Mannitol decreases neocortical epileptiform activity during early brain development via cotransport of chloride and water

Abstract

Seizures and brain injury lead to water and Cl⁻ accumulation in neurons. The increase in intraneuronal Cl⁻ concentration ([Cl⁻]_i) depolarizes the GABA_A reversal potential (E_GABA) and worsens seizure activity. Neocortical neuronal membranes have a low water permeability due to the lack aquaporins necessary to move free water. Instead, neurons use cotransport of ions including Cl⁻ to move water. Thus, increasing the extracellular osmolarity during seizures should result in an outward movement of water and salt, reducing [Cl⁻]_i and improving GABA_A receptor-mediated inhibition. We tested the effects of hyperosmotic therapy with a clinically relevant dose of mannitol (20 mM) on epileptiform activity, spontaneous multiunit activity, sIPSCs, [Cl⁻]_i, and neuronal volume in layer IV/V of the developing neocortex of C57BL/6 and Clomeleon mice. Using electrophysiological techniques and multiphoton imaging in acute brain slices (post-natal day 7-12) and organotypic neocortical slice cultures (post-natal day 14), we observed that mannitol: 1) decreased epileptiform activity, 2) decreased neuronal volume and [Cl⁻]_i through CCCs, 3) decreased spontaneous multi-unit activity frequency but not amplitude, and 4) restored the anticonvulsant efficacy of the GABA_A receptor modulator diazepam. Increasing extracellular osmolarity by 20 mOsm with hypertonic saline did not decrease epileptiform activity. We conclude that an increase in extracellular osmolarity by mannitol mediates the efflux of [Cl⁻]_i and water through CCCs, which results in a decrease in epileptiform activity and enhances benzodiazepine actions in the developing neocortex in vitro. Novel treatments aimed to decrease neuronal volume may concomitantly decrease [Cl⁻]_i and improve seizure control.

Introduction

The effect of the neurotransmitter GABA can be either inhibitory or excitatory depending on the relationship between the reversal potential of GABA (E_GABA) and the neuronal membrane potential. Values of E_GABA far above the neuronal membrane potential lead to excitatory actions of GABA; intermediate values of E_GABA lead to shunting inhibition, and values of E_GABA below the membrane potential result in inhibitory actions of GABA (Doyon et al., 2016; Glykys et al., 2017). E_GABA is mainly set by the concentration of intracellular and extracellular Cl⁻ with a small influence from bicarbonate, as Cl⁻ has a higher GABA_AR permeability and concentration compared to bicarbonate (Bormann et al., 1987).

Immobile anions have been proposed to set the intraneuronal Cl⁻ concentration ([Cl⁻]_i) while Cl⁻ co-transporters (CCCs) are key in the maintenance of this Cl⁻ set-point, or to return the [Cl⁻]_i to its baseline concentration following disruptions by synaptic transmission and/or pathological changes (Glykys et al., 2014a, 2017; for further discussion see Glykys et al., 2014b; Luhmann et al., 2014; Voipio et al., 2014). Importantly, CCCs move water besides Cl⁻ and cations making these cotransporters key players in the processes of regulatory volume decrease and increase observed in different cells, including neurons (Gauvain et al., 2011; Hamann et al., 2010; Lang et al., 1998; MacAulay and Zeuthen, 2010; Zeuthen, 1991; Zeuthen and Macaulay, 2012).

High neuronal [Cl⁻]_i is observed in different brain regions, including the neocortex, during early brain development (Ben-Ari et al., 2007; Glykys and Staley, 2016; Sulis Sato et al., 2017) and results in depolarizing actions of GABA (Glykys et al., 2009; Kirmse et al., 2015; Rheims et al., 2008). Elevated neuronal [Cl⁻]_i is also observed in pathological conditions including gliomas (Pallud et al., 2014), during reperfusion after oxygen-glucose deprivation (Blauwblomme et al., 2018; Pond et al., 2006), and seizures (Dzhala et al., 2010; Glykys et al., 2009; Huberfeld et al., 2007). Importantly, high neuronal [Cl⁻]_i contributes to anticonvulsant inefficacy during neonatal seizures (Dzhala et al., 2005; Glykys et al., 2009; Glykys and Staley, 2015; Mazarati et al., 2009; Nardou et al., 2011b), a condition which is difficult to treat in humans (Glass, 2014; Painter et al., 1999; Stevenson et al., 2016).

Cytotoxic edema is caused by neuronal water shifts coupled with ionic movements, especially of Cl⁻ (Allen et al., 2004; Dijkstra et al., 2016; Rungta et al., 2015), and it is observed in different neurological conditions including traumatic brain injury, seizures and ischemia (Glykys et al., 2017). Mannitol and hypertonic saline are used frequently in clinical practice to treat increased intracranial pressure which is a consequence of severe cytotoxic and vascular edema (Carney et al., 2016; Ropper, 2014). These osmotic agents are thought to lead to the movement of pure water out of cells. Here we test whether mannitol leads to the exit of pure water out of neocortical neurons, or if there is a link between water and Cl⁻ outward movement, as neuronal membranes have a low water permeability (Aitken et al., 1998; Andrew et al., 2007; Caspi et al., 2009) due to a lack of aquaporins that are present in other cells (Lang et al., 1998). If linked, mannitol should decrease [Cl⁻]_i, make E_GABA more negative, and improve GABAergic anticonvulsant efficacy. However, if water and [Cl⁻]_i movement are unlinked, hypertonic therapies should draw pure water out of the cytoplasm. This would increase [Cl⁻]_i, make E_GABA more positive and decrease the effectiveness of GABAergic anticonvulsants.

Animals

Postnatal CLM-1 Clomeleon (P7-14; C57bl/6 background, gift from G. Augustine) and C57bl/6 (WT, P7-63; RRID:IMSR_JAX:000664) mice of both sexes were anesthetized using inhaled isoflurane and decapitated per protocol, approved by the Institutional Animal Care and Use Committee of the Massachusetts General Hospital Center for Comparative Medicine. The brain was removed and placed in ice-cold artificial cerebrospinal fluid (aCSF) containing (in mM) NaCl (120), KCl (3.3), CaCl₂ (1.3), MgCl₂ (2), NaH₂PO₄ (1.25), NaHCO₃ (25), and D-glucose (10) with pH 7.3–7.4 when bubbled with 95% O₂ and 5% CO₂ (287 ± 4 mOsm, n=9; mean ± SD). Coronal brain slices, 350-500 μm thick, were cut using a vibratome (Leica VT1000S) while submerged in aCSF containing 2 mM kynurenic acid. The brain slices were placed in an interface holding chamber containing aCSF (1.3 mM MgCl₂) at room temperature for 30 minutes. Next, the temperature was slowly increased and held at 30°C. Slices were stored fort hour minimum before being transferred to the recording chamber. For some imaging experiments the slices were incubated in Low-Mg²⁺ aCSF for at least 1 hour. Organotypic slices: P6-7 coronal 400 μm thick neocortical slices (frontal region, pre-hippocampus) were cut using a McIlwain tissue chopper. Slices were mounted in clots of chicken plasma and thrombin on poly-L-lysine-coated glass coverslips. Slices were incubated in roller tubes at 36°C with 750 μL of Neurobasal A/B27 medium supplemented with 0.5 mM GlutaMAX, 30 μg/mL gentamicin, and 20 mM NaCl to reach a final media osmolarity of 279 ± 6 mOsm (n=11; mean ± SD). The culture media was changed bi-weekly.

