Problems with extracellular recording of electrical activity in gastrointestinal muscle

Abstract

Motility patterns of the gastrointestinal tract are important for efficient processing of nutrients and waste. Peristalsis and segmentation are based on rhythmic electrical slow waves that generate the phasic contractions fundamental to gastrointestinal motility. Slow waves are generated and propagated actively by interstitial cells of Cajal (ICC), and these events conduct to smooth muscle cells to elicit excitation–contraction coupling. Extracellular electrical recording has been utilized to characterize slow-wave generation and propagation and abnormalities that might be responsible for gastrointestinal motility disorders. Electrode array recording and digital processing are being used to generate data for models of electrical propagation in normal and pathophysiological conditions. Here, we discuss techniques of extracellular recording as applied to gastrointestinal organs and how mechanical artefacts might contaminate these recordings and confound their interpretation. Without rigorous controls for movement, current interpretations of extracellular recordings might ascribe inaccurate behaviours and electrical anomalies to ICC networks and gastrointestinal muscles, bringing into question the findings and validity of models of gastrointestinal electrophysiology developed from these recordings.

During the early part of the twentieth century, physiologists began utilizing techniques developed for cardiac and nerve recording to monitor action currents in the gastrointestinal tract. Two early attempts by Alvarez and Mahoney¹ and Richter² utilized different electrode designs and amplification techniques to record from the stomach. Recordings with considerably different waveforms were obtained. At the time of these experiments, muscular movements could be seen but the electrical and biochemical mechanisms responsible for smooth muscle contractions were unknown. Thus, the investigators were without prejudice regarding the source and cause of the biopotentials (electrical signals recorded from living tissues) they recorded. Richter’s electrodes contained a suction chamber to hold the electrodes to the gastric surface. Nevertheless, the multiphasic waves he observed were attributed, in part, to the differential contractions of the circular and longitudinal muscle layers².

Since those early recordings, and following developments in cardiac electrophysiology³, biopotential recording from gastrointestinal organs has advanced in sophistication with the advent of multichannel recording and digital data processing. Extracellular recording techniques — including active and passive electrodes, large-scale electrode arrays, multichannel analogue-to-digital signal conversion, digital filtering and sophisticated signal processing software, and computerized modelling of propagated events — have been applied to understand what goes wrong with the motor patterns in gastrointestinal organs in disease states⁴. Spatiotemporal mapping of biopotentials in patients with motility disorders has suggested fundamental problems in the propagation of electrical activity, and anomalies, such as arrhythmias, re-entry phenomena and ectopic pacemaking, have been reported^4–8. Recognizing and naming abnormalities in gastro intestinal organs similar to abnormalities in cardiac tissues has been considered to be an important advance in understanding the pathophysiology of motility disorders. However, a major problem plagues these efforts, reminiscent of one of the basic questions considered by Richter² — what part of the biopotentials recorded from visceral smooth muscles is caused by the movements of these muscles? An active debate on this question has reverberated through the literature for the past 5 years, without a consensus developing^9–13. The aim of this article is to discuss the problems with the extracellular recording techniques that are being used to evaluate frequency and propagation disorders in gastroenterology and to discuss the need for development of more reliable diagnostic methods.

Movements generate biopotentials

The lengths of muscle fibres decrease when muscles contract in gastrointestinal organs, resulting in movements of the tissue relative to surface electrodes. This movement might result in voltage changes that can be equal to the amplitude of the physiological signals, as in electromyograms from skeletal muscle tissues¹⁴. Another source of artefacts from movement is related to electrical potential changes between layers of tissue¹⁴. In gastrointestinal organs, this complicating factor is likely to be important because there are several layers of tissue in which electrical potential differences exist, and movements of muscles probably introduce dynamic changes in electrical potentials between these layers. Additionally, circular and longitudinal muscle layers are present in gastrointestinal muscles and contractions of these muscles at 90° angles to one another (shearing movements) might cause complex spatiotemporal changes in tissue impedance. For example, fixed markers placed on the surface of the stomach showed that elliptical movements occurred during each contraction cycle¹⁵. This finding is clearly not equivalent to the propagation of slow waves that move in an orthogonal plane within sheets of gastric muscle¹⁶. As tissues move under fixed electrodes (as in electrode array recordings), or even when electrodes are adhered to the tissues, the interface between the electrodes and the tissue might change and affect the impedance in series with the recording electrodes. Many electrode designs have been used for extracellular electrical recording in gastrointestinal muscles but, unfortunately, incorporation of new electrode designs has not been accompanied by rigorous tests of the effects of muscle movements, which would be necessary to optimize electrode design. To further understand the potential problems encountered when recording electrical activity from gastrointestinal muscles with extracellular techniques, it is necessary to first consider tissue morphology and the basic mechanisms behind the generation and propagation of electrical slow waves.

Gastrointestinal muscle physiology

Many regions of the gastrointestinal tract display autorhythmicity, a periodic depolarization–repolarization cycle that underlies the phasic contractile activity fundamental to peristalsis and segmentation¹⁷. The electrophysio logical events that determine the frequency of phasic contractions are electrical slow waves, which are periodic oscillations in transmembrane potential that vary in amplitude and frequency from organ-to-organ and species-to-species¹⁸. In general, slow waves are 10–60 mV in amplitude and one to many seconds in duration¹⁸. Slow waves depolarize gastrointestinal smooth muscle cells (SMCs) into the activation range of L-Type Ca²⁺ channels¹⁹. In some regions of the gastrointestinal tract (for example, small intestine and colon), slow waves initiate Ca²⁺ action potentials in SMCs, but in the stomach slow waves do not activate SMC action potentials except in the terminal antrum and pyloric sphincter¹⁷. Thus, excitation–contraction coupling is linked to slow-wave activity either through initiation of SMC action potentials or by sustained Ca²⁺ entry during the slow-wave depolarization²⁰. Slow waves are actively propagated in gastrointestinal muscle tissues, enabling recruitment of thousands of SMCs to contract together or in sequence to generate segmental and peristaltic contractions.

