Regular exercise may improve systemic markers of chronic inflammation, but direct evidence and dose-response information is lacking. The objective of this study was to examine the effect and time course of changes in markers of chronic inflammation in response to progressive exercise training (and subsequent detraining). Forty-one sedentary men 45–64 yr of age completed either a progressive 24-wk exercise intervention or control followed by short-term removal of the intervention (2-wk detraining). Serum IL-6 fell by −0.4 pg/ml (SD 0.6) after 12 wk and responded to moderate-intensity exercise. Serum alanine aminotransferase (ALT) activity fell −7 U/l (SD 11) at 24 wk although there was no evidence of any change by week 12 (and therefore ALT required more vigorous-intensity activity and/or a more prolonged intervention). The effect on IL-6 was lost after 2-wk detraining whereas the change in ALT was retained. The temporal fall and rise in IL-6 with training and subsequent detraining in men with high IL-6 at baseline provided a retrospective opportunity to examine parallel genomic changes in peripheral mononuclear cells. A subset of 53 probes was differentially regulated by at least twofold after training with 31 of these changes being lost after detraining (n = 6). IL-6 responded quickly to the carefully monitored exercise intervention (within weeks) and required only moderate-intensity exercise, whereas ALT took longer to change and/or required more vigorous-intensity exercise. Further work is required to determine whether any of the genes that temporally changed in parallel with changes in IL-6 are a cause or consequence of this response.
- cardiovascular disease
- Type 2 diabetes
- low-grade inflammation
- gene expression
chronic low-grade inflammation plays a central role in the etiology of diseases such as cardiovascular disease (22) and Type 2 diabetes (42). There appears to be a strong inverse relationship between physical activity/fitness and serum markers of inflammation such as IL-6 and C-reactive protein (CRP) (5, 19, 21, 31). However, causal evidence to support cross-sectional associations is far less consistent, with some intervention studies reporting a reduction in certain measures of inflammation in response to regular exercise or physical activity (14, 22, 26, 31, 35), while others report no effect of increased physical activity on the same or similar parameters (14, 16, 25, 50). These discrepant findings cannot be explained by the length of the exercise programs employed since interventions lasting several weeks have been both effective (10–11, 36, 45) and ineffective (16, 25) and the same is true for more prolonged interventions lasting 6–10 mo (14, 18). Some studies have seen a positive effect with two 30- to 80-min exercise bouts per week over 8 wk in overweight women (36), whereas others have seen no effect from five 30-min bouts of vigorous-intensity exercise per week over 16 wk in obese men and women (25). It is noteworthy that very few studies have incorporated a control group into their experimental design, and this may explain some of the inconsistencies in the evidence base (14, 16, 25). Clearly, more information is needed on the impact of regular exercise or physical activity on markers of inflammation.
There are various markers of inflammation available and each can provide subtly different information about the response to an exercise intervention. The most commonly used marker of chronic inflammation is CRP, which is produced by the liver primarily in response to IL-6 (12). We propose a central role for IL-6 as it is released from adipose tissue and skeletal muscle (10, 23, 25, 28) and may link changes in these tissues with downstream consequences elsewhere (e.g., liver). In this context, high resting serum concentrations of IL-6 are a potent risk factor for future myocardial events (43) and risk of future cardiovascular disease in healthy men (6). An increase in IL-6 provides approximately the same information about future risk as traditional risk factors for cardiovascular disease and may represent a therapeutic target (6). In addition to circulating cytokines such as IL-6, and given the fundamental role for leukocytes in inflammation in cardiovascular disease (27, 41), it may be possible to examine exercise-induced changes in the expression of key pro- or anti-inflammatory molecules in these cells. In this context, the enzyme heme oxygenase-1 (HO-1) is both oxidant responsive and anti-inflammatory, with effects ranging from the desensitization of adhesive responses in leukocytes through to the resolution of inflammation (33, 34). Research confirms that leukocyte HO-1 can play an important role in atheroprotection (37). Collectively, we propose that through the considered selection of different measures during the course of an exercise intervention it may be possible to reveal the temporal sequence of events and that this may point toward primary mechanisms and events.
