Title: Resistance but not endurance training suppresses glucocorticoid-induced leucine zipper (GILZ) expression in human skeletal muscle

The glucocorticoid-induced leucine zipper (GILZ) serves as an anti-inflammatory regulator of gene expression in different tissues and is also expressed in human skeletal muscle. GILZ mediates the anti-myogenic and myotoxic side effects of statins via a shift in the Akt/FoxO signaling pathways. Recent evidence suggests that GILZ suppression is regulated by physical exercise, with external load being the decisive factor. Interestingly, statin treatment is rarely tolerated by habitually exercising individuals due to statin-associated muscle symptoms (SAMS). The opposing regulation of GILZ underpins this detrimental interaction of key measures of cardiovascular prevention. This interaction hypothetically differs between diverging exercise modalities in a mechanosensitive manner. To verify emerging evidence, we conducted a systematic search of the Gene Expression Omnibus (GEO) repository for studies reporting the acute effects of either endurance (END), conventional resistance (RT), or eccentric resistance training (ECC). 15 studies with 204 participants (22 females; 182 males, 18 to 90 years of age) were included in the analysis. Participants’ activity levels ranged from sedentary to trained. RT resulted in the highest GILZ suppression, significantly differing from the expressional change after END (-0.46 ± 1.11 vs. -0.07 ± 1.08; p = 0.03), but not from ECC (-0.46 ± 1.11 vs. -0.46 ± 0.95; p = 0.19). Furthermore, subgrouping revealed that RT-experienced participants exhibited a more pronounced GILZ suppression than their inexperienced counterparts (-0.98 ± 0.66 vs. -0.34 ± 1.16; p = 0.001). Our results strengthen the assumption that mechanical loading can be considered a key mediator of exercise-induced changes in GILZ expression.


