Written by Dr. Edward Lau and Dr. Rob Beynon
"A mouse has a new liver every few days, but the lifespan of a mouse liver cell is hundreds of days” - Dr. Rob Beynon
In 1965, Robert Schimke led a team that examined the response of rat liver tryptophan oxygenase (then called tryptophan pyrrolase, the first enzyme in the breakdown of the gluconeogenic amino acid tryptophan) to glucocorticoids and to feeding of the substrate, tryptophan [ref 1]. – both treatments led to elevated enzyme levels in the liver. They concluded that glucocorticoids induced synthesis of the enzyme, whereas dietary tryptophan prevented degradation - clear evidence of the importance of the two ‘opposing’ processes of synthesis and degradation controlling the intracellular abundance of an enzyme. This was, parenthetically, one of the first examples of a connection between the transcriptome (controlling synthesis) and the metabolome (controlling degradation).
This example serves to illustrate that despite the high energetic cost of making and degrading proteins, the proteome is in a dynamic state of renewal. Even in the steady state, where the protein abundance is constant, new proteins are synthesised and at the same time, the pool is commensurately depleted through degradation. This process of protein turnover can account for a sizeable proportion of the energy budget. The rate at which any protein is replaced must have evolved through natural selection; some proteins are replaced with minute time scales, others are essentially static through the life of the individual. Moreover, the rate of protein replacement is not the same in different tissues, nor is it the same in different species – smaller mammals, with a higher basal metabolic rate replace their proteins at much higher rates than larger mammals. Indeed, the high rate of protein turnover in small mammals may be a way to elicit thermogenesis.
Although the phenomenon of protein turnover has been described since the pioneering work of Schoenheimer eight decades ago, our knowledge of how it is regulated in homeostasis and disease and how it contributes to the anatomy and physiology of the proteome has remained lagging. Early studies, mostly using radioisotopes, could only measure turnover of total protein, and the goal of measurement of individual protein turnover rates seems unattainable
In recent years this has changed. Advances in separation science and mass spectrometry delivered the ability to resolve proteins, peptides or amino acids labelled with stable isotopes.
We and others would argue that we need to understand the scope and scale of intracellular protein turnover at the level of the proteome – it is likely that the subtlety seen by Schimke and colleagues is manifest in many other biological systems. Many protein turnover studies have been conducted with mammalian cells in culture. With this experimental system, it is relatively straightforward to introduce label isotopologs of essential amino acids into the cell culture media to trace their incorporation into proteins (dynamic SILAC approaches).
Although many insights have been gained from such studies in vitro, it is increasingly evident that turnover of proteins in cell culture is very different from that in large, intact adult animals. In rapidly growing cells protein synthesis is driven by high rates of cell proliferation, with doubling times of a day or less. By contrast, in intact animals the doubling rate of cells is measured in hundreds of days (for example, the mouse hepatocyte). At such a low proliferation rate, protein abundance cannot be adjusted by cell number expansion (with commensurate dilution), and individual proteins can be expected to be replaced in time frames that sit within the lifespan of the cell, ranging from minutes to that lifespan. To avoid proteotoxic aggregates, damaged proteins will need to be removed by carefully regulated degradation, rather than simply diluted to daughter cells. These two differences between cells in culture and in tissues require us to adopt different approaches and analytical strategies.
To properly understand the dynamics of the proteome, whether in steady state or in flux states, we need to measure protein turnover rates in intact animals. But, the convenience of a simple, instantaneous medium change no longer exists, and whole animal labeling studies require a different approach, confounded by difficulties of isotope administration and reutilisation of labeled amino acids for new protein synthesis. Moreover, turnover can only feasibly be accessed through measurement of synthesis by tracking incorporation of stable isotope labels. It is almost obligatory to measure the rate of synthesis of a protein through isotope incorporation, administered over days, months or years.
Quantification of protein turnover in animals is further complicated by the slow precursor availability in the tissue of interest after the isotopic label is administered. For example, labeled proteins or amino acids supplied in the diet have to move through the digestive system, cross the intestinal mucosal barrier, pass through the hepatic system and be transported to peripheral tissues in the blood. Any delay in appearance of labeled amino acids in the precursor pool of a peripheral tissue would interfere with measurement of the true turnover rate. By contrast, heavy water ([2H]2O) crosses tissue barriers much more quickly. Thus, these two labelling methods could yield different apparent turnover rates for the same protein. But can the two approaches be brought to convergence? Intuitively, one would expect high turnover proteins to be most affected by slow precursor equilibration, and this is indeed the case.