Imaging

Two-photon imaging was performed using a Fluoview 1000MPE two-photon microscope with pre-chirp optics and a fast acoustic-optical modulator mounted on an Olympus BX61WI upright microscope body with a 25× water immersion objective (NA 1.05). A Ti:sapphire mode-locked laser (DeepSee Mai Tai; Spectra-Physics) generated two-photon fluorescence with 860 nm excitation. Emitted light was bandpass filtered through two emission filters: 460-500 nm for cyan fluorescence protein (CFP) and 520-560 nm for yellow fluorescence protein (YFP). Two photomultiplier tubes (Hamamatsu Photonics) were used to simultaneously acquire CFP and YFP signals. Slices were perfused with Low-Mg²⁺ aCSF, held at 32-34°C and aerated with 95% O₂–5% CO₂. Three-dimensional stacks (3D) of raster scans in the XY plane (282.072 μm × 282.072 μm) were imaged at a z-axis interval of 2 μm, resolution 512 × 512 pixels.

Neuronal Cl⁻ determination

Quantitative measurements of neuronal [Cl⁻]_i and area of 3D stacks were performed using ImageJ (National Institutes of Health, RRID:SCR_003070) as described previously (Glykys et al., 2009) with modifications for area determination. The CFP and YFP z-stack images were loaded and the corresponding background level was subtracted from the entire 3D stack. Next, 3D planes were median filtered. Maximum intensity Z-projections were created every 10 μm for neuronal area determination. This allowed us to have, in a single plane, the maximal area (MaxArea) projection for each neuron. A region of interest (ROI) was drawn manually around the cell bodies (MaxArea) based on an automatic neuronal edge detection plugin in ImageJ (Canny Edge Detector plugin, Figure 3A). The ratio of the YFP/CFP fluorescence intensity was obtained for each pixel in the ROI at the z-plane corresponding to the MaxArea. Next, the median YFP/CFP ratio of the ROI was calculated. The following equation converted each cell’s YFP/CFP ratio into [Cl⁻]_i:

$${[{\mathit{Cl}}^{-}]}_i=K_D^{\prime }\frac{(R\max -R)}{(R-R\min )}$$ (1)

where K’_D is the apparent dissociation constant, Rmax is the ratio when Clomeleon is not bound to Cl⁻, and Rmin is when Clomeleon is completely quenched. The following values were used: K′_D of 91 mM, Rmax of 1.026, and Rmin of 0.268 based on our calibrations (Glykys et al., 2014a, 2009). Recalibration in 2016 did not show significant difference in these values: K’_D =81.47 mM, Rmax=1.026, Rmin=0.244. The median [Cl⁻]_i was used as its distribution is skewed to lower values (non-Gaussian distribution). Cells with [Cl⁻]_i above 130 mM were excluded from analysis (5% of imaged cells).

Electrophysiology

Acute brain slices were placed in an upright microscope (Zeiss, Axioskop) or a custom-made interphase chamber and perfused with aCSF containing 0 mM MgCl₂ (Low-Mg²⁺) to induce epileptiform activity. They were held at 32–34°C and aerated with 95% O₂–5% CO₂ (flow rate of 5–8 ml/min). Glass electrodes were placed in neocortex layer IV/V and extracellular field potentials were recorded using a low-noise multichannel amplifier (EX 4-400, Dagan Corporation) with a 1,000× gain and digitized at 2 kHz using an analog to digital converter (DigiData 1321A, Molecular Devices). Analysis was performed using a custom-written macro in IgorPro v6.35 (Wavemetrics, RRID:SCR_000325) as previously described (Glykys et al., 2009). For each 30-sec epoch, the mean value was subtracted and Fast Fourier Transformed (FFT) was obtained using a Hanning window apodization. The FFT (30,000 points) was smoothed using a running-median window of seven points, divided by the total number of points, and the signal area (1–500 Hz) was calculated (wide-band power). Finally, the mean FFT power during equal baseline and drug condition epochs was calculated. Tungsten electrodes were used to record multiunit activity (MUA). Recordings were high-pass filtered at 10 Hz and events were detected in Clampfit (Molecular Devices). Drug effects were compared between epochs prior to the drug perfusion, at the end of the drug perfusion, and at the end of washout. Pyramidal cells recordings: Layer IV/V pyramidal cells were recorded in regular aCSF in whole-cell voltage-clamp mode using pipettes filled with (in mM) 140 CsCl, 4 NaCl, 0.5 CaCl₂, 10 HEPES, 5 EGTA and 2 MgATP Ionotropic glutamate receptors were blocked with 2 mM kynurenic acid. Spontaneous inhibitory post-synaptic currents (sIPSCs) were recorded at baseline and followed by application of 20 mM mannitol at room temperature (holding potential of −75 mV). Signal acquisition was performed using a Multiclamp amplifier (Multiclamp 700B; Molecular Devices) with Clamplex 10 software (Molecular Devices). Signals were sampled at 10 KHz and analyzed using Clampfit program (Clampfit 10.6, Molecular Devices).

Experimental Design and Statistical Analysis

[Cl⁻]_i was expressed as a median and an interquartile range (IQR), as [Cl⁻]_i follows a non-Gaussian distribution. Gaussian distribution of data was tested with the Shapiro-Wilk test. Paired and unpaired t-tests were used for parametric comparisons (2 conditions). The Wilcoxon Signed Rank Test (WSRT) was used for non-parametric paired data. One-way repeated analysis of variance (ANOVA) was used for multiple comparisons of parametric data with Holm-Sidak post-hoc test. Friedman Repeated ANOVA and Kruskal-Wallis One Way ANOVA on Ranks were used for multiple comparisons of non-parametric data with Tukey post-hoc test. Kolmogorov-Smirnov test was performed to test the peak cumulative distributions of MUA. Statistical significance was set to p<0.05. IgorPro v6.35 (Wavemetrics, RRID:SCR_000325) and SigmaPlot v.11 (Systat Software, Inc., RRID:SCR_003210) were used for data analysis.

