Cortical spreading depolarization is a potential target for rat brain excitability modulation by Galanin

The inhibitory neuropeptide Galanin (Gal) has been shown to mediate anticonvulsion and neuroprotection. Here we investigated whether Gal affects cortical spreading depolarization (CSD). CSD is considered the pathophysiological neuronal mechanism of migraine aura, and a neuronal mechanism aggravating brain damage upon afflictions of the brain. Immunohistochemistry localized Gal and the Gal receptors 1 – 3 (GalR1 – 3) in native rat cortex and evaluated microglial morphology after exposure to Gal. In anesthetized rats, Gal was applied alone and together with the GalR antagonists M40, M871, or SNAP 37889 locally to the exposed cortex. The spon-taneous electrocorticogram and CSDs evoked by remote KCl pressure microinjection were measured. In rat cortex, Gal was present in all neurons of all cortical layers, but not in astrocytes, microglia and vessels. GalR2 and GalR3 were expressed throughout all neurons, whereas GalR1 was preponderantly located at neurons in layers IV and V, but only in about half of the neurons. In susceptible rats, topical application of Gal on cortex decreased CSD amplitude, slowed CSD propagation velocity, and increased the threshold for KCl to ignite CSD. In some rats, washout of previously applied Gal induced periods of epileptiform patterns in the electrocorticogram. Blockade of GalR2 by M871 robustly prevented all Gal effects on CSD, whereas blockade of GalR1 or GalR3 was less effective. Although microglia did not express GalRs, topical application of Gal changed microglial morphology indicating microglial activation. This effect of Gal on microglia was prevented by blocking neuronal GalR2. In conclusion, Gal has the potential to ameliorate CSD thus reducing pathophysiological neuronal events caused by or associated with CSD.


Introduction
Pathological increases in brain cortical excitability play an important role in human central nervous diseases ranging from epilepsy to migrainic headache.Excitability of nerve cells is controlled by excitatory neurotransmission and by inhibitory interneurons.To maintain functional homeostasis in brain, excitatory and inhibitory processes are balanced by neurotransmitters/neuromodulators.Among them is the neuropeptide Galanin (Gal), a 29-or 30 amino acid peptide that was first identified by Tatemoto et al. (1983).Gal mediates neuronal effects via three Gal receptors (GalR1-3).GalR1 and GalR3 directly open potassium channels and inhibit the cAMP synthesis (Branchek et al., 2000;Xu et al., 2005).GalR2 activates phospholipase C and mobilizes intracellular calcium ions (Branchek et al., 2000).
Pleiotropic functions have now been assigned to Gal.It has wakepromoting effects (McGinty and Szymusiak, 2001), is obesogenic by controlling the release of hypothalamic hormones (Mills et al., 2021), and reduces the release of endogenous excitatory amino acids (Zini et al., 1993).The expression of GalRs in the heart compartments is regulated by stress ( Šípková et al., 2017).Pieribone et al. (1998)
Linked to pathological hyperexcitation is also the cortical spreading depolarization (CSD), a neuronal and glial mass depolarization, that was first observed as spreading depression of electroencephalic activity (Leao, 1944).During CSD, ionic and water homeostasis is transiently disturbed by a massive influx of sodium ions together with water into neurons and the outflow of potassium ions from neurons.The redistribution of ionic gradients and restitution of water homeostasis is energy demanding (Hartings et al., 2017).CSD is accepted as the electrophysiological correlate of migraine aura (Close et al., 2019).In a healthy brain, a single CSD is well-tolerated (Ayata and Lauritzen, 2015).However, after stroke or traumatic brain injury, repetitive CSDs can progressively compromise and destroy brain tissue and ultimately lead to a terminal depolarization (Andrew et al., 2022).Shen et al. (2003) have shown that CSD induced the expression of Gal mRNA 2 days after the event, and within an even longer time interval (7-28 days) also the expression of GalR1 mRNA, but not of GalR2 mRNA.Interestingly, after CSD, oligodendrocyte progenitor cells proliferated and transiently showed increased Gal expression (Shen et al., 2005;Shen et al., 2003).Apart from these anecdotic observations, no further studies investigated the relationship between Gal and CSD.Taking into account the mostly inhibiting actions of Gal in the brain, we hypothesized that Gal should diminish the brain's capability to initiate and/or to propagate CSD, provided that GalRs are widely expressed in cerebral cortex.
To address this hypothesis, we stained native rat brain slices for localization of Gal and GalR1-3.We investigated whether application of Gal to a treatment area of the exposed rat cortical surface influences the amplitude and propagation of CSDs, and the excitability of neurons by measuring the threshold for evoking CSD by local potassium injection at a remote site.To decipher which GalRs are involved in CSD modulation, GalR 1-3 antagonists (M40, M871, or SNAP 37889) were applied before or together with Gal.We also explored the effect of Gal on the morphology and number of glial cells.

Methods
Experiments were approved by the Thuringian Government (Thüringer Landesamt für Verbraucherschutz).They were performed according to the Protection of Animals Act of the Federal Republic of Germany, in accordance with the declaration of Helsinki and the guiding principles in the care and use of animals.Data sampling, evaluation, and presentation complied with the ARRIVE guidelines.