INTRODUCTION:
The glucocorticoid-induced leucine zipper (GILZ, gene name TSC22D3) was first described as an immunoregulatory protein induced by dexamethasone in murine thymocytes 1 .Thereafter, GILZ expression has been reported in numerous human and murine tissues, such as the thymus, lymph nodes, bone marrow, spleen, lung, and also skeletal muscle [2][3][4] .Within skeletal muscle, GILZ mediates the anti-proliferative and apoptosis-inducing effects of glucocorticoids (GC) 2,4 .Similarly to glucocorticoid-mediated GILZ induction in skeletal muscle, recent evidence shows that GILZ is also elevated due to statin application 5 .Statins are a major agent for the management and treatment of hyperlipidemia and for the primary and secondary prevention of cardiovascular disease (CVD) [6][7][8] .Preventing cardiovascular disease continues to be of high importance as it remains the leading cause of premature mortality and rising healthcare costs, killing over four million people in Europe and over 18 million people worldwide each year 6,9 .Furthermore, the steady increase in the prevalence and mortality of cardiovascular disease between 1990 and 2019 is expected to continue, cementing the status of this problem as one of the greatest challenges in medicine 9 .
While statins effectively lower LDL-cholesterol and thus reduce the risk of CVD 6,7 , they are also associated with notable muscle-specific side effects 7 .These statin-associated muscle symptoms (SAMS) or statin myopathies range from mild myalgia 8 to life-threatening cases of rhabdomyolysis 10 .The prevalence of SAMS is reported to vary between 5% -29% 7,11 .Besides the acceptable benefit-to-harm ratio of statins 8 , SAMS remain the main reason for treatment discontinuation 12 .In addition, SAMS appear to be more prevalent in regularly exercising individuals and professional athletes 13,14 , which goes hand in hand with a reduced exerciseintensity tolerance [15][16][17][18][19] and a reduced training adaptability in statin users 20,21 .Taken together, these findings suggest an underlying link between exercise intensity and the occurrence of SAMS.Recent research offers new insights into this presumed link, indicating that the statininduced impairment of myogenesis is accompanied by elevated levels of GILZ expression 5 .Furthermore, the knockout of GILZ results in a resistance to statin-induced myotoxicity and to statin-induced changes in myogenin expression, an important myogenesis regulating factor (MRF).To further substantiate the involvement of GILZ in SAMS, the anti-myogenic effects of statins were mimicked by the sole overexpression of GILZ in zebrafish embryos 5 , insinuating that GILZ mediates the anti-myogenic effects of statins.
Evidence suggests that increased FoxO signaling mediates statin myopathy via an elevated expression of MuRF-1, MaFbx-1, and CTSL 22,23 .The transcriptional factor FoxO3 is particularly interesting in this context since, in addition to the expression of MuRF-1 and MaFbx-1, it also upregulates the expression of GILZ 4,5,30 .
An opposite effect to the statin-induced shift towards FoxO signaling is mediated by the peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α, gene name: PPARGC1A), which is often described as a master regulator for endurance-associated adaptations 31,32 .PGC1α serves as a metabolic sensor of calcium signaling, which is promptly (2 h post exercise) upregulated after endurance-type exercise in rats and humans 25,31,33,34 .Sandri and coworkers observed that PGC1α transgenic mice were resistant to denervationinduced muscle atrophy and changes in muscle fiber type distribution.Moreover, the overexpression of PGC1α reduced the expression of MuRF-1, MAFbx, and CTSL by 40% 25 .These results are in line with later findings, indicating that PGC1α overexpression protects against hind limb unloading-induced muscle atrophy 32 .
Given that SAMS prevalence and habitual physical exercise appear to be linked to another 13,14 , and considering that GILZ mediates statin-induce myopathy 5 , it appears of critical importance to further investigate the relationship between physical exercise, GILZ, and SAMS on a mechanistic level.Recent evidence indicates that GILZ expression is regulated by physical exercise in a mechanosensitive manner 35 ; Hecksteden et al. demonstrated that GILZ expression was acutely (30' and 3 h post-exercise) downregulated after a single session of resistance training (RT) at an intensity of 80% 1 repetition maximum (1RM), but not after a single time-matched endurance training session (END) 35 .Thus, it is assumed that the absolute external load, as opposed to cardiovascular strain, can be regarded as the driving factor regulating exercise-induced changes in GILZ expression.Moreover, the acute downregulation of GILZ after physically demanding RT in combination with the statin-induced upregulation of GILZ may account for the harmful side effects of statins 35 .This mechanism would explain that trained individuals exercising at higher absolute loads rarely tolerate statin treatment 14 and is further supported by the fact that physical exercise is considered a risk factor for SAMS 12,13 .
Since available exercise guidelines for CVD patients include not only moderate intense, continuous endurance exercise but also high-intensity interval training (HIIT), modified team sports, and strength training 36 , it seems reasonable to investigate the interaction between mechanical loading and GILZ expression.An improved understanding of the interaction between cardiovascular versus muscular strain and GILZ-mediated SAMS may contribute to better exercise recommendations and pharmacological strategies in CVD.Against this background, we analyzed datasets from acute exercise trials published on the Gene Expression Omnibus (GEO) repository.We hypothesized that GILZ expression is mechanosensitively affected by physical exercise.

Eligibility criteria
Included data sets were required to have conducted at least one bout of either endurance (END), traditional resistance (RT), or eccentric resistance training (ECC) exercise.Studies implementing a mixture of both END and RT exercise (also called concurrent training) were excluded.Further, a sample of human skeletal muscle must have been obtained acutely (between 3-6 h) post the bout of exercise.Considered subjects were required to be described as healthy and of legal age.Studies were excluded from our analysis if the implemented exercise did not suit the requirements (i.e., concurrent exercise or insufficiently intense RT described as rehabilitation exercise).Subjects who were administered hormone-altering drugs (e.g., 17beta-estradiol) 37 were excluded from the analysis.In the case that a study did conduct unilateral exertion, and did not obtain a muscle biopsy before the bout of exercise, while post-exercise, biopsies from both the trained and untrained extremity were obtained [38][39][40] , the data was included, and the untrained extremity was considered as pre-measurement.However, if no pre-measurement was obtained and both examined extremities did perform the same bout of exercise (i.e., but at different intensities), or one extremity did perform endurance and the other resistance type exercise, the data was excluded.Expression data were required to be obtained via either microarray or high-throughput sequencing techniques.Additionally, if the transcript of GILZ, TSC22D3, was not represented on the microarray or high-throughput sequencing platform, the data set was also excluded.

