Imagine trying to eavesdrop on a room full of people by thrusting a microphone, attached to a ramrod the size of telephone pole, through the wall. Although this technique of listening might allow one to hear some noises, the message would not be representative of what normally goes on within that room. Our intrusive microphone would blatantly disrupt any conversation. That is the type of challenge that is faced by neurophysiologists who wish to study the details of the electrical activity within single nerve cells. These cells measure, on average, less than 30 or 40 microns across, and in many cases, have a diameter of less than 20 microns, one-fiftieth of a millimeter and a fraction of the breadth of a human hair. Compounding the challenge, the electrical signals are tiny, ranging from 1/10 of a volt at largest to less than 1/1,000 of a volt (1/10,000 of the voltage of an AAA battery) at smallest, depending on the type of electrical activity one wants to record.
The largest of these signals are the all-or-none action potentials, the nerve impulses produced by any given neuron. Each neuron communicates with other neurons by producing action potentials, which propagate along its axon, finally reaching synapses at the axon terminals where the message is passed on to other (“postsynaptic”) neurons by a process called synaptic transmission. The message of each neuron is encoded by the rate and pattern of its action potentials. Early neurophysiologists developed the capability to record the action potentials produced by single neurons in the brain and spinal cord, using tiny microelectrodes fashioned from stainless steel or tungsten, tapered to a very fine tip and insulated to within a few thousandths of a millimeter at the tip. These electrodes were inserted into the brain or spinal cord but did not enter neurons, so the nerve impulses were recorded without actually puncturing or touching the target nerve cells. This configuration was like placing a microphone just next to, but outside, of a room full of people. Nerve impulses from different neurons or “units” could be distinguished using this methodology on the basis of their size, with the signals from the neurons closest to the electrode being largest. This type of “single unit recording” turned out to be quite informative for researchers interested in determining what a neuron is saying. The rate and pattern of its action potentials are its message.
But what if one wants to understand how the neuron produces its message. In generating a nerve impulse or action potential, a neuron uses a rich repertoire of ion channels—several types of sodium channels, several types of potassium channels, and in some neurons, one or more types of calcium channels—which turn on and off in a complex but well-regulated way. If one wants to understand the molecular basis for neuronal signaling, or if one wants to target the molecules that generate nerve impulse activity for therapeutic purposes, one needs to discern the activity of these various types of ion channels. Imagine trying to dissect the contribution of one particular ion channel within the chatter of the nerve cell. Since a single neuron may contain dozens of types of ion channels, this is like trying to discern, microscopically, the contribution of one musical instrument within the music produced by an orchestra. How can neuroscientists accomplish this?
The electrical activity of nerve cells is produced by ion channels in the cell membrane and is most accurately measured by assessing the voltage drop across the cell membrane or the current flowing across it. These measurements require comparison of electrical activity inside and outside the cell. But how does one record electrical events inside single nerve cells? The first recordings from inside of nerve fibers took advantage of the presence in the squid of “giant” axons, which earned their name because they have diameters of nearly 1 millimeter, fifty times larger than the diameters of human nerve fibers. In the late 1940s, Hodgkin and Huxley, working at Cambridge University, succeeded at inserting tiny wires into squid giant axons and were able to record the actual currents that flowed across the nerve cell membrane to produce action potentials. This allowed them to predict the presence of sodium channels within the nerve cell membrane and to describe their basic properties. Generations of “axonologists” followed up on this seminal work, and, even today, some researchers continue to use the squid giant axon as a model and the Hodgkin–Huxley equations, developed on the basis of recordings in the squid giant axon, continue to form a basis for modern neurophysiological theory.
