DNA repair in terminally differentiated cells

This page summarizes research interests in the Nouspikel lab, and outlines current and future directions of research.
Click on any picture to enlarge it.

DNA repair is a daunting taskSome DNA lesions

Even though DNA is pretty stable, we have so much of it that its decay pauses quite a challenge: due to spontaneous depurination we loose about 3000 purines per cell per day. In addition, because of byproducts of the oxygen metabolism within each cell, several hundreds of pyrimidines become oxidized every day. As if this were not enough, DNA duffers multiple aggressions from the environment: from UV rays in sunlight, to chemical carcinogens in food, cigarette smoke, or polluted air.

To cope with this damage load, we have evolved a whole set of DNA repair systems:
Maintaining that many systems in working order does not come cheap: NER alone involves over 30 polypeptides that constantly scan our genome. To which one should add the energy cost of carrying the repair operation itself. The energy expense is critically important, though, as illustrated by the variety of horrible diseases (generally cancer prone) caused by defects in a given repair system.

More info
Instability and decay of the primary structure of DNA Lindahl,T. Nature 362:709-715 (1993)
Quality control by DNA repair  Lindahl,T. & Wood,R.D. Science 286:1897-1905 (1999)

The "parsimony" hypothesis

Terminally differentiated cells never replicate their DNA, so why should they bother repairing it? Think of all the energy they could save by dispensing with producing DNA repair enzymes, not to mention performing the actual repair.

Of course cells need to maintain the few genes that they are currently using, but this can be done by a dedicated mechanism called Transcription-Coupled Repair (TCR). It is generally believed that TCR is achieved by RNA polymerase II, which calls repair enzymes for help when it encounters a blocking DNA lesion.

More info
Subpathways of nucleotide excision repair and their regulation Hanawalt,P.C. Oncogene 21:8949-8956 (2002)

Repair phenotype of human neurons

This is precisely the phenotype that we have observed in human neurons, perhaps the epitome of differentiated cells. Because it is known to be subject to TCR, we initially focused on Nucleotide Excision Repair.

Upon differentiation neurons show a marked drop in NER at the global genome level. By contrast, transcribed genes are still proficiently repaired on both DNA strands. The latter is quite a surprise: TCR only works on the transcribed strand, as RNA polymerase II is not affected by lesions on the non-transcribed strand. The non-transcribed strand is normally repaired by NER like the bulk of the genome, but in neurons the bulk of the genome is barely repaired at all. So what is repairing the non-transcribed strand in neurons? We decided to call it Differentiation-Associated Repair (DAR) and elucidating its mechanism is one major goal of the lab.
Why we need DAR
But of course, DAR makes a lot of sense retrospectively. The way TCR (and global NER) work is by removing a ~30-nucleotide piece of DNA around the lesion, then filling the gap using the other strand as a template. If neurons were to let the non-transcribed strand fall in dereliction, sooner or later TCR would run into problems, when presented with a damaged template (click on picture for details).

More info
Terminally differentiated human neurons repair transcribed genes but display attenuated global DNA repair Nouspikel & Hanawalt, Mol. Cel. Biol. 20:1562-1570 (2000) (download pdf file 138 K)

Answers from macrophages

At this point, there were two main questions that need to be addressed:
Because neurons are difficult to work with (you can get so few of them), we switched to another type of terminally differentiated cells: macrophages. These have essentially the same repair phenotype as mature neurons: low global NER, proficient TCR, and presence of DAR. But the advantage is that the precursor cells can be grown to the billions, before being differentiated into macrophage. This allowed us to set up an in-vitro DNA repair assay that would not have been possible with neurons.
Ubiquitination pathway
Using this assay, we purified a protein that could complement the NER deficit in macrophage extracts. Mass spectroscopy identified it as the ubiquitin-activating enzyme E1, i.e. the first step on the ubiquitination pathway (it passes ubiquitin to an E2 enzyme, then an E3 enzyme transfers it to the target protein).

 As usual, answering one question raises even more questions. In this case:
  1. What's wrong with E1 in differentiated cells?
  2. How does E1 affect NER?