Muscles of the gastrointestinal tract contain at least three types of cells (SMCs, interstitial cells of Cajal (ICC) and platelet-derived growth factor receptor (PDGFR)α⁺ cells that are electrically coupled, forming a unit known as the SIP syncytium²¹. Among the SIP cells, ICC serve as pacemaker cells and express a specialized apparatus that includes Ca²⁺-activated chloride channels (CaCC; encoded by ANO1), T-Type Ca²⁺ channels and intracellular Ca²⁺ stores (endoplasmic reticulum-containing ryanodine and inositol trisphosphate receptors)¹⁸. Release of Ca²⁺ from the endoplasmic reticulum activates CaCC, producing inward currents that are the underlying pacemaker mechanism initiating slow waves²¹. ICC are electrically coupled to one another and can actively regenerate and propagate slow waves for many centimetres¹⁸. SMCs lack the necessary ion channels to generate or regenerate slow waves. Thus, slow waves conduct passively from ICC into SMCs, depolarize SMCs and initiate excitation– contraction coupling¹⁸. PDGFRα⁺ cells are also electrically coupled to ICC and SMCs and serve to transduce purinergic inhibitory neurotransmission, thereby regulating the excitability of SMCs²¹. No evidence suggests that PDGFRα⁺ cells can either generate or regenerate slow waves. Thus, ICC are the active cells in the generation and propagation of slow waves and the other cells of the SIP syncytium constitute a formidable capacitance (current sink) for the currents generated by ICC.

Slow waves are omnipresent in gastrointestinal organs and motor activity is controlled, in part, by modulation of the frequency, amplitude and duration of slow waves¹⁷. Coupling between slow waves and contractions is therefore of great importance in understanding gastrointestinal motility and for developing concepts about what might lead to motility disorders. Alterations in ICC density and/or function could be a central factor in several gastrointestinal motility disorders²², however, a cause-and-effect relationship has not been firmly established. Many years of research have been dedicated to understanding the mechanisms of slow-wave generation and propagation, but understanding the effect of changes in ICC on the motor behaviours of gastrointestinal organs requires techniques to record and model the patterns of slow-wave generation and propagation.

The statements here are not meant to suggest that SMCs do not possess active currents. Depolarization of SMCs activates a variety of voltage-dependent conductances and, in addition to L-Type Ca²⁺ channels, SMCs express a variety of K⁺ channels that are also regulated by membrane potential²³. Activation or suppression of voltage-dependent, receptor-operated, mechanosensitive and metabolically regulated conductances, all of which are expressed and functional in SMCs, can affect the space constants of SMCs (a value indicating how far potentials can spread passively within the syncytium), reduce or enhance the amplitude and duration of slow waves and modulate excitation–contraction coupling. Thus, effective modelling of gastrointestinal electrophysiology and understanding how electrophysiological events control motor activity requires knowledge of the diversity of ionic conductances, the properties of ion channels in cells of the SIP syncytium and how the channels respond to the dynamic conditions that occur during fasting and digestion.

Electrophysiological recordings

Intracellular recording techniques for gastrointestinal muscles.

A variety of electrophysiological techniques have been developed to monitor slow waves and action potentials in gastrointestinal muscles (see FIG. 1). The most reliable method for measuring electrical activity is to impale cells with microelectrodes and record transmembrane potentials directly via a high-input impedance amplifier (FIG. 1a). Investigators have used this technique since Edith Bülbring made recordings from guinea pig taenia coli in the early 1950s²⁴. Transmembrane potential recordings monitor the electrophysiological behaviour of the SIP syncytium. Typically, SMCs are impaled because these cells are the most abundant, but a few skilful and intrepid investigators have successfully impaled other cells of the SIP syncytium^25–29. These recordings have revealed unique and important properties of slow waves: the velocity of the initial depolarization of slow waves (upstroke phase) recorded directly from ICC is faster than the upstrokes of slow waves recorded from SMCs^27,29; and the amplitude of slow waves recorded from ICC is greater than in SMCs^25,27,29. Both observations occur because ICC generate and regenerate slow waves, and these events conduct passively into and through the smooth muscle portion of the SIP syncytium³⁰. Slow waves decay electronically in SMCs³¹, as voltage transients decay passively in cables. Slow waves recorded from ICC display the quantitative electrical parameters against which electrical recording methods must be tested, reconciled and standardized.

Figure 1 |. Schematic representations of electrophysiological recording techniques.

Each example shows an amplifier with an incorporated display to show events that ideally would be recorded. a | Intracellular recording using sharp electrodes to impale single cells. Transmembrane potentials are recorded with this technique. b | Pressure or suction electrode recordings. Large diameter tip electrodes press on muscles or adhere to muscles via suction. Membrane potentials are recorded between the depolarized region and the normally polarized region outside the area affected by pressure or suction. c | Sucrose gap technique in which partitions are created by flowing sucrose solution or Vaseline to isolate different regions of the muscle. The peripheral chambers are depolarized by elevated external K⁺ (labelled K⁺). d | Monopolar extracellular recording using a metal electrode placed on the surface of a muscle strip or whole organ. Ideally these recordings should be approximated by the first-time derivatives of the transmembrane potential changes in the interstitial cells of Cajal. e | Extracellular array electrode recording using metal electrodes arranged in a matrix and connected to a multichannel amplifier. The array is placed on the surface of a muscle sheet or organ. In this example monopolar recordings are made in reference to a bath electrode. Display terminal shows idealized recordings from electrodes 1, 3, 5 and 7.

