The global regulatory logic of organ regeneration: circuitry lessons from skin and its appendages

Abstract

In organ regeneration, the regulatory logic at a systems level remains largely unclear. For example, what defines the quantitative threshold to initiate regeneration, and when does the regeneration process come to an end? What leads to the qualitatively different responses of regeneration, which restore the original structure, or to repair which only heals a wound? Here we discuss three examples in skin regeneration: epidermal recovery after radiation damage, hair follicle fate choice after chemotherapy damage, and wound-induced feather regeneration. We propose that the molecular regulatory circuitry is of paramount significance in organ regeneration. It is conceivable that defects in these controlling pathways may lead to failed regeneration and/or organ renewal, and understanding the underlying logic could help to identify novel therapeutic strategies.

I. INTRODUCTION

Understanding organ regeneration is a major challenge in biomedical research. Much research has focussed on the competence of cells that enable regeneration, namely the identification of stem cells and progenitors (Tanaka, 2016; Chen & Poss, 2017; Gonzales & Fuchs, 2017; Wells & Watt, 2018). Lineage tracing further clarified these cell sources and revealed their plasticity in the regenerating blastema (Tanaka & Reddien, 2011; Tornini et al., 2016, 2017; Gerber et al., 2018; Mokalled & Poss, 2018). However, the fundamental logics that regulate the initiation and progression of regeneration remain unclear (Brockes & Kumar, 2008; Liu et al., 2013; Sikes & Newmark, 2013; Umesono et al., 2013; Goldman & Poss, 2020). Clearly, basic questions cannot be addressed without considering the molecular circuitry governing the regeneration process. For example, quantitatively, what defines the threshold that triggers regeneration, and when does it come to an end? Qualitatively, what drives the differential response after wounding to elicit repair, which only heals the damage, or regeneration which restores the original structure and function?

The integument forms the interface between the organism and the environment, and is very robust in its regenerative abilities. Skin and its appendages are among the most regenerative organs in the animal body, in particular, hair follicles (HFs) in mammals and feather follicles (FFs) in birds. The identification of epidermal stem cells and HF/FF stem cells has significantly improved our knowledge regarding the maintenance and renewal of this organ (Cotsarelis, Sun & Lavker, 1990; Morris et al., 2004; Yue et al., 2005; Gonzales & Fuchs, 2017). In an era of regenerative medicine, the skin and HFs offer attractive opportunities for tissue engineering. For instance, HFs may form through various possible routes: (i) morphogenesis via embryonic development; (ii) physiological cycling; (iii) wound-induced regeneration; (iv) wound-induced neogenesis (Ito et al., 2007; Gay et al., 2013; Lim et al., 2018); and (v) tissue engineering (Higgins et al., 2013; Lei et al., 2017; Abaci et al., 2018). The molecular mechanisms governing these events may or may not be the same, attracting extensive research interest.

Skin and its appendages have long been used as models to investigate how chemotherapy and radiation therapy damage the tissue (Malkinson & Keane, 1981; Paus et al., 1994; Chiu & Chuong, 2015; Gao et al., 2019). With recent progress, novel concepts are emerging regarding the molecular regulatory circuitries for organ regeneration from the perspective of systems biology. Specifically, here we discuss how epidermal regeneration is triggered following ionizing radiation (IR)-induced skin damage (Xie et al., 2017), how the HF/FF responds differentially to chemotherapy- and radiation therapy-induced damage (Chen et al., 2014; Xie et al., 2015; Huang et al., 2017, 2019, 2021; Haslam et al., 2021), and the feedback mechanisms involved in wound-induced FF regeneration (Chu et al., 2014; Lin et al., 2018). The lessons learned from skin and its appendages may be useful in other systems. In addition, we propose that under pathological conditions when the regulatory pathways are perturbed, regeneration may fail to initiate, leading to reduced or loss of organ function.

