For revealing the effect of acute and transient metabolic failure on glutamatergic synaptic transmission, we combined electrophysiological monitoring of glutamatergic synaptic transmission at CA3-CA1 Schaffer collateral synapses in acute hippocampal slices with a chemical ischemia model. Baseline synaptic transmission was monitored for >10 min before ATP production was disrupted by switching to a modified extracellular solution lacking glucose and containing sodium azide and 2-deoxyglucose to inhibit oxidative phosphorylation and glycolysis, respectively (schematic in Figure 1A). This protocol for ‘chemical ischemia’ has previously been shown to reliably induce a rapid and reversible decline of the intracellular ATP levels (Pape and Rose, 2023). We first tested the effect of acute and transient chemical ischemia lasting for 2–5 min and observed two qualitatively distinct results. A comparison of recordings with 2 and 5 min illustrates both scenarios (Figure 1B–F). Although the postsynaptic response transiently disappeared for both durations, they remained suppressed over the next 50 min of recording for 5 min of chemical ischemia but recovered when chemical ischemia lasted only 2 min, in which case it showed a significant potentiation at the end of the recording (Figure 1B, C, and E). In contrast, the axonal fiber volley, which reflects the axonal firing of action potentials after stimulation, was transiently suppressed and recovered fully for both durations (Figure 1D). Analyzing the entire data set including chemical stress durations of 3 and 4 min confirmed that the axonal fiber volley robustly recovered from chemical ischemia for the tested durations (Figure 1G). In contrast, the postsynaptic response was either potentiated for shorter durations or almost completely suppressed for longer durations (Figure 1H). For chemical ischemia with near-complete suppression of the postsynaptic response (Figure 1H, orange data points), the average duration of chemical stress was 4.46±0.207 min (n=11) and 2.74±0.20 min (n=19) for all other recordings. Because during chemical ischemia the cellular depolarization and ensuing increased neuronal synaptic glutamate release should lead to an extracellular potassium increase, we measured extracellular [K⁺] in parallel in a subset of experiments (example in Figure 1F). Correlating the maximum increase of the extracellular [K⁺] during chemical ischemia with the lasting effect on synaptic transmission revealed that a stronger extracellular [K⁺] increase (i.e. a severe metabolic failure) made postsynaptic failure more likely (Figure 1I). Together, these experiments reveal two distinct outcomes of acute chemical ischemia: either potentiation of the synaptic response or its near-complete suppression, which represents a postsynaptic failure of synaptic transmission as revealed by the following glutamate imaging experiments.

