Interleukin-6 Drives Mitochondrial Dysregulation and Accelerates Physical Decline: Insights From an Inducible Humanized IL-6 Knock-In Mouse Model

Abstract

Chronic activation of inflammatory pathways (CI) and mitochondrial dysfunction are independently linked to age-related functional decline and early mortality. Interleukin 6 (IL-6) is among the most consistently elevated chronic activation of inflammatory pathways markers, but whether IL-6 plays a causative role in this mitochondrial dysfunction and physical deterioration remains unclear. To characterize the role of IL-6 in age-related mitochondrial dysregulation and physical decline, we have developed an inducible human IL-6 (hIL-6) knock-in mouse (TetO-hIL-6^mitoQC) that also contains a mitochondrial-quality control reporter. Six weeks of hIL-6 induction resulted in upregulation of proinflammatory markers, cell proliferation and metabolic pathways, and dysregulated energy utilization. Decreased grip strength, increased falls off the treadmill, and increased frailty index were also observed. Further characterization of skeletal muscles postinduction revealed an increase in mitophagy, downregulation of mitochondrial biogenesis genes, and an overall decrease in total mitochondrial numbers. This study highlights the contribution of IL-6 to mitochondrial dysregulation and supports a causal role of hIL-6 in physical decline and frailty.

Graphical Abstract

Although the chronic activation of inflammatory pathways (CI) in older adults is strongly associated with the worsening of chronic disease and increased frailty, sarcopenia, and mortality (1,2), considerable gaps in knowledge remain as to how CI affects the integrity of cells and tissues and drives these adverse outcomes. Of all CI markers, interleukin 6 (IL-6) has long been touted as a biological driver of adverse health outcomes and has been nicknamed the “cytokine for gerontologists” (3). However, establishing a causal role for IL-6 in age-related decline remains a challenge. At the molecular level, CI is closely associated with mitochondrial damage, and both are considered primary hallmarks of aging with extensive crosstalk (4). Aging-related mitochondrial dysfunction leads to the release of mitochondria-derived damage-associated molecular patterns and contributes to the cycle of senescence and systemic inflammation. In turn, inflammation can also further lead to mitochondrial damage (5,6).

Genetic knock-out mouse models have been developed to recapitulate components of the morbidity and mortality seen in frail humans, allowing for the controlled study and development of treatments that can target adverse outcomes from frailty. For example, aged mice lacking the anti-inflammatory cytokine interleukin-10 (IL-10) have a propensity to develop age-associated elevation of serum proinflammatory cytokines, including IL-6; these mice also demonstrate phenotypes similar to frail humans and show significant alterations in mitochondrial function (1,2,7). Additionally, a mouse model of frailty and chronic inflammation in the absence of IL-6 (deficient in both IL-10 and IL-6, “double knockout”) demonstrates short-term physical and cardiac mitochondrial oxygen consumption improvements but higher mortality in aged mice (8), suggesting a protective role for IL-6. Despite these observations, the direct effects of age-related chronic increases in IL-6 on physical and mitochondrial function and how these changes contribute to aging phenotypes are unknown.

This study focuses on a humanized inducible IL-6 model due to the significant homology between mouse and human IL-6 (hIL-6) at the amino acid level, with minor differences that do not affect activity (9). Additionally, the hIL-6 cytokine has demonstrated activity in mice (10), and antibodies targeting hIL-6 have shown therapeutic effects in mice (11). The primary goal of this study is to better understand the role of IL-6 in frailty in light of the conflicting results from our previous work with the double knockout mouse. We hypothesize that induction of IL-6 will lead to increases in frailty and physical decline due to skeletal muscle changes that are mediated by changes in mitochondrial regulation and reduced mitophagy.

Results

Development of Humanized IL-6 Conditional Knock-in Mouse

The hIL-6 gene knock-in mouse (TetO-hIL-6^mitoQC) was developed utilizing clustered regularly interspaced short palindromic repeat/CRISPR-associated 9 (CRISPR/Cas9) technology targeting the Rosa26 locus as previously described (12). The transgene donor vector consisted of a tetracycline/doxycycline response element (TRE) promoter driving expression of hIL-6 cDNA and mitoQC element (mCherry linked to enhanced green fluorescent protein (EGFP) and mouse mitochondrial Fis1 cDNA, driven by a constitutive CAG promoter; Figure 1). This transgene was inserted between exons 1 and 2 of the Rosa26 gene. Polymerase chain reaction (PCR) confirmed hIL-6 expression upon doxycycline induction (Figure 1). The presence of the mitoQC system allows for further quantitative and qualitative studies of mitochondria and mitophagy (13). Under steady-state conditions, mitochondria fluoresce both red (mCherry) and green (EGFP), resulting in yellow, as shown in Figure 1. When mitochondria are transported to lysosomes during mitophagy, EGFP fluorescence is quenched in the acidic environment while mCherry fluorescence persists, resulting in red-only mitolysosomes. Transgene insertion was confirmed by PCR (Figure 1 and Supplementary Table 1). Electron microscopy imaging revealed the expression of mitochondrial EGFP and the uptake of EGFP-marked mitochondria in mitophagosomes and autophagosomes (Supplementary Figure 1).

Figure 1.

Design of TetO-hIL-6^mitoQC mice. Upper panel (1, 2, 3, and 4): details of clustered regularly interspaced short palindromic repeat/CRISPR-associated 9 (CRISPR-Cas9) technology to insert the human IL-6 (hIL-6) and mitoQC transgene into murine Rosa26 gene. Lower panel (5), nucleus: schematic detailing doxycycline induction of hIL-6. Image demonstrating expression of hIL-6 in heterozygous male mouse cardiac tissue after 6 weeks of induction. Lower panel (6), Polymerase chain reaction (PCR) confirmation of GFP and hIL-6 gene insertion in TetO-hIL-6^mitoQC mice. Lower panel (7, 8), cytoplasm: schematic detailing mitoQC function. EGFP and mCherry colocalization on outer surface of mitochondria in cytoplasm with corresponding image. With mitochondrial transport into lysosome and creation of mitolysosome, EGFP fluorescence is quenched in acidic environment. Corresponding image showing colocalization of lysosomal protein cathepsin B with mCherry.