Reagents and solution osmolarity

Mannitol (two different batches including one with 99.9999% purity), kynurenic acid, 4-aminopyridine, diazepam and furosemide were all obtained from Sigma-Aldrich. Solution osmolarity was measured with a vapor pressure osmometer (Vapro 5520, Wescor Inc) calibrated each day a measurement was made. Osmolarity of the aCSF used for incubation and recordings was 289 ± 6 mOsm (n=16; mean ± SD). The organotypic slice incubation solution had an osmolarity of 290 ± 2 mOsm (n=4; mean ± SD) on experiment day (when slices were removed from roller tubes and placed in the recording chamber). Slices equilibrated in aCSF for at least 10 minutes prior to an experimental recording. Changes in aCSF osmolarity by mannitol or hypertonic saline were verified by vapor pressure osmometer (Control: 289 ± 6; Hyperosmolar solution: 310 ± 6 mOsm, n=16; mean ± SD).

Results

A clinical dose of mannitol decreases neocortical epileptiform activity during early neocortical development

Mannitol increases extracellular osmolarity, decreases cell swelling, and is used to treat increased intracranial pressure in clinical practice (Bratton et al., 2007; Pitfield et al., 2012). Neurons do not have aquaporins to mov< free water (Amiry-Moghaddam and Ottersen, 2003) and have a low water permeability (Aitken et al., 1998; Andrew et al., 2007; Caspi et al., 2009; Lang et al., 1998; Risher et al., 2009). Instead, water shifts are associated with ionic movement including Cl⁻ (Hoffmann et al., 2009; Jourdain et al., 2011; Lambert et al., 2008; Rungta et al., 2015; Zeuthen, 2010). We hypothesized that a reduction in neuronal volume, due to mannitol perfusion, will lead to a concomitant reduction in neuronal [Cl⁻]_i. This reduction in [Cl⁻]_i should enhance inhibitory actions of GABA and decrease epileptiform activity if present.

First, we determined if the addition of 20 mM mannitol to the aCSF leads to a decrease of epileptiform activity in neocortical brain slices during early brain development (P7-12) when GABA has excitatory actions (Glykys et al., 2009; Rheims et al., 2008). This mannitol dose increases the aCSF osmolarity by 20 mOsm, which corresponds to the change obtained by a clinically relevant dose of mannitol (1 gr/Kg) commonly used in pediatric and adult populations for the treatment of increased intracranial pressure (Cloyd et al., 1985; Marshall et al., 1978; Polderman et al., 2003).

In vitro epileptiform activity in neocortical brain slices (Layer IV/V) of P7-12 WT mice was induced by Low-Mg²⁺ (Mody et al., 1987) and measured by field electrodes. Mannitol (20 mM) was perfused after a baseline period (Figure 1A), and epileptiform activity was measured by FFT power for equal time periods of baseline and drug application (Figure 1B). Mannitol perfusion led to a 14% decrease (SD=13) in epileptiform activity from 29.3 μV*Hz (interquartile range, IQR=22.1-52.7) to 25.6 μV*Hz (IQR= 20.2-36.5; n=12, p=0.001, WSRT; Figure 1C). The inter-event interval increased in the presence of mannitol from 5.4 secs (IQR=2.1-9.2) to 8.2 secs (IQR=5.1-16.5; n=11, p=0.01, WSRT; Figure 1D). To test reproducibility, we repeated this experiment using a different batch of mannitol and obtained the same decrease in epileptiform activity (29.3 μV*Hz [IQR=23.4-30.8] to 20.8 μV*Hz IQR 20.3-26.9; n=7, p=0.016 WSRT; 15% decrease [SD=12]).

Figure 1. A clinical dose of mannitol decreases neocortical epileptiform activity in acute neocortical slices during early brain development.

A) Top, in vitro epileptiform activity induced by Low-Mg²⁺ in the neocortex (P8, acute brain slice, Layer IV/V) recorded with extracellular electrodes. Mannitol (20 mM) perfusion shown by line. Middle, higher magnification of boxed segments from baseline (1) and during mannitol perfusion (2). Below, further magnification of box segments. B) FFT area from trace in A calculated every 30 seconds. C) Mannitol anti convulsive effect, P7-12 (open circles: individual recordings; line: median; n=12, Wilcoxon Signed Rank Test; Log scale). D) Interevent interval change during mannitol application (line: median; n=11, Wilcoxon Signed Rank Test).

We further verified the effect of mannitol on neocortical epileptiform activity in a different in vitro model, the neocortical organotypic slice culture. The organotypic preparation develops spontaneous epileptiform activity in regular aCSF (Glykys and Staley, 2015) with the advantage of not requiring the addition of pro-convulsive chemicals or altering its ionic composition. We recorded and analyzed epileptiform activity using the same methods as in acute brain slices. Like the effect observed in the Low-Mg²⁺ model, 20 mM mannitol perfused in regular aCSF (1.3 mM MgCl₂) produced a significant reduction in epileptiform activity at days in vitro 7-8 (DIV) from 28.9 μV*Hz (IQR=23.7-37.1) to 24.5 μV*Hz (IQR=21-30.2; n=9, p=0.004, WSRT; Figure 2A-C) corresponding to a 16% decline (SD=7). Mannitol produced a similar decrease in epileptiform activity in a subgroup of experiments where a washout was performed (Baseline: 26.1 ± 4.64 μV*Hz; Mann: 22.3 ± 3.50 μV*Hz; Washout: 26 ± 5.27 μV*Hz; n=6, F(2,10)=10.05, p=0.004, One-way repeated ANOVA; Holm-Sidak post-hoc test significant between Baseline vs. Mann and between Mann vs. Wash, p=0.003). There was no difference in the mannitol effect on epileptiform activity between acute and organotypic slices (p=0.769, unpaired t-test; Figure 2D). The epileptiform activity amplitude decrease during mannitol perfusion was not due to a reduction in extracellular space resistance (field effect; Jefferys, 1995) as mannitol did not lead to a fall in multi-unit activity amplitude in the absence of seizures (see below). Therefore, the change in epileptiform amplitude represents a decrease in neuronal recruitment.

Figure 2. A clinical dose of mannitol decreases neocortical epileptiform activity in neocortical organotypic slice cultures.

