Twenty-three adult male Wistar rats (282- to 335-g body weight at the beginning of the study; Charles Rivers Laboratories, Wilmington, MA, USA) were enrolled. All animals were housed on a 12-h light-dark cycle in a room with temperature and humidity control, and with ad libitum access to tap water and standard rodent food. Permanent occlusion of the left anterior descending (LAD) coronary artery was performed in 23 rats as previously described 10 12 . The choice of this model rather than a model of ischemia/reperfusion was based on preliminary experiments suggesting that adenosine may be more protective in large infarcts than small infarcts (not shown). Rats were anaesthetized by inhalation of an isoflurane/oxygen mixture (2% to 3%:1.5% v/v) and were intubated and ventilated with a rodent ventilator (Harvard Apparatus, Holliston, MA, USA). The heart was exposed through a left lateral thoracotomy of the fifth intercostal space. After pericardial incision, the proximal part of the LAD (2 to 3 mm from the top of the left atrium) was ligated with a 7-0 Prolene suture (Ethicon, Somerville, NJ, USA). Perioperative lidocaine (10 mg/kg, Aguettant, Lyon, France) was used to provide a local anaesthesia and to reduce the incidence of ventricular tachycardia and fibrillation. Finally, the chest was closed in layers with a 2-0 Vicryl suture (Ethicon, Somerville, NJ, USA), the lungs were reinflated using positive-end expiratory pressure, and the endotracheal tube was removed once spontaneous breathing had resumed. Amoxicillin was injected intramuscularly (60 mg/kg, GlaxoSmithKline, Marly-le-Roi, France) to avoid bacterial infections. The mortality of the procedure was <20%.

Two days after surgery, rats were assessed by ECG-triggered ¹⁸ F-FDG-PET. This first PET exam allowed for randomizing the animals into three treatment groups according to infarct size. This randomization was crucial to avoid any bias due to different infarct sizes in experimental groups. Infarct size is indeed a major determinant of LV remodelling. A three-point classification was used to determine MI size, as previously described by our team 10 : small MI (no more than 3 among the 17 LV segments), moderate MI (4 to 5 segments) and large MI (at least 6 MI segments). Groups were constituted in order to exhibit equivalent rates of rats with small, moderate and large MI. In each of the three groups, rats were assigned to treatment with NaCl (control group; n = 7) or 2-chloroadenosine (2 mg/kg/day), a stable analogue of adenosine, which was given alone (CADO group; n = 8) or with 8-sulfophenyltheophilline (10 mg/kg/day), an antagonist of adenosine receptors (8-SPT group; n = 8). NaCl, CADO and 8-SPT (Sigma-Aldrich, Bornem, Belgium) were administered intraperitoneally twice daily for 2 months, starting 7 days after the date of LAD occlusion.

Two days, 1 month and 2 months after surgery, LV function and infarct size were assessed in vivo by FDG-PET. As already described in detail elsewhere 10 13 , rats received an oral pre-medication of 50 mg/kg of acipimox, a potent nicotinic acid derivative yielding high-quality cardiac FDG-PET images. Around 74 MBq of ¹⁸ F-FDG (IBA, Nancy, France) were injected intravenously under a transient anaesthesia with isoflurane. A 16-min recording was started 1 h later on a high-resolution dedicated small animal PET system (Inveon, Siemens, Knoxville, TN, USA) and under continuous anaesthesia by isoflurane. Images were reconstructed in 16 cardiac intervals with a 3-D ordered subset expectation maximization algorithm leading to a voxel size of 0.8 × 0.4 × 0.4 mm.

FDG uptake was determined on the set of collapsed short-axis slices in each segment from the 17-segment LV division from the American Heart Association 14 with the QGS software 15 . LV end-diastolic volume (EDV), LV end-systolic volume (ESV) and ejection fraction (EF) were obtained from the set of contiguous ECG-triggered short-axis slices with the QGS software 15 16 . The accuracy of these volume measurements was previously demonstrated on phantoms from LV rats above the level of 100 μL, corresponding to the lower limit for the LV end-systolic volume in adult rats 13 . The QGS software was also used for assessing the contractility of each of the 17 LV segments according to the percentage of systolic increase in myocardial counts 15 . This parameter is strongly linked to the percentage of myocardial thickening with both myocardial SPECT and TEP imaging techniques 17 18 .

