当前位置:首页 >> 农林牧渔 >>


Pflugers Arch - Eur J Physiol (2002) 444:491–498 DOI 10.1007/s00424-002-0831-z


Troy W. Margrie · Michael Brecht · Bert Sakmann

In vivo, low-resistance, whole-cell recordings from neurons in the anaesthetized and awake mammalian brain

Received: 9 January 2002 / Revised: 15 February 2002 / Accepted: 19 February 2002 / Published online: 17 April 2002 ? Springer-Verlag 2002

Abstract A blind patch-clamp technique for in vivo whole-cell recordings in the intact brain is described. Recordings were obtained from various neuronal cell types located 100–5,000 ?m from the cortical surface. Access resistance of recordings was as low as 10 M? but increased with recording depth and animal age. Recordings were remarkably stable and it was therefore possible to obtain whole-cell recordings in awake, head-fixed animals. The whole-cell configuration permitted rapid dialysis of cells with a calcium buffer. In most neurons very little ongoing action potential (AP) activity was observed and the spontaneous firing rates were up to 50-fold less than what has been reported by extracellular unit recordings. AP firing in the brain might therefore be far sparser than previously thought. Keywords In vivo · Whole-cell · Olfactory bulb · Barrel cortex · Evoked · Spontaneous · Awake · Anaesthetized

Classically, neuronal activity in the intact nervous system has been documented by extracellular unit recording [17] or by impalement of neurons with sharp microelectrodes [6, 11]. More recently, a variant of the whole-cell patch technique has also been used for in vivo neuronal recording [19, 27, 36]. Patch-clamp recordings were developed originally to study single-channel currents in biological membranes [13], but the whole-cell variant of this technique has been applied to examine the intrinsic [10, 32] and synaptic properties of neurons in vitro [2, 29]. Using the whole-cell recording technique to
T.W. Margrie and M. Brecht contributed equally T.W. Margrie (?) · M. Brecht · B. Sakmann Abteilung Zellphysiologie, Max-Planck-Institut für medizinische Forschung, Jahnstrasse 29, Heidelberg 69120, Germany e-mail: tmargrie@sun0.mpimf-heidelberg.mpg.de Tel.: +49-6221-486249, Fax: +49-6221-486459

examine electrical activity of preparations in vitro is nowadays a standard procedure and methodological aspects of this technique have been discussed in detail [2, 10, 13, 29]. The application of the whole-cell recording technique to in vivo preparations is still much less common and the goal of this paper is to discuss this methodology and the potential it may hold for studying neuronal activity in the intact brain. Several aspects of the in vivo, whole-cell recording technique are highlighted including: (i) factors contributing to low access resistance; (ii) recording stability; (iii) perfusion rate of cells and (iv) differences between results obtained with the whole-cell recording technique and those obtained with extracellular unit recording. Although in most instances the access resistance of recordings is not reported explicitly, in vivo whole-cell recordings often appear to suffer from the fact that only relatively high resistance electrical access to the recorded cell is obtained [27]. Since the capacitance of the patch pipette together with the access resistance act as a low-pass filter for voltage signals, it is important to minimize access resistance to reduce such filtering. Low access resistance is also advantageous for intracellular loading with neuronal markers such as biocytin, Ca2+ indicators [14] and pharmacological agents [25]. We provide below a detailed description of the factors determining access resistance. An inherent problem of neuronal recordings in vivo is mechanical instability associated with respiratory and heartbeat movement in the preparation. We therefore also show that whole-cell recordings are mechanically very stable and can thus be employed in awake animals in which instability hinders intracellular recording [12]. Whole-cell recordings lead to the dialysis of the recorded neuron with the internal pipette solution. While this can be exploited to load cells with various agents, important internal constituents of the cell can be washed out or diluted [28]. With sequential cell-attached and whole-cell recordings we also examined the effects of dialysis on the on-going and evoked suprathreshold activity of the recorded neurons.


