Circulating biomarkers associated with walking performance in elderly subjects: exploring miRNAs, metabolic and inflammatory biomarkers

Abstract

Aging phenotype is characterized by musculoskeletal impairment that leads to diminished mobility and physical function. This study investigated whether circulating miRNAs and metabolic and inflammatory biomarkers may reflect the walking performance of the elderly. Elderly hospitalized for an acute condition and recruited from the ReportAge Biobank were grouped, based on their walking performance, in active subjects (n = 23, age: 83.0 ± 4.3), able to walk ≥ 1 km and who performed more than 1 h activity, and inactive subjects (n = 23, age: 85.0 ± 6.0), able to walk < 100 m and who performed < 1 h activity in the 3 days prior hospitalization. Plasma levels of 754 miRNAs were evaluated using OpenArray® platform, and miRNAs whose level was ± 2.5 fold (p < 0.05) were validated by qPCR. Target prediction for validated miRNAs was performed on MirWalk 3.0, Gene Ontology and pathway enrichment on Panther 19.0. Cytokines and metabolites associated with bone, muscle, and inflammation were evaluated from plasma samples using Luminex and ELISA. Among the 7 miRNAs found differentially expressed in active compared to inactive elderly after the initial screening, 4 miRNAs were validated: hsa-let7g-5p, hsa-miR-27a-3p, hsa-miR-361-5p, hsa-miR-574-3p, all upregulated in the active group. Gene Ontology and pathway enrichment analysis revealed the identified miRNAs potentially involved in muscle and bone metabolism during aging. Among cytokines, gp130 and IL-10 significantly differed between the two groups. This study suggests the potential association of specific circulating biomarkers with walking performance in elderly and their potential involvement in the molecular mechanism underlying age-associated musculoskeletal impairment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-025-01510-2.

Introduction

The decline of age-related mobility can be attributed to both primary aging processes and lifestyle factors, such as a significant reduction in the amount and intensity of physical activity. The impairment of physical performance is a crucial factor in establishing the frail phenotype [1]. In the absence of neurological diseases, the principal cause of physical performance worsening is the progressive musculoskeletal impairment, either primary or secondary to metabolic dysfunctions [2]. Such impairment may lead to changes in gait (walking ability), balance, and strength, increasing the number and severity of falls and limiting daily activities. Walking ability, i.e., the ability to walk without assistance, is critical for independence and autonomy, but also loss of mobility is associated with higher rates of morbidity, disability, and mortality. Additionally, the persistence of this condition often represents exclusion criteria for clinical trials [3].

In the context of mobility in aging, most studies focus on functional aspects related to musculoskeletal impairment and/or biomarkers of specific musculoskeletal dysfunctions [4]. Sarcopenia, characterized by loss of skeletal muscle mass, strength, and functionality, represents a relevant example in this field. Sarcopenia is defined in terms of skeletal muscle mass by DXA or bioelectrical impedance analysis (BIA) and function measurement, such as grip strength, chair stand test, and gait speed; no biochemical or molecular markers are included in this diagnostic definition [5]. However, the 2023 consensus statement from the European Society for Clinical and Economic Aspects of Osteoporosis Osteoarthritis and Musculoskeletal Diseases (ESCEO) provided a list of biochemical markers of musculoskeletal health and aging that can be recommended in clinical trials on sarcopenia, but none of these markers showed enough diagnostic relevance [6]. Recently, we identified two circulating micro-RNA signatures (hsa-miR-126-5p, hsa-miR-146a-5p, hsa-miR-425-5p, and hsa-miR-126-5p, hsa-miR-146a-5p, hsa-miR-145-5p, hsa-miR-25-3p) associated with skeletal muscle mass, with high specificity and sensitivity, in postmenopausal osteoporotic women [7]. Recently, the pathological condition called osteosarcopenia, given by the coexistence of muscle impairment with bone loss and dysfunctional bone metabolism (i.e., osteopenia/osteoporosis), is rising interest in the scientific community [8]. It is known that this pathological condition is also associated with an increased fracture risk either dependently or independently from the risk of fall [9]. Since skeletal muscle and bone are closely linked, these tissues interact biomechanically and through a multilevel endocrine-like crosstalk mediated by factors of different nature (e.g., peptides, proteins, metabolites, RNA species). This crosstalk also integrates signals from other tissues, leading to a multiple-system response [10]. The progressive decline of this mechanical and biochemical interaction occurring in aging is associated with adverse outcomes that may compromise life quality in the elderly [8]. Given this interconnection, markers of the musculoskeletal function may be more informative than aspects of the single components.

MiRNAs, as post-transcriptional regulators involved in several biological processes, are expressed by all cell types with some tissue specificities and are released in biological fluids where they are easily detectable and measurable. The circulating fraction of miRNA represents the result of systemic signals generated in response to specific stimuli [11]. We have already demonstrated that circulating miRNAs associated with musculoskeletal function are highly responsive to physical activity [12, 13]. Moreover, these circulating miRNAs’ baseline levels significantly differ between physically active and sedentary subjects [14, 15].

Based on this background, we hypothesized that physical capability and walking ability in the elderly can affect the circulating level of miRNAs. This study aimed to evaluate the circulating miRNAs profile and biomarkers to describe the fragility phenotype and the reduced walking ability typical of the elderly. To this purpose, the profile of 754 miRNAs and a panel of metabolic and inflammatory biomarkers were investigated in two cohorts of elderly with different walking ability.

Material and methods

Study cohort

The study cohort includes hospitalized patients recruited within the Project REPORT AGE at INRCA, Ancona [16], with age > 75 years and able to walk autonomously (n = 832) (Fig. 1).

Fig. 1.

