Richard A. Cohen, Mark E. McComb
From the Vascular Biology Unit (R.A.C.), Whitaker Cardiovascular Institute, BU NIH-NHLBI Cardiovascular Proteomics Center (R.A.C., M.E.M.), Mass Spectrometry Resource Center (M.E.M.), Boston University School of Medicine, Boston, Mass.
Correspondence to Richard A. Cohen, Vascular Biology Unit, Boston University School of Medicine, 650 Albany St, X720, Boston, MA 02118. E-mail firstname.lastname@example.org
See related article, pages 706–714
Key Words: preconditioning • proteomics • ischemia • mitochondria • mass spectrometry
Cardiac ischemic preconditioning was first described in 1986 by Murry et al,1 who reported that brief periods of ischemia dramatically reduced myocardial infarct size caused by a subsequent, more prolonged period of ischemia. Since the initial report, the phenomenon has been shown to occur in many cell types, including human cardiomyocytes,2 raising hopes that ischemic injury to human myocardium might be prevented. The discovery that pretreatment with several pharmacological agents including adenosine, diazoxide, bradykinin, and insulin could also precondition the heart and prevent subsequent ischemic damage, provided an as yet unfulfilled promise that drugs could be developed which limit cardiac cell death caused by myocardial infarction in patients.3
As is always true of complex physiological problems, it is likely that understanding the fundamental molecular mechanism of preconditioning will be necessary to develop effective medical therapies. In recent years the mitochondrion has become the subject of intense scrutiny for the key to preconditioning. The formation of the mitochondrial permeability transition pore (mPTP) is an essential step in ischemia-induced cardiomyocyte death because its formation accounts for collapse of the mitochondrial membrane potential and failure of oxidative phosphorylation to produce ATP which leads to cardiomyocyte death. Furthermore, inhibitors and activators of the mPTP inhibit or favor preconditioning, respectively, suggesting both the mPTP and the mitochondrion are key participants.3,4
Further evidence that the mitochondrion is key to the initial physiological response to preconditioning is provided by Arrell et al in this issue of Circulation Research, who used a proteomic approach to measure changes in the amount of cardiomyocytes undergoing preconditioning.5 Relatively few of the proteins they identified changed at all. Of the 19 that did, about half were classified as mitochondrial proteins. The purpose of this editorial is not to review the pathophysiology of preconditioning, nor the role of the mitochondrion, which has been reviewed recently by experts in the field,3,4 but rather to discuss the novel application of proteomics to a basic problem in physiology, to indicate the promises and problems in the approach, and to point out the future challenges. The principles of proteomics have been reviewed.6,7
Arrell et al used a 2 dimensional gel electrophoresis approach to assess the abundance of cardiac proteins after cultured cardiomyocytes were exposed for 1 hour to either control buffer or buffer containing one of the 2 preconditioning agents, diazoxide or adenosine. They chose cultured cardiomyocytes over intact preconditioned hearts to avoid contamination with protein from other cell types. To accomplish their goal, they used several protein separation methods, first separating the proteins into a neutral cytosolic and an acid soluble contractile protein fraction, and then separating the proteins in each fraction by 2 dimensional electrophoresis, first by a pH gradient, then by a size gradient. The authors used a rigorous approach, using several pH ranges for separation and loading the gels with several different amounts of protein in the attempt to maximize separation and identification of as many proteins as possible. They then identified the protein in each of 135 spots on the gels by using a technique called "peptide mass fingerprinting". This method employs mass spectrometry to very accurately measure the mass of peptides and their component amino acids from the digested proteins. A positive protein identification was made when multiple peptides from a single protein within a spot were identified by matching the masses of the peptides with those of each protein listed in an online database, or when the identification was ambiguous, by matching their derived sequence. From the changes in intensity of the spots on the silver stained gels performed on cell samples from each of 4 rabbits, they then deduced statistically significant changes in abundance of a particular protein. By limiting the number of protein spots to those that they were sure of, by demanding accuracy in mass spectrometric identification of the protein in the spot, and by estimating the change in density of the spot (protein abundance) for each of 4 animals per treatment group, the investigators have ensured the accuracy of the changes in abundance. Even so, the number of proteins examined was small.