Baseline (preinduction) and 6 weeks postinduction blood draws and physical measures were performed on 8-month-old heterozygous male TetO-hIL-6^mitoQC mice. Doxycycline (2 mg/mL) was administered via drinking water with 5% sucrose. Repeat physical testing was also done at 3 and 5 months postinduction and compared to uninduced mice.

Human IL-6 Expression Leads to Impaired Physical Function and Increased Frailty

To understand the effects of hIL-6 induction on physical function in TetO-hIL-6^mitoQC mice, changes in grip strength and treadmill performance (number of falls and time running on the belt) were assessed at baseline and after 6 weeks of hIL-6 induction. Induced mice demonstrated significant decreases in forelimb grip strength (Figure 2A) but no difference in total standing or walking time (Figure 2B). Treadmill testing revealed that TetO-hIL-6^mitoQC mice did not finish the running protocol postinduction compared to preinduction and had an overall decrease in the total amount of time running on the treadmill belt (p = .016; Figure 2D). Preinduction mice had a 4-point decline in percentage time on the belt for every 15-minute increment in the preinduction period (95% CI: from 0.6 to 7.6 point decline, p = .023) while postinduction mice had an 8-point decrease in percentage time on the belt for every 15-minute increment (95% CI: from 2.1 to 13.9 point decline, p = .008). Additionally, postinduction mice demonstrated a 40% increase in falls while running compared to preinduction mice (95% CI: 15%–70%, p = .001), and for every 15 minutes on the treadmill, the number of falls increased by 33% (95% CI from 7% to 65%, p = .01).

Figure 2.

TetO-hIL-6^mitoQC mice demonstrate physical impairment and higher frailty index score following human IL-6 (hIL-6) induction. N = 4–9. Male mice at baseline and following hIL-6 induction with doxycycline had (A) grip strength (adjusted for body weight) after 6 weeks of induction, (B) nonexercise activity (standing and walking) after 6 weeks of induction, (C) latency to fall in rotarod test after 5 months of induction, (D) exercise activity (treadmill testing) after 6 weeks of induction, and (E) 29-item clinical frailty index score after 5 months of induction *p < .05, **p < .01, all comparisons are to baseline before induction.

The rotarod test was used to assess overall motor coordination and balance assessment in mice following hIL-6 induction (14). Consistent with the decreased grip strength observed postinduction, the rotarod test demonstrated reduced total time on the rotarod before falling in the TetO-hIL-6^mitoQC mice after 5 months of induction compared to uninduced mice (p = .016; Figure 2C). Next, to assess changes in frailty with induction, we used a modified mouse clinical frailty index (mFI) using a total of 29 clinical signs of frailty (excluding individual weight and temperature) (15) and calculated mFI score at baseline and 5 months after induction. We observed significant increases in mFI scores postinduction compared to uninduced mice (p = .034, Figure 2E). The most prominent changes in frailty index at 5 months postinduction were gait disorder, forelimb grip strength, and the mouse grimace scale.

Whole Blood RNAseq Analysis Demonstrates Upregulation of Inflammatory Pathways After hIL-6 Induction

Whole blood from TetO-hIL-6^mitoQC mice at baseline, 2 weeks, and 6 weeks after induction of hIL-6 was obtained, and transcriptomics was performed using bulk RNAseq. Over 2 493 transcripts of interest were identified, of which 641 had significant differential expression at 2 weeks (FDR <0.05), and 1 946 had significant differential expression at 6 weeks (FDR <0.05) compared to baseline (Supplementary File 1).

Agglomerative hierarchical clustering of the differentially expressed genes yielded 6 gene clusters, of which 4 are predominant. The largest, cluster I (328 genes) contained transcripts upregulated after 6 weeks of hIL-6 expression. A second, cluster VI (100 genes) was composed of transcripts only upregulated after 2 weeks of hIL-6 expression. Finally, 2 clusters encompassed transcripts downregulated at 2 weeks (49 genes, cluster III) and at 6 weeks (70 genes, cluster IV), respectively, relative to baseline expression (Figure 3A). Functional association network analysis using STRING and parsed by Markov clustering demonstrated 7 principal modules within cluster I genes upregulated at 6 weeks (Figure 3B): lipid metabolism and uptake, binding proteins (1), cell cycle regulation (2), inflammatory markers produced by lymphocytes and natural killer (NK) cells (3), chemokines (4), NK cell lectin family proteins (5), metabolic pathways (6), and complement pathway and pattern recognition receptors (7).

Figure 3.

Whole blood RNAseq shows upregulation of inflammatory pathways and metabolic dysregulation in TetO-hIL-6^mitoQC mice. (A) Heatmap of whole blood transcript levels in male mice at 2- and 6-weeks post IL-6 induction (N = 4), filtered on significance at q-value <.05. Visual inspection revealed 6 distinct clusters corresponding to distinct expression trajectories. Upregulated transcripts are indicated in red and downregulated transcripts are noted in blue. (B) Functional association network of genes in cluster 1 (upregulated at 6 weeks only). Network Markov clustering revealed about half of the transcripts fell into 7 principal modules consisting of 7 or more nodes. Representative ontologies are: 1—lipid metabolism and uptake, binding proteins, 2—cell cycle and cell division, 3—inflammatory markers produced by lymphocytes and natural killer cells (IL-12, IFN-γ, granzyme A/B), 4—chemokines, 5—natural killer cell lectin family proteins, 6—metabolic pathways, 7—complement pathway and pattern recognition receptors. (C) Ingenuity Pathway Analysis (IPA) of cluster I (upregulated genes at 6 weeks). (D) IPA of cluster IV (downregulated genes at 6 weeks). (E) IPA of cluster VI (upregulated genes at 2 weeks). (F) IPA of cluster II (downregulated genes at 2 weeks).