Materials and methods
Animal preparation Mice (P21–35) and Wistar rats (P16–70) (where P indicates the animal’s age in days postnatal) were anaesthetized with urethane (1.2 and 1.4 g/kg i.p., respectively). All pressure points and incised tissues were infiltrated with lidocaine. Throughout all experiments rectal temperature was monitored and maintained at 37±0.5 °C by a homeothermic blanket (FHC, Bowdoinham, Me., USA). For rat barrel cortex experiments a craniotomy (0.75× 0.75 mm) above the somatosensory cortex was performed and the dura removed. Typically, electrodes were arranged to penetrate perpendicular to the cortical surface at the centre of the vibrissae/ barrel cortex. At the end of each neuronal recording, the subpial depth of the cell was estimated from the distance that the micromanipulator had advanced. Similar surgical procedures were used to record from the rat thalamus and the mouse olfactory bulb. Preparation for awake recordings All recordings in awake animals were performed in the somatosensory cortex of 4- to 7-week-old rats. Rats were handled extensively and familiarized with the recording environment several days prior to surgery. Water containing sucrose was offered constantly as a positive reward throughout this training period and during recording sessions. We used standard surgical techniques to prepare animals with implants for chronic experiments (1–3 weeks of recordings). Rats were implanted under general anaesthesia induced by ketamine and xylaxine (10 and 2 mg/kg i.m., respectively). Dental acrylic and 0.5 mm screws were used for fixation of the implant on the skull. A capped recording cylinder was fixed over a small trepanation above the barrel cortex and a head fixation bolt was included in the dental acrylic implant. Recording sessions began 5–7 days after recovery from the surgery and once the animals were accustomed to the experimental apparatus and head fixation. For recording sessions rats were head-fixed by connecting the head fixation bolt to a metal rod. The recording cylinder was opened and the dura removed. For multiple recording sessions, the recording cylinder was resealed and the animals intermittently re-housed. Electrophysiology For pipette fabrication we used both filament (0.25 mm) and nonfilament borosilicate glass (OD 2.0 mm, ID 1.5 mm, Hilgenberg, Malsfeld, Germany). The standard intracellular solution (internal A) was (in mM): K-gluconate 130, Na-gluconate 10, HEPES 10, phosphocreatine 10, MgATP 4, GTP 0.3, NaCl 4 and biocytin 0.3–1%, pH 7.2. In experiments in which the patch pipette was perfused with a high concentration of the calcium buffer BAPTA, the BAPTA internal solution (internal B) contained (in mM): K-BAPTA 80, Na-gluconate 10, HEPES 10, phosphocreatine 10, MgATP 4, GTP 0.3, NaCl 4 and biocytin 1–4%, pH 7.2. Tetrapotassium BAPTA was the charge carrier (instead of the normal Kgluconate) to maintain correct osmolarity (300 mOsm) and ionic balance of the internal solution. The junction potential difference of the two internal solutions (+7 mV and +29 mV respectively) was corrected according to [24] so as to determine the resting membrane potential (RMP) before and after perfusion with internal B. For perfusion experiments we adopted an approach previously reported [33] whereby a second port of the pipette holder was attached to a syringe pump for loading of a second internal solution. Through this second port, silicon tubing was attached to a quartz glass filament pulled over a propane flame to produce a diameter of less than 2 ?m. The tip of the quartz filament was positioned 300–500 ?m from the electrode tip. The surface of the cortex was superfused with standard Ringer’s solution (in mM): NaCl 125, KCl 2.5, NaHCO3 25, NaH2PO4 1.25, MgCl2 1, glucose

25, CaCl2 2. The agarose (Sigma) solution sometimes used was 1% in PBS (pH 7.2). Recordings were amplified using an appropriate amplifier (Axoclamp-2B, Axon Instruments, Foster City, Calif., USA), filtered at 3–10 kHz and digitized at 5–20 kHz (ITC16; Instrutech, New York, N.Y., USA) using HEKA (Lambrecht, Germany) software. Histology Following the completion of physiological recordings animals were perfused transcardially with 0.1 M PBS followed by a solution of 4% paraformaldehyde. The brain was removed from the skull and immersed in fixative for at least one more day, after which it was sectioned coronally in 100-?m-thick slices. Slices were then processed with the avidin-biotin-peroxidase method [15]. Whisker stimulation A piezoelectric bimorph wafer with an attached glass capillary served for quantitative single-whisker stimulation [31]. Steps elicited by the piezoelectric device had a 10–90% rise time of 1 ms. The deflection point was chosen to be 8–10 mm from the base of the vibrissa; the deflection amplitude and duration were 1 mm (roughly 6° deflection angle) and 200 ms respectively.

