Single-cell analysis of skeletal muscle macrophages reveals age-associated functional subpopulations

  1. Linda K Krasniewski
  2. Papiya Chakraborty
  3. Chang-Yi Cui  Is a corresponding author
  4. Krystyna Mazan-Mamczarz
  5. Christopher Dunn
  6. Yulan Piao
  7. Jinshui Fan
  8. Changyou Shi
  9. Tonya Wallace
  10. Cuong Nguyen
  11. Isabelle A Rathbun
  12. Rachel Munk
  13. Dimitrios Tsitsipatis
  14. Supriyo De
  15. Payel Sen
  16. Luigi Ferrucci
  17. Myriam Gorospe  Is a corresponding author
  1. Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, United States
  2. Flow Cytometry Core, National Institute on Aging Intramural Research Program, National Institutes of Health, United States
  3. Translational Gerontology Branch, National Institute on Aging Intramural Research Program, National Institutes of Health, United States

Abstract

Tissue-resident macrophages represent a group of highly responsive innate immune cells that acquire diverse functions by polarizing toward distinct subpopulations. The subpopulations of macrophages that reside in skeletal muscle (SKM) and their changes during aging are poorly characterized. By single-cell transcriptomic analysis with unsupervised clustering, we found 11 distinct macrophage clusters in male mouse SKM with enriched gene expression programs linked to reparative, proinflammatory, phagocytic, proliferative, and senescence-associated functions. Using a complementary classification, membrane markers LYVE1 and MHCII identified four macrophage subgroups: LYVE1−/MHCIIhi (M1-like, classically activated), LYVE1+/MHCIIlo (M2-like, alternatively activated), and two new subgroups, LYVE1+/MHCIIhi and LYVE1−/MHCIIlo. Notably, one new subgroup, LYVE1+/MHCIIhi, had traits of both M2 and M1 macrophages, while the other new subgroup, LYVE1−/MHCIIlo, displayed strong phagocytic capacity. Flow cytometric analysis validated the presence of the four macrophage subgroups in SKM and found that LYVE1− macrophages were more abundant than LYVE1+ macrophages in old SKM. A striking increase in proinflammatory markers (S100a8 and S100a9 mRNAs) and senescence-related markers (Gpnmb and Spp1 mRNAs) was evident in macrophage clusters from older mice. In sum, we have identified dynamically polarized SKM macrophages and propose that specific macrophage subpopulations contribute to the proinflammatory and senescent traits of old SKM.

Editor's evaluation

In this study, Krasniewski and colleagues describe important findings leveraging single-cell transcriptomics to identify subpopulations of macrophages in the skeletal muscle of aging mice. They present solid evidence for the existence of several new resident subpopulations of skeletal muscle macrophages, spanning a range of polarization states using novel markers. Additionally, they identify a shift in relative abundances of these subpopulations with age, leading to a functional shift in inflammatory marker expression and phagocytic capacity. This work will be useful to researchers in the field of immune aging as a resource.

https://doi.org/10.7554/eLife.77974.sa0

Introduction

Macrophages are heterogeneous innate immune cells (Shapouri-Moghaddam et al., 2018) that provide the first line of defense against pathogens, but are also deeply involved in inflammation, dead cell removal, wound healing, and tissue remodeling (Mills et al., 2014; Ross et al., 2021; Shapouri-Moghaddam et al., 2018). Macrophages adapt to individual tissues and acquire specific tissue-dependent functions (Wynn et al., 2013). Upon transplantation, tissue-resident macrophages quickly lose their original gene expression patterns and gain host organ markers (Lavin et al., 2014). The tissue environment contributes to determining the tissue-specific protein production by macrophages and thereby establishes tissue-dependent expression patterns and functions (Gautier et al., 2012; Lavin et al., 2014). Hence, the function of macrophages should be studied in the context of their tissue of residence.

Macrophages play diverse functions in tissues by differentiating into specific functional subgroups, a process usually defined as macrophage polarization (Yao et al., 2019). Most macrophages are known to polarize to proinflammatory M1 or anti-inflammatory M2 subgroups (Martinez et al., 2008; Mills et al., 2000; Rath et al., 2014). While such dichotomy largely explains the strikingly different actions of macrophages commonly seen in many tissues, macrophages appear to be more functionally heterogeneous than simply M1 or M2. In this regard, recent flow cytometry and single-cell studies have identified several new macrophage subgroups in arteries, lung interstitium, heart, adipose tissue, and other tissues and organs (Chakarov et al., 2019; Dick et al., 2022; Jaitin et al., 2019; Lim et al., 2018; Schyns et al., 2019) with distinct tissue-dependent polarization status. Dissecting polarization in each tissue is thus critical to elucidating shared and tissue-specific macrophage functions.

Skeletal muscle (SKM) contains large numbers of macrophages that play critical roles in injury repair and regeneration (Arnold et al., 2007; Tidball, 2011; Tidball, 2017). Macrophages assume different polarization to play distinct functions at different stages of repair after injury (Scala et al., 2021; Yang and Hu, 2018). In the absence of injury or infection, most macrophages residing in human and mouse SKM were shown to be MRC1 (CD206)+, M2-like macrophages (Cui et al., 2019; Wang et al., 2015). However, the full range of macrophage subgroups and their age-related changes in SKM is poorly understood (Cui and Ferrucci, 2020).

To better understand the complexity of the macrophage polarization status and their changes with aging in mouse SKM, we carried out single-cell transcriptomic analysis. We present evidence that SKM macrophages comprise 11 distinct clusters associated with specific proposed functions. Using a complementary classification based on the presence of membrane markers, SKM macrophages were divided into two large populations based on the presence of LYVE1 and was further classified into four functional subgroups by introducing MHCII as an additional surface marker. We further show that mRNAs that encode proinflammatory proteins and senescence- and aging-related proteins were significantly upregulated in specific macrophage clusters in old SKM. Our findings reveal a dynamic polarization of functional subpopulations of mouse SKM macrophages, including changes toward proinflammatory and senescent phenotypes with aging.

Results

Isolation of macrophages from mouse SKM and single-cell RNA sequencing

To isolate macrophages from SKM, we collected all muscles from hind limbs, including quadriceps, gastrocnemius, tibialis, and soleus, from C57BL/6JN male mice, combined and minced them into small cubes, and isolated mononuclear cells by digesting them with enzymes including collagenase and other proteases (Krasniewski et al., 2022; Liu et al., 2015; Figure 1A). To identify macrophage-rich fractions from the mononuclear cell preparation, we carried out flow cytometric analysis based on the presence of CD45, a pan-leukocyte marker, and CD11b, a pan-myeloid lineage marker. As we found previously, CD11b+ cells clearly separated from the rest of the mononuclear cell population (Krasniewski et al., 2022).

Figure 1 with 2 supplements see all
Macrophage isolation from mouse skeletal muscle (SKM) and single-cell RNA-seq analysis.

(A) Workflow of mononuclear cell collection from mouse SKM, CD11b+ cell isolation by FACS, and single-cell RNA-seq analysis using the 10× Genomics platform. (B) Cells isolated from mouse SKM that were CD11b+ and F4/80+. (C) Unsupervised clustering of SKM macrophages revealed 11 clusters. %, proportion of each cluster. (D) Dot plot shows featured mRNAs in each cluster. (E) Heat maps show enriched genes in Cl0, 2, 6, and 8.

For single-cell RNA-sequencing (scRNA-seq) analysis, we collected CD11b+ cells from three young (3 months old [3 m.o.]) and three old (23 m.o.) male mice as biological triplicates by fluorescence-activated cell sorting (FACS). From each mouse, 5000–10,000 CD11b+ cells were used for single-cell library preparation using the 3’ gene expression pipeline from 10× Genomics followed by RNA-seq analysis. We successfully obtained sequences from 2000 to 5000 single cells from each mouse, and a mean of ~80,000 RNA-seq reads per cell corresponding to a median of >2000 genes per cell (Materials and methods; GEO identifier GSE195507). Sequencing analysis showed that >80% of cells expressing Cd11b mRNA were also positive for F4/80 mRNA (Adgre1 mRNA), another common marker for mouse macrophages (Figure 1B). Those cells expressing both Cd11b mRNA and Adgre1 mRNA were considered SKM macrophages. Very few cells were positive for Ly6g mRNA or Siglecf mRNA (specific markers for neutrophils and eosinophils, respectively; Figure 1—figure supplement 1A), indicating minimal contamination from these cells in our macrophage population.

Identification of 11 macrophage clusters in SKM by unsupervised classification

To gain insight into the subpopulations of SKM macrophages, we pooled scRNA-seq data from young and old mice and performed unsupervised classification. By using FindClusters at a resolution of 0.3, we found 11 clusters (Cl0-10; Figure 1C). Given that we isolated macrophages on three different dates due to technical limitations (lengthy procedure) and mouse availability (Materials and methods), we compared the different datasets to ensure there were no batch effects. Overall, the distribution of macrophages across the 11 clusters was comparable among the biological replicates (Figure 1—figure supplement 1B), and the patterns of transcriptomes were also comparable among the replicates (Figure 1—figure supplement 1C). Those mRNAs that were expressed >1.5-fold higher in a given cluster relative to the other 10 clusters, p<0.05, and were expressed in >25% of macrophages in that cluster are shown in Supplementary file 1. Each cluster showed a distinct gene expression pattern (Figure 1D).

To investigate the functional features of these clusters, we carried out gene ontology (GO) enrichment analysis using g:Profiler (Materials and methods). Although all clusters shared functional terms general to macrophages, including ‘immune system process,’ ‘defense response,’ ‘response to stress,’ ‘cell migration,’ and ‘cell death,’ each cluster also displayed distinct functional associations (Table 1). The largest cluster, Cl0, showed a more reparative function, with high expression of M2-type genes (Mrc1, Cd163, Lyve1, and Folr2 mRNAs) and reduced proinflammatory function compared to the other clusters (Figure 1E and Table 1). The second largest cluster, Cl1, showed a similar expression pattern as Cl0 (Figure 1D), including the expression of M2-type mRNAs (Lyve1 and Folr2 mRNAs), but the expression levels of these mRNAs were lower in Cl1 than in Cl0 (Figure 1—figure supplement 2A, B). This resulted in fewer unique genes in Cl1 when compared to the other 10 clusters (Supplementary file 1). When we excluded Cl0 and compared Cl1 with Cl2-10 (Supplementary file 1, ‘Cl1 vs Cl2-10’), Cl1 showed strong enrichment of M2-type mRNAs (Figure 1—figure supplement 2C) and strong association with reparative functions (Table 1). Thus, the two largest clusters, Cl0 and Cl1, account for nearly one-half of total macrophages and displayed M2-like gene expression patterns.

Table 1
Gene ontology (GO) annotation of unsupervised clusters.
GO annotation
ClustersElevatedReducedFeatured membrane proteins
Cl0Vasculature development (7.8)
Amoeboidal-type cell migration (3.7)
Endocytosis (3.4)
Response to wounding (2.1)
Cytokine production (3.6)
Positive regulation of inflammatory response (1.9)
Cellular detoxification (1.5)
↑: LYVE1, FOLR2, MRC1,
CD163
↓: H2-AB1, H2-DMB1
Cl1Vasculature development (6.7)
Amoeboidal-type cell migration (4.3)
Endocytosis (2.9)
Response to wounding (1.6)
Translation (4.2)↑: LYVE1, FOLR2, MRC1,
CD163
Cl2Antigen processing and presentation via MHC class II (6.7)
Cytokine-mediated signaling pathway (4.3)
Cellular response to IL-1 (3.3)
Chemotaxis (2.2)
Defense response to virus (2.2)
Vasculature development (5.3)
Endocytosis (3.4)
Muscle cell proliferation (2.5)
Amoeboidal-type cell migration (1.4)
↑: H2-AB1, H2-EB1, H2-
DMB1, CCR2
↓: LYVE1, FOLR2, MRC1,
CD163
Cl3Cellular detoxification (4.0)
Lymphocyte activation (1.6)
Vasculature development (6.9)
Muscle cell proliferation (3.9)
Response to wounding (1.7)
↑: CCR2
↓: LYVE1, FOLR2, MRC1,
CD163
Cl4Fc receptor signaling pathway (4.2)
Regulation of phagocytosis (3.5)
Antigen processing and presentation via MHC class I (2.5)
Cell killing (1.5)
Angiogenesis (8.6)
IL-1 production (5.2)
Muscle cell proliferation (2.2)
Antigen processing and presentation via MHC class II (2.2)
↑: H2-K1, H2-D1
↓: LYVE1, FOLR2, MRC1,
CD163, H2-AB1, H2-
EB1, H2-DMB1, CCR2
Cl5Response to LPS (6.4)
TLR signaling pathway (2.9)
TNF production (2.0)
Viral entry into host cell (2.3)↑: TREM2
↓: LYVE1, FOLR2, MRC1,
CD163
Cl6Response to lipoprotein particle (3.3)
ATP metabolic process (2.6)
Long-chain fatty acid transport (2.0)
Regulation of transcription from RNA polymerase II promoter in response to stress (7.3)
Cell chemotaxis (5.1)
↑: GPNMB, TREM2
↓: LYVE1, FOLR2, MRC1,
CD163, CCR2
Cl7Translation (12.4)
Antigen processing and presentation via MHC class II (5.2)
Ribosome assembly (2.2)
Vasculature development (8.3)
IL-1β production (3.9)
Response to wounding (2.1)
↑: H2-AB1, H2-EB1, H2-
DMB1
↓: LYVE1, FOLR2, MRC1,
CD163
Cl8Positive regulation of cell death (4.2)
Phagocytosis (2.2)
Autocrine signaling (1.9)
Antigen processing and presentation via MHC class II (4.4)
response to IFN-γ (3.3)
↓: LYVE1, FOLR2, MRC1,
CD163, H2-AB1, H2-
EB1, H2-DMB1, CCR2
Cl9Cell cycle (20.3)
DNA replication (4.5)
DNA repair (3.2)
Myeloid cell differentiation (3.6)
IL-1 production (1.4)
↓: LYVE1, FOLR2
Cl10Vasculature development (19.9)
Extracellular matrix organization (9.2)
Response to wounding (2.2)
↑: LY6C1
  1. Parenthesis: negative log10 of adjusted p-value. ↑:elevated in the cluster. ↓:reduced in the cluster.

Clusters Cl2-9 showed very low expression of M2 marker genes (Table 1) and instead displayed more diverse functional associations. Cl2 expressed mRNAs related to inflammation and to the functions of antigen processing and presentation (Figure 1E and Table 1). The mRNAs present in Cl3 were associated with cellular detoxification, and Cl4 was associated with phagocytosis and expressed elevated MHC class I (MHCI) mRNAs. Cl5 expressed mRNAs strongly associated with the inflammatory response. Cl6 was enriched in mRNAs encoding proteins involved in the response to lipoprotein particles, ATP metabolism, and lipid transport; this cluster also expressed Gpnmb, Spp1, Ctsd, Trem2, and Gdf15 mRNAs, encoding proteins involved in senescence and aging (Henjum et al., 2016; Pazolli et al., 2009; Suda et al., 2021; Suda et al., 2022; Tanaka et al., 2018; Williams et al., 2022), and Fabp5 and Fabp4 mRNAs, encoding proteins implicated in atherosclerosis (Babaev et al., 2011; Furuhashi et al., 2007; Makowski et al., 2001; Figure 1E). The mRNAs expressed in Cl7 were strongly associated with translation and antigen processing and presentation via MHC class II, while those expressed in Cl8 were associated with cell death and phagocytosis, although M2-type markers and MHCII genes were reduced. Of note, S100a8 and S100a9 mRNAs, the most robustly elevated mRNAs in Cl8, encode proinflammatory proteins (Figure 1E, Supplementary file 1). Cl9 expressed cell cycle-related mRNAs, with elevated Top2a, Mki67, and Cdk1 mRNAs (Supplementary file 1), likely representing a group of reported proliferating macrophages (Wang et al., 2020). The smallest cluster, Cl10 (0.5% of total CD11b+/F4/80+macrophages [Figure 1C]) was associated with a reparative function, and one-half of Cl10 cells expressed Ly6c1 mRNA (Supplementary file 1).