As previously described and validated 10 , all segments for which the average FDG uptake was very low (<50% of the maximal voxel value) could be considered as predominantly necrotic, and the percentage of such segments was used to assess the extent of MI areas. For the segmental analysis of LV contractility, MI segments were defined as those showing a very low FDG uptake (<50% of the maximal voxel value) on the entire segment area, remote segments were defined as those which were not adjacent to any segment showing such a low uptake, and the remaining segments were considered to be within the border zone.

The average heart rate value during PET acquisition was extracted from the list mode recording data. Systolic blood pressure was calculated as the mean value of four recordings by the tail-cuff method during the PET acquisition (AD Instruments, Powerlab, Paris, France).

Rats were sacrificed 1 to 2 days after the last 2 months of PET acquisition with an overdose of sodium pentobarbital. The hearts were excised, snap frozen in liquid nitrogen, fixed and embedded in optimal cutting media (VWR, Fontenay-sous-Bois, France). Contiguous sections (8 μm), orientated along the vertical or horizontal long axis of the LV (depending on infarct location), were obtained in a cryostat at −22°C and stored at −80°C until analysis.

The degree of fibrosis in heart sections was assessed by Masson's trichrome staining. The collagen volume fraction was measured while omitting fibrosis of the perivascular, epicardial and endocardial areas 19 . The fibrosis area in the border zone was measured in three random fields per heart section (×400 magnification) by dividing the area of collagen to the total area and using ImageJ version 1.42 (http://rsbweb.nih.gov/ij/index.html).

Haematoxylin and eosin staining was performed to assess the cross-sectional area of cardiomyocytes in heart sections. For each section, at least five cardiomyocytes sectioned transversely were randomly chosen, and their area was quantified using ImageJ software. The average area was calculated for each experimental group.

Immunohistochemical staining was performed for annexin-5 apoptosis marker using a rabbit polyclonal antibody (Abcam, Cambridge, UK). Alexa Fluor® 635-coupled goat anti-rabbit antibody was used as secondary antibody (Invitrogen, Merelbeke, Belgium). Technical controls without primary antibody were performed for each marker to ensure staining specificity. DAPI was used to reveal nuclei. Images were recorded on a confocal microscope (Zeiss Laser Scanning Microscope LSM 510, Carl Zeiss Microscopy, Oberkochen, Germany) with a × 400 magnification using the LSM 510 META software (Carl Zeiss Microscopy, Oberkochen, Germany).

Analyses were performed using the SAS R9.3 software (SAS Institute, Cary, NC, USA), and the two-tailed significance level was set to p < 0.05.

Pairwise comparisons of quantitative variables between the CADO or 8-SPT group and the control group were carried out using the Mann-Whitney test, and corresponding variables were expressed with mean ± standard error of the mean (SEM).

Segmental contractility was analysed over time using repeated measures mixed ANOVA: (1) with treatment group and segment location as fixed factors and (2) with rat as a random factor nested within the ‘treatment’ variable (all segments from a single rat received the same treatment). The results of these analyses were expressed as adjusted means ± SEM. Validity conditions of the models were thoroughly checked (normality of residuals, homoscedasticity and absence of collinearity or interaction). These global analyses were conducted for assessing, first, the effect of CADO alone (CADO vs. control) and, thereafter, the effect of CADO associated with 8-SPT (8-SPT vs. control). In order to preserve the overall 5% alpha error rate, the comparisons of CADO or 8-SPT vs. control were carried out at the 2.5% level.

Representative FDG-PET images obtained at 48 h, 1 month and 2 months are shown in Figure 1. On the baseline FDG-PET scans recorded 2 days after coronary occlusion, necrotic segments were observed in all 23 animals. According to the three-point classification of MI size (small, moderate and large MI) 10 , rats were subsequently randomized into three treatment groups: control (n = 7), CADO (n = 8) and 8-SPT (n = 8). Thus, the mean infarct size was equivalent between the three groups (Table 1). The mean baseline values of LV volumes, LV ejection fraction, heart rate, blood pressure and body weight were also similar between the three groups (Tables 1 and 2).