II. THE INITIATION OF EPIDERMAL REGENERATION: THE CONCEPT OF CONTACT INHIBITION

The epidermis, the outer epithelial lining of vertebrate skin, is organized as proliferating units (Potten, 1974). Its location-dependent, developmentally predetermined thickness under physiological conditions is dictated by a stringently maintained balance between keratinocyte proliferation, terminal differentiation and apoptosis, which is disrupted by wounding. Recent advances in intravital imaging and clonal analysis further complement this concept by showing that cells compete for space in the epidermis, and the epidermal cells undergo competitive equilibrium (Rompolas et al., 2016; Lynch et al., 2017; Lei & Chuong, 2018; Mesa et al., 2018; Murai et al., 2018). When this equilibrium is perturbed by wounding or in the context of inflammatory skin diseases, signature shifts in epidermal gene expression were reported (Mills et al., 2018), leading to a new homeostatic status.

Historically, the clonal regeneration of epidermis after IR-induced damage was key evidence for the concept of epidermal stem cells, where patches of clonal epithelial growth eventually replace the damaged epidermis (Malkinson & Keane, 1981). Using TOPGAL mice where the activation of Wnt [wingless-type mouse mammary tumour virus (MMTV) integration site] signalling induces LacZ (bacterial beta-galactosidase gene) expression, we confirmed that epidermal regeneration indeed is clonal after IR-induced skin damage (Xie et al., 2017). Although detailed lineage tracing experiments are still missing for the epidermal regeneration process, these data highlighted a critical contribution of Wnt signalling. In addition, we found that the triggering mechanism for skin regeneration after IR damage involves contact inhibition.

Contact inhibition was initially used to describe contact-dependent cell growth in vitro (Fig. 1A). The molecular events involved include E-cadherin/β-catenin-mediated cell contact, and Hippo signalling which controls subsequent cell growth (Mendonsa, Na & Gumbiner, 2018). In IR-induced skin damage in murine skin, E-cadherin was degraded through reactive oxygen species (ROS)-triggered pathways, and activation of Wnt and Hippo signalling precedes clonal expansion of basal keratinocytes (Xie et al., 2017). Epidermal regeneration was triggered only after a single high dose of IR irradiation, which leads to the disruption of cell–cell contacts: basal keratinocytes physically lose their contacts with neighbouring cells. Moreover, by using a blocking antibody for E-cadherin which helps to break the adherens junctions, the regeneration process can be triggered at a lower IR dose; by contrast, by promoting the stability of cell contacts, epidermal regeneration can be prevented (Xie et al., 2017). Based on these cellular and molecular events, it is likely that loss of contact between basal keratinocytes, and thus loss of contact inhibition signalling, triggers epidermal regeneration (Fig. 1B, C).

Fig 1.

The logic of epidermal skin regeneration. (A) Contact inhibition is a phenomenon initially described in cell culture in vitro, where compact cells cease proliferation. Upon removal of a portion of the cells, they undergo proliferation and migration to reach confluence. (B) In ionizing radiation (IR)-induced skin damage, the compactness of basal keratinocytes is disrupted, leading to clonal expansion of epidermal stem cells which then re-build the skin barrier. (C) Pathways involved following IR-induced damage or wounding include the activation of Src (proto-oncogene Src, Rous sarcoma)/Abl (Abelson murine leukemia viral oncogene homolog) kinases, which then phosphorylate β-catenin (β-Cat), leading to the degradation of E-cadherin (E-Cad). Wnt and Hippo pathways are activated to drive cell proliferation. YAP, Yes-associated protein.

Epidermal keratinocytes are stacked compactly through adherens junctions, desmosomes and tight junctions, which support the mechanical stability of skin and its barrier function (Sumigray & Lechler, 2015; Yokouchi & Kubo, 2018). Interestingly, in ultraviolet (UV) radiation-induced skin damage, breaking down of tight junctions (Yuki et al., 2011), de-stabilization of desmosomes (Chouinard et al., 2001; Johnson et al., 2014), and down-regulation of E-cadherin (Qiang et al., 2016), have all been documented. It appears that loss of contact inhibition signalling also contributes to UV radiation-induced skin regeneration. Together, these data indicate the crucial role of breaking cell–cell contacts in the initiation of epidermal regeneration (Fig. 1B, C).

