The study was performed in accordance to the German laws for animal health and protection declaration and was approved by the local authorities, Government of Leipzig (Reg.-Nr.: 24-9168.11-11/03). We used male White New Zealand rabbits of 1500–2000 g body weight (at beginning of the study) (conventional, normally fed ad libitum) (Charles River, Kisslegg, Germany).
Animal model and investigation of contractility
Heart failure was induced by repeated i.v. injection of doxorubicine 3 mg/kg/week over 6 weeks, followed by a 14 days doxorubicine-free interval. Animals that survived 2 weeks after heart failure induction were randomized to either no treatment, sham operation or BMSC transplantation. After intramuscular anaesthesia with ketamin (50 mg/kg) and xylazine (5 mg/kg), we obtained autologous mesenchymal bone marrow stem cells from a femur punction, isolated the mononuclear stem cell fraction, and after subcultivation for 4 days (see below), the stem cells were injected directly into the wall of the left ventricle. For transplantation rabbits were anaestetized (0.8 vol% isofluran, protocol see . Left thoracotomy through the intercostal space was performed and the left ventricle was exposed. Either BMSC (1.5 – 2.0 Mio cells in 1 ml) or medium (1 ml) was injected at 4 positions into the tissue of the free lateral wall of the left ventricle in a circular manner with a 1 ml syringe within 2 minutes.
28 days after the operation, animals were subjected to the final experiment. One hour prior to surgery (anaesthesia as described above), we collected 2 ml venous blood in potassium EDTA monovettes (with addition of 1 μM glutathione/ml plasma) from the rabbits after at least 30 min at rest for determination of plasma catecholamines. After centrifugation (600 g/10 min/4°C), plasma was removed, quickly frozen in liquid nitrogen, and stored at -80°C until further use. We introduced a Millar catheter designed for rabbits (FMI, Germany) via the femoral artery and registered pressure-volume loops under the influence of 1.5 μg/kg dopamine, in order to assess the contractile function. dP/dtmax and dP/dtmin were determined from these measurements. Subsequently, we excised the heart for mapping experiments and radioligand binding assay. After the experiment, tissue was immediately placed in cardioplegic solution. Left ventricle, septum and right ventricle were excised and subsequently shock-frozen.
We investigated 4 groups of rabbits: healthy control animals (n = 10), doxorubicine-induced heart failure (n = 6), doxorubicine-induced heart failure with sham operation (n = 8), and doxorubicine-induced heart failure with stem cell treatment (n = 14). During the investigations and subsequent evaluations the experimentators were blinded.
Stem cell purification and subcultivation
Under sterile conditions we obtained 1.5 – 2 ml bone marrow in 2 mM EDTA containing phosphate buffered saline (PBS) from a femur punction. The mixture was initially centrifuged at 300 g, at room temperature (RT), for 5 min, and the resulting pellet was resuspended in PBS containing 2 mM EDTA, and separated in a Ficoll (1.073 g/ml) density gradient centrifugation (cell suspension : ficoll: 1.5: 1.0; 30 min centrifugation at 500 g, RT). The mononuclear fraction interphase was collected, washed twice in PBS/EDTA. The final pellet was resuspended in 12 ml Dulbeccos Modefied Eagle Medium (DMEM) cell culture medium (+ 10% fetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycine; 37°C, 5% CO2) and seeded on gelatine-coated petridishes. For the first 24 hours cells were incubated with additional 10 μM 5-azacytidine. Cells were cultured for 4 days. Cell culture medium was changed every 48 h, and non-adherent hematopoetic cells were discarded. Prior to transplantation cells were detached with trypsin/EDTA, and incubated with 2.5 μM Vybrand DiI cell labelling solution (Molecular Probes) for 0.5 h protected from light in the incubator at 37°C followed by 3 washing steps in PBS. Finally, the resulting mesenchymal mononuclear c-kit positive bone marrow stem cells (tested using a commercial anti-c-kit antibody) were suspended in DMEM for injection (1.5–2.0 Mio. cells in 1 ml).
Radioligand binding study
β-adrenoceptors were assessed by (-) [125I]-iodocyanopindolol (ICYP)-binding assay described elsewhere . Briefly, tissue samples from right ventricle, left ventricle and septum were homogenised in 10 volumes of ice-cold 1 mmol/l KHCO3 with an Ultra Turrax (Janke and Kunkel, Staufen, Germany), diluted to 20 ml with 1 mmol/l KHCO3, centrifuged at 500 g for 10 min, passed through 4 layers of cheesecloth, and centrifuged again at 50,000 g for 20 min. Pellets were washed once by resuspension and recentrifugation and finally resuspended in incubation buffer (Tris-HCl 10, NaCl 154, ascorbic acid 0.55 mmol/l, pH 7.4, 25°C) at a protein concentration of 0.1–0.2 mg/ml. Protein content was determined by the method of Lowry using bovine serum albumine as a standard.
The density of β-adrenoceptors in cardiac membranes was determined by (-) [125I]ICYP binding at six concentrations ranging from 5 to 200 pmol/l as detailed elsewhere . Non-specific binding of ICYP was defined as binding to membranes which could not be displaced by a high concentration of the non-selective β-adrenoceptor antagonist (±) CGP 12177 (1 μmol/l). Specific binding was defined as total binding minus non-specific binding and usually was about 70–80% at 50 pmol/l ICYP.
