Response to Bakhoum et al.

reflect differences in experimental conditions or divergence between cell line isolates. These are mitigated in our analysis by the use of positive controls (monastrol washout to induce lagging centrics and low dose aphidicolin to prevent efficient DNA replication and induce acentrics and bridges) and orthogonal methods (immunodetection of two centromere markers, and FISH using an allcentromere probe) to confirm our ability to identify lagging centrics, acentrics and bridges, and by examination of multiple cell lines of a single cancer type (colorectal) [7]. These data raise an important point about controls for CIN+ versus CINcomparisons. Our analysis of HCT116 and DLD1 lagging centrics and anaphase bridges is concordant with Bakhoum et al.’s data [1,7]. However, in three additional CINcolorectal cancer cell lines we observed low percentages of bridges (similar to DLD1 cells), indicating that bridges are significantly more frequent in CIN+ than CINcells within our larger, tissue-type specific cell line panel. Bakhoum et al. use cell lines from multiple cancer types, making comparisons more difficult, since absolute segregation error frequencies may vary between CIN+ breast, cervical and colorectal cancers. One conclusion of our study [7] was that replication stress can lead to deviation of centromere numbers determined by centromeric FISH. Importantly we demonstrated this deviation in cell clones grown for multiple generations following induction of replication stress by depletion of PIGN, ZNF516 or MEX3C, demonstrating ongoing change in centromere number as observed in CIN+ cells [1,6]. Our findings that a high proportion of chromosome segregation errors in CIN+ colorectal cancer cells are pre-mitotic in origin, combined with multiple additional experiments demonstrating elevated replication stress in CIN+ compared to CINcolorectal cancer cell lines, led us to conclude that replication stress-induced chromosome missegregation is an important contributor to CIN in colorectal cancer [7]. In our opinion, data presented by Bakhoum et al. in this issue of Current Biology [1] and previously, demonstrating reduced Response to Bakhoum et al.