Tcf4 belongs to the basic-helix-loop-helix (bHLH) family of proteins, which act as transcription factors. In the developing nervous system bHLH proteins regulate neurogenesis, and migration of postmitotic neurons: Initially, cycling progenitor cells are committed to a neuronal fate through activation of Notch signalling [1], while later proper localization of neurons is established [2]. In the brain Tcf4 interacts with the class II bHLH transcription factors Math1, HASH1, and neuroD2 [3]. Binding of the calcium sensor calmodulin to Tcf4 inhibits transcriptional activation through interaction with the DNA binding domain [3].

Mutations or deletions of the human Tcf4 gene cause Pitt-Hopkins syndrome, a rare developmental disorder which is characterized by severe intellectual disability, developmental delay and autistic behavior [4]. In mice, constitutive Tcf4 knockout disrupts normal brain development [2]. In contrast, a gain of function in Tcf4 is associated with a higher risk in developing schizophrenia [3, 5].

A number of signaling cascades involved in brain development have been shown to play important roles in synaptic function in adults. For instance, several proteins involved WNT signaling, which plays important roles in cortical development [6–8], were shown to modulate synaptic function in adults [9–12]. Similarly, proteins involved in CREB signaling, which play a role in cerebellar development [13] affect synaptic transmission [14–16]. All three signaling cascades (TCF4, CREB, WNT) have in common that they have been identified to mediate the risk for schizophrenia in humans [3, 5, 17–19], thus highlighting a possible role in the maintenance of synaptic function independently of their developmental functions. However, the synaptic function of Tcf4 in the adult brain has not yet been examined.

In order to study the effects of Tcf4 on the regulation of adult synapses, we generated mice with inducible knockout of Tcf4 in a sparse subset of fluorescently labelled neurons. These permit visualization of dendritic spines to determine alterations in their number or shape [12, 20]. Functionally, dendritic spines are the morphological correlates of excitatory postsynapses, where specialized synaptic proteins, such as scaffolding proteins and ion channels are clustered [21, 22]. The structural shape of spines typically reflects specific functional properties. For instance, increases in spine head size help accommodate larger numbers of neurotransmitter receptors, while shortening and widening of spine necks decreases the electrical resistance of the spine neck, thereby leading to larger excitatory postsynaptic potentials [23]. Thus, at dendritic spines neuronal information is received and integrated and the numbers and shapes of spines reflect biological function [24].

Dendritic spines are typically classified based on their morphology into three groups [25–27]: Mushroom spines possess a large head and thin, clearly discernible neck. Stubby spines similarly possess a large head but lack a discernible neck. Thin spines are long and slender and possess a smaller head than the previous two classes. These different shapes are thought to cause functional differences. For example, mushroom and stubby spines with their large heads, which provide a larger area for neurotransmitter receptors, are thought to be more stable than thin spines, which are thought to be more plastic [25–27].

In order to study dendritic spines in Tcf4 deficient mice, we crossed mice co-expressing tamoxifen-inducible cre recombinase with YFP in a sparse subset of neurons under the Thy1 promoter (Slick-V line) [28] with Tcf4^fl/fl mice (Fig 1A) [29], enabling us to visualize dendritic spines of cortical and hippocampal pyramidal cells (Fig 1B). Tcf4 knockout was induced by tamoxifen administration to 10–11 weeks old animals for 5 days, resulting in the deletion of a 4.3 kb fragment of the Tcf4 gene containing the bHLH exon and 3’ exons, which are responsible for DNA binding and dimerization [29].

First, we quantified the number of dendritic spines of basal dendrites in YFP-labelled layer V cortical neurons (Fig 1C): Dendrites of tamoxifen-treated wildtype controls showed average spine densities of 1.01±0.02/μm (n = 6 mice). Loss of a single allele of Tcf4 (Tcf4^fl/+) caused a significant reduction in spine density to 0.86±0.03/μm (n = 6 mice; p < 0.01, ANOVA with Tukey’s multiple comparison test). Homozygous loss of Tcf4 (Tcf4^fl/fl) caused a similar reduction in dendritic spines, to 0.85±0.03/μm (n = 6 mice; p < 0.01 vs. control, ANOVA with Tukey’s multiple comparison test; Fig 1D). We found similar results in CA1 hippocampal neurons: In stratum oriens dendrites wildtype controls showed average spine densities of 1.77±0.07/μm (n = 6 mice), which were reduced to 1.45±0.06/μm in Tcf4 heterozygous animals (n = 6 mice; p < 0.01, ANOVA with Tukey’s multiple comparison test) and to 1.34±0.06/μm in homozygous Tcf4 knockout (n = 6 mice; p < 0.001 vs. control, ANOVA with Tukey’s multiple comparison test; Fig 1D). In stratum radiatum dendritic spine density was reduced from 1.83±0.07/μm in wildtype (n = 6 mice) to 1.6±0.06/μm (n = 6 mice) in heterozygous and to 1.56±0.07/μm in homozygous Tcf4 knockout (n = 6; p < 0.05 vs. control, ANOVA with Tukey’s multiple comparison test; Fig 1D).

