Effects of carbonic anhydrase activity on the excitability of hippocampal axons during high-frequency firing

Epileptic activity is known to cause a lowering of intraneuronal pH, which has been suggested to serve as a feedback signal to terminate seizures. The mechanism of such signaling is unclear, but likely involves an altered function of several types of ligand-and voltage-gated channels in postsynaptic membranes caused by increasing cytosolic and extracellular [H + ]. In addition, axonal conduction properties may be altered by endogenous pH signals, but this has not been investigated. In the present study, we have recorded the axonal compound action potential (fiber volley) in hippocampal slices in the presence of glutamatergic and GABAergic antagonists. During high-frequency stimulation (HFS) of the Schaffer collaterals, the fiber volley was depressed and its latency from stimulus to peak increased. In the CA1 stratum radiatum these changes were enhanced when the carbonic anhydrase inhibitor acetazolamide (1 mM) was co-perfused. The enhancing effect of acetazolamide was absent after lowering of [Ca 2 + ] in the perfusion medium. Acetazolamide had no detectable effect on HFS-evoked fiber volleys recorded from a more proximal site along the Schaffer collaterals (at the CA2-CA3 border) or from axons in the alveus of CA1. Intracellular acidification imposed by washout of NH 4 Cl (5 mM) had qualitatively similar effects on fiber volleys evoked at low frequency as those observed with acetazolamide during HFS in CA1 stratum radiatum. The results suggest that carbonic anhydrase-dependent pH regulation counteracts activity-induced reduction of the excitability of Schaffer collateral axons in CA1. A possible influence from local synaptic terminals on this effect is discussed.


Introduction
Over the past decades, accumulating evidence has shown that H + suppress seizure activity (de Curtis et al. 1998, Tolner et al. 2011, Pavlov et al. 2013, Ruusuvuori and Kaila, 2014. Furthermore, activity-induced lowering of pH (Ballanyi and Kaila, 1998;Chesler, 2003;Kaila and Chesler, 1998;Siesjö et al., 1985) correlates with cessation of interictal and seizure-like spiking, suggesting an intrinsic feed-back regulatory role of H + in terminating epileptic activity (de Curtis et al. 1998, Pavlov et al. 2013, Raimondo et al. 2015. To understand the underlying mechanisms, it is important to identify which elements of neural transmission are critically influenced by endogenous pH shifts during high-intensity activity. Functional consequences of activityinduced pH perturbation has been reported in a number of key studies examining specific neuronal properties. These studies have demonstrated the involvement of both glutamatergic and GABAergic synaptic transmission in generating local extracellular pH transients, which, in turn, can alter transmitter-operated postsynaptic receptors, e. g. cause proton inhibition of the NMDA receptor channel (Gottfried and Chesler 1994, Taira et al. 1993, Taira et al., 1995, Makani and Chesler 2007, Ruusuvuori and Kaila 2014. In addition, extracellular and intracellular pH shifts in the pathological range can alter voltage-gated ion channel properties and gap junction coupling, which may lead to significant changes in network excitability during epileptic activity, including non-synaptic interactions (Tombaugh and Somjen 1996, de Curtis et al. 1998, Schweitzer et al. 2000. Apart from altering postsynaptic membrane properties, pH shifts during epileptic activity could possibly also involve axons, which may contribute to feedback control of the activity. Kraig et al. (1983) found that stimulus-evoked activity in parallel fibers in rat cerebellum caused an extracellular Abbreviations: CGP55845, (2S)-3-[[(1S)-1-(3,4-Dichlorophenyl)ethyl]amino-2-hydroxypropyl](phenylmethyl)phosphinic acid hydrochloride; CNQX, 6-cyano-7nitroquinoxaline-2,3-dione; HFS, high-frequency stimulation; LY341495, (2S-2-amino-2-[(1S,2S-2-carboxycycloprop-1-yl]-3-(xanth-9-yl propanoic acid; MK 801, (5R,10S-(+-5-methyl-10,11-dihydro-5H-dibenzo [a,d] cyclohepten-5,10-imine-maleate; str., stratum. acidification of metabolic origin, presumably caused by accumulation of lactate in the extracellular space. Such change occurred in the absence of synaptic transmission indicating an axonal origin. Similar activitydependent extracellular acidic shifts is observed in the optic nerve, which has no synapses (Ransom et al. 1986). It is thus conceivable that axonal-activity induced change in intracellular or extracellular [H + ] could affect the electrical properties of axons and possibly presynaptic terminals, but this has so far not been elucidated.
