LEGENDS TO THE SUPPLEMENTARYFIGURES

Supplementary figure S1: EGFP-based test system to monitor recombination

(A)MLLbcr (or other) spacer sequences were inserted between the two mutated genes encoding enhanced green fluorescent protein (EGFP). In the HR-EGFP, part of the chromophore region is exchanged by the I-SceI recognition site (Δ), in the 3’-EGFP the start-codon is replaced by stop-codons. A CMV-promoter is inserted upstream of the 5’end of the HR-EGFP. Selection of cells with chromosomally integrated spacer-construct is enabled by puromycin-resistance cassette (PURO).

Upon spontaneous or induced breakage within the spacer sequence or after I-SceI incision within HR-EGFP, the two dysfunctional EGFPs can recombine to reconstitute functional wild-type EGFP. The CMV-promoter then drives transcription of the reconstituted wild-type EGFP, thus enabling detection of recombination events by quantification of green fluorescent cells by FACS analysis.

(B) In T47D cells knockdown of Rad51 decreased the frequency of aphidicolin-induced recombination of the transiently-introduced MLLbcr-construct.Normalized mean values ±SD (n=8) are presented (* p<0.05).

Supplementary figure S2: Cellular responses to replication-stress inducing treatment

(A) In WTK1 and HeLa cells aphidicolin treatment increased the percentage of cells in G1/S while decreasing the percentage of cells in G2 (PI staining). (B) Aphidicolin treatment also increased the percentage of apoptotic/dying (Sub-G1) WTK1 and HeLa cells. (C)A trend towards increased early apoptosis was observed in HeLa cells treated with aphidicolin (AnnexinV/PI double staining). This non-significant increase was diminished with 25µM zVAD-fmk co-treatment. (D) Aphidicolin treatment increased the phosphorylation of Chk1 on Ser345 in HeLa cells. (E) In HeLa cells aphidicolin treatment increased the percentage of cells entering senescence (β-Gal staining). Co-treatment with zVAD-fmk had no effect on the number of senescent cells. Representative pictures of β-Gal staining (blue) are presented.(F)Treatment with zVAD-fmk had no effect on the recombination frequency of transiently introduced MLLbcr-construct in aphidicolin and mock-treated WTK1 cells. (G) Treatment with 5-fluorouracil in the presence of zVAD-fmk increased the recombination frequency of transiently introduced MLLbcr but not of SV40 construct in WTK1 cells.

Normalized mean values ±SD (n≥6) are presented (* p<0.05, ** p<0.01, *** p<0.001).

Supplementary figure S3: EndoG knockdown and characterization of the repair pathways of the stably integrated MLLbcr-construct

In T47D and HeLa cells transient EndoG knockdown had no statistically significant effect on (A) the cell cycle distribution (PI staining) or (B) cell death (Sub-G1) compared to the mock-transfected cells.(C)HeLa/MLL cells with stably integrated MLLbcr-construct were treated with aphidicolin to induce recombination of the construct, yielding a wild-type EGFP gene and thus green fluorescence of the cells that underwent recombination. The EGFP-sorted cells, cells that did not underwent recombination (white cells) and non-treated cells were compared based on 2 different PCRs to characterize the repair products after aphidicolin treatment. PCR-1 detects only initialHR-EGFPwith intact I-SceI cleavage site which is present in non-treated cells, to a lesser extend in white cells (possibly due to NHEJ repair) but completely absent in EGFP-sorted cells, suggesting loss of the I-SceI cleavage site by recombination of the HR-EGFP. PCR-2 detects spacer-containing fragment of the MLLbcr-constructwhich ends downstream of the 3’-EGFP. It is present in non-treated cells, white cells and also EGFP-sorted cells excluding that the latter ones underwent non-conservative recombination or single-strand annealing after aphidicolin-induced breakage. Genomic MLLintr20 was used as loading. (D)EndoG knockdown by siRNA (siENDOG) decreased aphidicolin-induced recombination frequencies of both transiently introduced MLLbcr-construct in T47Dand stably integrated MLLbcr-construct in HeLa/MLL cells.

Normalized mean values ±SD (n≥4) are presented.

Supplementary figure S4: EndoG knockdown and cleavage of the chromosomal MLLbcr

(A)EndoG knockdown (siENDOG) confirmed the specificity of EndoG signal (red) in immunofluorescence microscopic images. Nuclei were stained with DAPI (blue). (B)Western blot showing increased EndoG protein content in total cellular lysates of HeLa after etoposide treatment. (C)PCR analysis of the genomic DNA showing that etoposide treatment reduced the intensity of the MLLbcr-specific band, which was partially prevented by EndoG knockdown in HeLa cells. (D)PCR analysis of the genomic DNA showingthat zVAD-fmk treatmentmoderately reduced the intensity ofMLLbcr-specific band in HeLa. Concomitant aphidicolin treatment leaded to further decrease of the band intensity.

Supplementary figureS5: RNF20 knockdown andMLLbcr recombination

(A)RNF20 knockdown by siRNF20-3 and siRNF20-5as compared to control siRNA in HeLa/MLL cells was verified by qRT-PCR. (B)Silencing of RNF20by both siRNAs decreased the recombination frequency of stably integrated MLLbcr-construct in aphidicolin-treated HeLa/MLL cells leading to loss of recombination induction by aphidicolin.Normalized mean values ±SD (n=6) are presented (*p<0.05).