Impaired phagocytic function in CX3CR1+ tissue‐resident skeletal muscle macrophages prevents muscle recovery after influenza A virus‐induced pneumonia in old mice

Abstract Skeletal muscle dysfunction in survivors of pneumonia disproportionately affects older individuals in whom it causes substantial morbidity. We found that skeletal muscle recovery was impaired in old compared with young mice after influenza A virus‐induced pneumonia. In young mice, recovery of muscle loss was associated with expansion of tissue‐resident skeletal muscle macrophages and downregulation of MHC II expression, followed by a proliferation of muscle satellite cells. These findings were absent in old mice and in mice deficient in Cx3cr1. Transcriptomic profiling of tissue‐resident skeletal muscle macrophages from old compared with young mice showed downregulation of pathways associated with phagocytosis and proteostasis, and persistent upregulation of inflammatory pathways. Consistently, skeletal muscle macrophages from old mice failed to downregulate MHCII expression during recovery from influenza A virus‐induced pneumonia and showed impaired phagocytic function in vitro. Like old animals, mice deficient in the phagocytic receptor Mertk showed no macrophage expansion, MHCII downregulation, or satellite cell proliferation and failed to recover skeletal muscle function after influenza A pneumonia. Our data suggest that a loss of phagocytic function in a CX3CR1+ tissue‐resident skeletal muscle macrophage population in old mice precludes satellite cell proliferation and recovery of skeletal muscle function after influenza A pneumonia.

Infection with the influenza A virus is among the most common causes of pneumonia, and like pneumonia induced by the severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2), the mortality attributable to influenza A pneumonia disproportionately impacts the elderly (Jain et al., 2015;Kanegai et al., 2016;Ortiz et al., 2013;Zhou et al., 2020). Influenza A virus-induced pneumonia in mice is an attractive model to study the effects of aging on pneumonia-induced muscle function as it recapitulates the disproportionate skeletal muscle dysfunction reported in older survivors of pneumonia (Bartley et al., 2016). Furthermore, as the influenza A virus is trophic for the lung epithelium, there is little if any viremia (Radigan et al., 2019). In previous work foundational to the current study, we demonstrated that influenza A infection induces the expression of the muscle-specific E3 ligase FBXO32 (atrogin-1) and that Fbxo32 −/− mice were resistant to influenza A-induced muscle loss. Furthermore, we found that the administration of tocilizumab, an antibody targeting the IL-6 receptor, to mice, prevented the induction of atrogin-1 and the loss of muscle function after influenza A infection without affecting the severity of the lung injury or viral clearance. We concluded that the release of IL-6 and possibly other endocrine signals from the injured lung induce the active degradation of skeletal muscle during acute influenza A-induced pneumonia.
Because skeletal muscle fibers are post-mitotic, muscle myofibers lost during injury must be replaced through activation, proliferation, and differentiation of muscle satellite cells, a muscle-specific progenitor cell population (Brack & Rando, 2012;Wang & Rudnicki, 2011). After injury, muscle satellite cells undergo rapid proliferation, expand in number, and fuse with damaged myofibers, restoring the number of myonuclei, myofiber content, and volume (Munoz-Canoves & Michele, 2013;Sala et al., 2019;Tierney et al., 2014). The process of muscle repair has been studied in models of direct injury, for example, freezing, trauma, and toxin injection, but the mechanisms by which muscle fibers are regenerated after pneumonia are not known.
Given this background, the disproportionate loss of skeletal muscle mass in older pneumonia survivors could be attributable to enhanced muscle loss during the acute infection, impaired recovery, or both. We report that when the dose of influenza A administered to young and old mice is titrated to induce equivalent mortality, old and young mice suffer a nearly proportionate loss of skeletal muscle size and function. However, while young mice fully recovered muscle function to pre-pneumonia levels 60 days after injury, old mice did not. In young mice, a population of CX3CR1-expressing tissue-resident macrophages in the skeletal muscle expanded and downregulated MHCII expression after influenza A infection in the absence of recruitment of monocytes from the bone marrow. This was followed by proliferation of muscle satellite cells. All of these findings were absent in old mice and mice deficient in Cx3cr1. Transcriptomic profiling suggested a loss of phagocytic function in tissue-resident macrophages from the skeletal muscle of old mice, which was confirmed using in vitro assays. Knockout of Mertk, a tyrosine kinase involved in macrophage phagocytosis, phenocopied the molecular and physiologic findings in old mice. Our findings suggest phagocytosis-induced signaling in CX3CR1 + tissue-resident skeletal muscle macrophages is necessary for satellite cell proliferation during muscle recovery after influenza A virus-induced pneumonia.

