Article pubs.acs.org/jpr
Release of Tissue-specific Proteins into Coronary Perfusate as a Model for Biomarker Discovery in Myocardial Ischemia/Reperfusion Injury Stuart J. Cordwell,*,†,‡,§ Alistair V. G. Edwards,‡,§ Kiersten A. Liddy,† Lia Moshkanbaryans,‡ Nestor Solis,† Benjamin L. Parker,‡ Andy S. C. Yong,∥ Clement Wong,∥ Leonard Kritharides,∥ Brett D. Hambly,‡ and Melanie Y. White†,§ †
School of Molecular Bioscience and ‡Discipline of Pathology, School of Medical Sciences, The University of Sydney, New South Wales, Australia 2006 ∥ Department of Cardiology, Concord Repatriation General Hospital, Australia 2137 S Supporting Information *
ABSTRACT: Diagnosis of acute coronary syndromes is based on protein biomarkers, such as the cardiac troponins (cTnI/ cTnT) and creatine kinase (CK-MB) that are released into the circulation. Biomarker discovery is focused on identifying very low abundance tissue-derived analytes from within albumin-rich plasma, in which the wide dynamic range of the native protein complement hinders classical proteomic investigations. We employed an ex vivo rabbit model of myocardial ischemia/reperfusion (I/R) injury using Langendorff buffer perfusion. Nonrecirculating perfusate was collected over a temporal profile of 60 min reperfusion following brief, reversible ischemia (15 min; 15I/60R) for comparison with irreversible I/R (60I/60R). Perfusate proteins were separated using two-dimensional gel electrophoresis (2-DE) and identified by mass spectrometry (MS), revealing 26 tissue-specific proteins released during reperfusion post-15I. Proteins released during irreversible I/R (60I/60R) were profiled using gel-based (2-DE and one-dimensional gel electrophoresis coupled to liquid chromatography and tandem mass spectrometry; geLC−MS) and gel-free (LC−MS/MS) methods. A total of 192 tissue-specific proteins were identified during reperfusion post-60I. Identified proteins included those previously associated with I/R (myoglobin, CK-MB, cTnI, and cTnT), in addition to examples currently under investigation in large cohort studies (heart-type fatty acid binding protein; FABPH). The postischemic release profile of a novel cardiac-specific protein, cysteine and glycine-rich protein 3 (Csrp3; cardiac LIM domain protein) was validated by Western blot analysis. We also identified Csrp3 in serum from 6 of 8 patients postreperfusion following acute myocardial infarction. These studies indicate that animal modeling of biomarker release using ex vivo buffer perfused tissue to limit the presence of obfuscating plasma proteins may identify candidates for further study in humans. KEYWORDS: biomarkers, cysteine and glycine-rich protein 3 (Csrp3), ischemia/reperfusion, Langendorff perfusion, myosin-binding protein C
■
INTRODUCTION Cardiovascular disease affects an estimated one in three people in the USA, with approximately 1 million deaths per annum directly related to acute myocardial infarction (AMI).1 Clinically silent or unrecognized AMI, which is undetected by either patient or physician, is thought to account for many further cases.2 Early diagnosis of an ischemic episode may therefore open a wider range of effective treatment options to reduce ischemia/reperfusion (I/R) injury, including limitation of eventual infarct size and a reduction of contractile © 2012 American Chemical Society
dysfunction in surviving myocytes. Biomarkers (e.g., proteins) of disease must be reproducibly obtained by standardized methods readily acceptable to the patient, and easily interpreted clinically. Most importantly, the marker must be specific to the tissue and disease tested, detectable at a high level of confidence, and present at a point in disease progression where effective therapy can be applied, thus improving the Received: July 21, 2011 Published: January 18, 2012 2114
dx.doi.org/10.1021/pr2006928 | J. Proteome Res. 2012, 11, 2114−2126
Journal of Proteome Research
Article
reduces the dynamic range of samples while not reducing the sensitivity of detection, therefore allowing much simpler biomarker discovery. This approach has been used effectively in similar studies employing rodent models.33 The major shortcoming of previous studies has also been a lack of temporal resolution, an issue that requires addressing given the impact that release profiles have on the application of biomarkers in the clinic. In this study, we analyzed buffer perfusate proteomes from rabbit hearts subjected to brief, reversible (15 min ischemia) and prolonged, irreversible (60 min ischemia) I/R injury to produce a temporal profile of tissue-specific protein release. The identified proteins comprise known AMI biomarkers, as well as novel proteins with application for testing in patient samples for the early detection of AMI. A novel protein with near cardiac specificity (cysteine and glycine rich protein 3 [Csrp3]) was chosen for validation and was observed in samples from ST elevation myocardial infarction (STEMI) patients post-I/R injury.
