Sprague Dawley rats were obtained from Charles River Laboratories (Wilmington, MA). Experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee at Vanderbilt University Medical Center (Protocol ID#: M/09/269). The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health and the standards of the Association for Assessment and Accreditation of Laboratory Animal Care. All surgeries were performed under anesthesia as detailed below. Every animal received post-operative pain management, and all efforts were made to minimize suffering.
Induction of Myocardial Infarction (MI)
Male Sprague Dawley rats (175–200 g body weight) were anaesthetized intraperitoneally with 30 mg/kg of pentobarbital. Absence of a response to pinching the toe was used as an indicator of the appropriate level of anesthesia. Rats were then rapidly intubated and mechanically ventilated (tidal volume, 1 mL/100 g body weight; ventilation rate, 65 strokes/min) by a constant volume small animal ventilator (Model 683, Harvard Apparatus). A left thoracotomy was performed at the fourth intercostal space, and the left coronary artery was then ligated by irreversible tightening of a 6–0 suture loop. MI was confirmed by regional cyanosis of the myocardial surface distal to the suture, accompanied by S-T segment elevation on the electrocardiogram. Following successful induction of MI, the chest cavity was compressed to evacuate any air before being tightly sealed. Sham-operated animals underwent the same surgical procedure with the exception that the left coronary artery was not ligated. The rats were given buprenorphine 0.05 mg/kg post-operatively, then Q 8–12 hours post-operatively by subcutaneous injection. The rats that survived through day 7 post-MI were randomly assigned to various groups as outlined in the experimental design.
All rats were maintained on a normal chow diet of 12% calories from fat, 28% calories from protein, and 60% calories from carbohydrate (LabDiet) prior to coronary ligation surgery. Following surgery, rats in the study examining effects of GGF2 early after MI were also randomly assigned to either a normal diet or a high saturated fat diet of 43.5% calories from fat, 22.2% calories from protein, and 34.3% calories from carbohydrate (LabDiet) beginning 7 days post-MI and continuing until completion of the study. All rats in the study examining effects of GGF2 late after MI were continuously maintained on a normal chow diet.
Recombinant GGF2 (96.0% purity by SEC-HPLC) was provided by Acorda Therapeutics, Inc. and stored at 4°C. GGF2 was used at low dose (0.625 mg/kg) or high dose (3.25 mg/kg) based upon rat weights taken at each day of treatment. GGF2 was administered via the right lateral tail vein every second day to surviving MI rats starting one or 8 weeks after MI. Vehicle-treated rats received equal volume injections of sterile GGF2 diluent (20 mM histidine, 100 mM arginine, 100 mM sodium sulfate, 1% mannitol, pH 6.5).
Positron Emission Tomographic (PET) Imaging
Rats in the study examining effects of GGF2 late after MI underwent microPET imaging to evaluate myocardial metabolic changes and myocardial viability post-infarction. Anesthesia was induced using 2–3% isoflurane/97–98% oxygen supplied via nosecone. Fasting rats were administered an intraperitoneal injection of 18F-fluorodeoxyglucose (18FDG) and anesthesia was discontinued. After a sixty minute rest period to allow for uptake of radioisotope tracer into tissues, anesthesia was re-induced for the imaging procedure. A small-animal tomograph (Siemens MicroPET Focus 220) acquired full body PET images for approximately 30–45 minutes. Images were later analyzed with Amide software. Mean cardiac standard uptake values (SUVs) were calculated from transverse slices (0.47 mm thickness) by drawing 3-dimensional, free hand regions of interest that completely encircled the heart. Regions of interest approximately 0.5 cm in diameter were also drawn on each animal's right shoulder skeletal muscle. Mean cardiac SUVs were then normalized to shoulder muscle SUVs. After completing the PET procedure, animals were placed in a recovery cage and monitored until fully ambulatory. Rats were scanned at 8, 10, and 12 weeks post-MI. All scanning and interpretation was performed blinded to the treatment group.
