Intracortical astrocyte subpopulations defined by astrocyte reporter Mice in the adult brain

2.1 Animals

The eaat2-tdT transgenic mice (C57Bl6/FVB mixed background) were generated as previously described (Yang et al., 2011). Bac aldh1l1-eGFP (Cahoy et al., 2008) transgenic mice (C57Bl6/FVB mixed background) were obtained from the GENSAT project through The Jackson Laboratory (stock#: 030247 Bar Harbor, ME, USA). Both male and female mice were randomized in all experiments. All mice were maintained on a 12 hr light/dark cycle with food and water ad libitum. Both male and female mice were used for all experiments. All procedures were in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Tufts University Institutional Animal Care and Use Committee.

2.2 Tissue processing, immunohistochemistry, and confocal imaging

Bac aldh1l1-eGFP or eaat2-tdT reporter mice (P60–70) were deeply anesthetized with ketamine (100 mg/kg) plus xylazine (10 mg/kg) in saline by i.p. injection and perfused intracardially with 4% paraformaldehyde (PFA) in PBS. The brains were dissected and kept in 4% PFA overnight at 4 °C, then cryoprotected by immersion in 30% sucrose for 48 hr. Brains were embedded and frozen in Tissu-Tek OCT Compound (Sakura). Coronal sections (20 μm) were prepared with a cryostat (model HM525, Leica) and mounted on glass SuperFrost+ Slides (Fisher Scientific) with ProLong Gold Antifade Mountant with DAPI (Thermo Fisher Scientific). For immunostaining, slides were rinsed three times in PBS, then treated with blocking buffer (1% BSA, 5% goat serum, and 0.2% Triton-X 100 in PBS) for 30 min at room temperature. Primary antibodies for Iba1 (Wako, 1:500) and APC (cc1 clone, Calbiochem, 1:50) were incubated overnight at 4 °C in blocking buffer. After washing slides three times in PBS, corresponding secondary antibody (1:2000) was added for 90 min at room temperature (RT). The sections were rinsed three times in PBS before mounting. For representative images of the large field cortex, automated stitching of fluorescence images was performed using the Keyence BZ-X700 Series Microscope at 20× magnification (numerical aperture, 0.7). Confocal images were taken using the Nikon A1R confocal laser scanning microscope (15 μm Z stack with 0.5 μm step) magnified with 40× (numerical aperture 0.8) or 60× (numerical aperture 1.0) objectives (Nikon Instruments Inc., Melville, NY). For images, we focused mostly somatosensory cortex.

2.3 Astroglial domain analysis

Reconstruction and measurement of cortical astroglial domain size were performed using Imaris (Bitplane, Zurich, Switzerland). All confocal images of tdT fluorescence from cortex of eaat2-tdT transgenic mice (P60) for Imaris analysis were taken with a 40× oil immersion objective lens. Images were taken under optimized settings to best illustrate the astroglial morphology. The 3D reconstruction images were first made using original confocal Z-stack images in Imaris software, as previously described (Morel, Higashimori, Tolman, & Yang, 2014). Briefly, the surface tool was used to build the astroglial domain. This function uses an automatic smoothing of the image with the Gaussian filter. The tdT fluorescence negative area in each of the confocal stack images was used as the internal control to determine the background fluorescence. The sensitivity threshold (absolute intensity) was manually adjusted so that the generated astroglial domain in the 3D image matches with that in the original confocal image. The cell somas were then detected based on size (≥12 μm in diameter) and used as seeding points to build the 3D domain. The quality (intensity) threshold was also manually adjusted to ensure that all cell somas were detected in a given image. The seeded watershed algorithm enables the Imaris software to recognize and split the domains of neighboring cells. The cells that were only partially included in confocal and 3D images were excluded from analysis. Overlapped domains were also carefully examined to ensure accuracy. The volume size of individual astroglia can be directly measured from generated 3D domains in Imaris. The expression levels of tdT have no effect on the labeling and quantification of astrocyte domain size, as previously shown (Morel et al., 2014) by dye fill of astrocytes.