Glia staining and quantification
To assess the effects of Gal on glial cells, a subgroup of rats was subjected to the same surgical procedure as described in subsection 2.3., but no electrodes were inserted and no CSD was elicited.These animals were in one hemisphere either treated with a topical application of Gal at 10 − 6 mol/L for 3 h (n = 3) or with the GalR2 antagonist M871 at nmol/L for 2 h followed by Gal at 10 − 7 mol/L for another 2 h (n = 3).The other hemisphere remained as control.Perfusion, fixation and slicing of the brains were performed as described above.In 40-μm-thick cryostat slices microglia was stained using goat anti-Iba1 (1:200; GTX89792; GeneTex Inc., Irvine, CA) and astrocytes were stained using rabbit anti-GFAP (1:100, GTX108711; GeneTex) as primary antibodies.Alexa 488 donkey anti-goat (1:200, #A11055, Thermo Fisher Scientific) and Alexa 568 donkey anti-rabbit (1:200, #A10042, Thermo Fisher Scientific) were used as secondary antibodies.Stacking pictures acquired every 1 μm of each section were taken from the confocal laser scanning microscope (TCS SP5, Leica).For quantification, number of microglia cells and microglia morphology concerning endpoints/cell and process length/cell as a measure of microglia complexity was evaluated using the skeletonize plugin of Image J according to the protocol recommended by Young and Morrison (2018).

Surgical preparation of the rats
Adult male Wistar rats (n = 61; 350-450 g, aged 95 ± 4 days, housed in the Animal Facility of University Hospital Jena) were deeply anesthetized using sodium thiopental (Trapanal®; Inresa, Freiburg, F. Gimeno-Ferrer et al.Germany; initially 100-125 mg/kg intraperitoneally [i.p.]).During surgery, depth of anesthesia was constantly checked for the lack of reflexes to noxious squeezing of the interdigital skin and corneal blink.During the experiments, supplemental doses of Trapanal® i.p. (maximally 20 mg/kg) were administered, if necessary, to maintain the depth of anesthesia.The trachea was cannulated to improve oxygenation, the right femoral artery was cannulated to continuously monitor the mean arterial blood pressure, the femoral vein was cannulated for intraveneous injections.The electrocardiogram was continuously monitored.Body temperature was maintained at 37 • C using a feedback-controlled heating system.
Surgery of the skull with trephinations was performed following the protocol previously described (Gimeno-Ferrer et al., 2022;Richter et al., 2017).Using a stereotactic holder, the head was fixed, and two trephinations were made over the left hemisphere of the skull exposing the brain (one circular more caudal, with a diameter of 3-4 mm in front of Lambda, and one rectangular and frontal, spanning from 2 mm in front of Bregma over a length of 5-6 mm, 3-4 mm wide) using a minidrill and cooling with artificial cerebrospinal fluid (ACSF) during the procedure.The composition of ACSF was in millimoles per litre (mmol/L): NaCl 138.4,KCl 3.0, CaCl 2 1.3, MgCl 2 0.5, NaH 2 PO 4 0.5, urea 2.2, and glucose 3.4, warmed to 37 • C and equilibrated with 5% CO 2 in O 2 .Then, dura and arachnoidea were removed at the trephination sites to expose the cortex, which was kept moist with ACSF.A wall was built with dental acrylic on the skull around the frontal trephination, forming a trough with a capacity for 100-150 μL to apply topically compounds in a restricted cortical area.

Recording of intracortical direct current potentials, regional cerebral blood flow, and data processing
An Ag/AgCl reference electrode (containing 2 mol/L KCl) was located on the nasal bone.Electrodes with tip diameters of approximately 5 μm (resistance <10 MΩ) for direct current (DC) and electrocorticogram (ECoG) recordings were filled with 150 mmol/L NaCl.An electrode sensitive for changes in extracellular potassium concentration ([K + ] e ) was also used for recording.This electrode was filled with 100 mmol/L KCl, and the tip was silanized and filled with K + ion changer (WPI 190,Sarasota,FL).The electrode was calibrated with KCl solutions (3, 6, 12, 24, 48 mmol/L).Changes in voltage (millivolts) associated to the increase of the KCl concentration during the calibration were calculated, establishing a voltage of 0 mV to the lowest KCl concentration (3 mmol/L) (Kraig and Nicholson, 1978).The [K + ] e electrode was attached to an electrode for DC recordings.Following standard procedures for in vivo CSD recordings, two DC electrodes were glued together with a vertical and horizontal separation to simultaneously record CSD-related DC potential shifts at different cortical depths.Two sets of microelectrode arrays were built: one set with two DC electrodes and the electrode for CSD elicitation, and the other with two DC electrodes and the K + sensitive electrode.The sets of electrodes were lowered into cerebral cortex to a depth of 1200 μm (in untreated cortex: KCl injection and DC 1; in treated cortex [K + ] e , and DC 3) and 400 μm (in untreated cortex: DC 2, in treated cortex: DC 4).The electrode for CSD elicitation contained 1 mol/L KCl.CSD waves were ignited by injection of 0.5 μL 1 mol/L KCl solution with a pressure of 100 kPa using a microinjector (picoinjector PLI-100; Harvard Apparatus, Holliston, MA).Injection times ranged from 0.1 to 1 s.If the first KCl-microinjection 0.1 s did not ignite a CSD, the injection time was increased in steps of 0.25 s.Intervals of repeated microinjections were 3-5 min.The shortest injection time that elicited reproducibly 2 times a CSD wave was established as threshold.The DC signals were recorded using a 4-channel highimpedance amplifier (Meyer, Munich, Germany) and stored on a computer (sampling rate 2.048 Hz).The criteria to accept a CSD were a steep onset with a peak between 1 and 3 s, amplitudes >5 mV, and propagation from electrode 1 to 4. In each animal two CSDs were ignited at intervals of 20 min by KCl microinjection before application of the test compounds to the treated area to confirm the ability to ignite CSD.
CSD were evaluated using the following parameters: occurrence in treated areas, number of CSDs, maximal amplitudes related to baseline before the depolarization, duration at half-maximal amplitude and propagation time (velocity from the elicitation site at electrode DC 1 to the deepest electrode in the treated area, electrode DC 3), and amount of KCl required to ignite a CSD.
In addition to DC measurements, the regional cortical blood flow (rCBF) was measured continuously with a Doppler flowmetry system (LDF) (Laser Blood Flow Monitor DRT4, Moor Instruments, Millwey, Axminster, Devon, EX13 5HU, UK).A sensor with a tip diameter of approximately 1 mm was placed onto the cerebral surface in the area where Gal was applied, avoiding the major blood vessels but close to the DC recording electrodes.The other probe was placed on the untreated area using the same criteria.CSD-related changes in regional cortical perfusion during Gal application were measured between rCBF baseline before CSD and the maximal peak of the rCBF deflection, and they were expressed as a percentage of the CSD-related changes before Gal application (control rCBF deflection, 100%).
To analyze patterns of depressed ECoG activity associated with CSD, or other types of pathologic electroencephalographic activity (e.g.epileptiform patterns or seizures), DC recordings were resampled offline with a sample rate of 205 Hz and first detrended by appropriate adaptive filtering, followed by band pass filtering (bandpass 0.01-45 Hz).To reveal alternate current (AC)-ECoG activity, the signals were high-pass filtered with a lower frequency limit of 0.5 Hz.In addition, power of bandpass filtered recordings (0.5-45 Hz) and the integral of power of bandpass filtered recordings were calculated (Dreier et al., 2017).