A) Top, epileptiform activity recorded in a neocortical organotypic slice culture (DIV8, Layer IV/V) in the presence of regular aCSF and mannitol. Below, higher magnification of baseline (1), during mannitol perfusion (2), and washout (3). B) FFT area from trace in A calculated every 60 seconds. C) Mannitol anti convulsive effect, pooled data (open circles: individual recordings; line: median; n=9, Wilcoxon Signed Rank Test). D) Effect of mannitol on epileptiform activity between acute and organotypic slices (mean ± SD; p=0,769, unpaired t-test).

These results indicate that an increase in the extracellular osmolarity with a clinically relevant dose of mannitol leads to a decrease in neocortical epileptiform activity during early brain development in both an induced and a spontaneous seizure model.

Mannitol reduces neuronal volume and [Cl⁻]_i in neurons with high initial [Cl⁻]_i.

Neuronal [Cl⁻]_i increases during neonatal seizures (Glykys et al., 2009), it gradually decreases after seizure termination (Dzhala et al., 2010; Glykys et al., 2009; Nardou et al., 2011b), and the rise in [Cl⁻]_i is associated with an increase in neuronal volume (Glykys et al., 2014a). Therefore, we evaluated whether the decrease in epileptiform activity mediated by mannitol correlates with a reduction in neuronal volume and [Cl⁻]_i. We quantified neuronal [Cl⁻]_i using Clomeleon mice which express a genetically encoded ratiometric Cl⁻ sensitive fluorophore (Kuner and Augustine, 2000). It has the advantage of assessing the neuronal [Cl⁻]_i directly and simultaneously from many neurons. Neuronal volume (using the maximal neuronal area [MaxArea], see methods) and [Cl⁻]_i were measured by multiphoton imaging in the neocortical layer IV/V of acute brain slices (P11-12) exposed to Low-Mg²+ aCSF followed by a 20 mM mannitol perfusion (Figure 3A). Slices exposed to Low-Mg²+ aCSF resulted in high [Cl⁻]_i (Figure 3A,B; Supp. Figure 1) as previously observed (Dzhala et al., 2012, 2010, Glykys et al., 2014a, 2009).

Figure 3. Mannitol reduces neuronal volume and [Cl⁻]_i in neurons with high initial [Cl⁻]_i.

A) Top, Two-photon stack image (CFP) of a P10 neocortical slice (Layer IV/V) expressing Clomeleon during Low-Mg²⁺ (left) and after 20 minutes of mannitol perfusion (right). Contrast, brightness, gamma identical between both images. Below, ROI of neurons that completed the treatment. [Cl⁻]_i is pseudo-colored. B) The neuronal [Cl⁻]_i (left) and maximal area (MaxArea, right) decreased during mannitol perfusion (n=107 paired neurons, 5 slices, p<0.001 Friedman Repeated ANOVA on Ranks, * is p<0.05 by Tukey post-hoc test). C) Correlation between baseline [Cl⁻]_i and its change after mannitol perfusion. Line stands for linear regressions (R²=0.085 at 5’, p=0.002; and R²=0.395 at 20’, p<0.001). D) No correlation between baseline neuronal MaxArea and its change after mannitol perfusion. E) Correlation between baseline [Cl⁻]_i and the change in MaxArea (R²=0.085 at 5’, p=0.002 and R²=0.104 at 20’, p<0.001). F) Significant correlation between the change in neuronal MaxArea and [Cl⁻]_i after 20 minutes of mannitol (R²=0.088, p=0.002).

Neuronal [Cl⁻]_i decreased in neocortical layer IV/V neurons from 28.4 mM (IQR: 19-50.4) to 24.2 mM (IQR: 14.6-43.6) after 5 minutes of mannitol perfusion and to 19.0 mM after 20 minutes of mannitol (IQR: 10.5-30.8) (p<0.001, Friedman Repeated measures ANOVA on Ranks; significance between the 3 conditions, p<0.05 post-hoc Tukey test; Fig 3B). This change in [Cl⁻]_i corresponds to a change in E_Cl from −39 to −49 mV (Nernst equation calculation). MaxArea also decreased after 5 minutes of mannitol perfusion from 254 μm² (IQR: 179-312) to 245 μm² (IQR: 183-293) and to 226 μm² after 20 minutes (IQR: 176-297; n=107 paired neurons, 5 slices, p<0.001 Friedman Repeated ANOVA on Ranks; significance between initial condition and both mannitol perfusion at 5’ and 20’, p<0.05 post-hoc Tukey test, Figure 3B). Notably, neurons with the highest initial [Cl⁻]_i had the greatest decrease under mannitol (R²=0.085 at 5’, p=0.002; and R²=0.395 at 20’, p<0.001; Figure 3C). This is in contrast with the decrease of MaxArea by mannitol which is independent of the initial neuronal area (p>0.05 at 5’ and 20’, Figure 3D) due to the large variance of neuronal sizes. Notably, neurons that had a high initial [Cl⁻]_i had a larger decrease in MaxArea (R²=0.085 at 5’, p=0.002 and R²=0.104 at 20’, p<0.001; Figure 3E). A correlation was observed between the change in [Cl⁻]_i and neuronal MaxArea at 20 minutes (R²=0.088, p=0.002, Figure 3F) but no significant difference was observed at 5 minutes (p=0.198). Neurons exposed to only Low-Mg²⁺ aCSF and no mannitol maintained a stable [Cl⁻]_i and MaxArea for up to 3 minutes of sequential measurement (Supp. Figure 1).

Next, we addressed whether CCCs are involved in the co-transport of water and Cl⁻ during epileptiform activity induced by Low-Mg²⁺. CCCs were blocked with furosemide 300 μM (Pond et al., 2006) as higher doses of this drug can lead to a significant block of NMDA receptors (Lerma and Martin del, 1992). Perfusion of mannitol in the presence of furosemide resulted in a small but significant increase in [Cl⁻]_i (Figure 4, p<0.001; Friedman Repeated ANOVA on Ranks, p<0.05 by Tukey post-hoc test between furosemide and furosemide + mannitol). Mannitol was still able to decrease MaxArea in the presence of furosemide (Figure 4, p=0.003; Friedman Repeated ANOVA on Ranks, p<0.05 by Tukey post-hoc test between furosemide and furosemide + mannitol).

Figure 4. Blocking CCCs prevent the decrease of [Cl⁻]_i and volume by mannitol.

