pubmed.ncbi.nlm.nih.gov

Protein kinetic signatures of the remodeling heart following isoproterenol stimulation - PubMed

. 2014 Apr;124(4):1734-44.

doi: 10.1172/JCI73787. Epub 2014 Mar 10.

Protein kinetic signatures of the remodeling heart following isoproterenol stimulation

Maggie P Y Lam et al. J Clin Invest. 2014 Apr.

Abstract

Protein temporal dynamics play a critical role in time-dimensional pathophysiological processes, including the gradual cardiac remodeling that occurs in early-stage heart failure. Methods for quantitative assessments of protein kinetics are lacking, and despite knowledge gained from single-protein studies, integrative views of the coordinated behavior of multiple proteins in cardiac remodeling are scarce. Here, we developed a workflow that integrates deuterium oxide (2H2O) labeling, high-resolution mass spectrometry (MS), and custom computational methods to systematically interrogate in vivo protein turnover. Using this workflow, we characterized the in vivo turnover kinetics of 2,964 proteins in a mouse model of β-adrenergic-induced cardiac remodeling. The data provided a quantitative and longitudinal view of cardiac remodeling at the molecular level, revealing widespread kinetic regulations in calcium signaling, metabolism, proteostasis, and mitochondrial dynamics. We translated the workflow to human studies, creating a reference dataset of 496 plasma protein turnover rates from 4 healthy adults. The approach is applicable to short, minimal label enrichment and can be performed on as little as a single biopsy, thereby overcoming critical obstacles to clinical investigations. The protein turnover quantitation experiments and computational workflow described here should be widely applicable to large-scale biomolecular investigations of human disease mechanisms with a temporal perspective.

PubMed Disclaimer

Figures

Figure 1
Figure 1. 2H2O-labeling strategy to quantify protein turnover in normal and isoproterenol-stimulated mice.

(A) 2H2O-labeling scheme for the mouse experiments. Step 1: Mice were labeled to ≈4.4% body water 2H2O enrichment. Step 2: Normal, remodeling, and reverse-remodeling mice were euthanized at eight separate time points to harvest heart and plasma protein samples. Step 3: Newly made proteins containing 2H2O labels had a higher average mass and shifted the isotope distribution of peptides in the mass spectrum. Step 4: With the incorporation of heavier isotopes after 2H2O labeling, the gradual decrease in the proportion of unlabeled isotopomers (m0/mi) could be modeled to deduce the turnover rate. Step 5: Protein kinetics in normal mice and disease models were compared. (B and C) Isoproterenol-induced cardiac remodeling, as measured by heart weight/body weight ratio (HW/BW), ejection fraction (EF), and mitochondrial function. Solid lines denote local regression, dashed lines denote 95% confidence intervals. (D) Histograms of protein turnover rates in normal mouse heart cytosol and mitochondria. On average, cytosolic proteins turned over twice as fast as mitochondrial proteins. (E) Distributions of individual protein turnover rates in normal, remodeling, and reverse-remodeling hearts. Box: interquartile; whiskers: 1.5× interquartile; violin: data density. (F) Turnover rates of proteins belonging to ten selected complexes in normal (gray), remodeling (red), and reverse-remodeling (blue) mouse hearts. Associated proteins shared similar turnover rates, e.g., ATP synthase: 0.026 (0.017–0.037) d–1; 20S proteasome: 0.106 (0.088–0.140) d–1. Cardiac remodeling and reverse remodeling exert different effects on each complex. Analyzed protein count is given in parentheses. Box: interquartile; whiskers: 1.5× interquartile.

Figure 2
Figure 2. Protein kinetics in the remodeling heart.

(A) Isoproterenol increases the turnover rates of multiple annexins (left). The fraction of turned over ANXA2 over time was much higher in remodeling (red) versus normal (black) hearts (right). Lines represent the best-fitted curve from a single peptide. (B) Ratios of turnover rates of the glycolytic proteins in remodeling (R) and reverse-remodeling (RR) hearts, normalized to normal (Nor) hearts. Each data series represents the behavior of a single glycolytic enzyme. (C) Heatmap of turnover rate ratios of the glycolytic proteins in remodeling hearts versus normal hearts (left) and the fraction of turned over proteins in selected enzymes (right) over time in normal (black) and remodeling (red) hearts. (D–G) Isoproterenol regulated protein categories differentially, e.g., fatty acid oxidation, branched-chain amino acid (AA) metabolism, 20S proteasomal proteins, and 60S ribosomal proteins. (H) The β and F0 subcomplexes in respiratory complexes I and V had particularly elevated turnover in remodeling. Subcomplexes are heatmapped where possible; white represents no quantified ratios. (I) Scatter plot comparing the turnover rates of mitochondria-targeted proteins when measured in extracted mitochondrial samples versus when measured in a cytosolic sample in normal (left), remodeling (middle), and reverse-remodeling (right) mouse hearts. Only proteins that were quantified with high confidence were included (2+ fitted peptides, median absolute distribution <25%). We observed no systematic bias in the turnover rate of each cellular fraction, suggesting that protein import into the mitochondria is not a significant factor for turnover measurements.

Figure 3
Figure 3. Protein kinetics in reverse remodeling.

