Zhao et al. (2015) hypothesized that Gli1 expressing cells were the MSCs for craniofacial bones, as they were for the incisor mesenchyme they reported earlier (Zhao et al., 2014). The zinc finger protein GLI1 is a well-known transcriptional effector of Hedgehog signaling (Hooper and Scott, 2005). Seidel et al. (2010) had previously proposed it as a dental stem cell marker upon discovery of Hh responsiveness (Gli1^lacZ) from slow cycling cells of the dental epithelium (Seidel et al., 2010). The stem cell identity of these cells was later confirmed in the studies of Zhao et al. (2014). The AXIN2 protein, a negative regulator of Wnt signaling, was previously implicated in calvarial morphogenesis with Axin2 knockout mice showing craniosynostosis (Yu et al., 2005). Maruyama et al. (2011) described the essential role of AXIN2 in orchestrating the signaling network (Wnt, BMP, FGF) that regulates mesenchymal cell fate determination (Maruyama et al., 2011). The protein was recently proposed as a calvarial stem cell marker as Axin2-expressing cells in the midline of the suture mesenchyme were found slow-cycling in nature (Maruyama et al., 2016). Finally, the transcription factor Prx1 (also referred to as Prrx1) was previously shown to be highly expressed during limb bud formation and craniofacial development (Martin and Olson, 2000). Interestingly, PRX1 expression seemed to avert differentiation of early progenitors into committed osteoblasts, suggesting that the transcription factor could potentially mark SSCs in the calvarial bones (Lu et al., 2011). This hypothesis was successfully tested in the latest of the above-mentioned studies by Wilk et al. (2017).

Gli1+ cells of the cranium are found in the entire periosteum, dura and suture mesenchyme at birth. However, from postnatal day 21 onwards these cells are gradually restricted to the cranial sutures where they remain throughout adulthood (Zhao et al., 2015). Axin2 was previously found to be expressed in the midline of sagittal suture skeletogenic mesenchyme. In a more recent study, Axin2 expression was reported in all calvarial sutures, co-localizing with slow cycling cells that retained EdU (Maruyama et al., 2016). Unlike the Gli1+ cells, no obvious mention or visual evidence of early expression in the dura or periosteum is seen. In fact, Axin2 expression is reportedly restricted to the midline of the sutural mesenchyme from postnatal day 10, whereas Gli1+ tracing seems to target a larger area within the suture. Wilk et al. (2017) investigated postnatal Prx1+ cells and found them exclusively in the calvarial sutures from 2 weeks of age until late adulthood. Like Axin2-expressing cells, Prx1+ were also not found in the periosteum or dura mater, although no mention or visual evidence of early postnatal expression is given (as in Zhao et al., 2015). Overall, the markers described seem to identify cells in a common niche. Whether the populations overlap in identity and function, or represent distinct functional subsets of a main stem cell source is still under debate as further discussed below. No expression of the above mentioned markers was detected in the posterior frontal suture, which likely correlates with the fact that this is the only calvarial suture fused at that stage in mouse development.

As previously mentioned in this review, self-renewal ability and in vivo multipotency are the cardinal requirements of a bone fide stem cell. While the in vivo differentiation potential can be assessed in many ways, some authors argue that self-renewal should only be assessed via heterotopic transplantation of non-induced cultures (Bianco and Robey, 2015). A typical SSC after being cultured and clonally isolated should still be able to generate all the skeletal compartments (multipotency) and “a cell compartment anatomically, phenotypically and functionally equivalent to the one originally explanted” (self-renewal) (Bianco and Robey, 2015). This remains the mainstay of MSC investigation criteria, as it honors the concept of stem cell autonomy in making, without the need for any artificial cues, organized skeletal tissue, including a reservoir of stem cells. In the case of calvarial stem cells, this would mean the ability to make in vivo, after culture and clonal isolation, all the components of a membranous bone and a mesenchymal equivalent to the cranial sutures.