A) Neuronal [Cl⁻]_i (left) and MaxArea (right) measured with 2-photon imaging of Clomeleon expressing Layer IV/V neocortical neurons (P9-12) under Low-Mg²⁺ after sequential perfusion of furosemide (Fur), furosemide + mannitol (Fur+Man) and washout (Wash). Six slices, n=60 paired neurons, Friedman Repeated ANOVA on Ranks, p<0.001 for [Cl⁻]_i and p=0.003 for MaxArea. * isp<0.05 by Tukey post-hoc test. ([Cl⁻]_i in mM: Control 18.1 [IQR: 13.2-26], Fur: 18.8 [11-25.4], Fur+Man 22.6 [15.1-30.2], Wash 23.8 [19.4-31.3], MaxArea in μm: Control 277 [237-331], Fur 281 [233-328], Fur+ Man 277 [227-326], Wash 303 [239-352]). B) Absolute change in [Cl⁻]_i (left) and MaxArea (right) in the presence of drugs in A. C) Combined change in [Cl⁻]_i and MaxArea in the presence of drugs in A.

These observations indicate that when CCCs are blocked, co-transport of water and Cl⁻ become unlinked so that an increase in extracellular osmolarity induced by mannitol results in a net loss of free water and leads to an increase in intracellular Cl⁻. The decrease in neuronal MaxArea when CCCs are blocked suggests that water can escape through other non-CCC pathways (MacAulay and Zeuthen, 2010). However, when CCCs are operative, the water and Cl⁻ responses are determined by CCC activity. Together, these results indicate that the action of mannitol on epileptiform activity correlates with a concomitant decrease in neuronal MaxArea and [Cl⁻]_i mediated primarily by CCCs, especially in neurons with high initial [Cl⁻]_i.

Mannitol reduces spontaneous multiunit activity frequency but not amplitude

Next, we verified the functional changes in [Cl⁻]_i using a Cl⁻ imaging independent technique. We measured MUA in the neocortex (layer IV/V) of acute brain slices (P7-8) in the presence of 50 μM 4-aminopyridine (to increase the frequency of action potentials) plus 2-3 mM kynurenic acid (glutamatergic excitatory transmission blocker) to determine if a decrease in [Cl⁻]_i by mannitol causes a reduction in action potential firing as a result of an increase in GABAergic inhibition. These experiments also address if the decrease in epileptiform activity amplitude, observed in the presence of mannitol (Figs. 1 and 2), is due to a decrease in extracellular conductivity (Jefferys, 1995). MUA, as an assay of extracellular conductivity, is independent of the summation of ligand- and voltage-gated membrane currents. Mannitol perfusion (20 mM) produced a significant decrease in spontaneous MUA frequency from 2.08 Hz (IQR 0.42-2.66) to 1.10 Hz (IQR: 0.24-1.76) between 5-10 min and to 0.49 Hz (IQR: 0.08-1.39) between 10-15 minutes (n=12, p<0.001, Friedman Repeated ANOVA on Ranks, difference between baseline and both conditions p<0.05, Tukey post-hoc test Figure 5B, F). These results indicate that mannitol enhanced the inhibitory effect of spontaneously released GABA in brain slices as excitatory transmission is blocked by kynurenic acid. It is also an independent confirmation of the [Cl⁻]_i decrease observed with Clomeleon imaging. We did not find evidence that this concentration of mannitol in the developing neocortex decreased the extracellular conductivity (field effect) because there was no change in MUA peak amplitudes (Baseline: 21.8 μV [IQR: 18.8-26.3], Mann 5-10 min 20.4 μV [18.5-26-6], Mann 10-15 min 21.3 μV [17.6-25.2], p=0.472 Friedman Repeated ANOVA on Ranks; Figure 5C-E) nor a change in baseline root mean square noise (Baseline: 3.14 μV [IQR: 3.08-3.25], Mann 5-10 min 3.15 μV [3.06-3.2], Mann 10-15 min 3.12 μV [3.03-3.25], p=0.338 Friedman Repeated ANOVA on Ranks; Figure 5G).

Figure 5. Mannitol reduces spontaneous multiunit activity frequency but not amplitude.

A) Top, Multiunit activity (MUA) recorded in the presence of 2 mM kynurenic acid and 50 μM 4-Aminopyridine (P8) and in the presence of mannitol (bottom line) in acute neocortical brain slices (Layer IV/V). Below, higher magnification of dashed boxes from baseline (1), under mannitol (2), and washout (3). B) Top, MUA frequency over time. Below, MUA frequency counts in the regions represented by the dashed boxes: baseline (1), mannitol (2), and washout (3). C) Peak amplitude probability density of detected events from the same segments depicted in B (Baseline: 27.9; Man: 29.6; Wash: 23.83 μV; median values). D) Cumulative distribution of peak amplitudes from segments in C. Note no change between control and mannitol perfusion. K-S test between all groups p>0.05. E) MUA peak amplitudes are not different (P7-8) (open triangles: individual recordings; line: median; n=12; p=0.472, Friedman Repeated ANOVA on Ranks). F) MUA frequency decreases in the presence of mannitol (open circles: individual recordings; line: median; n=12; p<0.001; Friedman Repeated ANOVA on Ranks; *, Tukey post-hoc test p<0.05). G) No change in baseline root mean square with mannitol (p=0.338; Friedman Repeated ANOVA on ranks).

While it is unlikely that 20 mM mannitol can induce a change in the GABA release probability as one order of magnitude higher concentrations of sugars are needed (500 mM vs. 20 mM; Bekkers and Stevens, 1996; Staley et al., 1998; Stevens and Tsujimoto, 1995) we decided to test this possibility. Indeed, we observed no change in the frequency nor amplitude of sIPSCs recorded in layer IV/V neocortical neurons during whole cell voltage clamp experiments where the neuronal [Cl⁻]_i closely matches the pipette’s [Cl⁻]_i (Sup. Figure 2).

This data, combined with our Cl⁻ imaging, strongly indicates that a clinical dose of mannitol decreases neuronal [Cl⁻]_i leading to a more efficacious inhibitory effect of endogenous GABA.

A decrease in neuronal [Cl⁻]_i by mannitol potentiates the effect of benzodiazepines during early neocortical development but not at older ages

A clinical dose of mannitol decreased neuronal [Cl⁻]_i and produced a significant reduction in epileptiform activity (Figs. 1–3). Next, we explored whether this dose of mannitol enhances the anti convulsive action of a clinically relevant dose of diazepam (GABA_AR positive allosteric modulator), that by itself is not effective during early neocortical development (Connell et al., 1989; Glykys and Staley, 2015; Nardou et al., 2011a). We tested 0.5 μM of diazepam which corresponds to a concentration of 0.14 mg/L. A drug plasma concentration of 0.15 mg/L (range: 0.07–0.23 mg/L) is found in human neonates after receiving diazepam at an anticonvulsant dose of 0.1–0.3 mg/kg, using a volume of distribution of 1.3 L/kg (Anderson and Miller, 2002; Greenblatt and Sethy, 1990).