Diagram representing the process of patient enrollment

Based on their walking performance, the entire population has been divided into active and inactive groups. The active group (n = 154) includes elderly able to walk ≥ 1 km and to perform physical activity for 1–2 h, 3 days before the admission; the inactive group (n = 193) includes patients able to walk < 100 m and who have performed < 1 h physical activity, 3 days before the admission. The walking performance was assessed by self-report.

Three hundred and one patients have been excluded based on the presence of primary pathology: infective diseases, neoplasia, diabetes mellitus, endocrine dysfunctions, malnutrition, disorders in mineral metabolism, hematological disorders, alcohol or substances addiction, developmental disorders, neurodegenerative disorders, neuropathies, rheumatologic disorders, heart dysfunctions, chronic hepatic disease, kidney diseases, joint diseases, bone malformations and congenital anomalies, recent fractures, and recent trauma.

The final cohort includes 46 elderly (28 men, 18 women), divided into two groups: 23 active and 23 inactive patients. Study cohort characteristics are detailed in Table 1.

Table 1.

Characterization of the study cohort

	Total	Inactive	Active	p-value
	N = 46	N = 23	N = 23
Sex, n (%)				0.365
M	28(60.9%)	12(52.2%)	16(69.6%)
F	18(39.1%)	11(47.8%)	7(30.4%)
Age (years), mean ± sd	84 ± 5.2	85 ± 6.0	83 ± 4.3	0.277
Length of stay (days), mean ± sd	7.5 ± 3.4	7.3 ± 3.2	7.8 ± 3.6	0.658
3 + impaired ADLs, n (%)	6(13.0%)	6(26.1%)	0(0.0%)	0.022
3 + impaired IADLs, n (%)	16(34.8%)	16(69.6%)	0(0.0%)	< 0.001
5 + Drugs, n (%)	24(52.2%)	15(65.2%)	9(39.1%)	0.139
Blood parameters
eGFR (BIS1, mL/min), mean ± sd	48.1 ± 17.2	46.9 ± 18.1	49.2 ± 16.7	0.651
AST (U/L), mean ± sd	49.4 ± 100.8	20.3 ± 11.8	82.8 ± 141.8	0.041
ALT (U/L), mean ± sd	40.9 ± 92.6	16.8 ± 14.4	64.9 ± 126.9	0.078
RDW-SD (fL), mean ± sd	47.8 ± 5.7	48.8 ± 6.4	46.7 ± 4.8	0.220
RDW-CV (%), mean ± sd	14.5 ± 1.7	14.6 ± 1.8	14.4 ± 1.5	0.715
Hemoglobin (g/dL), mean ± sd	12.5 ± 2.1	12.2 ± 2.1	12.7 ± 2.2	0.360
Platelets (n/mm3), mean ± sd	220.6 ± 106.9	213.6 ± 113.3	227.6 ± 102.0	0.661
Albumin (g/L), mean ± sd	3.5 ± 0.5	3.6 ± 0.5	3.3 ± 0.4	0.128

Note: p-values are calculated with Fisher exact tests for categorical variables and Student’s t test for continuous variables

Abbreviations: ADL activities of daily living, AST aspartate transaminase, ALT alanine transaminase, eGFR estimated glomerular filtration rate, Hgb hemoglobin, IADL instrumental activities of daily living, RDW-SD red blood cell distribution width-standard deviation, RDW-CV red blood cell distribution width-coefficient of variation

The study protocol of the Report-Age project has been approved by the IRCCS INRCA hospital Ethics Committee, Ancona, ID number INRCA-01–2011. Trial Registration No. NCT01397682 on Clinical Trials.gov.

Blood collection and sample preparation

Venous blood samples were collected on the day of admission by antecubital venipuncture in tubes spray-coated with ethylenediaminetetraacetate dipotassium salt (K2EDTA, BD Vacutainer®, Becton Dickinson, Franklin Lakes, NJ, USA) in the non-fasted state, in subjects at rest in a sitting position for 10 min. Blood samples were inverted ten times and then centrifuged, according to the manufacturer’s instructions (1300 g, 10 min at room temperature, RT = 22 °C) to get plasma. Plasma aliquots were immediately frozen at − 80 °C until assayed.

miRNA profiling

RNA enriched in miRNAs was isolated using mirVana^TMParis™ Kit (Thermo Fisher Walthman, MA, USA) from a starting plasma volume of 350 µl, following the manufacturer’s protocol. The RNA enriched in miRNAs was then reverse-transcribed using the TaqMan MicroRNA Reverse Transcription (RT) Kit (Thermo Fisher) (40 cycles at 16 °C for 2 min, 42 °C for 1 min, 50 °C for 1 s, and one cycle at 85 °C for 5 min) and the Megaplex Primer Pool (Thermo Fisher) containing 754 miRNA primers divided in two sets of primers: Pool A and Pool B. For each sample, two different reverse-transcription have been performed, one with Pool A and one with Pool B primer set. Before miRNA profiling, reverse-transcribed miRNAs were pre-amplified using a set of specific primers, Megaplex PreAmp Primers Mix (Thermo Fisher), including Pool A and Pool B primers. Each pre-amplification reaction contained 7.5 µL of cDNA and 32.5 µL of PreAmp reaction mix that includes Megaplex PreAmp Primers Mix, TaqMan PreAmp master mix (Thermo Fisher), and nuclease-free water (one cycle at 95 °C for 10 min, 55 °C for 2 min, 72 °C for 2 min, 12 cycles at 95 °C for 15 s and 60 °C for 4 min, one cycle at 99.9 °C for 10 min). Each pre-amplification product was diluted 1:40 in 0.1 × TE pH 8.0 (Thermo Fisher). The diluted pre-amplification products were further diluted 1:1 in 2 × TaqMan OpenArray Real-Time PCR Master Mix (Thermo Fisher) prior to miRNA profiling with TaqMan OpenArray microRNA human panel (Thermo Fisher) on QuantStudio 12 K Flex Real-Time PCR system (Thermo Fisher). The OpenArray mix reaction was loaded into a 384-well plate and transferred on the OpenArray plate using a QuantStudio OpenArray AccuFill System. Data have been analyzed using GeneEx software. Only data with amplification quality score > 1.24 have been considered. A threshold cycle of 28 was applied, and miRNAs were considered undetected when quantitation cycle < 28 have been revealed in less than 35% of samples. Results have been calculated using the 2^−ΔΔCq, normalizing on the global mean. Before normalization on global mean, missing data have been imputed. The analysis was performed by considering separated miRNAs from Pool A and Pool B.