In hippocampal brain slices, repetitive stimulation of the Schaffer collaterals is known to cause an initial alkalinization followed by a slow acidification of the bulk extracellular space (Kaila and Chesler 1998, Gutschmidt et al. 1999, Chesler 2003. In the CA1 stratum (str.) radiatum, such pH shifts are almost completely abolished in the presence of antagonists of glutamate and GABA receptors, demonstrating that they are largely dependent on synaptic transmission (Grichtchenko and Chesler 1996). This latter observation does, however, not exclude contribution from axons or presynaptic terminals as pH transients can occur in local microdomains in the vicinity of active membranes without inducing a global change in extracellular pH (Makani and Chesler 2010).
In the present study, we have stimulated the Schaffer collateral axons in hippocampal brain slices at high frequency to emulate epileptic activity. We have quantified the activity-dependent modification of the local fiber volley under conditions where synaptic transmission was blocked pharmacologically (Andreasen and Nedergaard 2017). To examine if endogenous pH dynamics influences axonal excitability, we used the carbonic anhydrase inhibitor acetazolamide as the CO 2 /HCO 3 system is instrumental in buffering extra-and intracellular pH transients in the CNS (Kaila and Chesler, 1998;Obara et al., 2008;Ruusuvuori and Kaila, 2014;Tong et al., 2000). The results suggest that carbonic anhydrase-dependent pH regulation has a role in mitigating activityinduced decrease in the excitability of Schaffer collateral axons in area CA1.

Results
We examined the effect of the carbonic anhydrase inhibitor, acetazolamine (1 mM) on the fiber volley recorded in CA1 str. radiatum. Recordings were done in six slices where blockers of ionotropic glutamate receptors (CNQX (40 μM), MK-801 (10 μM)) and GABA B receptors (CGP (2 μM)) were present, and in another five slices where blockers of GABA A receptors (gabazine (10 μM)) and metabotropic glutamate receptors (LY341495 (100 μM)) were also present. We found no difference in the responses to acetazolamide between these groups, and data from all experiments were therefore pooled. Averaged data (1 min recording) obtained after 20 min perfusion of acetazolamide showed no effect on fiber volleys evoked by low-frequency stimulation of the Schaffer collaterals (0.2 Hz) when compared to data from a control period before the perfusion or 30 min after washout (Fig. 1a,b). In accordance with previous findings (Andreasen and Nedergaard, 2017), a HFS train of 100 stimuli at 100 Hz, induced a biphasic change in the fiber volleys evoked during the train, consisting of an initial increase in the area and amplitude peaking close to stimulus #10 followed by a large decrease (Fig. 1c). The latter change was accompanied by an increase in the latency from stimulus to peak. In 9/9 slices tested during perfusion of acetazolamide (20 min), we found that HFS trains induced a larger reduction of the size and a larger increase in the latency of fiber volleys evoked late in the train compared to control (Fig. 1c). Statistical comparison of normalized data showed that the initial enlargement at stimulus #10 was unaltered by acetazolamide, whereas the late changes, measured at stimulus #100, deviated significantly from controls taken before perfusion of the drug. This effect included both the area (reduction in control: 67 ± 3 %; acetazolamide: 77 ± 2 %, p < 0.01), amplitude (reduction in control: 79 ± 2 %; acetazolamide: 83 ± 2 %, p < 0.01), and latency (increase in control: 33 ± 4 %; acetazolamide: 38 ± 4 %, p < 0.01). The effect of acetazolamide on area (p < 0.05) and latency (p < 0.01), but not amplitude, showed significant reversal after wash out ( Fig. 1d). After terminating the train, we monitored the time course of recovery. Averaged data from eight slices tested showed a rightward shift of the recovery curves for both the area, amplitude, and latency in the presence of acetazolamide compared to