Adult (3–5 year old) female Romney sheep were housed in individual crates, and acclimatized to laboratory conditions (18 °C, 50% relative humidity, 12 h light–dark cycle) and human contact for 1 week before any experiments. Sheep were fed 2–2.5 kg/day (Country harvest pellets) and had access to water ad libitum. All animal experiments and surgical procedures followed relevant guidelines and were approved by the Animal Ethics Committee of the University of Auckland (#2082). Experiments were carried out over a period of 6 months in each animal; a summary is schematized in Fig. 1B.
Induction of heart failure—Embolization Surgical Procedure
Three groups of heart failure sheep were studied: RSA, monotonically paced, and a time control. In the conscious state and standing, echocardiography was performed (Phillips HD-11 Ultrasound) to measure ejection fraction in the healthy animals (typically 70–80%). Subsequently, sheep were anaesthetized which was induced with 2% Diprivan (Propofol) (5 mg/kg i.v. AstraZeneca, AUS), maintained with a 2% isoflurane-air-O2 mixture and were intubated for mechanical ventilation. Anaesthesia depth was monitored throughout the surgery by an absence of the corneal reflex and an absence of a withdrawal response to a noxious pinch. The left or right femoral artery was accessed percutaneously using an 8F (CORDIS®, USA) sheath and under fluoroscopic guidance, the left main coronary artery was then cannulated, and the catheter advanced into either the proximal left main coronary artery or left descending coronary arteries as described before . To induce heart failure with reduced ejection fraction sheep underwent sequential weekly (1–3 weeks) embolizations using polystyrene latex microspheres (45 μm; 1.2 mL, approximately 650 microspheres, Polysciences, Warrington, PA, USA). Prior to the injection of microspheres, β-blocker (metoprolol up to 20 mg/kg, IV) and lignocaine (2 mg/kg, IV) were injected intravenously to prevent ventricular arrhythmias. Electrocardiogram (ECG) was recorded from lead II prior to the infusion of the microspheres and for a further 5 min after infusion. The recordings were made on a dual bio amp electrocardiograph switch box with a power lab and LabChart (AD Instruments, NZ). A change in the ST segment (elevation or depression) and T wave (inversion) was taken as an indication of a successful embolization.
Three days post embolization, conscious sheep underwent echocardiography; once ejection fraction dropped from 70 to 80% to ~ 45%, no more embolizations were performed. A drop in ejection fraction to ~ 45% was selected as the criteria for successful induction of heart failure. Giving all sheep the maximum three embolizations without taking into account ejection fraction resulted in unnecessary animal loss in previous studies. There was a loss of three animals during or within 48 h post the embolizations in this study. Echocardiograms were repeated 1 week, and 3 months post embolization, then once weekly during pacing and the post-pace period. In the long-axis M-mode, diastole, systole, fractional shortening, and ejection fraction were obtained and calculated for the left ventricle. There were no differences in body weight or baseline ejection fraction between groups at the start of the experiment.
Three months after heart failure was induced by microembolization, left ventricle ejection fraction was again checked using echocardiography. No sheep was found to significantly increase in ejection fraction over the 3 months and there was a loss of one animal during this period. Once confirmed to be ~ 45%, sheep underwent instrumentation under general anaesthesia (induced with 2% Diprivan (Propofol) 5 mg/kg i.v., AstraZeneca, AUS) and maintained with a 2% isoflurane-air-O2 mixture. An intercostal nerve block was performed on ribs 2–6, using a 22G needle 1 mL sterile saline followed by 3 mL Bupivavaine which was injected into the intercostal space immediately cranially of each rib. Sheep were placed on their right side and instrumentation was carried out in full sterile conditions.
A 10 cm incision was made on the left side of the neck. To get an index of blood pressure a single-tip pressure probe (Millar Inc., Texas, USA) was inserted into the left common carotid artery. Cannulae were inserted into both the jugular vein for venous infusion and the common carotid artery for arterial blood sampling. The pressure probe and catheters were secured with a purse-string suture (Filasilk, 3.0 non-absorbable braided silk suture) to maintain blood flow through the vessel. The incision in the neck was closed with 1.0 suture.
To gain access to the heart a dorsal to ventral incision was made on the left side of the chest with a scalpel (size 20) and diathermy (Aaron 250, Bovie Medical), muscle layers were separated and either the fourth or fifth rib was removed. The chest was opened with a retractor and the pericardial sack opened. To measure cardiac output the aorta was separated from the pulmonary artery, and a doppler flow probe (Size 22, Transonic, AU) was placed around the ascending aorta for a direct measure of cardiac output on a beat-by-beat basis. For cardiac pacing two pacing leads (Biotonik, Berlin, Germany, Solia S 53- in case of failure in one) were secured externally to the left atrium with 3.0 Filasilk suture and silicone gel. The pericardial sack was closed with 3.0 Filasilk suture. To gain a real time measure of respiration electrodes were implanted into the diaphragm to give a measure of diaphragmatic EMG (‘DEMG’) as previously described . Two strips of seven-stranded Cooner Wires (AS 633-7SSF, Cooner Wire, CA, USA) were passed through the diaphragm using a needle and an exposed section of the wire was in contact with the diaphragm and secured with silicone gel.
