Figure 2. Whole-genome sequencing from defined population doubling Pol ε^wt/exo- mismatch repair-deficient cells.

(A) Whole genome sequencing (2.8 × 10⁹ bp, average 30X coverage) was performed on Pol ε^wt/exo- cells lacking mismatch repair at two defined population doubling levels, P0 and P14, as described in the Methods. P0 was used as the matched normal cells to define only those mutations arising during the 14 population doublings. The fraction of each type of base pair substitution from the PD 14 Pol ε^wt/exo- cells was plotted and compared to the fraction of each type of mutation from HCT116 ((Abaan et al., 2013) and this study) and HCC2998 cells (Abaan et al., 2013). Chi square tests with Yates correction were used to calculate p values relative to SNVs found in Pol ε^wt/wt mismatch repair-deficient cells in this study. Pol ε^wt/wt (Abaan et al.) χ² = 0.033, p=0.8551; Pol ε^wt/P286R (Abaan et al.) χ² = 872.341, p<0.0001; Pol ε^wt/exo-χ² = 2,3680.508, p<0.0001. ****p<0.0001; n.s., not significant. (B) The number of each indicated base pair substitution in a specific trinucleotide context was plotted from the PD 14 Pol ε^wt/exo- mismatch repair-deficient cells. The base pair substitutions shown (C > A and T > G transversions, left; C > T transitions, right) are those found enriched in POLE tumors. Chi square tests with Yates correction were used to calculate p-values relative to SNVs found in Pol ε^wt/wt mismatch repair-deficient cells in this study. C > A TCT χ² = 152.772, p<0.0001; T > G TTT χ² = 72.254, p<0.0001. ****p<0.0001.

Figure 2—figure supplement 1. Whole genome SNVs identified in Pol ε^wt/exo- (PDL = 14) in cells lacking functional mismatch repair identified.

Whole exome SNVs identified in HCT116 cells in the current study, as well as from HCT116 and HCC2998 cells previously by Abaan et al. were extracted and identified. All SNVs were then analyzed for their triplet nucleotide sequence context. Occurrences of each of the 96 possible trinucleotide base pair substitutions was then plotted as a percentage of the total number of SNVs. The mutation status of Pol ε is indicated for each cell line. The number of SNVs for each study is indicated.

Figure 2—figure supplement 2. POLE mutation signature extracted from POLE-mutant cell lines.

(A) Non negative matrix factorization (NMF) was used to extract six unique mutation signatures from POLE-mutant (n = 14) and POLE-wild type (n = 545) colorectal cancer patients, Pol ε-P286R HCC2998 cells (SNVs extracted from Abaan, Cancer Res. 2013) and the indicated cell lines with and without MMR from the current study. Cosine similarity to all 30 cancer mutational signatures (Alexandrov, Cell Reports 2013) showed significant overlap between our Signature ‘A’ set and mutation signature 10 (ascribed to Pol ε mutations) and 98% overlap between between our Signature ‘B’ and mutation signatures 5, 6 and 20 (ascribed to MMR mutations). (B) The relative contribution of each extracted mutation signature to the total mutations in the indicated sample was calculated. Signatures and their respective contributions to each sample were derived using NMF. NMF was performed on all samples together (including cell lines and patients) using the Scikit-learn Python library (Pedregosa et al., Jrnl Machine Research Learning 2011.).

Figure 2—figure supplement 3. Mutation counts in the indicated trinucleotide context (top) were plotted as a proportion of their occurrence (bottom) in WGS samples.

Figure 2—figure supplement 4. Mutation counts in the indicated trinucleotide context (top) were plotted as a proportion of their occurrence (bottom) in WES samples.

Figure 2—figure supplement 5. Relative contributions of Cosmic Mutation Signatures to individual patient mutation spectra were determined using deconstructSig.

The relative proportion of each signature is shown for several tumor types (bMMRD, blue; somatic endometrial, green; somatic colorectal, salmon) and for Pol ε (black) and Pol δ (gray) mutations.

Figure 2—figure supplement 6. Mean coverage was greater than 90x for each WES sample.

Figure 2—figure supplement 7. Alignment rate to the reference genome exceeded 99% for each WES sample.

Figure 2—figure supplement 8. Total reads exceeded 60 million for each WES sample.

Figure 2—figure supplement 9. Greater than 90% of the bases in the WES genome exceeded 30x coverage.

Figure 2—figure supplement 10. Greater than 85% of the bases in the WGS genome exceeded 20x coverage.

Figure 2—figure supplement 11. Average alternate base quality to reference base quality of ~1.0.