Early on, impalements of SMCs were difficult and hard to maintain for periods long enough to perform experimental procedures. Improved electrode fabrication has alleviated many of these difficulties but, because of the technical challenges, many early studies used modified techniques to measure electrical events in smooth muscle tissues. Such methods included the use of pressure electrodes and suction electrodes (FIG. 1b), which create a region of reduced membrane resistance caused by the pressure of the electrode tip on the muscle or suction of part of the syncytium into the electrode tip. This approach enables recording of pseudo-transmembrane potentials between the region of depolarized cells and the region of muscle around the pressure or suction electrode tip that is normally polarized^32,33. Another method used extensively was the sucrose gap technique, in which a portion of the syncytium is made permeable by depolarization and pseudo-transmembrane potentials are measured between the depolarized region and a region of normal polarization at some distance along the SIP syncytium³⁴. Signals proportional to transmembrane potentials can be recorded with these techniques.

It has been suggested that electrical slow waves recorded from strips of smooth muscle in vitro differ from the electrical events that occur in vivo due to loss of some as yet unknown factor, damage to muscle strips by dissection or factors induced by in vitro conditions and/or physiological solutions^10,35. However, although conditions in vitro could possibly alter the electrophysiology of gastrointestinal muscles, the criticisms of in vitro recordings are not supported by empirical comparisons between in vivo and in vitro recordings. In the few cases in which investigators have impaled cells in gastrointestinal organs with microelectrodes in vivo, slow waves of similar waveform and frequency to those recorded in vitro have been recorded^36,37.

Extracellular recording techniques for gastrointestinal muscles.

Extracellular electrodes of various designs and materials have been used, including wires piercing the muscles¹, metal electrodes sewn to the surfaces of organs^38,39, metal electrodes held to the surface by suction² and arrays of metal electrodes (or printed electrode arrays) laid upon or near the surface of organs^5–8,40. Signals are typically recorded via AC-coupled amplifiers to reduce baseline drift, and a variety of filtering techniques have been used to optimize signal-to-noise ratio (discussed later).

Both bipolar (recordings made via two closely spaced electrodes) and monopolar (recordings made via one electrode placed on the muscular region of interest and an indifferent electrode placed at some distance to the recording site; FIG. 1d) recordings have been utilized. Extracellular recording of biopotentials in gastrointestinal muscles is a simple procedure and can be accomplished with metal electrodes and a commercial AC amplifier. Some studies have utilized more sophisticated amplifiers, such as the ActiveTwo System (BioSemi, Netherlands), which has the advantages of digital data collection, broad flexibility in sampling rates, increased ranges of filtering and the ability to record simultaneously from many electrodes⁵ (FIG. 1e).

Examination of the recordings in the literature, from the time of Alvarez^1,2 in the 1920s to more modern recordings^5,6,8,41, raises questions about extracellular recording because of the wide variety of waveforms that have been attributed to slow waves in gastrointestinal muscles. Although researchers have speculated about what the ‘ideal’ waveform of slow waves should look like when recording with extracellular electrodes⁴², published recordings display substantial variability. Variations in waveform occur from moment-to-moment in recordings from the same electrode and from electrode-to-electrode along the axes of slow-wave propagation. TABLE 1 lists figures from some of the references quoted in this Review so the reader can appreciate the broad diversity of waveforms considered to be slow waves recorded from the stomach and small intestine. Variability in waveforms comes from a variety of sources, such as differences in recording techniques (monopolar versus bipolar), differences in filtering and data processing, differences in the electrical activity in different regions of the gut or in different species, differences in the state of excitability and differences in the direction of propagation and/or mechanical artefacts. However, intracellular recording shows that the basic waveforms of slow waves are similar, at least in stomach and small intestine, and comprise an upstroke depolarization of varying velocities and amplitudes, a plateau phase of varying amplitudes and durations, and repolarization¹⁸. Thus, one might expect similarities in the waveforms of the field potentials recorded from these organs, and certainly from those recorded at the same location using the same conditions during the same recording session.

Table 1 |.

Examples of possible movement artefacts in published extracellular recordings

Organ (species)	Type of recording	Waveform traits suggestive of movement artifacts	Ref.
Gastric corpus (human)	Array electrodes	Waveforms change from event to event at the same electrode, and at different electrodes along same axis	5
Gastric corpus (human)	Array electrodes	Waveforms variable at each electrode	6
Small intestine (species not given)	Array electrodes	Waveforms change from event to event at the same electrode	7
Small intestine (dog)	Unipolar and bipolar platinum	Waveforms change from event to event at the same electrode, and at different electrodes along same axis	39
Stomach (pig)	Array electrodes	Waveforms change from event to event at the same electrode, and at different electrodes along same axis	41
Stomach (dog)	Array electrodes	• Multiphasic events • Waveforms change from event to event at different electrodes along same axis	8
Stomach (human)	Array electrodes	• Time course of slow wave too long to correspond to upstroke • Waveforms change from event to event at different electrodes along same axis	64
Small intestine (rat)	Array electrodes	Waveforms change from event to event at the same electrode, and at different electrodes along same axis	67
Corpus (pig)	Array electrodes	• Time course of slow wave far too long to correspond to upstroke • Waveforms change from event to event at different electrodes along same axis	68

Extracellular field potentials are proportional to membrane ionic current (i_m). In terms of intracellular voltage (V), $i_m=-C_m\frac{dV}{dt}$ where C_m is the membrane capacitance and t is time in seconds. Thus, extracellular voltage transients should be proportional to the first-time derivative of the transmembrane voltage waveform, and the ‘ideal’ waveform for extracellular slow waves will be approximated by the first derivative of the slow waves recorded from the cells generating the transmembrane currents responsible for slow waves, the ICC.

