Somatic Epidermal Stem Cells Can Produce Multiple Cell Lineages During Development


  • Luchuan Liang,

    1. Departments of Anatomy and Cell Biology, The University of Iowa, Iowa City, Iowa, USA
    Search for more papers by this author
  • Jackie R. Bickenbach Ph.D.

    Corresponding author
    1. Departments of Anatomy and Cell Biology, The University of Iowa, Iowa City, Iowa, USA
    2. Dermatology, The University of Iowa, Iowa City, Iowa, USA
    • Departments of Anatomy and Cell Biology, and Dermatology, The University of Iowa College of Medicine, 1-457 BSB, 51 Newton Road, Iowa City, IA 52242, USA. Telephone: 319-335-6719; Fax: 319-335-7198
    Search for more papers by this author


It has been demonstrated that several types of somatic stem cells have the remarkable capacity to differentiate into other types of tissues. We demonstrate here that stem cells from the skin, the largest organ of the body, have the capacity to form multiple cell lineages during development. Using our recently developed sorting technique, we isolated viable homogenous populations of somatic epidermal stem and transient amplifying cells from the skin of 3-day old transgenic mice, who carried the enhanced green fluorescent protein transgene, and injected stem, TA, or unsorted basal epidermal cells into 3.5-day C57BL/6 blastocysts. Only the stem-injected blastocysts produced mice with GFP+ cells in their tissues. We found GFP+ cells in ectodermal, mesenchymal, and neural-crest-derived tissues in E13.5 embryos, 13-day-old neonates, and 60-day-old adult mice, proving that epidermal stem cells survived the blastocyst injection and multiplied during development. Furthermore, the injected stem cells altered their epidermal phenotype and expressed the appropriate proteins for the tissues into which they developed, demonstrating that somatic epidermal stem cells have the ability to produce cells of different lineages during development. These data suggest that somatic epidermal stem cells may show a generalized plasticity expected only of embryonic stem cells and that environmental (extrinsic) factors may influence the lineage pathway for somatic stem cells. Thus, the skin could be a source of easily accessible stem cells that are able to be reprogrammed to form multiple cell lineages.


It has been know for several decades that the epidermis of the skin contains a subpopulation of basal cells that exhibit the properties expected of somatic stem cells: slow cell cycle, high proliferative potential, location in a protected niche, capacity to maintain and repair the tissue in which they reside, and long life span [1-, 8]. Slowly cycling epidermal stem cells have been identified by long-term nuclear retention of tritiated thymidine or bromodeoxyuridine label [2,, 3,, 9]. These undifferentiated label-retaining stem cells (LRCs) have been shown to reside in the bulge area of the hair follicle [6,, 10,, 11] and in the interfollicular basal layer of the epidermis [2,, 3,, 12]. They are self-renewing and able to produce daughter transient amplifying (TA) cells that undergo a finite number of cell divisions before they differentiate and leave the proliferative basal compartment, a property similar to stem cells in other continuously renewing tissues [5,, 13].

Several techniques based on cell surface markers, size, or in vitro adhesion assays have been used to enrich for epidermal stem cells [9,, 14-, 19]. We have previously shown that long-term retention of tritiated thymidine or bromodeoxyuridine marks very slowly cycling cells, and we have shown that they can be identified for as long as 6 months after labeling [2]. Although this technique has been used extensively for the last two decades to identify label-retaining stem cells in several epithelia [3,, 6,, 20], it cannot be used to sort a viable population of epidermal stem cells. Three years ago, we published a method to enrich for these LRCs [21]. When examining the total enriched population, we found that only around 50% of the cells in this population fulfilled the criteria expected for stem cells [9,, 21,, 22]. Last year, we developed a modified sorting method [23] that used a combination of two stem cell enrichment techniques, selection by size [14,, 16], and a modification of a previously published method for isolating hematopoietic stem cells [24,, 25]. The latter technique was based upon Hoechst 33342 dye exclusion in combination with no expression of hematopoietic lineage markers. We combined the two techniques and adjusted the gates based upon where we found LRCs [2]. We showed that this new procedure yielded a virtually homogenous population of epidermal cells that contained >95% of the LRC population [23]. This population had all of the functional characteristics expected for epidermal stem cells. All of the cells expressed keratin K14, a marker for basal epidermal cells [26], and did not express K1, a marker for differentiated epidermal cells [27]. None of the cells expressed typical hematopoietic markers, such as CD34 or Sca-1. The stem cell population showed slow cell proliferation, long-term in vitro growth, long-term expression of recombinant genes, and complete recapitulation of the epidermis with long-term maintenance of the engineered tissue, whereas the TA population did not exhibit any of these characteristics [23]. In the present study, using the same sorting technique, we report that these somatic epidermal stem cells, but not the TA cells, can produce several cell lineages in a variety of tissues in the developing mouse, suggesting that these stem cells have retained a high level of tissue plasticity.

