A family business: stem cell progeny join the niche to regulate homeostasis

Structure of the ‘hub’

The D. melanogaster testis is an elongated tubular structure that is closed at the apical end and connected to the rest of the genital tracks at the basal side. The ‘hub’, which is composed of approximately 10–15 somatic cell clusters, resides at the apical end. Germline stem cells (GSCs) and somatic cyst stem cells (CySCs) connect to the hub through adherens junctions. Approximately 8–10 GSCs and twice as many CySCs encase the hub²⁸ (FIG. 2a).

GSCs divide asymmetrically to produce one daughter cell that maintains stem cell identity and another, the gonialblast, that initiates a differentiation programme. Gonialblasts divide synchronously with incomplete cytokinesis to form 16 interconnected germ cell cysts that will terminally differentiate and become mature sperm. CySCs also divide asymmetrically, producing cyst cells that enclose the differentiating germ cells. It is thought that cyst cells provide essential signals to promote germ cell differentiation through Epidermal growth factor (EGF) signalling²⁹.

Signalling events in the hub

The hub has long been established as a critical niche component for both types of stem cells. The hub secretes an interleukin-like cytokine called Unpaired (UPD), which activates Janus kinase (JAK) signalling in both GSCs and CySCs. The activated downstream effector, phosphorylated Signal transducer and activator of transcription (STAT), is necessary and sufficient for CySC self-renewal^18,19. Two downstream targets of JAK–STAT were identified in CySCs: the zinc-finger transcription repressor zfh1 (zinc-finger homeodomain 1)³⁰ and the BTB (BR-C, TTK and BAB) domain-containing nuclear factor chinmo (chronologically inappropriate morphogenesis)³¹. Both are necessary and sufficient to promote self-renewal of CySCs and to maintain their undifferentiated status. By contrast, JAK–STAT activation in GSCs is required for their anchorage to the hub but not for their self-renewal. Instead, GSC self-renewal relies on sustained signalling through Bone morphogenic protein (BMP)-like ligands Decapentaplegic (DPP) and Glass bottom boat (GBB), which are secreted from both the hub and the CySCs^32–34. The outcome is transcriptional repression of a differentiation factor, bag of marbles (bam), resulting in the suppression of GSC differentiation into gonialblasts³⁴ (FIG. 2b). The interactions between GSCs and CySCs and their interdependencies are intriguing paradigms for studying the coordinated behaviours of multiple stem cell types within a shared niche.

CySCs and the hub: more alike than different

It was assumed that, when fully developed, the hub remains static and permanent. However, recent experiments suggest that, under some circumstances, CySCs might contribute to the mature hub³⁵. The rationale behind this work was simple: if cells outside the hub are marked by either bromo-deoxyuridine (BrdU) or transgenes, then, following a chase period, markers will be observed within the hub only if its residents are dynamic rather than static. To test this possibility, researchers exploited the fact that hub cells are terminally differentiated and therefore do not divide. When β-galactosidase-expressing (βGAL⁺) CySC clones were generated through mitotic recombination³⁶, βGAL⁺ hub cells were initially absent but appeared following a chase period. Similarly, upon administration of a BrdU pulse, BrdU-positive hub cells were detected within several days. Thus, CySCs not only make the somatic cyst cells but also can contribute to their niche by making hub cells (FIG. 2c). Although CySC contribution to the hub might be rare during steady state, when CySCs carry mutations in the lines gene, they transform their fate to adopt characteristics of hub cells. Interestingly, lineage-tracing experiments further suggest that hub cells and CySCs originate from a common precursor pool during embryogenesis³⁷. Together, these data point to a close relationship between CySCs and hub. In the future, it will be interesting to see whether CySCs increase their contribution to hub cells during ageing or stress.