eventual prognosis. Human body fluids are therefore the most prevalent source of biomarkers currently applied clinically,3−5 and proteomics offers great potential for the discovery of new protein- or peptide-based markers.6 Current diagnosis of AMI relies on the use of blood-borne biomarkers accompanied by clinically determined symptoms (e.g., chest pain) and electrocardiographic perturbations. Several AMI biomarkers have been proposed, including myoglobin (MyG), creatine kinase-MB (CK-MB), heart-type fatty acid binding protein (FABPH), B-type natriuretic peptide signal peptide7 and the current gold standards cardiac troponin I (cTnI) and T (cTnT).8−10 MyG is an abundant cytoplasmic protein found in cardiac and skeletal muscle, and plasma levels are greatly increased within 1−3 h post-AMI.11 MyG is thus recommended for those presenting up to 6 h postchest pain onset.12,13 Specificity however is poor, with skeletal muscle injury contributing to false positives.14 Similarly, CK-MB is rapidly induced post-AMI, but is also plagued by poor specificity and assay sensitivity.15,16 FABPH is rapidly induced (20 min postischemia), but is also found in several other organs.17 The status quo is that AMI is diagnosed most frequently using cTnI and/or cTnT, since phosphorylation and cleavage during injury lead to release from the myofilament.18,19 Other proteins released from the contractile apparatus during injury, at least in animal models, include myosin light chain 2.20 The cTns have the advantage of cardiac-specificity and low circulating plasma levels in healthy individuals.21 Despite recent advances in sensitivity testing for cTnI22 and cTnT23 at earlier stages of AMI, their slow release typically hinders accurate diagnosis at a stage where effective myocardial salvage can be conducted and where subsequent contractile dysfunction may be limited. The cTns remain elevated for some 7−10 days, thus also limiting their effectiveness in cases of reinfarction. In these instances, CK-MB, due to a shorter circulating half-life, is often more suitable.24 While these markers have revolutionized diagnosis of AMI, milder contractile dysfunction following brief periods of nonlethal ischemia (myocardial stunning) is also clinically relevant. Stunning can occur as an adjunct to AMI in the relatively better-perfused areas of tissue, as well as independently of AMI as a consequence of a variety of perfusion supply/ demand mismatch situations.25,26 Stunning and resolved shortterm ischemia can be clinically indicative of a long-term increased risk of AMI26 and the ultimate aim diagnostically is to detect ischemic events prior to permanent damage. Therefore, the search for additional biomarkers of myocardial ischemia remains clinically relevant.27,28 The difficulties in identifying new tissue- or tumor-specific biomarkers in human plasma have been reviewed extensively.29−31 These typically include the inability to detect lowlevel biomarkers due to the dynamic range of protein abundance (suggested to span more than 10 orders of magnitude3). Current proteomic technologies can only successfully resolve proteins within 3 to 4 orders of magnitude,3,28 and consequently numerous studies have utilized techniques to remove highly abundant proteins22 of which comprise 99% of the plasma proteome28prior to analysis. With removal of the highly abundant classical plasma proteins, however, potential disease-specific biomarkers can be lost.32 To avoid these obstacles, ex vivo animal models of I/R injury can be used as an alternative in the biomarker discovery phase. The substitution of blood with an aproteinacious crystalloid buffer during ex vivo Langendorff perfusion greatly
■
MATERIALS AND METHODS
Langendorff Perfusion of Rabbit Hearts