Materials and Methods

Epidermal Stem Cell and TA Cell Isolation

Nontransgenic C57BL/6 mice and transgenic C57BL/6 mice carrying the enhanced green fluorescent protein (EGFP) transgene (TgN(ACTbEGFP)1Osb) were originally obtained from Jackson Labs (Bar Harbor, ME;, and breeding colonies of both were established and maintained in The University of Iowa's Animal Care Facility under the direct supervision of a certified veterinarian. Animals were humanely sacrificed when necessary and only according to U.S. federal guidelines. Epidermal stem cells were prepared and isolated as previously described [23]. Skin samples from neonate C57BL/6-EGFP mice were incubated in 0.25% trypsin overnight at 4°C. This allowed us to remove the interfollicular epidermis from the dermis, leaving the hair follicles from the infundibulum to the dermal papillae within the dermis. Using 0.25% trypsin overnight at 4°C works equally as well as using dispase overnight at 4°C or 0.3 mM EDTA for 1-2 hours at 37°C in cleanly separating the interfollicular epidermis as a sheet from the underlying dermal tissue containing the hair follicles, microvasculature, nerves, and adipose tissue. The interfollicular basal cells in the epidermal sheet were dissociated into a single cell suspension by gently shaking the epidermal sheet in low calcium medium (0.05 mM). This separation and dissociation method results in a virtually homogenous population of basal cells that are keratin K14+ [26] and K1 [27]. The dissociated basal cells were centrifuged and resuspended at 5 × 106 cells/ml in SMEM (Life Technologies; Grand Island, NY; with 0.05 mM Ca++, 7% chelexed fetal bovine serum (FBS), 1 mM Hepes, 1% antibiotic/antimycotic (Life Technologies), and 5 μg/ml Hoechst 33342 (Sigma; St. Louis, MO; Cells were incubated in the medium plus dye at 37°C for 90 minutes, centrifuged and resuspended in 1 μg/ml propidium iodide in medium, and kept on ice until the flow cytometric procedure. Based on more that 50 experiments, there has been virtually no variability in the Hoechst 33342 dye uptake of individual epidermal cell preparations.

Cell sorting was performed on a Coulter EPICS 753, which was configured to set nonrectilinear (bitmapped) sort gates precisely as depicted in the bitmap histograms in Figure 1 (note: these bitmap histograms show clearly the full range of cells that we sorted and clearly show where the stem and TA cells are located, unlike our previously published bitmap, which showed only the cells in Gate 1:box A [23]). These gates were defined based on where the LRC population was found [23], and debris and signal noise were gated out. Forward and orthogonal scatter signals were generated using 100 mW @ 488 nm. Hoechst 33342 and propidium iodide were excited with 100 mW of UV radiation (351-364 nm). Hoechst fluorescence was measured through a 440/60 nm bandpass filter while propidium iodide was measured through a 670/14 nm bandpass filter. All parameters were collected using linear amplification in listmode. Epidermal cells were sorted by first gating the smallest 25% of the basal population (Gate 1:box A in Fig. 1). To sort the stem cells out of this small cell population, we further gated to collect only the cells showing very low Hoechst fluorescence (Gate 2:box C in Fig. 1). Previously, we showed that these sorted cells contained >95% of the LRC population, could be clonally grown, showed very high proliferative potential, and had the capacity to completely recapitulate an epidermis with very long recombinant gene expression [23]. This also indicates that we are not isolating a diverse population of stem cells of different origins. The TA cells were collected as cells showing a medium forward and orthogonal scatter without regard to Hoechst fluorescence (Gate 1:box B in Fig. 1). These TA cells were shown to only have limited proliferative potential and could not maintain an epidermis [23]. More importantly, no LRCs were found in the population sorted via Gate B.

