Transcription Factor 4
Transcription factor 4 (TCF4) is a member of the E protein family of transcription factors that play a critical role in development. As a family, E proteins were named because they bind the Ephrussi-box (E-box) sequence (5′-CANNTG-3′) and were first investigated as regulators in B-cell development, but TCF4 specifically has been more closely tied to disorders of the central nervous system. Despite its medical relevance, mutations in the Tcf4 gene have been linked to intellectual disability, schizophrenia, obesity, and corneal dystrophy; little was known about the biochemical and neurological roles of TCF4 until quite recently. It is now known that TCF4 responds to calcium signaling, alters epigenetic modifications necessary for learning and memory, and regulates the transcription of genes that modulate plasticity in hippocampal and cortical neurons. Because TCF4 was initially investigated across multiple fields simultaneously, protein and gene synonyms are commonly used. TCF4 protein is often referred to as immunoglobulin transcription factor 2 (ITF2) and E2-2, dating back to the early work in B-cell development and its relationship to E2A (also known as TCF3). The official gene name for transcription factor 4 is now Tcf4, and it is important to distinguish between it and the more well-understood T-cell factor 4 (Tcf7l2) that for historical reasons is still often abbreviated TCF-4 and has resulted in a regrettable degree of confusion throughout the literature. For our purposes, TCF4 will refer to transcription factor 4, a protein critical for proper brain development and function.
Role in Disease
TCF4 has been implicated in neuropsychiatric diseases and disorders as well as in other physiological aberrations, which is consistent with its near ubiquitous expression during development in the central nervous system (Sweatt 2013). Of note, the major diseases and disorders that TCF4 is known to have a role are intellectual disabilities, schizophrenia, and Fuchs corneal dystrophy (Forrest et al. 2014; Kennedy and Sweatt 2016; Navarrete et al. 2013).
Mutations, deletions, or microduplications of Tcf4 that result in haploinsufficiency have been associated with Pitt-Hopkins syndrome (PTHS), a rare autism-spectrum disorder that presents with severe intellectual disability, little to no language development, and additional physiological abnormalities such as intermittent hyperventilation, facial dysmorphia, microcephaly, gastrointestinal disturbances, developmental delay, impaired motor skills, dysmorphia of the hands and feet, seizures, and ophthalmological abnormalities (Sweatt 2013; Whalen et al. 2012). While the intellectual disability and lack of language development in PTHS are most notable in regard to symptomology, gastrointestinal problems manifesting in the form of constipation and gastroesophageal reflux are a common cause of discomfort in patients with PTHS (Marangi et al. 2011; Sweatt 2013). Importantly, available animal models of PTHS (Tcf4 +/− mice) exhibit behaviors and characteristics that are consistent with the PTHS symptomology, including reduced ultrasonic vocalizations (rodent analogue of human language), reduced gastrointestinal transits, and impaired associative and spatial learning and memory (Grubisic et al. 2015; Kennedy et al. 2016), underscoring the role of Tcf4 in the disease etiology and affording a viable mechanism to study PTHS in vivo.
While Tcf4 haploinsufficiency is the underlying cause of PTHS, clinical studies have elucidated the diversity of mutations that occur in PTHS. Whalen and colleagues reported that while the majority of PTHS cases were due to point mutations, there were also small deletions/insertions and partial or whole gene deletions – most of which were de novo (Whalen et al. 2012). Further examination into the insertions/deletions and point mutations revealed that most of them resulted in premature stop codons, but there was also evidence of splice mutations and missense mutations, which occurred mostly in the basic helix-loop-helix (bHLH) domain of TCF4, which facilitates TCF4 dimerization and subsequent DNA binding. While most PTHS-related Tcf4 mutations do indeed occur de novo, there have been reports of PTHS in which the patients’ parents possessed a somatic mosaicism of a Tcf4 mutation but expressed an otherwise typical phenotype (Steinbusch et al. 2013), suggesting a possibility of PTHS heritability in certain cases.
