Isoform switch of T-cell factor7L2 during mouse heart development

Abstract

Canonical WNT signaling plays critical, often opposing roles in heart development and disease, but its context-dependent mechanisms remain unclear. We hypothesized that alternative splicing of Tcf7l2, a key nuclear partner of β-catenin, contributes to WNT signaling specificity in the heart. To investigate this, we cloned and sequenced 53 Tcf7l2 transcripts in ventricular tissues from embryonic day 17.5 (E17.5, 24/53) and postnatal day 8 (P8, 29/53) mice, identifying 32 distinct isoforms. Among 18 potential exons, exons 6 and 17 were absent, and over 80 % of transcripts lacked exon 4. Alternative splicing was prominent in the C-terminal exons (14, 15, and 16), with exon 14 inclusion significantly higher in P8 hearts (64.3 %) than E17.5 hearts (34.8 %). Variations in exon 15 and 16 combinations, along with reading frame shifts caused by the adenine insertion and deletion (indel) near the beginning of exon 18, affected C-terminal structures, altering the presence of the E-tail, C-clamp, and CtBP-binding motifs. Notably, exon 14 insertion introduced a redox-switch domain spanning the NLS and C-clamp regions in E and S isoforms, while adenine indels altered isoform lengths, driving transitions between E, S, and M isoforms. RT-PCR validation across multiple developmental stages confirmed these splicing patterns. Our findings suggest that a postnatal redox-sensitive isoform switch in Tcf7l2 modulates WNT signaling, potentially influencing cardiomyocyte maturation during the transition from proliferation to hypertrophy.

Graphical abstract

Postnatal Redox-Sensitive Tcf7l2 Isoform Switch during Mouse Heart Development. During the transition from embryonic to postnatal heart development, increased reactive oxygen species (ROS) levels arise due to exposure to higher oxygen concentrations, an immature antioxidant defense system, and a metabolic shift in cardiomyocytes. At embryonic day 17.5 (E17.5), the predominant Tcf7l2 isoforms lack exon 14, making them redox-insensitive. By postnatal day 8 (P8), exon 14 inclusion increases, introducing a redox-sensitive switch domain spanning the nuclear localization signal (NLS) and C-clamp regions in E and S isoforms or following the NLS in M isoforms. Additionally, adenine insertion-deletion (indel) events in exon 18 alter isoform length and redox sensitivity, leading to the loss of redox-responsive regions at E17.5 and the acquisition of redox sensitivity at P8. These splicing variations facilitate embryonic-to-postnatal transitions in Tcf7l2 isoforms, generating redox-active variants that may modulate canonical WNT signaling during cardiomyocyte maturation from cardiomyocyte proliferation to hypertrophy.

Highlights

• •Alternative splicing diversifies Tcf7l2 isoforms during mouse heart development.
• •Exon 14 inclusion increases postnatally, introducing a redox-switch domain.
• •Adenine indels in exon 18 alter the length and function of Tcf7l2 isoforms.

1. Introduction

Canonical WNT signaling plays a pivotal role in normal cardiac development and is critical for maintaining adult heart homeostasis [[1], [2], [3], [4]]. A key molecule in this pathway, β-catenin, is a multifunctional protein that mediates cell-cell adhesion through its interaction with cadherin family transmembrane proteins and regulates gene expression through its association with T-cell factors (TCF)/lymphoid enhancing factor (LEF) family transcription factors [1].

In vertebrates, the TCF/LEF family includes four members, all of which possess an evolutionarily conserved high-mobility group (HMG) DNA-binding domains (DBD). While these proteins share conserved functional domains, they also contain unique functional regions that mediate context-specific functions (Fig. 1). For instance, the first 69 N-terminal amino acids show ~60 % sequence homology and encode a high-affinity β-catenin binding domain (BCBD), which interacts with the armadillo repeat domain of β-catenin [5,6]. The DBD and the subsequent nuclear localization signal (NLS) also show high sequence conservation [7]. However, outside of these conversed regions (BCBD, DBD and NLS), the sequence homology drops to 15-20 %. Emerging evidences suggest that these divergent regions contribute significantly to the context-dependent gene regulation.

Fig. 1.

Potential exons, splicing variants, and functional domains of Tcf7l2. A: Tcf7l2 consists of at least 18 potential exons, multiple alternative promoters, and several splicing variants. Exon 1 encodes a high-affinity β-catenin binding domain (BCBD, dark green), while alternative promoters in intron 6 give rise to C isoforms lacking this domain. The high-mobility group (HMG) box, encoded by exons 11 and 12, is responsible for DNA binding. The nuclear localization signal (NLS) is primarily encoded by exon 13. Between the BCBD and HMG box lies the variable context-dependent regulatory domian (CRD), which facilitates low-affinity β-catenin binding and interactions with Groucho/TLE co-repressors and other transcriptional regulators. Exons 4 and 6 are alternative. Additionally, an alternative splicing donor site in exon 8 and an alternative splicing acceptor site in exon 10 generate isoforms where exon 9 is flanked by extra 12- and/or 15-nucleotide sequences, encoding the LVPQ and SXXSS motifs, respectively. Diverse C-terminal variants arise from different combinations of alternative exons 14, 15, 16, and 17, producing diverse C-termini (E-tail) with complete, incomplete, or absent C-clamp domains (encoded by exons 15, 16, and part of exon 18), and with or without CtBP-binding motifs, depending on the usage of stop codons. B: Eight distinct C-terminal Tcf7l2 isoforms frequently expressed in mouse hearts. E1–E4 isoforms are produced when exon 15 or 16 is individually spliced to exon 18, allowing full translation to the final stop codon and inclusion of the complete C-clamp domain and CtBP-binding motifs. M1–M2 isoforms result from the exclusion of both exons 15 and 16, leading to activation of a proximal stop codon in exon 18; these isoforms lack the C-clamp. S1–S2 isoforms are generated when both exons 15 and 16 are included, triggering a stop codon in exon 16 and producing a truncated C-clamp (C-clamp_ΔC). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

TCF/LEF isoform diversity is further expanded through alternative splicing and selective promoter usage [8]. This isoform diversity enables TCF/LEFs to mediate tissue- and developmental stage-specific WNT responses during embryogenesis, stem cell differentiation, and organ development. Among all TCF/LEF family members, Tcf7l2 exhibits the highest isoform diversity [8]. Notably, it is the only TCF/LEF gene with relatively abundant mRNA level in the mouse ventricle at later embryonic stages, and its protein is detectable in cardiomyocyte (CM) nuclei [3,9].