Figure Figure 1..

Sorting mouse epidermal stem and TA cells using a Coulter Epics 753.Hoechst stained epidermal basal cells were sorted by first gating the smallest 25% of the basal population using bitmap gating (Gate 1: box A). To sort stem cells from this small cell population, a second bitmap gate was set to collect only those cells in gate 1:box A with very low Hoechst red/blue fluorescence (Gate 2: box C). TA cells were collected as cells showing a medium forward and orthogonal scatter (Gate 1: box B) without regard to Hoechst fluorescence (Gate 3). The space between box A and box B (Gate 1) was designed to avoid any overlap or contamination from one population to the other. Cells from box C (Gate 2) showed the highest proliferative potential, while cells from box B (Gate 1) showed limited proliferative potential.

It is important to note that the stem cell population could not be sorted based on size alone because many of the sorted small cells did not show high proliferative potential and could not maintain an epidermis (cells outside box C in Gate 2 in Fig. 1). Stem cells also could not be sorted based on Hoechst dye expression alone because some of the TA cells showed very low Hoechst dye expression (cells in Gate 3 in Fig. 1), but these cells did not show high proliferative potential and could not maintain an epidermis [23]. To reanalyze this profile, we took some of the sorted sample, reran it to see how accurately the population was sorted, and found it to be 99% the same. Furthermore, the Hoechst 33342 dye exclusion was an active process of these cells, as verapamil blocked the exclusion. This technique proved reproducible for both neonatal and adult epidermal cells. The stem cells could be plated and clonally grown in culture for more that 19 population doublings, whereas the TA cells could only be grown in culture for three to four populations doublings. We established that both the sorted stem and TA cells for blastocyst injection were K14+ and K1, indicating that they were undifferentiated epidermal basal cells, since Langerhans cells, melanocytes, fibroblasts, endothelial cells, and any blood-derived cells do not express K14. Thus, both populations were homogenous, which makes them quite different from the Goodell side populations (SP) [24,, 25,, 28].

Blastocyst Injection

To test whether somatic epidermal stem cells could function in a similar manner to embryonic stem cells during development, we injected GFP+ epidermal stem, TA, or unsorted basal cells into 3.5-day blastocysts from C57BL/6 nontransgenic mice that did not carry the GFP gene. This allowed us to clearly follow the lineage of cells and tissues formed from the injected GFP+ epidermal cells, and to distinguish them from cells and tissues formed from the inner cell mass of the C57BL/6 nontransgenic blastocyst. The injection of epidermal cells was done similarly to injection of embryonic cells and was performed by personnel in The University of Iowa Gene Targeting Core. Each C57BL/6 blastocyst was injected with 15-20 sorted GFP+ epidermal stem, TA, or unsorted basal cells. Nine to 15 injected blastocysts were transferred to the uterus of each ICR pseudo-pregnant recipient mother (10 received stem-cell-injected blastocysts, nine TA-cell-injected blastocysts, and three unsorted basal-cell-injected blastocysts). Thirteen out of the 22 mothers became pregnant (eight implanted with stem-injected blastocysts, four implanted with TA-injected blastocysts, and one implanted with unsorted basal-cell-injected blastocysts). Seven embryos from one mother implanted with stem-cell-injected blastocysts and two embryos from one mother implanted with TA-cell-injected blastocysts were harvested 10 days after blastocyst implantation (developmental day E13.5). The rest of the mothers gave birth. Seven mothers implanted with stem-injected blastocysts gave birth to a total of 15 pups; all survived. Three mothers implanted with TA-injected blastocysts gave birth to a total of nine pups; all survived. One mother implanted with total basal-cell-injected blastocysts gave birth, but the mother killed all of the pups immediately after birth.

