Author + information
- Received June 1, 2020
- Revision received June 19, 2020
- Accepted June 24, 2020
- Published online August 17, 2020.
- Rostislav Bychkov, PhD,
- Magdalena Juhaszova, PhD,
- Kenta Tsutsui, MD, PhD,
- Christopher Coletta, MS,
- Michael D. Stern, MD,
- Victor A. Maltsev, PhD and
- Edward G. Lakatta, MD∗ ()
- Laboratory of Cardiovascular Science, Intramural Research Program, National Institute on Aging, National Institutes of Health, Baltimore, Maryland
- ↵∗Address for correspondence:
Dr. Edward G. Lakatta, Laboratory of Cardiovascular Science, NIA/NIH, Biomedical Research Center, 251 Bayview Boulevard, Baltimore, Maryland 21224.
Objectives This study sought to identify subcellular Ca2+ signals within and among cells comprising the sinoatrial node (SAN) tissue.
Background The current paradigm of SAN impulse generation: 1) is that full-scale action potentials (APs) of a common frequency are initiated at 1 site and are conducted within the SAN along smooth isochrones; and 2) does not feature fine details of Ca2+ signaling present in isolated SAN cells, in which small subcellular, subthreshold local Ca2+ releases (LCRs) self-organize to generate cell-wide APs.
Methods Immunolabeling was combined with a novel technique to detect the occurrence of LCRs and AP-induced Ca2+ transients (APCTs) in individual pixels (chronopix) across the entire mouse SAN images.
Results At high magnification, Ca2+ signals appeared markedly heterogeneous in space, amplitude, frequency, and phase among cells comprising an HCN4+/CX43− cell meshwork. The signaling exhibited several distinguishable patterns of LCR/APCT interactions within and among cells. Rhythmic APCTs that were apparently conducted within the meshwork were transferred to a truly conducting HCN4−/CX43+ network of striated cells via narrow functional interfaces where different cell types intertwine, that is, the SAN anatomic/functional unit. At low magnification, the earliest APCT of each cycle occurred within a small area of the HCN4 meshwork, and subsequent APCT appearance throughout SAN pixels was discontinuous and asynchronous.
Conclusions The study has discovered a novel, microscopic Ca2+ signaling paradigm of SAN operation that has escaped detection using low-resolution, macroscopic tissue isochrones employed in prior studies: synchronized APs emerge from heterogeneous subcellular subthreshold Ca2+ signals, resembling multiscale complex processes of impulse generation within clusters of neurons in neuronal networks.
The sinoatrial node (SAN), the heart’s primary pacemaker, exhibits a high degree of structural and functional complexity that is vital for generating flexible heart rates and robust rhythms (1). Following the discovery of the SAN in 1907 by Keith and Flack (2), numerous SAN cell characteristics have been described, including a diversity of: cell shapes and protein expression; arrangement of cells within SAN tissue (e.g., gradient vs. mosaic ); autonomic neuronal input (4); cell-to-cell electrical and mechanical interactions (5); and intranodal impulse initiation, spread within, and exit from, the SAN (6–9).
A generalized view of coordinated firing of pacemaker cells within the SAN emerged about 40 years ago, in which a dominant or “master” pacemaker cell or a leading pacemaker center dictates the excitation rate and rhythm of thousands of other SAN pacemaker cells by overdriving their intrinsic spontaneous excitation rates (6,10). Shortly thereafter, however, an idea of mutual entrainment of coupled oscillators (11) was applied to the coordinated firing of the entire population of SAN cells (12,13): individual SAN cells that are loosely connected generate spontaneous excitations that differ in phase, mutually entrain each other to fire with a common period. The respective intrinsic cell oscillation frequencies of individual cells and the degree of intercellular electrical coupling determined the common period at which all cells fire. In other terms, the mutual entrainment theory of SAN (12) proposed that whereas the ensemble of SAN cells can be entrained to operate at a given frequency, there can be marked phase differences among spontaneous excitations of individual cells, and the frequency of impulses that exit the SAN lies somewhere between the fastest and slowest spontaneous intrinsic excitation rates of resident SAN cells.
Thus, rather than mimicking classical electrical conduction by consecutively exciting each other, as in ventricular muscle tissue, SAN cells are in fact mutually entrained by phase resetting, and conduction within SAN is only “apparent” (13,14). The idea of apparent conduction was a major departure from the classic view of the SAN as a slowly conducting tissue in which the action potential (AP), initiated by a small cluster of dominant pacemaker cells, propagates radially at an accelerating pace. This new idea was supported by later studies that showed that the initiation of the cardiac impulse could be multimeric, that is, initiated from different locations (15). The early ideas of mutual entrainment were further elaborated in elegant studies by Verheijck et al. (16) using computer-controlled coupling conductance between individual pacemaker cells and found a critical coupling conductance for 1:1 frequency entrainment of <0.5 nS (i.e., generated by a few connexin molecules).
Similar to the evolution of ideas on how the SAN impulse is generated, the functional paradigm of pacemaker cells operating in isolation has also undergone substantial refinement from a simple to more complex origin: from what was described as the “pacemaker channel” that solely drives spontaneous diastolic depolarization (17), to a modern concept of a coupled system of chemical and ion current oscillators (18,19). According to this coupled oscillator theory, rhythmic local intracellular Ca2+ releases (LCRs) generated by an intracellular Ca2+ oscillator during diastole self-organize into a powerful and timely Ca2+ signal that drives the ensemble of membrane current oscillators to generate inward current culminating in membrane depolarization, sufficient to trigger a cell-wide AP.
More specifically, the interactions of spontaneous LCRs with Na+/Ca2+ exchanger (NCX) begin to occur near the maximum diastolic potential in the context of NCX current activation and anomalous rectification of hyperpolarization-activated cyclic nucleotide-gated channel 4 (HCN4) current both activated by hyperpolarization, and decay of K+ currents. These combined actions of LCRs, NCX, and HCN4 regulate cell surface membrane potential within a range that activates low-threshold L-type Ca2+ current (Cav1.3), or T-type Ca2+ current (Cav3.1) which, in turn, via Ca2+-influx–induced Ca2+ release via ryanodine receptors, increases and feeds forward the ensemble LCR Ca2+ signal and NCX current to accelerate the rate of the diastolic depolarization (20). The transition in LCR characteristics is steeply nonlinear, resembling a phase transition (21). Thus, simultaneous growth of the diastolic ensemble LCR signal and diastolic surface membrane depolarization are thus manifestations of a slowly evolving self-organized electrochemical gradient oscillation that reaches criticality in late diastole, followed by a phase transition manifest as the rapid AP upstroke that triggers a whole-cell cytosolic Ca2+ transient, that is, AP-induced Ca2+ transient (APCT). In other terms, spontaneous LCR emergence at a critical time of an AP cycle (late diastole) entrains membrane depolarization, and late diastole, therefore, might be considered to be an “intrinsic entrainment zone” (14) or a “time window” in which chemical and current oscillators become mutually entrained, controlling the rate and rhythm of AP firing of that cell.
A wide spectrum of spontaneous AP firing cycle lengths observed among single SAN cells in isolation under a large variety of experimental conditions is, in fact, predicted by the spectrum and degree of synchronization of LCR periods among these cells (19,22). Here, we hypothesized that on the larger scale of intact SAN tissue, LCRs not only occur within cells, but: 1) differ in spatial distribution, frequency, amplitude, and phase; and 2) heterogeneity of local Ca2+ signaling characteristics among cells translates into a different pattern of pacemaker cell excitation within the node.
