All animals used for the experimentation were treated in accordance with the accepted standards of animal care and with the procedures approved by the local Committee of Research and Ethics of the Instituto de Fisiología Celular, Universidad Nacional Autónoma de Mexico (protocol number: JMA175-21). The protocol used followed the Guidelines for the Care and Use of Mammals in Neuroscience as well as guidelines released by the Mexican Institutes of Health Research and the National Institutes of Health guide for the care and use of Laboratory animals. All efforts were made to minimize animal suffering and to reduce the number of animals used.

Cerebellar granule neurons (CGN) cultures were prepared as previously described (Moran and Patel, 1989). CGN were obtained from 8 -day- old Wistar rats or from 4-day-olds in ICR mice cerebellum and plated at a 265 × 10³ cell/cm² density in plastic dishes coated previously with poly-l-lysine (5 μg/mL). The culture medium contained basal Eagle’s medium supplemented with 10% (v/v) heat-inactivated fetal calf serum, 2 mM glutamine, 25 mM KCl, 50 μg/mL streptomycin, and 50 U/mL penicillin. Cytosine arabinoside (10 𝜇M) was added 24 h after seeding to prevent the proliferation of nonneuronal cells. Cultures were maintained under these conditions for 8 days and kept at 37°C in an atmosphere of CO² (5%) and saturated air with water vapor (95%). At the end of the preparation, CGN cultures contained approximately 95% neurons. Apoptotic death of CGN was induced by two different stimuli: serum-free medium containing 5 mM KCl (K5) or by 0.5 𝜇M staurosporine (Sts) administration. In the CGN viability assays with short exposure to K5 and Sts we used K25 conditioned medium to prevent neuronal death by fresh medium, as has been observed.

A human MSN neuroblastoma cell line was donated by Dr. Clorinda Arias from the Institute of Biomedical Science of the National Autonomous University of Mexico. The cell line was cultured in accordance with Ferrera et al. (2017). Adherent MSN cells were seeded in a monolayer with RPMI 160 medium supplemented with 10% of fetal bovine serum (Gibco, Life Technologies Corporation, Grand Island, NY, United States), 1% of non-essential MEM, and 1% glutamine-serine-asparagine. The cells were maintained in humified incubation with 95% of air and 5% of CO² and 37°C. The cells were seeded in a density of 1 × 10⁶ cells /100 mm dish (Reynolds et al., 1986).

ROS levels were quantified by dihydroethidium. DHE is a fluorogenic probe for the detection of intracellular superoxide that is oxidized to ethidium and intercalates into nucleic acids emitting fluorescence at 610 nm (Cho and Hwang, 2011). DHE is irreversibly oxidized and therefore the signal obtained is the result of the sum of the previous accumulation. CGN were seeded in a multi-well plate and treated with 10 μM dihydroethidium (DHEt) for 30 min at 37°C. Cells were observed in an epifluorescence microscope with a rhodamine filter with a wavelength of 488–515 nm. DHEt diffuses into cells and is oxidized by ROS in the cytosol producing ethidium and 2- hydroxy ethidium that binds to the DNA and emits bright red fluorescence. Fluorescence images were acquired, and fluorescence intensity was measured with the Image J Program. The data were normalized with respect to K25 conditions.

Cell viability was measured by calcein-AM accumulation and propidium iodide (PI) exclusion to stain live and dead cells, respectively. Endogenous esterases in living cells transform calcein-AM to calcein, a green-fluorescent product. The PI crosses the plasma membrane of damaged cells; it binds to DNA and double-stranded RNA by intercalation and emits an intense red fluorescence signal. This dye shows weak fluorescence when it is not bound and in an aqueous solution. Cells were incubated with calcein-AM (1 𝜇M) and PI (40 𝜇M) for 30 min at 37°C and were observed in an epifluorescence microscope. Stained cells were counted with Image J software.