Data acquisition
A systematic search of the GEO repository was conducted (Fig. 1).This search was structured according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement 2020 41 and ended on February 29, 2024.The search strings used are documented in the supplementary materials (supp.2.1.1).Screening was conducted manually and structured according to the aforementioned inclusion criteria.In the rare case where multiple microarray probes were assigned to only one RefSeq ID of interest (a total of five occurrences), all corresponding probe signals were averaged.If one RefSeq ID was represented by multiple probes, and the probes were denoted with versions, the signal of the latest probe version was considered (one occurrence).If included studies obtained multiple muscle biopsies within our determined time frame, the biopsy closest to 6 h post-exercise was chosen.This decision was based on recent evidence demonstrating that translation-and transcription-initiation factor expression, as well as global gene expression, can be expected to peak within the latter phase of the selected time span [42][43][44] .
If subjects were tested multiple times, for example, before and after an exercise intervention lasting several weeks 40,45 , or after a high-load and a low-load exercise bout of RT 46 , measurements of only one of the occasions were analyzed.Thereby we intended to avoid multiple inclusions of identical subjects.In these cases, the measurement after the exercise intervention and acutely after the high-load RT session were included in the analysis.This decision was based on the assumption that higher external loads are apparent post higher relative exercise intensities and post familiarization, ultimately resulting in a more pronounced suppression of GILZ expression.Additionally, in one particular placebo-controlled study, half of the subjects were assigned to a 17beta-estradiol treatment, lasting for eight days 37 .In this case, only the placebo group was included in our analysis.

Data processing
Data analysis and visualization were performed using R Statistical Software 47 in the R Studio IDE 48 .To account for variation between trials, within-trial robust scaling of the raw data was performed.Robust scaling is considered a non-parametric alternative to the z-score and is calculated by dividing the difference between the respective value and the median by the interquartile range.Consequently, the scaled values represent the difference between the current value and the individual 'baseline' in units of interquartile ranges.Given that gene expressional data generally cannot be assumed to be distributed normally 49 , and varies substantially on inter-and intraindividual level 50,51 , this scaling method seemed appropriate.Subsequently, changes in gene expression from pre-to post-exercise were calculated (Δ).Datapoints deviating more than three times the interquartile range from the median were considered outliers and excluded from the analysis.

Statistical analysis
Gene expression data is robust scaled and is presented in arbitrary units (AU) as mean ± standard deviation.Normal distribution of the robustly scaled values was assessed visually with the help of quantile-quantile plots.ANOVA was computed to compare changes in Δ gene expression between groups.Independent Welch's t-test was used to compare ΔGILZ expression between the subset of trained and untrained individuals within the RT group.Critical threshold of significance was set at p < 0.05 and is denoted with asterisks (p < 0.05: *; p < 0.01: **; p < 0.001: ***).

RESULTS:
A total of 15 studies and 204 participants (22 female; 182 male; see Tab. 1) were included in the analysis.Six studies conducted END, five RT, two ECC, and two split their subjects into either END or RT.Three of the seven RT protocols investigated RT-trained or familiarized participants 46,52,53 (Tab.1).Participant's habitual activity level ranged from sedentary to endurance and strength trained (V ̇O2peak ≥ 55 ml•min −1 •kg −1 , and barbell back squat ≥ 1.5 times bodyweight), and the age of examined individuals ranged from 18 to 90 years (Tab.1).A summary of included studies, implemented training regimes, and participants' characteristics is provided in table 1 (Tab.1).