The squid giant axon, and other invertebrate preparations, provided invaluable information about the processes by which nerve cells generate electrical signals. Nevertheless, neuroscientists wanted to understand these processes in the nerve cells of vertebrates including mammals. At the end of the 1940s, Ralph Gerard, a physician-researcher in the Department of Physiology of the University of Chicago, developed “sharp intracellular microelectrodes,” pulled from tiny glass capillary tubes so that they tapered to only several microns at the tip. He used these, as small harpoons, to directly impale nerve cells and record their electrical activity. This methodology propelled a generation of important investigations which included pioneering studies such as those of Bernard Katz, who, in seminal work at University College London, discovered that tiny, discrete packets of neurotransmitters are released at synapses by one neuron and act as signals on an adjacent nerve cell; and the studies of John Eccles in Canberra which demonstrated the ways in which excitatory and inhibitory synapses can both impinge on the same receiving nerve cell, which integrates them like a tiny computer chip. The following generation of electrophysiologists used the intracellular microelectrode to record the electrical activity within a multitude of types of neurons, and these studies gave us much of the basis for neurophysiology. Even the finely honed intracellular microelectrode, however, invaded the neurons from which it recorded, usually doing some damage and possibly distorting the message.
Against this background, Erwin Neher, Bert Sakmann, and their colleagues in Germany developed the “patch-clamp” recording method for studying electrical activity of nerve cells in the early 1980s (Neher and Sakmann 1992). Their innovative new method involved using finely crafted glass micropipettes that, rather than impaling neurons, fuse with their outer cell membranes, where they form a nearly perfect high-resistance “gigaseal.” This permits relatively noninvasive recording, with high fidelity, of electrical activity within these neurons. Their development of the patch-clamp method was honored with the Nobel Prize in 1991.
This brings us to the question of how one can study, in a precisely quantitative manner, the effect of a mutated ion channel on the behavior of a particular type of neuron. One approach (Dib-Hajj et al. 2009) is to grow the neuron of interest in culture and insert the gene for the mutated ion channel into the cell. Patch-clamp recording can then be used for a head-to-head comparison of neurons transfected with mutant channels and neurons transfected with normal channels (termed “wild-type” by geneticists), so that the properties of the cells containing mutant and wild-type channels can be compared. This method can be very informative in providing a qualitative or semi-quantitative assessment of the effect of the mutant channel on neuronal firing. However, in experiments of this type, the level of expression of the transfected channels (the number of functional ion channels produced in each cell) cannot be precisely controlled, and thus it is not possible to ensure that the number and density of mutant channels within the transfected cell are similar to those in a naturally occurring neuron. An alternative approach is to use genetic “knock-out” strategies to silence one particular gene so that cells containing the channel of interest can be compared with cells that do not contain it. This approach, however, also has a limitation: It is usually constrained to studying cells from mice where knock-out is most readily achieved, and it limits the assessment to an “all-or-none” comparison. Still another potential alternative, recording from actual neurons from humans carrying the mutation, presents several challenges: In one approach, the nerve cells would have to be removed from a living human subject, which is ethically unacceptable. Alternatively, cells could be obtained from humans carrying the mutation at postmortem, but, in order to maintain the fragile cells and the molecules within them in a healthy state, the postmortem would have to be carried out and the neurons transported to the research laboratory within several hours of death, which is a substantial challenge.
The Vasylyev et al. paper (Vasylyev et al. 2014) addresses this problem by using the powerful technique of “dynamic clamping.” The dynamic clamp method combines the strengths of patch-clamp recording with powerful computer simulation methods (Prinz, Abbott, and Marder 2004). We used it to study the effect of a precisely calibrated number of human NaV1.7 channels, either wild-type or mutant, by virtually placing the simulated channels within an actual healthy pain-signaling DRG nerve cell removed from a rat or mouse. In our iteration of this method, to study mutations of NaV1.7 we first used a patch-clamp electrode and powerful amplifiers to record the electrical currents (including the currents produced by NaV1.7 channels) from a native, unperturbed nerve cell. We then used computational algorithms to subtract the current produced by the normal NaV1.7 sodium channels and electronically injected, into the cell, electrical current that simulates the activity of a precisely calibrated number of mutated ion channels which substitute for the deleted normal channels.