| Old mice do not recover skeletal muscle function after influenza A-induced pneumonia
In mice, the course of influenza A follows a predictable and consistent pattern in which replicating viruses are detectable by plaque assay over the first 2-5 days of the infection accompanied by the recruitment of neutrophils and monocyte-derived alveolar macrophages (Peteranderl et al., 2016). Mice then develop progressive hypoxemia, respiratory failure, and death (Traylor, Aeffner, & Davis, 2013), or begin a long process of lung repair, which is incomplete even months after the injury (Kanegai et al., 2016) (Figure 1a). During the acute phase of viral replication, blood cytokines including IL-6 are dramatically increased coinciding with the loss of muscle mass. We reported that the IL-6-dependent activation of FBXO32 in the skeletal muscle was necessary for the acute loss of muscle mass during IAV infection (Radigan et al., 2019).
Critically ill elderly patients with pneumonia have been reported to have both enhanced muscle loss during their ICU stay and prolonged muscle dysfunction that persists more than a year after injury. Modeling this in mice poses challenges, as the LD50 for influenza A virus is dramatically different in old mice compared with young mice (Figure 1b). Accordingly, we developed a model in F I G U R E 1 Acute influenza A virus (IAV) infection is worse in old compared with young mice. (a) Timeline of influenza A-induced pneumonia in mice. Data in this figure focus on the acute phase (red). (b) Young (4 months old; green lines) and old (20-24 months old; purple lines) mice were infected with either 15 pfu (solid lines) or 25 pfu (dotted lines) of IAV (A/WSN/33). A Kaplan-Meier survival curve is shown (n = 9-14 mice per group). Log-rank (Mantel-Cox) tests were performed, and p values are shown on graph. Each dot represents a single animal. For all data (c-k), young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV (25 pfu for young mice, 15 pfu for old mice). Mice were harvested for analysis at 0, 4, and 10 days post-infection. A two-way ANOVA with Tukey's post hoc corrections for comparison with more than three groups was used to determine statistical differences, or Student's t test for comparisons between two groups, p values are shown on graph. Box plot center lines are median, and box limits are upper and lower quartiles. Each dot represents an individual animal. (c) Percentage of starting total body weight is shown for day 4 and day 10 post-IAV infection (n = 5-7 mice per group). (d) Total protein in bronchoalveolar lavage (BAL) fluid was measured by Bradford assay (n = 4-12 mice). (e) Number of live cells in the BAL was counted using hemocytometer, and dead cells were excluded using trypan blue (n = 5-7 mice). (f) IL-6 levels in the BAL were measured by ELISA (n = 5-7 mice). (g) IL-6 levels in the serum were measured by ELISA (n = 4-8 mice). (h) Soleus muscles were excised, and soleus wet muscle weights were determined. Values are presented as a percent of naive age-matched controls (n = 8-18 mice). (i) Extensor digitorum longus (EDL) muscles were excised, and EDL wet muscle weights were measured. Values are presented as a percent of naive age-matched controls (n = 5-7 mice). (j) Soleus cross-sectional areas were determined using transverse cross sections immunostained for laminin. Values are normalized to naive age-matched controls (n = 4-7 mice). (k) Quantification (top panel) and representative Western blot (bottom panel) of atrogin-1 expression in the mouse tibia anterior muscle. Values are normalized to GAPDH (loading control) (n = 5-11 mice)   Figure 1d, e) and increased levels of IL-6 in bronchoalveolar lavage fluid ( Figure 1f) and in the plasma ( Figure 1g) compared with young mice. The moderate age-related increase in systemic inflammation corresponds with injury to the hind limb muscles that is only slightly worse in old compared with young mice. The soleus muscle wet weight decreased similarly in both young and old mice (Figure 1h), but the loss of wet weight in the extensor digitorum longus (EDL) muscle was more pronounced in old mice (Figure 1i). In the soleus muscle, the average fiber cross-sectional area (CSA) was significantly reduced in both young and old mice after influenza A virus-induced pneumonia (Figure 1j), although the proportional reductions were smaller in old mice due to a lower starting volume ( Figure S1).
Consistent with the higher levels of plasma IL-6, the increase in skeletal muscle atrogin-1 occurred more rapidly in old animals, but was similar 10 days after infection in young and old mice ( Figure 1k).
As a physiologic measure of skeletal muscle function (Figure 2a), we continuously recorded voluntary running distance on monitored exercise wheels in young and old mice over the course of influenza A infection and recovery. Mice adapted to the wheels over 14 days, after which the voluntary distance run on the wheel was consistent and similar between animals (Figure 2b), although on average it was lower in old mice (5.2 ± 1.1 and 2.1 ± 1.2 km/night/mouse in old young and old mice, respectively, Figure S2a). Once baseline running distances were established, mice were infected with influenza A virus and then monitored continuously for up to 60 days. Both young and old mice almost stopped running after the influenza A virus infection, reaching a minimum at day 10 post-infection (Figure 2b, c). However, by day 20, the distance run returned to pre-infection F I G U R E 2 Old mice fail to recover skeletal muscle function after influenza A infection. (a) Timeline of influenza A-induced pneumonia in mice. Data in this figure focus on the recovery phase (blue). (b) Monitored running wheels were placed in cages, and the voluntary distance run was measured for 14 days before animals were infected with influenza A virus (IAV). At day 0, young (4 months old; green lines) mice were infected with 25 pfu of IAV and old (20-24 months old; purple lines) mice were infected with 15 pfu IAV. Voluntary wheel running was recorded for 30 days, and the 8 p.m. to 6 a.m. average wheel rotation/mouse, normalized to the 10-day average control, is shown (n = 12 mice per group). A graph of the distance run is shown with means ± SD See also Figure S2. (c) Three-day average voluntary wheel running distance per cage normalized to the 10-day average control is shown for young mice (4 months old; green bars) and old mice (20-24 months old; purple bars). The average distance run for days −3 to −1 prior to infection with IAV (25 pfu for young mice, 15 pfu for old mice), days 9-11 post-infection, and days 28-30 post-infection is shown (n = 12 mice per group). Box plot center lines are median, box limits are upper and lower quartiles, whiskers are minimal and maximal values, a two-way ANOVA with Tukey's post hoc corrections for comparison with more than three groups was used to determine statistical differences, and p values are shown on graph. Each dot represents a single animal. (d) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV at day 0 (25 pfu for young mice, 15 pfu for old mice). Forelimb grip strength was measured at the indicated times (n = 6-12 mice per group). Box plot center lines are median, box limits are upper and lower quartiles, whiskers are minimal and maximal values, a two-way ANOVA with Tukey's post hoc corrections for comparison with more than three groups was used to determine statistical differences, and p values are shown on graph. Each dot represents a single animal