Structure of the bulge: a shared stem cell niche

Epithelial hair follicle stem cells (HFSCs) reside in an anatomical region known as the bulge, which is located below the sebaceous glands at the base of the non-cycling portion of follicles. Slow-cycling HFSCs are located in the outermost layer and are marked by high levels of integrins (α6β4 and α3β1), keratins (keratin 5 (K5), K14 and K15) and cell surface markers (CD34 and Leu-rich repeat-containing G protein-coupled receptor 5 (LGR5))^3,38,39. They also express transcription factors, including SOX9, LIM homeobox 2 (LHX2), nuclear factor of activated T cells, cytoplasmic 1 (NFATC1), T cell factor 3 (TCF3) and TCF4. Loss-of-function studies show that all of these factors help to maintain features of ‘stemness’ (REFS 40–45).

The inner layer of bulge cells forms at the end of the hair cycle (see below). These cells do not express integrins or CD34 but they do express some HFSC markers, including NFATC1, SOX9, LHX2 and TCF3. They also exhibit features not seen in HFSCs, including unusual intercellular adhesions to the hair shaft and expression of K6 (REF. 46).

The bulge base features a small cluster of cells, known as the hair germ. Transcriptional profiling shows that hair germ cells are more similar to HFSCs than to their transient amplifying progeny, that is, to matrix cells (see below)⁴⁷. Except for NFATC1 and CD34, the hair germ expresses the HFSC markers mentioned above. They can be distinguished from the bulge HFSCs by their high levels of placental cadherin (P-cadherin).

The bulge niche hosts not only HFSCs but also melanocyte stem cells (MCSCs), which are interspersed between HFSCs⁴⁸. The behaviours of HFSCs and MCSCs are coordinated, enabling differentiating MCSCs to generate and transfer pigment to matrix cells as they begin to terminally differentiate to produce hair shaft cells^49,50. The bulge niche has also been implicated as a residence for smooth muscle progenitors responsible for generating the arrector pili muscle, which is located just above the bulge⁵¹.

Components of the HFSC niche

Why do so many stem cells remain in and near the bulge? One reason is that the bulge offers a rich milieu for various regulatory inputs coming from a diverse array of cell types. Converging on the upper bulge are four different types of sensory neurons, providing a means of coordinating sensory responses in the skin with stem cell behaviour⁵². Additionally, the bulge is encased by blood vessels, providing hormonal and nutrient inputs to stem cells.

Much of what we know about the other components of the niche and their signalling is focused on their relation to HFSC regulation. HFSCs themselves express several secreted factors that regulate their own behaviour^39,53. Additionally, several different cell types in and around the bulge niche are likely to participate in regulating HFSC activity (FIG. 3a). Within the niche, MCSCs are immediate HFSC neighbours and likely candidates for crosstalk⁵⁰. In the inner bulge layer, K6⁺ cells are particularly interesting, in that they are progeny of HFSCs and function in maintaining HFSC quiescence⁴⁶. On the other side of the extracellular matrix-rich basement membrane, which separates dermis from epithelium, are counteracting signals that promote stem cell activation and proliferation. A major source of these signals is the dermal papilla, which is located just beneath the hair germ and has long been established as an essential activation centre^54–56. The HFSC and hair germ niche is surrounded by a sheath of dermal fibroblasts, which can also provide local activating or inhibitory factors for HFSCs⁵⁷.

Two longer range signalling inputs participate in regulating HFSC behaviours. Subcutaneous adipocytes send out waves of alternating inhibitory and activating factors that help synchronize HFSC niches within the skin⁵⁷. Most recently, adipocyte precursors were implicated in transmitting regulatory factors that probably signal to receptors on the surface of the dermal papilla and/ or dermal sheath⁵⁸. Although we currently understand only a fraction of the extensive crosstalk within these many cell types, the intriguing picture emerging is that multiple niche components regulate stem cell behaviour at different hair cycle stages (see the discussion below on specific cell types and signals involved to drive or inhibit hair growth). It seems likely that many of these cells will also harbour regulatory functions towards other stem cell residents of the bulge.

Activation of HFSCs by dermal niche components

Hair follicles undergo cyclical bouts of growth (anagen), destruction (catagen) and rest (telogen), which are collectively known as the hair cycle⁵⁹. This unique feature has provided insights into the general process of tissue regeneration during homeostasis.