Ingenuity Pathway Analysis (IPA) of the heatmap clusters demonstrated upregulation in proinflammatory pathways at 6 weeks (Th1 and Th2 activation pathway, Stat3, and Pi3k/Akt signaling, Figure 3C) with downregulation in the anti-inflammatory cytokine IL-10 and metabolic pathways such as urea cycle and arginine degradation pathways (Figure 3D). Additionally, metabolic pathways, including histidine degradation and vitamin C transport, were upregulated after 6 weeks of treatment (Figure 3C).

Altered Mitochondrial Energetics Following Induction of hIL-6

Given the observed dysregulation in metabolism-related genes via RNAseq analysis, we next performed untargeted metabolomics on serum from mice at baseline and 6 weeks postinduction to better evaluate metabolic changes associated with hIL-6 elevation. ATP, pyruvate, lactate, N-acetyl aspartate, and alpha-ketoglutarate were decreased in the serum of mice postinduction (Figure 4). Succinate was increased approximately 3-fold following hIL-6 induction.

Figure 4.

Altered serum metabolic profiling in TetO-hIL-6^mitoQC mice following human IL-6 (hIL-6) induction. N = 7. Changes in serum tricyclic antidepressant (TCA) cycle and glycolysis intermediates in male TetO-hIL-6^mitoQC mice at baseline (pre) and following 6 weeks of hIL-6 induction (post). *p < .05, **p < .01, ***p < .001.

Increased hIL-6 Transcript Levels in Skeletal Muscle Following hIL-6 Induction

To validate the effectiveness of induction by the Tet-On expression system and doxycycline treatment in muscle, we quantified the levels of both mouse IL-6 and hIL-6 in the gastrocnemius muscle tissue of uninduced and induced TetO-hIL-6^mitoQC mice (Figure 5A). The absolute number of transcripts per microliter for each gene was quantified through digital PCR. Compared to the uninduced group, there was no change observed in the levels of mouse IL-6 with induction. In contrast, there was approximately 4-fold increase in transcript levels of hIL-6 upon induction (p = .001), affirming that the experimental treatment increases hIL-6 at the skeletal muscle level.

Figure 5.

Functional, morphological, and cellular characterization outcomes in gastrocnemius muscle tissues in TetO-hIL-6^mitoQC mice of human IL-6 (hIL-6) induction. (A) Quantification of mouse IL-6 and hIL-6 in gastrocnemius muscle tissue from TetO-hIL-6mitoQC male mice following 6 weeks of hIL-6 induction. N = 4 for uninduced group, N = 5 for induced group. Comparisons were conducted using Student t test. **p < .01. (B) Comparison of in vivo force-frequency curves of gastrocnemius muscle in induced TetO-hIL-6^mitoQC male mice at baseline, 3 months postinduction, and 5 months postinduction. N = 5. Data are presented as mean and standard error of the mean (SEM). Comparisons were conducted using 2-way repeated measures ANOVA and Tukey test for post hoc analysis. *p < .05. (C) Collagen type 1 and fibronectin expression in gastrocnemius of uninduced and induced TetO-hIL-6^mitoQC mice (following 6 weeks of hIL-6 induction). N = 4 for each group. Data are presented as mean and standard error of the mean (SEM). Comparisons were conducted using Student t test. *p < .05. Representative images of type 1 collagen deposition on gastrocnemius muscle in TetO-hIL-6^mitoQC (i) uninduced mice and (ii) induced mice. Representative images of fibronectin deposition on gastrocnemius muscle in TetO-hIL-6^mitoQC (iii) uninduced mice and (iv) induced mice. (D) Schematic depicting overall changes in mitochondria and mitophagy at baseline and postinduction (following 6 weeks of hIL-6 induction). From top to bottom, the first graph depicts total mytolysosomes, the second graph shows total intact mitochondria, and the third graph shows the ratio of mitolysosomes to intact mitochondria. *p < .05, **p < .01. Schematic was created with BioRender.com. (E) Protein association network of significantly downregulated genes in gastrocnemius tissue from uninduced and induced TetO-hIL-6^mitoQC male mice. N = 4 for each group. Significant fold changes (p < .05) are shown in colored nodes, with their respective fold regulation values depicted as a continuous gradient. Gene network constructed using STRING (v11.5) database via Cytoscape and manually organized to denote aging pathways specific to the RT2 array. Full gene list with fold changes and p values provided in Supplementary File 2.

Decreased Nerve-Evoked Isometric Force of Gastrocnemius Muscle With Induction of hIL-6

Given the observed physical decline upon induction of hIL-6, we next assessed the relationship between hIL-6 and neuromuscular function. Using an in vivo quantification of nerve-evoked function, we measured nerve-evoked isometric force and compared the frequency versus force relationship at baseline, 3 months, and 5 months after induction (16). There was a significant change in isometric force with time in both legs (p < .001 for left and p = .042 for right). In post hoc analysis, we found that peak isometric force was reduced at 3 and 5 months postinduction compared to baseline in vivo muscle strength testing in left gastrocnemius muscle (p = .006, p < .001, respectively). The force versus frequency relationship curve revealed small but significant differences in left leg low-frequency trains (10 and 20 Hz) at 5 months postinduction compared to baseline. In the right leg, the isometric force was reduced at 5 months postinduction compared to 3 months postinduction (p = .029); however, there was no significant difference at 3 months and 5 months postinduction from baseline (Figure 5B).