Validation

Validation was performed on those miRNAs found ≥ 2.5 fold up- or down-regulated in plasma samples of active and inactive groups. The identified deregulated miRNAs were isolated, as described above and reverse-transcribed using the TaqMan MicroRNA Reverse Transcription (RT) Kit (Thermo Fisher) using specific RT-primers for hsa-let7g-5p, hsa-miR-361-5p, hsa-miR-27a-3p, and hsa-miR-574-3p. qPCR was carried out on QuantStudio 12 K flex Real-Time PCR system (Thermo Fisher), using 2 × TaqMan Real-Time PCR Master Mix (Thermo Fisher). The thermal protocol was as follows: 10 min at 95 °C for polymerase activation and denaturation, 40 amplification cycles of 10 s at 95˚C and 1 min at 60 °C. Each sample was run in triplicates. The relative expression of each gene was calculated by the 2^−ΔΔCq method, using global mean to normalize data.

Bioinformatic analysis

An in silico analysis (November 2024) aimed to identify gene targets and the potential molecular pathways regulated by validated miRNAs was performed.

Target prediction of miRNAs was conducted using mirWalk ver. 3.0 (http://mirwalk.umm.uni-heidelberg.de/) [17], combining the other three prediction programs (TargetScan, TarBase, and miRDB). A score of 0.9 was chosen as a threshold. Only target genes listed by mirWalk and listed at least by two tools, always considering TarBase, the database of the experimentally validated miRNA-gene interactions, were considered. Panther 19.0 was used to perform Gene Ontology (GO) enrichment analysis, classifying target genes into biological process (BP), molecular function (MF), cellular component (CC), and pathway analysis (http://www.pantherdb.org/) [18–20].

Circulating biomarkers analysis

Circulating bone, skeletal muscle, inflammation, and metabolic biomarkers have been analyzed in the plasma of both active and inactive elderly as follows.

Bone alkaline phosphatase (BAP) (Quidel, San Diego, CA, USA) and cystatin C (BioTechne, Minneapolis, MN, USA) have been assayed by ELISA, displaying a sensitivity of 0.7 U/L and of 0.227 ng/mL, respectively.

C-terminal telopeptides of type I collagen (CTx-I) has been assayed with Serum CrossLaps® (ids, Boldon Business Park Tyne & Wear, UK), displaying a sensitivity of 0.3 ng/mL.

Creatinine has been assayed by a colorimetric detection kit (Thermo Fisher), with a sensitivity of 0.081 mg/dL.

Samples have been analyzed in duplicates and read on a Victor X3 (PerkinElmer, Waltham, MA, USA).

Adiponectin, leptin, and glycoprotein 130 (gp130) have been assayed by bead-based immunofluorescent assay (BioTechne), with a sensitivity of 148.0 pg/mL, 10.2 pg/mL, and 11.2 pg/mL, respectively.

The pro-inflammatory cytokines interleukin-6 (IL-6), interleukin-10 (IL-10), interleukin-13 (IL-13), and tumor necrosis factor-alpha (TNFα) have been assayed by high-sensitivity multiplex bead-based immunofluorescent assay (BioTechne), with sensitivities of 0.25 pg/mL, 0.38 pg/mL, 5.47 pg/mL, 3.39 pg/mL, and 0.62 pg/mL, respectively.

Myostatin/growth differentiation factor (GDF)8 and follistatin-like 1 protein (FSTL1) were analyzed with a Luminex® Myokine Magnetic Bead Panel (Millipore, Burlington, MA, USA), having a minimum detectable concentration of 163 and 548 pg/mL, respectively.

Samples have been analyzed in duplicates and read on a MagPix® Multiplex System (Luminex® Co., Austin, TX, USA). All assays were performed following the manufacturer’s instructions.

Creatinine and cystatin C were used to calculate the Sarcopenia Index (SI) [6, 21], as follows: (creatinine [mg/dL] / cystatin C [mg/L]) × 100.

Statistical analysis

Statistical analysis was performed with Prism® v6.01 (GraphPad Software Inc., Boston, MA, USA). The d’Agostino–Pearson omnibus normality test was performed for both miRNAs and circulating biomarkers to analyze data distribution. MiRNA profiling and cytokine levels between the two groups of active and inactive elderly were compared through a non-parametric Mann–Whitney test. Statistical analyses were considered significant when p-value < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001). Correlation analyses have been performed using Spearman’s correlation; only moderate and strong correlations (r ≥ 0.5) have been considered.

Two-sided Fisher’s exact test was used for the Gene Ontology (GO and pathway enrichment), and p-values were corrected by calculating the FDR. GO categories and pathways were considered significant when the corrected p-value was < 0.05.

