Final Report Summary - RNF4 IN THE DDR (Identifying the targets and mechanism of action of the SUMO targeted ubiquitin ligase RNF4 in response to distinct DNA lesions)
Identifying the targets and mechanism of action of the SUMO targeted ubiquitin ligase RNF4 in response to distinct DNA lesions
Background and Aims
The Ubiquitin (UB) and SUMO modification pathways have been shown to be important regulators of genome stability. However, despite their significance, we do not have a detailed understanding of which repair proteins they modify, nor how they promote DNA repair. This proposal aimed to address this by studying the human protein RNF4 which sits at the interface between the UB and SUMO pathways. RNF4 is a SUMO targeted ubiquitin ligase (STUbL) which binds poly-SUMOylated chains and ubiquitylates them. Moreover, polySUMO chains have been shown to trigger dimerization of RNF4 thereby stimulating RNF4 and enabling it to ubiquitinate substrates. To date, very few RNF4 targets are known. We attempted to use a novel approach enriching for hybrid SUMO-UB chains followed by quantitative proteomics to identify the targets of RNF4-mediated ubiquitylation. As DNA repair factors are targeted in cancer therapy, this work has the potential to lead to the development of novel therapeutics.
Aim 1. To identify the SUMO modified proteins RNF4 ubiquitylates in response to DNA damage
Aim2. Establish the role of such SUMO and UB modified proteins at distinct DNA lesions
Aim 3. Determine if the hybrid SUMO-UB modification of these proteins is required for efficient DNA repair.
Results and conclusions
The overall aim was to understand how RNF4 acts to promote the repair of different types of DNA damage. RNF4 depleted (siRNA) cells are sensitive to Ionizing radiation (IR) and hydroxyurea (HU). These induce distinct lesions: IR causes DNA SSBs and DSBs whereas HU stalls replication forks, exposing large stretches of ssDNA. However, RNA interference (using siRNA) just reduces and does not completely remove target protein from the cell. Therefore to generate clean reagents, completely lacking RNF4, we used CRISPR/Cas genome editing technology to delete RNF4 in two different human cell lines: HCT116 and U2OS. The first step was to follow the ability of these cells to proliferate in response to DNA damage drugs using clonogenic cell survival assays. Indeed, the HCT116 RNF4 null cells were found to be sensitive to HU, IR, Cisplatin and mitomycin C, however the defects observed were less pronounced than in chicken DT40 RNF4 null cells or in human RNF4 (siRNA). So far we have tested the response of the U2OS RNF4 null cells to HU and found the defect to be more comparable to chicken DT40 RNF4 null cells and human RNF4 (siRNA). The difference between the HCT116 and U2OS cells may be due to the fact that U2OS cells have wild type p53 whereas HCT116 do not.
To isolate the relevant RNF4 ubiquitylated target proteins I had proposed using pull-down purifications with the N-terminal fragment of RAP80, which contains tandem SUMO Interaction Motifs (SIMs) and Ubiquitin Interaction Motifs (UIMs). This fragment has been found to bind hybrid SUMO-UB chains with 80-fold higher affinity than either SUMO or UB chains alone. The idea behind the experiment was based on the published observation that RNF4 was critical for the recruitment of RAP80 and BRCA1 to sites of DNA damage. Using antibodies against RAP80 and BRCA1 we followed their recruitment to sites of DNA damage induced by laser lines and marked by yH2Ax. However, we found that in the absence of RNF4, RAP80 and BRCA1 are still recruited and co-localise with yH2Ax at sites of DNA damage, indicating that RNF4 activity is not strictly necessary for RAP80 binding. This suggests that our RAP80 pull-down experiments comparing wild type and RNF4 null cells would not identify RNF4 relevant substrates. In the absence of RNF4, it is possible that alternative factors gain access to the DNA lesion to enable back-up repair pathways. For example another STUbL may take over the synthesis of hybrid chains at sites of damage or it may be that hybrid chains are not absolutely required for the recruitment of the RAP80/BRCA1 complex. We have however confirmed that RAP80 and BRCA1 complexes do show greater preference for binding polySUMO2-polyUB chains over K63-poly UB, polySUMO2-mono UB, or polySUMO2. Furthermore, if we immunoprecipitate RAP80 and then treat with SUMO proteases we see loss of high molecular weight ubiquitin conjugates suggesting that RAP80 does interact with SUMOylation-dependent ubiquitin conjugates. Therefore the RAP80-resin we generated may be capable of identifying proteins modified with hybrid chains. However, unfortunately, they will not be solely RNF4 dependent substrates.
