Dynamic clamp

From Scholarpedia
Astrid A. Prinz and Robert H. Cudmore (2011), Scholarpedia, 6(5):1470. doi:10.4249/scholarpedia.1470 revision #138328 [link to/cite this article]
Jump to: navigation, search
Figure 1: Dynamic clamp operating principle. From Goaillard, Marder (2006).

Dynamic clamp is an electrophysiology method that uses a real-time interface between one or several living cells and a computer or analog device to simulate dynamic processes such as membrane or synaptic currents in living cells.

Contents

Basic operating principle

The operating principle of the dynamic clamp is illustrated in Fig. #F1 for examples of one (right) or several (left) contacted cells. Each living cell is contacted with one or several electrodes and its membrane potential \(V\) (or \(V_1\) and \(V_2\)) is amplified and fed into the dynamic clamp machine (grey box). The dynamic clamp machine contains a model of the membrane or synaptic conductance(s) to be inserted in the living cell(s), either in the form of equations (for digital dynamic clamp systems, top) or in the form of a dedicated electrical circuit (for analog systems, bottom). The dynamic clamp system computes the current(s) \(I\) (or \(I_1\) and \(I_2\)) generated by the modeled conductance(s) and outputs it in real-time. That current is injected into the living cell, which therefore receives the same current as if it contained the membrane or synaptic conductance modeled with the dynamic clamp. For optimal dynamic clamp operation, the cycle of reading the membrane potential and computing and injecting the dynamic clamp current needs to be completed with an update rate faster than the fastest dynamic rate present in the system.

Types of dynamic clamp applications

Depending on what types of conductances are simulated with the dynamic clamp, different applications fall into one or several of the following categories:

Addition or subtraction of membrane currents

For the addition or subtraction of membrane currents (currents mediated by postsynaptic ion channel proteins), first, a model describing the voltage and time dependence of the given membrane current needs to be developed. Once the model is specified, the addition or subtraction of this membrane current is simply the dynamic clamp connected to a real neuron to take the membrane potential in and to compute the current to inject. Some common model formalism used to introduce or negate membrane currents are: Hodgkin-Huxley, FitzHugh-Nagumo, and Markov models. In general the model is created by fitting it to experimental data of the membrane current one is interested in simulating with dynamic clamp.

The effect of inserting and removing currents have been examined by many groups in a wide range of preparations, for an example see Ma and Koester (1996, Fig. #F2).

Figure 2: Example experiment where dynamic clamp is used to insert and remove a voltage and time dependent conductance. First, a sequence of action potentials (APs) are evoked (top left) and the width of APs becomes broader. Next, the increase in AP width is blocked with pharmacology (top center). Finally the AP broadening is mimicked by the dynamic clamp insertion of current. In another recording the dynamic clamp is used to partially mimic the pharmacological block by subtracting the current (bottom center). Taken from Prinz et al. (2004).

Addition or cancellation of synaptic connections

Similar to the addition and subtraction of postsynaptic ion channels (membrane currents), dynamic clamp can be used to introduce or negate electrical gap junctions and chemical excitatory or inhibitory synaptic input between two or more recorded cells. To accomplish this goal, a model specifying the time and voltage dependence of a synaptic input must be created. In one form a presynaptic neuron is recorded and its AP firing controls an artificial dynamic clamp synapse injected into a postsynaptic neuron. This configuration of dynamic clamp allows the parameters of an existing synapse to by explored with fine detail or the introduction of a new synapse where there was not one before. Similarly, the effects of an existing synaptic connection can be canceled by adding a negative spike-triggered conductance to counter the biological connection.

This configuration of dynamic clamp has been used to explore how the timing of synaptic input and the balance of excitatory/inhibitory input determine postsynaptic response and firing behavior of real neurons (Chance et al. 2002).

Simulation of network input

In vivo, neurons receive spontaneous excitatory and inhibitory synaptic input that is stochastic and can modulate the response properties of individual neurons. This stochastic synaptic input can be simulated in vitro by using dynamic clamp. The process involves the construction of an appropriate model describing the activity of a number of presynaptic neurons. This presynaptic activity is then translated into postsynaptic current and injected into the real neuron by the dynamic clamp.