Application of substances
After establishing the baseline CSDs (control), ACSF was removed from the frontal brain opening, and Gal was applied onto the treated area at 10 − 6 , 10 − 7 , 10 − 8 , 10 − 9 or 10 − 10 mol/L (TOCRIS Bio-Techne GmbH, Wiesbaden-Nordenstadt, Germany, diluted in PBS) (one concentration per experiment).After Gal application, CSDs were elicited every 30 min until a maximal duration of 3 h after Gal application.Then Gal was washed from the treated area with ACSF for 1 h with washes every 15 min.During the wash phase, 2 KCl injections to ignite CSD were performed (one after 30 min and one after 1 h wash).
To antagonize the Gal effects, we applied either the Gal receptor antagonist M40 (TOCRIS, diluted in PBS) at 3 nmol/L, a concentration that blocks GalR1, or the GalR2 antagonist M871 (TOCRIS, diluted in PBS) at 3 nmol/L, or the GalR3 antagonist SNAP 37889 (MedChemExpress, MCE, Monmouth Junction, NJ, diluted in DMSO to 1 mmol/L stock solution and further thinned in PBS) at 30 nmol/L topically to the cortical surface in the treated area.We performed a topical pretreatment for 2 h with CSD inductions 30, 60, 90 and 120 min after antagonist application.Then the antagonist was washed away and Gal at 10 − 7 mol/L was applied to that brain area for another 2 h with CSD elicitation again after 30, 60, 90 and 120 min.In another subset of rats each GalR antagonist was tested in simultaneous application: the frontal brain area was topically treated with co-application of one of the three GalR antagonists and Gal at 10 − 7 mol/L for 4 h.Single CSDs were induced after 30,60,90,120,150,180,210 and 240 min of coapplication.

Data statistics
Data in bar diagrams are reported as mean ± standard error of the mean.In the box plot diagrams, down and up borders of the boxes represent the percentile 25 and 75 of the data respectively.The line inside the box gives the median of the data.Down and up whiskers represent percentile 10 and 90 of the data respectively.Scatter plots over bar or box diagrams show the single data distribution.For statistical analysis we used the InStat software package (Graph Pad, San F. Gimeno-Ferrer et al.Diego, CA), testing for normal distribution, and performing tests within groups (paired t-test, Student), one-sample t-tests for comparison with a standard value or testing between groups (Mann-Whitney U test).Alpha adjustment was performed when necessary.Significance was accepted at p < 0.05.

Expression of Gal, GalR1, GalR2 and GalR3 in rat naïve cerebral cortex
The neuropeptide Gal was present in all cortical neurons at all layers, including also inhibitory parvalbumin (parv) neurons.The distribution of GalRs differed between cortical depths: whereas GalR2 and GalR3 were identified in all cortical neurons in all layers (including inhibitory neurons), GalR1 was only expressed in layers IV/V by about half of the neurons.Part of GalR1-positive neurons were also parv-positive, showing GalR1 also in inhibitory neurons (For overview, see Fig. 1A, for detailed images with co-localization Fig. 1B,C).Since all Gal-or GalR-positive cells were neuronal, we did not see evidences for the expression of Gal or GalRs in other cells.