It is also important to note that Tcf4 has been implicated in non-syndromic mild forms of intellectual disability (Forrest et al. 2014). Previous studies have reported that patients presenting with mild intellectual disability phenotypically distinct from PTHS possess mutations involving exons that do not involve the bHLH domain – particularly within exonic regions upstream of exon 18 (Kharbanda et al. 2016). This study by the Timmusk group reported that their patients with mild intellectual disability also possessed fewer long Tcf4 transcripts compared to controls, and they hypothesize that this may partially explain the intellectual disability phenotype. This may be explained by that compared to shorter transcripts, longer transcripts are more affected by upstream mutations, since the protein will be encoded from parts of the transcriptome that possess those variances. Additionally, considering their study within the context of the mild intellectual disability literature, they also speculate that the number of intact transcripts may relate to the severity of the phenotype in mild intellectual disability. Future studies should investigate directly how mutations resulting in alternative splicing may relate to the phenotype. It is also important to note that different TCF4 isoforms activate transcription differently (Sepp et al. 2011). Efforts should be made to include some of the different TCF4 isoforms in the context of PTHS animal models to determine if physiological, behavioral, and molecular biological differences depend on the specific mutated transcript.
Variants of Tcf4 have also been associated with an increased risk for schizophrenia, a neuropsychiatric disorder that presents with psychotic symptoms and deficits in cognitive domains (Quednow et al. 2014). Along with other genes, mutations within intronic regions of Tcf4 have reached genome-wide significance for association with schizophrenia (Steinberg et al. 2011). In several studies, many different single-nucleotide polymorphisms (SNPs) on Tcf4 introns have been identified as a risk for a schizophrenic phenotype (Li et al. 2016; Stefansson et al. 2009).
Given the cognitive impairments with which schizophrenia presents, associations between several SNPs in Tcf4 and cognition have been studied in both clinical populations and animal models of schizophrenia (Quednow et al. 2014). Patients with first-episode psychosis and the T allele of the Tcf4 SNP rs9960767, one of the first variants to reach genome-wide significance for association with schizophrenia, exhibit significantly poorer performance with reasoning and problem solving, even when controlling for IQ (Albanna et al. 2014). Patients with schizophrenia carrying the C allele on the same SNP performed significantly better on recognition tasks (Lennertz et al. 2011). Additionally, the Tcf4 SNP rs2958182 within a schizophrenic cohort found that patients carrying the A allele exhibited poorer language skills, whereas healthy individuals carrying the A allele performed more poorly on delayed memory tasks (Hui et al. 2015), strengthening the relationship between Tcf4 genotype, language, and cognition.
Sensorimotor gating, a measure of one’s attentional filter which is known to be impaired in schizophrenia (Quednow et al. 2014), has also been shown to be influenced by certain allelic variants of Tcf4 SNPs. For example, the C allele of the SNP rs9960767 was shown to be associated with reduced pre-pulse inhibition (PPI) – a means to evaluate sensorimotor gating (Quednow et al. 2011). Similarly, Tcf4 SNPs rs9960767, rs10401120, rs17597926, and 17512836 were associated with decreased P50 suppression, a measure of sensory gating, which is another metric of attentional filter (Quednow et al. 2012). These findings indicate that certain allelic variances of Tcf4 may mediate symptom severity. Importantly, animal models of schizophrenia – namely, Tcf4 transgenic mice – exhibit deficits in cognitive domains such as trace fear memory, as well as impairments in sensorimotor gating (Brzozka et al. 2010; Brzozka and Rossner 2013), supporting the role of Tcf4 variants in schizophrenia and the viability of Tcf4 mouse models of the disorder.
Fuchs Corneal Dystrophy
Outside of the neuropsychiatric realm, mutations within Tcf4 have also been associated with Fuchs corneal dystrophy (FCD), a common disease of the corneal epithelium for which the main treatment is corneal transplantation (Navarrete et al. 2013). Several SNPs at the Tcf4 locus are associated with FCD, with certain allelic variants conferring different levels of risk (Baratz et al. 2010).