Technique for patch-clamp recordings from neurons in vivo Preparation Preparation of the cortical surface contributed significantly to the likelihood of successful recordings. Small craniotomies helped reduce heartbeat- and respirationinduced pulsation of the cortex. Excessive movement hindered gigaohm seal formation between pipette and neuron. The thorough removal of dura without bleeding was also of utmost importance for avoiding contamination of the patch pipette as it penetrated the tissue. Pipettes We used conventional, low-resistance patch pipettes (4–7 M?) [3, 27, 30] to obtain whole-cell recordings in vivo up to 5 mm from the brain surface. Typically, the tip size was approximately 1–2 ?m. The length of the shank was increased for deeper recording sites such as the thalamus. Voltage-clamp recordings In contrast to previous studies [7, 27, 36], we used the voltage- rather than the current-clamp mode to search for cells in vivo. In the voltage-clamp mode the evoked current in response to a square voltage step (10-mV, 10-ms steps, 4–10 Hz) was monitored and used as an indicator of pipette resistance, which increases when the pipette


interacts with neuronal membranes. This results in a corresponding decrease in the evoked current pulse amplitude. Searching for cells in the voltage- rather than the current-clamp mode offers two advantages. First, in voltage-clamp the membrane patches can be clamped to hyperpolarized command potentials to facilitate the formation of a gigaohm seal [13]. Second, the holding current required to clamp the cell to the holding potential is recorded online, thus offering a quantitative and immediate assessment of seal resistance between pipette and membrane. Pipette insertion Previous descriptions of whole-cell recording in vitro [2, 10, 29, 32] have indicated the importance of cleaning the membranes prior to recording. This facilitates the formation of tight seals and is usually achieved by application of positive pressure to the recording pipette [2, 29, 32]. In our in vivo experiments positive pressure (100–200 mbar) was applied to the pipette interior as it was lowered down to the cortical surface. Initially all artificial cerebrospinal fluid (ACSF) was removed from the cortical surface to establish when the pipette tip encountered the surface of the brain for referencing depth measurements. The surface was then superfused with ACSF to ensure that it remained moist. For successful recordings it was essential to assess the pipette resistance continuously while the pipette was being lowered to the region of interest. The resistance of each electrode was noted prior to entering the brain. With high positive pressure (100–200 mbar), pipettes were advanced through the tissue. We usually observed a DC shift of up to 0.5 nA and a transient increase in pipette resistance (up to 50%) while the pipette was penetrating the pia. Successful recordings were most likely to occur if the pipette resistance returned to the initial value. Establishing whole-cell recordings After arriving at the depth of interest, the positive pressure was reduced to 25–35 mbar and the pipette advanced in 2- to 3-?m steps (Fig. 1A). When a sudden decrease in the current pulse amplitude was observed (20–50%) the positive pressure was removed. The best predictor of the pipette having made contact with a neuronal membrane was pulsation of the reduced current pulse at heartbeat frequency (Fig. 1B). Slow changes in current pulse amplitude that lacked the rhythmic modulation rarely resulted in neuronal recordings. In general, removal of positive pressure resulted in a spontaneous seal of 1–10 G? resistance. However, sometimes suction up to 200 mbar was required to obtain tight seals. During seal formation the command potential was taken slowly to a more hyperpolarized potential (down to –70 mV; Fig. 1C) and the holding current monitored to assess the seal resistance. After achieving the cell-attached configuration, the patch of

Fig. 1A–D Establishing whole-cell recordings in vivo. Once the electrode has penetrated the upper cortical layers and is positioned at the target depth, the positive pressure (>200 mbar) on the pipette is reduced to approximately 30 mbar. The electrode resistance should be monitored to ensure that the pipette has remained debris-free during the initial positioning phase (A). Step sizes of 2–3 ?m were used to search for cells. A “strike” is apparent from the immediate and phasic shift in electrode impedance locked to the heartbeat frequency (B). Removing the remaining positive pressure and slowly clamping the patch to a hyperpolarized potential usually results in a gigaohm seal (C–D). The holding potential here is –70 mV and the seal resistance is approximately 10 G?. A slow ramp of negative pressure (up to 100 mbar) is applied to the pipette to achieve the whole-cell configuration

membrane was broken by applying a slow ramp of negative pressure (up to 20–250 mbar) to obtain the wholecell configuration (Fig. 1D). A slow ramp of negative pressure consistently yielded a higher success rate than steps or rapid ramps of negative pressure. Success rates The overall success rate of in vivo recordings was markedly lower than that of visually guided recordings in brain slices. Success rates of obtaining whole-cell recordings varied substantially from preparation to preparation and decreased with the age of the animal. In 4- to 5-week-old rats and mice the highest rates of success were in the order of one whole-cell recording for every second patch pipette. Under these optimal conditions gigaohm seals were obtained on approximately 80% of occasions. However on average one should expect success rates closer to 20% rather than 50% of penetrations. Success rates below 10% usually indicated that experimental parameters, such as pipette geometry or the preparation, were suboptimal. With the patch pipettes employed here we would expect to encounter a neuron in the rat barrel cortex within 50–100 steps of the electrode. Passive membrane properties The membrane properties of the recorded neurons differed in the various brain structures. The input resistanc-