Overall, unsupervised clustering revealed a wide functional heterogeneity of SKM macrophages. GO annotation identified clusters of macrophages expressing mRNAs that were particularly associated with reparative functions (Cl0, Cl1, and Cl10), the promotion of inflammation (Cl2 and Cl5), antigen processing and presentation via MHC class II (Cl2 and Cl7), cellular detoxification (Cl3), phagocytosis (Cl4 and Cl8), lipid homeostasis and cell senescence (Cl6), protein synthesis (Cl7), and proliferation (Cl9).

Identification of M2-like macrophages by membrane marker-based classification

Macrophage membrane markers, including MRC1, CD86, LYVE1, and MHCII, have been successfully used to functionally classify macrophage subgroups (Mantovani et al., 2002; Stein et al., 1992; Dick et al., 2022; Chakarov et al., 2019; Lim et al., 2018). To complement the unsupervised clustering and gain a more comprehensive view of the highly heterogeneous group of SKM macrophages, we further carried out supervised classification with membrane markers.

Initially, we attempted to subgroup SKM macrophages by traditional polarization markers: MRC1, CD86, or CD80. MRC1 is a widely used marker of M2 macrophages, whereas CD80 and CD86 are M1 markers (Mantovani et al., 2002; Stein et al., 1992). However, our scRNA-seq data showed that Mrc1 and Cd86 mRNAs were broadly expressed in ~80% of macrophages, Cd80 mRNA was expressed only in a small population, and most macrophages expressed Mrc1 and Cd86 mRNAs simultaneously (Figure 2—figure supplement 1A), suggesting they are not ideal to classify SKM macrophages at the transcriptomic level.

We therefore turned to other candidate membrane markers. Recently, LYVE1 and MHCII were successfully used to subgroup several tissue-resident macrophages (Dick et al., 2022; Chakarov et al., 2019; Lim et al., 2018). By unsupervised clustering, Lyve1 and MHCII mRNAs were differentially expressed in select clusters (Table 1); therefore, we classified SKM macrophages by LYVE1 expression levels first. LYVE1 status divided SKM macrophages into two large, similarly sized groups, LYVE1+ (46.6%) and LYVE1− (53.4%) (Figure 2A). LYVE1+ macrophages displayed an M2-like transcriptomic program, including mRNAs encoding proteins associated with functions in ‘vasculature development,’ ‘wound repair,’ and ‘endocytosis’ (Figure 2B; Buchacher et al., 2015; Stein et al., 1992). Interestingly, transcripts encoding proangiogenic proteins (Ang, Stab1, and Egr1 mRNAs) as well as transcripts encoding antiangiogenic proteins (Cfh and Hspb1 mRNAs) were upregulated in LYVE1+ macrophages. Transcripts encoding proteins implicated in wound healing (Igf1, Nrp1, and Gas6 mRNAs) were also elevated in LYVE1+ macrophages (Figure 2B); and mRNAs encoding endocytosis-related members of the CD209 family (including Cd209d and Cd209b mRNAs) as well as Cd36, Cd163, and Mrc1 mRNAs were also highly expressed in LYVE1+ macrophages. Other mRNAs, such as Timd4 and Fcna mRNAs, were almost exclusively expressed in the LYVE1+ macrophages and might be good candidate markers for this population (Figure 2—figure supplement 1B).

Figure 2 with 1 supplement see all
Functional clusters of genes differentially expressed in LYVE1+ and LYVE1− macrophages following single-cell RNA-sequencing (scRNA-seq) analysis.

(A) Lyve1 mRNA expression pattern in skeletal muscle (SKM) macrophages. (B) mRNAs highly expressed in functional clusters of LYVE1+ macrophages. (C) mRNAs highly expressed in LYVE1− macrophages. (D) Validation of select mRNAs differentially abundant as identified in panels (B and C). LYVE1+ and LYVE1− macrophages were isolated by fluorescence-activated cell sorting (FACS) from three male mice, 3 months old (m.o.), and mRNAs elevated in LYVE1+ macrophages (top and bottom left), and mRNAs predominantly elevated in LYVE1− macrophages (bottom right) were quantified by RT-quantitative PCR (qPCR) analysis. Data were normalized to the levels of Gapdh mRNA, also measured by RT-qPCR analysis. Data represent the means and SD from two different sorts for each group.

In contrast, LYVE1− macrophages expressed higher levels of mRNAs encoding antigen-processing and antigen-presenting proteins (H2-Eb1, H2-Ab1, H2-DMb1, and Cd74 mRNAs), proteins related to the NF-kB signaling pathway and implicated in cell death (Il1b, Bcl2a1b, Bcl2a1d, Cd14, Traf1, and Malt1 mRNAs), and proteins with function in antioxidant responses (Gsr, Prdx5, Prdx6, Prdx1, and Hp mRNAs). In addition, many mRNAs encoding ribosomal proteins were highly expressed in this group (Figure 2—figure supplement 1C). The full list of mRNAs differentially abundant in LYVE1+ and LYVE1− macrophages is in Supplementary file 2.

To validate the differences in gene expression programs, we separated each population (LYVE1+ and LYVE1− macrophages) by FACS. Reverse transcription (RT) followed by real-time quantitative (q)PCR analysis confirmed that Lyve1, Folr2, Timd4, CD209f, and Fcna mRNAs were almost exclusively expressed in LYVE1 + macrophages (Figure 2D, top, n=2 biological replicates). By contrast, Mrc1, Igf1, and Ang mRNAs were expressed in both LYVE1+ and LYVE1− macrophages, but at much higher levels in LYVE1+, while Il1b mRNA levels were significantly higher in the LYVE1− population (Figure 2D, bottom). The RT-qPCR results (Figure 2D) were consistent with the single-cell transcriptomic analysis (Figure 2B and C; Supplementary file 2), indicating that LYVE1 is an effective marker for subgrouping mouse SKM macrophages.

LYVE1 and MHCII further classify macrophages into four subgroups

By single-cell profiling analysis, the membrane marker MHCII (encoded by H2-Ab1 and H2-Eb1 mRNAs) divided SKM macrophages into two groups, MHCIIhi and MHCIIlo (Figure 3—figure supplement 1A). Considering the relative levels of LYVE1 and MHCII on the membrane allowed the classification of SKM macrophages into four subgroups: LYVE1+/MHCIIhi, LYVE1+/MHCIIlo, LYVE1−/MHCIIhi, and LYVE1−/MHCIIlo (Figure 3A). Among them, LYVE1+/MHCIIhi and LYVE1−/MHCIIhi were the largest subgroups, comprising 36.55 and 39.97% of all macrophages, respectively (Figure 3A), while LYVE1+/MHCIIlo and LYVE1−/MHCIIlo comprised 10.09 and 13.39%, respectively (biological replicates in Figure 3—figure supplement 1B, C). The overall distribution of cells among the biological replicates was comparable among the subgroups selected based on expression levels of Lyve1 and MhcII mRNAs in supervised analysis (Figure 3—figure supplement 1B, C). Notably, both LYVE1+ subgroups, LYVE1+/MHCIIlo and LYVE1+/MHCIIhi, largely overlapped with two reparative clusters, Cl0 and Cl1, from the unsupervised clustering (compare Figure 3A with Figure 1C). The LYVE1−/MHCIIhi subgroup comprised most macrophages from Cl2, Cl5, Cl6, Cl7, Cl9, and part of Cl3. LYVE1−/MHCIIlo contained Cl4 and part of Cl3 and Cl8. Overall, LYVE1− macrophages showed more heterogeneity than LYVE1+ macrophages (Figures 1C and 3A).

Figure 3 with 1 supplement see all
Classification of mouse skeletal muscle (SKM) macrophages into four functional subgroups according to surface markers.

(A) Subclassification of mouse SKM macrophages based on LYVE1 and MHCII levels: LYVE1+/MHCIIlo, LYVE1+/MHCIIhi, LYVE1−/MHCIIhi, and LYVE1−/MHCIIlo. Uniform Manifold Approximation and Projection (UMAP) analysis of the distribution and size of each of four subgroups individually (left) and combined (right). (B) Heat map analysis of the single-cell RNA-sequencing (scRNA-seq) data depicting distinct gene expression patterns of the four subgroups. (C) Gene ontology (GO) annotation of the functions of each subgroup. Brown box, LYVE1+/MHCIIlo; green box, LYVE1+/MHCIIhi; blue box, LYVE1−/MHCIIhi; purple box, LYVE1−/MHCIIlo.

Single-cell analysis (Figure 3B) revealed distinct gene expression patterns across the four supervised subgroups. Those mRNAs that were expressed >1.5-fold higher in a given subgroup relative to the other three subgroups (p<0.01) and were expressed in >25% of macrophages in that subgroup are shown in Supplementary file 3. Functional annotations of the genes showing higher expression in each subgroup revealed that LYVE1+/MHCIIlo macrophages (brown box, Figure 3C) expressed higher levels of mRNAs associated with vasculature development and wound healing, similar to the macrophages in Cl0 and Cl1 (Table 1) and M2 macrophages (Krzyszczyk et al., 2018). LYVE1−/MHCIIhi macrophages (blue box, Figure 3C) were associated with antigen processing and presentation, cytokine production, and responses to bacteria and were overall more M1-like (Mills, 2015). LYVE1+/MHCIIhi macrophages (green box, Figure 3C) were a more complex group; GO annotation suggested that they largely shared LYVE1+/MHCIIlo (M2-like) functions like vasculature development and wound healing, but also shared LYVE1−/MHCIIhi (M1-like) functions such as antigen processing and presentation and cytokine production. Finally, LYVE1−/MHCIIlo macrophages (purple box, Figure 3C) were associated with cytotoxicity and phagocytosis. Notably, among the four subgroups, LYVE1+/MHCIIhi and LYVE1−/MHCIIlo were not previously reported in SKM (Wang et al., 2020), and LYVE1−/MHCIIlo macrophages were not reported in any other tissue so far (Chakarov et al., 2019; Lim et al., 2018). Thus, in addition to the M2-like (LYVE1+/MHCIIlo) and M1-like (LYVE1−/MHCIIhi) subgroups, supervised classification revealed two new subgroups, LYVE1+/MHCIIhi and LYVE1−/MHCIIlo, in resting mouse SKM. The supervised classification thus complemented the unsupervised clustering, offering a more comprehensive understanding of the heterogeneity of SKM macrophages.

Confirmation of four SKM macrophage subgroups by flow cytometry

We further analyzed if the macrophage subgroups identified from scRNA-seq could be validated by cell-surface protein markers. As anticipated, flow cytometric analysis using antibodies that recognized LYVE1 and MHCII divided CD45+/CD11b+/F4/80+SKM macrophages from 3 m.o. male mice into four subgroups, LYVE1+/MHCIIlo, LYVE1+/MHCIIhi, LYVE1−/MHCIIhi, and LYVE1−/MHCIIlo (Figure 4A, n=4). Notably, the LYVE1+/MHCIIlo, LYVE1−/MHCIIhi, and LYVE1−/MHCIIlo subgroups showed clear clusters of cells, but LYVE1+/MHCIIhi macrophages spread across LYVE1+/MHCIIlo and LYVE1−/MHCIIhi (Figure 4A, bottom). The sizes of each subgroup identified by flow cytometry and those identified by single-cell transcriptomics were comparable (Figures 3A and 4A). While the present study focused on SKM macrophages from male mice, we assessed the overall influence of sex on macrophage polarization in SKM by performing flow cytometric analysis with SKM macrophages from 3 m.o. female mice. As shown, female mice also showed four SKM macrophage subgroups, comparable to male mice (compare Figure 4A with Figure 4—figure supplement 1A; n=4). However, when compared with male SKM macrophages, female SKM LYVE1+/MHCIIhi macrophages were ~17% lower, and LYVE1−/MHCIIlo macrophages were ~42% higher (Figure 4—figure supplement 1B). It was recently reported that mouse gender affects macrophage polarization, function, and morphology (Han et al., 2021; Jaillon et al., 2019). The biological significance of the sex-related differences in specific macrophage subgroups in SKM warrants further study.

Figure 4 with 1 supplement see all
Characterization of macrophage subgroups by flow cytometry and immunofluorescence staining.

(A) Flow cytometric analysis of the four subgroups in skeletal muscle (SKM). CD45+/CD11b+/F4/80+macrophages (top three panels show gating) were further classified by LYVE1 and MHCII (bottom right). LYVE1+/MHCIIlo, LYVE1−/MHCIIhi, and LYVE1−/MHCIIlo subgroups formed clear cell clusters, while LYVE1+/MHCIIhi spanned LYVE1+/MHCIIlo and LYVE1−/MHCIIhi. Note: the sizes of each subgroup by flow cytometric analysis (bottom left) were similar to those seen with single-cell RNA-seq analysis. Gating was based on FMO (fluorescence minus one) controls for each experiment. (B) Immunofluorescence analysis of the presence of LYVE1+/MHCIIhi macrophages in mouse SKM. Top, LYVE1+, MHCII+, and LYVE1+/MHCII+ double-positive cells in endomysium and perimysium areas of mouse SKM. Bottom, colocalization of LYVE1 (left) and MHCII (middle) with CD11b, a macrophage marker; secondary antibodies only (right). (C) LYVE1+ macrophages LYVE1+/MHCIIlo and LYVE1+/MHCIIhi, colocalizing with CD31+, depicting blood vessels (top). LYVE1+ and LYVE1− macrophages colocalizing with the nerve fiber marker TUBB3+ (bottom).

To validate these macrophage subgroups in intact mouse SKM, we performed immunofluorescence detection analysis (Figure 4B). As anticipated, both LYVE1+ cells (red) and MHCII+ cells (green) were found in intramuscular connective tissues, namely the endomysium and perimysium regions, as visualized with discontinuous lines demarking muscle cell membranes (Figure 4B). Importantly, many LYVE1+ cells were also MHCII+ (LYVE1+/MHCIIhi) in SKM (Figure 4B, yellow arrows, top), consistent with the flow cytometric and the single-cell transcriptomic analyses. Tyramide signal amplification (TSA) staining of CD11b confirmed that LYVE1+ and MHCII + cells were CD11b+ (Figure 4B, bottom). Thus, immunofluorescence analysis indicated that LYVE1+, MHCII+, and LYVE1+/MHCIIhi macrophages were constitutively present in mouse SKM. Further analysis revealed that both LYVE1+/MHCIIlo (white arrows) and LYVE1+/MHCIIhi (yellow arrows), but not LYVE1−/MHCIIhi macrophages, localized near CD31+ blood vessels (Figure 4C, top). However, both LYVE1+ (yellow and white arrows) and LYVE1− (red arrows) macrophages localized near nerve fibers, the latter positive for TUBB3 (Figure 4C).

Macrophage subgroups show distinct phagocytic capacities

To gain insight into the functional differences among the four subgroups, we assessed their phagocytic capacity, a fundamental function of macrophages, using a flow cytometry-based method that measures the uptake of labeled particles (pHrodo Red Escherichia coli Bioparticle assay, Materials and methods). As anticipated, all macrophage subgroups were strongly phagocytic (Figure 5A), with 97.2% of LYVE1+/MHCIIlo, 98.5% of LYVE1+/MHCIIhi, 86.4% of LYVE1−/MHCIIhi, and 49.6% of LYVE1−/MHCIIlo macrophages actively phagocytizing E. coli particles at 37°C; in control incubations, <17.7% macrophages were active at 4°C (Figure 5A and B, n=3). Significantly, fewer macrophages in the LYVE1−/MHCIIlo subgroup were actively phagocytic compared with the other three subgroups (Figure 5B, p<0.01), but those macrophages that were active showed greater phagocytic capacity than the other three subgroups.

Figure 5 with 2 supplements see all
Analysis of the phagocytic capacities of each macrophage subgroup.

