA), PA5-15181 (Thermo Fisher Scientific Cat# PA5-15181 RRID:AB_2207637), and HPA012128 (Sigma-Aldrich Cat# HPA012128 RRID:AB_1858226). Using MC-7403, PA5-15181 and HPA012128 (at a range of dilutions: 1:100 to 1:10,000), similar patterns of staining were observed for both WT and TNIK^Δ7/TNIK^Δ7 tissues; thus, these antibodies were not suitable for immunohistochemistry. Accordingly, HPA012297 was used for all data presented in this paper.

To identify the vesicular glutamate transporter VGLUT1, we used a guinea pig polyclonal antibody (Chemicon, Millipore Corporation, Billerica, MA, AB5905, lot# 24041061, RRID:AB_11213019) raised against rat VGLUT1. We have previously sequenced the immunogenic peptide (Chemicon, Millipore Corporation, Billerica, MA, #AG208), concluding that this antibody was raised against a C-terminal peptide (GATHSTVQPPRPPPPVRDY, (Melone et al., 2005)). The antibody recognizes a single band of ~ 60 KDa on immunoblots of synaptic membrane fractions from rat cerebral cortex. Furthermore, immunogold labeling shows that VGLUT1 immunoreactivity is selectively associated with axon terminals forming asymmetric synapses in cerebral cortex and hippocampus.

To identify GABAergic synapses, we use a mouse monoclonal antibody raised against GABA conjugated to BSA (Millipore Cat# MAB316, RRID:AB_94396, lot # 2080836, Millipore Corporation, Billerica, MA). This antibody does not cross-react with other amino acids.

For choline acetyltransferase (ChAT), a goat polyclonal antibody was used (Chemicon, cat. # AG208, RRID:AB_11212843). The antibody recognizes ChAT on Western blots, and immunostaining was not observed in tissue sections when the antiserum was preadsorbed with pure ChAT protein (Bruce et al., 1985; Shiromani et al., 1987). Furthermore, this antibody produced labeling that was similar to labeling produced by the well-characterized monoclonal antibody AB8 (Armstrong et al., 1983; Levey et al., 1983).

Tissue preparation

After inducing deep anesthesia with sodium pentobarbital (60 mg/kg, i.p.), mice were intracardially perfused with 50–100 ml of fixative: 4% freshly-depolymerized paraformaldehyde in phosphate buffer (PB, 0.1 M, pH 7.4) for light microscopy (LM); a mixture of 4% paraformaldehyde and 0.1% glutaraldehyde in PB, for LM labeling with GABA; or a mixture of 2% paraformaldehyde and 2% glutaraldehyde in PB for post-embedding electron microscopy (EM). Brains were sectioned at 50 μm and 100 μm on a Vibratome, and collected in cold PB.

Light microscopic immunohistochemistry

Free-floating sections were incubated in 10% normal donkey serum (NDS). The primary antibody (TNIK, 1:1000) was then applied overnight. For immunoperoxidase microscopy, sections were then incubated for 3 hours in biotinylated secondary antibody (1:200; Jackson ImmunoResearch, West Grove, PA) and for 1 hour in ExtrAvidin-peroxidase complex (1:5,000; Sigma, St. Louis, MO); peroxidase was histochemically visualized with nickel-intensified diaminobenzidine. Processed sections were mounted on gelatin-coated slides, air dried, and cleared with xylene before being coverslipped with D.P.X. mountant (BDH Chemicals, Poole, England).

For immunofluorescence microscopy, antigenic sites were visualized with donkey IgG conjugated to DyLight 549 (1:200, Jackson ImmunoResearch; West Grove, PA). For double labeling, the second primary antibody (1:5,000, guinea pig anti-VGLUT1, 1:5,000, mouse anti-GABA or 1:2,000 goat anti-ChAT) was applied overnight and visualized by a secondary antibody conjugated to Alexa 488 (1:200, Invitrogen). Some sections were then counterstained with Hoechst 33342 (Sigma, St. Louis, MO) to visualize nuclei and NeuroTrace 640-660 (Invitrogen, Thermo Fisher Scientific, Rockford, IL) to selectively visualize neuronal somata. Control experiments, in which the primary antibodies were omitted, were performed to control for nonspecific binding of the secondary antibody. Sections were examined with a Leitz DMR microscope (Leica, Wetzlar, Germany), and a Leica SP2 confocal microscope.