control before. As shown in Fig. 1e, the time to full recovery of the mean area was increased from ~5 s (control) to ~10 s (acetazolamide). The amplitude showed a slightly later recovery, which was also slowed by acetazolamide. Recovery of the latency occurred within 15-20 s in control condition, and was somewhat retarded with acetazolamide. These effects of acetazolamide showed partial or full reversal after wash out, but less so for the amplitude. Statistical comparison of data recorded at 1 s post train showed significant effects of acetazolamide on the fiber volley area (reduction in control: 9 ± 2 %; acetazolamide: 16 ± 3 %, p < 0.01), amplitude (reduction in control: 17 ± 2 %; acetazolamide: 22 ± 3 %, p < 0.05), and latency (increase in control: 8 ± 0.5 %; acetazolamide: 12 ± 0.5 %, p < 0.01). Wash out of acetazolamide (30 min) lead to significant reversal of the effect on the area and latency, but not on the amplitude (Fig. 1f). The discrepancy between wash out effects on the area and amplitude was not further explored. It is possible that the fiber volley amplitude, which, unlike the area, depends on synchronous discharge of individual axons, requires >30 min wash to recover. The effects of acetazolamide did not differ significantly between stimulus #100 during the HFS and 1 s post train (two-way ANOVA on fv area and latency data, Holm-Sidak post hoc test).
The above results suggested that blockade of carbonic anhydrase potentiates the activity-dependent reduction of excitability of axons in CA1 str. radiatum. We speculated that this effect depends on pH changes secondary to presynaptic Ca 2+ influx. To examine, we perfused the slice with ACSF containing low Ca 2+ (0.2 mM)/high Mg 2+ (10 mM). In all experiments, field EPSPs were completely abolished after 20 min perfusion with low Ca 2+ , and we therefore proceeded without adding antagonist of synaptic receptors. With the same electrode arrangement and stimulus protocol as above, a 100 Hz train still induced suppression of the fiber volley late in the train. However, the early enhancement, present in normal Ca 2+ , was small or absent in the slices exposed to low Ca 2+ , in line with recent observations by Owen et al. (2021). Under these conditions, perfusion of acetazolamide had little effect on the fiber volley during the train (Fig. 2a1). Averaged data (n = 8) showed no overall effects on either the area, amplitude, or latency, measured at stimulus #10 or #100 (Fig. 2a2). The time-course of post-train recovery ( Fig. 2a3) and the relative magnitude of depression at 1 s ( Fig. 2a4) was also unaltered by acetazolamide in these slices. These results suggested that the effects of acetazolamide depends on extracellular Ca 2+ .
Schaffer collaterals make a large terminal field within the CA1 area containing extensive axonal branching and numerous synaptic contacts (Ishizuka et al. 1990, Li et al. 1994. It is possible that the sensitivity of the fiber volley to acetazolamide depends on events specific to the CA1 terminal field. To examine, we made recording from a more proximal location along the Schaffer collaterals, in the area close to the border between CA2 and CA3. Recordings were done in the presence of CNQX, MK-801, CGP and gabazine. Fiber volleys were evoked by orthodromic stimulation (stimulus electrode placed in area CA3, four slices) or antidromic stimulation (stimulus electrode placed in area CA1, str. radiatum, five slices, Fig. 2b1). We found no difference between results from the two stimulation positions, and the data were therefore pooled. At the CA2/CA3 recording site, 100 Hz train stimulation resulted in depression of the fiber volley late in the train (Fig. 2b1). Little or no enhancement was present early in the train, in accordance with previous findings (Owen and Grover, 2015). Furthermore, acetazolamide had no detectable effect on the course or magnitude of the fiber volley depression induced during the train (Fig. 2b1-b2) or in the recovery period after the train (Fig. 2b3-b4).