Our experiments conformed to the Guide for the Care and Use of Laboratory Animals, published by the U.S. National Institutes of Health. The experimental protocols were approved by the Animal Care and Use Committee of the National Institutes of Health (protocol #034-LCS-2019). We used 1- to 3-month-old C57BL mice (Charles River Laboratories, Wilmington, Massachusetts) anesthetized with sodium pentobarbital (50 mg/kg). The adequacy of anesthesia was monitored until reflexes to tail pinch were lost.
To determine whether the Ca2+ signaling patterns in human SAN resemble those recorded in mouse SAN, we recorded Ca2+ signals from SANs of adult human hearts, acquired as part of our prior study in which we recorded Ca2+ signals in isolated human SAN cells (23). In short, hearts (not required for transplantation) were procured from Washington Regional Transplant Community (Washington, DC). Experimental protocols were approved by the George Washington University Institutional Review Board. Informed donor consent was obtained for all tissue used in this study. All methods related to human SAN preparation were performed in accordance with the National Institutes of Health guidelines on human research. Human SAN was isolated as previously described (23). Unlike mouse SAN, the human SAN has a fat layer that was carefully dissected in order to have clear optical access to SAN cells.
The SAN was dissected according to standard methods (24). The heart was removed quickly and placed in standard Tyrode solution containing (in mmol/l): 130 NaCl, 24 NaHCO3, 1.2 NaH2PO4, 1.0 MgCl2, 1.8 CaCl2, 4.0 KCl, 5.6 glucose equilibrated with 95% O2/5% CO2 (pH 7.4 at 35.5°C). The whole heart was pinned to a silicon platform under a surgical microscope in order to excise the right and left atria. A 10-ml tissue bath was perfused with standard solution at a rate of 10 ml/min. After removal of the ventricles, the right atrium was opened to expose the crista terminalis, the intercaval area, and the interatrial septum. The preparation was not trimmed, leaving the SAN region together with the surrounding atria and superior vena cava (SVC) and inferior vena cava (IVC) intact. The SAN preparation was pinned to the silicon bottom of the experimental chamber by small stainless-steel pins with the endocardial side exposed. Care was taken to provide the minimal amount of stretch required to flatten the SAN tissue. After mounting, the preparation was superfused with solution maintained at a temperature of 36 ± 0.3°C. Anatomic landmarks were used to locate the SAN (Figure 1A).
Optical system for SAN imaging
We developed a novel imaging system to assess intracellular Ca2+ dynamics within individual cells resident within the entire mouse intact SAN. We used a stationary fixed-stage upright microscope (AxioExaminer D1 equipped with zoom tube [0.5× to 4×], Carl Zeiss Microscopy LLC, Oberkochen, Germany) and edge 4.2 camera (PCO Imaging, Kelheim, Germany), featuring a scientific complementary metal-oxide semiconductor (sCMOS) sensor with a high spatial and temporal resolution (Figure 1B). The microscope was mounted on a motorized X-Y MT-2078/MT-2278 translator (Sutter Instruments, Novato, California). The experimental chamber with the SAN preparation was placed on a platform (Sutter Instruments) that was mounted onto a pressurized air table (Newport Corporation, Irvine, California).
Imaging of local Ca2+ signals in SAN tissue
The SAN preparation was incubated with a membrane-permeable Ca2+ indicator Fluo-4 AM (10 μmol/l) for 1.5 h. In some experiments, we also recorded Ca2+ signals specific to HCN4+ cells using SANs from mice (pCAGGS-GCaMP8) with HCN4-targeted expression of a Ca2+ probe, provided by Michael Kotlikoff, VMD, PhD, College of Veterinary Medicine, Cornell University (25).
The excitation light (CoolLED pE-300 ultra, CoolLED, Andover, United Kingdom) was directed to the microscope via a single-port epifluorescence condenser, providing uniform illumination of the object plane. To optimize the signal-to-noise ratio and durability of the preparation, we applied excitation light to the SAN at 40% of the maximum power of the light source CoolLED. The excitation light was reflected to the SAN preparation by a dichroic mirror with a central wavelength of 498 nm, and the emitted fluorescence signal was collected through a 530 ± 20-nm filter (Semrock, Rochester, New York). The fluorescence image of the SAN preparation was projected by air or water lenses onto the sCMOS camera sensor. To prevent interference of tissue motion during recordings from SAN tissue, we decoupled electrical excitation and mechanical contraction in some preparations by inhibiting the formation of the Ca2+-sensitive regulatory complexes within sarcomeres using 10 μmol/l cytochalasin B (26). Otherwise, residual mechanical artefacts produced during diastolic phase were compensated with ImageJ software version 1.52 (NIH, Bethesda, Maryland) during the post-processing of recorded images.
To image APCTs across entire SAN preparations (n = 7) from SVC to IVC, we used 2.5× and 5× magnification. To visualize local Ca2+ dynamics of individual cells within the SAN in a given region of interest (ROI), we used 10× and 20× magnification water immersion lenses (W Plan-Apochromat 10×/0.5 M27 75 mm, Plan-Apochromat 20×/1.0 DIC D = 0.17 M27 75 mm). The combination of our camera sCMOS image sensor of 2,200 by 2,200 pixel resolution within a 4,000 μm by 4,000 μm area observed via a 2.5× air objective permits a resolution of 1.8 μm per pixel. A 2,000 μm by 2,000 μm area covered by air 5× objective permits resolution about 0.9 μm per pixel. Water immersion objectives of 10× and 20× permitted resolution of about 0.45 μm and 0.225 μm per pixel. Ca2+ signals had a high signal-to-noise ratio (>10) over the entire SAN from SVC to IVC, even at a low optical resolution with a 2.5× (air) lens. This signal consistency, together with high temporal (500 to 700 Hz) and spatial resolution of SAN images (2,200 × 320 pixels of 2 × 4 mm tissue) allowed acquisition of data required to construct spatial and temporal maps of SAN cell network activation in each pixel, that is, chrono-pixel, or “chronopix” for short. Ca2+ indicator fluorescence signal intensity was stable for 20 to 35 min.
Because our Ca2+ imaging is not confocal, we cannot completely rule out the possibility that signals from different cells above and below the focal plane within SAN tissue can overlap and contribute to recording in the focal plane (i.e., out-of-focus signals). Although such overlap may occur in our recordings at low magnification (panoramic whole-mount SAN imaging), we did not examine LCRs in individual cells at low resolutions. When imaging at higher magnifications using water emersion objectives, signal overlap was a minor factor because of substantially higher resolving power and smaller point spread function of the optical system. The minor contribution of out-of-focus events was evidenced by the fact that bright LCRs (observed by eye and analyzed here) occurred within, but not outside, the cell contour. If out-of-focus Ca2+ signals were present, they would be seen randomly everywhere in the focal plane, regardless of the visible cell contours. Furthermore, our SAN preparation is flat, and individual cells are oriented mainly along the recording plane; their shapes are clearly seen within the focal plane. These shapes are similar to those observed in our confocal 3-dimensional (3D) measurements with 40× oil immersion objective used in our imaging of immunolabeled cells (described later in the text).