Electron microscopic immunocytochemistry

Brain sections (100 μm) were pre-treated in 0.1% calcium chloride in 0.1 M sodium acetate, rinsed, and cryoprotected in a graded series to 30% glycerol in 0.1 M sodium acetate. Sections were quick-frozen in methanol chilled with dry ice. Freeze substitution in 4% uranyl acetate in methanol was carried out in a Leica Automatic Freeze Substitution System; after rinsing in methanol, sections were infiltrated with Lowicryl HM-20, mounted between sheets of ACLAR plastic sandwiched between glass slides, and polymerized with ultraviolet light. After polymerization, regions of interest were cut from the sections and glued to plastic blocks. Sections were cut at ~70–90 nm with an ultramicrotome and collected on nickel grids, coated with Coat-Quick (Electron Microscopy Sciences). Grids were pre-treated 15 minutes at 60 °C in 0.01 M citrate buffer, pH 6, rinsed in water, blocked in 1% bovine serum albumin in TRIS-buffered saline with 0.005% Tergitol NP-10, then incubated overnight at 21–24°C with the primary antibody (TNIK, HPA012297, 1:1000). Grids were rinsed, blocked in 1% normal goat serum, and incubated in goat anti-rabbit IgG F(ab)2 conjugated to 10-nm gold particles (1:20, Ted Pella, Redding, CA). Grids were then rinsed and counterstained with 1% uranyl acetate, followed by Sato’s lead, and examined in a Philips Tecnai electron microscope (Hillsboro, OR) at 80 kV.

Image analysis

To further test TNIK antibody specificity for postembedding EM immunohistochemisty, and to evaluate the distribution of labeling in different subcellular compartments, we collected random EM photomontages (each ~ 10 × 10 μm) from neuropil of the cerebral cortex from WT and KO mice. We used ImageJ (Schneider et al., 2012) (ImageJ, RRID:nif-0000-30467) to outline all identifiable dendritic spines, PSDs, presynaptic terminals, mitochondria and dendritic shafts. Since these were from single sections, many profiles, especially neuroglial processes and the smallest axons and dendrites, could not be unambiguously identified, and were therefore excluded from further analysis. We then computed the density of gold particles for each compartment, analyzing tissue from WT and KO mice in a blinded fashion.

To investigate the organization of immunogold label associated with the PSD of asymmetric synapses (likely for their morphology to be glutamatergic), we took measurements from 35–40 immunolabeled synapses for each of the two brain regions studied, from each of three animals. To define “axodendritic” position, the shortest distance between the center of each gold particle and the outer leaflet of the postsynaptic membrane was measured. To define the “lateral” synaptic position of a gold particle along the synapse, we measured the distance from each end of the PSD to a line drawn perpendicular to the synaptic apposition, running through the center of the particle (Valtschanoff and Weinberg, 2001). Normalized lateral position of each gold particle within the axodendritic peak (from −10 nm to +50 nm from the postsynaptic membrane) was computed as L_N = |(a−b)|/(a+b), where a and b are tangential distances along the plasma membrane from the center of the gold particle to the lateral edges of the synaptic specialization; thus L_N = 0 for gold particles at the center of the PSD, and L_N = 1 for particles at its edge.

RESULTS

Immunoperoxidase-reacted sections displayed staining for TNIK in the grey matter throughout the brain; with little or no immunostaining in the white matter (Fig. 2A). TNIK staining was strong in olfactory bulb (2B), piriform cortex (2C), isocortex (2D), hippocampus (2E), striatum (2F), thalamus (2G), hypothalamus (2H) and cerebellum (2I). Immunostaining was weaker in midbrain, pons, and medulla. We focused our attention on neocortex and striatum, two brain regions implicated in multiple human psychiatric disorders.

Figure 2. Immunoperoxidase staining for TNIK in the mouse brain.

A: Parasagittal section of whole mouse brain. Staining is conspicuous in gray matter, largely sparing white matter. Higher magnification views show patterns of staining in specific regions of brain, including B: Olfactory bulb; C: Piriform cortex; D: Neocortex; E: Hippocampus; F: Striatum; G: Thalamus; H: Hypothalamus; I: Cerebellum.