In a subset of the recordings shown in Figure 1H (filled data points), we have combined electrophysiological recordings with monitoring of glutamate release using the fluorescent indicator iGluSnFR (Marvin et al., 2013; Figure 2). In this subset, iGluSnFR was virally expressed in astrocytes and its fluorescence intensity changes were monitored in response to paired stimulation of CA3-CA1 synapses by line scanning throughout the experiment as indicated in Figure 2A and B. Additional control experiments confirmed the stability of such combined recordings (Figure 2C and F). We grouped recordings combining iGluSnFR imaging and electrophysiology according to the effect of chemical ischemia on the synaptic response: ‘chemical ischemia with postsynaptic failure’ (Figure 1H, filled orange data points) if the postsynaptic response did not recover to above 50% of the baseline level and ‘chemical ischemia’ when it did (Figure 1H, filled dark gray data points). Similar to electrophysiological recordings, two response types could be identified in iGluSnFR recordings. When acute and transient chemical ischemia led to a persistent potentiation of synaptic transmission, this was associated with an increase of the iGluSnFR responses (ΔF/F₀) (Figure 2D). In contrast, if chemical ischemia led to postsynaptic failure, we observed a transient surge of the resting fluorescence of iGluSnFR (F₀) during acute chemical ischemia and reduced and gradually recovering iGluSnFR transients in response to synaptic stimulation after chemical ischemia (Figure 2E). Presented are six out of the nine experiments with postsynaptic failure that were performed. Three recordings were not further analyzed because the iGluSnFR fluorescence became undetectable after chemical ischemia. A more comprehensive statistical analysis of the responses before and after chemical ischemia can be found in Figure 2—figure supplement 1 and Figure 2F, G, and H. In a single experiment, we have been able to record the response to two subsequent periods of short chemical ischemia in a >3 hr recording (Figure 2—figure supplement 2), which highlights the persistence of the potentiation of both the postsynaptic responses and glutamate transients. In addition to the bidirectional changes of synaptic transmission and glutamate release depending on the severity of chemical ischemia, we also found a consistent decrease of the iGluSnFR resting fluorescence over the duration of the experiments in all conditions. In general, this could be caused by a gradual reduction of the resting extracellular glutamate levels and/or a reduction of fluorescent iGluSnFR, for instance due to bleaching, which appears to be a likely explanation given the repeated imaging over up to 90 min. It is important to note that the increase of the iGluSnFR transients (ΔF/F₀) associated with potentiated postsynaptic responses is unlikely due to the reduction of F₀ because the latter is observed in all experimental conditions. Similarly, it is unlikely that the increase of the iGluSnFR transients is because of the increased axonal excitability (increased fiber volley amplitude) because this was also observed in control recordings, where no change of the iGluSnFR response (ΔF/F₀) was detected. The increase of the iGluSnFR response after synaptic stimulation also points to an increase of extracellular glutamate transients, because the iGluSnFR response to saturating glutamate concentrations is unaffected in experiments with chemical ischemia without postsynaptic failure (Figure 2—figure supplement 3).

Taken together these experiments reveal that longer durations of chemical ischemia result in a persistent postsynaptic failure with slowly recovering glutamate release. A likely reason for the postsynaptic failure is the death or persistent depolarization of CA1 pyramidal cells. Although not quantified, we indeed observed irreversible swelling and/or loss of integrity of CA1 pyramidal cell bodies in parallel to postsynaptic failure. This is in line with previous studies using transient metabolic inhibition with similar durations, which reported irreversible ATP depletion in neocortical neurons after transient chemical ischemia (Pape and Rose, 2023) and persistent depolarization of CA1 pyramidal cells after hypoxia combined with glucose deprivation (Tanaka et al., 1997). The beginning of this postsynaptic failure coincides with a surge of extracellular glutamate (Figure 2E), which we attribute to the increase of synaptic glutamate release at the onset of metabolic stress observed before (Hershkowitz et al., 1993; Tanaka et al., 1997), and of extracellular potassium (Figure 1F and I). Both are likely driven by neuronal depolarization, Ca²⁺ accumulation, and the various other contributing mechanisms (Passlick et al., 2021). In contrast, shorter durations of chemical ischemia durations lead to a transient failure of synaptic transmission, possibly due to a transient depolarization of neurons, followed by an unexpected potentiation of synaptically driven glutamate transients and of postsynaptic responses.

We next asked what the underlying reason for the increase glutamate transients is. One potential explanation is a restriction of extracellular space (ECS): If a given amount of glutamate is released into the ECS leading to a corresponding glutamate concentration, then a reduction of the ECS would increase the glutamate concentration. Indeed, anoxia and ischemia consistently lead to a reduction of the ECS fraction (Syková and Nicholson, 2008). This scenario was investigated using the method of combining tetramethylammonium (TMA) iontophoresis and TMA-selective microelectrode recordings for estimating the ECS fraction and extracellular diffusion (Nicholson, 1993; Hrabětová and Nicholson, 2007; Syková and Nicholson, 2008) in parallel to chemical ischemia induction. We found that chemical ischemia with postsynaptic failure was accompanied by a decreased ECS fraction and a decreased effective diffusion coefficient, which is consistent with previous reports (Syková and Nicholson, 2008; Figure 2—figure supplement 4). However, this was not observed for chemical ischemia without postsynaptic failure (Figure 2—figure supplement 4), in which the increased glutamate transients were observed.