Recording stability It is well known that gigaohm seals are mechanically stable [13]. As alluded to in previous studies, whole-cell recordings offer great stability for in vivo recordings [27], presumably due to the initial step of gigaohm seal formation. This is consistent with our observation that gigaohm seal formation is associated with the disappearance of heartbeat- and respiration-associated movement artefacts. Before gigaohm seal formation heartbeat related changes in resistance were observed in virtually every electrode (Fig. 1B) whereas in hundreds of recordings we never observed such movement artefacts following the successful formation of a gigaohm seal (Fig. 1B, C). The stability of the recordings obtained using the technique described above permitted us to record for prolonged periods (up to 2 h, with two-thirds of our recordings lasting longer than 30 min). The stability and longevity of the recordings permitted us to record evoked responses to stimulation of the entire whisker array [4] in both anaesthetized and awake, behaving animals. Whole-cell recordings in awake, head-fixed animals were stable, lasting up to 30 min (see also [7]). Although some recordings were lost due to sudden and rapid movements of the animal, it was often possible to maintain the recording when the animal performed behaviours including licking, whisking or even grooming of the head with the paws. An example of whiskerevoked responses recorded in an awake animal is shown in Fig. 3. During this recording we were able first to determine the principal whisker (PW), (D2 in this case), by hand-mapping, and then to deflect the PW and most of the surrounding whiskers in a quantitative and controlled manner with a piezoelectric stimulator. These procedures not only required several minutes of recording time but also involved touching the animal with various somatosensory probes. Rapid perfusion of cells Whole-cell recordings permit rapid perfusion of cells with various agents. We investigated the time course of intracellular loading by perfusing mitral cells of the olfactory bulb with two internal solutions. We took advantage of the fact that back-propagating action potentials (APs) in mitral cell dendrites result in dendritic glutamate release onto the dendrites of inhibitory interneurons. These in turn release GABA back onto the mitral cell dendrite, leading to a recurrent IPSP [20, 22] (Fig. 4). As with axodendritic transmission, the release of vesicular glutamate from mitral cell dendrites is dependent on rises in intracellular Ca2+ [5, 18] and is therefore prevented by buffering intracellular Ca2+ [18]. We first patched a mitral cell using the control internal solution (internal A). After establishing a baseline period of the amplitude of the recurrent IPSP evoked by somatically initiated APs under control conditions, we perfused the patch pipette with a second internal (internal B) con-

Fig. 2 The estimated series resistance for whole-cell recordings in vivo is age and recording depth dependent. Recordings from mitral and granule cells of the mouse main olfactory bulb (P21–35, open triangles), L2/3 and L4 of rat barrel cortex (P21–36, open circles; P50–70, filled circles) and young rat thalamus (P16–25, hatched circles) (where P is the animal’s age in days postnatal). The plot shows the estimated series resistance as a function of electrode depth (i.e. distance electrode has moved through tissue) and animal age. The mean (±SE) recording depths and series resistance were: mouse olfactory bulb, 299±23 ?m, 32±6 M?, n=21; mature rat cortex, 543±32 ?m, 47±3 M?, n=47; adult rat cortex 873±91 ?m, 119±12 M?, n=10; young rat thalamus 4630±18 ?m, 62±3 M?, n=48. The series resistance for all groups was significantly different from one another (P<0.05, Student’s t-test)

es and resting membrane potentials for regular spiking cells in somatosensory cortex were 79.5±30.2 M? and –65.1±4.3 mV in the ventral posterior medial thalamic nucleus (VPM, n=31), 39±4 M? and –81.8±6.3 mV in cortical layer 4 (n=23) 61±12 M? and RMP=–83.8± 5.2 mV in cortical layers 2/3 (n=31), 470±36 M? and –72±2 mV for olfactory bulb granule cells (n=42) and –57 mV±2 mV and 116±6M? (n=48) for mitral cells. Factors determining series resistance After achieving the whole-cell configuration in the voltage-clamp mode, the amplifier was switched to the current-clamp mode for capacitance neutralization and balancing the bridge. The access resistance of our recordings depended on the recording depth and the age of the preparation (Fig. 2). In young animals we obtained recordings with low access resistance (10–50 M?) while access resistance in animals older than P50 was typically around 100 M?. The effect of recording depth on the access resistance can be appreciated by comparing recordings from the VPM (approximately 4,500 ?m below the surface) and cortex (L2/3 and L4 100–900 ?m). Despite the thalamic recordings being carried out on a younger sample of animals the access resistance was consistently greater than that observed for more superficial recordings. The higher access resistance of deep recordings was probably related to contamination of the pipette tip [14] and could be reduced by applying extra-high positive pressure during lowering of the pipette or advancing the pipette to the target structure through a guide tube.