The potential is great for new information to come from a proteomic study. For instance, by taking a broad survey of changes in the abundance of cell proteins one can ascertain where in the cell proteins are changing as a result of the physiological intervention, as Arrell et al did by discovering that mitochondrial proteins were among those to be changed by the preconditioning agents. In addition, one can determine what known cell functions are served by the proteins identified to change. Another example is the potential to identify novel protein participants in the response, thereby keying in on new biomarkers or unidentified members of signaling pathways that might serve as drug targets. Four spots in the study by Arrell et al were not matched to any known rabbit protein, and some of these were changed dramatically by the preconditioning agents, suggesting their potential importance. One of the most significant potentials from a proteomic study comes from the ability to examine a particular protein in great molecular detail. This is exemplified in the study by Arrell et al by their demonstration that the ATP synthase ß subunit is multiply phosphorylated by adenosine. This novel finding alone likely justifies the time, work, and expense of the study. It is known that several kinase signaling pathways, including Erk and Akt participate in preconditioning, but phosphorylation of which proteins is essential for preconditioning is uncertain.3,4
Some limitations of current proteomic methodology should be mentioned because they reflect on the ability of a proteomic study to aid in understanding a complex physiological problem. First, to detect proteins on gels and to identify them by mass spectrometry, a minimum amount of protein is required. There are also considerable difficulties in separating and identifying membrane proteins by this methodology. These issues may explain why Arrell et al were able to identify only 135 spots despite their thorough and varied separation strategies. This is obviously far fewer than the proteins produced by the genome that comprises on the order of 30 000 genes, and there is every likelihood that many of the proteins that were not evaluated in this study are important to preconditioning. An additional problem is that proteome databases for animal species like that for rabbit are not as complete as those for human and mouse, making the choice of experimental model important to enhance the outcome of proteomic studies. One method to improve the number of proteins detected is to study a less complex proteome by purifying proteins in a specific cell fraction or organelle. As a pertinent example, while the article by Arrell et al was under review, a study was published that examines changes in abundance of the proteins in mitochondria isolated from intact rabbit hearts subjected to ischemia with or without preconditioning.8 Changes were identified in 22 mitochondrial proteins, 7 from the respiratory chain including 2 ATP synthase subunits (but not the ß subunit).
Another major factor that limits the number of proteins identified is the degree to which the proteome undergoes posttranslational modifications (PTM’s). Because the mass of a particular peptide is altered by phosphorylation, glycation, prenylation, methylation, or oxidation of amino acid residues (eg, nitrotyrosine, S-nitrosation, S-glutathiolation), the mass of the peptide may fail to match that in online databases unless they are specifically queried. The myriad of possible PTM’s dramatically complicates the proteome. As an example, 1 cysteine residue in a serum protein was found to exist in at least 14 different adducts or oxidation states.9 PTM’s are of supreme physiological significance because many, like phosphorylation, change protein structure-function relationships. This makes solving the proteome of particular importance to understanding fundamental biological mechanisms. One potential solution is to concentrate on 1 particular modification such as phosphorylation. Immobilized metal affinity column chromatography (IMAC) was used by Arrell et al to concentrate for further study the phosphorylated ATP synthase subunit. In another new publication, Li et al10 used IMAC to look for phosphorylated proteins in cultured rat ventricular myocytes exposed to diazoxide. They found changes in phosphorylation of 5 proteins, but not in ATP synthase subunits. Because reactive oxygen and nitrogen species arising during ischemia have been implicated in preconditioning, oxidative PTM’s could introduce key protein function changes. Many oxidative PTM’s, eg, S-nitrosation and S-glutathiolation, are reversible and induce changes in protein function, and changes in these adducts would be missed in most studies because the proteins are typically prepared under thiol reducing conditions. Many new techniques are available, however to examine oxidative PTM’s.11
How does the data from the proteomic approach used by Arrell et al aid in understanding the physiology of preconditioning, and what other approaches would provide further information? As indicated by the authors, inhibitors of protein synthesis do not prevent preconditioning in cell systems. As the authors discuss, this does not mean that the changes in protein abundance observed are meaningless to preconditioning. Explanations include the possibility that PTM’s mentioned above, could have caused a portion of a particular protein to be shifted elsewhere on the gel, leading to an apparent decrease in abundance. Secondly, changes in the degradation of rapidly turning over proteins could have up or down regulated its abundance. A characteristic of proteomic studies is that they indicate some change in a protein, but not why it changes. Another characteristic is that, proteomic studies generate many new questions and spin-off studies, such as those that will likely address the importance of ATP synthase phosphorylation. Whether PTM’s or protein degradation are essential for preconditioning will have to be examined by future studies. Other studies will need to address whether mitochondrial proteins change similarly in cultured cardiomyocytes as they do in intact preconditioned ischemic hearts, and will determine whether the changes observed are essential for preconditioning. Preconditioning is a universal characteristic of ischemic cells, suggesting that much could be learned by establishing commonality among changes observed in different cell types. As an example, 2 new studies have been published on proteomic changes induced by pharmacological preconditioning of neurons in culture12 and ischemic brain.13
Another characteristic of proteomic studies is that they tend not to be hypothesis driven, but rather are of a descriptive nature. Although important discoveries may result, genomic and proteomic studies often are not enthusiastically supported by review committees of individual research grant applications. Four years ago, the NHLBI funded 10 cardiovascular proteomics centers that were charged with developing proteomics technology as applied to cardiovascular disease. It is notable that the authors of the study by Arrell et al (as well as the authors of this editorial) are supported by this program. Each of these centers are developing technology for different aspects of proteomic studies including, protein separation methods, mass spectrometry, PTM identification and detection, array technology, bioinformatics, and systems biology. Maximal quality and information coming from proteomics studies requires varied expertise and advanced technology, best ensured by a center approach. Physiological proteomics has a great potential to add significantly to our understanding of physiological mechanisms, and therefore disease and its treatment. Further technology development within centers will be necessary to deal with the complexity as well as the promise of the proteome.