Higher Type 1 Collagen Levels in Gastrocnemius Muscle Following hIL-6 Expression but no Difference in Fiber Type

Given the observed impairment in muscle contractility after hIL-6 induction, we sought to understand whether these changes were precipitated by IL-6-mediated changes in muscle fiber type. We quantified changes in gene expression of sarcomeric myosin heavy chain (Myh) genes associated with type I slow twitch (Myh6 and Myh7) and type 2 fast twitch fibers (Myh1, Myh2, and Myh4) after 10 weeks of hIL-6 induction in male mice gastrocnemius muscle tissue. Our data showed no significant difference in relative gene expression following induction (Supplementary Figure 2). We also measured elements of the extracellular matrix in male mice, specifically type I collagen and fibronectin, which are increased in muscle repair and fibrosis (17), and observed higher type I collagen content in gastrocnemius muscles after hIL-6 induction (p = .011). There was no change in fibronectin levels (Figure 5C).

Increased Skeletal Muscle Mitophagy Following hIL-6 Expression

The quantity of the mitochondrial population and the relative quality of mitochondria in cells can affect overall bioenergetic capacity, stress resistance, and frailty (18,19). We investigated the effect of hIL-6 induction on mitochondrial turnover using the mitoQC component of the transgene construct. Under steady-state conditions, mitochondria fluoresce red and green; however, when mitochondria are transported to lysosomes in mitophagy, EGFP fluorescent signal is quenched by the acidic milieu while mCherry fluorescence persists. Therefore, quantifying mCherry-only puncta in mitoQC expressing cells represents the number of mitolysosomes present. To understand the effect of hIL-6 on mitochondrial turnover and mitophagy, we quantified colocalized mCherry+EGFP-positive puncta to measure intact mitochondrial content and the density of mCherry-only puncta to quantify mitolysosomes in gastrocnemius from mice preinduction and after 6 weeks of induction. We calculated the ratio of mitolysosome content to intact mitochondria. Both mitolysosome and intact mitochondria content were significantly lower in gastrocnemius of induced mice (p = .002, p = .005, respectively). In contrast, the percentage of mitolysosomes relative to intact mitochondria was higher postinduction (p = .045; Figure 5D).

Downregulation of Mitochondrial Function Genes Following hIL-6 Expression

Since hIL-6 has been linked to the development and progression of frailty and has been hypothesized to exacerbate the effects of aging on different organ systems (3), we next performed a targeted array looking at genes commonly linked to aging, including genomic instability, mitochondrial dysfunction, apoptosis, cellular senescence, inflammatory response, and oxidative stress. Given the changes in mitophagy and fibrosis observed in gastrocnemius muscles following induction, we focused on gene expression changes in this muscle in age-matched uninduced and induced mouse groups. Mitochondrial function-related genes were the most differentially expressed following the induction of hIL-6 (Figure 5E). The mitochondrial protein sirtuin 3 (Sirt3), a protein deacetylase that has roles in almost every aspect of mitochondrial biology, including ATP generation, reactive oxygen species detoxification, and mitochondrial dynamics (20), was downregulated (fold change = 0.49, p = .024). Mitochondrial transcription factor A (Tfam), which has a crucial role in mitochondrial DNA (mtDNA) transcription, replication, and mitochondrial biogenesis, was significantly downregulated (fold change = 0.55, p = .009). Finally, Tfb1m (transcription factor B1, mitochondrial) and Tfb2m (transcription factor B2, mitochondrial), which also have roles in regulation of mtDNA transcription and replication were downregulated (fold change = 0.57, p = .039; fold change = 0.54, p = .008, respectively). Downregulation of Ndufb11 (NADH: ubiquinone oxidoreductase subunit B11) was observed postinduction (fold change = 0.44, p = .004). Additionally, our data show downregulation of the Ste24 (zinc metallopeptidase) gene in the genomic instability pathway, which can lead to nuclear architecture abnormalities and a shortened lifespan (21) (fold change = 0.42, p = .029).

Taken together, these results show that hIL-6 induction is associated with increased frailty index, impaired physical performance, decreased skeletal muscle contractility, increased measures of fibrosis (type I collagen), downregulation of mitochondrial function genes, and decrease in intact mitochondrial and mitolysosome content in gastrocnemius muscle.

Discussion

In this study, we showed that elevating hIL-6 expression in adult mice impairs physical function in the form of reduced grip strength, worse performance on the rotarod, and higher frailty index score. Along with this, we demonstrated that chronic inflammation induced by increased hIL-6 levels can activate innate and adaptive immune pathways and dysregulate metabolic pathways. Changes in whole blood RNAseq demonstrate increased inflammation in mice following hIL-6 induction, with enrichment in cytokine and innate immune system signaling as well as upregulation in complement-associated genes, similar to human studies showing elevations in classical complement pathway with aging (22). We observed several differences between the aging human and mouse immune systems. In particular, RNAseq analysis showed hIL-6-related increases in mouse NK cell lectin receptors, while prior studies examining NK cell receptors across the human lifespan demonstrated overall increases in cytotoxic NK cells but decreases in a subset of lectin receptors (23,24).

The changes observed via RNAseq and serum metabolomics underscore the complex relationship between inflammatory pathways and energy utilization. This study showed downregulation of several metabolites associated with healthy aging following hIL-6 induction, similar to what has been observed in prior studies of aged animals. We observed downregulation of the citrulline biosynthesis pathway-related genes and upregulation of histidine degradation-related genes compared to uninduced mice (Figure 3). Increased citrulline promotes synaptic plasticity in aged rats (25), and histidine has been observed to decrease in older humans (26). Although the relevance of increased vitamin C transport remains unclear; it may be a compensatory mechanism for the increased IL-6 as it has been shown that plasma levels of IL-6 can be reduced by vitamin C via downregulation of hepatic mRNA expression (27). Additionally, serum metabolomics demonstrated significantly decreased pyruvate and ATP production following hIL-6 induction, similarly observed in frail and prefrail older females compared to young controls (28).