The original plan was that proteins identified through RAP-80 mediated pull-down would be further studied to test whether they were recruited to and required for the repair of distinct DNA lesions. As we have been unable to identify RNF4-dependent substrates this has not been possible. We had additionally aimed to create reporter assays to follow DNA repair. One such novel DNA repair reporter involved monitoring repair of protein-DNA adducts using the Flp-nick assay. The Flp-nick system is based on the expression of a mutant FLP recombinase which initiates cleavage at a unique FRT site but which cannot complete its reaction cycle, leaving the FLP protein remaining covalently bound to the DNA adjacent to a single-strand DNA nick. This is a very unique DNA lesion that mimics the damage caused by the anti-cancer drug camptothecin (CPT). Importantly, this would only be relevant to study if RNF4 depleted cells are themselves sensitive to CPT. However, we have found that RNF4 null cells are not sensitive to CPT.
Our work has highlighted unexpected complexity in the involvement of RNF4 in DNA repair. The role of RNF4 is likely influenced by other DNA damage signalling pathways such as p53. In the absence of RNF4, other STUbLs may take over the synthesis of hybrid SUMO-UB chains at sites of damage. Finally the relative sensitivity of RNF4 to distinct types of DNA damage (very sensitive to HU but not CPT) may point to a specific role for RNF4 in replication fork stability, which we are currently investigating. Indeed, replication stress can induce DNA damage and genome stability, sometimes resulting in cancer and premature aging. If we uncover the role of RNF4 in DNA replication it could aid our understanding of the connection between replication stress and disease.
Background and Aims
The Ubiquitin (UB) and SUMO modification pathways have been shown to be important regulators of genome stability. However, despite their significance, we do not have a detailed understanding of which repair proteins they modify, nor how they promote DNA repair. This proposal aimed to address this by studying the human protein RNF4 which sits at the interface between the UB and SUMO pathways. RNF4 is a SUMO targeted ubiquitin ligase (STUbL) which binds poly-SUMOylated chains and ubiquitylates them. Moreover, polySUMO chains have been shown to trigger dimerization of RNF4 thereby stimulating RNF4 and enabling it to ubiquitinate substrates. To date, very few RNF4 targets are known. We attempted to use a novel approach enriching for hybrid SUMO-UB chains followed by quantitative proteomics to identify the targets of RNF4-mediated ubiquitylation. As DNA repair factors are targeted in cancer therapy, this work has the potential to lead to the development of novel therapeutics.
Aim 1. To identify the SUMO modified proteins RNF4 ubiquitylates in response to DNA damage
Aim2. Establish the role of such SUMO and UB modified proteins at distinct DNA lesions
Aim 3. Determine if the hybrid SUMO-UB modification of these proteins is required for efficient DNA repair.
Results and conclusions
The overall aim was to understand how RNF4 acts to promote the repair of different types of DNA damage. RNF4 depleted (siRNA) cells are sensitive to Ionizing radiation (IR) and hydroxyurea (HU). These induce distinct lesions: IR causes DNA SSBs and DSBs whereas HU stalls replication forks, exposing large stretches of ssDNA. However, RNA interference (using siRNA) just reduces and does not completely remove target protein from the cell. Therefore to generate clean reagents, completely lacking RNF4, we used CRISPR/Cas genome editing technology to delete RNF4 in two different human cell lines: HCT116 and U2OS. The first step was to follow the ability of these cells to proliferate in response to DNA damage drugs using clonogenic cell survival assays. Indeed, the HCT116 RNF4 null cells were found to be sensitive to HU, IR, Cisplatin and mitomycin C, however the defects observed were less pronounced than in chicken DT40 RNF4 null cells or in human RNF4 (siRNA). So far we have tested the response of the U2OS RNF4 null cells to HU and found the defect to be more comparable to chicken DT40 RNF4 null cells and human RNF4 (siRNA). The difference between the HCT116 and U2OS cells may be due to the fact that U2OS cells have wild type p53 whereas HCT116 do not.