Histology and Immunohistochemistry

Tissues were harvested at embryonic day 13.5, and at days 13 and 60 after birth. Tissues were soaked overnight in 30% sucrose, embedded in Tissue Tek O.C.T. compound (Electron Microscopy Sciences; Washington, PA;, frozen in liquid nitrogen, and stored at –80°C. Longitudinal sections were made at 8 μm on a cryostat, and the slides were stored at –80°C until used for antibody studies. Each slide containing three sections was fixed for 1 hour in 4% paraformaldehyde at room temperature. Then, one section was stained for expression of GFP, and one section was stained for mouse K14 to determine whether epidermal stem cells retained their original marker. The other continuous section was double stained for GFP and the appropriate antibody for the tissue (see Antibodies below for details). Blood from adult mice was rinsed several times in phosphate-buffered solution (PBS) + 3% FBS, dried as small spots onto glass slides, and stored at –80°C. Bone marrow cells were rinsed from the tibias and femurs of the adult mice by squirting PBS + 3% FBS through the marrow of the bones with a 23-gauge needle and syringe. The dilute bone marrow isolate was collected, rinsed several times with PBS + 3% FBS, centrifuged, dried as small spots onto glass slides, and stored at –80°C. Bone marrow and blood slides were rinsed first in PBS, then fixed and stained with antibodies as described above for tissue sections.


Each section was doubled stained with two primary antibodies (rabbit anti-GFP and a primary antibody specific for that tissue), except for K14. The antibodies for GFP and K14 were both made in rabbits and, thus, were used on continuous sections. After antibody staining, sections were mounted in VectaShield (VECTOR; Burlingame, CA;, and double-exposure photographs were taken on a Nikon Eclipse E600 fluorescent microscope. The primary antibodies used in this study were: rabbit anti-GFP (1:1,000; Medical and Biological Laboratories; Nagoya, Japan;; rabbit anti-mouse K14 (stains keratinocytes; 1:1,000; BAbCO; Richmond, CA); rat anti-pecam-1 (stains vessels and basement membranes; 1:1,000; PharMingen; San Diego, CA;; goat anti-vimentin (stains mesenchymal cells; 1:500; Sigma); rat anti-mouse CD34 (stains late progenitor hematopoietic cells; 1:2,500; PharMingen); rat anti-mouse lineage cocktail (Lin) CD5 (stains T lymphocytes), CD11b (stains [Mac-1] myelomonocytes), CD45R (stains [B220] B lymphocytes), GR1, 7-4 (stains granulocytes), TER119 (stains erythroid cells; 1:100; Stem Cell Technologies; Vancouver, BC, Canada;; rat anti-mouse Sca-1 (stains mouse hematopoietic progenitors; 1:100; PharMingen); mouse anti-glial fibrillary acidic protein (GFAP) (stains astrocytes and glial cells; 1:3,000; Sigma); mouse anti-neuronal nuclei (NeuN) (stains neuronal cells; 1:100; CHEMICON; Temecula, CA;; mouse anti-rat nestin (stains neuronal progenitor cells; 1:1,000; PharMingen). The secondary antibodies used in this study were: swine anti-rabbit fluorescein isothiocyanate (FITC) conjugated (1:100; DAKO; Carpinteria, CA;; goat anti-rabbit biotin conjugated (1:200; VECTOR); goat anti-rat FITC conjugated (1:100; CHEMICON); goat anti-rat biotin conjugated (1:1,000; CHEMICON); donkey anti-goat biotin conjugated (1:1,000; Jackson Immuno; West Grove, PA;; horse anti-mouse biotin conjugated (1:200; VECTOR); streptavidin-Texas Red conjugated (1:800; Sigma).

