Attenuated Induction of the Unfolded Protein Response in Adult Human Primary Astrocytes in Response to Recurrent Low Glucose

Abstract

Aims/hypothesis: Recurrent hypoglycaemia (RH) is a major side-effect of intensive insulin therapy for people with diabetes. Changes in hypoglycaemia sensing by the brain contribute to the development of impaired counterregulatory responses to and awareness of hypoglycaemia. Little is known about the intrinsic changes in human astrocytes in response to acute and recurrent low glucose (RLG) exposure.

Methods: Human primary astrocytes (HPA) were exposed to zero, one, three or four bouts of low glucose (0.1 mmol/l) for three hours per day for four days to mimic RH. On the fourth day, DNA and RNA were collected. Differential gene expression and ontology analyses were performed using DESeq2 and GOseq, respectively. DNA methylation was assessed using the Infinium MethylationEPIC BeadChip platform.

Results: 24 differentially expressed genes (DEGs) were detected (after correction for multiple comparisons). One bout of low glucose exposure had the largest effect on gene expression. Pathway analyses revealed that endoplasmic-reticulum (ER) stress-related genes such as HSPA5, XBP1, and MANF, involved in the unfolded protein response (UPR), were all significantly increased following low glucose (LG) exposure, which was diminished following RLG. There was little correlation between differentially methylated positions and changes in gene expression yet the number of bouts of LG exposure produced distinct methylation signatures.

Conclusions/interpretation: These data suggest that exposure of human astrocytes to transient LG triggers activation of genes involved in the UPR linked to endoplasmic reticulum (ER) stress. Following RLG, the activation of UPR related genes was diminished, suggesting attenuated ER stress. This may be a consequence of a successful metabolic adaptation, as previously reported, that better preserves intracellular energy levels and a reduced necessity for the UPR.

Keywords: ER stress; human primary astrocytes; recurrent low glucose; transcriptome (RNA-seq); unfolded protein response.

Figure 1

Schematic of the recurrent low glucose model. Human primary astrocytes were exposed to 0, 1, 3, or 4, three-hour long bouts of 0.1 mmol/l glucose; control (C), acute low glucose (LG), antecedent recurrent low glucose (aRLG), and recurrent low glucose (RLG) respectively. Each day cells were first incubated in 2.5 mmol/l glucose for 2 hours as a step down from overnight/stock media of 5.5 mmol/l glucose. Adapted from (18).

Figure 2

Glucose variation alters expression of genes involved in endoplasmic-reticulum stress. Volcano plots on the pairwise differential expression analysis between control cells (C) versus (Ai) low glucose (LG), (Aii) antecedent RLG, and (Aiii) recurrent low glucose (RLG), the red points on the plots represent genes padj<0.05. (B) Heatmap of hierarchical clustering of LRT analysis FDR ≤ 0.1 indicates differentially expressed genes (rows) between the four groups (padj<0.1). Orange indicates up-regulation and blue indicates down-regulation. The LG and RLG groups cluster together. TXNIP (C), XBP1 (D), HSPA5 (E), MANF (F), ND4L (G), ND4 (H) expression profiles, selected for their functional relevance to hypoglycaemia (p-value is the adjusted result of the likelihood ratio test). n=5; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data presented as Mean ± SD.

Figure 3

Effect of glucose variation on DNA methylation. (A) The most differentially methylated genes (ANOVA p ≤ 0.001) in pairwise comparison (red points are p<0.0001) between control treated HPA cells (C) versus (Ai) low glucose (LG), (Aii) antecedent recurrent low glucose (aRLG), and (Aiii) recurrent low glucose (RLG). (B) Heatmap of hierarchical clustering using probes ANOVA p<0.001 indicates differentially methylated cg sites (rows) between the four groups. Orange indicates hypermethylation and yellow indicates hypomethylation. Box plots of some of the most differentially methylated CpG sites labelled by their associated gene, selected for functional importance (C), cg07417745/SLC19A3, (D), cg16777181/GRID1, (E), cg11022541/NIPA1, (F), cg11692715/SLC8B1 (G), cg22467827/CLHC1 (p-value is the adjusted result of the ANOVA). (H), CLCH1 gene expression increases. Error bars represent standard deviation (I), Venn diagram of differentially methylated cg sites in yellow and differentially expressed genes (blue) and overlap between the two data sets, 28 genes. Data presented as Mean ± SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. n=6 for methylation data and n=5 for gene expression changes.