Results

Effect of walking performance on miRNA profiling in elderly

The circulating level of 754 miRNAs (377 from panel A and from panel B) was analyzed in the plasma of active and inactive elderly (Fig. 2). The overall undetected miRNAs were 662 in the active cohort and 666 in the inactive cohort, while 661 miRNAs were undetected in both groups (Supplementary Table 1).

Fig. 2.

miRNA expression profile in active and inactive elderly. Heat map showing the differential expression of miRNAs between active and inactive elderly. The colour scale (red to blue) represents the relative abundance of each miRNA. The heatmap was generated using the available online tool (http://www.heatmapper.ca/expression/)

Six miRNAs were ≥ 2.5-fold upregulated (hsa-let-7g-5p, hsa-miR-193a-5p, hsa-miR-27a-3p, hsa-miR-361-5p, hsa-miR-574-3p, hsa-miR-151-3p) and 1 miRNA was ≥ 2.5-fold downregulated (hsa-let-7d-5p) in active compared to inactive elderly (Fig. 3). The fold changes of all miRNAs are shown in Table 2.

Fig. 3.

Circulating miRNAs differentially expressed between active and inactive elderly identified by OpenArray analysis. Scatter dot plot of differentially expressed miRNAs with a fold-change > ± 2.5 (p < 0.05), between active and inactive elderly. Comparison were performed using the non-parametric Man-Whitney test. The differences were considered statistically significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Table 2.

Fold change of plasma miRNAs in the active group compared to the inactive group. Fold change and relative p-value of the expression level of miRNA in plasma of active elderly compared to inactive elderly. Comparisons were performed using the non-parametric Mann–Whitney test. Differences were considered statistically significant when p < 0.05. miRNAs significantly deregulated are reported in bold

Panel A			Panel B
miRNA	Fold change	p-value	miRNA	Fold change	p-value
hsa-let-7a_000377	2.51	0.518	hsa-miR-126#_000451	1.15	0.538
hsa-let-7b_002619	0.69	0.352	hsa-miR-1260_002896	0.99	0.725
hsa-let-7d_002283	0.03	0.002	hsa-miR-1274A_002883	1.02	0.786
hsa-let-7e_002406	1.03	0.961	hsa-miR-1274B_002884	1.11	0.789
hsa-let-7g_002282	40.06	< 0.0001	hsa-miR-151-3p_002254	64.47	0.001
hsa-miR-103_000439	0.70	0.721	hsa-miR-30a-5p_000417	0.84	0.411
hsa-miR-106a_002169	0.96	0.496	hsa-miR-30d_000420	0.98	0.822
hsa-miR-106b_000442	1.02	0.856	hsa-miR-30e-3p_000422	1.57	0.982
hsa-miR-122_002245	2.07	0.739	hsa-miR-409-3p_002332	1.44	0.855
hsa-miR-126_002228	0.67	0.114	hsa-miR-572_001614	1.43	0.948
hsa-miR-130a_000454	1.10	0.983	hsa-miR-601_001558	1.44	0.530
hsa-miR-130b_000456	0.91	0.796	hsa-miR-657_001512	0.64	0.816
hsa-miR-132_000457	1.28	0.144	hsa-miR-720_002895	0.87	0.524
hsa-miR-139-5p_002289	0.93	> 0.9999	hsa-miR-942_002187	0.75	0.702
hsa-miR-142-3p_000464	0.86	0.119	rno-miR-7#_001338	1.02	0.615
hsa-miR-146a_000468	1.05	0.890
hsa-miR-146b_001097	0.82	0.432
hsa-miR-148a_000470	0.59	0.107
hsa-miR-150_000473	0.95	0.755
hsa-miR-152_000475	2.00	0.227
hsa-miR-15b_000390	1.12	0.471
hsa-miR-16_000391	0.98	0.994
hsa-miR-17_002308	1.01	0.675
hsa-miR-186_002285	0.90	0.735
hsa-miR-18a_002422	1.21	0.206
hsa-miR-191_002299	1.22	0.723
hsa-miR-192_000491	0.57	0.294
hsa-miR-193a-5p_002281	59.22	< 0.0001
hsa-miR-195_000494	0.80	0.715
hsa-miR-197_000497	0.25	0.810
hsa-miR-199a-3p_002304	1.43	0.222
hsa-miR-19a_000395	0.96	0.806
hsa-miR-19b_000396	1.00	0.772
hsa-miR-20a_000580	1.03	0.805
hsa-miR-20b_001014	0.88	0.716
hsa-miR-21_000397	1.18	0.612
hsa-miR-210_000512	1.00	0.711
hsa-miR-215_000518	1.43	0.193
hsa-miR-22_000398	1.36	0.912
hsa-miR-221_000524	1.10	0.553
hsa-miR-222_002276	1.63	0.054
hsa-miR-223_002295	1.12	0.430
hsa-miR-24_000402	1.16	0.167
hsa-miR-25_000403	1.16	0.303
hsa-miR-26a_000405	1.24	0.394
hsa-miR-26b_000407	1.28	0.263
hsa-miR-27a_000408	47.37	< 0.0001
hsa-miR-27b_000409	1.52	0.560
hsa-miR-29a_002112	1.22	0.339
hsa-miR-29c_000587	0.79	0.545
hsa-miR-30b_000602	1.01	0.744
hsa-miR-30c_000419	1.10	0.839
hsa-miR-320_002277	0.80	0.135
hsa-miR-328_000543	1.38	0.074
hsa-miR-331_000545	1.68	0.186
hsa-miR-335_000546	1.00	0.928
hsa-miR-342-3p_002260	1.04	0.987
hsa-miR-34a_000426	1.12	0.834
hsa-miR-361_000554	25.12	0.002
hsa-miR-367_000555	0.69	0.542
hsa-miR-374_000563	0.86	0.714
hsa-miR-423-5p_002340	0.79	0.353
hsa-miR-425-5p_001516	0.91	0.551
hsa-miR-483-5p_002338	0.70	0.323
hsa-miR-484_001821	1.07	0.805
hsa-miR-486_001278	0.85	0.337
hsa-miR-520e_001119	0.93	0.406
hsa-miR-532_001518	0.97	0.959
hsa-miR-551b_001535	2.41	0.536
hsa-miR-574-3p_002349	2.50	0.010
hsa-miR-590-5p_001984	0.56	0.993
hsa-miR-636_002088	1.19	0.251
hsa-miR-660_001515	0.82	0.196
hsa-miR-885-5p_002296	1.37	0.837
hsa-miR-92a_000431	0.95	0.553
mmu-miR-140_001187	1.03	0.910
mmu-miR-451_001141	1.03	0.628
mmu-miR-93_001090	0.95	0.405