DISCUSSION
This is the first study that systematically screened data from the GEO repository to investigate the effect of different training modalities on GILZ expression.We assumed that physical exercise affects GILZ expression and hypothesized that concentric and eccentric strength training would lead to a more pronounced GILZ suppression than endurance training.Our results indicate notably suppressed GILZ expression after traditional and eccentric RT.Furthermore, the GILZ expression post RT differed significantly from the GILZ espression following endurance training (Fig. 2).This finding supports our hypothesis that GILZ expression is influenced by physical exercise, particularly mechanical loading.Based on a comparatively small sample (Tab. 1) and, consequently, diminished study power, changes in GILZ expression after eccentric training did not significantly differ from endurance training.Although insignificant, eccentric training resulted in the downregulation of GILZ, which was more than six-fold greater than post-endurance training.
Interestingly, the participants' training status seems to play a critical role in the magnitude of GILZ suppression post-exercise.We found that RT-trained participants experienced a more significant GILZ suppression compared to their untrained counterparts (Fig. 3).This finding further strengthens our hypothesis that absolute external load is the main determinant influencing GILZ expression post-physical exercise since resistance-trained participants performed the same relative intensity in an RT session with absolutely higher external loads, as compared to strength training inexperienced individuals 54 .This assumption is in line with the findings from Hecksteden and colleagues, demonstrating a significant downregulation of GILZ expression acutely after resistance training, only in RT-trained, but not in strength training inexperienced participants 35 .It thus seems that absolute mechanical external load is the main driver of exercise-induced GILZ expression changes.This interaction may explain why exercising individuals rarely tolerate statin therapy and further suggests that regular strength-related exercise may increase the risk of statin myopathy [12][13][14] .
When examining post-exercise atrogene expression, it appears that mechanically demanding exercise regimes generally tend to suppress MuRF1, MAFbx, and CTSL expression, which again, stands in contrast to the statin-induced elevation of FoxO downstream targets 22,23 and the heightened expression levels post-END (Fig. 4 A-C).More specifically, our results show that a single bout of endurance exercise results in a positive change of MuRF1, MAFbx, and CTSL expression (ΔMuRF1 = 0.59 ± 1.08; ΔMAFbx = 0.47 ± 1.07; ΔCTSL = 0.64 ± 0.87).This observation is in line with findings from Louis and Coffey et al., demonstrating that both running (30 minutes at 75% V ̇O2peak), and cycling (60 min at 70% V ̇O2peak) induced an acute (1 to 4 h post-exercise) upregulation of MuRF1 and MAFbx 43,55 .Similarly, Schwalm and colleagues found an immediate (1 h post) upregulation of CTSL after 2 h of cycling at an intensity of 70% V ̇O2peak 56 .Interestingly, 2 h of cycling at lower intensities (55% V ̇O2peak) did not significantly change CTSL expression 56 .Thus, an intensity-dependent CTSL expression seems likely.
Our observations expand on these findings, showing that only mechanically demanding eccentric resistance exercise led to a meaningful MuRF1 suppression, which significantly differed from the positive ΔMuRF1 expression after END and traditional RT (Fig. 4A).These findings are at least partly in line with available evidence, suggesting that RT, like END, acutely upregulates MuRF1 expression.It appears that moderately intense (65-70% of 1RM) RT, consisting of only 30 repetitions of knee extensions, is a sufficient stimulus to acutely (1 to 4 h post-exercise) upregulate MuRF1 43,57 .Further, Nedegaard et al. reported an upregulation of MuRF1 3 h post 300 concentric contractions (i.e., unilateral box step-ups), while the MuRF1 expression in the contralateral leg, performing the eccentric contractions (i.e., stepping down from the box), remained unchanged 58 .Compelling evidence on MuRF1 expression post-ECC is still lacking as gene expression data post eccentric resistance exercise are scarce.Considering the available evidence, it might be reasonable to assume that traditional RT acutely upregulates MuRF1 expression, while exercise implementing higher external loads (potentially mediated via eccentric contractions) leads to no acute change in MuRF1 expression or even to a slight suppression.
Our results show that both RT and ECC induce an acute downregulation of MAFbx, which differs significantly from the upregulation of MAFbx expression following END (Fig. 4B).The majority of published articles did not observe changes in MAFbx expression acutely (1 to 4 h) after RT 43,44,55,57 .In contrast to these findings, when expanding the time frame after exercise cessation (5-12h post), Louis et al. did find a significant suppression in MAFbx expression after moderately intense (70% of 1RM, 30 reps) resistance exercise 43 .In this regard, Stefanetti and colleagues also observed a decreased MAFbx expression 2.5 h post resistance exercise, which continued to decrease to ultimately reach its minimum at 5 h post-RT 59 .Consequently, MAFbx expression seems acutely (1-4 h post) unchanged after RT, and then, over the time course of 5-12 h post-exercise, decreases to levels below the baseline values.Our results strengthen the assumption that changes in MAFbx expression are heavily time-dependent (Tab.1 & Fig. 4B).Taking existing evidence and our results together, it appears that RT-induced downregulation of MAFbx can be expected to occur within a time frame of 4-12 h after resistance exercise.Thus, it might be assumed that not only the opposing regulation of GILZ, but also the opposing regulation of MuRF1, MAFbx, and CTSL, post-mechanically demanding exercise and statin application, might lead to elevated risk of SAMS occurrence in habitually exercising individuals.All considered exercise modalities (i.e., END, RT, and ECC) led to a positive change in PGC1α expression of a similar magnitude, with no significant between-group differences.Since PGC1α is considered an important sensor for motor neuron-induced calcium signaling 25 , it seems reasonable that all exercises containing sufficient motor neuron-induced muscle contractions would lead to a PGC1α upregulation.This assumption is supported by the fact that metabolic perturbations, such as intracellular Ca ++ oscillation, acutely upregulate PGC1α expression and activity 60 , in both animal and human models 33,61 .Moreover, the total number of consecutive muscle contractions may be essential in PGC1α upregulation and, subsequently, the PGC1α-mediated suppression of FoxO signaling 62 .Although PGC1α expression is known to be upregulated by a low-amplitude and low-frequency Ca ++ oscillation, typically induced by high-volume and low-intensity endurance exercise 60 , recent evidence shows that PGC1α expression is also upregulated post resistance type exercises 63,64 .Specifically, resistance exercise seems to upregulate the expression of a particular PGC1α splice variant called PGC1α4 64 .This splice variant ultimately promotes muscle fiber hypertrophy and enhances anaerobic glycolysis 63,64 .
Our retrospective analysis of GEO datasets comprises limitations that need to be addressed: A substantial amount of between-trial heterogeneity regarding participants' characteristics and performed exercise regimes needs to be stated.This applies to the training status of the examined individuals, which affects absolute external loads during RT sessions and the resulting implemented absolute exercise intensities.Further, only four 45,46,52,53 out of seven RT trials 39,65,66 (~57.1%)examined trained or familiarized subjects.This discrepancy in familiarization level may affect the magnitude of RT-induced changes in GILZ expression when RT trials are examined collectively.The same limitation holds true for the included END trials, where five 53,[67][68][69][70] out of eight 40,66,71 (62.5%) studies examined endurance-trained or familiarized subjects.Furthermore, the intensity of the exercises implemented in the examined trials varied substantially.For example, two END trials investigated gene expression changes in endurance-trained males post high-intensity interval training (HIIT) 67,69 .In both of these studies, subjects performed exercise bouts of maximal (100%) or close to maximal (90%) intensities (peak power output; PPO) interspersed by active recovery segments at lower intensities (50% PPO).Our analysis revealed that these particular exercise trials resulted in a downregulation of GILZ that was exceptionally high compared to the other END trials, which involved less intense and continuous endurance exercise (results not shown).Nonetheless, our findings clearly demonstrate a significant downregulation of GILZ expression post-RT, specifically when subjects are familiarized and thus perform the same relative intensity at absolute higher external loads.These findings have been obtained despite the aforementioned heterogeneity between implemented exercise regimes and the training status of included studies.

