The dermal papilla dilemma and potential breakthroughs in bioengineering hair follicles

Abstract

The generation and growing of de novo hair follicles is the most daring hair replacement approach to treat alopecia. This approach has been explored at least since the 1960s without major success. Latest in the 1980s, the realization that the mesenchymal compartment of hair follicles, the dermal papilla (DP), is the crucial signaling center and element required for fulfilling this vision of hair follicle engineering, propelled research into the fibroblasts that occupy the DP. However, working with DP fibroblasts has been stubbornly frustrating. Decades of work in understanding the nature of DP fibroblasts in vitro and in vivo have led to the appreciation that hair follicle biology is complex, and the dermal papilla is an enigma. Functional DP fibroblasts tend to aggregate in 2D culture, while impaired DP cells do not. This fact has stimulated recent approaches to overcome the hurdles to DP cell culture by mimicking their natural habitat, such as growing DP fibroblasts in three dimensions (3D) by their self-aggregation, adopting 3D matrix scaffold, or bioprinting 3D microstructures. Furthermore, including keratinocytes in the mix to form hair follicle-like composite structures has been explored but remains a far cry from a useful and affordable method to generate human hair follicles in sufficient quantity and quality in a practical time frame for patients. This suggests that the current strategies may have reached their limitations in achieving successful hair follicle bioengineering for clinical applications. Novel approaches are required to overcome these barriers, such as focusing on embryonic cell types and processes in combination with emerging techniques.

Introduction

Hair follicles are complex mini-organs within the skin that undergo constant regenerative cycling, which is driven by epithelial-mesenchymal interactions (EMIs) mainly between keratinocytes in the hair matrix/germinative epithelium and specialized fibroblasts in the dermal papilla (DP). Hair follicles contribute to many physiological functions (Yu et al. 2014), including sensory inputs, protection, wound healing (Stevenson et al. 2008), and thermoregulation, as well as psychological and social well-being (Moattari and Jafferany 2022). In recent decades, hair follicle bioengineering has advanced significantly due to increased demands for effective cures for hair loss and skin injuries. For instance, hair loss and graying, which can be psychologically devastating to individuals, affect a growing number of people worldwide. Approximately 50% of the male population aged 30–50 years exhibits male pattern baldness (Krupa Shankar et al. 2009). At least 10% of the female population is affected as well (Mu et al. 2021). Hair transplantation is often used in the treatment of androgenetic alopecia. However, patients themselves are the source of hair follicles, which apparently is not only painful but limits the supply of healthy hair follicles.

Another significant, clinically relevant example is tissue-engineered skin substitutes (ESS) (Niehues et al. 2018). ESS have delivered substantial medical benefits for patients with acute and large skin losses, such as large-scale burn injuries (Boyce et al. 2017), when using autologous split-thickness skin grafts harvested from non-burned parts of the patient’s body for wound repair is not sufficient or an option (Wainwright 2009). However, ESS have compromised physiological and mechanical functions due to missing hair follicles and other skin appendages (Castro and Logarinho 2020). Therefore, new research efforts are being made to produce ESS containing cutaneous appendages, which is important for patients’ long-term quality of life.

The DP is the core signaling center of hair follicle growth and regeneration

As shown in Fig. 1, the hair follicle is rebuilt by hair follicle stem cells (HFSCs) located in the hair bulge region during each hair cycle. HFSCs are slow cycling, possess the ability to self-renew, and become activated when elicited by multilayered signals from the surrounding macroenvironment, i.e., the DP (Fuchs 2009). The dermal sheath (DS), a connective tissue sheath, lines the epithelium of the hair follicle from the bulge downward and connects with the DP at the base of the hair follicle. The cells within the DP and DS are highly specialized fibroblasts of mesenchymal origin, collectively forming a dermal macroenvironment encircling the epithelial compartment (Yang and Cotsarelis 2010). DP fibroblasts display unique characteristics, such as aggregative behavior and specific gene expression, and most importantly, they have remarkable roles in hair follicle morphogenesis and regenerative cycling (Botchkarev and Kishimoto 2003; Jahoda and Oliver 1984a; Messenger 1984). In addition, it was shown that DP fibroblasts can potentially be regenerated in vivo by DS fibroblasts (Rahmani et al. 2014). To induce hair follicle formation, DP fibroblasts or fibroblasts with DP characteristics are needed for the onset of morphogenetic events by influencing the fate decision in keratinocytes, which will switch from making an epidermis to making a hair follicle. During hair follicle regenerative cycling, DP fibroblasts are indispensable for the initiation and maintenance of a series of essential EMIs with HFSCs, secondary hair germ cells, and matrix keratinocytes, which are normally mediated by diffusible molecules, including WNTs (Enshell-Seijffers et al. 2010), FGFs (Greco et al. 2009; Harshuk-Shabso et al. 2020), TGF-β/BMPs (Andl et al. 2004; Oshimori and Fuchs 2012), and SHH (Liu et al. 2022). Some of those molecular signals are essential for maintaining DP hair inductivity, including BMP signaling (Rendl et al. 2008) and SHH signaling (Woo et al. 2012). Theoretically, gaining an improved understanding of the interactions among these DP signaling pathways can shed light on the creation of a supportive dermal environment and the maintenance of DP functionalities to promote hair follicle bioengineering in vitro and in vivo.