Gal effect on CSD
CSDs were elicited in the untreated area by microinjection of KCl and recorded by DC electrodes in the untreated and the treated area.The topical application of Gal to the treated area (Fig. 2A) induced a reduction of the amplitudes of propagating CSDs until a complete abolishment in some cases in the treated area.A typical example of almost complete abolition of CSD by Gal at 10 − 6 mol/L in the treated area, which did not recover after the washout, is shown in Fig. 2B.A decrease in CSD amplitudes by Gal at 10 − 7 mol/L in the treated area is shown in Fig. 2B.In this example, the decrease in CSD amplitudes even continued during the 60 min washout phase.Together with the decline in CSD amplitudes, the [K + ] e magnitude at electrode DC 3 was decreased.In the untreated areas (kept moist with ACSF during the entire experimental course), no significant change in amplitudes was observed in the same time range (Fig. 2B and C, traces of electrode 2).Fig. 3A shows the summarized data of the impact of all concentrations of Gal on CSD amplitudes related to the baseline before application (normalized to 100%).The decline in amplitudes in the treated area was insignificant at doses of 10 − 9 and 10 − 10 mol/L.First effects of Gal became visible with the extinguishment of CSD in some animals after a h treatment with a dose of 10 − 8 mol/L.However, the reduction in amplitudes was still insignificant.A significant reduction of CSD amplitudes was only observed at doses of 10 − 7 mol/L and 10 − 6 mol/L (paired t-test).We consider the concentration of 10 − 8 mol/L Gal as a threshold dose, since first Gal effects on CSD amplitude were clearly visible at this Gal concentration.However, a big inter-individual variability between the rats was noted (abolishment of CSD versus small or no changes in amplitudes).
To exclude that the reduction in CSD amplitudes could be explained by a simple decay of the exposed brain areas over time, we compared i) the CSD amplitudes in the untreated brain areas and did not find any significant differences (Fig. 3A, white boxes), and ii) performed in another subgroup of rats the same experimental protocol but applied only ACSF to both exposed brain areas.These rats did not show a significant decrease in CSD amplitudes (Fig. 3A, far right).
A CSD wave is usually accompanied by a transient increase in regional cortical blood flow (rCBF) (Mayevsky and Weiss, 1991).We measured the CSD-related changes in rCBF in both cortical areas (see Fig. 2) during the control phase when only ACSF was applied during all CSDs, and set the mean values as 100%.The CSD-related changes in rCBF after 3 h application of Gal and after 1 h washout were related to these baseline values as percentages.Fig. 3B shows that the application of ACSF in the untreated brain area had no significant effect on the changes in rCBF (white bars).In some animals low baseline rCBF values during controls led to higher variability.However, the application of Gal, not only reduced the amplitudes of CSD but also diminished the CSD-related changes in rCBF (light grey bars).As seen in the CSD amplitudes, a washout of Gal only partially restored the changes in rCBF (dark grey bars).
Together with the decline in CSD amplitudes, we observed a slowing down of CSD spreading from untreated to treated area after topical application of Gal (Fig. 3C, light grey bars).During the control phase of the experiment where both areas were treated by ACSF (Fig. 3C, white bars), and in the animal group treated at both exposed brain areas only with ACSF (Fig. 3C, far right) stable propagation velocities were observed.The slowing in CSD propagation in groups treated with Gal was statistically significant at concentrations of Gal at 10 − 6 , 10 − 7 , 10 − 8 , and 10 − 10 mol/L (paired t-test, control vs 3 h Gal).The washout of Gal did not always restore CSD propagation velocity, only at 10 − 7 mol/L the Gal effect on propagation velocity was reversed (Fig. 3C, dark grey bars).
Gal had also an impact on CSD excitability outside of the treated area.The threshold of ignition (application time of KCl to induce the CSD) was increased by the higher concentrations of Gal.This effect was expressed as X-fold of the threshold in the control phase which was set to 1.The threshold increased from 1 to 2.61 after Gal 10 − 6 mol/L, from 1 to 2.52 after Gal 10 − 7 mol/L, and from 1 to 1.47 at Gal 10 − 8 mol/L.However, threshold remained unchanged when Gal was applied at 10 − mol/L and 10 − 10 mol/L.Gal also reduced the number of ignited CSDs that propagated from the ignition site to the treated area.After Gal 10 − mol/L the percentage of propagating CSD was reduced to 89.3%, after Gal 10 − 7 mol/L to 90.5%, and after Gal 10 − 8 mol/L to 90.5%, respectively.No change in number of propagating CSD was observed after Gal In sum, Gal induced a consistent decrease in propagation velocities.CSD amplitudes were reduced by Gal only at high concentrations.A remarkable increase in KCl threshold to ignite CSDs in the untreated areas was observed at high concentrations of Gal.
Interestingly, after washout of Gal with ACSF some rats showed cortical epileptiform activity characterized by sharp waves and/or ictal discharging activity.Fig. 4 shows a case in which ictal discharges were observed 30 and 45 min after washout, only in the treated area.Such cortical hyperactivity was seen in 5 out of 22 animals (22.7%) tested with Gal at concentrations from 10 − 6 mol/L to 10 − 8 mol/L.