*p < 0.05 for unpaired comparisons vs. control group.

During the 2-month follow-up, all rats showed an increase in body weight, a decrease in heart rate and a stable level of systolic blood pressure. At 2 months, however, body weight and heart rate were slightly but significantly lower in CADO-treated rats than in control rats (Table 2).

As detailed in Table 1, control rats exhibited marked LV remodelling during the 2-month follow-up, as attested by increased EDV and ESV, and decreased EF. These changes were not significantly different in the CADO group or in the 8-SPT group, even though there were trends toward a beneficial effect on LV volumes and EF in the CADO group and a detrimental effect on LV volumes and EF in the 8-SPT group (Table 1).

Seventeen segments were analysed in each of the 23 animals, leading to a total of 391 analysed segments. On the 2-day FDG-PET exams, 43 segments were totally necrotic (MI segments), 109 were considered as remote segments and the 239 remaining segments were considered to be within the border zone. Myocardial thickening, assessed through the percentage of systolic increase in myocardial counts, exhibited a continuous decline between remote (40% ± 9%), border zone (23% ± 13%) and MI (12% ± 4%) segments (p < 0.001).

Analyses of variance revealed that the evolution of myocardial thickening over 2 months was enhanced in CADO-treated rats, compared to control rats, with a significant interaction by segment location (remote, border zone or MI area) (p = 0.03). In contrast, no difference was observed when 8-SPT rats were compared to control rats using the same model.

Changes in segmental contractility from baseline were also compared between the three groups at the specific time-points of 1 and 2 months, as detailed in Figure 2. The main observation was that, on average, the contractile function of the overall segments from CADO rats was enhanced when compared to controls at 1 and 2 months (Figure 2A). After 1 month, the increase of myocardial thickening was +3.8% ± 0.7% in the CADO group compared to +0.9% ± 0.7% in the control group (p < 0.01). After 2 months, contractile function was impaired in control rats, as attested by a decrease of myocardial thickening of −2.3% ± 0.8%, and preserved in CADO-treated rats (+1.6% ± 0.8%, p < 0.001). This effect of CADO was blunted by 8-SPT (Figure 2A).

Interestingly, the beneficial effect of CADO on contractile function varied according to segment location, both after 1 month (Figure 2B) and after 2 months (Figure 2C). In particular, at 2 months, this effect was obvious (1) in the border zone where CADO rats exhibited a significant increase in contractile function (+5.6% ± 0.8%), whereas control rats did not (+1.5% ± 0.8%, p < 0.001 vs. CADO), and (2) in remote segments where a decline in contractile function was documented in all groups but to a lower extent in the CADO group (−4.0% ± 1.0%) than in the control group (−10.4% ± 1.3%, p < 0.001 vs. CADO). By contrast, the contractile function of MI segments was stable and comparable in the two groups at 2 months (CADO +3.4% ± 2.0% vs. control +2.0% ± 1.8%, p = NS).

The beneficial effect of CADO on the contractile function of the border and remote segments was lost for rats additionally treated with 8-SPT (Figure 2A,B).

We examined the extent of fibrosis outside the MI areas (Figure 3). CADO appeared to reduce fibrosis in remote areas, when compared to controls, although this difference did not reach the level of statistical significance. However, this fibrosis was markedly increased by the administration of 8-SPT (eightfold increase compared to the CADO group, p < 0.001).

The cardiomyocyte cross-sectional area in healthy parts of the heart was lower in the CADO than in the control group, and this effect was blunted by 8-SPT (Figure 4).

Annexin-5 staining was reduced in CADO rats compared to control rats and was robustly increased by 8-SPT (Figure 5). These differences were evident in the remote and border zones.

Therefore, activation of adenosine receptors reduces MI-induced fibrosis, cardiomyocyte hypertrophy and apoptosis.