2.4 Brain slice preparation

Cortical brain slices were prepared from young adult (P21–28) Bac aldh1l1-eGFP⁺ eaat2-tdT⁺ double transgenic mice. Animals were anesthetized with a ketamine/xylazine cocktail (110 mg/kg, 10 mg/kg); the cortex was quickly removed and 300 μm cortical slices were cut using a vibratome (Leica VT1200, Leica Microsystems) in ice-cold artificial cerebrospinal fluid (aCSF) (in mM): KCl 3, NaCl 125, MgCl₂ 1, NaHCO₃ 26, NaH₂PO₄ 1.25, glucose 10, CaCl₂ 2, and 400 μM l-ascorbic acid, with osmolarity at 300–305 mOsm equilibrated with 95% O₂–5% CO₂. Slices were incubated at RT until needed.

2.5 Electrophysiology and dye-fill of astrocytes

For astrocyte recordings, cortical astroglial cells were identified by eGFP or tdT florescence and unique electrophysiological properties (−75 mV to −85 mV resting membrane potential and no action potential firing with −120 mV to +100 mV ramp sweep). For astrocyte patch-clamp, we focused mostly somatosensory cortex. Astroglial somatic whole-cell recordings were obtained following the procedure described above at a holding potential of −75 mV. For the application of BaCl₂, BaCl₂ (100 μM) was bath applied to cortical slices. Resting membrane potentials (RMPs) of three different cortical astrocyte populations were measured and manually recorded prior to and after BaCl₂ application (5 min). I–V curves were generated by plotting capacitance-normalized current amplitudes as a function of applied voltage, from −100 mV to +250 mV, in 50 mV increments, as per voltage step protocol. For whole-cell recordings, input resistance and capacitance were measured in a voltage-clamp with 500 ms 5 mV step pulse from −75 mV holding potential. Cell capacitance was calculated by first obtaining the decay time constant of the transient current by the 5 mV step pulse and dividing this number by the series resistance. For the intracellular labeling of astrocytes, 5% Lucifer yellow (LY) potassium salt (Invitrogen) was added to the pipette solution. In voltage-clamp mode, the clamp potential was set at the resting membrane potential of astrocytes and the dye was injected into the cell by applying 0.5 s negative current pulse (1 Hz) until astroglial processes were completely filled. After the cells were filled, slices were immersed in 4% PFA overnight at 4C, rinsed in PBS, and mounted using Invitrogen ProLong Gold mounting medium.

2.6 Preparation of cortical cell suspension, fluorescence activated cell sorting of astrocytes, and RNA isolation

The total cortex of double Bac aldh1l1-eGFP⁺ eaat2-tdT⁺ transgenic mice (P60–90) was used for FAC sorting to increase yield. Animals were deeply anesthetized with Ketamine (100 mg/kg) + Xylazine (10 mg/kg) in saline by i.p. injection and perfused intracardially with Hanks Buffered Salt Solution (HBSS; Thermo Fisher Scientific). The total cortex was immediately dissected in cold Hanks buffer supplemented with glutamate receptor antagonists, 3 mM DNQX and 100 mM APV (Sigma, St. Louis, MO), and cut into small pieces. The cell suspension was prepared by following the manufacturer's instructions in the neural tissue dissociation kit (Miltenyi Biotech, Auburn, CA). Briefly, small pieces of tissue were treated with papain enzymatic mix (37 °C, 15 min) and then digested with DNase I (37 °C, 10 min), followed by careful trituration. Cell mixtures were then filtered through a cell strainer (40–70 mm) and resuspended in cold HBSS solution (5–10 × 10⁶ cells/ml) for FACS. One mouse was used for preparing each FAC sorted sample. Cells were sorted using MoFlo MLS high-speed cell sorter (Beckman Coulter) with Summit version 4.3 software. The whole procedure for cell suspension preparation and FAC sorting process was completed within 2–3 hr. FAC sorted cells were spun down and RNA was isolated from the cell pellet using the standard TRIzol reagent.