Simultaneous Ca2+ imaging and AP recording
We used spontaneous APCTs to inform on the occurrence and locations of spontaneous APs within the SAN. To validate APCTs as reporters of APs, we simultaneously measured APs and APCTs in a subset (n = 3) of SAN preparations. To record the transmembrane potential, we used sharp microelectrodes (40 to 70 MΩ) fabricated from aluminosilicate glass capillaries (1.5-mm outer diameter, 0.86-mm inner diameter) with a horizontal pipette puller (Sutter Instruments), and back-filled with 3 mol/l KCl. The neck of the recording sharp microelectrode was pulled long enough to be sufficiently “springy” in order to follow tissue movement without compromising the recording. A sharp glass microelectrode was inserted from the endocardial side of the tissue. APs were recorded with a high impedance amplifier (A-M Systems, Sequim, Washington) with a virtual bridge circuit for current injection. The electrical signal was digitized with Digidata 1440 version 1440A and analyzed with PCLAMP10 software version 10.4 (both from Molecular Devices, San Jose, California).
Microsoft Excel (version EXCEL from MSOffice 2019) (Microsoft, Redmond, Washington) and OriginLab software (version ORIGIN 9) (OriginLab Corporation, Northampton, Massachusetts) were used to analyze datasets of membrane potential recordings, amplitudes of intracellular Ca2+ transients measured within the ROIs, and delay times, to normalize to the maximum of recorded traces and to make linear fit. We then also used OriginLab to generate power spectra in time series of local Ca2+ signals. The time series were generated from the average signal within each ROI for each frame in the sequentially recorded stack of images. Note that although we present our data in terms of chronopix depicting the time in milliseconds at which Ca2+ signals follow the occurrence of the earliest Ca2+ signal, they can be also expressed in radians or percentage of the cycle length as stated in the coupled oscillator theory.
Whole-mount SAN immunolabeling
We combined Ca2+ imaging with immunolabeling to correlate Ca2+ dynamics with cytoarchitechture. Another subset (n = 7) of SAN preparations was fixed in 4% paraformaldehyde overnight at 4oC. The SANs were washed 3 times in phosphate-buffered saline (PBS) and permeabilized overnight in PBS containing 0.2% Triton X-100 and 20% dimethyl sulfoxide. After blocking the nonspecific binding sites by incubation for 8 h with 0.2% Tween-20 in PBS containing 3% normal donkey serum, SAN whole-mount preparations were incubated for 3 days with the primary antibodies diluted in PBS containing 0.2% Tween-20 and 3% normal donkey serum. They were then washed 3 times with PBS containing 0.2% Tween-20, incubated overnight with appropriate secondary antibodies, then washed 3 times with PBS containing 0.2% Tween-20. Whole-mount SAN preparations were mounted in VECTASHIELD (Vector Laboratories, Burlingame, California) and sealed with a coverslip. The SAN preparations were mounted with the endocardium uppermost. Immunolabeling of whole-mount SANs was imaged with a Zeiss-LSM510 confocal microscope (Carl Zeiss, Oberkochen, Germany) and a Zeiss-AxioExaminer D1 fluorescence microscope with a zoom tube (0.5× to 4×) equipped with appropriate filters for fluorescence spectra. Parts of the whole-mount SAN were imaged individually, and these images were concatenated into 1 output image to portray the entire SAN preparation. To obtain high-resolution images of the entire SAN, the preparations were imaged in a tile-scanning mode. Fluorescence of immunolabeled cells within SAN tissue in 3D was visualized by confocal optical slicing (Z stacking) to a depth up to 100 μm from the endothelial surface.
Antibodies: HCN4+ cells were identified by rabbit polyclonal antibodies for hyperpolarization-activated, cyclic nucleotide-gate cation channels HCN4 (1:250, Alomone Labs, Jerusalem, Israel). Mouse monoclonal antibody to connexin 43 (1:250, Invitrogen, Carlsbad, California) was used to label the gap junctions of striated cells and Alexa Fluor 633 phalloidin was used to visualize the F-actin filaments.
Immunolabeling of whole-mount SAN for HCN4, connexin 43, and labelling of F-actin
Previous studies in thin slices from central SAN demonstrated that numerous cells have strong HCN4 immunolabeling and lacked connexin 43 (CX43) (27). To explore the expression patterns of key molecules/markers of pacemaker cells, we immunolabeled and reconstructed nearly the entire cellular network within intact whole-mount SAN preparations. We analyzed immunolabeling of the entire SAN at low optical magnification and analyzed immunolabeling within regions in which cells with rhythmical LCRs preceded APCTs at higher magnification. We employed confocal imaging to acquire a 3D reconstruction of the optically dissected intact SAN tissue (of note, in previous studies, 3D reconstruction was performed by physical dissection of SAN tissue with a microtome into separate tissue slices). The resultant 3D networks of intact cells expressing HCN4, F-actin, and/or CX43 were reconstructed and further examined with Zeiss imaging software ZEN2 (version ZEN 2.5). Images were processed with Image J software.
Simultaneous AP recording and intracellular Ca2+ imaging
We imaged spontaneous APCTs to report the occurrence of spontaneous APs. In order to ensure the fidelity between the occurrence of APCTs and the occurrence of APs, we simultaneously acquired Ca2+ signal fluorescence, recorded at lower optical power, and APs, recorded with conventional sharp microelectrodes, in a subset of experiments (Figure 2). In each experiment, a ROI within a Ca2+ image at low magnification was selected around the tip of the sharp microelectrode. Figure 2A shows a representative example of simultaneously recorded APs and APCTs. The AP trace and Ca2+ transient recordings, normalized to their peak amplitudes, are superimposed in Figure 2B. Note that as expected, an APCT shortly followed the corresponding AP onset. APCT cycle lengths plotted against AP cycle lengths showed a tight linear relationship (Figure 2C), validating the use of whole-mount APCT images to report the occurrence of APs within SAN tissue.
Panoramic imaging (at 2.5×) of the entire SAN
The mouse SAN averages 100 to 200 μm in thickness and up to 1,000 μm in width and 4,000 μm in length, enabling visualization of the entire SAN preparation within a microscopic field of view at low optical magnification. We quantified Ca2+ signals in defined ROIs within a 2D SAN image.
Figure 3 illustrates a complete spatiotemporal pattern of the earliest detectable APCTs and of subsequent APCT occurrences within chronopix across the entire SAN within a given focal plane (Video 1). Two ROIs in Figure 3A that are separated by approximately 300 μm show the location in which the earliest APCTs emerge and the region where the next APCTs emerged. When the APCTs from the 2 ROIs (Figure 3A) were normalized to their maxima, superimposed, and plotted as a function of time, their phase difference was 7 ms (Figure 3C). The earliest APCTs occurred within a region of about 20 by 200 μm (Figure 3A), near the SVC. This can be clearly seen as a small red spot at the 2-ms chronopix, and subsequent APCTs emerged 4 to 18 ms thereafter (Figure 3B).
Heterogeneous LCR-APCT patterns among clusters of SAN cells localized within the central SAN
We further inspected the area where the earliest APCT occurred (Figure 4A) (near the APCT ROI in Figure 3) at a higher optical resolution using 10× and 20× water immersion lenses. We discovered Ca2+ signals that were markedly heterogeneous within and among cells, ranging from highly synchronized to less synchronized APCTs, manifest as variably synchronized increases in intracellular Ca2+ throughout the entire volume of a cell (Figure 4B). LCRs, by contrast, were manifest as an increase in Ca2+ signal in 1 part of the cell, while at the same time, Ca2+ in other parts of the same cell remained at its basal level (Figure 4B and Video 2). Importantly, many cells manifested only low-amplitude LCRs, but not APCTs.