| Satellite cells and fibroadipogenic progenitors fail to expand in old mice during recovery from influenza A virus-induced pneumonia
We questioned whether the failure to recover muscle strength and function in old mice was due to a failure to activate regenerative cell populations. We used flow cytometry to simultaneously quantify muscle satellite cells (MuSCs) and fibroadipogenic progenitors (FAPs) as well as lymphoid and myeloid immune cells during muscle recovery after influenza A-induced lung injury. We identified MuSC as CD45 -CD31 -CD34 + Sca-1 -Itga7 + and FAP as CD45 -CD31 -CD34 + Sca-1 + Itga7 + (Figure 3a). In young, but not in old mice, we observed an expansion of MuSC and FAP after influenza A virus-induced pneumonia (Figure 3b-G). This suggests that muscle dysfunction induced by influenza A pneumonia is sufficient to induce regenerative cell populations in young mice, but that this response is impaired in old animals.

| Tissue-resident CX3CR1+ skeletal muscle macrophages expand and downregulate MHCII during recovery from influenza A virus-induced pneumonia
We measured myeloid and lymphoid populations in the skeletal muscle after influenza A infection using flow cytometry (Figures 4a and S3m). In young mice, we observed increases in the number of skeletal muscle macrophages (CD45+Ly6G-Siglec-F-NK1.1-CD11b+Ly6C-CD64+) after influenza A infection ( Figure 4b). This increase in skeletal muscle macrophages was not observed in old mice ( Figure 4b). In contrast, we did not observe influenza A virusinduced increases in the numbers of neutrophils, eosinophils, natural killer cells, CD11b+ dendritic cells, Ly6C-high classical monocytes or Ly6C-low non-classical monocytes, CD4 and CD8 T cells, regulatory T cells, or B cells in the skeletal muscle of either young or old mice after influenza A infection ( Figure S3).
As we observed a significant expansion in macrophages after influenza A infection in young mice, we focused our attention on these cells. In naïve mice, the majority of macrophages were represented by an MHCII high population ( Figure 4c). During recovery from F I G U R E 4 Tissue-resident skeletal muscle macrophages expand and downregulate MHCII expression after influenza A pneumonia in young, but not in old mice. (a) Gating strategy to identify myeloid cell populations in skeletal muscle tissue using flow cytometry. (b) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with influenza A virus (IAV) (25 pfu for young mice, 15 pfu for old mice). Graph shows number of macrophages per gram of tissue, before infection (day 0) and 4 and 10 days post-infection (n = 9-14 mice). (c) Representative flow cytometry showing MHCII staining in skeletal muscle macrophages from young (top two panels) or old (bottom two panels) mice before and 10 days after IAV infection. (d) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV (25 pfu for young mice, 15 pfu for old mice). Graph shows number of MHCII high macrophages in skeletal muscle per gram of tissue before infection (day 0) and 4 and 10 days after influenza A infection (n = 9-14 mice). (e) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV (25 pfu for young mice, 15 pfu for old mice). Graph shows number of MHCII low macrophages in skeletal muscle per gram of tissue before infection (day 0) and 4 and 10 days after influenza A infection (n = 9-14 mice). (f) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV (25 pfu for young mice, 15 pfu for old mice). Graph shows percentage of EdU-positive MHCII high macrophages, of total MHCII high macrophages, in skeletal muscle before infection (day 0) and 10 days after influenza A infection (n = 3-4 mice). (g) Young (4 months old; green bars) and old (20-24 months old; purple bars) mice were infected with IAV (25 pfu for young mice, 15 pfu for old mice). Graph shows percentage of EdU-positive MHCII low macrophages, of total MHCII low macrophages, in skeletal muscle before infection (day 0) and 10 days after influenza A infection (n = 3-4 mice). (h) Representative plot of macrophages from Cx3cr1 Gfp/WT mice showing heterogeneous expression of MHCII and CX3CR1 (GFP). (i) Cx3cr1 ERCre /zsGreen mice were administered tamoxifen (10 mg via oral gavage, twice, 24 hr apart). Mice were harvested 7, 28, 35, 42, and 56 