Unravelling myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS): Gender‐specific changes in the microRNA expression profiling in ME/CFS

Abstract Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a multisystem illness characterized by medically unexplained debilitating fatigue with suggested altered immunological state. Our study aimed to explore peripheral blood mononuclear cells (PBMCs) for microRNAs (miRNAs) expression in ME/CFS subjects under an exercise challenge. The findings highlight the immune response and inflammation links to differential miRNA expression in ME/CFS. The present study is particularly important in being the first to uncover the differences that exist in miRNA expression patterns in males and females with ME/CFS in response to exercise. This provides new evidence for the understanding of differential miRNA expression patterns and post‐exertional malaise in ME/CFS. We also report miRNA expression pattern differences associating with the nutritional status in individuals with ME/CFS, highlighting the effect of subjects' metabolic state on molecular changes to be considered in clinical research within the NINDS/CDC ME/CFS Common Data Elements. The identification of gender‐based miRNAs importantly provides new insights into gender‐specific ME/CFS susceptibility and demands exploration of sex‐suited ME/CFS therapeutics.


| INTRODUC TI ON
The underlying mechanisms associated with the onset and progression of myalgic encephalomyelitis/chronic fatigue syndrome (ME/ CFS), a condition characterized by an abrupt or delayed onset of persistent or relapsing symptomatology, including memory problems, muscle and joint pain, gastrointestinal issues, neurological problems, hormonal imbalance and debilitating fatigue or easy fatigability, remains unclear. Still, a decade of research into ME/CFS disease activity has provided evidence of immune dysfunction described by persistent immune activation associated with the onset and progression of the illness. [1][2][3][4][5][6][7][8][9][10] During the last decade, miRNAs have been emerging as biomarkers in ME/CFS due to their regulatory role in the development, maturation and proliferation of immune cells. [11][12][13][14][15][16][17] Brenu et al reported miR-127-3p, miR-142-5p and miR-143-3p as potential plasma and miR-146a, miR-223 and miR-21 as cytotoxic lymphocyte biomarkers for ME/CFS diagnosis, 15,16 and due to this illnesses' similarity with fibromyalgia, a complex neuroimmune condition characterized by similar symptomatology, miR-223-3p, miR-451a, miR-338-3p, miR-143-3p and miR-145-5p have come into focus as well. 17, 18 Petty et al 19 reported up-regulated miR-99b, miR-330, miR-126 and miR-30c as potential diagnostic markers for individuals with ME/CFS among the 34 miRNA which were found to be differentially expressed in individuals with ME/CFS. A recent study reported significantly diminished miR-let-7i-5p, miR-93-3p and miR-200a-5p in cerebrospinal fluid of ME/CFS subjects who underwent exercise challenge 20 which is important due to the post-exertional malaise (PEM) being a unique characteristic of ME/CFS. Altogether, despite of the evidence of gender-related differences in this illness, 21,22 a comprehensive analysis of the effects of an exercise testing on differential miRNA expression between males and females with ME/CFS has not been performed so far.
The present study aimed to investigate peripheral blood mononuclear cells (PBMCs) miRNA expression in ME/CFS subjects in comparison with sedentary healthy controls (HCs) under an exercise challenge. The differential expression of miRNAs observed in ME/CFS was further evaluated in females and males. Our results indicate that miRNA expression differs between female and males with ME/CFS both at the baseline, and in response to the exercise challenge. In addition, we found that miRNA expression was different in PBMCs of fasting as compared to non-fasting subjects with ME/CFS, which emphasizes the importance of the National Institute of Health (NIH), National Institute of Neurological Disorders and Stroke (NINDS) and Centers for Disease Control and Prevention (CDC)'s newly adopted ME/CFS Common Data Elements (CDE) principles for detailed reporting of subjects' conditions.