Sources of Funding
The authors are supported in part with funds from the National Heart, Lung, and Blood Institute sponsored Boston University Cardiovascular Proteomics Center (contract N01-HV-28178), the National Center for Research Resources (grant P41RR10888-6), and the National Institutes of Health.
The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association.
Murry CE, Jennings RB, Reimer KA. Preconditioning with ischemia: a delay of lethal cell injury in ischemic myocardium. Circulation. 1986; 74: 1124–1136.[Abstract/Free Full Text]
Yellon DM, Alkhulaifi AM, Pugsley WB. Preconditioning the human myocardium. Lancet. 1993; 342: 276–277.[CrossRef][Medline] [Order article via Infotrieve]
Downey JM, Cohen MV. Reducing infarct size in the setting of acute myocardial infarction. Prog Cardiovasc Dis. 2006; 48: 363–371.[CrossRef][Medline] [Order article via Infotrieve]
Halestrap AP. Calcium, mitochondria and reperfusion injury: a pore way to die. Biochem Soc Trans. 2006; 34: 232–237.[CrossRef][Medline] [Order article via Infotrieve]
Arrell DK, Elliott ST, Kane LA, Guo Y, Ko YH, Pedersen PL, Robinson J, Murata M, Murphy AM, Marban E, Van Eyk JE. Proteomic analysis of pharmacological preconditioning: novel protein targets converge to mitochondrial metabolism pathways. Circ Res. 2006; 99: 706–714.[Abstract/Free Full Text]
Loscalzo J. Proteomics in cardiovascular biology and medicine. Circulation. 2003; 108: 380–383.[Free Full Text]
Domon B, Aebersold R. Mass spectrometry and protein analysis. Science. 2006; 312: 212–217.[Abstract/Free Full Text]
Kim N, Lee Y, Kim H, Joo H, Youm JB, Park WS, Warda M, Cuong DV, Han J. Potential biomarkers for ischemic heart damage identified in mitochondrial proteins by comparative proteomics. Proteomics. 2006; 6: 1237–1249.[CrossRef][Medline] [Order article via Infotrieve]
McComb ME, Huang H, Huang H, Skelton TP, Steinberg MH, Chui DHK, Lim A, Prokaeva T, Connors LH, Skinner M, Cohen R, and Costello CE. Probing protein cysteine post-translational modifications. Proceedings of the 52nd ASMS Conference on Mass Spectrometry and Allied Topics; 2004 May; Nashville: available at http://www.asms.org/aspfolder/ASMSAbstracts.html.
Li H, Xiao YB, Gao YQ, Yang TD. Comparative proteomics analysis of differentially expressed phosphoproteins in adult rat ventricular myocytes subjected to diazoxide preconditioning. Drug Metabol Drug Interact. 2006; 21: 245–258.[Medline] [Order article via Infotrieve]
Redox Proteomics. 2006. John Wiley & Sons, Inc., Hoboken, N.J.
Meloni BP, Tilbrook PA, Boulos S, Arthur PG, Knuckey NW. Erythropoietin preconditioning in neuronal cultures: signaling, protection from in vitro ischemia, and proteomic analysis. J Neurosci Res. 2006; 83: 584–593.[CrossRef][Medline] [Order article via Infotrieve]
Chen DM, Xiao L, Cai X, Zeng R, Zhu XZ. Involvement of multitargets in paeoniflorin-induced preconditioning. J Pharmacol Exp Ther. 2006. In press.