days after the tamoxifen pulse. A separate cohort of mice was infected with 25 pfu IAV 28 days after the tamoxifen pulse and harvested 7, 14, and 28 days later. Representative flow cytometry plots of macrophage GFP expression at each of these time points are shown. (j) Refers to the schematic in (i). Cx3cr1 ERCre /zsGreen mice were administered a pulse of tamoxifen (10 mg via oral gavage, twice, 24 hr apart) at day 0. The percentage of GFP + cells from the indicated cell populations is shown. Some mice were infected with 25 pfu IAV 28 days after the tamoxifen pulse and harvested 7, 14, and 28 days later. Black symbols indicate naïve mice. Red symbols indicate IAV-infected mice. (k) Young wild-type mice (4 months old; green bars) and age-matched Cx3cr1 Gfp/Gfp mice (gray bars) were infected with 25 pfu of IAV. Graph represents skeletal muscle macrophage number before infection (day 0) and 7 days after influenza A infection. Values are expressed per gram of tissue (n = 4-5 mice). (l) Young wild-type mice (4 months old; green bars) and age-matched Cx3cr1 Gfp/Gfp mice (gray bars) were infected with 25 pfu of IAV. Graph represents muscle satellite cell (MuSC) number before infection (day 0) and 7 days after influenza A infection. Values are expressed per gram of tissue (n = 4-5 mice). (m) Young wild-type mice (4 months old; green bars) and age-matched Cx3cr1 Gfp/Gfp mice (gray bars) were infected with 25 pfu of IAV. Graph represents fibroadipogenic progenitor (FAP) number before infection (day 0) and 7 days after influenza A infection. Values are expressed per gram of tissue (n = 4-5 mice). For all experiments, a two-way ANOVA with Tukey's post hoc corrections for comparison with more than three groups was used to show statistical differences, and p values are shown on the graph. Each dot represents an individual animal. Also, see Figure S4  influenza A infection, the majority of macrophages were MHCII low in young, but not in old mice (Figure 4c-e). Furthermore, incorporation of EdU into MHCII low macrophages in after pneumonia-induced muscle atrophy increased in young but not old mice, indicating an age-dependent proliferation specific to this population (Figure 4f,g).
Using reporter mice in which GFP is knocked into the Cx3cr1 gene (Cx3cr1 GFP/+ ), we found that approximately 25%-30% of skeletal muscle macrophages expressed Cx3cr1 in naïve mice ( Figure S4), and both CX3CR1 + and CX3CR1macrophages included populations that were MHCII high and MHCII low (Figure 4h).
To determine whether these skeletal muscle macrophage populations are maintained by local proliferation or are replenished by monocytes recruited from the circulation, we performed fate-mapping experiments using Cx3cr1 ER-Cre mice crossed to a ZsGreen LSL reporter. We administered a tamoxifen pulse (two doses of tamoxifen, one day apart via oral gavage) to Cx3cr1 ER-Cre /ZsGreen LSL  We used the Cx3cr1 ER-Cre /ZsGreen LSL mice to determine whether the increase in skeletal muscle macrophages we observed after influenza A virus-induced pneumonia was attributable to the recruitment of monocyte-derived macrophages or the proliferation of tissue-resident macrophages. We treated Cx3cr1 ER-Cre /ZsGreen LSL with two pulses of tamoxifen one day apart, waited 28 days for GFP-labeled monocytes to disappear, infected the mice with influenza A virus, and then harvested mice 7, 14, and 28 days after infection (Figure 4i).
We did not observe any reduction in the percentage of GFP-positive macrophages, excluding the possibility that the increase in skeletal muscle macrophages following influenza A infection represented an influx of circulating cells (Figure 4i,j). Further, the finding that the number of macrophages expanded after influenza A infection combined with the constant percentage of GFP-positive cells suggests both CX3CR1-positive and CX3CR1-negative cells expand proportionately after influenza A infection (Figure 4b,j). CX3CR1 is required for the adhesion and migration of leukocytes, and in the microglia, CX3CR1 is necessary for synaptic pruning during development (Paolicelli et al., 2011). We therefore sought to determine whether CX3CR1 is necessary for the expansion of macrophages and recovery of muscle function after influenza A infection. We treated