| Participants and samples
A community-based cross-sectional study included 63 clinically diagnosed ME/CFS subjects and 55 healthy controls matched for age (±5 years) and BMI (±2). All individuals with ME/CFS and HCs were recruited from three different geographical areas: Miami/

| Isolation of PBMCs and RNA extraction
Up to 8 mL of whole blood/sample were collected in K 2 EDTA tubes (Becton Dickinson) and processed within 2 hours by dilution at 1:1 (v/v) ratio in phosphate-buffered saline solution (PBS) with layering on top of 1 volume of Ficoll-Paque Premium (GE Healthcare) and separation by density centrifugation at 500× g for 30 minutes (20°C, brakes off). PBMC layers were isolated and washed with PBS. Isolated PBMC pellets were resuspended in 1 volume of red blood cell lysis buffer (155 mmol/L NH 4 Cl, 10 mmol/L NaHCO 3 , 0.1 mmol/L EDTA and pH 7.4), kept on ice for 5 minutes, and centrifuged (20°C at 500× g for 10 minutes). Washed pellets were resuspended in freezing medium (90% FBS, 10% DMSO) adjusting their concentration to 10 7 cells/ml, aliquoted and frozen in liquid nitrogen until use. Total RNA was extracted using RNAzol (Molecular Research Center, Inc.) according to the manufacturer's instructions.
All RNA samples had RNA Integrity number (RIN) above 7.

| miRNA expression analysis
We used NanoString nCounter platform (NanoString Technologies) to evaluate miRNA expression with the nCounter Human miRNA recommendations are those of the author and are not necessarily endorsed by either by the Department of Defense or National Institute of Health.

K E Y W O R D S
exercise, gender, immune function, inflammation, microRNAs, myalgic encephalomyelitis/ chronic fatigue syndrome, peripheral blood mononuclear cell Panel v3 (NanoString Technologies, GXA-MIR3-12). The panel included unique oligonucleotide tags for over 800 highly curated human miRNAs (from miRbase v21) and five housekeeping genes for normalization of expression between samples (ACTB, B2M, GAPDH, RPL19 and RPLP0). Each sample was analysed by using 100 ng of total RNA for processing and consecutive hybridization (21 hours at 65°C) to the probe pairs consisting of Reporter Probe, which carry the signal on its 5′ end, and Capture Probe, which holds biotin on its 3′ end. After hybridization, sample cleanup and counting were performed according to the manufacturer's instructions.

| Data analysis
Raw counts were analysed using NanoString nSolver version 4.0 software. We calculated geometric means of negative ligation controls plus two standard deviations for all samples. All count values below this threshold were excluded from normalization. After that, all the normalization steps were performed according to manufacturers' instructions.
Only miRNAs that were expressed in 75% or more of samples in at least one group were further analysed. We then selected miR-NAs that showed statistical significance with P < .05. Differentially   expressed miRNAs in each contrasted group were detected by oneway ANOVA using Partek Genomic Suite v7, and false discovery rate (FDR) was calculated for multiple comparisons using the q-value. 26 MiRNAs were differentially expressed if they met the following selection criteria: FDR ≤ 0.1, and fold change (FC) of at least 1.5 in either direction. For some further analyses, we also considered miR-NAs that did not pass FDR ≤ 0.1 but passed unadjusted P < .05 (as indicated in Tables S1-S5 and explained in the Section 3). Data were plotted using GraphPad Prism 5 (GraphPad Software The significance level for all analyses was set at P < .05.

| Participant characteristics
The study compared 37 female and 26 male ME/CFS subjects with 33 female and 22 male-matched HCs. A total of 24 female and 11 male ME/CFS subjects, and 21 female and 13 male-matched HCs underwent exercise challenge (Table 1a). Blood was drawn from these participants for PBMC isolation at T0 (baseline before the exercise), T1 (VO 2 max, peak of exercise challenge) and T2 (four hours after T1, recovery). These individuals had unified breakfast (one banana and a yogurt) 30 minutes before the exercise challenges and the first blood draw. T0 was designated for participants at baseline that did not undergo exercise challenge and fasted overnight before blood draw. In the fasting group, there were 13 female and 15 male ME/CFS subjects with 12 females and 9 males as HCs (Table 1b).
No significant difference was observed for age, gender and body mass index (BMI) between individuals with ME/CFS and HCs, regardless of fasting status. ME/CFS cohort had poor (self-reported, P < .05) outcomes in all reported domains of SF-36 except mental health (P = .595), as compared to healthy controls, regardless of fasting status (Table 1).