Scale bar = 2 mm in A; 500 μm in B, 200 μm in C; 250 μm in D, E, F; 500 μm in G; 250 μm in H; 500 μm in I.

TNIK was expressed through all layers of the neocortex, most prominent in layer II (Fig. 2D); no obvious differences in the pattern of labeling were seen across neocortical areas. TNIK was detected in somata and dendrites, and in small puncta throughout the neuropil. Counterstaining with NeuroTrace (a fluorescent Nissl stain) and Hoechst 33342 (a nuclear stain) showed that TNIK was generally restricted to neurons. Most neuronal somata throughout the neocortex contained TNIK (Fig. 3A–H), although scattered somata exhibited little or no staining (arrows in Fig. 3E–H). Our suspicion that these might represent inhibitory interneurons was confirmed by double labeling with GABA, which showed that most GABA-positive somata were immunonegative for TNIK (Fig. 3I–K).

Figure 3. Immunofluorescence labeling for TNIK in neocortex.

A–D: Sections were counterstained with Hoechst 33342 to visualize nuclei, and with NeuroTrace 640-660 to visualize neuronal somata. Most NeuroTrace-identified neurons are also immunopositive for TNIK, though to varying degrees. E–H: Enlargement of boxed area from upper panel; white arrow points to a small neuron immunonegative for TNIK (extended-focus confocal images). Insets: enlarged single-focus images from the same field; note that many somatic TNIK “blobs” in E correspond to Nissl bodies as defined by Neurotrace (F); some examples are outlined with white circles. I–K; Double labeling for TNIK and GABA; a GABAergic soma (asterisk) is immunonegative for TNIK.

Scale bar = 30 μm in A–D; 20 μm in E–H; 10 μm in I–K.

In the striatum, TNIK was prominently expressed in the gray matter between fascicles of myelinated fibers, which themselves were devoid of staining (Fig. 4). TNIK staining was strong in somata and dendrites of cells likely for their number, size, and distribution to be medium spiny neurons; numerous puncta in the neuropil were also staining. Scattered larger somata exhibited little or no TNIK staining. Double labeling with choline acetyltransferase (ChAT) showed that most of these TNIK-negative neurons were large aspiny cholinergic interneurons (Fig. 4I–K).

Figure 4. Immunofluorescence labeling for TNIK in striatum.

A–H: Sections were counterstained with Hoechst 33342 to visualize nuclei, and NeuroTrace 640-660 to visualize neuronal somata. Most neurons in the field were immunopositive for TNIK. Boxed area in upper panel is shown at higher magnification in middle panel (E–H). The asterisk marks a neuron immunonegative for TNIK. I–K: Double labeling for TNIK and ChAT; white arrows point to ChAT-positive neurons that are immunonegative for TNIK.

Scale bar = 30 μm in A–D; 10 μm in E–H; 25 μm in I–K.

In both cortex and striatum, TNIK was excluded from the nucleus. Immunostaining within the somatic cytoplasm was typically organized into patches or blobs; many of these blobs colocalized with Nissl bodies (as defined by Neurotrace), suggesting an association of TNIK with rough endoplasmic reticulum (Fig. 3E–H; Fig. 4E–H. TNIK staining was weak and diffuse in dendrites, which consequently were barely visible. It was our impression that in many cases these dendrites were “coated” with brightly-stained puncta. Abundant punctate TNIK staining was present throughout the neuropil in both neocortex and striatum. Double labeling experiments revealed a close relationship between these puncta and the presynaptic marker VGLUT1 (Fig. 5), suggesting an association of TNIK with excitatory synapses in both cortex (whose spiny principal neurons are glutamatergic) and striatum (whose spiny principal neurons are GABAergic).

Figure 5. Colocalization of TNIK with VGLUT1.

Double labeling in mouse neocortex (A–C) and striatum (D–F) shows that many TNIK puncta (green) are closely apposed to, or partially overlap with, the vesicular glutamate transporter VGLUT1 (magenta), suggesting that TNIK concentrates at synapses.

Scale bar = 4 μm in A–F.

In summary, TNIK in the forebrain was expressed mainly in spiny neurons, especially at puncta likely to represent glutamatergic synapses.

