Up to eight trabeculae were dissected from one appendage. Experiments were performed in modified Tyrode’s solution containing (mM): NaCl 126.7, KCl 5.4, CaCl₂ 1.8, MgCl₂ 1.05, NaH₂PO₄ 0.42, NaHCO₃ 22, EDTA 0.04, ascorbic acid 0.2 and glucose 5.0. The solution was maintained at pH 7.4 by bubbling with a mixture of 5% CO₂ and 95% O₂. Atrial trabeculae were mounted in pairs, attached to SWEMA 4–45 strain gauge transducers in an apparatus containing above solution at 37°C and paced at 1 Hz. Trabeculae were pre-stretched to 50% of the length associated with maximum developed force. Sample sizes were chosen based on previous experience with experiments with positive inotropic substances and concomitant availability of myocardial and vascular preparations. The trabeculae were distributed randomly in four organ baths, usually two per bath. The assignment of experimental groups to the baths was randomized by drawing lots. In all experiments, unless otherwise indicated, we followed a protocol aimed to minimize effects of endogenous catecholamines. To this end, tissues were incubated with 6 μM phenoxybenzamine for 90 min. Phenoxybenzamine is an unselective α-adrenoceptor antagonist and increases release of noradrenaline (Enero et al., 1972). Trabeculae were washed from released catecholamines and let stabilized over additional 30 min. Force was recorded using Chart Pro for Windows version 5.51 analysis program (ADI Instruments, Castle Hill, NSW, Australia).

The adherent connective tissue was carefully dissected, and the artery was cut in up to eight rings of 3 mm width. The IMA segments were suspended on wire hooks in the organ bath described above (same Tyrode’s solution). Resting tension was increased stepwise from 2 up to 20 mN (four steps, every step lasts 7 min). KCl (100 mM) was applied and washed out six times to confirm proper function of the vessel rings. All experiments with IMA segments were performed in the absence of phenoxybenzamine.

Akrinor^TM is a mixture of cafedrine hydrochloride (200 mg) and theodrenaline hydrochloride (10 mg) in a 2 ml solution. Pharmacologically relevant concentrations were estimated at 42 mg/l (based on an injection of a single ampoule Akrinor^TM assuming 5 l blood volume). 42 mg/l Akrinor^TM contain 5.7 μM noradrenaline conjugated to theophylline, 101 μM norephedrine conjugated to theophylline, and 106.7 μM conjugated theophylline. Therefore, we performed experiments with theophylline at concentrations of 10, 100, and 1000 μM to compare effects of 4.2, 42, and 420 mg/l Akrinor^TM. Akrinor^TM, theophylline, norephedrine, and cafedrine were provided by TEVA ratiopharm (Ulm, Germany). Noradrenaline, phenoxybenzamine, forskolin, CGP 20712A (2-hydroxy-5-[2- [[2-hydroxy-3-[4-[1-methyl-4-(trifluorometyl)-1H-imidazol-2- yl]phenoxy]propyl]amino]ethoxy]-benzamide), ICI 118,551 (1-[2,3-dihydro-7-methyl-1H-inden-4-yl]oxy-3-[(1-methylethyl) amino]-2-butanol) and all other chemicals were obtained from Sigma-Aldrich (Darmstadt, Germany).

Data are expressed as mean ± SEM. When more than one tissue from a patient was available for one experimental group, mean values were calculated for individual patients. LogEC₅₀ were obtained by fitting sigmoidal concentration-response curve to data points from individual experiments. Paired t-test was used to compare logEC₅₀ values under PDE-inhibition to the respective controls obtained from the same patients. If more than two experimental groups were present, we compared maximum effects and logEC₅₀ values by one-way ANOVA followed by Bonferroni post hoc test. Curve fitting and all statistics were done by Prism GraphPad 5.0 (La Jolla, CA, United States).

In a first set of experiments, we investigated if Akrinor^TM evokes a positive inotropic effect in human atrial trabeculae via stimulation of β-adrenoceptors (AR). Experimental concentrations were varied from 4.2 up to 420 mg/l in a cumulative manner in order to construct concentration-response curves. We measured Akrinor^TM effects in the presence of β-AR subtype selective antagonists to elucidate the involvement of β-AR subtypes. CGP 201712A (300 nM) was used to block β₁-AR and ICI 118,551 (50 nM) to block β₂-AR. Proper inotropic reaction of the muscles was confirmed at the end of each experiment by increasing Ca²⁺ concentration from 1.8 to 8 mM to provoke maximal inotropic responses (Figures 1A,B).

