Optimization of a transplant model to assess skin reconstitution from stem cell-enriched primary human keratinocyte populations


  • Normand Pouliot,

    1. Epithelial Stem Cell Biology Laboratory, Peter MacCallum Cancer Centre, St Andrew's Place, East Melbourne, Victoria, Australia;
    Search for more papers by this author
    • *

      Both the authors made equal contributions.

  • Richard P. Redvers,

    1. Epithelial Stem Cell Biology Laboratory, Peter MacCallum Cancer Centre, St Andrew's Place, East Melbourne, Victoria, Australia;
    Search for more papers by this author
    • *

      Both the authors made equal contributions.

  • Sarah Ellis,

    1. Microscopy Research Laboratory, Peter MacCallum Cancer Centre, St Andrew's Place, East Melbourne, Victoria, Australia;
    Search for more papers by this author
  • Nicholas A. Saunders,

    1. Epithelial Pathobiology Group, Princess Alexandra Hospital, Ipswich Road, Brisbane, Queensland, Australia
    Search for more papers by this author
  • Pritinder Kaur

    Corresponding author
    1. Epithelial Stem Cell Biology Laboratory, Peter MacCallum Cancer Centre, St Andrew's Place, East Melbourne, Victoria, Australia;
    Search for more papers by this author

Pritinder Kaur
Epithelial Stem Cell Biology Laboratory
Peter MacCallum Cancer Centre
St Andrew's Place, East Melbourne,
Victoria 3002, Australia
Tel.: +61 39 656 3714
Fax: +61 39 656 3738
e-mail: pritinder.kaur@petermac.org


Abstract:  Given that an important functional attribute of stem cells in vivo is their ability to sustain tissue regeneration, we set out to establish a simple and easy technique to assess this property from candidate populations of human keratinocyte stem cells in an in vivo setting. Keratinocytes were inoculated into devitalized rat tracheas and transplanted subcutaneously into SCID mice, and the epithelial lining regenerated characterized to establish the validity of this heterotypic model. Furthermore, the rate and quality of epidermal tissue reconstitution obtained from freshly isolated unfractionated vs. keratinocyte stem cell-enriched populations was tested as a function of (a) cell numbers inoculated; and (b) the inclusion of irradiated support keratinocytes and dermal cells. Rapid and sustained epidermal tissue regeneration from small numbers of freshly isolated human keratinocyte stem cells validates the utilization of this simple and reliable model system to assay for enrichment of epidermal tissue-reconstituting cells.


It is well accepted that in vivo, stem cells are responsible for the lifelong production of epidermal keratinocytes of the skin, although transit amplifying (TA) cells represent an important pool of actively dividing albeit short-lived progenitor cells (1). Thus, an important functional attribute of any putative keratinocyte stem cell (KSC) population must be its capacity to exhibit sustained epidermal tissue regeneration in long-term repopulation assays. The ability to identify stem cells of the murine interfollicular epidermis and hair follicle as slow-cycling DNA label-retaining cells in situ(2–5) has been instrumental in devising techniques for their viable isolation, culminating recently in the development of genetic strains of mice bearing GFP (green flourescent protein)-positive label-retaining cells in the hair follicle bulge region. These mice provide an elegant means of furthering our biological understanding of murine follicular stem cells (6,7), while confirming previous work demonstrating that the hair follicle bulge region is enriched for slow-cycling cells that retain DNA label (5); capable of extensive long-term cell regeneration in vitro (8,9); and reconstituting both hair follicles and interfollicular epidermis (10–12). The recent development of a competitive in vivo repopulation assay for murine interfollicular epidermal cells (13) also represents an important advance with the potential to provide further insights into the tissue-regenerative ability of candidate KSCs. However, while murine skin represents an accessible and manipulable model for studying epidermal stem cell biology, the development of clinical applications in patients requires complete biological characterization of human KSCs.

Unequivocal identification of human KSCs has been hampered given that for ethical reasons, one cannot generate label-retaining cells in humans. Importantly, the development of culture techniques for keratinocytes (14) has led to the establishment of a variety of surrogate in vitro assays, i.e. clonogenicity, long-term proliferative output and short-term tissue reconstitution (15–18), believed to reflect the extensive capacity for self-renewal and superior proliferative potential expected of KSCs in vivo. Although the ability of autologous grafts of cultured epidermal cells to rescue patients with extensive full-thickness burns for over a decade (19,20) demonstrate the immense regenerative capacity of keratinocytes, much of this work has been performed with mass cultures. Similarly, experimental long-term epidermal tissue reconstitution studies have been performed with transduced bulk cultures of human keratinocytes for up to 40 weeks (21). Consequently, it is not clear whether the activities measured in various assays can be attributed solely to stem cells. Particularly in the clonogenic assays (15), the status of stem or TA cells could only be assigned retrospectively based on expected behaviour of stem cells vs. TA cells. Thus, whilst there is a strong case in support of the high clonogenic and replicative potential of KSC in vitro, the functional properties of TA cells remain unclear and difficult to distinguish from that of KSCs when relying on retrospective analysis. Clearly, a combined approach using prospective isolation of keratinocyte progenitors and short- vs. long-term assays for both in vitro and in vivo clonogenic and tissue-regenerative ability will be valuable in devising an assay capable of distinguishing KSCs from other proliferative keratinocytes, while permitting us to refine our enrichment strategies for the stem cell population.