In a recent study [ref 2], we compared two common methods of measuring protein turnover in animals either using heavy labeled amino acids in diet or provision of [2H]2O in drinking water. The two strategies differ in precursor availability and metabolism, as well as the mass spectral features of peptides following label incorporation. Our question was very simple: what were the optimal data analysis strategies and when applied, do the two methods yield comparable turnover rate results?
We set up mouse labelling studies in which the only significant variable was the labeling protocol. Two groups of adult mice of identical strain, sex, age, maintained with identical husbandry, were labeled either with [13C6]lysine or heavy water ([2H]2O) over about a month. Animals were sampled at different times over this period, and the proteomes of the heart, liver, kidney, and skeletal muscle were analyzed by mass spectrometry. These tissues differ in their median turnover rate, allowing our analysis to extend over a broader range (liver and kidney are higher than heart, and in turn all three are higher than skeletal muscle). To ensure consistency in data processing, one of us (EL; https://ed-lau.github.io/riana) wrote Riana, new Python software that quantitates peptide labelling, recovers isotope abundance as a function of labelling time and fits these data to recover the first order rate of replacement, equivalent to steady state half-life.
Because heavy water is known to rapidly equilibrate across tissues and compartments, it would give data closest to the ground truth at least where label utilization is concerned. This experimental design therefore allows us to use water data as a reference for optimizing the analysis of lysine labelling data.
As anticipated, with [13C6]lysine labeling there was a delay in precursor equilibration in all tissues. Simple exponential models of protein turnover are compromised by this delay, and the rate of turnover of high turnover proteins is underestimated. This can be corrected by using a suitable two-compartment kinetic model that also models the delay in the precursor pool, the protein turnover rate constant (kd) and the precursor availability rate constant (kd).
Surprisingly, finding suitable kp values to adjust for labeling delay was not straightforward. Although the ratio of [13C6]lysine vs. [12C6]lysine can be measured in a tissue by LC-MS, an empirical sampling of tissue lysine isotope enrichment over time led to an apparent underestimation of the true precursor pool. Moreover, the best strategy for finding kp is dependent on the tissue being examined. Global parameter optimization method can effectively find the best kp value that explains the data sets in slow turnover tissues but is less effective in high turnover tissues. Therefore although the complications of slow precursor equilibration can be overcome with the proper strategy, careful considerations must be given based on the tissue and animal under study.
Such complications make a compelling case for heavy water as a turnover label in intact animals. Heavy water is inexpensive, virtually all peptides demonstrate isotopic incorporation, and the speed with which water equilibrates in the body mitigates complications due to precursor pool delay and reutilisation. That being said, heavy water is not without drawbacks. The pathways of heavy water labeling of different essential and non-essential amino acids is incompletely understood, and measurement of isotope incorporation in precursor spectra is more prone to errors and isobaric contamination in the mass spectra.
Accurate measurement of whole animal, proteome-wide protein turnover is still difficult, and there are several largely unresolved issues.
● What are the specific considerations in data analysis strategy when using different labeling protocols?
● Do existing analytical workflows and software packages give comparable turnover rates and profiles when analyzing a common data set?
● Each protein yields multiple peptides, and the rate of labeling of each peptide yields a measure of turnover. How are these data aggregated - is it better to combine all peptide data and fit once curve, or fit each peptide data individually? If the latter, are there objective criteria that can be applied to eliminate outlier turnover values?
● What are the optimal practices for error estimation in turnover measurements, and statistics in comparing turnover across conditions (e.g., homeostasis vs. disease)?
● Is it possible to establish a set of 'gold standard' turnover rates in different tissues and in different species?
We’d like to propose that those who are interested in this challenge bring together their knowledge and expertise in a community effort to resolve some of these questions. Should anyone be interested they are invited to contact us for further information.
Reference 1: Schimke et al. 1965 PMID 14253432
Reference 2: Hammond et al. 2022 PMID 35636728
.png)