To this end, the primary aim of the present randomized-controlled trial was to examine the effect of a 6-mo exercise intervention on the magnitude and time course of changes in markers of low-grade inflammation in sedentary middle-aged men. Importantly, we include robust measures of physical activity and compliance to quantify the dose of prescribed exercise within the context of habitual physical activity. A secondary aim was to examine the extent to which any effects observed over 6 mo would be reversed on short-term removal of the exercise intervention. One of the most striking findings in the present investigation was that the fall in serum IL-6 was most pronounced in men with high IL-6 at baseline. The observed temporal change in serum IL-6 in response to an exercise intervention (a fall with exercise and a rise with subsequent detraining) provided a retrospective opportunity to examine parallel molecular changes within peripheral blood mononuclear cells. Peripheral mononuclear cells are of considerable interest since monocyte-derived macrophages and lymphocytes play a particularly important role in atherosclerosis (41, 44), and changes in these cells will have a powerful effect on their stimulation of other cells such as adipocytes (49).
MATERIALS AND METHODS
Sedentary male volunteers were recruited from the local community. A total of 152 telephone screening interviews were conducted (Fig. 1). Individuals with known disease (e.g., heart disease, diabetes, arthritis) or who self-reported they engaged in structured physical activity lasting ≥30 min on two or more occasions per week were excluded from the study, as were volunteers who smoked, had a body mass index (BMI) ≥35 kg/m2, or took regular medication. Eligible participants completed a health questionnaire and were fitted with a physical activity monitor to further establish that they were sufficiently inactive (see below for details), before being randomly allocated to one of two groups using a sealed envelope (Fig. 1). The envelopes were numbered with the sequence generated and known only by a third party. Each volunteer provided written informed consent, and the investigation was approved by the local ethics committee.
Participants in both groups reported to the laboratory at baseline and then 4, 12, 24, and 26 wk later, each time following a 12-h overnight fast.
Exercise intervention and control.
Individuals in the exercise group completed a 24-wk exercise training program followed by 2 wk of detraining (removal of the intervention). Intensity and duration were increased in a progressive manner (Table 1). The exercise (attendance, exercise duration, and heart rate) was recorded using a monitoring system (Fitronics, Bath, UK). Once every fortnight, volunteers reported to the laboratory to complete one of their weekly exercise sessions so that intensity and heart rate response could be monitored and the exercise prescription altered accordingly. Men in the control group were asked to maintain their current lifestyle and levels of physical activity during the 26-wk study period.
Height, body mass, submaximal oxygen uptake, and blood pressure were measured as described previously (26). Maximal oxygen uptake (V̇o2max) was determined using a progressive incline test to volitional fatigue on a treadmill, consisting of 3-min exercise stages with the gradient increasing by 3% at the end of each stage (Woodway, ELG 70, Weiss, Germany). Expired gas samples, heart rate, and ratings of perceived exertion were collected in the final minute of each stage. Participants were required to meet two of four criteria for having achieved V̇o2max: heart rate ≥ age-predicted maximum heart rate, respiratory exchange ratio ≥ 1.10, rating of perceived exertion ≥ 19, or an increment in V̇o2 ≤ 5 ml·kg−1·min−1 in response to an increased gradient.
Assessment of physical activity.
Daily physical activity energy expenditure was estimated using combined accelerometry and heart rate with branched-equation modeling (Actiheart, Cambridge Neurotechnology, Cambridge, UK). More information is provided in Supplemental Information available with the online version of this article.
Blood collection, processing, and analysis.
Before blood sampling, volunteers were asked to refrain from strenuous physical activity for at least 36 h. Participants were also asked whether they had experienced any symptoms of illness in the past 3 days (e.g., upper-respiratory tract infection) or had taken any medication in the past 48 h. If a participant indicated that this was the case, a new appointment was made 3 days later.
Venous blood samples were collected after participants had been in a supine position for 15 min. Whole blood differential leukocyte counts and lymphocyte and monocyte heme oxygenase-1 (HO-1) expression were measured as described previously (26). Serum was analyzed for IL-6, CRP, soluble intercellular adhesion molecule-1 (sICAM-1), and insulin using commercially available solid-phase ELISAs [IL-6 and sICAM-1: Quantikine, R and D Systems, Abingdon; insulin and CRP: DSL, Oxford Bio-Innovation, Oxon; oxidized low-density lipoprotein (LDL): Mercodia, Uppsala, Sweden]. Plasma was analyzed for triglycerides, cholesterol, high-density lipoprotein cholesterol (HDL-C), nonesterified fatty acids (NEFA), glucose, and alanine aminotransferase (ALT) using commercially available kits (Randox Laboratories, UK; and Wako Diagnostics, Alpha Laboratories, UK). Samples were frozen at −80°C, and all samples from a single individual were analyzed in a single run at the end of the trial.
