Physical activity and diurnal variation have no impact on plasma microRNA-140 expression levels in patients with knee osteoarthritis and healthy controls

Abstract

Objective

This study aimed to determine whether physical activity or diurnal variation influence circulating microRNA-140-3p (miR-140-3p) and miR-140-5p expression levels in patients with knee osteoarthritis (OA) compared to healthy controls.

Method

Twenty-one patients with knee OA and ten healthy controls ran on a treadmill for 20 ​min and blood samples were taken prior to and after running. To assess diurnal variation, blood samples were drawn at six different times during a 24-h period. RNA was extracted from plasma and used for cDNA synthesis. Expression levels were assessed with real-time quantitative polymerase chain reaction and compared between patients and controls.

Results

All participants had detectable expression levels of miR-140-3p and miR-140-5p. A paired analysis could not demonstrate statistically significant differences between the groups.

Conclusions

Plasma miR-140-3p and miR-140-5p expression levels were found in all samples; however, the expression levels were not affected by physical activity and did not show diurnal variation. The observation of stable miR-140 expression supports its potential as a reliable biomarker for OA, providing a methodological foundation for future diagnostic and translational studies.

1. Introduction

The impact and economic burden of osteoarthritis (OA) on society are escalating due to increased longevity and higher activity levels across all age groups [1]. Identifying a biomarker for OA is of global interest as this can possibly enable preventive treatment strategies to be employed earlier [2]. The small non-coding microRNAs (miRs) have been proposed as potential biomarkers for OA [3,4], and are easily accessible for analysis [5]. Of the vast numbers of miRs, the cartilage-specific miR-140 has aroused particular interest as a potential biomarker for OA [3,4].

MiR-140 is a cartilage-specific microRNA that plays a central role in joint homeostasis. It is highly expressed in chondrocytes and regulates genes involved in extracellular matrix turnover and inflammation, such as matrix metalloproteinase-13 and ADAMTS-5 [6,7]. Animal studies have shown that deletion of miR-140 accelerates OA-like changes, whereas overexpression protects against cartilage degradation [8,9]. In human OA cartilage, miR-140 expression is consistently downregulated, contributing to the dysregulated gene expression profile seen in diseased chondrocytes [10,11]. Notably, miR-140 is detectable in synovial fluid and circulation, and reduced levels have been associated with OA severity, suggesting its potential as a non-invasive biomarker [12].

When assessing potential biomarkers, certain factors are of importance. Physical activity has been shown to affect miR levels in plasma. Polakovičová et al. demonstrated that miR-21 often increased with intense or chronic exercise [13], whereas Biggish et al. reported that miR-1 and miR-133a/b increased after acute and endurance exercise [14]. Further, many biochemical molecules demonstrate a cyclic change over 24 ​h, indicating a diurnal variation which may be of significance [13,15]. Previous studies have reported the influence of physical activity and diurnal variation on miRs in plasma, but no study has assessed the effect of physical activity and diurnal variation in knee OA patients. We hypothesized that miR-140 would be affected by physical activity and/or diurnal variation. To test this hypothesis, we compared plasma expression levels of miR-140-3p and miR-140-5p between knee OA patients and healthy controls.

2. Methods

2.1. Participants

Twenty-one patients with knee OA and ten healthy controls aged 42–65 years were recruited between 2017 and 2023. Both men and women with a body mass index of 20–35 ​kg/m² were enrolled. Participants underwent a broad clinical examination, followed by weight-bearing radiographs of both knees, and venous blood samples were obtained.

Patients were required to fulfill both the clinical and radiographic criteria for knee OA, based on the guidelines of the American College of Rheumatology (ACR) and the Kellgren-Lawrence (K-L) grading scale (grades 0–4) [16,17]. The clinical examination was conducted according to the ACR criteria, and radiographic imaging was performed using the SynaFlexer™ X-ray positioning frame (Synarc, Newark, CA) [18]. The diagnosis of OA was established through the combination of clinical findings and radiographic evidence, with a K-L grade ≥1 considered indicative of OA. The control group consisted of healthy volunteers who did not meet the ACR criteria for knee OA and exhibited no radiographic signs of OA, defined as K–L grade 0 (Table 1). Exclusion criteria for the control group included acute injuries, symptoms or pain in the lower extremities and a history of prior knee surgery or trauma. Exclusion criteria for both groups included kidney and liver dysfunction, systemic disorders, cardiovascular disease, psychiatric illness and use of anti-platelets and/or anti-coagulants (Table 1).