HFSCs remain quiescent throughout most of the hair cycle^2,3. At the start of the hair growth phase, HFSCs become activated briefly^47,60. Stem cell activation is complex, and the factors involved are still being discovered. Two critical pathways are activation of WNT signalling and inhibition of the BMP pathway. Together, these stabilize nuclear β-catenin in the hair germ and promote hair follicle regeneration^47,61–69.

During the quiescent phase of the hair cycle, hair germ cells produce increasing levels of WNTs while the dermal papilla produces increasing levels of BMP inhibitory factors^47,56. At the telogen–anagen transition, transforming growth factor-β2 (TGFβ2) is produced and transmitted by the dermal papilla. Hair germ cells respond by inducing TMEFF1 (transmembrane protein with EGF-like and two follistatin-like domains 1), which inhibits the BMP pathway, thereby activating HFSCs⁷⁰. In addition, intradermal adipocyte precursors secrete platelet-derived growth factor-α (PDGFα), which further stimulates the mesenchyme surrounding the bulge⁵⁸. Together, these accumulating activation forces transmitted to HFSCs finally overcome strong inhibitory thresholds from other cells within the niche, thereby triggering a new hair cycle. When this happens, the hair germ begins to proliferate and reform the new hair follicle, followed by HFSC proliferation in the bulge 1–2 days later.

HFSCs divide several times as they exit the bulge, forming a downward trail that extends from the bulge along the exterior (the outer root sheath (ORS)) of the regenerating hair follicle. Meanwhile, the dermal papilla is pushed further downward, away from the bulge. As the dermal papilla becomes more distant, bulge and upper ORS cells return to quiescence. Lower ORS cells become increasingly more proliferative as they approach the dermal papilla and change into transient amplifying matrix cells.

In the mature hair follicle, matrix cells envelope the dermal papilla, forming a bulb-like structure at the base. There, transient amplifying matrix cells divide rapidly but transiently before terminally differentiating to produce the new hair and its channel⁷¹. The dynamic movement of the dermal papilla from the bulge to the transient amplifying matrix during the hair cycle may contribute to the prompt return of the HFSCs to quiescence and result in a distinct pool of highly proliferative transient amplifying cells to fuel this tissue-producing factory.

Interactions that resemble those of stem cells in the D. melanogaster testis were reported recently as occurring between MCSCs and HFSCs. During anagen onset, WNT signals activated at the bulge base promote MCSC proliferation and differentiation⁵⁰. As activated MCSCs move downwards with the HFSC progeny, they generate pigment, transferring it to the transient amplifying precursors of the hair shaft. This collaboration between HFSCs and MCSCs within the hair follicle offers a valuable mammalian model to probe the interactions and coordination of multiple stem cell types within the same niche.

Feedback signals from progeny switch HFSCs off

Upon exhausting their proliferative potential, the matrix population undergoes apoptosis, triggering catagen that stops hair growth. During this time, most cells within the lower ORS also undergo apoptosis, as the remaining epithelial strand and basement membrane retracts upward, bringing the dermal papilla with it. HFSCs in the upper portion of the ORS are spared and become the new bulge for the next cycle. HFSCs that progressed a bit further along the ORS become the hair germ.

Surprisingly, a few committed cells in the lower ORS also survive catagen. With matrix cells undergoing apoptosis, these ORS cells bypass the transient amplifying step, undergoing an atypical differentiation route and settling in the inner K6⁺ bulge layer (FIG. 3b). Despite retaining many stem cell markers, K6⁺ cells do not function as stem cells during homeostasis or upon wounding. Rather, K6⁺ bulge cells form specialized intercellular junctions that anchor the old hair shaft and prevent the animal from losing all its hairs at the end of the hair cycle. In addition, these cells produce high levels of fibroblast growth factor 18 (FGF18) and BMP6 (REF. 46), which are quiescence factors that are also expressed at low levels by HFSCs³⁹. Notably, if K6⁺ bulge cells are either removed through hair plucking or ablated through preferential expression of diphtheria toxin receptor (DTR) followed by diphtheria toxin administration, bulge HFSCs become prematurely activated and enter a new round of hair growth. Importantly, injections of BMP6 or FGF18 prevent this early activation, suggesting that the K6⁺ bulge restricts HFSC activity in part through these factors. This discovery is an example in which stem cell progeny can act as a key signalling centre to stem cells and maintain their quiescence in the niche⁴⁶.