Role of Gal receptors in influencing the Gal effect on CSD parameters
To explore the putative role of the three GalRs influencing CSD parameters, different antagonists were used: M871 for GalR2 (3 nmol/L) (Sollenberg et al., 2010;Sollenberg et al., 2006), SNAP 37889 for GalR3 (30 nmol/L) (Swanson et al., 2005) and M40 (non-specific GalRs antagonist) in a concentration of 3 nmol/L that blocked preferably GalR1 instead of GalR2 and GalR3 (Lu et al., 2005).We applied the antagonists either in a pre-treatment protocol in which the antagonist was applied for 2 h followed by Gal at 10 − 7 mol/L for 2 h (Fig. 5A), or we used co-application of the antagonist and Gal for 4 h (Fig. 5B).As shown in Fig. 5A, Gal alone at 10 − 7 mol/L caused a reduction of CSD amplitude in the treated area in some rats, but such an effect was not observed after pre-treatment with any of the three antagonists.The application of the antagonists alone (value after 2 h) had no own effect on CSD amplitude.In the co-application protocol, only the GalR2 antagonist prevented significantly (Mann-Whitney U test, p < 0.05) the reduction of CSD amplitude by Gal whereas the GalR1 and GalR3 antagonists did not always prevent the Gal effect (Fig. 5B).
For unkown reasons, the pre-treatment with any of the three GalR antagonists slowed the CSD propagation velocity similarly as the treatment with Gal alone (data not shown).Interestingly, after a pretreatment with the GalR2 or the GalR3 antagonists Gal did not further reduce CSD propagation velocity.In the co-application groups, a slowing of CSD propagation velocity was recorded, but this effect may result either from Gal or from the GalR antagonists or from both together.
We further compared the thresholds for KCl microinjection in the untreated areas to ignite CSD in the co-application and in pre-treatment experiments with those after application of Gal at 10 − 7 mol/L alone.The blockade of GalR2 with M871 prevented any threshold changes by Gal in both types of application.By contrast, in pre-treatment during blockade of GalR3 by SNAP 37889 the threshold for KCl microinjection increased, since 2 of 3 animals required longer KCl microinjections for CSD ignition.This effect was not observed in the co-application of SNAP 37889 and Gal.We conclude that the effect of Gal on ignition threshold is mediated by GalR2.

Gal effect on glial cells
Because glial cells play a role in CSD, we also studied whether nonneuronal cells in the cerebral cortex are affected by Gal.Staining of astrocytes with anti-GFAP did not reveal a noticeable change in astrocytes after 3 h of Gal 10 − 6 mol/L, nor did the pre-treatment with M871 3 nmol/L for 2 h and the subsequent application of Gal 10 − 7 mol/L for 2 h (Fig. 6A).However, microglia cells labelled by anti-Iba1 showed a visible reaction after 3 h of Gal 10 − 6 mol/L, manifested with a significant decrease in the number of endpoints/cell and branch length/cell.No significant change in number of cells was detected between treated and untreated areas (Fig. 6 B,C).These effects of Gal were prevented by the blockade of GalR2 with M871 3 nmol/L for 2 h applied before the subsequent application of Gal 10 − 7 mol/L for 2 h (Fig. 6B,D).Therefore, these data suggest an indirect reaction of the microglia to Gal mediated by neuronal GalR2.

