A destabilised metabolic niche provokes loss of a subpopulation of aged muscle stem cells

Ageing is a multi‐factorial condition that results in a gradual decline in tissue and organ function. Systemic, local and intrinsic factors play major roles in this process that also results in a decline in stem cell number and function. In this issue of The EMBO Journal, Li et al (2019) show that a subpopulation of mouse muscle stem cells is depleted in aged mice through loss of niche‐derived granulocyte colony‐stimulating factor (G‐CSF).

A new study reveals that Granulocyte‐Colony Stimulating Factor secreted from muscle fibers of aged mice safeguards satellite cell integrity.

Skeletal muscle satellite (stem) cells (MuSCs) undergo progressive loss of regenerative potential during ageing, and in extremely aged mice, they enter a pre‐senescent state where several pathways (Evano & Tajbakhsh, 2018) including autophagy (Garcia‐Prat et al, 2016) and JAK/STAT signalling (Price et al, 2014; Tierney et al, 2014) are perturbed.

Although the transcription factor Pax7 marks the MuSC population that is quiescent during homeostasis, Pax7^Hi and Pax7^Lo expressing cells (GFP intensity from transgenic Tg:Pax7‐nGFP mice) were shown to have distinct properties. Pax7^Hi cells are in a deeper state of quiescence (dormant state), which is characterised by a longer lag‐time to execute the first mitosis and lower metabolic activity, compared to Pax7^Lo cells (Rocheteau et al, 2012). Therefore, the quiescent state exists as a continuum from Pax7^Hi (dormant) to Pax7^Lo (poised) cells (Fig 1A). These findings suggest that Pax7^Hi cells might be solicited less frequently to effect routine regeneration of damaged fibres owing to their dormant state, although this has not been formally proven. Comparative studies of stem cell properties between young and aged mice are sometimes confounded by the observation that aged MuSCs appear to break quiescence more frequently (Evano & Tajbakhsh, 2018). Examination of Pax7^Hi cells might reduce this variability if they have a lower propensity to activate spontaneously. In a recent study, it was shown that even aged Pax7^Hi MuSCs exhibit a global increase in uncoordinated transcriptional heterogeneity compared to those in young mice (Hernando‐Herraez et al, 2019). The identification of distinct quiescent cell states raises the question of plasticity and determinism within the stem cell pool—do all MuSCs have equivalent potential, or are some (e.g. Pax7^Hi) more stem‐like (see Fig 1A)?

Figure 1. Scheme depicting heterogeneities within the muscle stem cell pool and response to ageing.

(A) Muscle stem cells form a continuum between a dormant and poised state within the quiescent state. Model for deterministic and stochastic selection of subpopulations of MuSCs that will self‐renew during regeneration is indicated. (B) G‐CSF is a glycoprotein that stimulates bone marrow to produce granulocytes and stem cells and release them into the bloodstream. This molecule is secreted by myofibres, and it is depleted during ageing. G‐CSF plays a role in maintaining stemness features (Pax7^Hi) of MuSCs and asymmetric cell divisions. Some of these features are partially restored to the youthful state following exercise.

In a recent study, Li et al (2019) extended these observations by isolating Pax7^Hi and Pax7^Lo cells from Tibialis anterior (TA) limb muscle (based on GFP intensity from Tg:Pax7‐nGFP mice) and performing single‐cell RNAseq analysis. They showed that distinct transcriptional profiles characterise Pax7^Hi and Pax7^Lo subpopulations, with mitochondrial genes being enriched in Pax7^Hi cells. They then examined FACS profiles of MuSCs and noticed a diminished number of Pax7^Hi cells in old compared to young mice. Accordingly, lower levels of Pax7^Hi marker genes—the stemness‐related gene CD34—and higher levels of myogenic commitment genes (MyoD and MyoG) were noted in aged TA muscles compared to the young, suggesting that the Pax7^Hi state was being impoverished during ageing. Whether depletion of this cell subpopulation compromises muscle function depends on the plasticity of other subpopulations to replenish the former.

To identify the mechanism responsible for this loss, the authors then turned to factors secreted by the niche, specifically the muscle fibres. Given that different fibre types exhibit different metabolic activities, the authors asked whether MuSCs assume properties of their associated fibres, and if these properties are altered during ageing, as reported for some metabolic indicators (Pala et al, 2018). Indeed, more Pax7^Hi cells were observed in TA (predominantly glycolytic) compared to Soleus (mainly oxidative) muscle fibres from the same Tg:Pax7‐nGFP mouse. Accordingly, stemness markers and genes enriched in the Pax7^Hi subpopulation were elevated in TA‐derived satellite cells. In contrast, a smaller number of Pax7^Hi cells was observed in PPARβ transgenic mice, which are enriched in oxidative muscle fibres and have reduced numbers of glycolytic muscle fibres.