Several studies have examined how this dynamic clamp introduction of background synaptic input determines the input-output of biological neurons (Chance et al. 2002) and to determine the detailed role of postsynaptic ion channels in this transfer (Desai and Walcott 2006). One example involving the injection of stochastic inhibitory and excitatory synaptic network input to recreate a high-conductance state in cortical neurons in vitro is described here.

Hybrid networks

Network behavior depends on the complex interaction between intrinsic neuronal properties, synaptic connections, and network architecture. Dynamic clamp can be used to construct hybrid networks of real and model neurons in order to systematically isolate and examine these properties to better understand network behavior (Le Masson et al 1995). In its simplest form, a hybrid network is constructed by connecting an isolated biological neuron to a model neuron via a modeled synapse. Other configurations of the dynamic clamp hybrid network include adding a new computational neuron to an existing biological network and rescuing an ablated neuron from a biological network.

This implementation of dynamic clamp is particularly powerful as it closes the loop between a real neuron and a simulated neuron (via a simulated synapse), allowing one to examine network behavior with exceptional experimental control.

Strengths and limitations of dynamic clamping

Strengths

In an ideal situation, one is able to introduce or negate an ionic conductance, and synaptic input. This is very powerful to aid in understanding the function of both individual neurons and networks of neurons.

In many situations, dynamic clamp can be used to mimic pharmacology in that the computational insertion or negation of a current (ion channel or synapse) can have the same effect as activating or inactivating the current with drugs. The dynamic clamp can sometimes have an advantage because of the fine control of the strength and kinetics of the current. In addition, dynamic clamp does not have the non-specific effects of pharmacological treatment.

Dynamic clamp is also an exceptional hands-on educational tool. By using a dynamic clamp setup and replacing the real neuron with a model neuron, all of the equipment for a real experiment can be used for teaching the concepts of electrophysiology.

Limitations

Two limitations of Dynamic clamp are entirely technical: it has to be fast and temporally consistent.

The time required for reading the membrane potential and calculating the current to inject has to be faster than the fastest time constant in the real neuron. This becomes problematic when the model is computationally intensive (as with stochastic Markov models). With ever increasing computer processor speeds, the speed limitation of dynamic clamp is constantly being pushed back.

The update interval of the dynamic clamp has to be reliable and reproducible. Modern operating systems (OS) such as Linux, Mac OS, and Windows suffer from the inability of a program running on the OS to be guaranteed a requested operation will be performed precisely when it is requested. Operating systems that can guarantee a request is performed within a well defined time window are Linux with the RTLinux kernel extension, LabView with the real time module, and DOS. A version of dynamic clamp developed by R. Pinto solves the time step problem by implementing a version that uses a variable time step in the calculation of current (Pinto et al. 2001).

Another limitation is an experimental limitation, dynamic clamp suffers from space-clamp problems. If one wants to introduce or negate an ion channel or synapse, dynamic clamp injects the proper current but this current is limited to a space around the recording electrode. This is especially problematic when the membrane current or synapse to be modeled is far from the recording/stimulating electrode. As occurs in CNS neurons with large dendritic trees where the membrane proteins which generate the current are located on distant dendrites. It also occurs when trying to simulate synapses as the location of synapses can be beyond the distance of the dynamic clamp space clamp and experimentally unreachable via an electrode.

Finally, dynamic clamp injects current and thus mimics the electrical effects of an ion channel. It is important to note that the proteins which mediate these electrical effects can also be linked to many other biological signaling pathways mediated by protein kinases, calcium concentration changes, and protein-to-protein interaction. These non-electrical effects are not mimicked by the dynamic clamp. Although lack of chemical effects often limits the physiological realism of dynamic-clamp simulated currents, it can also be perceived as an asset if electrical and chemical effects of the same membrane protein are to be teased apart experimentally.