An alternative explanation for the increased glutamate transients is reduced glutamate uptake. To investigate this possibility, we analyzed the decay time constant of the iGluSnFR fluorescence transients after synaptic stimulation, because it is increased when glutamate uptake is decreased (Armbruster et al., 2016; Romanos et al., 2019). However, we did not detect a significant change of the iGluSnFR decay time constant after chemical ischemia (Figure 3A and B). In addition, we tested if a deficit of glutamate uptake could be unmasked by increasing the amount of synaptically released glutamate. This was tested by high-frequency stimulation of CA3-CA1 synapses in the presence of glutamate receptor blockers (to prevent induction of synaptic plasticity, Figure 3C). Again, no differences of the decay time constant were observed between control recordings and recordings with chemical ischemia (Figure 3E). Finally, we verified the sensitivity of our experimental approach by inhibiting glutamate uptake pharmacologically at the end of the experiments. This manipulation indeed increased the decay time constant by a factor of ~10–20 (Figure 3D, F, and G). We therefore conclude that glutamate uptake is surprisingly resilient in the chemical ischemia model and that a reduction of glutamate uptake does not explain the increased extracellular glutamate transients after chemical ischemia without postsynaptic failure.

Our results point toward an increased synaptic glutamate release when chemical ischemia is shorter and followed by postsynaptic potentiation but not when it is associated with postsynaptic failure. Investigating the paired-pulse behavior of synaptic responses is a common approach for probing changes of presynaptic glutamate release (Debanne et al., 1996). Analyzing the paired-pulse ratio (PPR) of postsynaptic response and iGluSnFR transients revealed no consistent changes after chemical ischemia (Figure 4—figure supplement 1). However, even manifold presynaptically induced increases of synaptic strength were recently shown to result in relatively small changes of the PPR, especially of the iGluSnFR signal at single CA3-CA1 synapses (Dürst et al., 2022). This suggests that a more moderate presynaptic increase of glutamate release may escape its detection using the PPR. Therefore, we used an electrophysiological approach that directly detects changes of the glutamate concentration in the synaptic cleft. γ-D-glutamylglycine (γ-DGG) is a rapidly equilibrating competitive antagonist of AMPA receptors, and its inhibitory effect on AMPA receptor mediated postsynaptic responses depends on the ambient glutamate concentration because glutamate and γ-DGG compete for AMPA receptor binding (Liu et al., 1999; Christie and Jahr, 2006): The lower the glutamate concentration in the synaptic cleft, the stronger the inhibitory effect of γ-DGG, and vice versa. For performing this test after chemical ischemia without postsynaptic failure, we combined our previous experimental approach with whole-cell patch clamp recordings from CA1 pyramidal cells starting ~45 min after induction of chemical ischemia (Figure 4A). Analyzing the effect of γ-DGG application on simultaneously recorded field excitatory postsynaptic potentials (fEPSPs) and excitatory postsynaptic currents (EPSCs), we found that the percentages of fEPSP slopes and EPSC amplitudes remaining after γ-DGG were significantly higher in experiments with chemical ischemia compared to control experiments (Figure 4B and C). This indicates that chemical ischemia led to an increased peak glutamate concentration in the synaptic cleft, which is consistent with an increased synaptic glutamate release. Thus, shorter durations of chemical ischemia led to a potentiation of presynaptic glutamate release and of the postsynaptic response.

