255 Chapter 10 Patch clamp recording from cells in sliced tissues ALASDAIR J. GIBB and FRANCES A. EDWARDS 1. Introduction Before the introduction of gigohm seal patch clamp techniques (Hamill et al. 1981) the best resolution that could be achieved when measuring ionic currents in cell membranes was of the order of 100 pA. The combination of the gigohm seal and the placement of the current-voltage converter clamping amplifier in the headstage gave a dramatic improvement in signal resolution allowing currents of around a pA or less to be resolved (Neher, 1992; Sakmann, 1992). This great improvement in technique and instrumentation was followed by a tremendous expansion in the variety of cell types accessible to electrophysiological investigation. This was particularly true in relation to experiments on small cells (e.g. adrenal chromaffin cells: Fenwick et al. 1983; and central neurones in tissue culture: Nowak et al. 1984; Cull-Candy & Ogden, 1985; Bormann et al. 1987) and should also have been true for neurones in brain slices. However, the problem of obtaining a clean access for the patch electrode to the surface of neurones in brain slices seemed insoluble and so the potential advantages of applying patch clamping to brain slices were not immediately achieved. Signal resolution in patch clamp recordings The signal resolution achieved in any particular experiment depends on several related factors but the basic point in this relates to the noise inherent in any resistor (e.g. Neher, 1992; see also Chapters 4 and 16). The rms thermal noise in a simple resistor is given by s = (4kTD f/R)0.5where s is the rms of the current through the resistor, kis n n Boltzmann’s constant, T the absolute temperature, D f is the bandwidth and R is the resistance. Basically, what this equation says is that the current noise in the recording is inversely proportional to the source resistance. In order to measure currents of the order of a pA at a bandwidth of 1 kHz, the resistance of the source needs to be of the order of 2 GW (2· 109 W ). The ‘source resistance’ in this case is mainly the A. J. GIBB,Department of Pharmacology, University College London, Gower St, London WC1E 6BT, UK F. A. EDWARDS, Department of Pharmacology, University of Sydney, NSW 2006, Australia 256 A. J. GIBB AND F. A. EDWARDS combination of the feedback resistor in the amplifier, the seal resistance, and the resistance of the preparation itself (cell or isolated patch). In addition, noise associated with the electrode and cell or patch capacitance add to the total (see Rae & Levis, 1992 for a discussion of noise sources). The noise contributions from each source add together as the square root of the sum of the squared rms noise for each component. Some background to patching cells in slices In principle the small cell size and small membrane currents of cells in the central nervous system make them ideal for patch clamp recording. However, in the years after patch clamping was first described, cells in intact central nervous system tissue were thought to be inaccessible for patch clamping because there was no apparent way to maintain the intact structure of brain tissue and yet achieve the clean cell membrane that is essential for forming a gigohm seal between glass pipette and cell (Hamill et al. 1981). In the meantime neurones and glia were studied in primary culture. Unfortunately, primary cultures have the disadvantage that changes in gene expression and synaptic connections, and even the identity of cells all become unknown factors. An alternative was to study neurones and glia after acute dissociation using enzymes such as papain or trypsin (Numann & Wong, 1984; Gray & Johnston, 1986; Kay & Wong, 1986; Barres et al. 1990). Although these preparations allowed the study of receptors and ion channels on adult central neurones (e.g. Kay & Wong, 1987; Huguenard et al. 1988; Sah et al. 1988; Gibb & Colquhoun, 1992), the possibility remained that the receptors or ion channels of interest could be altered by the enzymes used and synaptic transmission, of course, could not be studied. Meanwhile, during the 1980s the techniques for recording from brain slices with intracellular electrodes were greatly improving, aided by the development of the discontinuous single electrode voltage clamp (Finkel & Redman, 1985). Several studies were published where simultaneous recordings from pairs of synaptically connected neurones in hippocampal slices were achieved (e.g. Knowles & Schwartzkroin, 1981; Miles & Wong, 1984; Scharfman et al.1990; Sayeret al. 1990; Mason et al. 1991). However, these recordings were still limited by the lower resolution of intracellular microelectrode recording. The problems of cell isolation and signal resolution were overcome by the introduction of techniques to allow patch clamp recordings to be made from brain slices (Edwards et al. 1989, adapted by Blanton et al. 1989) and even from the in vivo brain (Xing Pei et al. 1991; Ferster & Jagadeesh, 1992). The tremendous significance of this advance is seen by the fact that these techniques have rapidly been applied to many fundamental questions in neurobiology such as the mechanism of synaptic transmission at inhibitory (Edwards et al. 1990; Takahashi, 1992) and excitatory synapses (Hestrin et al. 1990; Silver et al. 1992; Stern et al. 1992, Hestrin, 1992a), as well as allowing single channel recording of the properties of receptors and ion channels in identified neurones from the brain (e.g. NMDA receptors, Gibb & Colquhoun, 1991; GABA receptors, Edwards et al. 1990; glycine receptors, A Takahashi & Momiyama, 1991). Patch clamp recording from cells in sliced tissues 257 2. Preparation of brain slices for patch clamp recording Brain slices have been widely used for both biochemical and electrophysiological studies (for review see Alger et al. 1984), and the methods used to prepare brain slices are widely documented (e.g. Langmoen & Andersen, 1981; Cuello & Carson, 1983; Alger et al. 1984; Madison, 1991). Although different labs have their own individual variants, below is a brief description of the procedures that we use, which seem to give good results with a variety of brain areas (see also Edwards et al. 1989; Konnerth, 1990; Edwards & Konnerth, 1992). (i) The animal is decapitated and the brain removed and placed in ice-cold physiological solution within 60 seconds of decapitation (the solution should be so cold that it contains a few ice crystals and to maintain this temperature the container should be sitting on ice). (ii) Pause for 3-5 minutes while the tissue cools down. (iii) Trim or block off the tissue using clean cuts with a sharp scalpel blade in preparation for gluing to the stage of the tissue slicer. Avoid squeezing or otherwise deforming the tissue at this stage. (iv) Apply a thinlayer of cyanoacrylate glue (Super-glueR) to the stage of the tissue slicer and then gently place the tissue at the correct orientation onto the glue (the most common orientation is to have the region of interest near to the blade, or at least try to avoid cutting through white matter before reaching the area to be sliced). Immediately, pour ice-cold solution over the tissue until it is submerged. (v) Cut slices (100-300 m m thick) with vibrating slicer. Fine dissecting scissors or two hypodermic needles can be used to dissect out an area of brain from each whole brain slice. (vi) Using a Pasteur pipette cut and fire-polished to an opening of 3-5 mm across, transfer each slice as it is produced to the holding chamber which should be in a water bath at 32-35°C with a good steady flow of O /CO bubbling through the solution. 2 2 (vii) Incubate the slices at 32-35°C for at least 30 minutes before beginning recordings. Equipment check-list About 250 ml of ice-cold ‘slicing Krebs’ sitting on crushed ice Large scissors for decapitation Small scissors to cut open the skull Curved, blunt forceps to remove top of skull No. 11 scalpel to hemisect the brain Small spatula to remove brain halves from skull Large weighing boat or similar shallow container of ice-cold Krebs sitting on crushed ice to cool the brain halves Cyanoacrylate glue Large spatula to lift blocked-off piece of brain onto tissue block Two fine hypodermic needles or fine dissecting scissors for dissecting small regions from the brain slice Broken and fire-polished pasteur pipette (opening 3-5 mm) 258 A. J. GIBB AND F. A. EDWARDS 3. Notes on making slices The time factor It is critical that the time from decapitation till immersion of the brain in cold solution is kept short (<1 min). Partly because of this, it seems easier to make healthy slices from younger animals (e.g. less than 3 weeks) where the skull is soft and can be removed more rapidly. In addition, the smaller brain of younger animals will cool more rapidly than a larger adult brain and may be more resistant to anoxia. To improve cooling some people remove the skull with the whole head submerged in ice-cold Krebs. Bubbling the Krebs during the cooling period may also improve cooling. The whole process of making slices should preferably not take more that about 30 minutes. However, we have observed that tissue kept ice-cold for half an hour (e.g. the second half of the brain when making hippocampal slices) can still be glued to the slicer and good slices prepared from it. This can be useful if two people wish to slice different parts of the brain or as a backup if something goes wrong during slicing such as the tissue block coming off the slicer stage during slicing (this may happen occasionally although less often with practice: perhaps the block was moist before the glue was applied, or too thick a layer of glue was used, or the glue itself was too thick in consistency, or the surface of the tissue block is not flat). Tissue slicers The tissue slicer should be able to vibrate at sufficient frequency (around 10 Hz) and with a long enough stroke (1-2 mm) to cut cleanly through