(A) Phagocytic activity was measured for mouse skeletal muscle (SKM) macrophages at 4°C (control, low phagocytosis) and 37°C (active phagocytosis, right boxes). Phagocytic capacity was divided into groups that were negative (Neg; intensity <103), low (Lo; 103–104), medium (Med; 104–105), and high (Hi; >105), depending on signal intensities. Gating was established using fluorescence minus one (FMO) controls for each experiment. (B) Quantification of the macrophages showing active phagocytosis (Lo + Med + Hi) in the four subgroups. (C) Signal intensities of macrophages in each capacity group (Lo, Med, and Hi). (D) Quantification of number of active phagocytic macrophages in each subgroup of the three intensity groups. Data are representative of three independent experiments.

As macrophages showed a range of phagocytic capacities, we divided them into four groups by their geometric mean fluorescence intensity (gMFI): negative (Neg; intensity <103), low (Lo; 103–104), medium (Med; 104–105), and high (Hi;>105; Figure 5A). The phagocytic capacities of the four macrophage subgroups were similar (Figure 5C, n=3), and LYVE1+/MHCIIlo, LYVE1+/MHCIIhi, and LYVE1−/MHCIIhi subgroups showed similar numbers of active macrophages in each of the low-, medium-, and high-capacity groups (Figure 5D). However, LYVE1−/MHCIIlo macrophages showed significantly fewer active macrophages in the Lo group and strikingly more in the Hi capacity group compared to the other three subgroups (Figure 5D). This finding suggested that the LYVE1−/MHCIIlo group comprised two macrophage subpopulations with different phagocytic capacity: a silent group and a highly phagocytic group, each with roughly the same number of macrophages (Figure 5A, top). We performed efferocytosis assays to further assess the capacity of the macrophage subgroups in phagocytizing apoptotic cells. All four macrophage subgroups showed lower efferocytosis than phagocytosis, but LYVE1−/MHCIIlo macrophages again showed relatively greater capacity (Figure 5—figure supplement 1A, B).

These observations prompted us to further subclassify the LYVE1−/MHCIIlo subgroup by unsupervised clustering, which yielded six subclusters (SubCl; Figure 5—figure supplement 2A). GO annotation showed clustering of phagocytosis-related terms only in SubCl0 (Figure 5—figure supplement 2B). GO annotation suggests that SubCl0 may represent macrophages with higher phagocytic capacity in the LYVE1−/MHCIIlo subgroup (Figure 5A, B and D), although further studies are required for clarification.

Elevated proinflammatory and senescence-related mRNAs in old SKM macrophages

To investigate if there are aging-related changes in SKM macrophages, we further analyzed the genes differentially expressed in macrophages from young and old mouse SKM. The number of live macrophages isolated from SKM was comparable between young and old mice, both in males and females (Figure 6—figure supplement 1A-D, n=5), and the number of differentially abundant mRNAs was rather small, likely reflecting the lower sensitivity of scRNA-seq analysis. Therefore, we used slightly less strict criteria to find differentially expressed mRNAs: those expressed in >10% of total macrophages in young or old, p<0.01, and fold change >1.3. By these criteria, 41 mRNAs were more abundant, and 47 mRNAs were less abundant in macrophages from old SKM (Figure 6A). GO annotation suggested that mRNAs encoding proteins involved in chemotaxis of granulocytes (e.g. Cxcl1 and Cxcl2 mRNAs; Girbl et al., 2018) and monocytes (e.g. Ccl2 and Ccl7 mRNAs; Deshmane et al., 2009), and the cellular response to IFN-γ (e.g. Tnf, Cxcl10, and Zfp36 mRNAs) were less abundant in old SKM macrophages (Figure 6B and C). Some mRNAs encoding M2-like markers (e.g. Lyve1, Folr2, and Mrc1 mRNAs) were also significantly lower in old SKM macrophages (Figure 6A and C). By contrast, mRNAs encoding proteins related to cellular detoxification (e.g. Gsr, Hp, Prdx1, Prdx5, and Prdx6 mRNAs), inflammation (e.g. S100a8, S100a9, Fabp4, and Il1b mRNAs), senescence (Gpnmb and Spp1 mRNAs), and long-chain fatty acid transporters (Fabp4 and Fabp5 mRNAs; Wang et al., 2018; Babaev et al., 2011; Furuhashi et al., 2007; Pazolli et al., 2009; Suda et al., 2021; Suda et al., 2022) were elevated in old SKM macrophages (Figure 6B and C; full list in Supplementary file 4).

Figure 6 with 1 supplement see all
Analysis of gene expression programs in skeletal muscle (SKM) macrophages from young and old mice before clustering.

(A) In single-cell RNA-sequencing (scRNA-seq) analysis, a total of 88 mRNAs were differentially expressed between old and young SKM. Arrows indicate featured mRNAs upregulated (red) or downregulated (blue) in old SKM macrophages. (B) Gene ontology (GO) annotation depicting the functional categories that were upregulated and downregulated in the old SKM macrophages relative to young SKM macrophages. (C) Fold changes in the abundance of select mRNAs (O/Y), as determined from the scRNA-seq analysis.

We next analyzed the relative abundance of macrophage subgroups as a function of age. scRNA-seq indicated that LYVE1+ macrophages decreased, while LYVE1− macrophages increased in old SKM (Figure 7A). Flow cytometric analysis confirmed this trend, as LYVE1+ macrophages decreased and LYVE1− macrophages increased in old SKM (Figure 7B —supporting Figure 1A and B, n=4). Thus, both scRNA-seq and flow cytometric analysis confirmed the changes in numbers of LYVE1+ and LYVE1− macrophages in old SKM, consistent with the changes in Lyve1, Folr2, and Mrc1 mRNAs during aging (Figures 6A, C, 7A and B). All four macrophage subgroups displayed differentially expressed mRNAs. The top 15 elevated and top 10 reduced mRNAs in each subgroup were shown (Figure 7—figure supplement 1C). S100a9 mRNA, encoding a proinflammatory marker, was upregulated in all four subgroups, the senescence-related Gpnmb and Spp1 mRNAs and the fatty acid transporter Fabp5 mRNA were elevated in two MHCIIhi subgroups, LYVE1+/MHCIIhi and LYVE1−/MHCIIhi, while Apoe and Fabp4 mRNAs were only abundant in LYVE1−/MHCIIhi macrophages, and Il1b mRNA was elevated only in LYVE1−/MHCIIlo macrophages in old SKM (Figure 7—figure supplement 1C).

Figure 7 with 2 supplements see all
Identification of changes in macrophage subpopulations in old (O) relative to young (Y) skeletal muscle (SKM).

(A) Single-cell RNA-sequencing (scRNA-seq) analysis showing altered numbers of LYVE1+ and LYVE1− macrophages in old SKM. (B) Flow cytometric analysis showing comparable changes with scRNA-seq in old SKM. (C) Changes in macrophage numbers in unsupervised Cl0, 3, 6, and 8. (D) Top, UMAP plots showing Gpnmb, Spp1 and Fabp5 mRNAs in old (O) and young (Y) SKM (arrow, Cl6); violin plot representing Gpnmb mRNA (number of macrophages and expression levels) in the different clusters. Bottom, S100a8 and S100a9 mRNAs in O and Y SKM (arrow, Cl8).

In unsupervised clustering, Cl0 macrophages, mostly Lyve1+/Folr2+/Mrc1+, were less abundant in old SKM, while macrophages in Cl3, 6, and 8 increased in old SKM (Figure 7C). Gpnmb, Spp1, and Fabp5 mRNAs were largely concentrated in Cl6, a cluster that was strikingly enriched in old SKM (Figure 7D), and S100a9 and S100a8 mRNAs were elevated mainly in Cl8 in old SKM (Figure 7D). Biological replicates of the expression patterns of these genes in young and old SKM (Cl6 and Cl8) are shown (Figure 7—figure supplement 2A,B, respectively).

Overall, gene expression changes suggest that mRNAs related to chemotaxis and responses to pathogens were reduced, but mRNAs encoding proinflammatory, senescence, and cellular detoxification were elevated in macrophages from old SKM. In old SKM macrophages, senescence-related mRNAs were enriched in Cl6 and proinflammatory mRNAs in Cl8.

Discussion

Heterogeneity and functional versatility are critical characteristics of macrophages. Derived from embryonic and/or adult hematopoietic system (Cox et al., 2021), macrophages adapt their gene expression profiles to the tissues in which they reside and play diverse functions by polarizing to different subgroups. In this study, we identified functional subgroups of mouse SKM macrophages by single-cell transcriptomic analysis. Using unbiased clustering, we found 11 clusters, each comprising macrophages associated with reparative, proinflammatory, phagocytic, proliferative, and lipid homeostasis and senescence/aging functions, revealing the striking heterogeneity of SKM macrophages. An alternative classification based on membrane markers further revealed populations that expressed or lacked LYVE1 on their plasma membrane and could be further divided into four subgroups by the levels of cell-surface MHCII proteins. These four subgroups included the well-known M2-like and M1-like macrophages and two additional new subgroups that were confirmed by flow cytometry and immunohistology. Thus, our study has characterized diverse subpopulations of macrophages in resting mouse SKM.

A recent study comprehensively evaluated mouse SKM (Wang et al., 2020) and identified five clusters that largely overlapped with our findings. For example, the ‘CD209,’ ‘CCR2,’ and ‘proliferating’ clusters were very similar to our Cl0, Cl2, and Cl9, respectively (Figure 1 and Table 1). Moreover, the expression of M2-like markers (e.g. Lyve1, Mrc1, Folr2, and Cd163 mRNAs) suggested that the macrophages in ‘unspecified cluster 0’ are equivalent to our Cl1 macrophages, which also expressed many M2-like genes, although at lower levels than our Cl0. Excluding Cl0 from the comparison allowed us to identify M2-like features of the Cl1 (Table 1, Supplementary file 1). Furthermore, by analyzing both young and old SKM, we identified important new differences in macrophage clusters, including those associated with senescence and inflammation (Cl6 and Cl8, respectively). Gene expression patterns suggested that clusters ‘CD209’ and ‘CCR2’ resembled our LYVE1+/MHCIIlo and LYVE1−/MHCIIhi subgroups (Figures 3 and 4; Wang et al., 2020).

Among the supervised four subgroups, the new LYVE1+/MHCIIhi subgroup showed both M1- and M2-like gene expression patterns and functional capabilities (Figure 3C and Figure 4A). We hypothesize that this subgroup may have distinct functions or may have the potential to shift to M2-like LYVE1+/MHCIIlo or M1-like LYVE1−/MHCIIhi subgroups depending on surrounding conditions. The gene expression heat map showed that LYVE1+/MHCIIhi macrophages express features of both LYVE1+/MHCIIlo and LYVE1−/MHCIIhi, but these patterns are not prominent (Figure 3B). In flow cytometric analysis, LYVE1+/MHCIIhi macrophages spanned two distinct cell clusters, LYVE1+/MHCIIlo and LYVE1−/MHCIIhi (Figure 4A), possibly suggesting that LYVE1+/MHCIIhi macrophages represent an intermediate stage, even if they stand alone as an independent population (Figure 4B). The function of LYVE1+/MHCIIhi macrophages relative to LYVE1+/MHCIIlo and LYVE1lo/MHCIIhi macrophages requires further study.

By contrast, the new LYVE1−/MHCIIlo subpopulation, which clearly separated from the other three subgroups by flow cytometric analysis (Figure 4A), was predicted to have a more distinct ‘killing’ capacity and may be directly implicated in innate immunity. In phagocytosis assays, the LYVE1−/MHCIIlo subgroup showed fewer active macrophages (Figure 5A and B), but those that were active had strikingly greater phagocytic capacity compared to the other three subgroups (Figure 5D). Unbiased further clustering suggested that this specific subgroup consists of strong (SubCl0) and weak (SubCl1-5) phagocytic subclusters (Figure 5—figure supplement 2), consistent with phagocytosis assays. Importantly, Ly6c mRNA, known to be highly expressed in circulating monocytes (Wolf et al., 2019), was expressed in <3% of LYVE1−/MHCIIlo and the other subgroups (not shown), while CD11c, a dendric cell (DC) marker (Singh-Jasuja et al., 2013), and CD49 and CD122, candidate markers for lymphoid lineage natural killer (NK) cells (Nabekura and Lanier, 2016), were not detected in LYVE1−/MHCIIlo or the other subgroups (Supplementary file 3). These data strengthen the view that LYVE1−/MHCIIlo macrophages are distinct from circulating monocytes or DC and NK cells. Additional studies are also needed to characterize the function of LYVE1−/MHCIIlo subgroup in SKM.

Our study further revealed aging-related expression changes in macrophages in SKM. Overall, LYVE1+ macrophages were less abundant, and LYVE1− macrophages were more abundant in aged SKM (Figure 7A and B). Consistent with these observations, S100a8 and S100a9 mRNAs, encoding proinflammatory biomarkers, were significantly elevated in macrophages from aged SKM. Unlike neutrophils, macrophages were reported to express S100A8 and S100A9 at low levels in the absence of stimulation (Hessian et al., 1993; Wang et al., 2018). Often forming heterodimers, S100A8 and S100A9 serve as biomarkers for the diagnosis and therapeutic responses in inflammatory diseases like inflammatory arthritis and inflammatory bowel disease, while blocking their activity resulted in reduced inflammation in mouse models (Wang et al., 2018). S100a8 and S100a9 mRNAs were in very low abundance in macrophages from young SKM but were strikingly more abundant in old SKM (Figures 6C and 7D). The levels of Fabp4, Fabp5, and Il1b mRNAs, encoding additional proinflammatory proteins, were also upregulated in macrophages from old SKM (Figures 6C and 7D). This finding is important because macrophage-derived FABP4 and FABP5 were shown to promote a proinflammatory state in the vasculature during atherosclerosis development (Babaev et al., 2011; Furuhashi et al., 2007; Makowski et al., 2001), in keeping with the proinflammatory status of old SKM. We propose that the expression levels of S100A8 and S100A9 in macrophages can be essential indicators of the inflammatory status of SKM, and possibly other tissues (Wang et al., 2018). Several markers of senescence and aging, including Gpnmb and Spp1 mRNAs (Pazolli et al., 2009; Suda et al., 2021; Suda et al., 2022), were also elevated in old SKM macrophages (Figure 6A–C), suggesting the presence of senescent macrophages. We also found increased expression of mRNAs encoding antioxidant enzymes in old SKM macrophages, possibly reactive to elevated reactive oxygen species (ROS) in aged SKM (Jackson and McArdle, 2011).

By contrast, several mRNAs encoding neutrophil and monocyte/macrophage chemoattractants (Deshmane et al., 2009; Girbl et al., 2018) were expressed in lower amounts by old SKM macrophages (Figure 6B and C). In pathological conditions, like injury or infection, neutrophils are the earliest effector cells to infiltrate into the injury site followed by monocytes/macrophages (Forcina et al., 2020). At the same time, it is well known that injury repair and regeneration are slower in old SKM, perhaps due to a delay in leukocyte infiltration at early stages and to reduced CCAAT enhancer-binding protein β function toward regeneration after muscle injury (Blackwell et al., 2015). Thus, the reduced production of chemoattractants in macrophages may contribute to the delayed repair of older SKM.

Finally, unsupervised classification identified specific macrophage clusters significantly affected during SKM aging, particularly Cl6 and Cl8. Gpnmb mRNA, encoding the senescent membrane marker GPNMB (Suda et al., 2021; Suda et al., 2022), was concentrated in Cl6 and was significantly elevated in old SKM macrophages; similarly, senescence- and aging-related Spp1 mRNA and lipid transporter Fabp5 mRNA were highly enriched in Cl6 in old SKM macrophages (Figures 6A, C, 7D). On the other hand, S100a8 and S100a9 mRNAs were highly concentrated in Cl8 and significantly elevated in the old (Figures 6A, C, 7D). Thus, unsupervised clustering identified distinct subpopulations specifically altered during aging.

In closing, aging impacts all tissues and organs. Intrinsic and extrinsic factors, including DNA damage, endoplasmic reticulum stress, mitochondrial dysfunction, and a systemic inflammatory environment in aged individuals, inevitably affect the characteristics of macrophages (van Beek et al., 2019). A recent study suggested that macrophages from old SKM contributed to axonal degeneration and demyelination in the neuromuscular junction, and depletion of macrophages led to increased muscle endurance (Yuan et al., 2018). We propose that the age-associated SKM macrophage gene expression patterns identified here represent an important first step toward elucidating how macrophage subpopulations influence the pathophysiology of old SKM.

