Wistar male rats, 16–24 weeks old, were used for this study. The rats were handed in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Pub. No. 83­23, Revised 1996). The protocol of the study was approved by the Animal Care and Use Committee of Department of Pharmacology, Medical School, National and Kapodistrian University of Athens (license 120946/18-04-2019, ΕL 25BIOexp 10).

A protocol of intermittent 5 cycles of 40 min low-flow perfusion followed by 20 min normal flow perfusion was applied as an intervention mimicking ischemic preconditioning. This intervention is designated as PC in the present study.

In order to investigate the potential effects of PC with and without the addition of T3 on ex vivo rat heart normothermic perfusion, the following experiments were performed a.

Hearts excised and subjected to an initial period of 30 min normothermic perfusion (stabilization period) in Langendorff apparatus with Krebs-Henseleit (KH) followed by 330 min of normal flow perfusion with KH buffer supplemented with vehicle (group NP, n = 9). Coronary flow was kept constant throughout the experiment.

Rats were anaesthetized with ketamine HCl and heparin 1,000 IU was given intravenously. The hearts were rapidly excised, placed in ice-cold Krebs­-Henseleit (KH) buffer and mounted on the aortic cannula in Langendorff perfusion system within 60 s. Retrograde perfusion is already used in the existing perfusion machines approved for human heart perfusion (2, 5). Perfusion in the isolated rat heart system was achieved with oxygenated (95% O₂/5% CO₂) KH buffer as previously described (21, 24). The composition of KH buffer (in mmol/L) was the following: sodium chloride 118, potassium chloride 4.7, potassium phosphate monobasic 1.2, magnesium sulphate 1.2, calcium chloride 1.4, sodium bicarbonate 25, and glucose 11.

The pressure signal of the left ventricle (LV) was monitored in real time via a water filled balloon which was inserted in the LV cavity via the left atrium and connected to a pressure transducer. Left ventricular balloon volume was adjusted to produce an average initial left ventricular end-diastolic pressure of 6–8 mmHg in all groups and was held constant thereafter throughout the experiment. Thus, measurement of LV pressure was performed under isovolumic conditions. Data analysis software (IOX, Emka Technologies) was used in order to analyze and record the LV pressure and the perfusion pressure signals (24).

The perfusion apparatus was heated to ensure a temperature of 37 ^oC throughout the course of the experiment. Hearts were paced at 320 beats/min with a Harvard pacemaker. Pacing was stopped during periods of low flow and re-established 5 min after coronary flow restoration.

At the end of perfusion period, the right ventricle was cut from the heart and the left ventricle (LV) was weighed. The ratio of LV weight to body weight was calculated as an index of tissue edema (26).

3,5,3'-triiodothyronine (T3) was purchased from Sigma Chemicals (St Louis MO, USA). T3 was dissolved in ethanol with the addition of NaOH which results in T3 sodium salt. Then T3 was diluted in 0.9% sodium chloride buffer to create a stock solution with a concentration of 1 mg/ml and was kept at −20 °C. Before each experiment 40μl of this stock solution was added per liter of KH buffer to reach a final concentrationof 60nM T3. This dose is several times higher compared to doses used in clinical practice to treat hypothyroidism and restore T3 levels in blood to 1–2nM. The dose of 60nM T3 has previously been shown to be cardioprotective against hypoxic injury in an isolated rat heart experimental model and during normothermic perfusion (21, 24). T3 administration or vehicle (T3 diluent) was added to the perfusate after the first 30 min of perfusion (stabilization period) in the isolated heart apparatus.

Left ventricular function was assessed by recording the left ventricular developed pressure (LVDP, mmHg) and the positive and negative first derivative of LVDP (+dp/dt and –dp/dt). Diastolic function was assessed by monitoring isovolumic left ventricular end-diastolic pressure (LVEDP) as a measure of diastolic chamber distensibility. Perfusion pressure under constant flow conditions was used to assess coronary vessel resistance (PP, mmHg). Parameters were also expressed as percentage of change from baseline values.

Values are presented as mean (standard deviation). Normal distribution of variables was estimated with Shapiro–Wilk test of normality. Normally distributed data between groups were compared using one-way ANOVA. Paired samples t-test was used to compare measurements between stabilization and end of perfusion within the same group. Serial measurements of LVEDP and PP were also compared by mixed, repeated measures analysis of variance (mixed ANOVA) to test for the effect of treatment, time and the interaction (tests for “within-subjects” factor and “between-subjects” factor); the respective non-linear fit-curves were produced with non-linear fit analysis. When significant, differences within and between each group were tested by a post hoc analysis using Bonferroni correction for multiple comparisons. A two-tailed test with a p value less than 0.05 was considered significant. Analysis was performed using SPSS 23.0 and GraphPad 8 software.

In this study, rat hearts were perfused for 6 h with constant flow, in a Langendorff mode and Krebs-Henseleit as perfusate with glucose being the only metabolic substrate (NP), while PC treated hearts were subjected to 5 cycles of 40 min low flow perfusion followed by 20 min normal flow perfusion with either vehicle (PC) or 60nM T3 (PC + T3). Figure 1. The selection of the PC protocol and the perfusion duration was based on pilot studies (data not shown). The rationale was to establish a long duration protocol of perfusion with periods of low flow ischemia and reperfusion which is protective and does not result in irreversible damage. Flow was adjusted during the first 30 min of stabilization in order to achieve a mean perfusion pressure of 65–70 mmHg (normal flow) and was kept constant thereafter in group NP. In groups PC and PC + T3, flow was kept constant during the normal flow periods after the initial adjustment period. During periods of low flow perfusion, the rate of the peristaltic pump was decreased to 20% of normal flow and perfusion pressure was reduced to 20–24 mmHg. In accordance with the above, mean normal flow (L/min/g body weight) was 33 (9.8) for NP, 32 (4.5) for PC and 30 (3) for PC + T3 groups. During low flow periods, mean flow was 6.4 (0.9) for PC and 6.0 (0.6) for PC + T3 groups.

