Aging augments the impact of influenza respiratory tract infection on mobility impairments, muscle-localized inflammation, and muscle atrophy

Following influenza infection aged mice have prolonged weight loss and elevated lung viral titers

Young and aged mice were infected intranasally with a sublethal dose (500 EID₅₀) of influenza A/PR/8/34 (PR8). Percent weight loss following infection is more severe and prolonged in aged mice (Fig 1A). While young mice begin to gain body mass by 10 days post infection (DPI), aged mice do not recover as quickly and differences between young and aged mice exist on 10-15 DPI. In the aged groups, increased weight loss is accompanied by slower viral clearance (Fig 1B) as measured by influenza polymerase (PA) copies in whole lung tissue via RT-qPCR. Although weight loss is a common marker used for pathogenicity in mice, it is rarely considered as a relevant outcome measure and mechanisms involved remain unknown.

Influenza infection induces impairments in voluntary locomotor activity and gait parameters in young and aged mice

To assess functional decrements associated with flu infection, we examined both voluntary activity levels, as well as more sensitive postural and kinematic gait alterations. Decreased voluntary locomotor activity was evident by 3 DPI in both young and aged mice as assessed by beam breaks in the open field test (Fig 2A). Diminished activity persisted through 20 DPI. Aged mice had fewer beam breaks per minute than their young counterparts on 11 (77.9% fewer), 15 (67.5% fewer), and 20 DPI (63.1% fewer) indicating that flu-induced decreased voluntary locomotor activity is more pronounced and prolonged in aged mice.

More detailed analysis of walking patterns was performed using the DigiGait system, which employs ventral plane videography to assess both spatial and temporal indices of gait at a given speed. Preliminary studies determined that 16cm/s was a speed that both young and aged mice could complete without difficulty and have consistent gait patterns for analysis (data not shown). Throughout the course of flu infection, no significant changes in stride, swing, or stance duration were observed (data not shown); however, alterations in postural components and acceleration parameters existed. By 7 DPI mice reduced stance width of both the fore (Fig 2B) and hind (Fig 2E) limbs by 15% and 13%, respectively, compared to baseline. Interestingly, aged mice initially had a wider fore limb stance, but these differences were concealed later in the infection. More pronounced differences in flu-induced gait alterations between the young and aged mice were observed in the midline distance of the fore (Fig 2C) and hind limbs (Fig 2F) later in infection, with hind limb midline distance being 46% narrower in aged mice compared to young mice on 20 DPI. This indicates aged mice are reaching less from their center with every step, perhaps due to increased muscle and joint pain limiting mobility. Additionally, later in the infection aged mice have increased stride length variability in the hind limbs compared to the young mice (Fig 2J). Gait symmetry index of the fore/hind limbs (Fig 2D) was increased in both young and aged mice at 11 DPI and 15 DPI as well, indicating a greater number of steps were taken with the fore limbs compared to the hind limbs.

Acceleration and deceleration measures were also affected by flu infection. The maximal rate of change of paw area contact during the braking phase (Max dA/dt), or how rapidly the mouse decelerates, is decreased in the fore (Fig 2H) and hind limbs (Fig 2K) later in the infection, though more pronounced in the hind limbs. Similarly, the maximal rate of change in paw area contact during the propulsion phase (Min dA/dt), or how rapidly the mouse propels itself into the next step, is decreased in the hind limbs by 7 DPI, but recovered by 20 DPI (Fig 2L). Both acceleration and deceleration are important parameters indicating the rate of force development, an important component of muscle health and quality. Interestingly, decreased Max dA/dt was still evident at 20 DPI.

Taken together, alterations in gait include decreased force development accompanied by decreased reaching distance and a narrower stance, more marked in the hind limbs and in aged mice. Diminished voluntary locomotor activity is prolonged in the aged mice as well. Functional impairments may be indicative of flu-induced muscle inflammation and damage, and that this is more severe and prolonged in aged mice.