The aim of this study was to establish a technically simple in vivo model of long-term human skin regeneration providing a reliable assay for epidermal tissue reconstitution from low numbers of fluorescence-activated cell sorter (FACS)- fractionated primary human KSCs, transplanted without prior culture. The model relies on the repopulation of devitalized rat tracheas by human keratinocyte suspensions following subcutaneous transplantation into immunodeficient mice and is a modification of a method originally developed to evaluate carcinogen-induced neoplastic progression in rat tracheal cell lines (22) and is used more recently to demonstrate hair follicle reconstitution (7,23). We describe complete optimization and characterization of this model for robust regeneration of epithelium from cell suspensions of cultured, primary and stem cell-enriched fractions of human keratinocytes. In particular, the strategies described herein make it technically feasible to assay rare cell populations in vivo.

Materials and methods

Isolation and culture of primary human basal keratinocytes

Human primary basal keratinocytes (HFK) were isolated from neonatal foreskin tissue as described previously (18,24). Keratinocytes were cultured in keratinocyte growth medium (KGM) consisting of keratinocyte basal medium (Clonetics, San Diego, CA, USA) supplemented with 10 ng/ml murine epidermal growth factor (EGF), 5 µg/ml insulin, 0.5 µg/ml hydrocortisone (Sigma, St Louis, MO, USA), 70 µg/ml bovine pituitary extract (Hammond Cell Tech, Windsor, CA, USA), 6 µg/ml penicillin, 80 µg/ml gentamycin (Life Technologies, Gaithersburg, MD, USA) and 6 µg/ml fluconazole (Pfizer, West Ryde, NSW, Australia).

Isolation and culture of human dermal fibroblasts

The dermal foreskin tissue was digested in 4 mg/ml dispase and 3 mg/ml of collagenase (Worthington Biochemical Corporation, Lakewood, NJ, USA) in phosphate-buffered saline (PBS) at 37°C for 1.5 h and the resultant cell suspension filtered through a 70 µm cell strainer (BD Biosciences, San Jose, CA, USA). Dermal cells (HFF) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), l-glutamine (2 mM; Multicel, Castle Hill, NSW, Australia), penicillin (1.2 µg/ml), gentamycin (16 µg/ml) and sodium pyruvate (1 mM, Gibco BRL, Melbourne, Victoria, Australia) (DMEM-10).

Enrichment for KSCs by FACS

Immunofluorescent staining and FACS of primary keratinocytes were performed as described previously (18) and resuspended at 2–3 × 106 cells/ml in KGM prior to sorting on a Becton Dickinson FACSVantage flow cytometer. Total unfractionated (UF) viable keratinocytes and fractions enriched for KSCs defined by the phenotype inline image(18,25) were collected into KGM for immediate inoculation into tracheas. At least 10 000 events per sample were saved for analysis, and the stem cell fraction was routinely re-analysed to ensure purity. The viability of keratinocyte subpopulations after sorting was assessed by Trypan Blue exclusion and exceeded 90%.

Preparation of rat tracheas for transplantation

Rat tracheas were aseptically removed from 200 to 250 g Sprague-Dawley rats (Institute of Medical and Veterinary Science, Adelaide, South Australia, Australia) trimmed off excess connective tissue and devitalized by three cycles of freeze–thawing (−70°C and 37°C). Tracheas were stretched and secured onto microbore PTFE (polytetrafluoroethylene) tubing (Cole-Parmer Instrument Company, Vernon Hills, IL, USA) and their ends sealed by LT200 ligaclips (Ethicon Endo-Surgery, Cincinnati, OH, USA). A small incision was made in sealed tracheas and inoculated with keratinocytes resuspended alone or together with foreskin dermal fibroblasts in a 1 : 1 mixture of KGM and DMEM supplemented with 10% FCS, 20 ng/ml EGF, 0.4 mg/ml hydrocortisone and 10 ng/ml cholera toxin. Tracheas were re-sealed with a ligaclip and placed in KGM : DMEM on ice until transplantation. Where indicated, 10 µg/ml laminin-10/11 (LN-10/11) (Life Technologies) was used to precoat the denuded tracheas overnight at 4°C and added to the cell suspensions (also at 10 µg/ml) prior to inoculation. Female SCID mouse recipients (9–12 weeks) were anaesthetized by intraperitoneal injection of 10 µl/g body weight of a solution of Rompun (2 mg/ml xylazine, Bayer Australia, Pymble, Australia) and ketamine (10 mg/ml, Parnell Laboratories, Sydney, NSW, Australia) prepared in PBS and two tracheas implanted subcutaneously into each mouse. Two mice were transplanted for each condition tested and harvested at various time points, and each experiment replicated two to three times.