III. TO REGENERATE OR NOT: THE MOLECULAR DETERMINANTS OF THE BIFURCATION BETWEEN REPAIR AND REGENERATION FOLLWOING HAIR FOLLICLE DAMAGE

There are two possible fates for the organ post-wounding: repair, which heals the wound but does not always restore the structure and function to its original level, or regeneration, which re-builds a structure that is fully functional. Using the HF and FF as models, novel rules are emerging that dictate whether the repair or regeneration process will be initiated after wounding.

HFs undergo cyclic growth involving active growing (anagen), regression (catagen) and resting (telogen), which may or may not be synchronized in an adult animal (Chen et al., 2016; O’Sullivan et al., 2021). To provide a reliable and reproducible model system, it is important to synchronize their growth – this can be achieved by plucking-induced regeneration (Malkinson, Griem & Morse, 1961; Paus et al., 1994), or by using their natural, synchronized first cycle of post-natal growth (Huang et al., 2021). Similarly, the life cycle of FFs in adult chicken is divided into a growth phase and resting phase (Chen et al., 2015), and plucking-induced regeneration can be used to synchronize their growth (Chen et al., 2014; Xie et al., 2015; Yue & Xu, 2017).

(1). Two possible fates for damaged HFs after chemotherapy

Chemotherapy-induced alopecia (hair loss; CIA) is a common adverse effect in clinical oncology (Paus et al., 2013; Freites-Martinez et al., 2019). The massively proliferating keratinocytes of the hair matrix are typically most sensitive to chemotherapy-induced damage, leading to HF dystrophy as a consequence of HF keratinocyte apoptosis, cell cycle arrest, and various degrees of stem cell damage, along with major disruption of the HF pigmentary unit, ultimately leading to hair loss (Paus et al., 1994, 2013; Slominski et al., 1996; Lindner et al., 1997; Hendrix et al., 2005; Bodo et al., 2007; Purba et al., 2019).

Chemotherapy-exposed HFs choose between two distinct damage response pathways with very different clinical outcomes (Paus et al., 2013), which persuasively demonstrate the key differences between organ regeneration and tissue repair (Fig. 2): (i) dystrophic catagen – the HFs rapidly involute by entering apoptosis-driven catagen, followed by rapid hair shaft shedding, and then progress through an extremely shortened resting phase (telogen), followed by rapid reactivation of HF epithelial stem cells to initiate a new round of normal hair growth (regeneration); (ii) dystrophic anagen – the HFs remain in anagen which is typically longer than normal, while progenitors are mobilized in the outer root sheath/lower hair bulb to repopulate cell mass loss (repair).

Fig 2.

Fate choice between repair and regeneration of the damaged hair follicles. In chemotherapy-induced hair follicle (HF) damage, a key event is the reduction of Shh (sonic hedgehog) signalling. If the reduction does not breach a threshold, anagen is maintained and the HF is repaired to produce a damaged hair fibre (the dystrophic anagen response). Paradoxically, this retards the time point at which full HF regeneration would be accomplished during a new anagen. If the level of Shh signalling drops below the threshold, the HF immediately regresses followed by catagen induction and a greatly shortened telogen, and quickly re-enters anagen, thus regenerating a new hair fibre (the dystrophic catagen response). There are clinically relevant agents that are known to favour each of these two pathways, although the molecular mechanisms remain to be systematically dissected. DP, dermal papilla; ORS, outer root sheath; α-MSH, alpha-melanocyte stimulating hormone.

Even though the dystrophic catagen response is associated with the highest degree of hair loss, it also leads to the fastest re-growth of normally pigmented hair shafts, hence representing a classical regeneration phenomenon. Instead, the dystrophic anagen response, which is clinically associated with much milder hair loss, only engages in tissue repair without proper organ remodelling, and thus produces a defective, often depigmented hair shaft (Haslam et al., 2021) (Fig. 2).