Influenza infection induces altered inflammatory gene expression in skeletal muscle

Since alterations in gait were primarily in hind limbs, the gastrocnemius (gastroc) muscle gene expression was further examined over the course of flu infection in young and aged mice. As hypothesized, inflammatory gene expression in the gastroc was altered over the course of the flu infection. By 7 DPI flu induced increased gene expression of interleukin (IL)-6 (IL6) and IL-6 receptor alpha (IL6RA) (Fig 3A and Fig 3B).

IL6RA expression remained elevated by 3.6 fold in the aged mice at 11 DPI while the young mice returned to baseline expression levels. Aged gastroc had 2.7 fold increased expression of tumor necrosis factor (TNF) at baseline and these differences were intensified on 15 DPI with 6.7 fold greater expression of TNF in aged gastroc (Fig 2C). IL-6 and TNF are two key inflammatory mediators in skeletal muscle degeneration and repair, signaling through STAT3 and NFκB, respectively, that have vast effects on pro-inflammatory signaling, protein degradation, and atrophy gene induction (reviewed in [29]). Additionally, a dramatic 43 fold increase in expression of chemokine (C-X-C Motif) Ligand 10 (CXCL10), a predominant player in T helper (Th) 1 responses that recruits immune cells, particularly T lymphocytes expressing its receptor CXCR3, into tissue, was observed in the aged gastroc at 11 DPI (Fig 3D). Prolonged and exaggerated levels of CXCL10 would increase immune cell recruitment and muscle inflammation, and potentially impair muscle regeneration processes.

Taking these results collectively, flu induced lingering inflammation in the aged muscle. As it is known that these inflammatory mediators signal through NFκB and other pathways that induce muscle atrophy and protein degradation, we next examined the expression of genes involved in these processes.

Increased expression of protein degradation and muscle atrophy genes post influenza infection is prolonged in aged mice

Although there are many pathways involved in protein degradation and atrophy, we focused on the ubiquitin proteasome pathway, primarily the muscle-specific E3 ubiquitin ligases atrogin-1 (also known as muscle atrophy F-box (MAFbx)) and muscle RING finger 1 (MuRF1), as the majority of literature to date shows increased atrogin-1 and/or MuRF1 expression at some point during almost all conditions of muscle wasting and atrophy [30]. As part of the ubiquitin proteasome pathway, atrogin-1 and MuRF1 control the ubiquitination and therefore degradation of specific target proteins in response to key signals. Key initiators of these pathways include inflammatory cytokines, as well as myostatin, glucocorticoids, FoxO transcription factors, and others. Following flu infection, we observed increased gastroc gene expression of both myostatin (MSTN) and FOXO1 at 7 DPI (Fig 4C and 4D), by 2.0 and 2.3 fold change, respectively. FOXO1 remained elevated at 11 DPI, while MSTN elevation was only transient. Interestingly, FOXO1 was elevated in the young compared to aged mice at baseline and 3 DPI.

Corresponding with elevated cytokines, myostatin, and FOXO1 gene expression, both ATROGIN1 and MuRF1 gene expression were elevated with flu infection at 7 DPI by 2.1 and 5.3 fold, respectively. Moreover, in accordance with the prolonged inflammation, aged mice ATROGIN1 and MuRF1 expression remained elevated at 11 DPI (Fig 4A and Fig 4B). While in young mice ATROGIN1 and MuRF1 expression was already decreased to below baseline levels (−2.6 and −2.3 fold, respectively) at this time point, aged mice expression remained elevated 2.3 and 2.5 fold, respectively. Interestingly, ATROGIN1 was downregulated by 15 DPI, perhaps indicating an attempt to limit muscle degradation and begin repair processes. Baseline expression of MuRF1 was decreased in aged mice, while no significant baseline differences existed in ATROGIN1 expression. The influence of atrogin-1 and MuRF1 mRNA and protein expression on solely age-related muscle loss has yet to be resolved; some murine studies agree with our findings and show baseline suppression [31], while others have shown increased expression [32], and human studies have shown no differences [33, 34]. Nonetheless, induction of atrogin-1 and MuRF1 in response to flu infection indicates increased ubiquitination and proteolysis within the muscle. Furthermore, expression of two ubiquitin proteasome encoding genes, ubiquitin B (UBB) and ubiquitin C (UBC), was also increased with flu infection with a 2.6 and 3.6 fold increase at 7 DPI (Fig 4E and Fig 4F). Similarly, these increases were more dramatic and prolonged in aged mice. In young mice UBC expression peaked at 7 DPI, while aged mice continued increasing expression to a staggering 9.8 fold increase at 11 DPI. The increased levels of UBB and UBC, as well as ATROGIN1 and MuRF-1 indicate a catabolic environment of increased proteolysis, promoting muscle atrophy.