Polymerase Chain Reaction (PCR) Analysis for EGFP Expression

DNA was isolated using a modification to the protocol recommended by Jackson Labs. Bone marrow or blood cells were resuspended in lysis buffer (0.32 M sucrose, 10 mM Tris-HCl [pH 7.5], 5 mM MgCl, 1% [v/v] TritonX-100), centrifuged, then resuspended in PBND buffer (50 mM KCl, 10 mM Tris-HCl [pH 8.3], 1.5 mM MgCl, 0.45% [v/v] Nonidet P40, 0.45% [v/v] Tween 20) with 60 μg/ml proteinase K in 10 mM Tris-HCl (pH 7.5), incubated at 55°C for 60 minutes, then heated to 97°C for 10 minutes to inactivate the proteinase K. Five hundred ng (for blood) or 1,000 ng (for bone marrow) of DNA were used per 20 μl PCR. To ensure that we detected even a low number of GFP+ cells in the total cell population, we used nested primers in two separate PCRs. The first set (5′ AAG TTC ATC TGC ACC ACC G 3′, 5′ TGC TCA GGT AGT GGT TGT CG 3′) was designed to prime 475 bp of the EGFP gene. The second set of primers (5′ CGA CCA CAT GAA GCA GCA CG 3′, 5′ GTT CTG CTG GTA GTG GTC GG 3′) was just inside the first set of primers and designed to prime 331 bp of the EGFP gene. The first PCR was 37 cycles (90 seconds at 94°C, then 12 cycles of 20 seconds at 94°C, 30 seconds at 62°C [with a reduction of 0.5°C per cycle], 35 seconds at 72°C, then 25 cycles of 20 seconds at 94°C, 30 seconds at 58°C, 35 seconds at 72°C, then 5 minutes at 72°C). To run the second PCR, 2 μl of the first PCR product were used. The second PCR was 37 cycles (90 seconds at 94°C, then 12 cycles of 20 seconds at 94°C, 30 seconds at 68°C [with a reduction of 0.5°C per cycle], 35 seconds at 72°C, then 25 cycles of 20 seconds at 94°C, 30 seconds at 62°C, 35 seconds at 72°C, then 5 minutes at 72°C).


Epidermal Stem Cells Survive the Blastocyst Injection and Incorporate Into Developing Tissues

To test whether somatic epidermal stem cells can function in a similar manner to embryonic stem cells during development, GFP+ epidermal stem or TA cells, isolated by our novel method (Fig. 1), or total epidermal basal cells, were injected into 3.5 day blastocysts. In order to be able to follow the injected epidermal cells long term, we isolated them from the skin of C57BL/6 transgenic mice carrying the enhanced GFP gene [29] (from Jackson Labs) and injected them into blastocysts isolated from C57BL/6 nontransgenic mice (no GFP). Thus any GFP+ cell or tissue found in developing embryos or adult mice had to have been derived from the injected GFP+ epidermal cells. No pups survived from the total-basal-cell-injected blastocysts. Both TA-cell-injected blastocysts and stem-cell-injected blastocysts yielded viable, apparently normal mice, with no visible abnormalities. All mice were of normal size. None of the nine mice derived from TA-injected blastocysts expressed the GFP marker, suggesting that neither the TA cells nor their progeny survived the development process (Figs. 2G-2K). Thus, the TA cell population did not display tissue plasticity. All seven embryos and 15 adult mice derived from stem-cell-injected blastocysts showed GFP+ cells in most of their tissues with a varying degree of mosaicism among mice (Fig. 2). Since only 20 GFP+ cells were injected into each blastocyst and many more GFP+ cells were found, this indicates that somatic epidermal stem cells survived and proliferated after injection into blastocysts. Furthermore, GFP+ cells were present in several developing tissues in E13.5 embryos, including the ectodermally derived epidermis (Fig. 3A) and several mesodermally derived tissues, such as developing vertebrae (Fig. 2C), liver (Fig. 2D), and connective tissue (Fig. 2E), and in the brain (Fig. 2F). These data suggest that our sorting method yields a population of somatic epidermal cells that show the tissue plasticity expected of stem cells. They also suggest that somatic epidermal stem cells have the ability to cross developmental boundaries and give rise to a variety of tissues from all germ layers during development.

Figure Figure 2..

GFP+cells are present in E13.5 embryonic tissues.GFP+cells were found only in E13.5-day embryos derived from 3.5-day C57BL/6 blastocyst-injected GFP+epidermal stem cells (C-F). No GFP+cells were found in embryos derived from blastocysts injected with GFP+TA cells (G-K). Serial sections were stained with hematoxylin and eosin (A, B), and adjacent sections observed under a fluorescent microscope for GFP expression (C, D, E, F). A) Longitudinal section of mid torso. B) Longitudinal section of the crown of the head. Inset boxes C, D, and E in panel A and inset box F in panel B represent the areas shown in panels C, D, E, and F, respectively. GFP+cells are shown in developing vertebrae (C), liver (D), connective tissue (E), and brain (F). Tissues from embryos derived from blastocysts injected with GFP+TA cells show no GFP expression in brain (G, H), developing vertebrae (I, J), or liver (K). Note that to help with orientation by showing the epidermis overlying the GFPtissues, panels H and J are double stained for GFP and K14, and represent the same section shown in panels G and J, respectively. These data indicate that only the GFP+epidermal stem cells survived the blastocyst injection and proliferated and migrated to various tissues during development. Green = GFP; red = K14. Magnification 40× (A, B); 200× (C-E, G-I); 400× (F).

