L3MBTL2 orchestrates ubiquitin signalling by dictating the sequential recruitment of RNF8 and RNF168 after DNA damage

Abstract

Cells respond to cytotoxic DNA double-strand breaks (DSBs) by recruiting DNA repair proteins to the damaged site. This recruitment is dependent on ubiquitylation of adjacent chromatin areas by E3 ubiquitin ligases such as RNF8 and RNF168, which are recruited sequentially to the DSBs. However, it is unclear what dictates the sequential order and recruits RNF168 to the DNA lesion. Here, we reveal that L3MBTL2 (lethal(3)malignant brain tumour-like protein 2) is the missing link between RNF8 and RNF168. We found that L3MBTL2 is recruited by MDC1 and subsequently ubiquitylated by RNF8. Ubiquitylated L3MBTL2, in turn, facilitates recruitment of RNF168 to the DNA lesion and promotes DNA DSB repair. These results identify L3MBTL2 as a key target of RNF8 following DNA damage and demonstrates how the DNA damage response pathway is orchestrated by ubiquitin signalling.

Figure 1. ATM-mediated phosphorylation of L3MBTL2 recruits it to the double strand break site

(a–b) L3MBTL2 localizes to the DSB site in U2OS I-SceI cells where one DSB is induced per cell using triamcoline acetonide (red). This L3MBTL2 focus (green) overlaps with γH2AX (blue). The yellow box locates the size of the cut. Inset, schema of the U2OS I-SceI reporter system. (b) Quantification of U2OS I-SceI cells with both H2AX and L3MBTL2 foci. (c–d) L3MBTL2 forms radiation-induced puncta that overlaps with γ-H2AX in U2OS cells. (d) Quantification of the indicated foci with and without irradiation in U2OS cells. (e–g) ATM phosphorylates L3MBTL2 at the S335 residue following DNA damage. ATM inhibitor (ATMi) KU55933 was used as control. Protein sequence of L3MBTL2 is shown in f. Putative ATM phosphorylation sites are underlined and bolded. (h–j) Phosphorylation of L3MBTL2 is required for recruitment of L3MBTL2 to double strand break sites in U2OS cells. Mutation of the phosphorylation site on L3MBTL2 abrogates its localization to DSB sites. U2OS cells in which endogenous L3MBTL2 had been knocked out were transfected with the indicated GFP-tagged plasmids. Cells were exposed to 2GY IR and stained for γ-H2AX foci (red). Nucleus is stained with DAPI (blue). (i) Quantification of L3MBTL2 foci in U2OS cells expressing the indicated plasmids with and without irradiation. Data are represented as the mean ± SEM of n = 3 independent experiments. Circles depict individual data points. Statistical significance was calculated using 2-way ANOVA. **p= 0.000000005 for L3MBTL2 WT vs S335A with IR treatment. Source data are provided in Supplementary Table 1. (j) Expression level of the L3MBTL2 constructs in L3MBTL2 knockout and control U2OS cells. L3MBTL2 was knocked out in U2OS cells using CRISPR. GFP-tagged WT or S335A mutant of L3MBTL2 was transfected into these cells. Shown is the comparable expression level of L3MBTL2 in these cells. Representative images of three independent experiments are shown in a, c and h. Scale bars, 10 µm. Representative western blots in e, g, and j are provided from 3 biologically independent experiments. Unprocessed blots are provided in Supplementary Figure 5.

Figure 2. ATM-mediated phosphorylation of L3MBTL2 promotes its interaction with MDC1 and recruits it to the double strand break site