Processing of transplants for histology and immunohistochemistry

Mice were killed and the tracheas excised from recipients and fixed in 4% (w/v) buffered formalin for 2 h. Half of each trachea was decalcified in 10 ml of 20% (w/v) ethylenediaminetetraacetic acid (EDTA) pH 8.5 for 24 h at 4°C and embedded for cryosections. The other half of each trachea was fixed for a further 18 h at 4°C, decalcified in 20% (w/v) EDTA as above and processed for paraffin embedding and routine histology [haematoxylin and eosin staining].

Immunofluorescence and immunohistochemistry

Transverse cryosections (5 µm) were blocked with 3% (v/v) preimmune goat serum in PBS 0.05% Tween-20 (PBS-T) for 30 min prior to incubation with primary antibodies (2 h at room temperature or overnight at 4°C). For immunofluorescence, sections were incubated with Mab V9 (IgG1, DAKO Corporation, Carpinteria, CA, USA) against vimentin or 5B5 (IgG1, DAKO) against the beta subunit of human prolyl 4-hydroxylase at 1 : 3000 and 1 : 100, respectively, fluorescein isothiocyanate-conjugated antimouse F(ab′)2 secondary at 1 : 200 (Chemicon, Temecula, CA, USA) and counterstained with propidium iodide (1 µg/ml, Sigma). For immunohistochemistry, monoclonal antibodies to keratins (K), K10 (LHP2, IgG1, 1 : 20 dilution), K14 (LL001, IgG2a, dilution 1 : 200) and K15 (LHK15, IgG2a, 1 : 5 dilution) were used as hybridoma supernatants kindly provided by Dr Irene Leigh (Royal London Hospital, London, UK). The α5 chain of LN-10/11 was detected with Mab 4C7 (IgG2a, 1 : 200 dilution of ascites, Chemicon). Mab P1E1-E4 to LN-5 (mouse IgG1) and P5D2 to β1 integrin (IgG1) were used as undiluted hybridoma supernatants, kindly provided by Dr William Carter (Fred Hutchinson Cancer Research Center, Seattle, WA, USA). Mab 4F10 (IgG2b, Biodesign International, Saco, ME) directed against the α6 integrin was used at 5 µg/ml. Rabbit polyclonal antibodies to involucrin (a kind gift from Dr Robert Rice, University of California, San Francisco, CA, USA) and filaggrin (kindly supplied by Dr Richard Presland, University of Washington, Seattle, WA) were used at a 1 : 3000 dilution. Rabbit polyclonal antibody AF62 against loricrin (BAbCO, Richmond, CA, USA) was used at a 1 : 500 dilution. Isotype-matched negative control Mabs 1A6.11 (IgG2b), 1D4.5 (IgG2a) and 1B5 (IgG1) used to determine non-specific staining were provided by Dr Leonie Ashman (University of Newcastle, Newcastle, NSW, Australia). Sections were then incubated with biotinylated goat antimouse or goat antirabbit secondary antibodies (1 : 250 dilution, Vector, Burlingame, CA, USA); inactivation of endogenous peroxidase was performed by incubation with 0.3% H2O2 in methanol for 30 min at room temperature. Sections were incubated with streptavidin– horseradish peroxidase (ABC kit, Vector Laboratories) and developed with the AEC substrate kit (Vector Laboratories).

Ultrastructural analysis

Tracheas containing skin grafts were fixed in 2.5% glutaraldehyde buffered with 0.08 M Sorensen's phosphate buffer, dissected into 1 mm cubes and postfixed in 2% osmium tetroxide. The tissue was then dehydrated through a graded series of alcohols and passed through two changes of acetone before embedding in Spurrs resin. Sections approximately 70 nm thick were cut on a Reichert-Jung Ultracut E (Leica) ultramicrotome using a Diamond knife (Diatome) and examined in a Hitachi H600 Transmission Electron Microscope.


