Ghrelin Regulates Expression of the Transcription Factor Pax6 in Hypoxic Brain Progenitor Cells and Neurons

The nature of brain impairment after hypoxia is complex and recovery harnesses different mechanisms, including neuroprotection and neurogenesis. Experimental evidence suggests that hypoxia may trigger neurogenesis postnatally by influencing the expression of a variety of transcription factors. However, the existing data are controversial. As a proof-of-principle, we subjected cultured cerebral cortex neurons, cerebellar granule neurons and organotypic cerebral cortex slices from rat brains to hypoxia and treated these cultures with the hormone ghrelin, which is well-known for its neuroprotective functions. We found that hypoxia elevated the expression levels and stimulated nuclear translocation of ghrelin’s receptor GHSR1 in the cultured neurons and the acute organotypic slices, whereas ghrelin treatment reduced the receptor expression to normoxic levels. GHSR1 expression was also increased in cerebral cortex neurons of mice with induced experimental stroke. Additional quantitative analyses of immunostainings for neuronal proliferation and differentiation markers revealed that hypoxia stimulated the proliferation of neuronal progenitors, whereas ghrelin application during the phase of recovery from hypoxia counteracted these effects. At the mechanistic level, we provide a link between the described post-ischemic phenomena and the expression of the transcription factor Pax6, an important regulator of neural progenitor cell fate. In contrast to the neurogenic niches in the brain where hypoxia is known to increase Pax6 expression, the levels of the transcription factor in cultured hypoxic cerebral cortex cells were downregulated. Moreover, the application of ghrelin to hypoxic neurons normalised the expression levels of these factors. Our findings suggest that ghrelin stimulates neurogenic factors for the protection of neurons in a GHSR1-dependent manner in non-neurogenic brain areas such as the cerebral cortex after exposure to hypoxia.


Introduction
Stroke is a medical condition of impeded blood supply to the brain and oxygen shortage (hypoxia). The recovery from hypoxia requires intra-and extracellular processes recapitulating nervous system development such as extracellular matrix reorganisation, neurogenesis, and stimulation of neuronal plasticity. A prolonged hypoxia state may result in ischemia followed by neuronal death and reduction of neuronal density [1]. Interestingly, hypoxia can also trigger neurogenesis within the surrounding tissue [2,3] prenatally as well as postpartum [4,5]. Neurogenesis in the adult brain is essential and occurs during the entire postnatal life. It is believed to be restricted to the subventricular zone and the subgranular zone of the dentate gyrus, though it has been observed in some other areas

Acute Cerebral Cortex Slices
Acute cerebral cortex slices were obtained from a mixed population of 4-day-old female and male Wistar rats. The animals were anaesthetised by isoflurane inhalation and decapitated. Brains were isolated, the brain hemispheres were dissected in DMEM medium (Merck) and sliced at a thickness of 400 µm perpendicularly to their longitudinal axis using the McIlwain tissue chopper (Plano, Wetzlar, Germany). Slices were placed onto Millicell membrane inserts (Merck) and transferred into six-well plates with 1 mL of nutrition medium per well (25% heat-inactivated horse serum, 25% HBSS, 50% DMEM, 2 mM glutamine, pH 7.2). Slices were maintained under standard conditions of 37 • C, 100% humidity and 5% CO 2 for a period of 24 h prior to hypoxia.

Induction of Hypoxia and Treatment with Ghrelin
One-third of the cultured cells/acute slices were kept under normoxic conditions (37 • C, 100% humidity and 5% CO 2 ) and the rest of the cultures were exposed to hypoxia for 6 h. Hypoxia was achieved by subjecting the cultures and slices to air evacuation [23], which took~3-4 s. The induction of hypoxia begins with lowering the partial oxygen pressure (pO 2 ) from~160 mmHg to less than 25 mmHg within 30 min. This process was visually inspected with an oxygen indicator. The specimens were air-evacuated and sealed in a plastic bag, and then incubated at 37 • C and in a humidified atmosphere for 6 h (for further details see also [23]). After six hours of hypoxia, the cultures and slices were maintained under normoxic conditions (37 • C, 100% humidity and 5% CO 2 ) for 24 h. Thereby, half of these cultures/slices were supplemented with human ghrelin peptide (cat.# ab199421, Abcam, Cambridge, UK) for the duration of 24 h (final concentration of 0.5 µM as described elsewhere [24][25][26][27]). The other half of the cultures were kept in a plain medium, also for 24 h, and used as a control.
To test the specificity of the GHSR1 antibody, we conducted two control staining procedures as follows. Prior to staining, we incubated the murine primary GHSR1 antibody with murine pituitary/hypothalamus tissue homogenate to bind the antibody with detergent-solubilised GHSR1. In particular, tissue of the pituitary gland and hypothalamus from an 11-month-old male mouse was freshly isolated and homogenised in RIPA buffer (150 mM sodium chloride, 50 mM tris-HCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, pH 8.0). The homogenate was then centrifuged at 13,000 rpm (room temperature) for 10 min and the supernatant was used for further incubation steps. Approximately 120 µg of the supernatant protein were mixed with 4 ng of the GHRS1 antibody at room temperature for 30 min. This mixture was applied to fixed cerebral cortex cells at 4 • C for 10 h, followed by the standard steps of immunocytochemistry as described above. For the second control staining procedure, we applied only the Alexa 594-conjugated secondary antibody (without the primary GHSR1 antibody) to the cells (using a dilution of 1:1000) for 2 h. In both cases, the GHSR1 fluorescence signal was barely detectable with the set-up of the spinning disk microscope as described in the imaging acquisition section.

Homogenate Preparation for Dot Blot Analysis
Cerebral cortex slices (described above) were homogenised in lysis buffer containing 100 mM Tris-HCl, pH 7.4 (Carl Roth), 12 mM magnesium acetate tetrahydrate (Merck), and 6 M urea (Sigma-Aldrich, St. Louis, MO, USA) under several freezing-refreezing rounds in liquid nitrogen. The total protein amount was measured on the Genova Nano microvolume Life Science Spectrophotometer (Jenway, Staffordshire, UK) using a direct UV light detection set-up following the manufacturer's instructions. For each condition, 20 µg total protein per dot were dropped onto a nitrocellulose membrane (cat.# 10600015, 0.2 µm, GE Healthcare, Chicago, IL, USA). The membrane was rinsed in Vilber washing buffer (cat.# PU4000500, Vilber, Collégien, France) at room temperature for 30 min and then incubated with a murine antibody recognising GHSR1 (cat.# sc-374515, RRID:AB_10987651, Santa Cruz Biotechnology, dilution 1:1000, incubation overnight) or mouse β-actin antibody (cat.# A5441, RRID:AB_476744, Sigma-Aldrich, dilution 1:10,000, incubation for 1 hour) which were diluted in the Vilber Purity TM anti-mouse HRP reagent (cat.# PU4200100, Vilber) following the manufacturer's instructions. The membranes were then incubated in the Vilber Purecl TM Dura substrate (cat.# PU4400125, Vilber) according to the manufacturer's instructions and subjected to chemiluminescence detection using the Vilber Fusion FX Imager (Vilber). The duration of the exposure for the GHSR1 and β-actin signal were 10 s and 20 s, respectively. For quantification of chemiluminescent signals, the ImageJ2 software (version 2.3.0/1.53f, Fiji) was used. The membranes were then stained with Ponceau S solution (cat.# P7170-1L, Sigma-Aldrich) for 4 min, washed in distilled water for 15 min and air-dried.
Transmitted light microscopic images were captured using a fully motorised widefield microscope Zeiss AxioImager Z.2 (Carl Zeiss, Jena, Germany) with an AxioCam Mrm rev.3 monochrome CCD camera (Carl Zeiss), and AxioVision v.4.9 software through an EC Plan-Neofluar objective 5×. Shading from the irregular illumination field was corrected during the acquisition via the camera's built-in shading correction. The images were savedin the zvi-format (Carl Zeiss) of the AxioVision software, stitched together, and exported to an 8 bit TIFF format for further processing and analysis with the ImageJ2 software (version 2.3.0/1.53f, Fiji).