2.7 RNA-Seq analysis

The 100 bp paired-end sequencing for all prepared sequencing libraries was carried out at the Tufts Genomics Core Facility using an Illumina HiSeq 2500 sequencer. RNA-sequencing libraries were assigned randomly to sequencing lanes to avoid lane bias. FastQC program was used to examine the quality of generated reads. For mouse samples, we usually obtained 25–50 million reads for each library with a Phred quality score > 32. The total number of mapped reads for each region was at least 50 million, sufficient for calculation of reliable Fragments Per Kilobase of Mapped Reads (FPKM) values and subsequent differential gene expression analysis. The Tuxedo suite of tools (Tophat/Cufflinks/Cuffdiff) was used to analyze RNA-Seq reads (Trapnell et al., 2012). Reads were aligned to the reference mouse genome (Jul 2007 NCBI37/mm9) using the Tophat (v 2.0.8) software. Greater than 50% of both read pairs were mapped for all samples. Cufflinks (v 2.1.1) was used to calculate FPKM values with –G option for all mapped transcripts.

Cuffdiff2 was used to identify significant changes in mRNA expression among different samples using the biological replicates with a default q-value (FDR-adjusted p value) cut-off of 0.05 (q < 0.05). The heatmaps were generated from the Qlucore software based on the output lists from Cuffdiff2 with a more stringent q value (q < 0.001). For generating the pie chart, genes that were differentially expressed between tdT⁻eGFP⁺ versus tdT^higheGFP⁺ or between tdT⁻eGFP⁺ versus tdT^loweGFP⁺ astrocytes were overlapped first, then upload to Ingenuity pathway analysis (Qiagen) for annotating their functions. The output gene functional categories are summarized in the pie chart.

2.8 qRT-PCR

For qRT-PCR of mouse genes, RNA isolated from FAC sorted cells was converted to cDNA using a high archive cDNA synthesis kit (Applied Biosystems, Foster City, CA). The relative expression levels of selected genes were measured using SYBR Green (Life Technologies, Carlsbad, CA) reagent and specific primers for analyzed genes were chosen from PrimerBank (https://pga.mgh.harvard.edu/primerbank/).

2.9 Experimental design and statistical analysis

Sample size and statistical approach used for each experiment are described in each method section and in figure legends. Briefly, for FACS and RNA-seq, two replicates were included from separate mice; for astrocyte recording, 16–20 astrocytes from 4 to 6 mice/group were used. The G*Power (version 3.1) was used to estimate the reasonable sample size. All statistical analyses were performed using GraphPad Prism6. All values were plotted as the mean ± SEM. For multiple groups (> 2), one-way ANOVA was used to analyze the variance, followed by a Tukey post hoc test to compare multiple groups. For two-group comparison, Student's t-test was used. Statistical significance was tested at a 95% (p < .05) confidence level and was denoted with an asterisk (*p < .05; **p < .01; ***p < .001).

3.1 Identification of astroglial subpopulations in the adult cortex

The majority of cortical astrocytes are produced in an unusual way by local proliferation from newly born and differentiating protoplasmic astrocytes during postnatal 2–3 weeks (Ge et al., 2012), although radial glia from the ventricular or subventricular zone are the initial (astro)glial progenitors during (astro)gliogenesis (Freeman, 2010). Previous observations have found that conventional GFAP immunostaining only labels a limited number of astrocytes (especially in layer I) in physiological conditions (Cahoy et al., 2008) even though astrocytes are abundantly present in the adult cortex, leading to speculation that there are astroglial subpopulations in the cortex. The laminal organization of the cerebral cortex from layers I–VI also makes it interesting to explore whether there are layer-based astrocyte subpopulations in the cortex.