A brief history of the dynamic clamp

The dynamic clamp was developed in parallel - and with less than desirable inter- and even intra-discipline communication - in two fields that deal with excitable cells, neurophysiology and cardiac physiology. The regretable lack of awareness and communication between researchers with similar goals in the two fields is presumably at least in part due to the use of different terminology by different researchers and developers: what we here subsume under the term dynamic clamp has in different versions and on different occasions been labeled "coupling clamp", "artificial synapse", synthesized conductance injection, hybrid network method, reactive current clamp, electronic pharmacology, and electronic expression, and perhaps other terms that the authors may not be aware of.

The following brief timeline highlights major steps in the development of the dynamic clamp in the context of cardiac physiology and neurophysiology. The timeline is not intended to be comprehensive and focuses on technical advances rather than attempting to list all published dynamic clamp applications. Corrections or suggestions for additional entries are appreciated:

  • 1990: Cardiac physiologists Tan and Joyner (1990) publish experimental results that use a simple analog circuit to couple two isolated cardiac ventricular cells through a coupling resistance that mimics a non-rectifying gap junction between the cells, and to couple a single ventricular cell to a passive resistance and capacitance (RC) circuit. The coupling resistance can be varied. Tan and Joyner call their system coupling clamp. Also see Joyner et al. (1991) for a second report using the same system.
  • 1992: Sharp et al. (1992) introduce the first dynamic clamp application in the neurosciences. They use an analog circuit functionally identical to that used by Tan and Joyner (1990) to create an artificial electrical synapse between two isolated neurons, and create an artificial chemical synapse by triggering a brief iontophoretic pulse of neurotransmitter onto one (postsynaptic) cell on the rising phase of the membrane potential oscillation in another (presynaptic) cell. In a note added in proof, Sharp et al. (1992) acknowledge learning of Joyner et al.'s earlier work after submission of their own paper.
  • 1993: Two groups of neurophysiologists independently report the first uses of digital circuits in dynamic clamp applications. Robinson and Kawai (1993) use a dedicated signal processing board to multiply a pre-defined digital synaptic conductance trace with the driving force derived from the continuously recorded membrane potential of an isolated neuron and inject the resulting artificial synaptic current back into the neuron. Simultaneously, Sharp et al. (1993a,b) use a digital processor to continuously perform the computations and differential equation integrations necessary to inject an artificial voltage-dependent membrane conductance into a neuron and create an artificial chemical synapse with voltage-dependent activation dynamics between two unconnected neurons. Sharp et al. (1993a,b) emphasize that their system effectively amounts to a conductance clamp (as opposed to the simpler methods of current clamp and voltage clamp), and for the first time use the term dynamic clamp. Neither Robinson and Kawai (1993) nor Sharp et al. (1993a,b) appear to be aware of each other's efforts or of the earlier cardiac work by the Joyner group.
  • 1995: Le Masson et al. (1995) use the dynamic clamp to construct three-cell hybrid neuronal networks from biological neurons and digital and hardware model neurons. At the time, these constitute the largest and most complex dynamic-clamp constructed hybrid networks.
  • 1999: Christini et al. (1999) introduce the first Real-Time Linux based dynamic clamp system for use in cardiac electrophysiology. The same system is later adapted and extended by Dorval et al. (2001) for use in neuronal electrophysiology.
  • 2001: While previous digital dynamic clamp systems were limited to the recording from and control of at most two neurons simultaneously, a new system introduced by Pinto et al. (2001) uses signal multiplexing to interface with up to four neurons.
  • 2004: Raikov et al. (2004) introduce MRCI, which includes a high level scripting interface that solves arbitrary systems of differential equations with minimal coding effort.
  • 2007: Hughes et al. (2007) introduce NeuReal, a dynamic clamp system designed for the implementation of extensive artificial dendrites and large hybrid networks that pushes the envelope by simulating over 1,000 Hodgkin-Huxley type conductances simultaneously.
  • 2007: Butera, Christini, and White introduce RTXI, an open-source software development project based on RTLDC (above, 1999).
  • 2008: Milescu et al. (2008) develop QuB, the first dynamic clamp system to include Markov-type kinetic models of voltage-gated ion channels, and to allow real-time modeling by simultaneously fitting the action potential waveform and pre-recorded voltage clamp data.