Analogous to heart, skeletal muscles and nerves, transmembrane currents flowing across the membranes of active cells of the SIP syncytium should generate field potentials that can be monitored by extracellular electrodes, as in electrocardiogram, electromyogram and electroe ncephalogram recordings. Several structural and biophysical features of slow waves and gastrointestinal muscles, however, confound measurements of field potentials. The first problem is that the velocity of slow-wave depolarization (dV/dt) is about two orders of magnitude less than the upstroke velocities of action potentials generated by cardiac and skeletal muscles⁴³, which might limit resolution of authentic slow-wave activity. This difference is due to the slower kinetics of activation of the T-type Ca²⁺ channels and CaCC in ICC, which are responsible for the upstroke depolarizations of slow waves²¹, compared with the fast kinetics of Na⁺ channels during action potentials in nerves and cardiac and skeletal muscles⁴³. A second problem is related to the driving force for inward currents during slow waves. Nerve and muscle action potentials overshoot 0 mV, as membrane potential trends toward the equilibrium potential for Na⁺ ions (~60 mV) during the upstroke phase of the action potential⁴³. The equilibrium potential for chloride ions in ICC is negative to 0 mV, so slow waves depolarize only to about −10 mV (REF. 21). Thus, the driving force for the inward current in ICC is about half that in nerves and cardiac or skeletal muscles. A third problem stems from the low density of cells that regenerate slow waves in the SIP syncytium (FIG. 2). In cardiac muscles, skeletal muscles and nerves, the majority of cells are active and can regenerate action potentials, providing robust field potentials. SMCs were once thought to actively regenerate slow waves⁴⁴, but now we know that SMCs lack the conductances necessary to regenerate these events¹⁸. In fact, the enormous capacitance that SMCs and PDGFRα⁺ cells contribute to the SIP syncytium actually represents a sink for the currents generated by ICC, and a portion of the charge entering ICC during the slow-wave upstroke is shunted to SMCs and PDGFRα⁺ cells to depolarize this portion of the SIP syncytium. In gastrointestinal muscles, ICC represent only a fraction of the cells (<10%) within the tunica muscularis⁴⁵. If <10% of cells actively regenerate slow waves, in contrast to thick muscles of the heart, skeletal muscles and large nerves that actively propagate action potentials, then the field potentials generated by gastrointestinal muscles would be accordingly smaller in amplitude. All of these issues make it difficult to resolve field potentials in gastrointestinal muscles above the noise inherent to electrophysiological recordings.

Figure 2 |. Morphological considerations in gastrointestinal muscles.

Confocal immunohisto-chemical images show the sparse distribution of interstitial cells of Cajal (ICC) within the circular muscle layer of the human gastric antrum. a | c-Kit⁺ ICC (arrows). b | Smooth muscle myosin heavy chain (SMMHC)⁺ circular smooth muscle (cm). ICC were located within and around circular muscle bundles and at the level of the myenteric plexus (my) that demarcates the circular muscle from the longitudinal muscle (lm). c | Merged image of part a and part b. From work on large laboratory animals, it seems that pacemaker (regenerative) ICC are distributed around muscle bundles and are not limited to the myenteric region. However, in human muscles it is unclear to what extent ICC possess the ability to generate and regenerate slow waves. In this image a few rounded c-Kit⁺ mast cells (*) were also observed in the muscle layers. Scale bar in part a = 20 μm and is representative of all panels. Images courtesy of Y. Bayguinov, University of Nevada.

Another factor that might complicate the interpretation of electrical signals from gastrointestinal muscles is the contribution of field potentials from other excitable cells within the same tissues. For example, SMCs also express voltage-dependent conductances that might increase the complexity of field potentials²³. Enteric neurons form a dense reticulated network within gastrointestinal muscles, and some fire tonically with action potentials increasing during gastrointestinal reflexes⁴⁶. Nerve action potentials are due to fast Na⁺ and K⁺ currents and would be expected to produce substantial background activity superimposed upon slow-wave activity. However, no studies seem to have noted nerve action potential activity during extracellular recordings of slow-wave activity from gastrointestinal muscles, suggesting that the summed field potentials from nerve action potentials are too small in magnitude, or the recording conditions used are inadequate to resolve these events. Ca²⁺ action potentials generated by SMCs in the colon and small intestine should be better resolved than slow waves owing to the transmembrane currents that generate these events, yet unequivocal evidence demonstrating the extracellular recording of authentic smooth muscle action potentials in the small intestine and colon has not been provided. For gastric muscles, action potentials are observed only in the most terminal antrum and pyloric canal⁴⁷. In the corpus and antrum, Ca²⁺ currents leading to excitation–contraction coupling during slow waves are of low amplitude and occur throughout the plateau phase^20,21. Such transmembrane currents would contribute negligibly to field potentials.

