Distinct muscarinic acetylcholine receptor subtypes mediate pre- and postsynaptic effects in rat neocortex.

Background Cholinergic transmission has been implicated in learning, memory and cognition. However, the cellular effects induced by muscarinic acetylcholine receptors (mAChRs) activation are poorly understood in the neocortex. We investigated the effects of the cholinergic agonist carbachol (CCh) and various agonists and antagonists on neuronal activity in rat neocortical slices using intracellular (sharp microelectrode) and field potential recordings. Results CCh increased neuronal firing but reduced synaptic transmission. The increase of neuronal firing was antagonized by pirenzepine (M1/M4 mAChRs antagonist) but not by AF-DX 116 (M2/M4 mAChRs antagonist). Pirenzepine reversed the depressant effect of CCh on excitatory postsynaptic potential (EPSP) but had marginal effects when applied before CCh. AF-DX 116 antagonized the depression of EPSP when applied before or during CCh. CCh also decreased the paired-pulse inhibition of field potentials and the inhibitory conductances mediated by GABAA and GABAB receptors. The depression of paired-pulse inhibition was antagonized or prevented by AF-DX 116 or atropine but only marginally by pirenzepine. The inhibitory conductances were unaltered by xanomeline (M1/M4 mAChRs agonist), yet the CCh-induced depression was antagonized by AF-DX 116. Linopirdine, a selective M-current blocker, mimicked the effect of CCh on neuronal firing. However, linopirdine had no effect on the amplitude of EPSP or on the paired-pulse inhibition, indicating that M-current is involved in the increase of neuronal excitability but neither in the depression of EPSP nor paired-pulse inhibition. Conclusions These data indicate that the three effects are mediated by different mAChRs, the increase in firing being mediated by M1 mAChR, decrease of inhibition by M2 mAChR and depression of excitatory transmission by M4 mAChR. The depression of EPSP and increase of neuronal firing might enhance the signal-to-noise ratio, whereas the concomitant depression of inhibition would facilitate long-term potentiation. Thus, this triade of effects may represent a “neuronal correlate” of attention and learning.


Background
Considerable evidence indicates a crucial role of acetylcholine (ACh)-mediated transmission in learning and memory. Impairment of cholinergic signalling contributes to cognitive deficits in animal models [1] and diseases such as schizophrenia and Alzheimer's Disease [2][3][4]. Conversely, drugs that augment cholinergic transmission (e.g. donepezil or galatamine) have been shown to improve certain cognitive functions [5,6]. In this context metabotropic ACh receptors, termed muscarinic ACh receptors (mAChRs) [7] play a crucial role. Five mAChR subtypes have been cloned (M 1 -M 5 ); each consisting of seven highly conserved transmembranal segments. A large and more variable intracellular domain [8] couples the receptor to second messengers and ion channels via heterotrimeric G-proteins [1,7,9,10].
The mAChRs are now recognized as key targets for the treatment of different neuropathologies such as Alzheimer's Disease, schizophrenia and Parkinson's Disease [10]. In particular, agonists of M 1 mAChR or antagonists at M 2 or M 4 mAChRs are promising substances [10,11].
However, the similarity in ligand-binding sites between mAChR subtypes, the paucity of selective agonists/antagonists as well as the expression of different subtypes of mAChRs in a given area constitute major obstacles to delineate cellular effects of a given subtype [1]. The improvement of therapeutic approaches using specific mAChR modulators is further impeded by a lack of established and validated protocol to screen efficacy of different mAChR agonist/antagonist at native receptors in neocortical slices. Characterization of receptors in a native system is a crucial issue. For instance, at D 2 -type dopamine receptors [12] or AMPA receptors [13], the pharmacology at recombinant receptors critically depends on the experimental conditions. Thus, the pharmacology of mAChR in heterologous systems may not necessarily relate to the natively expressed mAChR [14]. Therefore, to delineate the pharmacology of mAChRs in the neocortex, the data must be obtained in the native neocortical slice.
We tried to define simple paradigms to quantify routinely the efficacy of different mAChR agonists. This should facilitate further screening for novel drugs in the neocortex. We investigated the effects of CCh and various mAChR antagonists on several parameters of neuronal excitability and the synaptic transmission in neocortical layer II/III. We have chosen layer II/III of neocortex because of the established expression of M 1 , M 2 and M 4 but not M 3 and M 5 mAChRs at the mRNA [15] and the protein levels [15][16][17][18].
As a first step we tested whether the published effects obtained under various experimental conditions can be reproduced. Firstly, in neocortical pyramidal neurones, ACh or the prototype cholinergic agonist carbamylcholine-chloride (carbachol, CCh) depolarizes the membrane potential and increases action potential (AP) firing [19][20][21]. Secondly, CCh reduces excitatory postsynaptic potentials (EPSPs) in neocortical neurons [21,22]. This depressant effect was reduced by the preferential M 1 /M 4 and M 2 /M 4 mAChRs antagonists (pirenzepine and AFDX 116, respectively) and was not due to activation of nicotinic receptors [22]. Thirdly, ACh reduces the GABA release in neocortex via activation of M 2 mAChR and to a smaller extent by activation of M 1 mAChR [23].
Here we show that these 3 effects are mediated by 3 different subtypes of mAChRs allowing screening for mAChRs subtype-specific agonists/antagonists.