Regulating Gene Expression
The bHLH domain of TCF4 mediates two important biophysical functions, namely, the dimerization of TCF4, either with itself or another E protein, and the localization of the dimer to an E-box along the genome. Comprised of basic residues that chelate the phosphate backbone of DNA and other residues that mediate the dimerization necessary to straddle the major groove of DNA and recognize the E-box, mutations in the bHLH domain of TCF4 are associated with loss of function and PTHS. The homodimerization and heterodimerization of TCF4 with other E proteins complicate understanding its regulatory function as a transcription factor. For example, it is entirely possible that some dimers activate while others inhibit transcription of the same gene targets. Adding an additional layer of complexity, TCF4 can bind and localize epigenetic modifiers to its genome targets, providing a secondary mechanism for long-term changes in gene transcription.
Interactions with Other E Proteins
To further complicate matters, TCF4 has multiple transcripts that are generally categorized into long and truncated transcripts referred to as canonical and alternative, respectively. Other than the act of being inhibited by ID proteins, there are no known cytosolic targets of TCF4 signaling, yet the more highly expressed alternative transcripts lack the nuclear localization sequence (NLS) present in the canonical transcripts (Sepp et al. 2011). This means either a large percentage of TCF4 transcripts are nonfunctional, serve solely to saturate ID protein inhibitors, or are brought into the nucleus to affect transcription in the form of a dimer with a canonical copy or some other NLS-containing E protein. Needless to say, the kinetic considerations of TCF4 homo- and heterodimerization and subsequent transcriptional regulation are complicated to model.
TCF4 does not simply regulate the activation or repression of gene transcription; it also binds the histone acetyltransferases CBP and p300. Histone acetylation is an epigenetic mark associated with transcriptional activation, and histone acetylation levels in neuronal tissue can be manipulated to affect synaptic plasticity (Gräff and Tsai 2013). One downstream consequence of reduced TCF4 activity, such as in the case of PTHS (Tcf4 +/−), might be a reduction of histone acetylation levels at genes that regulate neuronal development and plasticity. Interestingly, inhibitors of histone deacetylases (HDACs), enzymes that remove acetyl marks from histones, and the selective knockdown of HDAC2 were both capable of rescuing normal learning and memory in Tcf4 +/− mice (Kennedy et al. 2016). This suggests that the regulation of the epigenome in the hippocampus, where it is highly expressed, is a major function of TCF4 (Fig. 1).
Tcf4 haploinsufficiency also affects DNA methylation, specifically at genes associated with memory formation in the hippocampus (Kennedy et al. 2016). Reduced Tcf4 expression induces the hypomethylation of genes that have altered transcription shortly after experiential learning. It is not clear whether this activity is mediated by TCF4 directly by binding a modifier of DNA methylation or indirectly by affecting the expression of DNA methylation modifiers. For example, TET2, an enzyme that catalyzes the oxidative conversion of DNA cytosine methylation to hydroxymethylation, was also dysregulated in Tcf4 +/− hippocampal tissue. It was also determined that DNA methylation states affect TCF4 function. The methylation of E-boxes with (5′-CACGTG-3′) and (5′-CATGTG-3′) sequences inhibits TCF4 binding, while hydroxymethylation resulted in an enhanced affinity of TCF4 for the E-box (Khund-Sayeed et al. 2016).