In both human and mouse, Tcf7l2 gene comprises 18 potential exons (Fig. 1A), several of which are subject to alternative splicing (exons 4, 6, 9, and 14–17). Additionally, alternative splice donor and acceptor sites at exons 8 and 10 allow exon 9 to be flanked by 12- and/or 15- nucleotide extensions, encoding LVPQ and SXXSS motifs, respectively [10]. Moreover, alternative splicing of exons 14, 15, 16, and 17 gives rise to Tcf7l2 isoforms with highly divergent C-terminal regions, which are categorized into three major isoform types: E, M, and S in mice[8,11]. Combinations of exons 14, 15, 16, and 18 generate eight distinct C-terminal isoforms [8,12]. When either exon 15 or 16 is individually spliced with exon 18, a distal stop codon is utilized, producing four E isoforms that contain a full C-clamp and two CtBP-binding motifs: E1 (13–15-18), E2 (13–14–15-18), E3 (13–16-18), and E4 (13–14–16-18). In contrast, when both exons 15 and 16 are excluded, exon 18 is only partially translated due to the activation of a proximal stop codon, resulting in two M isoforms: M1 (13–18) and M2 (13–14-18). These M isoforms lack both the C-clamp and CtBP motifs. When exons 15 and 16 are both included, the stop codon in exon 16 is activated, leading to the production of two S isoforms: S1 (13–15–16-18) and S2 (13–14–15-16-18). These S isoforms contain a truncated C-clamp (C-clamp_ΔC), which lacks the final conserved cysteine encoded by exon 18, and do not include the CtBP motifs (Fig. 1B).

CMs shift from proliferation to hypertrophy shortly after birth [13], but the molecular mechanisms remain unclear. The dynamic splicing of Tcf7l2 may help specify canonical WNT activity during this transition. Understanding its isoform regulation is key to uncovering how WNT signaling influences cardiomyocyte growth and differentiation.

2. Materials and methods

2.1. Animals

Ventricular tissues, dissected free of atria and connective tissues, were collected from embryonic day 13.5 (E13.5), E17.5, postnatal day 1 (P1), P4, P8, P12, P21, and 3-months-old (3 M) C57BL/6 mice. Animals were housed under specific pathogen-free standards with free access to food and water. All animal experiments were performed in accordance with the Ethical and Safety Guidelines for Animal Experiments and approved by the Animal Use Committee of the University of Minnesota Medical Center and the University of Texas Health Science Center at San Antonio.

2.2. RNA isolation, reverse transcription polymerase chain reaction (RT–PCR), and cloning of Tcf7l2 RT–PCR products

Total RNA was extracted from mouse ventricles using TRIzol Reagent (Invitrogen, Waltham, MA). For Tcf7l2 cloning, RNA was isolated from four pooled E17.5 hearts and two pooled P8 hearts. For RT-PCR, RNA was extracted from either two pooled embryonic hearts or from individual hearts each time, depending on the developmental stage. The RNA was treated with DNase I, and 0.5 μg of total RNA was reverse transcribed using Superscript II reverse transcriptase (Invitrogen, Waltham, MA), following the manufacturer's protocol. To evaluate alternative splicing across the full length of Tcf7l2, we focused on its 18 potential exons, noting that exons 1 and 18 are commonly included in full-length transcripts. Tcf7l2 transcripts for cloning were amplified from E17.5 and P8 hearts using primers targeting the end of exon 1 and the beginning of exon 18. PCR products were resolved on 2.5 % agarose gels, and DNA fragments were purified using the DNA and Gel Band Purification Kit (Promega, Madison, MI). Purified fragments were cloned into the pCR2.1-TOPO vector using the TOPO TA Cloning Kit for Sequencing (Invitrogen, Waltham, MA). To validate specific splicing variants, specific primer pairs targeting alternative exons, exon donor sites, and acceptor sites (Suppl Table S1) were designed and used to verify Tcf7l2 splicing variants in the heart of E13.5, E17.5, P1, P4, P8, P12, P21 and adult 3 M mice.

2.3. Sequencing Tcf7l2 transcripts and data analysis

Sequence analyses were conducted using the BigDye cycle sequencing Kit (Applied Biosystems, Waltham, MA). In total, 56 individual clones were sequenced, including 24 from E17.5 and 32 from P8. Sequences were analyzed using BioEdit software, and exon compositions of each clone were compared and aligned to NCBI RefSeq RNA database to identify alternative splicing events and isoform diversity.

2.4. Protein sequence analysis and identification of redox-sensitive residues

The primary amino acid sequences of different Tcf7l2 isoforms were generated using CLC Sequence Viewer (version 8.0) and searched in the AlphaFold Protein Structure Database. Identical sequences were visualized using UniProt, while sequences not found in the database were further analyzed using InterPro, the Eukaryotic Linear Motif (ELM) resource, and AIUPred [14,15]. InterPro and ELM were used to predict domains and key functional sites in proteins, while AIUPred was employed to assess disordered protein-binding regions and potential redox-sensitive regions.

AIUPred prediction identifies disordered protein-binding regions—segments that are likely to lack stable tertiary structure but participate in protein–protein interactions. By default, it contains disorder predictions from IUPred2 (red line, AIUPred-disorder) and binding site predictions from ANCHOR2 (blue line, AIUPred-bing). Residues with ANCHOR2 scores ≥0.5 were considered disordered binding regions.