Modulation of neuronal properties and firing behaviour
In control conditions, neurones had a resting membrane potential (E m ) of −76.2 ± 0.4 mV, an input resistance of 20.4 ± 0.8 MΩ and the AP amplitudes averaged 92.4 ± 0.7 mV (n = 92). The slope of neuronal AP firing was 57.3 ± 1.6 APs/nA in control conditions.

Pharmacological delineation of the mAChR subtypes involved
Atropine (n = 7) or pirenzepine (n = 7) reversed the CChinduced increase of slope of neuronal firing (p < 0.05 vs CCh alone and p > 0.05 vs. control; not shown). This antagonism was almost complete and indistinguishable from the reversibility observed after washout (n = 10; p > 0.05; Figure 2A). However, AFDX did not reverse the CChinduced increase of slope of neuronal firing (p > 0.05 vs. CCh and p < 0.05 vs. control). In fact, the antagonism observed with AFDX (n = 7) was much smaller than the reversibility observed after washout (n = 10, p < 0.01; Figure 2A). In the continuous presence of CCh, the antagonism observed during a co-application of "pirenzepine + AFDX" (n = 8) was not different to that obtained during application of pirenzepine alone (n = 7, p > 0.05; Figure 2A). This suggests that the slight antagonism observed during application of AFDX alone is due to some antagonistic effect of AFDX on M 1 mAChR.
The slope of neuronal firing during application of CCh in atropine-containing (n = 9) or pirenzepine-containing artificial cerebrospinal fluid (ACSF) (n = 7) is significantly smaller than that obtained during application of CCh in standard ACSF (n = 40, p < 0.01; Figure 2B). However, the slope of neuronal firing during application of CCh added to an AFDX-containing ACSF (n = 7) is not different to that obtained during application of CCh in standard ACSF (n = 40, p > 0.05; Figure 2B), i.e. AFDX fails to modify the CCh-induced effects on firing. To further test the involvement of M 1 mAChR in the effects on firing, we applied xanomeline, a M 1 /M 4 preferring mAChRs agonist [24]. Xanomeline increased the slope of neuronal firing from 57.8 ± 6.5 APs/nA to 80.4 ± 7.7 APs/nA (n = 11, p < 0.001).
Together these data suggest that CCh exerts its action on slope of neuronal firing via mAChRs, and in particular M 1 but not M 2 or M 4 mAChRs.  Pharmacology of CCh-induced increase in slope of neuronal APs firing. A. Plot of the reversibility (in %) of the CCh-induced increase in slope of neuronal APs firing after washout or addition of different mAChR antagonists to a CCh-containing ACSF as indicated. Note the reversibility observed with atropine (n = 7) or pirenzepine (n = 7) was maximal since not different to that observed after washout (n = 10). However, the reversibility observed with AFDX was much smaller to that observed after washout. The reversibility observed during a coapplication of CCh with "AFDX + pirenzepine" (n = 8) was not different to that obtained during a co-application of CCh with "pirenzepine alone" (n = 7). These experiments were performed in STRC. *: p < 0.05 vs. reversibility after washout. B. Plot of the slope of neuronal AP firing during CCh addition in standard ACSF or in ACSF containing different mAChR antagonists as indicated. Note the slope of neuronal firing during application of CCh in a atropine-containing (n = 9) or a pirenzepine-containing (n = 7) ACSF is significantly smaller to that obtained during application of CCh in standard ACSF (n = 40). However, slope of neuronal firing during application of CCh in an AFDX-containing ACSF (n = 7) is not different to that obtained during application of CCh in standard ACSF (n = 40). These experiments were performed in STRC. The symbols represent: *: p < 0.05 vs. slope of neuronal firing during application of CCh in standard ACSF.

