This study was carried out in accordance with the guidelines of the National Institutes of Health with a protocol approved by the Institutional Animal Care and Use Committee of University of California San Francisco (UCSF). Eighteen Yorkshire-Landrace pigs weighing 35–43 kg were obtained (Pork Power, Turlock, CA) for this study.

MI was induced by a 90 min balloon occlusion of the mid left anterior descending coronary artery (LAD) as previously published by our group [8, 15, 19]. Briefly, general anesthesia was induced by intramuscular injection of ketamine (20 mg/kg), xylazine (2 mg/kg) and atropine (0.04 mg/kg) then maintained with 2% isoflurane. The levels of anesthesia were kept the same at all study time points. Continuous blood pressure, oxygen saturation, and telemetry monitoring were performed during all procedures. A 6F sheath was placed in the femoral artery and after systemic heparinization (100–200 U/kg), the coronary artery was selectively engaged with a 6F HS0.75 guide catheter. A standard guide wire was placed in the LAD, and a 2.5 to 3.5 × 15 mm coronary angioplasty balloon delivered to the mid LAD just distal to the second diagonal branch. Balloon inflation (~4 atm) for 90 min was performed to induce MI. Complete occlusion with balloon inflation and LAD patency after balloon deflation was confirmed angiographically. Intravenous (IV) amiodarone (75 mg over 10 min) and lidocaine (1 mg/kg IV bolus, followed by 1 mg/min infusion) were started prior to balloon occlusion, with additional lidocaine (1–3 mg/kg IV bolus) given at the discretion of the operator for significant ventricular arrhythmias. Animals were medicated with atenolol (25 mg orally) daily starting 3 days post-MI. Prior to sacrifice, all animals underwent repeat coronary angiography.

Animals were assigned to one of two treatment groups using the same protocol previously published by our group testing EPO and GCSF monotherapy [8, 15]: 1) Long-acting EPO analog (Aranesp, Amgen, Thousand Oaks, CA) was given as IV bolus at the time of reperfusion (0.9 ug/Kg), then as weekly SC injections for 4 weeks (0.4 ug/Kg), and GCSF (Neupogen, Amgen, Thousand Oaks, CA) was given as IV bolus at time of reperfusion and then daily SC injections from day 5 to 9 post MI; and 2) control group (normal saline in equivalent volume given IV and SC to match the administration of the EPO therapy group). EPO dosages were selected following clinical data showing safety and feasibility with a single bolus of a fixed dose of darbepoetin [20], clinical studies addressing safety of chronic use of darbepoetin in patients with chronic renal failure [21], and pre-clinical data showing improvement in cardiac function and neovascularization using prolonged EPO therapy in a dose that did not increase the hematocrit [14]. GCSF dosages were selected following clinical data showing safety and feasibility of similar dose post MI [9].

Behavior, excreta, attitude (alertness, responsiveness, appetite), signs of respiratory distress (respiratory rate, respiratory pattern), motor weakness, and swelling extremities were monitored daily for the length of study. Arterial blood pressure was monitored at baseline, immediately post reperfusion, one, and 6 weeks post MI.

Whole blood samples were collected at baseline (prior to MI), at 2 h, on week one to four after MI induction, and then at sacrifice. Hemoglobin, hematocrit, white blood count, creatinine levels, creatine kinase MB fraction (CK-MB), and troponin I (TnI) were measured by the animal core laboratory (IDEEX, Sacramento, CA).

Transthoracic echocardiography (TTE) was performed at baseline, 1 and 6 weeks after MI using an Acuson 128XP machine with S3 (1–3 MHz) and S8 (3–8 MHz) probes (Siemens, Malvern, PA) as previously published by our group [8, 15, 19]. Long- and short-axis parasternal views and 4- and 2- chamber apical views were acquired. Left ventricular end-diastolic volume (LVEDV), end-systolic volume (LVESV), and ejection fraction (LVEF) were measured using the area/length method. The wall motion index (WMI) was calculated, using the method previously described by the American Society of Echocardiography [22], by grading the standard 17 myocardial segments (normal = 1, hypokinesis (reduced endocardial motion and wall thickening in systole) =2, akinesis (absence of inward endocardial motion or wall thickening in systole) =3, dyskinesis (outward motion or “bulging” of the segment in systole, usually associated with thin, scarred myocardium) =4, aneurysm = 5), and dividing the sum of the scores by the number of segments visualized. For all above parameters, at least three loops per scan were selected and the results presented as an average of the readings. Readings were made by blinded operators. The inter-observer variability (made from different readings of recorded loops) expressed using coefficient of variation in the measurement of LVEDV and LVESV was 4.3 +/- 5.7 mL and 1.4 +/- 2.9 mL, respectively, corresponding to variability in absolute LVEF of 2 +/- 3%. The intraobserver variability for WMSI (mean difference between measures 1 and 2) was 1.9%, while the interobserver variability (mean difference between observers 1 and 2) was 2.5%.

Left ventricular pressure-volume (PV) data were collected at baseline, 1 week, and 6 weeks after MI as previously published by our group [8, 15, 19]. Conductance and pressure signals were acquired using a dual field 5F 12-electrode pigtail PV catheter (Millar Instruments, Houston, TX) connected to a Leycom CFL-512 console (CD Leycom, Zoetermeer, Netherlands) via a TC-510 (Millar) pressure control unit and a patient module (CD Leycom) as previously described [19].