the tissue. Most commercially available slicers will do this when set at their maximum settings. Care must be taken that there is an absolute minimum of play or vibration in the mechanism driving the cutting blade. Use of a rotating blade for cutting slices, rather than an oscillating blade, has recently been described and a rotating blade slicer is now marketed by Dosaka (Model DTY 8700). However, we have no personal experience of rotating blade slicers as compared to oscillating blade slicers. Depending on the brain area being sliced, it may be useful to view the slicing using a low-magnification dissecting microscope or large magnifying glass (some slicers come fitted with a magnifying glass). It is always useful to have a good light source available to illuminate the tissue block (e.g. using fibre optic light guides). The simplest slicers have a manual movement of the tissue block towards the oscillating blade (e.g. Cambden Vibroslice, UK) and in our experience these work very well for a variety of different brain areas. Some slicers have a Peltier-cooled stage to maintain the tissue close to 0°C during slicing. However, it is perfectly adequate to have frozen Krebs in the bottom of the slicing chamber (or make Krebs ice-cubes) to ensure the tissue stays cool during slicing. Some slicers have a motor drive to advance the blade or tissue during cutting (e.g. Dosaka 1500E, Japan; Camden Vibroslice, UK; FTB Vibracut, Germany; Technical Products Inc. Vibratome 1000, USA). An annoying feature is that some slicers Patch clamp recording from cells in sliced tissues 259 automatically reverse at the end of the cut, when what is often needed is to stop the blade in that position until a piece of tissue of interest is dissected free from the whole brain slice. The Vibracut has a useful innovation in that the tissue bath mounts on a magnet allowing it to be rotated to any angle, which avoids the difficulty of placing the tissue on the glue at exactly the right angle. There are quite a variety of tissue slicers available with the tissue block inside a tissue bath and so suitable for cutting living slices. These slicers vary in sophistication and price. However we find that a simple slicer such as the Camden Vibroslice works very well. If a more sophisticated slicer is preferred, we recommend the vibrating Dosaka slicer or the Vibracut. Slicer blades The blades used for slicing should be as sharp as possible. High-carbon steel blades are preferable to stainless-steel (a high-carbon steel is magnetic and brittle and will break with a sharp snap). Stainless-steel razor blades are probably not as sharp. Slicing different brain areas Different brain areas are more or less difficult to slice and in the original description of the technique a large variety of different brain regions were successfully recorded from (Edwards et al. 1989). As well as taking the age of the animal into account, a high degree of myelination and vascularization of a particular area tends to make slicing more difficult (areas like this seem to require a particularly slow forward speed during cutting). The spinal cord and brain stem are regarded as difficult to slice, particularly in older animals, but in the last few years successful patch clamp experiments have been made with both young (e.g. Takahashi, 1992) and adult spinal cord (Yoshimura & Nishi, 1993) and with several parts of the brain stem (e.g. Forsythe & Barnes-Davies, 1993; Kobayashi & Takahashi, 1993). Getting the right angle Many neurones have their dendritic tree angled in a particular orientation or plane. A big improvement in cell survival is obtained if care is taken to cut the slices at an angle that will preserve the dendrites (e.g. transverse slices of hippocampus to maintain the pyramidal cell apical dendrites). Of course, the angle of slicing could also be important in maintaining synaptic connections. Obviously, if a particular input is to be stimulated then the angle of slicing must be arranged to avoid cutting the incoming axons. Alternatively, it may be necessary to stimulate locally (e.g. with a patch electrode pushed into the slice) if the desire is to stimulate a local interneurone. In general it is harder to obtain healthy, large neurones in slices compared to obtaining healthy small neurones, probably because of the problem of neuronal death if some of the dendrites are cut, but perhaps also due to differences in resistance to anoxia between different cell types. Thus, in hippocampal slices, even when CA1 and CA3 cells look poor, it is often possible to see healthy granule cells. Likewise in cerebellar slices it is more difficult to obtain healthy Purkinje cells than healthy granule cells. 260 A. J. GIBB AND F. A. EDWARDS Slicing solutions Slices are made with a standard extracellular Krebs solution. For example we use (in mM) NaCl 125, KCl 2.5, CaCl 2, NaHCO 26, NaH PO 1.25, MgCl 1, Glucose 25, 2 3 2 4 2 of pH 7.4 when bubbled with 95% O and 5% CO . Although the exact composition 2 2 of the Krebs solution varies between laboratories, particularly in the concentrations of Ca2+, NaHCO and glucose, it is generally considered important that the K+ 3 concentration is less than 3 mM (to avoid epileptiform activity in the slice) and that a high glucose concentration is used. Efforts to improve cell survival during slicing include the substitution of sucrose for 50% of the NaCl in the Krebs (Aghajanian & Rasmussen, 1989), inclusion of Hepes buffer as well as HCO - buffer in the Krebs, raised extracellular Mg2+, use of 3 NMDA channel blockers and excitatory amino-acid antagonists. Slice incubation chamber A good incubation chamber must be able to provide a good circulation of freshly oxygenated solution since the slices must be kept in good condition in the chamber for the whole of the experimental day (10-12 hours). It must be stable not only to prevent mechanical disturbance of the slices but also so that slices can be placed in or removed from the chamber without disturbing any of the other slices. Preferably it should be simple to make and clean. Different laboratories use different types of incubation chamber. Here we describe a simple construction illustrated in Fig. 1 which we find easy to make and use. This incubation chamber uses a standard 100 ml beaker. It contains a piece of light cotton clamped across two rings made for example using the base and lid of a 35 mm Petri dish which have had the top and bottom broken out (Falcon dishes seem to work best). This makes a tight net of cotton on which the slices will rest. The cotton clamp is then wedged halfway down the beaker using a piece of stiff plastic tube about 3-4 cm long. This plastic tube should reach from almost the bottom of the beaker to about 5 mm belowthe surface of the Krebs. A gas bubbler is inserted into the tube to near the bottom and generates a stream of bubbles which by rising to the top of the tube draws the Krebs from the bottom of the beaker, so generating a circulation of Krebs which will act to hold the slices down on the net. The incubation chamber is placed in a heated water bath (a large water tank, 5-10 litres, heated with a standard aquarium heater is sufficient) and covered (e.g. with a Petri dish lid) to prevent evaporation. The incubation chamber can be dismantled every night and reassembled next day with a new piece of cotton (standard white muslin is cheap enough that a meter of material bought at the local drapers shop will last for years!). However, if the chamber is rinsed with distilled water and then left to soak overnight in distilled water acidified with a few drops of HCl, then the same chamber can be used for several days at a time. Immobilizing the slice in the recording chamber It is necessary to immobilize the slice during recording so that no movements occur as a result of the solution flowing through the bath. Typical flow rates would be Patch clamp recording from cells in sliced tissues 261 Fig. 1. A simple incubation chamber for maintaining brain slices. The chamber is made using a 100 ml beaker containing a cotton support which allows the slices to be held in a gentle circulation of oxygenated Krebs solution. The cotton support is made from standard cotton muslin stretched across two rings made from the top and bottom of a 35 mm Petri dish. This is then wedged halfway down the beaker using a stiff plastic tube which extends well below and a little above the cotton support. The tube should reach from near the bottom of the beaker to about 5 mm below the surface of the Krebs. When a bubbler is placed near the bottom of the tube the bubbles rise up the tube drawing solution with them and generating a current which flows down over the slices. (From Edwards & Konnerth, 1992). between 1 ml and 3 ml per minute for bath volumes of less than 1 ml and stability is improved if the inflow and outflow are in separate chambers connected to the recording chamber by small, submerged passages. This has the disadvantage of tending to slow solution exchange around the slice so a good compromise is to have the outflow in a separate chamber and place the inflow on a ramp running directly into the recording chamber. Several methods have been described for immobilizing the slice including the use of fibrin clots (Takahashi, 1978; Blanton et al. 1989) and pieces of netting. However, many people find that a grid (described in Edwards et al. 1989) made of flattened platinum wire with single nylon strands glued across it with cyanoacrylate glue works well for holding the slice firmly on the bottom of the recording chamber. 4. Visualizing cells in living brain slices How healthy is the slice? In the past the health of the slice was generally determined from physiological parameters such as resting membrane potential of impaled cells, size of the action potential, population spike etc. Comparison of these parameters measured in vitro with the same parameters measured in vivo suggests that healthy sliced brain tissue behaves in a remarkably similar way to the in vivo state. However, in the past only a few studies used high-resolution optics to allow cells in slices to be visualized 262 A. J. GIBB AND F. A. EDWARDS (Yamamoto, 1975; Takahashi, 1978; Llinas & Sugimori, 1980) so that a direct visual assessment could be made of the health of the sliced tissue. One great advantage of visualizing the cells in slices is that it allows an immediate assessment of the health of the slice to be made and cells suitable for recording to be carefully picked. The first examination of a healthy slice under the microscope is a fantastic sight! In a hippocampal slice for example, lots of bright shiny cells should be visible with a variety of cell dendrites and different cell morphologies present. Compared to blindly inserting the electrode into the slice, a great deal of time can be saved by first picking out the good cells particularly, for example, if the cell of interest is a relatively rare interneurone. If the slice does not contain many bright cells, but instead is uniformly dark with many round opaque cells with clearly visible nuclei evident, then the slice should be discarded (see also Edwards & Konnerth, 1992 for a discussion of visually assessing the slice). Labelling and identifying cells It was partly as a result of the desire to record from identified cells (Takahashi, 1978) that the techniques for patch clamping visually identified cells in brain slices were developed (Edwards et al. 1989). Retrograde transport of fluorescent dyes or fluorescent beads (e.g. Takahashi, 1978; Katz et al. 1984; Gibb & Walmsley, 1987) has been used to allow subsequent identification of living cells in slices or following dissociation. However, these procedures can only be used where cells have a definite projection (e.g. motor neurones) and require expensive fluorescent optics on the microscope. Instead, cell bodies and parts of the dendritic tree can be easily observed using differential interference contrast (DIC) Nomarski or Hoffmann modulation optics. The identification of the cell to be studied then depends on the use of information about the local anatomy, and the size and morphology of the cells of interest. For most purposes this is sufficient to identify a cell clearly. Microscope requirements The particular brand of microscope used is not critical. Zeiss, Olympus and Nikon all make upright microscopes which can be used for visualizing cells in slices (Micro- Instruments in Oxford make a good customized microscope fitted with Nikon optics). Ideally the microscope should have a fixed stage so that focusing occurs without moving the preparation relative to the patch electrode. Although the standard Olympus BHS is not a fixed-stage microscope, it can easily be converted and Olympus will now do this conversion if requested. This is a cheap and satisfactory option. It is also important that the microscope is not mounted on rubber feet but instead is firmly fixed to the vibration isolation table or the electrode will crash into the slice every time the focus is adjusted! Use of a high numerical aperture (e.g. 0.75) · 40 water immersion objective on a standard upright microscope preferably fitted with Nomarski optics allows visualization of neurones and their dendrites with a resolution of about 1-2 m m, if the cell lies within 20 m m of the surface of the slice (looking from above). For Nomarski optics to be effective, however, the maximum Patch clamp recording from cells in sliced tissues 263 slice thickness is around 300 m m. The thicker the slice, the more the light is scattered passing through the slice and the dimmer and lower the resolution of the image. On the other hand, it is more difficult to obtain healthy thin slices: slices 200-300 m m are usually a good compromise. The choice of objective is a compromise between the need for high resolution (high numerical aperture) and the need for a reasonable working distance (at least 1.5 mm) to allow access to the surface of the slice with a normal patch electrode. The Zeiss · 40 achromat (numerical aperture 0.75, working distance 1.9 mm) fitted to the Zeiss Axioskop is a good example. In most countries the Zeiss Axioskop is considerably more expensive than the Olympus BHS fitted with the newly released Olympus 40· water immersion objective (NA 0.7, WD 3 mm). Unlike the Axioskop, the Olympus does not have infinity-corrected optics and so the new Zeiss and Olympus objectives are not interchangeable. There is also a Nikon 40· water immersion objective (NA 0.55, WD 2 mm) but, although this is a little cheaper, the image resolution seems to be not as good presumably because of the lower numerical aperture. The Zeiss Axioskop microscope gives excellent image quality. This may be partly because infinity-corrected optics are superior to standard optics (at least in principle) but could also be the result of a very stable condenser and Nomarski system combined with a very good light source. For patch clamping very small cells it may be an advantage to fit the microscope either with an octovar giving variable intermediate (1.0· , 1.25· and 1.6· ) magnification or 16· eyepieces to give an overall magnification of more than 600· . In principle, it might be expected that, when the water-immersion objective is in contact with the bath solution, a ground loop will occur because the objective will also be in electrical continuity with the rest of the microscope which is usually earthed. In practice we know of varying experiences on this where some objectives did, and some did not need insulating from the microscope, perhaps because some objectives are coated, which effectively insulates them anyway. If necessary, a solution is to manufacture an insulating collar to insert between objective and nose- piece. It should be noted that for best results the numerical aperture of the condenser lens should always be as high or higher (0.9 for example) than that of the objective. This generally means that the working distance of the condenser will allow only a thin glass cover slip or glass base for the recording chamber, if the light from the condenser is to be focused properly on the surface of the slice (plastic chambers, although fine for phase contrast optics, destroy Nomarski imaging). Whatever the precise optical arrangement it is essential for best results that good microscopic practice is followed (see e.g. Bradbury, 1989). In particular, good Köhler illumination must be set up with the condenser adjusted to focus the light source diaphragm exactly in the plane of the cells of interest. For DIC optics, the polarizers should be 90° to each other and the analyzer adjusted for optimum image quality. Secondary diaphragms in the condenser are then used to cut down the light entering the tissue and so improve the sharpness of the image. It is often less tiring to view the image using a CCD camera in combination with a 264 A. J. GIBB AND F. A. EDWARDS standard monitor. These are relatively cheap (e.g. from Radio Spares) and the smaller cameras (<300 g) are light enough to mount directly on top of the microscope trinocular head without applying too much weight to the microscope focusing mechanism. It is generally important to ensure that the camera is insulated from the microscope to avoid conducting interference into the patch clamp signal. When using a CCD camera, a better image may be achieved if the secondary diaphragms on the condenser are left open and the gain and contrast of the camera controller used to optimize the image, although this will tend to make the image down the eye pieces look very washed-out (Levis & Rae, 1992). Dodt & Zieglgänsberger (1990) have described the use of Nomarski optics in combination with an infra-red filter placed in the normal light path to give infra-red DIC imaging of cells in brain slices. The infra-red image is then visualized with an infra-red-sensitive CCD camera (specialist infra-red CCD cameras are expensive but even an ordinary CCD camera is quite sensitive to infra-red light up to about 1000 nm wavelength) and the image can then be stored on video tape or on computer using a frame grabber. Analogue or digital image enhancement techniques can then be applied to the image. The improved resolution achieved with infra-red DIC may be partly due to reduced scattering of infra-red light during transmission through the slice. Although the infra-red imaging may add considerably to the cost of the microscope, it allows imaging much deeper in the slice and may be useful for specialist applications such as patching directly on to dendrites in slices (Stuartet al. 1993). 5. Recording from cells in slices Electrodes The electrodes used for patching cells in slices are fabricated in the normal way. Thick-walled glass and coating with SylgardR are useful in minimizing the noise associated with the fact that the electrode is immersed quite deep in solution under the objective. For clamping large or fast currents where it is important to minimize the series resistance, it may be better to use thin-walled glass. The choice of glass can make a big difference with a thick-walled Aluminosilicate glass (e.g. Clark Electromedical SM150F 7.5) having a much lower noise than a thin-walled borosilicate glass (e.g. Clark Electromedical GC150TF 7.5). Rae & Levis (1992) discuss in detail a wide range of glass types for patch clamping. Selecting a healthy cell Selecting the best cell for patch clamping requires experience of the particular brain slice in use under the conditions presented by the way the microscope is adjusted. A good guide, however, is that the cells should be smooth with a clear outline and have a ‘soft’ appearance (see also Edwards & Konnerth, 1992). Cells that appear very shiny or ‘hard’ in appearance do not make seals easily and if observed for some time, appear to die gradually. Dead cells are opaque with visible nuclei and are often