Modulation of GABAergic transmission
CCh decreases the paired-pulse inhibition of field potentials Paired-pulse protocols are often used to study GABAergic inhibition in field potentials [27]. We therefore employed a paired-pulse protocol as a first step towards studying CCh modulation of inhibition.
The input-output curves obtained in ITRC (n = 49; not shown) were qualitatively similar to that obtained in STRC, except for larger amplitudes of potentials (p < 0.0001), as expected [28]. The amplitude of the field potentials decreased significantly in the presence of CCh (10 μM) at stimulus intensities larger than 0.3 mA (not shown; n = 49, p < 0.05).
Paired-pulse stimulations (interpulse intervals: 10 ms -1 s) were used to determine the magnitude of pairedpulse depression of the second field potentials (FP2). Since the apparent time-to-peak of the intracellular GABA A -mediated response is about 20-30 ms [29], we focus here on the results obtained for interpulse intervals of 10-30 ms to investigate GABA A receptor-mediated inhibition. A high stimulus intensity (1 mA) was used to maximally activate GABAergic transmission. When two pulses were delivered 20 ms apart, the mean amplitude of the FP2 was 25 ± 4% of the amplitude of the first (FP1; n = 59).
Application of linopirdine had no effect on the PPR (n = 10, p > 0.68; Figure 3H) suggesting that the M-current blockade is not involved in the CCh-induced increase of PPR. Additionally, we verified that retigabine (10 μM) did not alter the PPR (not shown, n = 8, p > 0.05). This suggests that modulation of GABA release, underlying the paired-pulse depression, is dominated by M 2 mAChR effects. To further rule out the involvement of M 1 /M 4 mAChRs in the depression of inhibition, we applied xanomeline. Xanomeline did not alter the PPR (n = 6, p > 0.05).

CCh decreases inhibitory synaptic conductances via M 2 mAChR
To more directly evaluate the effects of mAChRs on synaptic inhibition, we evaluated the synaptic conductances underlying the IPSP A and IPSP B in the presence of various mAChR related compounds. Application of CCh greatly reduced the amplitudes of both, IPSP A and IPSP B components (Figure 4 A,B). Estimating the underlying conductances (g IPSP ) revealed substantial decreases; g IPSP-A decreased from 110 ± 31 nS to 34 ± 11 nS, g IPSP-B decreased from 41 ± 9 nS to 12 ± 3 nS, i.e. by 69.1 and 69.8%, respectively ( Figure 4E; n = 17, p < 0.05). To evaluate which subtypes of mAChRs are involved in this modulation of inhibition we first tested the M 2/ M 4 mAChRs antagonist AFDX in the presence of CCh. In these neurones, CCh virtually abolished the conductances of both IPSPs (g IPSP-A from 32.8 ± 13.6 to 0.0 ± 5.4 nS; n = 5, p < 0.05 and g IPSP-B from 18.7 ± 4.6 nS to 3.5 ± 1.4 nS; n = 5, p < 0.05). Application of AFDX caused a considerable reversal in the depression of inhibition recovering to 63 and 72% of the control values (p > 0.05 vs. control).
Next we tested the effects of the M 1 /M 4 mAChRs antagonist pirenzepine. Application of CCh considerably reduced the synaptic conductances, g IPSP-A from 104.3 ± 39.4 nS to 58.9 ± 29.0 nS, g IPSP-B from 25.0 ± 8.2 nS to 13.0 ± 5.3 nS i.e. on average by 44 and 48%. On addition of pirenzepine, the depression of inhibitory conductances was essentially unaltered (g IPSP-A : 74.9 nS, p > 0.05 vs. CCh; g IPSP-B : 15.7 nS p > 0.05 vs. CCh).
To further substantiate our hypothesis of a M 2 mAChR-mediated depression of inhibition, we next  (p < 0.05), but just not significant for the g IPSP-A (p = 0.057).