(2). The degree of cell apoptosis alone cannot predict the fate of the HFs

The HF’s choice between these two response pathways can be manipulated pharmacologically. For example, while cyclosporine A, tacrolimus, and alpha-melanocyte stimulating hormone (α-MSH) favour the dystrophic anagen pathway, glucocorticoids, oestradiol, and calcitriol promote the dystrophic catagen pathway (Paus et al., 1996, 2013; Maurer, Handjiski & Paus, 1997; Ohnemus et al., 2004; Bodo et al., 2009; Bohm et al., 2014). This implies that there are candidate pathways whose activation or inhibition by these agents drives the regeneration versus repair response. However, these pathways remain to be systematically dissected and characterized.

It may be the case that the severity of the damage to the HF dictates its fate: more severe damage leads to stem cell-mediated regeneration, whereas less-severe damage leads to simple repair by local progenitors. This is true for chemotherapy-induced hair follicle damage (Paus et al., 1994; Huang et al., 2017, 2019, 2021). However, in IR-induced alopecia (RIA), one of the most frequent and feared adverse effects of cancer radiotherapy, the situation appears more complex. Like chemotherapy, high-dose IR induces extensive apoptosis in the hair matrix keratinocytes, leading to HF dystrophy and hair shaft defects or shedding (Malkinson & Keane, 1981; Huang et al., 2017). Moreover, IR causes massive damage to the HF stem cells (Sotiropoulou et al., 2010; Schuler, Timm & Rube, 2018). Yet, HFs typically remain in dystrophic anagen without resorting to stem cell-mediated organ regeneration (Huang et al., 2019). Since the degree of hair matrix apoptosis induced by chemotherapy or IR is not significantly different under total hair loss conditions, as judged by the percentage of TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling)-positive cells in the hair matrix keratinocytes (Lindner et al., 1997; Hendrix et al., 2005; Huang et al., 2017), the degree of injury as judged by cell apoptosis alone does not explain the fate choice between the repair and regeneration responses.

(3). Involvement of Shh signalling in controlling the fate of the damaged HFs

Sonic hedgehog (Shh) signalling is an evolutionarily conserved pathway in vertebrate development. It is also essential for HF morphogenesis (St-Jacques et al., 1998; Woo, Zhen & Oro, 2012). In post-natal hair growth and cycling, it is critical for anagen initiation and maintenance, because inhibition of this pathway arrests hair growth (Wang et al., 2000; Oro & Higgins, 2003; Haslam et al., 2021).

Using a comparative approach, we recently defined a role of Shh signalling in chemotherapy-induced tissue damage in avian feathers (Chiu & Chuong, 2015; Xie et al., 2015; Gao et al., 2019). Feathers grow and regenerate robustly, similarly to mouse hair. However, there is an important distinction in the role of Shh signalling: in mouse hair, inhibition of Shh signalling arrests hair growth and leads to dystrophy/regression (Wang et al., 2000; Haslam et al., 2021), whereas in avian feathers, only the branching is perturbed but the feather axis remains normal because the pathway is not active in this region (Xie et al., 2015). Therefore, chemotherapy may perturb the murine hair growth cycle, but only induces transient damage in avian feathers, as might be deduced from the distinct role of Shh signalling in these two appendages (Gao et al., 2019).

Subsequent study confirmed the down-regulation of Shh signalling in the HFs after chemotherapy, and that reactivation of Shh signalling maintains the HFs in anagen, thus partially rescuing CIA (Haslam et al., 2021). These data are consistent with the notion that Shh signalling regulates the hair growth cycle. However, in IR-induced HF damage, the perturbation of Shh signalling is not significant (Huang et al., 2017), thus the RIA response is mainly repair/dystrophic anagen. On the other hand, IR can prematurely terminate feather growth, followed by a rapid regeneration response with only minor pigmentary defects (Chen et al., 2014) – this phenotype differs significantly from the response to chemotherapy in FFs where only feather branching was perturbed (Xie et al., 2015). These feather phenotypes are consistent with the suggestion that Shh signalling does not control the transition from growth phase to resting phase in FFs. Together, it appears that a critical level of Shh signalling dictates whether HFs will regenerate or not (Fig. 2).

