Physical exercise is associated with a reduction in plasma levels of fractalkine, TGF-β1, eotaxin-1 and IL-6 in younger adults with mobility disability

Mobility disability (MD) refers to substantial limitations in life activities that arise because of movement impairments. Although MD is most prevalent in older individuals, it can also affect younger adults. Increasing evidence suggests that inflammation can drive the development of MD and may need to be targeted for MD prevention. Physical exercise has anti-inflammatory properties and has been associated with MD prevention. However, no studies to date have examined whether exercise interventions affect the peripheral inflammatory status in younger adults with MD. To this end, we used blood samples from young and middle-aged adults with MD (N = 38; median age = 34 years) who participated in a 12-week intervention that included aerobic and resistance exercise training. A pre-post assessment of inflammatory biomarkers was conducted in plasma from two timepoints, i.e., before the exercise trial and at follow-up (3–7 days after the last exercise session). We successfully measured 15 inflammatory biomarkers and found that exercise was associated with a significant reduction in levels of soluble fractalkine, transforming growth factor beta 1 (TGF-β1), eotaxin-1 and interleukin (IL) 6 (corrected α = 0.004). We also found significant male-specific effects of exercise on (i) increasing IL-16 and (ii) decreasing vascular endothelial growth factor-A (VEGF-A). In line with our results, previous studies have also found that exercise can reduce levels of TGF-β1, eotaxin-1 and IL-6. However, our finding that exercise reduces plasma levels of fractalkine in younger adults with MD, as well as the sex-dependent findings, have not been previously reported and warrant replication in larger cohorts. Given the suggested role of inflammation in promoting MD development, our study provides additional support for the use of physical exercise as a treatment modality for MD.

Introduction samples from young and middle-aged adults with MD. Here, we present the results from the successful pre-post measurements of 15 inflammatory biomarkers, some of which have not previously been well-studied in relation to physical exercise.

Baseline characteristics of participants with mobility disability
The baseline characteristics of the study participants with MD (N = 38) are presented in Table 1. The participants reached moderate-to-vigorous physical activity levels corresponding to 43 (31-56), 48 (40-63) and 48 (35-61) min/day (median [25 th -75 th percentiles]) as measured on the 1 st , 6 th and 12 th week of the intervention, respectively. As shown in Table 2, there was no change in metabolic markers in the study participants following the exercise intervention.

Differences between males and females in inflammatory biomarker changes following exercise
As shown in Fig 2, when we stratified for sex in our analyses, we found significant differences in pre-post change between sexes for IL-16 (p = 0.0053), sIL-2Rα (p = 0.013), TRAIL (p = 0.012), and VEGF-A (p = 0.0032). Post-hoc assessments showed that for males, but not for females, IL-16 levels increased (p = 0.00098) and VEGF-A levels decreased (p = 0.0029) following exercise. Conversely, females but not males, had a trend toward decreased levels of sIL-2Rα (p = 0.0065) following exercise.

Discussion
The anti-inflammatory properties of exercise [19,20,26,27], which are observed "chronically", i.e., following regular training [28], are of particular importance to MD because the presence of inflammation has emerged as a possible biological mechanism that can drive MD