Fig. 1.

Illustration of hair follicle structure. The hair follicle contains the epithelial compartment, which is composed of keratinocytes, and the mesenchymal compartment, which includes the DS lining the epithelium of the hair follicle and the DP at the base of the hair follicle. The cells within the DP and DS are highly specialized fibroblasts of mesenchymal origin. HFSCs located at the hair bulge region are responsible for hair follicle regenerative cycling

Historically, the recognition of the importance of DP fibroblasts for hair and feather growth can be traced back to Lillie and Wang, who demonstrated that the DP possesses a vital morphogenetic and regenerative function in feather follicles (Lillie and Wang 1941, 1944; Wang 1943). If the DP is removed from the follicle, feather production stops. However, if a DP is transplanted, the “truncated” follicles can reinitiate feather growth. Chase et al. performed a detailed analysis of human hair growth and suggested that the DP is the inducer of hair growth at the end of telogen (Chase 1955). Oliver used rat vibrissa follicles as a model and showed that the regeneration of a functional whisker is dependent on rebuilding a new DP first after the removal of the vibrissa follicle DP (Oliver 1966a, b, c). When isolated DPs were implanted into transected whisker follicles, whiskers regenerated, strongly suggesting that the DP is indeed the inductive center for whisker growth (Oliver 1967). Later, Oliver further showed that vibrissa DPs can induce the formation of hair follicles and hair growth, even in the epidermis (Oliver 1970). In 1981, Jahoda and Oliver reported in vitro culture of rat DP fibroblasts isolated from vibrissa follicles (Jahoda and Oliver 1981), which could be used to rescue the growth of hair follicles “from which the lower halves had been removed” (Jahoda et al. 1984). These data showed for the first time that DP fibroblasts can potentially maintain their inductive capacity in vitro and can be used in vivo if the correct conditions are met. In 1984, Messenger first reported the successful isolation and culture of DP fibroblasts from human hair follicles (Messenger 1984). Since then, human DP fibroblasts have become a focal point of research in hair follicle biology and practical applications (Yang and Cotsarelis 2010).

DP fibroblast bottleneck in hair follicle bioengineering

Human dermal fibroblasts and keratinocytes do not lose the memory for initiating required EMIs and can self-assemble a bilayered skin construct after ex vivo culture as shown by the success of constructing different types of ESS (Fig. 2). It was described in the 1970s that keratinocytes by themselves can terminally differentiate in vitro without inputs from fibroblasts and that cultured epithelial cells can produce a stratified squamous epithelium without fibroblasts (Green 1977; Lillie et al. 1980) although intercellular crosstalk between epidermal keratinocytes and dermal fibroblasts is still needed for adequate skin differentiation and the development of a functional skin barrier (Jevtic et al. 2020).

Fig. 2.

Keratinocytes interact with dermal fibroblasts and dermal papilla fibroblasts to generate the epidermis and hair follicle, respectively. Human keratinocytes and dermal fibroblasts self-assemble to form a skin structure; however, keratinocytes and DP fibroblasts fail to form a HF due to long-lasting bottlenecks

However, it is not as simple to form a hair follicle from epidermal keratinocytes and DP fibroblasts. Previously, Zhang et al. showed that human epidermal and dermal cells prepared from the fetal scalp and foreskin tissues after in vitro expansion could produce mature hair follicles in the patch assay (Zhang et al. 2017). Nevertheless, such efforts to generate human hair follicles from isolated adult patient cells have not been successful on a large and consistent basis owing to inadequate manipulation of isolated DP fibroblasts ex vivo.