Modulation of glutamatergic transmission
Effects of carbachol on excitatory synaptic transmission: Field potentials and intracellular recordings In control ACSF, the amplitudes of EPSP and field potentials increased progressively with increasing stimulus intensities to reach a plateau ( Figure 5C, D). Application of CCh (10 μM) decreased the amplitude of both EPSP (n = 45, p < 0.001; Figure 5A, B, C) and field potentials (STRC; n = 42, p < 0.05; Figure 5A, B, D) in a similar range of stimulus intensities. The input-output curves of EPSP were well fitted by the Boltzmann equation yielding a maximal EPSP amplitude (EPSP max ) of 22.6 ± 0.9 mV, a half maximal stimulus intensity (I 50 ) of 6.71 ± 0.36 V and a slope factor of 1.66 ± 0.14 (n = 45). CCh significantly decreased EPSP max to 16.0 ± 1.1 mV and increased both I 50 and slope factor to 9.08 ± 0.44 V and 2.82 ± 0.24, respectively (n = 45; p < 0.001), i.e. the curve became more shallow and I 50 shifted to higher values.
The depression of field potentials by CCh (n = 8) was completely reversible (p > 0.05 vs. control), whereas that of EPSPs (n = 11) was partially reversible after washout. The EPSP max recovered on average by 66 ± 11% after washout (p < 0.05 vs. control and CCh), whereas the effects on I 50 and slope factor were completely reversible (p > 0.05 vs. control, n = 11).

CCh-induced depression of synaptic responses is due to activation of mAChRs
Atropine (1 μM) alone had no effect on the amplitude of EPSP or field potentials at all stimulus intensities (2-20 V; n = 9, p > 0.05; not shown). Also, I 50 or slope factor were unaffected by atropine (not shown), indicating no obvious steady state modulation of synaptic responses by endogeneous mAChRs. Pre-application of atropine prevented the CCh-induced decrease of EPSP max as well as the CChinduced increase of I 50 and slope factor (n = 7, p > 0.05 vs. atropine or vs. control; Table 1). Conversely, addition of atropine fully reversed the CCh-induced decrease of EPSP max and field potential amplitude (n = 7, p > 0.05 vs. control and p < 0.05 vs. CCh; Table 1). The CCh-induced increase of I 50 and slope factor were also reversed (n = 7, p > 0.05 vs. control and p < 0.05 vs. CCh; Table 1). This suggests that the CCh effects on excitatory synaptic transmission are mediated by activation of mAChRs.
The CCh-induced depression of synaptic responses is sensitive to pirenzepine and AFDX Applied alone, neither pirenzepine (1 μM; n = 10) nor AFDX (2 μM; n = 11) had detectable effects on the EPSP amplitude, I 50 or slope factor of the input-output curve (not shown; p > 0.05).
Application of pirenzepine in the presence of CCh antagonized the CCh-induced decrease of EPSP max (n = 9, p > 0.05 vs. control and p < 0.05 vs. CCh; Table 2). Pirenzepine also antagonized the increase of slope factor (n = 9, p > 0.05 vs. control and p < 0.05 vs. CCh) and reduced the increase of I 50 (CCh: 189 ± 39% of control; CCh + Pir:124 ± 12% of control; n = 9, p < 0.05 vs. control and CCh; Table 2). Application of pirenzepine prior to CCh, however, failed to prevent the CCh-induced decrease of EPSP max (n = 7, p < 0.05 vs. control or vs. pirenzepine alone; Table 2). In fact, CCh added to a pirenzepine-containing ACSF depressed the EPSP max to 76 ± 6% of control (n = 7), indistinguishable from the depression observed with CCh alone (to 69 ± 4% of the control; n = 45; p > 0.05). This failure of pirenzepine to prevent a CCh-induced depression, when applied before CCh, is in sharp contrast to the pirenzepine effect on neuronal firing, where pirenzepine is effective irrespective of the sequence of application. Application of AFDX in the presence of CCh partly antagonized the CCh-induced decrease of the EPSP max (by 52.3 ± 3.1%; n = 8, p < 0.05 vs. control and CCh; Table 3). AFDX did not antagonize the effect on I 50 or slope factor (n = 8, p > 0.05 vs. CCh and p < 0.05 vs. control; Table 3). Also, application of AFDX before addition of CCh prevented the depressant effect of CCh on EPSPmax . Indeed, the EPSP max in the presence of AFDX + CCh was not different from that measured in the control condition (n = 7, p > 0.05; Table 3). Pre-application of AFDX also prevented the CCh-induced increase of I 50 (p > 0.05 vs. control or AFDX alone) but not the increase of slope factor (n = 7, p < 0.05 vs. control or AFDX alone). To further determine which subtype of mAChR mediates the effects of CCh on EPSP, we applied xanomeline. Xanomeline decreased the EPSP max from 25.6 ± 1.8 mV to 23.2 ± 2.7 mV (n = 12, p < 0.05).

Discussion
The key findings of this study are: (i) activation of M 1 mAChR increased the neuronal AP firing via blockade of M-current, (ii) activation of M 2 mAChR increased the paired-pulse ratio of field potentials and decreased both components of GABAergic inhibition, and (iii) activation of M 4 mAChR depressed glutamatergic transmission presumably via activation of presynaptic receptors.