Regeneration of fully stratified human epidermis in the in vivo tracheal transplantation model

To establish the suitability of the rat trachea model for epidermal tissue reconstitution, initial transplantation experiments were performed with primary cultured (p0) neonatal human foreskin keratinocytes (HFKs) and the optimal requirements for tissue reformation determined including number of keratinocytes transplanted, inclusion of human dermal cells (HFF) and the time taken to establish a mature stratified epidermis. Control tracheas inoculated with culture medium alone consistently failed to yield epidermal tissue at all time points analysed following transplantation (Fig. 1a; 4 weeks). Progressive granulation was observed at 2 and 4 weeks partially or completely invading the tracheal lumen. In contrast, seeding the tracheas with HFKs gave rise to epidermal tissue. The extent of luminal coverage was predictably dependent on the number of keratinocytes transplanted: inoculation with 3 × 104 cells resulted in clonal cyst-like epidermal growth even after 4 weeks (Fig. 1b); complete epithelialization of the tracheal lumen was observed with cell numbers of 5 × 105 and above. A continuous but immature epidermis was established by 2 weeks (Fig. 1c), which matured into a multilayered sheet containing basal, spinous, granular and cornified layers by 4 weeks (Fig. 1d). Notably, epidermal proliferation and differentiation was not dependent on the inclusion of dermal cells in the transplant, and the absence of exposure to air was not a limiting factor in progressing through the maturation programme with abundant production of terminally differentiated squames. We investigated the effects of cotransplanting cultured human foreskin fibroblasts (HFFs) and the basement membrane (BM) protein LN-10/11 given our recent work illustrating a role for this extracellular matrix component in stimulating keratinocyte proliferation in vitro (25,26). The inclusion of LN-10/11 did not have a demonstrable effect on the rate of epidermal regeneration at the 2- and 4-week time points analysed (data not shown), but the inclusion of both HFFs and LN-10/11 accelerated the rate of epidermal regeneration and maturation at 2 weeks (Fig. 1e) compared to control grafts seeded with keratinocytes alone (Fig. 1c); a thicker epithelium was also obtained at 4 weeks post-transplant in the presence of HFFs and LN-10/11 (Fig. 1f) compared to controls (Fig. 1d).

Figure 1.

Epidermal regeneration in the rat trachea transplant model from cultured human primary keratinocytes. Denuded rat tracheas were inoculated with (a) medium only; (b) 3 × 104 primary cultured (p0) neonatal human foreskin keratinocytes (HFKs) alone; and (c–f) 5 × 105 p0 neonatal HFK alone (c, d) or in the presence of laminin-10/11(10 µg/ml) and 104 p7 human foreskin fibroblasts (e, f). Epidermal regeneration was evaluated 2 weeks (c, e) or 4 weeks (a, b, d, f) post-transplantation by haematoxylin and eosin staining of transverse sections of harvested tracheas embedded in paraffin. The asterisk (*) in panel a denotes the cartilage ring, while the arrow indicates the fibroelastic membrane of the luminal surface upon which epithelium may be generated. Scale bars = 50 µm.

Infiltration of host mesenchymal cells into the tracheal submucosa

Close histological examination of the grafts revealed that the tracheal submucosa was infiltrated by mesenchymal-like cells in all transplants. We therefore stained the tracheal transplant tissue with two antibodies directed against mesenchymal markers: V9 which recognizes both mouse and human vimentin (27) and the human-specific 5B5 antibody which detects the β-subunit of prolyl 4-hydroxylase, an enzyme involved in collagen synthesis and commonly used as a fibroblast marker (28). As expected, both V9 and 5B5 antibodies reacted strongly with dermal cells in normal human neonatal foreskin (Fig. 2a,b). Analysis of 4-week tracheal grafts inoculated with HFKs and HFFs confirmed the presence of vimentin-positive mesenchymal cells immediately beneath the engrafted epidermis, but also within the tracheal submucosa (Fig. 2c). In contrast, reactivity of the human-specific 5B5 antibody was restricted to the mesenchymal cells directly beneath the regenerating epidermis (Fig. 2d), and was not with the submucosal vimentin-positive mesenchymal cells, the latter presumably being of host origin. This data show that while HFFs persist in transplants, the tracheal tissue is also populated with host (murine) mesenchymal cells that may assist in epidermal tissue regeneration in the absence of donor fibroblasts. Interestingly, keratinocytes in the basal and spinous layers of the regenerating epidermis, although clearly epithelial in origin (Fig. 2c, vimentin negative), also reacted strongly with the 5B5 antibody, suggesting that epidermal keratinocytes may contribute to collagen synthesis transiently, given that 5B5 reactivity was diminished significantly in the basal layer at 12 weeks (data not shown).

Figure 2.

Infiltration of donor mesenchymal cells into the subtracheal mucosa of transplanted tracheas. Five-micron cryosections of (a, b) neonatal foreskin and (c, d) 4-week-old tracheal transplants derived from 5 × 105 primary cultured (p0) neonatal human foreskin keratinocytes and 104 p7 human foreskin fibroblasts (HFFs) were stained with antibodies to vimentin (V9 antibody – a, c), and human prolyl 4-hydroxylase (5B5 antibody – b, d) by indirect immunofluorescence and counterstained with propidium iodide to visualize nuclei. The presence of donor vimentin-positive cells is evident in panel c, while HFFs are identifiable in panel d. Scale bar = 50 µm.