Figure Figure 3..

GFP+cells alter their phenotype and express tissue-specific markers in developing tissues.GFP+cells found in 13.5-day embryonic tissues express markers typical of the tissue in which they reside. A) A cluster of GFP+epidermal cells in the limb. B) An adjacent section to A stained for mouse K14, a marker of epidermal cells. Note: most of the GFP+cells show K14 expression. C) Area of developing dermal tissue double stained with anti-GFP and anti-pecam-1, a marker for endothelial cells in vessels. D) Liver double stained with anti-GFP and anti-CD34, a marker for late hematopoietic progenitor cells in mouse tissues. Note: in the embryo, the liver is the main hematopoietic tissue. E) Dermal tissue double stained with anti-GFP and anti-CD34. F) Brain tissue double stained with anti-GFP and anti-nestin, a marker for neuronal progenitor cells. G) Brain tissue double stained with anti-GFP and anti-NeuN (neuronal nuclei), a marker for neuronal cells. H) Brain tissue double stained with anti-GFP and anti-GFAP, a marker for astrocytes and glial cells. Green = GFP; red = tissue-specific marker; yellow = double expression. Magnification 200× (A, B); 400× (C-H).

Epidermal Stem Cells Have the Capacity to Alter Their Phenotype During Development

To determine whether somatic epidermal stem cells can alter their lineage pathway in response to extrinsic developmental signals, we examined the contribution of GFP+ cells to the formation of embryonic tissues. All tissues that had GFP+ cells showed a varying mosaic pattern of GFP+ cells intermixed with GFP cells (Figs. 2 and 3, Figure 3.). In the developing epidermis, we found large patches of GFP+ cells, and these cells expressed mouse K14 (Figs. 3A, 3B), a marker for all basal cells, including stem cells, that is first detected at E13.5 [30]. In developing mesenchymal tissues, we found that GFP+ cells had altered their phenotype and no longer expressed K14 (dermis in Figs. 3A, 3B). Instead these cells now expressed markers typical of the specific developing tissue. In the skin, we found GFP+ cells that expressed vimentin (data not shown) and pecam-1 (Fig. 3C). In the liver, we found GFP+ cells that expressed CD34 (Fig. 3D). It should be noted that in the embryo the liver is the main hematopoietic tissue. This alteration of phenotype was not limited to the developing mesenchymal tissues, but was also present in the developing brain tissues, where GFP+ cells were found to express nestin (Fig. 3E), NeuN (Fig. 3F), and GFAP (Fig. 3G). These results indicate that epidermal stem cells altered their original epidermal phenotype, and suggest that they have the capacity to alter their original function according to their surroundings.

Altered Cell Lineages Formed From Epidermal Stem Cells During Development Persist into Adulthood

To determine whether GFP+ cells persisted into adulthood, we sampled skin, brain, bone marrow, and blood from 13-day neonates and 60-day adults derived from both stem- and TA-injected blastocysts. We found no GFP+ cells in the tissues from the nine mice derived from TA-injected blastocysts (Figs. 4H, 4I). However, we found GFP+ cells in the tissues from all of the 15 mice derived from stem-cell-injected blastocysts, in both neonatal and adult tissues. In the epidermis, GFP+ cells were clustered in a mosaic pattern of 10 to approximately 100 cells (Figs. 4A, 4D), and expressed K14. GFP+ cells found in tissues other than the epidermis did not retain their original epidermal phenotype and thus did not express K14, instead they altered their phenotypic expression to tissue-specific markers. We also found GFP+ cells in the associated hair follicles, including the bulge region where hair follicle stem cells are known to reside (Fig. 4A) [6,, 10,, 11]. Many GFP+ cells found in the dermis of 13-day and 60-day mice expressed vimentin (Figs. 4A, 4D), pecam-1 (Fig. 4B), or CD34 (Fig. 4G). GFP+ cells found in the brain expressed the brain-specific markers nestin (data not shown), NeuN (data not shown), or GFAP (Fig. 4E). GFP+ cells found in the blood showed positive staining to Lin (Fig. 4C). In the bone marrow, a few GFP+ cells showed Sca-1 expression (Fig. 4F). Only bone marrow and blood from animals derived from the stem-injected blastocysts showed GFP expression via PCR (lanes 6 and 7 in Fig. 5). However, since only a few GFP+ cells were found in the cell smears (Figs. 4C, 4F), we had to perform nested PCR. We used 2 μl of the first PCR reaction product as a template in the second PCR reaction (note: this resulted in two bands of 475 bp and 331 bp in the positive control because the total added amount of the first PCR product [475 bp] was not used up in the second PCR reaction). Overall, these findings suggest that somatic epidermal stem cells retain the ability to produce cells of different cell lineages during development, and that environmental factors may influence or reprogram the lineage pathway for somatic stem cells.

