Deep learning model of somatic hypermutation reveals importance of sequence context beyond hotspot targeting

Abstract

B cells undergo somatic hypermutation (SHM) of the Immunoglobulin (Ig) variable region to generate high-affinity antibodies. SHM relies on the activity of activation-induced deaminase (AID), which mutates C>U preferentially targeting WRC (W=A/T, R=A/G) hotspots. Downstream mutations at WA Polymerase h hotspots contribute further mutations. Computational models of SHM can describe the probability of mutations essential for vaccine responses. Previous studies using short subsequences (k-mers) failed to explain divergent mutability for the same k-mer. We developed the DeepSHM (Deep learning on SHM) model using k-mers of size 5-21, improving accuracy over previous models. Interpretation of DeepSHM identified an extended WWRCT motif with particularly high mutability. Increased mutability was further associated with lower surrounding G content. Our model also discovered a conserved AGYCTGGGGG (Y=C/T) motif within FW1 of IGHV3 family genes with unusually high T>G substitution rates. Thus, a wider sequence context increases predictive power and identifies features that drive mutational targeting.

The landscape of somatic mutations in lymphoblastoid cell lines

“… 20 , 21 SHM initiates with the deamination of cytosine to deoxyuracil via activation-induced cytidine deaminase (AID) at sequence motifs such as WR C (Y)/(R) G YW. 22 , 23 Repair of deoxyuracils then often involves DNA synthesis by the low fidelity translesion DNA polymerase η, leading to mutations. 24 Polymerase η synthesis can extend to nearby nucleotides to produce proximal A>G/C substitutions with a context preference of 3′ A/T.…”

Somatic hypermutation patterns in immunoglobulin variable regions are established independently of the local transcriptional landscape

“…However, WRCH motifs are not equally mutated, suggesting that the DNA sequence context determines whether ssDNA containing a WRCH motif is mutated and to what extent, a notion which is supported by biochemical studies (101). Indeed, analyses of variable regions has suggested that sequence-intrinsic features may regulate the SHM spectra (47,(50)(51)(52). Given that neither ssDNA patches nor nascent transcriptional features correlate with SHM spectra, we favor the notion that sequence context of a WRCH motif determines its mutability and propose that sequence context regulates the residence time of AID on ssDNA and thereby modulates the catalytic efficiency of AID in situ (Fig.…”

“…A major, unsolved problem in SHM biology is the differential mutability of WRCH motifs within variable regions which implies the existence of local, sequence-intrinsic mechanisms regulating AID activity postrecruitment (46)(47)(48)(50)(51)(52). We asked whether SHM spectra are dictated by local features of nascent transcription landscape that could transiently lead to increased ssDNA exposure, such as during transcription initiation, pausing and/or convergent/antisense transcription.…”

“…AID preferentially deaminates cytidine residues within WRCH motifs (where W = A or T, R = A or G and H = A, C or T) (43)(44)(45). However, not all WRCH residues are targeted and the mutation frequencies of those that are mutated can differ substantially (46)(47)(48)(49)(50)(51)(52). Remarkably, even identical WRCH motifs within a given variable region can differ significantly in mutation frequency (47)(48)(49)(50)(51)(52).…”

“…However, not all WRCH residues are targeted and the mutation frequencies of those that are mutated can differ substantially (46)(47)(48)(49)(50)(51)(52). Remarkably, even identical WRCH motifs within a given variable region can differ significantly in mutation frequency (47)(48)(49)(50)(51)(52). This differential mutability is a ubiquitous feature of SHM and implies the existence of as-yet-unknown mechanisms that determine which residue is mutated and to what extent.…”