Committed MCSC lineage cells are not known to return back to the bulge at the end of catagen. However, repeated hair plucking, which depletes K6⁺ bulge cells, also results in early hair greying. This could be a sign of MCSC exhaustion owing to overactivation⁷². In the future, it will be interesting to probe deeper into the possible crosstalk between K6⁺ bulge cells and both HFSCs and MCSCs.

Joint efforts of microenvironment and macroenvironment

Once hair follicles re-enter the quiescent phase, both HFSCs and MCSCs can remain dormant for months. During this time, the stimulatory dermal papilla abuts against the HFSCs, hair germ and MCSCs, necessitating strong inhibitory signals from the surrounding environment to maintain stem cell quiescence. In addition to K6⁺ bulge cells, the global dermal macroenvironment, including fibroblasts and subcutaneous adipocytes, secretes BMP2 and BMP4 and contributes to the BMP-high, WNT-low state of the niche at this time. This status progressively changes so that, by the end of telogen, dermal BMP levels decline while WNTs rise^57,68. Thus, together with local crosstalk between stem cells and neighbouring niche components, long-range signals from the macroenvironment shift the BMP-high and WNT-low state to a BMP-low and WNT-high one, thereby favouring the telogen to anagen transition. These macroenvironmental cues reinforce synchronized stem cell behaviour across the animal’s coat.

In summary, the hair follicle system is an example of how stem cell progeny and heterologous niche cell types either counterbalance each other’s effects or cooperate to ensure proper stem cell activity during homeostasis. Moreover, the ability of HFSC progeny to return to the niche and impose regulatory signals to stem cells may provide a paradigm for stem cell niches, which must orchestrate when to make tissue and when to stop. The communication network in this busy corner of the hair follicle appears to be handling quite a cacophony of signalling circuits, and it remains a wonder that the ‘wires’ do not get crossed.

Coexistence of slow-cycling and fast-cycling ISCs

The G protein-coupled receptor LGR5, is expressed at the bottom of the crypt. A tamoxifen-inducible Cre–oestrogen receptor (Cre–ER) recombinase under control of the Lgr5 promoter was used to excise a floxed stop codon that otherwise prevents a genetically manipulated Rosa26–LacZ locus from being ubiquitously expressed. After Cre–ER activation, βGAL was detected in LGR5⁺ cells and, soon after, in all of their progeny. All cell types within crypts and villi were marked in a long-term manner, suggesting that LGR5⁺ cells are bona fide ISCs and not a transient amplifying population²⁵. Despite proliferative differences between fast-cycling LGR5⁺ ISCs and slow-cycling HFSCs, they share many key stem cell genes, including Lgr5, Sox9 and Tcf4. As seen in hair follicles, BMP signalling imposes a quiescence force, whereas WNT signalling is crucial for ISC activation^73,74.