| Transcriptomic profiling identifies downregulation of phagocytic and proteostatic pathways and upregulation of inflammatory pathways in tissue-resident skeletal muscle macrophages from old compared with young mice
To identify possible mechanisms responsible for the lack of a reparative response in tissue-resident skeletal macrophages from and S4, list of DEGs and corresponding GO processes). These transcriptomic data suggested impaired phagocytosis and increased inflammatory response in skeletal muscle macrophages from old F I G U R E 5 Tissue-resident macrophages from skeletal muscle from old mice show impaired phagocytosis. (a) Skeletal muscle macrophages were flow cytometry-sorted from naïve young (4 months old) and old (20-24 months old) mice and analyzed using RNA-seq (n = 5 mice per group). Differentially expressed genes (FDR q < 0.05) between young (green dots) and old (purple dots) mice are shown with representative genes and GO biological processes (see Tables S1 and S2 for complete list of genes and GO processes). (b) Young (4 months old) and old (20-24 months old) mice were infected with influenza A virus (25 pfu for young mice and 15 pfu for old mice) and harvested 60 days later. Skeletal muscle macrophages were flow cytometry-sorted and analyzed using RNA-seq (n = 5 mice per group). Differentially expressed genes (FDR q < 0.05) between young (green dots) and old mice (purple dots) are shown with representative genes and GO biological processes (see Tables S3 and S4 for complete list of genes and GO processes). (c) Single-cell suspensions prepared from skeletal muscle of young (4 months old) and old (20-24 months old) mice were incubated with antibodies to detect myeloid cells and serum-opsonized Fluoresbrite microspheres. Uptake of the fluorescently labeled microspheres by macrophages was measured using flow cytometry. Non-opsonized beads were used as a negative control. Representative contour plots are shown, and numbers indicate percentage from the parent population (gated on macrophages; see Figure 4a

| The phagocytic receptor Mertk is necessary for skeletal muscle repair after influenza A-induced pneumonia
Our transcriptomic data suggested impaired phagocytosis in old compared with young animals. Furthermore, the persistence of MHCII expression in skeletal muscle macrophages from old mice after influenza A infection might reflect impaired phagocytosis (Lemke, 2013). MERTK is a member of the TAM family (Tyro3/ Axl/Mer) of tyrosine kinases that is expressed on macrophages, and is involved in recognizing and clearing apoptotic cells through phagocytosis (Lemke, 2013). We found that the level of mRNA encoding Mertk and its phosphatidylserine binding co-receptor Gas6 were reduced in flow cytometry-sorted macrophages from old compared with young mice (Figure 6a). Accordingly, we infected Additionally, we found increased accumulation of age-related autofluorescent granules representing either lipofuscin or ceroid in the muscle of young Merkt −/− mice compared with young wild-type mice, and a similar accumulation of lipofuscin or ceroid was observed in old mice (Figures 6f and S5). Similar to Cx3cr1 knockout mice, we observed increased numbers of skeletal muscle macrophages in young Mertk −/− mice at baseline. However, the number of skeletal muscle macrophages in young Mertk −/− mice did not increase during recovery from influenza A virus-induced pneumonia (Figure 6g) and there was no increase in the proportion of macrophages that were MHCII low (Figure 6h), similar to old mice. Consistent with these findings, Mertk −/− mice lacked expansion of MuSC and FAP during recovery from influenza A-induced pneumonia (Figure 6i,j). These data support a requirement for functional phagocytosis in recovery from influenza A pneumonia-induced skeletal muscle atrophy.

| D ISCUSS I ON
As vividly illustrated by the COVID-19 pandemic, pneumonia disproportionately affects the elderly (Zhou et al., 2020). While mortality from pneumonia is higher in older compared with younger individuals, most elderly patients with access to modern healthcare survive their pneumonia (Jain et al., 2015;Ortiz et al., 2013). It is well established that skeletal muscle dysfunction is common in survivors of severe pneumonia requiring intensive care (Herridge et al., 2003).
This skeletal muscle dysfunction after pneumonia is disproportionately prolonged and severe in the elderly when compared to younger