Akrinor^TM already increased force of contraction at concentrations 10-fold lower than expected from intravenous injection of a single ampoule Akrinor^TM (4.2 mg/l). Maximum responses were reached at concentrations of 420 mg/l and were not smaller than effects of high calcium concentration indicating full agonist activity of Akrinor^TM. The presence of the β₁-AR antagonist CGP 201712A (300 nM) completely blunted Akrinor^TM inotropic effects, while responses to Ca²⁺ were preserved. The lack of any CGP 20712A-resistent positive inotropy indicates exclusive mediation of Akrinor^TM effects via β₁-AR. ICI 118,551 shifted the concentration-response curve from an EC₅₀ value of 40.8 ± 3.1 mg/l to 142 ± 4.4 mg/l (p < 0.01; n = 11/9 vs. 11/8, unpaired t-test). Effects of ICI118,551 further substantiate the latter interpretation. The small shift cannot be interpreted as a β₂-AR-contribution to Akrinor^TM positive inotropic effect. ICI 118,551 preferentially binds to β₂-AR and 50 nM ICI 118,551 should shift the curve for β₂-AR-mediated effects more than two log units. However, there is also some affinity of ICI 118,551 to β₁-AR. This small shift of the Akrinor^TM concentration-response curve by ICI 118,551 fits nicely to the known affinity data of ICI 118,551 to β₁-AR and is in line with recently measured dose shift in β₂-AR-KO mice (Pecha et al., 2015).

In a next set of experiments, we examined if Akrinor^TM exhibits PDE-inhibition in human atrial trabeculae. The classical approach to measure PDE-inhibition is to determine the leftward shift of the concentration-response curve for the inotropic effects of catecholamines (Rall and West, 1963). Since Akrinor^TM directly activates β₁-AR, we needed to modify the protocol and performed all the following experiments in the presence of the β₁-AR antagonist CGP 20712A (300 nM). To activate cAMP production we used the direct adenylyl cyclase activator FSK. FSK produced the expected positive inotropic effect in control experiments and in the presence of different concentrations of theophylline and Akrinor^TM (Figure 2A). To make changes in sensitivity more clear we normalized force responses to its individual maximum (Figure 2B). In order to minimize data scattering based on variability between patients we compared data obtained with theophylline or Akrinor^TM to untreated trabeculae from the same patient as controls. Therefore, a paired t-test was used to compare the effect of the intervention. First, we checked for a leftward shift of the concentration-response curve for the positive inotropic effect of forskolin by the prototypical PDE-inhibitor theophylline. 1 μM theophylline did not shift the concentration-response curve for subsequent exposure to forskolin. There was a trend with 10 μM theophylline. However, 100 μM theophylline were necessary to shift the FSK concentration-response curve by half a log unit from -logEC₅₀ of 5.52 ± 0.14 to 6.01 ± 0.2 M (p < 0.001, n = 11/6 each group, paired t-test). Next, we repeated experiments with Akrinor^TM. The sample size for experiments with AkrinorTM was adapted to experiments necessary to confirm theophylline-induced leftward-shift of concentration-response curve for forskolin. Only very high, clinically irrelevant concentrations of Akrinor^TM (420 mg/l) produced significant potentiation of FSK effects, from -logEC₅₀ value of 5.36 ± 0.31 to 5.7 ± 0.38 M (p < 0.05, n = 9/4 vs. 4/4, paired t-test), conceivable by PDE-inhibition. On the other hand, a matching concentration of theophylline (1 mM) alone generated a sustained increase in force, but FSK effects were blunted (Supplementary Figure S1 and Table S1). We would interpret this finding as an indication of theophylline toxicity.

Akrinor^TM contains theodrenaline and cafedrine. The latter drug is a conjugate of theophylline and norephedrine. Norephedrine is known to induce indirect sympathomimetic effects releasing noradrenaline from nerve endings (Liebmann, 1961; Trendelenburg et al., 1962). Therefore, we examined if an indirect sympathomimetic effect of cafedrine may contribute to the overall effects of Akrinor^TM. Classification of sympathomimetic effects as “indirect” critically depends on demonstration that catecholamine stores involved (Trendelenburg, 1969). For this purpose, we used phenoxybenzamine, a drug that non-selectively blocks α-AR (Langer, 1977). Micromolar concentrations of phenoxybenzamine inhibit neuronal uptake of noradrenaline by more than 50% (Enero et al., 1972). Therefore, we compared norephedrine effects in the absence and presence of phenoxybenzamine (Figures 3A,B). In the absence of phenoxybenzamine norephedrine increased force significantly from 2.9 ± 0.4 to 5.7 ± 1.1 mN (p < 0.05, n = 9/3, paired t-tests) with a calculated -logEC₅₀ of 6.1 M. This effect was completely abolished by phenoxybenzamine. It should be noted that under both conditions a clear negative inotropic effect occurred at concentrations >10 μM, indicating some toxic effects of norephedrine not related to its indirect sympathomimetic effects. In contrast to norephedrine, cafedrine did not show any positive effect (Figures 4A,B). Decline in force in cafedrine-treated trabeculae was not significantly different from TMC. Taken together our data suggest that the (indirect) sympathomimetic effect of norephedrine is lost when conjugated to theophylline.