To isolate the relevant RNF4 ubiquitylated target proteins I had proposed using pull-down purifications with the N-terminal fragment of RAP80, which contains tandem SUMO Interaction Motifs (SIMs) and Ubiquitin Interaction Motifs (UIMs). This fragment has been found to bind hybrid SUMO-UB chains with 80-fold higher affinity than either SUMO or UB chains alone. The idea behind the experiment was based on the published observation that RNF4 was critical for the recruitment of RAP80 and BRCA1 to sites of DNA damage. Using antibodies against RAP80 and BRCA1 we followed their recruitment to sites of DNA damage induced by laser lines and marked by yH2Ax. However, we found that in the absence of RNF4, RAP80 and BRCA1 are still recruited and co-localise with yH2Ax at sites of DNA damage, indicating that RNF4 activity is not strictly necessary for RAP80 binding. This suggests that our RAP80 pull-down experiments comparing wild type and RNF4 null cells would not identify RNF4 relevant substrates. In the absence of RNF4, it is possible that alternative factors gain access to the DNA lesion to enable back-up repair pathways. For example another STUbL may take over the synthesis of hybrid chains at sites of damage or it may be that hybrid chains are not absolutely required for the recruitment of the RAP80/BRCA1 complex. We have however confirmed that RAP80 and BRCA1 complexes do show greater preference for binding polySUMO2-polyUB chains over K63-poly UB, polySUMO2-mono UB, or polySUMO2. Furthermore, if we immunoprecipitate RAP80 and then treat with SUMO proteases we see loss of high molecular weight ubiquitin conjugates suggesting that RAP80 does interact with SUMOylation-dependent ubiquitin conjugates. Therefore the RAP80-resin we generated may be capable of identifying proteins modified with hybrid chains. However, unfortunately, they will not be solely RNF4 dependent substrates.
The original plan was that proteins identified through RAP-80 mediated pull-down would be further studied to test whether they were recruited to and required for the repair of distinct DNA lesions. As we have been unable to identify RNF4-dependent substrates this has not been possible. We had additionally aimed to create reporter assays to follow DNA repair. One such novel DNA repair reporter involved monitoring repair of protein-DNA adducts using the Flp-nick assay. The Flp-nick system is based on the expression of a mutant FLP recombinase which initiates cleavage at a unique FRT site but which cannot complete its reaction cycle, leaving the FLP protein remaining covalently bound to the DNA adjacent to a single-strand DNA nick. This is a very unique DNA lesion that mimics the damage caused by the anti-cancer drug camptothecin (CPT). Importantly, this would only be relevant to study if RNF4 depleted cells are themselves sensitive to CPT. However, we have found that RNF4 null cells are not sensitive to CPT.
Our work has highlighted unexpected complexity in the involvement of RNF4 in DNA repair. The role of RNF4 is likely influenced by other DNA damage signalling pathways such as p53. In the absence of RNF4, other STUbLs may take over the synthesis of hybrid SUMO-UB chains at sites of damage. Finally the relative sensitivity of RNF4 to distinct types of DNA damage (very sensitive to HU but not CPT) may point to a specific role for RNF4 in replication fork stability, which we are currently investigating. Indeed, replication stress can induce DNA damage and genome stability, sometimes resulting in cancer and premature aging. If we uncover the role of RNF4 in DNA replication it could aid our understanding of the connection between replication stress and disease.