Discussion
In this study, we show that Gal is expressed in all neurons throughout the cerebral cortex.However, the distribution of the three GalRs subtypes differed: whereas GalR2 and GalR3 are expressed in all cortical neurons including inhibitory ones, GalR1 is preponderantly located in half of the neurons in cortical layers IV and V. Some of these neurons are also inhibitory.We did not find any GalR staining in glial nor in vascular structures.From that, we conclude that the observed effects of Gal are neuronally mediated.Application of Gal to the cortical surface diminished CSD amplitudes, slowed CSD propagation velocity from the untreated into the treated brain area and increased the threshold for KCl to ignite CSD in the remote brain area.These Gal effects were robustly blocked by an antagonist at GalR2, in pre-treatment and co-application protocols.By contrast, GalR1 and GalR3 blockers prevented Gal effects only in a pre-treatment protocol.Interestingly, the application of Gal changed the morphology of microglia cells mediated by GalR2 although there is no evidence for the expression of GalRs by microglia cells themselves.In some cases, rapid removal of Gal resulted in epileptiform discharges in the ECoG.Therefore, Gal may keep cortical excitability at a stable level and may attenuate pathological cortical hyperexcitability.
The neuronal location of Gal has been extensively investigated e.g., in the locus coeruleus, hippocampus, corpus callosum, ascending systems, dorsal raphe, dorsal root ganglia, and spinal cord (Lang et al., 2015).Interestingly, to our knowledge, the only report about Gal in cerebral cortex was provided by Shen et al. (2003) and Shen and Gundlach (2010) who found very few neurons expressing Gal in naïve neocortex, but a strong expression of Gal immunoreactivity 2 days after CSD.In contrast, we observed extensive Gal immunoreactivity in all cortical neurons of naïve cerebral cortex.
There is little information in the literature on the physiological role of a basal release of Gal in the cerebral cortex.Merchenthaler et al. (1993) reviewed effects of Gal in reducing acetylcholine release in the ventral hippocampus and in controlling the release of neuroendocrine hormones such as gonadotropin, prolactin, growth hormone and adrenocorticotropin.In the hippocampus of freely moving rats, Consolo et al. (1994) recorded a spontaneous release of Gal of 1.8 ± 0.3 fmol/mL per 20 min using microdialysis.In the spinal cord the basal Gal release was close to the detection level of the radioimmunoassay used (Hygge Blakeman et al., 2001).A different picture was seen in pathophysiological situations when nerve injury induced a release of Gal in the spinal cord (Colvin and Duggan, 1998;Duggan and Riley, 1996).
The distribution of the three GalRs differs again between animal species: GalR1 and GalR2 are present in all vertebrates, GalR3 only in some mammals (Liu et al., 2010).Since there is only sparse information of the GalR expression in rat neocortex, we stained for all subtypes and found a homogeneous distribution of GalR2 and GalR3 throughout all cortical layers, but only a minority of neurons expressed GalR1 in deeper Fig. 2. Electrophysiological recordings from the rat cortex in vivo before and after Gal application.A) Schematic drawing of the rat skull (not to scale) with the two trephinations (treated and untreated areas), inserted electrodes and LDF probes in each area.B) Representative CSD recorded simultaneously with 5 electrodes (3 for DC-current and 1 for the associated increase in [K + ] e , the DC electrode showing the KCl microinjection is left off) from treated and untreated brain areas before (control), 3 h after topical application of Gal at 10 − 6 mol/L, and after 1 h final wash with ACSF.Arrows mark the microinjection of KCl to elicit CSD.Dotted lines accentuate CSD propagation times from site of elicitation to treated area that was slowed after Gal.The panels of each electrode show the DC recordings and the respective highpass filtered electrocorticographic data with a lower frequency limit of 0.5 Hz (ECoG) thus indicating that there is spreading depression of electroencephalographic activity.Regional cerebral blood flow (rCFB) recordings show transient increases induced by the propagating CSDs.C) Representative recordings of the effect of Gal at 10 − 7 mol/L and the subsequent washout with ACSF.All other explanations as in panel B.
F. Gimeno-Ferrer et al. (caption on next page) F. Gimeno-Ferrer et al. cortical layers (resembling IV and V).Whereas both GalR1 and GalR3 are linked to G i/o -protein to block the adenylate cyclase and GIRK ion channels, GalR2 is coupled to G i/o , to G 12/13 , and to G q/11 and the respective signaling pathways (Lang et al., 2015).
The reported electrophysiological effects of Gal are mostly inhibitory.Gal hyperpolarized neurons in locus coeruleus and dorsal raphe neurons via outward potassium currents and decreases of membrane resistance, and it reduced the glutamate release in hypothalamus, the magnitude of excitatory Schaffer collateral activation and the longlasting depolarization in hippocampus (Pieribone et al., 1998).Moreover, Gal inhibited epileptic activity in rat hippocampus after status epilepticus (Mazarati and Lu, 2005).GalR2 knockout mice had an increased number of seizures, but such an effect was not observed in GalR3 knockout rats (Drexel et al., 2018).GalR1 knockout mice exhibited more severe seizures in epilepsy models (Mazarati et al., 2004) and spontaneous epilepsy in hippocampus (McColl et al., 2006).In turn, Gal overexpression increased the resistance against seizure induction in hippocampus (Mazarati et al., 2000).