Blood pressure is the product of cardiac output and vascular resistance. Any increase in blood pressure can result from one or more elements of positive inotropy, positive chronotropy, or vasoconstriction. Here, we found three pharmacodynamic pathways that could explain in vivo effects of constituents of Akrinor^TM and norephedrine, eventually metabolized in vivo: direct and indirect sympathomimetic actions and potentiation of cAMP-evoked inotropy by PDE-inhibition. All three mechanisms involve β-AR in the myocardium. In contrast, α-AR mediate vasoconstriction in response to sympathetic nervous stimulation and therefore vascular resistance, while β₂-AR-stimulation has opposing effects. In order to estimate whether arterial vasoconstriction contributes to Akrinor^TM’s blood pressure rising effects we measured the effects of Akrinor^TM and some of its constituents on tension of human arterial rings. In addition, we estimated the impact on vasoconstriction evoked by the natural agonist on α-AR, noradrenaline.

In human internal A. mammaria preparations, we observed a continuous decline in tension over time not only in TMC but also in preparations exposed to Akrinor^TM and theophylline. In contrast norephedrine-treated preparations showed stable tension (p < 0.05 vs. TMC, ANOVA, followed by Bonferroni). We would interpret this finding as some evidence for α-adrenoceptor-mediated vasoconstriction (Figure 5). However, the effect size of relevant concentrations (30 μM) was modest. Sensitivity to subsequent noradrenaline exposure was not affected by pretreatment with norephedrine. Theophylline (maximum concentration 32 μM) alone did not affect tension (Figure 5A), but blunted vasoconstriction to subsequent noradrenaline challenge (Figure 5B). It should be noted that Akrinor^TM did not evoke any vasoconstriction alone, but shifted the concentration-response for subsequent noradrenaline challenge to the right (from -logEC₅₀ 6.18 ± 0.08 to 5.23 ± 0.05 M, p < 0.05 vs. TMC, ANOVA, followed by Bonferroni).

From our in vitro findings, one could assume that cafedrine has no major effect at all. We measured cafedrine effects over a large concentration range. We extrapolated a peak concentration of 42 mg/l, resulting from an injection of 210 mg Akrinor^TM in 5 l blood, as no further data on Akrinor^TM distribution in plasma are available. This should result in cafedrine concentrations of about 100 μM. Cafedrine half-life is about 60 min (Koch and Wenzel, 2006). Exact metabolites are not known. Detailed knowledge of cafedrine pharmacokinetics is the basis of understanding of cafedrine in vivo pharmacodynamics. Interestingly and in contrast to catecholamines, maximum effects of cafedrine were observed with some delay but decrease rather slowly (Sternitzke et al., 1984). This finding could indicate that cafedrine, unable to evoke indirect sympathomimetic actions by itself, is metabolized to its active congener norephedrine or other active metabolites. In addition, it seems conceivable that theodrenaline could be also metabolized to noradrenaline. Interpretation about Akrinor^TM as a pro-drug are at present pure speculation. Further pharmacokinetic studies should help to clarify the contribution of cafedrine to the overall effect of Akrinor^TM and better understand the time course of effect of Akrinor^TM.

The direct and indirect effects of Akrinor^TM and some of its constituents described in this study do not allow direct interpretation of the effects of the drug in vivo. Any intervention that affects the uptake or storage of noradrenaline could increase noradrenaline concentration in the systemic circulation and evoke effects in organs with limited storage capacity for noradrenaline. For example, clinically relevant concentrations of aminophylline can increase plasma concentrations of epinephrine and to a lesser extent noradrenaline in humans (Vestal et al., 1983). While the potencies of noradrenaline to evoke inotropy and sensitivity to PDE-inhibition are very similar in human ventricular and atrial tissue (Christ et al., 2006; Molenaar et al., 2013), the effect size of indirect sympathomimetic agents like norephedrine and cafedrine may differ. Vascular resistance is regulated by arterioles and not by arteries investigated here. However, we cannot access to human arterioles. Effect size of indirect sympathomimetic activation demonstrated here may reflect noradrenaline stored within the tissue under investigation. Finally, while our data suggest reduced binding affinities of noradrenaline and norephedrine when conjugated to theophylline direct binding data are lacking. During the last decades investigation of the principles of indirect sympathomimetic drug action has lost some attraction. Therefore, we had to refer to older papers. Some of them may no longer represent state of the art. Nevertheless, the principles of indirect sympathomimetic actions still represent textbook knowledge (Westfall and Westfall, 2011). Probably there is a continuum of activity from predominantly direct-acting to predominantly indirect-acting drugs. Thus, this classification has to be interpreted as a relative rather than absolute one (Westfall and Westfall, 2011).