We previously generated astrocyte reporter eaat2-tdTomato (tdT) mice in which the tdT reporter is widely expressed in cortical astrocytes (Morel et al., 2014; Yang et al., 2011), facilitating the identification of differentially labeled cortical astrocytes when bred with other astroglial reporter mice, such as Bac aldh1l1-eGFP mice (Cahoy et al., 2008). Based on somatic eGFP and tdT reporter expression in adult (P60–90) eaat2-tdT⁺ Bac aldh1l1-eGFP⁺ double positive mice, we characterized three distinct expression patterns of tdT/eGFP reporters in cortical astrocytes: tdT⁻eGFP⁺ (red arrows, Figure 1a), tdT^loweGFP⁺ (yellow arrows, Figure 1a), and tdT^higheGFP⁺ (white arrows, Figure 1a). Interestingly, tdT⁻eGFP⁺ astrocytes were selectively localized only at cortical layers I–II, while both tdT^loweGFP⁺ and tdT^higheGFP⁺ astrocytes were distributed across layers III–VI. Layer-specific distribution of astrocytes in the cortex was also recently observed using the cell surface antigen CD51 (John Lin et al., 2017). In addition, tdT^higheGFP⁺ astrocytes (white arrows, Figure 1b) appeared to exhibit a more ramified astroglial domain with increased process arborization than tdT^loweGFP⁺ astrocytes (yellow arrows, Figure 1b). We have previously developed a method to quantitatively measure astrocyte domain size, by converting the astrocytes into 3D surface images using Imaris software (Morel et al., 2014). We also showed that the expression levels of the tdT reporter have no significant effect in labeling the full morphology of astrocytes, following the comparison of dye-filled and tdT labeled astrocyte morphology (Morel et al., 2014). Using this approach, the ramification differences of tdT^loweGFP⁺ and tdT^higheGFP⁺ astrocytes were evidently illustrated (Figure 1b). Subsequent quantification of tdT^loweGFP⁺ and tdT^higheGFP⁺ astrocyte domains showed that the average domain size of tdT^higheGFP⁺ astrocytes (19,326 ± 822, volume generally >10,000 μm³) is two times larger than that of tdT^loweGFP⁺ astrocytes (9,987 ± 234, volume generally <10,000 μm³) (Figure 1c). To reveal the full morphology and quantitatively determine the domain size of tdT⁻eGFP⁺ astrocytes, we prepared cortical slices and dye-filled layer I tdT⁻eGFP⁺ astrocytes based on eGFP⁺ fluorescence. Although tdT⁻eGFP⁺ astrocytes show no detectable tdT fluorescence, their domain sizes are in a similar range (average domain size of 15,289 ± 1,430) as that of tdT^higheGFP⁺ astrocytes. Our results are consistent with the idea that cortical astrocytes are morphologically diverse, and their morphology is not solely determined by their layer-specific localization (Lanjakornsiripan et al., 2018).