| MicroRNA expression profile differences between ME/CFS individuals and HCs without stratification
We analysed the difference in miRNA expression between all individuals with ME/CFS and HCs regardless of gender and fasting status (breakfast vs fasting). At T0, the expression of miR-150-5p, miR-4443, miR-423-5p and miR-342-3p was higher, whereas, for miR-199-3p, miR-126a-3p, let-7i-5p and miR-130a-3p, the expression was lower in ME/CFS than HCs. However, out of a total of eight miRNAs that were differentially expressed with P < .05, only miR-150-5p passed criteria of FDR < 0.1 (Figure 1, Table S1a).
Only miRNAs that showed P < .05 in non-stratified and at least once in gender-stratified analysis were shown in figures.

| Response to exercise without stratification by gender
For these comparisons, we only selected data from individuals who completed all three time-points in exercise and did not undergo extended fasting before blood draw. For figures and discussion, we focused on miRNAs highlighted across analyses showing P < .05 in non-stratified analysis and at least once in sex-stratified analysis (P < .05).

| Comparisons for miRNAs expression in response to exercise
A subgroup of individuals who underwent exercise challenge was selected to compare differences in miRNA expression between ME/ CFS and HCs in response to exercise (Figure 3), as detailed in 3.1.
Interesting to note is that regardless of time-points being evaluated, individuals with ME/CFS showed higher expression of miR-150-5p and miR-423, and reduced miR-199-3p levels, although not all miRNAs could reach FDR < 0.1 (Figure 3, Tables S2a-c).
The expression of some miRNAs was detected or fluctuated in response to exercise, namely (a) miR-4443 expression fluctuated in response to exercise; higher at T0, reduced at T1 and back to elevated at T2, (b) miR-125a-5p which was overexpressed at T1, F I G U R E 2 miRNA expression profile differences between diagnosed ME/CFS individuals and HCs at T0. Data are shown as normalized counts for males (♂) and females (♀) ME/CFS (red) and healthy controls (HC; Blue) at T0 for (A) miR-150-5p, (B) miR-4443, (C) miR-423-5p, (D) miR-199-3p and (E) miR-130a-3p, regardless of fasting status. * denotes P < .05 and # denotes FDR < 0.1 underexpressed at T2, but did not show significant differences at T0 and 3) miR-374-5p and miR-148a-5p were underexpressed both at T1 and T2 but were not significantly different at T0. It shows that expression of these miRNAs is affected by exercise in ME/CFS (Figure 3, Tables S2a-c).

| Comparisons of miRNA expression between time-points in exercise
The differential expression of miRNA in response to an exercise challenge was evaluated between ME/CFS and HCs for intertime-point comparisons ( Figure 3). Ten miRNAs reached statistical significance of P < .05 for differential expression between T1 and T0 (T1/T0) in HCs (Table S3a), as compared to one in ME/ CFS (Table S3b). No miRNA passed criteria of FDR < 0.1; however, miR-181a-5p showed increased expression at T1 regardless of disease status ( Figure 3, Table S3a,b). Six miRNAs were differentially expressed between T2 and T1 (T2/T1) in HCs (P < .05); however, only miR-363-3p passed criteria of FDR < 0.1 with its expression being lower at T2 than at T1 (Table S3c). No miRNA passed criteria of FDR < 0.1 for T2/T1 comparison for ME/CFS, although miR-4516 was significantly underexpressed at T2 (Table S3d). Interestingly, miR-181a-5p did not pass criteria of FDR < 0.1, the higher expression of miR-181a-5p seen in HCs and ME/CFS between T1 and T0 (P < .05), but lowered expression between T2 and T1, for HCs only (P < .05, Figure 3).