Several changes in RNAseq and serum metabolomics analysis after 6 weeks of hIL-6 induction can likely potentiate chronic inflammation. Finally, succinate, which was increased approximately 3-fold after hIL-6 induction (Figure 4), is elevated by chronically activated macrophages, and this can induce production of IL-1β (29).

Elevated hIL-6 and its downstream effects were associated with decreased isometric force and higher collagen I expression in gastrocnemius muscles. A mechanism by which this may occur is via changes to collagen I following hIL-6 expression. The extracellular matrix of skeletal muscle consists of several different collagens, integrins, proteoglycans, and glycoproteins, forming a complex architectural network designed to transmit myofibrillar forces throughout the muscle fiber and provide structural integrity (30). Of these components, collagen is the most abundant and appears to be highly sensitive to mechanical loading (31,32).

Mitochondrial dysfunction is a major contributor to the physical frailty syndrome in older adults, and appropriate regulation of mitophagy is essential to preserve mitochondrial efficiency (33). Consistent with this, we observed mitochondrial dysregulation and downregulated genes related to mitochondrial function after induction of hIL-6. Down-regulation of the Sirt3 gene has been shown to be related to several age-related diseases, including cancer, diabetes, and neurodegenerative disorders. Furthermore, an accelerated aging phenotype and loss in mitochondrial integrity were shown in the Sirt3 knock-out mouse model (34). Prior work suggests that Sirt3 attenuates aging-associated mitochondrial dysfunction by regulating metabolism and protective antioxidant pathways, and its reduced expression is associated with frailty in humans (20,35). Another downregulated gene, Tfam, a downstream transcription factor of PGC-1alpha and a possible mitochondria biogenesis marker, has been shown as one of the main factors affecting mitochondrial biogenesis in aging (36,37). The findings of mitochondrial dysfunction observed in our TetO-hIL-6^mitoQC mouse model are consistent with a recent study showing increased mitophagy following IL-6 treatment in cerebral vessels (38).

Our data suggest a connection between CI, mitochondrial dysregulation, muscle fibrosis, physical decline, and frailty, all of which can be induced by hIL-6 in our TetO-hIL-6^mitoQC mouse model. Some of our physical and molecular findings are supported by a recent study that found loss of muscle mass and decreased grip strength following IL-6 induction (39). However, in this study, Jergović et al. observed an increase in IL-6 levels in spleen homogenates, but not in muscle or fat tissue. In contrast, our study found a significant upregulation of hIL-6 transcripts in skeletal muscle, but no changes in mouse IL-6 levels. Also, our RNAseq analysis revealed an upregulation of inflammatory pathways and a downregulation of anti-inflammatory and metabolic pathways, and we observed increased skeletal muscle mitophagy and a decrease in mitochondrial function, as suggested by the downregulation of genes tied to mitochondrial operation. Our study further uncovered signs of impaired neuromuscular function and increased muscle repair and fibrosis activity. We also noted a downregulation of Ste24, a gene linked to genomic instability offering additional insights into the age-accelerating nature of higher levels of hIL6.

While the findings of this study provide valuable insights into the role of hIL-6 in physical function, frailty, and mitochondrial function, it is important to acknowledge the inherent limitations, including small sample size, lack of female mice, and that while our results indicate a certain trend, the potential effects of aging (accelerated), particularly in the 5-month induction group, could have contributed to the observed changes. The induction of hIL-6 was experimentally done via doxycycline administration, which may have its own side effects (40) and also may not replicate the natural and complex regulatory pathways of hIL-6 expression in human tissues.

Furthermore, while hIL-6 is a significant player in the inflammatory response and mitochondrial function, it is part of a large network of interacting genes and proteins. The study does not consider the effects of other genes or proteins that could potentially be influencing the observed results. The study only measures certain parameters at specific points (baseline, 2 weeks, 6 weeks, 3 months, and 5 months postinduction). Also, the relative proportion of leukocyte subpopulations in circulation was not assessed in this study, which may contribute to the changes observed in gene expression in the blood.

In the current study, we focused on gastrocnemius tissue as gastrocnemius muscle is predominantly composed of fast-twitch muscle fibers, which are likely to be more susceptible to atrophic signals. The gastrocnemius muscle is a large and easily accessible muscle that is commonly studied in aging research, which allows for comparability across studies. However, the effects of hIL-6 may not be limited to the gastrocnemius muscle, and assessing the affect on other muscles could yield more comprehensive insights into the systemic effects of hIL-6. Additionally, the study was conducted using a mouse model, and although the mouse model used in this study was humanized, one must consider the physiological differences between mice and humans that can affect translation of these findings to human physiology and pathophysiology.

Despite these limitations and given that CI is typically seen in aging and frailty, an optimal animal model in which to study the mechanisms contributing to chronic inflammation-mediated physical changes is one that can recapitulate many of the physical and metabolic changes observed in frail humans. The inducible TetO-hIL-6^mitoQC model has advantages over other genetically modified knock-out or knock-in mouse models in that the timing of hIL-6 induction can begin in midlife instead of throughout the entirety of the organism’s developmental period and lifespan, providing a closer representation of what is observed in frail humans (41). Additionally, and perhaps more importantly, the inducible TetO-hIL-6^mitoQC mouse is also a suitable model to test pharmacological and physical interventions to help slow the development and progression of frailty-related physical decline in older adults.