(a–c) MDC1 is required for L3MBTL2 recruitment to the DSB site. (a) Knockdown of MDC1 using siRNA inhibits L3MBTL2 (green) accumulation at DSB site in U2OS I-SceI cells. γ-H2AX (blue) was used as a positive control. The yellow box locates the site of the cut. (b) Knockdown efficiency of MDC1 with siRNA. (c) Quantification of U2OS I-SceI cells with L3MBTL2 foci with and without MDC1 knockdown using siRNA. Data are represented as the mean ± SEM of n= 3 independent experiments. Circles depict individual data points. Statistical significance was calculated using 1-way ANOVA. **p= 0.000002 for control siRNA vs MDC1 siRNA 1 and 2. (d) L3MBTL2 interacts with MDC1 following DNA damage. (e) DNA damage-induced phosphorylation of L3MBTL2 at S335 is required for its interaction with MDC1. (f) L3MBTL2 interacts with the FHA domain of MDC1 upon DNA damage. Deletion of the FHA domain (ΔFHA) abrogates the interaction between MDC1 and L3MBTL2. (g) The FHA domain of MDC1 is required for the interaction between MDC1 and L3MBTL2. (h) L3MBTL2 interacts with the FHA domain of MDC1 in vitro. GST-tagged constructs of MDC1 were incubated with lysates from cells expressing L3MBTL2 and treated with or without radiation. (i) L3MBTL2 interacts with the FHA domain of MDC1 in vitro following DNA damage. GST-tagged FHA domain of MDC1 was incubated with lysates from L3MBTL2 knockout cells expressing WT or phosphorylation mutant of L3MBTL2 (S335A) and treated with or without radiation. (j) The FHA domain of MDC1 directly interacts with phosphorylated L3MBTL2. The indicated peptides of L3MBTL2 (control and pS335) were incubated with GST-tagged FHA domain of MDC1. (k–m) Mutation in the FHA domain of MDC1 abrogates the interaction between MDC1 and L3MBTL2 both in vitro and in vivo. Representative images of three independent experiments are shown in a. Scale bars, 10 µm. Representative western blots in b, d–m are provided from 3 biologically independent experiments. Source data and unprocessed blots are provided in Supplementary Table 1 and Supplementary Figure 5, respectively.

Figure 3. L3MBTL2 recruits RNF168 to the double strand break site

(a–c) shRNA mediated knockdown of L3MBTL2 in U2OS cells prevents formation of RNF168 foci following DNA damage. Shown are (a) quantification and (b) representative images of cells with γ-H2AX, MDC1, RNF8, and RNF168 foci after 2GY irradiation. (c) Knockdown of L3MBTL2 using shRNA does not affect RNF168 protein level in U2OS cells. (d–f) Phosphorylation of L3MBTL2 is required for RNF168 foci formation following DNA damage. Shown are (d) quantification and (e) representative of γ-H2AX, MDC1, RNF8, and RNF168 foci after 2GY irradiation in L3MBTL2 knockout U2OS cells expressing wild-type (WT) or the phosphorylation mutant of L3MBTL2 (S335A). (f) Western blot showing the expression level of WT and S335A mutant of L3MBTL2. (g) MDC1, RNF8 and RNF168 interact with endogenous L3MBTL2 following DNA damage. Cells were exposed to the indicated doses of irradiation. Lysates were collected after an hour and proteins interacting with L3MBTL2 were assessed. (h) DNA damage-induced interaction between RNF168 and L3MBTL2 is dependent on RNF8. Knockdown of RNF8 using shRNA does not affect the interaction between endogenous MDC1 and L3MBTL2. Cells were exposed to the indicated doses of irradiation. Lysates were collected after an hour and proteins interacting with L3MBTL2 were assessed. Data in a and d are represented as the mean ± SEM of n = 3 independent experiments. Circles depict individual data points. Statistical significance was calculated using 2-way ANOVA. *p<0.05, **p<0.01 (vector vs all other groups). Please refer to Supplementary Table 1 for exact p values in a and d. The experiments in b and e were repeated 3 independent times with similar results. Representative western blots in c and f to h are provided from 3 biologically independent experiments. Scale bars, 10 µm. Source data and unprocessed blots are provided in Supplementary Table 1 and Supplementary Figure 5, respectively.

Figure 4. RNF8 mediated K63 linked ubiquitylation of L3MBTL2 following DNA damage is critical for the interaction with UDM1 of RNF168 and subsequent histone ubiquitylation

(a–b) RNF8 ubiquitylates L3MBTL2 following DNA damage and forms K63 linkage. (a) RNF8 was knocked down using shRNA in MDA-MB-231 cells expressing L3MBTL2. Cells were subjected to control or 10GY irradiation. Lysates were collected after 1 hr. The interaction between the indicated proteins was analyzed. Blots were probed with the indicated antibodies. (b) MDA-MB-231 cells were transfected with the indicated constructs. Cells were exposed to the indicated doses of radiation and lysed after an hour. Immunoprecipitation was performed using nickel (His) beads. Blots were probed with the indicated antibodies. (c) RNF8 ubiquitylates L3MBTL2 in vitro. The indicated recombinant proteins along with their required cofactors were incubated at 30°C for 1 hour. The reaction was analyzed by western blot. Blots were probed with the indicated antibodies. (d) L3MBTL2 interacts with recombinant UDM1 domain of RNF168 in vitro in an RNF8 dependent manner. The indicated GST-tagged constructs of RNF168 were incubated with the reaction product of (c) with and without RNF8 for an hour. The result was analyzed by western blot and blots were probed with the indicated antibodies. (e) Mutations in UDM1 domain of RNF168 reduces its interaction with L3MBTL2. The reaction product from (c) was incubated with the indicated GST-tagged UDM1 constructs for an hour. The result was analyzed by western blot and blots were probed with the indicated antibodies. (f) L3MBTL2 interacts with UDM1 via K63-linked ubiquitin chains. The indicated recombinant proteins (without UDM1 GST) along with required cofactors were incubated at 30°C for 1 hour. The reaction product was incubated with GST-tagged UDM1 construct for an hour. The reaction was analyzed by western blot. Blots were probed with the indicated antibodies. Representative western blots are provided from 3 biologically independent experiments in a to f. Unprocessed blots are provided in Supplementary Figure 5.