Experimental studies were performed using male adult New Zealand White rabbits (2.5 to 3.5 kg), according to protocols approved by the Animal Research Ethics Committee of The University of Sydney (approval no. K20/6−2009/3/5078). Following heparin administration (2,500 units IV), rabbits were euthanased with sodium pentobarbital (150 mg/kg IV). The aorta was cut approximately 1 cm above the aortic valve, and the heart was rapidly removed from the thoracic cavity and placed into ice-cold saline solution (0.9% (w/v) NaCl, pH 7.4) to minimize injury during transition to a Langendorff perfusion apparatus (ADInstruments, Australia). Hearts were attached to the Langendorff system via insertion of a stainless steel cannula into the severed aorta, which was firmly tied in position with a 3.0 silk suture. Retrograde perfusion was performed at a constant pressure of 80 mmHg with modified Krebs-Henseleit buffer containing 16 mM glucose (37 °C (temperaturecontrolled organ-bath); pH 7.4, bubbled with 95% O2 and 5% CO2). The performance of the heart was monitored using a saline-filled latex balloon inserted into the left ventricle (LV) and passing through the mitral valve. The balloon was attached to a calibrated pressure transducer and the heart rates and LV developed pressure were recorded (LabChart, ADInstuments). The balloon was inflated with 0.9% saline solution (37 °C) to obtain a LV end-diastolic pressure of 8−10 mmHg under baseline conditions. The protocols for the induction of reversible injury (stunning; 15 min ischemia) and irreversible injury (infarction; 60 min ischemia) in isolated rabbit hearts were as previously described34,35 (detailed methods can be found in Supplementary Methods File 1, Supporting Information). At the end of the perfusion protocol, the heart was rapidly removed from the Langendorff apparatus. The presence of irreversible injury in perfused hearts was assayed by triphenyltetrazolium chloride (TTC) staining, as described previously36 (Supplementary Methods File 1). Collection of Perfusate
Perfusate was collected during baseline perfusion or during reperfusion (postischemia). An incision was made in the pulmonary trunk and a thin piece of tubing approximately 5 cm in length inserted to allow coronary effluent to flow. After 12 min of baseline perfusion, the next (final) 3 min of baseline 2115
dx.doi.org/10.1021/pr2006928 | J. Proteome Res. 2012, 11, 2114−2126
Journal of Proteome Research
Article
comparison to other database entries. Where inconclusive protein identification was obtained from MALDI-TOF MS, the spots were reanalyzed by MALDI MS/MS using the Q-STAR Elite (Supplementary Methods File 1). Amino acid sequences were deduced by the mass differences between y- and b-ion series using the Mascot script contained within Analyst and confirmed by manual interrogation. Sequences were then used to search Mascot and were confirmed in the Swiss-Prot and TrEMBL databases using the program BLASTP “short nearly exact matches”.39 A minimum of two matching peptides was required to provide confidence in a given identification.
perfusate was collected. Perfusate was collected during reperfusion at a constant volume of 30 mL. The time points varied during reperfusion from 3 to 9 min depending on both the contractile function and flow-rate required to maintain 80 mmHg. During ischemia, no perfusate was collected due to the no-flow model employed. Samples were stored at −80 °C. Perfusate Protein Preparation
Protein precipitation was required to remove salts present in the modified Krebs-Henseleit buffer and to concentrate proteins prior to analysis. The perfusate samples (30 mL) were precipitated in 8 volumes of ice-cold acetone. Samples were left at 4 °C for 16 h and then collected by centrifugation (4 °C, 17100× g for 50 min37). Precipitated proteins were left to air-dry (7−10 min) and then resuspended in either 100 μL 6 M urea, 20 mM dithiothreitol (DTT) for SDS-PAGE analysis and total protein digestion, or 460 μL of 2-DE buffer solution (5 M urea, 2 M thiourea, 2% (w/v) CHAPS, 2% (w/v) sulfobetaine 3−10, 2 mM tributylphosphine, 65 mM DTT, 1% (v/v) pH 3−10 carrier ampholytes (Bio-Rad, Hercules, CA), 40 mM Tris and a trace amount of bromophenol blue) containing phosphatase inhibitors (0.2% (v/v) okadaic acid), protease inhibitors (0.2% (v/v) protease cocktail inhibitor; Sigma, St. Louis MO) and endonuclease (300 U/mL).