To quantitatively define cortical astrocyte subpopulations, we analyzed reporter expression and the relative distribution of these astrocyte subpopulations in the cortex of adult (P60–90) eaat2-tdT⁺ Bac aldh1l1-eGFP⁺ double transgenic mice using flow cytometry. The tdT⁻eGFP⁺ (in green, tdT A.U. < 25, Figure 2a), tdT^loweGFP⁺ (in magenta, 25 < tdT A.U. < 1,250, Figure 2a), and tdT^higheGFP⁺ (in red, tdT A.U. > 1,250, Figure 2a) cortical astrocytes were identified based on the drastic intensity differences of the tdT reporter. These subpopulations can be consistently gated based on the tdT intensity described above with minimal variability. These astrocyte subpopulations are distributed in an approximate 1:7:2 (tdT⁻eGFP⁺: tdT^loweGFP⁺: tdT^higheGFP⁺) ratio in the flow cytometry scatter plot (Figure 2a). We also analyzed a smaller number of cortical astrocytes to better illustrate the separated tdT^loweGFP⁺ and tdT^higheGFP⁺ subpopulations (Figure 2b). Meanwhile, the eGFP reporter fluorescence was generally constant with an intensity range of 100–400 A.U. (between black and magenta dashed lines, Figure 2a). We did observe a very small fraction of astrocytes with increased eGFP fluorescence (beyond the magenta dashed line, 400 < A.U. < 1,000), which were too few to be collected for analysis. It also appears that there might have a narrow band of cells with reduced eGFP levels (50 < A.U. < 100, between the blue and the black dashed lines, Figure 2a), however, these cells are intermingled with background debris fluorescence, making it less likely to be a distinct cell population. To confirm the differential expression levels of the tdT reporter in tdT^high and tdT^low astrocytes, we quantified the total tdT fluorescence intensity of typical tdT^high and tdT^low astrocytes (yellow and white arrows, Figure 1b) and indeed found drastically different total tdT intensity in tdT^high (mean intensity: 10.44 ± 0.49) and tdT^low astrocytes (mean intensity: 1.59 ± 0.28, Figure 2c). Importantly, the fold differences of the tdT fluorescence between these two populations of astrocytes quantified by either flow cytometry or Image J are comparable, further validating the intensity cut-off applied in the flow cytometry to define tdT^high and tdT^low astrocytes. In addition, only background auto-fluorescent signals from both tdT and eGFP channels (indicated by both dashed lines, generally with A.U. < 30) were observed in the cortex of wild-type control mice with the same gating cut-off, laser power, and equivalent number of cells analyzed (Figure 2d). It is intriguing that the expression of tdT is largely undetectable in the majority of layers I–II astrocytes as compared to cortical astrocytes in other layers, even though the tdT reporter is directed by the same human eaat2 promoter in all astrocytes. We then determined the GLT1 mRNA levels in tdT⁺eGFP⁺ and tdT⁻eGFP⁺ astrocytes isolated by FAC sorting from cortex of adult (P60–90) eaat2-tdT⁺ Bac aldh1l1-eGFP⁺ mice and found that GLT1 mRNA levels are drastically reduced (~5 times) in tdT⁻eGFP⁺ astrocytes as compared to tdT⁺eGFP⁺ (both tdT^high and tdT^low) astrocytes (Figure 2e). The significantly reduced GLT1 mRNA levels in tdT⁻eGFP⁺ astrocytes provide appealing evidence that the eaat2/glt1 promoter is less activated in these astrocytes than in tdT⁺eGFP⁺ astrocytes, suggesting that tdT⁻eGFP⁺ astrocytes represent a functionally distinct astrocyte subpopulation.

3.2 Whole-genome transcriptome analysis of cortical astroglial subpopulations

The observation that tdT⁻ eGFP⁺ astrocytes are mainly localized in layer I–II and tdT^higheGFP⁺ and tdT^low eGFP⁺ astrocytes are intermingled in other cortical layers makes it unfeasible to selectively isolate subpopulation-specific mRNAs using direct isolation approaches, such as translational ribosome affinity purification (TRAP) approach from these astrocytes. Although the cytoplasmic content can be extracted through the patching pipette for RNA-seq analysis after whole-cell patching using the “Patch-seq” technique (Cadwell et al., 2016), the much smaller astrocytic soma size (compared to neurons) makes it highly challenging to isolate the intracellular contents of astrocytic soma. A recent study physically dissected different cortical layers and then isolated layer-specific cortical astrocytes from Bac aldh1l1-eGFP mice (Lanjakornsiripan et al., 2018), which may introduce layer-specific bias due to the inconsistent separation of cortical layers. Alternatively, because these astrocyte subpopulations are differentially labeled with tdT or eGFP reporters, we isolated these subpopulations using fluorescence activated cell sorting (FACS) and then extracted total RNA for RNA-seq based transcriptome analysis. We first examined the specificity of RNA-seq datasets of astroglial subpopulations and found generally very low FPKM values (log₂ FPKM <0) of representative genes from neurons, oligodendrocytes, microglia, and NG2 glial cells in these subpopulations, while FPKM values of known astrocyte genes, slc1a2, aqp4, aldh1l1, and gfap were generally high (Figure 3a) in all astroglial subpopulations, confirming that the RNA-seq datasets were specific to the astroglial transcriptome. We did observe that the expression levels of several nonastrocytic genes, such as mobp, iba1, cd11b, and pdgfrα, are elevated in tdT⁻eGFP⁺ astrocytes, indicated by higher FPKM values (Figure 3a), compared to tdT⁺eGFP⁺ astrocytes. We and others have previously demonstrated that the eGFP expression driven by the aldh1l promoter is specific to cortical astrocytes with no (neurons, microglia, and NG2 cells) or minimal (< 4%, oligodendrocytes) expression in other CNS cells (Cahoy et al., 2008; Yang et al., 2011). Therefore, the elevated expression levels of typical mRNA transcripts from multiple CNS cell types (microglia, oligodendrocytes, and NG2 cells) in tdT⁻eGFP⁺ astrocytes are unlikely to be a result of contamination during FACS. We further performed immunostaining of Iba1 and APC that specifically labels microglia and oligodendrocytes, respectively. Consistent with our previous results (Yang et al., 2011), very rare (< 2%) APC and no Iba1 staining signals (Figure 3b) were overlapped with eGFP⁺ cortical astrocytes.