Although heterotopic transplantation of putative calvarial SSCs was not the approach taken, the evidence presented in the studies reviewed herein strongly supports the ability of the putative stem cells to generate all calvarial compartments in vivo. Firstly, single pulse labeling (induced by tamoxifen in Gli1-CreERT2 and Prx1-creER mice, or doxycycline in Axin2-cre-Dox) provides that only the cells initially labeled within the sutural mesenchyme and their progeny are being tracked (Zhao et al., 2015; Maruyama et al., 2016; Wilk et al., 2017). In all studies, the co-expression of osteogenic markers like runx2, osterix, collagen type 1, osteocalcin and sclerostin was not observed in the labeled suture mesenchyme, confirming that the proposed stem cell populations were undoubtedly distinct from any postnatal preosteoblasts and mature osteoblasts. Gli1+ cells were permanently labeled at 1 month of age and found increasingly abundant in the suture mesenchyme, periosteum, dura mater and parts of the calvarial bones up to 8 months of age (Zhao et al., 2015). In a similar manner, derivatives of the Axin2+ cells permanently labeled at 1 month, accumulated continuously in the sagittal suture until after 1 year of development, with a small number of cells found embedded in the bone matrix near the osteogenic fronts. However, this study did not report any contributions of the cells in question to the mature bone and surrounding tissues (dura mater and periosteum). Conversely, a large contribution of Prx1+ cells to all calvarial tissues, except bone marrow osteoblasts, was observed in 6–8 week old mice (Wilk et al., 2017). Prx1 expression is detected as early as embryonic day 15, if not earlier, in the developing skull (Ouyang et al., 2013). Unfortunately, the conclusions about postnatal Prx1+ contribution to adult tissues were drawn from constitutive Prx1-cre lineage tracing analysis. This could include, although unlikely, cells that were specified immediately before the formation, or turn-over of the calvarial components analyzed, without definitive proof that they came from the sutural niche. Moreover, although Zhao and Wilk make no specific claims about self-renewal properties of the stem cell populations reported, Maruyama goes so far as to say that the accumulation of Axin2+ cells within the 1 year period reflects a self-renewal capability of those cells (Maruyama et al., 2016); however, this could also be due to a high proliferation capacity (Bianco and Robey, 2015). Nonetheless, Maruyama provides rigorous and more compelling evidence of in vivo stem cell behavior by transplanting isolated multi-colored (R26R^Confetti) Axin2⁺ cells to an extra-skeletal site (kidney capsule), which the other authors did not attempt. Ectopic bones labeled by single fluorescent colors were observed 3 weeks after transplantation, indicating that these cells can clonally expand in vivo and differentiate autonomously into the osteogenic lineage (Maruyama et al., 2016). However, contrary to bone marrow residing skeletal stem cells, Axin2⁺ SCs from the suture were not able to generate cartilage, unless artificially stimulated by BMP2, which drifts away from the principle of non-cued multipotency toward skeletal lineages expected from bona fide SSCs. While this could be a specific feature of membranous bone residing stem cells, ectopic chondrogenesis has been previously reported in the sutural mesenchyme, suggesting that they possess the intrinsic ability to differentiate into the chondrogenic lineage (Grigoriadis et al., 1988; Maruyama et al., 2011).

Aside from rigorously established criteria, MSCs are popularly defined by their in vitro culture properties, including the ability to differentiate into various cell types (osteoblasts, chondrocytes, adipocytes and, arguably, neurons) (Keating, 2012). Likewise, although with considerable disagreement within the stem cell field, MSCs are loosely defined based on the expression profile of numerous surface markers (Dominici et al., 2006). Typical MSC markers CD90, CD73, CD44, Sca1, and CD146 were not detected in the majority of Gli1⁺ cells in vivo. However, after 1 week of culture, fluorescent-activated Gli1⁺ sorted cells express high levels of such MSC markers. Gli1⁺ cells are also clonogenic in vitro and subcultured clones were successfully differentiated into osteo-, chondro-, and adipogenic lineages under appropriate conditions (Zhao et al., 2015). With the exception of the recently reported Leptin receptor (Zhou et al., 2014), none of the bone marrow SSC markers investigated by Maruyama was distinctively expressed in the Axin2⁺ cells of the suture (Mcam/CD146, Nestin and Gremlin1). Moreover, the in vitro differentiation ability of these cells was not assessed in this particular study. Axin2⁺ isolated cells were, nevertheless, able to form colonies in vitro and to differentiate, after heterotopic transplantation, into the osteogenic lineage and into chondrogenic, although only upon external stimulation (BMP2), as previously mentioned (Maruyama et al., 2016). Postnatal Prx1⁺ cells are also clonogenic and exhibit a SSC-like profile ex vivo, including: upregulation of SSC markers Pdgfrα and Mcam/CD146; downregulation of cell cycle and DNA replication markers Ccne2, Mcm4, and Pcna—typical of quiescent cells; and high levels of transcripts associated with stem cell homing (Itga2, Itga3, and Itga6). Upon in vitro and in vivo stimulation with recombinant WNT3A, these cells are pushed toward osteogenic differentiation, judging by upregulated Osx and Col1 and reduced expression of chondrogenic markers Sox9 and Col2 (Wilk et al., 2017). Altogether, the variable approaches, contrasting profiles and requirement of cueing factors for multi-differentiation in these studies points to the unreliability of loosely defined criteria to identify SSCs, based on which virtually every tissue can provide “stem cell” equivalents.