SDS-PAGE-LC−MS/MS (geLC−MS) and Gel-free LC−MS/MS of Perfusate Proteins
SDS-PAGE gel lanes containing perfusate proteins were “sliced” into 12 bands of equal width, reduced, alkylated and trypsin digested overnight (Supplementary Methods File 1, Supporting Information). For gel-free LC−MS/MS, precipitated proteins were resolublised in 6 M urea, 2 M thiourea, 1% (w/v) SDS, 50 mM ammonium bicarbonate (pH 8.0) prior to reduction and alkylation. Proteins were then diluted in 5 volumes of 50 mM ammonium bicarbonate (pH 8.0) and digested with 1% (w/v) trypsin at 25 °C for 24 h. Prior to mass spectrometric analysis, the peptide mixtures from both approaches were concentrated and desalted with the use of prefabricated microcolumns containing Poros R2 (Perseptive Biosystems, Framingham MA) resin.40 Peptides were eluted from R2 columns with 1% (v/v) TFA and subjected to reversed phase LC−MS/MS (Supplementary Methods File 1). Approximately 3000 MS/MS spectra were generated from each replicate LC−MS/MS experiment. Tandem mass spectra were extracted, charge state deconvoluted and deisotoped in Mascot and analyzed using Mascot and X!Tandem (thegpm.org; version 2007.01.01.1). Data were searched against Oryctolagus cuniculus (Rabbit) and mammalian entries included in the Swiss-Prot and TrEMBL databases with the significance threshold set at p < 0.05. Searches were conducted with a fragment ion mass tolerance of 0.6 Da and a parent ion tolerance of 0.2 Da. Methionine sulfoxide and carbamidomethyl-cysteine were specified as variable modifications. Scaffold (vers. Scaffold_3_00_02, Proteome Software Inc., Portland, OR) was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted if they exceeded specific database search engine thresholds. Mascot identifications required ion scores greater than the associated identity scores of 30, 40, and 40 for doubly, triply and quadruply charged peptides, respectively. X!Tandem identifications required -Log (Expect Scores) scores of greater than 2.0. Protein identifications were accepted if they contained at least 1 identified peptide matching the above criteria. Proteins containing identical peptides that could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Within each experiment, many individual peptides were sequenced on multiple occasions. Proteins that were not present in technical replicates were excluded, as were peptides that were only sequenced once within a single individual data set. Both geLC−MS and LC− MS/MS replicate data sets generated approximately 200−300 peptides that satisfied these criteria. Since the data sets were comparatively small, all MS/MS spectra were also manually validated. To determine semiquantitative differences in proteins identified by LC−MS, we employed a “crude” spectral counting
Two-Dimensional Gel Electrophoresis (2-DE)
Proteins were separated using 17 cm precast immobilized pH gradient (IPG; Bio-Rad) strips across the pH range 3−10. Isoelectric focusing was conducted according to published techniques34 and proteins were focused for a total of 85 kVh. Proteins in IPG strips were reduced, alkylated and detergent exchanged and separated by second dimension SDS-PAGE, as previously described.34 Gels were fixed in 10% (v/v) methanol, 7% (v/v) acetic acid for 1 h, and then stained with Sypro Ruby (Bio-Rad) overnight. Gels were destained for 2 h in 10% (v/v) methanol, 7% (v/v) acetic acid and scanned using a Bio-Rad Molecular Imager FX. A total of 3 biological replicates and technical duplicates were analyzed for each time-point (6 gels per time-point). Changes in protein release during reperfusion from the two ischemic groups were compared from: (i) baseline perfusion; (ii) time-matched NITC samples; and (iii) 3 collection time-points during reperfusion (early (0−8R), midphase (30−38R) and late (55−60R)) postischemia. Image analysis was performed using PD-Quest software (Bio-Rad) as previously described34 (Supplementary Methods File 1, Supporting Information). Mass Spectrometry and Protein Identification following 2-DE
Protein spots were excised from 2-DE gels, processed by trypsin digest and subjected to MALDI-TOF MS peptide mass mapping and/or MS/MS as previously described38 (Supplementary Methods File 1, Supporting Information). Peptide mass mapping data were used to perform searches of Oryctolagus cuniculus (Rabbit) and mammalian entries included in the SWISS-PROT and TrEMBL databases (January 2010 release; 64998 entries), via the program Mascot (v.2.2; Matrix Science Ltd., London UK). Identification parameters included peptide mass tolerance of 0.1 Da, a maximum of one missed tryptic cleavage per peptide, and with the methionine sulfoxide and cysteine-acrylamide variable modifications checked. Identifications were based on the Mascot score and E-value, the number of matching peptide masses and the total percentage of the amino acid sequence that those peptides covered, in 2116
dx.doi.org/10.1021/pr2006928 | J. Proteome Res. 2012, 11, 2114−2126
Journal of Proteome Research
Article
Lack of Tissue-Specific Protein Release during Baseline Perfusion and in NITC
approach, based upon the total of identified peptides for a particular protein expressed as a proportion of all identified peptides from the relevant LC−MS experiment, as previously described.41,42