Epithelial morphogenesis and maturation of human epidermal tissue in the tracheal transplant model

To determine whether the heterologous tracheal environment altered the physical and biochemical properties of the epidermal tissue regenerated in this transplant model, we analysed the epithelial sheets for ultrastructural integrity and the appropriate temporal and spatial expression of keratinocyte differentiation markers. Semi-thin sections of 4-week-old skin derived from cultured HFKs and HFFs showed that the overall architecture of the epidermal tissue regenerated in the transplant model closely resembled that of native foreskin tissue, comprised of a basal layer, six to eight spinous cell layers, two to three granular layers and abundant cornified layer (Fig. 3a). Transmission electron microscopy (TEM) of the dermo-epidermal junction revealed highly polarized basal keratinocytes (Fig. 3b) separated from the underlying stroma by a continuous BM with a distinct lamina lucida and lamina densa (Fig. 3d) and numerous adjacent hemi-desmosomes (Fig. 3d,e) containing keratin filaments. Other features of normal epidermis included intercellular desmosomes in the spinous layers (Fig. 3c), electron-dense keratohyalin granules and lamellar bodies in the granular layer (Fig. 3f) and cornified cell envelopes in the uppermost layers (Fig. 3f).

Figure 3.

Untrastructural characterization of human epidermis regenerated in tracheal transplants. Human epidermis derived from 4-week-old tracheal xenografts inoculated with 5 × 105 primary cultured (p0) neonatal human foreskin keratinocytes. (a) Semi-thin section of reconstituted epidermis stained with toluidine blue and examined under light microscopy shows normal architecture of reconstituted tissue comprising basal layer and suprabasal differentiating granular and cornified layers. (b–f) Transmission electron microscopy analysis demonstrating the regeneration of a polarized basal layer (b) with numerous intercellular desmosomes (c, arrows) and tonofilament bundles (c, arrowheads) in the spinous layer. The formation of a complete basement membrane under the basal layer (d) with continuous lamina lucida and lamina densa (Ld) and numerous mature hemi-desmosomes connected with intracellular keratin filament bundles (e) is evident. Keratohyalin granules (f) were present in the granular layer, and terminal differentiation to form cornified cells occurred in the uppermost layers (f, arrows). Scale bars: panel a = 25 µm; panels b and f = 0.5 µm; panels c, d and e = 1 µm.

Consistent with the TEM data, immunohistochemical localization of epidermal proteins demonstrated that the transplanted keratinocytes were able to re-establish a normal homeostatic programme of epidermal growth and maturation. Specifically, the expression of β1 integrin (Fig. 4a,d), K14 (Fig. 4b,e), K10 (Fig. 4c,f), involucrin (Fig. 4g,j) and loricrin (Fig. 4h,k) was observed in the appropriate layers analogous to normal skin, irrespective of the inclusion of HFFs. We observed the deposition of human LN-5 α3 chain (data not shown) and LN-10/11 α5 chain in the BM region, again irrespective of the inclusion of HFFs (Fig. 4i,l). The data shown in Fig. 4 are representative of 4-week post-transplant; however, no significant differences in the expression of epidermal markers was observed at 2 weeks (data not shown). These data demonstrate that a normal programme of epidermal differentiation is established relatively early in this model system with or without donor dermal support, despite the heterologous tracheal environment. Interestingly, differences in the onset of K15 re-expression were regulated by microenvironmental factors, specifically the inclusion of LN-10/11 in the transplants. K15 normally found in homeostatic skin in basal keratinocytes is lost in hyperproliferative states such as psoriasis and in organotypic cultures (25,29). K15 was undetectable in epithelial sheets generated by HFKs alone (data not shown), or in the presence of HFFs (Fig. 4m), at 2-week post-transplant and re-expressed in the basal layer by 4 weeks in all sheets examined (Fig. 4n; HFF + HFK). The addition of exogenous LN-10/11 together with HFKs and HFFs in the transplants resulted in the earlier onset of K15 re-expression at 2 weeks (Fig. 4o). These data suggest that homeostasis appears to be established earlier in the presence of HFFs and LN-10/11, consistent with the more rapid maturation observed histologically (Fig. 1e) at this time point, compared to HFKs alone.

Figure 4.