The hurdles are that we still do not have an accurate understanding as to how to expand hair-inducing DP fibroblasts ex vivo while maintaining their functional properties, which of the DP-associated signature genes are relevant for their ex vivo expansion, and how to use the existing knowledge for hair follicle bioengineering. First, the isolation of DP fibroblasts from human hair follicles involves complicated techniques, such as microdissection, which are proven to be technically inefficient (Gledhill et al. 2013; Limbu and Higgins 2020). Second, DP fibroblasts proliferate slowly in ex vivo culture compared to dermal fibroblasts, and generating a sufficient number of cells needed for hair follicle regeneration is difficult (Warren et al. 1992). Third, it has been repeatedly demonstrated that fibroblasts isolated from microdissected human DPs quickly lose hair inductivity during ex vivo expansion and fail to induce hair follicle neogenesis (Horne et al. 1986; Jahoda et al. 1984), possibly due to the loss of microenvironmental cues and EMIs. Kang et al. showed that human DP fibroblasts in two-dimensional (2D) culture were unable to induce mouse epidermal cells to form hair follicles (Kang et al. 2012). Higgins et al. performed a global gene expression analysis of human DP fibroblasts in 2D and three-dimensional (3D) cultures and discovered that the cells quickly lost DP molecular signatures in 2D cultures, which is consistent with the early loss of hair-inducing capacity (Higgins et al. 2013).

Novel approaches are needed to create a scalable in vitro system that can faithfully replicate the in vivo DP microenvironment. Such a system can not only amplify DP fibroblasts but also simultaneously maintain, restore, or even enhance their trichogenicity. One way to achieve this goal is to replenish epithelial cues that are generated in the epithelial compartment of a hair follicle. Inamatsu et al. cultured rat vibrissa DP cells with rat keratinocytes or medium conditioned by keratinocytes and found that rat DP fibroblasts sustained hair follicle inductive ability even after 70 passages at a level that is comparable to that of original DP fibroblasts (Inamatsu et al. 1998). Although an astonishing feat, similar results have unfortunately not been reported elsewhere. Abreu et al. showed that the characteristics and inductive phenotype of human DP fibroblasts isolated from scalp hair follicles can be partially restored by conditioned medium collected from interfollicular keratinocytes (Abreu et al. 2021). The EMI signals may also be supplied, at least in part, by diffusible signaling molecules that are known to function in hair follicles, including Wnt and BMP proteins. Wnt3A, Wnt7A, Wnt10B, and BMP6 have been shown to prolong hair inductivity in cultured mouse DP cells (Kishimoto et al. 2000; Ouji et al. 2012; Rendl et al. 2008).

Another option to maintain and restore DP inductivity is to culture human DP fibroblasts in 3D architecture akin to its in vivo conformation. The feasibility of this approach has been demonstrated by an array of interesting studies in recent years. Higgins et al. reported that intact 3D DPs isolated from human adult scalp hair follicles were able to induce hair follicle neogenesis in nonhair-bearing neonatal human foreskin at an average success rate of 56%, while 2D DP cell “slurries” had no success at all (Higgins et al. 2013). Due to its potential, this review focuses on a variety of 3D model systems that have been attempted to propagate human DP fibroblasts and maintain their inductive capabilities for hair follicle formation.

The 3D cell aggregation model

Initial observations by Messenger et al. reported that human DP cells display a unique morphology in 2D culture, appear polygonal at the stationary growth phase, and tend to form aggregates when reaching confluency (Messenger et al. 1986). Considering that DP fibroblasts assume an aggregative conformation mixed with fibronectin-, collagen IV- and laminin-rich extracellular matrix (ECM) in the DP in vivo (Jahoda et al. 1992; Messenger et al. 1991; Tobin 1992; Young 1980) and tend to self-aggregate in vitro, it can be assumed that DP fibroblasts have an inherent tendency to replicate their natural habitat by self-organization. Providing the ability to form a 3D conformation that structurally resembles a DP may allow DP cells to better maintain their DPness in vitro (Betriu et al. 2020; Higgins et al. 2013; Lin et al. 2016; Miao et al. 2014; Topouzi et al. 2017). Many studies have utilized the self-organizing abilities of DP fibroblasts to promote 3D culture for hair follicle induction. Osada et al. compared the hair inductivity between cultured DP fibroblasts dissociated from mouse vibrissae and their aggregates in the form of spheres at different passages (Osada et al. 2007). Surprisingly, they found that cultured mouse DP cells past passage 4 could no longer induce hair follicles, but spheres generated by passage 10 cells were still able to induce new hair follicles. Qiao et al. showed that aggregates formed by dermal and epidermal cells crudely isolated from embryonic day 18 (E18) mouse skin formed hair-like structures in vitro and developed into mature hair follicles after being implanted intradermally into nude mice (Qiao et al. 2008b).

Higgins et al. cultured human DP fibroblasts isolated from scalp hair follicles using the hanging drop method and found that the cells aggregated and formed spherical structures within 20 h. These spheroids showed similarities to in vivo DPs and retained AP activity and the expression of many DP signature genes, while cells grown in 2D lost them (Higgins et al. 2010). Vahav et al. constructed DP spheroids by allowing adult human scalp DP fibroblasts to self-assemble in ultra-low-attachment plates. The DP spheroids resembled human DPs in size and gene expression and expressed the same characteristic DP ECM proteins, including chondroitin sulfate, collagen IV, laminin V, and fibronectin. The formed DP spheroids were then incorporated into human ESS and cultured for 10–14 days. Interestingly, DP spheroids stimulated epidermal down-growth, which engulfed DP spheres and produced immature hair follicle structures (Vahav et al. 2020). Lin et al. cultured high-passage DP fibroblasts isolated from human scalp hair follicles in a 3D hanging drop system and found that this culture method restored DP fibroblasts at passage 8 to a state comparable to passage 2 DP cells (Lin et al. 2016).