Assessment of Cardiac Function by Echocardiography
Transthoracic echocardiographic images of hearts from all groups of rats were obtained using a 12-MHz ultrasound probe (model s12, Agilent Technologies) and an echograph (SONOS 5500, Hewlett Packard) while rats were immobilized under anesthesia (1.5–2.0% isoflurane). For M-mode recordings, the parasternal short-axis view was used to image the heart in two dimensions at the level of the papillary muscles. End-diastolic and end-systolic LV cavity dimensions were measured using software resident on the ultrasonograph. LV fractional shortening (FS) was calculated from M-mode-derived left ventricular inner diameters in diastole (LVIDd) and systole (LVIDs) using the formula (LVIDd – LVIDs)/LVIDd x100%. All imaging and interpretation of results was done blinded to treatment group. In all studies rats underwent serial echocardiograms at baseline prior to initiation of treatment, and subsequently at two week intervals until sacrifice.
All histological evaluation was performed in a blinded fashion. Paraffin-embedded sections (4 µm) of hearts from all of the groups were stained with Masson's trichome. Cardiac fibrosis in the LV remote from the site of infarction was measured in using NIH ImageJ software and was expressed in arbitrary units as a percentage of the total area of the tissue section stained.
Measurement of Myocardial Oxidative Stress
We examined myocardial oxidative stress by measuring myocardial protein carbonylation as previously described 
using OxyBlot protein oxidation detection kit (Millipore) according to manufacturers instructions. Tssue lysates were prepared from non-infarcted region of the left ventricle representing 15 µg of protein. Blots were developed using fast tablet (BCIP/NBT; Sigma-Aldrich) and quantified using Scion Image (PC version of Macintosh-compatible NIH Image) software. No non-specific background binding of the primary or secondary antibodies was found.
Western Blot Analysis of ErbB2 and ErbB4 Expression
Remote, non-infarct left ventricular tissue from all groups was homogenized using a liquid nitrogen chilled mortar and pestle, in a liquid nitrogen chilled CryoGrinder. Homogenization was performed using RIPA buffer (Millipore), supplemented with protease inhibitor cocktail (Roche). Proteins were quantified with Bradford assay (Bio-Rad). 50 micrograms of protein in SDS sample buffer was heated using a Dry Bath incubator (Fisher scientific) for 10 minutes and after centrifugation, loaded onto 4–12% NuPage gels (Invitrogen). After electrophoresis, proteins were transferred to an immunoblot PVDF membrane, using the Semi-Dry transfer apparatus at 20 V for 30 mins. Membranes were blocked using PBST-5% BSA and then incubated overnight at 4°C with antibodies directed against ErbB2 and ErbB4 (clone F-11 and clone C-18 respectively, Santa Cruz Biotechnology), each diluted 1
500. Secondary horseradish peroxidase-conjugated antibody (Cell Signaling) was applied for 1 hour at room temperature. Blots were visualized using the SuperSignal West Pico chemiluminiscent substrate (Pierce) and analyzed with NIH ImageJ densitometry software.
Total RNA was extracted from the non-infarcted LV using Trizol (Invitrogen Co.) and RNeasy columns (Qiagen, Valencia, CA), according to the manufacturers' instructions, followed by DNAse treatment (Qiagen). RNA quality measurements using a 2100 Bioanalyzer (Agilent, Santa Clara, CA) and further sample processing were performed in the Genome Sciences Resource (GSR) Core at Vanderbilt. Processed and labeled samples were subsequently hybridized to the rat Gene 1.0 ST whole-transcriptome array (Affymetrix, Santa Clara, CA). Three biological replicates were used for each sample type as follows: Sham-operated, early MI rats (vehicle and low dose [0.625 mg/kg] GGF2-treated), and late MI rats (vehicle and high dose [3.25 mg/kg] GGF2-treated). Therefore, there were a total of 15 samples (5 sample types, n
3). Only rats on the standard diet were chosen for transcriptome analysis, due to minimal differences in GGF2-mediated responses in rats fed a high versus low-fat diet.
Subsequent to import of raw data (.CEL) files, Robust Multi-chip Average (RMA) normalization 
followed by statistical analysis was performed using Partek Genomics Suite version 6.6 (Partek Incorporated, St. Louis, MO). Quality assessment was performed based on Affymetrix internal controls, Box-and-Whisker plots, histograms, PCA (principal components analysis), and standard deviation of biological replicates. Based on these criteria, one sample (one of the low-dose, early GGF2-treated biological replicates) was identified as an outlier. We were unable to determine the source of the variation, which could have been technical or biologically relevant, and therefore performed analyses with and without the outlier sample. Following normalization and quality assessment, Partek was used to perform pairwise comparisons of average group values and one-way ANOVA. Only transcripts that resulted in a fold-change of at least 1.5 and p value of less than 0.05 were considered as significantly altered.