| Comparisons of miRNA expression between time-points in exercise stratified by gender
The patterns of miRNA expression were further analysed in ME/CFS and HCs for differences between time-points in both females and males. Eight miRNAs showed differential expression in healthy females for T1/T0, and only overexpressed miR-363-3p passed criteria of FDR < 0.1 (Table S5a). Both increased expression of miR-181a-5p and the reduced expression of miR-125a-5p could not reach FDR criteria (P < .05, Table S5a). For T2/T1 comparison in HC females, nine miRNAs were differentially expressed where miR-125a-5p was overexpressed and miR-181a-5p, miR-363-3p and miR-4443 were reduced in expression. However, only miR-181a-5p, miR-363-3p passed criteria of FDR (<0.1) significance (P < .05, Table S5b).
Moreover, the increased expression of miR-150-5p seen in ME/ CFS regardless of fasting status, presented higher levels in fasting as compared to non-fasting suggesting possible modulation in ME/ CFS by dietary factors.
Three miRNAs were highlighted here that showed opposite pattern of expression in ME/CFS between fasting and non-fasting: (a) the fasting ME/CFS females and males showed overexpressed miR-223-3p opposite to consistently underexpressed pattern in non-fasting ME/CFS, even in response to exercise, (b) miR-191-5p was overexpressed in both fasting females and males but it was underexpressed in non-fasting ME/CFS males, (c) the consistent higher expression of miR-343-3p in fasting ME/CFS which was not identified in other group and (d) underexpressed miR-548q in both fasting ME/CFS females and males.