Because the DP is encompassed by keratinocytes in the hair bulb region and interacts reciprocally with them to promote hair growth, it is plausible that the hair-inducing capacity of DP fibroblasts could be greatly improved if keratinocytes and DP fibroblasts form 3D spheres together to restore essential EMIs (Leng et al. 2020). Co-cultured spheroids can be formed using two major methods, i.e., a “mixed” or a “layered” spheroid. In the “mixed” model, DP fibroblasts and keratinocytes are mixed to form aggregates. Such mixtures of keratinocytes and fibroblasts isolated from embryonic or adult rodent skin can form aggregates using either the hanging drop method or low-attachment plates and retain their ability to induce hair follicle neogenesis (Asakawa et al. 2012; Balana et al. 2015; Guo et al. 2019; Lei et al. 2017; Qiao et al. 2008b; Toyoshima et al. 2012; Yen et al. 2010). Lin’s group reported that rat keratinocytes and DP fibroblasts could self-assemble into spheroidal microtissues on biomaterials and facilitate EMIs for hair formation in vivo (Yen et al. 2010; Young et al. 2008, 2009). Different types of keratinocytes isolated from skin and hair follicles, including HFSCs, outer root sheath (ORS) keratinocytes, and skin epidermal keratinocytes, were also tested together with human adult scalp DP fibroblasts for hair follicle induction (Atac et al. 2020; Havlickova et al. 2009; Jang et al. 2020; Kalabusheva et al. 2017; Lindner et al. 2011; Tan et al. 2019). Kalabusheva et al. used skin epidermal keratinocytes with adult human scalp DP fibroblasts in a hanging drop culture and found the formation of hair germ structures (Kalabusheva et al. 2017). Jang et al. reported that adult human ORS keratinocytes cultured together with human DP fibroblasts in ultra-low attachment plates developed a “polar elongated structure,” which is similar to the hair germ structure (Jang et al. 2020).

The “layered” model is built by adding human keratinocytes to form a shell wrapping the DP fibroblast “core”. Tan et al. layered immortalized human scalp DP fibroblasts, HaCaT keratinocytes, and dermal fibroblasts sequentially in 3D microwells fabricated from polyethylene glycol diacrylate hydrogel and found that the expression of DP signature genes was better preserved compared with 2D culture. HaCaT cells formed the outer layer, while DP fibroblasts formed a core in the aggregates, mimicking the normal spatial organization among these cells (Tan et al. 2019). Atac et al. and Lindner et al. reported that they created layered hair-producing microfollicles in vitro by co-cultivating DP spheroids formed by adult human DP fibroblasts with follicular ORS keratinocytes and melanocytes isolated from human scalp samples and foreskin (Atac et al. 2020; Lindner et al. 2011). Both groups reported the formation of hair follicle-like structures in vitro, which expressed specific lineage markers of hair follicle keratinocytes, DP fibroblasts, and melanocytes. Hair-like fibers began to form in both studies but only in a small fraction of organoids.

One reason that explains the failure to form a hair shaft in the vast majority of organoids generated by various groups is likely the failure to form a real and highly proliferative matrix cell compartment. The epithelial portion in those organoids exhibits a confused and mixed differentiation that does not correlate well with the in vivo situation. One of the most detailed studies is the study reported by Ataç et al., in which they showed that the cells adjacent to the DP-like structure are non-proliferating and KRT15 positive and therefore have little resemblance to matrix cells. This suggests that essential EMIs are missing and matrix cell differentiation does not occur (Atac et al. 2020). In 2021, Fukuyama et al. reported a new method to reconstitute hair follicle-like structures from DP fibroblast and keratinocyte aggregates in vitro (Fukuyama et al. 2021). They first seeded DP fibroblasts in ultra-low attachment plates to form aggregates and then placed the aggregates in Matrigel on a cell culture insert. Under a microdissection microscope, a highly condensed keratinocyte solution was cylindrically injected above DP aggregates using a micropipette to form a DP-keratinocyte hair follicle-like aggregate. Afterward, a nylon fiber was inserted into the keratinocyte compartment that functioned as a structural support and a guide for the formation of a hair follicle-like construct. After 14 days of culture, a human hair follicle-like structure with a DP aggregate at the bottom and three histologically distinct epithelial layers, including the outer root sheath, inner root sheath, and keratin 40-positive hair shaft, was formed. However, it appears that this new method still needs to be further improved for clinical application.