The chest was closed using sutures in each of the tissue layers including the intercostal muscles (Covidien, Surgipro, 1.0 monofilament polypropylene suture) and skin (1.0 braided silk suture). Negative pressure was re-established in the chest, flow probe, pacing and DEMG leads were tunnelled sub-cutaneously and exited percutaneously on the dorsum of the sheep for connecting to chronic recording devices after recovery.
Sheep were given antibiotic injections (6 mL i.m.; Oxytetra, Phenix, NZ), and analgesia (Ketofen 10%, 1 mL i.m.; Merial, Boehringer Ingelheim, NZ) at the start of surgery, and for the first 3 days post-operatively. Animals were allowed to recover for 5–7 days.
Hemodynamic measurements and analysis
All analyses were repeated independently by two experienced members of the group. Cardiovascular and respiratory parameters were recorded from conscious sheep in heart failure on a desktop computer with a CED micro 1401 interface and a data acquisition program (Spike2 v8, Cambridge Electronic Design, UK). A baseline recording period was acquired when heart rate and cardiac output had stabilized post-operatively (between 5 and 7 days). Continuous arterial blood pressure, cardiac output from the ascending aorta blood flow probe and heart rate were recorded 24 h per day for 5 weeks. Heart rate was calculated from the inter-pulse interval of the blood flow in the ascending aorta. DEMG signal was amplified (X10, 000), and filtered (band pass 0.3–3.0 kHz).
Sheep were randomized into three different groups before instrumentation surgery. There was a loss of four animals in total (out of 24) during the instrumentation surgery but this was not different between the groups. The first three animals were time controls, following which animals were randomly assigned to the different groups. The sinusoidal group was conducted once the results from the RSA and monotonic groups were collected.
Monotonic—rate fixed pacing (n = 6)
Pacing leads were connected to a stimulator (Grass Instruments) and pacing was set at 10–15 beats per minute above the resting heart rate of each sheep (Fig. 1E). To achieve the desired heart rate the frequency of the stimulation pulse (1.5–2.5 V, 2 ms pulse width) was adjusted and visualized in Spike2 before connecting to the pacing lead of the animal. Pacing voltage was adjusted during the 4 weeks pacing period if pacing became intermittent.
RSA pacing (n = 6)
RSA pacing was achieved using a biofeedback device described previously [37,38,39]. This device performed real-time integration of the diaphragmatic electromyographic (EMG) activity , used to define the inspiratory phase (Ceryx Medical, UK; Fig. 1A). The device exploits the excitatory response of neuronal oscillators to increase and decrease heart rate during inspiration and expiration, respectively (Fig. 1A, F) [36, 37]. DEMG input was split to enable both raw DEMG to be recorded and provide an input into the pacing device. Pacing was set 10–15 beats per minute above resting heart rate with an RSA magnitude (peak-to-trough) of 12 beats per minute. Pacing voltage was set at 1.5 V and pulse width at 2 ms. Pacing voltage was increased if and as needed during the four-week pacing period. RSA pacing was visually checked against the DEMG channel to ensure the rising phase of heart rate correlated with inspiration, and the falling phase of heart rate with expiration (Fig. 1F).
Time control—no pacing (n = 5)
Time control animals were set up identically to the other two groups but had no pacemaker connected to the pacing lead. Our protocols ensured that both the RSA and Mono paced animals received the same number of heartbeats (Fig. 2C).
Pacing under periodic modulation (n = 3)
Variable heart rate pacing, not phase locked to respiration, was achieved by feeding a steady-state square wave signal as the input into the RSA pacing device (instead of the animal DEMG input). Frequency was set to produce a peak-to-tough amplitude of 12 bpm, comparable to the RSA group (stimulation pulse 1.5–2.0 V, 2 ms pulse width).
To determine the percentage of the day the animals were being paced, pacing efficacy was calculated. For monotonically paced sheep the number of heartbeats with a beat interval outside the target paced range was calculated as a percentage of the total number of heart beats. For RSA paced animals, a threshold was placed on the heart rate channel at a value just below the pre-set peak heart rate change during inspiration. This value was subtracted from the total number of breaths to give RSA pacing efficacy.
Breath rate and apnoea analysis
Breathing parameters were assessed from 24 h DEMG recordings. Breath rate was determined as average breaths per minute over 24 h. Apnoea was defined as cessation of diaphragmatic activity. Apnoea incidence was determined by calculating the total number of apnoeas of given lengths (> 3 secs, > 4 secs, > 5 secs, > 6 secs) in 24 h.
