CHAPTER TEN

Dynamic Metabolic Control of an Ion Channel Bertil Hille*, Eamonn Dickson*, Martin Kruse*, Bjoern Falkenburger† *Department of Physiology and Biophysics, University of Washington, Seattle, Washington, USA † Department of Neurology, RWTH Aachen University, Aachen, Germany

Contents 1. Background and History of Channel Modulation and KCNQ Current 2. Modeling Approach 2.1 Goals and programming environment 2.2 Tools to obtain quantitative kinetic data 3. Modeling Receptor and G-Protein Activation 3.1 General considerations 3.2 Ligand, receptor, and G-protein: A ternary complex 3.3 G-protein activation 3.4 Unresolved problems with the ternary complex model and G-protein activation 4. Activation of PLC, Modulation of Channels, and Deactivation of G-Proteins 5. PLC Messengers 5.1 Production of DAG and activation of PKC 5.2 IP3 and Ca2 þ signals 5.3 Concentration–response relations and spare receptors 6. Phosphoinositide Metabolism and Compartments 6.1 The surface density of PI(4,5)P2 6.2 Lipid pools 6.3 Relative rates of the lipid kinases 6.4 Stimulated phosphoinositide kinases 6.5 Compartments 6.6 Closing words References

220 223 223 224 229 229 230 232 234 234 236 236 237 238 239 239 239 240 240 241 243 243

Abstract G-protein-coupled receptors mediate responses to external stimuli in various cell types. We are interested in the modulation of KCNQ2/3 potassium channels by the Gq-coupled M1 muscarinic (acetylcholine) receptor (M1R). Here, we describe development of a mathematical model that incorporates all known steps along the M1R signaling cascade and accurately reproduces the macroscopic behavior we observe when KCNQ2/3 currents are

Progress in Molecular Biology and Translational Science, Volume 123 ISSN 1877-1173 http://dx.doi.org/10.1016/B978-0-12-397897-4.00008-5

#

2014 Elsevier Inc. All rights reserved.

219

220

Bertil Hille et al.

inhibited following M1R activation. Gq protein-coupled receptors of the plasma membrane activate phospholipase C (PLC) which cleaves the minor plasma membrane lipid phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) into the second messengers diacylgycerol and inositol 1,4,5-trisphosphate, leading to calcium release, protein kinase C (PKC) activation, and PI(4,5)P2 depletion. Combining optical and electrical techniques with knowledge of relative abundance of each signaling component has allowed us to develop a kinetic model and determine that (i) M1R activation and M1R/Gb interaction are fast; (ii) Gaq/Gb separation and Gaq/PLC interaction have intermediate time constants; (iii) the amount of activated PLC limits the rate of KCNQ2/3 suppression; (iv) weak PLC activation can elicit robust calcium signals without net PI(4,5)P2 depletion or KCNQ2/3 channel inhibition; and (v) depletion of PI(4,5)P2, and not calcium/CaM or PKC-mediated phosphorylation, closes KCNQ2/3 potassium channels, thereby increasing neuronal excitability.

This chapter concerns receptor modulation of potassium channels of the KCNQ family. By modulation, we mean changes of the properties of ion channels due to receptor-initiated biochemical signaling within the cell. Modulation is typically much slower than the very rapid action of membrane voltage on channel voltage sensors or of extracellular neurotransmitters on the fast ligand-gated receptor channels of synapses. Ongoing modulatory signals continually alter the firing decisions and the input–output relations of neurons and nerve circuits. They underlie major changes of mental state and are the target of most of the major drugs of psychiatry. The classical work of Hodgkin and Huxley1 introduced neurobiologists to highly deterministic computation of stereotyped electrical responses of neurons from kinetic models of ion-channel gating. With the recognition of ion-channel modulation, it became clear that channel gating does not have fixed stable properties. Instead the Hodgkin–Huxley equations must be continuously changing to reflect the dynamic impact from slower modulatory signals. Thus, full simulations of neuronal excitability would need to incorporate the kinetics of modulatory signals that act through receptors on intracellular enzymes and metabolism. Neurobiologists frequently call the underlying receptors, metabotropic receptors. Most of them are G-protein coupled. We describe the history and background of receptormediated ion-channel modulation and then outline our work toward a fuller mechanistic and kinetic analysis of one example.

