CBS yeast complementation

Classical homocystinuria–characterized by excessive levels of total homocysteine in plasma–is caused by primarily-rare variants in the cystathionine-beta-synthase (CBS) gene. With early detection, existing therapies are highly effective, e.g., most patients respond to vitamin B6 therapy. Functional CBS variants, or deleterious variants that respond to vitamin B6, can be detected based on their ability restore growth in yeast cells lacking CYS4 (the yeast ortholog of CBS). Here we describe a comprehensive missense variant effect map for CBS.

We reimplemented a previously validated humanized yeast model (Kruger & Cox, PNAS, 1994), confirming that expression of human CBS from the hORFeome collection restores the ability of a yeast cys4Δ strain to grow without supplementation of glutathione. Coupling this functional complementation with our recently developed framework for exhaustively mapping functional coding variants (Weile & Sun et al, MSB, 2017), we attempted to test the functional impact as well as the vitamin B6 remediability of all possible missense CBS variants in parallel.

First, we constructed a library of CBS variants using our previously described POPCode mutagenesis method (Weile & Sun et al, MSB, 2017). The variant library, initially generated as a pool of amplicons, was transferred en masse into the appropriate yeast expression vector via two steps of recombinational subcloning. The resulting library of variant expression clones was then transformed en masse into the yeast cys4 mutant strain.

Next, pools of transformed yeast cys4 mutant strains were grown competitively in cysteine-lacking medium supplemented with low (0, 1ng/ml) or high (400 ng/ml) concentrations of vitamin B6. Allele frequencies of CBS variants before and after selection were determined by next-generation sequencing. To ensure accuracy of allele frequency estimates, we following the TileSeq approach (Weile & Sun et al, MSB, 2017) in which sequencing is applied to a tiling set of ~100 nt segments of the gene library, enabling sequencing reads from both forward and reverse strands of each template cluster on the flow cell. Sequencing was performed such that both forward and reverse strands of each nucleotide position were covered by ~2M reads.

The sequence analysis code can be found at https://bitbucket.org/rothlabto/tileseq_package

See https://www.biorxiv.org/content/early/2018/11/19/473983 for more information.