Subsequent RNAseq analysis of whole TA muscle from old and young mice revealed metabolic alterations in gene expression that correspond to a shift from a glycolytic to an oxidative state. Consistently, the Pax7^Hi subpopulation and the gene expression profile corresponding to this cell state were significantly reduced in TA muscle of old mice compared to the young. Examination of secreted factors that are downregulated in old versus young TA muscle pointed to several candidates, including granulocyte colony‐stimulating factor (Csf3; G‐CSF) as a candidate niche regulator. Interestingly, Csf3 was more highly expressed by the glycolytic TA compared to oxidative Soleus muscle, and reduced in TA muscle of PPARβ transgenic mice. In a series of experiments, including supplementing cultured satellite cells with G‐CSF, examination of Csf3r ^−/− (Csf3 receptor), MyoD null mice and ChIP assays, the authors concluded that G‐CSF is secreted by myofibres and regulated by MyoD. Further, its receptor is expressed by MuSCs and this pathway is required to maintain the Pax7^Hi subpopulation.

Exercise significantly increases glycolytic activity in TA muscles of aged mice compared to those of sedentary aged mice. Therefore, the authors subjected mice to an exercise regime and observed that the percentage of Pax7^Hi cells dramatically increased in TA muscles of exercised aged mice compared to sedentary aged mice, and almost to the levels seen in untrained young mice. In keeping with a role for G‐CSF in this process, restoration of Pax7^Hi cells was not observed in Csf3r ^−/− mice under these conditions.

Collectively, these experiments point to plasticity among MuSCs where they respond differentially to ageing, exhibit metabolic activities reflecting that of their associated muscle fibre, and respond to the secreted niche factor G‐CSF. However, it remained unclear why only a subset of MuSCs were responsive. To address this point, the authors followed in the footsteps of previous reports showing that asymmetric cell divisions (ACD) of MuSCs impact their self‐renewal (Evano & Tajbakhsh, 2018) and that G‐CSF is distributed asymmetrically in a subpopulation of MuSCs (Hayashiji et al, 2015). Specifically, they followed the asymmetric segregation of old and new DNA strands between daughter cells during muscle regeneration (Rocheteau et al, 2012) and noted that daughter cells inheriting new DNA strands had lower expression of Pax7 and higher levels of G‐CSFR. Addition of G‐CSF to cultured myofibres during the first cell divisions resulted in a slight increase in ACD events, but only on Extensor digitalis longus (EDL, glycolytic) and not oxidative Soleus fibres. These findings suggest a link between ACD and responder MuSCs that preferentially express G‐CSFR.

The authors then asked how the Pax7^Hi subpopulation was replenished, i.e. from the Pax7^Mid or Pax7^Lo pool. Single‐cell RNAseq of Pax7^Mid cells isolated during regeneration showed a transcriptome profile resembling that of the Pax7^Hi fraction. In keeping with their relatedness, asymmetric distribution of G‐CSFR was more prominent in the Pax7^Mid subpopulation compared to Pax7^Lo cells. Furthermore, addition of G‐CSF to cultured Pax7^Hi and Pax7^Mid cells tended to yield higher numbers of Pax7^Hi cells and more ACD events compared to Pax7^Lo cells. G‐CSF acts through STAT3 signalling, and treatment of the Pax7‐nGFP subpopulations with a Stat3 inhibitor blocked the increase in the number of Pax7^Hi cells.

Taken together, this study shows that the niche‐secreted factor G‐CSF, which is expressed preferentially by glycolytic muscle fibres, regulates MuSC stemness properties (Fig 1B). These effects are more accentuated on the Pax7^Hi and Pax7^Mid cells, and G‐CSF appears to increase the frequency of ACDs in these subpopulations. These interesting findings raise other questions. Notably, how this secreted factor impacts on the regenerative and engraftment properties of MuSCs remains to be determined. Whether G‐CSF is secreted by other cell types in muscle during homeostasis, injury or ageing, and when during ageing the depletion of G‐CSF occurs remain open questions. It was recently reported that MuSCs in aged mice uptake the thymidine analogue BrdU less frequently during homeostasis compared to those in young mice suggesting that they are in a deeper state of quiescence (Hernando‐Herraez et al, 2019). This observation contrasts with the preferential loss of Pax7^Hi (dormant) cells reported by Li et al (2019), suggesting that other factors are in play. The link between the frequency of ACD and ageing to replenish the stem cell pool is intriguing, and it needs further exploration, notably during chronic injury such as a myopathy, where ACD was reported to be compromised (Dumont et al, 2015). Interstitial cells also play a central role in MuSC function (Evano & Tajbakhsh, 2018). Whether ageing or fibre type differentially affect interstitial cells and their interactions with subpopulations of MuSCs is another interesting avenue to explore, now that G‐CSF is added to the increasing list of molecules that contribute to cell and organismal ageing.

The EMBO Journal (2019) 38: e103924