To identify unique mRNA expression profiles from different cortical astrocyte subpopulations, we performed multi-group cluster analysis with a stringent q value (FDR q < 0.001) and found that tdT⁻ eGFP⁺ astrocytes showed a drastically different mRNA expression profile than that of tdT^high eGFP⁺ or tdT^low eGFP⁺ astrocytes (Figure 3c), clearly suggesting that this cortical subpopulation is molecularly unique. The representative top differentially expressed genes from each subpopulation are shown in Table 1. We next performed pairwise cuffdiff analysis of RNA-seq datasets to compare the transcriptome differences of any given two astroglial subpopulations. Consistently, we found only a small number (59) of genes that showed differential FPKM values (FC > 2, FDR q < 0.05) between tdT^loweGFP⁺ and tdT^high eGFP⁺ astrocytes (Figure 3d, also in Supporting Information Table S1), among which only ttr and prm1 expression levels were dramatically different. In contrast, a large number of genes showed significantly different FPKM values (FC > 2, FDR q < 0.05) between tdT⁻eGFP⁺ and tdT^higheGFP⁺ (528 genes, Figure 3e, also in Supporting Information Table S2), and between tdT⁻eGFP⁺ and tdT^loweGFP⁺ astrocytes (480 genes, Figure 3f, also in Supporting Information Table S3), further suggesting that the molecular identity of tdT^loweGFP⁺ and tdT^higheGFP⁺ astrocytes is highly similar, but different from tdT⁻eGFP⁺ astrocytes. Interestingly, the vast majority (90%) of these genes show a reduced expression in either tdT^loweGFP⁺ or tdT^higheGFP⁺ astrocytes when compared to tdT⁻eGFP⁺ astrocytes. Particularly, expression levels of nonastrocyte glial marker genes, such as mog and mobp (oligodendrocytes), iba1 and cd11b (microglia), and pdgfrα (NG2 cells) in tdT⁻eGFP⁺ astrocytes are significantly higher than in tdT^loweGFP⁺ and tdT^higheGFP⁺ astrocytes (Figure 3a), suggesting a potential mixed glial lineage molecular signature in tdT⁻eGFP⁺ astrocytes.

To determine functional pathways that are potentially distinct between tdT⁻eGFP⁺ and tdT⁺eGFP⁺ astrocytes, differentially expressed genes between tdT⁻eGFP⁺ and tdT^higheGFP⁺ or between tdT⁻eGFP⁺ and tdT^loweGFP⁺ were first overlapped and underwent functional pathway analysis using Ingenuity Pathway Analysis program (Qiagen). These genes are mostly clustered as transporters, enzymes, transcriptional regulators, and kinases (Figure 3g, detailed lists are shown in Supporting Information Table S4). We further performed qRT-PCR using RNA samples prepared from FAC sorted astrocyte subpopulation samples to specifically validate the expression differences of genes identified between tdT⁻eGFP⁺ and tdT⁺eGFP⁺, as tdT^higheGFP⁺ and tdT^loweGFP⁺ astrocytes share highly similar expression profiles (Figure 3c,d). The qRT-PCR confirmed that enpp2 and ptgds mRNA levels, two of the highly expressed genes in tdT⁻eGFP⁺ astrocytes and the most differentially expressed genes between tdT⁻eGFP⁺ and tdT⁺eGFP⁺ astrocytes based on RNA-seq data, are indeed highly elevated in tdT⁻eGFP⁺ astrocytes (Figure 3h,i). In addition, we also found that mRNA levels of the kir4.1 gene, one of the major inward-rectifying potassium channels that determine astrocyte membrane potential (Olsen et al., 2007), were significantly lower in FAC sorted tdT⁻eGFP⁺ astrocytes than in tdT^loweGFP⁺ and tdT^high eGFP⁺ astrocytes (Figure 3j).