Peripheral blood mononuclear cell gene expression.
Peripheral blood mononuclear cells (PBMCs) were isolated from EDTA-treated blood using a one-step centrifugation technique (Lymphoprep, Nycomed, Norway). Harvested cells were washed twice in 20 ml PBS and then counted using an automated hematology analyzer (SF-3000, Sysmex UK). PBMCs were lysed in Trizol (Invitrogen, Paisley, UK), with the volume added dependent on cell concentration (1 ml per 5 × 106 cells) and then frozen at −80°C until subsequent analysis. Samples from six of the men with high serum IL-6 at baseline were included for microarray analysis. RNA was not available for all subjects at all time points. Based on rank order for serum IL-6, the six men included for arrays were ranked 1, 2, 4, 5, 6, and 8 (from 10).
Total RNA was isolated using Trizol, followed by the RNeasy kit (Qiagen, Crawley, UK). The integrity of the RNA was confirmed with analysis by the Agilent 2100 bioanalyzer (Palo Alto, CA) using the RNA 600 LabChipTM kit. Four-hundred nanograms input RNA and 3.0 μl (1:2,500 dilution) Agilent One-Color RNA Spike-In RNA were labeled with the Agilent Low RNA Input Linear Amplification Kit PLUS, One-Colour (Agilent Technologies UK, Wokingham, Berkshire, UK) according to manufacturer's instructions as previously described (29). The Agilent Hybridisation Kit (catalog no. 5188–5242) was used in conjunction with Agilent Human Oligo Arrays (catalog no. G4112F, Agilent Technologies UK). Hybridization was performed as previously described (29) according to the manufacturer's protocol (using 2 μg of the labeled sample cRNA as input). The slides were scanned with the Agilent G2565BA Microarray Scanner System. The Agilent G2567AA Feature Extraction Software v.9.1 (Agilent Technologies UK) was used for data extraction and quality control. In compliance with MIAME standards, data files were deposited into the NCBI Gene Expression Omnibus (GEO). The following link was created to allow review of these data: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=pxgjdeyeiomgify&acc=GSE12385.
Extracted data were analyzed using GeneSpring GX 7.3.1 (Silicon Genetics). Agilent standard scenario normalizations for FE1-color arrays were applied to all data sets.
To permit meaningful comparisons with previous research that has not reported intermediate time points between pre- and postintervention measures, preplanned contrasts between treatment groups were conducted in relation to the overall change in primary outcome measures from baseline to 24 wk (i.e., completion of the intervention period). To examine the time course of responses, a two-way mixed-model ANOVA was used to compare results between treatment groups over time (treatment × time) irrespective of minor deviations from a normal distribution (28) but with the Greenhouse-Geisser correction applied to intraindividual contrasts for ε < 0.75 and the Huynh-Feldt correction adopted for less severe asphericity (1). Where significant interactions were observed, multiple t-tests were applied to determine the location of variance both between treatments at each time point and between time points within each treatment relative to baseline, with both methods subject to a Holm-Bonferroni correction (1). Relationships between outcome variables were examined using a Pearson's product-moment correlation, and any meaningful associations (i.e., r ≥ 0.7) between baseline status and the magnitude of response to treatment were explored further via stratification of the exercise group according to baseline status.
Sample size calculations were based on our unpublished data indicating that physically active middle-aged men display significantly lower serum IL-6 concentrations than sedentary middle-aged men similar to those recruited for the present study [0.61 pg/ml (SD 0.38) vs. 1.18 pg/ml (SD 0.67), respectively]. Therefore, assuming a reduction in IL-6 over 24 wk of 0.57 pg/ml with a pooled SD of 0.55, it was estimated that 42 participants divided equally between treatment groups would provide ∼90% probability of detecting such an effect at an α-level of 0.05. To allow for drop-out, 54 men were randomly allocated (Fig. 1). All data are reported as means (SD).