Table 1.

Inclusion and exclusion criteria.

Inclusion criteria	Exclusion criteria
35–65 years	Systemic diseases
BMI 20–35	Rheumatic
Signed informed consent	Diabetes
	Cancer
	Hematologic
	Cardiovascular disease
	Psychiatric illness
	Ongoing use of
	Immunodepressants
	Anti-platelets and/or anti-coagulants
	NSAIDs
	Kidney and liver dysfunction (pathological levels of)
	Creatinine
	ASAT
	ALAT
	Albumin
	PT-INR
	Fossa cubiti
	Skin lesions
	Medical implant or device
	Infection

BMI – body mass index, NSAIDs – non-steroid anti-inflammatory drugs, ASAT – aspartat aminotransferase, ALAT – alanin aminotransferase, PT-INR – Prothrombin time – International normalized ratio.

2.2. Ethical approval and informed consent

The study was approved by the Regional Committee for Medical Research Ethics Southeast Norway (Ref. No.: 2015/1000). Prior to inclusion, all participants were asked to sign a written informed consent in accordance with the 1957 Declaration of Helsinki as revised in 2020.

2.3. Experimental protocols

Participants underwent two distinct protocols for blood sample collection: one involved physical activity and the other focused on diurnal variation. The physical activity involved a treadmill run at a speed of 6 ​km/h and 0° inclination for 20 ​min (Fig. 1a). The first blood sample (baseline) was collected before the run, following a 30-min rest while seated. After completing the exercise session, participants were seated again, and subsequent blood samples were collected. The diurnal variation protocol followed the standard set by Lottenburger et al. [19]. The first blood sample (baseline) was collected at 13:00 (Fig. 1b). Subsequent blood samples were taken every 3 ​h until 22:00, with two additional samples collected the following morning.

Fig. 1.

a) Scheme of the physical activity protocol. b) Scheme of the diurnal variation protocol. c) Schematic overview of the experiment.

BS – blood sample, min – minutes, hr – hours.

2.4. Collection, storage, and shipment of plasma

At each sampling, 5 ​mL of venous whole blood was drawn into EDTA tubes, which contain EDTA as an anticoagulant to prevent whole blood or plasma clotting, resulting in a total of 12 blood samples from each participant. The samples were centrifuged at 2500 ​g for 15 ​min and stored and transported in an upright position at −80 ​°C to Akershus University Hospital, Norway for isolation and analysis.

2.5. RNA extraction and complementary DNA (cDNA) synthesis

100 ​μl of plasma per sample was used for RNA extraction utilizing the MagMax miRVana total RNA kit and the KingFisher Apex automated extraction system (Thermo Scientific, Waltham, MA, USA). All samples were eluted in 50 ​μl of elution buffer from the kit containing the total RNA (including small RNA). Random RNA samples were chosen and checked for quality control, running the Agilent Small RNA kit (Agilent Technologies, CA, USA). All samples were satisfactory, showing a normal electrophoresis profile. Further, each RNA sample was synthesized to cDNA using the miRCURY LNA RT kit (Qiagen, Hilden, Germany). UniSp6 (0.5 ​μl) was added into the reaction mix as a spike-in control. The temperature cycling protocol was 42 ​°C for 60 ​min followed by 95 ​°C for 5 ​min for enzyme inactivation. The cDNA synthesis was performed on a Mastercycler X50s thermal cycler (Eppendorf, Hamburg, Germany). Absolute quantification using standard curves was not feasible due to the unknown starting RNA concentration in plasma. Therefore, we applied relative quantification (ΔΔCt method) for group comparisons. However, to ensure qPCR assay quality, we constructed standard curves (four-point dilution: 1:5, 1:25, 1:125, 1:625) using a pooled cDNA sample comprising patients and healthy controls. The 1:1 dilution was excluded due to saturation. The standard curves demonstrated r² values ranging from 0.985 to 0.999 (miR-140-3p: 0.997, miR-140-5p: 0.985, miR-16-5p: 0.999, UniSp6: 0.998), and amplification efficiencies ranged from 91.2 ​% to 102.3 ​%, indicating acceptable performance. All experimental samples had Ct values within the range of these standard curves. Melt curve analysis confirmed specific amplification with single, well-defined peaks for each primer set.