Cysteine residues are pivotal in protein structure and function due to their unique ability to form disulfide bonds under oxidizing conditions and remain in a reduced thiol state under reducing conditions. This reversible redox behavior significantly influences protein folding, stability, and activity. Under oxidizing conditions, the thiol groups (-SH) of two cysteine residues can oxidize to form a covalent disulfide bond (-S-S-). These disulfide bonds play a key role in stabilizing protein structures, with disruption strongly associated with loss of protein function and activity [[16], [17], [18]]. Key amino acids susceptible to redox modifications were specifically examined. Additionally, the ‘Redox state’ option in AIUPred was used to analyze redox potential-dependent disordered regions. This analysis incorporated AIUPred prediction profiles to calculate disorder scores under both reducing and oxidizing environments and generate ‘Redox plus’ and ‘Redox minus’ lines. The ‘Redox plus’ line corresponds to conditions where cysteine stabilization occurs through disulfide bond formation. Whereas the ‘Redox minus’ line models a state without cysteine stabilization. In most cases, no changes in the overall order-disorder tendency would be observed. If sequences that undergo disorder-to-order transition upon changes in redox conditions, a pronounced shift could be observed between the redox-plus and the redox-minus profiles. Redox sensitivity scores are defined as the absolute difference between disorder scores under these two conditions for each residue. The actual region is calculated based on a heuristic segmentation approach, and the area between the ‘Redox plus’ and ‘Redox minus’ lines is filled with orange color, indicating potential redox-sensitive structural changes [15,19].

2.5. Statistical analysis

Categorical data were presented as numbers and percentages in E17.5 and P8 hearts. The Chi-square (χ²) test or Fisher's exact test (for expected cell counts less than five) was used to compare categorical variables. A P value of <0.05 was considered statistically significant. All analyses were performed using SPSS 19.0 and Prism 9.0.

3. Results

The mouse Tcf7l2 gene contains at least 18 potential exons, several of which undergo alternative splicing with documented splicing donor and acceptor sites (Fig. 1). However, the specific exons included in Tcf7l2 transcripts within the heart remain unclear. Furthermore, it is unknown whether Tcf7l2 exon composition changes during cardiac development. CMs in E17.5 hearts are highly proliferative, but after P3, they exit the cell cycle and transition to hypertrophy growth [4,20,21]. Therefore, we analyzed the exon composition and combination of Tcf7l2 transcripts in RNA extracted from E17.5 and P8 ventricles to determine if different Tcf7l2 isoforms exist in proliferative and hypertrophic hearts. PCR was performed using a forward primer at the end of the first exon (Tcf7l2 1F) and a reverse primer near the beginning of the final exon 18 (Tcf7l2 18R). Products were cloned into a TA vector and sequenced. A total of 56 clones (24 from E17.5 and 32 from P8) were sequenced, of which 53 clones (24 from E17.5 and 29 from P8) contained Tcf7l2 sequences.

At P8, one clone lacked exon 3, while E17.5 contained one clone with an 8 base pairs (bp) deletion in exon 3. Both changes led to a reading frame shift with a premature stop codon; thus, these 2 clones were excluded from further analysis.

3.1. Alternative exon usage and splicing variation in the context-dependent regulatory domains (CRD)

The CRD is located between the high affinity BCBD and the HMG box and includes functional modules encoded by alternative exons and splicing variants (Fig. 1). Exon 6 was not detected in any of the clones. >80 % of clones from E17.5 and P8 hearts lacked exon 4 (Fig. 2B). Both E17.5 and P8 hearts contained one clone missing exon 9, which did not alter the reading frame. The LVPQ motif, generated by an alternative donor site in exon 8, was present in >80 % of clones from E17.5 and P8 hearts. In contrast, the SXXSS motif, produced by an alternative acceptor site in exon 10, exhibited inclusion rates of 34.8 % in E17.5 hearts and 17.9 % in P8 hearts (Fisher’s exact test, P = 0.2071). Notably, the SXXSS motif was not observed in S forms (Fig. 2 D and E).

Fig. 2.

Exon combinations in context dependent regulatory domains. A: RT-PCR confirms alternative splicing of exon 4, while exon 6 is absent. A forward primer at exon 3 and a reverse primer at exon 5 generate two bands: the upper band includes exon 4, while the lower band excludes it. A forward primer at exon 5 and a reverse primer at exon 7 yield a single band, consistent with the absence of exon 6. Three biological replicates were analyzed per time point, using single postnatal and pooled embryonic (two per replicate) hearts. B: Quantification of the inclusion rates for exon 4 and exon 6 in TCF7L2 transcripts identified by RT-PCR cloning. The frequency of exon 4 inclusion at E17.5 and P8 is below 20 %, whereas exon 6 is not detected. C: RT-PCR analysis of LVPQ and SXXSS motif inclusion. The inclusion of LVPQ (right panel; 7F/LVPQ) remains constant, whereas SXXSS (left upper panel; 7F/SXXSS) is more prevalent at E17.5 compared to P1 and P8. A longer gel run (left lower panel) distinguishes isoforms with (upper band) or without (lower band) LVPQ. Three biological replicates were analyzed per time point, using single postnatal and pooled embryonic (two per replicate) hearts. D and E: Quantification of LVPQ and SXXSS motif inclusion in Tcf7l2 transcripts in E17.5 and P8 hearts (D) and frequency of LVPQ and SXXSS in Tcf7l2 E, M and S isoforms (E) using RT-PCR cloning.