What then generates the biopotentials recorded from gastrointestinal muscles with extracellular electrodes? We sought to record electrical activity of gastric muscles from various mouse genetic models while monitoring gastric contractile patterns by video imaging. Biopotentials were readily recorded from the gastric surface, but they occurred at lower frequencies than commonly observed during intracellular recordings¹⁵. When contractile activity was partially blocked downstream of slow waves, the multiphasic biopotentials often recorded with extracellular electrodes became biphasic, the waveform claimed to be ‘ideal’⁴¹. When contractile activity was strongly blocked (that is, movements were suppressed), biopotentials were abolished but slow waves still persisted in these muscles, as shown by recordings made with intracellular microelectrodes¹⁵. The same observations were made with gastric muscles of humans (FIG. 3), suggesting that mechanical artefacts contaminate the signals recorded with extracellular electrodes and the only way to objectively validate the fidelity of events recorded in this manner is to clearly demonstrate the persistence and characteristics of the biopotentials in the absence of movement.

Figure 3 |. Stabilization of movement abolishes biopotentials in human gastric muscles despite the persistence of electrical slow waves.

Simultaneous intracellular and extracellular electrical recordings and isometric force measurements were made from human gastric antral muscle strips⁷². a | Intracellular electrical activity recorded via impalement of a smooth muscle cell (top trace) and extracellular electrical activity (biopotentials) recorded by a monopolar electrode placed on the muscle within 3 mm of the intracellular electrode (middle trace). Isometric force was also recorded by attaching a force transducer at the end of the muscle strip to measure contractions (bottom trace). The intracellular electrode recorded slow waves, which were followed by phasic contractions as described previously⁷². b | Nifedipine (3 μM) failed to block contractions (not shown), but contractions were blocked by exposure to nominally Ca²⁺-free solutions for at least 20 minutes. Reducing contractions reduced the biopotentials to within the noise level while electrical slow waves persisted in recordings made via the intracellular electrode.

Our examination of the literature showed that few studies using extracellular recordings on gastrointestinal organs and tissues have included careful experiments to control for the effects of movement, a known artefact in extracellular recordings from muscle tissues¹⁴. In some studies in which contractile activity was reduced, the amplitude of the biopotentials also decreased^48,49. Another study claimed that little difference was noted in extracellular recordings upon suppression of movement⁴². However, a closer investigation of the evidence presented indicates that movements were not fully stabilized as claimed. The method used to visually analyse movements examined distortion in the longitudinal axis, but the technique did not quantify circumferential movements. Thus, the full extent of wall movements was probably underestimated. Movements before and after the calcium channel antagonist nifedipine was introduced might also have been misrepresented by using brightness and contrast settings that obscured all but the strongest of longitudinal movements. When brightness and contrast are optimized to resolve smaller amplitude contractions, substantial movements become apparent in the presence of nifedipine. Movements persisting after nifedipine had a frequency close to the frequency of the recorded biopotentials (10–11 cycles per minute). Despite the obvious need to evaluate the contributions of movements to extracellular recording, control experiments to suppress movements of gastrointestinal muscles have been inadequate.

Biopotential recording problems

Field potentials recorded from excitable tissues have characteristic and distinctive waveforms that are based upon the transmembrane currents flowing during the excitable events. As discussed previously, many extracellular recordings from gastrointestinal muscles do not display consistent waveforms, which might suggest that something other than authentic electrophysiological events are recorded with this technique. In general, electrical events recorded with electrodes on the surface of muscles are approximated by the first-time derivative of membrane potential changes³². An example of this relationship is shown by comparing the intracellular action potential recorded from mouse heart, the extracellular (electrocardiogram) action potential recorded simultaneously and the first derivative of the action potential recorded with the intracellular electrode (FIG. 4). Several important points are apparent in these recordings: this demonstration confirms that the electrocardiogram is approximated by the waveform of the first derivative of intracellularly recorded action potentials; the action potential upstroke is the dominant event in the electrocardiogram and its relative rate-of-rise and time course are reflected by the first derivative; and the major transient of the electrocardiogram (QRS complex) occurs within approximately the same time course as the action potential upstroke — the action potential upstroke phase is responsible for the QRS complex.

Figure 4 |. Comparison of action potentials recorded simultaneously from mouse heart with intracellular and extracellular electrocardiogram electrodes.

a | Intracellular recordings of cardiac action potentials. b | The same events recorded simultaneously with extracellular electrodes. c | The first-time derivative (dV/dt) of the transmembrane potential changes occurring during the action potentials. Note how the major transient of the electrocardiogram (QRS complex) overlaps the time course of the upstroke of the action potential (dashed lines). The waveform of the electrocardiogram recording of action potentials is approximated by the first-time derivative of the voltage change during the action potential. The relationship between these recording modalities should also hold true for transmembrane voltage changes that occur during slow waves in gastrointestinal muscles. However, waveforms of extracellularly recorded biopotentials, claimed to be slow-wave events in the literature, vary substantially even within the same recording sessions and/or at the same point of recording (TABLE 1).

Slow waves propagating in ICC networks in gastrointestinal muscles have waveforms with the same key features as cardiac action potentials (upstroke and plateau phases), however, slow waves have far slower upstroke velocities, smaller amplitudes and longer durations than cardiac action potentials¹⁸. Analogous to cardiac action potentials, the first derivatives of slow waves recorded from ICC should approximate the waveforms of slow waves that would be recorded with extracellular electrodes, and the upstroke depolarization of the slow wave (that is, the portion of the slow wave that most resembles the QRS complex of the electrocardiogram) is most likely to be the best-resolved component in extracellular recordings. However, if a comparison is made between the first derivative of authentic slow waves recorded from ICC in intact muscles with intracellular microelectrodes and the variety of waveforms claimed to represent slow waves in the literature (see examples in TABLE 1), many discrepancies are obvious: the time courses of extracellular events do not come close to the kinetics of the slow wave upstroke or the first derivative; some recordings show biphasic or triphasic events that some might suggest are equivalent to the first derivatives, but these are too slow to represent the kinetics of the upstroke phase, sometimes by a factor of 10 or more; sinusoidal components lasting the duration of the slow-wave cycle and having the appearance of the events attributed to circular muscle contraction by Richter² are often noted; and some events along the axis of propagation of peristaltic contractions in the stomach display substantially different waveforms.