We recorded in vitro epileptiform activity induced by Low-Mg²⁺using field electrodes in the neocortex layer IV/V of P10-11 mice. We assessed the effect of diazepam (0.5 μM; DZP) followed by the addition of 20 mM mannitol (Mann+Dzp). Mann+Dzp significantly decreased epileptiform activity but diazepam alone did not (Baseline: 37 ± 6.45 μV*Hz, Dzp 32.2 ± 9.21 μV*Hz, Mann+Dzp 24.6 ± 5.15 μV*Hz; n=8, F(2,14)=9.409, p=0.003, One-way repeated ANOVA, significance between Mann+Dzp and Baseline and between Mann+Dzp and Dzp, p<0.001 and p=0.019 respectively, Holm-Sidak post-hoc test; Figure 6A-C). Also, the ratio effect of Mann+Dzp is significantly greater than mannitol alone (Mann: 0.89 IQR: 0.72-0.97, n=12; Mann+Dzp: 0.69 IQR: 0.5-0.84, n=8; p=0.041, Mann-Whitney Rank Sum Test; Figure 6D). The ineffectiveness of diazepam at this age is consistent with prior publications in the organotypic neocortical slice cultures (Glykys and Staley, 2015), in cultured hippocampal neurons and in the intact hippocampus (Deeb et al., 2013; Nardou et al., 2011a). In contrast, the same dose of diazepam (0.5 μM) decreased epileptiform activity in adult (P31-63) neocortical brain slices (9/9 slices), but it was not enhanced by mannitol (Baseline: 27.7 ± 0.88 μV*Hz, Dzp: 24.5 ± 6.73 μV*Hz, Man+Dzp: 23.3 ± 5.96 μV*Hz, Washout: 25.7 ± 8.15 μV*Hz; n=9, F(3,24)=7.07, p=0.001, one-way repeated ANOVA; significant between Baseline and both Dzp and Man+Dzp p<0.001 and p=0.004 respectively, Holm-Sidak post-hoc test; Figure 6F, G).

Figure 6. A decrease in neuronal [Cl⁻]_i by mannitol potentiates the effect of benzodiazepines during early neocortical development.

A) Top, in vitro epileptiform activity induced by Low-Mg²⁺ in the presence of diazepam (Dzp) and mannitol plus diazepam (Mann+Dzp) at P10 (acute brain slice, Layer IV/V). Below, higher magnification of areas: baseline (1), diazepam (2), and Mann+Dzp (3). B) FFT area from trace in A calculated every 30 seconds. C) Change in FFT area by the different drugs (line: mean; P10-11; n=8; p=0.03, ANOVA; Holm-Sidak post-hoc test). D) Ratio change between mannitol alone and Mann+Dzp (open circles: individual recordings; line: median). E) Same as A but recorded in an adult neocortical brain slice (P63). F) FFT area from trace in E calculated every 30 seconds. G) Change in FFT area by the different drugs (line: mean, P31-63; n=9; p=0.001, ANOVA; Holm-Sidak post-hoc test; percentage decrease between 2-25% for Dzp and 5-33% decrease for Dzp+Man).

These results indicate that an ineffective dose of diazepam can become effective when [Cl⁻]_i is lowered by mannitol during early brain development. In contrast, as adult neuronal [Cl⁻]_i is already lower than during early brain development (Glykys et al., 2009; Glykys and Staley, 2016; Owens et al., 1996; Rheims et al., 2008; Sulis Sato et al., 2017), a clinical dose of diazepam is already effective and mannitol cannot further enhance it. These results also agree with our prior data in which a clinical dose of diazepam is ineffective when the baseline neuronal [Cl⁻]_i is high but is effective when [Cl⁻]_i is low (Glykys and Staley, 2015). The ineffectiveness of mannitol in the adult neocortex could be due to a lower number of neurons with high [Cl⁻]_i compared to the neonatal brain (Glykys et al., 2009) or by the re-establishment of low [Cl⁻]_i after seizures by CCCs, mainly K⁺-Cl⁻ cotransporters (Moore et al., 2017).

Finally, hypertonic saline is another clinical therapy often used to treat increased intracranial pressure. We studied if adding an additional 10 mM of NaCl to the aCSF (a similar 20 mOsm increase as produced with 20 mM mannitol) would also result in a decrease in neocortical epileptiform activity. Hypertonic saline did not decrease epileptiform activity in acute neocortical brain slices (P9-13) exposed to Low-Mg²⁺. The same recording and analysis methods were used as before (Baseline: 60.9 ± 2.05 μV*Hz, +NaCl: 63.1 ± 2.45 μV*Hz, n=12, p=0.814, paired t-test). The ineffectiveness of increasing the extracellular osmolarity by a higher concentration of extracellular NaCl may be due to an enhanced neuronal import of Cl⁻ by the Na⁺-K⁺-Cl⁻ co-transporter (NKCC1); however, this was outside the scope of our present study.

Discussion

Summary

Our results performed in the neocortex show: 1) A clinical concentration of mannitol decreases neocortical epileptiform activity during early brain development in two different seizure models. 2) The anticonvulsant effect of mannitol is associated with a decrease in neuronal volume and [Cl⁻]_i in neurons with high initial [Cl⁻]_i. 3) The linked movement of water and Cl⁻ out of neurons is mediated primarily by CCCs. 4) Mannitol leads to a decrease in MUA frequency but not amplitude. 4) Mannitol potentiates the effect of a clinical dose of diazepam on neocortical epileptiform activity during early brain development, and 5) a similar increase in aCSF osmolarity by NaCl did not reduce epileptiform activity. We conclude that enhancing the outflux of water in neurons with pathologically elevated [Cl⁻]_i results in reduced neuronal volume and [Cl⁻]_i, decreased epileptiform activity, and enhanced actions of positive allosteric GABA_ARs modulators (Figure 7). These results are not consistent with the independent movement of water and Cl⁻ across neuronal membranes because: 1) neurons have a low water membrane permeability (Aitken et al., 1998 Andrew et al., 2007; Caspi et al., 2009; Lang et al., 1998); and 2) increased extracellular osmolarity would have drawn pure water out of the neuronal cytoplasm, raising the neuronal [Cl⁻]_i (with no need for dissipation of the increased [Cl⁻]_i because neurons can sustain elevated levels in physiological and pathological conditions, Ben-Ari, 2002; Huberfeld et al., 2007; Pond et al., 2006) and further compromising GABAergic inhibition.

Figure 7. Mannitol decreases neuronal [Cl⁻]_i and volume leading to improved inhibitory actions of GABA.