Validation of the 7 identified miRNAs confirmed the deregulated expression of only 4 miRNAs: hsa-let-7g-5p, hsa-miR-27a-3p, hsa-miR-361-5p, and hsa-miR-574-3p. All these miRNAs were upregulated in the active group compared to the inactive group, as shown in Fig. 4.

Fig. 4.

miRNAs validation by qPCR analysis. Box plot of 7 identified miRNAs in plasma samples in active and inactive elderly. Comparison were performed using the non-parametric Mann–Whitney test. The differences were considered statistically significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

miRNA target prediction and Gene Ontology

Based on target prediction by mirWalk (ver. 3.0), a total number of 235 genes were expected to be targeted by the 4 validated miRNAs, whose circulating levels were affected by the effort (Supplementary Table 2).

To investigate the biological function of the target genes predicted for miRNAs modulated by the physical status, the Gene Ontology enrichment analysis was performed using Panther (Fig. 5, Fig. 6, Fig. 7, Supplementary Table 3).

Fig. 5.

Gene Ontology (GO) of target genes predicted for validated plasma miRNAs between active and inactive elderly: biological process. The first 20 most enriched categories are shown. GO was performed with Panther ver. 19.0 (http://www.pantherdb.org/), and the related statistic is shown in Supplementary Table 3

Fig. 6.

Gene Ontology (GO) of target genes predicted for validated plasma miRNAs between active and inactive elderly: cellular component. The first 20 most enriched categories are shown. GO was performed with Panther ver. 19.0 (http://www.pantherdb.org/), and the related statistic is shown in Supplementary Table 3

Fig. 7.

Gene Ontology (GO) of target genes predicted for validated plasma miRNAs between active and inactive elderly: molecular function. The first 20 most enriched categories are shown. GO was performed with Panther ver. 19.0 (http://www.pantherdb.org/), and the related statistic is shown in Supplementary Table 3

In the biological process (Fig. 5, Supplementary Table 3), the predicted target of validated miRNAs was mainly enriched in the regulation of metabolic process (GO: 0019222), regulation of biosynthetic process (GO: 0009889), intracellular signal transduction (GO: 0035556), regulation of gene expression (GO: 0010468), cellular response to endogenous stimulus (GO: 0071495), and mesenchyme development (GO: 0060485). In the cellular component (Fig. 6, Supplementary Table 3), the most enriched terms were related to the nucleus (GO: 0005634), organelle (GO: 0043226), cytoplasm (GO: 0005737), protein-containing complex (GO:0032991), and chromatin (GO: 0000785). In molecular function, predicted target genes were mainly enriched in protein binding (GO: 0005515), DNA-binding transcription repressor activity, RNA polymerase II-specific (GO: 000122), mRNA regulatory element binding translation repressor activity (GO: 0000900), transcription regulator activity (GO: 0140110), and G protein-coupled receptor activity (GO: 0004930).

Analysis of pathways

Targets predicted for validated miRNAs were enriched in 14 pathways, as shown in Fig. 8 and Supplementary Table 4. Targets were significantly enriched in pathways involved in EGF receptor signaling pathway (P00018), gonadotropin-releasing hormone receptor pathway (P06664), insulin/IGF pathway-mitogen activated protein kinase kinase/MAP kinase cascade (P00032), FGF signaling pathway (P00021), insulin/IGF pathway-protein kinase B signaling cascade (P00033), PI3 kinase pathway(P00048), inflammation mediated by chemokine and cytokine signaling pathway (P00031), interleukin signaling pathway (P00036), angiogenesis (P00005), Ras Pathway (P04393), integrin signaling pathway (P00034), p53 pathway feedback loops 2 (P04398), T-cell activation (P00053), and p53 pathway (P00059).

Fig. 8.

Pathway enrichment analysis of target genes predicted for validated plasma miRNAs between active and inactive elderly. The analysis was performed with Panther 19.0. All the enriched pathways are shown. Related statistic are shown in Supplementary Table 4

Elderly mobility-dependent modulation of plasma biomarkers

In order to define whether mobility affects bone, skeletal muscle, inflammation, and energy metabolism, a set of biomarkers was analyzed (Fig. 9): BAP, CTx-I, adiponectin, leptin, IL-6, IL-10, IL-13, TNFα, gp130, FSTL-1, and myostatin. gp130 and IL-10 significantly differed between the two groups: gp130 was downregulated (p = 0.035), while IL-10 was upregulated (p = 0.043) in active compared with inactive elderly. IL-1β and IL-13 were undetectable in both groups.

Fig. 9.