To confirm alternative exons and domain-specific variants in CRD region of mouse Tcf7l2 isoforms, RT-PCR was performed on embryonic, postnatal, and adult hearts using various combinations of exon and domain-specific primer pairs. The RT-PCR for exons 3 and 5 produced two bands with expected sizes: 217 bp with exon 4 and 148 bp without exon 4 (Fig. 2A, 3F/5R). Exons 5 and 7 RT-PCR confirmed the absence of exon 6, showing a single band of 169 bp (Fig. 2A, 5F/7R), while no 313 bp including exon 6 was detected. Exon 7 and LVPQ RT-PCR revealed a single band of 160 bp (Fig. 2C, 7F/LVPQ), demonstrating no significant difference between embryonic and postnatal hearts. In contrast, exon 7 and SXXSS RT-PCR showed a detectable band (Fig. 2C, 7F/SXXS) that decreased in intensity from embryonic to postnatal hearts. A longer gel run revealed two distinct bands: 254 bp (with LVPQ) and 242 bp (without LVPQ) (Fig. 2C, bottom).

Fig. 3.

Various combinations of C-terminal exons of Tcf7l2 transcripts. A: Conservation of alternative exons of Tcf7l2 across 100 vertebrates. Pairwise alignments of each species to the human genome are displayed below the conservation histogram as a wiggle that indicates alignment quality. B: Predicted splicing variations of alternative exons 14, 15, 16, and 17, illustrating different isoform possibilities. C: Frequency of C-terminal alternative exons detected by RT-PCR cloning. The Chi-square (χ²) test or Fisher's exact test was used. P < 0.05 was considered statistically significant. ns, not significant. D: C-terminal exon combinations detected by RT-PCR cloning and their frequency in 17.5E and P8 hearts. S1 (without exon 14) is E17.5-specific, whereas S2 (with exon 14) is observed at both stages. E: RT-PCR verification of C-terminal exon combinations. RT-PCR using a forward primer in exon 12 (12F) and reverse primers in exon 14 (14F), 15 (15R) or 16 (16R) confirm exon 14 and 15 inclusion/exclusion. 12F and 14R (12F/14R) produces a single band, indicating no alternative splicing in exon 13. 12F and 15R (12F/15R) reveals two bands, corresponding to transcripts with (225 bp, upper) or without (174 bp, lower) exon 14. 12 F and 16R (12/16R) produces three bands: Top (304 bp), which contains both exons 14 and 15; Middle (253/231 bp), which lacks either exon 14 or 15; Bottom (180 bp), which lacks both exons 14 and 15. The upper bands (exon 14 inclusion) are enriched during early neonatal stages, gradually decreasing to baseline levels in adult hearts, while the bands without exon 14 remain relatively unchanged during development. GAPDH is included as a housekeeping gene. Three biological replicates were analyzed per time point, using single postnatal and pooled embryonic (two per replicate) hearts.

Fig. 5.

Structural impact of alternative splicing and adenine indels in the C-terminal region of Tcf7l2 isoforms. A-D: Top panel, AIUPred-disorder (red) and AIUPred-binding (blue) scores predict intrinsic disorder and binding propensities across the Tcf7l2 C-terminus. Scores above 0.5 indicate disordered regions, while binding sites are highlighted in blue peaks. Bottom panel: the output of AIUPred using the redox state modeling option, where the estimated sensitivity of the disorder tendency is marked in orange. The plot shows two lines corresponding to redox_plus (maroon) and redox_minus (red) scenarios. An identified redox-sensitive disordered region will be indicated by shading the region (orange) between the two lines (does not appear if there are no sensitive regions predicted). The x-axis denotes amino acid positions, and the lower section displays annotated Pfam domains.

A: AIUPred analysis of the Tcf7l2 E1 isoform, which includes exons 13, 15, and 18 in its C-terminal region. The E1 isoform exhibits disordered N-terminal and C-terminal regions, indicated by the high AIUPred-disorder values (red line), while the central region fluctuates between disorder (red line) and binding propensity (blue line), indicating a largely ordered DNA-binding domain overlapping with the HMG-box. The N-terminal region contains the β-catenin binding domain (BCBD), while the C-terminal disordered region harbors several known binding sites for nuclear localization signals (NLS) and CtBP-binding motifs which are recognized by the AIUPred-binding and highlighted by blue peaks. Known binding regions are indicated by arrows. The NLS, primarily encoded by exon 13 (basic tail), is located at residues 401–410 (yellow arrows). The C-clamp is located at residues 424–446 (brown arrows). The E1 isoform also features two divergent CtBP-binding motifs (black arrows). The corresponding residues for these sites are listed in the rectangle with same color. B: Effect of exon 14 insertion. Top: Inclusion of exon 14 along with exons 13, 15, and 18 results in the E2 isoform. There are several regions spanning residues 400–500 that the AIUPred prediction dips slightly below the 0.5 threshold after exon14 insertion (red line). Bottom: Exon 14 insertion introduces a redox-switch domain spanning residues 396–477 (orange area). This region encompasses both the NLS and the C-clamp. C: Impact of adenine deletion in exon 18: An adenine deletion in exon 18 of Tcf7l2 E1 (E1_delA) produces a truncated isoform (466aa), altering the redox state of residues 416–466. This region contains an atypical C-clamp domain lacking the final conserved cysteine (green area), shifting its structure closer to that of the Tcf7l2 S1 isoform (panel D), which features a truncated C-clamp (C-clamp_ΔC, light brown arrows). E: Schematic comparison of the C-terminal amino acid sequences from TCF4 E1, E1_delA, and S1 variants. The four conserved cysteines of the full-length C-clamp domain are highlighted in black (E1 numbering). Amino acid sequences derived from exon 15 are underlined in black, while the NLS sequence is underlined in yellow. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.2. C-terminal variations due to alternative exon usage

C-terminal alternative exons, 14, 15 and 16 are highly conserved across vertebrates compare to exons 4 and 6 (Fig. 3A). These conserved alternative exons are enriched in genes involved in transcriptional regulation, and frequently contribute to tissue-specific or developmental-stage-specific gene expression, which is often preserved across species [[22], [23], [24], [25]].