2.6. Real time quantitative polymerase chain reaction analysis

Each cDNA sample was run in duplicate in 384 well plates using the QuantStudio™ 6 Pro System (Applied Biosystems, Waltham, MA, USA). Primer assays targeting miR-140-3p (YP00204304) and miR-140-5p (YP00204540) (Qiagen, Hilden, Germany) were used. miR-16-5p (YP00205702) was selected as the reference microRNA for normalization based on prior literature demonstrating relatively stable expression in plasma and serum across a range of clinical conditions, including OA. A primer assay targeted at UniSp6 (YP00203954), (Qiagen, Hilden, Germany) expression was also utilized as a technical control. All four primer assays were run for the same patient samples in the same plate for optimal comparison. For all plates, the same two cDNA samples were run in duplicate as inter-plate controls and used for inter-plate calibration. The following temperature cycling was performed: an initial heat activation step (95 ​°C/5 ​min) followed by a 2-step temperature cycling for 45 cycles (95 ​°C/10 ​s and 56 ​°C/60 ​s). A melting curve analysis was added at the end. All samples were given cycle threshold values from the runs, and average Ct values (for the duplicates) were calculated using the delta-delta Ct method to obtain the miRNA expression profiles. The calibrator sample was set as the average of all control blood samples at baseline. Although no universally accepted housekeeping miRNA exists for extracellular miRNA qPCR, miR-16 has been commonly used due to its consistent performance in several studies [20,21]. All qPCR reactions were run in duplicate due to budgetary limitations. Technical replicates showed minimal variability, with low standard deviation (SD)s, supporting the reliability of the duplicate measurements. While triplicates are considered best practice, the consistency across duplicates suggests that measurement error was likely negligible. The study complied with the Minimum Information for publication of Quantitative real-time PCR Experiments (MIQE) checklist. An overview of the experiment is shown in Fig. 1c.

2.7. Sample size

Sample size calculations were based on data from an unpublished pilot study involving 10 patients with knee OA and 10 healthy controls. The OA group consisted of 5 men and 5 women (mean age 64.8 years; mean BMI 30.9), with a median K-L grade of 3 (range 2–4). The control group included 10 women (mean age 68.4 years; mean BMI 24.4), with a median K-L grade of 0 (range 0–2). Experimental procedures mirrored those used in the current protocols for physical activity and diurnal variation. The pilot study demonstrated a pooled SD of 0.2 in miR-140 fold change across groups. Based on these data, we estimated the sample size needed to detect a fold change difference of 0.2 in miR-140 levels with 80 ​% power and a 5 ​% significance level, assuming a conservative correlation coefficient of 0.1. The calculation indicated that a total of 21 patients with knee OA would be required. As no prior studies have evaluated circulating miR-140 in relation to physical activity or diurnal variation in this context, a control group of 10 participants was deemed adequate for exploratory comparison.

2.8. Statistical analysis

Mean, median and SD were calculated for continuous variables, and categorical data were expressed as frequencies and cumulative frequencies. The groups were compared using the control group's baseline as a reference for both physical activity and diurnal variation. Comparisons were made using multiple unpaired t-tests between controls and patients at corresponding time points. We used a false discovery rate approach to handle the multiple comparisons with the false discovery rate (q-value) set to 1 ​%. Within each series we tested for different expression levels using one-way analysis of variance, comparing each subsequent time point with the baseline using Dunnett's test adjustment for multiple comparison. A p-value of <0.05 was considered statistically significant. The analyses was performed using the Statistical Package for the Social Sciences, v. 29 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 10.0 (Prism Software, Irvin, CA, USA).

3. Results

Twenty-one patients with knee OA and ten healthy controls aged 42–65 years were included. There were no statistically significant differences in sex, age, body mass index or excretion ability between the two groups (Table 2). The groups were statistically significantly different in clinical and radiological OA assessments, and the control group reported statistically significant higher activity levels than the knee OA patients (Table 2).

Table 2.

Demographics of the study population.