Bone, skeletal muscle, inflammation, and energy metabolism biomarkers levels in plasma of active and inactive elderly. Level of glycoprotein 130 (gp130), leptin, interleukin-6 (IL-6), interleukin-10 (IL-10), tumor necrosis factor α (TNFα), bone alkaline phosphatase (BAP), cross-linked C-telopeptide of type I collagen (CTx-I), adiponectin, myostatin, and follistatin like I (FSTL1), in plasma samples of active and inactive elderly. Data are expressed as min to max. Comparison were performed using the non-parametric Man-Whitney test. The differences were considered statistically significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001)

In the active group, IL-6 was positively associated with IL-10 and TNFα (r = 0.66, p = 0.003; r = 0.50, p = 0.034, respectively), while it was negatively associated with myostatin (r = − 0.61, p = 0.007) (Table 3). Negative correlations have been observed also between CTx-I and leptin (r = − 0.52, p = 0.027) and myostatin and FSTL1 (r = − 0.61, p < 0.001) (Table 3).

Table 3.

Spearman’s correlation analysis among cytokines in active and inactive groups. The table shows correlation coefficients: correlation for inactive group are highlighted in grey; statistically significant (p < 0.05) correlation coefficients are reported in bold. gp130 glycoprotein 130, IL-6 interleukin-6, IL-10 interleukin-10, TNFα tumor necrosis factor α, BAP bone alkaline phosphatase, CTx-I cross-linked C-telopeptide of type I collagen, FSTL1 follistatin-like 1

		ACTIVE
		gp130	leptin	IL-10	IL-6	TNFα	Adiponectin	BAP	CTX-I	FSTL1	Myostatin	Sarcopenia index
INACTIVE	gp130		0.22	− 0.24	− 0.08	− 0.27	0.23	− 0.10	− 0.12	0.18	0.03	− 0.12
	leptin	− 0.26		− 0.27	− 0.44	− 0.06	− 0.29	0.04	− 0.52	0.33	0.29	0.03
	IL-10	0.27	− 0.21		0.66	0.13	0.21	− 0.11	0.06	− 0.08	− 0.24	− 0.23
	IL-6	0.48	− 0.18	0.26		0.50	0.20	− 0.21	0.27	− 0.35	− 0.61	− 0.21
	TNFα	0.04	0.26	0.49	0.38		0.10	− 0.17	0.19	− 0.42	− 0.44	− 0.40
	Adiponectin	0.53	− 0.29	0.20	0.28	− 0.08		0.02	0.27	0.20	− 0.05	− 0.56
	BAP	− 0.02	0.18	− 0.30	0.04	− 0.24	0.17		0.05	0.06	0.13	0.11
	CTX-I	0.45	− 0.25	− 0.19	0.55	− 0.10	0.13	0.41		− 0.08	− 0.29	− 0.15
	FSTL1	− 0.17	0.21	− 0.34	0.26	− 0.06	− 0.12	0.51	0.54		0.81	0.11
	Myostatin	− 0.05	0.38	− 0.22	0.21	0.05	− 0.29	0.31	0.30	0.67		0.24
	Sarcopenia index	− 0.33	0.09	0.39	− 0.66	0.03	− 0.27	− 0.15	− 0.65	− 0.33	− 0.14

In the inactive group, positive associations have been observed between gp130 and adiponectin (r = 0.53, p = 0.016), IL-6 and CTx-I (r = 0.55, p = 0.014), FSTL1 and CTx-I (r = 0.54, p = 0.019), and FSTL1 and BAP (r = 0.51, p = 0.030) (Table 3).

From target prediction analysis, the only cytokine which was predicted to be target of a miRNA was the gene encoding for IL13 found to be target of hsa-let-7g-5p.

To investigate on the association of cytokines with validated miRNAs, correlation analysis have been performed. Results have shown a positive correlation between hsa-miR-27a-3p and both BAP and FSTL-1, and between TNF-α and both hsa-miR-574-3p and hsa-let-7g-5p in inactive group. A negative correlation was observed between adiponectin and both hsa-miR-27a-3p and hsa-miR-574-3p in active group (Supplementary Table 5).

Sarcopenia index

Plasma level of creatinine and cystatin C was measured to define the sarcopenia index (SI), calculated as (creatinine [mg/dL] / cystatin C [mg/L]) × 100 [21]. Despite the concentrations of both creatinine and cystatin C remained unchanged between active and inactive elderly (p = 0.223, p = 0.313, respectively), SI resulted significantly higher in active (117.10 ± 46.57) than inactive elderly (91.16 ± 20.32) (p = 0.049) (Fig. 10).

Fig. 10.

Sarcopenia Index in active and inactive elderly. Sarcopenia index (SI) has been calculated as (creatinine [mg/dL] / cystatin C [mg/L]) × 100, according to [21]. Data are shown as mean to max. Comparisons were performed using the non-parametric Man-Whitney test. Statistical analyses were considered significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001)

SI correlated positively with adiponectin in the active group (r = 0.56, p = 0.017) and negatively with CTx-I (r = − 0.65, p = 0.005) and IL-6 (r = − 0.66, p = 0.005) in the inactive group (Table 3); SI correlated with none of validated miRNAs (Supplementary Table 5).

Discussion

This study provides the first attempt to explore circulating miRNA profile and metabolic and inflammatory biomarkers to define walking behavior in the elderly. One of the main characteristics of the aging phenotype is represented by the impairment of the musculoskeletal apparatus, which leads to diminished mobility and physical function and, consequently, increased risk of falls [22, 23]. Preserving the fitness capacity (e.g., strength, endurance, agility, and balance) and walking ability is crucial to performing everyday-life activities and maintaining a high quality of life and independence in the community [3, 24]. The impaired physical performance, with the consequent mobility loss, is associated with increased rates of morbidity, disability, and mortality [3]. Several studies have attempted to define the fitness level or threshold, based on the measurement of the physical performance required for physical independence in later life, to determine the risk of premature mobility loss [24–27]. However, identifying circulating biomarkers may represent a valid and easy-to-perform alternative for assessing physical capabilities, fitness level, and walking ability in the elderly to optimize appropriate intervention strategies. Moreover, despite the decline in the amount and intensity of physical activity and mobility with aging has been well debated, the molecular mechanisms involved in these impairments and the potential ability to predict them are not entirely understood.