Immunogold Transmission Electron Microscopy
Formaldehyde-fixed cultures (described above) were incubated with 1% bovine serum albumin in PBS at room temperature for 30 min, followed by an incubation with the mouse monoclonal antibody against GHSR1 (diluted 1:1000 in PBS) at 4 • C for 72 h. Specimens were washed in PBS three times (20 min for each round) and incubated with a donkey polyclonal antibody coupled to 10 nm gold (cat.# ab39593, Abcam, RRID:AB_954429, dilution 1:100 in PBS) at room temperature for 1 hour. After intensive rinsing in PBS (five rounds, 5 min for each round), sections were fixed with 1% glutaraldehyde in PBS at room temperature for 10 mins. Specimens were then incubated in 0.8% NaCl and 8% glucose at room temperature for 30 min and chilled on ice for a further incubation in an aqueous mixture of 2% OsO 4 and 1% potassium hexacyanoferrate (III) (cat.# 31251, Sigma-Aldrich) for 2 h. The osmicated sections were washed in an ice-chilled aqueous mixture of 0.8% NaCl and 8% glucose for 30 min and dehydrated in 70% ethanol (2 × 15 min), 90% ethanol (2 × 15 min), 95% ethanol (2 × 15 min), 100% ethanol (2 × 30 min), and pure propylene oxide (2 × 20 min) on ice. Sections were then incubated in a propylene oxide-Araldite mixture (1:1) containing 3% accelerator on ice for 2 h and then at room temperature overnight. Sections were then transferred into Araldite containing 2% accelerator for 1 hour (at room temperature) and then the Araldite mixture was refreshed for the final embedding. The embedded sections were cured at 65 • C overnight. Semithin sections (500 nm) were cut on an ultramicrotome (Ultracut R, Leica, Wetzlar, Germany), stained with 1% toluidine blue in borax (4 min) and inspected under the Leica DME light microscope (10× and 40× objectives). Ultrathin sections (55 nm) were then cut, mounted on 100-meshed Nickel grids (Ted Pella, Redding, CA, USA) and stained with an aqueous solution of 4% uranyl acetate (cat. # E22400, Science Services, München, Germany) at room temperature for 20 min. The sections were rinsed in distilled water and then stained in aqueous 2% Pb(NO 3 ) 3 (cat. # 228621, Sigma-Aldrich). After rinsing in distilled water and air drying, the sections were subjected to low voltage electron microscopy using the LVEM25 (Delong Instruments, Brno, Czech Republic).

Data Collection and Statistical Analysis
Twenty light microscopic images of cerebral cortex sections of at least 4 animal brains with transient cerebral ischemia were analysed for GHSR1 expression. The intensity of the GHSR1 signals was measured within the cells in arbitrary units (AU) and normalised on the measured cell area. The data sets were subjected to Welch's t-test and are presented as medians ± standard deviation (SD). The sample size of each group is indicated within the corresponding figure legend. For in vitro cultures, images of 200 to 800 cells from each marker per group were taken and subjected to analysis of the intensity signals for each condition. At least four independent cultures obtained from six animals were used per condition. The percentage of cells positive for each marker was calculated and presented as means ± standard error of the mean (SEM). The intensity of the GHSR1 fluorescence signals was measured in AU and normalised on the measured cell area-the data are presented as medians ± SD. All images used for comparison were taken at the same setting on the confocal spinning disc microscope. For dot blot analysis, the chemiluminescent signals of homogenates from organotypic slices of four animals (four independent experimental samples per condition) were processed with the ImageJ software as described above and each GHSR1 signal was referenced to the signal of the corresponding ß-actin band. The data are presented as ratios (medians ± SD). The data sets were subjected to One-Way ANOVA statistical analysis using the post hoc Tukey's Multiple Comparison Test or Kruskal-Wallis Multiple Comparison Test to assess the statistical significance of differences between the various conditions. The values of p < 0.05 were regarded as statistically significant. All p-values and sample sizes are indicated in the corresponding figure/figure legends. The statistical analyses were performed with the GraphPad Prism software (version 9.3.0, GraphPad Software Inc., San Diego, CA, USA).

Experimental Ischemia In Vivo Leads to Increased GHSR1 Expression Levels in Cerebral Cortex Cells
There is a growing body of evidence indicating that ghrelin ameliorates neuroregeneration upon injury. Therefore, we became interested in testing whether transient cerebral ischemia, which was induced experimentally in mice, may alter the GHSR1 expression in cerebral cells within the ischemic region. To this aim, we occluded the middle cerebral artery of adult mice for 45 min and allowed afterwards reperfusion to take place, thus inducing transient cerebral ischemia. Animals were then sacrificed on day 28 after induction of stroke and subjected to immunostaining for the ghrelin receptor GHSR1. Analysis of immunostaining intensity signals revealed enhanced GHSR1 immunoreactivity in cortical cells of the stroke area when compared to the expression observed at the contralateral side ( Figure 1A,B). In the ipsilateral hemisphere (occlusion site), the GSHR1 signal was detected across the cerebral cortex cell bodies, whereas contralaterally, the signal was mainly restricted to the cytoplasm ( Figure 1A). The quantification analysis indicated that the condition of stroke (hypoxia) stimulated the expression of GHSR1 in cerebral cortex cells.
Test or Kruskal-Wallis Multiple Comparison Test to assess the statistical significance of differences between the various conditions. The values of p < 0.05 were regarded as statistically significant. All p-values and sample sizes are indicated in the corresponding figure/figure legends. The statistical analyses were performed with the GraphPad Prism software (version 9.3.0, GraphPad Software Inc., San Diego, CA, USA).