Fig. 3A–E Whole-cell recordings from barrel cortex in the awake rat. A Schematic representation of the whisker (w) arrangement on the rat’s snout. B Two successive responses to backwards deflection of the principal whisker (PW) (whisker D2). C Two successive responses to backwards deflection of the surround whisker (whisker D3). On and off stimulus artefacts are seen in all records as small deflections. The cell showed stronger responses to stimulus off than to stimulus on, possibly due to preference for forwards whisker deflection. D, E Average (ten trials) response amplitudes for stimulus off responses. The labelling of the plots corresponds to the whisker arrangement in A. D Sub-threshold responses to stimulus off. E Action potential (AP) responses to stimulus off. Whisker positions indicated by the grey colour in D and E refer to the responses shown in B and C

Fig. 4A, B Rapid perfusion of cells with whole-cell patch pipettes. A A plot of the amplitude of the recurrent IPSP evoked in a mitral cell as a function of time. After establishing a stable baseline recording of the recurrent IPSP in a cell, the electrode was perfused with a second internal solution (internal B), containing BAPTA (see Materials and methods). B Within 5 min the calciumdependent release of glutamate from the mitral cell (as measured by the recurrent IPSP) was prevented. The numbers 1 and 2 correspond to the single sweeps shown below recorded at the times indicated

Whole-cell recordings indicate low rates of AP activity Consistent with previous whole-cell recording studies the spontaneous and evoked AP rates in the somatosensory system [4, 23, 36] and the olfactory bulb [22] were substantially lower than those reported in extracellular unit recording studies. For example, in our recordings the average frequency of spontaneous AP activity of 22 regular-spiking and 2 fast-spiking, layer-4 neurons in the rat somatosensory barrel cortex was 0.056 Hz. Unit recording studies employing the same anaesthetic have reported spontaneous firing rates of 0.8–1.54 Hz [1, 16]. Unit recording studies also have reported much higher rates of stimulus-evoked AP activity than have wholecell recording studies. In our sample of whole-cell recordings from layer-4 neurons the average rate of evoked APs per PW-stimulus was 0.15, much less than the 1 and 1.4 APs per PW-stimulus reported by unit recording studies [1, 9] despite our use of a stronger whisker stimulus (6° compared with 1.1° deflection amplitude). As with anaesthetized animals, whole-cell recordings of cortical neurons in awake animals generally indicated lower rates of AP activity compared with reports using extracellular unit recording techniques (average rates of spontaneous activity <1 Hz). The low estimates of AP activity based on whole-cell recording are not specific to cortical layer 4 but are observed in other layers of rat somatosensory cortex [23, 36]. Likewise in the olfactory bulb extracellular single-unit recordings under urethane anaesthesia have reported spontaneous mitral cell firing

taining the Ca2+ chelator BAPTA. Within 5 min (4.2± 0.45 min, n=3) of perfusion with internal B, the recurrent IPSP was abolished (–8.2±1.9 mV vs. 0.2±0.3 mV, n=3; Fig. 4A). The blockade of the IPSP was paralleled by slowing of the after hyperpolarising potential (data not shown) and is consistent with effective buffering of the AP-evoked Ca2+ influx by the BAPTA containing internal [18]. These low-resistance, whole-cell recordings therefore permit rapid loading of neurons in vivo. The percentage of morphologically identified cells recovered by the avidin-biotin-peroxidase method [15] after recordings varied considerably across processing sessions within the range of 40–100%. The likelihood of recovering the cell morphology was improved by taking a number of steps: (i) limiting the recording time to less than 30 min; (ii) slow withdrawal of the patch pipette and (iii) rapid transcardial perfusion of the animal following termination of the experiment.


