1. Volume 63, Issue 4, Pages 633-646 (August 2016)
DNA Repair Profiling Reveals Nonrandom Outcomes at Cas9-Mediated Breaks 
Megan van Overbeek, Daniel Capurso, Matthew M. Carter, Matthew S. Thompson, Elizabeth Frias, Carsten Russ, John S. Reece- Hoyes, Christopher Nye, Scott Gradia, Bastien Vidal, Jiashun Zheng, Gregory R. Hoffman, Christopher K. Fuller, Andrew P. May 
Molecular Cell 
Volume 63, Issue 4, Pages (August 2016)
DOI: /j.molcel
Copyright © 2016 Elsevier Inc. Terms and Conditions

2. Molecular Cell 2016 63, 633-646DOI: (10.1016/j.molcel.2016.06.037)
Copyright © 2016 Elsevier Inc. Terms and Conditions

3. Figure 1 Profiling DNA Repair Outcomes after Cas9 Cleavage
(A) (i) Cell editing workflow. (ii) The sequencing library steps are shown. (iii) Indel class visualization following Cas9 cleavage at a target in JAK1 (Spacer 54, Table S1) is shown. The position of each insertion class (blue) and deletion class (black) is plotted relative to the cut site (green line). Throughout the text, indel visualizations are annotated as follows: each class is annotated with frequency (fraction of total reads and fraction of mutant reads) and the number of reads observed (right). The classes are ranked by frequency (classes with frequency <0.01 [fraction of mutant reads] are not displayed).
(B) Visualization of the five most frequent indel classes and WT at the same target in three cell lines (the first replicate, R1, of each is displayed).
(C) An indel frequency heatmap by length for each cell line (three replicates and WT control shown). Throughout the text, heatmaps are annotated as follows: insertions of 1 to 8 nucleotides are displayed (green). Single-base insertions are separated by nucleotide (A, T, C, and G). The deletion lengths of 1 to 50 nucleotides are displayed (blue). The color intensity scales with frequency as a fraction of mutant reads up to 0.2. The bar graph on the right (gray) displays the mean frequency of each indel. The bar graph above (orange) displays editing efficiency (Edit. Eff.) as a fraction of total reads (related to Figures S1 and S2).
See also Tables S1 and S2.
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

4. Figure 2 DNA Repair Profiles Are Unique to Each Spacer Sequence
(A and B) A matrix of the (Jaccard/Tanimoto) similarity of the top ten indel classes across pairs of 69 target sites (Table S1) in HEK293, K562, and HCT116 cell lines comparing (A) sgRNP-only delivery of reagents and (B) sgRNP and constitutive delivery of reagents. The targets with the same spacer label (within the ticks) are different experimental replicates of each cell type targeted by the same sgRNA. A similarity score of 1 represents complete overlap of the top ten indel classes between two sites, while 0 represents no overlap of the top ten indel classes between two sites.
(C) ARI values from cluster analysis (see Experimental Procedures) of (A) and (B).
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

5. Figure 3
DNA Repair Outcomes at Cas9-Mediated DSBs Are Sequence Dependent
(A) Genomic coordinates (hg38) that contain exact copies of the Spacer 15 sequence (Table S3).
(B) A heatmap of the frequencies of indels by length for the seven targets listed in (A) from replicate experiments plus WT controls 48 hr after nucleofection into HEK293 cells.
(C) A heatmap of the frequencies of indels by length in HEK293 cells for 22 different spacer groups (outlined in gray boxes) (Table S3). Each target sequence occurs at 2–14 times in the genome. For each target site within each spacer group, three experimental replicates and a WT control are displayed (within the minor ticks).
(D) A heatmap of the frequencies of indels by length in K562 cells for the same 22 spacer groups (outlined in gray boxes) as described in (C).
(E and F) A matrix of the (Jaccard/Tanimoto) similarity of the top ten indel classes across pairs of target sites in HEK293 cells (E) and K562 cells (F). The targets with the same spacer label (within the ticks) are different genomic loci targeted by the same sgRNA (related to Figures S3–S5).
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

6. Figure 4
The Distribution of DNA Repair Outcomes after Cas9 Cleavage Changes over Time
(A) Heatmap of DNA repair outcomes in HEK293 cells for 96 different spacers, each with three experimental replicates for multiple time points (4, 8, 16, 24, and 48 hr) and a WT control (within the minor ticks) (Table S1).
(B and F) Heatmaps for the indicated spacer: Spacer 13 (B) and Spacer 54 (F) showing single experiments at each time point for three cell lines.
(C and G) Bar graphs of indel frequencies by length for the indicated spacer: Spacer 13 (C) and Spacer 54 (G) displayed as a fraction of mutant reads (mean and SD across triplicates) in three cell lines at the 48 hr time point.
(D and H) A heatmap of Spacer 13 (D) or Spacer 54 (H) showing single experiments at each time point for three cell lines after applying a stringent MH mask (MH_score >3, see Figure S5D).
(E and I) Bar graph of indel frequencies by length displayed as a fraction of mutant reads (mean and SD across triplicates) at the 48 hr time point for Spacer 13 (E) or Spacer 54 (I) after applying a stringent MH mask (MH_score >3) (related to Figures S6 and S7).
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