For comparison, we also examined fiber volleys from the CA1 alveus (Fig. 2c1), which contains axonal projections devoid of synaptic contacts. To exclude interference from synaptic signaling in the deeper strata of CA1, we perfused CNQX, MK801, and CGP. Consistent with our previous findings (Andreasen and Nedergaard 2017), trains of 100 Hz delivered to an isolated part of the alveus (see Methods) led to a progressive and monophasic decrement of the fiber volley (Fig. 2c1), which recovered within approximately 5 s (area and amplitude) or 20 s (latency) (Fig. 2c3). Addition of acetazolamide had no detectable effect on the course or magnitude of the HFS-induced fiber volley depression in the alveus nor on the recovery after HFS (Fig. 2c2-4).
The above effects of acetazolamide in area CA1 is likely to reflect an intracellular acidification. To substantiate, we tested if acute intracellular acidification by NH 4 + retraction could alter the fiber volley properties. We added 5 mM NH 4 Cl to the ACSF and perfused for 7 min before wash out. Fiber volleys evoked by low frequency stimulation (0.2 Hz) in the CA1 str. radiatum showed a transient decrement during wash out. The effect peaked around 5 min after removal of NH 4 Cl, at which time the area, amplitude, and latency deviated significantly from controls before NH 4 Cl. Recordings close to the CA2/CA3 border yielded similar results with NH 4 Cl (Fig. 3). In the alveus, the fiber volley displayed, unexpectedly, little overall change at 5 min wash out. The area was largely unchanged whereas the amplitude showed a slight decline, which turned out to be significant. However, the area and amplitude were substantially reduced during the preceding 7 min of NH 4 Cl perfusion, which were highly significant for both. These effects were accompanied by a slight, but significant, increase in the latency of the fiber volley, which persisted after 5 min washout (Fig. 3).

Discussion
This study shows that the carbonic anhydrase inhibitor acetazolamide reversibly potentiates HFS-induced reduction of the fiber volley in the CA1 str. radiatum. This effect existed in the presence of blockers of ionotropic and metabotropic receptors, suggesting that it is independent of activation of postsynaptic or presynaptic receptors for glutamate or GABA, and the underlying processes likely reside in the stimulated axons. The presynaptic volley in str. radiatum largely represents the activity of the excitatory Schaffer collateral axons (Andersen et al. 1978), but some contribution from axons of local inhibitory interneurons cannot be excluded despite recording at a distance of ≥ 500 μm from the site of stimulation. It should be noted that the paradigm used here (100 Hz, 1 s) simulates seizure-like activity and thus reflects a non-physiological, hyperexcited state.
Activity-dependent modification of neuronal membrane properties has been reported in a large number of studies, mostly focusing on network or whole cell activity. It is known that repetitive stimulation of axons or spontaneous seizures can generate changes in the extracellular concentration of K + , Na + , and Ca 2+ in sufficient scale to influence neuronal excitability (Heinemann et al. 1990, Raimondo et al. 2015. Furthermore, increased [K + ] o has depressant effect on the presynaptic afferent volley in CA1 (Meeks and Mennerick, 2004;Poolos et al., 1987) contributing to synaptic depression through membrane depolarization and Na + channel inactivation. Effects of perturbations of major ion concentrations are likely pivotal for the HFS-induced decrement of the fiber volley in CA1. However, the present findings indicate that pH shifts induced by activity in local axons can have significant impact on their excitability. The observed inhibitory effects of acetazolamide most likely reflect an intracellular acidosis or a combined intra-and extracellular acidosis. Lowering of the intracellular or extracellular pH is known to cause decreased network excitability due to changes in the functional state of ion channels, ion pumps, and transporters (Tombaugh and Somjen 1997, Balestrino and Somjen 1988, Raimondo et al. 2015. In accordance, we found that withdrawal of NH 4 + , which induces acute