Materials and methods

Collection of SKMs from young and aged C57BL/6JN mice

Request a detailed protocol

All mouse work was done under an Animal Study Proposal (ASP #476-LGG-2023) that was reviewed and approved by the Animal Care and Use Committee of the National Institute on Aging (NIA), National Institutes of Health (NIH). Young (Y, 3 m.o.) and aged (O, 22–24 m.o.) male and female inbred C57BL/6JN mice were purchased from the NIA aged rodent colony (https://ros.nia.nih.gov/). The mice were sacrificed, and all hind limb muscles, including quadriceps, hamstring, gastrocnemius, soleus, and tibialis anterior muscles, were harvested. Collected samples were directly used for mononuclear cell isolation or frozen in isopentane chilled by liquid nitrogen and stored at –80°C for immunohistology.

Mononuclear cell isolation from SKM

Request a detailed protocol

Tendons, blood vessels, and fat tissues were removed under a dissection microscope. Muscle tissues were finely chopped and minced using dissection scissors to form a slurry. For scRNA-seq analysis, we isolated mononuclear cells with Miltenyi’s SKM dissociation kit (#130-098-305) with GentleMACS Octo Dissociator (#130-096-427), as described previously (Krasniewski et al., 2022). For further flow cytometric analysis, we also used an established method (Liu et al., 2015) with slight modifications. Briefly, the muscle slurry was digested with 1000 U/mL Collagenase type II (Gibco, Cat# 17101015) in 10 mL of complete Ham’s F-10 medium (Lonza, Cat# BE02-014F) for 70 min with 70 rpm agitation at 37°C. Partially digested muscles were washed in complete Ham’s F-10 medium and centrifuged at 400 rcf speed for 5 min, and cell pellet with 8 mL of the remaining suspension (pellet 1) was collected; 42 mL of the supernatant was collected in two tubes (21 mL each) that were filled up to 50 mL with Ham’s F-10 media and centrifuged again at 500 rcf for 8 min, and the pellet (pellet 2) was collected. Pellet 1 was subjected to a second round of digestion in 1 mL of 1000 U/mL Collagenase type II and 1 mL of 11 U/mL Dispase II (Thermofisher, Cat# 17105041) along with the 8 mL of the remaining cell suspension, for 20 min with 70 rpm agitation, at 37°C. Digested tissues were aspirated and ejected slowly through 10-mL syringe with 20-gauge needle followed by washing in complete Ham’s F-10 media at 400 rcf for 5 min. The supernatant was collected and centrifuged again at 500 rcf for 8 min, and the pellet obtained (pellet 3) was pooled with the pellet 2 above. The suspension of pellets 2+3 was filtered through 40-μm cell strainer (Fisher scientific, Cat # 22363547), followed by final wash in complete Ham’s F-10 medium. Cell pellets were resuspended in 1 mL complete Ham’s F-10 medium. Cell counting was performed using trypan blue (Invitrogen, Cat# T10282) at a 1:1 ratio in Countess cell counting chamber slides (Invitrogen, Cat# C10228) using Countess II FL Automated Cell Counter (Invitrogen).

Flow cytometric analysis and FACS

Request a detailed protocol

Flow cytometric analysis and CD11b+ cell sorting by FACS for scRNA-seq were described in detail in our previous report (Krasniewski et al., 2022). For further flow cytometric validation studies and RT-qPCR analysis, mononuclear cell suspensions were incubated with BD Horizon Fixable Viability Stain 780 (FVS780, BD Biosciences, Cat# 565388, dilution: 1:4000) in PBS (Ca+ and Mg+ free, Thermofisher) for 30 min at 4°C in the dark. Fc receptors were blocked using TruStain FcX (anti-mouse CD16/32) Antibody (Biolegend, Cat# 101320, Clone: 93, dilution 1:1000) for 5 min at 4°C in FACS staining buffer (1% BSA and 10 mM EDTA in Miltenyi’s Auto MACS Rinsing Solution). For macrophage sorting, mononuclear cells were further stained in FACS staining buffer for 40 min at 4°C in the dark, with fluorochrome conjugated antibodies specific to mouse as indicated: BUV395 Rat anti-mouse CD45 (BD Biosciences, Cat# 564279, Clone: 30-F11, dilution: 1:100), PE anti-mouse/human CD11b Antibody (Biolegend, Cat# 101208, Clone: M1/70, dilution: 1:100), PE/Cyanine7 anti-mouse F4/80 Antibody (Biolegend, Cat# 123114, Clone: BM8, dilution: 1:40), Brilliant Violet 711 anti-mouse I-A/I-E Antibody (Biolegend, Cat# 107643, Clone: M5/114.15.2, dilution: 1:40), and APC Rat Anti-Mouse Lyve1 Antibody (Thermofisher, Cat# 50-0443-82, Clone: ALY7, dilution: 1:20; see Supplementary file 5 for a full list of antibodies). Stained cells were fixed using BD Cytofix Fixation buffer (BD Biosciences, Cat# 554655) for 20 min on ice in the dark for analysis (but not for sorting). Compensation matrices were created using single color controls prepared using COMPtrol Kit, Goat anti-mouse Ig (H&L) coated particles, with negative and high in separate vials (Spherotech, Cat# CMIgP-30–2 K), combining one drop from each vial in equal ratio. Gating was based on FMO (fluorescence minus one) controls for each experiment. The cells were acquired on a BD FACSAria Fusion (BD Biosciences) instrument and analyzed with Flowjo software (Tree Star, Inc).

Macrophage scRNA-seq by 10× Genomics

Request a detailed protocol

Macrophages isolated from three 3 m.o. and three 23 m.o. C57BL/6JN male mice (biological triplicates) were stained with CD11b antibody and isolated by FACS analysis. Given that the lengthy collection protocol made it impossible to process all the mice on the same day, we isolated cells in three consecutive weeks: from two young mice (Y1 and Y2) the first week, from two old (O1 and O2) the second week, and from one young mouse (Y3) and one old mouse (O3) the last week. Isolated SKM macrophages were immediately subjected to single-cell library construction without culture to minimize differences related to batch effects. Single-cell libraries were prepared with 10× Genomics Chromium Single Cell 3ʹ Reagent Kits v3 (10× Genomics Cat# PN-1000092) with Chip B (10× Genomics, Cat# PN-1000073) following the manufacturer’s protocol. Briefly, 5000–10,000 single macrophages were used for GEM (Gel Bead-in-Emulsion) generation. The cDNAs were then synthesized, and their qualities were assessed on the Agilent Bioanalyzer with High-Sensitivity DNA kit (Agilent Cat# 5067–4626). cDNAs were then used for library preparation and the quality of the final libraries assessed on the Agilent Bioanalyzer with DNA 1000 kit (Agilent, Cat# 5067–1504). The libraries were sequenced with an Illumina NovaSeq 6000 sequencer with a mean depth of ~80,000 (70,876–156,962) RNA-seq reads per cell, corresponding to ~2000 (2027–2256) genes per cell. The numbers of cells from each mouse successfully sequenced and subjected to statistical analysis are as follows: Y1, 3730; Y2, 3325; Y3, 2033 and O1, 3391; O2, 5338; O3, 4097. RNA-seq data were deposited in GEO with identifier GSE195507.

scRNA-seq data analysis

Request a detailed protocol

scRNA-seq samples were demultiplexed and mapped to the mm10 mouse reference genome using the Cell Ranger software version 3.0.2 (10× Genomics). Further analysis of the matrices of read counts obtained was carried out in R (version 4.1.3) with the Seurat package, version 4.1.0 (Hao et al., 2021), using default parameters in all functions, unless specified otherwise. To exclude empty droplets, poor-quality cells, and potential doublets from downstream analysis, quality control filtering was applied for each sample, which removed cells containing more than 7.5% mitochondrial genes, cells expressing <300 or >7000 transcripts, and below 500 or above 60,000 counts. Genes that were detected in less than 10 cells were eliminated from the analysis. Cells expressing Itgam (Cd11b) and Adgre1 (F4/80) mRNAs, two key macrophage markers, were subjected to further analyses.

Each sample was normalized with the LogNormalize method, and the top 2000 variable genes were selected with the FindVariableFeatures function. The SelectIntegrationFeatures function was applied to find shared variable features across the samples, and the FindIntegrationAnchors function was used to identify inter-sample anchors for integration. Then, the samples were integrated with the IntegrateData function, scaled, and subjected to principal component analysis (PCA).

For supervised cluster analysis, the macrophage dataset was divided into four cell subgroups based on the log-normalized expression values of Lyve1 and H2-Ab1 (MHCII) mRNAs, as follows: LYVE1+/MHCIIlo (Lyve1 >0 and H2-Ab1 <2), LYVE1+/MHCIIhi (Lyve1 >0 and H2-Ab1 ≥2), LYVE1−/MHCIIhi (Lyve1 ≤0 and H2-Ab1 ≥2), and LYVE1−/MHCIIlo (Lyve1 ≤0 and H2-Ab1 <2). For unsupervised cell clustering, a shared nearest neighbor graph was generated with the FindNeighbors function (using the first 30 principal components) and clustered with Louvain algorithm in the FindClusters function with a resolution of 0.3. To visualize and explore cell clusters in a two-dimensional space, the Uniform Manifold Approximation and Projection (UMAP) analysis was performed using the first 30 principal components, as determined by the ElbowPlot method. To identify subpopulations of LYVE1−/MHCIIlo cells, the analysis was rerun on the LYVE1−/MHCIIlo subgroup, and clusters were visualized with resolution set to 0.3.

Differentially expressed marker genes for each cluster were identified with FindAllMarkers function, and the FindMarkers function was used to find differentially expressed genes across conditions. Those mRNAs that were expressed in at least 25% of cells per cluster were considered for differential gene expression analysis among clusters. mRNAs were defined as differentially expressed if they had an absolute fold change >1.5 and adjusted p-value<0.01. All R processing scripts are included in Supplementary file 6.

Functional annotation of the differentially expressed genes was performed using the web-based tool g:Profiler (Raudvere et al., 2019) (https://biit.cs.ut.ee/gprofiler/gost). The analysis was done with differentially expressed genes in corresponding subpopulations with ‘g:SCS threshold’ as a ‘significance threshold’ and 0.05 as the ‘user threshold’, and functional terms for ‘GO biological process’ were collected. In addition, we used 14,542 genes detected from young and old macrophages in our scRNA-seq analysis as the background gene set for GO annotation.

RT-qPCR analysis

Request a detailed protocol

For RT-qPCR analysis, CD11b+/F4/80+/LYVE1+ and CD11b+/F4/80+/LYVE1− macrophages were isolated by FACS. Sorted LYVE1+ and LYVE1− macrophages were lysed with lysis buffer (RNeasy Mini Kit, Qiagen, Cat# 74104) and stored at –80°C. RNA was then isolated with a QIAcube (Qiagen) instrument following the manufacturer’s protocol, using a column for RNase-Free DNase I (Qiagen, Cat# 79254) digestion. The quality of isolated RNAs was assessed on the Agilent TapeStation with RNA Screen Tape (Agilent, Cat# 5067–5576). RT was performed by synthesizing cDNAs from the LYVE1+ and LYVE1− mRNAs with the Superscript III First-Strand Synthesis System (Invitrogen, Cat# 18080051), and qPCR amplification was carried out using ready-to-use Taqman probe/primer sets (Applied Biosystems) to detect expression levels for Lyve1 (Mm00475056_m1), Folr2 (Mm00433357_m1), Cd209f (Mm00471855_m1), Fcna (Mm00484287_m1), Timd4 (Mm00724713_m1), Mrc1 (Mm01329362_m1), Igf1 (Mm00439560_m1), Ang (Mm01316661_m1), Il1b (Mm00434228_m1), and Gapdh (Mm99999915_g1) mRNAs. Two biological replicates (n=2 per replicate) were used for the LYVE1+ and LYVE1− macrophages and assayed in triplicate. The relative RNA levels were calculated after normalizing to Gapdh mRNA using the 2−ΔΔCt method, and the data were analyzed for significance using Student’s t-test.

Phagocytosis assays

Request a detailed protocol

Macrophages were isolated from the hind limb muscles of C57BL/6JN male mice as described above. Mononuclear cells from three animals were pooled for each set of experiments, and cells were aliquoted for necessary treatment conditions and technical replicates. Three biological replicates (total nine mice) were analyzed. The phagocytic activity of macrophages was measured by red fluorescence from pHrodo E. coli bioparticles (Invitrogen, Cat# P35361). Briefly, 6×106 macrophages were resuspended in 200 μL of Ham’s F-10 complete media (Lonza, 12–618 F) containing 10% horse serum (Gibco, 16050114) for each sample. Aliquots of 20 μL of pHrodo E. coli bioparticles, resuspended in live-cell imaging buffer (1 mg/mL, Invitrogen, Cat# A14291DJ) and sonicated for 2 min × 3, with 2 min intervals on ice between each sonication, were added to each cell tube, including appropriate FMO control tubes. Cell suspensions were gently and thoroughly mixed to ensure a homogenous distribution of the E. coli bioparticles. One set of samples was immediately transferred to a CO2 incubator for 2 hr at 37°C, and another set (negative control) was incubated on ice for 2 hr. After incubation, cells were washed with live cell imaging solution at 400 rcf for 5 min, followed by another wash with PBS. All steps were performed in the dark.

After the phagocytosis assay, cells were stained with viability dye followed by primary antibody staining as described above. Fluorochrome-conjugated antibodies used for staining the cells are as follows: BUV395 Rat anti-mouse CD45, PE-Cyanine7 anti-mouse/human CD11b Antibody, BUV737 Rat anti-mouse F4/80, Brilliant Violet 711 anti-mouse I-A/I-E Antibody, APC Rat anti-mouse LYVE1 Antibody. The cells were acquired on a BD FACSAria Fusion instrument on the same day and analyzed with Flowjo. For all the samples, including controls, CD11b+/F4/80+macrophages were further categorized as high (Hi, >105), medium (Med, 104–105), low (Lo, 103–104), and negative (Neg, <103) intensity groups based on their ability to engulf labeled bacteria. The relative phagocytosis levels for each group were calculated using gMFI. For statistical analysis, we performed a Shapiro-Wilk test (Mishra et al., 2019) first to assess if our data were normally distributed (GraphPad Prism 8). We found that all data shown in Figure 5B and D, and Figure 4—figure supplement 1B, and Figure 5—figure supplement 1B were normally distributed (not shown). Therefore, we performed parametric tests, one-way ANOVA (Dunnett’s multiple comparisons test) for Figure 5D, and two-way ANOVA (Sidak’s multiple comparisons test) for Figure 5B and Figure 4—figure supplement 1B, and Figure 5—figure supplement 1B using GraphPad Prism 8.

Efferocytosis assay

Request a detailed protocol

To study the engulfment of apoptotic cells by SKM macrophages, hind limb muscles from three male mice were combined as one biological replicate, followed by digestion to generate a mononuclear cell suspension as mentioned above, and pooled for study of phagocytic cells. To generate apoptotic cells, Jurkat T cells, cultured in RPMI 1640 (Thermo Fisher, Cat# 11875–093) with 10% heat-inactivated FBS (Thermo Fisher, Cat# 10438025) at 37°C, 5% CO2 10 mL were collected from a cell culture flask, washed with PBS, pelleted gently, and resuspended in 1 mL PBS. For labeling Jurkat cells, CFSE (CellTrace CFSE Cell Proliferation Kit, Thermo Fisher, Cat# C34554) was added to cells at final 5 µM concentration and incubated at 37°C for 20 min; 10 mL of RPMI 1640 (10% heat-inactivated FBS) was then added, mixed by vortexing, and further incubated at 37°C for 5 min. After washing, cells were resuspended in 1 mL serum-free RPMI 1640 medium. Apoptosis was induced by treatment with 1 µM staurosporine (Millipore Sigma, Cat# 19–123) for 5 hr at 37°C, 5% CO2, followed by washes in RPMI 1640 (10% heat-inactivated FBS) and resuspension in 1 mL RPMI 1640 (10% heat-inactivated FBS) for use in efferocytosis assays. Mononuclear cells from SKM and apoptotic Jurkat T cells were counted and combined at a 1:1 ratio in 2 mL RPMI 1640 (10% heat-inactivated FBS) and incubated for 18 hr at 37°C, 5% CO2. Mononuclear cells from SKM without Jurkat cells were used as controls. After incubation for 18 hr, cells were assayed by flow cytometry, as explained above.

