Wistar rat pups (RjHan:WI; Janvier Labs, Hannover, Germany) were used for cell cultures. Animals were kept according to the recommendations of the Federation of European Laboratory Associations (FELASA). Pups were decapitated, and brains were removed. In ice-cold HEPES buffer, the hemispheres were separated, and meninges, diencephalon, and mesencephalon were removed. Tissue was transferred to 10 mL Dulbecco’s modified Eagle medium (DMEM, Thermo Fisher Scientific, MA, USA) supplemented with 10% foetal calf serum (FCS; Thermo Fisher Scientific, MA, USA) and 50 U/mL penicillin and 50 µg/mL streptomycin (Thermo Fisher Scientific, MA, USA). Brain tissue was mechanically homogenized with 10 mL and 1 mL pipette tips and strained through a 70-µm cell strainer. After centrifugation, the supernatant was discarded, and cells resuspended in DMEM with 10% FCS. Cells were cultured in 75 cm2 cell culture flasks pre-coated with poly-L-ornithine (PLO; Sigma-Aldrich, MO, USA). Flasks were kept at 37 °C and 5% CO2. After the preparation, flasks were not moved for 4 days to assure cell adherence to the flask surface. Afterwards, medium was changed twice a week and cells were passaged 2 times (1:2–1:3) until the final passage during which cells were seeded for experiments to the corresponding PLO-coated plates. Before the first passage, non-adherent cell types were eliminated by placing the cell culture flasks in an orbital shaker for 2 h at 37 °C, aspirating the medium and adding new medium to the remaining adherent cells. Astrocyte purity was determined through regular extractions and counting of GFAP-positive cells after immunofluorescence staining to assure an astrocyte purity of 95% or higher.

Experimental treatment started 1 day after seeding the cells in standard DMEM (10% FCS). A semi-starvation protocol was previously established to provoke cell stress with a limited amount of cell death (Kogel et al. 2021). In the semi-starvation condition, cells were cultured in medium with 2 mM glucose, while the control condition was treated with a standard amount of 25 mM glucose. In both conditions, FCS was reduced to 0.5% to induce cell cycle synchronization. For the experimental medium, glucose was added to a glucose-free DMEM (Thermo Fisher Scientific, MA, USA) and then sterile filtered. Cells were kept in starvation medium for 15 days. Medium change and sample collection were performed every 3 days (Fig. 2). To test for reversibility of previously observed effects with 2 mM glucose, a recovery period of 6 days was added to the present study. During the recovery period, starting at day 15, both groups were supplied with 25 mM glucose medium.

Experimental study set-up. The investigated cells were primary rat cortical astrocytes. After harvesting, cells were proliferated for several weeks. Cells were seeded for experiments one day before experimental begin. The experimental protocol had a duration of 21 days with medium changed and samples taken every 3 days. In the semi-starvation period from day 0 to day 15, cells in the starvation condition were semi-starved with 2 mM glucose. In the recovery period from day 15 to day 21, cells from the starvation period were cultured with 25 mM glucose. The control condition was continuously cultured with 25 mM glucose. All conditions were cultured with 0.5% FCS. FCS, fetal calf serum; d, day

Cells were seeded with eight technical replicates on PLO-coated black 96-well plates with 15,000 cells per well. CellTiter-Blue® Cell Viability (CTB) Assay (Promega, WI, USA, G8081) was used to determine the metabolic activity of the cells. Two biological replicates were employed for this study. According to the manufacturer’s instructions, at the designated sample collection time points (Fig. 2), CellTiter-Blue® reagent was added to the cells and incubated for 2 h at 37 °C. The assay is based on the conversion of the redox dye resazurin into resorufin, a fluorescent end-product. The resulting fluorescence intensity (excitation: 560 nm, emission: 590 nm) was recorded with a Tecan infinite M200 plate reader and processed with the i-control 3.4.2.0 software (Tecan, Männedorf, Switzerland). The CytoTox 96® Non-Radioactive Cytotoxicity (LDH) Assay (Promega, WI, USA, G1782) was used to determine cell cytotoxicity. According to the manufacturer’s instructions, at the designated sample collection time points, CytoTox 96® working solution was added to wells and incubated for 30 min at room temperature. The assay is based on the quantification of extracellular LDH through the directly proportional formation of a dye. Absorbance was recorded at 490 nm with a Tecan infinite M200 plate reader and processed with the i-control 3.4.2.0 software (Tecan, Männedorf, Switzerland).