might have been the result of a larger number of nonspiking cells in the sample of the cortical whole-cell recordings. This prominent difference in AP firing rates between extracellular unit and whole-cell recording prompted us to examine whether low rates of AP activity resulted from wash-out of cells in the whole-cell configuration. To this end we quantified the spontaneous and evoked firing rates in the cell-attached mode before establishing the whole-cell configuration (Fig. 5). Regular spiking (RS) neurons in rat barrel cortex, regardless of whether recorded in the cell-attached or whole-cell configuration showed little or no spontaneous AP activity. Figure 5A shows that in fact the spontaneous activity of a given cell in the whole-cell mode of recording was actually significantly higher than in the cell-attached recording mode. Similarly, low rates of sensory-evoked AP activity were observed in both the cell-attached (CA) and whole-cell (WC) configurations (Fig. 5B, C). In the few fast-spiking (FS) neurons (putative inhibitory neurons) we recorded from, the whole-cell configuration reported a dramatically elevated level of spontaneous (CA 0.619 vs. WC 3.49 Hz, n=5) and evoked (CA 0.28 vs. WC 3.6 APs/ stimulus, n=3) firing rates when compared to the cellattached mode. It is therefore unlikely that washout of the internal constituents of the cell can explain the comparatively low firing rates observed during whole-cell recordings. If anything, the whole-cell recording configuration may over-estimate the AP rates of cortical neurons in vivo.

Fig. 5A–C Cell-attached and whole-cell recordings of spontaneous and evoked activity in regular spiking (RS) neurons of the rat barrel cortex. A Left, upper trace: recording from a urethaneanaesthetized rat (P20) 735 ?m below the pial surface. A representative trace of spontaneous AP activity recorded in the cellattached configuration. Traces were recorded in the voltage-clamp mode while the membrane patch was held at 0 mV. Left, lower trace: a single trace showing the spontaneous AP activity after establishing the whole-cell recording configuration. Right: grouped data showing the mean spontaneous AP firing rate for regular spiking cells (n=28) during cell-attached (CA) and whole-cell (WC) recording (P=0.027). B Left, upper trace: representative traces in voltage clamp of evoked AP activity in the cell-attached configuration recorded from the cell in A. Left, lower trace: a representative trace of membrane voltage (Vm) showing the spontaneous AP activity after establishing the whole-cell recording configuration. Right: group data of evoked AP rates of seven RS cells during cell-attached and whole-cell recording (P>0.05). C A population peristimulus time histogram (PSTH) of the whiskerevoked activity for all RS cells recorded in the cell-attached and whole-cell recording mode. The data for the fast-spiking cells are not included in the PSTH. The bin width of the PSTH is 5 ms

Here we have described a technique for obtaining stable, low-resistance, whole-cell recordings from neurons in the mammalian brain. In contrast to previous studies [7, 25, 27] we used the voltage-clamp technique for seal formation and obtaining the whole-cell configuration. It is known that voltage clamping to negative potentials facilitates the formation of seals [26] and thus provides increased recording stability. The stability provided by this technique is perhaps the most attractive aspect for in vivo intracellular recording. As shown above, one of the trademark characteristics of the “strike” of the pipette against neuronal material is pulsation of the recorded current at heartbeat frequencies. In our experience this is the best indicator of the patch pipette making contact with neuronal material. While there were instances in which this pulsation was due to contact with non-neuronal membranes, presumably glia or blood vessels, this occurred less than 5% of the time. We also observed that dampening the cortical pulsation with topographical application of agar reduced the rhythmic changes in electrode resistance associated with heartbeat and respiration during seal formation. To preserve the integrity of the cortical surface it was advantageous not to remove the agar with each new pipette but rather leave the agar in place throughout the experimental procedure. Advancing

rates up to 30 Hz [21, 34, 35]. This is in marked contrast to the 0.6±0.1 Hz (n=26) observed using whole-cell recordings. While the AP rates between cortical extracellular unit and whole-cell recordings differ 10–50 fold, only a twofold lower firing rate was observed in cells of the somatosensory thalamus. This region specific difference