Here we show for the first time that Gal has the potential to control CSD parameters.We applied Gal at different concentrations topically (10 − 6 to 10 − 10 mol/L) to cortical surface and found an inhibition of features of CSD, including the CSD-related change in rCBF.But several facts have to be noticed.First, statistically significant inhibitory effects of Gal on CSD amplitude were only found at Gal concentrations of 10 − to 10 − 6 mol/L.For comparison, the EC 50 of Gal in rat hippocampal slice was 1.1 × 10 − 9 mol/L (Saar et al., 2002).Since Gal was applied to the surface of the cortex, we do not know the exact concentrations at different cortical sites.However, even with the effective concentrations, CSDs were not equally modified by Gal in all animals tested.While some animals showed a total failure of CSD after Gal, other animals did not exhibit an effect of Gal on CSD even at higher Gal concentrations.Thus, susceptibility to Gal, concerning reduction of CSD amplitude, was not a     similar robust finding throughout the rats.In line with this, in few animals we observed an increase in KCl threshold to ignite CSD outside of the Gal-treated brain areas.We can only speculate whether the diffusion of Gal over time reached not only the deeper cortical layers but also brain areas surrounding the Gal-treated ones thereby interfering with CSD elicitation.Another explanation is that Gal induced an inhibition in the treated brain area that spread to the untreated brain areas.
Interestingly, only in a minor number of animals the washout of Gal at doses ranging from 10 − 6 down to 10 − 8 mol/L induced the development of spontaneous ictal discharging activity in the ECoG.There is evidence in the literature that Gal reduces the presynaptic glutamate release in the hippocampus via GalR1 (Zini et al., 1993).We can only speculate that such a mechanism also operates in the cerebral cortex, but this would explain both the slowing of CSD propagation which is also driven by glutamate, and conversely the development of ictal discharging activity after removal of Gal by washout as a sign of overexcitation.Surprisingly, ictal activity for some time coexisted with propagating CSD waves, as we had previously observed (Gimeno-Ferrer et al., 2022).Therefore, at least at the onset of epileptogenic activity in the cortex, CSD waves could coexist as a state of hyperexcitability.We suppose that the previously ignited CSDs in a minor group of animals still had maintained the increased excitability.Therefore, altogether, Gal inhibited CSD amplitudes particularly in some very susceptible animals.In this particular group, the effect of Gal was pronounced because the inhibitory effect of Gal was barely reversed after washout of Gal.A more homogeneous picture was seen in the reduction of the propagation velocity of CSDs.
In this respect, it should be noted that a previous study reported the induction of Gal mRNA 48 h after CSD, and of GalR1 and GalR2 mRNA immunoreactivity and receptor binding even later (Shen et al., 2003).Thus, effects of Gal may be more homogeneous after preceding events evoking CSDs.On the other hand, our immunohistochemical data show a marked expression both Gal and GalRs in the naïve cortex, suggesting that the Gal system is expressed under normal basal conditions.
To differentiate which GalR subtypes affect CSDs, we used the most specific receptor antagonists.The GalR2 antagonist M871 in a dose of 3 nmol/L (Sollenberg et al., 2006) prevented Gal effects on CSD amplitudes and threshold in pre-treatment as well as in co-application protocols.The GalR3 blocker SNAP 37889 (Swanson et al., 2005) prevented the Gal-evoked decrease in CSD amplitudes but not the Gal-evoked increase of CSD threshold.However, this inhibitory SNAP 37889 effect was only observed upon pre-treatment, not upon co-application with Gal.Since no specific GalR1 antagonist is available, we used the antagonist M40 in a dose that preferentially binds to GalR1 (Lu et al., 2005).M40 had similar effects as SNAP 37889 on CSD amplitudes but prevented in addition the Gal-evoked threshold increase of CSD ignition.Taken together, the antagonist at GalR2 was most efficient in blocking the Gal effects on CSD whereas antagonists at GalR1 and GalR3 were only efficient upon pre-treatment.
The precise mechanism of the robust GalR2 effect on CSD threshold is unclear.According to Lang et al. (2015) the activation of calciumactivated (big) potassium channels and calcium-dependent chloride channels hyperpolarizes neuronal membranes.Thus, this cellular pathway may be more important for CSD amplitude reduction than the GalR1 and GalR3 pathways, which block the effects on adenylate cyclase and GIRK ion channels.
Upon pre-treatment, all three GalR antagonists reduced CSD propagation velocity.However, Gal application after pre-treatment with M871 and SNAP 37889 did not further reduce CSD propagation velocity suggesting that both substances blocked further Gal effects.The minor blocking effect of the GalR1 antagonist may correspond to the restricted localization of GalR1 in deeper cortical layers.The particular mechanism how GalR antagonists interfere with CSD propagation is unclear yet.Possibly, GalR antagonists impair via presynaptic GalRs the release of excitatory neurotransmitters required for CSD propagation.Alternatively, they may interact with the extracellular matrix building up a hindrance for volume transmission required for CSD movement (Syková, 2004).
We did not find an expression of Gal nor of GalRs in non-neuronal cells (microglia, astrocytes, and endothelial cells) in the native brain.This contrasts to earlier studies (Shen et al., 2005) that found Gal production induced by CSD in oligodendrocyte progenitor cells.This is not necessarily a contradiction.In previous studies Gal was expressed after challenges to the brain, e.g.colchicine application (Ubink et al., 2003) or focal ischemia (Gill et al., 1992), and this Gal expression was only seen 48 h or even later after the challenge.
Although there was no GalR expressed at the microglia, we found a change of microglial morphology 3-4 h after Gal application, suggesting that Gal can affect indirectly microglia.Whether the activation is pro-, or anti-inflammatory cannot be determined from morphological changes which we observed within 3-4 h of Gal application.The participation of neuronal GalR2 is likely since the treatment with the GalR2 antagonist maintained normal microglial morphology.Notably morphological changes after Gal were observed in native brains without CSDs or mechanical challenges (insertion of electrodes) indicating that the changes were in fact evoked by Gal.Most likely Gal causes release of mediators from neurons, which affect microglia.Candidates are the neuropeptide TGF-β or the cytokine IL-10 ( Butovsky et al., 2014;Salvi et al., 2017).