| D ISCUSS I ON
This is the first ME/CFS exercise challenge study which evaluated gender differences in ME/CFS utilizing miRNA expression profiles.
The results of this study have highlighted: (a) miRNAs that are differentially expressed in ME/CFS, (b) altered miRNAs induced by exercise challenge in ME/CFS, (c) gender influenced miRNAs and their differential expression in response to exercise in disease state and finally (d) fasting induced miRNA difference, emphasizing the possible role they may play in the ME/CFS-associated disease biology.
In this study, miR-150-5p, miR-130a-3p, miR-199-3p, miR-223-3p, miR-126a-3p, miR-4443, miR-374-5p, miR-146a-5p and miR-423 have emerged as prominent miRNAs whose expression was altered in ME/CFS. These miRNAs with genes involved in cellular processes and immunity as their molecular targets have been detected in other ME/CFS studies. [10][11][12][13][14][15][16][17][18][19][20]27 Brenu et al had detected miR-146a-5p, miR-223-3p and miR-126a-3p in NK and CD8 + cells, and in plasma of individuals with ME/CFS, respectively. 15,16 Bjersing et al detected reduced expression of both miR-223-3p and miR-374-5p in cerebrospinal fluid and serum of ME/CFS subjects. 17,27 The up-regulation of miR-150-5p is seen in both T-cell and B-cell maturation and differentiation via Notch pathway and c-Myb, respectively, and it influences the release of pro-inflammatory cytokines suggesting its critical involvement in the modulation of immunity and inflammatory response. [28][29][30] MiR-199-3p, regulated by free fatty acids and adipokines, is a negative regulator of NF-κB and IL-8. 31 Additionally, low miR-199-3p expression, seen in ME/CFS subjects, is linked with poor survival outcome in carcinomas possibly affecting the disease-related physiological burden. 32,33 Another dysregulated miR-223 identified in this study is reported in other inflammatory conditions, infections and carcinomas. 34 It modulates the TLR4/TLR2/NF-κB/STAT3 signalling pathway consequently affecting inflammatory cytokine expression 35 and controls inflammatory bowel disease (IBD)-associated intestinal inflammation by inhibiting NLRP3 inflammasome. 36 It was shown that the cytokines released in response to inflammatory assault, particularly TNF-α, is directly suppressed by miR-130a-3p, reducing inflammation and associated oxidative stress. 37 The higher inflammatory cytokine expression seen in ME/CFS could therefore be induced by diminished miR-130a-3p expression in ME/ CFS subjects, as identified in the above-mentioned study. 37 Another miRNA presenting consistent reduced expression both at baseline and at recovery post-exercise is miR-126a-3p, which is linked with inflammation. However, Petty et al 19 reported that individuals with ME/CFS had up-regulated miR-126. The cause of this conflicting finding could be attributed to the subgroup differences within the ME/CFS subject population, possible gender influences due to higher proportion of females in Petty' study or lifestyle habits. On the other hand, we observed elevated expression of miR-423 in ME/CFS which is associated with inflammation and is also reported to be overexpressed by Petty et al. 19 In a study by Gharbi et al, 38 miR-423-3p was found to be elevated in veterans exposed to sulphur mustard (SM), a vesicant agent responsible for SM-induced necrosis and inflammation.
Of the highlighted miRNAs in this study, miR-146a is among the first-studied miRNA for its role as key regulator of the immune and The overexpressed miR-4443 increases pro-inflammatory cytokines by activated NF-κΒ pathway via targeting TRAF4. 41 In fact, a consistent feature of ME/CFS is significantly reduced natural killer (NK) cell function, unique pro-and anti-inflammatory cytokines, hypersensitivity responses and viral infections prior to ME/CFSassociated symptom onset.
In relation to exercise response, we successfully identified key miRNAs which could be critical to the post-exertional recovery-associated pathways in ME/CFS. Consistently in ME/CFS, miR-150-5p and miR-423 (3p and 5p) were overexpressed while miR-199a-3p, miR-130a-3p, miR-146a-3p and miR-223-3p were underexpressed regardless of stage in the exercise challenge. Additionally, the expression of miR-4443 was transiently reduced in ME/CFS after exercise.
Most of the miRNAs which were found to be reduced in ME/ CFS, regardless of gender and time-point otherwise confer protective effect against inflammation. [42][43][44] Inter-time-point comparisons highlighted miR-181a-5p, miR-125a-5p and miR-363-3p to be influenced by gender. These miRNAs are associated with TNF-α levels, regulation of differential activation of macrophages and inflammation and mast cells function, respectively, whose expression is affected by exercise. [44][45][46][47] In this sense, the observed miRNA expression pattern in ME/CFS could be interpreted as an "echo" of the imbalance in immune molecules reported in former ME/CFS biomarker studies. 1-4 The reduction of miR-22-3p has been reported in ME/CFS at re- and ME/CFS. 20 We saw this reduction only in males, and interestingly, the group that displaced this reduction in response to exercise in Baraniuk study was mostly males. 20 MiR-22 directly targets NLRP3 (Nod-like receptor protein 3), a key protein in the NLRP3 inflammasome. 48 The HCs recovered from exercise induced down-regulation of both miR-130-3p and miR-223 miRNAs in response to exercise challenge, but it was also only reflected in males.
Few studies have evaluated the relationship of fasting with differential expression of miRNAs, but the risk diet imposes on chronic disease development is well known. In fasting ME/CFS, we identified key miRNAs that may play a critical regulatory role in ME/CFS metabolic state. The inconsistencies in miRNA expression profiles between studies could have been, at least partly, a result of fasting-induced processes influencing miRNA expression. It is, therefore, extremely important to align assessment and reporting with Common Data Elements (CDE) in human subject research that improves data quality and allows for comparisons across multiple studies. NIH has recently adopted CDE principles in detailed reporting of ME/CFS subjects' conditions and encourages the use of these carefully selected CDEs in ME/CFS research which should be utilized for proper registry towards generalizability of findings. 49 To the best of our knowledge, we are the first to characterize the gender effects on miRNA profiles in ME/CFS compared with matched healthy controls, particularly in response to exercise. The findings of this study should be useful not only for the understanding of gender-dependent disease-specific ME/CFS pathology, but also in the identification of therapeutic regimes suited to ME/CFS gender-specific biology demands.

CO N FLI C T O F I NTE R E S T
The authors confirm that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in the Appendix S1 of this article.