On the genetic basis of tail-loss evolution in humans and apes

Primates exhibit wide variability in tail morphology and function, influencing locomotion and ecology. While New World monkeys evolved prehensile tails, hominoids (humans and apes) lost the external tail around 25 million years ago, retaining only a coccyx. Tail loss has been hypothesized to facilitate orthograde posture and bipedal locomotion, but the underlying genetic mechanism remained unknown. Advances in primate genomics enable the identification of hominoid-specific genetic changes linked to phenotypes, and developmental genetics has elucidated gene networks controlling posterior body and tail development. The Mouse Genome Informatics database lists over 100 genes whose perturbation causes absent or shortened tail phenotypes, including core mesoderm/endoderm regulators such as Tbxt (T/Brachyury), Wnt3a and Msgn1, whose expression is enriched in the primitive streak and posterior body formation. Despite this knowledge, the causal genetic changes driving tail-loss evolution in hominoids had not been determined. This study addresses that gap by searching for hominoid-specific genetic variants in tail-development genes and testing their functional consequences.

Prior work established the role of TBXT (T/Brachyury) in mesoderm formation and tail development across vertebrates; heterozygous coding mutations in TBXT orthologues in mouse, cat (Manx), dog, and zebrafish cause reduced or absent tails, while homozygous loss is typically lethal. Comparative primate genomics has revealed extensive lineage-specific variation and provides resources to connect genotype to phenotype. Alu elements are abundant primate-specific SINEs that can influence gene expression and splicing, and inverted Alu pairs are known to affect alternative splicing and promote circular RNA biogenesis. Studies have linked human TBXT variants to neural tube defects and sacral agenesis. These literature foundations support the hypothesis that intronic transposable elements could modulate TBXT splicing to influence tail development and human disease.

Comparative genomics: - Compiled 140 vertebrate tail-development genes from MGI phenotype terms (absent, vestigial, short tail) and literature. - Extracted gene bodies plus ±10 kb flanks from Multiz30way alignments across 27 primates; used six hominoids to form a hominoid consensus and two outgroup monkeys (pig-tailed macaque and marmoset). - Identified hominoid-specific variants: 85,064 SNVs, 5,533 deletions, 13,820 insertions. Predicted functional impacts with Ensembl VEP; manual cross-species checks filtered to nine coding variants of uncertain relevance to tail loss. - Noted a hominoid-specific AluY insertion in intron 6 of TBXT flanking exon 6 with an existing inverted AluSx1 in intron 5, suggesting potential RNA secondary structure-mediated exon 6 skipping. RNA structure and in vitro splicing assays: - Predicted RNA secondary structure (ViennaRNA RNAfold) for TBXT exon5–intron5–exon6–intron6–exon7 showing likely pairing of AluSx1 (intron 5) with AluY (intron 6) forming a stem-loop that traps exon 6 and promotes exon 6 skipping (TBXTΔexon6). - Differentiated human ES cells (H1) to primitive streak-like states to induce TBXT; RT-PCR validated TBXTΔexon6 in human but not mouse. - Used CRISPR-Cas9 in human ES cells to delete AluY (intron 6) or AluSx1 (intron 5). Deletion of either element largely abolished TBXTΔexon6; also observed a minor Δexon6-7 transcript likely from a secondary AluSx1–AluSq2 pairing. Mouse genetic models and assays: - Generated TbxtΔexon6 mice via zygotic CRISPR deletion of exon 6 to force exon 5–7 splicing, mimicking human TBXT co-expression of full-length and Δexon6 isoforms. Validated splicing by Sanger sequencing. - Performed Capture-seq covering Tbxt ±200 kb to exclude off-target edits in multiple founders. - Engineered mouse ES cell lines inserting either human AluSx1+AluY into introns 5 and 6 (TbxtinsASAY) or a reverse complementary intronic sequence (RCS; 297 bp from intron 5) into intron 6 (TbxtinsRCS) to create inverted repeats flanking exon 6. Differentiation of ES cells showed induction of TbxtΔexon6 in both designs (higher ratio in RCS than Alu pair). - Generated mouse lines: obtained one TbxtinsASAY line and, serendipitously, a TbxtinsRCS2 line with a 220 bp reverse-complement insertion from intron 6 into intron 5 forming an inverted pair. Validated by Capture-seq and Sanger sequencing. - Phenotyping: measured adult tail lengths, categorized phenotypes (no tail, short, kinked, long). Collected E10.5 tailbud RNA for isoform RT-PCR to quantify relative abundance of full-length vs Δexon6 transcripts. - Crosses: Intercrossed heterozygotes; generated compound genotypes (TbxtinsRCS2/Δexon6) to modulate isoform ratios. Transcriptomics and pathology: - Bulk RNA-seq of day-1 differentiated mouse ES cells across genotypes (WT, insASAY/insASAY, Δexon6/+, Δexon6/Δexon6) analyzed with STAR and DESeq2; examined differentially expressed TBXT target genes. - Embryo analyses at E11.5 assessed viability and neural tube closure; documented NTD-like phenotypes. - Standard cell culture, CRISPR, nucleofection, selection, Cre-mediated cassette excision, and genotyping protocols were employed as detailed in Methods.