F I G U R E 6
Mertk is necessary for recovery of skeletal muscle function after influenza A-induced pneumonia. (a) Skeletal muscle macrophages were isolated from naïve young (4 months: green bars) and old (20-24 months: purple bars) mice, and RNA was analyzed for TAM family members via real-time qPCR with SYBR Green (n = 5 mice). (b) Young (4 months old: orange bars) Mertk −/− mice and agematched wild-type control mice (green bars) were infected with influenza A virus (IAV) (25 pfu) and harvested before the infection (day 0) and 7 days after the infection. IL-6 was measured by ELISA in the serum (n = 4-6 mice). (c) Young (4 months old: orange bars) Mertk −/− mice and age-matched wild-type control mice (green bars) were infected with IAV (25 pfu) and harvested before the infection (day 0) and 7 days after the infection. Quantification (top panel) and representative Western blot (bottom panel) of atrogin-1 expression in the mouse tibialis anterior muscle. GAPDH was used as a loading control (n = 3-6 mice). (d) Prior to being infected with IAV, young (4 months) Mertk −/− mice (orange lines) and age-matched wild-type control mice (green lines) were exposed to voluntary wheel running for 14 days. They were then infected with 25 pfu of IAV, and voluntary wheel running was recorded for 50 days. The 8 p.m. to 6 a.m. average wheel rotation/mouse, corrected to the initial 10-day average control, is shown (n = 15 mice per group). Graph shows distance run with means ± SD (e) Mice were treated as in ( While a subpopulation of macrophages expressing the fractalkine receptor CX3CR1 in the skeletal muscle has been previously reported, nothing was known regarding the function or ontogeny of these cells (Arnold et al., 2015;Jin, Warunek, & Wohlfert, 2018;Zhao, Lu, Wang, Ransohoff, & Zhou, 2016). In young mice recovering from influenza A pneumonia, we observed expansion and downregulation of MHCII in skeletal muscle macrophages followed by mus- guides monocytes and macrophages to areas of injury or inflammation, and during brain development, CX3CR1 is necessary to guide microglia to synapses where they are pruned from living neurons via a process referred to as trogocytosis (Paolicelli et al., 2011;Weinhard et al., 2018).
We used transcriptomic profiling to compare tissue-resident skeletal muscle macrophages in young and old mice under naïve conditions and during recovery from influenza A infection. These data suggested the hypothesis that skeletal muscle macrophages from old mice lose phagocytic function, which was confirmed by reduced uptake of serum-opsonized polystyrene beads in flow cytometry-sorted skeletal muscle macrophages from old compared with young mice. MERTK is a member of the TAM family of scavenger receptors (MERTK, AXL, and TYRO3), which are activated during injury by interactions with Protein S or Gas6 bound to Annexin V on the surface of apoptotic cells (Lemke, 2013). Ligation of MERTK in macrophages results in the induction of phagocytosis through protein kinase C and Rac-mediated regulation of the cytoskeleton and the activation of AKT, leading to cellular proliferation and survival (Lemke & Rothlin, 2008). In addition, the activation of MERTK results in its association with the type I interferon receptor (IFNAR) to induce the Jak1/Stat1-mediated activation of suppressor of cytokine signaling 1 and suppressor of cytokine signaling 3 (SOCS1 and SOC3) (Lemke, 2013). These transcription factors inhibit inflammatory mediators and further reduce expression of MHC class II molecules on the cell surface (Dalpke, Opper, Zimmermann, & Heeg, 2001). We found that during recovery from influenza A infection, young Mertk −/− mice largely phenocopied old mice, in that they failed to expand tissue-resident macrophages, downregulate MHCII on tissue-resident macrophages, or expand MuSC and they failed to muscle function after influenza A infection. Our findings are consistent with previous reports that MERTK is necessary for cardiac repair after myocardial infarction and that mesenchymal stem cells promote cardiac repair by activating phagocytic pathways in cardiac macrophages, strengthening the causal association between tissue-resident skeletal muscle macrophage function and muscle sat-  (Cantini & Carraro, 1995;Cantini et al., 2002;Chazaud et al., 2003;Merly, Lescaudron, Rouaud, Crossin, & Gardahaut, 1999;Saclier et al., 2013). In our model, however, we cannot determine whether impaired MuSC proliferation is related to the loss of paracrine signals induced in macrophages upon uptake of cellular debris or an inhibition of cellular proliferation by the debris that is not cleared by macrophages.
Our molecular findings are consistent with the studies of muscle biopsies from patients with myopathy after critical illness. In these biopsies, investigators observed MuSC proliferation in the absence of inflammatory cell infiltration (Dos Santos et al., 2016). This differs from many murine models used to study muscle recovery, most of which involve direct injury to the muscle including trauma, freezing, or cardiotoxin injection. In all of these injuries, inflammatory cells are recruited to the injured muscle including neutrophils, monocytes,