To elucidate the organization of TNIK in the vicinity of the synapse, we analyzed its distribution with electron microscopy, using postembedding immunogold labeling (Fig. 6). Gold particles coding for TNIK labeled somata and dendrites, concentrating in dendritic spines; particles were only occasionally seen in presynaptic terminals. In spines, TNIK labeling concentrated over the PSD. Labeling was also detected within the cytoplasm of the spine head, though at lower concentrations. To ensure that the preferential PSD labeling seen here reflected the distribution of antigen, rather than simply arising from nonspecific adsorption to the protein-rich environment of the synapse, we performed the same immunogold analysis on tissue from TNIK^Δ7/TNIK^Δ7 mice.

Figure 6. Ultrastructural analysis using postembedding immunogold electron microscopy reveals association of TNIK with the postsynaptic density (PSD).

A: an immunopositive spine in cerebral cortex; arrows point to two particles lying close to the edge of the PSD. B: micrograph shows a thin dendritic shaft in cortex receiving two asymmetric synapses; both are immunopositive for TNIK. The lower arrow points to a gold particle lying in the middle of the synapse, whereas the upper arrow points to a particle lying at the edge of the PSD. C: view of a large immunopositive spine in striatum; two gold particles (arrows) lie close to the postsynaptic membrane. D: histograms show the distribution of immunolabeling in the axodendritic axis (20 nm bins, cerebral cortex: N=135 synaptic particles, striatum: N=133 particles). Diagram at bottom illustrates measurement technique; positive values correspond to particles lying inside the postsynaptic plasma membrane. The pattern of labeling was consistent for cortex (left) and striatum (right), peaking just inside the postsynaptic membrane. E: histograms show distribution of label tangentially along the synapse, normalized such that 0 corresponds to the center of the PSD, and 1.0 to its edge (see diagram at bottom; cortex: N=120 synaptic particles, striatum N=116 particles). Labeling in both cortex (left) and striatum (right) concentrated in the vicinity of the edge of the PSD. An additional pool of TNIK lay at the center of the PSD in striatum, but not in cortex. X-axis for each histogram has been normalized so that the largest bin corresponds to 1.0 units, and the origin corresponds to zero.

Scale bar = 200 nm in A, B, C

Labeling over the protein-rich mitochondria was twice as high in the TNIK^Δ7/TNIK^Δ7 as in WT material (Table 2). We interpret this staining as nonspecific, and have often encountered similar nonspecific labeling over mitochondria in other postembedding immunocytochemical studies. In contrast, the density of gold particles over the PSD was more than 6-fold higher in WT than in TNIK^Δ7/TNIK^Δ7 tissue, implying that at least 85% of the staining at the PSD of WT mice represented genuine TNIK protein (Table 2). Labeling densities in dendritic spines and shafts were also higher in the WT than in TNIK^Δ7/TNIK^Δ7, though background noise in the face of a weaker signal made interpretation difficult. Quantification of label densities in different subcellular compartments showed that label over the PSD was enriched by a factor of at least 60, compared to the entire neuropil, consistent with the high levels of enrichment seen in biochemical PSD fractions (Wang et al., 2011). We conclude that TNIK is indeed expressed at high levels within the anatomical domain of the PSD.

Table 2.

	WT	KO	WT/KO
PSDs	32.16 ± 4.81	5.32 ± 0.71	6.1
Spines	0.86 ± 0.40	0.42 ± 0.19	2.0
Dendritic shafts	0.38 ± 0.07	0.26 ± 0.02	1.5
Axon terminals	0.55 ± 0.24	0.52 ± 0.17	1.0
Mitochondria	1.49 ± 0.36	3.06 ± 0.36	0.5

Number of gold particles/μm²

To better define the spatial organization of TNIK in the PSD, we measured the distance of synapse-associated immunogold particles to the plasma membrane. We found that labeling for TNIK concentrated very close to the postsynaptic membrane, lying at a mean distance of 6 nm inside the membrane, in both neocortex and striatum (Fig. 6D). To determine the tangential distribution of TNIK along the length of PSD, we computed the “normalized lateral position” (L_N) of gold particles along the membrane (confining our attention to particles lying within a window between −10 nm and +50 nm from the postsynaptic plasma membrane, to restrict attention to label associated with the PSD). We found that TNIK labeling in both neocortex and striatum concentrated in the vicinity of the lateral edge of the PSD (Fig. 6E). The striatum, but not cortex, also exhibited a peak of label (~21% of the total synaptic pool) at the center of the PSD (Fig. 6E).