Gene functions, as presented in significant gene list tables, were determined using publicly available records using NCBI Entrez Gene, Stanford SOURCE, Aceview, and Pubmed databases. Statistical analyses (including correction for multiple hypothesis testing) for identification of overrepresented ontologies and functions were performed using Ingenuity Pathway Analysis (IPA) software (ingenuity Systems, Redwood City, CA).
For comparison of our gene expression results to previous studies, raw microarray data (CEL or text files) from four rat MI studies were downloaded from Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo/
). Gene expression data from autopsied human hearts and failing hearts collected during heart transplant surgeries were also obtained from ArrayExpress Archive database (www.ebi.ac.uk/arrayexpress
under accession number E-TABM-480 
, and the GEO database under accession numbers GSE1145 
, GSE1869 
, and GSE5406 
. Combined, our study, the four other rat studies, and the human studies included 363 individual biological samples. Raw data from other studies were processed similarly to ours, including RMA normalization using Partek Genomics Suite and fold-change of 1.5 or greater and p value ≤0.05 considered as altered, to make the various studies more comparable.
To link our gene expression results (i.e., gene expression profiles of GGF2-treated early and late MI rats) to results obtained from the various other animal and human studies, a perl script was written to match genes across the various lists by accession number (for same species results) or gene symbol (for inter-species results comparisons). If neither the accession number nor gene symbol matched perfectly, then the gene descriptions were compared and manually checked if the difference did not exceed 25%. If a data set contained more than one match for a single data point in a published study, only the top three matches were recorded, with the best match ordered first. This perl script was later modified to compute averages for the multiple matches. This method would be expected to fail to identify some homologous genes in which there are name designation differences between species (e.g., the mouse version of human IL-8 is referred to as KC or GRO). However, our intention was to greatly minimize false positives and thus compare only those genes that were truly the same between the various data sets. To ensure that this was indeed the case, all final gene lists representing overlap between study results were also manually examined and any genes that were computationally misidentified were subsequently removed.
Real-time RT-PCR (qRT-PCR)
Real-time RT-PCR was performed to quantify transcript levels for a subset of genes using the CFX96 C-1000 (Bio-Rad, Hercules, CA) using SYBR Green I dye (QIAGEN), per manufacturer's instructions. Briefly, each 25-μl reaction contained 100 ng of RNA, 2.5 μl of primers (Quantitect Primer Assays; QIAGEN), 12.5 μl of SYBR Green PCR master mix and 0.25 μl of reverse transcriptase. Negative controls containing water instead of RNA were concomitantly run to confirm that the samples were not cross-contaminated. Targets were normalized to reactions performed using Quantitect TPT1 primers (QIAGEN), which amplify a translationally-controlled transcript. Fold change was determined using the comparative threshold method 
Difference Gel Electrophoresis and Differential-Display Proteome Analysis
Myocardial tissue originating from the remote, viable LV in all groups of rats was immediately frozen in liquid nitrogen and stored at –80°C until use. Frozen viable LV tissues were homogenized in lysis buffer consisting of 7 M urea, 2 M thiourea, 30 mM Tris, 5 mM magnesium acetate, 4% CHAPS, and 58 mM DTT using a ratio of 1 g tissue:10 ml lysis buffer. The lysate was centrifuged at 12,000×g for 1 h at 10°C. The pellet was discarded and the protein concentration of the supernatant was determined using a 2-D Quant protein assay kit (Amersham Biosciences/GE Healthcare, Piscataway, New Jersey).
