Donuts and Zippers for DNA Repair

How the protein RAD52 conducts homologous search and rescue of damaged DNA. And what we can do with that knowledge.

Our bodies are full of wonders. Despite being about fifty years into the age of molecular biology, we are still only scratching the surface of the complexity that lies within. Remember that when you hear that someone has found a gene "for" a disease. That gene and the protein it encodes have a lengthy and intricate story to tell about what it does in the normal organism, how it got to that point in evolution, and maybe a multitude of ways in which mutations might drive it off track. Linking it to one disease is only a small part of a much larger tale.

BRCA2 is an example of this: a pre-eminent gene "for" breast cancer. But what does it do? The point of its existence is hardly to make us miserable. No, it participates in DNA repair, and thus some mutations in its sequence, or full deletions, can leave cells open to a cascade of further mutations, which can lead to cancer.

Recent work has shown that BRCA2 is synthetically lethal with a better understood protein called RAD52. Being synthetically lethal (in mice in this case, where genetics can be done) means that a mutation or deletion of either one of the genes is not lethal by itself, but the combination of both *is lethal. Such an observation implies that the two proteins participate in the same process, at least partially. One can "cover" for defects in the other one, in a pinch. But take away both, and the organism can not survive.

So BRCA2 and RAD52 do similar things, which is helpful to know, since RAD52 is actually better understood. Yeast cells, which are even easier to study than mice, have RAD52 but no BRCA2, which seems to have arisen more recently in evolution. We'll get back to BRCA2 briefly below, but I will focus on RAD52- a protein that conducts that amazing process that is homology-driven DNA repair.

Homologous DNA repair happens when a mutation is detected, such as a double-strand break, a missing nucleotide (and consequent single strand break), or even just where a mis-matching nucleotide was inserted by the DNA polymerase by mistake. Perhaps the sequence goes CAGCT on one strand, and GGTTG on the other, antiparallel strand. There is a G:T pair in here that is an error- not part of the official Watson-Crick regime!

A mismatch happens (orange).

What was the right sequence? The sequence of this duplex does not tell you, unfortunately, anymore. In some cases, there are chemical marks like methylation that can be used to tell the difference between the newly synthesized strand and the other one that has been around the block, and this is used in some forms of post-replication repair. But if the mutation slipped through that process, or happened at a different time, again we would not know where to turn for the right way to correct such an error.

But there is one place to turn ... another copy of that DNA, if the organism has one. Diploid organisms always have another copy of every gene lying around, and other organisms, even bacteria, do as well, if they have already replicated their DNA but not divided the cell, quite yet. This is called homologous DNA repair, because the other copy is called a homolog of the first. Organisms have developed ornate methods of homology detection, strand invasion, copying, and repair to use the information from a good version to repair a defective one. And one of the key actors is RAD52, or its colleague, BRCA2.

Structure of RAD52. The ring is composed of multiple subunits of the same protein, each colored differently here. DNA, in single stranded form, can wind around the outside.

RAD52 looks like a big donut. Well, in relation to DNA, at least. The DNA doesn't thread though the middle, as it does in some other important machinery, but rather winds around the outside, in single-stranded form. That way, RAD52 can expose the single stranded DNA to other donuts with their respective single stranded DNA, and figure out which ones match. It is evident that the energetics of DNA-DNA binding and DNA-RAD52 binding are engineered to be very similar, so that DNA engaged in mismatches does not get tangled up in unproductive knots, but prefers to wrap around these massive protein donuts. But then on the other hand, when a true DNA-DNA duplex match is found, it smoothly zippers up, out of the Rad52-wrapped state.

Model of the energetics of SS DNA binding to RAD52, with subtle transitions from SS and protein-bound, to double-stranded if a proper match is found (pink).

This is the mechanistic secret of homology search between DNA stands, at least in principle. How this happens at the scale of full super-coiled duplex genomes, in the crowded milieu of the cell, with the DNA studded with countless other proteins and complexes, plus a few chemical alterations, remains a bit hard to understand, however. And the rest of the repair story, of directing a large invasion of homologous DNA into the defective duplex, editing out the defective strand, re-duplicating the donor DNA, flushing the ends, and all the other odds and ends, are of course another story entirely.