	Patients, n ​= ​21	Controls, n ​= ​10	p-value
Sex, n (%) women	11 (52)	7 (70)	0.07
Age, mean (SD)	57.5 (6.4)	54.5 (5.4)	0.82
BMI, mean (SD)	27.5 (4.6)	24.9 (3.0)	0.09
Tegner score, mean (SD)	2.5 (2.0)	4.1 (1.2)	0.03
Lysholm score, mean (SD)	53.8 (12.5)	95.4 (4.4)	<0.001
Fulfill ACR criteria knee OA, n (%)	21 (100)	0 (0)	<0.001
K-L grade, median (range)	3 (1–3)	0 (0)	<0.001
Creatinine, mean (SD)	71.2 (10.0)	71.0 (12.3)	0.39
ASAT, mean (SD)	24.0 (7.7)	20.7 (4.1)	0.49
ALAT, mean (SD)	23.3 (12.6)	18.9 (5.7)	0.17
Albumin, mean (SD)	45.7 (2.9)	45.0 (2.2)	0.32
PT-INR, mean (SD)	1.0 (0.1)	1.2 (0.1)	0.37

n – numbers, SD – standard deviation, BMI – body mass index, ACR – American College of Rheumatology clinical evaluation, OA – osteoarthritis, K-L – Kellgren-Lawrence radiologic evaluation, ASAT – aspartat aminotransferase, ALAT – alanin aminotransferase.

All participants had detectable expression levels of miR-140-3p and miR-140-5p. Although miR-140-3p and miR-140-5p were detectable in all participants, expression levels exhibited inter-individual variability that did not show a consistent pattern in relation to physical activity (Fig. 2). Further, no statistically significant diurnal changes were observed (Fig. 3). A comparison at corresponding time points could not demonstrate statistically significant differences in expression levels between knee OA patients and controls in relation to physical activity or diurnal variation. However, differences in miR-140-3p in the diurnal variation experiment were flagged as possible discoveries in blood samples 1, 2 and 3 but with small fold changes (e.g., 1.21 (SE ​= ​0.10) vs. 0.9 (SE ​= ​0.04)) in blood sample 1 (Fig. 3). The raw Cq-files to the individual miRs are found in the supplement file.

Fig. 2.

Paired analysis of expression levels of miR-140-5p and miR-140-30 following physical activity protocol.

PA – physical activity, BS – blood sample, nd – not detectable.

Fig. 3.

Paired analysis of expression levels of miR-140-5p and miR-140-30 after diurnal variation protocol.

DV – diurnal variation, BS – blood sample, nd – not detectable.

4. Discussion

The most important finding of the present study was that physical activity and diurnal variation did not influence miR-140-3p and miR-140-5p expression levels in knee OA patients compared with healthy controls. Despite small, inconsistent differences in expression levels at different time intervals, no meaningful differences were detected between patients and controls.

The effect of physical activity on circulating miR expression levels has been summarized in a review by Polakovicová et al. [22]. The authors reported that expression levels are sensitive and specific to the type and intensity of physical activity. However, inconsistent variances were reported, but no study reported expression levels of miR-140. Although we were unable to detect any differences in miR-140 expression levels, our study is the first to report miR-140 expression levels after physical activity.

Like many biochemical molecules, miRs are assumed to have diurnal variation. Heegaard et al. examined 92 miRs in plasma over a 24-h period and in contrast to our study, they found that miR-140-5p showed a statistically significant reduction from daytime to nighttime [15]. One major difference between the studies is the study population. Our participants were both men and women with knee OA and healthy controls with a mean age of 55, whereas the study by Heegaard et al. consisted of 24 Caucasian males with a mean age of 26 years. Based on the methodological differences, we suggest that the study by Heegaard and colleagues lacks generalizability to an OA population.

Similar to our findings, Poulet et al. failed to detect any influence of the circadian rhythm in cell-free DNA [23]. Their blood sampling followed a 4-h regime, similar to our protocol, adding further strength to the results. However, their reported results should be treated with caution in the OA setting as they only examined healthy male subjects aged 18–40 years.

The findings of this study align with prior evidence highlighting the central role of miR-140 in cartilage homeostasis and OA pathophysiology. miR-140 is predominantly expressed in chondrocytes and regulates a range of genes involved in matrix remodeling and inflammatory signaling, including MMP-13 and ADAMTS-5 [6,7]. Its downregulation in OA cartilage has been consistently observed and is thought to contribute to the upregulation of catabolic pathways that drive cartilage degeneration [10,11]. Experimental models have further substantiated its protective role: mice lacking miR-140 exhibit spontaneous OA-like joint pathology, while intra-articular overexpression attenuates cartilage damage in induced OA models [8,9]. Importantly, the presence of miR-140 in synovial fluid and circulating blood components supports its utility as a non-invasive biomarker. Reduced miR-140 levels in the synovial fluid of OA patients, as reported by Yin et al., were inversely correlated with radiographic disease severity [12], underscoring its potential for disease stratification and monitoring. Collectively, these observations reinforce the dual significance of miR-140 as both a mechanistic mediator and a translational target in OA.