Influenza infection decreases expression of positive regulators of muscle growth

Increased atrogin1/MuRF-1 pathway components, as well as increased inflammatory cytokines, are associated with diminished muscle growth; however, following injury or atrophy muscle generally has a remarkable capacity to regenerate and repair. Muscle regeneration through myogenesis is regulated at many key steps by myogenic regulatory factors (MRFs), myocyte enhancer binding factors 2 (Mef2), and other growth factors. Thus, we next examined the gene expression of positive regulators of muscle growth following flu infection to determine if repair processes are suppressed during this time and/or if muscle regeneration follows flu-induced atrophic responses. At 7 DPI when atrophy and ubiquitin genes are upregulated, gastroc IGF1 expression is reduced in both young and aged mice (Fig 5A). This was transient in the young mice, but remained downregulated in the aged mice through 11 DPI. During flu infection gastroc PAX7, a marker of satellite cells which are critical for adding myonuclei and regenerating muscle tissue, expression is decreased at 7 DPI in both young and aged mice, but to a greater degree in aged mice, less than one fold reduction compared to ∼3.5 fold reduction in the aged (Fig 5D). Similarly at this time the expression of both MYOD1 and MYOG, key MRFs, was decreased to a greater degree, with approximately 7 and 2 fold reductions, respectively, in the aged mice (Fig 5E and 5F). MEF2C, which acts in concert with the MRFs to control DNA binding and transcriptional regulation, was also suppressed at this time point and remained suppressed at 11 DPI in the aged mice (Fig 5B).

In summary, flu suppressed positive regulators of muscle mass and regeneration with concurrent increases in negative regulators. Additionally, these responses were greater and/or prolonged in the aged mice.

Influenza-induced weight loss is correlated with gastroc gene expression of ubiquitin proteasome components

In order to determine if flu-induced weight loss was associated with muscle degradation, we performed correlation analysis on all variables that exhibited a time effect over the course of the flu infection. Indeed, multiple variables had significant correlations and age interactions were evident (Fig 6). In both young and aged mice FOXO1 (Fig 6A), IL6RA (Fig 6B), UBB (Fig 6C), UBC (Fig 6D), and MuRF1 (Fig 6E) were significantly correlated with percent body mass where UBC accounted for the greatest variation with R² = 0.487 and 0.541 in young and aged mice, respectively (Fig 6D). Interestingly, a significant correlation was observed with IGF1 expression in young mice that was not evident in aged mice (Fig 6G), and the opposite was observed with ATROGIN1 expression (Fig 6F).

Further, all variables that had any significant correlations per age group were placed in a step-wise multiple regression analysis to determine if multiple variables could account for greater variability. In young mice UBC, IGF1, and MuRF1 expression accounted for approximately 71% of variability seen in body mass changes (adjusted R²= 0.714, p=0.007), with additional variables not adding significantly to the model. In contrast, in aged mice UBC expression accounted for approximately 58% of the variability in body mass (adjusted R²= 0.576, p<0.001) with no other variables tested (ATROGIN1, MuRF1, IL6RA, UBB, and FOXO1) adding significantly to the model.

Negligible viral copies detected in the gastrocnemius post influenza infection in vivo

Since flu-induced weight loss was correlated with gastroc gene expression, we next examined a potential mechanism; direct infection of the skeletal muscle in vivo. RNA harvested from the gastroc was probed for the flu PA similar to as performed on whole lung tissue. Though rare case studies have identified virus particles in muscle biopsies of influenza infected humans with myalgia, this highly unusual finding may be limited to critically ill and preterminal cases [12, 21-23]. Indeed, flu PA was not detectable in the majority of gastroc tissue and the few samples that had detectable levels of flu PA had only negligible levels (Fig 7A). Since there was an upregulation of muscle degradation genes in all mice by day 7 post infection, the mechanism of direct infection occurring during a natural infection is not supported by these results.

