Quantification of the evolutionary constraint on intronic branch point sequences in the bovine and human genome

We predict the branch point sequence in 180,892 bovine introns and investigate the variability of this cis-acting splicing element at nucleotide resolution. A degenerate heptamer "nnyTrAy" with conserved thymine and adenine residues at positions 4 and 6 constitutes the bovine branch point sequence. This motif is located between 18 and 37 bp upstream of the 3 splice acceptor site. More than 90% of the introns contain canonical branch point sequence motifs with thymine and adenine residues at positions 4 and 6. Using a catalogue of 29.4 million variants detected in 266 cattle, we show that the conserved thymine and adenine residues are strongly depleted for mutations. These two intronic residues harbor 39% and 41% less variants than coding sequences suggesting extreme purifying selection. The evolutionary constraint is higher at the thymine residue than the branch point adenine itself in heptamers with a canonical branch point sequence motif. We replicate these observations in human branch point sequences with a catalogue of more than 115 million SNPs detected in 3,942 genomes from the gnomAD database. Our study suggests that the evolutionary constraint is stronger on intronic branch point sequences than coding regions in the bovine and human genome.