3.3 Physiological differences among cortical astroglial subpopulations

Encouraged by the molecular differences among cortical astroglial subpopulations, we next determined whether these astroglial subpopulations exhibit different physiological properties by performing astroglial recording on cortical slices (P21–28) of eaat2-tdT⁺ Bac aldh1l1-eGFP⁺ mice (Figure 4a). Interestingly, we found that the resting membrane potential (RMP) decreases gradually and significantly from tdT⁻eGFP⁺ (−73 ± 0.7 mV) to tdT^loweGFP⁺ (−78.3 ± 0.8 mV, p < .001), and to tdT^higheGFP⁺ astrocytes (−84.3 ± 0.9 mV, p < .001, Figure 4b). In addition, the membrane resistance of tdT⁻eGFP⁺ astrocytes (7.2 ± 0.6 MΩ) was significantly higher than that of tdT^low eGFP⁺ (5.0 ± 0.3 MΩ) and tdT^high eGFP⁺ astrocytes (4.3 ± 0.2 MΩ, Figure 4c). Because the resting membrane potential (RMP) of a mature astrocyte is primarily determined by the activity of inward-rectifying potassium channel 4.1 (Kir4.1; Olsen & Sontheimer, 2008), we examined the functional differences in Kir4.1 activity on these astrocytes. Interestingly, the application of BaCl₂ (100 μM), a well-established inhibitor of Kir4.1, on cortical slices resulted in only a moderate depolarization of tdT⁻eGFP⁺ astrocytes (11.2 ± 0.7 mV) (Figure 4d). By comparison, BaCl₂ application had a much stronger effect on both tdT^loweGFP⁺ and tdT^high eGFP⁺ astrocytes (17.3 ± 1.1 or 0.8 mV, respectively, p < .001, Figure 4d), indicating that tdT^loweGFP⁺ and tdT^high eGFP⁺ astrocytes express significantly larger inward Kir-mediated current, presumably as a result of higher Kir4.1 expression levels. Because the expression of Kir4.1 and K⁺ buffering is one of the central functions of mature astrocytes, the significantly smaller inward Kir-mediated current in tdT⁻ eGFP⁺ astrocytes suggests that these astrocytes are likely to have a reduced buffering capacity for extracellular K⁺ than tdT^low eGFP⁺ and tdT^high eGFP⁺ astrocytes. The reduced inward Kir-mediated current results are also consistent with our observation above that kir4.1 mRNA levels are significantly lower in tdT⁻eGFP⁺ astrocytes than in tdT^low eGFP⁺ and tdT^high eGFP⁺ astrocytes. Moreover, we generated the I–V plot using capacitance (size)-normalized recorded currents as a function of voltage in Figure 4f. Representative current recording tracers for tdT⁻eGFP⁺ and tdT^higheGFP⁺ astrocytes (tdT^loweGFP⁺ astrocytes are similar to tdT^higheGFP⁺ astrocytes) in response to stepwise increased voltage are shown in Figure 4e. Although current amplitude was slightly larger in tdT^high eGFP⁺ astrocytes than tdT⁻eGFP⁺ astrocytes, all three astroglial subpopulations demonstrated passive membrane currents typical of astrocytes with the reversal potential close to the K⁺ equilibrium potential (~ − 80 mV), confirming that they all carry typical astroglial membrane properties.