Historically, a population of slow-cycling cells (BOX 2) were identified around the fourth position from the bottom of the crypt (‘position +4’), above the LGR5⁺ cells⁷⁵. Several proteins mark these slow-cycling cells, such as Polycomb complex subunit BMI1 (REF. 76), telomerase reverse transcriptase (TERT)⁷⁷ and the atypical homeodomain-containing protein HOPX1 (REF. 78). Interestingly, analogous lineage-tracing experiments using Bmi1–Cre–ER, Tert–Cre–ER or Hopx1–Cre–ER to mark the +4 cells have provided compelling evidence that these cells can also generate all cell types within crypts and villi, including the LGR5⁺ cells. Importantly, when the LGR5⁺ cells were ablated, these slow-cycling ISCs expanded in number and efficiently reconstituted the LGR5⁺ population⁷⁹. This apparent functional equivalency and lack of hierarchy suggests that, like the stem cells of the hair follicle and bone marrow, ISCs can transition between slow-cycling and fast-cycling states. In the future, it will be important to determine the relative contributions of these states to tissue formation during homeostasis and upon injury. Given the rapid turnover requirements of the intestinal epithelium, it seems probable that fast-cycling ISCs may contribute more to homeostasis, whereas slow-cycling ISCs are reserved mainly for injury conditions, such as ionizing radiation. A similar situation has previously been described for the haematopoietic system^26,80.

Paneth cells regulate the LGR5⁺ ISCs

Which cells are the niche cells for ISCs? Neighbouring mesenchymal cells were initially proposed to be crucial ISC niche stimulants but, thus far, in situ hybridization experiments have failed to map specific ISC stimuli to these cells⁸¹. By contrast, recent studies have brought Paneth cells, which are interspersed between the LGR5⁺ ISCs, to the forefront of the candidate list. Paneth cells secrete antimicrobial molecules and have long been known for their defensive roles⁸². Lineage-tracing analyses confirmed that Paneth cells are produced directly by ISCs and, in this regard, they constitute ISC progeny that are located at the crypt niche²⁵.

The relationship between the Paneth cells and the ISC niche has long been postulated. However, it was only recently that researchers demonstrated that Paneth cells might function in ISC maintenance^83,84. The first evidence that Paneth cells may be critical ISC niche cells came from in vitro studies. By combining Paneth and LGR5⁺ cells in culture, the efficiency of forming organoids that resemble the gut was increased significantly relative to that of LGR5⁺ cells alone⁸³. To show that Paneth cells function similarly in vivo, three different mouse models known to affect Paneth cell numbers were used⁸³: a mutation in the zinc-finger protein growth factor-independent 1 (Gfi1)⁸⁵; a conditional knockout of Sox9 (REFS 86,87); and expression of diphtheria toxin under a Paneth cell-specific promoter⁸⁸. When Paneth cell numbers were reduced, the numbers of ISCs were also reduced accordingly (FIG. 4). Interestingly, the remaining ISCs clustered around the surviving Paneth cells⁸³.

It is not yet clear whether Paneth cells act by secreting paracrine factors that sustain ISC survival and/or proliferation or whether they function to suppress ISC differentiation. Transcriptional profiling has revealed that Paneth cells express Wnt3, Egf, Tgfa and Delta-like ligand 4 (Dll4), which, at least in culture, are all factors that are important for ISC proliferation and maintenance^81,89,90. Among the pathways controlled by these factors, the canonical WNT pathway appears to be the primary signal that promotes intestinal epithelial maintenance, as evidenced by Tcf4 loss-of-function mutations⁷³ and by recent molecular and genetic links established between WNT receptors and the ISC G protein-coupled receptors LGR5 and LGR4 (REF. 91). The LGR ligands are roof plate-specific spondins (R-spondins), which are known to stimulate WNT–β-catenin signalling^84,92. Future studies should aim to demonstrate the in vivo relevance of these myriad Paneth-secreting factors and to firmly link these candidate factors to the importance of Paneth cells as key LGR5⁺ ISC niche components. Depletion of candidate factors specifically in the Paneth cells might help to advance this area. However, the capacity of Paneth cells to stimulate the proliferation of ISCs seen in vitro suggests that Paneth cells can no longer be viewed merely as immune surveillance guardians of ISCs.

Although Paneth cells might facilitate ISC self-renewal, this does not exclude the importance of underlying mesenchymal cells. Mesenchymal cells at the crypt–villus border express BMP2 and BMP4, which may halt proliferation and promote differentiation of ISCs at this juncture^74,93. Other signals from mesenchymal cells, such as the secreted Frizzled-related proteins (sFRPs), might also attenuate and modulate WNT signalling to prevent over-proliferation of ISCs⁸¹.