(A) Scatter plots of protein turnover rate ratios in remodeling and reverse-remodeling hearts over normal hearts. Each data point represents a protein species. (B) Proteins in quadrant 1 have elevated turnover rates in remodeling hearts (greater than 1.25-fold versus normal), but depressed turnover (less than –1.25-fold versus normal) in reverse-remodeling hearts compared with normal hearts and are significantly enriched for ribosomal proteins, suggesting that ribosomal proteins have higher turnover during remodeling, but slower turnover in reverse remodeling. (C) Proteins in quadrant 2 (greater than 1.25-fold in remodeling/normal and reverse-remodeling/normal) and (D) quadrant 3 (less than –1.25-fold in remodeling/normal and reverse-remodeling/normal) are enriched for MAPK signaling and proteasomes, respectively.

Figure 4
Figure 4. Temporal dynamics of the human plasma proteome.

(A) Gradual labeling in human clinical studies complicates 2H2O-labeling data analysis and curve fitting. (B) Labeling and data analysis schemes for the four healthy human subjects. Step 1: The subjects were labeled for 14 days. Step 2: Three milliliters of blood was collected at 10 to 15 time points. Step 3: Plasma proteins were analyzed by MS. Step 4: The changing isotopomer patterns were modeled with a nonlinear function to deduce protein turnover rates. (C) The gradual enrichment of 2H2O in the body water of the subjects was modeled by a first-order exponential decay function to deduce the enrichment rate (kp) and level (pss) that are used in turn to model the turnover kinetics of peptide isotopomers. (D) Experimental data and kinetic curve fitting of a human plasma peptide (EQLGEFYEALDCLCIPR3+). The fractional abundance of the unlabeled isotopomer (m0/mi) decreased in a sigmoidal curve that reflects the two rate constants of 2H2O ramping (kp) and protein turnover (k). Dashed lines indicate the upper and lower limits of fitting. (E) The turnover rates of 182 plasma proteins that were quantified as in D in three or more subjects. Selected proteins are marked. The full dataset is provided in Supplemental Excel file 1.

Figure 5
Figure 5. Measurement of protein turnover from a single biopsy.

(A) Simulation of the mass isotopomer abundance curves of the same peptide sequence as shown in Figure 4D, but with different hypothetical turnover rates (k), according to the nonlinear model. Puncta denote the corresponding mass isotopomer abundance (m0/mi) that would be measured from a single time-point experiment on day 8 of labeling. A single (m0/mi) value is therefore sufficient to deduce k. (B) Experimental data and fitting of the same peptide sequence as shown in Figure 4D, but from our single-point experiment on the day 8 plasma sample from subject 1. Triplicate data points acquired from the single sample define the kinetic curve to the same effect as the multiple data points from different time points as in Figure 4D, demonstrating the feasibility of acquiring protein kinetics information without a time-course experiment. (C) Correlations in turnover rates between peptides commonly analyzed from single-point sampling and 15-point time-course experiments from subjects 1 (top) and 2 (bottom). The average peptide relative standard error was ≈20% compared with the time-course experiment.

Figure 6
Figure 6. Comparison of protein turnover and expression changes.

(A) Scatter plot of individual protein turnover ratios versus abundance ratios in remodeling versus normal hearts, as measured by either spectral count (top) or iBAQ (bottom). Right panels: Analysis performed only on proteins with significant changes in turnover in the remodeling hearts showed a similar lack of correlation. The lack of correlation between protein turnover and abundance changes is consistent with previous observations in the literature (41) and further corroborate the effective independence of the two measurable parameters. (B) Immunoblots (IB) against annexin II (ANXA2), annexin V (ANXA5), and lactate dehydrogenase (LDHA/C) show that for these proteins, increased turnover is associated with increased abundance after 7 or 14 days of isoproterenol (Iso). On the other hand, other tested proteins including the glycolytic enzymes aldolase A (ALDOA), GAPDH, hexokinase 1 (HK1), hexokinase 2 (HK2), pyruvate dehydrogenase (PDH), phosphoglycerate mutase 1 (PGAM1), and pyruvate kinase (PKM1/2) did not show increased abundance, suggesting that their accelerated turnover is independent of upregulated expression. kIso/k, fold-change of turnover rates in remodeling versus normal hearts. (C) The independence between changes in protein turnover and abundance is consistent with the notion that both synthesis and degradation contribute to the overall protein pool size.

Similar articles

Cited by

References

    1. Balch WE, Morimoto RI, Dillin A, Kelly JW. Adapting proteostasis for disease intervention. Science. 2008;319(5865):916–919. doi: 10.1126/science.1141448. - DOI - PubMed
    1. Hinkson IV, Elias JE. The dynamic state of protein turnover: It’s about time. Trends Cell Biol. 2011;21(5):293–303. doi: 10.1016/j.tcb.2011.02.002. - DOI - PubMed
    1. Claydon AJ, Beynon R. Proteome dynamics: revisiting turnover with a global perspective. Mol Cell Proteomics. 2012;11(12):1551–1565. doi: 10.1074/mcp.O112.022186. - DOI - PMC - PubMed
    1. Katz AM. Physiology of the Heart. 5th ed. Philadelphia, Pennsylvania, USA: Lippincott Williams & Wilkins; 2010.
    1. Goonasekera SA, et al. Decreased cardiac L-type Ca(2)(+) channel activity induces hypertrophy and heart failure in mice. J Clin Invest. 2012;122(1):280–290. doi: 10.1172/JCI58227. - DOI - PMC - PubMed

Publication types

MeSH terms

Substances