250 µg from each of the samples were precipitated with methanol/chloroform 
and resuspended in 40 µL labeling buffer (7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris, 5 mM magnesium acetate). One-third of each sample (83 µg) was removed and pooled into a separate tube for the mixed-sample internal standard. The remaining 167 µg of each sample was separately labeled with 200 pmoles of either Cy3 or Cy5 NHS-ester minimal labeling reagents (2 µL of a 100 pmol/µL in NN di-methyl foramide) for 30 min on ice in the dark, followed by quenching with 2 µL of 10 mM lysine for 10 min on ice in the dark. The pooled sample mixture (830 µg total) was similarly labeled with 1,000 pmol of Cy2 and quenched with 10 µL of 10 mM lysine. Individual Cy3- and Cy5-labeled samples were combined with an equal portion of the Cy2-labeled internal standard for a total of approximately 300 µg on each gel.
All equipment was manufactured by GE Healthcare/Amersham Biosciences (Piscataway, New Jersey) unless otherwise noted. The samples were separated by standard 2D gel electrophoresis using a manifold-equipped IPGphor first-dimension isoelectric focusing unit and 24 cm 4–7 immobilized pH gradient (IPG) strips, followed by second-dimension 12% SDS-PAGE homogenous on hand-cast gels that had one plate pre-silanized to ensure accurate robot picking in subsequent steps, using an Ettan DALT 12 unit according to the manufacturer's protocols. The Cy2 (mixed standard), Cy3 (sample X) and Cy5 (sample Y) components of each gel were individually imaged at 100 µm resolution with mutually-exclusive excitation/emission wavelengths using a Typhoon 9400 Variable Mode Fluorescence Imager. A Sypro Ruby total protein post-stain (Invitrogen/Molecular Probes) was used to ensure accurate protein excision, as the low stoichiometry of Cy-dyes label only 1–3% of the total protein.
DeCyder software v6.5 was used for simultaneous comparison of abundance changes across all sample pairs with statistical confidence and without interference from gel-to-gel variation 
. Normalized volume ratios were calculated for each resolved protein on every gel, and the internal standard signals were used to normalize and compare these ratios across all five gels to allow for the calculation of average abundance changes and p-values for the variance of these ratios for each protein-pair across all five independent gels.
Mass Spectrometry and Database Interrogation
Proteins of interest were robotically excised, digested into peptides in-gel with modified porcine trypsin protease (Trypsin Gold, Promega) and peptides extracted using the automated Ettan Spot Handling Workstation (GE Healthcare) using 20 μL 20 mM NH4HCO3 containing 0.01 μg/μL Trypsin Gold (Promega) for 3 h at 37°C for protein digestion. Peptides were extracted by two rounds of incubation with 60% acetonitrile, 0.1% trifluoroacetic acid, dried and reconstituted in 15 μL 0.1% formic acid and placed into autosampler vials. 5 μL of each peptide hydrosylate was separately analyzed by C18 reverse-phase LC-MS/MS using a Thermo LTQ ion trap mass spectrometer equipped with a Thermo MicroAS autosampler and Eksigent HPLC nanoLC pump system, nanospray source, and Xcalibur 2.0 instrument control using standard data-dependent methods. Tandem MS data were analyzed with the Sequest algorithm, searching the IPI_rat-v342 database (Apr 2008) that contained a concatenated reverse decoy database to estimate false-discovery rates. Search results were filtered by cross-correlation scores (<1.0 for singly-charged peptides, <1.8 for doubly-charged, and <2.5 for triply-charged) with an overall 2.4% false-discovery rate. Protein identifications were based on a minimum of 2 peptides passing these criteria for each protein. The number of unique peptides and the total number of peptide identifications (spectral counts) for each protein are listed, and the predicted MW and pI were correlated with the relative gel position from which the protein was excised.
Gene expression and real-time RT-PCR statistical analyses were performed as described above. All other statistical analyses were completed using statistical software R version 2.13.0 (2011–04–13). Echocardiography data were analyzed by using the generalized least squares model. This linear regression model accounts for correlation across time within animals and was fit to investigate whether GGF2 treatment and diet affected the 35-day outcome (FS %). Estimated mean values were fit using restricted maximum likelihood or REML and are presented with its 95% confidence interval. Non- linearity was addressed by applying a restricted cubic spline term, and transformation was applied when the normality assumption did not hold. Time by treatment and time by diet interactions were added to the model, and Wald statistics were used to assess the individual and joint significance of regression coefficients. Other data were analyzed by ordinary least squares and one-way ANOVA as indicated.