1. BACKGROUND AND HISTORY OF CHANNEL MODULATION AND KCNQ CURRENT Since the late 1800s, physiologists have noted that hormones and plant extracts change the heart rate, contraction of the gut and arteries, and

221

Modulation of a K Channel

secretion from glands. During the 1960s and 1970s, it was recognized that cell membranes have ion channels and G-protein-coupled receptors and that many physiological actions of hormones and neurotransmitters are accounted for by modulation of the channels via receptor activation. Probably, the pioneering example was augmentation of cardiac calcium currents by beta-adrenergic stimulation of the heart.2 Eventually, that augmentation was explained by cyclic AMP-dependent phosphorylation of the calcium channel itself.3 Table 10.1 shows additional, very early examples of ionchannel modulation through G-protein-coupled receptors.4–10 Judging from the pace of current research, we can suggest that every type of ion channel has several kinds of modulation that change its gating properties in response to physiological inputs. Such modulation, prominent in neurons, is also present in all other types of cells of the body, muscle, epithelia, etc. Brown and Adams (1980) discovered that cholinergic agonists such as acetylcholine acting on muscarinic receptors turn off a novel Kþ channel in frog sympathetic neurons.4 Regulation of this channel is the focus of this chapter. Because of the muscarinic regulation, they called the current IM. Subsequent work showed that the receptor is the M1 muscarinic receptor and that many other receptors that couple to the G-protein now called Gq regulate the M-current channel as well as several Ca2þ channels in various neurons.11,12 Already partially open at resting potentials, this channel regulates the excitability of neurons. When the Kþ current is suppressed in a graded way by receptor action, the cell depolarizes and becomes more excitable. The channel was cloned and the principal subunits given the gene name KCNQ, a family with five gene subtypes. A heterotetramer of KCNQ2 and KCNQ3 subunits forms the classical neuronal M-current channel. Regulation of KCNQ channels by Gq-coupled receptors had an Table 10.1 Early examples of ion-channel modulation Current Modulator

ICa heart"

Adrenaline (cAMP)2

IM sympathetic neuron#

mACh (several messages)4

ICa DRG neuron#

Adrenaline (several messages)5

IS aplysia#

5-HT (cAMP)6

IGIRK heart"

mACh (Gi/o)7,8 mACh (Gbg)9

IAHP #

Adrenaline (cAMP)10

222

Bertil Hille et al.

Figure 10.1 Muscarinic modulation of KCNQ2/3 current and simulation by the model. (A) KCNQ potassium current recorded under whole-cell patch clamp. A tsA-201 cell has been transfected with KCNQ2 and KCNQ3 subunits and the M1 muscarinic receptor so that it expresses the M-current ion channel. When the muscarinic agonist oxotremorineM is applied to the cell, the KCNQ current is nearly fully suppressed. It recovers slowly when agonist is removed. This example is from a single cell. Each cell is slightly different from the next. (B) The output of our kinetic model mimics this behavior.7,13,15–21 The model simulates the average of our observations and is not optimized to match the cell in panel (A).

unexpected mechanism. Activation of Gq canonically activates the enzyme phospholipase C (PLC), which cleaves the minor plasma membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2). The wellknown products of this cleavage are the two second messengers diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP3). However, rather than depending on the signaling of these two messengers, the strong muscarinic modulation of the channel is primarily due to the depletion of the lipid PI(4,5)P2.13,14 The channel requires plasma membrane PI(4,5)P2 to function and turns off when that phospholipid is depleted (Fig. 10.1A). A dependence of some ion channels and transporters on PI(4,5)P2 was first proposed by Donald Hilgemann.8,22 All members of the KCNQ family need this membrane lipid to function. Here, we describe our modeling of the muscarinic modulation of KCNQ channels. The experiments and development of the model have been published in a series of papers.13,15–20,23 In conceptual terms, the model and this chapter deal with the following modular components: Agonist action turns on the PLC enzyme, the lipid is cleaved, and the channel turns off. In parallel, the messenger products IP3 and DAG initiate signaling of