CONCLUSION
Our results strongly support the hypothesis that particularly mechanical loading serves as a key mediator of training-induced suppression of GILZ acutely post exercise.Notably, this result has been obtained despite the substantial differences in age, level of habitual exercise, and exercise dose across studies.If confirmed, these findings may contribute to a more harmonized treatment of CVD, consisting of both statin medication and specifically adapted exercise training.To further clarify the underlying relationship between exercise, GILZ expression and SAMS, further research is needed that considers the use of statins and different exercise modalities.

Figure Captions:
Figure 1 Flow chart of the systematic search process in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement 41 .

Figure 2
Group comparison of ΔGILZ expression post different exercise modalities.Black line depicts the mean, outer lines: SD, and whiskers: twofold SD.Red dashed zero-line indicates the reference for no expression change from pre-to post-exercise.*p < .05,**p < .01,***p < 0.001, ns: non-significant.

Figure 4
Figure 4 A-D Group comparison of ΔAtrogenes and ΔPGC1α expression post different exercise modalities.Black line depicts the mean, outer lines: SD, and whiskers: twofold SD.Red dashed zero-line indicates the reference for no expressional change from pre to post exercise.*p < .05,**p < .01,***p < 0.001, ns: non-significant.

Figures Figure 1
Figures

Figure
Figure 2