In this context, we selected a cohort of active elderly subjects and their age- and sex-matched inactive counterparts to identify miRNAs that may help to clarify the molecular mechanism of aging and that may be studied as potential biomarkers of physiology in the elderly.

From a panel of 754 miRNA analyzed in plasma samples obtained from these subjects, 7 miRNA resulted differentially expressed between the two cohorts; of these, only 4 were further validated: hsa-let-7g-5p, hsa-miR-27a-3p, hsa-miR-361-5p, and hsa-miR-574-3p. None of the differentially expressed miRNAs has been previously identified as a predictor of fall risk or associated with walking performance. However, most of these miRNAs are related to skeletal muscle, bone, or adipose tissue homeostasis, metabolism, and function. The best-characterized miRNA, among those found differentially expressed in our setting, is miR-27a-3p. miR-27a regulates myofiber metabolism by improving lipid use while reducing glycogen breakdown and, consequently, glycolysis [28]. Myostatin is a target of miR-27a-3p [29, 30]; indeed, the lower the level of miR-27a in fast fibers compared to slow fibers, the higher the expression of myostatin [31]. MiR-27a is also involved in the crosstalk between adipose tissue and skeletal muscle. Exosome-derived miR-27a, released from adipocytes, induces insulin resistance in C2C12 skeletal muscle cells by repressing the peroxisome proliferator-activated receptor γ (PPARγ) [32]. Additionally, it was found downregulated in serum of exercised-obese mice, which associated with the promotion of adipose tissue browning [33]. MiR-27a has a role also in bone metabolism, and it may be crucial in osteoporosis: it is involved in the differentiation of mesenchymal stem cells, promoting osteogenic differentiation rather than adipogenic differentiation, by targeting myocyte enhancer factor 2C (mef2c) [34], phosphatidylinositol 3-kinase (PI3K) [35], and secreted frizzled-related protein 1 (Sfrp1) [36]. Moreover, miR-27a has been observed to inhibit osteoclastic differentiation of bone marrow-derived macrophages [37, 38]. Another miRNA, miR-361-5p, upregulated in our study, may be associated with bone health and function since it has been upregulated in postmenopausal osteoporotic women’s serum compared with their healthy counterparts [39]. These findings suggest these miRNAs are potentially involved in skeletal muscle and bone tissue homeostasis, further suggesting a potential role in impaired walking performance and physical capability in the elderly. Indeed, age-associated alterations in body composition, particularly in skeletal muscle and bone, as in sarcopenia and osteoporosis, are among the leading causes of reduced physical function and increased disability in the elderly [40].

The differences in the walking performance in the active and inactive elderly cohorts may be explained by differences in body composition, particularly in skeletal muscle mass, as evidenced by the sarcopenia index [6, 21], which is higher in the active cohort. Although this index differ between the two groups, it was not significantly correlated to any of the validated miRNA.

Besides skeletal muscle weakness and mass loss, infiltrating adipose tissue and fat mass accumulation represent two variables that negatively influence functional performance in the elderly [41, 42]. Conflicting results arise from previous studies on the role of bone mineral density (BMD) as a predictor of physical performance; however, results on the positive correlation of BMD with physical performance have shown a more substantial relationship in women than men [40]. In our study, no significant differences arose in the circulating concentration of bone and adipose tissue metabolites between active and inactive subjects, although different relationships among these molecules characterized the two groups. Indeed, SI negatively correlated with adiponectin in the active group and CTx-I in the inactive group. A recent study has shown that plasma adiponectin levels are negatively associated with muscle mass in middle-aged and elderly individuals [43], supporting our results. The negative correlation between CTx-I and SI highlighted the close mechanical and functional relationship between skeletal muscle loss and osteoporosis, characterized by increased bone resorption during aging [44]. In the active group, the bone resorption biomarker CTx-I negatively associated with leptin, an adipocyte-derived hormone involved in regulating appetite and energy expenditure [45]. Although no association between the plasma level of these two biomarkers has been observed in inactive subjects, a study by Roux C. et al. found a negative correlation between serum leptin and urinary CTx-I in postmenopausal healthy women [46]. The paradoxical positive association between the two antagonists, FSTL1 and myostatin, observed in both groups might be explained by age-associated inflammatory processes.

The homeostasis of bone, skeletal muscle, and adipose tissue is linked to immune system activity, and the relationships among cytokines and inflammatory factors observed in the two groups may represent some aspects of this crosstalk. Although adiponectin is a multifunctional cytokine with anti-inflammatory properties [47], our study found it positively associated with the IL-6 receptor subunit, gp130, in the inactive group. In the same group, IL-6 also positively associated with CTx-I. This correlation might be explained by the role of IL-6 in stimulating osteoclast differentiation and activity and, thus, the consequent stimulus towards bone resorption [48]. The same explanation could be argued for the positive correlation of CTx-I with FSTL1 in the inactive group. Although anti-catabolic, FSTL1 was observed to promote osteoclast formation via NF-κB and MAPK signaling pathways [49]. FSTL1 is a pro-inflammatory factor and can act as an osteogenic suppressor in inflammatory conditions by inhibiting the osteogenic differentiation of BMSCs [50]. This contrasts with our results on the positive correlation between FSTL1 and BAP observed in the inactive group.