Method

Generation of TetO-hIL-6mitoQC Mice

Mice were generated by Cyagen (Santa Clara, CA). The gRNA to ROSA26 gene, the donor vector (Figure 1A), and Cas9 mRNA were co-injected into fertilized C57BL6/N mouse oocytes to generate targeted knock-in offspring. F0 founder animals were identified by PCR and sequence analysis (Supplementary Table 1). F0 mice were bred with wild-type mice to test germline transmission and F1 animal generation. F1 mice were bred with wild-type C57BL6/N mice to obtain positive F2 mice. Male heterozygous mice were treated with 6 weeks of doxycycline (2 mg/mL in water containing 5% sucrose), which was changed weekly throughout the study.

RNAseq Library Preparation and Sequencing

Whole blood was obtained from 4 male mice at baseline, 2 weeks, and 6 weeks postinduction. RNA was isolated using RNeasy Mini Kit from Qiagen (Germantown, MD) according to the manufacturer’s instructions, and the quality and yield of RNA were measured via Nanodrop 2000 from Thermo Fisher Scientific (Waltham, MA). The libraries were prepared by poly(A) RNA isolation and fragmentation, complementary DNA synthesis and end repair, sequencing adapter ligation, and PCR amplification. The library insert size was 200–500 bp and peaked at around 300 bp. After libraries were uniquely barcoded and pooled, NovaSeq6000 sequencing was performed. Reads were aligned to the mouse genome mm10 and a sequence consisting of the human TRE3G promoter and IL6 transcript variant 1 (NM_00600.4), using Tophat v.2.1.1 (42), and then assembled into transcripts with Stringtie v.1.3.6 (43). Differentially expressed genes were obtained with DESeq2 (44) and used for subsequent pathway analysis with Ingenuity Pathway Analysis from Qiagen (Germantown, MD). Library preparation, sequencing, and data analysis were performed at the Johns Hopkins University Computational Biology Consulting Core.

RNAseq Analysis

Differentially expressed genes were identified by DESeq2 package in R and had Benjamini and Hochberg false discovery rate (FDR) criterion applied for multiple testing corrections of the q-value, with a threshold of DEGs set at FDR <0.05. DESeq2 data were imported to Ingenuity Pathway Analysis (IPA) to identify Gene Ontology terms, metabolic pathways, upstream regulators, and functional networks among the DEG. Functional association and interaction networks were constructed by loading identifiers of significantly regulated genes into StringApp v1.4.2 (45) embedded in Cytoscape v3.8.0 (46) then searching the STRING v11 database (https://string-db.org/). The default association/interaction threshold (STRING score >0.4) was used to map relationships between genes, gathered from protein–protein interaction databases, pathway knowledge from curated databases, automated text mining, predicted interactions, coexpression, and inferred interactions from other species. Network modules were extracted using MCL Cluster (Markov clustering) (47) in the ClusterMaker2 v1.3.1 app (48), using the Markov Cluster score for edge-weighting. The final network is presented in an edge-weighted, spring-embedded layout. Modules were repositioned manually for clarity.

Grip Strength

Fore-limb grip strength was measured as maximum tensile force using a mouse Grip Strength Meter from Columbus Instruments (Columbus, OH) with a sensor range of 0–5 000 g and accuracy of 0.15% (7). A maximum of 3 pulls of the Grip Strength Meter was recorded between 9 and 11 am for all mice; the test was performed daily over 3 consecutive days, and the daily maximum values were averaged.

Traveling Activity

A single mouse was placed in a new cage, and the number of times the mouse crossed completely over the midline of the cage over a 5-minute period was counted as walking activity. Standing activity (rearing behavior) was measured by counting each time a mouse balanced itself on its hind paws while extending its body vertically over a 5-minute period, either leaning on the wall of the cage or without cage support.

Treadmill Testing

Mice completed a 3-day exercise protocol using a treadmill with an incline of 15° from Columbus Instruments (Columbus, OH) using the protocol detailed in Ma and Nidadavolu et al. (8). Briefly, the treadmill was started at 0 m/min and increased to 18 m/min over 90 minutes. When a mouse fell onto the electrified grid at the end of the treadmill from fatigue, they received an electric shock of low intensity. The number of shocks administered to each mouse was quantified at every 15 minutes. Number of falls were used for the analysis in each time point.

Frailty Index

The frailty index assessment and scoring followed a previously described protocol (15). In brief, seven hIL-6 male mice were evaluated at baseline and 5-months postinduction for integumentary, physical, musculoskeletal, vestibulocochlear, auditory, ocular, nasal, digestive, urogenital, and respiratory decreases commonly observed in frail subjects. A total of 29 clinical signs of frailty were observed, excluding individual weight and temperature. Scoring severity of frailty constituted a simple 0–1 score per physical composition item, in which 0 indicated no sign of deficit, 0.5 eluted mild deficit, and 1 indicated severe deficit. Additional materials included a training clicker to evaluate auditory perception and a paintbrush to conduct grimace scale and menace reflex examinations.

Rotarod Balance Measurements

Rotarod ENV-578M from Med Associates, Inc. (Fairfax, VT) procedures followed those outlined by Stanford Behavioral and Functional Neuroscience Laboratory (49). Briefly, seven hIL-6 male mice were acclimated to the room environment and introduced to the rotarod before recording. After acclimation, subjects were placed onto the apparatus at the default speed (5 rpm). Using RotaRod Version 1.4.1 software, speed was then increased to 40 rpm over the course of 300 seconds (1 rpm per 8.3 seconds). The trial period began after acceleration and ended when the animal fell off the rotarod onto a padded surface below. After 300 seconds, all mice were placed in their home cages, and testing repeated after a 15-minute rest period.