monocyte-derived macrophages, regulatory T cells (Tregs), and
other T-cell populations (Arnold et al., 2007;Heredia et al., 2013;Kuswanto et al., 2016;Teixeira et al., 2003;Varga et al., 2016;J. Zhang et al., 2014). Inflammatory cell recruitment in those models precluded the specific study of tissue-resident skeletal muscle macrophages. For example, two independent groups reported that deletion of monocyte-derived macrophages recruited in response to cardiotoxin-induced injury resulted in impaired satellite cell proliferation. In both of these studies, the strategies used to deplete monocyte-derived macrophages (CD11b-DTR and an antibody to M-CSF) would target both monocyte-derived macrophages and the skeletal muscle tissue-resident macrophages we identified in this study (Arnold et al., 2007;Segawa et al., 2008). While Tregs have been reported to be important for skeletal muscle recovery after cardiotoxin injury (Kuswanto et al., 2016), we found that Treg number was reduced in both young and old mice after influenza A infection.
Nevertheless, Treg proliferation was higher in young compared with old mice and it is possible that factors released from resident tissue Tregs are important for recovery.
Our study has some limitations. First, while voluntary wheel running in mice is an intuitive and compelling measure of the skeletal muscle dysfunction that impairs mobility in survivors of pneumonia, it is a composite measure that may be altered by factors other than muscle function. Second, while Mertk −/− mice recapitulated the lack of cellular responses in old animals, they transiently regained muscle function after injury. This may be explained by the high basal levels of MuSC in Mertk −/− relative to wild-type mice, which allows short term, but unsustainable recovery of muscle function. Third, while we observed expansion of FAPs during recovery from muscle injury in young, old, and Mertk-deficient mice, their role in muscle recovery is not certain. FAP cells have been described as important facilitators of MuSC behaviors, supporting their myogenic differentiation (Fiore et al., 2016), yet FAP cell persistence has been shown to have a detrimental effect on muscle tissue by contributing to fibrosis (Gonzalez et al., 2017) and ectopic fat accumulation (Uezumi, Fukada, Yamamoto, Takeda, & Tsuchida, 2010). Thus, if tissue-resident macrophages contribute to FAP clearance, as has been reported for inflammatory, monocyte-derived macrophages (Lemos et al., 2015) some of the effects of aging or Mertk loss in macrophages might be attributed to the persistence of these cells. Finally, while the expression of Cx3cr1 in the skeletal muscle is limited to macrophages, Cx3cr1 is also expressed on monocytes and LYVE1expressing macrophages associated with the vasculature and nerve fibers (Lim et al., 2018;Schaum et al., 2018). We therefore cannot completely exclude off target effects related to global deletion of Cx3cr1 in our system.
Skeletal muscle represents approximately 40% of the total body mass (Janssen, Heymsfield, Wang, & Ross, 2000), and maintaining its function is critical for posture, breathing, locomotion, nutrient storage, and overall well-being. As a result, loss of muscle function and mass with age-related sarcopenia is a major driver of frailty (Janssen et al., 2000;Topinkova, 2008

| CONTAC T FOR RE AG ENT AND RE SOURCE S HARING
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contacts: Jacob Sznajder (j-sznajder@northwestern.edu) and Alexander Misharin (a-misharin@northwestern.edu).

| Mouse models
All mouse procedures were approved by the Institutional Animal Care and Use Committee at Northwestern University (Chicago, IL, USA). All strains including wild-type mice are bred and housed at a barrier-and specific pathogen-free facility at the Center for  (Yona et al., 2013) and ZsGreen reporter (Madisen et al., 2010) mice were obtained from Jackson laboratories (Jax stocks 000664, 020940, and 007906, correspondingly) as were Cx3cr1 gfp/gfp mice (Jax stock 005582).  NJ), and then instilled with either sterile, phosphate-buffered saline (S), or mouse-adapted influenza A virus (IAV) as previously described (Radigan et al., 2019). Various doses in a range of 10-100pfu were tested to find an optimal sublethal dose matched to the age of the animal. Based on survival curves, we determined that 25 pfu was optimal for young animals, but a lower dose was required for old animals, as a 25 pfu dose resulted in excessive mortality of old mice ( Figure S1). We continuously observed mice infected with influenza A virus for signs of distress (slowed respiration, failure to respond to cage tapping, failure of grooming, huddling, and fur ruffling). Mice that developed these symptoms were sacrificed, and the death was recorded as an influenza A-induced mortality. Weight was measured prior to influenza infection and prior to harvest at 4 or 10 days postinfection, and recorded as percentage of weight loss from baseline (Radigan et al., 2019).

| Measurement of muscle dysfunction
Immediately prior to muscle harvest, forelimb skeletal muscle strength was assessed using a digital grip strength meter (Columbus Instruments) as described (Radigan et al., 2019). Grip strength was measured in each animal six successive times, and the average of the highest four values for each mouse was used. The same operator performed all tests. The mice were then terminally anesthetized with EUTHASOL (pentobarbital sodium/phenytoin sodium). The soleus and EDL muscles were excised, and tendon was trimmed under a microscope to assure optimal accuracy for weight measurement.
The muscles were then blotted dry and weighed. Muscles were either frozen in liquid nitrogen-cooled isopentane for cryosectioning or snap-frozen in liquid nitrogen for protein extraction. Alternatively, muscles were minced for flow cytometric analysis as described below.