Consistent with findings in other organs, extensive alternative usage and combinations of C-terminal exons 14 to 18 were observed in mouse hearts. Exon 17 was not detected in any of the clones. Exon 14 inclusion was more frequent in P8 hearts (18 of 28, 64.3 %) compared to E17.5 hearts (8 of 23, 34.8 %), making it the only C-terminal alternative exon showing statistically significant differential expression between E17.5 and P8 hearts (Chi-square, P = 0.0360; Fig. 3C). The frequencies of exon 15 inclusion were 60.9 % and 50.0 %, while exon 16 inclusion was 17.4 % and 17.9 % respectively in E17.5 and P8 hearts. Additionally, in E17.5 hearts, exon 16 was exclusively present together with exon 15. The frequencies of exon 14 sliced together with exon 15 were 17.4 % in E17.5 and 32.1 % in P8 hearts (Fisher’s exact test, P = 0.3355; Fig. 3C).

From E13.5 to 3-month-old mice, a 196 bp amplicon was produced using RT-PCR with exon 12 and exon 14 primers (Fig. 3, 12F/14R). This confirmed that there was no alternative splicing of exon 13. Further RT-PCR using exon 12 and exon 15 primers showed two bands (Fig. 3, 12F/15R): 225 bp with exon 14 and 174 bp without exon 14. Similarly, using exon 12 and exon 16 primers, three bands were seen (Fig. 3, 12F/16R): 304 bp with exons 14 and 15, 253/231 bp without exon 14 or 15, and 180 bp without either exon 14 or exon 15.

3.3. Protein isoform conversion by adenine insertion or deletion in exon 18 of Tcf7l2 E and M isoforms

A stretch of nine consecutive adenines (9a) near the beginning of exon 18 (located at chr19:55,919,869–55,919,877 in the mouse reference genome mm39) normally supports the in-frame translation of exon 18, enabling the usage of the distal or proximal stop codon for E and M isoforms, respectively. Exon 14 is 51 base pairs long, and its inclusion via alternative splicing is in-frame; thus, its presence in the E2 and M2 isoforms does not affect downstream amino acid sequences or stop codon selection.

However, an insertion (insA; 10a) or deletion (delA; 8a) of a single adenine in this poly-A stretch causes a frameshift, introducing an upstream stop codon (the first stop codon) before the normal proximal stop codon (Fig. 4A). The deletion (delA) in E isoforms shifts the reading frame and activates the second stop codon in exon 18, resulting in a truncated isoform more structurally similar to the S isoforms. In contrast, the insertion (insA) triggers the use of the first stop codon, generating an even shorter E isoform compared to the delA variant [11,12]. These indels were commonly observed in E isoforms—except for E3, likely due to its low transcript abundance in mouse hearts (Fig. 3; 3.6 %, 1/51). In the M2 isoform, delA also causes a frameshift that leads to a shorter protein, differing by 22 amino acids at the C-terminus, though this results in only a minor length change. This M2 variant was detected exclusively in E17.5 mouse hearts. Conversely, insA in M2 alters the reading frame and activates the third (distal) stop codon, producing a longer protein that lacks the C-clamp (ΔC-clamp) but gains CtBP-binding motifs, rendering it structurally similar to TCF7L1. Notably, no adenine indels were observed in M1, S1, or S2 isoforms in mouse hearts (Fig. 4B and Table 1). The final stop codon usage of Tcf7l2 isoforms, after adjustment for adenine indels, is summarized for E17.5 and P8 in Supplemental Table S2.

Fig. 4.

The effect of an adenine indel in exon 18 on Tcf7l2 C-termini. A: Variability in adenine (A) repeats near the start of exon 18 generates stretches of 8, 9, or 10 adenines (8a, 9a, 10a). The abundant 9a variant allows translation of exon 18 to the third (distal) stop codon in E isoforms and the second (proximal) stop codon in M isoforms. In E isoforms, deletion of one adenine (delA, 8a) shifts termination to the proximal second stop codon, producing a shorter protein without CtBP-binding motifs. Conversely, insertion of one adenine (insA, 10a) activates the first stop codon, yielding an even shorter isoform. In M isoforms, delA has no significant effect, whereas insA activates the third stop codon, generating a longer product with CtBP-binding motifs. B: Frequency of 8a, 9a, and 10a variants in Tcf7l2 transcripts at E17.5 and P8, determined by RT-PCR cloning.

Table 1.

Classification of Tcf7l2 transcript and protein C-terminal isoforms.

Transcript isoform	Transcript C-tail	Adenine indels in exon 18	C-clamp	CtBP motif	Redox-switch domain	Protein isoform	Protein length	E17.5	P8
E1	13–15-18	No	full	2	No	E_C-clamp	581	17.4 %	7.1 %
E1_delA	13–15-18	deletion	C-clamp_ΔC	0	C-clamp	S_C-clamp_ΔC_Redox	466	8.7 %	10.7 %
E1_insA	13–15-18	insertion	C-clamp_ΔC	0	No	S_C-clamp_ΔC	452	4.3 %	0.0 %
E2	13–14–15-18	No	full	2	NLS, C-clamp	E_C-clamp_Redox	598	4.3 %	21.4 %
E2_ delA	13–14–15-18	deletion	C-clamp_ΔC	0	NLS, C-clamp	S_C-clamp_ΔC_Redox	483	0.0 %	3.6 %
E2_ insA	13–14–15-18	insertion	C-clamp_ΔC	0	No	S_C-clamp_ΔC	469	8.7 %	0.0 %
E3	13–16-18	No	full	2	No	E_C-clamp	581	0.0 %	3.6 %
E3_ delA	13–16-18	deletion	C-clamp_ΔC	0	C-clamp	S_C-clamp_ΔC_Redox	466	0.0 %	0.0 %
E3_ insA	13–16-18	insertion	C-clamp_ΔC	0	No	S_C-clamp_ΔC	452	0.0 %	0.0 %
E4	13–14–16-18	No	full	2	NLS, C-clamp	E_C-clamp_Redox	598	0.0 %	0.0 %
E4_ delA	13–14–16-18	deletion	C-clamp_ΔC	0	NLS, C-clamp	S_C-clamp_ΔC_Redox	483	0.0 %	3.6 %
E4_ insA	13–14–16-18	insertion	C-clamp_ΔC	0	C-clamp	S_C-clamp_ΔC_Redox	469	0.0 %	3.6 %
M1	13–18	No	No	0	No	M	442	21.7 %	14.3 %
M1_ delA	13–18	deletion	No	0	No	M	427	0.0 %	0.0 %
M1_ insA	13–18	insertion	No	2	No	E_ΔC-clamp	553	0.0 %	0.0 %
M2	13–14-18	No	No	0	411–459	M-Redox	459	4.3 %	25.0 %
M2_ delA	13–14-18	deletion	No	0	No	M	444	8.7 %	0.0 %
M2_ insA	13–14-18	insertion	No	2	No	E_ΔC-clamp	570	4.3 %	0.0 %
S1	13–15–16-18	No	C-clamp_ΔC	0	No	S_C-clamp_ΔC	454	13.0 %	0.0 %
S2	13–14–15-18	No	C-clamp_ΔC	0	NSL, C-clamp	S_C-clamp_ΔC_Redox	471	4.3 %	7.1 %