Experimental Ischemia In Vivo Leads to Increased GHSR1 Expression Levels in Cerebral Cortex Cells
There is a growing body of evidence indicating that ghrelin ameliorates neuroregeneration upon injury. Therefore, we became interested in testing whether transient cerebral ischemia, which was induced experimentally in mice, may alter the GHSR1 expression in cerebral cells within the ischemic region. To this aim, we occluded the middle cerebral artery of adult mice for 45 min and allowed afterwards reperfusion to take place, thus inducing transient cerebral ischemia. Animals were then sacrificed on day 28 after induction of stroke and subjected to immunostaining for the ghrelin receptor GHSR1. Analysis of immunostaining intensity signals revealed enhanced GHSR1 immunoreactivity in cortical cells of the stroke area when compared to the expression observed at the contralateral side ( Figures 1A,B). In the ipsilateral hemisphere (occlusion site), the GSHR1 signal was detected across the cerebral cortex cell bodies, whereas contralaterally, the signal was mainly restricted to the cytoplasm ( Figure 1A). The quantification analysis indicated that the condition of stroke (hypoxia) stimulated the expression of GHSR1 in cerebral cortex cells. Immunostaining with an antibody recognising GHSR1 (black staining) on a representative coronal brain section showing strong GHSR1-immunoreactivity at the occlusion site (ischemia/ipsilateral) in comparison to the signal of the contralateral side (control hemisphere). Note that the contralateral signal was restricted mainly to the cytoplasm (crescentshaped appearance). Abbreviations: cc-corpus callosum, dg-dentate gyrus, cp/lv-choroid Immunostaining with an antibody recognising GHSR1 (black staining) on a representative coronal brain section showing strong GHSR1-immunoreactivity at the occlusion site (ischemia/ipsilateral) in comparison to the signal of the contralateral side (control hemisphere). Note that the contralateral signal was restricted mainly to the cytoplasm (crescent-shaped appearance). Abbreviations: cc-corpus callosum, dg-dentate gyrus, cp/lvchoroid plexus/lateral ventricle. Scale bar, 300 µm. The GHSR1 signal intensity within the cells was measured in arbitrary units [AU]. Data are presented as medians ± SD analysed by Welch's t-test.
The p-value is indicated; n = 203 measured ipsilateral cells; n = 197 measured contralateral cells.