The 3D scaffold model

The critical roles of DP fibroblasts in hair follicle induction are regulated by both inherent DP characteristics and the external inductive microenvironment. In vitro, DP fibroblasts self-organize to form tight aggregates partly due to their abilities to produce ECM molecules, e.g., collagens and proteoglycans, which provide both structural support and molecular signals to DP fibroblasts and surrounding keratinocytes and facilitate EMIs. However, the exact molecular nature that drives DP self-aggregation is unknown (Jahoda and Oliver 1984b), even though 3D architecture maintains “good” DP cells better than 2D culture. As such, 3D scaffolds can theoretically provide a better environment for DP fibroblasts to survive, proliferate, and form a DP conformation that maintains hair inductivity. Another advantage of adopting complex multi cell-type 3D scaffolds is the achievement of active reciprocal communication among keratinocytes, DP fibroblasts, and the ECM niche. Therefore, efforts have been made to use different types of scaffolds to stimulate the DP aggregation process and hair follicle formation.

The ESS can provide a 3D regenerative environment that allows skin formation and hair follicle morphogenesis to occur at the same time. Sriwiriyanont et al. inoculated human foreskin dermal fibroblasts, mouse DP fibroblasts, and human foreskin keratinocytes sequentially into collagen-glycosaminoglycan (GAG) scaffolds. Because the DP ECM is rich in glycosaminoglycans, CAG scaffolds theoretically provide an improved microenvironment for DP fibroblasts and hair follicle induction. The scaffolds with inoculated cells were then incubated for nine days at the air–liquid interface and grafted onto nude mice. Pigmented hairs were formed in chimeric skin substitutes without sebaceous glands (Sriwiriyanont et al. 2013). Wang et al. also tested collagen-GAG scaffolds and inoculated human epidermal stem cells and mouse skin-derived precursors (SKPs). The inoculated scaffolds were incubated for 2 h at 37°C before being grafted onto nude mice. Black hairs were generated within three weeks (Wang et al. 2021). Interestingly, Casale et al. reported an endogenous human skin equivalent (Endo-HSE) in which fibroblasts were placed onto a temporary gelatin scaffold (Casale et al. 2016). The embedded fibroblasts were allowed to expand and form a new ECM scaffold to replace the original gelatin scaffold. Keratinocytes were plated onto Endo-HSE containing the newly assembled ECM and formed a hair follicle-like structure in vitro. However, no thorough characterization of the epithelial invaginations was performed, which would confirm the assertion that the “invasive” structure truly represents the attempt by the artificial skin to make hair follicles.

Natural or synthetic materials can be used to build 3D scaffolds, including Matrigel, collagen, acellular dermal matrices, and human amniotic membranes. Kageyama et al. prepared collagen-embedded cell aggregates, named hair beads (HBs), by mixing human DP fibroblasts in a collagen type IA solution. They found that the expression of alkaline phosphatase (AP) and versican was higher in human DP fibroblasts in HBs than those in “traditional” 3D spheroid culture. HBs containing human DP fibroblasts and mouse epithelial cells also produced more hairs than those in 3D spheroids with mouse epithelial cells in the patch assay, although no statistically significant increase was achieved. An important but often neglected achievement of the work accomplished by Kageyama et al. was not just improved hair inductivity but scalability with the goal of simultaneously producing more than 5000 hair-inducing units (Kageyama et al. 2019). Thangapazham et al. constructed a dermal-epidermal composite (DEC) model system to evaluate the hair inductivity of cultured adult human DP fibroblasts. DECs were built by embedding DP fibroblasts in rat tail collagen type 1 and adding neonatal foreskin keratinocytes on top. DECs were kept at the air–liquid interface for two days before grafting onto nude mice. The results showed human hair follicle neogenesis in grafts using DECs formed by DP fibroblasts expressing a high level of alkaline phosphatase eight weeks after grafting (Thangapazham et al. 2014a, b).

A drawback of using collagen I scaffolds is their poor mechanical properties (Dong and Lv 2016). Crosslinking collagen scaffolds or modifying them using natural/synthetic polymers or inorganic materials, such as chitosan, silk fibroin, hyaluronic acid, and alginate, can increase their mechanical properties (Zhong et al. 2010). Havlickova et al. developed a human folliculoid microsphere (HFM) system to coculture human DP fibroblasts and ORS keratinocytes in a matrix consisting of collagen I and Matrigel, the “magic powder” of organoid biology. For DP fibroblasts, Matrigel seems to be favorable due to its high content of basement matrix components and at least on paper imitates a matrix environment that is closer to the DP niche (Havlickova et al. 2009). They showed that HFMs retained essential hair follicle-specific EMIs, and the cells in the aggregates expressed several major markers of human hair follicles, including K6 for ORS cells and versican for DP cells. The results further demonstrated the advantages of aggregating keratinocytes and DP fibroblasts together in initiating hair-inducing EMIs.