For microarray data, a subset of probes for data interrogation was generated that excluded controls, spots of poor quality, and gene probes that were present in <50% of samples. From these selected probes, mean expression of each probe was calculated (n = 6), and relative expression was determined comparing expression profiles at week 24 with baseline. Genes differentially regulated by >2.0-fold (comparing week 24 with baseline) were selected. One-way ANOVA was performed followed by Benjamini and Hochberg multiple-test correction with a false-discovery rate of 0.05. Microsoft Excel templates were prepared containing the 53 probes that were differentially regulated following the 24-wk exercise intervention. Expression of these 53 probes relative to baseline was also determined following the detraining period (week 26). Of these 53 probes, 31 were no longer significantly differentially expressed following detraining (compared with baseline). Ingenuity Pathway Analysis 3.0 (Ingenuity Systems) was utilized to assemble functional networks and pathways associated with the 53 and 31 probes found to be regulated by training and detraining in men with high IL-6 at baseline, respectively.
Exercise prescription and energy expenditure.
Of the 49 volunteers who received the intervention, 41 completed the study (Fig. 1). There were no baseline differences between groups with regard to any measured variable (Tables 2 and 3). Mean adherence to the exercise program using the gym-based electronic monitoring system was 95% (SD 5%), i.e., participants completed 89 (SD 5) of the 94 prescribed sessions. The exercise intervention successfully increased the amount of time and energy engaged in physical activity that complies with current physical activity recommendations (please see Supplementary Information available online).
There was a small significant reduction in body mass (P = 0.01) and body mass index (P < 0.01) over the 24-wk intervention in the exercise group (Table 2), and there was no change in the control group (treatment × time: F = 10, P < 0.01, both variables). Notably, the effects on body mass and BMI persisted after 2 wk of detraining in the exercise group (both P < 0.01).
The exercise intervention significantly increased maximal oxygen uptake by 8.4% (SD 7.7%) at the 24-wk time point relative to baseline (P < 0.01), an effect that was retained after the 2-wk detraining period (P = 0.02; Table 2). The control group displayed no change in this variable (treatment × time: F = 13, P < 0.01), with specific differences identified between the exercise and control groups at both the 24- and 26-wk time points (P = 0.02 and P = 0.04, respectively).
Glycemic control, blood lipids, and blood pressure.
Fasting plasma glucose, insulin, and blood pressure did not change from baseline in either group (Table 2). Similarly, there was no significant difference between groups with regard to blood lipids (Table 2), although the preplanned contrast between groups in relation to the change in plasma NEFA from baseline to 24 wk did approach statistical significance [i.e., control group = 0.0 mmol/l (SD 0.2) and exercise group = −0.1 mmol/l (SD 0.2); P = 0.09].
Fasting serum IL-6 concentrations responded differently to the two treatments (treatment × time: F = 3.2, P = 0.02; Table 3). There was no change in the control group but significantly reduced IL-6 relative to baseline in the exercise group at both the 12- and 24-wk time points (P = 0.02 and P = 0.01, respectively). The preplanned contrast between groups in relation to the change in IL-6 from baseline to 24 wk was different between the exercise and control groups [control group = 0.1 pg/ml (SD 0.5) and exercise group = −0.4 pg/ml (SD 0.5); P = 0.01], an effect that was apparent even after 12 wk of the intervention (P = 0.01). Serum ALT activity was also different between treatment groups (treatment × time: F = 4.0, P = 0.01; Table 3), again with stable values over time in the control group but with significantly reduced ALT activity relative to baseline after 24 wk of the exercise intervention (P = 0.03). The preplanned contrast between groups in relation to the change in ALT from baseline to 24 wk was different [i.e., control group = 0 U/l (SD 10) and exercise group = −7 U/l (SD 11); P = 0.03], although there was no evidence of any change at the 12-wk time point relative to baseline for ALT. Notably, the effects of the exercise intervention on both IL-6 and ALT were no longer significantly different either between the treatment groups or within each group relative to baseline following 2 wk of detraining, although there was a tendency for ALT to remain lower than baseline at 26 wk (P = 0.09) and the change in ALT during detraining did not differ between treatments (P = 0.39). In contrast, detraining lead to a significant increase in IL-6 in the exercise group relative to the change in the control group over the same period (P = 0.04). All other inflammatory markers did not respond to the exercise intervention (Table 3).
Correlations and stratified analyses.