We further explored the underlying reason for this increase of glutamate release by computational modeling. First, we tested the hypothesis that presynaptic conditions expected for metabolic failure could increase vesicle loading with glutamate. This was analysed using a previously established model for simulating presynaptic vesicle loading (Kolen et al., 2023). As shown in Figure 4—figure supplement 2, accumulation of glutamate in the synaptic vesicle under control conditions reaches a steady state of close to 3500 molecules. Cytosolic conditions consistent with metabolic failure (decreasing pH, increasing [Na⁺] and [Cl^-], decreasing [K⁺]) all lead to a smaller number of glutamate molecules in the vesicle. Similarly, a decrease in the extracellular pH and a reduced vesicular ATPase activity (due to reduced ATP levels) also cause a decrease in the luminal glutamate concentration. As a control, we increased the cytosolic pH from 7.2 to 7.4 and observed a significant rise in the number of glutamate molecules. However, we are not aware of any reports regarding a presynaptic alkalinization during ischemia or chemical ischemia. Overall, these simulations show that the conditions that are typically observed during ischemia or metabolic failure tend to lower the vesicular glutamate concentration and therefore cannot explain the observed increased glutamate release. An alternative explanation is an increased probability of presynaptic vesicular release, which could result from increased presynaptic Ca²⁺ influx, for instance, due to a presynaptic broadening of action potentials. For testing this, we analyzed the axonal fiber volley recorded in previous experiments (Figure 3C and D), in which glutamate receptors were blocked and the fiber volley was therefore isolated (Figure 4D). We found that the amplitude of the fiber volley was not different between control recordings and recordings after chemical ischemia without postsynaptic failure (Figure 4E, left panel), which confirms previous conclusions (Figure 2). However, the fiber volley was significantly broadened by ~20% (Figure 4E, middle panel) without a significant change of its delay after stimulation (Figure 4E, right panel). This could indicate that short transient chemical ischemia broadens the presynaptic action potential, which increases presynaptic Ca²⁺ influx and presynaptic glutamate release and, thereby, the postsynaptic response.

Experiments were performed using 2- to 4-month-old male C57BL/6N mice (Charles River) that were kept under 12 hr light/dark conditions and had ad libitum access to food and water. All animal procedures were conducted in accordance with the regulations of the European Commission Directive 2010/63/EU and all relevant national and institutional guidelines and requirements. Procedures were further approved by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV, Recklinghausen, Germany) where required.

In order to express the glutamate sensor iGluSnFR (Marvin et al., 2013) on the surface of astrocytes, 4-week-old mice were injected bilaterally with an AAV expressing iGluSnFR under control of the GFAP promoter (AAV1.GFAP.iGluSnFr.WPRE.SV40, PennCore) into the CA1 region of the dorsal hippocampus. Mice were deeply anesthetized by intraperitoneal injection (i.p.) of fentanyl/midazolam/medetomidine (0.05/5.0/0.5 mg/kg bodyweight, injection volume 0.1 ml/10 g bodyweight). After deep anesthesia was confirmed, the head was shaved, the skin disinfected, and the head fixed in a stereotactic frame (Model 901, David Kopf Instruments). An incision was made, bregma localized, and a small hole was drilled with a dental drill (coordinates for the dorsal hippocampus, relative to bregma: anterior −1.8 mm, lateral ±1.6 mm, ventral −1.6 mm). Next, a beveled needle nanosyringe (nanofil 34 G BVLD, WPI) was slowly inserted into the brain and 0.5–1 µl of viral particles were injected into the hippocampus using a microinjection pump (50 nl/min, WPI). The injection needle was left in place for about 3 min and then slowly retracted. After repeating the procedure for the other hemisphere, the incision was sutured using absorbable thread (Ethicon) and an antibiotic was applied (Refobacin 1 mg/g, Gentamicin). Finally, anesthesia was terminated by i.p. injection of naloxone/flumazenil/atipamezole (1.2/0.5/2.5 mg/kg bodyweight, injection volume 0.1 ml/10 g bodyweight). Analgesia was applied 30 min before terminating the anesthesia and 24 hr after surgery by subcutaneous injection of carprofen (5 mg/kg bodyweight, injection volume 0.1 ml/20 g bodyweight). Animals were used for experiments 2–5 weeks after virus injection.