Many other types of 3D scaffolds have also been tested for hair follicle induction. Dong et al. cultured rat DP fibroblasts in a biomimetic silk fibroin/sodium alginate scaffold. The results indicated that the silk fibroin/sodium alginate scaffold allowed rat DP fibroblasts to maintain their normal morphology and characteristics in vitro. Furthermore, it promoted hair follicle neogenesis in vivo (Dong et al. 2021). Lim et al. fabricated 3D fibrous hydrogel scaffolds in which human DP fibroblasts and epidermal keratinocytes were embedded using a chitin solution and sodium alginate. By implanting cell-laden scaffolds in immunodeficient SCID mice, the formation of hair follicle-like structures was observed (Lim et al. 2013). Leiros et al. seeded human scalp DP fibroblasts with HFSCs into the porcine acellular dermal matrix and maintained them at air–liquid interphase for 14 days to generate skin constructs (Leiros et al. 2014). The cultured matrices were then grafted onto a full-thickness wound in nude mice. With DP fibroblasts, neovascular network maturation was improved, and hair bud-like structures, which contained cells of human origin and expressed a hair follicle differentiation marker, type II epithelial keratin 6hf (Wang et al. 2003), were found in the newly formed skin. Wang et al. tested the ability of “bio-functional self-assembling peptide nanofibers” to produce 3D scaffolds for hair follicle neogenesis. The scaffolds were made of peptide RADA16 (Ac-(RADA)4-CONH2) and PRG (Ac-RADARADARADARADAGPRGDSGYRG DSCONH2) (Zhang et al. 1993). They function to mimic the natural ECM and are able to form hydrogels surrounding embedded cells. The scaffolds promoted the proliferation of mouse neonatal epidermal and dermal cells in vitro and de novo hair follicle neogenesis (Wang et al. 2016). These artificial scaffolds may be the preferential material for hair follicle engineering because they can be generated in a controlled and reliable fashion for clinical applications, are animal-free, and can be at least as effective as the gold standard Matrigel (Aisenbrey and Murphy 2020).

3D microstructure model

The hair follicle needs to produce a canal for the hair shaft to emerge at the skin surface after morphogenesis (Mesler et al. 2021). However, the production of the hair shaft and the hair follicle canal is not only the expected outcome of hair follicle formation but also an integral part of hair follicle morphogenesis. Therefore, new attempts are made to guide hair follicle morphogenesis and differentiation, such as creating vertical hair channels in artificial scaffolds before inoculating either cells or aggregates. Leng et al. produced 1- or 2-mm-diameter punch biopsy wounds on the back skin of nude mice and inoculated dissociated epidermal and dermal cells isolated from neonatal mice or human fetal scalp for hair follicle induction. After three months, human follicles were seen in the graft site with 2 mm-size punches on the top of reconstituted full-thickness human skin, albeit the efficiency was very low (Leng et al. 2020).

To produce a large amount of 3D microstructures for studying communication between mesenchymal and epithelial cells in hair follicles, Pan et al. fabricated hair follicle-like microwells in polyethylene (glycol) diacrylate (PEGDA)-based hydrogels using soft lithography. They seeded human dermal fibroblasts and HaCaT keratinocytes in the microwells in a way that mimics mesenchymal and epithelial compartmentalization in native hair follicles and observed cell proliferation and survival over 14 days as well as the spreading of dermal fibroblasts (Pan et al. 2013). A similar approach was taken by Tan et al. using a human dermal papilla cell line. However, they had little success with PEGDA and some feasibility with gelatin methacrylate (GelMA) hydrogels (Tan et al. 2021). Overall, these microstructure approaches are still in their infancy and must prove that they can produce hair follicles.