Modulation of a K Channel

223

Figure 10.2 Schematic cartoon of early steps in the signaling pathway to channel modulation. (A) The Gq-coupled muscarinic receptor binds the agonist ligand and activates the G-protein. The G-protein turns on the enzyme PLC. (B) Active PLC cleaves membrane PI(4,5)P2, the KCNQ channel is inhibited as it loses PI(4,5)P2, and several second messengers are generated. Modified from Ref. 17.

their own, and during recovery PLC is turned off again and lipid synthesis restores the pools of PI(4,5)P2. Most of the underlying signaling steps represent enzymatic modifications of membrane lipids and G-proteins. The individual steps are discussed in detail below. Many steps in the onset of agonist action are summarized in Fig. 10.2, which will serve as a point of reference to guide the discussion. While developing the model, we tried to measure the time course of all the intermediate steps using optical and electrical methods.15–20,23,24 The measurements helped to refine the assumptions of the model, and the model helped to specify further measurements.

2. MODELING APPROACH 2.1. Goals and programming environment Our goals were to understand and describe the time course of suppression and recovery of KCNQ channels during addition and removal of agonists like acetylcholine (Fig. 10.1A). The overall model output would have to

224

Bertil Hille et al.

simulate current measurements (Fig. 10.1B) and would enable inclusion of ion-channel modulation in any simulations of the firing activity of sympathetic neurons or of other central neurons expressing KCNQ channels. Ideally, on a cell biological level, every identifiable enzymatic step would be represented explicitly. The detailed steps within the model would test and describe hypotheses about G-protein-coupled signaling and phosphoinositide lipid metabolism. They should predict what would happen in different neurons that express different quantities and subtypes of the signaling enzymes. In practice for neurobiological computations, the many internal mechanisms could be lumped appropriately into fewer components to allow faster simulations. For implementing the calculations, we chose the Virtual Cell software environment hosted by the University of Connecticut Health Center.25,26 Virtual Cell allows simulation of multicompartment, one-, two-, and three-dimensional models with large numbers of explicit reaction mechanisms that generate ordinary- and partial-differential equations. The user places reactants, enzymes, and products in different compartments using a graphical interface and specifies the rate laws. Typically, the user interface is on a personal computer and the actual calculations are done at high speed on the hosting mainframe. The calculation engine generates the differential equations and integrates them in time in a manner transparent to the user. The system handles multiphase models that, for example, combine membrane surface compartments and cell volume compartments. Thus, the transformation from surface units to volume units in hydrolysis of membrane PI(4,5)P2 to yield membrane DAG and cytoplasmic IP3 is handled automatically. Finished models can be posted in a public library for public use and simulated results exported in various formats including graphs, spreadsheet data (comma separated value .csv), image files (GIF, NRRD), or movies (Quicktime or Animated GIFs). Equations from Virtual Cell can be exported to MATLAB or as SBML files. A public version of our latest model is available and can be run by logging into Virtual Cell and loading the biomodel: FalkenburgerDicksonHille2013 from shared: Hillelab.