Experimental Hypoxia In Vitro Leads to Increased GHSR1-Immunoreactivity in Cultured Cerebral Cortex Cells, Cerebellar Granule Neurons and Organotypic Cerebral Cortex Slices
Following the results from the experimental in vivo stroke model, we became interested in testing whether the stimulatory effect of hypoxia on the GHRS1 expression could be seen in neuronal cultures deprived of oxygen. As a proof-of-principle, we used an in vitro model of hypoxia consisting of neonatal dissociated rat cortical neurons that had been cultured for one week and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. Half of the cultures were supplemented with ghrelin during these 24 h, while the other half remained untreated as a control. All samples were fixed and processed for immunostaining with antibodies recognising calretinin and GHSR1. We tested the specificity of the GHSR1 antibody in two control staining set-ups: prior to staining of the cell cultures, we incubated the murine primary GHSR1 antibody with murine pituitary/hypothalamus tissue homogenate to opsonise the antibody with detergent-solubilised GHSR1 (Supplementary Figure S1A and Methods). For the second control staining procedure, we applied only the secondary antibody, without the primary GHSR1 antibody (Supplementary Figure S1B and Methods). The signal of calretinin immunofluorescence revealed arborised neurons, which were surrounded by nonneural (fibroblasts) and glial cells (Figure 2A, arrows). The GHSR1 fluorescence signal in calretinin-positive neurons was measured ( Figure 2B). Under normoxic conditions, the neuronal immunoreactivity for GHSR1 was weak at the plasma membrane, cytoplasm and nucleus ( Figure 2A). Hypoxia imposed on the cell cultures led to increased GHSR1immunoreactivity in calretinin-positive neurons (Figure 2A,B). The GHSR1 signal was increased in the neurites and nuclei of the neurons. Of note, in the nuclei as well as in the cytoplasm of hypoxic glia and fibroblasts the GHSR1 signal was also increased (Figure 2A, arrows, middle panel). When ghrelin was applied to hypoxic neurons, the fluorescence signals of GHSR1 in neurons were reduced to those measured under normoxia (Figure 2A,B). recovery period of 24 h under normal oxygen supply. Half of the cultures were supple-mented with ghrelin during these 24 h, while the other half remained untreated as a control. All samples were fixed and processed for immunostaining with antibodies recognising calretinin and GHSR1. We tested the specificity of the GHSR1 antibody in two control staining set-ups: prior to staining of the cell cultures, we incubated the murine primary GHSR1 antibody with murine pituitary/hypothalamus tissue homogenate to opsonise the antibody with detergent-solubilised GHSR1 (Supplementary Figure S1A and Methods). For the second control staining procedure, we applied only the secondary antibody, without the primary GHSR1 antibody (Supplementary Figure S1B and Methods). The signal of calretinin immunofluorescence revealed arborised neurons, which were surrounded by nonneural (fibroblasts) and glial cells (Figure 2A, arrows). The GHSR1 fluorescence signal in calretinin-positive neurons was measured ( Figure 2B). Under normoxic conditions, the neuronal immunoreactivity for GHSR1 was weak at the plasma membrane, cytoplasm and nucleus (Figure 2A). Hypoxia imposed on the cell cultures led to increased GHSR1immunoreactivity in calretinin-positive neurons (Figures 2A,B). The GHSR1 signal was increased in the neurites and nuclei of the neurons. Of note, in the nuclei as well as in the cytoplasm of hypoxic glia and fibroblasts the GHSR1 signal was also increased ( Figure  2A, arrows, middle panel). When ghrelin was applied to hypoxic neurons, the fluorescence signals of GHSR1 in neurons were reduced to those measured under normoxia (Figure 2A  (A) Immunostaining with antibodies recognising calretinin (green) and GHSR1 (red) on neonatal dissociated rat cortical neurons that had been cultured for one week and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. DAPI (blue) was used to stain the nuclei. Calretinin immunofluorescence revealed the arborisation pattern of the cerebral cortex neurons, which were surrounded by nonneural and glial cells (arrows). Under normoxic conditions, neurons expressed GHSR1 at the plasma membrane, cytoplasm and nucleus (inlets). Hypoxia imposed on the neurons led to increased expression levels of GHSR1 (A,B). Note that in the nuclei as well as in the cytoplasm of hypoxic glia and fibroblasts the GHSR1 signal was also increased. Application of ghrelin to hypoxic cultures reduced the expression levels of GHSR1 in neurons to normoxic levels (A,B). Scale bar, 30 µm. Data are presented as medians ± SD analysed by Kruskal-Wallis' Multiple Comparison Test. p-values are indicated; n = 97 for normoxia; n = 119 for hypoxia and n = 70 for hypoxia + ghrelin. See also Supplementary Figure S1.
We next asked whether these effects hold true for other types of neurons and studied, therefore, dissociated rat cerebellar granule neurons that had been cultured for 24 h and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. Similar to the experimental design with cortex neurons mentioned above, half of the cultures of cerebellar granule neurons were supplemented with ghrelin during the recovery phase, while the other half remained untreated as a control. All samples were fixed and immunostained for calretinin and GHSR1 ( Figure 3A). The calretinin immunostained cerebellar granule neurons appeared unipolar, often growing in clusters ( Figure 3A). In normoxic cerebellar granule neurons, the expression of GHSR1 was weak and predominantly restricted to the nucleus ( Figure 3A, inlets). Hypoxia led to increased GHSR1-immunoreactivity in the nucleus ( Figure 3A,B). When ghrelin was applied to recovering hypoxic cerebellar granule neurons, the GHSR1 signals were reduced to those measured under normoxia ( Figure 3A,B).
ditions, neurons expressed GHSR1 at the plasma membrane, cytoplasm and nucleus (inlets). Hypoxia imposed on the neurons led to increased expression levels of GHSR1 (A,B). Note that in the nuclei as well as in the cytoplasm of hypoxic glia and fibroblasts the GHSR1 signal was also increased. Application of ghrelin to hypoxic cultures reduced the expression levels of GHSR1 in neurons to normoxic levels (A,B). Scale bar, 30 µm. Data are presented as medians ± SD analysed by Kruskal-Wallis' Multiple Comparison Test. p-values are indicated; n = 97 for normoxia; n = 119 for hypoxia and n = 70 for hypoxia + ghrelin. See also Supplementary Figure S1. We next asked whether these effects hold true for other types of neurons and studied, therefore, dissociated rat cerebellar granule neurons that had been cultured for 24 h and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. Similar to the experimental design with cortex neurons mentioned above, half of the cultures of cerebellar granule neurons were supplemented with ghrelin during the recovery phase, while the other half remained untreated as a control. All samples were fixed and immunostained for calretinin and GHSR1 ( Figure 3A). The calretinin immunostained cerebellar granule neurons appeared unipolar, often growing in clusters ( Figure 3A). In normoxic cerebellar granule neurons, the expression of GHSR1 was weak and predominantly restricted to the nucleus ( Figure 3A, inlets). Hypoxia led to increased GHSR1-immunoreactivity in the nucleus (Figures 3A,B). When ghrelin was applied to recovering hypoxic cerebellar granule neurons, the GHSR1 signals were reduced to those measured under normoxia ( Figures 3A,B). Immunostaining for calretinin (green) and GHSR1 (red) on cerebellar granule neurons that had been cultured for 24 h and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. DAPI (blue) was used to stain the nuclei. Cerebellar granule neurons were unipolar, partially growing in clusters. In normoxic cerebellar Immunostaining for calretinin (green) and GHSR1 (red) on cerebellar granule neurons that had been cultured for 24 h and then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. DAPI (blue) was used to stain the nuclei. Cerebellar granule neurons were unipolar, partially growing in clusters. In normoxic cerebellar granule neurons, the expression of GHSR1 was predominantly restricted to the nucleus (inlets). Hypoxia led to increased GHSR1-immunoreactivity in the nucleus (A,B). When ghrelin was applied to recovering hypoxic cerebellar granule neurons, the GHSR1 signals were similar to the normoxic levels (A,B). Scale bar, 30 µm. Data are presented as medians ± SD analysed by Kruskal-Wallis' Multiple Comparison Test. p-values are indicated; n = 99 for normoxia; n = 151 for hypoxia and n = 83 for hypoxia + ghrelin.
