Cryo-EM sheds light on how cells repair DNA damage
A new study has revealed the sequence of key events in an important DNA damage pathway.
Cryo-EM structure of human FANCD2/FANCI complex showing the Tower domain of FANCD2 and main body
The work on the Fanconi anemia (FA) DNA repair pathway, from the labs of Dr. Martin Cohn in the department and Dr. Catherine Vénien-Bryan at the Pierre and Marie Curie University in Paris, has been published in Nature Communications (1). It explains how mutations in a key protein in the FA pathway disrupt the repair process, leading to devastating consequences for patients with the FA disorder.
Cryo-EM structure of human FANCD2/FANCI complex
Mammalian cells need efficient ways of repairing DNA damage, which can occur at a frequently of up to 1 million individual lesions per cell per day and threatens the integrity of the genome. One type of lesion is an interstrand DNA crosslink (ICL), which forms between the two strands of the DNA helix. ICLs are repaired by the FA repair pathway, largely during DNA replication as the replication fork passes along the DNA and arrests at the ICL. Many proteins are involved in the pathway, and the FANCD2/FANCI protein complex is central to it.
Some of the details of the pathway have been described, such as the recruitment of the FANCD2/FANCI complex to the ICL, which is critical for repair. Transfer of ubiquitin onto FANCD2 and FANCI, by a process of monoubiquitination, is known to be a key step, likely because it recruits nucleases to the complex that are needed for repair. However, exactly how this process takes place - in particular, whether the complex is recruited to the ICL before or after monoubiquitination - is unclear.
EGFP-FANCD2 is recruited to the DNA interstrand crosslinks (the stripes). Scale bar: 20 μm
Martin Cohn and his students - Eric Liang, who carried out much of the work, and David Lopez-Martinez - wanted to understand this better. They started by determining the structure of the full-length human FANCD2/FANCI complex using cryo-electron microscopy (cryo-EM) in collaboration with Dr Vénien-Bryan. As the complex is large, it is difficult to crystallise, and cryo-EM was more likely than crystallography to yield results as well as provide insight into a more native-like form of the complex.
The researchers found that the human FANCD2/FANCI complex looked similar to the mouse FANCD2/FANCI complex, whose crystal structure had been previously described, but with one striking difference. The EM structure revealed a protruding domain in the human complex that appeared horizontal in the mouse structure. The group named this new domain, part of FANCD2, the Tower domain.
Intrigued by the finding, they investigated what the Tower domain might be doing. Using live-cell imaging set up at the advanced imaging facility Micron, they showed that full-length FANCD2, but not FANCD2 with the Tower domain deleted, was recruited to chromatin when ICLs were introduced into HeLa cells.
They then tested the Tower domain further by carrying out in vitro DNA binding experiments with the FANCD2/FANCI complex and synthesized DNA molecules. The complex bound to a DNA structure mimicking a replication fork, but could do so only weakly when the Tower domain was deleted. They found that the full-length complex bound more strongly when they used DNA with a fork that contained an ICL.
Alanine substitution of the positively charged residues in the Tower domain of FANCD2 (third row down) diminishes its recruitment to the DNA interstrand crosslinks (the stripes). Labelled wild-type FANCD2 and UHRF1 (a sensor for ICLs) are also shown. Scale bars: 20μm (Click to enlarge)
As Eric Liang explains, these results confirm the importance of the Tower domain: 'We used to think that FANCD2/FANCI is recruited to ICL after monoubiquitination, but we have now shown that FANCD2/FANCI by itself has a high affinity for the replication fork and the crosslink. We demonstrated that the Tower domain is required for docking the complex to DNA.'
Next, the group investigated at what step ubiquitination takes place by reconstituting this reaction in vitro using recombinant proteins in the presence or absence of DNA. 'We showed that if the FANCD2/FANCI complex does not bind to DNA, there is very weak ubiquitination,' says Eric Liang. 'So it is only after the FANCD2/FANCI complex is recruited to DNA that it is ubiquitinated, and recruitment requires the Tower domain.'
Given these results, the group was interested to know whether there might be a link between the Tower domain and known genetic mutations in FA patients. 'We started looking through the database for mutations in the domain and found that several disease-causing mutations are located here,' says Dr. Cohn. 'Now that we know the structure, we can say why these known deletions cause the disease.'
Having identified the sequence of events necessary to activate the FANCD2/FANCI complex, the researchers looked for key residues in the Tower domain that might be important for FANCD2 binding to DNA. This led them to two patches of positively charged residues, which they tested by mutating to neutral residues followed by in vivo and in vitro analyses. They found that these residues are responsible for protein-DNA electrostatic interactions and showed that they are needed for recruitment of the FANCD2/FANCI complex to DNA and for its ubiquitination.
Why is FANCD2 ubiquitinated only after it has been recruited to DNA? Dr. Cohn speculates that it may be because the complex undergoes a conformational change upon binding to DNA, opening up the ubiquitination site. 'High-resolution cryo-EM of the individual stages of recruitment and ubiquitination will provide insight into this,' he comments. 'The current resolution of our structures is about 20 Å, but with the new Titan Krios cryo-EM at the Diamond Light Source, we expect to increase the resolution to higher than 5 Å.'
Model showing how the FANCD2/FANCI complex is recruited to a crosslinked replication fork, triggering the subsequent monoubiquitination of the complex (Click to enlarge)
A high-resolution structure will also help to probe all the interactions that hold the FANCD2/FANCI complex onto the DNA; since the interaction cannot be sequence-specific, there are likely to be many contacts that stabilize it. 'The Tower domain is important, but not sufficient, for DNA binding,' comments Eric Liang. 'We think that it may embrace the DNA and in this way help to stabilize the interaction. We don't have any data yet to suggest this, but hope to confirm it with a higher resolution structure.'
One line of study that the group is pursuing together with the Micron facility, which will give further insight into the mechanisms involved, is single-molecule imaging, says Dr. Cohn. 'Live-cell imaging has been very powerful for our research to study single cells, but it looks at an average of many ICLs in the cell. With single-molecule imaging, we can find out information about timing - we can look at one particular crosslink and observe which proteins are recruited and when. We'll be able to answer big questions about what is happening mechanistically.'