intracellular acidification, resulted in a rapid decrement of the fiber volley. Seizures, or prolonged neuronal depolarization, has been shown to result in cytosolic acidosis combined with a late extracellular acid shift. Evidence suggest that these changes, observed in whole cells in situ, primarily result from i) metabolic production of lactate and CO 2 , and ii) net entry of H + through pumps or channels. Either of these mechanisms seem to be linked to a rise in intracellular Ca 2+ leading to a fall in intracellular pH due to increased energetic demands associated with Ca 2+ loading or as a direct consequence of Ca 2+ /H + exchange by the plasmalemmal Ca 2+ ATPase (Wang et al. 1994, Trapp et al. 1996, Schwiening and Thomas, 1998, Chesler, 2003. Consistent with a Ca 2+dependent acidifying process, we found that lowering of [Ca 2+ ] in the perfusate abolished the inhibitory effect of acetazolamide. This result supports the idea that the effect of blocking carbonic anhydrase activity in the Schaffer collateral terminal field was due to enhancement of an activity-induced, Ca 2+ dependent cytosolic and/or local extracellular acidosis leading to decreased axonal excitability. The acidosis is likely predominantly H + -dependent, as CO 2 /bicarbonate-dependent components of a pH shift would expectedly be opposed by carbonic anhydrase inhibition (Kaila and Chesler 1998). The primary site of this effect is uncertain, as acetazolamide is a membrane permeant inhibitor (Saarikoski and Kaila 1992), and thus does not discriminate between intracellular and extracellular carbonic anhydrase. We found no effect of acetazolamide on the dimension of fiber volleys evoked at low frequency, suggesting that a possible decrease in baseline interstitial pH, as previously shown during perfusion of acetazolamide in hippocampal slices (Chen and Chesler 1992), was insufficient to affect background electrical properties of Schaffer collateral axons.
Acetazolamide had no detectable effect on HFS-evoked fiber volleys recorded from Schaffer collaterals in the CA2/CA3 border region or from axons of CA1 pyramidal neurons located in the alveus. The results with NH 4 + showed that there was little response to imposed acidification in the alveus, the reason for which is unclear, but the finding seems consistent with a lack of effect of acetazolamide on these fibers. However, NH 4 + withdrawal consistently reduced the fiber volley in str.
radiatum. This effect was present both in CA1 and in the CA2/CA3 border region, making it less likely that differences in pH sensitivity could explain the different response to acetazolamide between these areas. As CA1 str. radiatum contains a high density of synaptic terminals in addition to axonal elements, it is tempting to suggest that the effects of acetazolamide was exerted primarily on synaptic terminals or, at least, was dependent on their presence. Involvement of synaptic  1, 10, and 100) shown enlarged from a control period, and during perfusion of acetazolamide. CNQX, MK-801, and CGP was present throughout. Calibrations: 1 mV, 1 ms. Bottom: from the same experiment, fiber volley area, amplitude, and latency are plotted as function of stimulus number. d, summary histogram of the normalized change in fiber volley area, amplitude, and latency at stimulus number 10 and 100 from control period, during acetazolamide perfusion, and after wash out. * p < 0.05, ** p < 0.01, non-significant differences are not indicated. e, top: voltage trace showing fiber volleys evoked at 0.5 Hz stimulation beginning 1 s after the last stimulus of a 100 Hz train. Middle: enlarged records of the fiber volleys evoked by the first stimulus of the train (# 1) and 1 s after the train. Records in the presence of acetazolamide (broken lines) are superimposed on control records (full lines). Calibrations: 1 mV, 1 ms. Bottom: summary plots of fiber volley area, amplitude, and latency (normalized) in the post-train period during control and during perfusion of acetazolamide (n = 9). f, summary histogram of the change in fiber volley area, amplitude, and latency at 1 s after the train (normalized) from control period, in the presence of acetazolamide, and after wash-out. * p < 0.05, ** p < 0.01.