PLOS ONE
Physical exercise, inflammatory biomarkers, and mobility disability development [12][13][14][15][16][17]. In line with these "chronic" anti-inflammatory properties of regular exercise, our measurements (taken 3-7 days following the last exercise session of a 12-week intervention program) suggested that regular aerobic and resistance training led to an improvement in the peripheral inflammatory status of younger adults with mild-to-moderate MD. Specifically, following the exercise intervention, we found a significant reduction in plasma levels of four inflammatory biomarkers: (i) sFKN, (ii) TGF-β1, (iii) eotaxin-1 and (iv) IL-6. Moreover, (v) sVCAM-1 levels showed a trend for exercise-associated reduction, and sex-stratified analyses indicated different responses in males, compared to females, with postexercise increases in IL-16 and decreases in VEGF-A levels in males only.
(i). Fractalkine (FKN) is a chemokine, primarily expressed by vascular endothelial cells, which exists in a cell-surface-bound form and a soluble form (sFKN). FKN is made soluble by metalloproteinase-dependent cleavage of its membrane-bound form [29]; a Exercise is associated with a reduction in sFKN, TGF-β1, eotaxin-1 and IL-6. The differences in plasma inflammatory biomarker levels between baseline and follow-up (i.e., before and after the 12-week physical exercise intervention program), were determined using Wilcoxon signed-rank tests. For CRP, the dotted red (3000 ng/ml) and the dotted green (10 000 ng/ml) lines represent the cut-offs for low-grade and mild-acute inflammation, respectively. Abbreviations: CRP = C-reactive protein, sFKN = soluble Fractalkine, GRO-α = Growth-regulated oncogene-alpha, IL-12/IL-23p40 = Interleukin (IL)-12/IL-23p40, IL-16 = Interleukin-16, IL-18 = Interleukin-18, sIL-2Rα = soluble Interleukin-2 receptor subunit alpha, IL-6 = Interleukin-6, SAA = serum amyloid A, sICAM-1 = soluble Intercellular adhesion molecule-1, sVCAM-1 = soluble Vascular cell adhesion molecule-1, TGF-β1 = Transforming growth factor beta 1, TRAIL = Tumor necrosis factor-related apoptosis-inducing ligand, VEGF-A = Vascular endothelial growth factor A.  (16). Thus, the reduction in plasma levels of sFKN in MD participants, which was observed in our study, may indicate an exercise-induced reduction in vascular inflammation. While aerobic exercise is known to reduce vascular inflammatory activity [23,34], only few studies have previously investigated the effect of physical exercise on levels of sFKN. These previous studies, however, found that 12 months of combined aerobic and resistance training, or 12 weeks of balance exercise programs, were associated with slightly elevated circulating levels of serum sFKN in either patients with type 2