7. Figure 5
Chemical Perturbation of c-NHEJ Promotes a Subset of DNA Repair Outcomes after Cas9 Cleavage
(A) Heatmap representing DNA repair classes present 48 hr after sgRNP introduction into HEK293T cells (Spacer 93; Table S1). The cells were treated with the DNA-PK inhibitor NU7441 (Leahy et al., 2004) in a 2-fold dilution series ranging from 1.56 μM to 25 μM. Each concentration was tested in duplicate. The untreated control replicates are shown on the right. Mean frequency plots are shown on the left comparing untreated samples with samples treated with DNA-PK inhibitor NU7441 (average mean frequency of entire dilution series 1.56 μM–25 μM displayed). The arrows indicate the repair classes that change frequency after treatment with NU7441 (red indicates a decrease in mean frequency in the presence of inhibitor, and green indicates an increase in mean frequency in the presence of inhibitor).
(B) Visualization of the five most frequent indel classes and WT of the same target shown in (A) (the first replicate, R1, of each is displayed). The NU7441 inhibitor concentrations 1.56 μM, 3.13 μM, and 6.25 μM are shown.
(C) Same as in (A) for Spacer 54 (Table S1).
(D) Same as in (B) for Spacer 54.
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

8. Figure 6
High-Frequency In-Frame Mutation after Cas9 Cleavage of a Target in CD34
(A) Genomic location (hg19 coordinates) of a target in the CD34 gene with the PAM boxed in yellow and Cas9 cut site indicated by an arrow. The entire protospacer sequence is boxed in purple (Spacer 16; Table S1). Deletion of the three nucleotide sequence located in the light blue box is the most frequent indel after either sgRNP delivery or constitutive expression of sgRNA to direct Cas9 cleavage activity to this site.
(B) A visualization of a subset of the indel classes (the five most frequent) and WT at the CD34 target in three different cell lines using sgRNP delivery and in two different cell lines for constitutive expression of Cas9/sgRNA as indicated on the left (a single replicate of each is displayed).
(C) A heatmap of the frequencies of indels by length at the CD34 target in HEK293, K562, and HCT116 cell lines (three experimental replicates of each and a WT control) 48 hr post sgRNP delivery.
(D) A heatmap of the frequencies of indels by length at the CD34 target in HEK293 and HCT116 cell lines 11 and 14 days post constitutive expression of Cas9/sgRNA (single replicate of two different time points and a WT control, Cas9-only).
(E) Bar graph of indel frequencies by length displayed as a fraction of mutant reads (mean and SD across three experimental replicates) in three cell lines 48 hr post sgRNP delivery.
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions

9. Figure 7
DNA Repair Outcome Profiling in Cell Lines Is Predictive for Human Primary Cells
DNA repair outcome after cleavage of a BRCA2 target results in two dominant products.
(A) Genomic location (hg19 coordinates) of a target in the BRCA2 gene (PAM, yellow and protospacer, purple). The Cas9 cut site is indicated by an arrow. The entire protospacer sequence is boxed in purple (Spacer 19; Table S1). The deletion of the four nucleotide sequence located in the light blue box and a single nucleotide insertion represent the most frequent indels after Cas9 cleavage.
(B) Visualization of a subset (the five most frequent) of the indel classes and WT at the BRCA2 target in three different cell lines and HSCs for sgRNP and three different cell lines for constitutive expression of Cas9/sgRNA as indicated on the left (a single replicate of each is displayed).
(C and D) Heatmaps of the frequencies of indels by length at the BRCA2 target in various cell types (three experimental replicates and a WT control) 48 hr post sgRNP delivery (C) or 11 and 14 days post constitutive expression of Cas9/sgRNA (D).
(E) Bar graph of indel frequencies by length displayed as a fraction of mutant reads (mean and SD across three experimental replicates) in three cell lines and HSCs 48 hr post sgRNP delivery.
(F) Use of DNA repair profiles to restore the reading frame of specific mutant alleles of BRCA2. 1. The WT BRCA2 locus (hg19 coordinates) (same as Figure 7A) is shown. 2. The sequence of BRCA2 mutant allele (dbSNP ID: ) missing a “C” nucleotide shifting the frame of BRCA2, resulting in a premature stop codon (red box with asterisk) is shown. 3. The genomic location of a target site on the BRCA2 mutant allele shown in (2) with the PAM boxed in yellow and the Cas9 cut site indicated by an arrow is shown. The entire protospacer sequence is boxed in purple. 4. An insertion of a single nucleotide after Cas9-cleavage at the target site shown in (3) would restore the frame of the BRCA2 mutant allele. Depending on the nucleotide that was inserted during the DNA repair reaction, the fifth amino acid from the left would be a phenylalanine, isoleucine, valine, or leucine (WT). 5. Sequence of a BRCA2 mutant allele (dbSNP ID: ) containing a duplication of “CTTA”, resulting in a frameshift and a premature stop codon (red box with asterisk) is shown. 6. The genomic location of a target site on the BRCA2 mutant allele shown in (5) is shown. 7. A four base deletion after Cas9-cleavage at the target site shown in (6) would restore the frame of the BRCA2 mutant allele. The resulting allele would differ from the WT allele by a single amino acid (leucine instead of a threonine, fourth amino acid from the left).
Molecular Cell  , DOI: ( /j.molcel )
Copyright © 2016 Elsevier Inc. Terms and Conditions