IV. A Wnt/Dkk-Frzb-Sfrp FEEDBACK LOOP AS THE CORE SIGNALING MODULE FOR FEATHER/HAIR REGENERATION

It is well established that hair growth and regeneration are organized by a specialized population of mesenchymal cells, the dermal papillae (DPs) which reside at the bottom of the follicle and are surrounded by the keratinocyte matrix (see Fig. 2). Although early surgical experiments demonstrated that rat vibrissae follicles can regenerate even when their DP are ablated (Oliver, 1966a,b), and subsequent studies confirmed that the mouse DP can be regenerated from the dermal sheath cells (Rahmani et al., 2014), it is nonetheless essential to have fully functional DP cells for HF reconstitution or tissue engineering (Higgins et al., 2013; Lei et al., 2017; Abaci et al., 2018). By contrast, the feather DP is absolutely required for feather regeneration, and its ablation rends the follicle unable to regenerate (Lillie & Wang, 1941, 1944; Wang, 1943). Therefore, extensive efforts have been made to characterize this cell population molecularly, and define its inductive capability. A group of key signalling molecules including fibroblast growth factor (FGF), bone morphogenetic protein (BMP) and Wnt family members have been shown to be critical. In this section, we discuss recent advances in our understanding of the involvement of these pathways, and propose that a feedback loop involves Wnt and Dkk (dickkopf-related protein)–Frzb (frizzled-related protein)–Sfrp (secreted frizzled-related protein) drives feather/hair regeneration.

(1). Profiling gene expression in the dermal papilla

The mouse DP cells express high level of BMP ligands (Bmp4, Bmp6) and antagonists [Noggin, Gdf10 (growth differentiation factor 10), Bambi (BMP and activin membrane bound inhibitor)], Wnt agonists (Wnt5a, Wnt11, R-spondin2/3) and antagonists [Wif (Wnt inhibitory factor), Sfrp1, Sfrp2, Frzb], FGF ligands (Fgf10, Fgf7) and receptors (Fgfr1), Shh receptors [Ptch1 [Patched 1)] and inhibitors [Hhip (Hedgehog interacting protein)] (Rendl, Lewis & Fuchs, 2005; Hagner et al., 2020). Similar profiling has also been performed in human DP cells (Ohyama et al., 2012; Higgins et al., 2013). It must be noted that Wnt5a and Wnt11 are not the typical ligands for canonical Wnt/β-catenin signalling, therefore other ligands must exist that trigger this pathway. The matrix keratinocytes are the most likely candidate, which was shown to be the case for mouse HFs (Lim et al., 2016) and more recently for human HFs (Hawkshaw et al., 2018). Consistently, Wnt, BMP and FGF signalling have been shown to maintain the inductive capability of DP cells (Ohyama et al., 2012); results from 3D culture provide some support for this (Higgins et al., 2013).

The molecular signature of the feather DP shows similarities and differences in several respects (Chu et al., 2014). Bmp4, Bmp7 and Noggin2 are enriched, with the transforming growth factor beta (TGF-β) signalling pathway as a distinctive feature. Fgfr1 and Fgf7 are also present. The Wnt receptors [Fzd8 (Frizzled 8), Fzd7] and inhibitors (Dkk2, Frzb, Sfrp1) but not the ligands are expressed in the feather DP – the Wnt ligands are mainly expressed in the feather epithelium. The feather DP is also characterized by smooth muscle-related gene expression, including Laminin, Desmin, Acta2 (smooth muscle actin α2), Actg2, Mylk (myosin light chain kinase), Myh11 (myosin heavy chain 11), Myl9 (myosin light chain 9), NCAM (neural cell adhesion molecule), and Tenascin-C. This expression pattern may suggest a smooth muscle-related origin of feather DP cells (Chu et al., 2014).