Within the exercise group, there was a strong positive linear correlation between baseline IL-6 concentration and the magnitude of the change in IL-6 over 24 wk of the exercise intervention (r = 0.9, P < 0.01). Similar baseline-dependent relationships were also noted for ALT and NEFA, with correlations between baseline levels and the magnitude of changes over 24 wk of r = 0.8 and r = 0.7, respectively (P < 0.01). To explore these correlations further, all outcome variables were reexamined within the exercise group alone but stratified simply into two equal subgroups based on baseline status for IL-6, ALT, and NEFA (Fig. 2). For IL-6 and NEFA, only the men with high baseline values responded to the exercise intervention (IL-6 strata × time: F = 4.2, P = 0.02; NEFA strata × time: F = 5.3, P < 0.01; Fig. 2). Men with both high and low ALT responded to the exercise intervention, and there was no significant difference between strata (strata × time: F = 1.4, P = 0.3), although the greatest response was observed in men with high ALT at baseline (Fig. 2). Notably, the overall changes in IL-6, ALT, and NEFA were mutually independent because there were only weak associations between the changes for these parameters (e.g., IL-6 and NEFA r = 0.2; IL-6 and ALT r = 0.1; NEFA and ALT r = 0.1). The men with high IL-6 did not show a universally poorer profile in terms of other measures relative to the men with low IL-6 (e.g., there were no significant differences between men with high and low IL-6 in terms of age, mass, glycemic control, and so on; data not shown). Despite the fact that 12 of the 20 men in the exercise group had “high” serum CRP concentrations at baseline (>3 mg/l), there was no evidence that the men with high values responded to the exercise intervention whereas those with low values did not.
It is noteworthy that participants with higher baseline IL-6 and NEFA both tended to revert to baseline levels on removal of the intervention (i.e., after 2 wk detraining). This factor, coupled with the finding that the men in the control group with high baseline IL-6 and NEFA showed no change over time (data not shown), negates the possibility that the observed baseline-dependent exercise-induced effects were an artefact of regression to the mean.
Training-induced changes in PBMC gene expression.
We used microarrays to examine the change in gene expression profiles in PBMCs that accompanied the fall and rise of serum IL-6 in response to the exercise training-detraining intervention in six of the men with high IL-6 at baseline (i.e., which genes responded to the exercise intervention and persisted or reverted to baseline following detraining). Serum IL-6 in these six men fell from 2.1 pg/ml (SD 0.7) to 1.3 pg/ml (SD 0.2) after training (week 24) and increased to 1.8 pg/ml (SD 0.9) after 2 wk of detraining (week 26). The other responses to the intervention in the men with high IL-6 were reflective of the group as a whole (an increase in maximum oxygen uptake, fall in mass, and no change in blood measures such as blood lipids and fasting glucose/insulin; data not shown).
After exclusion of absent calls and probes that were expressed in less than half of experimental samples, 35,282 of an initial 41,267 probes (85%) were considered for analysis. Only probes with a mean differential expression of at least 2.0-fold at week 24 (compared with baseline) and that were significant (P < 0.05) were considered to be truly differentially expressed. Data analysis revealed differential expression of 53 probes using these criteria (Table 4). To validate microarray results we compared expression of six genes by microarray and qRT-PCR and, although the absolute magnitude of differential expression differed between the methods, both microarray and RT-PCR detected similar patterns of expression for five of six genes studied (please see Supplementary Information available online).
To better understand the genomic responses to the 24-wk exercise intervention we examined these 53 differentially expressed probes in further detail using the Ingenuity Pathway Analysis (IPA) tool. Molecular networks of direct physical, transcriptional, and enzymatic interactions were computed from the Ingenuity knowledge base. The resulting two significant networks (P < 0.001) contain molecular relationships with a high degree of connectivity and every gene in the network is supported by published literature. The molecules included in these two networks and main cellular functions or conditions/diseases associated with these two networks are listed in Table 5. Figures for these networks are also presented (please see Supplementary Information available online).
Detraining-induced changes in PBMC gene expression.
To determine genes that may be associated with increased serum IL-6 levels following the cessation of the exercise intervention (i.e., at week 26), we examined expression of the 53 probes differentially regulated by training following a 2-wk detraining period (Table 4). Differential expression of 22 of the 53 physical activity-regulated probes was retained at week 26 after 2 wk of detraining (P < 0.05 compared with baseline) while expression of 31 probes returned to baseline (Table 4). Since serum IL-6 levels also returned to baseline following the 2-wk detraining period (week 26). it is possible that the 31 probes whose expression concomitantly returned to baseline may be mechanistically linked with this response. Thus we used IPA to investigate the interactions between these 31 probes, and this assembled one network from the 20 genes/proteins known to be encoded by these 31 probes. The molecules included in this network and main cellular functions or conditions/diseases associated are listed in Table 5 and depicted visually in the online Supplementary Information.