Plasma brain natriuretic peptide measurement
Arterial blood samples (20 mL) were collected into EDTA tubes (BD Vacutainer, NJ, USA). Plasma was separated from whole blood by centrifugation (4 °C, 3000 rpm, 15 min), rapidly frozen and stored at − 80 °C before processing. All samples were coded and measurements were made in blinded fashion by the Christchurch Heart Institute; results were returned for decoding. The assay for brain natriuretic peptide has been described previously.
Direct recordings of renal sympathetic nerve activity
An incision was made in the left flank and the left renal artery and renal nerve exposed using methods routine in the lab. Renal sympathetic nerve activity (RSNA) was differentially recorded between a pair of electrodes. The signal was amplified (× 20,000) and filtered (bandpass 400–1200 Hz). Mean arterial pressure was obtained by the pressure probe already implanted. Changes in blood pressure and renal sympathetic nerve active were measured in response to phenylephrine and sodium nitroprusside. For both drugs, the concentration increased at 1-min intervals and doses were 25, 50, 100, 200, and 400 mg/min. All the parameters were recorded on a desktop computer with a CED micro 1401 interface and a data acquisition program (Spike2 v8, Cambridge Electronic Design, UK).
At the end of these experiments, the sheep were euthanized with an overdose of sodium pentobarbitone (0.5 mL/kg, intravenously) (Provet NZ Pty Ltd., New Zealand). Once death was confirmed, heart and lungs were weighed, in relation to tibia length and cardiac tissue collected for further analysis.
All tissue samples were collected from the LV mid-free wall, approximately half-way between the atria and the apex. Care was taken to ensure all tissue samples collected were from the same location between animals, and, therefore, should represent a similarly ischemic area of tissue.
Cell membrane staining
Freshly dissected samples from the LV were cut into 0.5 cm thick 1 cm strips spanning the epi-to endocardial surfaces. Samples were collected from all sheep from the same region of the LV. Prior to fixation samples were placed in Tyrode’s solution containing BDM to prevent contraction artifacts from rapid cooling. LV samples were fixed in 1% paraformaldehyde for 1 h 4 °C, before being moved through three solutions of increasing sucrose gradient (10%, 20%, 30%) over 48 h. Once cryoprotected with sucrose, tissue samples were frozen in liquid nitrogen chilled methylbutane (Thermofisher), and stored at -80 °C before processing.
Fixed frozen samples were cut into 12 µm sections on a Leica CM 1900 cryostat and mounted on Superfrost Plus cover slips. Slides were rinsed with phosphate-buffered saline (PBS), and incubated in Fx signal enhancer (Thermofisher) for 1 h at room temperature. Slides were then washed 3 × with PBS, followed by incubating in a 1:100 concentration of wheat-germ agglutinin (WGA, Alexa Fluor 594, Fisher. cat# W11262) for 2 h at room temp. Slides were again washed 3 × with PBS before applying VectaShield and cover slipping. All analysis of cell size and t-tubule power was conducted by experienced members blinded to the group status.
Fluorescent images of labelled tissue sections were imaged on an Olympus FV1000 confocal microscope (diode-pumped 559 nm laser, 20×/0.8 NA oil, FluoView 4.2). 5–8 different regions from each stained LV slice were imaged at 20×. Analysis was performed using ImageJ (Fiji). A two-pixel median filter was applied to all images. Individual cardiomyocytes deemed to be lying parallel to the imaging, in full longitudinal view were traced around the circumference to measure cell area (µm2). For each region, 5–10 different cells were measured and an average cell size presented.
T-tubule power was determined by previously described methods. The WGA stained LV samples were imaged on an Olympus FV1000 confocal microscope (diode-pumped 559 nm laser, 60×/1.35 NA oil, FluoView 4.2) to a pixel size of 80 nm. 10–12 areas were imaged at 60 × per LV section. Images were analyzed in ImageJ (Fiji). All images underwent a 2 pixel median filter, and 0.3 pixel background subtract. FFT was performed along a selected 20 µm line (Extended Data 3). 2–4 individual myocytes were analyzed per image. T-tubule power was determined from a power spectrum obtained from the FFT.
All data are expressed as mean ± SEM, except where indicated. All-time course data (cardiac output, heart rate, mean arterial pressure, systemic vascular resistance, stroke volume, breath rate) were analyzed using a repeated measures 2-way ANOVA, or mixed-effects model if any missed data points were ‘missing at random’. Posthoc analysis was completed if there was an effect of time and Dunnetts two-sided post hoc tests were used to compare the time points to baseline. Renal sympathetic nerve recordings were analyzed in SigmaPlot. Heart weight, and lung weight were analyzed between monotonic and RSA paced sheep using an unpaired two-tailed students t-test. Histology analysis was analyzed between three groups, time control, monotonic paced, and RSA paced using an ordinary one-way ANOVA. All statistical analysis was performed in SPSS (v8.1). Data were considered significant if p < 0.05.