The notion that skeletal stem cells work as postnatal contributors for tissue regeneration represents not only a general understanding of the organ physiology, but the very motivation for cautious and rigorous investigation of the stem cell populations, aiming at their translational potential. Although classically established requirements for SSCs do not include their regenerative properties, more than it does their multi-lineage differentiation ability in the context of organogenesis and postnatal turn-over, investigating the role of these cells during injury repair is essential, as the pursuit for stem cell-based therapies demand such knowledge. Therefore, the mechanisms of SSC recruitment and contribution to calvarial wound repair should accompany the research of any reported populations, as the three current studies highlighted here have done well to address. Gli1⁺ cells respond immediately to injury in the suture by rapidly activating proliferation within 24 h post incision at the sagittal midline. Contribution of Gli1⁺ cells to calvarial repair is observed in the vast infiltration of conditionally labeled cells into a 1 mm wide parietal wound situated 2 mm away from the sagittal suture. Two weeks after the injury is made, the majority of cells in the area are strongly labeled and, at 1 month, the repaired bone is mostly composed of Gli1⁺ osteocytes as well as positively labeled periosteum and dura mater. Orthotopic transplantation of a lineage traced calvarial bone piece containing sagittal suture into a 4mm defect, revealed a significant number of labeled cells within 1 month-regenerated periosteum, dura mater and bone, whereas an explant lacking the suture failed to generate any of these components. Even with intact dura mater and periosteum, calvarial explants devoid of suture were incapable of regenerating the wounded bone in the recipient mice. The results indicate that, contrary to the previous idea that progenitors involved in cranial repair mainly resided in the periosteum, the sutural mesenchyme and not the surrounding membranes possesses the regenerative capacity (Zhao et al., 2015). Similarly to the Gli1+ population, Axin2+ lineage traced cells here drastically expanded in the sutural mesenchyme in response to a 1.4 mm parietal bone defect. Four weeks after the surgical procedure, large infiltration of the injury site by Axin2⁺ cells is observed and co-localization with OSX (osteoprogenitors) and SOST (osteocytes) expression strongly indicates direct contribution to bone repair. Axin2⁺ cells were also shown to improve injury repair when directly transplanted into the wound site. While Axin2 negative cells from the suture mesenchyme provide no better healing than that seen in non-transplanted injuries, Axin2⁺ cell transplantation significantly increased the repaired area 2 and 4 weeks after surgery. As in the previous experiment, the lineage labeling overlapped with markers of osteoprogenitors and osteocytes, confirming direct contribution of the Axin2+ population to repair of the wounded area (Maruyama et al., 2016).