Extracellular recordings from mouse stomach muscles¹⁵ were criticized as being atypical of slow waves because rather than being biphasic, which was considered ‘ideal’⁴², they were multiphasic. In fact, multiphasic recordings have been recorded from the gastrointestinal tract in many studies using extracellular recording techniques, and these waveforms are likely to result from the contractile activities of the muscles^8,39,50,51. Events with a biphasic waveform have also been recorded from dog stomach^8,38,50 and it should be noted that when movements of gastric muscles were partially inhibited, biphasic waveforms could be recorded¹⁵. In older studies, the multiphasic waveforms were thought to be due to superposition of action potentials upon the slow waves⁵⁰, but there is no empirical verification that biphasic events were due only to slow waves, or that the multiphasic waveforms were a result of the generation of action potentials. In fact, as shown in the study of Lammers et al.⁸, multiphasic events were recorded from the orad and mid corpus, yet this region of the stomach does not generate action potentials, even when stimulated with excitatory agonists⁵².

Our recordings of biopotentials from the gastric smooth muscles of several species, including humans (example shown in FIG. 3), make it clear that more careful evaluation of the effects of movement on these recordings is essential¹⁵. Slow waves determine the frequency and amplitude of gastric phasic contractions¹⁷, but if contractions contaminate biopotentials then interpretations about the origin and spread of slow waves could be distorted. One might surmise that because there is coupling between slow waves and contractions, even if contractile activity was the actual event being monitored by extracellular recording, information about slow-wave frequency and direction might be obtained. The problem with this assumption is that contractile activity is complex and complicated by contractions of both circular and longitudinal muscles. In fact, the local trajectories of contractile movements in the mouse stomach were found to be approximately elliptical or circular¹⁵, which is not the pattern of slow-wave propagation in ICC networks⁵³. Contractions occurring at some distance from a recording electrode might cause tissue distortions at the recording site owing to the elastic properties of muscle tissues. Thus, mechanical artefacts can initiate the onset of biopotentials from sites far away from the site of recording, and claims that biopotentials precede local contractions are not a valid reason to exclude contractions as the source of signals¹⁰. Finally, arrhythmias, apparent ectopic pacemaking and other electrophysiological anomalies might be a result of uncoordinated contractile patterns. Such contractile patterns might result from abnormal pacemaking and slow-wave propagation, but it would be difficult to deconvolve the source and pattern of pacemaking and slow-wave propagation from the complex movements associated with this activity.

Electrophysiological filtering

Filtering parameters have been debated as part of the discussion over the validity of extracellular recording for measurements of field potentials in visceral smooth muscles⁵⁴. Some investigators have reasoned that to monitor slow waves, which are thought to occur at 3 cycles per minute in the human stomach and at ~10 cycles per minute in the small intestine¹⁷, filtering parameters must be set to capture these low frequencies using low-pass filtering with a window including 0.05 and 0.17 Hz, respectively. Slow waves might occur at 3 cycles per minute, but this aspect does not mean they are approximated by 0.05 Hz sine waves. If three action potentials occur in a nerve axon per minute, filtering should not be set at 0.05 Hz; appropriate filtering parameters would need to include a window that could capture the frequency response of a nerve action potential. As noted previously and highlighted in FIG. 5, slow waves have much faster frequency components (upstroke phase) than are found in 0.05 Hz sine waves. If there is any hope of recording the field potentials caused by slow waves in gastrointestinal muscles, the primary event that might be detected would be the upstroke phase, because it results from the highest transmembrane current density, causing the fastest change in potential (dV/dt). To capture events that have a dV/dt equal to or exceeding 1 V per second^27,29 (see FIG. 5), a filter window that enables frequencies between 3–100 Hz to pass through has been used¹⁵. These parameters are similar to the filtering used by the Lammers group in the majority of studies during the past three decades^8,40. By contrast, some investigators have used low-pass filtering with a cutoff at 2 Hz (REF. 6), which would be inappropriate for capturing the upstroke phase of authentic slow waves¹⁵. In response to this analysis, various filtering parameters were tested on the biopotentials obtained by extracellular recordings, and it was reported that 2 Hz low-pass filtering optimized the display of the data⁵⁴. From this analysis, the filtering parameters used by Lammers would attenuate most of the signal attributable to slow waves⁵⁴.

Figure 5 |. Effects of different filtering parameters on slow waves recorded directly from interstitial cells of Cajal.

Slow-wave recordings from myenteric interstitial cells of Cajal (ICC) in mouse (parts a, c, e, g) small intestinal muscle strips and rabbit²⁷ (parts b, d, f, h). Parts a, b show membrane potential changes in myenteric ICC that occur during propagating slow waves in intact muscle strips. By analogy with the heart, one might assume that extracellular recordings of slow waves made with monopolar electrodes would be similar in waveform to the first-time derivatives (dV/dt) of slow waves recorded with intracellular electrodes. Parts c, d show the first-time derivatives of the raw slow-wave recordings (smoothing ± 10ms, Δt = ± 10ms). Note that the dominant event corresponds to the upstroke depolarization and near quiescence occurs in the dV/dt traces between the upstroke and repolarization phase. The dV/dt traces were filtered with a band pass window of 3–100 Hz (parts e, f) or with a low-pass filter with a cutoff at 2 Hz (parts g, h). Note that the band pass filter between 3–100 Hz better preserves the expected field potentials, but low-pass filtering dramatically attenuates the upstroke potential.