Expression of keratinocyte markers in epithelial sheets derived from human foreskin keratinocytes (HFKs) following transplant. Epidermal tissue obtained at 4 weeks from 5 × 105 p0 HFKs transplanted alone (a–c; g–i), or together with 104 p7 human foreskin fibroblasts (HFFs) (d–f; j–l), were stained for β1 integrin (a, d), keratin 14 (b, e), keratin 10 (c, f); involucrin (g, j), loricrin (h, k) and laminin-10/11) (LN-10/11)_α5 chain (i, l). Appropriate temporal and spatial expression of epidermal markers was observed in the reconstituted tissue irrespective of the inclusion of HFFs. K15 expression (m–o) was absent at 2 weeks (m) but present at 4 weeks post-transplant (n) when HFKs were transplanted with HFFs. Interestingly, the inclusion of exogenous LN-10/11 and HFFs together with HFKs resulted in re-expression of K15 at only 2 weeks after transplant (o). Scale bar = 50 µm.

Epidermal tissue reconstitution from freshly isolated primary keratinocytes of neonatal and adult human skin

The ultimate goal of this work was to optimize the transplant model to assay for long-term tissue regeneration from low numbers of uncultured primary keratinocytes to avoid altering their biological properties upon culture – an important step in establishing the parameters for transplantation of small numbers of candidate stem cells. Thus, tracheal transplants were performed with primary HFKs directly after isolation and harvested at 3, 5 and 12 weeks (Fig. 5). Under these conditions, we found that seeding 5 × 105 uncultured keratinocytes alone was sufficient to regenerate a stratified epidermis, although the rate of tissue regeneration was slower than cultured, primary keratinocytes needing up to 5 weeks to form a mature epidermis (Figs 5a,b). Complete epidermal regeneration was also observed when freshly isolated HFKs were inoculated together with low numbers (5 × 103) of non-irradiated HFFs; however, higher numbers of HFFs resulted in excessive intraluminal fibroblast growth and lower epidermal regeneration (data not shown). Primary (uncultured) keratinocytes from adult skin (HAKs) could also reform an epidermis in this model system although greater number of keratinocytes (106) were required to regenerate a complete lining of stratified epidermis within 3–5 weeks (Figs 5d,e).

Figure 5.

Long-term in vivo epidermal regeneration by freshly isolated human neonatal and adult primary keratinocytes. (a–c) Epidermal tissue regeneration following transplantation of 5 × 105 freshly isolated neonatal primary human foreskin keratinocytes (HFKs) alone, harvested at 3 (a), 5 (b) and 12 (c) weeks; (d–f) Epidermal tissue regenerated from 106 adult breast skin keratinocytes alone, harvested at 3 (d), 5 (e) and 12 (f) weeks. Better long-term tissue regeneration was obtained from neonatal rather than adult keratinocytes (compare panel c vs. panel f). (g–i) Long-term epidermal tissue reconstitution (12 weeks post-transplant) from 104 primary HFKs inoculated alone (g), together with 5 × 105 15 Gy irradiated p0 HFKs as fillers (h), or with fillers and 105 20 Gy irradiated human foreskin fibroblasts as feeders. Good long-term tissue maintenance was observed when high numbers of primary HFKs were transplanted alone (c). Optimal long-term tissue regeneration from small numbers of HFKs was obtained by the inclusion of keratinocyte fillers and dermal feeder cells (i). Scale bar = 50 µm.

Long-term epidermal tissue reconstitution from primary human keratinocytes

To ascertain the suitability of this model for long-term epidermal tissue regeneration in vivo, we analysed tracheal grafts at various time points up to 12 weeks. High numbers of primary HFKs (5 × 105) transplanted either alone (Fig. 5c), or in combination with HFFs (data not shown), maintained a fully stratified epidermis at 12 weeks with an accumulation of mature squames in the lumen. Notably, good long-term epidermal renewal was not obtained from adult keratinocytes (Fig. 5f) even with the greater cell numbers inoculated (106); at 12 weeks, HAKs formed a circular stratified epidermis covering approximately one-fifth of the tracheal surface, while the rest of the lumen was filled with host granulation tissue (Fig. 5f).

We therefore hypothesized that the incidence of keratinocytes capable of long-term tissue reconstitution in the inoculate may determine long-term maintenance of the grafts and/or that the absence of skin-derived dermal cells may prevent self-renewal of such stem cells, thereby resulting in poor epidermal replacement in the long term as proposed by Leary et al. (30), while not affecting short-term tissue regeneration (i.e. over 5 weeks). Titration of keratinocyte cell numbers in this model (Fig. 1) also suggested that a critical factor in establishing these grafts was to avoid invasion by host cells immediately after transplant. We devised two strategies to promote the evaluation of long-term epidermal tissue reconstitution in this model: (a) the inclusion of lethally irradiated keratinocytes or support cells that would act as space fillers to prevent invasion by granulation tissue, thereby assisting small numbers of keratinocytes to repopulate the trachea; and (b) the inclusion of irradiated HFFs (feeder cells) to promote the survival and/or maintenance of transplanted long-term repopulating keratinocytes (stem cells). Thus, 104 primary HFKs were transplanted alone in the presence of fillers (5 × 105 irradiated p0 HFKs), or in the presence of fillers and feeders (irradiated p0 HFKs and 105 irradiated HFFs). 104 primary neonatal HFKs transplanted alone did not engraft at any time point tested, including 12 weeks after transplant (Fig. 5g). The inclusion of filler keratinocytes resulted in improved short-term (data not shown) and long-term engraftment of HFKs (Fig. 5h), while the addition of fillers and fibroblast feeders resulted in excellent repopulation with the maintenance of a complete lining of stratified epidermis within the tracheal lumen for 12 weeks (Fig. 5i; compare with Fig. 5c– HFK alone at 12 weeks). These data suggest that the inclusion of irradiated fillers helps establish the epidermal grafts from small numbers of cells, while the irradiated feeders improve long-term engraftment.