Serum Metabolomics Analysis

Metabolomics assessment was performed at the Johns Hopkins Medical Institution’s Metabolomics Facility. Serum from mice at baseline and 6 weeks postinduction was processed according to the previously published protocol (8). Briefly, following methanol extraction, the supernatant containing the metabolite fraction resuspended in 50% acetonitrile in mass-spectrometry grade water was measured using mass spectroscopy. The protein fraction was used to standardize metabolite intensities by protein concentration. Metabolomics data were obtained with an Agilent 6490 triplequadrupole (QQQ) mass spectrometer equipped with an Agilent 1260 HPLC system (Santa Clara, CA). Samples were maintained at a stable sample temperature of 4°C throughout the analysis.

RT-qPCR Analysis

Under sterile conditions, gastrocnemius tissue was harvested and homogenized with bead mill Precellys Evolution tissue homogenizer using Cryolys Evolution cooling system (Bertin Technologies, Montigny-le-Bretonneux, France) from 4 male mice for each group. Samples were homogenized using 1.4 mm ceramic (zirconium oxide) beads at 4°C with the following parameters: 7 200 rpm for 20 seconds 3 times with a 15-second pause in between at 4°C. For RNA isolation and preparation of cDNA, 25–30 mg of the sample were utilized. RNeasy Fibrous Tissue Mini Kit (Qiagen) was used to isolate RNA, and samples were digested with DNase using RNase-Free DNase Set (Qiagen) to remove genomic DNA contamination. The RNA concentration and purity of the samples were assessed by UV–Vis spectrophotometer (NanoDrop, Wilmington, DE). An RNA absorbance ratio of A260/280 ≤2.1 to ≥1.8 was accepted as satisfactory, ensuring high-quality samples were used. RNA was stored at −80°C.

cDNA was synthesized using Applied Biosystems High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Polymerase chain reaction cycling parameters were 10 minutes at 25°C, 120 minutes at 37°C, and 5 minutes at 85°C. Polymerase chain reaction cycling parameters and all primers were optimized using QuantStudioTM 6 Pro Real-Time PCR Systems (Applied Biosystems, Foster City, CA). The TaqMan Fast Advanced Master Mix (Applied Biosystems) was used according to instructions provided by the manufacturer. TaqMan assay primers were used for Myh2 (Mm01332564_m1), Myh4 (Mm01332541_m1), Myh6 (Mm00440359_m1), Myh7 (Mm00600555_m1), and RPL13A (Mm05910660_g1) gene expression. The relative gene expressions for Myh2, Myh4, Myh6, and Myh7 were calculated with the 2^−ΔΔCT method, and mouse RPL13A was used as housekeeping control for normalization (50).

RT2 Profiler PCR Array

Total RNA was isolated from gastrocnemius muscles as described previously in our RT-qPCR methods using the RNeasy Fibrous Tissue Mini Kit (Qiagen). Four male mice were used for each group. RNA samples were assessed by UV–Vis spectrophotometer (NanoDrop) and demonstrated consistent quality with absorbance ratios of A260/280 ≤2.1 to ≥1. 8 and concentrations >40 ng/μL. Preparation of cDNA was performed using the RT2 First Strand Kit (Qiagen), including a genomic DNA elimination buffer step. Cycling parameters were 15 minutes at 42°C and 5 minutes at 95°C. Comparative gene expression analysis was performed using the aging-pathway mouse-specific RT² Profiler PCR Array (Qiagen, PAMM-178Z). Real-time PCR was completed on the QuantStudioTM 6 Pro Real-Time PCR Systems (Applied Biosystems) with the cycling parameters of 10 minutes at 95°C and 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. All plates passed quality control assessing genomic contamination, PCR efficiency, and quality of reverse transcription reaction. Array data were analyzed using the web-based GeneGlobe Data Analysis Center by Qiagen, and fold changes of gene expression were obtained with the 2^−(ΔΔCT) method normalized to the housekeeping GAPDH. The geometric mean was used for normalization.

Quantification Analysis of hIL-6

Total RNA was isolated from gastrocnemius muscles as described previously in our RT-qPCR methods using the RNeasy Fibrous Tissue Mini Kit (Qiagen). Four and five male mice were used for the uninduced and induced groups, respectively. RNA quality control, generation of cDNA, and PCR cycling parameters are described in RT-qPCR analysis methods. Prepared cDNA samples were sent for processing at the Genetic Resources Core Facility, Johns Hopkins Department of Genetic Medicine (RRID:SCR_018669). The digital PCR assay for absolute quantification was conducted using the QIAcuity (Qiagen) firmware version 2.0. Briefly, undiluted sample was added to a mastermix solution and partitioned across each well of the instrument plate. Samples were thermocycled using the cycling parameters of 120 seconds at 95°C and 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. Assay data were analyzed using the software QIAcuity Suite v.2.1.8 by Qiagen.

Fluorescent Microscopy

Muscle tissues were embedded in Tissue-Tek O.C.T. Compound (Electron Microscopy Sciences), and multiple thin sections (10 μm) were cut using a Leica cryostat (Leica CM3050S, Wetzlar, Germany). Four male mice were used for each group. Subsequently, the sections were stained with Masson’s Trichrome (Polysciences, Inc., Warrington, PA) or using standard immunofluorescence techniques. Masson’s Trichrome staining was carried out according to the manufacturer’s protocol with the addition of a 1-hour 10% formalin fix at room temperature (RT) before the fixation in Bouin’s solution. For immunostaining, the sections were brought to RT for 30 minutes for fixation with ice-cold 4% paraformaldehyde for 5 minutes, then blocked with 5% BSA/0.3% TritonX-100/PBS for 1 hour at RT, and incubated with the primary antibodies overnight at 4°C.