In parallel, we prepared acute organotypic slice cultures from brain hemispheres of 4-day-old Wistar rats and exposed the slices to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. One half of the cerebral cortex slices were supplemented with ghrelin during the recovery phase, while the other half remained untreated as a control. After fixation, the samples were immunostained for calretinin and GHSR1 ( Figure 4A). In slices, the calretinin immunostained cerebral cortex neurons appeared multipolar, surrounded by a huge population of calretinin-negative cells expressing GHSR1 ( Figure 4A). Under normoxia, the GHSR1 signal was homogeneously distributed across the sliced tissue ( Figure 4A). Hypoxia imposed on the slices led to increased GHSR1-immunoreactivity ( Figure 4A,B). When ghrelin was applied to recovering hypoxic slices, the GHSR1 signals were slightly reduced (in neurons almost to those seen under normoxia) ( Figure 4A,B). However, many calretinin-negative cells were still displaying strong GHSR1 signals.
In parallel, we prepared acute organotypic slice cultures from brain hemispheres of 4-day-old Wistar rats and exposed the slices to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. One half of the cerebral cortex slices were supplemented with ghrelin during the recovery phase, while the other half remained untreated as a control. After fixation, the samples were immunostained for calretinin and GHSR1 ( Figure 4A). In slices, the calretinin immunostained cerebral cortex neurons appeared multipolar, surrounded by a huge population of calretinin-negative cells expressing GHSR1 ( Figure 4A). Under normoxia, the GHSR1 signal was homogeneously distributed across the sliced tissue ( Figure 4A). Hypoxia imposed on the slices led to increased GHSR1-immunoreactivity ( Figures 4A,B). When ghrelin was applied to recovering hypoxic slices, the GHSR1 signals were slightly reduced (in neurons almost to those seen under normoxia) ( Figures 4A,B). However, many calretinin-negative cells were still displaying strong GHSR1 signals. (green) and GHSR1 (red) immunostaining of cerebral cortex slices. Slices were obtained from 4-dayold rats and cultured for 24 h to be then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. DAPI (blue) was used to stain the nuclei. In slices, calretinin-positive cerebral cortex neurons appeared multipolar, surrounded by many calretininnegative cells expressing GHSR1. Normoxic expression of GHSR1 was homogeneously distributed across the sliced tissue. Hypoxia imposed on the slices led to increased GHSR1-immunoreactivity (A,B). When ghrelin was applied to recovering hypoxic slices, the GHSR1 signals were slightly reduced (in neurons almost to normoxic levels). Many calretinin-negative cells were still strongly positive for GHSR1. Scale bar, 30 µm. Data are presented as medians ± SD analysed by Kruskal-Wallis' (green) and GHSR1 (red) immunostaining of cerebral cortex slices. Slices were obtained from 4-dayold rats and cultured for 24 h to be then exposed to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. DAPI (blue) was used to stain the nuclei. In slices, calretinin-positive cerebral cortex neurons appeared multipolar, surrounded by many calretininnegative cells expressing GHSR1. Normoxic expression of GHSR1 was homogeneously distributed across the sliced tissue. Hypoxia imposed on the slices led to increased GHSR1-immunoreactivity (A,B). When ghrelin was applied to recovering hypoxic slices, the GHSR1 signals were slightly reduced (in neurons almost to normoxic levels). Many calretinin-negative cells were still strongly positive for GHSR1. Scale bar, 30 µm. Data are presented as medians ± SD analysed by Kruskal-Wallis' Multiple Comparison Test; p-values are indicated; n = 562 for normoxia; n = 142 for hypoxia and n = 274 for hypoxia + ghrelin.
Furthermore, we prepared homogenates from the slices and subjected the homogenates to dot blot analysis using the GHSR1 antibody. Protein loading was controlled by immunodetection for β-actin and by staining with Ponceau S (Figures 5A and S2). The GHSR1 signal was normalised on the β-actin signal. The quantitative analysis of the dots confirmed that the expression of GHSR1 was increased in the homogenates of the slices which were subjected to hypoxia when compared to normoxic slices ( Figures 5A,B and S2). Ghrelin application to slices recovering from hypoxia decreased the levels of GHSR1 ( Figures 5A,B and S2).
immunodetection for -actin and by staining with Ponceau S (Figures 5A and S2). Th GHSR1 signal was normalised on the -actin signal. The quantitative analysis of the dot confirmed that the expression of GHSR1 was increased in the homogenates of the slice which were subjected to hypoxia when compared to normoxic slices ( Figures 5A,B an  S2). Ghrelin application to slices recovering from hypoxia decreased the levels of GHSR ( Figures 5A,B and S2). Considering the GHSR1 immunoreactivity found on the cell surface and within th cell interior, we asked whether the receptor is present in the cell nuclei and associates wit the chromatin. We first mapped the murine and human GHSR1 protein sequence for nu clear localisation signals in the cNLS Mapper [28]. The predicted NLS in the mous GHSR1 sequence ( Figure 6A) and that of the human ( Figure 6B) received the scores 3. and 4.2-4.7, respectively. Such scores suggest a potential cytoplasmatic and nuclear pres ence of GHSR1 [28]. Following these predictions, we performed immunogold transmis sion electron microscopy for GHSR1 on cerebral cortex cells under normoxia, hypoxia an hypoxia with ghrelin treatment ( Figure 6C). Under normoxic conditions, we found a few immunogold grains/precipitates within the cytoplasm and the nucleus of the cells ( Figur  6C, first panel), whereas, under hypoxia, the amount of the immunogold precipitates wa per se higher in both the cytoplasm and nucleus ( Figure 6C, second panel). In the hypoxi Considering the GHSR1 immunoreactivity found on the cell surface and within the cell interior, we asked whether the receptor is present in the cell nuclei and associates with the chromatin. We first mapped the murine and human GHSR1 protein sequence for nuclear localisation signals in the cNLS Mapper [28]. The predicted NLS in the mouse GHSR1 sequence ( Figure 6A) and that of the human ( Figure 6B) received the scores 3.2 and 4.2-4.7, respectively. Such scores suggest a potential cytoplasmatic and nuclear presence of GHSR1 [28]. Following these predictions, we performed immunogold transmission electron microscopy for GHSR1 on cerebral cortex cells under normoxia, hypoxia and hypoxia with ghrelin treatment ( Figure 6C). Under normoxic conditions, we found a few immunogold grains/precipitates within the cytoplasm and the nucleus of the cells ( Figure 6C, first panel), whereas, under hypoxia, the amount of the immunogold precipitates was per se higher in both the cytoplasm and nucleus ( Figure 6C, second panel). In the hypoxic cells that had received ghrelin, the amount of cytoplasmic and nuclear immunogold precipitates appeared reduced ( Figure 6C, third panel). Of note, in all three conditions, nuclear immunogold precipitates were found in the regions of the heterochromatin ( Figure 6C, red arrows). Interestingly, under hypoxia, a few gold grains were found also within the euchromatin. These combined findings confirmed the nuclear presence and the chromatin association of GHSR1. cells that had received ghrelin, the amount of cytoplasmic and nuclear immunogold precipitates appeared reduced ( Figure 6C, third panel). Of note, in all three conditions, nuclear immunogold precipitates were found in the regions of the heterochromatin ( Figure  6C, red arrows). Interestingly, under hypoxia, a few gold grains were found also within the euchromatin. These combined findings confirmed the nuclear presence and the chromatin association of GHSR1. Red arrows indicate nuclear GHSR1-immunogold precipitates (predominantly associated with the heterochromatin). Note the increased number of cytoplasmic and nuclear gold grains under the condition of hypoxia. Scale bar, 100 nm.