Immunofluorescent staining of macrophages in mouse SKM

Request a detailed protocol

Frozen sections from rectus femoris muscle from 3 m.o. C57BL/6 J mice were cut, fixed in cold acetone, and subjected to regular double immunofluorescent staining or double TSA staining (Tyramide SuperBoost kit, Thermo Fisher, Cat# B40932) as performed previously (Cui et al., 2019). Primary antibodies recognizing LYVE1 (Abcam, Cat# ab14917, 1:200 dilution), MHCII (Invitrogen, Cat# 14-5321-82, 1:100), CD31 (Millipore, Cat# MAB1398Z, 1:100), and TUBB3 (Biolegend, Cat# 801201, 1:200) worked well for regular immunostaining. Secondary antibodies were used for LYVE1 (Invitrogen, Cat# A-11012, 1:1000 dilution) and MHCII staining (Invitrogen, Cat# A-11006, 1:1000 dilution). Detection of CD11b (Santa Cruz, Cat# sc-1186, 1:50 dilution) required TSA staining. To identify LYVE1+/MHCIIhi macrophages in SKM, we carried out three sets of double staining: LYVE1 with MHCII by regular immunofluorescence staining, CD11b with LYVE1, and CD11b with MHCII by TSA staining. Micrographs were taken on a DeltaVision microscope using a 20× lens.

Data availability

The single-cell RNA-seq analysis was uploaded to GEO with identifier GSE195507.

The following data sets were generated

References

    1. Hessian PA
    2. Edgeworth J
    3. Hogg N
    (1993)
    Mrp-8 and MRP-14, two abundant Ca (2+) -binding proteins of neutrophils and monocytes
    Journal of Leukocyte Biology 53:197–204.

Decision letter

  1. Bérénice A Benayoun
    Reviewing Editor; University of Southern California, United States
  2. Carlos Isales
    Senior Editor; Augusta University, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Single-cell analysis of skeletal muscle macrophages reveals age- associated functional subpopulations" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Carlos Isales as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The reviewers and I discussed the manuscript, and we believe that while some major revisions are needed, this manuscript could be appropriate for eLife after major revisions. The most salient are summarized below:

1. The choice of a supervised clustering approach for macrophage heterogeneity analysis was puzzling to reviewers, especially with the choice of using noncanonical markers – the rationale for not using an unsupervised approach was not well explained. In addition, the chosen non-standard nomenclature is confusing, and it is recommended to avoid shorthand to improve the reading experience. In a revised manuscript, the authors should start with an unsupervised approach, as is the gold standard of the field, which can then reveal specific markers that may be used to segregate functional subpopulations.

2. The manuscript needs to provide additional information about both wet and dry methods used to generate results for improved reproducibility: (i) better clarification of potential batch processing, (ii) more information on the functional enrichment analysis, (iii) support for using parametric tests (i.e. tests for normality), (iv) extensive information on package and software versions used for analysis, (v) catalog details for flow antibodies and information on the sorting schemes, (vI) deposition of all R processing scripts, etc.

3. Information about the provenance of the macrophages in the tissue (i.e. associated to blood vessels, inside the parenchyma, etc.) in relation to the diverse phenotypes identified using scRNAseq.

4. Many conclusions need to be toned down as most of the functional information is derived from genomic annotation and not functional assays (i.e. the "wound healing" discussion in the absence of healing assays or of efferocytosis assays) to reflect the degree of evidentiary support. It is also noted that lower numbers of DE genes in scRNAseq datasets is a common caveat of the method that should not be overinterpreted.

5. The reviewers were concerned about the lack of consideration of sex as a biological variable. It was discussed that, in the absence of scRNAseq data in female animals, either some functional experiments should be performed in females to confirm the broader applicability of results, or the male-specific nature of the paper should be explicitly discussed.

In general, all discussed data should be shown and conclusions should not reach beyond that which is directly supported by data unless explicitly stated to be speculation.

Reviewer #1 (Recommendations for the authors):

1. Aging and immunity are both very sex dimorphic. Since the authors profiled exclusively male animals, it will be important to explicitly discuss how these results may differ in female animals in the Discussion section.

2. When describing results from the flow cytometry-based phagocytosis assay (Figure 4), the authors find that LnHl macrophages are phagocytic to a lower proportion than other described subgroups (~49% compared to >85%), although the phagocytic macrophages show strikingly higher levels of phagocytosed cargo by MFI analysis. Although we agree that the significance of this is unclear with the current evidence, this suggests further heterogeneity in the LnHl group. It would thus be important to try to use unbiased SNN clustering of LnHl macrophages (not just of all macrophages as in Figure 6) and identify potential subpopulations that may explain this functional heterogeneity using the generated scRNAseq data (for instance as relating to phagocytosis-gene related mRNAs).

3. Based on experimental flow/description, it sounds like young and old samples may not have been processed in parallel, which may be problematic due to the known impact of batch effects in genomics. Can the authors clarify and explicitly discuss whether batching may be a problem?

4. There needs to be additional provided information about some of the bioinformatic tools and/or analyses.

a. Although the authors generally provide information about software/package versions or dates of access, some are missing (e.g. R, g:profiler). This needs to be updated for reproducibility.

b. Although a GO analysis with g:profiler is described in the text and figures, the method is not described in the method section. Since the nature/use of background lists in functional enrichment analysis is crucial, the authors should clarify the list of genes used as background for enrichment analysis (ideally all detected/expressed genes), as well as the FDR threshold for considering a term significantly enriched. A supplementary table with all enriched terms would also be invaluable.

c. For long-term reproducibility, it would be important to either deposit all R scripts to a public repository such as github or provide it as a supplemental archive to accompany the manuscript.

d. In the methods, the authors mention using Student's t-tests, but not tests to verify that data was normally distributed before use of the t-test. Please include the reference to any normality test used, or, if normality of data cannot be verified, please update to use non-parametric tests.

5. Please provide catalog numbers for the antibodies used in the flow phagocytosis assay, as is needed for reproducibility (methods, page 20).

Reviewer #2 (Recommendations for the authors):

1) Pg5 the authors rely on a limited number of markers to determine the macrophage polarization status (M1 vs M2) of the different subsets of macrophages characterized in the study. Can the authors investigate more exclusive markers (PMID: 26699615) to determine if MHCII and Lyve macrophage subsets are indeed skewed to one state or the other?

2) Lp+ macrophages were described in the study to express a transcriptomic program characterized by M2-like gene program involved in wound repair and healing. Thus, beyond investigating the phagocytic capacity of the macrophage using labeled E. coli, can the authors test the uptake of apoptotic cells using an efferocytosis assay, which seems more relevant for wound repair.

3) Pg7 "We found that MHCII mRNAs (encoding H2-Ab1, H2-Eb1) divided SKM macrophages into two groups, MHCII-high (Hh) and MHCII-low (Hl) in single-cell profiling analysis." Please provide a figure reference.

4) Pg 8, in the immunostaining are different subsets of macrophages localized to different sections of muscle tissues? i.e. more associated with endothelial cells, muscle cells or innervated areas?

5) Can the authors provide more insight into both old and young skeletal muscle to determine if an absolute number of macrophages change during aging (for example via flow cytometry) and also to determine using transcriptomics if the subpopulations of macrophages in the study are resident vs non-resident macrophages.

6) The study lacks analysis of skeletal muscle macrophages in female young and old mice, and would be more informative if they were included in the study to determine any sex differences. Perhaps qPCR or flow cytometry analysis of the four major subsets can be investigated in female mice, since performing single cell would be cost prohibitive.

7) Pg 11. "Cdk1 and Top2a mRNAs were expressed in an even lower number of macrophages (data not shown)." I am confused by data not shown since the data appears to be in table. If not please show.

Reviewer #3 (Recommendations for the authors):

1) Manuscript provides no functional insights of how changes in macrophages affect muscle-physiology/pathology in young and aged mice. Conclusion about functions are based on associations.

2) Please clarify what the percentage of macrophages in muscle in comparison to other CD45+ cells is.

3) Please clarify wheter the analyzed myeloid cells are present in vessels or in muscle parenchyma. How that affects muscle function is also unclear.

4) While the scRNAseq of skeletal muscle macrophages reveals interesting findings about their diversity and how aging affects their transcriptomic profile, the authors chose an unusual approach to analyze the data by defining subsets based on extracellular marker expression (Lyve1 and MHCII) and not transcriptomic profile. By starting with a supervised approach, the authors have missed an important part of their data regarding the high diversity of skeletal muscle macrophages, that cannot only be described through MHC-II and Lyve1 expression. The expression of extracellular markers does not necessarily allow to define functionally distinct subsets. A deeper and more detailed unsupervised analysis is required for scRNA seq data. Moreover, as a starting point, the clusters should be defined based on gene expression and then propose candidate markers to characterize the subpopulations and their functional properties

5) Several conclusions are not supported by the data – e.g.

a) Naming Lyve1+ and Lyve1- macrophages "healing" and pro-inflammatory" is premature as this has not been functionally tested.

b) Unless the authors have data showing the general health of the animals used for the scRNAseq, the low number of differentially expressed genes between young and old shouldn't be interpreted as a result of a "healthier" status of the aging cohort. It is rather a common caveat of scRNAseq that lacks sequencing depth. A bulk RNAseq of specific sorted population that changes in aging may be important.

6) Characterization of macrophages with polarization markers is inadequate and based on old literature. Hence, not very informative.

7) The unsupervised analysis of the scRNAseq lacks depth. Each cluster should be described independently of whether they belong to LpHl, LpHh, LnHh and LnHl. This analysis should come as the first figure. For example, do not dismiss the possibility of non-macrophage clusters as cells were only sorted on CD11b marker. Cluster 6 could be a granulocyte cluster.

8) Overall, the logic behind the direction taken for the analysis of the data is unclear:

– It is not clear why the authors chose to perform a scRNAseq. If the goals of the authors were to study macrophages subsets based on extracellular markers, sorting these subsets, and performing a bulk RNAseq would have been more adequate and would have provided and better sequencing depth.

– The choice of MHC-II and Lyve1 markers to divide macrophage subset seems arbitrary. It is not clear why the authors chose to use markers described for lung macrophages specifically.

– The data showing phagocytic capacities are interesting. Though, it is not clear why the authors chose to investigate this function. Is phagocytosis relevant to the physiological function of skeletal muscle tissues? Are there other functions highlighted in the GO annotations that can be tested? (endocytosis, inflammatory response, antigen presentation)

9) It is difficult to evaluate the solidity of the data. For quantification of populations by flow cytometry analysis and using the scRNAseq data, we suggest adding histograms representing n numbers and statistical significance when comparing the abundance of populations (Figure 3D, S3C, 5A, 6A). Moreover, instead of using tables to depict differentially expressed genes, heatmaps or volcanoplots are recommended (Figure 5C, S3C).

Figure 1C and Figure 2: Ln cells appear to be multiple clusters – with same features? How is that possible?

Figure 1D: The bioinformatic analyses used to characterize Lp/Ln cells lacks statistical validity.

Figure 3D: biological purpose to reveal new insights is lacking.

Figure 3E; Fails to provide spatial clarity of localization in muscle.

Figure 4: phagocytic analyses are superficial and fail to provide new insights.

Figure 6: is the most important figure. However, it is poorly put together in terms of data analyses and presentation. Identity of cluster 5 and 6 which change with age is the main finding – but their relevance is unknown

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Single-cell analysis of skeletal muscle macrophages reveals age-associated functional subpopulations" for further consideration by eLife. Your revised article has been evaluated by Carlos Isales (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Notably, after a new round of review/discussion, the reviewers felt like the issue of batching in the data needed further information/analyses to make sure that the reported results are not the result of batch and biology being confounded in 2 out of the 3 replicates for each age.

Upon discussion, the reviewers thought that this could be addressed with:

1. Doing an analysis paralleling the main manuscript but using only the 2 samples that were processed "unbatched" (i.e. the samples where 1 young and 1 old mouse were processed in parallel), and only these. If the main results of the study are conserved in this unbatched subset of the data, this would strengthen the likelihood that the batching did not grossly impact the conclusions. We would then recommend including this analysis as a supplement.

2. As highlighted by reviewer #3, it is crucial that the batching/experimental collection scheme be discussed explicitly in the manuscript.

3. Finally, please address Reviewer #1's remaining concern on the use of a background gene list for the g:profiler analysis.

Thank you!

Reviewer #1 (Recommendations for the authors):

Although the authors have addressed most of my concerns, some large concerns remain at this point.

1. A very large concern was revealed by their answer to one of my questions about the batchiness of the data. Indeed, the author's response revealed 3 batched: (i) only young samples, (ii) only old samples and (ii) one old and one young. Unfortunately, since batch and biological groups are confounded for groups i and ii, that data is meaningless (i.e. batch cannot be properly accounted for when it is confounded with biology). Since I understand that the authors may not be able to redo the entire experiment the way it should have been done, I believe it is imperative that all analyses also be done exclusively on batch 3 (the one where both groups were represented), to show that all results would hold in the absence of batch. The results should then be included and compared/discussed in the context of the paper as this is a big problem.

2. The authors still did not address the background list used for GO enrichment in g:profiler. This leads me to believe they used the default (all genes instead of detected genes in the dataset), which is incorrect and would lead to spurious enrichments. These analyses should be rerun with the correct background list.

Reviewer #2 (Recommendations for the authors):

We believe the authors have responded to the concerns of the reviewers sufficiently and the paper is significantly improved. Thus, in our opinion, the paper is suitable for publication.

Reviewer #3 (Recommendations for the authors):

The authors have addressed most of my prior concerns. Some issues remain, but in general given the importance of the topic, the manuscript is ready to forward in the process.

An important issue that remains unaddressed is that the scRNA analyses and cell sorting for young/old groups were done on different days. The authors responded to this issue and acknowledged this caveat, but do not describe the consequence of this on the data generation and conclusions.

Authors need to provide this information. This reviewer could not find it in the beginning of the Results section where the description of data generation is provided.

https://doi.org/10.7554/eLife.77974.sa1

Author response

Essential revisions:

The reviewers and I discussed the manuscript, and we believe that while some major revisions are needed, this manuscript could be appropriate for eLife after major revisions. The most salient are summarized below:

1. The choice of a supervised clustering approach for macrophage heterogeneity analysis was puzzling to reviewers, especially with the choice of using noncanonical markers – the rationale for not using an unsupervised approach was not well explained. In addition, the chosen non-standard nomenclature is confusing, and it is recommended to avoid shorthand to improve the reading experience. In a revised manuscript, the authors should start with an unsupervised approach, as is the gold standard of the field, which can then reveal specific markers that may be used to segregate functional subpopulations.

We sincerely appreciate your advice after consulting the reviewers. Following your recommendation, we have reorganized our revised manuscript to start with unsupervised clustering skeletal muscle (SKM) macrophages by single-cell RNA-sequencing analysis, followed by supervised classification using membrane markers LYVE1/MHCII, and ending with the functional analysis. We agree that this approach provides a better flow and better appreciation of the complementary between the two methods for analysis of SKM macrophages.

We agree that we had not explained well why our original manuscript began with the classification of macrophages using noncanonical membrane markers. Briefly, while the traditional classification of M1 and M2 macrophages was largely based on membrane markers CD206, CD86, and CD80, LYVE1 had recently emerged as a promising membrane marker for M2 macrophages, and MHCII proteins for M1 macrophages, and therefore we examined them here. As indicated in our manuscript, for functionally subgrouping skeletal muscle macrophages, the LYVE1/MHCII combination appeared to be more informative than the traditional CD206/CD86 or CD206/CD80 combination.