Methodological aspects
The previous studies investigating the mAChRs in slice preparations used either sharp microelectrode recordings [20,30] or whole cell patch clamp recordings [22,23,31,32]. The latter disturb the intracellular milieu and a massive rundown of M-current (modulated by mAChRs) was observed [33], precluding a reliable pharmacology. Intracellular recordings in cortical slices with sharp microelectrodes hence present a valuable method to quantify the effects at native mAChRs in slices.
However, reports in the neocortex focussed on the role of mAChRs or M-current either on the neuronal firing [20,30,32], on the GABAergic transmission [23,31] or on the glutamatergic transmission [22,34]. Only few studies in cortex e.g. [21] or in hippocampus e.g. [35] investigated the role of mAChRs on both neuronal firing and synaptic transmission. The comparison of data obtained in different preparations using different experimental conditions is problematic, though. Therefore, in the present study we investigated the 3 aspects, namely change of firing, depression of EPSPs and of IPSPs, in the same preparation under identical experimental conditions.
Although some authors used different concentrations of mAChRs agonists/antagonists, only one or few stimulus intensities e.g. [22] or current injections e.g. [35] were used. Therefore, the results were mostly qualitative from a physiological point of view, precluding quantitative comparisons of different agonists/antagonists.

Effect of mAChR activation on neuronal excitability
The consistent increase in the slope of neuronal firing by CCh in the somatosensory cortex is similar to the hippocampus [35] and the nucleus gracilis [36]. However, our determination of slope and intercept provides a more quantitative analysis of changes of firing compared to previous observations [35,36]. In particular, these authors compared the neuronal firing in response to a single current intensity before and during substance application. The interpretation of such data would be complicated by concomitant alterations of membrane potential and/or neuronal input resistance.
In any case, alterations of firing are mediated by mAChRs, as evidenced by the antagonism by atropine. Moreover, the antagonism by pirenzepine (M 1 /M 4 mAChRs antagonist) but not by AFDX (M 2 /M 4 mAChRs antagonist) further suggests an effect via M 1 mAChR, similar to the M 1 mAChR-mediated repetitive spontaneous neuronal firing observed in rat prefrontal cortex [21]. Our conclusion is in line with previous data indicating that ACh application increases the neuronal firing of neocortical pyramidal neurones in mice lacking M 3 or M 5 mAChRs but not in mice lacking M 1 mAChR, confirming further an involvement of the M 1 mAChR in this effect [37].
The cellular mechanisms underlying the CCh-induced increase of AP firing can be inferred from the effect of linopirdine which is an established blocker of K V 7 channels [25]. Linopirdine increased the neuronal firing as in hippocampus [25] and entorhinal cortex [32]. Together, these data indicate a M 1 mAChR mediated depression of M-current [38] as the underlying mechanism of CChinduced firing increase.

Effect of mAChR activation on GABAergic transmission
We studied the mAChR mediated effect on neurotransmitter release by using the paired-pulse protocol with field potentials. CCh increased the PPR for interpulse intervals ranging from 10 to 400 ms. This suggests that the CCh effect involves both GABA A and GABA B receptor-mediated effects. The CCh-induced increase of the PPR is most likely due to mAChR-mediated depression of GABA release [23], and evoked IPSCs [31], during the first stimulus. Normally, GABA would attenuate the response to a second stimulus by decreasing release via presynaptic GABA B receptors [39,40]. Hence a mAChRmediated depression of GABA release might contribute to the altered paired pulse properties of synaptic responses.
The involvement of M 2 mAChR in the CCh-induced depression of GABA release underlying increase of PPR can be inferred from the antagonism by AFDX (M 2 /M 4 mAChRs antagonist) and lack of antagonism by pirenzepine (M 1 /M 4 mAChRs antagonist). Similar to the pharmacology at M 1 mAChR, the effects of atropine and AFDX on this M 2 mAChR-mediated increase of PPR were identical whether the antagonist was applied in the presence of, or before, CCh.
Further evidence was obtained by evaluating the alterations in inhibitory conductance mediated by GABA A and GABA B receptors. Firstly, the similarity in the depression of both g IPSP-A and g IPSP-B , indicates a common denominator for both effects i.e. depression of GABA release. The effects of the selected compounds revealed which mAChR is involved in this depression of GABA release. The CCh effects were considerably antagonized by the M 2 /M 4 mAChRs antagonist AFDX but slightly by the M 1 /M 4 mAChRs antagonist pirenzepine. Moreover, the M 1 /M 4 mAChRs agonist xanomeline had no effects on g IPSP-A or g IPSP-B , yet subsequent addition of CCh reduced g IPSP-A and g IPSP . This observation strongly suggests that neither M 1 nor M 4 mAChRs are involved in the modulation of GABA release. Together these data support our view that M 2 mAChR are involved in the modulation of GABA release.
This view is also supported by histological data. M 2 mAChRs are predominantly localized at presynaptic axons of GABAergic neurones and are associated with asymmetric as well as symmetric cortical synapses [16][17][18]23], suggesting that M 2 mAChR can modulate GABA release in the neocortex. In addition, M 1 mAChRs are mainly expressed on cortical pyramidal neurones rather than by GABAergic neurones [41].
ACh reduces the GABA release in neocortex by two separate mechanisms i) by decreasing Ca 2+ currents via activation of a M 2 mAChR/PI3K/Ca 2+ -independent PKC pathway and to a smaller extent by activation of a M 1 mAChR/PLC/Ca 2+ -dependent PKC pathway [23]. This combination might explain the relatively large contribution of the M 2 mAChR and the absence of detectable effects of the M 1 mAChR in the CCh-induced increase of PPR observed here.

