Cells show their individuality in a new study of DNA damage responses
A new paper from postdoctoral fellow Stephan Uphoff in the Biochemistry Department has revealed that random variation in the DNA repair capacity of cells can lead to genetic variation.
The results are published in Science (1) and are the fruition of a collaborative project between Dr Uphoff in Professor David Sherratt's lab and the lab of Professor Johan Paulsson at Harvard Medical School. They provide insight into how phenotypic variation can lead to genetic variation - a new twist on studies exploring the impact of variability in gene expression between cells.
E. coli cells treated with DNA methylation damage induce the adaptive response by activating Ada protein expression. The microscopy image shows fluorescently tagged Ada in yellow. Despite identical genetic makeup and treatment, a fraction of cells fails to induce the Ada response (in grey). Scale bar: 5 µm (Click to Enlarge)
A physicist by training, Dr Uphoff has spent the last few years developing and applying live cell imaging techniques. Currently funded by a Sir Henry Wellcome Postdoctoral Fellowship from the Wellcome Trust and a Junior Research Fellowship at St John's College in Oxford, he has been using single-molecule imaging to study mechanisms of DNA repair in bacteria, in both the Sherratt and Paulsson labs.
The newly published study explores the consequences of heterogeneity in a bacterial DNA repair process. Whilst there has been lots of discussion about noise in gene expression giving rise to phenotypic heterogeneity in genetically identical cells, there have been few studies that go beyond transient variations in gene expression. In the case of DNA repair, however, any transient heterogeneity could persist over long timescales in the form of mutations.
In the bacterium Escherichia.coli, the adaptive response protects cells against the toxic and mutagenic effects of DNA methylation damage. This requires Ada protein, which as well as directly repairing methylated DNA, also activates ada gene expression. It does this via a positive feedback mechanism - ada expression is increased a thousand-fold by methylated Ada which acts as a transcriptional activator after transfer of a methyl group from damaged DNA onto the protein during the repair process.
Another feature of the DNA damage response is that Ada protein is present in low numbers in cells before DNA damage. 'We hypothesised that there should be substantial heterogeneity in the adaptive response between cells because positive feedback tends to amplify the noise that is inherent in low molecule numbers,' says Dr Uphoff.
To resolve the heterogeneity and visualise the dynamics of the response, Dr Uphoff joined Professor Paulsson's lab for one year at the beginning of the project. He took advantage of a single-cell microscopy technique there. This technique uses microfluidics to follow single cells over many generations. Dr Uphoff and colleagues found that although most cells rapidly induced the damage response after treatment with the DNA-damaging agent methyl methanesulfonate (MMS), some cells failed to respond to MMS for several generations.
Microfluidic imaging shows stochastic activation of Ada expression (yellow) in individual cells
When the group followed this up with single-molecule microscopy to count individual Ada protein molecules in live cells before MMS treatment, they observed that the abundance of Ada was extremely low and that a significant proportion of the cells did not contain a single Ada molecule. The percentage of cells without a single Ada molecule matched the percentage of cells with a delayed response in the microfluidics study.
These experiments confirmed that the damage response is very sensitive to the presence of Ada. 'Cells need to wait until stochastically they make a single molecule,' explains Dr Uphoff. 'Once there is one, it senses damage and this is enough to make the switch.'
For the next part of the study, Dr Uphoff returned to the Biochemistry Department at Oxford to establish whether heterogeneity in Ada abundance affects mutation rates. To directly measure this in single cells, he turned to a super-resolution single-molecule microscopy method that he had previously developed during his DPhil and subsequent six- month postdoc at Micron, the bioimaging facility at Oxford. DNA methylation lesions are highly mutagenic and only their rapid repair by Ada prevents the formation of mutations. Dr Uphoff reasoned that DNA mismatches - the precursors of mutations - could be detected with this microscopy method by tracking the movement of the DNA mismatch recognition protein MutS.
Dr Uphoff and colleagues measured whether MutS proteins were mismatch-bound or mobile while also imaging Ada proteins in the same live cells. They found that after damage, cells that had switched on the adaptive response effectively repaired the damage whereas cells with a delayed response accumulated damage that was converted to DNA mismatches.
Delayed activation of the adaptive response leads to mutations. DNA mismatches are visualised by single-molecule imaging of the DNA mismatch recognition protein MutS (red tracks). Cells that have high levels of Ada (yellow) show few MutS molecules bound to mismatches, whereas cells with a delayed response have many mismatches (red tracks). Scale bar: 2 µm
'We identified a fraction of cells not inducing the adaptive repair response for several generations despite high doses of damage and found that this was due to the extremely low expression of the proteins that sense the damage,' explains Dr Uphoff. 'The cells that failed to induce the response had no such molecules at all and therefore could not sense damage until they finally made one molecule which triggered the response.'
Given that the potential consequences of having a low average abundance of Ada are serious, the group considered why cells produce so little before damage. In light of their observations that high Ada expression is toxic and that cells minimize variation in Ada levels as much as possible, they suggest that cells do not appear to be exploiting the variability as a strategy to diversity the cell population. Instead, cells may be obliged to express Ada protein at low levels and must cope with the unavoidable consequences of stochastic fluctuations.
So what are the wider implications of the work? Dr Uphoff comments: 'People think about phenotypic variation as a consequence of genetic variation, but we have shown that the opposite can happen - cells can start off genetically identical and because of random variation in DNA repair capability, can become genetically different.'
In future work, Dr Uphoff would like to explore if random variation might affect the fidelity of other DNA repair pathways as well. He adds that the phenomenon could be relevant to bacterial antibiotic resistance and cancer cell evolution. 'Bacteria activate stress responses that modulate DNA repair and mutagenesis when challenged by antibiotics, and cell-to-cell variation in these responses could have similar effects on genetic variation as we have observed for the adaptive response.' Like all cells, cancer cells will show random variation in gene expression, and a cell that has particularly low levels of a DNA repair protein may accumulate mutations that favour its growth.
- Uphoff S, Lord ND, Potyin-Trottier L, Okumus B, Sherratt DJ, and Paulsson J. (2016) Stochastic activation of a DNA damage response causes cell-to-cell mutation rate variation. Science 351 (6277), pp1094-1097