- Discovery of a hominoid-specific AluY insertion in intron 6 of TBXT that pairs with an ancestral, inverted AluSx1 in intron 5 to create a primate-specific RNA structure promoting exon 6 skipping, yielding TBXTΔexon6. - Deletion of AluY or AluSx1 in human ES cells abolished the TBXTΔexon6 isoform, demonstrating both elements are required for this alternative splicing event. - In human ES cells, a minor TBXTΔexon6-7 transcript likely results from a lower-probability AluSx1–AluSq2 interaction; distances AluSx1–AluY (~1,448 bp) vs AluSx1–AluSq2 (~4,188 bp) help explain preferential Δexon6 formation. - Mouse TbxtΔexon6/+ (heterozygous exon 6 deletion) mice, which co-express full-length and Δexon6 isoforms, showed heterogeneous tail phenotypes: 21 of 63 heterozygotes displayed no-tail or short-tail phenotypes, versus 0 of 35 wild-type littermates. - Capture-seq found no off-target mutations at the Tbxt locus in founder mice, supporting causality of the Δexon6 edit. - Engineered inverted intronic sequence pairs in mouse Tbxt induced the same exon 6 skipping: both TbxtinsASAY and TbxtinsRCS ES cells produced TbxtΔexon6; TbxtinsRCS expressed a higher Δexon6 proportion than TbxtinsASAY. - In vivo, TbxtinsASAY/insASAY mice showed no overt adult tail phenotype and low Δexon6 abundance in E10.5 tailbuds, whereas TbxtinsRCS2/insRCS2 mice had ~10% shorter tails and higher Δexon6 abundance than full-length in E10.5 tailbuds. - Relative abundance of Tbxt isoforms in embryonic tailbuds correlates with phenotype severity: higher Δexon6 relative to full-length associates with shorter tails; compound heterozygotes TbxtinsRCS2/Δexon6 (n=19) exhibited complete absence of external tail. - TbxtΔexon6/Δexon6 homozygotes were non-viable: embryos arrested around E9 or showed neural tube closure defects (NTDs) leading to perinatal death; one Δexon6/+ pup also showed NTDs, linking the Δexon6 isoform to NTD risk. - Comparative screen across 140 tail genes identified 85,064 hominoid-specific SNVs, 5,533 deletions, 13,820 insertions; nine coding variants were found but did not specifically implicate tail reduction, prioritizing the TBXT Alu-mediated splicing mechanism.

This study provides a plausible genetic mechanism for hominoid tail-loss: an intronic AluY insertion in TBXT pairing with a pre-existing inverted AluSx1 induces a hominoid-specific alternative splicing event that creates TBXTΔexon6. Functional experiments demonstrate that this splicing isoform, when co-expressed with full-length Tbxt in mice, is sufficient to cause tail shortening or complete tail loss. The phenotypic outcomes scale with the relative abundance of the Δexon6 transcript in embryonic tailbuds, suggesting tail development requires a threshold level of full-length TBXT or, conversely, is suppressed when Δexon6 exceeds a threshold. The work illustrates how interactions between intronic transposable elements can modulate splicing of a key developmental regulator to impact a complex morphological trait. Given the prevalence of intronic Alu elements in the human genome, similar mechanisms may influence other developmental processes and diseases, including through generation of circular RNAs. The link between TBXTΔexon6 expression and neural tube defects implies an evolutionary trade-off: selection for tail reduction may have increased NTD risk that persists in humans. Although the AluY insertion could have initiated tail shortening in early hominoids, additional lineage-specific variants likely contributed to fixation and stabilization of tail loss.

The authors identify a hominoid-specific AluY insertion in TBXT that, via pairing with an ancestral inverted AluSx1, drives exon 6 skipping and produces a TBXTΔexon6 isoform. Mouse models reveal that expression of this isoform alongside full-length Tbxt shortens or eliminates the tail, with phenotype severity determined by isoform ratios in embryonic tailbuds. Homozygous expression of Δexon6 is lethal and associated with neural tube defects, suggesting an adaptive trade-off in tail-loss evolution. The study advances an evolutionary mechanism connecting transposable element–mediated splicing regulation to a defining hominoid trait. Future work should: (1) delineate the molecular functions and genomic binding of the Δexon6 protein isoform; (2) map splicing regulation across relevant embryonic lineages and stages; (3) survey genomes for additional inverted intronic repeat configurations causing exon skipping; and (4) identify other hominoid-specific variants contributing to tail-loss stabilization and their potential interaction with TBXT splicing.

- The engineered Alu pair mouse model (TbxtinsASAY) did not produce high Δexon6 levels in E10.5 tailbuds despite in vitro ES cell evidence, limiting direct in vivo recapitulation; splicing regulation may vary by cell type or developmental stage. - Phenotypes in TbxtΔexon6/+ mice showed incomplete penetrance, indicating sensitivity to genetic background, stochastic splicing variation, or environmental factors. - The comparative genomics screen, while extensive, may contain false positives due to outgroup-specific variants and annotation limitations; only TBXT was functionally validated. - The precise biochemical functions of the Δexon6 protein isoform (e.g., co-factor interactions, transcriptional regulation) remain unresolved. - Additional hominoid-specific variants likely contributed to tail-loss fixation; these were not comprehensively functionally tested. - Neural tube defect observations are based on limited numbers of affected embryos/pups; larger cohorts would refine penetrance estimates.