Effect of mAChR activation on glutamatergic transmission
Our observation of CCh decreasing the amplitudes of evoked synaptic potentials is in line with a wealth of evidence from various areas of the rodent brain [22,[34][35][36]. Both field potentials and intracellularly recorded EPSPs were depressed by CCh to a similar degree. Considering the marginal changes in E m , this decrease cannot be accounted for by a membrane depolarization, confirming results obtained in entorhinal cortex [34]. These authors reported that the depression of synaptic transmission persisted when neurones were returned to their initial resting potential. However, previous observations in hippocampus e.g. [35] or neocortex e.g. [21,22] compared the synaptic response following a single stimulus intensity before and during substance application. Our determination of I 50 , EPSP max and slope factor provide a more quantitative analysis of changes of synaptic transmission, corroborating and extending previous findings. In particular, the consistent effects over a wide range of stimulus intensities exclude the possibility of threshold phenomena.
Our data indicate that this CCh-induced depression of EPSP is 1) fully prevented when CCh is applied in the presence of atropine or AFDX and 2) only partially prevented when CCh was applied in the presence of pirenzepine. This corroborates and extends previous data from layer V of the somatosensory cortex [22]. Unlike these authors, applying CCh in the presence of antagonists, we also applied the mAChRs antagonists in the presence of CCh, revealing marked differences with different sequence of application. The CCh-mediated depression of EPSP was fully reversed when pirenzepine or atropine was added in the presence of CCh, i.e. pirenzepine is more effective when applied in the presence of the agonist. AFDX, however, only partially reversed this depression, i.e. is less effective when applied in the presence of the agonist. This peculiarity may contribute to the discrepancies between previous reports. Usually, the authors have either applied the antagonist in the presence of agonist or vice versa, the agonist in the presence of antagonists. To the best of our knowledge, the two complementary protocols had not been used in the same study so far e.g. either addition or pre-application was used, see [22,42].
We propose that pirenzepine, a M 1 /M 4 prefering mAChRs antagonist [43], affects the EPSP amplitude via M 4 mAChR for the following reasons: 1) The increase of neuronal firing by CCh was prevented and reversed by pirenzepine. However, the decrease of EPSP max by CCh was reversed but not prevented by pirenzepine. This difference may suggest that the pirenzepine-sensitive effects on EPSP max and on slope of neuronal firing are mediated by two distinct mAChRs, i.e. the pirenzepine-sensitive effect on EPSP max is not mediated by M 1 mAChR.
2) Only M 2 and M 4 mAChRs (but not M 1 , M 3 and M 5 mAChRs) are coupled to G-proteins G i /G o family involved in the inhibition of neurotransmitter release [7,44,45]. In addition, recent evidence indicates that M 4 mAChR is the major mAChR subtype responsible for direct cholinergic modulation of EPSP [46]. Therefore, the pirenzepine effect on CCh-induced EPSP depression is probably due to an action of pirenzepine on M 4 rather than M 1 mAChRs [43]. This view is supported by immunohistochemical data [17], showing that cortical M 1 mAChRs are located exclusively at postsynaptic sites.
3) The effects of atropine and AFDX on the M 2 mAChR-mediated increase of PPR were identical regardless of the sequence of application. However, the effects of pirenzepine and AFDX on the CCh-induced decrease of EPSP were different whether this antagonist is applied in the presence or before the CCh application. This difference may suggest that the CCh-induced decrease of EPSP max and increase of PPR are mediated by two distinct mAChRs, i.e. the CCh-induced decrease of EPSP max is not mediated by M 2 mAChR. Moreover, the increase of PPR by CCh (mediated by M 2 mAChR) was not prevented or reversed by pirenzepine. This suggests that pirenzepine is not active at M 2 mAChR in our conditions. 4) AFDX applied before CCh or pirenzepine applied in the presence of CCh can both fully prevent and reverse, respectively, the CCh-induced depression of EPSP max . The similar potency of pirenzepine and AFDX on CChinduced EPSP max depression therefore suggests an involvement of M 4 mAChR as the most parsimonious explanation.