The adenine indels in exon 18 have been reported in the literature and are also reflected in the human genome database (GRCh38). An adenine insertion has been reported under the genomic variant ID rs769592153, characterized by a G > GA substitution at position 113,165,557. Meanwhile, adenine deletion is associated with the genomic variants rs745872748 (GA > G at positions 113,165,557–113,165,558). Frameshift mutations of Tcf7l2 induced by the delA and insA in exon 18 have been also observed and reported in gastric carcinomas, brain tumors and colorectal carcinomas [11,[26], [27], [28]].

3.4. Exon 14 inclusion primarily dictates redox potential sensitive conditional disorder in the C-terminus of Tcf7l2 protein isoform

To gain deeper insights into the structural and functional variations among different Tcf7l2 isoforms with diverse C-terminal compositions, InterPro, ELM and AIUPred analyses were performed to assess domain and motif organization, intrinsic disorder propensity and to identify disordered binding regions. The full-length amino acid sequences of the observed Tcf7l2 C-terminal isoforms (Suppl Table S3) were subjected to domain prediction and disorder analysis. Prediction profiles were generated using ANCHOR2 and the Redox-state options to assess potential structural and functional changes.

Fig. 5 shows that the Tcf7l2 E1 isoform, which contains exons 13, 15, and 18, has 581 amino acids (aa) and features a disordered N-terminal and C-terminal region, with a structured DBD (HMG-box) in the center. The N-terminal region contains the BCBD, while the C-terminal disordered region harbors multiple experimentally validated regions for the NLS (⁴⁰¹GKKKKRKRDK⁴¹⁰) and two potential CtBP binding motifs, PLSLSLK and PLSLVTK [29]. The NLS, located between residues 401–410 (E1 numbering, indicated by yellow arrows), is primarily encoded by the exon 13 (basic tail). Additionally, two CtBP-binding motifs are found at residues 493–499 and 572–578 (E1 numbering, black arrows). These motifs are in disordered regions (red line, scores >0.5), as confirmed by ANCHOR2 (AUIPred-binding), appearing as separate blue peaks in the prediction profile.

The in-frame alternative exon 14 encodes 17 residues (EHSE⁴²¹CFLNP⁴²⁶CLSLPPIT) containing two cysteines in disordered regions, and is included in E2 isoforms. When exon 14 is inserted between exons 13 and 15 in the E2 isoform, several regions spanning residues 400–500 become more structured and folded in the prediction profile (Fig. 5B, upper panel; red line), suggesting that the NSL and C-clamp regions may be buried within a folded core formed by exon 14 inclusion. Most importantly, AIUPred modeling predicts that E2 is highly susceptible to redox state–driven phase transitions due to the presence of cysteines at 421 and 426. This introduces a redox-sensitive domain spanning residues 396–477, which includes both the NLS and C-clamp, but is located before the two CtBP binding peaks, contributing to the formation of a C-terminal redox-switch domain in Tcf7l2 isoforms (Fig. 5B). These findings suggest that exon 14 insertion modifies the conformation and function of the NLS and C-clamp domains, allowing them to exhibit different binding activities under unstressed and oxidative stress conditions.

The delA in exon 18 of the Tcf7l2 E1 isoform (E1_delA) produces a truncated 466 aa protein without CtBP-binding motifs, altering the redox state within residues 416–466 (Fig. 5C-E). This region contains an atypical C-clamp domain that lacks the final conserved cysteine residue (C-clamp_ΔC). Since the NLS of E1_delA is not located within the redox-switch region, the conversion of E1 to the S isoform may not impact its nuclear localization in both normal and oxidative conditions.

Without considering adenine variations in exon 18, the NLS domains of E2, E4, and S2 isoforms reside within redox-sensitive regions, making them susceptible to oxidation and reduction. This leads to varying abilities of their NLS for binding to importin proteins under redox-minus and redox-plus conditions, which in turn could impact their nuclear import, overall functions, and most importantly, subject these isoforms controlled by the redox signaling (Table 1). The NLS domain of M2 remains unchanged; however, the potential protein-binding activity of residues 411–459 is predicted to vary significantly between unstressed and oxidative stress conditions, according to the redox-state prediction model of AIUPred (Suppl Fig. S1). The total percentage of Tcf7l2 isoform with a redox-switch C-terminus was 21.7 % (13.0 % + 8.7 %) in E17.5 compared to 75.0 % (64.3 % + 10.7 %) in P8 (Fisher’s exact test, P = 0.0002). In the P8 stage, 64.3 % of redox-switch isoforms contained exon 14. Conversely, at E17.5, 21.7 % of isoforms originally possessing a redox-switch C-tail lost this feature due to either adenine deletion or insertion (Fig. 6A).