Experimental Hypoxia In Vitro Leads to Decreased Numbers of Pax6 and NeuN Expressing Cerebral Cortex Cells, but Increases the Number of Ki67-Positive Cells
Based on the findings obtained from the transmission electron microscopy, we hypothesised that nuclear GHSR1 affected the expression of neurogenic and mitotic factors such as Pax6 and Ki67. To test this hypothesis, we prepared neonatal dissociated rat cerebral cortex neurons that had been cultured for 3 weeks to a mature stage to then expose them to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. Half of the cultures were supplemented with ghrelin during 24 h of recovery, while the other half remained untreated as a control. We immunostained the cultures for Pax6 ( Figure 7A). Because of the relative difference in the cell density, we calculated the Pax6-positive cells as a percentage of all cells (counterstained with DAPI). In mature 3week-old cultures under normoxic conditions, 55.5 ± 3.0% (mean ± SEM) of the cells expressed Pax6 (Figures 7A,B). Their proportion was decreased under hypoxia to 38.9 ± 2.1% (mean ± SEM), whereas ghrelin treatment upregulated the expression levels of Pax6 (53.2 ± 1.4%, mean ± SEM) and brought them closely to levels seen under normoxia ( Figure 6B).

Experimental Hypoxia In Vitro Leads to Decreased Numbers of Pax6 and NeuN Expressing Cerebral Cortex Cells, but Increases the Number of Ki67-Positive Cells
Based on the findings obtained from the transmission electron microscopy, we hypothesised that nuclear GHSR1 affected the expression of neurogenic and mitotic factors such as Pax6 and Ki67. To test this hypothesis, we prepared neonatal dissociated rat cerebral cortex neurons that had been cultured for 3 weeks to a mature stage to then expose them to severe hypoxia for 6 h, followed by a recovery period of 24 h under normal oxygen supply. Half of the cultures were supplemented with ghrelin during 24 h of recovery, while the other half remained untreated as a control. We immunostained the cultures for Pax6 ( Figure 7A). Because of the relative difference in the cell density, we calculated the Pax6-positive cells as a percentage of all cells (counterstained with DAPI). In mature 3-week-old cultures under normoxic conditions, 55.5 ± 3.0% (mean ± SEM) of the cells expressed Pax6 (Figure 7A,B). Their proportion was decreased under hypoxia to 38.9 ± 2.1% (mean ± SEM), whereas ghrelin treatment upregulated the expression levels of Pax6 (53.2 ± 1.4%, mean ± SEM) and brought them closely to levels seen under normoxia ( Figure 6B). Cells 2022, 11, x FOR PEER REVIEW 13 of 25 Additionally, staining of 3-week-old cerebral cortex cell cultures for Pax6 (produced in rabbit) and GFAP was performed ( Figure 8A). Under normoxic conditions, 25.6 ± 3.9% (mean ± SEM) of the cultured cells expressed both Pax6 and GFAP ( Figure 8A, upper panel). In the hypoxic group 14.7 ± 2.1% (mean ± SEM) of the cells were positive for Pax6 and GFAP ( Figure 8A, middle panel). In hypoxic cultures supplemented with ghrelin during the recovery period, 30.9 ± 6.0% (mean ± SEM) of all cells were double-stained for Pax6 and GFAP ( Figure 8A, lower panel). In a parallel set-up of the same conditions, we used another Pax6 antibody (produced in the mouse) in combination with calretinin ( Figure  8B). Normoxic neurons stained for calbindin were strongly immunoreactive for Pax6 (mouse), whereas hypoxic neurons displayed weak immunoreactivity signals ( Figure 8B, upper and middle panels). Ghrelin-treated hypoxic neurons showed strong Pax6-immunoreactivity signals ( Figure 8B, lower panels).  Additionally, staining of 3-week-old cerebral cortex cell cultures for Pax6 (produced in rabbit) and GFAP was performed ( Figure 8A). Under normoxic conditions, 25.6 ± 3.9% (mean ± SEM) of the cultured cells expressed both Pax6 and GFAP ( Figure 8A, upper panel). In the hypoxic group 14.7 ± 2.1% (mean ± SEM) of the cells were positive for Pax6 and GFAP ( Figure 8A, middle panel). In hypoxic cultures supplemented with ghrelin during the recovery period, 30.9 ± 6.0% (mean ± SEM) of all cells were double-stained for Pax6 and GFAP ( Figure 8A, lower panel). In a parallel set-up of the same conditions, we used another Pax6 antibody (produced in the mouse) in combination with calretinin ( Figure 8B). Normoxic neurons stained for calbindin were strongly immunoreactive for Pax6 (mouse), whereas hypoxic neurons displayed weak immunoreactivity signals ( Figure 8B, upper and middle panels). Ghrelin-treated hypoxic neurons showed strong Pax6-immunoreactivity signals ( Figure 8B, lower panels). We further immunostained the cultures for Ki67 (Figures 9A,B) to detect active cell proliferation. The pattern of staining varied significantly between the cells; in some of them, the staining signal was restricted to the nucleoli, whereas in the large nuclei the signal appeared dispersed throughout the entire nucleus ( Figures 9A,B). Moreover, some of the nuclei showed speckled patterns of staining (e.g., glial cells, Figure 9A). Proliferating neuronal precursors displayed Ki67-positive nuclei smaller than those of the proliferating glial cells, localised in the close proximity of the GFAP-positive glial cell nets. Under normal oxygen supply, the Ki67 index (percentage of neuronal precursors positive for the Ki67 antigen) was 9.7 ± 0.8% (mean ± SEM). Interestingly, in control cultures exposed to hypoxia for 6 h and followed by 24 h of re-oxygenation, the proportion of proliferating neuronal precursors was 59.8 ± 1.4% (mean ± SEM). In the experimental group supplemented with ghrelin during the recovery period, Ki67 was detected in 15.9 ± 1.6% (mean ± SEM) of the neuronal precursor nuclei (Figures 9A,B). We further immunostained the cultures for Ki67 ( Figure 9A,B) to detect active cell proliferation. The pattern of staining varied significantly between the cells; in some of them, the staining signal was restricted to the nucleoli, whereas in the large nuclei the signal appeared dispersed throughout the entire nucleus ( Figure 9A,B). Moreover, some of the nuclei showed speckled patterns of staining (e.g., glial cells, Figure 9A). Proliferating neuronal precursors displayed Ki67-positive nuclei smaller than those of the proliferating glial cells, localised in the close proximity of the GFAP-positive glial cell nets. Under normal oxygen supply, the Ki67 index (percentage of neuronal precursors positive for the Ki67 antigen) was 9.7 ± 0.8% (mean ± SEM). Interestingly, in control cultures exposed to hypoxia for 6 h and followed by 24 h of re-oxygenation, the proportion of proliferating neuronal precursors was 59.8 ± 1.4% (mean ± SEM). In the experimental group supplemented with ghrelin during the recovery period, Ki67 was detected in 15.9 ± 1.6% (mean ± SEM) of the neuronal precursor nuclei ( Figure 9A,B). Further immunostaining for visualisation of the marker for post-mitotic neurons, NeuN, revealed immunoreactivity not only in the nuclei but also in the cytoplasm of many neurons, including the initial part of the neurites ( Figure 10A). Some cells exhibited a strong immunoreactivity for the antigen ( Figure 10A). The comparison between the quantitative data pointed to a significant decrease in the ratio of the NeuN-positive neurons under hypoxia ( Figure 10B). Post-hypoxic supplementation of ghrelin increased the number of NeuN-positive neurons to normoxic values ( Figures 10A,B). In particular, we observed that hypoxia reduced significantly the ratio of NeuN-positive post-mitotic neurons (61.5 ± 3.7%, mean ± SEM) compared with the ratio which was assessed under normoxia (77.1 ± 2.8%, mean ± SEM). Ghrelin treatment during the post-hypoxic period elevated the fraction of neurons positive for NeuN up to numbers observed under normoxia (85.1 ± 3.2%, mean ± SEM).  Further immunostaining for visualisation of the marker for post-mitotic neurons, NeuN, revealed immunoreactivity not only in the nuclei but also in the cytoplasm of many neurons, including the initial part of the neurites ( Figure 10A). Some cells exhibited a strong immunoreactivity for the antigen ( Figure 10A). The comparison between the quantitative data pointed to a significant decrease in the ratio of the NeuN-positive neurons under hypoxia ( Figure 10B). Post-hypoxic supplementation of ghrelin increased the number of NeuN-positive neurons to normoxic values ( Figure 10A,B). In particular, we observed that hypoxia reduced significantly the ratio of NeuN-positive post-mitotic neurons (61.5 ± 3.7%, mean ± SEM) compared with the ratio which was assessed under normoxia (77.1 ± 2.8%, mean ± SEM). Ghrelin treatment during the post-hypoxic period elevated the fraction of neurons positive for NeuN up to numbers observed under normoxia (85.1 ± 3.2%, mean ± SEM).

Discussion
The main findings of the present study are: (i) The expression of ghrelin's receptor GHSR1 is enhanced after hypoxia-induced in vivo and In vitro; (ii) Ghrelin treatment of hypoxic cultures counteracts the increase of GHSR1; (iii) hypoxia reduces the Pax6 levels In vitro, simultaneously increasing neural progenitor proliferation.
In some areas of the nervous system, such as the hypothalamus and the spinal cord, ghrelin has been shown to play an important role in (adult) neurogenesis [29,30]. Interestingly, ghrelin is the only known endogenous ligand activating the GHSR1 [31,32]. In contrast to ghrelin, GHSR1 is highly expressed in the nervous system [33][34][35][36][37][38]. Does the expression level of GHSR1 change upon injury or stress imposed on the nervous system? Cabral et al. [39] have suggested that neuronal remodelling is GHSR1-mediated and depends on the energy balance. Moreover, experimental evidence from the past two decades has revealed that ghrelin mediates neuroregeneration upon injury (for a detailed review see [40]). Therefore, we became interested in testing whether neurons subjected to metabolic stress may alter their GHSR1 expression. We demonstrated that hypoxia imposed on the cell cultures led to increased GHSR1-immunoreactivity in the neurites and nuclei of the neurons, as well as in the cytoplasm and nuclei of hypoxic glia and fibroblasts. On the contrary, previous studies have reported downregulation of GHSR1 expression after ischemia [18,41], and controversial effects of ghrelin on GHSR1′s expression-while the research group of Miao [18] reported that postischemic treatment of rats with ghrelin led to an elevation of the receptor levels, Huang et al. [41] did not observe any change. These discrepancies could be due to the methods of sampling or/and the way of exposure of the