Figure 5. DNA damage induced RNF8 mediated ubiquitylation of L3MBTL2 is critical for DNA DSB repair

(a) L3MBTL2 is ubiquitylated at K659 following DNA damage. The indicated plasmids and His-tagged wild-type ubiquitin (His WT Ub) were transfected into L3MBTL2 knockout cells. Lysates were acquired an hour following 10GY irradiation and subjected to immunoprecipitation. (b) RNF8 is required for the interaction between the UDM1 domain of RNF168 and L3MBTL2 in vitro. The indicated recombinant proteins were incubated with L3MBTL2 purified from L3MBTL2 knockout cells expressing the indicated plasmids. Immunoprecipitation was performed with GST-tagged UDM1 protein. (c) Ubiquitylation of L3MBTL2 is required for its binding with recombinant UDM1 domain of RNF168 in vitro. L3MBTL2 knockout cells expressing the indicated L3MBTL2 constructs were exposed to the indicated doses of irradiation, lysed after an hour. L3MBTL2 was purified from these lysates and incubated with GST-tagged UDM1 domain of RNF168 for an hour. (d) Ubiquitylation is important for the interaction between RNF168 and L3MBTL2 but not MDC1. (e) The ubiquitylation mutant (K659R) fails to form radiation-induced RNF168 foci. (f) Analysis of irradiation induced H2A ubiquitylation by RNF168 in chromatin fraction of L3MBTL2 knockout cells expressing the indicated plasmids. (g) Loss of L3MBTL2 sensitizes cells to irradiation. Survival assays of L3MBTL2 knockout MDA-MB-231 cells treated as indicated and exposed to the indicated doses of irradiation. Data are represented as the mean ± SEM of n = 3 independent experiments in e and g. Circles depict individual data points. Statistical significance was calculated using 2-way ANOVA. *p<0.05, **p<0.01 (vector vs all other groups) in e. Please refer to Supplementary Table 1 for exact p values. p=0.0001 for vector vs all other groups at each time point in g. Representative western blots are provided from 3 biologically independent experiments in a–d and f. Source data and unprocessed blots are provided in Supplementary Table 1 and Supplementary Figure 5, respectively.

Figure 6. L3MBTL2 induces class switch recombination and chromosome end fusion

(a–b) L3MBTL2 is required for class switch recombination. CH12F3-2a cells were treated as indicated. Class switch recombination was stimulated using ligand. IgM to IgA conversion was assessed by immunostaining followed by flow cytometry. Inset is shown a representative image of IgM to IgA conversion with treatment. (b) Representative western blot showing the knockdown efficiencies of shRNAs. (c–d) L3MBTL2 is required for telomere fusion. TRF2^F/F MEFs were treated as indicated. Cells were infected with Cre lentivirus, treated with KaryoMAX Colcemid, fixed, and stained with PNA FISH. Percent of cells with telomere fusion was assessed. Inset is shown a representative image of telomere fusion (white arrows). DNA was stained with DAPI (blue). (d) Representative western blot showing the knockdown efficiencies of shRNAs. (e) Model for RNF8, L3MBTL2 and RNF168 function in ubiquitin dependent signaling after DNA DSBs. We propose that ATM-mediated phosphorylation of L3MBTL2 promotes its interaction with MDC1 and recruits it to the DSB site. It is subsequently ubiquitylated by RNF8 which recruits RNF168 to the damage site. RNF168, in turn, monoubiquitylates H2A-type histones to amplify the DNA damage response and recruit downstream DNA repair proteins for proper DSB signaling. Data are represented as the mean ± SEM of n = 4 independent experiments for (a). In (c) data from two independent experiments are shown with a line indicating the mean. Circles depict individual data points. Scale bars, 10 µm. Statistical significance was calculated using 1-way ANOVA in a (p=0.0001 for all groups vs control shRNA). Representative western blots in b and d are from 4 and 2 biologically independent experiments, respectively. Source data and unprocessed blots are provided in Supplementary Table 1 and Supplementary Figure 5, respectively.