PLOS ONE
Physical exercise, inflammatory biomarkers, and mobility disability diabetes and coronary artery disease, or in those without a diagnosis or with Parkinson's disease [35,36]. Elevated levels of sFKN have also been found in serum of healthy individuals immediately following maximal exercise on a treadmill [37]. However, all of these previous studies measured sFKN levels in serum, which does not contain platelets, and FKN has a known role in mediating platelet adhesion [38]. Thus, since we measured sFKN levels using plasma, our findings of exercise-induced reductions in sFKN may indicate specificity to this liquid portion of blood, which warrants future investigation.
(ii). TGF-β1 is a multifunctional cytokine with an immune-modulatory role by inhibiting lymphocyte proliferation, stimulate IgA class switching of B cells and regulate tissue repair. Together with Il-6 it promotes the differentiation of pro-inflammatory CD4 + cells expressing IL-17. Higher plasma TGF-β1 levels have been associated with atherosclerosis, hypertension, hyperlipidemia and diabetes [39]. Moreover, there is evidence suggesting that increased TGF-β1 protein activity in skeletal muscle contributes to an impaired exercise response through inhibition of mitochondrial regulators and insulin signalling [40]. The expression of TGF-β1 is induced by mechanical loading in various tissues, including bone and cardiac tissue [41]. A previous study, using healthy individuals, found that TGF-β1 plasma levels increased transiently after two weeks of heavy resistance training, but that these chronic point-wise levels dropped below baseline following another two weeks of exercise [28]. These results are in agreement with our findings showing reduced TGF-β1 plasma levels after 12 weeks of regular exercise.
(iii). Eotaxin-1 is a chemokine expressed by the vascular smooth muscle cells located beneath the vascular endothelium, and also by, e.g., bronchial epithelial cells and dermal fibroblasts [42]. Eotaxin-1 expression is induced by proinflammatory cytokines (e.g., TNF-α) and it acts as a chemoattractant mainly for eosinophils, basophils, mast cells and Th2 lymphocytes. Higher plasma eotaxin-1 levels have been associated with asthma and allergy disorders but also neurodegeneration and accelerated aging [42]. In line with our findings of reduced eotaxin-1 levels following the 12-week exercise intervention, obese individuals who underwent eight-to-twelve week programs of aerobic and/or resistance exercise were also found to have reduced levels of eotaxin-1 [43][44][45].
(iv). IL-6 is a pleiotropic molecule with multiple immunologic and metabolic effects, which is both a pro-inflammatory cytokine and an anti-inflammatory myokine [46]. Higher levels of IL-6 have been found to predict, or to be associated with, physical impairment, disability and mortality in older persons [13,15,17,47,48]. Muscle-derived IL-6 is known to be released into the bloodstream during exercise, and is immediately followed by a transient increase in levels of anti-inflammatory IL-1 receptor antagonist (IL-1RA) and IL-10 [49][50][51]. Plasma levels of these three interleukins are back to pre-exercise levels in about 2 hours after a single exercise bout [52]. In line with our findings of reduced IL-6 levels following the 12-week exercise intervention, regular aerobic exercise has consistently been reported to decrease IL-6 levels in healthy adults [53,54]. The latter effect has also been proposed to be moderated by a genetic polymorphism in the IL6 gene promoter, which regulates the beta-adrenergic activation of IL6 transcription, in individuals with impaired glucose tolerance [55,56].
(v). One of the adhesion molecules measured, sVCAM-1, showed a trend for post-exercise reduction. sVCAM-1 levels correlate with levels of membrane-bound VCAM-1, which is expressed specifically by vascular endothelial and smooth muscle cells and is central to leukocyte-endothelial cell adhesion and leukocyte extravasation into the surrounding tissue [57,58]. Elevated levels of sVCAM-1 have been associated with vascular inflammation, endothelial dysfunction and atherosclerosis, and-similar to our findings -sVCAM-1 has been found to be reduced by low-to-moderate aerobic exercise in a variety of study populations, but without any effect of resistance exercise [34].
Although the mechanistic connections between fractalkine, TGF-β1, eotaxin-1 and IL-6 were not explored in our study, some assumptions can be made based on previous literature and our correlation analyses. For instance, TGF-β1 has previously been found to stimulate fractalkine expression in human cancerous cell lines and rat microglia [59,60]. This could indicate that the exercise-induced reductions in levels of fractalkine found in our study, are mediated by a primary effect of exercise on lowering TGF-β1 levels. However, we found no significant correlations between plasma fractalkine and TGF-β1 at either baseline or follow-up (S1 Fig), or between fractalkine level change (follow-up-baseline) and change in TGF-β1 (S3A Fig). Moreover, other studies found that TGF-β1 actually reduced fractalkine mRNA and protein expression in glioma cells [61], which supports a complex interplay between the levels of these two molecules that may be dependent on the tissue or disease of interest. Similarly, studies using smooth muscle cells or fibroblast cultures, found that TGF-β1 (together with IL-13) can stimulate the expression of eotaxin-1 [62,63]. Indeed, we detected a correlation between change in TGF-β1 and change in eotaxin-1 levels (r = 0.33, p = 0.046, S3B Fig). Along the same lines, cell culture experiments have demonstrated a crosstalk between TGF-β1 and IL-6, with TGF-β1 controlling aspects of early innate inflammation through promotion of IL-6 secretion and/or induction of IL-6 protein and mRNA levels [64][65][66], but without any significant correlations between these two molecules found in the plasma samples of our study. Interestingly, however, studies have found that plasma fractalkine levels correlate positively with plasma levels of IL-6 [67], which is in line with our correlation data for these two molecules that were significant at baseline (Spearman's r = 0.42; S1 Fig), including the change in fractalkine that correlated with the change in IL-6 levels (r = 0.39, p = 0.015, S3C Fig). Indeed, previous studies using CD16+ monocytes suggested that membrane-bound fractalkine stimulates the production of IL-6 [68], providing a possible mechanistic link between the levels of these two molecules, which warrants additional investigation in physical exercise settings.
Besides our primary pre-post analyses, we also conducted sex-stratified analyses that revealed significant increases in IL-16, and significant reductions in VEGF-A, in male individuals only. IL-16 is known to stimulate the expression of pro-inflammatory cytokines [69] and has been associated with inflammatory diseases, such as rheumatoid arthritis [70]. By contrast, there are also reports suggesting that IL-16 may possess protective effects against other inflammatory disorders, such as atherosclerosis [71,72]. However, to our knowledge, no previous studies have reported sex-specific effects of physical exercise on IL-16 levels. VEGF-A, on the other hand, is produced by many different cell types, including vascular endothelial cells, macrophages and cardiomyocytes, and has a key role in mediating angiogenesis and vasodilation. Although VEGF-A is essential for the cardiovascular system, aberrantly elevated levels have been implicated in inflammatory processes of cardiovascular disease and pathology following atherosclerosis [73], as well as in inflammatory diseases, such as inflammatory bowel disease [74] and psoriasis-like disease [75]. Similar to our findings, plasma VEGF-A levels have previously been found to decrease for at least 2 hours, following a single bout of exercise, in healthy men [76]. By contrast, in overweight postmenopausal women, there was no effect of regular moderate-to-vigorous aerobic exercise on plasma levels of VEGF-A [77].