Fig. 6.

Redox-sensitive Tcf7l2 isoforms in mouse hearts. A: Percentages of redox-sensitive and nonredox-sensitive Tcf7l2 isoforms in E17.5 and P8 hearts, determined by exon 14 inclusion or exclusion and adenine indels. B: The classification of Tcf7l2 protein isoforms according to their C-terminal length, the presence of a full (C-clamp) or truncated C-clamp (C-clamp_ΔC), and redox-sensitive region. C: Most common Tcf7l2 transcript isoforms in 17.5E and P8 hearts. There are three fetal isoforms that are exclusively found in E17.5 hearts: E1, which includes LVPQ, SXXSS, and exon 15 but excludes exons 4 and 14 (XM_030250847); M1, which includes exon 4, LVPQ, and SXXSS, but excludes exons 14, 15, and 16 (new); and S1, which includes LVPQ, exons 15 and 16, but excludes exon 4, SXXSS and exon 14 (XM_017318129). The most common neonatal isoform is E2 which contains LVPQ, exons 14 and 15, and a redox-switch domain, but lacks exon 4, SXXSS, and exon 16 (NM_001142922).

The exon 14 insertion adds a redox-switch domain to all Tcf7l2 isoforms, including E, M, and S. In contrast, the delA/insA in exon 18 affects only the E and M isoforms, as the S isoform utilizes a stop codon in exon 16. Consequently, the adenine deletion/insertion does not alter the amino acid sequence in the S isoform. Additionally, the delA and insA in the M1 can alter the amino acid sequence but cannot change the redox sensitivity at the C-terminus (Suppl Fig. S1A, C and E). These may explain why we observed an abundance of S-isoform and M1-isoform transcripts in both E17.5 and P8 hearts, but none with adenine indels. Moreover, this finding indirectly confirms that the delA/insA in exon 18 represents a genuine modification in Tcf7l2 transcripts rather than an artifact introduced by PCR.

Considering protein length and C-terminal features, including the C-clamp (full C-clamp or C-clamp_ΔC), CtBP binding domains, and the redox-switch region, we classified Tcf7l2 protein isoforms into seven distinct isoforms (Fig. 6B). Among these, the M and S isoforms with truncated C-clamp (S_C-clamp_ΔC) are the two most abundant isoforms in the E17.5 stage (30.4 % and 26.1 %, respectively). In contrast, at P8, the M and S isoforms with a redox-switch C-terminus (M_Redox, S_C-clamp_ΔC_Redox) dominate as the top two isoforms (Fig. 6B). The detailed results for C-terminal isoforms, including features introduced by alternative exon splicing and adenine indels in exon 18, are summarized in Table 1.

3.5. Major full-length Tcf7l2 transcripts in fetal and postnatal hearts

Any Tcf7l2 C-terminal combination can potentially have different alternative exons and splicing variants in the CRD to create unique individual Tcf7l2 transcripts. After the sequence alignment and exclusion of two isoforms with exon 3 alterations, we identified 32 individual Tcf7l2 transcript isoforms with distinct compositions of alternative exons and splicing variants in the CRD and C-terminus and half of them were novel (Suppl Table S4). Most of isoforms showed differential expression in E17.5 and P8 hearts. Among these Tcf7l2 isoforms, only four were commonly expressed in both E17.5 and P8 hearts, while the others were exclusively detected in either the embryonic or postnatal hearts. Postnatal isoform (Tcf7l2–10), a Tcf7l2 E2 (NM_001142922) which has LVPQ but without exon 4 and SXXSS (Fig. 6C, Suppl Table S4) was 14.3 % in P8 and 0.0 % in E17.5 hearts. Three fetal isoforms (Fig. 6C) were 0 % in P8 and 13.0 % in E17.5 hearts: Tcf7l2 E1 with LVPQ and SXXSS but without exon 4 (Tcf7l2–03, XM_030250847); Tcf7l2 M1 with LVPQ, SXXSS and exon 4 (Tcf7l2–22, new); Tcf7l2 S1 with LVPQ but without SXXSS and exon 4 (Tcf7l2–30, XM_017318129). Four isoforms, one E, two M, and one S, were shared by both fetal and postnatal hearts. The new isoform, E1 with the delA (Tcf7l2–04), was present in 4.3 % and 7.1 % of E17.5 and P8 hearts, respectively. Tcf7l2 M1 (Tcf7l2–19, NM_001142923) was also present in 4.3 % and 7.1 % of E17.5 and P8 hearts. Tcf7l2 M2 (Tcf7l2–24, NM_009333) was 10.7 % in P8 compared to 4.3 % in E17.5 hearts. Tcf7l2 S2 (Tcf7l2–32, NM_001331137) showed a similar percentage in both E17.5 (4.3 %) and P8 (3.6 %) hearts.

4. Discussion

Canonical WNT signaling exhibits multiphasic, often opposing roles in cardiac development. While activation before gastrulation promotes heart morphogenesis, it inhibits cardiac differentiation during and after gastrulation [30]. The molecular mechanisms underlying these divergent effects remain unclear. Another critical transition during heart development occurs soon after birth, when CMs switch from hyperplasia to hypertrophy [13]. Whether and how canonical WNT signaling modulates this transition is not well understood. Tcf7l2, a nuclear partner of β-catenin, produces diverse isoforms through promoter usage and alternative splicing, which may either activate or repress downstream gene expression. However, stage-specific expression of Tcf7l2 isoforms in the heart has not been well defined. Understanding Tcf7l2 isoform switch during heart development could provide valuable insights into its stage-specific activity and its potential role in modulating WNT signaling specificity.