To date, only a few studies have assessed circulating miRs as potential biomarkers for OA [[24], [25], [26], [27]]. Common to all studies are inconsistent findings and the inability to identify a biomarker for OA. This can be explained by the fact that the detected miRs are abundantly found inside the cell cytoplasm or tightly bound to proteins or lipoproteins. A recent study showed a decrease in concentration of cell-free DNA after meal ingestion [23], which was explained either by a post-prandial effect or detection bias due to higher plasma levels of lipids and triglycerides. Unfortunately, our study did not take meal ingestion into account when assessing diurnal variation of miR-140.

While qPCR remains the most commonly used method for miRNA quantification in plasma, variability in some samples suggests that additional methodological approaches, such as digital PCR or small RNA sequencing, may improve sensitivity and reproducibility in future studies. Validation using an orthogonal method could strengthen the reliability of miR-140 detection under low-concentration conditions. Our observation of stable miR-140 expression under realistic clinical conditions enhances the translational relevance of our findings. Clinically, the stability of miR-140 in plasma irrespective of physical activity or daytime sampling increases its potential utility as a reliable biomarker should future studies confirm its diagnostic value. Our study thereby provides an essential methodological foundation for the standardized measurement of miR-140 in future OA biomarker research.

Our study has limitations. It has been demonstrated that miRs can exist not only in their canonical form, but also through isomiRs [28]. In fact, isomiRs can outnumber canonical miRs [28]. The importance of this finding has yet to be clarified, but one must assume that they affect biological function and that the presence of isomiRs could affect expression levels of detected miRs. Unfortunately, our study did not include assessment of isomiRs. We also ran duplicates and not triplicates when analysing expression levels, but since the SDs were very small, it is unlikely that this would have had any effect on our results. When we planned this study, we opted to use miR-16-5p for normalization as this has previously been reported to be stably expressed across similar sample cohorts [24]. However, there are also publications demonstrating that the expression of miR-16 tends to decrease in OA patients compared to healthy controls which could bias our results [25]. The relatively small sample size, particularly when combined with biological variability in circulating miRNA levels, may have limited the power to detect subtle expression differences. Future studies should aim to validate these findings in larger and more stratified OA cohorts. The sample size was calculated for OA patients, not for controls. Ten healthy controls might not have been sufficient, but several studies on miRs lack any control group. Hence, we chose to proceed with ten controls for convenience, and better than none. The study protocol for diurnal variation did not contain any analysis during nighttime. It is possible that expression levels of miR-140 show variation during the night, but this was not examined for practical and economic reasons.

In conclusion, our study identified a stable expression of miR-140 in all plasma samples, which was not influenced by physical activity or diurnal variation and showed no difference between OA patients and healthy controls. The observation of stable miR-140 expression supports its potential as a reliable biomarker for OA, providing a methodological foundation for future diagnostic and translational studies.

Author contributions

TFA: interpretation of data, drafting and revision of manuscript. RBJ: statistical analysis, interpretation of data, revision of manuscript. MBD: statistical analysis, interpretation of data, revision of manuscript. AA: interpretation of data, revision of manuscript. PHR: interpretation of data, revision of manuscript. MSH: interpretation of data, revision of manuscript. OBL: original idea, interpretation of data, revision of manuscript.

Data availability statement

The participants of this study did not give written consent for their data to be shared publicly; therefore, due to the sensitive nature of the research, the data are not available.

Declaration of Generative AI in scientific writing

The authors report that AI and AI-assisted technologies has not been used in the writing process.

Funding

This work was supported by a research grant from Møre and Romsdal Hospital Trust (Ref. No.: 2023/5413).

Competing interest statement

The authors report there are no competing interests to declare.

Handling Editor: Professor H Madry

Abbreviations

miR

MicroRNA

OA

Osteoarthritis

Appendix A. Supplementary data

The following is the Supplementary data to this article.