Finally, Prx1⁺ cells actively contributed to regeneration of subcritical size defects in both frontal and parietal bones, although in this study a smaller 100 μm wide injury was made. Nevertheless, lineage traced Prx1⁺ cells were increasingly present in the wound area 5, 10, and 30 days post-surgery, when they co-localized to osteoblasts and osteocytes embedded in the newly formed bone. Moreover, the regeneration of, questionably, critical-sized (2 mm) defects was greatly improved by heterotopic transplantation of minced sutures into the wound, where Prx1 labeling was seen in the majority of cells integrating the repaired bone 4 weeks after surgery. Interestingly, the repair of parietal subcritical defect did not occur in suturectomized mice after 4 weeks, whereas the 0.5 mm wound is fully repaired in mice with intact sagittal and coronal sutures (Wilk et al., 2017). Taken together, these studies confirm the unequivocal and potentially exclusive contribution of the sutural mesenchyme to calvarial injury repair.

Overall, the studies described provide compelling evidence for direct (by providing osteoprogenitors) as well as indirect (via paracrine signaling) contribution of the putative stem cell populations to calvarial repair. Predominant infiltration of Gli1⁺/Axin2⁺/Prx1⁺ cells is seen in all the subcritical size defect assays, whereby the healing process should be accomplished by natural mechanisms occurring during normal bone physiology. Moreover, transplantation of these cells to the wound site showed significant repair improvement in all the attempts described in the previous section, supporting the translational potential of calvarial SSCs in a number of clinical scenarios. In two of the studies, notwithstanding, the authors found it appropriate to investigate whether the stem cell population described was indispensable for skull development, homeostasis and injury repair. While this is not a particular requirement for determining a resident stem cell, the findings provide valuable information on the ontogeny of the sutural niche and other calvarial compartments, as well as potential mechanisms of congenital disorders like craniosynostosis. Conditional ablation of Gli1⁺ cells, for instance, led to severe craniofacial phenotypes in Gli1-Cre^ERT2; DTA^flox/flox mice (diphtheria toxin A). When ablation was induced at 1 month of age, fusion of ordinarily patent coronal and frontal-premaxilla sutures was observed after 1 month, leading to general reduction of DTA-mice body size at 2 months post-induction. At this stage, all the craniofacial sutures were fused and the bones of Gli1⁺ ablated mice presented severe osteoporosis. Repair of 1mm parietal defects was also compromised in DTA mice when compared to fully regenerated controls. Altogether, the results indicate that Gli1⁺ cells are largely required for craniofacial bone turnover and injury healing. Moreover, pathological fusion of the cranial sutures in Gli1-ablated mice suggests that a deficient supply of undifferentiated mesenchymal cells hinders the patency expected in calvarial sutures of adult mice (Zhao et al., 2015).

Two different strategies confirmed Prx1⁺ SSCs requirement for injury repair. Firstly, 8 week-old Prx1-cre^ER; Rosa26^DTA/+ mice were used to perform global ablation of Prx1⁺ cells during 5 days before and 5 days after 0.5 mm parietal wound surgery. Twenty 8 days later, micro-computed tomography and histological analysis showed drastic healing impairment in Prx1⁺ ablated mice against complete regeneration in the controls. Even in mice injured 2 months after ablation, parietal defects failed to regenerate as non-ablated mice. Secondly, excision of sagittal and right coronal sutures yielded non-regenerated wound in the right parietal bone, whereas removal of the opposing sutures (frontal and left coronal) did not affect the healing of parietal subcritical defect. This strongly suggests that calvarial bones depend on the SSCs of adjacent sutures to perform injury repair. Unlike Gli1⁺ SSCs, however, global ablation of postnatal Prx1⁺ cells did not result in craniosynostosis or any other major craniofacial phenotype. Indeed, Gli1 expression is thought to identify a broadly distributed population of stem cells in the suture mesenchyme, of which Prx1+ cells constitute a subset seemingly dispensable for postnatal development and turnover. Ablation of embryonic Prx1+ cells leads to incomplete calvarial bone formation, suggesting that this specific subpopulation might only be required in earlier stages of development (Wilk et al., 2017). Whereas Maruyama did not perform any Axin2⁺ SSC ablation studies, Axin2-null mice are known to exhibit premature fusion of the cranial sutures (Yu et al., 2005), although this does not necessarily correlate with any particular requirement of the population concerned. Reduction of Axin2-expressing cells is seen upon ossification of the frontal suture (Maruyama et al., 2016). However, this is likely an effect on the entire undifferentiated mesenchyme at the interfrontal suture as it commits to the osteogenic lineage and promotes the fusion of the osteogenic fronts.