An increase in extracellular osmolarity by mannitol leads to a decrease in neuronal [Cl⁻]_i and volume in neurons experiencing injury or seizures through CCCs (Figure 3, 4). The decrease in [Cl⁻]_i leads to an increase GABAergic driving force. Two scenarios are illustrated: a depolarized membrane potential (MP) during seizures and a resting membrane potential (RMP) when neurons are not depolarized, with neurons having a minimal leak conductance for Cl⁻ (Chamberlin and Dingledine 1988; Yelhekar et al., 2017). The increase in GABAergic driving force results in decreased action potential firing (Figure 5) and seizure activity (Figures 1, 2, 6). The darker background represents an increase in extracellular osmolarity mediated by mannitol.

Neuronal Cl⁻ and volume

There is a positive correlation between [Cl⁻]_i and neuronal swelling. Neuronal swelling (Caspi et al., 2009; Glykys et al., 2014a, 2009; Steffensen et al., 2015) can arise by changes in the ionic concentration (especially for K⁺ and Cl⁻) and changes in the extracellular space that happen during prolonged neuronal activity and seizures (Dietzel, I. Heinemann, U., Hofmeier, G., and Lux, 1980; Dietzel et al., 1989, 1982; Dzhala et al., 2010; Hochman, 2012; Raimondo et al., 2015). During neonatal seizures, there is an increase in neuronal volume and [Cl⁻]_i (Glykys et al., 2014a, 2009). A rise in [Cl⁻]_i is also observed in cytotoxic edema which can be prevented by decreasing the extracellular Cl⁻ concentration (Dijkstra et al., 2016; Rungta et al., 2015; Steffensen et al., 2015). Clinically, cytotoxic edema is observed in neonatal hypoxic-ischemic encephalopathy (HIE) and severe traumatic brain injury in humans, both of which are associated with seizures (Gutierrez et al., 2010; Nedelcu et al., 1999; Olsson et al., 2006; Unterberg et al., 2004; Wei et al., 2012). Finally, neurons from patients with mesial temporal epilepsy also demonstrate cytotoxic edema and increased [Cl⁻]_i (Huberfeld et al., 2007; Hufnagel et al., 2003; Szabo et al., 2005).

Modifying the extracellular osmolarity alters seizure activity (Andrew, 1991) which points to its important role in this pathological condition. Hyperosmotic therapy with mannitol is one of the main treatments for elevated intracranial pressure which can be caused by severe cytotoxic and vascular edema (Carney et al., 2016). Mannitol creates a concentration gradient between the intravascular, interstitial, and intercellular compartments. It also improves cerebral perfusion pressure in hypoperfused brain regions, and alters blood viscosity (Muizelaar et al., 1983; Scalfani et al., 2012). Several mechanisms have been postulated to explain the actions of mannitol and other hypertonic solutions on epileptiform activity including changes in field effects (by increasing the extracellular space) and reducing excitatory post-synaptic potentials (Andrew et al., 1989; Andrew and MacVicar, 1994; Dudek et al., 1990; Hochman, 2012; Jefferys, 1995; Reed and Woodbury, 1964; Roper et al., 1992; Saly and Andrew, 1993; Traynelis and Dingledine, 1989).

Our results support a novel mechanism for mannitol, as prior studies did not address neuronal [Cl⁻]_i. An increase in osmolarity by 20 mOsm induced by mannitol (similar to what is obtained with a clinical dose of 1 gr/Kg, Cloyd et al., 1985; Marshall et al., 1978; Polderman et al., 2003) decreases neuronal volume and [Cl⁻]_i (Figure 3) which results in an enhancement of the inhibitory actions of GABA and of anticonvulsants that prolong the open time of GABA_ARs such as benzodiazepines (Glykys et al. 2017). Therefore, our data argues against a pure field effect of mannitol in the neocortex during the neonatal period to explain its anticonvulsant effect. However, we do not negate that in areas with a higher neuronal density (e.g. hippocampus) or with significantly higher doses of mannitol than used in clinic, a change in field effect might become an additional explanation of the actions of mannitol on neuronal excitability. Our in vitro results support prior studies which demonstrated that intravascular mannitol decreases seizure-like activity in murine species as well as humans (Baran et al., 1987; Haglund and Hochman, 2005).

Neuronal Cl⁻ homeostasis and seizure control

Low [Cl⁻]_i is paramount for the inhibitory actions of GABA. A small increase in [Cl⁻]_i can alter GABA actions from hyperpolarization to depolarization (Doyon et al., 2016; Glykys et al., 2017). Therefore, neuronal [Cl⁻]_i needs to be closely regulated. Our data supports the hypothesis that CCCs are involved in the movement of Cl⁻, cations, and water to restore the [Cl⁻]_i set point as well as neuronal volume (Delpire and Staley, 2014; Doyon et al., 2011; Glykys et al., 2014a; Hoffmann et al., 2009; Kahle et al., 2015). During seizure activity, the role of CCCs in neuronal Cl⁻ homeostasis is particularly important due to the elevated levels of synaptic activity that can lead to an excess entry of Cl⁻, cations and water (Glykys et al., 2014a, 2009; Raimondo et al., 2015).

Modulating neuronal [Cl⁻]_i is an important therapeutic strategy for the treatment of seizures. Blocking NKCC1, a conduit for Cl⁻ accumulation during seizures (Dzhala et al., 2010), increases the efficacy of GABA_ARs positive allosteric modulators in vitro and in vivo (Dzhala et al., 2005; Erker et al., 2016; Glykys et al., 2009; Mazarati et al., 2009; Sivakumaran and Maguire, 2016). A small, uncontrolled human trial did not support this concept. However, this trial was not designed to test efficacy and the results are controversial (Pressler et al., 2015; Thoresen and Sabir, 2015). A controlled clinical trial has just completed their recruitment (NCT00830531, results pending), and the development of new NKCC1 blockers with better blood-brain permeability is underway (Töllner et al., 2014). Enhancing the activity of K⁺-Cl⁻ co-transporter KCC2 (canonical exporter of Cl⁻) is an additional proposed treatment for seizures by decreasing pathologically elevated [Cl⁻]_i (Friedel et al., 2015; Moore et al., 2017; Silayeva et al., 2015).

