Extensive trimming of short single-stranded DNA oligonucleotides during replication-coupled gene editing in mammalian cells

Abstract

Through transfection of short single-stranded oligodeoxyribonucleotides (ssODNs) small genomic alterations can be introduced into mammalian cells with high precision. ssODNs integrate into the genome during DNA replication, but the resulting heteroduplex is prone to detection by DNA mismatch repair (MMR), which prevents effective gene modification. We have previously demonstrated that the suppressive action of MMR can be avoided when the mismatching nucleotide in the ssODN is a locked nucleic acid (LNA). Here, we reveal that LNA-modified ssODNs (LMOs) are not integrated as intact entities in mammalian cells, but are severely truncated before and after target hybridization. We found that single additional (non-LNA-modified) mutations in the 5'-arm of LMOs influenced targeting efficiencies negatively and activated the MMR pathway. In contrast, additional mutations in the 3'-arm did not affect targeting efficiencies and were not subject to MMR. Even more strikingly, homology in the 3'-arm was largely dispensable for effective targeting, suggestive for extensive 3'-end trimming. We propose a refined model for LMO-directed gene modification in mammalian cells that includes LMO degradation.

Fig 1. Differential effects of single AMMs in the 5’- and 3’-arm of LMOs.

(A) Schematic representation of the stably integrated neo reporter in mESCs and sequences of LMOs that generate a functional neo start codon and contain single AMMs. Blue capital characters indicate mismatches with respect to the reporter, underlined characters indicate LNA modifications. (B) Relative neo targeting efficiency of LMOs with single AMMs at the indicated positions in MMR⁺ and MMR^- cells. Bars indicate the mean with SD of at least three experiments. Significance was determined using a corrected two-way ANOVA. (C, D) Proportion of G418^R colonies from MMR^- (C) and MMR⁺ (D) cells in which the indicated AMM was integrated as determined by Sanger sequencing. Error bars represent 95% confidence interval.

Fig 2. Differential effects of multiple AMMs in the 5’- and 3’-arm of LMOs.

(A) Sequence of LMOs with three AMMs in the 5’- or 3’-arm. Blue capital characters indicate mismatches with respect to the reporter, underlined characters indicate LNA modifications. (B) Relative neo targeting efficiency of LMOs with three AMMs in MMR^- and MMR⁺ cells. Bars indicate the mean with SD of three experiments. Significance was determined using a corrected multiple t-test. (C) Sequence of LMOs with consecutive tracts of nine identical bases in the 3’-arm. (D) Targeting efficiency at the Gfp reporter of LMOs containing mononucleotide tracts in the 3’-arm.

Fig 3. The 5’-arm of LMOs is degraded during the process of targeting through endonuclease activity.

(A) Sequence of 5’-arm modified LMOs with and without AMMp5. Blue capital characters indicate mismatches with respect to the reporter, underlined characters indicate LNA modifications, Acr indicates modification at 5’-terminus with 6-chloro-2-methoxyacridine. (B, C) Relative neo targeting efficiency of 5’-arm modified control LMOs (B) and AMMp5 LMOs (C) in MMR^- and MMR⁺ cells normalized to control LMOs. Bars indicate the mean with SD of at least three experiments. Significance was determined using a corrected two-way ANOVA. (D, E) Proportion of G418^R colonies from MMR^- (D) and MMR⁺ (E) cells in which AMMp5 was integrated as determined by Sanger sequencing. (F) Sequence of AMMp5 LMOs with PTO modifications in the 5’-arm. Asterisks indicate PTO bonds. (G) Relative neo targeting efficiency of PTO-modified LMOs with AMMp5 in MMR^- cells. Bars indicate the mean with SD of three experiments. Significance was determined using a corrected one-way ANOVA. (H) Proportion of G418^R colonies from MMR^- cells in which AMMp5 was integrated as determined by Sanger sequencing. Error bars for AMMp5 integration rate (C, D, G) represent 95% confidence interval.

Fig 4. Suppression of 3’-5’ LMO degradation increases 3’-arm integration but does not increase targeting efficiencies.

(A) Sequences of control and AMMp17 LMOs with a second LNA modification in the 3’-arm. Blue capital characters indicate mismatches with respect to the reporter, underlined characters indicate LNA modifications. (B, C) Relative neo targeting efficiency of 3’-arm modified control LMOs (B) and AMMp17 LMOs (C) in MMR^- and MMR⁺ cells normalized to control LMO. Bars indicate the mean with SD of at least four experiments. ND indicates not determined. Significance was determined using a corrected two-way ANOVA. (D, E) Proportion of G418^R colonies from MMR^- (D) and MMR⁺ (E) cells in which AMMp17 was integrated as determined by Sanger sequencing. Error bars represent 95% confidence interval.

Fig 5. LMOs with a centrally placed LNA-protected mismatch provide the optimal targeting efficiency.

(A, B) Sequence (A) and neo targeting efficiency in MMR⁺ cells (B) of LMOs with repositioned LNA-protected mismatch. Blue underlined capital characters indicate LNA-protected mismatch. Bars indicate the mean with SD of four experiments. Significance was determined using a corrected one-way ANOVA.

Fig 6. Degradation steps during LMO-directed gene modification.

(A) After transfection to mammalian cells, LMOs are partially degraded by 3’-5’ exonucleases; the centrally positioned LNA provides protection from degradation. (B) The 3’-arm truncated LMOs anneal to their ssDNA target site during DNA replication. (C) DNA MMR scans DNA for mismatches and removes the nascent strand carrying a non-matching base. LNA-protected mismatches evade MMR. (D, E) After annealing, the 5’-arm of the LMO is removed through endonucleolytic activity before it becomes fully integrated (E) into the newly synthesized DNA strand.