Cholinergic transmission and cognition
Ample evidence suggests that mAChR-mediated signalling in the cortex is critical for learning and memory, yet the mechanisms of ACh facilitating cognitive processes remained elusive. The effects reported here, namely the depression of EPSP and the increased neuronal firing mediated by different mAChRs would provide a very interesting system for "attention" at the cellular level. The M 4 mAChR-mediated depression of EPSP reduces the noise of ongoing synaptic activity while the M 1 mAChR-mediated increase of neuronal firing augments the neuronal response to a given synaptic input, i.e. these two effects would improve the signal to noise ratio. Moreover, the shallower input-output curve of synaptic responses induced by M 4 mAChR observed here extends the subthreshold range for temporal summation, which is augmented by a CCh-induced reduction of Kir-type potassium current [21]. In addition, the cholinergic depression of GABA release corresponds to a depression of inhibition by higher frequencies of stimulation [40] implicated in long-term potentiation [47]. Together, these data corroborate and extend a theoretical framework proposed by Hasselmo and McGaughy [48].

Conclusions
Our data provide evidence that the triade of cellular CCh effects are mediated by three subtypes of mAChR, i) M 1 mAChR-mediated increase of firing, ii) M 2 mAChR-mediated depression of GABA release and iii) M 4 mAChR-mediated depression of glutamatergic synaptic responses. Activation of these effects by physiologically released ACh provides an interesting combination of neuronal attention and facilitation of LTP. In addition, the three parameters delineated here may present useful tools to test new M 1 , M 2 or M 4 mAChRs modulators at native receptors in the human neocortex from epilepsy surgery tissues.

Tissue handling and preparation
Coronal slices containing the sensorimotor cortex were made from male Wistar rats (age: 30-42 d). Experiments were approved by the Berlin health protection agency (T 026/96). Rats were deeply anesthetized with ether, decapitated and a block of brain containing the sensorimotor cortex was quickly removed. The tissue was cut into slices (400 μm) with a vibratome (HM 650 V, MICROM International, Walldorf, Germany) in artificial cerebrospinal fluid (ACSF, 5 ± 1°C). The slices were stored submerged at room temperature in continuously oxygenated (carbogen: 95% O 2 -5% CO 2 ) ACSF. Slices were allowed to recover for at least 1 hour before recordings commenced. Individual slices were transferred to the recording chambers and given 30 min to adapt.
The following mAChRs related compounds were used: pirenzepine (M 1 /M 4 mAChRs antagonist), AFDX 116 (M 2 /M 4 mAChRs antagonist), xanomeline (M 1 /M 4 mAChRs agonist) and atropine (broad spectrum mAChRs antagonist) [24,43,49]. We refrained from using mamba toxin 7, a proposed selective blocker of M 1 mAChR, because the application via perfusion would be prohibitively expensive and the stability under our recording conditions (oxygen, 32°C) is uncertain [communicated by a manufacturer]. We chose established concentrations of agonist and antagonist based on a literature survey and a few pilot experiments. A concentration of 10 μM CCh and 1 μM atropine is established in slices [21,22], and at 1 μM, atropine does not interfere with nicotinic receptors [50]. The concentrations of pirenzepine (1 μM) and AFDX (2 μM) were chosen from binding experiments [43,49] and are effective on the CCh-induced depression of PSP in neocortex or hippocampus slices for example [22,51]. Pirenzepine at 1 μM should have a maximal effect at M 1 and M 4 mAChRs but not at M 2 , M 3 and M 5 mAChRs. AFDX at 2 μM should have a maximal effect at M 2 and M 4 mAChRs but not at M 1 , M 3 and M 5 mAChR [49].
From the above considerations, the following predictions can be made with a minimum of compounds: 1) a M 1 mAChR-mediated effect should be antagonized by pirenzepine but not by AFDX, 2) a M 2 mAChR-mediated effect should be antagonized by AFDX but not by pirenzepine and 3) a M 4 mAChR-mediated effect should be antagonized by either pirenzepine or AFDX.
To reveal more subtle pharmacological differences of the mAChRs, i.e. effects of activation state, we applied antagonists either before or during the CCh application.