Response to Calcium Signaling
One of the most interesting aspects of TCF4 function is that calcium signaling directly regulates its activity (Fig. 1). Through a series of impressive protein NMR and protein dynamics simulations, it was determined that calmodulin (CaM), a calcium-binding protein that is critical in numerous cellular processes including learning and memory, will dimerize and chelate a TCF4–TCF4 homodimer (Larsson et al. 2001, 2005). This 2:2 stoichiometric binding of CaM to the TCF4 in the presence of calcium serves to prevent TCF4 from localizing to its genomic targets. Therefore, an influx of calcium binds CaM and temporally causes a reduction of active TCF4 cellular concentrations. In the case of a neuron, this would suggest that TCF4 should negatively regulate the gene targets of calcium signaling, and neurons with reduced levels of TCF4 should have gene expression levels similar to that of neurons recently activated. Indeed, Tcf4 (+/−) hippocampal tissue has gene expression profiles that more closely align with wild-type mice that have recently undergone experiential learning, and Tcf4 (+/−) hippocampal neurons elicit increased plasticity after theta burst stimulation (Kennedy et al. 2016).
Given the expressional patterns of TCF4 within the central nervous system as well as its important role in learning, memory, and cognition more generally, it is unsurprising that TCF4 modulates aspects of neurophysiology which together underpin these phenomena. However, this is indeed an understudied facet of TCF4 function, with only two published studies examining neuronal electrophysiology in the context of TCF4 (Kennedy et al. 2016; Rannals et al. 2016), which investigated the roles of TCF4 in neuronal intrinsic excitability and synaptic plasticity.
To characterize the neurodevelopmental role of TCF4 in neuronal intrinsic excitability, Rannals et al. (2016) knocked down TCF4 expression in rat prefrontal cortex by employing in utero electroporation just prior to neurogenesis, given that TCF4 is important during brain development. In a series of experiments during which whole-cell recordings from prefrontal cortical neurons were taken, TCF4 knockdown resulted in reduced action potential (AP) output and firing frequency (including maximum frequency), which was also accompanied by an increase in the resting membrane potential (RMP) and in peak amplitude of the AP. Furthermore, there was also increased variability in AP amplitude in neurons with TCF4 knockdown.
Further examination into the neurons’ afterhyperpolarization (involved in regulating AP firing, AHP) revealed that cells with TCF4 knockdown exhibited increased medium AHP (mAHP) and slow AHP (sAHP), which are known to be regulated by various calcium-activated potassium channels. Furthermore, TCF4 knockdown resulted in increased capacitance-normalized charge transfer during voltage clamping (to isolate potassium channel currents), compared to controls.
Examination of expressional patterns of select genes revealed an upregulation of Kcqn1 and Scn10a, which encode specific potassium and sodium channels, respectively. Importantly, these two genes have been previously implicated in the regulation of neuronal excitability and possess bHLH-binding domains, indicating possible direct regulation by TCF4. Further experimentation during which the investigators overexpressed Scn10a and pharmacologically blocked the specific potassium and sodium channels in normal neurons resulted in rescued AP firing frequency during blocking as well as rescuing and mimicking of other excitability deficits (i.e., normalization of AHP with potassium blocking), depolarized RMP/decreased spiking frequency, with the overexpression of Scn10a. Furthermore, knockout of Kcqn1 and Scn10a in normal neurons resulted in a normalization of the AP output. Taken together, these data suggest that TCF4 regulates AP spiking frequency in prefrontal cortical neurons through negatively regulating the expression of Kcqn1 and Scn10a and, therefore, the amount of potassium and sodium channels expressed at the membrane, which affect the AHP and the RMP, respectively.
Investigation of intrinsic excitability in a mouse model of PTHS (Tcf4 +/−) revealed that similarly to the rat prefrontal cortical neurons, brain slices exhibited reduced AP output and maximum firing frequency as well as a depolarized RMP. In contrast to the rat prefrontal neurons, there were no significant differences in the AHP in the Tcf4 (+/−) mice brains. Gene expressional analysis also indicated an increase and decrease in Scn10a and Kcnq1 expression, respectively. Application of channel antagonists revealed only a rescue of AP output for sodium channel blocking. These experiments, using a more realistic model of PTHS, provide a novel characterization of the neurophysiological differences associated with the etiology of this specific disease, distinct from the TCF4 knockdown in rat prefrontal cortex.