Electrophysiology and two-photon excitation fluorescence microscopy in acute hippocampal slices were combined as previously described (Minge et al., 2017). Briefly, coronal slices were obtained from mice and virus-injected mice (see above) with a thickness of 300 µm. Slices were prepared in an ice-cold slicing solution containing (in mM): NaCl 60, sucrose 105, KCl 2.5, MgCl₂ 7, NaH₂PO₄ 1.25, ascorbic acid 1.3, sodium pyruvate 3, NaHCO₃ 26, CaCl₂ 0.5, and glucose 10 (osmolarity 300–310 mOsm/l), and kept in the slicing solution at 34°C for 15 min before being stored at room temperature (21–23°C) in an artificial cerebral spinal fluid (ACSF) containing (in mM): NaCl 131, KCl 2.5, MgSO₄ 1.3, CaCl₂ 2, NaH₂PO₄ 1.25, NaHCO₃ 21, and glucose 10 (osmolarity adjusted to 295–305 mOsm/l). Slices were allowed to rest for at least 60 min. For recordings, slices were transferred to a submersion-type recording chamber and perfused with extracellular solution (ACSF). All recordings were performed at 33–35°C. All solutions were continuously bubbled with 95% O₂/5% CO₂. The flow rate of the extracellular solution was kept constant at 6–7 ml/min throughout all experiments.

Extracellular recordings of fEPSPs were performed using patch pipettes filled with extracellular solution placed in the stratum radiatum of the CA1 region. For stimulation of CA3-CA1 Schaffer collateral axonal connections, a bipolar concentric stimulation electrode was placed in the stratum radiatum at the border between CA2/3 and CA1. Stimulation intensities (DS3, Digitimer Ltd., UK) were set to obtain ~50% of the maximum fEPSP amplitude. The duration of individual stimuli was 100 µs. In some experiments, field recordings were combined with whole-cell patch clamp recordings from CA1 pyramidal cells using standard patch pipettes (3–4 MΩ) filled with an intracellular solution containing (in mM): KCH₃O₃S 135, HEPES 10, di-tris-phosphocreatine 10, MgCl₂ 4, Na₂-ATP 4, Na-GTP 0.4, 0.2 BAPTA, 5 QX314-Cl (pH adjusted to 7.2, osmolarity 290–295 mOsm/l). CA1 pyramidal cells were identified using infrared differential interference contrast or Dodt contrast optics by their typical morphology and location. Data were recorded using MultiClamp 700B (Molecular Devices) amplifiers, digitized (10 kHz) and stored for offline analysis. Whole-cell patch clamp recordings were rejected if at any time the access resistance exceeded 25 MΩ or changed by more than 20%.

Acute chemical ischemia was induced by switching from normal ACSF to a modified ACSF with 0 mM glucose, 2 mM deoxyglucose (2-DG) to inhibit ATP production through glycolysis, and 5 mM sodium azide (NaN₃) to inhibit ATP production through oxidative phosphorylation. The modified ACSF was applied for variable durations ranging from 2 to 5 min. In some experiments, the inhibitors D-AP5 (dAP5 in figures, 50 µM, Abcam), NBQX disodium salt (10 µM, Abcam), DL-TBOA (100 µM, Tocris), LY341495 (50 µM, Tocris), MPEP (10 µM, Tocris), and/or γ-DGG (1 mM, Tocris) were added to the extracellular solution as indicated.

Glutamate imaging using iGluSnFR (Marvin et al., 2013) in combination with electrophysiology was performed similarly as previously described (Herde et al., 2020). Slices with astrocytes expressing iGluSnFR after viral injection (see above) were transferred to a submersion-type recording chamber mounted on a Scientifica two-photon excitation fluorescence microscope with a ×60/1.0 NA objective (Olympus). iGluSnFR was excited using a femtosecond pulse laser (Vision S, Coherent, excitation wavelength of 910 nm). During imaging, line scans of iGluSnFR fluorescence were repeatedly obtained. The scanned line was positioned in the periphery of an iGluSnFR-expressing astrocytes (e.g. Figure 2B). The line was scanned at a frequency of 378.8 Hz to obtain a fluorescence profile over time. Line scans were repeated four to six times for each individual time point and experiment and averaged before the background fluorescence was subtracted to obtain an average iGluSnFR fluorescence time course. The resting iGluSnFR fluorescence before synaptic stimulation (F₀), the maximum increase of iGluSnFR fluorescence (ΔF), and the normalized iGluSnFR time course (ΔF/F₀) were determined. iGluSnFR imaging was combined with paired synaptic stimuli (Figure 2), high-frequency synaptic stimulation (Figure 3), and iontophoretic glutamate application (Figure 2—figure supplement 2). For the latter, an iontophoretic application system (MVCS-02C-150, NPI) was used with iontophoretic pipettes (60–80 MΩ resistance) filled with 150 mM glutamic acid (pH adjusted to 7.0 with NaOH) and 40 µM Alexa Fluor 633 to localize the pipette.