the patch pipette through the agar did not interfere with seal formation. Importantly, the topographical application of agar provides greater stability of brain structures for in vivo imaging studies [14]. The access resistance of recordings increased with electrode depth and the age of the animal. Since it is well established that a clean patch pipette improves the likelihood of tight seal formation [13] we assume that as the electrode moved through more material along its trajectory for deep recordings the tip would become increasingly contaminated by various membranes. The number of pipettes required to obtain gigaohm seal formation also increased markedly with recording depth. As with whole-cell recording in slices, the recording quality deteriorated with increased age of the preparation. Whether this is due to developmental changes in membrane structure remains to be determined. We have shown that these low-resistance, whole-cell recordings in vivo enable rapid perfusion of internal solutions. This is highly advantageous for fluorescence imaging, recovering cell morphology and, as shown above, pharmacological manipulation of the biophysical properties of single cells. Whole-cell recordings provided substantial continuance and could even be employed under conditions of increased mechanical instability, such as in awake, headfixed animals [7]. We assume that this recording stability results from the mechanically stable connection between patch pipette and cell membrane [13]. It is well known that once a gigaohm seal has been established, the patch pipettes can be moved many microns away from the cell without compromising the attachment between membrane and pipette [13]. In our view the stability and endurance of the whole-cell recording is one of the most promising aspects of this recording technique. It will allow assessment of subthreshold activity under conditions in which previously exclusively extracellular recordings have been employed. Finally, we showed that with whole-cell recordings the firing rate of cortical neurons was significantly lower compared with previous reports using extracellular unit recordings. This result was not due to washout of intracellular constituents as cell-attached recordings showed even lower firing rates. One possible explanation for the comparatively increased firing rates in the whole-cell recording mode is that a leak conductance might be generated when achieving the whole-cell configuration. Such a leak conductance may lead to a more depolarized membrane potential and thus an elevated firing rate compared with the cell-attached firing rates. Since we compare our whole-cell firing rates with unit data recorded in animals under the same anaesthesia, it seems the overall reduced firing rates observed with whole-cell recordings is unlikely to be accounted for by significantly different brain states. One hypothesis that accounts for the observed difference between the whole-cell and extracellular single-unit recorded firing rates is that there exists a significant sampling bias of the unit recordings against neurons that do not show high rates of spontaneous or evoked AP activi-

ty. In fact we observed that the difference in estimates of AP activity between whole-cell and unit recording correlated well with the percentage of silent cells encountered in different brain regions using the whole-cell technique [1, 4, 8, 9]. In addition, our recordings from mitral cells suggest that the differences in activity estimates cannot be explained simply by sampling biases associated with different cell types [21, 34, 35]. Given the magnitude of the differences in activity reported with whole-cell and unit recording techniques it will be imperative to resolve the origin of such discrepancies in order to develop a quantitative understanding of suprathreshold activity in the mammalian brain.
Acknowledgements We would like to thank Fritjof Helmchen for comments on the manuscript. This work was supported by the Max-Planck Geselleschaft, Alexander von Humboldt Foundation and the NHMRC of Australia.

1. Armstrong-James M, Fox K (1987) Spatiotemporal convergence and divergence in the rat S1 “barrel” cortex. J Comp Neurol 263:265–281 2. Blanton MG, Lo Turco JJ, Kriegstein AR (1989) Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. J Neurosci Methods 30:203–210 3. Borg-Graham LJ, Monier C, Fregnac Y (1998) Visual input evokes transient and strong shunting inhibition in visual cortical neurons. Nature 393:369–373 4. Brecht M, Sakmann B (2002) Whisker maps of neuronal subclasses in the rat ventral posterior medial (VPM) thalamus identified by whole-cell voltage recording and morphological reconstruction. J Physiol (Lond) (in press) 5. Chen WR, Ziong W, Shepherd GM (2000) Analysis of relations between NMDA receptors and GABA release at olfactory bulb reciprocal synapses. Neuron 25:625–633 6. Coombs JS, Eccles JC, Fatt P (1955) Excitatory synaptic action in motoneurones. J Physiol (Lond) 130:374–395 7. Covey E, Kauer JA, Casseday JH (1996) Whole-cell patchclamp recording reveals subthreshold sound-evoked postsynaptic currents in the inferior colliculus of awake bats. J Neurosci 16:3009–3018 8. Diamond ME (1995) Somatosensory thalamus of the rat. In: Jones EG, Diamond IT (eds) The barrel cortex of rodents. Plenum Press, New York, pp 189–219 9. Diamond ME, Armstrong-James M, Ebner FF (1993) Experience-dependent plasticity in adult rat barrel cortex. Proc Natl Acad Sci USA 90:2082–2086 10. Edwards FA, Konnerth A, Sakmann B, Takahashi T (1989) A thin slice preparation for patch clamp recordings from neurones of the mammalian central nervous system. Pflugers Arch 414:600–612 11. Fatt P, Katz B (1951) An analysis of the end-plate potential recorded with an intra-cellular electrode. J Physiol (Lond) 115: 320–370 12. Fee MS (2000) Active stabilization of electrodes for intracellular recording in awake behaving animals. Neuron 27:461– 468 13. Hamill OP, Marty A, Neher A, Sakmann B, Sigworth FJ (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflugers Arch 391:85–100 14. Helmchen F, Svoboda K, Denk W, Tank DW (1999) In vivo dendritic calcium dynamics in deep-layer cortical pyramidal neurons. Nat Neurosci 2:989–996 15. Horikawa K, Armstrong WE (1988) A versatile means of intracellular labelling: injection of biocytin and its detection with avidin conjugates. J Neurosci Methods 25:1–11