If the desire is to determine the appropriate filtering parameters for electrophysiological events, then authentic electrophysiological signals should be used to test the filtering parameters. As shown in FIG. 4, field potentials generated by cardiac action potentials (electrocardiogram recordings) are approximated by the first derivatives of the transmembrane potential changes. By analogy, first derivatives of slow waves recorded from ICC would be the most appropriate test potentials for determining suitable filtering parameters for the extracellular electrical slow waves in gastrointestinal organs, as the field potentials generated by transmembrane currents in the ICC network are the source of the signals one hopes to record with extracellular electrodes. If filtering parameters are tested with biopotentials recorded with extracellular electrodes⁵⁴, and these events are not due to authentic slow waves, then the wrong filtering parameters might be chosen. For example, if the biopotentials recorded were contaminated with tissue movements, as we have suggested¹⁵, then filtering with a <2 Hz cutoff filter would be suitable because contractions of gastrointestinal muscles have far slower kinetics than the upstrokes of slow waves¹⁷. FIG. 5 demonstrates this point, showing reasonable preservation of the first derivatives of slow waves recorded directly from mouse and rabbit ICC using intracellular electrodes and a filter window between 3–100 Hz, but substantial attenuation of these events with a 2 Hz low-pass filter. By contrast, contractions of gastric muscle strips, with much slower events than slow waves, are preserved using a 2 Hz low-pass filter (not shown). Thus, suitable preservation of the extracellular signals using a 2 Hz low-pass filter further supports the hypothesis that the extracellular recordings are heavily contaminated, if not caused, by movement artefacts.

Modelling electrical activity

Cellular models of gastrointestinal pacemaker activity.

Conceptual and mathematical models of slow waves have been proposed for many years. Different modelling approaches, deterministic and phenomenological, have been utilized^55,56 but it has been difficult to develop deterministic models because there continues to be uncertainty about the mechanism of pacemaker activity. A hybrid model, containing both deterministic and stochastic qualities, will probably be required to describe gastrointestinal pacemaker activity because known molecular entities with well-understood properties (for example, inositol trisphosphate receptors, CaCC channels) are involved. However, other factors must still be treated as a ‘black box’, such as ion concentrations in microdomains created by the close proximity of endoplasmic reticulum and plasma membranes¹⁸ and the stochastic nature of Ca²⁺ release from stores. Owing to a lack of precise information about the ion channels and transporters involved in gastrointestinal pacemaker activity, Youm et al.⁵⁵ utilized many features developed for cardiac pacemaking to simulate the activity of mouse small intestinal ICC. Several realistic features were included in this model: cell geometry, Ca²⁺ binding proteins, Ca²⁺ release dynamics from the ER, ion channels (including the autonomous current described by Goto et al.⁵⁷), and ion transporters.

Simulations of slow-wave activity using this model were fairly consistent in waveform with the slow waves recorded from myenteric ICC⁵⁵.

Hirst and Edwards⁵⁶ also produced realistic simulations of the patterns of rhythmic electrical activity in myenteric ICC and circular and longitudinal muscles in the stomach using a three-compartment-equivalent circuit model. For pacemaker activity (slow waves), the authors used two separate mechanisms in the component simulating myenteric ICC — a voltage-dependent conductance and a second-messenger-linked mechanism to release Ca²⁺ from stores. This approach generated slow-wave activity that was similar to the activity obtained by intracellular recording⁵⁶. As discussed earlier, the low-input impedance of the non-ICC components of the SIP syncytium would impose a substantial load on pacemaker cells. Hirst and co-workers measured the coupling between ICC and SMCs and concluded that while ICC and SMCs are well-coupled to cells of the same type, coupling between ICC and SMCs is rather weak³⁰. Weak coupling between ICC and SMCs is probably important for ICC to function as pacemakers, so that depolarizations caused by activation of inward currents in ICC can reach the threshold for initiation of slow waves.

Other approaches to conceptualize multicellular slow-wave propagation have considered the Ca²⁺ stores in electrically coupled cells as a system of coupled oscillators⁵⁸. In this model, release of Ca²⁺ from stores is coupled through a voltage-dependent mechanism. The investigators simulated pacemaker activity based on the following processes: cyclical release of Ca²⁺ from stores, in which release is regulated by inositol trisphosphate and Ca²⁺; Ca²⁺ released from stores activates an inward current that causes depolarization; membrane potential regulates the excitability of the stores by modulation of inositol trisphosphate or Ca²⁺ concentrations; and cells are electrically coupled by gap junctions. These concepts generally fit well with current ideas about the mechanism of pacemaker activity in ICC, except that release of Ca²⁺ and changes in Ca²⁺ concentration and/or inositol trisphosphate are probably localized in microdomains¹⁸.

Models of slow­wave propagation in gastrointestinal organs.

Models that attempt to explain propagation of slow waves in gastrointestinal organs are more complicated than the cellular models referenced earlier, as they must be based on reliable electrical recordings from multiple sites and contain realistic variables for transmission of information between cells. ICC networks can actively propagate slow waves and experiments have demonstrated that a voltage-dependent mechanism is required to sustain active propagation^29,59,60. Computer simulations of the macroscopic activity in ICC networks have been used to estimate the effects of altered slow-wave propagation^61–63. More modern techniques that purportedly measure slow-wave propagation in the stomach (extracellular electrode arrays or magnetogastrography) potentially enable more detailed models to be developed that are based upon the slow-wave dynamics identified by these techniques^64–68. However, the ultimate strength and predictive power of these models depends on the specificity and quality of the data upon which they are based.