Following your advice, we also changed our designation of the macrophage subgroups to follow more conventional nomenclature across the text and figures: ‘Lp’ is now LYVE1+, ‘Ln’ is LYVE1-, ‘LpHl’ is LYVE1+/MHCIIlo, ‘LpHh’ is LYVE1+/MHCIIhi; ‘LnHp’ is LYVE1-/MHCIIhi; and ‘LnHl’ is LYVE1-/MHCIIlo.

2. The manuscript needs to provide additional information about both wet and dry methods used to generate results for improved reproducibility: (i) better clarification of potential batch processing, (ii) more information on the functional enrichment analysis, (iii) support for using parametric tests (i.e. tests for normality), (iv) extensive information on package and software versions used for analysis, (v) catalog details for flow antibodies and information on the sorting schemes, (vI) deposition of all R processing scripts, etc.

We fully agree with these requests and have implemented them as follows:

(i) Better clarification of potential batch processing:

To avoid batch effects, we had adopted the following measures. For wet-lab methods, we isolated cells in 3 consecutive weeks, from 2 young mice one week, from 2 old mice on the second week, and from 1 young and 1 old mice on the third week. Single-cell libraries were prepared on the same day of each isolation using the 10x Genomics reagents and methods. We have included this information in the ‘Materials and methods’ section.

For dry-lab methods, we did not expect substantial batch effects in the data, as the sequencing of all the samples was performed at the same time, and the absence of batch effects was confirmed by careful visual inspection of cell distributions in each sample. Additionally, batch-effect corrections are part of the integration steps in the Seurat package we used in our analysis, which eliminates potential technical differences across samples (https://satijalab.org/seurat/articles/integration_introduction.html). In the revised version, we described these steps more thoroughly in the Material and Methods section and present cell alignments across all samples (Figure 1—figure supplement 1B, Figure 3—figure supplement 1B, and Figure 7—figure supplement 2).

(ii) More information on functional enrichment analysis:

The functional enrichment analysis was carried out using g:Profiler. This information is included in the revised Materials and methods section.

(iii) Support for using parametric tests (i.e. tests for normality):

In the revised Materials and methods section, we have updated the tests used for normality. We performed a Shapiro-Wilk test first to confirm the normal distribution of our data. We then performed parametric tests (one-way or two-way ANOVA). We have described these details in the revised Materials and methods section and have included the results of the Shapiro-Wilk test in our response to Reviewer 1 (below).

(iv) Extensive information on package and software versions:

In the revised Materials and methods section, we have updated the information on tests, packages, and software versions used.

(v) Catalog details for flow antibodies and information on the sorting schemes:

We have added a supplementary Table S6 in which we include catalog numbers for the antibodies used in the flow cytometry and immunofluorescence analyses. For information on the sorting schemes, we have reported details in our protocol paper, which is in press in Bio-Protocols now (and we have appended as a supplemental item in this revision). In the revised manuscript, we have stated this information more clearly, and have made clearer the sorting procedure, under ‘Flow cytometric analysis and FACS’ section.

(vi) Deposition of all R processing scripts, etc.

Following the advice of Reviewer 1, we have added a supplementary archive (Table S7) that contains all the scripts used.

3. Information about the provenance of the macrophages in the tissue (i.e. associated to blood vessels, inside the parenchyma, etc.) in relation to the diverse phenotypes identified using scRNAseq.

We appreciate this important request. To address it, we have carried out additional immunostaining to identify cells from other tissues associated with the muscle. These findings have been added to the new Figures 4B and 4C. Briefly, all macrophages were localized in the perimysium or endomysium areas (i.e., between muscle fibers, but not inside muscle fibers) as shown in Figure 4B. We further found that LYVE1+ macrophages (LYVE1+/MHCIIlo and LYVE1+/MHCIIhi), but not LYVE1-/MHCIIhi macrophages, were closely localized with blood vessels (Figure 4C, top), similar to previous findings in lung tissue, reported by Chakarov et al., 2019. However, all macrophage subtypes were located around nerve fibers. We have revised the text to highlight these results when we describe Figure 4C.

4. Many conclusions need to be toned down as most of the functional information is derived from genomic annotation and not functional assays (i.e. the "wound healing" discussion in the absence of healing assays or of efferocytosis assays) to reflect the degree of evidentiary support. It is also noted that lower numbers of DE genes in scRNAseq datasets is a common caveat of the method that should not be overinterpreted.

We fully agree with these recommendations. We have toned down the interpretations of the data throughout the text, particularly when the relevant functional assays were not available. Regarding the lower number of differentially expressed genes in scRNAseq datasets, we appreciate the Reviewer’s excellent advice and have made explicit note of this limitation in the revised Results.

5. The reviewers were concerned about the lack of consideration of sex as a biological variable. It was discussed that, in the absence of scRNAseq data in female animals, either some functional experiments should be performed in females to confirm the broader applicability of results, or the male-specific nature of the paper should be explicitly discussed.

We agree with and appreciate these recommendations. To address them, we now clearly state that the work was carried out using male mice in the Abstract, Results, Discussion, and Materials and methods sections. In addition, to understand possible male-vs-female influences on macrophage polarization in SKM, we carried out flow cytometric analysis using macrophages from female mice. We found statistically decreases in the LYVE1+/MHCIIhi subgroup and increases in LYVE1-/MHCIIlo macrophages in females compared to males. We included these data in the supplement (Figure 4—figure supplement 1A,B). These results have encouraged us to initiate more detailed studies to understand the biological significance of the differences between male and female SKM macrophages, and we hope to report them in the near future as a separate investigation.

In general, all discussed data should be shown and conclusions should not reach beyond that which is directly supported by data unless explicitly stated to be speculation.

We appreciate and agree with this advice. We have toned down our conclusions across the text.

Reviewer #1 (Recommendations for the authors):

1. Aging and immunity are both very sex dimorphic. Since the authors profiled exclusively male animals, it will be important to explicitly discuss how these results may differ in female animals in the Discussion section.

We appreciate this important reminder. We fully agree with the Reviewer and have revised the text in two ways. First, we state in the revised text (Abstract, Results, Discussion, and Materials and methods sections) that the study was done using male mice. Second, we isolated macrophages from female SKM, as suggested by Reviewer 2, and studied them by flow cytometry. Although female SKM macrophages distributed into the same 4 subgroups and these were comparable in size to those of males, the LYVE1+/MHCIIhi subgroup was relatively smaller and the LYVE1-/MHCIIlo subgroup larger than those of males. We included these data in the supplement (Figure 4—figure supplement 1A,B) and briefly described them in the Results section. In light these unexpected findings, we have begun dedicated studies to understand the biological significance of the differences between male and female macrophages in SKM, and we hope to report them in the near future as a separate investigation.

2. When describing results from the flow cytometry-based phagocytosis assay (Figure 4), the authors find that LnHl macrophages are phagocytic to a lower proportion than other described subgroups (~49% compared to >85%), although the phagocytic macrophages show strikingly higher levels of phagocytosed cargo by MFI analysis. Although we agree that the significance of this is unclear with the current evidence, this suggests further heterogeneity in the LnHl group. It would thus be important to try to use unbiased SNN clustering of LnHl macrophages (not just of all macrophages as in Figure 6) and identify potential subpopulations that may explain this functional heterogeneity using the generated scRNAseq data (for instance as relating to phagocytosis-gene related mRNAs).

We appreciate the Reviewer’s helpful observations. We performed unsupervised clustering specifically for the LnHl subgroup (now renamed LYVE1-/MHCIIlo) and found 6 subclusters within this subgroup of macrophages. Among them, subclusters 0, 2 and 3 showed strong correlation with phagocytosis/endocytosis. Subcluster 1 showed only one endocytosis term with low negative p-value, and subclusters 4 and 5 did not show any terms related to phagocytosis or endocytosis. We believe that within the LYVE1-/MHCIIlo subgroup, subclusters 0, 2 and 3 represent strong phagocytotic macrophages, and subclusters 1, 4 and 5 represent weakly phagocytotic macrophages. We included the unsupervised clustering and GO annotation in Figure 5—figure supplement 2A,B, listed the genes related to phagocytosis/endocytosis in the corresponding subclusters in Extended File 4, and described these data in the Results section.

3. Based on experimental flow/description, it sounds like young and old samples may not have been processed in parallel, which may be problematic due to the known impact of batch effects in genomics. Can the authors clarify and explicitly discuss whether batching may be a problem?

We apologize for the poor description of our experiment in the original submission. We isolated macrophages from two young mice in the first week, from two old mice in the second week, and from one young mouse and one old mouse in the third week. Single-cell libraries were prepared on the same day that we performed each isolation. We have added text describing these details in the Results and the Materials and methods sections. In the revised version, we also provide the biological replicates in supplement (Figure 1—figure supplement 1, Figure 3—figure supplement 1, Figure 7—figure supplement 2) and have modified the text descriptions.

4. There needs to be additional provided information about some of the bioinformatic tools and/or analyses.

We appreciate these specific requests. We have addressed them as explained below.

a. Although the authors generally provide information about software/package versions or dates of access, some are missing (e.g. R, g:profiler). This needs to be updated for reproducibility.

Regarding the use of g:Profiler, we added more details to the information in Materials and methods section.

b. Although a GO analysis with g:profiler is described in the text and figures, the method is not described in the method section. Since the nature/use of background lists in functional enrichment analysis is crucial, the authors should clarify the list of genes used as background for enrichment analysis (ideally all detected/expressed genes), as well as the FDR threshold for considering a term significantly enriched. A supplementary table with all enriched terms would also be invaluable.

We appreciate the Reviewer’s request and fully agree that more information and clarification are needed. First, we have included a description of the GO analysis with g:profiler, including significance thresholds and user threshold, as mentioned above, in response to section (a) of this query. Second, we have included all those genes that are significantly differentially expressed in each cluster/subgroup used for GO annotation in Supplementary Files 1 through 6. Third, the GO annotation usually provides more than a hundred functional terms, many of which are general functions of macrophages shared by most macrophages, as mentioned in the revised Results section, and many represent similar functional terms. We tried to select significant but non-redundant functional terms from them and displayed them in the figures and text. In addition, as the program is frequently updated, we felt that providing the list of featured genes (rather than the very long list of enriched terms), in the Supplementary Files 1 through 6 would be more informative. Interested readers could then easily access g:profiler (freely available) and can reanalyze GO annotation with upgraded g:profiler versions in the future. If the Reviewer prefers that we proceed differently, we ask for his/her kind advice.

c. For long-term reproducibility, it would be important to either deposit all R scripts to a public repository such as github or provide it as a supplemental archive to accompany the manuscript.

We appreciate this suggestion and have provided the R scripts as a supplemental archive, uploaded with this submission (Supplementary File 7).

d. In the methods, the authors mention using Student's t-tests, but not tests to verify that data was normally distributed before use of the t-test. Please include the reference to any normality test used, or, if normality of data cannot be verified, please update to use non-parametric tests.

We appreciate these valuable comments. In the revised manuscript, we have used the ‘Shapiro-Wilk test’ program in GraphPad Prism 8 to test for normal distribution. We observed that all of our data/comparisons are normally distributed. Although we mention these results as ‘data not shown’, we have displayed the results for the benefit of the Reviewer (Author response image 1). Based on the normality test, we proceeded to perform parametric tests and used one-way ANOVA (Dunnett's multiple comparisons test) for Figure 5D, and two-way ANOVA (Šídák's multiple comparisons test) for Figures 5B, Figure 5—figure supplements 1B and 2B.

Author response image 1

5. Please provide catalog numbers for the antibodies used in the flow phagocytosis assay, as is needed for reproducibility (methods, page 20).

We appreciate this request. In the new Supplementary File 6, we list the antibodies (and catalog numbers) used for flow cytometric analysis and tissue immunofluorescence staining.

Reviewer #2 (Recommendations for the authors):

1) Pg5 the authors rely on a limited number of markers to determine the macrophage polarization status (M1 vs M2) of the different subsets of macrophages characterized in the study. Can the authors investigate more exclusive markers (PMID: 26699615) to determine if MHCII and Lyve macrophage subsets are indeed skewed to one state or the other?

We thank Reviewer 2 for bringing to our attention the study by Jablonski et al. (PLoS ONE, 2015). We compared our data with the data in this paper while preparing our initial manuscript. However, there were major differences in macrophage processing, as macrophages were cultured and treated in the PLoS ONE study but were not cultured or treated in our study, and also major differences in detection of the expressed transcriptome, as microarrays were used in the PLoS ONE study, RNA-seq analysis in ours. Consequently, the gene expression patterns in SKM macrophages in our analysis were quite different from those published in PLoS ONE, except for very typical M2 markers like Mrc1 mRNA and M1 markers like Il1b mRNA. We picked top M1 markers, such as Cd38, Fpr2, Gpr18, Hp, Cfb, Cxcl10, Ccr7, and Il6 mRNAs, and top M2 markers like Egr2, Myc, Cd83, Ptgs1, Flrt2, Mmp12, Fn1, Arg1, Chi3l3, and Renla mRNAs from the PLoS ONE paper, and compared them with our data. The above genes were expressed at very low levels in general in our data, and showed no clear skewed expression pattern in either supervised or unsupervised clustering. Thus, we felt that discussing this paper would not have been very informative. However, another paper that characterized SKM macrophage origin and subgroups (Wang et al., Proc. Natl. Acad. Sci. USA, 2020) with an experimental setup more similar to ours showed profiles more similar to those in our study; we discussed this paper in the revised manuscript.

2) Lp+ macrophages were described in the study to express a transcriptomic program characterized by M2-like gene program involved in wound repair and healing. Thus, beyond investigating the phagocytic capacity of the macrophage using labeled E. coli, can the authors test the uptake of apoptotic cells using an efferocytosis assay, which seems more relevant for wound repair.

We appreciate the Reviewer’s excellent suggestion. We carried out efferocytosis experiments, as advised, included the results in supplementary Figures S6A and B, and describe them in the Results and Methods sections. As shown, the efferocytosis capacity was less prominent than the phagocytosis capacity, but LYVE1-/MHCIIlo macrophages were still more capable of efferocytosis than other subgroups.

3) Pg7 "We found that MHCII mRNAs (encoding H2-Ab1, H2-Eb1) divided SKM macrophages into two groups, MHCII-high (Hh) and MHCII-low (Hl) in single-cell profiling analysis." Please provide a figure reference.

The Reviewer makes another great point. We have included UMAPs for H2-Eb1 and H2-Ab1 mRNAs in Figure 3—figure supplement 1A and B.

4) Pg 8, in the immunostaining are different subsets of macrophages localized to different sections of muscle tissues? i.e. more associated with endothelial cells, muscle cells or innervated areas?

We appreciate this important question and have addressed it by performing additional experiments and by demarking the muscle cell borders. First, we indicated the muscle fiber outlines in the revised Figure 4B; this addition helped to identify those macrophages closely located with muscle cells, but only in intermuscular regions, that is, the endomysium and perimysium areas. Second, additional immunofluorescence analysis revealed that the two LYVE1+ macrophage subgroups, LYVE1+/MHCIIlo and LYVE1+/MHCIIhi, but not LYVE1-/MHCIIhi, were localized close to blood vessels, similar to previous findings in lung interstitial macrophages (Chakarov et al., 2019; Figure 4C, top). Third, the macrophage subgroups LYVE1+/MHCIIlo, LYVE1+/MHCIIhi, and LYVE1-/MHCIIhi were detected around nerve fibers (Figure 4C, bottom). These findings are described in the Results section.

5) Can the authors provide more insight into both old and young skeletal muscle to determine if an absolute number of macrophages change during aging (for example via flow cytometry) and also to determine using transcriptomics if the subpopulations of macrophages in the study are resident vs non-resident macrophages.

We thank the Reviewer for these questions. The absolute number of live macrophages from SKM for flow cytometry analysis were comparable across young and old, male and female mice. We have included these data in Figure 6—figure supplement 1A–D. Old male mice showed a tendency to have slightly higher levels of macrophages compared to young mice, but the differences were not statistically significant (Figure 6—figure supplement 1A). We described this finding in the Results section. We were unable to unequivocally distinguish resident macrophages from non-resident macrophages, but we believe that our macrophages, selected using markers CD11b and F4/80, are mostly resident macrophages.