(caption on next page)
terminals would also seem to be in line with the observed Ca 2+ dependency of the effect. In motor nerve terminals, activity-induced cytosolic acidification is paralleled by an increased [Ca 2+ ] i , is blocked in zero [Ca 2+ ] o , and increases in the presence of carbonic anhydrase inhibitors (Zhang et al. 2010, Caldwell et al. 2013, Rossano et al. 2013, consistent with the present observations. In this context, it is worth considering that the synaptic terminals in the CA1 str. radiatum are arranged in strings of boutons en passage (Ishizuka et al. 1990, Li et al. 1994. Such structural arrangement can significantly increase the tendency of axon conduction failures during high-frequency firing (Lüscher and Shiner, 1990). It is conceivable that acidic shifts originating in synaptic boutons, via local spread into juxtaposed axonal segments, could reduce conduction fidelity in these segments. An alternative reason for the difference in sensitivity to acetazolamide between Schaffer collaterals in CA1 and CA2/CA3 border region could be that the distribution of carbonic anhydrase or membrane transporters influencing pH differ between these subfields. In conclusion, our results suggest that carbonic anhydrase-dependent pH regulation attenuates activity-induced decrease in excitability of the distal parts of the Schaffer collateral axons. To our knowledge, such effect has not previously been observed. As rapid pH shifts may not be optimally buffered despite the presence of carbonic anhydrase (Kaila and Chesler 1998), it seems likely that seizure activity even with intact buffering could lower axonal conduction properties through acidic shift.
The present findings open the possibility that the mechanisms of feedback regulatory effect of pH on epileptic activity (de Curtis et al. 1998, Pavlov et al. 2013, Raimondo et al. 2015) may involve presynaptic elements. More detailed studies will be needed to clarify how pH dynamics in neuronal subdomains influence epileptic activity (and vice versa) under conditions where synaptic transmission is intact.

Preparation of brain slices
Animal care and protocol for euthanasia were in accordance with Danish and European law and approved by the Animal experimentation Board under the Danish Ministry of Justice. Male Wistar rats (4 -5 w, n = 44) were anesthetized with isoflurane and decapitated. After removal, the brain was placed in dissection medium at 4 • C (see below), and horizontal slices containing the hippocampus (400 μm) were cut on a vibrotome. One slice was immediately transferred to the recording chamber, where it was placed on a nylon-mesh grid at the interface between warm (31 -32 • C) standard perfusion medium (see below) and warm humidified carbogen (95% O 2 , 5% CO 2 ). Perfusion flow rate was 1 ml/min. The slice rested for at least one hour before electrophysiological recordings began. The remaining slices were stored in dissection medium bubbled with carbogen at room temperature until use.

Electrophysiology and analyses
Extracellular recordings were obtained using borosilicate glass electrodes (1.2 mm OD; Clark Electromedical, Pangbourne, UK) filled with 1 M NaCl (tip resistance 10 -20 MΩ). Conventional recording techniques were employed, using a high-input impedance amplifier (Axoclamp2A, Molecular Devices, USA). Signals were digitized on line using a Digidata 1440 interface and transferred to a computer for analysis employing pCLAMP (version10, Molecular Devices). For axonal stimulation, a bipolar Teflon-insulated platinum electrode (tip diameter: 50 μm, intertip distance: 25 μm) was placed on the slice surface. Constant-current pulses (50 μs) was delivered from a stimulus-isolation unit (Axon-11, Molecular devices, US). In all experiments, we first used a standard electrode arrangement with the stimulus electrode placed in str. radiatum of CA1 to stimulate the Schaffer collaterals orthodromically. The recording electrode was placed in str. radiatum of CA1 ≥ 500 µm from the stimulation site (Fig. 1a). Slices were accepted for recording if a single stimulus (400 μA) evoked a normal field EPSP with a single population spike of 5 -15 mV amplitude. The field EPSP was subsequently blocked by perfusing antagonists of synaptic transmission (see Results) for 25-30 min. After blockade, we confirmed that a stimulusinduced compound action potential (fiber volley) was present and isolated from the stimulus artifact. We used either of three different Fig. 2. Effects of acetazolamide on HFS-evoked fiber volleys in hippocampal