(2). The role of FGF signalling

The involvement of FGF signalling in hair growth control has long been documented. Fgf5 expression marks the onset of catagen; Fgf5 knockout leads to prolonged anagen and the formation of elongated hair (Hebert et al., 1994). Fgf18 is highly expressed in telogen and helps maintain the quiescence of stem cells; paradoxically, FGF18 administration promoted anagen transition and hair growth (Kawano et al., 2005; Kimura-Ueki et al., 2012). On the other hand, Fgf7 promotes the HF transit to anagen (Greco et al., 2009). These opposing results suggest that different FGF ligands elicit different responses within the HF.

More recently, ablating Fgfr1 and Fgfr2 in the mouse DP led to prolonged anagen and elongated hair formation (Harshuk-Shabso et al., 2020). Detailed analysis revealed cooperative changes in Wnt agonists (R-spondin 1/2/3/4) and antagonists [Dkk2, Notum (notum, palmitoleoyl-protein carboxylesterase)] expression within the DP, which in turn controls hair growth (Harshuk-Shabso et al., 2020). Thus, it appears that the role of FGF signalling in hair growth relies on the fine tuning of Wnt signalling.

A more dramatic phenotype for FGF signalling in the DP was found in FFs (Yue et al., 2012). Overexpression of FGF10 leads to expansion of the DP at the expense of distal components of the FFs (mesenchymal pulp and epithelial branching; see Fig. 3A), whereas blocking FGF signalling via Sprouty4 overexpression leads to loss of DP functionality, and eventually to loss of the follicle structure (Yue et al., 2012). FGF10 promotes the formation of long, elongated filopodia from the basal keratinocytes which project deep into the feather DP, possibly with a function of vesicle transport (Cheng et al., 2018). Reduced FGF signalling is required for the fate of more distal structures, whereas high levels of FGF signalling block feather branching (Cheng et al., 2018). Together, it appears that FGF signalling maintains the proximal fate of the FF and thus the DP properties; its down-regulation is required for the formation of distal feather structures.

Fig 3.

Feedback regulation of feather regeneration. (A) Schematic of feather regeneration. After plucking, the dermal papillae (dp) and the follicle wall with the thin covering layer of follicle sheath (fos) remain. This is followed by the expansion of epithelium and proliferation of the DP cells to form the pulp. A new feather germ is formed at day 4 (T4). ds, dermal sheath; fos, follicle sheath; dp, dermal papilla; pe, papillae ectoderm. (B) Wnt (wingless-type mouse mammary tumour virus [MMTV] integration site) ligands expressed in the epithelium and inhibitors Dkk2 (dickkopf-related protein 2)/Frzb (frizzled-related protein)/Sfrp1 (secreted frizzled-related protein 1) in the mesenchyme form a feedback loop. The equations describe their dynamic expression levels, as derived from this interactive module. $\left[T\right]$, $\left[L\right]$, concentrations of the molecules; $m$, maximum synthesis rate; $n$, Hill coefficient; α, background synthesis rate; $l$, degradation rate; $t$, time. (C) The kinetics of Wnt ligands and inhibitors during feather regeneration. Plucking removes most of the epithelium and thus the Wnt ligands, followed by their gradual accumulation during wound healing and regeneration. Since the inhibitors are downstream target genes of Wnt signalling, their expression also decreases initially, followed by a delayed increase. This interactive crosstalk drives feather regeneration.

(3). The role of BMP signalling

It is well established that BMP signalling is required for the maintenance of DP cell properties and their hair-induction capability (Rendl, Polak & Fuchs, 2008). In addition, BMP signalling maintains HF stem cell quiescence, and promotes proper differentiation of matrix keratinocytes (Genander et al., 2014). However, overexpression of Noggin, a potent antagonist of BMP signalling, promoted hair formation rather than blocked it (Plikus et al., 2004). In FFs, overexpression of Noggin promoted feather epithelial branching, whereas overexpression of BMP ligands inhibited it without perturbing feather regeneration (Yu et al., 2002).