Models that attempt to reproduce patterns of slow-wave propagation observed with extracellular array recordings might be misleading if the recordings are contaminated by smooth muscle contractions and movement-induced electrical artefacts. As such, arrhythmias, re-entry and abnormal propagation described in these models could well be related to, or caused by, abnormal contractile activity. For example, a model to simulate gastric slow-wave re-entry in the ICC network⁶⁵ required anisotropic propagation biased towards the circumferential direction (1.6–1.8 × compared to longitudinal conductivity) based on observations from extracellular array recordings⁶⁸. Although there is good evidence that coupling between SMCs in the circumferential direction is higher than in the longitudinal direction⁶⁹, studies examining the spread of slow waves in gastric ICC networks using Ca²⁺ indicator dyes, or after removal of the circular muscle, have concluded that propagation in the ICC network is isotropic (same in all directions, 3.5 mm per second^53,69). Understanding the integrated multicellular activities that result in slow-wave generation and propagation in organs will require accurate information from each of the cell types involved. Models of slow-wave propagation based on recordings contaminated by movements might inaccurately ascribe behavioural features to ICC and other cells of the SIP syncytium.

Summary and future directions

Concluding that all extracellular electrophysiological studies of gastrointestinal muscles published in the past are flawed or inaccurate due to contamination from tissue movements is impossible. However, in the absence of appropriate controls to stabilize movements it is also not possible to conclude that movements did not affect these recordings. The absence of important control experiments (see BOX 1) always clouds interpretation of results. Surface electrical recordings might have provided information about contractile frequency and the direction and rate of propagation of segmental or peristaltic contractions, but ambiguities remain about the electrical events driving phasic contractions in vivo. At the present time, doubts remain in gastrointestinal motility research about: the normal range of slow-wave frequencies in vivo; the directionality and organization of slow-wave propagation; the effects of regulatory inputs on electrophysiological parameters in vivo; and the contributions of electrical abnormalities in motility disorders. Thus, it seems necessary for investigators in the field to develop better and more accurate electrical recording methods and/or to utilize specific and safe pharmacological means of uncoupling excitation–contraction coupling in gastrointestinal muscles. We hoped this Perspective will stimulate innovation in this field and emphasize the need to validate extracellular recording techniques before studies of this type are applied to animal models and human patients.

Box 1 |. Advice for adequate stabilization of movement.

Testing the effects of movement on extracellular recordings is an essential control for this technique. Establishing conditions that block contractions but do not affect slow-wave activity should be the goal for rigorous controls. Contractions are initiated by Ca²⁺ entry, usually attributed to L-Type Ca²⁺ channels in smooth muscles. However, other Ca²⁺ entry mechanisms are present in gastrointestinal muscles (for example, T-Type channels, receptor-operated channels, non-selective cation channels). Effective means of blocking Ca²⁺ entry is specific to the muscle layer, organ region and species under investigation. Tests should be performed on muscle strips cut parallel to circular and longitudinal muscle layers because both layers of muscle can contribute to movements that are not necessarily controlled by the same Ca²⁺ channel blocking drugs. Inhibitors of myosin light-chain kinase can also be useful for blocking contractions downstream of slow-wave activity. After effective drugs have been established for blocking contractions, their effects must be tested on slow waves using intracellular recording from the muscle strips, which is the best way to show that slow-wave activity survives the drugs or ionic conditions used to block contractions. As movement is the key contaminant of extracellular recordings, it is also best to test the drugs chosen on movements of tissue sheets in vitro. We have found that movements less than 50 μm can produce movement artefacts of approximately the same amplitude as signals reputed to be slow waves (unpublished work). Contractions even several centimetres from the site of electrode placement can cause distortions of the tissue near the electrode due to the elastic properties of the muscles. Analysis of movement by eye is insufficient. Video imaging with adequate magnification to resolve tiny movements and analysis via construction of spatiotemporal maps can be a good means of documenting contractile responses in muscle sheets.

Certainly, it would be beneficial to have diagnostic techniques that are capable of evaluating defects in pacemaker activity and propagation of electrical activity that lead to dysmotilities in visceral smooth muscle organs. Data derived from noninvasive techniques, such as electrogastrography, are probably even more problematic than recording from the surface of organs. This technique attempts to record gastric electrical activity from the abdominal surface, but having electrodes positioned even farther from the cells generating electrical slow waves makes it that much more unlikely that signals are electrophysiological in nature. Modelling based on data contaminated by movement is unlikely to provide mechanistically accurate or therapeutically useful information. Thus, the quest to understand slow-wave propagation and how it organizes muscle contractions into organ-level motility patterns is at a watershed moment in the history of gastrointestinal motility research. As the desire is to understand motility patterns, direct monitoring of movements via imaging techniques might prove most accurate and practical⁷⁰. Suborgan mechanisms and behaviours are best investigated by in vitro experimentation, as has been the case for every major organ system. In vitro experiments also enable high-resolution imaging techniques to be used without compromise. In the case of laboratory animals, the use of genetically expressible fluorescent indicators to monitor the excitable behaviour of specific populations of cells is being pursued⁷¹, as these techniques can enable direct and simultaneous visual monitoring of slow-wave propagation and organ-level movements. Deterministic models of excitation and propagation derived from such efforts might also help to develop a better understanding of slow-wave propagation in intact organs.