Cerebellar Granule Neurons cultured for 7 days were used after treatments with K5, staurosporine, LY4002902 (30–100 μΜ), or hydrogen peroxide 100 μM. After harvesting the cells, total RNA was purified using TRIzol reagent and 0.5 μΜ of total RNA was reverse transcribed to cDNA by using Reverse aid Reverse Transcriptase (Thermo Scientific) and Oligo DT. A measure of 50 ng cDNA was used for qPCR and the MasterMix qPCR ROx PyroTaq EvaGreen 5x. Primer sequences were used at 10 μM. TXNIP forward 5’ GGCTGGGGGTGTTGTTTA 3’ Reverse 5’ GCGCAGGAAAATAAGATGA GAPDH Forward 5’CTCATGAC CACAGTCCATGC 3’ Reverse 5’TTCAGCTCTGGGATGACCTT 3′. PCR assays were performed in three replicates in each group. All results were analyzed by the Delta Ct method.

Promoter analysis was performed by using the NCBI database for the Txnip gene sequence. Two motifs reported by the JASPAR database for FOXO1 and FOXO3 in mice were searched. We analyzed the promoter by using NCBI upstream sequence for the Txnip gene in mice (accession number NM_023719). First, we localized the E-box sequences in the promoter and determined the transcription start site (TSS) using JASPAR-reported FOXO motifs in mice¹ and we found two motifs close to the TSS -1,000 nt upstream. The first is located at -560 nt and is specific for FOXO1 and we named it motif 1; the second, at -260 nt is common for FOXO1 and FOXO3 and we named it motif 2. We then analyzed promoter sequences in humans and rats (NCBI accession number NM_006472 and NM_001008767) and we found that these motifs are conserved across these species.

We used 1 × 10⁷ neurons of 7 DIV for each experimental condition: K25, K5, for 30 and 60 min and 100 μM hydrogen peroxide for 1 h. Cells were crosslinked with 1.1% formaldehyde for 10 min. The formaldehyde linking was quenched by adding 125 mM glycine for 10 min. Cells were pelleted and rinsed once in cold phosphate-buffered saline (PBS). Nuclear pellets were obtained by using RIPA buffer (50 mM Tris HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate 0.1% SDS, and Complete protease inhibitor cocktail (Roche)) and sonicated with a sonicator (Branson Analog) set at 400 w power and ½ mm point out the power of 10% with 8 cycles of 15 s ON/30 s OFF. Sonicated chromatin was centrifugated at 13,000 rpm for 15 min. A binding buffer was used (10 mM Tris–HCl, pH 8.0, 300 mM NaCl, 5 mM EDTA, pH 8.0, 0.5% SDS). Chromatin was incubated overnight at 4°C with 5 μg/sample of FOXO 1 polyclonal anti-rabbit (ab39670) and FOXO 3 polyclonal anti-rabbit ab ab12162 or with TFII ab as a control. The antibody was incubated overnight at 4°C in rotation. Dynabeads (Thermo Scientific) was used to immunoprecipitate. The beads were washed three times with binding buffer, precipitated on the magnetic strip, and resuspended in the same volume with binding buffer. The beads were washed with the following buffers: low salt buffer (50 mM Tris–HCl, pH 8.1, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 0.1% sodium deoxycholate), high salt buffer (50 mM Tris–HCl, pH 8.1, 500 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% triton X-100, 0.1% sodium deoxycholate), LiCl salt buffer (50 mM Tris–HCl, pH 8.1, 1 mM EDTA, pH 8.0, 1% NP-40, 0.7% sodium deoxycholate, 500 mM LiCl₂) and TE buffer (10 mM Tris–HCl, pH 8.0, 1 mM EDTA, pH 8.0). All samples had16 μl of 5 M NaCl added. We incubated samples overnight at 65°C with shaking to remove protein-DNA interaction. DNA was purified with phenol: chloroform including the input samples. PCR was performed as described above. Motif 1 FOXO1 forward 5’ GGCTGGGGGTGTTGTTTA 3’ Reverse 5’GCGCAGGAAAATA AAGATGA 3’ Motif 2 FOXO’s forward 5’ TAAAACAAGGGCCA AGTAGCC 3’ Reverse 5’ ATATAGCCGCCTGGCTTG 3’.