1) It is known that E1 is phosphorylated on at least 4 different serines. We have evidence that one of these sites (we don't know which one yet) is de-phosphorylated in differentiated cells. The assumption is that this prevents proper interaction with the E2 used by NER (we don't know yet which E2 it is), without affecting the E2s used by other pathways.

2) Ubiquitination is best known for its ability to tag a protein for degradation, but it is by no means its only function. Depending on the number of ubiquitin moieties attached to the target, and on the type of chain they form, it is possible to control the activity of the target protein, positively or negatively. An obvious possibility is thus that one of the NER enzymes is activated by ubiquitination. We have indications as to which one it is, but still need a formal proof...

More info
The catalytic mechanisms of protein ubiquitylation Passmore & Barford, Biochem. J. 379:513-525 (2004)

Mechanisms of DAR

Pei-hsin Hsu, a very good student who worked with me at Stanford, studied the NER phenotype of four different leukemia cell lines, at various stages of differentiation. His goal was to find out whether there was a progressive loss of NER, correlating with the degree of differentiation. The answer was no, but Pei confirmed that DAR exists in these cells, and most importantly that it can exist in proliferating cells also. Which means that we may have to change the name "DAR". since it's not unique to differentiated cells...

As for the mechanisms of DAR, we found out that it happens in areas of a transcribed gene that RNA polymerase II never reaches (click here for details). In addition, it also happens when RNA polymerase II is stalled half-way through the gene by transcription inhibitors. We thus believe that DAR is nothing more than a subset of NER, concentrated in the nuclear sub-compartments where transcription takes place. Bringing a gene into these "transcription factories" grants it preferential access to NER enzymes, whether it is actually transcribed or not. Knocking down NER enzymes with siRNA confirmed that DAR depends on the same enzymes than global NER, but not on those specific for TCR.

Directions of research

Filling in the blanks

There are many questions that remain to be answered, besides those outlined above:

Generalization of the model

I would like to know how universal these mechanisms are. In particular:

Potential for cancer therapy

What we have here is a natural control mechanism that shuts down DNA repair at the global genome level, while preserving it in transcribed genes. If we could learn how to trigger it in the whole organism, rather than just in differentiated cells, it could be a tremendous help for cancer therapy.

For instance, by combining it with anti-cancer drugs like cisplatin, which work by damaging the DNA. It would allow for lower doses of the drug, hence reducing the nasty side effects. It may also prevent relapses due to the emergence of a resistant clone of cancer cells overexpressing NER enzymes.

Potential for neuro-regeneration

It has been observed in many neurodegenerative diseases, among which Alzheimer's disease, that neurons attempt to re-enter the cell cycle. They even manage to replicate their DNA, but die shortly afterwards. This is good news and bad news: it's good to know that there is a molecular switch allowing to trigger neuro-regeneration, and that this switch must be fairly accessible for so many diseases to trigger it. The bad news is, of course, that it doesn't work because neurons die and we don't know why.

My (so far totally unproved) hypothesis is that they die because they have neglected DNA repair over decades and are now attempting to replicate and transcribe a crippled genome. If this were true, and if we could learn how to control NER via ubiquitination, then we might be able to trigger a bout of DNA repair in neurons, prior to causing them to replicate. This would open the gate to neuro-regeneration approaches!

More info
When parsimony backfires Nouspikel & Hanawalt: BioEssays 25:168-173 (download pdf file 169K)

E1 as a master switch

As far as I know, this is the first time that control of ubiquitination has been observed at the level of the E1 enzyme. Because there is only one E1, versus dozens of E2s and hundreds of E3s, it has always been assumed that control mechanisms would operate on the E3s. But if we are right that differential phosphorylation of E1 affects the way it interacts with the various E2, this may constitute a control mechanism affecting many systems other than NER. It may even be a "master switch" used to trigger all at once many of the phenotypic changes required by differentiation.

I am very interested in identifying other ubiquitination pathways regulated at the E1 level, and I believe that we have a very good system at hand, macrophage extracts, to undertake this task.

More browsing

Techniques used in the lab
Thierry Nouspikel's CV