Throughout the skeleton, as a local niche is established, mesenchymal stem cells are subject to recruitment, proliferation, induction of mitotic quiescence and lineage commitment. Over the years, extensive research of bone marrow SSCs has shown that these steps are regulated by largely convoluted signaling pathways including: platelet-derived growth factor (PDGF) during perivascular recruitment (Sacchetti et al., 2007); parathyroid hormone (PTH) during expansion of the bone marrow stroma (Kuznetsov et al., 2004); TGF-β1 as the cells are maintained in a quiescent state (reviewed in Pepper, 1997; Jain, 2003) and extensive crosstalk between TGFβ/BMP, Wnt/β-catenin, FGF and Hedgehog pathways, which generally control MSCs lineage fate decision (reviewed in Cook and Genever, 2013). Given the complexity of the network necessary for developing and maintaining a stem cell niche, it is no surprise that regulatory signaling receives little attention in recent papers that attempt to describe calvarial SSCs. In fact, most of the preliminary evidence regarding regulation of calvarial SSCs arose from in vitro or unconvincing in vivo assays that were based on mechanisms previously established for stem cells in the long bones. Zhao et al. set out to investigate whether Gli1⁺ cells were, like BM-SSCs, regulated by hedgehog signaling in the context of lineage commitment. No Sonic hedgehog (Shh) signal was detected in the suture region using a R26Tdtomato reporter. However, Indian hedgehog (Ihh), a regulator of skeletal development, was found in cells of the osteogenic fronts flanking the cranial sutures. Deletion of the hedgehog receptor Smoothened (Smo) in Gli1⁺ cells did not affect suture patency, proliferation of Gli1⁺ cells, or their ability to generate periosteum, dura and osteocytes; nor did it induce apoptosis in the sagittal suture. Nonetheless, all the craniofacial bones in SmoKO^Gli1−cre mice exhibited severe osteoporosis and reduced bone volume after 8 months. In vitro analysis of sutural mesenchyme cells revealed up-regulated Gli1 expression, increased osteogenic activity and upregulation of osteodifferentiation markers upon IHH treatment, whereas the hedgehog inhibitor GDC0449 induced the opposite effects. Neither treatment affected proliferation or apoptosis of cultured mesenchymal cells, suggesting that Hedgehog signaling might be important for regulation of Gli1⁺ SSC differentiation, but not their maintenance (Zhao et al., 2015). Maruyama and Wilk's insights on regulation of Axin2⁺ and Prx1+ cells, respectively, are merely suggestive, as these cells present in vivo expression of canonical Wnt inhibitors in their quiescent state and lineage fate decision changes upon exogenous activation of Wnt or BMP signaling. Transplanted Axin2⁺ cells were able to generate ectopic mineralized bone, but no cartilage after 7 days, contrasting the multi-lineage potential of long bone marrow SSCs. However, exogenous stimulation with BMP2 shifted Axin2⁺ cells from an osteogenic to a chondrogenic fate, shown by detection of Alcian blue (cartilage), but not von Kossa (bone) staining in the structure generated (Maruyama et al., 2016). Expression of Wnt inhibitors Dkk1 and Sost by Prx1⁺ sorted cells supports the well-established idea that skeletal progenitors are maintained in an undifferentiated status, or yield chondrogenic differentiation in the absence of canonical Wnt activation (Day et al., 2005; Hill et al., 2005). Indeed, in vivo expression of chondrogenic markers Sox9 and Col2 was diminished in pnPRX1⁺ cells upon subperiosteal injection of Wnt agonist WNT3A to the cranial sutures, whereas osteogenic markers Osx and Col1 were significantly upregulated. Commitment to the osteogenic fate has also observed when these cells were treated in culture with recombinant WNT3A, endorsing the previously described role of Wnt/β-catenin pathway in differentiation of mesenchymal progenitors (Wilk et al., 2017). Altogether, recently published studies provide informative, yet inconclusive findings to characterize molecular regulation of calvarial SSCs, leaving exciting open prospects for future stem cell research.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer FW and handling Editor declared their shared affiliation.