The time course of temporal changes in response to an exercise intervention may point toward primary mechanisms and events and also provide information on the type of exercise stimulus that is required. The main finding of the present investigation was that serum IL-6 responded relatively early to regular exercise and required only moderate-intensity exercise to elicit this effect, whereas plasma ALT did not change until 12–24 wk into the exercise intervention. This suggests either that ALT takes longer to respond (perhaps being secondary to other changes) or that the increased intensity and duration of each exercise bout during weeks 12–24 was responsible for this delayed response. With just 2 wk of detraining (removal of the intervention), the reduction in IL-6 that had been observed during the 24-wk intervention was lost.
This is the first report of a time course for changes in resting serum IL-6 following exercise training. IL-6 was reduced after 12 wk of moderate-intensity exercise (with no further improvement with an additional 12 wk of progressive exercise training). It is noteworthy that the fall in IL-6 was reversed 2 wk after removal of the intervention. While these effects were statistically significant for the whole exercise group, the change in IL-6 concentration was mostly explained by a fall in men with high IL-6 at baseline. Some investigations have reported changes in IL-6 after 12 wk of regular exercise (7), whereas others have reported no change after 8–14 wk (7, 11, 12, 31). There are numerous potential explanations for these discrepancies, including different assays and very different IL-6 values, different exercise prescriptions, different baseline values and populations (from children to people with Type 2 diabetes), and postintervention follow-up blood samples taken up to 14 days postintervention. From a mechanistic perspective, it has been proposed that some of the anti-inflammatory effect of exercise may be associated with an acute upregulation of parameters such as IL-6 and IL-10 during and after each bout of exercise (40). However, we feel that this is unlikely to explain the reduction in IL-6 observed in the present study since this was initiated and occurred after regular exercise that we have previously shown does not lead to an acute upregulation of IL-6 or IL-10 (26).
IL-6 potentially has both pro- and anti-inflammatory actions (13). This is context specific but it has been proposed that, in contrast to potential beneficial short-term actions of raised IL-6, the chronic systemic elevation of IL-6 would have more detrimental than beneficial effects (13). This makes sense because cross-sectional studies report that high IL-6 is a risk factor for future cardiovascular events (6, 43) and low leisure-time physical activity is associated with high IL-6 (39). It is estimated that adipose tissue contributes up to 35% of circulating IL-6 at rest (30) and anti-inflammatory pharmacological intervention has been shown to reduce the secretion of IL-6 from adipose tissue (35). It is therefore tempting to speculate that an exercise-induced change in IL-6 is partly mediated by a reduced secretion of IL-6 that is secondary to reduced inflammation in adipose tissue. If this is shown to be the case, then the rapid increase during 2 wk of detraining in the present study might suggest that this is not simply explained by an expansion of adipose tissue mass. However, this is not the only possible explanation for the present findings. Endurance training leads to an elevated basal IL-6 receptor expression in skeletal muscle (15) and a lower upregulation of IL-6 in response to a single bout of exercise (9). It is therefore possible that regular exercise improves the sensitivity to IL-6 and that the observed fall in the present investigation could be related to a reduced need to produce as much IL-6 (15). Whatever the mechanism, given the heightened risk associated with high IL-6 (6, 43), a fall in IL-6 is likely to be coupled to reduced risk of future cardiovascular disease and it is noteworthy that in the present study only a few weeks of moderate-intensity exercise was required to observe this effect.
Serum ALT activity is associated with hepatic lipid accumulation (8), and ALT is raised in men with hepatic inflammation (2). Some previous studies have shown a fall in serum ALT after relatively short-term exercise interventions (2, 36), while others have shown no effect following 12 wk of exercise training (8). In the present study, ALT was unchanged after 12 wk of moderate-intensity exercise but responded to a further 12 wk of intensified training. Notably, ALT fell by ∼30% in the men with the highest ALT and this is consistent with findings that ALT falls dramatically in response to exercise interventions when baseline values are high (2). Furthermore, unlike IL-6, changes in ALT induced over the 24-wk exercise intervention were largely retained during the detraining period. Collectively, these findings indicate varied kinetics for different inflammatory markers in response to regular exercise.