Conclusion
Taken together, our data show that Gal attenuates CSD.In susceptible animals, Gal decreased CSD amplitude, increased the ignition threshold in a remote brain area, and slowed CSD propagation.The effects of Gal were best antagonized by a GalR2 antagonist.Since CSD, a neuronal and glial mass depolarization, has pathophysiological relevance, the observed Gal effects on CSDs are likely to protect the brain from further damage in situations which evoke CSDs, e.g., stroke.We could show that in healthy rats, Gal was effective to interact with situations of hyperexcitablity suggesting the potential importance of this neuropeptide in the control of cortical excitability.We elicited CSDs in the normal brain by KCl ignition and did not study the effect of Gal in a damaged brain.Since inhibitory effects of Gal were particularly visible in the context of epilepsy, where Gal was supposed to be an "endogenous anticonvulsant" (Mazarati et al., 2001), the Gal system may be even more effective in modulating CSDs in pathophysiological situations with hyperexcitability.

Fig. 1 .
Fig. 1.Localization of Gal and its receptors (GalR1, GalR2 and GalR3) in naïve rat cortex.A) Distribution of Gal, GalR2 and GalR3 in all cortical neurons at all cortical layers.GalR1 is only expressed by about half of the neurons in layers IV/V.Scale bars 25 μm.B) Confirmation of the distribution depicted in A by the co-localization with NeuN.In green: immunolabeling of Gal in left panel, GalR1 in middle left panel, GalR2 in middle right panel and GalR3 in right panel; in blue: cellular nuclei stained by Hoechst 34580; in red: immunolabeling of neuronal somata (nucleus and perinuclear cytoplasm) with anti-NeuN; in yellow: co-localization of Gal, GalR1, GalR2 and GalR3 with NeuN.C) Expression of Gal, GalR1, GalR2 and GalR3 by parvalbumin (parv) inhibitory GABAergic neurons.In green: immunolabeling of Gal in left panel, GalR1 in middle left panel, GalR2 in middle right panel and GalR3 in right panel; in blue: cellular nuclei stained by Hoechst 34580; in red: immunolabeling of parv; in yellow: colocalization of Gal, GalR1, GalR2 and GalR3 with parv.Scale bars in B and C 10 μm.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3 .
Fig. 3. Effect of Gal on CSD amplitude and propagation velocity.A) Amplitude values are normalized as percentage of control amplitude, marked with the dashed line at 100%.No significant change of CSD amplitudes within 3 h in untreated areas (white boxes).Significant reduction of CSD amplitude in treated areas (cortical depths 1200 μm and 400 μm) at Gal 10 − 6 , and 10 − 7 mol/L.Non-significant reduction of CSD amplitude in treated areas at Gal 10 − 8 mol/L.No reduction after 3 h of Gal at 10 − 9 and 10 − 10 mol/L and ACSF.Boxplot represents distribution of data with down and up borders of the boxes representing the percentile 25 and 75 of the data respectively.Line inside the box represents the median of the data.Down and up whiskers represent percentile 10 and 90 of the data respectively.Dots represent single datapoints.B) Effect of Gal on the CSD-related increase in rCBF.The rCBF changes during the control phase are normalized to 100%.No significant change within 3 h in untreated areas (white bars), but reduction of the CSD-related changes in rCBF after Gal.The bars show mean values ± standard errors.Dots show single datapoints.Statistical comparisons versus control (100%) were made with the one-sample t-test, ** p < 0.01, *** p < 0.001.C) Reduction of the spreading velocity of CSD 3 h after Gal.Bars represent mean ± standard error of propagation velocity in mm/min.Non-propagating CSDs are excluded.The wash out of Gal with ACSF for 1 h barely recovered the velocity (only at 10 − 7 mol/L).In the ACSF group all bars show CSDs under ACSF at the appropriate time points.Dots represent single datapoints.In A) and C) paired t-test control versus 3 h Gal or wash, * p < 0.05, ** p < 0.01; n gives number of tested animals per group.Discrepancies between n and number of datapoints are due to electrode failure (in A and C), or to LDF failure in B.

Fig. 4 .
Fig. 4. Induction of cortical ictal activity after the washout of Gal from cortex with ACSF.Example of ictal activity on the EEG induced after the abrupt washout of Gal 10 − 8 mol/L restricted only to the treated area.The left panel shows the last CSD ignited by KCl during Gal treatment.The middle panel shows that in the first min of washout, a CSD can still be elicited by KCl, but the amplitude in the treated area is reduced probably due to the ongoing ictal activity.The right panel shows ongoing ictal activity in the treated area only.At that time no CSD was elicited.

Fig. 5 .
Fig. 5. Deciphering the role of the three GalRs on CSD amplitudes.A) Pre-treatment with GalR1 (M40), GalR2 (M871) and GalR3 (SNAP 37889) antagonists prevented the effect of Gal on CSD amplitudes.In each experimental group, two sets of three boxes are displayed: in the first set, the effect of ACSF (white) or the effect of the antagonist alone for 2 h (light and dark grey, corresponding to a cortical depth of 1200 and 400 μm, respectively) is shown; the second set displays the effect of the subsequent application of Gal 10 − 7 mol/L for another 2 h.B) Effect of GalR1 (M40), GalR2 (M871) and GalR3 (SNAP 37889) antagonists in coapplication with Gal on CSD amplitudes.Only the GalR2 antagonist consistently blocked the reduction of amplitudes induced by Gal.The GalR2 antagonist significantly prevented the reduction of CSD amplitudes by Gal at 10 − 7 mol/L (Mann-Whitney U test, 3 h Gal versus 4 h Gal+GalR2 antagonist, # p < 0.05).In A) and B) Significant reduction of CSD amplitudes after Gal at 10 − 7 mol/L (paired t-test control versus 3 h Gal, * p < 0.05).Down and up borders of the boxes represent the percentile 25 and 75 of the data respectively.Line inside the box represents the median of the data.Down and up whiskers represent percentile 10 and 90 of the data respectively.n gives number of tested animals per group.Discrepancies between n and number of datapoints are due to electrode loss.

Fig. 6 .
Fig. 6.Effects of Gal on glial cells.A) Gal 10 − 6 mol/L treatment for 3 h did not induce a visible change in astrocyte morphology labelled with anti-GFAP antibody.M871 GalR2 antagonist at 3 nmol/L as pre-treatment for 2 h and the subsequent application of Gal 10 − 7 mol/L for 2 h did not induce effects on astrocytes.B) Iba1immunofluorescence after application of Gal at 10 − 6 mol/L for 3 h revealed a change in microglia morphology.This change in microglia cells was prevented with the application of M871 GalR2 antagonist at 3 nmol/L as pre-treatment for 2 h.In A) and B) scale bars 50 μm.C) Quantification of endpoints/cell of microglia, process length/cell of microglia and number of microglia cells in areas treated with Gal 10 − 6 mol/L for 3 h and untreated areas in a ROI of 750 × 750 μm.D) Quantification of endpoints/cell of microglia, process length/cell of microglia and number of microglia cells in areas treated with M871 3 nmol/L for 2 h and the subsequent treatment of Gal 10 − 7 mol/L for 2 h and untreated areas in a ROI of 750 × 750 μm.In C) and D), bar graphs show the comparison between untreated and contralateral treated area.The columns show mean values ± standard errors.Dots show single datapoints.Statistical comparisons versus controls were made with the paired t-test, ** p < 0.01, *** p < 0.001.t-tests between control in C and control in D were performed without any statistical difference.