Extracellular [K⁺] was recorded using K⁺-sensitive microelectrodes as previously described (Breithausen et al., 2020). Briefly, these microelectrodes were built using theta-glass capillaries and pulled with a horizontal pipette puller. The reference and the K⁺-sensitive barrel were filled with an NaCl (154 mM) and KCl (150 mM) solution, respectively. The tip of the K⁺-sensitive barrel was filled with a valinomycin-based K⁺ ionophore (K⁺ ionophore 1, cocktail B, Sigma-Aldrich). The K⁺-sensitive microelectrodes were calibrated before each experiment with solutions containing 154 mM NaCl and 3 or 30 mM KCl. Electrodes were used if they responded to this 10× increase of K⁺ with a voltage response >50 mV.

Measurements of the relative changes of the ECS fraction and of extracellular diffusion during experiments with acute chemical ischemia were obtained using the method of combining TMA iontophoresis and TMA-sensitive electrode recordings (Nicholson, 1993; Hrabětová and Nicholson, 2007; Syková and Nicholson, 2008). TMA-sensitive, double-barrel microelectrodes were built like K⁺-sensitive microelectrodes except that the TMA-sensitive ionophore solution (IE190, WPI) and 150 mM TMA chloride, as the intrapipette solution, were used for the TMA-sensitive barrel. The reference barrel was filled with 154 mM NaCl and 60 µM Alexa Fluor 594 to visualize the electrode. Recordings were performed using an ION-01M amplifier (NPI, Germany). TMA-sensitive microelectrodes were tested for their sensitivity to TMA before each experiment by two calibration solutions containing either (in mM) TMA 1, KCl 3, NaCl 150 or TMA 10, KCl 3, NaCl 141. Only electrodes that responded to this 10-fold increase of TMA with a potential increase of >50 mV were used for the experiments. A subset of electrodes was also tested for their sensitivity over a larger range of TMA concentrations, which revealed Nernst-like behavior over higher concentrations and only a departure from that in the low sub-millimolar range. To correct for that, the calibration curve was fitted with the Nicolsky-Eisenman equation and the conversion of the TMA-electrode potential into a TMA concentration was adapted accordingly (Nicholson, 1993). For recordings in chemical ischemia experiments (Figure 2—figure supplement 4), the TMA-sensitive microelectrode was placed into the CA1 stratum radiatum and an iontophoresis pipette filled with 100 mM TMA and 60 µM Alexa Fluor 594 was placed at distance of 100–200 µm from the TMA-sensitive microelectrode. At each time point during the recording, TMA was iontophoretically injected into the tissue (100 nA, 30 s) and the corresponding [TMA] time course was calculated. For example, see Figure 2—figure supplement 4. Each time course was fitted using equations 10.2–10.4 from Hrabětová and Nicholson, 2007, using a MATLAB script to obtain the ECS fraction and the effective TMA diffusion coefficient (D*). D* deviates from the free diffusion coefficient of TMA (D) because the tortuosity of the ECS reduces TMA diffusion over the distances investigated here (Hrabětová and Nicholson, 2007). Therefore, changes of D* can indicate a change of ECS tortuosity. In these experiments, the extracellular solution always contained 1 mM TMACl to provide a known basal extracellular TMA concentration for the conversion of the recorded microelectrode potential into [TMA]. The reference barrel of the TMA-sensitive microelectrode was also used for recording fEPSPs in response to axonal stimulation. We have verified these recordings to report accurate values by recording [TMA] transients in agarose gel (ECS fraction ~1, tortuosity ~1) (Nicholson, 1993) and in acute brain slices, in which we obtained an ECS fraction of 0.19±0.023 (n=5) and a tortuosity of 1.54±0.046 (n=5) and which matches previous measurements in this region (Nicholson, 1993; Syková and Nicholson, 2008).