Limitations
There are a number of limitations to this exploratory study that need to be acknowledged. (i) The low number of individuals included in our study (N = 38) render the significant findings preliminary and exploratory in nature, and all results need to be interpreted with caution prior to replication in additional MD cohorts. Moreover, for those analytes that we could not detect a significant pre-post exercise change, we cannot exclude the presence of true small effect sizes. Finally, the number of participants was even lower in the sex-stratified analyses. Hence, only relatively large effect sizes could have been detected. (ii) The intervention study, from which the current data and samples derived from, included an additional N = 50 individuals with MD who donated blood but were not included in our study. Comparisons of all known baseline characteristics between the two groups showed only slight sex differences, i.e., with some male overrepresentation in our group (p = 0.046, S2 Table). (iii) The study lacks long-term post-intervention data on MD status, repeated sampling without active intervention in MD, as well as a control group without MD. (iv) The blood samples were collected in a time window of 3-7 days after the last exercise session. The transient inflammatory-resembling responses often observed "acutely", i.e., during or immediately after an exercise bout have been reported to last no longer than up to 24-36 h [37,52]. However, we cannot exclude the possibility of longer acute effects in some individuals. (v) As indicated by the levels of CRP in Fig 1, and in line with the hypothesis of inflammation in MD [12-17], a significant portion of the participants in our study had lowgrade inflammation. However, CRP was not among the analytes that was significantly affected by exercise. Moreover, due to the lack of standardized reference levels for most of the remaining analytes, we cannot evaluate to which extent the participants' levels deviated from normal ones. (vi) Although only few of the study's MD participants reported a chronic disease, we cannot exclude the presence of undetected pathologies that could influence biomarker levels.

Conclusion
Our findings support a beneficial role of regular moderate aerobic exercise, combined with resistance training, on both CRF [24] and peripheral inflammatory status in young and middle-aged adults with MD. Importantly, the post-exercise reduction in plasma sFKN levels has to our knowledge not previously been reported, but is in agreement with reduced plasma levels of vascular adhesion molecules reported by others [34], including the trend for reduced sVCAM-1 levels observed in our study. The post-exercise reductions that we found in plasma levels of TGF-β1, eotaxin-1 and IL-6, are in agreement with findings from other study groups [28, [43][44][45]53,54], although there is to our knowledge no report on these markers in younger adults with MD. The male-specific exercise effects that we detected in levels of IL-16 and VEGF-A are also novel for MD subjects, although sex-specific effects on VEGF-A are supported by previous reports in non-MD groups [76,77]. However, given the aforementioned limitations of the present pilot study, future studies are warranted to confirm these findings and translate them into outcomes of clinical significance. Moreover, future studies need to take into consideration putative confounding parameters such as disease comorbidity and medication. Nonetheless, given the proposed role of inflammation in promoting MD development [12][13][14][15][16][17], our study provides support for the continued evaluation of physical exercise as a preventive and treatment modality for MD.