Anti-FOXO3 cell signaling antibody (D19A7) and secondary antibody with Alexa Fluor 488 anti-rabbit cell signaling (4412S) were used. Cells were grown on glass coverslips in a 12-well plate over a 5 x PLL layer, maintained in K25 medium for 7DIV, and treated with the PI3K inhibitor LY294002 (100 μM for 4 h) were treated with K5 medium or Sts (0.5 μM) for 30, 60 or 120 min or treated with 100 μM hydrogen peroxide for 1 h. Subsequently, the cells were washed three times with PBS, fixed with 4% paraformaldehyde for 20 min at room temperature, washed three times with PBS, and blocked with 3% serum-0.2% triton-PBS for 2 h at room temperature. The primary antibody was incubated (1:500) in 3% goat serum PBS-T solution overnight. The secondary antibody coupled to Alexa 488 (1:1000) was incubated for 1 h at room temperature in the dark. Vechtashied mounting medium with DAPI was used. The images were captured in a Zeiss LSM800 confocal microscope and images were taken in Z Pearson’s colocalization coefficient was measured by using the JACoP plugin for Image J.

The MSN neuroblastoma cells were transfected before plating with shRNA TXNIP using the nucleofection electroporation technique with the Amaxa VP1003 kit using the C-13 program. GFP reporter-TXNIP shRNA and a scramble control sequence (Scr) were transfected. For every 4 million cells, 5 μg of the Maxiprep-purified plasmid was used. Cells were plated in 35 mm Costar dishes. The transfection efficiency was evaluated at 48 h with an epifluorescence microscope as the total number of GFP-positive cells relative to the total number of cells. The transfection efficiency was 28% for both scramble and shRNA transfected cells.

Neuroblastoma MSN cells were cultured and transfected as previously described and after 48 h cells were washed with cool PBS buffer and homogenized with RIPA lysis buffer (25 mM Trizma, 50 mM NaCl, 2% Igepal, 0.2% SDS, pH 7.4) with complete protease inhibitor (Roche), phosphatase inhibitor (Roche) and 50 mM NaF. Cells were sonicated for 30 s with 30% of sonication potency and centrifuged at 4,5000 rpm for 5 min. Protein content was determined by Lowry assay. Electrophoresis was performed on 10% SDS-PAGE with 20 μg per well. Gels were transferred onto PVDF membranes at 300 mA for 1 h. Membranes were blocked with TTBS buffer (100 mM Tri-HCl, 150 mM NaCl and 0.1% Tween, pH 7.4) with 5% fat-free milk for 1 h under shaking at room temperature. Cell signaling TXNIP rabbit antibody (D5F3E) (1:1000) and peroxidase-conjugated anti-rabbit (1:10,000) were used.

Data are presented as the mean ± the standard error of the mean. The Prism Graph 8.1 program was used to perform the statistical analysis of the experiments. The normal distribution of the data was determined with a Shapiro-Wilks test. Viability experiments are presented as a percentage normalized to K25. PCR data are normalized to the GAPDH levels and the K25 control relative to 1 and ChIP data were normalized to the background. Statistical significance was determined as p < 0.05. p values are represented as NEJM style *p < 0.033, **p < 0.002, ***p < 0.001. Data were analyzed by using the ANOVA test followed by Dunnett’s post hoc test. For multivariable data such as ChIP and TXNIP shRNA viability assay, a two-way ANOVA followed by Tukey’s test or Sidak’s test was used.

CGN was treated with low potassium (5 mM) or staurosporine (0.5 μM) during 6–24 h and then neuronal viability was evaluated with calcein and propidium iodide as indicated in Methods. In line with previous studies (Valencia and Morán, 2001; Guemez-Gamboa and Morán, 2009; Ramiro-Cortés and Morán, 2009; Zaragoza-Campillo and Morán, 2017; Cid-Castro and Morán, 2021) we found a decrease in cell viability of 20% after 6 h of treatment with low potassium (K5) and by 60% after 24 h. In the case of staurosporine, we observed about 60% of cell death after 24 h of treatment. Therefore, the temporary window in which death is executed is from 18 to 24 h after the treatments (Figure 1). To know the minimal time required to activate the death process, we exposed CGN to K5 for short periods of time and then evaluated cell death after 24 h. When we evaluated calcein fluorescence we found that 1–3 h of treatment induce about 40–50% cell death 24 h later. However, when cells were treated for 5 h the cell death increased to 75–80%, which was similar to the cell death observed for continuous treatment during 24 h (Figure 2). Similar results were observed with Sts treated cultures; however, unlike K5, due to the chemical nature of Sts, it could be more difficult to thoroughly wash off the drug and prevent it from remaining longer in the preparation. When cells were treated with 100 μM H₂0₂ for 1 h, cell death was about 90% 24 h later.