Figure 4D contrasts the large variety of Ca2+ signals generated in cells within the area in which the earliest APCT appears (Figure 4A). SAN cells in ROIs 1, 2, 8, 9, 10, 11, and 12 in Figure 4C generated both APCTs and LCRs; cells in ROIs 6, 7, 13, and 14 generated only APCTs, but not LCRs; whereas cells in the ROIs 3 and 4 persistently produced LCRs, but not APCTs.
Furthermore, spontaneous intrinsic frequencies and amplitudes of Ca2+ signals within different ROIs were markedly heterogeneous:
• Cells within ROIs 7, 8, and 14 generated only APCTs of relatively constant amplitude at a frequency of 8.1 Hz, which was also the frequency of APs recorded in the right atrium (not shown).
• Cells within ROIs 5, 6, and 13 generated APCTs at frequencies that were also close to those in ROIs 7, 8, and 14, but exhibited APCT amplitude alternans.
• Cells in ROIs 2, 9, and 10 generated APCTs at low frequencies of 0.9 Hz, 0.6 Hz, and 2.8 Hz, respectively, much lower than that in ROIs 6, 13, 7, 8, and 14.
• Cells within ROI 11 generated APCTs at a low frequency of 4.7 Hz that alternated in amplitude.
• And importantly, the cell in ROI 1 was initially silent, but then began to generate spontaneous APCTs at a frequency of 3.7 Hz during the course of sequential image recording. Thus, cells can generate APCTs both steadily and in bursts.
Furthermore, we also detected similar multimodal Ca2+ signals of variable amplitudes and durations in other SAN preparations (Supplemental Figure 1, Video 3), as well as in SANs from genetically manipulated mice (pCAGGS-GCaMP8) with HCN4-targeted expression of a Ca2+ probe, that is, Ca signals specific to HCN4+ cells (Video 4) and in SANs isolated from human heart (see a representative example in Video 5).
LCRs precede APCTs in some cells within the central SAN
To uncover patterns of the relationships between LCRs and APCTs in cells, we focused on diastolic phases of APCT cycles. In some cells, we observed a temporal relationship between LCRs and APCTs that is characteristic of single SAN cells in isolation (28), that is, diastolic LCRs in a given cell occurred before the APCT firing in that cell. A typical example is shown in Figure 5 and Video 6. LCRs and APCTs measured within chronopix in the ROI in the green box in Figure 5A are observed as small bright spots occurring during the diastolic phase (Figure 5B). Line scan images provide a convenient, instant view of Ca2+ dynamics (i.e., Ca2+ signal vs. time) localized along a selected line within the movie, that is, in the corresponding sequence of TIF files. The sequence of line scans (Figure 5C) clearly illustrates LCRs occurrence before APCTs in the same cell during several cycles. Overlaid plots of LCRs and APCTs show that LCRs onset occurs during an intrinsic, diastolic “entrainment zone” (14) before the onset of APCTs within each cycle (Figure 5D).
Some SAN cells generate only LCRs that precede APCTs in adjacent cells
We observed a novel type of LCR-APCT relationship (Figure 6A) in which some cells within a cell cluster did not generate APCTs, but generated only LCRs (Figure 6B) that preceded APCT firing in 1 or several adjacent cells that did not manifest their own LCRs (Figure 6B and Video 7). Note that the cell with only LCRs generated the earliest signal within the cluster. In this cell, having LCRs only, the LCRs emerged during the diastolic period at 16 ms that preceded AP firing of the adjacent SAN cells at 24 ms (Figure 6B). Note also that the cells adjacent to those cells generating only APCTs did not manifest intrinsic LCRs before generating a synchronous APCT (Video 7). A sequence of line scans recorded from the leading cell firing only LCRs and from an adjacent cell that fired APCTs, but without LCRs (Figure 6C), confirmed that LCRs in the cell without APCTs (Figure 6C) occurred prior to APCTs in the adjacent cell (Figure 6C). Superimposed time series of the chronopix representing LCRs from the cell with LCRs only, and APCTs from an adjacent cell (Figure 6D) also show that LCRs onset in the cell with only LCRs occurred during diastolic phase of adjacent cells. Note that the amplitude of the ensemble LCR Ca2+ signal continued to organize until the APCT in the adjacent cell reached its peak amplitude, after which the LCR ensemble Ca2+ signal decayed to basal level, indicating that the LCRs Ca2+ dynamics in the leading cell, in this case, are not externally reset by AP occurrence in the adjacent neighboring cells. This temporal relationship was observed during all recorded cycles in this preparation, and similar behavior was observed in 7 of 7 SAN preparations.
Variable cycle-to-cycle temporal relationships of Ca2+ signals occurring within some SAN cells
We observed yet another type of heterogeneity of LCRs and APCTs within and among cells within the central SAN: LCRs generated in cells within the ROI in the green box in Figure 7A that did not fire APCTs (Figure 7B and Video 8) preceded APCTs in an adjacent cell that generated both APCTs and LCRs.
LCRs within the yellow ROI (red dots at 6, 12, and 18 ms) were generated during the diastolic phase of the adjacent cell within the red ROI that generated APCT at 18 ms. Line scan images (Figure 7C) created for the 2 cells (Figure 7B) confirmed that the LCRs in the cell generating no APCTs (yellow ROI) occurred before APCTs in the adjacent cell (red ROI). Note also that this pattern occurred during every diastolic phase. Superimposed time series of the LCRs from the cell with LCRs only (Figure 7D) and APCTs from the adjacent cell having both LCRs and APCTs (bold red line) also showed that the onset of majority of LCRs in the cell having only LCRs occurred during the diastolic phases (i.e., “entrainment zones” ) of the adjacent cell firing both APCTs and LCRs.
A panorama of tiled images of a whole-mount SAN preparation from SVC to IVC obtained with a 2.5× objective (Figure 8) revealed the entire HCN4 expression pattern (red) within the intact SAN, extending along the crista terminalis within the central SAN from the SVC and the septum toward the aperture of the IVC. HCN4 immunoreactivity was much stronger within the SAN than in the right atrium (to the left of the crista terminalis), in line with previous reports (29). HCN4 immunolabeling covered more than 80% of the surface of the central SAN between the SVC and IVC (Figure 8). Some HCN4+ cell clusters extended medially toward the septum, and HCN4+ cells were also observed close to the IVC, but none were detected within the crista terminalis.
Typical HCN4+ cells in the central SAN visualized at higher magnification had shapes similar to those in which cell Ca2+ was imaged using Ca2+ indicator Fluo-4 AM (Figure 9). Three types of cell shapes—elongated, spindle, and spider-type—reported previously were observed: Elongated cells had relatively uniform thickness of 2 to 5 μm from end to end and were 100 to 200 μm in length. Spindle cells were shorter and thicker than elongated, with a 7- to 15-μm-thick center thinned to edges. Spider-type cells had thick cell bodies of 7 to 15 μm and lengths of 50 to 80 μm, and remarkably projected multiple thin branches towards neighboring cells (resembling telopodes). We also identified a fourth type of cell, not described previously, having pyramidal-like cell bodies with a 15- to 20-μm base from which thin processes, similar to those of spider-type cells, projected toward neighboring cells (Figure 9). Thirteen spindle cells, 16 spider cells, 9 elongated cells, and 4 pyramidal-like cells were identified within a visual field of 220 μm by 220 μm imaged with a 40× objective from 7 optical slices taken from 3 SAN preparations.