Comparisons between the HFSC niche and the ISC niche

When taken together, several interesting commonalities emerge between HFSCs and ISCs. Both are influenced similarly by WNTs and BMPs. In addition, both seem to be responsible for making critical niche components — HFSCs produce the K6⁺ inner bulge cells, and ISCs make the Paneth cells. In this regard, mammalian HFSCs and ISCs seem to differ significantly from D. melanogaster CySCs, which might contribute to their own niche only after hub formation.

Despite these similarities, stem cell progeny of the hair follicle and intestine display distinct effects on their self-renewing stem cell parents: K6⁺ bulge cells enhance HFSC quiescence, whereas Paneth cells promote ISC proliferation. These differences also reflect the stereotypical behaviours of the stem cells in their resident compartments: the HFSCs in the bulge are usually in a state of quiescence, whereas ISCs at the crypt base rapidly cycle. Although additional studies are still necessary, it is tempting to speculate that, by designing key niche components from downstream committed progeny, stem cells may receive a blueprint for the cycling behaviour needed for sustaining homeostasis within the tissue. Given that these epithelial tissues must also respond to injury, the ability of stem cells to receive cues from committed progeny adds flexibility to the blueprint, further enabling the stem cells to strike the requisite balance between proliferation and quiescence within their niche.

It is also worth noting that HFSC or ISC populations with different cycling properties are positioned at distinct locations, which might reinforce their cycling behaviours: in contrast to the bulge HFSCs, the cells of the hair germ do not make direct cell–cell contact with inhibitory K6⁺ bulge cells, but rather they are adjacent to the dermal papilla activation centre. Similarly, the +4 slow-cycling ISCs reside above the proliferation-promoting Paneth cells and are closer to the BMP-expressing mesenchymal cells at the crypt–villus border. Such features suggest that microenvironment has an impact on stem cell activity.

The HSC niche

Accompanying the unusually mobile and dynamic nature of its key residents is a complexity to the HSC niche that has not been seen in other stem cell systems. In addition to well-established niche components, such as osteoblasts and bone marrow vasculature^21,96–100, osteoblast progenitors, adipocytes, mesenchymal cells and non-myelinating Schwann cells within the bone marrow have recently been shown to influence HSC behaviour^101–105.

It has also been postulated that the location of HSCs directly affects their cycling behaviour, presumably because of the different environmental stimuli they encounter at each location. In this regard, live-imaging studies suggest that HSCs are more quiescent when they are closer to the endosteal lining, which is enriched for osteoblasts, osteoclasts and stromal fibroblasts. By contrast, more active HSCs tend to be closer to the perivascular niche, which is more centrally located within the bone marrow and marked by perivascular mesenchymal stem cells (MSCs), sinusoid endothelial cells and neural cells^23,24.

As the number of cell types known to regulate HSCs continues to increase, dissecting the complex crosstalk involved and the differentiation status of other niche lineage cells becomes a major challenge. Osteoblasts were among the first niche residents to be characterized. Paralleling the Paneth cell–ISC connection, when osteoblast numbers are increased, HSCs expand concomitantly^96,97. Conversely, depletion of osteoblasts results in a loss of HSCs^106,107. By contrast, osteoclasts seem to be dispensable for HSC maintenance¹⁰⁸. The most dormant, long-lived HSCs seem to reside next to osteoblast progenitor cells (which are known as preosteoblasts). Further supporting a functional role for these progenitors in regulating HSC dormancy, mutational loss of microRNA processing in preosteoblasts, but not in mature osteoblasts, causes HSC hyperproliferation, aberrant haematopoiesis and myelodysplasia¹⁰⁴.