6) The study lacks analysis of skeletal muscle macrophages in female young and old mice, and would be more informative if they were included in the study to determine any sex differences. Perhaps qPCR or flow cytometry analysis of the four major subsets can be investigated in female mice, since performing single cell would be cost prohibitive.

We appreciate these helpful comments and suggestions. To address them, we carried out flow cytometric analysis of macrophages from female SKM. We found that female mice also have 4 macrophage subgroups, although the relative ratios were modestly different. We found a statistically significant decrease of SKM macrophage numbers in the LYVE1+/MHCIIhi subgroup, and an increase in the LYVE1-/MHCIIlo subgroup in female compared to male. We included these data (Figure 4—figure supplement 1A,B), and we explain that we have undertaken dedicated studies to understand the biological significance of the differences in macrophages from male vs female SKM.

7) Pg 11. "Cdk1 and Top2a mRNAs were expressed in an even lower number of macrophages (data not shown)." I am confused by data not shown since the data appears to be in table. If not please show.

We appreciate the careful reading by this Reviewer and apologize for the mistaken description. We have revised the text and indicate that the data are included in the revised Supplementary File 1, Cluster 9.

Reviewer #3 (Recommendations for the authors):

1) Manuscript provides no functional insights of how changes in macrophages affect muscle-physiology/pathology in young and aged mice. Conclusion about functions are based on associations.

We appreciate the Reviewer’s remarks. We agree that our characterization of macrophage subpopulations in resting skeletal muscle (SKM) and the changes in subpopulations with age are primarily descriptive. In the revised manuscript, we have toned down our interpretation of the results and have acknowledged that our conclusions are based on associations. In addition, given that we identified new SKM subpopulations and found that their relative abundance changes with age, we propose to expand on these findings. To investigate in depth the functional roles of these macrophage subgroups, we are developing specific genetic mouse models and setting up a number of necessary methodologies. The detailed analysis of these macrophage subgroups represents a long-term effort in our laboratory.

2) Please clarify what the percentage of macrophages in muscle in comparison to other CD45+ cells is.

We thank the Reviewer for this request. In the revised manuscript, Figure 6—figure supplement 1 includes the absolute number of CD45+, CD45+/CD11b+, and CD45+/CD11b+/F4/80+ cells, and the percentage of CD45+/CD11b+/F4/80+ cells in CD45+ cells in young and old mice, males and females, as quantified by flow cytometry.

3) Please clarify wheter the analyzed myeloid cells are present in vessels or in muscle parenchyma. How that affects muscle function is also unclear.

We thank the Reviewer for this important question. To address it experimentally, we have carried out additional immunofluorescence staining and have added more information in Figure 4. First, we have highlighted the muscle membranes in the revised Figure 4B to help visualize the location of macrophages, all of which were found in the intermuscular regions (i.e., endomysium or perimysium areas), and not in the muscle parenchyma. Second, additional immunofluorescence staining revealed that two LYVE1+ macrophage subgroups, LYVE1+/MHCIIlo and LYVE1+/MHCIIhi (but not LYVE1-/MHCIIhi), localized close to blood vessels, similar to the previous findings by Chakarov et al., (2019) for lung interstitial macrophages (Figure 4C, top). Third, all 3 detectable macrophage subgroups, LYVE1+/MHCIIlo, LYVE1+/MHCIIhi, and LYVE1-/MHCIIhi were found around nerve fibers (Figure 4C, bottom). We describe these findings in the Results section. As our studies move forward, we will explore the function of these subgroups and their impact on SKM physiology using appropriate methodologies and genetic mouse models.

4) While the scRNAseq of skeletal muscle macrophages reveals interesting findings about their diversity and how aging affects their transcriptomic profile, the authors chose an unusual approach to analyze the data by defining subsets based on extracellular marker expression (Lyve1 and MHCII) and not transcriptomic profile. By starting with a supervised approach, the authors have missed an important part of their data regarding the high diversity of skeletal muscle macrophages, that cannot only be described through MHC-II and Lyve1 expression. The expression of extracellular markers does not necessarily allow to define functionally distinct subsets. A deeper and more detailed unsupervised analysis is required for scRNA seq data. Moreover, as a starting point, the clusters should be defined based on gene expression and then propose candidate markers to characterize the subpopulations and their functional properties

The Reviewer makes excellent suggestions. We have incorporated his/her helpful advice in the revised manuscript by reanalyzing our scRNA-seq data with updated software, and we present first the unsupervised clustering followed by supervised classification. The unsupervised clustering identified 11 clusters; GO annotation found several clusters that were more reparative, proinflammatory, and phagocytotic, and other clusters that were more proliferative and related to senescence. The unsupervised clusters and featured genes are shown in the revised Figure 1, Table 1, and Supplementary File 1, and are described in the revised text. We then complemented the unsupervised clustering with supervised classification using traditional membrane markers that have been employed to functionally classify macrophages for many years. We agree that the revised approach (first unsupervised classification, and afterwards supervised classification) offers more logical, unbiased, and comprehensive information about SKM macrophages.

5) Several conclusions are not supported by the data – e.g.

a) Naming Lyve1+ and Lyve1- macrophages "healing" and pro-inflammatory" is premature as this has not been functionally tested.

We fully agree with the Reviewer and have toned down our conclusions across the manuscript, as advised. For example, we focus on the mRNAs expressed in each macrophage subgroup and simply mention the functions suggested by the corresponding GO annotations.

b) Unless the authors have data showing the general health of the animals used for the scRNAseq, the low number of differentially expressed genes between young and old shouldn't be interpreted as a result of a "healthier" status of the aging cohort. It is rather a common caveat of scRNAseq that lacks sequencing depth. A bulk RNAseq of specific sorted population that changes in aging may be important.

The Reviewer’s advice is spot on. We have modified our description of the scRNA-seq analysis to acknowledge that this type of analysis is shallow and may miss differences that require deeper RNA-seq analysis. Bulk RNA-seq analysis of specific subpopulations is a key component of our future plan, after we have optimized our protocol to harvest enough macrophages of each subgroup, and we have the necessary genetic mouse models in place.

6) Characterization of macrophages with polarization markers is inadequate and based on old literature. Hence, not very informative.

We fully agree with the Reviewer. Following the collective advice from Reviewers and Editors, we started the revised manuscript with unsupervised clustering data yielding 11 clusters of SKM macrophages (Figure 1, Table 1, and Supplementary File 1) To complement this classification, we undertook supervised classification by first assessing traditional markers like MRC1 (CD206) and CD86, but found that more recently identified markers LYVE1 and MHCII (Dick et al., 2022, Chakarov et al., 2019, Lim et al., 2018) were more informative in SKM macrophages. LYVE1 in particular divided SKM macrophages into two similarly sized groups (LYVE1+ and LYVE1-), which showed distinct functions. Therefore, we used LYVE1, along with the common membrane marker MHCII, for supervised classification to complement the unsupervised clustering.

7) The unsupervised analysis of the scRNAseq lacks depth. Each cluster should be described independently of whether they belong to LpHl, LpHh, LnHh and LnHl. This analysis should come as the first figure. For example, do not dismiss the possibility of non-macrophage clusters as cells were only sorted on CD11b marker. Cluster 6 could be a granulocyte cluster.

We appreciate the helpful suggestions from Reviewer 3. We have included more depth in our unsupervised clustering analysis and have moved these data to the beginning of the manuscript (Figure 1, Table 1, and Supplementary File 1), and we later discuss the complementarity of unsupervised and supervised analyses. Regarding the Reviewer’s point on the CD11b marker, we apologize for not explaining this point clearly. We isolated CD11b+ cells by FACS, and performed scRNA-seq analysis on them. But when analyzing the sequencing data, we sought an additional marker, F4/80, in order to be more confident about their identity as macrophages. Thus, we restricted our consideration of macrophages to only those CD11b+/F4/80+ double-positive cells and performed downstream statistical analysis on this group. We addressed the possible contamination by granulocytes in the revised Figure 1—figure supplement 1A; we found that Ly6G+ neutrophils and SiglecF+ eosinophils comprised less than 1% of the total CD11b+/F4/80+ double-positive population. Therefore, we believe that, overall, granulocyte contamination in our macrophage population is very minor.

8) Overall, the logic behind the direction taken for the analysis of the data is unclear:

– It is not clear why the authors chose to perform a scRNAseq. If the goals of the authors were to study macrophages subsets based on extracellular markers, sorting these subsets, and performing a bulk RNAseq would have been more adequate and would have provided and better sequencing depth.

We appreciate the Reviewer’s question. Given that the polarization of macrophages is largely tissue-dependent, and that the polarization of SKM macrophages with aging is poorly understood, our primary goals were to identify macrophage subpopulations and investigate how they changed with aging. Given the small number of these macrophages, scRNA-seq analysis has been the best way to answer this question at the moment. To follow better logic, as advised by the Reviewer, we displayed the unsupervised clustering first (Figure 1, Table 1, and Supplementary File 1) and complemented it with supervised analysis afterwards (Figures 2-4). We identified more than 10 subpopulations by unsupervised clustering and supervised classification, many of which were not previously reported. We further observed altered genes during aging, and identified specific subpopulations, particularly Cl6 and Cl8 clusters, and LYVE1+ and LYVE1- subgroups in old SKM.

As Reviewer 3 correctly points out, bulk RNA-seq analysis with sorted macrophage subgroups will be informative, especially because the sensitivity of scRNA-seq is low. We plan to expand this analysis as our studies progress and we are able to collect sufficient numbers of macrophages and have appropriate mouse models in hand to address these questions.

– The choice of MHC-II and Lyve1 markers to divide macrophage subset seems arbitrary. It is not clear why the authors chose to use markers described for lung macrophages specifically.

We appreciate this question. We investigated LYVE1 and MHCII because they effectively classified many tissue-resident macrophages in recent published reports – not only in lung, but also in artery, heart, brain, kidney, and liver (Lim et al., 2018, Chakarov et al., 2019, Dick et al., 2022). Empirically, we observed that LYVE1 (but not other markers examined) divided SKM macrophages into two similarly sized groups that showed distinct functions in association analyses in SKM. Traditional polarization markers, e.g., MRC1 (CD206) and CD86 were expressed in the vast majority of our SKM macrophages, and most macrophages were positive for both as shown in Figure 1—figure supplement 1C [similar expression patterns were shown by Kosmac et al. (2018) in human SKM macrophages]. These observations indicated that traditional polarization markers may not be suitable for classification of SKM macrophages specifically. Therefore, for SKM macrophages, we used LYVE1 and MHCII as membrane polarization markers.

– The data showing phagocytic capacities are interesting. Though, it is not clear why the authors chose to investigate this function. Is phagocytosis relevant to the physiological function of skeletal muscle tissues? Are there other functions highlighted in the GO annotations that can be tested? (endocytosis, inflammatory response, antigen presentation)

The Reviewer brings up important questions. After the scRNA-seq analysis revealed an interesting heterogeneity and polarization status of SKM macrophages, we sought to study the biological function of the macrophage subgroups. We focused on phagocytosis because it is a key function of macrophages during the repair and defense of injured SKM. However, prompted by the advice of Reviewer 2 and questions from Reviewer 3, we carried out experiments that focused on efferocytosis (the capacity of macrophages to engulf apoptotic cells). Efferocytotic capacities were generally less robust than phagocytic capacities (Figure 5—figure supplement 1A,B), but here too, efferocytosis was more vigorous in the LYVE1-/MHCIIlo subgroup than the other 3 subgroups, similar to the findings with phagocytosis. We included the efferocytosis data in Figure 5—figure supplement.

9) It is difficult to evaluate the solidity of the data. For quantification of populations by flow cytometry analysis and using the scRNAseq data, we suggest adding histograms representing n numbers and statistical significance when comparing the abundance of populations (Figure 3D, S3C, 5A, 6A). Moreover, instead of using tables to depict differentially expressed genes, heatmaps or volcanoplots are recommended (Figure 5C, S3C).

We appreciate the Referee’s helpful advice. We agree that the suggested representations are far more intuitive. In the revised manuscript, we display the data using histograms, heatmaps, UMAPs, violin plots, etc., as shown in the revised Figures 1E, 4A, 6C, 7A-D, and Figure 1—figure supplement 2B,C, Figure 5—figure supplements 1 and 2, Figure 6—figure supplement 1, Figure 7—figure supplements 1 and 2.

Figure 1C and Figure 2: Ln cells appear to be multiple clusters – with same features? How is that possible?

We thank the Reviewer for his/her keen eye. Unsupervised clustering revealed that LYVE- cells (Ln by the old nomenclature) comprised ~7 smaller clusters, Cl3-Cl9 (Table 1). By contrast, unsupervised clustering of LYVE1+ yielded only two subgroups (Cl0 and Cl1). As discussed in the Results section, these findings suggest that LYVE1- (M1-like) macrophages appear to be more diverse than LYVE1+ (M2-like) macrophages. In keeping with this notion, M2 markers were quite robust (e.g., LYVE1, MRC1, CD163 etc.), while M1 markers were generally less useful. In this regard, providing gene expression details of the unsupervised and supervised subgroups helps readers understand the heterogeneity of SKM macrophages.

Figure 1D: The bioinformatic analyses used to characterize Lp/Ln cells lacks statistical validity.

We thank the reviewer for this comment. We removed the old Figure 1D, and merged those data with data that are now displayed in the revised Figures 2B,C. We also listed featured genes in Supplementary File 2, so that future readers that can easily access the most current version of g:Profiler to find GO annotations.

Figure 3D: biological purpose to reveal new insights is lacking.

We appreciate the request to elaborate on the biological purpose (note: the original Figure 3D is now Figure 4A). Identifying two new subgroups, LYVE1-/MHCIIlo and LYVE1+/MHCIIhi, provided important insights that we plan to expand into future studies. The LYVE1-/MHCIIlo subgroup formed a separate cell cluster, as shown in the new Figure 4A, and GO annotation further suggested a cell killing function of this subgroup (Figure 3C); interestingly, the size of this subgroup increased in old SKM (Figure 7A,B). The possible involvement of this subgroup in cleaning up senescent cells in aging SKM is an exciting future direction for our team.

The LYVE1+/MHCIIhi subgroup showed an ‘intermediate’ status by flow cytometric analysis. This subgroup may change phenotypes depending on environmental cues and is another important subgroup that we will investigate as our studies advance. In sum, our study offers important insight into the function of macrophage subpopulations in SKM aging.

Figure 3E; Fails to provide spatial clarity of localization in muscle.

In the revised Figure 4B, we marked the contour of muscle cells, so that readers can recognize more easily that the macrophages appear to be located in the intermuscular perimysium and endomysium areas. In Figure 4C, we have included additional immunofluorescence analysis showing that LYVE1+ macrophages are closely localized with blood vessels, while both LYVE1+ and LYVE1- macrophages are located near nerve fibers.

Figure 4: phagocytic analyses are superficial and fail to provide new insights.

Thank you for the opportunity to expand on the phagocytic analysis and discuss the results. We anticipated seeing strong phagocytic activity across the macrophage subgroups, but it was unexpected to find the particularly strong phagocytotic capacity of LYVE1-/MHCIIlo macrophages. The same subgroup also showed stronger efferocytotic capacity in an additional experiment that was carried out during the revision (Figure 5—figure supplement 1). These results offer strong evidence that the function of LYVE1-/MHCIIlo macrophages could include ‘cleaning-up’ during SKM aging, an exciting trait that we plan to pursue with appropriate mouse models.

Figure 6: is the most important figure. However, it is poorly put together in terms of data analyses and presentation. Identity of cluster 5 and 6 which change with age is the main finding – but their relevance is unknown

We appreciate the Reviewer’s comment. Most of the content in the original Figure 6 is now included in Figure 7. The new Figure 7 highlights the changes in macrophage numbers and expression levels of key genes in clusters during aging. As the Reviewer notes, changes in Cl5 and Cl6 (now Cl6 and Cl8) during aging were more pronounced. Many senescence-associated mRNAs including Gpnmb, Spp1, Ctsd mRNAs were found in the new Cl6, and S100a8 and S100a9 mRNAs, encoding non-classical proinflammatory markers, were enriched in the new Cl8. The elevated expression of these and other mRNAs in these two clusters in old SKM suggests that they are involved in SKM aging. As the Reviewer points out, their relevance is not known at present, and further studies are needed to characterize them fully.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Upon discussion, the reviewers thought that this could be addressed with:

1. Doing an analysis paralleling the main manuscript but using only the 2 samples that were processed "unbatched" (i.e. the samples where 1 young and 1 old mouse were processed in parallel), and only these. If the main results of the study are conserved in this unbatched subset of the data, this would strengthen the likelihood that the batching did not grossly impact the conclusions. We would then recommend including this analysis as a supplement.