Participants
Study participants were the first 38 subjects (of the total N = 110) that participated in a 12-week parallel-group randomized controlled trial that took place between May 2018 -December 2018, and which was designed to compare the effects of an app-based program with a supervised exercise and health program on levels of physical activity and changes in CRF in individuals with mild-to-moderate MD [24]. As there was no difference in effects between the two motivation arms on amount of psychical activity or change in CRF or body composition, we did not account for intervention arm in this study. Participants were recruited from rehabilitation, primary care and occupational health care centres in Stockholm, Sweden. The eligibility criteria for study participation were as follows: male or female aged 18-45 years old with any mobility-related problems affecting everyday life, such as problems with transportation, work, dressing, and performing household or personal hygiene tasks, experienced during the three years prior to trial enrolment. Exclusion criteria included being bound to a wheelchair or having a medical condition preventing moderate-intensity walking (for a detailed description of the trial see also Berglind D et al. [78]). Only a few of the 38 participants in this study reported a chronic disease diagnosis. Of the 38, 19 were restricted in running, heavy lifts, and demanding sports, and 2 were restricted in household activities such as vacuum cleaning, walking in forest or gardening. Comparisons of the baseline characteristics of this study's participants (N = 38) with the baseline characteristics of those individuals who contributed blood in the trial but were not included in the study (N = 50), are presented in S2 Table and show an overrepresentation of males in the present study (p = 0.046). All participants gave written informed consent before entering the trial, which was approved by the Ethical Review Board in Stockholm (Dnr: 2017/1206-31/1) and was registered in the ISRCTN registry (registration number: ISRCTN22387524). The present study was also approved by the same ethical committee. Moreover, all methods were carried out in accordance with relevant guidelines and regulations, and all experimental protocols were approved by the Ethical Review Board in Stockholm.

Intervention, body composition and cardiorespiratory fitness (CRF)
The treatment arms were designed with an intrinsic motivation-to-change strategy in order to encourage participants to maintain sustained changes in moderate-to-vigorous physical activity and CRF [79]. In both treatment arms, participants were encouraged to engage in a minimum of 30 minutes of moderate-to-vigorous physical exercise every day, counting both aerobic and resistance exercise. The actual amount of physical activity was measured using Actigraph GT3X+ accelerometers, worn on the hip during all waking hours, for 7 consecutive days at each assessment. The assessments were at baseline, after 5 weeks and after 11 weeks of intervention. Management and analyses of the accelerometer data have been described elsewhere [24]. Briefly, valid measurements included �10 hours wear-time per day for �4 days. Wear-time and classification of bouts were computed using ActiLife v.6.13.3. Moderate-tovigorous physical activity was defined as more than 3,208 counts per minute which in adults corresponds to a VO 2 /resting VO 2 � 3 or walking at~4.5 km per hour [80,81]. The exercise programs were designed in collaboration with personal trainers and individualized based on baseline levels of CRF (VO 2 max), illness or physical symptoms. Body composition and CRF measurements in the study participants have been described in detail elsewhere [24]. In brief, fat mass (in kilograms; kg) and fat-free mass (kg) were measured through bioelectrical impedance [82] using an Omron body composition monitor (model HBF-511B-E/HBF-511T-E). Physical tests also included measurements such as height, weight and waist circumference. CRF measurements were performed using a submaximal VO 2 max test on a stationary bicycle according to the Ekblom-Bak cycle ergometer test [25] and are presented in relative numbers (ml/min per kg body weight). VO 2 max categories are based on age and sexstratified reference values from~25,000 Swedish males and females in working age (available at https://www.gih.se/ekblombaktest).

Statistical analyses
To detect an effect of regular exercise on inflammatory biomarkers with the effect sizes previously reported [28,43,53,76,83], at an α = 0.05 and power = 0.80 for paired differences, a sample size of N = 10-35 individuals was estimated to be needed. Specifically, as several of the analytes measured by us were not previously included in studies of regular aerobic exercise, the sample size calculation was based only on eotaxin-1, IL-6, sICAM-1, sVCAM-1, TGF-β1 and VEGF-A. Group differences between those with and those without blood samples were tested using Mann-Whitney U tests (for continuous variables) and chi-square tests (for categorical variables). The bivariate correlations between clinical characteristics and the analyte levels were estimated with a non-parametric Spearman's rank correlation test. Differences between baseline and follow-up were tested using non-parametric Wilcoxon tests and (as described in the main text) the corrected significance was set at α = 0.004 (0.05/12) since there were 12 independent analyte groups. Sex differences in pre-post analyte level response were tested using an ANCOVA model on the analyte levels at follow-up, adjusting for baseline levels. For analytes with a sex difference at α = 0.05, the post-hoc pre-post analyte level difference in each sex was assessed using Wilcoxon tests and with a significance set at α = 0.004. No possible outlier was excluded from the analyses. The detected pre-post differences were confirmed not to be biased by different time-of-the-day sampling at baseline compared to follow-up by removing from the analysis the nine samples with different pre-post time-of-the-day samplings (data not shown).