Immunosenescence, i.e., age-associated alterations of immune function, represents another disabling aspect occurring in aging and is characterized by the increased release of inflammatory cytokines, such as IL-6, TNFα, and C-reactive protein, and decreased levels of anti-inflammatory mediators [51–53]. Changes in the circulating profile of pro-inflammatory cytokines have been associated with a declining ability to perform daily activities in aging [54, 55]. In our study, IL-6 correlated negatively with SI in the inactive group, indicating a pro-inflammatory, pro-catabolic potential. Additionally, IL-6 positively correlated with TNFα in the active group. However, while both IL-6 and TNFα circulating levels did not differ in the two groups, in the active cohort, IL-6 positively correlated with IL-10, thus supporting its anti-inflammatory potential when expressed and secreted in a physically active behavior [10]. This is further supported by the up-regulation of IL-10, an anti-inflammatory cytokine, and the down-regulation of gp130, a co-receptor of IL-6 receptor (IL-6R), in the active compared with the inactive elderly. IL-10 suppresses the systemic inflammatory response by inactivating macrophages through the blockage of antigen presentation and the release of inflammatory cytokines, such as IL-6, TNFα, and IL-1β [51]. During physical activity, the circulating level of IL-10 increases in response to IL-6 to prevent inflammation and tissue damage by inhibiting the secretion of pro-inflammatory cytokines, such as IL-1, TNFα, IL-8, and MIP [56]. In the active cohort, although reasonably low walking ability due to age, the higher circulating levels of IL-10 may limit the age-associated chronic low-grade inflammation.

The ubiquitous glycoprotein gp130 is a co-receptor of IL-6R, responsible for the intracellular signaling after association of the IL-6/IL-6R complex in both classical and trans-signaling. Besides the membrane receptor, a circulating soluble IL-6R (sIL-6R) can bind IL-6, thus activating gp130 in all cells [57]. Moreover, gp130 has a soluble portion that can bind and inactivate the IL6/sIL-6R complex, thus exerting an anti-inflammatory activity [58]. However, both sIL-6R and sgp130 are more stable in circulation than IL-6, and their concentration fluctuates much less than those of the cytokine, giving a more accurate indication on IL-6 pathway activation [59]. Overall results suggest the two altered cytokines between active and inactive elderly, i.e., IL-10 and gp130, may have a role in modulating walking performance in aging. However, no association of both IL-10 and gp130 with validated miRNAs has been observed.

Changes in the inflammatory cytokines profile, and thus in the overall immune response, are reflected in the bioinformatic analysis, where cytokines represent one of the enriched pathways associated with the differentially expressed miRNAs. Other enriched pathways relevant for bone and skeletal muscle homeostasis and, potentially, for walking performance and physical capability involve Ras and p53. Ras proteins belong to the GTPases superfamily that regulates multiple cellular processes like proliferation, differentiation, apoptosis, senescence, and metabolism in response to different stimuli, such as cytokines, hormones, and growth factors [60]; p53 is a tumor suppressor involved in the regulation of cell cycle, apoptosis, and senescence. Since enrichment analysis showed the significant involvement of both Ras and p53 signaling in analyzed samples, it can be argued that, in aged bone and skeletal muscle, Ras and p53 may be involved in cellular senescence, apoptosis, oxidative stress, and altered signaling pathways that collectively contribute to the aging phenotype of cells. Moreover, Ras is a downstream effector of the IL-6 pathway. The dimerization of gp130, upon binding with IL-6/IL-6R, activates the intracellular signaling via JAK/STAT. Phosphorylated tyrosine residues on gp130 by JAK1 recruits the phosphatase SHP-2, which, in turn, activates the Ras-Raf-MAPK signaling pathway [61].

Conclusion

To our knowledge, our study represents a first attempt to identify circulating miRNAs and metabolic and inflammatory biomarkers associated with the walking performance of elderly subjects, with several implications including potential identification of new biomarkers. Our results highlight differences in the miRNA profile between two cohorts characterized by different walking capabilities. In strict association with this, the two cohorts also differed in skeletal muscle mass and inflammatory profile.

However, this study was limited to hospitalized patients and further parameters to test mobility are needed to define better the association of circulating miRNAs with the walking performance of these subjects. Moreover, in this study, the function of differentially expressed miRNAs was derived from the bioinformatics analysis and the available evidence from the literature, while a functional analysis should be needed to confirm the role of the identified miRNAs in tissue homeostasis, especially for those tissue-relevant in walking and physical capability, such as skeletal muscle, bone, and adipose tissue.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors contribution

Conceptualization: MP, RM, FL, GL. Methodology: MF, MP, MDR, RM, VS. Formal analysis and investigation: MF, GL. Writing—original draft preparation: MF, GL. Writing—review and editing: MP, MDR, MR, VS, MG, LG, AMM, FL, BG. Funding acquisition: BG, FL, GL. Supervision: GL.

Funding

This study was supported by the National Science Center (Poland), project no. NCN 2018/29/B/NZ7/02094; by the Italian Ministry of Health, Project RCR-2022–23682286—Rete AGING—Next Generation Promising (NGP): a new network approach to frailty, multimorbidity and age-related disease in the Era of Precision Medicine—Fondo progetti Reti—Ricerca corrente 2022; and by the Italian Ministry of Health – “Ricerca Corrente.”

Data availability

The dataset supporting the conclusions of this article is available in the Zenodo repository (https://zenodo.org/records/11634917).

Declarations

Ethics approval

All procedures performed in this study were in accordance with the ethical standards of the Helsinki declaration and approved by the Ethics Committee of the IRCCS INRCA hospital, Ancona (INRCA-01–2011).

Consent to participate

Written informed consent was obtained from all individual participants included in the study.

Competing interests

The authors declare no competing interests.