Rapid epidermal tissue regeneration from FACS enriched neonatal human KSCs

We next hypothesized that stem cell-enriched epidermal suspensions may allow rapid and complete regeneration even at a low cell density, given the predicted potency of these cells. KSCs were isolated from neonatal HFKs based on their cell-surface phenotype (inline image) by FACS as described previously (26) and their regenerative potential compared to that of UF HFKs (p0) after 2 and 4 weeks in vivo. In initial experiments, UF and KSCs (inline image) were expanded in culture (p0) to obtain a sufficient number of cells to inoculate 3 × 104 HFKs/trachea. Consistent with previous data (Fig. 1b, 4 weeks), at 2 weeks, UF keratinocytes gave rise to clusters of epidermal growth invaded by abundant granulation tissue (Fig. 6a); in contrast, the KSC fraction rapidly reconstituted a stratified and polarized epidermis covering the entire tracheal lumen (Fig. 6b). The epidermis regenerated by 3 × 104 KSCs at 2 weeks exhibited greater cellularity (Fig. 6c) than grafts obtained with 5 × 105 UF HFKs at the same time point (Fig. 1c). The KSC-derived epidermis was more mature at 4 weeks with a thicker spinous layer and more developed cornified layer (Fig. 6d).

Figure 6.

Rapid in vivo epidermal regeneration by human keratinocyte stem cells compared to unfractionated human foreskin keratinocytes (HFKs). Epidermal tissue reconstitution obtained from (a) 3 × 104 unsorted HFKs at 2 weeks; or (b–d) inline image stem cell fraction (b, c: 2 weeks; d: 4 weeks). Both unsorted and fluorescence-activated cell sorter-enriched populations were placed in culture for expansion and transplanted without subculturing (i. e. p0). (e, f) Epidermal tissue regeneration obtained from freshly sorted R1 primary HFK population (unfractionated control, e) and inline image stem cell fraction (f) transplanted without subjecting to culture, harvested at 6 weeks. Enrichment for epidermal tissue-reconstituting cells is evident in the inline image fraction of primary HFKs (b, f). Scale bars: panels a–d = 50 µm; panels e and f = 100 µm.

In subsequent experiments, we transplanted inline image KSCs directly after their isolation and FACS fractionation into the tracheas without subjecting them to culture and lowered the number of cells transplanted to 104 to further determine their potency. Control transplants performed with 104 primary HFKs (subjected to the same antibody labelling and FACS selection for viable cells as for the KSC fraction, but without regard for their cell-surface phenotype) assessed at 6 weeks demonstrated that these UF keratinocytes were not capable of forming an epidermis (Fig. 6e). In the same time frame, an equivalent number of KSCs regenerated an impressive epithelial lining (Fig. 6f) demonstrating for the first time the potent in vivo tissue-regenerative potential of isolated human epidermal stem cells.


Recent work from our laboratory has revealed that although KSCs derived from neonatal foreskin tissue exhibit a greater capacity to regenerate epidermal tissue in short-term in vitro assays, equivalent tissue regeneration could also be obtained from the progeny of stem cells, i.e. TA and even differentiating keratinocytes under specific experimental conditions (25). While that study led to the identification of important regulatory factors found in the microenvironment of epidermal progenitors such as LN-10/11, it also served to highlight the inadequacy of short-term tissue-regeneration assays to clearly distinguish KSCs from their more committed progeny. We therefore set out to optimize the transplantation model system to provide a means to assay for the enrichment of tissue-regenerating cells, i.e. keratinocyte progenitors isolated prospectively on the basis of their cell-surface phenotype.