Currently available dynamic clamp systems

Below is a list of links to existing and available dynamic clamp systems. Further details about the hardware requirements etc. of most of these systems are provided under the respective links and/or in Prinz et al. (2004). The list is likely not comprehensive, but suggestions for additions and edits are highly welcome.

  • A Windows-based dynamic clamp software package with a user-friendly graphical interface is available here and described in Rabbah et al. (2005).
  • Another Windows-based system with the capability of contacting up to four neurons simultaneously, as described by Pinto et al. (2001), is available for download here. A newer version of the same system that allows the simulation of spike-timing dependent plasticity in artificial chemical synapses is described in Nowotny et al. (2006) and available on sourceforge.
  • NetClamp addresses the interrelated yet distinct purposes of modeling neural networks and performing dynamic clamp experiments, allowing users to construct networks with varying numbers of model cells and biological cells as well as varying synaptic connections among them. Dynamic clamping is robust and operated via National Instruments data acquisition cards.
  • QuB is a dynamic clamp system focused on the kinetic modeling of voltage-gated ion channels that runs under Microsoft Windows and achieves real-time performance by exploiting parallel processing on multi-core or multi-processor machines. The system is documented and available at QuB and described in detail in Milescu et al. (2008).
  • G-clamp is a dynamic clamp system based on Real-Time LabView and described by Kullmann et al. (2004).
  • RTXI is a Real-Time Linux based system that arose from the NSF-supported merger of three earlier systems described in Christini et al. (1999), Dorval et al. (2001), Butera et al. (2001), and Raikov et al. (2004).
  • The commercially available ITC-16 data acquisition board allows the implementation of artificial synaptic conductances based on programmable gate arrays.
  • Signal is a commercially available general-purpose data acquisition and analysis application that provides high-performance dynamic clamping (up to 100 KHz) using a fast RISC processor embedded in the Power1401 data acquisition hardware. The system is described at the CED web site.
  • National Instruments ([NI PXI-8176]) driven by Labview with the real time module.
  • Digital signal processing board (DS1104; dSpace, Novi, MI) driven by Matlab with Simulink.