The cytoarchitecture of intercellular connections between HCN4+ cells resembled a mesh-type alignment (30) in which cells within a web have numerous fine cellular branches, creating a high-density mesh of interlacing branches. Surface membranes of neighboring HCN4+ cells within the dense part of the continuous HCN4 meshwork were so close to each other that it was often difficult to discern cells. HCN4 labeled both the cell body and peripheral branches of these cells resembling a neurite-like arborization. Confocal imaging showed that HCN4+ cells were equally distributed between the epicardium and endocardium (Figure 10), indicating that the HCN4+ cell meshwork cytoarchitecture permeated the entire thickness of the SAN.
HCN4 immunolabeling and colabeling with F-actin
Panoramic views of a SAN at low optical magnification (Figure 11A) of colabeled whole-mount SAN preparations with HCN4 and F-actin (using phalloidin, a specific F-actin marker) uncovered intertwining between HCN4-expressing cells (red meshwork) and F-actin containing cells (Figure 11B). (F-actin was not immunolabeled because we did not use antibodies.) Although some SAN cells colabeled positively for both HCN4 and phalloidin, the majority of HCN4+ cells were F-actin−. Note that under our imaging conditions, due to a very few F-actin filaments within HCN4+ cells, F-actin labeling with phalloidin was close to the background fluorescence compared with high F-actin concentration in neighboring SAN cells. Many cells, however, were F-actin+ and HCN4−, clearly exhibiting 2 distinctly labeled networks within SAN tissue: HCN4+/F-actin− and HCN4−/F-actin+ (Figure 11B).
Confocal optical slicing of SAN tissue to create a series of images over sequential increments of 1 μm in depth (Figure 11) revealed the fine structural details of the intertwining of the HCN4+ cell meshwork and F-actin–labeled cell network. Side views of the reconstructed z-stack images in Figure 11C demonstrated penetration of HCN4−/F-actin+ SAN cells into the HCN4+/F-actin− meshwork. The apparent absence of intercellular spaces between the 2 types of cells suggests that surface membranes of each cell are adjacent to each other. HCN4+/F-actin− cells also penetrated the HCN4−/F-actin+ network of cells, often aligning together with HCN4−/F-actin+ cells in a 3D orientation (Figures 11B and 11C).
HCN4 and CX43 coimmunolabeling
Dual HCN4 and CX43 immunolabeling of whole-mount SAN preparations (Figure 12) revealed that the central SAN HCN4+ cell meshwork was largely devoid of CX43 (HCN4+/CX43−) and that striated HCN4− cells were CX43+ (HCN4−/CX43+). Importantly, images taken in a bright-field fluorescent microscope showed that the HCN4+ cell meshwork (Figure 12) and CX43 cell network (green) become intertwined throughout the central SAN from the SVC to IVC. CX43 antibody–labeled membrane proteins observed as green dots (Figure 12) align at the perimeter or at the ends of cells and clearly define the cell borders, revealing points of intercellular communication between CX43-expressing SAN cells. Of note, CX43 immunolabeling was not observed on the cell membranes of HCN4+ cells. The proximity of cells within the HCN4+/CX43− meshwork and the HCN4−/CX43+ network can be observed in optical slices of the SAN obtained from intertwining areas (Figure 12) in which HCN4+ cells come close to CX43+ cells, but do not overlap. The continuity of the CX43+-coupled cell network is therefore not disrupted by HCN4+ cells that do not express CX43 proteins, although 2 cell types come close to each other.
Spatial cytoarchitecture of SAN cells coimmunolabeling with HCN4, CX43, and simultaneous labeling with F-actin (phalloidin)
Triple immunolabeling of whole-mount SAN preparations with HCN4, CX43, and phalloidin antibodies revealed that intertwining between HCN4 meshwork and network of F-actin–expressing cells is not limited to specific parts of the SAN. Spatial cytoarchitecture of SAN within whole-mount preparations reconstructed from 36 tiled confocal images of the area of 1,350 μm by 1,350 μm demonstrates that areas of meshwork/network intertwining are scattered throughout the SAN from the SVC to IVC (Figure 13). Panoramic images showed that CX43 was expressed on F-actin–containing cells, but was not detected on HCN4+ cells. Panoramic images of large areas from intact whole-mount preparations that were optically dissected into series of thin slices show that SAN cytoarchitecture cannot be described as either mosaic or gradient: the loosely coupled HCN4+/F-actin− meshwork is surrounded by and intertwined with an F-actin+/HCN4− network that is strongly electrically coupled through CX43 gap junctions.
CX43 gap junctions connect cells within the F-actin+/HCN4− network in a way such that the electrical signal and corresponding APCTs can propagate in directions demarcated by the CX43 protein. CX43 gap junctions appear to be randomly scattered among cells when they are colored monochromatically. However, when the position of gap junctions is color coded (Figure 14) by depth and plotted within the z-stacks reconstructed from optical slices, alignment of CX43 gap junctions making a pathway (like railroad tracks) for the transmission of electrical or Ca2+ signal from coupled cells is revealed. Importantly, although cells within the HCN4+/F-actin− meshwork do not express CX43 protein, they may be weakly coupled through other types of gap junctions or through other nondetected intercellular coupling structures and processes that promote apparent conduction.
Heterogeneity of calcium signals within SAN recorded on panoramic calcium imaging
Because the HCN4+/F-actin− meshwork and F-actin+/HCN4− network would be likely to have different electrical coupling capacities with respect to the presence of absence CX43, we applied a new method to represent phase heterogeneity of Ca2+ signaling within each pixel. Image pixels across the entire SAN were color-coded relative to the time at which the earliest APCTs occurred (i.e., chronopix). We assigned colors to each chronopix recorded at low optical magnification (2.5× objective) and defined the image as a color-coded chronopix map (Figure 15A).
This pattern of discontinuous APCT occurrence within the panoramic chronopix image of the entire SAN demonstrated in Figure 15A is in antithesis to a concentric continuous spread observed at low resolution in previous studies (7–9) but can be interpreted as “apparent” conduction. Discontinuous APCT occurrence is due to the complicated architecture of the mutually intertwined HCN4+/F-actin− meshwork and F-actin+/HCN4− network. The HCN4+/F-actin− meshwork manifests apparent propagation that is consistent with the idea of weakly entrained oscillators operating out of phase (12,13), whereas the F-actin+/HCN4− network harbors both apparent and true conductional propagation.
We applied Fast Fourier transform (Figure 15C) to detect frequency heterogeneity of low-amplitude Ca2+ signals in pixels within HCN4+/CX43− meshwork near to where the earliest APCTs were detected. Although the power spectrum of the signal within the red ROI in Figure 15B revealed peaks at both 4 Hz and 7.3 Hz, the power spectrum of the signal within black ROI, close to crista terminalis, revealed a peak at 7.3 Hz, but a peak at 4Hz was not present, indicating heterogeneous frequencies of Ca2+ signals within the 2 ROIs. It is important to note that at this low optical magnification, whereas Ca2+ imaging and Fast Fourier transform analysis resolve APCT signals and peaks of cumulative LCR activity, individual LCRs cannot be clearly detected (i.e., they are close to the noise level).
What appears to be true conductional propagation within the SAN can be observed in areas that densely stain for the F-actin+/HCN4− network (Figure 8), that is, F-actin+/HCN4−CX43+ (Figure 13). We defined ROI within a 2D SAN image in which we quantified Ca2+ signals. Analyses of time series of APCTs recorded within different ROIs permitted assessment of heterogeneity of times of APCT occurrences among cells in which the earliest APCT and subsequent APCTs were generated (Figure 16). Ca2+ signal fluorescence intensity in each ROI was normalized to signal amplitude ranging from baseline fluorescence and the APCT maximum peak amplitude. The earliest detectable APCTs emerged within an area close to the SVC (Figure 16A). Subsequent APCTs were recorded along the crista terminalis (red, blue, and orange circles) close to the point at which APCTs occurred in the right atrium. These APCTs appeared with a delay of 9.7 ± 2.3 ms (n = 9) with respect to the time of the earliest AP occurrence. APCTs near the septum (Figure 16) appeared with a delay of 16.1 ± 2.7 ms (n = 9), respectively.
