Investigations of Peptide and Protein Deamidation

Citation

Robinson, Noah Edward (2004) Investigations of Peptide and Protein Deamidation. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/JHAP-JG47. https://resolver.caltech.edu/CaltechETD:etd-09212004-154502

Abstract

This work concerns the peptide and protein deamidation of asparaginyl, Asn, and glutaminyl, Gln, residues to aspartyl and glutamyl residues, respectively. This is accompanied by a change in charge and increase of one unit in mass. Isomerization also often takes place. This reaction--which occurs spontaneously and non-enzymatically in many proteins under physiological conditions--has been hypothesized to be a general molecular clock, which could be used as a timer for protein turnover, development, or aging, and for the controlled conversion of one protein into another. Since this hypothesis was first proposed, amide molecular clocks have been demonstrated in several biological systems. These include the timing of protein turnover in rat cytochrome c and aldolase; the counting of individual enzyme molecule catalytic cycles in triosephosphate isomerase; and the resetable time-dependent monitoring of DNA repair in Bcl-XL.

The objective of this work is to increase the understanding of peptide and protein deamidation, especially the prediction of deamidation rates for specific amides in proteins. This capability has been developed for Asn residues in proteins of known three-dimensional structure, and the sequence information required to do this for Gln residues has also been obtained. This work provides stability information to protein chemists and needed information to those who continue to investigate the molecular clock hypothesis.

The problem was attacked in two phases. The first consisted of experimental work to determine the rate of deamidation as a function of primary sequence near the amide. All 400 possible peptides of the type GlyXxxAsnYyyGly were synthesized and the deamidation rates of 323 of these were measured. An additional 67 asparaginyl containing peptides that had a variety of lengths and forms were also measured in order to assess the efficacy of pentapeptides in this work. Also, a similar set of 400 peptides were synthesized with Gln in place of Asn. The deamidation rates of 52 of the Gln peptides have been measured and, from these rates, the rates of the other Gln peptides have been inferred. With the addition of rates for 34 peptides with blocked Cys, a total of 476 individual rate experiments on peptides have been carried out in 0.15 M Tris buffer at pH 7.4, 37.00 [degrees]C. Additional rate experiments were conducted on solution effects and on properties of the experimental system.

The second phase involved using experimental data from the literature--on relative instabilities of specific amides in proteins under specified solvent conditions and on protein three-dimensional structures--to quantify the effect that secondary, tertiary, and quaternary protein structure has on primary structure deamidation rates.

Factors that determine deamidation rates include structural components as well as electronic effects. It happens that the primary sequence component is so large that it is possible to separate it from the rest of the problem and measure it alone. The secondary, tertiary, and quaternary effects were then considered. It was found that, on average, 60% of the rate is accounted for by primary sequence, while the remaining 40% is ascribable to secondary, tertiary, and quaternary structure in those proteins for which suitable deamidation observations are available. For all proteins, the estimated values are 50% and 50%. For individual amides, these percentages vary over a wide range.

A procedure was developed through which the quantitative deamidation rate of any amide in a protein for which the three-dimensional structure is known can be calculated. These estimated rates are accurate to within about a factor of two or better for most amides with half-times less than 100 days. They can be used to predict the fastest amides in a protein to greater than 95% reliability. This reliability is remarkable, considering the fact that deamidation half-times vary from less than 1 day to over 100 years in 37.00 [degrees]C, pH 7.4, 0.15 M Tris buffer. The procedure was calibrated with only relative amide instabilities. It was then evaluated with absolute deamidation rates.

Details of this work are given in the pages which follow. The result was a computerized procedure through which deamidation rates for every Asn of every protein in the PDB Protein Database were calculated and analyzed.

This work provides two discoveries of biological importance.

First, the assumption of genetic determination of deamidation rates, which has until now been dependent upon scattered and largely qualitative observations, has been placed upon firm quantitative foundation with a thorough understanding of the effects of primary, secondary, and tertiary structure.

Second, this quantitative understanding has permitted reliable estimation of the distribution functions of deamidation rates in thousands of proteins. These rates show that a substantial fraction of proteins are genetically programmed to deamidate in biologically relevant timed intervals even though most of the genetically available structures deamidate more slowly. This quantitative preference for fast deamidation times relative to slow deamidation times is an entirely new discovery. The fact that this property is present within the thousands of proteins of currently known three-dimensional structure markedly strengthens the molecular clock hypothesis.

Thesis Files