In vitro myoblasts do not harbor active influenza infection

To further investigate the possibility that direct infection of skeletal muscle could occur, we harvested leg skeletal muscle (gastroc, vastus medialis, vastus lateralis, soleus, and anterior tibialis) of uninfected young and aged mice. Myoblasts were harvested, grown, purified, and re-plated prior to infection. Myoblasts were incubated for 1 hour with 0 EID₅₀, 10 EID₅₀, or 100 EID₅₀ PR8. Supernatants were collected at 24, 48, and 72 hours post infection and cytokine/chemokine concentration were determined via multiplex (Fig 7C, 7E, and 7F). IL-6 secretion was significantly greater from the aged myoblasts, (age effect: p<0.001), but was not affected by infection or infection over time (infection effect: p=0.600, infection*time interaction effect: p=0.583, Fig 7C). The same pattern was observed for CXCL10 (age: p<0.001, infection: p=0.609, infection*time: p =0.684, Fig 7D), CXCL1 (age: p<0.001, infection: p=0.865, infection*time: p =0.121, Fig 7E), and CCL2 (age: p<0.001, infection: p=0.921, infection*time: p =0.263, Fig 7F). Thus, cytokine secretion was affected by age, but not by flu infection. Myoblasts were harvested and RNA was extracted at 96 hours post infection to probe for flu PA copies as previously done. Though some studies [18-20] show that myoblasts can be infected by influenza in vitro, our results indicate that both young and aged myoblasts are not susceptible to a productive flu infection (Fig 7B).

Mice

Young (2.5-4 month old) C57BL/6 male mice were obtained from Charles River Laboratories and aged (19-22 month old) C57BL/6 male mice were obtained through the National Institute on Aging rodent colony. All mice were housed in a climate controlled environment with 12:12 light:dark cycle and fed standard rodent chow and water ad libitum. All procedures were approved by the University of Connecticut Medical School IACUC (protocol 100705) and carried out in accordance with these regulations. All mice underwent gross pathological examination at time of sacrifice and animals with obvious pathology were excluded from the study.

Viral infection

Mice were anesthetized with isoflurane and intranasally inoculated with 50μl of 500 EID₅₀ of influenza virus A/PR/8/34 (PR8). Mice were weighed daily to monitor infection progression. At time points indicated, whole lung tissue was homogenized and RNA was isolated via RNeasy Mini Kit (Qiagen Inc., Valencia, CA). RNA was reverse transcribed with iScript cDNA synthesis Kit (Bio-Rad Laboratories, Inc., Hercules, CA) and flu viral copies were detected via reverse transcription quantitative PCR of flu acid polymerase (PA).

Voluntary locomotor activity

Spontaneous voluntary locomotor activity was measured via open field test at time points indicated. All tests were performed between 6-8am to control for diurnal variations. Following acclimation to the dim-lit testing room (at least 1 hour), mice were placed in the center of the photobeam activity system-open field (PAS-OF, 16”x16”x15” acrylic animal enclosure, San Diego Instruments, San Diego, CA) and their activity was recorded for 20 minutes. The first 5 minutes was excluded as this is generally considered to be exploratory behavior rather than general voluntary locomotor activity. The number of beam breaks per minute during the last 15 minutes was then used to assess voluntary locomotor activity.

Gait analysis

Gait analysis was performed using the DigiGait instrument (Mouse Specifics, Inc, Quincy, MA) and software (DigiGait Imager 4.0.0 and DigiGait Analysis 11.5, Mouse Specifics, Inc). The DigiGait instrument consists of a clear treadmill with a high-speed camera mounted underneath that collects images at 147 frames per second for high resolution of postural temporal gait parameters. Mice run within a 2” wide acrylic running chamber at set speeds. The ventral plane videos are analyzed with the DigiGait software which identifies portions of the paw that are in contact with the treadmill belt to produce both postural and kinematic gait parameters. Mice were introduced to the DigiGait system at a low speed (10cm/sec) briefly (30 seconds) prior to the initial testing. Mice were allowed to acclimate to the dim-lit room for 1 hour prior to each testing period and all tests were performed between 6-8am. Mice ran at the testing speed (16cm/sec) until approximately 5 seconds of consecutive walking was recorded and this video segment was analyzed via DigiGait software.