Since spine shape is intimately linked to function, we classed spines into three major morphological groups, mushroom, stubby and thin [30] (Fig 2A). In cortical layer V pyramidal cells, spine loss was carried mostly by a reduction in thin and stubby spines, while mushroom spines were affected to a lesser degree (Fig 2B). In hippocampal CA1 neurons, both on apical and basal dendrites, the overall reduction in spine was mostly caused by reductions in mushroom spines, which were the most abundant group in CA1 neurons (Fig 2B).

The number and shapes of dendritic spines are intimately linked to synaptic function. For instance, learning tasks in experimental animals lead to increased turnover and formation rates of dendritic spines. For instance, spines get formed and stabilized during motor learning [32] and repetitive motor learning caused newly formed spines to appear in clusters [33]. Neurodegenerative diseases, on the other hand are typically characterized by loss of dendritic spines [34], reflecting cognitive impairment and memory loss. Interestingly, however, reduced turnover of dendritic spines also causes cognitive impairment. This is the case in fragile X syndrome, where dendritic spines numbers are increased [35]. Furthermore, alterations in dendritic spines also occur in psychiatric diseases such as schizophrenia or autism-spectrum disorders, where alterations in spine density or turnover in the prefrontal cortex are thought to underlie cognitive and behavioral symptoms [19].

In the present study, we inquired whether Tcf4 plays a role in the maintenance of dendritic spines in adult animals, independently of its developmental functions. To achieve this, we generated a mouse line with inducible knockout of Tcf4 in a sparse subset of fluorescently labelled neurons. Our results suggest that heterozygous loss of Tcf4 suffices to substantially alter both spine density and morphology. These alterations may be the consequence of a disrupted negative feedback loop: Under normal conditions, elevated intracellular calcium, which may be the result of increased neuronal activity, inhibits the transcriptional activity of Tcf4 [36]. This may in turn dampen neuronal activity by removing or reshaping dendritic spines, in particular mushroom spines which, because of their large head surface, can generate stronger postsynaptic signals [23, 26]. Knockout of Tcf4 may thus precipitate these dampening effects even in the absence of increased neuronal activity.

The present results favor a postsynaptic, or cell-autonomous, role of Tcf4: In Slick-V mice, only a small fraction of neurons, both in the hippocampus and cerebral cortex, express cre (cf. Fig 1B). Therefore, the vast majority of presynaptic inputs into any given neuron are expected to be made up from processes of neurons not expressing cre recombinase, thus always expressing wildtype Tcf4. The genotype of the fluorescent neurons, in contrast, matches the induced alteration, i.e. wildtype, heterozygous loss of Tcf4 after tamoxifen administration (Tcf4^fl/+), or homozygous loss (Tcf4^fl/fl), with spine densities clearly correlating with these genotypes.

These results suggest that Tcf4 plays an important role in the control of synaptic plasticity, in adult animals, independent of the developmental function of Tcf4. Mouse models of Pitt-Hopkins syndrome, lacking functional Tcf4, typically show hyperactivity, reduced anxiety, and deficient spatial and associative learning [37, 38]. A reduced number of cortical synapses in humans was shown to correlate with decreased cognitive performance in the context of Alzheimer’s disease [39, 40]. Similarly, in various animal models, spine numbers and turnover have been shown to correlate with learning and memory and pathological alterations typically lead to decreased cognitive performance [24, 41]. Thus, based on our findings, we hypothesize that constitutive lack of functional Tcf4 causes cognitive and memory defects even in the absence of its neurodevelopmental effects. Thus, the clinical effects of loss of Tcf4 in Pitt-Hopkins syndrome may, at least in part, also be mediated by the synaptic function of Tcf4.