To provide confidence that the identified perfusate proteins from brief and prolonged I/R injury were specific to those protocols, we conducted analysis of two additional controls: (i) perfusate collected during the final 3 min of 15 min baseline perfusion (12−15BP) prior to the beginning of NITC, 15I/60R and 60I/60R; and (ii) perfusate collected from NITC hearts (75 min perfusion). Baseline perfusate samples were collected and analyzed by gel-based (2-DE and geLC−MS) and gel-free (total protein digest followed by LC−MS/MS) methods. Identification of proteins by MALDI-TOF MS (following 2DE; Supplementary Figure 2, Supporting Information) and LC−MS/MS showed that all baseline perfusate proteins were plasma-associated (Supplementary Table 3, Supporting Information) and demonstrated that no tissue-specific protein leakage occurred prior to the beginning of the I/R protocols. To further ensure myocardial proteins were not being released into the coronary vasculature simply in response to prolonged periods of ex vivo perfusion, a group of hearts was perfused continuously for 75 min (NITC) postbaseline. Following the baseline period, 30 mL aliquots of perfusate were collected at 3 distinct time points; early (15 min postbaseline); midstage (30 min postbaseline) and late (55 min postbaseline). Perfusate proteins were precipitated and analyzed by gel-based and gel-free methods. 2-DE analysis of perfusates from the 3 time points (Supplementary Figure 3, Supporting Information) revealed that no additional proteins appeared compared to gels derived from 12 to 15BP, indicating no tissue-specific protein leakage in NITC hearts. In addition, 2-DE revealed that high abundance plasma proteins, such as albumin, were gradually removed from the perfusate throughout the protocol. Both geLC−MS and LC−MS/MS supported these findings with no identifications of tissue-derived proteins and a reduction in the number of peptides identified corresponding to individual plasma proteins (data not shown). The reduction in protein abundance and peptide matches associated with plasma proteins throughout the NITC protocol is largely related to the continuing replacement of blood with nonproteinacious buffer, resulting in a “wash-out” effect. Importantly, however, this observation also provides additional confidence that even low levels of tissue-specific proteins do not leach into the perfusate, since their presence would be more readily detectable at lower concentrations of abundant plasma proteins.
Western Blotting
Perfusate protein samples for Western dot-blot probing were spotted on to a prewetted polvinylidene difluoride (PVDF) membrane. All PVDF membranes were blocked overnight in 5% bovine serum albumin (Sigma) and probed with a 1/20000 dilution of goat polyclonal anti-Csrp3 antibody (Abcam, Cambridge UK) raised against the peptide 150CGKSLEDTNVTDKD-163 from human Csrp3. Proteins reactive to goat IgG were detected using a 1/40000 dilution of donkey-antigoat immunoglobulin (Abcam) followed by incubation in Supersignal West Pico Chemiluminescent substrate according to the manufacturer’s instructions (Pierce, Rockford IL). Tissue homogenates (neat and 1/10 dilution) were used as a positive control. We also performed Western dot-blots using an anti-cTnI antibody (Abcam, dilution 1/ 20000) as a positive control. The resulting blots were visualized using Hyperfilm ECL (GE LifeSciences, Amersham UK) and a CP100 film processor (AGFA, Mortsel Belgium). Densitometry was performed on images from triplicate Western blots using PD-Quest as described above. To determine whether Csrp3 could be detected in human patient serum samples, we conducted a small pilot Western dotblot study against 8 patient sera obtained from Concord Hospital, Sydney. All clinical samples (collection and use was approved by the Concord Hospital Institutional Ethics Committee) were from ST elevation myocardial infarction (STEMI) patients collected 1−3 days postreperfusion (Supplementary Table 1, Supporting Information; median age = 49 yrs). Negative controls (n = 5; median age = 38 yrs) were collected from non-STEMI volunteers with no prior history of AMI. Patient sera (200 μL) were precipitated by TCA/acetone and the proteins collected by centrifugation at 17100× g (4 °C, 50 min). Proteins were resuspended in 100 μL 6 M urea, 2 M thiourea, 10 mM DTT and 1 μL dotted to prewetted PVDF membrane. Anti-Csrp3 antibody was added and blots processed as described above.
■
RESULTS
LV Function and Necrosis
Hemodynamic data were collected during crystalloid perfusion for each experimental group. The rate pressure product (RPP), an indicator of myocardial viability, was significantly reduced following both 15 and 60 min ischemia/60 min reperfusion (Supplementary Table 2, Supporting Information). RPP reached 62.0 ± 7.0 at 60 min reperfusion post 15I, and only 29.0 ± 3.0 at 60 min reperfusion post 60I. These data reflect reversible and irreversible I/R injury, respectively. We also examined horizontal sections of the myocardium using TTC staining to confirm the presence/absence of necrosis (Supplementary Figure 1, Supporting Information). Decreased TTC staining (59.3 ± 5.8% of the LV mass was TTC negative) was observed in hearts subjected to prolonged ischemia (60I/ 60R), indicative of irreversible injury. In hearts subjected to brief ischemia followed by reperfusion (15I/60R) and NITC hearts, an insignificant amount of TTC negative myocardium was observed (