The L858H mutation had been found in one of the early patients reported with inherited erythromelalgia (Yang et al. 2004), and we previously had used traditional patch-clamp methods to study the effect of the mutation on the behavior of the channel (Cummins, Dib-Hajj, and Waxman 2004). Now, we wanted to move closer to understanding the clinical effect of the mutation, and to understand the details of how the mutant channel altered the behavior of pain-signaling DRG nerve cells. The computational power of the dynamic clamp allowed us to precisely titrate the number of wild-type, normal NaV1.7 channels, or mutant NaV1.7 channels, that we sequentially placed in a real pain-signaling neuron. Thus, we could ask this question: What happens if you remove the normal NaV1.7 channels in a DRG neuron and replace them in the same cell with a precisely calibrated number of mutant channels?
Dynamic clamp analysis required modeling of the NaV1.7 channel. This, in turn, required development of a computer program that precisely simulated the dynamic behavior of the channel, so that we had to go back to the laboratory to assess, in detail, the properties of the channel in multiple functional states:
Next, these measurements were used to build a computer program which simulated the characteristics of the channel. Dmytro Vasylyev, a biophysicist who had received advanced mathematical training at the Bogomoletz Institute in his native Kiev, spent weeks constructing the model and writing the code for the program, which required calculating the current produced at a particular time by the channel, then recalculating again and again at intervals of 10 μsec (ten millionths of a second). Altogether, his computer had to repeat the calculations in an iterative manner for each 1/100,000 of a second of real time in a neuron, which was a substantial computational challenge. But he persevered, and, when this was done, we were finally ready to begin the experiment.
As a first step, while recording from a real cell, we subtracted the current produced by the endogenous NaV1.7 channels. We then asked how the presence of normal NaV1.7 channels alters the excitability of DRG neurons. To answer this question, we used the computer, which had virtually removed all NaV1.7 channels from the cell, to add precisely calibrated amounts of the NaV1.7 channel back into cell so that we could assess the effect of the cell’s having no NaV1.7 channels, or 20%, 40%, 60%, 80%, or 100% of its actual complement of NaV1.7. These experiments taught us that NaV1.7 channels act in a remarkably linear manner to increase the excitability of the cell. The more NaV1.7 channels, the more excitable the cell became, with the relationship between amount of NaV1.7 and threshold of the neuron falling on a straight line. The linear relationship showed us that the regulation of the number of NaV1.7 channels in a cell can precisely control the cell’s responsiveness. This is a beautiful example of elegance in the molecular architecture of the neuron. The strong effect of the number of NaV1.7 channels within DRG neurons on the degree of excitability of these cells may also be relevant to disease because sensory neurons produce extra NaV1.7 channels in response to inflammation and injury.
We next used the dynamic clamp to ask questions about mutant NaV1.7 channels. We wanted to understand, for example, “Will a small number of mutant channels have an effect on the behavior of the cell?” and “How large an effect does the mutant channel—virtually inserted into a real cell at a density that mimics the situation in humans with inherited erythromelalgia—have on the behavior of the cell?” We learned, in terms of the first question, that “gain-of-function” changes, which increase the activity of NaV1.7 mutant channels, are so powerful that even a small number of mutant NaV1.7 channels will markedly increase the excitability of a pain-signaling DRG neuron. And we learned, in terms of the second question, that deployment of mutant channels at levels expected in humans has an even larger effect on the cells, making them even more overactive. These experiments showed very clearly that mutant NaV1.7 channels, expressed at a physiologically relevant level, produce persistent sodium currents that, while tiny, are sufficient to depolarize neurons and reduce the threshold for nerve impulse generation, thus teaching us how, in a very fundamental sense, the mutant channels produce hyperexcitability of DRG neurons that underlies pain.
Shortly after completing our dynamic clamp study on NaV1.7, Vasylyev returned to Kiev. The dynamic clamp method, however, was up and running. A few months thereafter, we used this methodology to study the human NaV1.8 channel (Han et al. 2015) and the human NaV1.9 channel (Huang et al. 2014), two cousins of NaV1.7. Dynamic clamp gave us a quantitative, high-resolution picture of all three of the “peripheral” sodium channels and some of their mutants.