On the perivascular side of HSC niche components are nestin⁺ MSCs, which influence HSC migration and maintenance (see below)¹⁰². CXCL12-abundant reticular cells (CAR cells) also affect HSC numbers¹⁰³. These cells may partially overlap with nestin⁺ MSCs on the perivascular side but have also been seen in the endosteum. Finally, non-myelinating Schwann cells appear to maintain HSC quiescence by regulating latent TGFβ activation¹⁰⁵. This is in contrast to the function of TGFβ in the hair follicle niche, where active TGFβ seems to activate HFSCs⁷⁰. Although much is still to be learned about these signalling networks, it is already clear that different cell types within both endosteal and perivascular compartments influence HSC behaviour in multifaceted ways (FIG. 5a).

HSC progeny are HSC regulators

Regulatory T cells (T_Reg cells) are a type of HSC progeny in the lymphoid lineage. They are critical for suppressing immune responses and confining tolerance to self-antigens. T_Reg cells also mediate immune response suppression at specific immune-privileged sites, where introduction of foreign antigens is tolerated without eliciting an inflammatory immune response.

Immune privilege is thought to be an adaptation that protects vital structures from deleterious damage caused by inflammation. As stem cells are crucial for tissue regeneration, it has been speculated that they might reside within immune-privileged sites. At least for HSCs, this seems to be the case. Thus, transplanted HSCs survive in the bone marrow of non-irradiated syngeneic and allogeneic recipients without eliciting immune response. This effect seems to be attributable to T_Reg cells within the bone marrow, as their depletion results in fewer surviving donor HSCs. In this regard, T_Reg cells may act as a shield for HSCs, protecting them from immune attack¹⁰⁹.

On the side of myeloid lineage progeny, macrophages also provide interesting feedback to HSCs. It appears that the movement of HSCs is regulated by macrophages located in the bone marrow^110,111. Two independent groups used conceptually similar but technically different ablation approaches to study these macrophages. Winkler and colleagues¹¹⁰ targeted ablation of macrophages by two methods: intravenous injection of clodronate-loaded liposomes, which are taken up by macrophages; and a transgenic model that triggered macrophages to undergo FAS-induced apoptosis. Chow and colleagues¹¹¹ used a CD169–DTR model that expressed DTR in a subset of macrophages, enabling them to induce depletion upon injection with diphtheria toxin. Although different ablation approaches might target different subtypes of macrophages, both groups found that, when macrophages are depleted, large numbers of HSCs mobilize from their bone marrow niche and enter the peripheral bloodstream (FIG. 5b).

How do macrophages influence the localization of HSCs? Although the exact mechanism remains to be elucidated, current evidence suggests that they do so through an indirect fashion. On the perivascular side, this effect seems to involve the nestin⁺ MSCs in the bone marrow. Purified HSCs home near these nestin⁺ MSCs when transplanted into hosts and, conversely, depletion of nestin⁺ MSCs rapidly diminishes HSC numbers¹⁰². Upon depletion of macrophages, several genes implicated in HSC retention or maintenance are specifically downregulated within nestin⁺ MSCs residing within the bone marrow¹¹¹. These include the genes encoding the secreted factors CXC chemokine ligand 12 (CXCL12; also known as SDF1), angiopoietin 1 (ANGPT1), KIT ligand (KITL; also known as stem cell factor and steel factor) and vascular cell adhesion molecule 1 (VCAM1), and their downregulation suggests that macrophages may affect HSC retention through regulating critical retention factors in nestin⁺ MSCs. On the endosteal side, ablation of macrophages may affect osteoblast numbers, thereby making the bone marrow cavity a less favourable site for HSCs^110,112. The next important step will be to delineate the signals sent from macrophages to nestin⁺ MSCs and/or osteoblasts to control HSC retention within the bone marrow.

The example of macrophages and HSCs unveils yet another intriguing mechanism for how progeny can influence stem cells. In this case, instead of affecting stem cells directly, the progeny first have an impact on other heterologous niche cell types within the microenvironment and this then affects the behaviour of stem cells. Given the coexistence of multiple heterologous cell types within various stem cell niches, additional examples of these intricate circuits are likely to unfold in the future.