Our results add additional insight into the structural and functional complexity of SAN tissue (1,31) (Central Illustration and Video 9; to download Video 9 in PowerPoint format, see the Supplemental Appendix):
1. A dense meshwork of HCN4+ cells in the central intercaval region differs in cytoarchitecture from a striated F-actin network. The HCN4+/CX43− and F-actin+/CX43+ cells intertwine within a narrow interface zones to create an anatomic unit in which the intertwining areas appear to be the functional interfaces at which electrical and chemical signals are transmitted between the 2 cellular networks. This functional interface extends nearly the entire length of the SAN from the SVC to the IVC.
2. Some areas within the SAN manifest true electrical conduction, whereas others have only apparent conduction. The HCN4+/F-actin− meshwork manifests apparent propagation that is consistent with the idea of weakly entrained oscillators operating out of phase (12,13), whereas the F-actin+/HCN4− network harbors both apparent and true conductional propagation. In the intertwining areas, both real and apparent conduction are likely to occur.
3. Oscillatory Ca2+ signals, including both those occurring locally, that is, spontaneous LCRs, and those induced by APs, that is, APCTs that occur within clusters of cells comprising the HCN4 meshwork, are markedly heterogeneous. In some resident SAN cells, LCRs generated in the diastolic phase appear to ignite APCTs in SAN cells, suggesting that a coupled-oscillator mechanism found in isolated pacemaker cells also operates in some cells embedded within SAN tissue. The LCR–APCT coupling among cells embedded in intact SAN tissue, however, is more complex than observed in single isolated SAN cells.
4. CX43 gap junctions connect cells within the F-actin+/HCN4− network in such a way that electrical signal and corresponding APCTs can propagate in directions demarcated by the CX43 protein. When the position of the gap junctions is color coded (Figure 14) by depth and plotted within the z-stacks reconstructed from optical slices, alignment of CX43 gap junctions making a pathway (like railroad tracks) for the transmission of electrical or Ca2+ signal from coupled cells is revealed.
5. At low magnification, the earliest APCT of each cycle occurred within a small area of the HCN4 meshwork, and subsequent APCT appearance throughout SAN pixels was discontinuous and asynchronous.
Dual immunolabeling of whole-mount SAN preparations indicated that in addition to a the HCN4+/CX43− meshwork, there is a network of F-actin+/CX43+ that intertwines with the HCN4 meshwork in relatively narrow interface zones. The cytoarchitecture of the HCN4 meshwork and F-actin+–expressing cell network, however, could not be categorically defined as a gradient or mosaic network (1). In contrast to HCN4-expressing cells, the striated F-actin–expressing cells do not form a meshwork, but rather a network type of cells connected via Cx43, that is characterized by a repetitive pattern equal across the entire central SAN. Striated CX43-expressing cells do not appear to be inserted between HCN4+ cells and thus do not interrupt continuity of HCN4+ cells within meshwork, and vice versa. 3D reconstruction of optically sliced intertwining areas in whole-mount SAN preparations showed that cells from the HCN4+/F-actin meshwork and HCN4−/F-actin+ network appear to integrate into a single structural unit. In other words, F-actin+ and HCN4+ cells were located so close to each other that an intracellular space between them could not be resolved, even with a 40× objective.
A distinction between cells within the intertwining HCN4 meshwork and F-actin network in the present study is that the majority of cells within the HCN4 meshwork are not striated and are CX43−, in line with the results of prior studies describing properties of the central SAN cells (27,32).
In addition to detecting 3 shapes—elongated, spindle, and spider-type HCN4+ cells described previously (33)—a novel finding of the present study is the identification of SAN cells having a pyramidal-like cell body from which thin branches extended to neighboring cells within the HCN4 meshwork (Figure 9).
Ca2+ signals within SAN tissue
We devised a specific experimental approach to record Ca2+ signals to query whether pacemaker cells imbedded in SAN tissue operate via an LCR-linked AP firing paradigm discovered in isolated cells, but on a higher, more complex, functional scale within the multicellular ensemble. We quantified Ca2+ signals across the SAN at both low and higher magnification to study fine details of LCRs with respect to AP generation manifest as APCTs.
Ca2+ signals within cells embedded within HCN4 meshwork are highly heterogeneous in spatial distribution, amplitude, frequency, and phase
Our results are, in part, consistent with the theory of mutual entrainment of SAN cell oscillators via a democratic process (12,13), but that theory considered that all cells oscillate at the same frequency, but out of phase, and have the same amplitude; whether subthreshold subcellular oscillations differ in amplitudes was not addressed. In other terms, the mutual entrainment theory postulated that all signals must be all or none (i.e., a full-scale AP) while firing at a common rate with the differences in phase. Thus, a second novel finding of our study is that visualization of HCN4+ cells in the central SAN tissue at higher magnification revealed Ca2+ signals that were highly heterogeneous among cells, not only in phase, but also in frequency and amplitude (Figure 4). Importantly, local Ca2+ dynamics of cells within the HCN4 meshwork differed within clusters of cells: some cells generated only LCRs and did not fire APCTs; some only generated APCTs, manifested as synchronized bright flashes throughout the cell; and in some cells, LCRs appeared during the diastolic phase before an APCT occurrence in that cell (Figure 5) resembling those generated by a coupled-oscillator system described in single SAN cells in isolation (19). Importantly, cells can generate APCTs both steadily and in bursts (Figure 4).
A novel type of LCR and APCT interaction was also revealed in cells within the HCN4+/CX43− meshwork, in which APCTs within a cluster of cells appear, with a delay following LCRs occurrence in adjacent cells that generated only LCRs, but not APCTs (Figures 6 and 7). This spectrum of cell behaviors observed within the central SAN suggests that some cells within clusters coordinate their activity via intercellular entrainment, and different clusters of cells via intercluster entrainment. This crucial difference between the democratic theory of mutual entrainment (13) and our results is that we have demonstrated that cell Ca2+oscillations differ, not only in phase, but also in frequency and amplitude. This pattern of Ca2+ signals differing both in phase, frequency, and amplitude among clusters of cells within SAN tissue adds a new dimension of complexity to factors involved in AP generation within SAN cell networks.