We sincerely appreciate the helpful guidance from Reviewers and Editors. We shared their concern about possible batch effects, as the lengthy protocol for isolating macrophages from mouse skeletal muscle allowed us to process only one mouse at a time, and two mice total in one day by the same persons. Following the requested additional analyses, we gained evidence to support the notion that batch effects did not grossly impact the conclusions. We have provided the following additional analyses in this revision:

1. In the new Figure 1—figure supplement 1B, we now show the distribution of macrophages in the 11 clusters across all replicates individually (Y1, Y2, Y3, O1, O2, O3) and the combined replicates (Y1-3, O1-3) both young and old. We hope these side-by-side comparisons help the readers appreciate the similarities between batches/replicates.

2. In the new Figure 1—figure supplement 1C, we have included UMAP analyses of all biological replicates for both young and old. Again, these analyses can help the readers evaluate the similarities in gene expression patterns across batches/replicates.

3. In the new Figure 3—figure supplement 1B, we now display separately the biological replicates of macrophage distribution in four subgroups following a supervised classification. We show each individual sample (Y1, Y2, Y3, O1, O2, O3) and the combined samples (Y1-3, O1-3). In Figure 3—figure supplements 1B and 1C, the readers can see side-by-side the replicates in the supervised classification analysis, for both cell distribution and gene expression patterns.

4. In the revised Materials and methods section, we explicitly indicate that the isolated SKM macrophages were immediately subjected to single-cell library construction without culture or treatment. Given the methodological hurdles with the lengthy protocol, we believe that this strategy helped to maintain the in vivo characteristics and reduce batch-related problems.

It is worth noting the consistency in expression of other specific genes across replicates in young and old macrophages (Figure 7—figure supplement 2). Similarly, flow cytometric analysis using antibodies against LYVE1 and MHCII showed patterns very similar to those observed by the supervised classification of single-cell transcriptomics (percentages in Figure 7A,B). In sum, any potential batch effects by the collection scheme did not influence the results, and the main conclusions remain unchanged.

2. As highlighted by reviewer #3, it is crucial that the batching/experimental collection scheme be discussed explicitly in the manuscript.

We fully agree with this important request. In the revised Results, when describing Figures 1 and 3 (pages 5 and 9), we have discussed the additional analysis carried out to alleviate concerns related to batch effects.

3. Finally, please address Reviewer #1's remaining concern on the use of a background gene list for the g:profiler analysis.

We appreciate the request for additional analysis requested by Reviewer #1 and have addressed it by using genes detected from our SKM macrophages as background gene set, as requested. Our response to the reviewer is as follows:

“We appreciate the Reviewer’s following up on this request and apologize for previously misunderstanding this question, as indeed we had mistakenly used all mouse genes as the background in our earlier analysis. Following the Reviewer’s suggestion, we now used instead the 14,542 genes detected in our young and old skeletal muscle macrophages (CD11b+/F4/80+) as the background gene set. We included this information in the revised Materials and methods section (page 20). We ran again all GO annotations for this revision and obtained slightly different functional terms with the new background gene set. For example, we found that only SubCl0 in the LYVE1-/MHCIIlo subgroup shows phagocytosis-related terms in the new analysis (Figure 5—figure supplement 2). We included the new analysis as a table in Figure 5—figure supplement 2, and we removed the Supplementary File 4 we had included in the previous submission. We also updated Figures 2B, 2C, 3C, and 6B, as well as Table 1, the Results section, and Figure Legends section.”

Reviewer #1 (Recommendations for the authors):

Although the authors have addressed most of my concerns, some large concerns remain at this point.

1. A very large concern was revealed by their answer to one of my questions about the batchiness of the data. Indeed, the author's response revealed 3 batched: (i) only young samples, (ii) only old samples and (ii) one old and one young. Unfortunately, since batch and biological groups are confounded for groups i and ii, that data is meaningless (i.e. batch cannot be properly accounted for when it is confounded with biology). Since I understand that the authors may not be able to redo the entire experiment the way it should have been done, I believe it is imperative that all analyses also be done exclusively on batch 3 (the one where both groups were represented), to show that all results would hold in the absence of batch. The results should then be included and compared/discussed in the context of the paper as this is a big problem.

We appreciate the helpful guidance from the Reviewer. We shared his/her concern about possible batch effects, stemming from the fact that the lengthy isolation of macrophages from mouse skeletal muscle allowed us to process only one mouse at a time, and two mice total in one day by the same persons. By performing the requested additional analyses, we gained evidence supporting the notion that batch effects did not grossly impact the conclusions. We have provided the following additional analyses in this revision:

1. In the new Figure 1—figure supplement 1B, we now show the distribution of macrophages in the 11 clusters across all replicates individually (Y1, Y2, Y3, O1, O2, O3) and the combined replicates (Y1-3, O1-3) both young and old. We hope these side-by-side comparisons help the readers appreciate the similarities between batches/replicates.

2. In the new Figure 1—figure supplement 1C, we have included UMAP analyses of all biological replicates for both young and old. Again, these analyses can help the readers evaluate the similarities in gene expression patterns across batches/replicates.

3. In the new Figure 3—figure supplement 1B, we now display separately the biological replicates of macrophage distribution in four subgroups following a supervised classification. We show each individual sample (Y1, Y2, Y3, O1, O2, O3) and the combined samples (Y1-3, O1-3). In Figure 3—figure supplements 1B and 1C, the readers can see side-by-side the replicates in the supervised classification analysis, for both cell distribution and gene expression patterns.

4. In the revised Materials and methods section, we explicitly indicate that the isolated SKM macrophages were immediately subjected to single-cell library construction without culture or treatment. Given the methodological hurdles with the lengthy protocol, we believe that this strategy helped to maintain the in vivo characteristics and reduce batch-related problems.

It is worth noting the consistency in expression of other specific genes across replicates in young and old macrophages (Figure 7—figure supplement 2). Similarly, flow cytometric analysis using antibodies against LYVE1 and MHCII showed patterns very similar to those observed by the supervised classification of single-cell transcriptomics (percentages in Figure 7A,B). In sum, any potential batch effects by the collection scheme did not influence the results, and the main conclusions remain unchanged.

2. The authors still did not address the background list used for GO enrichment in g:profiler. This leads me to believe they used the default (all genes instead of detected genes in the dataset), which is incorrect and would lead to spurious enrichments. These analyses should be rerun with the correct background list.

As mentioned above in our response to the Editors, we appreciate the Reviewer’s following up on this request and apologize for previously misunderstanding this question, as indeed we had mistakenly used all mouse genes as the background in our earlier analysis. Following the Reviewer’s suggestion, we now used instead the 14,542 genes detected in our young and old skeletal muscle macrophages (CD11b+/F4/80+) as the background gene set. We included this information in the revised Materials and methods section (page 20). We ran again all GO annotations for this revision and obtained slightly different functional terms with the new background gene set. For example, we found that only SubCl0 in the LYVE1-/MHCIIlo subgroup shows phagocytosis-related terms in the new analysis (Figure 5—figure supplement 2). We included the new analysis as a table in Figure 5—figure supplement 2, and we removed the Supplementary File 4 we had included in the previous submission. We also updated Figures 2B, 2C, 3C, and 6B, as well as Table 1, the Results section, and Figure Legends section.

Reviewer #2 (Recommendations for the authors):

We believe the authors have responded to the concerns of the reviewers sufficiently and the paper is significantly improved. Thus, in our opinion, the paper is suitable for publication.

We appreciate the positive comments from Reviewer #2.

Reviewer #3 (Recommendations for the authors):

The authors have addressed most of my prior concerns. Some issues remain, but in general given the importance of the topic, the mansucript is ready to forward in the process.

An important issue that remains unaddressed is that the scRNA analyses and cell sorting for young/old groups were done on different days. The authors responded to this issue and acknowledged this caveat, but do not describe the consequence of this on the data generation and conclusions.

Authors need to provide this information. This reviewer could not find it in the beginning of the Results section where the description of data generation is provided.

We appreciate the helpful comments from the Reviewer. Following the advice and suggestions from the Editors and Reviewers, we have incorporated analysis that explicitly addresses if possible batch effects modify the overall conclusions. As we explain above in our responses to the Editor and Reviewer 1, the lengthy protocol for isolating macrophages from mouse skeletal muscle made it impossible for us to process more than one mouse at a time, and only two mice total could be processed per day by the same persons. Therefore, we have done the requested additional analyses to ensure that such effects did not grossly impact the conclusions. We have provided the following additional analyses in this revision:

1. In the new Figure 1—figure supplement 1B, we now show the distribution of macrophages in the 11 clusters across all replicates individually (Y1, Y2, Y3, O1, O2, O3) and the combined replicates (Y1-3, O1-3) both young and old. We hope these side-by-side comparisons help the readers appreciate the similarities between batches/replicates.

2. In the new Figure 1—figure supplement 1C, we have included UMAP analyses of all biological replicates for both young and old. Again, these analyses can help the readers evaluate the similarities in gene expression patterns across batches/replicates.

3. In the new Figure 3—figure supplement 1B, we now display separately the biological replicates of macrophage distribution in four subgroups following a supervised classification. We show each individual sample (Y1, Y2, Y3, O1, O2, O3) and the combined samples (Y1-3, O1-3). In Figure 3—figure supplements 1B and 1C, the readers can see side-by-side the replicates in the supervised classification analysis, for both cell distribution and gene expression patterns.

4. In the revised Materials and methods section, we explicitly indicate that the isolated SKM macrophages were immediately subjected to single-cell library construction without culture or treatment. Given the methodological hurdles with the lengthy protocol, we believe that this strategy helped to maintain the in vivo characteristics and reduce batch-related problems.

It is worth noting the consistency in expression of other specific genes across replicates in young and old macrophages (Figure 7—figure supplement 2). Similarly, flow cytometric analysis using antibodies against LYVE1 and MHCII showed patterns very similar to those observed by the supervised classification of single-cell transcriptomics (percentages in Figure 7A,B). In sum, any potential batch effects by the collection scheme did not influence the results, and the main conclusions remain unchanged.

https://doi.org/10.7554/eLife.77974.sa2

Article and author information

Author details

  1. Linda K Krasniewski

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Conceptualization, Investigation, Methodology, Writing – original draft
    Competing interests
    No competing interests declared
    Additional information
    Co-first authors
  2. Papiya Chakraborty

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Conceptualization, Investigation, Methodology, Writing – original draft
    Competing interests
    No competing interests declared
    Additional information
    Co-first authors
  3. Chang-Yi Cui

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    cuic@grc.nia.nih.gov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9856-9427
  4. Krystyna Mazan-Mamczarz

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Christopher Dunn

    Flow Cytometry Core, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Formal analysis, Investigation, Visualization
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7899-0110
  6. Yulan Piao

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  7. Jinshui Fan

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  8. Changyou Shi

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Supervision, Investigation, Methodology
    Competing interests
    No competing interests declared
  9. Tonya Wallace

    Flow Cytometry Core, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Resources, Investigation
    Competing interests
    No competing interests declared
  10. Cuong Nguyen

    Flow Cytometry Core, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Resources, Investigation
    Competing interests
    No competing interests declared
  11. Isabelle A Rathbun

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  12. Rachel Munk

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Resources, Investigation, Methodology
    Competing interests
    No competing interests declared
  13. Dimitrios Tsitsipatis

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Formal analysis, Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  14. Supriyo De

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Resources, Software, Supervision
    Competing interests
    No competing interests declared
  15. Payel Sen

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Supervision, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2809-0901
  16. Luigi Ferrucci

    Translational Gerontology Branch, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Resources, Supervision
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6273-1613
  17. Myriam Gorospe

    Laboratory of Genetics and Genomics, National Institute on Aging Intramural Research Program, National Institutes of Health, Baltimore, United States
    Contribution
    Conceptualization, Investigation, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    myriam-gorospe@nih.gov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5439-3434

Funding

National Institutes of Health (Z01-AG000511)

  • Linda K Krasniewski

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the Intramural Research Program of the National Institute on Aging. The authors thank Marc Michel for technical assistance.

Ethics

All animal study protocols were approved by the NIA Institutional Review Board (Animal Care and Use Committee). (ASP #476-LGG-2024).

Senior Editor

  1. Carlos Isales, Augusta University, United States

Reviewing Editor

  1. Bérénice A Benayoun, University of Southern California, United States

Publication history

  1. Received: February 17, 2022
  2. Preprint posted: February 24, 2022 (view preprint)
  3. Accepted: October 10, 2022
  4. Accepted Manuscript published: October 19, 2022 (version 1)
  5. Version of Record published: November 2, 2022 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 867
    Page views
  • 186
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Linda K Krasniewski
  2. Papiya Chakraborty
  3. Chang-Yi Cui
  4. Krystyna Mazan-Mamczarz
  5. Christopher Dunn
  6. Yulan Piao
  7. Jinshui Fan
  8. Changyou Shi
  9. Tonya Wallace
  10. Cuong Nguyen
  11. Isabelle A Rathbun
  12. Rachel Munk
  13. Dimitrios Tsitsipatis
  14. Supriyo De
  15. Payel Sen
  16. Luigi Ferrucci
  17. Myriam Gorospe
(2022)
Single-cell analysis of skeletal muscle macrophages reveals age-associated functional subpopulations
eLife 11:e77974.
https://doi.org/10.7554/eLife.77974
  1. Further reading

Further reading

    1. Cell Biology
    Joris P Nassal, Fiona H Murphy ... Matthijs Verhage
    Research Article

    Different organelles traveling through neurons exhibit distinct properties in vitro, but this has not been investigated in the intact mammalian brain. We established simultaneous dual color two-photon microscopy to visualize the trafficking of Neuropeptide Y (NPY)-, LAMP1-, and RAB7-tagged organelles in thalamocortical axons imaged in mouse cortex in vivo. This revealed that LAMP1- and RAB7-tagged organelles move significantly faster than NPY-tagged organelles in both anterograde and retrograde direction. NPY traveled more selectively in anterograde direction than LAMP1 and RAB7. By using a synapse marker and a calcium sensor, we further investigated the transport dynamics of NPY-tagged organelles. We found that these organelles slow down and pause at synapses. In contrast to previous in vitro studies, a significant increase of transport speed was observed after spontaneous activity and elevated calcium levels in vivo as well as electrically stimulated activity in acute brain slices. Together, we show a remarkable diversity in speeds and properties of three axonal organelle marker in vivo that differ from properties previously observed in vitro.

    1. Cell Biology
    2. Neuroscience
    Ge Gao, Shuyu Guo ... Gang Peng
    Research Article Updated

    Unbiased genetic screens implicated a number of uncharacterized genes in hearing loss, suggesting some biological processes required for auditory function remain unexplored. Loss of Kiaa1024L/Minar2, a previously understudied gene, caused deafness in mice, but how it functioned in the hearing was unclear. Here, we show that disruption of kiaa1024L/minar2 causes hearing loss in the zebrafish. Defects in mechanotransduction, longer and thinner hair bundles, and enlarged apical lysosomes in hair cells are observed in the kiaa1024L/minar2 mutant. In cultured cells, Kiaa1024L/Minar2 is mainly localized to lysosomes, and its overexpression recruits cholesterol and increases cholesterol labeling. Strikingly, cholesterol is highly enriched in the hair bundle membrane, and loss of kiaa1024L/minar2 reduces cholesterol localization to the hair bundles. Lowering cholesterol levels aggravates, while increasing cholesterol levels rescues the hair cell defects in the kiaa1024L/minar2 mutant. Therefore, cholesterol plays an essential role in hair bundles, and Kiaa1024L/Minar2 regulates cholesterol distribution and homeostasis to ensure normal hearing.