Discussion
The main findings of the present study are: (i) The expression of ghrelin's receptor GHSR1 is enhanced after hypoxia-induced in vivo and In vitro; (ii) Ghrelin treatment of hypoxic cultures counteracts the increase of GHSR1; (iii) hypoxia reduces the Pax6 levels In vitro, simultaneously increasing neural progenitor proliferation.
In some areas of the nervous system, such as the hypothalamus and the spinal cord, ghrelin has been shown to play an important role in (adult) neurogenesis [29,30]. Interestingly, ghrelin is the only known endogenous ligand activating the GHSR1 [31,32]. In contrast to ghrelin, GHSR1 is highly expressed in the nervous system [33][34][35][36][37][38]. Does the expression level of GHSR1 change upon injury or stress imposed on the nervous system? Cabral et al. [39] have suggested that neuronal remodelling is GHSR1-mediated and depends on the energy balance. Moreover, experimental evidence from the past two decades has revealed that ghrelin mediates neuroregeneration upon injury (for a detailed review see [40]). Therefore, we became interested in testing whether neurons subjected to metabolic stress may alter their GHSR1 expression. We demonstrated that hypoxia imposed on the cell cultures led to increased GHSR1-immunoreactivity in the neurites and nuclei of the neurons, as well as in the cytoplasm and nuclei of hypoxic glia and fibroblasts. On the contrary, previous studies have reported downregulation of GHSR1 expression after ischemia [18,41], and controversial effects of ghrelin on GHSR1 s expression-while the research group of Miao [18] reported that postischemic treatment of rats with ghrelin led to an elevation of the receptor levels, Huang et al. [41] did not observe any change. These discrepancies could be due to the methods of sampling or/and the way of exposure of the cultures or model systems to ischemia/hypoxia, or/and the concentration of ghrelin utilised for administration.
The methods of hypoxia inducement and ghrelin application in vivo reported in the literature vary significantly and lead to controversial experimental outcomes. The concentration of 0.5 µM used in our experiments complies with that reported by us [27,42] and other groups [24,26,[43][44][45], yet differs from the concentrations utilised by Miao et al. [18] and Huang et al. [41]. Nevertheless, we abode by our previously reported concentrations because of the probability that the acute slices might require more ghrelin to recover after hypoxia: i.e., being placed on membranes, they are nourished through diffusion from the medium underneath, where ghrelin was applied. For the sake of consistency, we used 0.5 µM of ghrelin in all experiments.
Ghrelin has been shown to exert neuroprotection via attenuation of oxidative stress [41], blockade of apoptosis [18], and stimulation of synaptic plasticity [46]. Suda et al. [47] have suggested that defective GHSR1 activity of dopaminergic neurons causes marked motor impairment. Remarkably, mice deficient in GHSR1 did not display any signs of neurodegeneration [48]. We interpret the upregulation of GHSR1 upon hypoxia to be a response to hypoxia/injury/stress.
The plasma levels of ghrelin are relatively low [49], therefore, the reaction of neurons upon hypoxia with increased levels of GHSR1 could have an amplifier-like effect for low concentrations of ghrelin (see also [40]). To the best of our knowledge, the finding that hypoxia upregulates the expression levels of GHSR1 in neurons is novel. So is also the finding that GHSR1 is present in/translocates to the nucleus and associates with the chromatin. The chromatin association of GHSR1 might affect the expression of transcription factors.
Pax6 is a crucial factor determining the fate of neuronal progenitors in vitro [14] as well as in vivo, in adult brains [50]. However, the capacity of Pax6 to efficiently reprogram nonneuronal cells into neurons, observed in vitro, is limited in situ in the adult brain due to the natural postnatal glycogenic environment [51][52][53]. Under pathological conditions, such as hypoxia, Pax6 expression is upregulated in the neurogenic niches but not in the cortex, as shown in rats enduring prenatal hypoxia [2]. Thus, the reduced expression of Pax6 in dissociated mature cortical cells in vitro under hypoxic conditions (38.9 ± 2.1%) observed in our study is in line with the published phenomenon in vivo. On the other hand, post-hypoxic exposure to ghrelin for 24 h significantly elevated the percentage of cells expressing Pax6 (53.2 ± 1.4%), bringing it to the level at the pre-hypoxic stage (55.5 ± 3.0%). To our knowledge, this is the first study demonstrating such a modulatory effect of ghrelin on Pax6 expression in mature cortical networks. Of note, Fong et al. [54] have shown that proliferation and maturation of neuroblast cell lines can be stimulated via neurotrophins and neuroprotective factors derived from cocultured C17.2 neural precursor cells. In our study, the marker of proliferating precursor cells Pax6 was found in GFAP and calretinin-immunoreactive cells. The number of Pax6-positive cells significantly exceeded the number of Pax6/GFAP-double-labelled cells, which indirectly indicates that not only astrocytes express the transcription factor but other cell types are positive as well. Indeed, double labelling for Pax6 and calretinin confirmed that neurons in mature cortical cultures express Pax6. These results are in consent with the findings of Nacher et al. [55] that Pax6-expressing cells are also present in the adult hippocampal dentate gyrus and in the subventricular zone/rostral migratory stream, and they are proliferating precursors as well as nonproliferating resting progenitor cells or granule neurons at a very early developmental stage. Strikingly, hypoxia reduced the number of Pax6-positive cells (glial cells and neurons/neuronal precursors), whereas ghrelin treatment restored their levels. These observations raised two questions, which need further attention: does ghrelin act as a trophic factor; is the expression of Pax6 temporally and spatially regulated by ghrelin?
During development, for instance, Pax6 expression is graded in time and space: the highest level of expression in the cortical neuroepithelial cells is at the onset of neurogenesis and lack of it in the basal progenitors of the subventricular zone and in the post-mitotic neurons [56]. The same grading also refers to the Pax6-mediated control of cell cycle's duration in cortical progenitors. Primarily, Pax6 has a repressive effect on the cell cycle progression [5]. Lack of Pax6 at early stages of mice development (embryonic day 12.5) results in shortening of the progenitor cell cycle and a temporary increased production of post-mitotic neurons [57,58], whereas at later stages (embryonic day 15.5), Pax6-deficient progenitors proliferate more slowly [59]. Thus, even in the same cell type, Pax6 can obviously do both: the factor promotes and inhibits the proliferation of cerebral progenitors [17,58,60]. In the adult pancreas, Pax6-deficient beta-and alpha-cells have been shown to lose their maturation characteristics and convert into ghrelin positive cells [61]. These observations suggest that the effects of Pax6 on the cell cycle (e.g., of the cortical progenitors) are contextdependent as supported by the concept that the transcription factor affects proliferation in opposite manners via its DNA-binding subdomains [62].
The Pax6 protein comprises two distinct DNA-binding domains, one of which is regulated by alternate splicing and is able to interact with a number of co-factors [5]. Recently, Brg1/Brm associated factors complex (BAF) and Meis2 have been identified as new co-factors modulating Pax6 s direct regulation of target gene expression during embryonic and adult neurogenesis [63][64][65]. Importantly, Pax6 is also expressed in glial progenitors in the adult brain [66,67]. We speculate that ghrelin functions as a co-factor modulating Pax6 expression, possibly affecting neural progenitor cell fate or cell cycle. However, to explore whether this modulatory effect of ghrelin is mediated directly via the transcription factor Pax6, or whether there is another yet unknown mechanism, additional experiments are needed.