Recent numerical studies of the SAN network suggest the importance of Ca2+ clock activation for robust impulse generation and resistance against annihilation (34). Our results show that LCRs in pacemaker cells embedded within SAN tissue have different entrainment patterns with respect to AP generation informed on by APCTs. One pattern is consistent with that previously discovered in single SAN cells operating in isolation in which LCRs precede APCT, confirming that the coupled oscillator theory discovered in isolated single SAN cells (19) is operational in cells within the intact SAN. Here, we show that the coupled oscillator theory, however, operates at a higher and much more complex level within the network of cells comprising the SAN tissue. In other terms, we observed greater freedom and diversity with respect to Ca2+ cycling in cells embedded in SAN tissue than in single isolated SAN cells: many cells within SAN tissue did not fire APCTs, but generated substantial number of LCRs that varied in amplitude, frequency, and rhythmicity (see the next section). Although it cannot be proven from our analyses of Ca2+ signals that LCRs in 1 cell cause an APCT in an adjacent cell, the fact that we observe the same temporal relationship of LCRs preceding APCTs in the adjacent cell over many cycles (Figure 6 and Video 7) suggests that the LCRs occurring in 1 cell can signal to adjacent cells and have a role in APCT generation in these cells. Also of note, with respect to LCRs generation within large cell clusters, the earliest Ca2+ signal within the respective ROIs did not originate from the same cell in every cycle (Figure 4B and Video 2), pointing to complex information processing,
An additional discovery of the present study, that some cells within the HCN4+/CX43− meshwork do not fire APs, is in line with the idea put forth by Opthof et al. (35), that is, only a fraction of cells (∼25%) embedded in SAN tissue participate in generating AP in any given electrical impulse that emanates from the SAN node. We have also recently observed similar non–AP-firing behaviors in a population of enzymatically isolated guinea pig and human SAN cells: these cells have LCRs, but do not fire APs (23,36). A large population of these dormant isolated SAN cells begin to fire spontaneous APs in response to increases in intracellular cAMP (36,37). Such dormant cell behavior within SAN tissue can potentially contribute to pacemaker function via: 1) having a role in entrainment of oscillators that are heterogeneous both in phase and amplitude to self-organize into synchronized signals that underlie rhythmic impulses that emanate from the SAN; and 2) their recruitment to fire APs in response to adrenergic receptor activation or AP silencing in response to cholinergic receptor stimulation.
Immunolabeled cells embedded within the HCN4 meshwork and cells loaded with Fluo-4 within the central SAN had similar shapes. It is noteworthy that regardless their soma shape, all manifested LCRs. The topology of the HCN4 meshwork delineated by immunolabeling may be a requirement for crescendo-like self-organization of heterogeneous local Ca2+ signals observed in our study. In computer research, meshwork properties are categorized by physical and logical topologies (38): physical topology relates to how the various components are placed within a network, for example, ring, bus, mesh, or star networks; whereas logical topology illustrates how data flows among network components. Within a meshwork, the infrastructure nodes (or infrastructure devices) connect directly, nonhierarchically to as many other nodes as possible, and cooperate with 1 another to route signals to the meshwork outputs. In this context, the soma of HCN4-expressing cells appear as nodes within a network with cytoarchitectural features connected via multiple branches, resembling neuronal arborization (39) that are variable and show similar topology throughout the central SAN. Obscure cell borders between HCN4+ cells likely create contiguous pathways from cell to cell within the HCN4 meshwork. This structure is, in fact, similar to interstitial telocyte networks recently discovered in numerous tissues, including heart (40). Of note, the original name for telocytes, in fact, was interstitial cells of Cajal.
Self-organization of local signals among cells in a crescendo-like manner generates an electrical impulse that exits the SAN. This self-organization of local Ca2+ signals among cells within SAN tissue recapitulates the self-organization of local Ca2+ signals that leads to AP firing within individual pacemaker cells studied in isolation (19). This self-similar organization of local Ca2+ signals across scales from cells to tissue can be envisioned as a fractal-like behavior. In other terms, although each LCR is a small, subthreshold signal, the emergent LCR ensemble signals critically contribute to generation of spontaneous APs within and among SAN cells.
Similarities of microscale signaling within SAN and that within other tissues
The fine details of the novel microscale signaling paradigm within SAN tissue discovered here is reminiscent of complex information processing among clusters of neurons that create spatiotemporal synchronization of signals that drive neuronal network functions. Examples of this type of behavior have been observed within the autonomic neural–visceral axis via its modulation of rhythms of peripheral organ function.
A characteristic of HCN4 channels, anomalous rectification, keeps the diastolic membrane potential positive to the potassium equilibrium potential. Anomalous rectification is crucial for the functions of coupled oscillator systems, not only for SAN cells in isolation (17,41) and in pacemaker cells embedded in SAN tissue, but also for similar systems operative in brain (42). Cells that exhibit coupled oscillator behavior within the central SAN, like many brain neurons, express HCN4.
One example occurs within the brain stem: 1 type of neuron, cells of preBötzinger complex (preBötC), create spontaneous signals that activate adjacent cells (e.g., Bötzinger cells) to generate impulses that travel within nerve from the brain to the diaphragm to regulate automatic breathing (43). It has been hypothesized that only a small fraction of preBötC cells within the network is required to initiate each excitatory cycle and that the initiating preBötC cells differ from cycle to cycle (44). A major premise of this hypothesis is that spontaneous activity is initiated within a few preBötC cells and induces activity in other preBötC cells, and this excitation percolates throughout the network, building to a crescendo that initiates inspiration signals. We may envision that networks of pacemaker cells within SAN tissue generate local Ca2+ signals that are similar to those of preBötC in conjunction with signals from the heart’s little brain, that is, the network of autonomic ganglia embedded within the atrial epicardium (45). Another example is the network of interstitial cells of Cajal that generate spontaneous Ca2+ signals to activate adjacent smooth muscle cells to effect gut motility (46). Our discovery of heterogeneous Ca2+ signals in cells comprising cardiac SAN are also reminiscent of those observed in studies in which Ca2+ was imaged in uterine smooth muscle (47).
Asynchronous and discontinuous APCT occurrence across SAN pixels (chronopix)
Another novel finding of our study was that in viewing the entire set of SAN pixels (at a low optical magnification of 2.5×), following the earliest appearing APCTs near the SVC, subsequent APCTs appeared asynchronously and discontinuously across the pixels of other parts of the SAN (Figures 3 and 15). This pattern of asynchronous and discontinuous APCT occurrence at remote cites following their earliest appearance is consistent with the idea of self-organized synchronization of a loosely coupled network (48) of cell oscillators within different parts of SAN tissue (also know as pacemaker cell mutual entrainment, suggested by Jalife ). This idea of self-organized synchronization within the SAN pacemaker cell network differs from interpretations of many previous studies, that is, that pacemaker cells within an initiation site generate an impulse that spreads in a “concentric” pattern, overdriving other SAN cells that oscillate at slower spontaneous frequencies (6–9).
Future directions of studies of SAN structure/function
Further intense study is required to understand how HCN4+ cells devoid CX43 signal to each other within the HCN4 meshwork and how these cells signal to CX43+/F-actin+–expressing cells. The lack of structural continuity defined by cell-to-cell connections via 1 type, CX43, of gap junctions suggests a degree of autonomy of the HCN4+ cells from electrical behavior throughout the central SAN. Moreover, a discontinuity of CX43 expression suggests that complex Ca2+ dynamics within some HCN4+ cells may not be externally reset by AP occurrence in neighboring cells. The failure of APCTs to reappear within the area near the initiation site during a given AP cycle may be explained on the basis of lack of CX43 in the HCN4+ SAN cells that prompted the initiation of the impulse.
Whether another type of connexin besides CX43 may connect HCN4+ cells is a moot issue awaiting specific antibodies to other connexins, for example, Cx45 and Cx30.2, thought to be present within the center of the SAN (27). Even if cells within the HCN4 meshwork are connected via an unidentified type of gap junction protein, lack of structural continuity between the CX43− meshwork to CX43+ network would remain. The close proximity of 2 different cell types creates an interface of possible electrical or chemical communication between 2 networks (Figures 11 and 14). Transmission of Ca2+ and electrical signals from the HCN4+ meshwork to the F-actin+ network would be expected to occur in these intertwining areas.