In the current paper, the researchers looked for inhibitors of RAD52, intended for breast cancer treatment. On of the traditional treatments is to break the patient's DNA with toxic chemicals, because the cancer cells typically are missing the function of some DNA repair proteins like BRCA2, or BRCA1 which acts in the same pathway. Missing this repair function has generated the cancer in the first place by creating new mutations, but there can be too much of a good thing. That is, an excess of DNA damage causes cells- even cancer cells- to kill themselves or to die outright. So it would be helpful in this mode of therapy to be able to turn off what turns out to be, for humans, a backup DNA repair pathway- RAD52.

"RAD52 forms an oligomeric ring, where the primary ssDNA binding site is located in the narrow groove spanning the ring circumference. We designated this ssDNA-binding groove as the feature to be targeted by small molecule inhibitors. While disrupting the protein-ssDNA interaction with small molecules presents a formidable challenge that has only been overcome in a handful of cases, the ssDNA binding groove of RAD52 is a promising target and is distinct from the ssDNA binding sites of other ssDNA binding proteins."

The technique to screen for RAD52 inhibitors used fluorescence energy transfer. Donor and acceptor fluorophores were put on opposite ends of 30-nucleotide DNA strands. This happens to be exactly the length that fits around one RAD52 donut, so when the DNA binds to RAD52, the ends meet and the acceptor efficiently absorbs fluorescent photons from the donor, and re-emits at its own characteristic wavelength. Wow!

Compound found by screening for impaired fluorescence energy transfer between ends of DNA wrapped around RAD52 protein complexes.

Example of screening data, showing that binding of double-stranded DNA (gray) is unaffected by the compound "1", while the binding of single-stranded DNA is strongly impaired, at very low concentrations. The assay (Y axis) is fluorescence by the red fluorophore, which is excited by the green fluorophore only when the green fluorophore is nearby and also excited by the researchers shining appropriate (green) wavelength light onto the experiment.

After screening a bunch of chemicals, they came up with one that really works well, at low concentration. It clearly inhibits single strand DNA winding around RAD52. It also, in other work, interacts directly with RAD52, indicating that, as it looks quite a bit like a couple of DNA bases, it probably just binds better than DNA itself. By computational modelling they find that this compound binds into the single-stand DNA binding groove of RAD52, very snugly positioned in a way that is tightly bound and will keep anything else at bay.

 Model, using the known structure of both RAD52 and compound "1", of how they fit together. The chemical fits tightly in a groove that would otherwise bind single stranded DNA.

Finally, they ask whether this candidate drug works in cells. In tissue culture cells, the very low concentration of 500nM makes the cells behave as though they had no RAD52 at all, which is very promising indeed. Indeed, it kills cells lacking BRCA2, if treated with a chemotherapy regime, validating the idea of the work. How all this will work in whole animals and in humans is, of course another story entirely, in a long road to drug development. But this is an example of beautiful work with its origins in studies in model organisms (yeast) and structures in basic molecular biology.

"We also show that [compound] ‘1’ selectively kills cells depleted of BRCA2, further supporting the importance of the RAD52-ssDNA interaction in BRCA deficient cells and the potential therapeutic value of RAD52 inhibition."

• It is OK to be just OK?
• Religion as fully imaginary... and for some, that is OK. "Secret knowledge that everyone else was too blind to see" "God is the ultimate transitional object".
• The decades of lies and meanness that led to this.
• Who are you gonna believe?
• Of ~400 districts in Afghanistan, the Taliban controls 40 and is actively fighting to control 44 more.
• Costa Rica- a beautiful, just, peaceful, and low-carbon country... "Traffic is offensive."
• Is there a problem with the sources of Islam, and our understanding of its history?
• Our friends the Saudis, cont... balkanizing the Balkans.
• Krugman detects evil and greed on the GOP side.
• In return, one of the steps of grief ... denial and lashing out.
• Is America breaking bad?
• Obama leaves a great deal undone.