Here we provide an alternate approach: altering the extracellular osmolarity to simultaneously decrease neuronal [Cl⁻]_i and volume (Glykys et al., 2017). Neurons are unique in that volume and [Cl⁻]_i regulation are directly tied to the polarity of GABA actions, and there is a correlation between cytotoxic edema, Cl⁻ and seizures (Glykys et al., 2017). There is strong evidence of Cl⁻, cations, and water being co-transported through CCCs (Gauvain et al., 2011; Hamann et al., 2010; Jourdain et al., 2011; MacAulay and Zeuthen, 2010; Zeuthen, 1994a; Zeuthen and Macaulay, 2012) since neurons do not have aquaporins and hence, a poor membrane water permeability (Aitken et al., 1998; Andrew et al., 2007; Lang et al., 1998) compared to epithelial cells, artificial membranes and astrocytes (Fettiplace and Haydon, 1980; Risher et al., 2009). The transport of Cl⁻ by NKCC1 and KCCs is coupled with a significant movement of water (MacAulay et al., 2004). KCCs and NKCC1 roughly transport 500 and 600 water molecules per cycle of ion movement respectively (Zeuthen, 1994b, 1994a; Zeuthen and Macaulay, 2012). Our results support the notion of the linked movement of water and Cl⁻ out of neurons, as when CCCs were blocked with furosemide, mannitol produced an increase in [Cl⁻]_i while continuing to decrease neuronal volume (Fig 4) similarly to what we observed in organotypic hippocampal slices in the past (Glykys et al., 2014a). Whether the movement of Cl⁻ and water is through KCCs and/or the reversal of NKCC1 pump direction is beyond this study and will require further research.

Strengths and limitations

Strengths of our study include the use of a clinically relevant dose of mannitol and diazepam, as well as the use of two distinct models of epileptiform activity: the Low-Mg²⁺ model in acute brain slices and the neocortical organotypic slice culture model which has spontaneous epileptiform activity. An additional strength includes the use of electrophysiology as well as 2-photon fluorescence microscopy to obtain complimentary data regarding the neuronal effects of osmotic manipulations. This dual approach robustly compensates for the limitations of either technique; for example, the use of a slightly pH-sensitive fluorophore, Clomeleon, to study [Cl⁻]_i (Kuner and Augustine, 2000).

A limitation of this study is the use of in vitro models which permit the application of osmotic agents directly into the intercellular space rather than the intravascular space. Currently, it is unclear how an increase in intravascular space osmolarity alters intercellular space osmolarity. An intact blood brain barrier, as seen in the early post-injury phase when cytotoxic edema is predominant (Liang et al., 2007), would limit the flux of therapeutic osmolytes (either mannitol or NaCl) from the vascular to the intercellular space. In this circumstance, the effect of hyperosmolar therapy in the extracellular space would be the loss of free water and a concomitant increase in the intercellular NaCl concentration, more closely resembling our experiments with the perfusion of NaCl rather than the administration of mannitol. However, HIE status epilepticus and seizures are known to disrupt the blood-brain-barrier leading to vasogenic edema (Gutierrez et al., 2010; Kim et al., 2001; Rumpel et al., 1995). After the blood-brain barrier breaks down, mannitol and NaCl administered in the intravascular space may become distributed throughout the intercellular space. Under these conditions, a differential effect of mannitol and NaCl on epileptiform activity could be observed, as in our experiments. Interestingly, a dose of mannitol resulted in a decrease in epileptiform activity in human patients (Haglund and Hochman, 2005). The effects of mannitol observed in our experiments were modest but significant. Whether a clinically meaningful anticonvulsant effect of hyperosmolar therapy occurs will require a clinical correlation of electrographic seizure activity after hypertonic therapy at sequential times post-injury. Also, we concentrated our studies on the effect of mannitol during epileptiform activity. How healthy neurons activate regulatory volume increase (RVI) and decrease (RVD) mechanisms (Hoffmann et al., 2009) in response to altered extracellular osmolarity is of great interest, but beyond the scope of our current study. Lastly, we concentrated our studies in neurons, yet all central nervous cell types suffer cytotoxic edema. Astrocytes, in particular, swell during brain injury (Liang et al., 2007). Further studies are needed to determine how astrocytic Cl⁻ and its cell volume change in the presence of mannitol.

In conclusion, our results strongly support the hypothesis that an increase in extracellular osmolarity leads to the linked movement of Cl⁻ and water out of neurons. While we have focused on CCCs, there are alternate and redundant pathways that mediate the outward movement of water and Cl⁻ including the volume-regulated anion channel (VRAC) and other Cl⁻ channels (Hoffmann et al., 2009; Lambert et al., 2008; Pedersen et al., 2016; Qiu et al., 2014; Voss et al., 2014; Zeuthen, 2010) as it is suggested by the decrease in neuronal volume in the presence of mannitol when CCCs were blocked (Figure 4) as well as our prior results (Glykys et al., 2014a).

A better understanding of volume and salt regulation, specifically in neurons, will lead to the development of new medical compounds that will not only improve neuronal cytotoxic edema but may lead to improved anti convulsive treatments.

Supplementary Material

1

Supplemental Figure 1. Neuronal [Cl⁻]_i and volume are stable in Low-Mg²⁺

A) Neuronal [Cl⁻]_i (left) and MaxArea (right) measured with 2-photon imaging of Clomeleon expressing Layer IV/V neocortical neurons (P14) at different time points up to 30 minutes in the presence of Low-Mg²⁺. Slices incubated in Low-Mg²⁺ for at least 1 hour. ◯, individual neurons. ●; median value. No significant change between time points ([Cl⁻]_i in mM: 0’ = 51.9 [IQR: 34.3-79.2]; 5’ = 43 [29.7-72.6]; 10’ = 47.8 [29.9-74.2]; 20’ = 45.6 [33.3-73.3]; 30’= 42 [30.1-71.3]; MaxArea in μm²: 0’ = 262 [196-316]; 5’ = 258 [198-297]; 10’ = 254 [201-290]; 20’ = 252 [ 210-284]; 30’ = 250 [ 208-289]. n=71 cells, 3 slices; p=0.576 for [Cl⁻]_i and p=0.606 for MaxArea, Kruskal-Wallis One Way ANOVA on Ranks). B) Histogram distribution of changes in [Cl⁻]_i (left) and MaxArea (right) compared to time point 0. Thin lines with symbols represent distribution, thick lines represent Gaussian fit. Black color represents difference between 30’ and 0’.

2

Supplemental Figure 2. Mannitol does not increase frequency of sIPSCs. A) Representative traces of sIPSCs recorded during whole cell voltage clamp experiments of a pyramidal cell at baseline (top panel) and after application of 20 mM mannitol (bottom panel), same neuron. B, C. No difference in frequency (n=7, p=0.10, paired t-test) nor amplitude of sIPSCs (n=7, p=0.06, paired t-test).

Highlights.

• Mannitol reduces neuronal volume and [Cl⁻]_i in neurons with high initial [Cl⁻]_i.

• Chloride co-transporters primarily mediate the linked movement of water and Cl⁻ out of neurons.

• Mannitol decreases neocortical seizure activity during early brain development.

• Mannitol potentiates the anticonvulsive effect of diazepam in the young neocortex.