Adding to the already complex role of BMP signalling in HFs, it was shown that dermal BMP ligands, which are mainly secreted from the adipocytes, control HF growth and regeneration on a global scale (Plikus et al., 2008; Geyfman et al., 2015). This is considered a key step to coordinate the HF growth cycle in skin, particularly in wild animals. Local macrophages may also contribute to the regulation of hair growth and regeneration after wounding as well as during the normal hair growth cycle (Rahmani et al., 2018; Abbasi & Biernaskie, 2019; Wang et al., 2019). The regulatory circuits involved in these diverse mechanisms are likely to be complex and remain a major challenge in this field.

(4). The Wnt/Dkk-Frzb-Sfrp feedback loop

Early functional evidence for the involvement of Wnt signalling in the maintenance of DP properties and hair-induction capability came from the hair reconstitution assay (Kishimoto, Burgeson & Morgan, 2000). Later, it was shown that overexpression of Dkk1 blocks HF formation (Andl et al., 2002), and HF regeneration requires β-catenin activity in the DP (Enshell-Seijffers et al., 2010). DP cells show reduced expression of Wnt receptor signalling pathway genes in culture (Higgins et al., 2013), whereas overexpression of Lef1 (Lymphoid enhancer binding factor 1) improves DP properties and hair-induction capability (Abaci et al., 2018).

The avian feather is a regenerative structure that undergoes continuous physiological cycling (Chuong et al., 2012; Chen et al., 2015; Lai & Chuong, 2016; Chang et al., 2019). After plucking, only the DP remains, plus an epithelial layer covering the follicle wall (the follicle sheath; Fig. 3A). It is likely that key factors controlling FF regeneration can be inferred from the kinetics of gene expression during this process: whereas initially the expression levels of key regulators may decrease because of the damage, they will eventually recover during regeneration. By contrast, genes that show little expression change may not be critical because they are not acutely involved.

Whole-genome expression analysis revealed that during feather regeneration, only Wnt signalling activity changed significantly among the major pathways (Lin et al., 2018). These include the abrupt removal of Wnt ligands after plucking, followed by their gradual re-accumulation in the feather epithelium. Wnt inhibitors including Dkk2/Frzb/Sfrp1 also are initially down-regulated, followed by a gradual increase in the DP and the newly formed pulp mesenchyme (Fig. 3B, C). Since these inhibitors not only block Wnt signalling, but are themselves downstream target genes of the Wnt signalling, a feedback regulatory loop develops. In this loop, feather removal by plucking (entailing loss of Wnt ligands) reduces Dkk2/Frzb/Sfrp1 transcription in the feather mesenchyme, whereas the gradual accumulation of Wnt ligands during progressive regeneration of the feather epithelium slowly up-regulates their level again (Lin et al., 2018). Thus, a balance of Wnt signalling in the follicle will be restored by this reciprocal epithelial–mesenchymal crosstalk. This also explains how feather regeneration comes to an end – theoretically, the nature of this feedback crosstalk will lead to cyclic growth, thus physiological moulting. Functional tests involving overexpression of either Wnt ligands or inhibitors in FFs led to a retarded regeneration process (Chu et al., 2014), attesting to the importance of balanced signalling for feather growth.

It is interesting to note that the control of Wnt activity in FFs by Wnt inhibitors expressed in the mesenchyme is remarkably similar to the human system: in scalp HFs, SFRP1 is secreted from the DP to regulate Wnt activity in the hair matrix, thereby determining the duration of anagen (Hawkshaw et al., 2018), or to suppress anagen development in telogen HFs (Hawkshaw et al., 2020). Future work will be able to investigate, using the powerful genetic tools now available, if this feedback loop also operates in mice to control hair regeneration and physiological cycling.