3D bioprinting is another potentially powerful method to produce hair follicle microstructures in a highly precise manner (Jorgensen et al. 2020; Weng et al. 2021). However, many parameters, including biosafety, biocompatibility, mechanical properties, and stability, need to be considered when biomaterials are selected for bioprinting to ensure that the embedded cells can survive, proliferate, and differentiate. In 2020, Zhou et al. reported that they developed a new approach to rapidly print functional living skin (FLS) using a newly designed GelMA/HA-NB/LAP bioink, which is composed of GelMA, N-(2-aminoethyl)-4-(4-(hydroxymethyl)-2-methoxy-5-nitrosophenoxy) butanamide (NB)–linked hyaluronic acid (HA-NB) and photo-initiator lithium phenyl-2,4,6- trimethylbenzoylphosphinate (LAP), and digital light processing (DLP)-based 3D printing technology (Zhou et al. 2020). The printed skin construct possesses a dense upper layer mimicking the epidermis and a porous lower layer resembling the dermis. Interconnected micro-channels in two layers facilitate cell migration, proliferation, and neo-tissue formation. In vivo, the study demonstrated that hair follicles can be generated from the grafted 3D-printed skin in both rats and pigs. In addition, Kang et al. constructed a multilayer composite scaffold using a 3D bioprinting technique based on a gelatin/alginate hydrogel, which mimics the in vivo hair follicle microenvironment (Kang et al. 2022). In their model, human umbilical vein endothelial cells (HUVECs), mouse fibroblasts, mouse DP fibroblasts, and mouse epidermal cells were printed in different layers in gelatin/alginate hydrogel with different densities and patterns. DP fibroblasts and epidermal cells formed hair follicle-like structures in the scaffolds in vitro after seven days in culture. Unfortunately, no histology was presented in the paper to support their findings. An in vivo skin reconstruction assay also showed that 3D-printed multilayer composite scaffolds developed hair follicles that seemed to be better organized and oriented than those in the control group that used a traditional DP/epidermal cell mixture.

One of the most advanced systems to produce human hair follicles at high density was presented by Abaci et al. (2018). Although they did not bioprint cells or scaffolds, they used 3D-printed silicon molds that contain hair follicle-shaped extensions to produce microwells in a type I collagen matrix containing dermal fibroblasts. Human DP fibroblasts were then added to the base of the microwells for spontaneous aggregation, and keratinocytes were seeded over the dermal construct to fill up the microwells and engulf the DP aggregates. This resulted in crude hair follicle-like structures in vitro, especially when using DP cells that overexpressed LEF1. To produce higher hair follicle density in in vivo transplanted constructs, HUVECs had to be included in the artificial dermis in addition to dermal fibroblasts. HUVECs greatly improved the vascularization of the transplanted construct. Clearly, 3D printing can play an important role in DP-centered hair follicle bioengineering. In particular, the goal of producing 3D microstructure templates with controlled size and spacing of hair follicle structures can be better achieved by 3D printing.

Ex vivo and in vivo skin models

Other possible approaches to improve human hair follicle neogenesis are to utilize ex vivo skin organ culture and in vivo mouse models, which provide a real 3D skin environment that is missing in the aggregates, such as supportive structure or essential ECM. Human skin explant models are not yet widely used, although the system was introduced in the 1960s (Keaven and Cox 1965). Their attractiveness to bridge the translation gap among animal models, in vitro models, and clinical applications is not to be overlooked (Neil et al. 2020). For example, Krugluger et al. injected human DP fibroblasts and ORS keratinocytes together into cultured retroarticular skin specimens for hair follicle induction. In four weeks, vellus-like hairs were observed in the skin specimens, and multiple miniaturized hair follicles with sebaceous glands were found after eight weeks (Krugluger et al. 2005). Qiao et al. developed a flap-graft assay in which cultured mouse epidermis was placed on the underside of a skin flap incised on the back of the mouse, and dermal cells could be applied to the basal surface of the epidermis for hair follicle induction (Qiao et al. 2008a). Using this graft assay, they claimed that DP fibroblasts isolated from human adult scalp hair follicles could be expanded using a revised culture medium recipe while still maintaining the hair-inducing ability (Qiao et al. 2009).

Higgins et al. inoculated aggregates of human DP fibroblasts between the split foreskin dermis and epidermis to make skin grafts. After being incubated in vitro for 24 h, the grafts were sutured onto SCID mice. New hair follicles were found in five out of seven grafts. Among them, the nonpigmented hair shaft protruded from two experimental foreskin tissues six weeks after grafting. No sebaceous glands were seen in spheroid-formed follicles, as sebaceous glands were seen in follicles induced by transplanted intact DP (Higgins et al. 2013). Kageyama et al. and Lin et al. both observed hair follicle neogenesis in nude mouse skin after implanting spheroids formed by human scalp DP fibroblasts together with neonatal or embryonic mouse epidermal cells (Kageyama et al. 2019; Lin et al. 2016). Mi et al. and Su et al. generated cell aggregates from human fetal scalp dermal cells and foreskin epidermal cells and grafted them onto nude mice for a hair reconstitution assay (Mi et al. 2019; Su et al. 2019). In both cases, human hair follicles were formed in vivo. In addition, Ohyama et al. placed human DP fibroblast aggregates between enzymatically separated murine sole epidermis and dermis. The composite grafts were then inserted subcutaneously into nude mice. In the group of aggregates of DP fibroblasts cultured using DP activation culture (DPAC) medium, they observed signs of hair follicle morphogenesis, including epidermal invagination or papillary mesenchymal body formation, in 9 out of 21 grafted mice (Ohyama et al. 2012). Overall, this is a powerful model to test whether in vitro cultured human keratinocytes and DP fibroblasts can be used to treat hair loss.