To evaluate ROS dynamics under cell death conditions, neuronal cultures were maintained in K25 medium for 7 DIV and then cells were switched to K5 medium or treated with staurosporine (0.5 μM) from 30 min to 5 h and ROS levels were measured with dihydroethidium (DHE) according to Material and methods. As previously reported (Valencia and Morán, 2001; Ramiro-Cortés and Morán, 2009; Zaragoza-Campillo and Morán, 2017), neurons treated with K5 showed a tendency to increase ROS levels from 30 min; however, we observed a statistically significant increase of 1,500% after 3 h that continued increasing after 5 h (Figure 3A). Similar results were observed for Sts. In this case, an increase of ROS production was also observed from 30 min, and after 3 h it was detected a significant increase of 900%, but the maximal ROS production was observed after 4 h (Figure 3B). We used 100 μM H₂0₂ for 1 h as a positive control and we observed that H₂0₂ increased the DHE signal by 2000% (Figure 3A). These data indicate that ROS are produced early, several hours (15 h) before neuronal death (Figure 3).

Our group previously found that TXNIP protein expression was induced by K5 and Sts (Zaragoza-Campillo and Morán, 2017). Furthermore, in that study, we demonstrated that TXNIP was also upregulated by ROS and that the antioxidant treatment decreased TXNIP levels induced by K5 and Sts (Zaragoza-Campillo and Morán, 2017). Here, we found that Txnip mRNA levels are also regulated by K5, Sts, and H₂0₂ (Figure 4). Cells were treated with K5 or Sts for 30 min to 5 h or with 100 μM H₂0₂ for 1 h. For K5, we found that Txnip levels showed a maximal increase after 2 h (about 20-fold) that decreased at 3 h, reaching an 8-fold increase after 4 and 5 h (Figure 4A).

When cells were treated with Sts, Txnip levels increased after 3 h (about 12-fold) and remained without change after 4 and 5 h of treatment (Figure 4B). To explore a possible involvement of Akt in the expression of Txnip we tested the effect of LY294002, an inhibitor of the IP3/Akt pathway. The inhibitor showed a dose-dependent increase of the Txnip levels (about 2 to 8-fold) after 4 h of treatment, supporting the idea that Akt is part of the signaling pathway (Figure 4C). Finally, we found that Txnip is 4-fold upregulated after 1 h of H₂0₂ treatment. To determine if there is an earlier response to the H₂0₂ we performed a temporal curve with H₂0₂ from 10 to 60 min. Under these conditions, Txnip levels showed a 17-fold increase after the first 10 min and a 24-fold increase at 20 min and after this time Txnip levels remained without change (4 to 5-fold increase) (Figure 4D). These data confirm that Txnip regulation is dependent on oxidant conditions and Akt inactivation.

As mentioned before, Akt inactivation induces Txnip mRNA expression (Figure 4). It is also known that Txnip promoters share common motifs for different transcription factors, including FOXO (Wang et al., 2006; Baker et al., 2008; Tsubaki et al., 2020), whose activity seems to depend on the inactivation of Akt. To test a direct regulation of Txnip gene by FOXO, we performed a Chromatin immunoprecipitation (ChIP) assay for FOXO1 and FOXO3 in mice CGN cultures as detailed in Methods. We analyzed the promoter sequences in humans and rats (NCBI accession number NM_006472 and NM_001008767), and we found that these motifs are conserved across these species (Figure 5A). Neurons were switched to K5 medium or treated with 100 μM H₂0₂ for 1 h. Amplification regions were designed for the two putative FOXO binding sites. The ChIP experiment revealed that endogenous FOXO3 associates with the Txnip promoter in motif 2 after 1 h of K5 treatment (Figures 5B,C). We used primers for Bdnf motifs and antibodies against TFII as a control. The untreated neurons showed a small enrichment in comparison with K5 after 1 h. We have not found conclusive data for FOXO 1 (data not shown). Interestingly, the enrichment levels with the treatment of 100 μM H₂0₂ for 1 h did not show a significant difference with K25 (control; Figure 5C). Since the TXNIP protein is increased by the same treatment with H₂0₂ (Zaragoza-Campillo and Morán, 2017), ChIP data suggest that the gene is induced earlier and is a transient response.