The ventricular end-systolic and end-diastolic pressure-volume relationships are considered as gold standards in the characterization of intrinsic ventricular pump properties [23‐25]. In this study, the PV catheter was inserted in the long axis of the left ventricle and oriented with segment one in the apex and segment seven in the aortic outflow as previously reported [8, 15, 19]. Inferior vena cava (IVC) occlusion was performed with a 7F, 34 mm Amplatzer sizing balloon (AGA Medical, Plymouth, MN) introduced via the femoral vein, inflated for 6–10 s to achieve ~50% drop in arterial blood pressure. Continuous data were acquired at a sampling frequency of 250 Hz during the steady state and IVC occlusion.

PV data were analyzed offline using the Conduct NT software (CD Leycom) by a blinded operator with a 10 Hz filter as previously described [19, 26]. Briefly, data were calibrated for parallel conductance (V_c) and alpha (α) based on volumes derived from transthoracic echocardiographic images collected at the beginning of each case. Using conductance data, and the α and V_c calculations, the Conduct NT software calculated ventricular volumes as previously described [27]. Steady state data included heart rate (HR), maximum rate of pressure change in systole (dP/dt_max), decline with relaxation (dP/dt_min), left ventricular end-diastolic pressure (LVEDP), end-systolic pressure (LVESP), LVESV, and LVEDV. Stroke volume (SV) was recorded as LVEDV—LVESV, cardiac output (CO) as SV × HR, LVEF as SV/LVEDV, and stroke work (SW) as the area enclosed by the PV loop.

Diastolic function was evaluated during steady state by the time constant of isovolumic relaxation, τ, and the dP/dt_min. τ was computed as described by Raff and Glantz using pressure recorded during the isovolumetric relaxation period which is the period from the time of dP/dtmin to the time when left ventricular pressure falls to 5 mmHg above the end-diastolic pressure of the following beat [28]. τ was calculated as the negative inverse of the linear slope of dp/dt vs. pressure during this period.

Data obtained during IVC occlusion were used to calculate the linear end-systolic pressure-volume relation (characterized by the slope; also called end-systolic elastance (E_es) and an intercept, V₀, and the preload recruitable stroke work (PRSW, or slope of SW versus LVEDV curve) [29].

All animals were sacrificed 6 weeks after MI and their hearts excised, weighed, and any gross surface abnormalities recorded. Histological and immunohistochemistry was performed in a blinded manner by CV Path Institute, Inc, MD. The ventricles were serially sliced at approximately 1 cm intervals parallel to the posterior atrioventricular sulcus from the apex to the base as previously describe by our group [8, 15, 19]. The thickness of each slice was measured and recorded. Digital images were taken for morphometric analysis of LV area, infarct size, and thickness, using IPLab software (Scanalytics, Rockville, MD). Infarct size was defined as a thinned and pale region of the anterior LV wall [30] and did not account for areas of viable tissue. Myocardial tissue for paraffin embedding was taken from the basal, mid, and apical-cavity levels. The three levels were then divided clockwise, beginning with the interventricular groove, into sixteen segments as described before [31]. Myocardial sections were transferred to 15% sucrose, dehydrated in a graded series of alcohols, and embedded in paraffin. Sections (4-5 μm) were mounted on charged slides and stained with hematoxylin and eosin, Masson’s Trichrome to evaluate for fibrosis using the IPLab software. The fibrotic areas from the 16 ventricle segments were summed and presented as a percent of the total LV area. The infarct zone (IZ) was defined as the arc of the left ventricle containing scar tissue.

For vessel density, myocardial tissue for paraffin embedding was taken from the basal, mid, and apical levels, and sectioned circumferentially into 4 adjacent transmural areas to include the central area of infarction, 2 adjacent border areas (border zone for capillary and arteriole measurements were defined at sites 1.2 mm and 2.0 mm outside the zone of infarction), and a control non-infarcted region (remote zone).

Immunohistochemical staining with a biotinylated lectin antibody (Dolichos biflorus, DBA, Sigma, St. Louis, MO) was used for the identification of capillaries. A monoclonal antibody against smooth muscle actin clone 1A4 (dilution 1:2000, Sigma) was used for the identification of vascular smooth muscle cells for identification of arteries and arterioles. Vascular density was measured at mid-ventricle region, in six to nine high power fields per section. Capillary (200X magnification), artery and arteriolar (100X) density were measured and expressed as the mean (± SEM) number of vessels per mm². The measured total tissue area was corrected for remaining interstitial space.

All results are expressed as means±standard error of the mean (SEM). A repeated measures ANOVA model (SigmaStat 3.5, Systat Software, San Jose, CA) was used to test the responses of examined parameters (measured at baseline, week1 and 6) in the experimental variants (Control, EPO+GCSF). The within-subject design included an overall F test of the main effects and then a post-hoc pairwise comparison of the values measured at 1 week and 6 weeks against the baseline, and 6 weeks against 1 week, using the Holm-Sidak method. Significance of differences between groups (Control, EPO+GCSF) was tested by an unpaired t test. A historical analysis using repeated measures ANOVA model (SigmaStat 3.5, Systat Software, San Jose, CA) was used to test the responses of examined parameters (measured at baseline, week1 and 6) in all experimental variants (EPO+GCSF). Correlations were performed using the Person method. A p < 0.05 was considered statistically significant in all the employed tests.