| Voluntary wheel running
Mice, housed three per cage, were provided access to a trackable, low-profile saucer wheels (Med Associates, Inc.) 24 hr per day for 14-28 days to establish consistent running patterns. After establishing baseline, all mice were infected with IAV at an age-appropriate dose ( Figure S1). Wheel rotations were gathered in one-minute increments via wireless hub and analyzed using wheel manager software (Med Associates, Inc. from inner border to inner border, passing through the center of the muscle fiber. This parameter has been shown to be very insensitive to deviations from the "optimal" cross-sectioning profile, as compared with direct measure of fiber CSA (Radigan et al., 2019).
CSA was calculated using this diameter, and results were expressed as mean CSA ± SE

| Bronchoalveolar lavage
Saline fluid (0.8 ml) was lavaged and then aspirated from the lungs of mice through tracheal angiocatheter. Cell counts were measured by trypan blue exclusion and the automated countess system.
To measure the cell differential, 200 μl of BAL fluid was placed in cytospin funnels and spun 450 rcf 5 min onto a glass slide using Shandon Cytospin 3 centrifuge. Cells were stained by Wright-Giemsa and manually counted and identified.

| IL-6 measurements
IL-6 expression assays were performed using a kit from Life technologies (KMC0061C) according to the manufacturer's instructions.
Assays were performed on either BAL fluids (acquired as described above) or from 50 μl/well blood plasma drawn from posterior vena cava using a 23-gauge needle, centrifuged to remove cells, and stored at −80°C. Analysis for each mouse was done in triplicate, and a minimum of five mice were analyzed/condition.

| Flow Cytometry and Cell Sorting
Mice were fully perfused through the right ventricle with 20 ml of PBS, and hind limb muscle was removed. Prior to cytometric analysis, 50 μl 123count eBeads (Invitrogen) were added to 200 μl each sample to allow accurate cell counts/ gm of tissue. "Fluorescence minus one" controls were used when necessary. Cell populations were identified using sequential gating strategy (Figures 3a,4a,5a). For Foxp3 detection, a staining kit for intracellular antigens was used (eBiosciences). Cells were fixed and permeabilized according to directions, before incubation with Foxp3 antibody prior to analysis. Cell sorting was performed at Northwestern University RLHCCC Flow Cytometry core facility on SORP FACSAria III instrument (BD) with the same optical configuration as LSR II, using a 100-µm nozzle and 40 psi pressure.
To measure proliferation, EdU was diluted in sterile saline from 100 mg/ml stock in DMSO, and injected intraperitoneally at 2 mg/ mouse, 16 hr prior to tissue harvest (Misharin et al., 2017). Singlecell suspensions were divided into thirds and stained with antibody panels as described in (Figures 3a,4a,5a)  cDNA was generated using iScript cDNA Synthesis Kit (Bio-Rad) in a Bio-Rad T100 thermal cycler, followed by qPCR with SYBR Green

| Transcriptome profiling via mRNA-Seq
Ten thousand skeletal muscle macrophages were sorted on a periments exceeded 6x10 6 per sample and over 94% of reads had a Q-score greater than 30. For RNA-seq analysis, reads were demultiplexed using bcl2fastq (version 2.17.1.14). Read quality was assessed with FastQC. Low-quality base calls were trimmed using Trimmomatic (version 0.36). Reads were then aligned to the Mus musculus reference genome (with mm10 assembly) using the STAR aligner (version 2.6.0). Counts tables were generated using HTSeq (version 0.6.1). Raw counts were processed in R (version 3.4.4) using edgeR (version 3.20.9) to generate normalized counts. Negative binomial likelihood with default setting followed by generalized linear models fitting was used to estimate differentially expressed genes.
FDR q-values were used to correct for multiple comparisons, and a value of 0.05 was used as a threshold to indicate statistical significance (see online code). Gene Ontology analysis was performed using GOrilla on two unranked gene lists. The RNA-seq datasets are available at GEO: GSE14 2580.

| In Vitro phagocytosis
After enzymatic digestion, 50 μl of the single-cell suspensions were incubated with 35 μl of myeloid antibody cocktail with 15 μl/ tube Fluoresbrite YG 1.0 micron microspheres (Polysciences Inc., Warrington, PA) that had been opsonized for 30 min at 37°C in by mixing 1 drop beads with 1ml 50% FBS/HBSS, followed by 2X wash/resuspension in HBSS 1 ml (non-opsonized beads were used as a negative control). Cell suspensions were allowed to uptake beads for 30 min at 37°C in the dark with shaking, followed by two washes in HBSS to remove any unbound beads, and fixation in 2% paraformaldehyde, followed by flow cytometry with a FITC negative myeloid cocktail to detect bead uptake in myeloid cell populations.

| Statistical analysis
Differences between groups were determined according to ANOVA.
When ANOVA indicated a significant difference, individual differences were examined using t tests with a Tukey correction for multiple comparisons, as indicated. All analyses were performed using GraphPad Prism, version 8.4.2 (GraphPad Software, San Diego CA).
Data are shown as means ± SEMs. p < .05 was considered statistically significant in all tests.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.