2.2. Tools to obtain quantitative kinetic data To enable quantitative kinetic modeling, we needed experimental techniques that measured the speed, concentrations, and buffering capacity of many signaling steps in living tsA-201 cells. Overexpressed fluorescent

Modulation of a K Channel

225

Figure 10.3 Expanded illustration of step 4 from Fig. 10.2, showing two examples of the use of real-time FRET. Schematic illustration of the overexpressed fluorescent reporter proteins used as real-time FRET monitors to track Gaq dissociation from Gb1 (4a) and subsequent activation of PLC (4b). In step 4a, Gb is the FRET acceptor and in 4b, PLC is the FRET acceptor.

reporter proteins were used as real-time monitors to track dynamic intermolecular or intramolecular interactions along the activated muscarinic receptor signaling cascade.16–19,23 Transfecting CFP- and YFP-labeled signaling proteins into cells allowed the progression of the muscarinic signal through each intermediate signaling step (Fig. 10.2) to be measured via Fo¨rster resonance energy transfer (FRET). The fundamental concept for FRET is that blue light normally evokes blue fluorescence from an isolated CFP, but when a YFP-labeled probe is in the close vicinity, energy can be transferred from CFP (donor) to YFP (acceptor), which then emits yellow light. FRET is a nonlinear measure of the proximity of donor molecules to acceptor molecules. One example is shown in Fig. 10.3 (step 4a), where two subunits of the G-protein have donor and acceptor fluorescent proteins. When they are together, the fluorescence is at longer wavelength because of FRET, and when signaling steps move them apart, the fluorescence changes to shorter wavelength. The amount of FRET is often quantified simply as the ratio of yellow emission YFPC to blue emission CFPC (YFPC/CFPC) during blue excitation.27–31 We call this the FRET ratio (FRETr). Thus, from early in the signaling pathway, the following donor and acceptor pairs were constructed (numbered as in Fig. 10.2A): (1) CFP and YFP attached at two different positions on the M1 muscarinic receptor protein reported the kinetics of conformational changes within the receptor upon ligand binding. (2) CFP attached to the receptor and YFP attached to a G-protein subunit reported the changing distance or orientation between receptor and G-protein.

226

Bertil Hille et al.

(3) CFP attached to Gaq and YFP to Gb1 reported the kinetics of G-protein separation as in Fig. 10.3. (4) Gaq–CFP and PLC–YFP reported the time course of Gaq/PLCb1 interaction. To measure FRETr, we made two-wavelength photometric measurements on single cells with an epifluorescence microscope. Our evidence that the quantitative data gathered from each CFP–YFP FRET pair represented actual energy transfer between CFP and YFP fluorophores, rather than some other optical change is that: (i) during perfusion of agonist, CFPC and YFPC values typically changed in opposite directions with identical time courses; (ii) fluorescence changes were reversible following removal of agonist; and (iii) when strong 500 nm illumination was used to bleach the YFP fluorophore (known as acceptor photobleaching), CFPC increased and FRETr fell to near zero. We measured FRETr at different sampling intervals depending on the kinetics of the reaction in question. For example, the time constant for M1 receptor activation is very fast, requiring sampling at 20 Hz, whereas the time constant for PI(4,5)P2 depletion is slow, allowing sampling at 0.25 Hz. Quantitative aspects of signaling depend on the absolute amounts of signaling molecules present. Therefore, transfection of fluorescently tagged signaling molecules, although extremely informative in terms of the kinetic information, also can perturb the system. The new proteins can reach much higher densities than the endogenous proteins and may alter the steady-state and kinetic properties of signaling. For instance, transfecting G-proteins increases the percentage of G-protein-bound receptors, and transfecting PLC accelerates G-protein deactivation (see below). Thus, amounts of endogenous and overexpressed signaling components needed to be measured. Conversely, changing the amounts of signaling molecules in this way also is an interesting intervention that further probes the signaling system. To quantify the density of overexpressed fluorescent molecules, the fluorescence intensities of cells were compared (i) to fluorescent beads, (ii) to solutions of recombinant fluorescent proteins, and (iii) to levels of transfected and endogenous proteins measured using Western blot analysis.17–19,23 These three independent assessments of protein expression levels indicated that proteins overexpressed enough to give good optical signals have densities around 3000 mm2 at the plasma membrane, whereas endogenous proteins have typically only

Dynamic metabolic control of an ion channel.

G-protein-coupled receptors mediate responses to external stimuli in various cell types. We are interested in the modulation of KCNQ2/3 potassium chan...
1MB Sizes 2 Downloads 3 Views