Biomarkers of ageing
A key use case of our ProQuant® platform technology is to look for biomarkers of disease. To exemplify this, we applied ProQuant® to investigate potential biomarkers of ageing, using serum samples from subjects of different ages. While it has a lower depth of coverage, the approach we used was our patented bottom-up DDA approach as it is better at investigating multiple PTM classes in a non-hypothesis driven manner.
An initial investigation of the data showed that none of the c. 300 proteins we quantified using at least two unique peptides were found to be associated with age after correction for multiple testing. However, the strength of our patented DDA platform is its reproducible and accurate measurement of peptides, and hence its ability to quantify PTMs. We therefore used the peptide data to calculate PTM fractions for a wide variety of PTMs detected using an error-tolerant approach implemented using Mascot. We were at first disappointed to find that no individual sites of modification were associated with age (unlike our results in diabetes. However, when we examined the data more closely there appeared to be patterns of changes associated with different classes of modification.
We therefore investigated potential methods to test for enrichment of p values within entire classes of PTMs and set up a method using Tukey’s Higher Criticism as implemented by Donoho and Jin (ref). This secondary analysis identified four classes of PTMs that were significantly associated with ageing.
Firstly, we found a strong association between age and acetaldehyde modification of proteins, but established that to be a confounding influence of alcohol, which was being enjoyed more by our older subjects!
More interestingly, we also found a strong association between age and the deamidation of asparagine residues on serum proteins, which is to the best of our knowledge a novel insight into ageing. It has been known for decades that long-lived proteins (such as crystallins in the eye and other extracellular matrix proteins) show greater deamidation in older proteins, and this has also been shown to occur in vitro.
However, this is an unexpected observation in serum proteins, where proteins typically have a much shorter half-life. Indeed, the deamidation may be indicative not of increased rate of deamidation but in an extension of the protein half-lives. Irrespective of the underlying mechanism, it is due to the power of ProQuant® that such subtle differences across multiple proteins can be detected.
Thirdly, levels of iron (Fe3+) cations associated with proteins are markedly reduced in older subjects, consistent with previously published work investigating iron metabolism in ageing.
Finally, conversion of cysteine to dehydroalanine is strongly elevated in older subjects, and is strikingly enriched in cysteines that were originally part of a disulphide bond. These changes are widespread, are probably part of a system-wide change in redox, and due to their localisation to disulphide-bonded cysteines are likely to affect structure and therefore function of proteins.
Overall, this study showed the power of a non-hypothesis driven proteomics protocol that is not just focussed on depth of coverage, but has the precision and accuracy to investigate aspects of the proteome not typically investigated in non-hypothesis studies. Coupling that with custom bioinformatics has unearthed multiple novel observations relating post-translational modification of proteins with the ageing process.