To identify FOXO3 localization during the death conditions and considering the Txnip expression data obtained (Figure 4), we performed a FOXO3 immunofluorescence of CGN treated with K5 and Sts during 30-, 60-, 120-, or 180-min. Figure 6A shows that under control conditions (K25) FOXO3 is mainly distributed in the cytoplasm. This is more evident in the inset where nucleus amplification is shown (Figure 6A). Pearson’s colocalization coefficient in this condition is 0.2 (Figure 6A graph). When cells were switched to a K5 medium, the FOXO3 signal increased in the nucleus, being evident after 2 h of K5 treatment (Figure 6A, arrows) with a colocalization coefficient of 0.4 (Figure 6A graph). The localization of FOXO3 in the nucleus after treatment with 100 μM H₂0₂ for 1 h showed a high Pearson’s coefficient (0.7) (Figure 6A, panel, graph). The translocation of FOXO3 correlates with the increase in ROS content.

When CGN was treated with staurosporine, cells showed an increase in FOXO3 nuclear localization after 3 h compared to K25. These data coincide with the expression of Txnip observed at 2 h with K5 and at 3 h with Sts (Figures 4A,B). To determine Akt participation in FOXO3 nuclear localization, we treated CGN with 100 μM LY294002 and we found a remarkable increase in FOXO3 nuclear distribution with a high Pearson’s coefficient (Figure 6B, arrow, and graph), which suggests that Akt inhibition regulates FOXO3 nuclear localization. Taking these results together we conclude that FOXO3 is a candidate to promote the Txnip expression in response to K5 and Sts in these cells.

We evaluated the participation of TXNIP in cell death in human MSN neuroblastoma cells treated with Sts. Potassium deprivation does not induce cell death in these cells. Txnip from humans and rats share 89% of identity and similar functions have been reported in both species. We found that MSN cells express Txnip under basal conditions and that shRNA decreases TXNIP protein levels by 10% 24 h after transfection (Figure 7A). MSN cells are less sensitive to staurosporine (data not shown), therefore we used 1 μM Sts to induce cell death. Control cells were treated with vehicle (DMSO) and transfection control was made by using an shRNA scramble sequence (Scr). Transfection efficiency was around 30% (data not shown) and no changes in the morphology and viability were found in control Scr or shRNA cells (Figure 7C).

When cells transfected with Scr were incubated for 24 h with Sts we observed a decrease of approximately 75% of cell viability measured as calcein staining (Figures 7A,C). In contrast, in cells transfected with shRNA against Txnip, Sts induced a decrease in cell viability of only 25%, which represents a 3-fold difference as compared with cells transfected with Scr sequence (Figures 7A,C). Data obtained with propidium iodide were in close agreement with calcein data (Figure 7C). Interestingly, Scr cells treated with Sts that survived after 24 h (stained with calcein) showed a rounder morphology compared to shRNA cells treated with Sts, which maintained the original morphology (Figure 7B amplification). We interpret this observation as an indication that the surviving Scr cells treated with Sts have already activated the apoptotic death pathway and therefore they will die in the following hours. These results demonstrate that TXNIP is required for cell death and that the lack of this protein increases neuronal viability under death conditions.

The role of JNK activation in apoptosis induced by K5 and Sts was confirmed in CGN by using SP600125, an inhibitor of the JNK protein in dose-dependent concentration (40 nM to 5 μM). Cell viability was measured 24 h after treatments. The reduction in the number of viable cells with K5 treatment was detected by calcein/propidium iodide technique. When cells were treated with K5 and the JNK inhibitor, CGN were markedly rescued from cell death. K5 treatment resulted in 26% of viable cells, in contrast to 80% of viable cells observed in the presence of SP (5 μM) (Figure 8). We also observed morphological changes associated with K5 treatment, including soma constriction and axonal degeneration, which were not evident in K5/SP treated cells (Figure 8, panel arrows).