Type I collagen and fibronectin quantification in muscle tissue was performed as previously described (51). The following antibodies were purchased commercially: anticollagen type 1 antibody (EMD Millipore, Billerica, MA, #Ab745) and antifibronectin antibody (Abcam, Boston, MA, #ab2413). Sections were then incubated with secondary IgG (H) antibody, Alexa Fluor 647 Conjugate (Invitrogen, Carlsbad, CA, #27040) at RT for 1 hour. Slides were mounted with ProLong Diamond Antifade Mountant with DAPI (Invitrogen, # P36966). Tiled images were taken at 10×, 0.3 NA, PlanNeoflaur on a Zeiss LSM 800 (Oberkochen, Germany). Three to five z-stacks were collected for each tissue sample for quantification at 20×/0.8 NA Plan Apo on the Zeiss LSM 80. Imaging and quantification were done at the JHU microscope facility by blinded examiners. Automated quantification of the mean intensity of the fluorescent signal was done using Volocity image analysis software (v6.3, PerkinElmer, Waltham, MA).

Confocal Imaging and Semiautomated Quantitation

Images were processed with Volocity 3D Image Analysis Software (PerkinElmer) using algorithms developed to analyze object overlap and count individual structures as previously described (13). The density of mCherry mass in tissue sections was counted with a threshold offset of 110 to select only the brightest units using the Automatic method with Volocity. Then, EGFP-positive bright structures were found to be similar to mCherry. Finally, mCherry-positive objects, which were also EGFP-positive, were excluded from the count using the “Exclude touching objects” command with Volocity to quantify mCherry-only structures.

In Vivo Contractility

To characterize the association between hIL-6 and neuromuscular decline, TetO-hIL-6^mitoQC gastrocnemius muscle strength was evaluated using in vivo, nerve-evoked contractile function assessment as previously described (16). Five male mice were used in this experiment. All contractile testing was performed under isoflurane inhalation anesthesia delivered by Kent Scientific SomnoSuite Small Animal Anesthesia system (Torrington, CT). Mice were induced using 3%–5% anesthesia and maintained using 1%–2% anesthesia, with a flow rate of 0.5–1 L/min. Anesthetized mice were placed in a supine position on a Far Infrared warming pad (37°C), and the knee was fixed so that a 90° angle was secured between the femur and tibia. The foot was subsequently secured to the 300C-FP footplate of an Aurora Scientific Incorporated (Aurora, ON, Canada) High Power Bi-Phase Current Stimulator (701B) and Dual-Mode Lever System (305C-LR-FP). The foot was positioned to create a 90° angle between the tibia and the foot itself. Disposable Subdermal Needle Electrodes (0.5”27G) were placed in the sciatic nerve region (stimulatory) and within adjacent muscle tendons. The current was individually adjusted for each mouse before force testing to ensure maximal isometric force. The force versus frequency relationship was determined using 500-ms trains of pulses between 1 and 150 Hz frequencies. All contractile force data were first analyzed using DMAv5.501 High Throughput analysis software (Aurora Scientific).

Statistical Analysis

Within and between mouse outcome changes were first explored using visual displays and then using regression models to account for within-mouse repeated measures. Generalized linear mixed effect models with random intercepts were used to compare outcome measures before versus after induction. The models included robust variance estimates. When each period (ie, pre- vs postinduction) had multiple measurements, the base model was extended to include measurement time, group and measurement time by group interaction or period (induction vs not), and measurement time by period interaction. Wald test for the interaction term significant at .05 level indicated that changes in measures differed by group or period. Paired and unpaired Student t tests were used for pairwise comparisons. Two-way repeated measures analysis of variance (ANOVA) (frequency × time) was used to compare the force. When the p value of the ANOVA achieved significance, a post hoc Tukey test was used. A significant difference was accepted when p < .05. Statistical data analysis was conducted using GraphPad Prism9 (Boston, MA).

Study Approval

All animal-related experimental procedures performed in this study were approved by the Johns Hopkins University animal care and use committee (protocol number MO21M319).

Supplementary Material

Supplementary data are available at The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences online.

Funding

This study was supported by the Johns Hopkins Older Americans Independence Center National Institute on Aging (P30 AG021334), National Institutes of Health (NIH; grant R01AG046441), Glenn Foundation for Medical Research and AFAR Grants for Junior Faculty (L.S.N.), Bright Focus Foundation Research Award (P.M.A.), Nathan W. and Margaret T. Shock Aging Research Foundation, and Nathan Shock Scholar in Aging (P.M.A.). D.B.F. was supported by NIH grant R01 HL134821 and American Heart Association grant 18TPA34170575.

Conflict of Interest

None.

Author Contributions

L.S.N., T.C., A.L., J.D.W., and P.M.A participated in study concept and design. L.S.N., R.M.-R, H.Y., J.D.W., and P.M.A. maintained the aging animal colonies for the mice used in this study. A.L., L.S.N., C.Z., and T.N were responsible for the acquisition and analysis of the metabolomic data. C.C., G.Y., L.S.N., T.C., J.D.W., and P.M.A participated in the analysis, interpretation of the data, and drafting the manuscript and approved the final version. T.C., L.S.N., and T.B. were responsible for in vivo muscle physiology experiments and muscle analysis. B.S. was responsible for sectioning, staining, and microscopy for muscle sections. Y.W. and A.M.G. were responsible for tissue preparation and qRT-PCR studies of skeletal muscle. C.A., L.F., C.C.T., D.B.F., and L.S.N. were responsible for the RNAseq experiments and analysis. J.E.F. was responsible for artwork. L.S.N., C.C, A.L., T.C., J.D.W., and P.M.A. were responsible for the critical revision of the manuscript for important intellectual content and approval of the final version. L.S.N., C.C., T.C., J.D.W., G.Y., and P.M.A. had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis.