1Electrical and Computer Engineering Department, The George Washington University, 2Pharmacology and Physiology Department, The George Washington University
Asfour, H., Wengrowski, A. M., Jaimes III, R., Swift, L. M., Kay, M. W. NADH Fluorescence Imaging of Isolated Biventricular Working Rabbit Hearts. J. Vis. Exp. (65), e4115, doi:10.3791/4115 (2012).
Since its inception by Langendorff1, the isolated perfused heart remains a prominent tool for studying cardiac physiology2. However, it is not well-suited for studies of cardiac metabolism, which require the heart to perform work within the context of physiologic preload and afterload pressures. Neely introduced modifications to the Langendorff technique to establish appropriate left ventricular (LV) preload and afterload pressures3. The model is known as the isolated LV working heart model and has been used extensively to study LV performance and metabolism4-6. This model, however, does not provide a properly loaded right ventricle (RV). Demmy et al. first reported a biventricular model as a modification of the LV working heart model7, 8. They found that stroke volume, cardiac output, and pressure development improved in hearts converted from working LV mode to biventricular working mode8. A properly loaded RV also diminishes abnormal pressure gradients across the septum to improve septal function. Biventricular working hearts have been shown to maintain aortic output, pulmonary flow, mean aortic pressure, heart rate, and myocardial ATP levels for up to 3 hours8.
When studying the metabolic effects of myocardial injury, such as ischemia, it is often necessary to identify the location of the affected tissue. This can be done by imaging the fluorescence of NADH (the reduced form of nicotinamide adenine dinucleotide)9-11, a coenzyme found in large quantities in the mitochondria. NADH fluorescence (fNADH) displays a near linearly inverse relationship with local oxygen concentration12 and provides a measure of mitochondrial redox state13. fNADH imaging during hypoxic and ischemic conditions has been used as a dye-free method to identify hypoxic regions14, 15 and to monitor the progression of hypoxic conditions over time10.
The objective of the method is to monitor the mitochondrial redox state of biventricular working hearts during protocols that alter the rate of myocyte metabolism or induce hypoxia or create a combination of the two. Hearts from New Zealand white rabbits were connected to a biventricular working heart system (Hugo Sachs Elektronik) and perfused with modified Krebs-Henseleit solution16 at 37 °C. Aortic, LV, pulmonary artery, and left & right atrial pressures were recorded. Electrical activity was measured using a monophasic action potential electrode. To image fNADH, light from a mercury lamp was filtered (350±25 nm) and used to illuminate the epicardium. Emitted light was filtered (460±20 nm) and imaged using a CCD camera. Changes in the epicardial fNADH of biventricular working hearts during different pacing rates are presented. The combination of the heart model and fNADH imaging provides a new and valuable experimental tool for studying acute cardiac pathologies within the context of realistic physiological conditions.
1. Setting Up for the Study
2. Heart Excision
3. Biventricular Cannulation
4. Signal Acquisition: Pressures, Monophasic Action Potentials, and fNADH
5. Off-line Processing of fNADH Images
6. Representative Results
Anterior and basal views of a biventricular working rabbit heart preparation are shown in Figure 1. Left ventricular pressure was measured by navigating a pressure transducer catheter (Millar SPR-407) past the aortic valve and into the left ventricle. Aortic, pulmonary artery, and left ventricular pressures (LVP) are shown in Figure 1C. Diastolic LVP is usually between 0 and 10 mmHg. The minimum diastolic aortic pressure is approximately 60 mmHg. Peak systolic LVP is dependent upon filling pressure (the preload or LA pressure) and contractility and, optimally, should be between 80 and 100 mmHg. The maximum aortic pressure and maximum LVP should closely match, as shown in Figure 1C.
Monophasic action potentials (MAPs) with a fast depolarization phase and a repolarization phase that are typical for rabbit hearts are shown in Figure 1D. MAPs can be recorded relatively easily from a contracting heart but will usually have small motion artifact during diastole, as shown in Figure 1D. MAPs are useful for confirming successful entrainment of the heart (capture) during pacing and can also be used to measure local electrophysiological changes due to ischemia or other acute perturbations. An ECG could also be measured by submerging the heart in a bath of warm superfusate and placing an electrode in the bath on the left and right sides of the heart. A third indifferent electrode is either placed in the bath, away from the heart, or is attached to the aorta. An ECG will provide information regarding the global excitation and repolarization process, which is useful for evaluating overall electrical function and for revealing the presence of ischemia.
fNADH imaging reveals changes in the mitochondrial redox state of the heart, which can be used to measure the spatiotemporal progression of ischemic or hypoxic regions. For this study, epicardial fNADH was measured to monitor changes in redox state during three pacing rates at cycle lengths (CLs) of 300, 200, and 150 msec. Average fNADH values from a region of interest (red box, Figure 2) show that baseline fNADH levels increase as the cycle length is shortened. When pacing rate is close to sinus rhythm (CL=300 msec) baseline fNADH level is relatively constant. As cycle length is shortened below 300 msec, baseline fNADH levels increase, with the largest increase at the shortest CL (150 msec). High resolution fNADH imaging of the full anterior surface at 200 and 400 bpm is shown in Figure 3. fNADH levels at 200 bpm were constant and spatially homogeneous. At 400 bpm, fNADH levels increased substantially throughout the epicardium. Significant spatial heterogeneity was observed with the largest increases occurring within the septal regions of the RV and LV.
The fNADH signal oscillates with contraction (motion artifact) and the frequency of oscillation corresponds to heart rate (Figure 2). In biventricular cannulation, the base of the heart is held by 4 cannulae, which helps to prevent the heart from swinging during contraction. Therefore, oscillation amplitude is always less than any longer time scale (5-10 sec) trends in fNADH that are caused by ischemia or hypoxia.
Figure 1. Typical pressures and monophasic action potentials from an isolated biventricular working rabbit heart. A. Basal view of the heart showing the four cannulae: 1, aortic; 2, pulmonary artery; 3, left atrial; and 4, right atrial. B. Anterior view of the heart showing the left ventricle (LV) and the right ventricle (RV). C. Representative pressures. Top: left ventricular pressure (solid line) and the aortic pressure (dotted line). Bottom: pulmonary pressure. D. Representative monophasic action potentials. The signal is aligned with the pressures shown in panel C. Click here to view larger figure.
Figure 2. fNADH imaging of an isolated biventricular working rabbit heart. Top: A cartoon of the field of view (left) and three fNADH images are shown. The corresponding pacing cycle length (CL) is indicated on each image. The region of interest for the fNADH signal in the bottom panel is indicated by the red box. The tip of the monophasic action potential electrode is seen to the right of the region of interest. The epicardium was illuminated using the mercury lamp and light guide, as shown in Figure 5. Only the epicardial surface surrounding the region of interest was illuminated. Bottom: Average fNADH for the region of interest indicated by the red box in the top panel. Average fNADH increases with reduced cycle length.
Figure 3. fNADH images of the full anterior surface of an isolated biventricular working rabbit heart. The heart was paced from the RA at 200 bpm and 400 bpm. fNADH was imaged (2 fps, 128x128 pixels at a resolution of 0.4 mm) while illuminating the entire anterior epicardium using two high power LEDs (Mightex PLS-0365-030-S, 365 nm, 4% intensity, 50 mW max).
The isolated Langendorff perfused heart remains a prominent tool for studying cardiac physiology2. It is especially useful in studies of cardiac arrhythmias, particularly those that use fluorescence imaging of transmembrane potential20. An advantage is that the entire epicardium of the isolated heart can be observed21, 22. Another advantage is that, in contrast to blood, perfusion with a clear crystalloid buffer solution does not interfere with fluorescence signals. A limitation is that the Langendorff technique is not well-suited for studies of cardiac metabolism, which often require the heart to perform work within the context of physiologic preload and afterload pressures.
To elevate the relevance of isolated heart preparations for metabolic studies, Neely introduced modifications to the Langendorff technique to establish appropriate left ventricular (LV) preload and afterload pressures3. The model is known as the isolated LV working heart model and has been used extensively to study LV performance and metabolism4-6. The LV working heart model is superior to the Langendorff model for functional evaluations, yet it does not provide a properly loaded right ventricle (RV). Demmy et al. first reported a biventricular model (LV & RV) as a modification of the LV working heart model7, 8. They found that stroke volume, cardiac output, and pressure development improved in hearts converted from working LV mode to biventricular working mode8. A properly loaded RV also improves septal function by diminishing abnormal pressure gradients across the septum. Biventricular working hearts have been shown to maintain aortic output, pulmonary flow, mean aortic pressure, mean pulmonary pressure, heart rate and myocardial ATP, and creatine phosphate levels for up to 3 hours8. Biventricular working heart studies typically use hearts from small animals, such as rats and rabbits, because the cardiac output and the required volume of perfusate are much less than that for hearts of larger animals. However, biventricular working heart studies have been conducted using hearts from swine, canines, and even humans23, 24.
The metabolic demand of isolated hearts in biventricular working mode is considerably higher than that of Langendorff perfusion. It is important that the perfusate solution provide enough oxygen and metabolic substrate to support biventricular heart function. Standard crystalloid buffer solutions, such as Krebs-Henseleit16, 17, 25 or Tyrodes26, 27, have oxygen solubilities as high as 5.6 mg/L. When these solutions are gassed with carbogen (a gas blend of 95% O2 and 5% CO2) and contain suitable metabolic substrate (glucose, dextrose, and/or sodium pyruvate), they are appropriate for biventricular working hearts beating at normal sinus rates (approximately 180 bpm for a rabbit).
Metabolic demand increases for fast rhythms and the amount of oxygen dissolved in standard perfusates might not be enough to fully support a biventricular working heart that is contracting at high rates. Crystalloid buffer solutions containing erythrocytes or mixed with whole blood have been used in working heart preparations to ensure adequate oxygen availability. Previous studies have shown that adding erythrocytes to a Krebs-Henseleit solution improved working heart function during rigorous pacing protocols and also reduced the incidence of ventricular fibrillation16. A limitation of using erythrocytes or mixtures of whole blood is that hemoglobin interferes with light wavelengths that are used for fluorescence imaging13. Other substrates, such as albumin, may also be added to perfusate solutions to prolong heart viability and reduce edema28.
During fluorescence imaging the intensity of excitation light should be high and the light distribution should be uniform. Achieving uniform illumination is not always easy due to the curvature of the epicardial surface. In our studies, we image fNADH by filtering light (350±25 nm) from a mercury lamp. A bifurcated fiber optic light guide is used to direct the UV light onto the epicardial surface. Uniform lighting can be achieved by appropriate positioning the two output ferrules. UV LED light sources could also be used, as we have demonstrated in Figure 3. LED sources are relatively inexpensive so multiple sources could be incorporated into an imaging system. LEDs can also be cycled on and off at high rates to synchronize excitation light with image acquisition.
Photobleaching of NADH should be minimized29 by reducing the time of tissue illumination. This can be done by cycling the illumination on and off using an electronic shutter and a lamp or with an LED lighting system and a controller. If illumination is synchronized with the cardiac cycle, then fNADH image acquisition could be confined to diastole, which would reduce motion artifact in the fluorescence signals. Trigging illumination and image acquisition using a pressure signal, such as LV pressure, would be one way to do this.
In our studies we have observed that changes in fNADH per unit time can be more than 5X higher at 400 bpm than at 200 bpm. This indicates that fast rhythms elevate the redox state of the heart. Whether or not this is caused by hypoxia or the inability of myocytes to oxidize NADH to NAD+ quickly enough to avoid the accumulation of NADH is still an unanswered question.
The performance of a biventricular working heart preparation is contingent upon multiple factors. One of the most important is to set appropriate preload and afterload pressures to mimic the physiological conditions that are under investigation. In particular, the LV afterload (aortic pressure) must be adjusted to represent systemic pressure. If it is too high, the LV will not be able to overcome the pressure, resulting in regurgitation. Pressure that is too low will adversely affect coronary perfusion. The LV preload pressure (left atrial pressure) should also be adjusted to provide an end diastolic volume that is appropriate for the experimental protocol.
fNADH imaging of living tissue is an established mode of fluorescence imaging13. Its application to cardiac tissue was illustrated by Barlow and Chance when they reported striking elevations of fNADH within regionally ischemic tissue after ligation of a coronary vessel14. Their fNADH images were recorded on film using a Fairchild oscilloscope camera and UV flash photography. Coremans et al. expanded upon this concept using the NADH fluorescence/UV reflectance ratio to measure the metabolic state of the epicardium of Langendorff blood-perfused rat hearts30. A videofluorimeter was used for imaging and data was recorded using a video recorder. Later, Scholz et al. used a spectrograph and photodiode array to measure average fNADH from a large area of the LV. This approach reduced the effects of epicardial fluorescence heterogeneities and local variations in circulation while revealing macroscopic work-related variations of fNADH31. This approach is similar to computing average fNADH levels for a region of interest across all frames of an fNADH imaging dataset, as illustrated in Figure 2. As we have presented in this article, today's technology provides high-speed CCD cameras and digitally controlled high-power UV spotlights. These technologies enable the spatiotemporal dynamics of fNADH and cardiac metabolism to be studied from many new perspectives. The relatively low-cost of the optics and light source makes fNADH imaging a useful accessory for conventional cardiac optical mapping systems.9, 32
No conflicts of interest declared.
This work was supported by a grant from the NIH (R01-HL095828 to M. W. Kay).