Towards developing an organotypic model for the preclinical study and manipulation of human hair matrix-dermal papilla interactions

Current ex vivo and in vitro assays for human HF research include the HF organ culture model [16, 21], 3D organotypic models [1, 7, 9] and 2-dimensional (2D) co-culture of outer root sheath keratinocytes (ORSK) and DP fibroblasts [3]. Human HF culture is a highly relevant system for evaluating hair growth, and, using pharmacological and gene knockdown approaches, specific pathways regulating hair growth can be interrogated ex vivo [5, 10, 23]. However, surplus human scalp HFs from cosmetic procedures are becoming increasingly difficult to obtain, as more advanced surgical techniques have reduced the amount of skin available for research. This growing limitation demands the development of alternative models, which are also amenable for use in pre-clinical drug discovery [16]. This methodological study attempts to do that by combining cultured HF mesenchyme with plucked anagen scalp hair shafts (HS) in vitro.

As a minimum, in vitro hair models should contain cell types with key biological roles within the HF: follicular epithelial cells of the matrix, epithelial stem cells, mesenchymal cells of the DP and connective tissue sheath (CTS), melanocytes and immunocyte populations [13]. At its most basic, an in vitro model may attempt to recreate the signalling events between 2 distinct cell types in co-culture, such as inductive DP fibroblasts and ORSK [3]. Monolayer co-cultures are simple to achieve, inexpensive and can be scaled up for high-throughput analysis [19]. However, 2D culture does not reflect the contextual behaviour of native DP cells, as DP cells cultured in monolayer rapidly lose their inductivity [12]. In contrast, 3D culture of DP fibroblasts results in DP cells that are phenotypically and functionally akin to native DP [12]. In addition, 3D culture of ORSK in the presence of DP cells and basement membrane (matrigel) has been shown to induce ORS-like epithelial differentiation [9].

Here, we report the initial stages in the development of a novel in vitro model that balances the advantages of HF organ culture with the flexibility of 3D cell-based organotypic models. This model combines plucked human anagen hair shafts (HS) containing vital HF epithelium, with DP fibroblasts pre-cultured as 3D spheroids. Both epithelial and mesenchymal components of this model are derived from material that is easily obtained in relatively large quantity (plucked anagen HSs) and can be stored and/or expanded in the lab when required (DP spheroids).

Freshly plucked human anagen scalp HSs retained part of the proximal portion of the hair matrix and the inner root sheath as demonstrated histologically and by the presence of Ki-67 + proliferative matrix keratinocytes (Fig. 1a, b). Keratin-85 (K85) immunoreactivity (IR) was clearly localised to the pre-cortical matrix and proximal HS (Fig. 1c). No residual mesenchyme was present in plucked HS, as demonstrated by negative staining for anti-versican (VN) IR (Fig. 1d, e). As expected, DPS cultured for 48 h aggregated into single spheroids that expressed markers of DP inductivity, such as ALP activity [15], versican [26] and noggin [27] IR (Fig. 1f–h). Interestingly, the distribution of versican within cultured DPS was very heterogeneous (Fig. 1g). DP fibroblasts migrated out of DPS within 48 h of being re-seeded onto tissue culture plastic, demonstrating viability of the cells in 3D (Fig. 1i).

Within 24 h of co-culture, individual DPS coalesced and surrounded the remnant matrix of the plucked HS. At 24 h, HSs could be removed from culture along with the attached DPS, indicating that DPS had successfully formed a tissue unit with the matrix (Fig. 2a–c). Robust attachment between DPS and plucked HS occurred in 100% of co-cultures. Versican (VN) showed heterogeneous distribution in the attached DPS after 24 h of co-culture (Fig. 2d), as was observed in single DPS prior to co-culture (Fig. 1g), making it difficult to define the junction between DPS and plucked HS. The attached DPS was more easily distinguished from the plucked HS by the expression of fibronectin (FN, Fig. 2e), allowing accurate quantification of Ki67 + and TUNEL + cells in the HS epithelium alone. Interestingly, the ability of DPS to attach to plucked HSs in vitro after 24 h of culture was not observed with spheroids cultured from human dermal fibroblasts (DFS), as attachment to HS by DFS was significantly diminished compared to DPS (p < 0.01, Fig. 2e–g).

Importantly, the presence of attached DPS significantly increased the intensity of K85 IR (p < 0.0001; Fig. 3a–c), and borderline significantly decreased the percentage of TUNEL + cells (p = 0.0508; Fig. 3d–f), in the proximal epithelium of the plucked anagen HS after 24 h culture, compared to plucked HSs cultured in medium alone. However, attached DPS did not prevent the rapid and complete loss of Ki67 + cells in the HS matrix (data not shown).

The pragmatic and physiologically relevant HF model we report here offers an instructive approach for investigating HF differentiation in the human anagen hair matrix. The experiments were designed primarily as a proof-of-principle study for the benefit of colleagues working in the field. It is, therefore, expected that future additional parameters will be assessed in this system, such as Wnt activity in the hair matrix, the gene expression profile of hair matrix keratinocytes and the impact of different culture media on supporting DP–hair matrix interactions.

This model contains the minimal cellular machinery for sustaining some aspects of epithelial–mesenchymal interactions that characterize human scalp HFs ex vivo. The increase in K85 in HS pre-cortical matrix cells with attached DPS attests to the vitality of pre-cortical human hair matrix keratinocytes, which are extremely difficult to culture, but which are shown here to continue synthesizing this major hair keratin [10, 17, 22]. The decrease in apoptosis in HF matrix cells with attached DPS suggests that some secreted components (“papilla morphogens”) [20] suppress hair matrix keratinocyte apoptosis under assay conditions; however, attached DPS did not prevent the complete loss of Ki67 + cells in the HS matrix, indicating that the secretory activity of DPS and the chosen assay conditions does not sufficiently support the extremely high proliferative activity of matrix keratinocytes. Thus, this co-culture system is, to the best of our knowledge, the first one to experimentally monitor human DP–hair matrix keratinocyte interactions in vitro.

Of interest is the heterogenous distribution of versican in DPS, which has previously been observed in human DP spheroid culture [11], and which suggests the presence of distinct cellular subpopulations in the human DP, as in murine DPs [28]. The same pattern of versican expression observed between single DPS and co-cultured DPS suggests that co-culture does not influence the distribution of versican in aggregated DP cells. DPS show significantly higher adhesion to the surface of the plucked HS compared to DFS, suggesting that DPS have an intrinsic affinity for the HS and can re-establish a unique epithelial–mesenchymal interface [18]. This is consistent with previous reports demonstrating that DP fibroblasts and fibroblasts from interfollicular dermis display key differences in their adhesion to basement membrane proteins, such as collagen-IV [2], and distinct expression of ECM components and ECM adhesion receptors [6, 25]. Differential adhesion of dermal fibroblast populations to the HF may also play a role in supporting HS survival, as human hair follicle survival in vitro is enhanced by activation of surface integrins [14]. However, critical components of DP-hair matrix interaction which sustain proliferation, such as P-cadherin expression, in the most proximal hair matrix cells at the DP interface [23] are clearly lacking in this system and may not easily be replaced.

In conclusion, this study describes the set-up of an organotypic HF model that provides the sophistication of the HF organ culture model with the flexibility of simpler in vitro organotypic HF models. As a tool for basic research, this pragmatic model provides a workable platform for investigating how exactly the DP controls cell proliferation and the differentiation of epithelial lineages in human hair.