Conclusion and outlook

Hair follicle morphogenesis and development rely on the proper and timely interplay among epithelial keratinocytes, mesenchymal cells, and the hair follicle niche microenvironment. Consequently, attempts in the lab and clinical studies to just inject DP fibroblasts or similar cell slurs have only resulted in failures (Tsuboi et al. 2020). The DP compartment is known as the only in vivo habitat in which DP fibroblasts can maintain hair inductive capability. The realization that in vivo conformation determines DP functions has resulted in a glut of attempts to grow DP fibroblasts in 3D structures, such as spheroids or aggregates, either by themselves or mixed with keratinocytes. Surprisingly, little is known about the factors that may confer the advantages of 3D growth. In contrast to the developing DP that starts out as a dense cluster of cells, the adult human DP is a relatively loose collective of cells embedded in a highly specialized matrix, which markedly changes during hair growth (Jahoda et al. 1992; Young 1980). These changes have led Sengel to postulate that the collagen-deprived and fibronectin-rich matrix of growing anagen hair follicles resembles “morpho-genetically active” stroma observed during development, which may constitute part of the morphogenetic message that is needed for the development of cutaneous appendages (Mauger et al. 1987). These findings highlight the importance of the 3D ECM microenvironment for DP function and faithfully replicating this environment for bioengineering purposes.

Furthermore, there is a massive gap between taking this laboratory approach and translating it into a workable, scalable production line of active, safe, and inductive DPs for clinical applications. Even in the laboratory, only a minority (approximately 15%) of human DP spheroids possess hair follicle-inducing capacity (Higgins et al. 2013). A related issue involves the cost and time needed to produce human hair follicles for the clinic that fulfills FDA Current Good Manufacturing Practice (CGMP) regulations. A series of companies (dNovo, Follica, Han Bio, Rapunzel, Replicel, Shiseido, Stemson, etc.) have entered the field to tackle these issues and to produce effective cell-based therapies for hair loss but have little to show for thus far.

The current DP-centric dogma is based on the seminal observation that the DP is the driver of hair follicles (Horne et al. 1986) and that the only way to make new hair follicles is to conquer DP fibroblasts. This dogma may eventually turn out to be insufficient, and additional aspects of hair follicle neogenesis need to be explored. Several scientists have wondered whether, e.g., antlers may offer valuable answers to our question on how to make a skin with all its adnexa (Price et al. 2005). Furthermore, at least in genetic mouse models of overactive WNT signaling in the epidermis, de novo new hair follicles were produced (Gat et al. 1998). One interpretation of this mouse model is that epidermal keratinocytes themselves can transform into hair follicle-producing cells and build a new DP from surrounding dermal fibroblasts, suggesting signals released by the DP to make a hair follicle can actually be reprogrammed in keratinocytes for self-instruction. Therefore, to successfully produce functional hair follicles, multiple missing pieces in current hair follicle engineering strategies may need to be targeted simultaneously, including the availability of potent keratinocytes and DP fibroblasts (e.g., from iPS cells or normal patient fibroblasts), 3D biostructures that mimic the in vivo hair follicle conformation, and intercellular inductive signals that promote the interactions between keratinocytes and DP fibroblasts (Fig. 3).

Fig. 3.

Components of hair follicle engineering. Hair follicle bioengineering can be approached from three major aspects, including the sources of human keratinocytes and DP fibroblasts, 3D biostructures and biomaterials, and intracellular inductive signals. Created with https://BioRender.com

Innovative technologies and ideas are being developed, such as 3D bioprinting, to create and optimize a favorable microenvironment, which accurately recapitulates the 3D architecture and milieu of the hair follicle and allows proper EMIs. Organoids have been trending, and hair-producing organoids appear to be an interesting avenue to explore (Lee et al. 2022). A better understanding of ECM components and their functions in the human hair follicle niche will improve the design of a human hair follicle in vitro model. In addition, the future of hair follicle engineering may be related to the events that occur during embryogenesis before DP fibroblasts are derived. The idea to exploit the potential of embryonal dermal precursor cells as a source for a novel hair follicle engineering approach can no longer be ignored.

In summary, the myriad of scientific findings eventually has to be integrated into novel strategies of multi cell-type 3D hair follicle engineering in which epithelial cells and mesenchymal cells start a series of interactions that eventually lead to mature hair follicle keratinocytes and a mature DP. The next few years will show whether current approaches show any signs of significant improvements and whether the wave of building human DPs in 3D structures has any potential for clinical applications.