Although a number of in vivo models exist that permit skin regeneration from single-cell suspensions, i.e. direct transplantation into the subcutis (16), into silicon chamber implants (13), and grafting in vitro reconstituted skin equivalents onto mouse muscle fascia (21,31,32), they frequently involve complex surgical procedures. Attracted by the simplicity of the rat trachea transplant model, we adapted this technique to permit tissue reconstitution from small numbers of FACS fractions of primary keratinocytes, despite the fact that earlier attempts to regenerate human epidermis met with limited success (33). Our data demonstrate that successful engraftment of epidermal cells is consistently obtained with complete recapitulation of the normal processes of epidermal proliferation, differentiation and morphogenesis particularly when large numbers of keratinocytes are transplanted, even in the absence of donor dermal cells. This process occurs more rapidly with the inclusion of dermal cells and the BM extracellular matrix proteins LN-10/11 and can most likely be attributed to the proliferative induction provided by these microenvironmental elements consistent with our earlier studies (25,26). Our data demonstrate that this model system is amenable to the study of microenvironmental regulators of tissue regeneration. Interestingly, host mesenchymal cells are recruited into the tracheal submucosa, and the transplanted keratinocytes produce LN-10/11 (α5 chain) themselves, establishing a normal epithelium albeit at a slower rate. Our data indicate that the inclusion of irradiated donor support keratinocytes is important in establishing epidermal grafts from small numbers of cells such as candidate stem cells, and most likely prevent invasion by host granulation tissue as previously reported (34). Interestingly, the rapid repopulation obtained with the stem cell-enriched fraction inline image suggests that this strategy may not be necessary when transplanting keratinocyte populations containing a high proportion of tissue-reconstituting cells. From this data, we infer that rapid tissue reconstitution from small numbers of keratinocytes in the absence of support cells provides a means for identifying KSC populations with high tissue forming potency. In other experiments, we have transplanted the progeny of KSCs, i.e. TA (inline image) and early differentiating (ED) (inline image) cells derived from neonatal human foreskin, and found inconsistent tissue regeneration when small numbers (i.e. 104) of freshly sorted TA and ED populations were inoculated in the absence of irradiated support cells (Pouliot et al., unpublished data). Interestingly, cotransplantation of the same low number of KSCs, TA and ED cells with irradiated support keratinocytes demonstrated equivalent tissue regeneration from all three fractions over a 6- to 10-week period (25). Thus, optimization of transplant conditions revealed significant and relatively long-term epidermal tissue-regeneration capacity, indicating that serial transplantation may be required to distinguish KSCs from their progeny. The rat trachea model lends itself to this approach – indeed Klein-Szanto et al. (33) have successfully used this technique to serially transplant fetal human tracheal epithelial cells up to six times. In ongoing studies, we have been able to harvest approximately 105 viable keratinocytes per trachea transplanted with unfractionated HFKs (Gangatirkar & Kaur, unpublished data), thus making it feasible to undertake serial transplantation experiments.

Our data indicate that short-term epidermal reconstitution (4–5 weeks) is not compromised in the absence of donor dermal cells, and presumably the host mesenchymal or vimentin-positive cells which appear in the submucosa under the epidermal tissue play a compensatory role. However, it is difficult to dissect the role of these cells in sustaining the transplanted keratinocytes in the broader context of circulating factors and autocrine regulatory factors in promoting epidermal tissue regeneration. Cotransplantation of dermal fibroblasts with keratinocytes appears to promote long-term epidermal tissue maintenance (up to 12 weeks). Whether this can be attributed to the self-renewal of epidermal stem cells within the grafts (30), or improved growth conditions due to factors provided by donor dermal cells, remains to be determined.

The data obtained with primary adult breast skin keratinocytes provide us with an indication of the tissue-regenerative ability of these cells relative to neonatal keratinocytes. Consistent with the observed poorer growth capacity of adult keratinocytes (14), we found that greater numbers of adult keratinocytes were required to obtain tissue regeneration equivalent to that of neonatal keratinocytes. This observation combined with the poor long-term epidermal tissue maintenance obtained at 12 weeks suggests that the incidence and/or potency of cells capable of both short-term and long-term tissue reconstitution is lower in adult vs. neonatal epidermis. This model will be valuable in further investigating the intrinsic vs. microenvironmental factors regulating epidermal renewal.

In conclusion, we describe a technically simple model to investigate the biological properties of keratinocyte progenitors of both murine and human origin. This model system has applications in testing putative markers for human KSCs and can be used to evaluate competitive repopulation so long as the test populations can be identified retrospectively as described recently for murine keratinocytes (13). Although it is technically feasible to genetically tag human keratinocytes in vitro following separation by FACS, the inevitable selection and activation of subpopulations by culture conditions has to be taken into account when interpreting the results obtained, compared to analysing freshly isolated primary keratinocyte subpopulations.


We thank Andrew Fryga and Ralph Rossi for their invaluable cell-sorting expertise. We thank Rebecca Morris for technical advice and bringing our attention to the transplant model. We thank Amy Li for technical advice. This work was supported by Project Grant funding from the National Health & Medical Research Council of Australia (145654) to PK. Richard Redvers was supported by the Anti-Cancer Council of Victoria Postgraduate Research Scholarship.