Left ventricular developed pressure (LVDP) and the rate of increase and decrease of LVDP (+dp/dt and -dp/dt) at the end of stabilization (baseline parameters) or the end of normothermic perfusion were not statistically different between groups Table 1. In NP group, LVDP decreased by 32.3% (11) % from baseline vs. 31.2% (8)% in PC group, p = 0.95. In PC + T3 group, LVDP decreased by 37.7% (11)% from baseline, p = 0.74 vs. NP.

LVEDP baseline parameters were similar in all groups Table 2. At the end of perfusion, LVEDP was significantly increased in both in NP and PC group as compared to baseline, p < 0.05. Interestingly, in PC + T3 group, LVEDP at the end of perfusion was similar to baseline. The magnitude of LVEDP change from baseline (LVEDP %) was significantly less in PC and PC + T3 compared to NP group. In NP group, LVEDP increased 188 (94) % from baseline vs. 73 (98)% in PC group, p = 0.021 Table 2. In PC + T3 group, LVEDP increased only 8 (58)% from baseline, p = 0.0003 vs. NP Table 2.

Mixed Repeated Measures ANOVA analysis for LVEDP (dependent variable) at different time points between the groups showed that the main effect of interventions (PC and PC + T3, “between-subjects” factor) on LVEDP was statistically significant (F = 9.2, p = 0.001). Using Tukey post hoc test, LVEDP was found to be significantly lower for PC group (p = 0.048) and for PC + T3 (p = 0.0001) compared to NP. Thus, both PC and PC + T3 treatment resulted in a lower LVEDP over time. Furthermore, there was statistical significance (F = 32.5, p = 10⁻¹³) regarding the main effect of time (“within-subjects” factor) on LVEDP. There was also a statistically significant interaction effect between time and group (F = 8.9, p = 10⁻⁶). Figure 2A.

Mean flow rate at the end of stabilization period was similar in all groups. Baseline coronary perfusion pressure was similar in all groups. At the end of perfusion, PP was significantly increased in all groups as compared to baseline (p < 0.05, Table 2). In NP group, PP increased 120 (63)% from baseline vs. 58 (35)% and 39 (27)% in PC and PC + T3 group, p = 0.02 and p = 0.002, respectively Table 2.

Mixed Repeated Measures ANOVA analysis for PP (dependent variable) at different time points between the groups showed that the main effect of interventions (PC and PC + T3, “between-subjects” factor) on PP was statistically significant (F = 6.6, p = 0.005). Using Tukey post hoc test, PP was found to be significantly improved for PC group (p = 0.035) and for PC + T3 (p = 0.005) compared to NP). Thus, both PC and PC + T3 treatment induced a sustained improvement of PP over time. Furthermore, there was a statistically significant effect (F = 46.9, p = 10⁻¹⁵) regarding the main effect of time (“within-subjects” factor) on deterioration of PP. Α statistically significant interaction effect between time and group-was found (F = 5.6, p = 0.00009). Figure 2B.

The ratio of LV weight to body weight was found to be 2.3 (0.18) for NP group vs. 2.05 (0.15) for PC group (p = 0.006) and 2.1 (0.18) for PC + T3 group (p = 0.035 vs. NP).

Some limitations in the present study should be taken into account. In this study, an acellular crystalloid based perfusate was used rather than a blood-based perfusate currently used in clinical practice. Crystalloid buffers have traditionally been used in Langendorff perfusion to study rodent heart physiology and pharmacology, but their use is limited in human heart perfusion due to low oxygen solubility particularly with increasing temperature in the perfusate (11). Although perfusion with blood-based solutions is considered superior to crystalloid based perfusion, an increased risk of microvascular obstruction may occur in the setting of ischemia -reperfusion due to erythrocyte aggregation (11, 39, 40). In accordance with the above, in a Langendorff rat heart normothermic perfusion for 4 h, the use of packed red blood cells induced a pro-inflammatory response associated with a greater cardiac damage compared to acellular perfusion (26). Blood perfusion imposes several technical challenges and make the perfusion system more complex. Thus, the interest for crystalloid based normothermic perfusion modalities has recently been revived highlighting the importance of the present findings. It should be noted that from a translational perspective, a normothermic acellular perfusate-based ex vivo system does not yet exist for heart transplantation. The present study has been designed to provide evidence for translational studies in large animals and humans. In this context, molecular mechanisms have not been investigated. However, the mechanisms of actions of ischemic preconditioning and T3 on reperfused myocardium have extensively been studied in previous investigations (22, 41). Finally, our experimental model used hearts from animals after anaesthesia and not after brain death. Ischemic preconditioning effect may be limited in hearts harvested from brain death. However, previous investigations have shown that the ischemic preconditioning response was preserved in hearts obtained after brain death and transplanted to the recipient animals (42).

In conclusion, intermittent low flow mimicking ischemic preconditioning appears to optimize crystalloid normothermic rat heart perfusion by limiting tissue edema and diastolic and vascular dysfunction. This study may be a paradigm for further investigations to explore the potential of simple crystalloid solutions in combination with means of cardioprotection to optimize normothermic perfusion.