Mitochondrial dysfunction and fatigue in Sjögren's disease

Abstract

Objective: Sjögren's disease (SjD) is a chronic exocrine disorder typified by inflammation and dryness, but also profound fatigue, suggesting a pathological basis in cellular bioenergetics. In healthy states, dysfunctional mitochondria are recycled by mitophagic processes; when impaired, poorly functioning mitochondria persist and produce inflammatory reactive oxygen species. Employing a case-control study, we tested our hypothesis that mitochondrial dysregulation in T cells is associated with fatigue in SjD.

Methods: We isolated pan T cells from peripheral blood mononuclear cells of 13 SjD and 4 non-Sjögren's sicca (NSS) subjects, who completed several fatigue questionnaires, along with 8 healthy subjects. Using Seahorse, we analysed T cells for mitochondrial oxygen consumption rate (OCR) and extracellular acidification rate, which we assessed for correlation with fatigue measures. Using public microarray data available for 190 SjD and 32 healthy subjects, we identified a mitophagic transcriptional signature that stratified SjD patients into 5 discrete clusters. Comparisons between the SjD subjects in these clusters to healthy individuals identified differentially expressed transcripts, which we subjected to bioinformatic interrogation.

Results: Basal OCR, ATP-linked respiration, maximal respiration and reserve capacity were significantly lower in SjD and NSS subjects compared with healthy individuals, with no differences in non-mitochondrial respiration, basal glycolysis or glycolytic reserve. Scores related to a sleep questionnaire and Bowman's Profile of Fatigue and Discomfort showed correlation with altered OCR in SjD. Subgroup differential expression analysis revealed dynamic transcriptional activity between mitophagy subgroups, expanding the number of differentially expressed transcripts tenfold.

Conclusions: Mitochondrial dysfunction and fatigue are significant problems in SjD warranting further investigation.

Keywords: Autoimmunity; Fatigue; Inflammation; Sjogren's Syndrome; T-Lymphocytes.

Figure 1. Mitochondrial metabolic characterization in T cells from SjD and healthy subjects fixed to the wells of tissue culture plates and analyzed using the Seahorse XF24 Extracellular Flux Analyzer. (A) Mitochondrial basal respiration (mitochondrial function under normal physiological conditions); (B) Non-mitochondrial respiration (the lowest oxygen consumption reading after adding antimycin A); (C) ATP-linked respiration (the proportion of basal respiration linked to energy production); (D) Maximal respiration (assesses the maximum phosphorylation capability of electron transport chain, after uncoupling electron transport chain from oxidative phosphorylation with carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP); (E) Mitochondrial reserve capacity (the ability of the cell to increase ATP production in response to stress); (F) Glycolytic reserve (representing highest extracellular acidification measured after adding FCCP). Values are expressed as the mean ± standard error of the mean.

Figure 2. Correlative relationships between fatigue scores assessed using validated questions and mitochondrial metabolic measures (basal oxygen consumption rate, ATP linked respiration, maximal respiration, and reserve capacity). Correlations are shown between oxygen consumption rate (OCR) and fatigue questionnaire scores for: (A) Sleep1; (B) Bowman’s PROFAD (Domain: General Discomfort, Facet: 8 [Ache all over]); (C) Bowman’s PROFAD (Domain: Mental, Facet: 5); (D) Bowman’s PROFAD (Domain: Mental, Facet: 6). Statistically significant correlations (p<0.05) are shown directly on the plots.

Figure 3. Heatmaps showing differentially expressed transcripts from the comparisons between SjD and healthy subject for: (A) all transcripts meeting padj<0.05 and |log2FC|>1 from results of overall analysis (S1); (B) extracted mitophagy transcripts meeting padj<0.05 for SjD cases alone revealed 5 discrete SjD mitophagy clusters (as indicated) when unsupervised average hierarchical clustering was applied.

Figure 4. Plots for selected mitophagy transcripts showing padj<0.05 stratified by group according to mitophagy cluster (M1-M5). The displayed log2FC results and padj values were taken from the results for subgroup differential expression analyses comparing those SjD patients comprising each mitophagy cluster to all healthy subjects (see online supplemental methods). Results are displayed for TANK-binding kinase 1 (TBK1; (A)), KRAS proto-oncogene, GTPase (KRAS; (B)), autophagy related 9B (ATG9B; (C)), mitogen-activated protein kinase 9 (MAPK9; (D)), PTEN induced kinase 1 (PINK1; (E)), and forkhead box O3 (FOXO3; (F)). All results with padj<0.05 are shown in bold, with the mean +/- standard deviation shown for each cluster.

Figure 5. Venn diagram (A) and UpSet plot (B) showing overlap off differentially expressed transcripts (padj<0.05 and |log2FC|>1) between the S1 and M1-M5 differential expression analyses. Few transcripts were unique to S1, suggesting that the M1-M5 analyses adequately capture most represented transcripts from the overall results.

Figure 6. Distribution of fatigue across mitophagy clusters from the replication dataset (GSE66795) employed. (A) Heatmap showing data for mitophagy transcript probes extracted from GSE66795 and subjected to unsupervised clustering showing similarly discrete clustering of SjD samples alone, as in figure 3; (B) Frequency of fatigue gradations across mitophagy clusters in SjD replication cohort; (C) Hybrid violin plot/bar chart showing distribution of fatigue data across mitophagy cluster with statistical analysis showing trending p-values (adjusted for multiple testing) between certain mitophagy clusters.