There was no change in other markers of inflammation and oxidative stress measured in the present investigation (leukocyte count, sICAM-1, CRP, oxidized LDL, and lymphocyte and monocyte HO-1). One simple explanation is that these parameters do not respond to regular exercise. However, it should be taken into account that the present investigation superimposed a structured exercise intervention over and above “other” physical activity in a group of middle-aged men who were not meeting current physical activity recommendations but who did accumulate considerable nonstructured physical activity (please see Supplementary Information online). Indeed, although structured physical activity was low in these participants at baseline, they accumulated a reasonable amount of total physical activity (e.g., physical activity level was relatively high). This may also explain why many of the more traditional risk factors did not change in response to the intervention (e.g., HDL cholesterol). The possibility that the accumulation of low-level physical activity may be sufficient to capture most of the anti-inflammatory benefits of physical activity has numerous implications and certainly reinforces the importance of including adequate measurements of total physical activity before and during physical activity interventions. Equally, however, it is important to highlight that if participants are recruited based on a “sedentary” behavioral criterion alone then this does not mean that they will have elevated risk factors for disease or, indeed, that they will necessarily respond to an exercise intervention. This may be explained by complications in the interpretation of physical activity recommendations as we have discussed in detail elsewhere (47). Alternatively, this may reflect the fact that physical activity is only one of a number of factors that will contribute to individual risk. It is likely that a greater and more uniform response would have been observed had we included elevated risk factors as a criterion for inclusion, but this would also have changed the research question. In summary, in the present study, we recruited based on physical activity behavior so that all men were defined as sedentary at baseline. Those sedentary men with high IL-6 and ALT (along with NEFA) responded with a more marked change than those sedentary men with low IL-6 and ALT.
It is interesting that CRP was unaffected by the exercise intervention even though IL-6 decreased because this cytokine is proposed to be the primary stimulus for hepatic CRP release (12). It may be that the concentration of IL-6 in the superficial venous circulation (as was measured in the present study) does not reflect IL-6 concentration in portal blood. An alternative explanation is that hepatic production of inflammatory mediators such as CRP is not simply influenced by cytokines received in blood from other tissues but, instead, reflects CRP produced as a consequence of localized hepatic inflammation (17). Within this context, it is noteworthy that serum ALT activity did not fall until sometime between weeks 12 and 24 of the exercise intervention and this opens the possibility that a longer and/or more progressive intervention may be necessary to produce a downstream reduction in CRP. Another explanation is that, given the ∼18 h half-life of CRP (3), acquisition of blood samples 36–48 h after the last bout of exercise may have captured a small part of the acute exercise-induced increase in CRP, potentially masking any training-induced reduction in this variable. However, this is only relevant for the 24-wk time point as our previous work indicates that moderate-intensity activity would not acutely increase serum CRP (26). Some of the effects of exercise last only hours and days and so choosing an appropriate follow up time point is always going to be difficult if there are multiple outcome measures.
The observation that men with high IL-6, ALT, or NEFA at baseline showed a more pronounced response to exercise than men with lower baseline values is perhaps unsurprising and has been noted in previous investigations for similar parameters (10, 20). There has been one report that the IL-6 response to an exercise intervention is influenced by a polymorphism in the IL-6 promoter (34). However, we found no evidence that the men with high IL-6 formed a distinct cluster based on genotype (both C and G alleles were found in men with high and low IL-6; data not shown). It is noteworthy that those men with high baseline IL-6 and/or NEFA also responded most rapidly to the removal of the intervention (i.e., detraining). Although speculative, this might suggest that physical activity or exercise behavior per se is important or that there are other highly responsive phenotypic characteristics that have the capacity to change IL-6 and NEFA.
PBMC gene expression following training and detraining in men with high IL-6 at baseline.
The fall in IL-6 with training and rise following the cessation of training (detraining) provided a retrospective opportunity to examine potential mechanisms and/or consequences of changes in IL-6 in terms of gene expression in PBMCs. As described earlier, these cells play an important role in cardiovascular disease (41, 44) and can modify the function of other cells such as adipocytes (49). Fifty-three probes were differentially expressed after the training period with the expression of most (31 probes) being reversed by the cessation of training (2 wk of detraining). Many of these genes are involved in inflammation and immune response, including upregulation of IL-4, carboxypeptidase A2 (CPA2), IL-8, and IL-2. These genes are involved in diverse immune functions such as chemotaxis, homing, migration, and recruitment. Network analysis showed that differential expression of these immunomodulatory molecules was linked via stress-activated protein kinase signaling through the central ERK, Akt, PI3K, and p38/MAPK molecular nodes. These data provide some mechanistic information on the well-described immunomodulatory effects of physical activity in men who were at increased risk of cardiovascular disease because of their relatively higher IL-6 at the start of the intervention (6, 43).