The Ki67 antigen is expressed in the nuclei exclusively in proliferating cells [68,69]. Quiescent or resting cells in the G 0 phase of the cell division cycle do not express it, therefore Ki67 is considered as a marker for the so-called growth fraction of given cells, i.e., progenitor cells [70,71]. During the interphase, cells express Ki67 exclusively in the nucleus, where it is required for the normal distribution and nucleolar association of the heterochromatin [72]. During mitosis, Ki67 covers the surface of chromosomes and represents approximately one-third of their protein mass [73]. Ki67 prevents chromosomal aggregation [74] and asymmetric distribution in daughter cells [75]. The transcription factor increases during the S phase, with further escalation in the G 2 phase, and maximal intensity in the metaphase [76,77]. The half-life of Ki67 is estimated to be approximately 1 h [68]. Intriguingly, cells entering the G 0 phase when treated with growth factors can re-increase their Ki67, to re-enter the S phase [76]. During the early G 1 phase, Ki67 is multi-focally expressed throughout the nucleoplasm, while during the S and the G 2 phases, larger foci overlapping the nucleoli and the heterochromatin regions are formed [77]. After the disintegration of the nuclear envelope, part of the Ki67 protein could be observed dispersed in the cytoplasm [77]. These patterns, including the chromosomes covered by the Ki67 protein, were also observed in our experiments. Before the onset of hypoxia, we found that 9.7 ± 0.8% of the cultured neuronal progenitor cells expressed Ki67, and their proportion was surprisingly increased up to 59.8 ± 1.4% after hypoxia. Ghrelin administration downregulated the percentage of Ki67-positive neural progenitors to 15.9 ± 1.6%. Our findings contradict the results from other studies showing that ghrelin administration in animal models of chronic neurodegenerative diseases, as well as endogenous ghrelin in other in vivo and in vitro experiments, triggers adult neurogenesis: ghrelin has been shown to stimulate proliferation of progenitor cells and increase the number of immature neurons in the hippocampus and the subventricular zone [78][79][80]. However, to the best of our knowledge, the current proof-of-principle study is the first one demonstrating that ghrelin attenuates the proliferation of progenitor cells in the cerebral cortex. Nevertheless, it is important to mention in this respect that Belaev and collaborators [81] have observed continuous proliferation of newly born neural precursors after occlusion of the middle cerebral artery of rats for 2 h. Treatment of the animals with docosahexaenoic acid (DHA) (member of the omega-3 polyunsaturated fatty acids family) 1 hour after the occlusion, followed by immunostaining two weeks post-injury, showed a neuroprotective and neurogenic effect, resulting in 88% increase of the BrdU/Ki67-positive and BrdU/NeuN-positive cortical cells in the penumbra. On the other hand, it seems that the Ki67 protein expression also depends on the degree of hypoxia, as established in experiments with glioma-derived neurospheres. Hypoxia limited to oxygen levels lower than 2% induces high proliferation rates [82], yet if the values are lowered to less than 1% oxygen, a decrease of the Ki67 antigen expression occurs [83].
Additionally, we have to carefully extrapolate the results from rodent models to non-human-and human primates due to the potential interspecies differences in adult neurogenesis. Studies on brains of adult macaque monkeys subjected to global cerebral ischemia have indicated that in both the striatum and neocortex putative newly generated neurons with long-term survival are slightly above 1% [84,85]. In adult human brains, hypoxia triggers some proliferation underneath the ependymal layer, but more than 30% of the Ki67 immunoreactivity is expressed in astrocytes, as double labelling for GFAP and Ki67 has indicated [86]. These GFAP-expressing neural stem cells can differentiate into astrocytes, oligodendrocytes and neurons in the presence [87,88] or absence of exogenous mitogens [89], thus contributing locally to the generation of new neurons as a reaction to ischemic injury. The rapid changes in the Ki67 protein expression, when comparing the normoxic neurons with those of the post-hypoxic experimental condition, can explain the significant fluctuations in the percentage of Ki67-positive cells in a relatively short time frame (24 h). Our findings are consistent with those obtained from human brains studies, where a strong upregulation of Ki67 expression has been observed in the subventricular zone of patients who died just a few days after ischemic stroke, but not in those who died six months after ischemic injury [86]. We interpret the increase in cell proliferation after ischemic injury as a transient event. Of note, Pax6 can be expressed in both neuronal and glial progenitors, and thus, the postischemic reduction of Pax6-positive cells observed in our study might include both cell types.
As Ortega et al. have reported [90], BrdU incorporation did not change after 6 h of hypoxia upon human cortical radial glial cells, but 24 h reduction of oxygen supply limited it, because hypoxia beyond 20 h of duration inflicts irreversible changes within the cells [91]. It appears that the observed effects of hypoxia vary in dependence on the experimental parameters such as location (i.e., cerebral cortex vs. subventricular zone), the severity of hypoxia, and duration of recovery.
Notably, the results from studies on NeuN in stroke models are quite controversial. Some of them report a substantial decrease of NeuN-positive cells in the infarction core and relate it to neuronal death [92], while others have related this decrease to depletion of NeuN's expression or loss of antigenicity [93]. In the present study, there was a significant change in the NeuN expression when comparing the three experimental groups, i.e., normoxia (77.1 ± 2.8%), hypoxia (61.5 ± 3.7%) and hypoxia + ghrelin (85.1 ± 3.2%). Of note, the nuclear NeuN staining was substantially reduced or even undetectable in some neurons. Mild cerebral ischemia has the same effect in in vivo experiments; however, it seems that the NeuN protein levels do not change within 24 h post-reperfusion, merely the NeuN antigenicity is reduced. Of note, the proportion of NeuN-positive neurons in the penumbra have been found to partially restore after antigen retrieval [93]. It is also possible that some special cell types or neurons with a distinct physiological/pathological status can be distinguished through the differences in NeuN/Rbfox3 expression [94]. In our experiments, ghrelin supplementation to recovering hypoxic cultures increased the proportion of neurons with nuclear expression of NeuN to pre-hypoxic values. Only more severe injuries, such as axotomy, have been shown to significantly abolish the NeuN protein levels, which then begin to restore within 7 days post-injury to reach the uninjured levels after 28 days [95]. Injured, but still viable neurons, may lose NeuN-protein expression due to downregulated protein synthesis and/or protein overconsumption [96]. Lower production of NeuN/Rbfox3 and the translocation of NeuN/Rbfox3 from the nucleus to the cytoplasm might lead to downregulation of the alternative splicing of the RNA of its target genes, and thus, change the complement of neuronal specific gene expression [97]. Differential splicing is an important mechanism for the evolutionary dynamics of the central nervous system's complexity [98] and its regular development. NeuN/Rbfox3regulated splicing is crucial for the final neuronal differentiation during development [99], but mutations in NeuN/Rbfox3 are also associated with neuropsychiatric disorders [100]. Thus, the up-regulating effect of ghrelin on the nuclear expression of NeuN observed in this study suggests that ghrelin can influence neuronal differentiation operating on its master switch [101].

Conclusions
In conclusion, our in vitro findings suggest that ghrelin is involved in postischemic neuroplasticity in non-neurogenic brain areas such as the cerebral cortex. Expression of the ghrelin receptor GHSR1 is increased in hypoxic neurons and normalised whenever ghrelin is supplemented to these neurons. Moreover, ghrelin regulates Pax6 and Ki67 expression, and stimulates cytoplasmic/nuclear shuttling of NeuN in hypoxic neurons. Since these factors are key players in proliferation, neurogenesis and differentiation, their stimulation could ameliorate recovery after hypoxia. This work emphasises the relevance of additional studies on the regulatory role of ghrelin in the cell cycle of cortical progenitors and neurons exposed to hypoxia, and corroborates the therapeutic potential of ghrelin for the treatment of stroke patients.

Data Availability Statement:
The datasets in support of the findings of this study are available from the corresponding authors, upon reasonable request.