V. DISCUSSION AND FUTURE DIRECTIONS

Skin and its appendages have long been used as model systems for regenerative medicine. Their tremendous regenerative capability often leads to the concept of feedback regulation (Al-Nuaimi et al., 2010; Kunche et al., 2016; Lei & Chuong, 2018). Recent progress in studying how these organs respond to damage and their repair/regeneration responses has led to interesting discoveries: the loss of contact inhibition signalling triggers epidermal regeneration (Xie et al., 2017), the level of Shh signalling dictates HF fate choice following damage (Gao et al., 2019; Haslam et al., 2021), and epithelial–mesenchymal crosstalk between Wnt ligands and their inhibitors controls feather/hair growth and regeneration (Chu et al., 2014; Hawkshaw et al., 2018, 2020; Lin et al., 2018). Although much remains to be discovered, we propose that feedback regulatory circuitries are of paramount significance in controlling the regeneration process, offering a mechanism to sense damage globally and determine organ fate.

Progress in regenerative medicine has defined many factors that may regulate organ regeneration, such as the composition of the extracellular matrix, biomechanical forces/tension, tissue electric current, endoplasmic reticulum stress/mitochondrial activity, exosomes/microvesicles, and inflammation/immunological cues (Brockes & Kumar, 2008; Sehring & Weidinger, 2020; Zuppo & Tsang, 2020). The concept of cell competence has been proposed to play a key role. Yet a global regulatory logic operates above the role of specific elements or cell populations. For example, through manipulating positional information (a cell’s identity based on its location within the body which is of ten conveyed by gene expression and/or combination of signalling gradients), the normally non-regenerating planarian can gain the capability to regenerate (Liu et al., 2013; Sikes & Newmark, 2013; Umesono et al., 2013). Progress has also been made in identifying the global enhancers and transcriptional programs involved in regeneration (Kang et al., 2016; Goldman & Poss, 2020; Suzuki & Ochi, 2020; Thompson et al., 2020). In a common form of hair disorder, androgenetic alopecia, where the HFs exhibit reduced growth and regeneration capability, the loss of stem cells is not the driving force; indeed, the stem cells are mostly well preserved and, given an appropriate micro- and macro-environment, can rebuild a HF (Garza et al., 2011; Guerrero-Juarez & Plikus, 2018; Wier & Garza, 2020). Thus it is critical to define the global regulatory logic for organ growth and regeneration, which combines the impacts of environmental and physiological factors.

A gradual decrease of regenerative capability after IR damage has long been documented (Ryan, 2012; Acauan et al., 2015; Bray et al., 2016; De Ruysscher et al., 2019; McBride & Schaue, 2020; Rocchi & Emmerson, 2020). In addition to loss of stem cells, the stroma cells that form the stem cell niche may also be damaged and fail to support tissue regeneration. IR may induce permanent hair loss, which is related to stem cell damage and/or collapse of the immune privilege of the HFs (Bertolini et al., 2020; Scoccianti et al., 2020). In IR-induced damage to the oral mucosa and salivary glands, the regeneration process often fails, leading to permanent xerostemia and other related symptoms (Acauan et al., 2015; Marmary et al., 2016; Rocchi & Emmerson, 2020). Although the epidermal stem cells/salivary gland progenitors are still in place, the regeneration process is nonetheless not triggered or is defective. This may be due to increased local inflammatory responses, overall senescence of the tissue, or perturbation of other yet-unknown parts of the regenerative process. Future work should attempt to define the details of these defects to enable the design of interventions to allow regeneration of these important tissues and organs.

VI. CONCLUSIONS

1. The initiation and progression of organ regeneration are under the control of genetic programs, which often encode feedback regulatory circuitries to sense the damage globally and dictate the fate of the organ.

2. In radiation-induced epidermal skin damage, regeneration is triggered by loss of cell–cell contacts, a mechanism similar to contact inhibition in cell culture.

3. In chemotherapy- or radiation therapy-induced hair follicle damage, the level of Shh signalling dictates the fate choice between repair and regeneration.

4. In both feather and hair follicles, a feedback loop between Wnt ligands in the epithelium and inhibitors in the DP senses the damage and drives the regeneration process.

5. It is important to define the global regulatory logic, rather than focusing on specific cell populations, to enable the design of better strategies to restore organ function.