Recording chambers
Experiments were carried out either in an interface-type recording chamber (ITRC) or in a submersion-type recording chamber (STRC). The ITRC enables maximal oxygen supply and improves recordings of evoked field potentials and was therefore chosen for the paired-pulse protocol. The STRC provides more stable conditions and allows a faster equilibration of drugs in the slices [28]. Therefore, STRC was chosen for intracellular recordings.
In ITRC, slices were perfused with oxygenated ACSF (1.5-2 ml/min; 34 ± 1°C) and the atmosphere was maintained by a continuous flow of pre-warmed and humidified carbogen. In STRC experiments slices were held between two nylon grids and perfused with a higher flow rate of oxygenated ACSF (4-5 ml/min; 31 ± 1°C).
All measurements were made at least 30 min (ITRC) or 15 min (STRC) after drug application to ascertain steady-state of applied drugs.

Extra-and intracellular recordings
The electrodes were positioned under visual control in neocortical layers II/III. Signals obtained with extracellular recordings were fed to a high-impedance preamplifier (EXT-01 C, npi electronic, Tamm, Germany) and processed through second-stage amplifiers with filtering capabilities (DPA 2 F; npi electronic, Tamm, Germany). The signals obtained with intracellular recordings were fed to an appropriate amplifier (SEC-05 L, npi electronic, Tamm, Germany).
We evaluated the current voltage relationship from families of current injections (between −0.5 nA and +1.5 nA, increment 0.05 nA, duration 600 ms), to estimate the neuronal input resistance and the firing behaviour. Linear regression of the number of APs vs. injected current in the initial linear range was used to calculate the slope and the intercept. The former parameter provides a quantitative index of AP firing behaviour in a given condition. The latter represents the current just below the rheobase and was termed sub-rheobase.
Synaptic responses were elicited by electrical stimuli (0-20 V, increment 2 V, 100 μs duration at 0.1 Hz, in triplicate) applied via a bipolar tungsten electrode placed in the deeper cortical layers (V/VI). The peak amplitude of the initial component represents the excitatory postsynaptic potential and will be referred to as EPSP, although it might be slightly affected by the inhibitory postsynaptic potential (IPSP) with a slightly later time to peak (see Figure 4A). The input-output curves of EPSP were fitted by the Boltzmann equation: yielding the maximal EPSP amplitude (A 2 , EPSP max ), the half maximal stimulus (x 0 , I 50 ) and the slope factor (dx). Synaptic conductance mediated by GABA A and GABA B receptors was estimated similar to previous methods [40,52]. In brief, linear regressions of the current voltage relation (usually between −0.3 and +0.3 nA) was carried out before the stimulus and at the apparent peak of the early and late inhibitory synaptic responses (near 20 ms and 150 ms poststimulus). Assuming synaptic conductances parallel to "resting" membrane conductance, the subtraction of resting membrane conductance from the total conductance at the peak of GABA A and GABA B receptor-mediated events yields an estimate for the two synaptic responses see [40].
To evaluate the effects of mAChR agonists/antagonists on neurotransmitter release supposedly via presynaptic sites, we recorded the field potentials during a pairedpulse protocol in an ITRC. The ratio of amplitude of the second field potential in relation to the first in response to identical stimulations (100 μs, 1 mA) was calculated for different interstimulus intervals (10 ms-1 s) and was termed paired-pulse ratio (PPR). An intensity of 1 mA was chosen for this protocol since it induced a maximal response with pronounced paired-pulse depression in all slices tested. In addition, we performed input-output curves, as in STRC. In ITRC, field potentials were elicited in cortical layers II/III by electrical stimuli (0-1 mA, increment 0.1 mA, 100 μs duration at 0.1 Hz, in triplicate) delivered via a bipolar tungsten electrode placed in cortical layers V/VI.

Data acquisition and analysis
Recorded signals were digitized on-line (10 kHz) with a PC based system (Digidata 1200 and Clampex 9.3 or Digidata 1440A and Clampex 10.1 software, Molecular Devices, Sunnyvale CA, USA), stored on hard disk and analysed off-line (Clampfit 10.1).
Data are presented as mean ± s.e.m., n indicates the number of cells (only one cell per slice was recorded). For comparisons, we used paired or unpaired Student's t-test. Differences with p < 0.05 were considered significant.