Trp1-Cre control-stained RPE. The blue stain on the RPE indicates Cre activity in the nuclei of individual cells. The black ovals on the RPE outline areas of the RPE where Cre was not active, as indicated by lack of staining. The inset (outlined in red) depicts an HR event (brown melanin in cytoplasm) in which Cre has acted on the reporter allele (blue nucleus) (a) and an HR event (brown melanin in cytoplasm) in which Cre has not acted on the reporter allele (clear nuclei) (b).

Conditional loss of null alleles correlates with changes in HR frequency. (A) The conditional loss of Blm, as measured by the percentage of blue-stained RPEs (Cre activity), positively correlates with an increase in HR frequency (eye spots) (Spearman r = 0.6667; P < 0.0001). (B) The conditional loss of full-length Brca1, as measured by the percentage of blue-stained RPEs (Cre activity), positively correlates with a decrease in HR frequency (eye spots) (Spearman r = −0.388; P = 0.0375). (A, B) The dotted gray line emphasizes the average frequency of HR events per RPE (∼5).

Loss of Blm or full-length Brca1 results in an increase or decrease in HR frequency, respectively. (A, B) Eye spots with blue nuclei from RPEs exhibiting ≥80% blue staining (Cre activity) were analyzed for HR frequency. Conditional loss of Blm results in a significant increase (P < 0.0001; Mann-Whitney test) of HR (A), and conditional loss of full-length Brca1 results in a significant decrease (P < 0.0001; Mann-Whitney test) of HR compared to that of the combined conditional (CC) control (B) (see the text for details). Data are represented as Tukey box and whisker plots.

Loss of Blm or full-length Brca1 effects replication-tied HR events. (A and B) Eye spots with blue nuclei from RPEs exhibiting ≥80% blue staining (Cre activity) were analyzed for the occurrence of single-cell versus multicell HR events. Loss of Blm leads to a significant increase (P = 0.0043; Fisher exact test) in the percentage of multicell HR events compared to that of the CC (combined conditional) control (A), and the absence of full-length Brca1 leads to a significant decrease (P = 0.0076; Fisher exact test) (B). Data are represented as the percentages of single-cell HR events (open bars) and multicell HR events (gray bars) for each genotype.

Loss of Blm increases HR frequency early during mouse embryonic development. Multicell events from RPEs with ≥80% blue staining (Cre activity) were analyzed for the positional effect of HR frequency. A position of 0 is equivalent to the optic nerve, compared to a position of 1.0, which is equivalent to the outer edge of the RPE. A significant increase of multicell HR events (P = 0.015; Fisher exact test) was detected in the region of the developing RPE, corresponding to the interval 0.21 to 0.40 of Blm null mice compared to the CC (combined conditional) control. Data are represented as fold changes of Blm null HR over CC control HR.

Two-cohort breeding strategy used for each gene of interest (G.O.I.). Mice in the constitutive cohorts were heterozygous for a targeted null mutation (neo disrupted) of the G.O.I. and were homozygous for the Trp1-Cre transgene. Mice in the conditional cohorts were homozygous both for a targeted floxed (or conditional) allele of the G.O.I. (usually loxP sites flanking one or more exons of the G.O.I. [exon “x”]) and for the RC::PFwe β-galactosidase reporter. Noted are nuclear localized β-galactosidase (nLacZ), loxP sites (black pentagons), and transcription (arrows above the schematics).

Haploinsufficiency of either Blm or full-length Brca1 does not alter HR frequency. Eye spots with blue nuclei from RPEs exhibiting ≥80% blue staining (Cre activity) were analyzed for HR frequency. No significant difference (P = 0.371; Kruskal-Wallis test) was detected when either Blm or Brca1 conditional heterozygotes were compared to controls. Data are represented as Tukey box and whisker plots.

Model for DNA replication restart in the context of a repeated DNA segment such as p^un in either a BLM-dependent or -independent manner. Presented is the occurrence of DNA damage that blocks lagging-strand DNA replication at the first repeat of DNA duplication, though similar models can be drawn for damage in the second repeated DNA element or that affect leading-strand synthesis. Complementary parental DNA strands (dark red and blue) and their nascently replicated leading and lagging strands (light blue and light red, respectively) are shown. DNA replication is indicated by a half arrowhead, DNA damage as a purple circle, and the repeated DNA elements by outlined black arrows. (A) Following damage-induced stalling of the replication fork, regression can occur (or template switching) by annealing the nascent strands to form a “chicken foot” structure possibly mediated by BLM (in parentheses, as other proteins such as Werner helicase can promote this activity). Using the nascent leading strand as a template, the lagging strand can now be extended beyond the DNA lesion found on the blue parental DNA strand. (B) Topological rearrangement. The “chicken foot” structure is topologically the same as a Holliday junction (HJ). (C, D) HJ branch migration (C), which can be mediated by BLM helicase, among others, can reverse the regressed fork, allowing replication restart (D). Of note, this mechanism would inhibit repeat duplication deletion, would not involve sister chromatid exchange, and would be limited by the amount of leading strand template available to bypass the lesion present on the parental strand. (E) Alternatively, the annealed nascent strands at the regressed replication fork would have a 3′-end overhanging tail that can be used for RAD51-dependent strand invasion promoted by BRCA1, resulting in strand displacement (a “D-loop”) and, consequently, the production of two additional HJs once the D-loop is captured by the nascent lagging strand. If the 3′-end tail is complementary to the repeated DNA element, then there is the chance that it can invade into the “correct repeat” or the “incorrect repeat.” (F) Irrespective of which repeat is invaded, BLM can dissolve a double HJ (DHJ) by migrating two HJs toward each other. Depending on which DHJ is dissolved, either a regressed replication fork is reformed that can be resolved by HJ branch migration or the replication fork is restored, without any possibility of either chromatid exchanges or repeat deletions in either case. (G) An alterative means to dissolve a DHJ is by resolution. The orientation of HJ resolution (cutting red HJ/red HJ or blue HJ/blue HJ versus cutting red HJ/blue HJ) will dictate whether or not resolution results in strand exchange. Irrespective of the resolution orientation, invasion of the 3′-end tail into the “correct” or “incorrect” repeat will dictate the possibility of repeat deletion (e.g., p^un reversion, in the case of our assay).