Transition of the signal between intertwined cells from HCN4+/F-actin− cells to HCN4−/F-actin+ cells that express connexin 43 would be expected to occur with a loss of signal amplitude and duration as reported for electrical synapses. Indeed, sharp transitions in AP waveforms over small distances of 100 μm that occur within the mouse central SAN (49,50) are similar to differences in shapes of APs that occur during the transfer of electrical signal from pre- to post-synaptic neuron through the electrical synapses (51).
Although the field of SAN biology has been dominated by the idea that electrotonic or electrophasic impulses mediate communication among SAN cells, there is a plethora of evidence that other types of signaling occur between cells. One is mechanical (52). Other types of cell-to-cell communications, though not yet demonstrated in SAN tissue, include: ephaptic (53); photons emitted from intracellular chromophores (54,55); cellular vibrations transmitted or reflected as electromagnetic waves (56); a gas, for example, nitric oxide, hydrogen sulfide, or carbon monoxide (57); a lipid signal generated acutely by arachidonic acid degradation; as in retrograde–grade endogenous cannabinoid signaling in the brain (58); and the secretion of Ca2+ buffers into interstitial spaces between cells (59–63). Because any of these types of cell-to-cell communications may occur among cells within SAN tissue, the foremost frontier for the field of cardiac pacemaker biology is to understand how HCN4+ cells communicate with each other within the HCN4 meshwork and how they communicate signals to the CX43 cells.
One limitation of our study is that we imaged Ca2+ signals with a sCMOS camera in 2 dimensions in a given focal imaging plane, whereas real signals emerge over time within a complex 3-dimensional meshwork of HCN4+ SAN cells. Thus, our 2D imaging provides only partial information about 3-dimension interactions among HCN4 cells that actually occur. As a result, our imaging may miss the cells in which the earliest APCT appears because they may be located on the SAN surface or deeply within the tissue and evade the focal Imaging plane. Based upon our optical 3D sectioning of HCN4 meshwork, however, functional clusters and their self-organized signals are expected to be present throughout the bulk of the SAN tissue, and therefore techniques that can penetrate the tissue are required to capture intercellular interactions throughout all part of the SAN. Another limitation of our study is the absence of an optical map of membrane potential recorded simultaneously with the Ca2+ signal chronopix map. Membrane potential recordings from cells that do not generate APCTs would be also helpful to provide additional information relating to how HCN4 cells with subthreshold activity that do not generate APCTs interact with cells that do generate APCTs.
Our results demonstrate that signals generated by the SAN cell oscillators are more heterogeneous than previously formulated, including differences in amplitudes, frequencies, and discontinuities. They indicate the need for modification of the previously described elegant operational paradigm depicting resident SAN cells as a democracy involving mutual entrainment that effects a common single rate of full-amplitude APs with different local phases. Our results demonstrate novel complexity of signals of cells embedded within SAN tissue at a lower (deeper) level of events including both subcellular and whole-cell origin (i.e., within and among SAN cells), resembling neuronal networks (42-45,64). Synchronized macroscopic signals within the SAN, including full-scale APs, emerge from heterogeneous microscopic subthreshold Ca2+ signals. This signal complexity underlies the robust impulse generated by the SAN and appears as a democracy at the higher level of synchronized AP generation described at a lower resolution in prior electrophysiological studies or imaging of Ca2+ and electrical signals.
COMPETENCY IN MEDICAL KNOWLEDGE: Based on our results we propose a new view of how the SAN operates.
TRANSLATIONAL OUTLOOK: Our hypothesis will lead to future studies on the specific new mechanisms discovered, that is, on the emergence of rhythmic impulses generated by the SAN from heterogeneous local signaling: mechanisms of cell-to-cell interactions and of signal synchronization within and among cells. This paradigm shift and related future basic studies will bring new insights into therapies for cardiac diseases specifically in the realm of cardiac arrhythmias and sick sinus syndrome. These novel treatments will target the interactions of cells residing within the SAN rather than mechanisms of single pacemaker cells per se.
The authors thank Tracy Oppel and Loretta Lakatta for editorial assistance.
This work was supported by the Intramural Research Program of the National Institutes of Health, National Institute on Aging. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- action potential
- action potential–induced Ca2+ transient
- connexin 43
- hyperpolarization-activated cyclic nucleotide-gated channel 4
- inferior vena cava
- Local Ca2+ release
- Na+/Ca2+ exchanger
- phosphate-buffered saline
- preBötzinger complex
- region of interest
- sinoatrial node
- scientific complementary metal-oxide semiconductor
- superior vena cava
- Received June 1, 2020.
- Revision received June 19, 2020.
- Accepted June 24, 2020.
- Zhang H.,
- Holden A.V.,
- Boyett M.R.
- Inokaitis H.,
- Pauziene N.,
- Rysevaite-Kyguoliene K.,
- Pauza D.H.
- Bleeker W.K.,
- Mackaay A.J.,
- Masson-Pevet M.,
- Bouman L.N.,
- Becker A.E.
- Efimov I.R.,
- Nikolski V.P.,
- Salama G.
- Li N.,
- Hansen B.J.,
- Csepe T.A.,
- et al.
- Sano T.,
- Sawanobori T.,
- Adaniya H.
- Michaels D.C.,
- Matyas E.P.,
- Jalife J.
- Anumonwo J.M.,
- Delmar M.,
- Vinet A.,
- Michaels D.C.,
- Jalife J.
- Verheijck E.E.,
- Wilders R.,
- Joyner R.W.,
- et al.
- DiFrancesco D.
- Maltsev V.A.,
- Lakatta E.G.
- Lakatta E.G.,
- Maltsev V.A.,
- Vinogradova T.M.
- Lyashkov A.E.,
- Behar J.,
- Lakatta E.G.,
- Yaniv Y.,
- Maltsev V.A.
- Vinogradova T.M.,
- Brochet D.X.,
- Sirenko S.,
- Li Y.,
- Spurgeon H.,
- Lakatta E.G.
- Tsutsui K.,
- Monfredi O.,
- Sirenko-Tagirova S.G.,
- et al.
- Vinogradova T.M.,
- Zhou Y.Y.,
- Bogdanov K.Y.,
- et al.
- Vinogradova T.M.,
- Zhou Y.Y.,
- Maltsev V.,
- Lyashkov A.,
- Stern M.,
- Lakatta E.G.
- Oren R.V.,
- Clancy C.E.
- Verheijck E.E.,
- Wessels A.,
- van Ginneken A.C.,
- et al.
- Li K.,
- Chu Z.,
- Huang X.
- Kim M.S.,
- Maltsev A.V.,
- Monfredi O.,
- et al.
- Tsutsui K.,
- Kim M.S.,
- Wirth A.N.,
- et al.
- ↵(2002) Networking Complete (Sybex, San Francisco, CA), 3rd edition.
- Menon S.,
- Gupton S.
- Kondo A.,
- Kaestner K.H.
- Kopell N.,
- LeMasson G.
- Kam K.,
- Worrell J.W.,
- Ventalon C.,
- Emiliani V.,
- Feldman J.L.
- Lee M.Y.,
- Ha S.E.,
- Park C.,
- et al.
- Del Negro C.A.,
- Funk G.D.,
- Feldman J.L.
- Boyett M.R.,
- Honjo H.,
- Yamamoto M.,
- Nikmaram M.R.,
- Niwa R.,
- Kodama I.
- Michalikova M.,
- Remme M.W.H.,
- Schmitz D.,
- Schreiber S.,
- Kempter R.
- Albrecht-Buehler G.
- Albrecht-Buehler G.
- Facchin F.,
- Canaider S.,
- Tassinari R.,
- et al.