To determine the relative amounts of β1- and β2-adrenoceptors, membranes were incubated with increasing concentrations of the β2-adrenoceptor antagonist ICI 118,551 (10-10 – 10-4 M) and a constant concentration of ICYP (100 pmol/l), and the specific binding assessed as described above (using CGP12177). Binding curves were analysed using the iterative curve fitting program GraphPadPrism (GraphPad Software, San Diego, CA, USA).
High Pressure Lquid Chromatography (HPLC)
Plasma norepinephrine content was assessed by high-pressure liquid chromatography (HPLC) and electrochemical detection  using a commercial HPLC assay (Chromsystems, Martinsried, Germany) based on the protocol by Goldstein  and by Hjemdahl et al. . Briefly, catecholamines were adsorbed to aluminiumoxide (in prepacked columns; Chromsystems, Martinsried, Germany) by shaking 1 ml plasma with 0.5 ml extraction buffer (Tris-buffer) and aluminiumoxide for 10 min. As internal standard 600 pg dihydroxybenzylamine were added to the plasma sample. After 10 min, the probe was washed twice and washing buffer was removed. Thereafter, the bound catecholamines and internal standard were eluted using 120 μl elution buffer (Chromsystems, Martinsried, Germany) and filtration (at 700 g). 40 μl probe volume were injected at a flow of 1 ml/min (HPLC autosampler (GINA50) and pump (P580): Gynkotec, Germering, Germany; mobile phase 5001; Chromsystems, Martinsried, Germany) on a pre-equilibrated RP18 column (Chromsystems, Martinsried, Germany). The HPLC system was controlled by the Chromeleon software 4.10 (Gynkotec, Germering, Germany). Catecholamines were detected using an electrochemical detector (DECADE; flow cell: VT-03; Fa. ANTEC Comp., Leyden, Netherlands) (a working potential of 0.55 V yielded maximum signals with lowest noise). An external calibration standard containing 5 ng/ml norepinephrine, 5 ng/ml dihydroxybenzylamine, 2.5 ng/ml epinephrine and 2.5 ng/ml dopamine was also used for each experiment. Each sample was injected three times and the concentration was determined as the mean of these three detections.
Tissue samples of the hearts were embedded in paraffin and 5 μm sections of the hearts were stained with picrosirius red according to standard protocols. Using fluorescence microscopy we identified the BMSC or their remnants by Vybrand-DiI-red fluorescence and counted the number of these BMSC (or remains) per visual field at 200 x magnification. Moreover, we looked for signs of inflammation, i.e. leukocyte/lymphocyte infiltration.
In order to investigate, whether BMSC injection might produce inhomogeneities in the cardiac electrical activation pattern or autonomic areas, we submitted 4 hearts of each group to epicardial mapping as previously described . Briefly, rabbits were anaesthetized by isoflurane, the heart was excised and prepared according to the Langendorff technique (constant pressure: 70 cm H2O, perfusion with Tyrode solution (Na+ 161.02, K+ 5.36, Ca++ 1.8, Mg++ 1.05, Cl- 147.86, HCO3- 23.8, PO42- 0.42 and glucose 11.1 mM, equilibrated with 95% O2 and 5% CO2 (pH = 7.4); the surface temperature of the heart was 37°C). The hearts, which were beating at their spontaneous rate, were connected to a 256 channel mapping system HAL4 (Ing. Buero Peter Rutten, Hamburg, Germany, temporal resolution: 20 kHz per channel; amplitude resolution: 0.04 mV, interchannel coupling <-60 decibel; bandwidth of the system: 0.5 Hz – 100 kHz, data were not filtered) as described previously . 256 AgCl electrodes (1 mm interelectrodes distance), which were attached to the heart surface in an elastic manner, so that they could follow the heart movements easily without dislocation. The four plates were located at (a) the right wall (64 channels) (b) the front wall (64 channels) (c) the left wall (64 channels) and (d) the back wall (64 channels), so that both ventricles were mapped with a total of 256 electrodes. Activation and repolarization time points at each electrode were determined as t(dU/dtmin) or t(dU/dtmax), respectively [33–35]. After automatic determination activation and repolarization timepoints were verified (or corrected if necessary) manually by the experimentator. Total activation time (TAT, [ms]) was calculated as the delay between activation of the first and activation of the last electrode. Standard deviation of activation times (SD(ACT)) was used as a measure for local inhomogeneity. For each electrode an activation-recovery-interval (ARI, reflecting epicardial potential duration) was calculated. Inhomogeneity of ARI was analyzed calculating standard deviation of ARI at 256 electrodes (=ARI-dispersion).
All continuous variables are presented as means ± S.E.M. of n experiments. Experimental data to β-adrenoceptors were fitted and analysed by computer-supported iterative non-linear regression analysis using the GraphPadPrism program (GraphPAD Software; San Diego, CA, USA). Statistical significance of differences was analysed by unpaired two-tailed Student's t-test or, if appropriate, by repeated measures ANOVA followed by t-test using Bonferroni corrections for multiple comparisons. P < 0.05 was considered to indicate a significant difference. The statistical evaluation was performed using SYSTAT program (Systat Inc., Evanston, IL, USA).
CGP12177, ICI118,551 were from Sigma (Taufkirchen, Germany), (-) [125I]iodocyanopindolol (ICYP, specific activity: 2200 Ci/mmol) (Perkin Elmer, Boston, MA, USA), all other chemicals were of the purest commercially available grade and obtained from Sigma.