The accumulation of glutamate in synaptic vesicles is modeled as previously described (Kolen et al., 2023) and the code for the model is archived along with it. Briefly, the temporal evolution of luminal glutamate molecule number, [H⁺], pH, and [Cl^-] are modeled by the following rate equations:, (1)d[Glut]Ldt=JVGLUT, (2)d[H+]Ldt=JH++JVATPase−JVGLUT, (3)dpHLdt=−1NA(dHL+dt1V)(1βpH)(4)d[Cl−]Ldt=JVGLUTCl.

where N_A, β_pH (=40mMpH), and V represent Avogadro’s number, the pH-buffering capacity of the vesicle, and the vesicle volume assuming a radius of 20 nm, respectively. The equations for glutamate/H⁺ flux through glutamate transporters (J_VGLUT), Cl^- flux through glutamate transporters (JVGLUTCl), proton leak (JH+), and V-type ATPase (JVATPase) are described in Kolen et al., 2023.

The equation for the membrane potential of the vesicle (∆ψ) is(5)Δψ=F×VCH+L+K+L+Na+L-Cl-L-GlutL-B

with C=C0×S being the total capacitance of the vesicle with C0=1 μF/cm², S the surface area of the vesicle, and F Faraday’s constant. Luminal K⁺ ([K+]L) and Na⁺ ([Na+]L) concentrations were fixed to extracellular values, i.e., to 5 and 145 mM, respectively. B is the luminal concentration of impermeant charges given by the conservation of charge in the vesicle under initial conditions (indicated by the subscript 0).(6)B=H+L,0+K+L,0+Na+L,0-Cl-L,0-GlutL,0-CF×Vψin-ψout

Finally, the effect of surface charge on various ion concentrations at the membrane was also taken into account, with the inner (ψin) and outer (ψout) leaflets potentials set to 0 and –50 mV, respectively.

Starting conditions for the mathematical modeling reflect the ion concentrations immediately after endocytosis of synaptic vesicles: the vesicular lumen contained solutions resembling the external solutions, with high [Cl^-], neutral pH, and negligible glutamate concentration ([K+]L,0=5 mM, [Na+]L,0= 145 mM, [Cl−]L,0 = 110 mM, [Glut]L,0 = 1 μM, and pH_L,0 = 7.2); cytoplasmic pH, Cl^-, K⁺, Na⁺, and glutamate concentration was set to 7.2, 10 mM, 140 mM, 10 mM, and 10 mM, respectively. We also incorporated changes in luminal Na⁺ and K⁺ concentrations, however, both did not affect our results.

Analyses were performed using ImageJ (NIH), Origin (OriginLab), and MATLAB (MathWorks). Numerical data are reported as mean ± s.e.m. with n being the number of samples. In all electrophysiological and imaging experiments in acute hippocampal slices, n refers to the number of recordings. Because the success rate of experiments combining electrophysiology, two-photon excitation imaging, and pharmacology is low, a single successful experiment was typically performed per day and animal. For all other experimental designs, detailed information can be found in the figure legends. The Shapiro-Wilk test was used to establish if data were normally distributed. Comparisons were then performed using the appropriate parametric and non-parametric tests (e.g. Student’s t-test or Mann-Whitney U-test). The statistical tests used are indicated throughout. In figures, asterisks indicate statistical significance. Exact p values are provided in the figure legends. Error bars represent s.e.m. and dots represent individual data points, which are connected by lines for paired data points.