The formation of long-term memories is believed to have a synaptic correlate, which pertains to the reorganization of neuronal circuits following learning – formally known as synaptic plasticity. One route of synaptic reorganization is known as long-term potentiation (LTP), which refers to the strengthening of synaptic connections between communicating neurons. Given the emerging role of TCF4 in cognition and learning/memory, the role of TCF4 in LTP as well as its potential regulatory functions on plasticity-regulated genes has been investigated (Kennedy et al. 2016).
While LTP induction is believed to relate directly with the formation of long-term memories (Sweatt 2016), the Tcf4 (+/−) mice were found to have enhanced LTP in Schaffer collaterals between hippocampal areas CA3 and CA1, despite deficits in spatial and associative learning and memory. Investigation into differentially expressed genes between the Tcf4 (+/−) mice and wild-type controls revealed that Tcf4 (+/−) mice exhibited significant dysregulation of genes involved in synaptic plasticity, including the upregulation of klotho, which has previously been characterized as both a cognitive and LTP enhancer in klotho-overexpressing mice (Dubal et al. 2014; Kennedy et al. 2016). Conversely, Arc, another gene involved in synaptic plasticity, was significantly downregulated in Tcf4 (+/−) mice compared to controls. It was concluded that Arc downregulation may be responsible for the enhanced LTP, given its role in negatively regulating the movement of AMPA receptors in the synapse (Kennedy et al. 2016; Shepherd et al. 2006), a critical component of the glutamatergic role in LTP induction.
TCF4 is critically important for brain development and function, as highlighted by its associations with a rare autism-spectrum disorder, intellectual disability, and schizophrenia. This is likely caused by the roles TCF4 plays in regulating the expression of plasticity-related genes during neuronal development and in modulating plasticity in the adult brain. Biochemically, TCF4 functions as both a transcription factor that can activate or repress transcription depending on its bHLH-binding partner and by altering the local epigenetic environment by attracting epigenetic modifiers to its gene targets. TCF4 activity is closely regulated by a family of inhibitor bHLH-containing ID proteins and by CaM and calcium that sequester and reduce the effective cellular concentrations of TCF4. Together, these two regulatory mechanisms provide a genetic and an extracellular switch to regulate TCF4 activity and neuronal plasticity.
The authors' work is supported by the Pitt-Hopkins Research Foundation and an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under grant number P20GM0103423.
- Kennedy AJ, Sweatt JD. Drugging the methylome: DNA methylation and memory. Crit Rev Biochem Mol Biol. 2016: 1–10.Google Scholar
- Larsson G, Schleucher J, Onions J, Hermann S, Grundström T, Wijmenga SS. Backbone dynamics of a symmetric calmodulin dimer in complex with the calmodulin-binding domain of the basic-helix-loop-helix transcription factor SEF2-1/E2-2: a highly dynamic complex. Biophys J. 2005;89(2):1214–26.PubMedPubMedCentralCrossRefGoogle Scholar
- Li J, Chen Z, Wang F, Ouyang Y, Zhang N, Yang M, et al. Polymorphisms of the TCF4 gene are associated with the risk of schizophrenia in the Han Chinese. Am J Med Genet B Neuropsychiatr Genet. 2016;171(8):1006–1012.Google Scholar
- Navarrete K, Pedroso I, De Jong S, Stefansson H, Steinberg S, Stefansson K, et al. TCF4 (e2-2; ITF2): a schizophrenia-associated gene with pleiotropic effects on human disease. Am J Med Genet B Neuropsychiatr Genet: the official publication of the Int Soc Psychiatr Genet. 2013;162b(1):1–16.CrossRefGoogle Scholar
- Sweatt JD. Neural plasticity & behavior – sixty years of conceptual advances. J Neurochem. 2016;139(Suppl 2):179–199.Google Scholar