Gastrocnemius Reverse-Transcription quantitative PCR (RT-qPCR)

At the time points indicated for gastrocnemius gene expression mice were fasted with the exception of water for 4-6 hours prior to sacrifice to minimize potential confounding results due to postprandial muscle protein synthesis. The gastrocnemius muscle was dissected and placed in RNAlater (Qiagen Inc.) overnight at 4°C. RNAlater was removed and gastrocnemius was frozen at 80°C until RNA extraction. The muscle was homogenized and RNA was extracted via RNeasy Fibrous Tissue Mini Kit (Qiagen Inc.). RNA quantity and quality was assessed with Nanodrop 2000c (Thermo Scientific, Waltham, MA) and was reverse transcribed via iScript Advanced cDNA synthesis Kit (Bio-Rad Laboratories, Inc.). RT-qPCR was performed using custom designed PCR plates with predesigned commercially available primers (Bio-Rad Laboratories, Inc.). Gene expression was calculated via a modified Pflaffl method utilizing multiple reference genes (RPS18 and TBP, which showed the least variability between conditions and thus suitable reference genes) and normalized to gene expression of young mice at Day 0 prior to infection to give comparable fold changes.

In vitro myoblast culture and infection

Harvested leg skeletal muscle (gastroc, vastus medialis, vastus lateralis, soleus, anterior tibialis) from uninfected young and aged mice was incubated in collagenase type IIA and dissociated in growth media (Dulbecco modified Eagle's medium (D-MEM) supplemented with 20% fetal bovine serum (FBS), 1% penicillin/streptomycin, 5ng/mL basic fibroblast growth factor (bFGF)) and grown on extracellular matrix-coated plates (extracellular matrix, Sigma-Aldrich Corp., St. Louis, MO). When myoblasts reached 50-60% confluency, they were purified via plating on an uncoated Petri dish to remove any adherent fibroblasts. Myoblasts were then seeded at 150,000 cells per well in coated 6-well plates and allowed to grow to 50-60% confluency prior to infection. For infections, myoblasts were washed twice with phosphate buffered saline (PBS) and incubated for one hour with 0, 10, or 100 EID₅₀ PR8 in D-MEM supplemented with 1% penicillin/strepto-mycin. Then myoblasts were washed twice with PBS and cultured with growth media for the remaining days. Culture media at time points indicated was analyzed for cytokine/chemokine concentrations via multiplex (25Plex Magnetic Bead Panel, EMD Millipore, Billerica, MA). Most cytokine/chemokines were below detectable limits (GCSF, GMCSF, IFNγ, IL10, IL12p40, IL12p70, IL13, IL15, IL17, IL1β, IL1α, IL2, IL4, IL5, IL7, IL9, MIP1β, MIP1α, MIP2, RANTES, TNFα). Total RNA was extracted via TRIzol (Ambion, Life Technologies, Grand Island, NY) according to manufacturer's recommendation. RNA was reverse transcribed and flu viral copies were detected via real-time quantitative PCR of flu PA.

Statistical analysis

Weight loss, viral titers, functional performance, and log-transformed gastroc gene expression results were analyzed via 2-way ANOVA (age x time point) with Bonferroni post hoc corrections when indicated with significance set a p<0.05. Genes were considered differentially expressed if fold changes were ≥ 2 and p<0.05. In vitro myoblast supernatant and RNA were analyzed via 3-way ANOVA (age x infection condition x time point) and 2-way ANOVA (age x infection condition), respectively, with Bonferroni post hoc corrections when indicated with significance set at p≤0.05. Univariate linear regression was used to compare the relationship between gastroc gene expression and percentage weight loss independently for each variable that showed time effects with significance set at p<0.05. Step-wise multiple regression analysis was used to determine if multiple variables could better predict percent weight loss.