498 16. Huang W, Armstrong-James M, Rema V, Diamond ME, Ebner FF (1989) Contribution of supragranular layers to sensory processing and plasticity in adult rat barrel cortex. J Neurophysiol 80:3261–3271 17. Hubel DH, Wiesel TN (1959) Receptive fields of single neurones in the cat’s striate cortex. J Physiol (Lond) 148:574–591 18. Isaacson JS, Strowbridge BW (1998) Olfactory reciprocal synapses: dendritic signalling in the CNS. Neuron 20:749–761 19. Jagadeesh B, Wheat HS, Ferster D (1993) Linearity of summation of synaptic potentials underlying direction selectivity in simple cells of the cat visual cortex. Science 262:1901– 1904 20. Jahr CE, Nicoll RA (1982) An intracellular analysis of dendrodendritic inhibition in the turtle in vitro olfactory bulb. J Physiol (Lond) 326:213–234 21. Jiang M, Griff ER, Ennis M, Zimmer LA, Shipley MT (1996) Activation of locus coeruleus enhances the responses of olfactory bulb mitral cells to weak olfactory nerve input. J Neurosci 16:6319–6329 22. Margrie TW, Sakmann B, Urban NN (2001) Action potential propagation in mitral cell lateral dendrites is decremental and controls recurrent and lateral inhibition in the mammalian olfactory bulb. Proc Natl Acad Sci USA 98:319–324 23. Moore CI, Nelson SB (1998) Spatio-temporal subthreshold receptive fields in the vibrissa representation of rat primary somatosensory cortex. J Neurophysiol 80:2882–2892 24. Neher E (1992) Correction for liquid junction potentials in patch clamp experiments. Methods Enzymol 207:123–131 25. Nelson S, Toth L, Sheth B, Sur M (1994) Orientation selectivity of cortical neurons during intracellular blockade of inhibition. Science 265:774–777 26. Ogden D, Stanfield P (1987) Patch clamp techniques for single channel and whole-cell recording. In: Microelectrode techniques: the Plymouth workshop handbook, 2nd edn. The Company of Biologists, Cambridge Pei X, Volgushev M, Vidyasagar TR, Creutzfeldt OD (1991) Whole cell recording and conductance measurements in cat visual cortex in-vivo. Neuroreport 2:485–488 Rozov A, Burnashev N (1999) Polyamine-dependent facilitation of postsynaptic AMPA receptors counteracts paired-pulse depression. Nature 401:594–598 Sah P, Hestrin S, Nicoll RA (1990) Properties of excitatory postsynaptic currents in interneurones in rat hippocampus in vitro. J Physiol (Lond) 430:605–616 Sakmann B, Neher E (1983) Geometric parameters of pipettes and membrane patches. In: Sakmann B, Neher E (eds) Single channel recording. Plenum Press, New York Simons DJ (1983) Multi-whisker stimulation and its effects on vibrissa units in rat SmI barrel cortex. Brain Res 276:178–182 Stuart GJ, Dodt H-U, Sakmann B (1993) Patch-clamp recordings from the soma and dendrites of neurons in brain slices using infrared video microscopy. Pflugers Arch 423:511–518 Velumian AA, Zhang L, Carlen PL (1993) A simple method for internal perfusion of mammalian central nervous system neurones in brain slices with multiple solution changes. J Neurosci Methods 48:131–139 Walsh RR (1956) Single spike activity in the olfactory bulb. Am J Physiol 186:255–257 Young TA, Wilson DA (1999) Frequency-dependent modulation of inhibition in the rat olfactory bulb. Neurosci Lett 276:65–67 Zhu JJ, Connors BW (1994) Intrinsic firing patterns and whisker-evoked synaptic responses of neurons in the rat barrel cortex. J Neurophysiol 81:1171–1183

27. 28. 29. 30. 31. 32. 33.

34. 35. 36.