Our findings reveal that multiple Tcf7l2 isoforms arise from various combinations of alternative exons and splicing variants. In the CRD region, alternative exons 4 and 6 are less conserved across vertebrates and may exhibit tissue- and cell-specific alternative splicing, but their function remains unclear. Notably, exon 6 was undetected in the mouse heart, and over 80 % of isoforms excluded exon 4, as in the brain. In contrast, intestinal epithelial cells commonly include exon 4, which may suppress WNT target gene expression [8,12]. Variations in exons 8, 9, and 10 are functionally significant. Groucho/TLE (transducin-like enhancer of split) suppressors bind this region to repress WNT signaling [31,32]. Alternative splice donor and acceptor sites in exons 8 and 10 generate Tcf7l2 isoforms flanking exon 9 with different combinations of LVPQ and SXXSS motifs [10]. In Xenopus, the present of LVPQ and SXXSS motifs together form a repressing element in Tcf/Lefs, but their role in mammals is unclear [10,33]. We observed that exon 9 was mostly flanked by LVPQ (>80 % of transcripts at E17.5 and P8). SXXSS inclusion was lower than LVPQ in E17.5 (34.8 %) and P8 (17.9 %).

Alternative C-terminal exons 14, 15, 16, and 17 are highly conserved throughout evolution, suggesting their functional significance and crucial roles in the diversification of gene function. It has been proved that they harbor multiple functional regions which play critical roles in protein interaction and DNA binding. Mouse hearts showed extensive alternative splicing in exons 14–16, generating E, M, and S isoforms with highly diverging C-termini [8,11].

In E17.5 hearts, Tcf7l2 transcripts comprised 43.4 % E, 39.1 % M, and 17.3 % S isoforms, whereas in P8 hearts, the proportions shifted to 53.6 % E, 39.3 % M, and 7.1 % S isoforms. Exon 14 inclusion significantly increased in P8 hearts (64.3 %) compared to E17.5 (34.8 %). Exon 15 inclusion was 60.9 % in E17.5 and 50.0 % in P8 hearts, and was more frequently spliced with exon 14 in P8 hearts (32.1 %) compared to E17.5 (17.4 %). Exon 16 inclusion remained comparable between E17.5 (17.4 %) and P8 (17.9 %) hearts. However, at E17.5, exon 16 was exclusively spliced with exon 15 to generate S isoforms. Exon 17 was not detected in mouse hearts.

Despite these observations, drawing meaningful conclusions solely from transcript-level analysis remains challenging without considering the corresponding protein products. We used InterPro, ELM, and AIUPred to analyze domain structure and intrinsic disorder propensity in different Tcf7l2 C-terminal isoforms. For the first time, we show that exon 14 inclusion, when combined with other alternative C-terminal exons, generates a redox-sensitive center spanning the NLS and C-clamp domains in E2 and S2 isoforms. In the M2 isoform, the redox-sensitive center is positioned after the NLS. Importantly, the CtBP binding sites remain unchanged after exon 14 inclusion. These redox-sensitive domains may regulate nuclear import and DNA binding of E2 and S2. Furthermore, the redox-sensitive C-clamp domain may exhibit differential binding activities under non-oxidative and oxidative conditions, giving these isoforms distinct functional properties compared to E1 and S1.

At E17.5, redox-sensitive isoforms with exon 14 made up 13.0 % of transcripts, increasing to 64.3 % at P8. Adenine indels in exon 18 further modulate isoform function by causing frame shifts that either introduce or eliminate redox-switch domains. These indels also change isoform length, producing E-to-S or M-to-E transitions. At E17.5, 21.7 % of Tcf7l2 transcripts lost their redox domains due to adenine indels; at P8, 10.7 % gained redox sensitivity from these modifications. Overall, our findings reveal a significant postnatal shift in cardiac Tcf7l2 isoform composition, with the proportion of redox-sensitive isoforms increasing from 21.7 % at E17.5 to 75.0 % at P8. This transition suggests a potential regulatory mechanism in which redox-sensitive Tcf7l2 isoforms become predominant postnatally, possibly influencing WNT signaling and cardiomyocyte maturation during the critical switch from proliferation to hypertrophy.

5. Final perspective

At birth, increased oxygen exposure triggers a physiological oxidative burst, leading to reactive oxygen species (ROS) generation and activation of redox signaling. Our findings reveal that Tcf7l2 isoforms in the mouse heart are dynamically regulated postnatally through alternative splicing and adenine indels, introducing redox-sensitive domains that may modulate WNT signaling during the critical transition from cardiomyocyte proliferation to hypertrophy. These structural changes expand the functional repertoire of Tcf7l2 and suggest a potential role for redox-sensitive isoforms in postnatal cardiac maturation. Additionally, the adenine indel-induced frame shifts that impact redox domain may also have broader relevance in other biological contexts, including cancer [11,27,28,34].

The diversity generated by CRD and C-terminal exon variation highlights the complexity of Tcf7l2 isoform regulation. Subtype-specific elements, including exon 4, LVPQ, SXXSS, and exons 14–18, may contribute to the distinct functional properties of Tcf7l2 isoforms. Future studies focusing on the functional properties of subtype-specific elements will be essential for determining their roles in cardiac development and WNT/β-catenin regulation.

The following are the supplementary data related to this article.

Supplementary material

Table S4

Tcf7l2 isoforms.

CRediT authorship contribution statement

Bo Ye: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Lu Xiao: Investigation, Data curation. Yuyong Xu: Investigation, Data curation. Haodong Xu: Writing – review & editing. Faqian Li: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

Declaration of Generative AI and AI-assisted technologies in the writing process

The authors did not use generative AI or AI-assisted technologies in the development of this manuscript.

Declaration of competing interest

The authors declare no competing interests.

No financial or non-financial support was received from any third party for the work reported in this manuscript.

The authors have no relevant financial relationships, advisory roles, consulting fees, equity interests, or non-financial support outside this work within the past three years.

There are no patents, copyrights, or other intellectual property relevant to the content of this manuscript.

The authors are not aware of any other activities or relationships that could be perceived to influence the reported work.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.