Cells were seeded on PLO-coated glass coverslips in 24-well plates with 50,000 cells per well. Cells were fixated with 3.7% paraformaldehyde (Carl Roth Karlsruhe, Germany), and cell membranes were permeabilized with 0.2% Triton X-100 (Carl Roth Karlsruhe, Germany). Unspecific binding was blocked with blocking buffer (1% BSA and 2% FCS in PBS) for 1 h. Cells were incubated with respective antibodies in blocking buffer overnight (Table 1). Then, cells were incubated with the respective secondary antibodies for 1 h and counterstained with Hoechst 33342 solution (1:10,000, Thermo Fisher Scientific, MA, USA) for 5 min. Coverslips were mounted with Epredia™ Shandon™ Immu-Mount™ (Thermo Fisher Scientific, MA, USA) onto glass slides. For picture acquisition, cells were imaged with a DMI6000 B fluorescence microscope (Leica Biosystems, Wetzlar, Germany).

For RNA isolation and gene expression studies, cells were seeded on PLO-coated 6-well plates with 400,000 cells per well. Four biological replicates were employed in this study. RNA was extracted by removing cell culture medium from wells and solving cells in 600 µL RNA-Solv® Reagent (Omega Bio-tek, Inc., GA, USA). The isolation is based on a phenol-chloroform phase separation, whereby the upper aqueous phase is used to extract total RNA content. RNA concentration and purity were measured with a NanoDrop 1000 spectrometer (Thermo Fisher Scientific, MA, USA). Reverse transcription to cDNA was performed with a M-MLV Reverse Transcriptase (Thermo Fisher Scientific, MA, USA) with random primer (Thermo Fisher Scientific, MA, USA) according to the manufacturer’s instructions. Semi-quantitative real-time PCRs (qRT-PCR) were performed with an AceQ SYBR Green qPCR Master Mix (Nanjing Vazyme Biotech Co, Nanjing, China) with the listed target primers (Table 2) and measured with a CFX Connect™ Real-Time system (BioRad, Feldkirchen, Germany). Data were analysed through the ΔΔCt-method. For each sample, values of the gene of interest were normalized to the relative quantity of the reference gene Cyclophilin A.

Cells were seeded on PLO-coated 10 cm dishes with two million cells per dish. Four biological replicates were employed in this study. After treatment, protein isolation was started by detaching cells with trypsin. Medium was added for neutralisation. Cell suspension was collected and centrifugated at 4 °C. The cell pellets were collected and resuspended in 50 µL RIPA buffer. Protein concentration was measured with the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA) which was used according to manufacturer’s instructions. Protein concentration was measured with a Tecan infinite M200 plate reader and processed with the i-control 3.4.2.0 software (Tecan, Männedorf, Switzerland). For the SDS-Page, 12% (v/v) SDS-gels were used. Samples were adjusted to 20 µg protein per 20 µL reaction volume with RIPA buffer and Laemmli buffer. The Spectra Multicolor Broad Range Protein Ladder (Thermo Fisher Scientific, MA, USA) was used for band size reference. For blotting in a semi-dry setup, gels were transferred to PVDF Western blotting membranes (Roche Holding AG, Basel, Switzerland). Membranes were blocked for 1 h in 5% milk or 5% BSA depending on the solvent of the primary antibody (see Table 3). Membranes were incubated overnight with the corresponding primary antibodies. Afterwards, they were incubated with a secondary antibody for 2 h. Membranes were developed with an ECL chemiluminescence kit (Thermo Fisher Scientific, MA, USA). For evaluation of the individual protein signal, a densitometric analysis of bands was performed and later normalized to the GAPDH reference bands using the ImageJ software (NIH, Bethesda, MD, USA). For full-length western blots, please refer to Supplementary Fig. 1–3.

Statistical analysis of experimental data was performed with GraphPad Prism 9.1.1 and IBM SPSS Statistics 22. Data are indicated as arithmetic mean±standard deviation unless stated otherwise. The statistical evaluation for semi-starvation and refeeding was divided; thus, two-way ANOVA was calculated on both conditions individually to only include two parameters in the comparison (time and treatment). Normal distribution was tested with the Shapiro-Wilk test. In case of non-normal distribution, a Box-Cox transformation with an adequate lambda was applied. When the data tested positive for normal distribution, statistical significance was tested with a two-way ANOVA test. A p ≤ 0.05 was considered as statistically significant. For western blot analysis, a normalization to the control group per each timepoint was performed. Therefore, a statistical evaluation was not performed.