References

  • Butera RJ, Wilson CG, DelNegro C, Smith JC (2001) A methodology for achieving high-speed rates of artificial conductance injection in electrically excitable biological cells. IEEE Trans Biomed Eng. 48:1460-1470. doi:10.1109/10.966605 PMid:11759927
  • Chance FS, Abbott LF, Reyes A (2002) Gain modulation from background synaptic input. Neuron 35: 773-82. doi:10.1016/S0896-6273(02)00820-6
  • Christini DJ, Stein KM, Markowitz SM, Lerman BB (1999) A practical real-time computing system for biomedical experiment interface. Ann Biomed Eng. 27:180-186. doi:10.1114/1.185 PMid:10199694
  • Desai NS, Walcott EC (2006) Synaptic bombardment modulates muscarinic effects in forelimb motor cortex. J Neurosci. 26:2215-26. doi:10.1523/JNEUROSCI.4310-05.2006 PMid:16495448
  • Dorval AD, Christini DJ, White JA (2001) Real-Time Linux Dynamic Clamp: A fast and flexible way to construct virtual ion channels in living cells. Ann Biomed Eng. 29:897-907. doi:10.1114/1.1408929 PMid:11764320
  • Hughes SW, Lorincz M, Cope DW, Crunelli (2007) NeuReal: An interactive simulation system for implementing artificial dendrites and large hybrid networks. J Neurosci Methods. 169:290-301. doi:10.1016/j.jneumeth.2007.10.014 PMid:18067972 PMCid:3017968
  • Joyner RW, Sugiura H, Tan RC (1991) Unidirectional block between isolated rabbit ventricular cells coupled by a variable resistance. Biophys J. 60:1038-1045. doi:10.1016/S0006-3495(91)82141-5
  • Kullmann PHM, Wheeler DW, Beacom J, Horn JP (2004) Implementation of a fast 16-bit dynamic clamp using LabVIEW-RT. J Neurophysiol. 91:542-554. doi:10.1152/jn.00559.2003 PMid:14507986
  • Le Masson G, Le Masson S, Moulins M (1995) From Conductances to Neural Network Properties: Analysis of Simple Circuits Using the Hybrid Network Method. Prog Biophys Molec Biol. 64:201-220. doi:10.1016/S0079-6107(96)00004-1
  • Ma M, Koester J (1996) The role of potassium currents in frequency-dependent spike broadening in Aplysia R20 neurons: a dynamic clamp analysis. J Neurosci. 16:4089–4101. PMid:8753871
  • Milescu LS, Yamanishi T, Ptak K, Mogri MZ, Smith JC (2008) Real-time kinetic modeling of voltage-gated ion channels using dynamic clamp. Biophys J. 95:66-87. doi:10.1529/biophysj.107.118190 PMid:18375511 PMCid:2426646
  • Nowotny T, Szucs A, Pinto RD, Selverston AI (2006) StdpC: A modern dynamic clamp. J Neurosci Methods. 158:287-299. doi:10.1016/j.jneumeth.2006.05.034 PMid:16846647
  • Pinto RD, Elson RC, Szucs A, Rabinovich MI, Selverston AI, Abarbanel HDI (2001) Extended dynamic clamp: controlling up to four neurons using a single desktop computer and interface. J Neurosci Methods. 108:39-48. doi:10.1016/S0165-0270(01)00368-5
  • Rabbah P, Nadim F (2005) Synaptic dynamics do not determine proper phase of activity in a central pattern generator. J Neurosci. 25:11269-11278. doi:10.1523/JNEUROSCI.3284-05.2005 PMid:16339022
  • Raikov I, Preyer AJ, Butera RJ (2004) MRCI: A flexible real-time dynamic clamp system for electrophysiology experiments. J Neurosci Methods. 132:109-123. doi:10.1016/j.jneumeth.2003.08.002 PMid:14706709
  • Robinson HPC, Kawai N (1993) Injection of digitally synthesized synaptic conductance transients to measure the integrative properties of neurons. J Neurosci Methods. 49:157-165. doi:10.1016/0165-0270(93)90119-C
  • Sharp AA, Abbott LF, Marder E (1992) Artificial Electrical Synapses in Oscillatory Networks. J Neurophysiol. 67:1691-1694. PMid:1629771
  • Sharp AA, O'Neil MB, Abbott LF, Marder E (1993a) Dynamic Clamp: Computer-Generated Conductances in Real Neurons. J Neurophysiol. 69:992-995. PMid:8463821
  • Sharp AA, O'Neil MB, Abbott LF, Marder E (1993b) The dynamic clamp: artificial conductances in biological neurons. Trends Neurosci. 16:389-394. doi:10.1016/0166-2236(93)90004-6
  • Tan RC, Joyner RW (1990) Electrotonic influence on action potentials from isolated ventricular cells. Circ Res. 67:1071-1081. PMid:2225348

The following PhD thesis seems to be the first implementation of the dynamic clamp (although we were unable to get the text):

  • Scott S (1979) Stimulation Simulations of Young Yet Cultured Beating Hearts. PhD Thesis, State University of New York at Buffalo.

Further reading

  • Destexhe A, Bal T, eds. Dynamic clamp: from principles to applications. Springer, 2009.
  • Goillard J-M, Marder E (2006) Dynamic clamp analyses of cardiac, endocrine, and neural function. Physiology. 21:197-207.
  • Economo MN, Fernandez FR, White JA (2010) Dynamic clamp: alteration of response properties and creation of virtual realities in neurophysiology. J Neurosci. 30:2407-2413.
  • Prinz AA, Abbott LF, Marder E (2004) The dynamic clamp comes of age. Trends Neurosci. 27:218-224.

External links

See also

High-conductance state

Personal tools
Namespaces

Variants
Actions
Navigation
Focal areas
Activity
Tools