DISCUSSION

We used immunohistochemistry to examine the distribution of TNIK in the mouse brain. The protein was widely expressed in neurons and neuropil throughout the gray matter; in somata it was excluded from the nuclear compartment. Focusing on cerebral cortex and striatum, we found that TNIK concentrates in the dendritic spines of principal neurons (where it associates with the postsynaptic density), whereas local circuit interneurons (including GABAergic neurons in cortex and cholinergic neurons in striatum) often failed to stain above background. At least some of the somatic labeling likely reflected newly-synthesized protein awaiting transport, but we cannot exclude that TNIK may also play a functional role in the cytoplasm.

In model systems, TNIK interacts with the adhesion and signaling molecule β-catenin to modulate cell proliferation via the canonical Wnt signaling pathway (Mahmoudi et al., 2009; Satow et al., 2010; Shitashige et al., 2010; Schurch et al., 2012). In contrast to this well-documented effect in cancer biology and in normal embryonic development, the role of TNIK in the adult brain is poorly understood. Impaired neurogenesis has been reported in TNIK KO mice (Coba et al., 2012), suggesting that TNIK may also be important for cell proliferation in the brain. However, the presence of TNIK in excitatory synapses points to a role beyond neurogenesis.

The present results show that TNIK in spines concentrates at the lateral edge of the PSD in both excitatory pyramidal cells of cerebral cortex and inhibitory medium spiny neurons (MSNs) of striatum; TNIK also concentrated in the PSD center in synapses onto MSNs, but not in synapses onto pyramidal cells. This result, along with previous data suggesting differences in the organization of PSD proteins between the excitatory spiny cells of cortex and hippocampus and the inhibitory spiny cells of striatum (Kharazia et al., 1996; Kharazia and Weinberg, 1997; Bernard and Bolam, 1998; Clarke and Bolam, 1998; Racca et al., 2000; Burette et al., 2014), raise the possibility that there may be fundamental differences in synaptic structure between glutamatergic synapses onto excitatory vs inhibitory neurons.

Notwithstanding this subtle difference, TNIK in both types of synapses concentrates mainly at the peripheral edge of the PSD. This lateral “perisynaptic” zone is also a preferential target for the actin filaments that link the synapse to the spine core (Frost et al., 2010; Svitkina et al., 2010; Burette et al., 2012). The perisynaptic zone plays a special role in the trafficking of AMPA receptors (Lu et al., 2007; Petrini et al., 2009; Freche et al., 2011), which represents a principal mechanism underlying long-term synaptic plasticity. Evidence from both knockdown experiments and knockout mice shows that disruption of TNIK impairs AMPAR signaling (Hussain et al., 2010; Wang et al., 2011; Coba et al., 2012), apparently via activation of Rap2, a regulator of the actin cytoskeleton (Hussain et al., 2010; Coba et al., 2012); TNIK may also regulate actin by phosphorylating the actin-binding protein gelsolin (Fu et al., 1999).

We find that TNIK lies very close to the plasma membrane, where it can interact with integral membrane proteins (Valtschanoff and Weinberg, 2001), especially those concentrating at the edge of the synapse. Intriguingly, the cadherin/catenin complex is also enriched at synapses (Benson and Tanaka, 1998; Togashi et al., 2002), especially in the perisynaptic zone (Uchida et al., 1996; Petralia et al., 2005; Petralia et al., 2010). Besides its role in synapse assembly and dendritic spine formation, this complex is involved in actin remodeling, and also regulates synaptic AMPARs (Nuriya and Huganir, 2006; Okuda et al., 2007; Silverman et al., 2007; Tai et al., 2008; Peng et al., 2009; Brigidi and Bamji, 2011). This correspondence in both location and function leads us to speculate that TNIK at the synapse is linked to β-catenin.

Pathologies in both neocortex and striatum are strongly implicated in psychiatric disease. Especially considering TNIK’s association with AMPA receptors, with actin signaling pathways, and with DISC1 (variants of which have been linked to multiple psychiatric diseases), we suggest that that understanding the biology of TNIK represents an important avenue of research to understand psychiatric illness.