slices. Results from three experimental settings are shown: a, recording and stimulation in CA1 str. radiatum during perfusion of ACSF with low [Ca 2+ ]. b, recording close to the border between CA2 and CA3 while stimulating Schaffer collaterals in CA3 (ortodromic) or CA1 (antidromic). c, recording and stimulating within a physically isolated part of the alveus of CA1. a1-c1 top: schematics showing electrode placements. Middle: example records of fiber volleys evoked during 100 Hz trains, at stimulus #1, #10, and #100, before (control) and during perfusion of acetazolamide (1 mM). Calibrations: 1 mV, 1 ms. Bottom: from the same experiment, fiber volley area, amplitude, and latency are plotted as function of stimulus number. a2-c2: summary histograms of the normalized area, amplitude and latency of fiber volleys evoked at stimulus #10 and #100 before (con), during acetazolamide perfusion (actz), and after wash out. Data obtained with acetazolamide did not differ significantly from control before or after wash out in any of the recording conditions. a3-c3, top: fiber volleys evoked by stimulus #1 and 1 s post-train from control period (full line) and during acetazolamide (broken line). Calibrations: 1 mV, 1 ms. Bottom: summary plots of fiber volley area, amplitude, and latency as function of time after the 100 Hz train in control period and during perfusion of acetazolamide. a4-c4: Summary histograms of the normalized change in area, amplitude, and latency of fiber volleys evoked 1 s after a 100 Hz train compared to the first fiber volley in the train. Changes observed with acetazolamide did not differ significantly from the controls before or after wash out in any of the recording conditions. Fig. 3. Effects of NH 4 Cl (5 mM) on fiber volleys evoked at low frequency (0.2 Hz). Values are represented as normalized deviation from controls obtained before adding NH 4 Cl. Samples were averaged over a period of 1 min immediately before NH 4 Cl, at the end of a 7 min wash in period (NH 4 + ), and after 5 min washout (wash). n = 8---9 from each of the three locations indicated. Asterisks denotes statistically significant deviation from control values. * p > 0.05, ** p > 0.01. electrode setups, one of which was the standard arrangement described above. In the second setup, the recording electrode was moved to str. radiatum close to the border between CA2 and CA3 (Masukawa et al. 1982). The stimulation electrode was placed either in CA1 (≥500 μm from the recording electrode) for antidromic stimulation, or in CA3 (≥300 μm from the recording electrode) for orthodromic stimulation of the Schaffer collateral pathway (Fig. 2b). In the third setup, the stimulation and recording electrodes were both placed in the alveus of CA1 (≥300 μm separation) to record fiber volleys from the axons of CA1 pyramidal neurons. Prior to these recordings, a knife cut was made along the border between alveus and str. oriens, separating the two layers. In some of the slices an additional cut was made across the alveus, and the free edge was subsequently moved away from the str. oriens (Fig. 2c).
Stimulation strength was set to generate a fiber volley of ≥ 1.5 mV (range: 150 -800 μA). Experiments were included in the analyses if the fiber volley was free of the stimulus artifact and consisted of a smooth negative deflection with a single peak surrounded by positive deflections (Fig. 1a). High frequency stimulation (HFS) was delivered in trains of 100 stimuli at 100 Hz. The interval between trains was set to 50 s, which allowed full recovery of the fiber volley. Beginning at 1 s after the HFS train, fiber volleys were evoked at low frequency (0.5 Hz) to monitor the time course of recovery. Signals were filtered at 2 kHz (low pass) before measurement. Data from three consecutive trains were used for analysis. Fiber volley amplitudes were measured as the voltage difference between the negative peak and the peak of the following positive deflection. Fiber volley areas were taken as the integral of the negative voltage trajectory under a fictive line between the peaks of the positive deflections on either side. Latencies were measured from the beginning of the stimulus artifact to the negative peak (Fig. 1a). In each experimental condition, values obtained during and after HFS trains were normalized to the corresponding values from the first stimulus in the train.
Unless otherwise noted, we used the one-way ANOVA combined with Holm-Sidak post hoc test for statistical comparisons.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.