We used gene arrays to identify genes that responded to the exercise intervention in the context of the changes in serum IL-6. Thus, of the 53 probes that were differentially expressed at week 24, the subset of these probes whose expression was no longer significant following detraining (i.e., comparing expression at week 26 with baseline) has the potential to be mechanistically linked to changes in serum IL-6. Pathway analysis of these 31 probes whose differential expression was not retained following detraining shows that IL-4 and IL-2 are centrally involved. IL-4 and IL-2 are known to interact in both a synergistic and antagonistic manner, depending on the cell type in which the responses are measured (32). Increased expression of IL-4 mRNA in response to physical activity is of particular interest as IL-4 is well known to regulate IL-6 levels, while a direct relationship between IL-2 and IL-6 has not been established. IL-4 is a product of activated T-cells that is known to act on a number of immune cells, such as T-cells, monocytes, and mast cells (4, 46). Additionally, IL-4 is thought to possess immunosuppressive properties via its inhibition of IL-6 production (23–24, 33). For example, overexpression of IL-4 by adenovirus reduced IL-6 production concomitant with the prevention of bone erosion (24). Furthermore, in human mast cells, IL-4 completely abrogated IL-6 production (23). In another study, while IL-2 had no effect on IL-6 release from T-cells, IL-4 was critical for inhibiting IL-6 release following stimulation (33). Mononuclear cells are not a primary source of IL-6 in the circulation and this probably explains why expression of IL-6 in PBMCs did not change in response to the intervention (data not shown). Of course, we should not exclude the possibility that altered expression of IL-4 and other similar molecules in mononuclear cells could lead to the altered release of IL-6 from other cell types (49).
In addition to the regulation of IL-4, network analysis revealed that several other molecular factors were upregulated at week 24 but not different from baseline at week 26 (e.g., RET, CLC, INSL6, and PARD6G). These are known positive regulators of transcription factors such as ERK and STAT3. These findings suggest probable activation of the ERK signaling pathway and further provide a direct link between IL-4 and IL-6 because it has been demonstrated that IL-4 affects proliferation and cytokine responses via ERK signaling (23). Furthermore, ERK signaling is known to be altered by exercise (4, 38, 48). However, in this context, it is worth bearing in mind that in the present study samples were taken approximately 36–48 h after the last exercise bout. Furthermore, we do not have protein measurements and so our observations are limited to changes that were induced and sustained at the transcriptional level at the time of sampling. As a result, we might only expect to see changes in genes that either have a long half-life or that are regulated in response to chronic (rather than acute) exercise training. Therefore, it is possible that because of the varied kinetics of individual genes there may have been some changes that we have not measured and were overlooked because of the blood sampling framework. This is the first investigation of gene expression arrays in peripheral blood mononuclear cells with exercise training and we report multiple interesting changes in these cells, although future studies may seek to consider repeated follow-up time points after training and in different populations and groups.
In conclusion, a carefully monitored exercise intervention superimposed above other physical activity reduced fasting IL-6 and ALT in middle-aged men who did not meet current physical activity recommendations, particularly in those with high values at baseline. IL-6 responded quickly (within the first few weeks) and required only moderate-intensity exercise to elicit this effect, whereas ALT either takes longer to change and/or requires more vigorous-intensity exercise. The removal of regular exercise (detraining) quickly reverses the changes in IL-6 but not ALT. The expression of many genes in PBMCs occurred in parallel with the change in IL-6 in men with high IL-6 at baseline, although it is too early to tell whether this temporal relationship is causal. One particularly interesting possibility is that changes in mononuclear cells (e.g., IL-4) could downregulate IL-6 production from other cells such as adipocytes.
This study was supported by funding from the British Heart Foundation (PG/04/124/17944).
No conflicts of interest are declared by the authors.
We acknowledge the technical assistance of Fei Ling Lim and Sarah E. Askew with the microarray experiments.
The gene expression arrays were conducted by Unilever Discover.
- Copyright © 2010 the American Physiological Society