Figure Figure 4..

GFP+cells persist and retain their altered phenotype in adult tissues.GFP+cells were found in all 13-day and 60-day adult tissues derived from stem-injected blastocysts (A-G), but not in tissues derived from the TA-injected blastocysts (H, I). A) 13-day mouse skin from stem-injected blastocyst, anti-GFP and anti-vimentin, a marker for mesenchymal cells. Note the mosaic pattern of GFP+cells in overlying epidermis (arrow); also note scattered cells in the dermis that are both GFP+and vimentin+ (yellow cells). B) 60-day adult skin from stem-injected blastocyst, anti-GFP and anti-pecam-1 (arrow), a marker for endothelial cells in vessels. C) Blood from 60-day adult mouse derived from stem-injected blastocyst, anti-GFP and anti-Lin (lineage cocktail of CD5, CD11b, CD45R, GR1, 7-4, and TER119), markers of blood cell lineages. Note the small green cells (arrow) that do not stain with any of the lineage antibodies in the cocktail. D) 60-day adult skin from stem-injected blastocyst, anti-GFP and anti-vimentin; note the overlying GFP+epidermis and scattered cells in the dermis that are both GFP+and vimentin+(yellow cells, arrow). E) 60-day brain derived from stem-injected blastocyst, anti-GFP and anti-GFAP, a marker for astrocytes and glial cells. F) Bone marrow from 60-day adult mouse derived from stem-injected blastocyst, anti-GFP and anti-Sca-1 (arrows), a marker of mouse hematopoietic progenitors. G) 60-day skin derived from stem-injected blastocyst, anti-GFP and anti-CD34, a marker for late progenitor hematopoietic cells and some endothelial cells in mouse tissues. H) 60-day skin from TA-injected blastocyst, anti-GFP; note the autofluorescence of hair follicles, but no GFP staining in the epidermis or the dermis, including vessels, nerves, sweat glands, or connective tissue cells. I) Same section as in B, doubled stained for GFP and K14, which stains the epithelial cells in the epidermis, hair follicles, and sweat glands. Green = GFP; red = tissue-specific marker; yellow = double expression. Magnification 100× (A); 200× (D, E, H, I); 400× (B, G); 800× (C, F).

Figure Figure 5..

GFP expression is found only in the bone marrow and blood of adult tissues derived from stem-injected blastocysts.Nested PCR was performed on DNA isolated from the bone marrow and blood from 60-day-old mice derived from stem- or TA-injected blastocysts; positive expression yielded a product of 331 base pairs. Lane 1, M = markers; Lane 2, - = negative control; Lane 3, + = positive control, blood from GFP transgenic mouse (note, there are two bands due to leftover product from the first PCR reaction; the top band is 475 bp); Lane 4, BM-TA = bone marrow (TA-cell-injected blastocyst); Lane 5, BL-TA = blood (TA-injected blastocyst); Lane 6, BM-SC = bone marrow (stem-cell-injected blastocyst); Lane 7, BL-SC = blood (stem-cell-injected blastocyst).


The recent explosion in interest in stem cell research is due in large part to the recognition that a broad variety of adult tissues contain stem cells, and that these somatic stem cell populations may be pluripotent, i.e., they retain the ability to produce more than one cell lineage. The first evidence that somatic adult stem cells might be pluripotent was reported in 1961 for the hematopoietic system [31]. Currently, hematopoietic progenitor cells are identified by expression of the cell surface antigen CD34 for human or Sca-1 for mouse and the lack of hematopoietic lineage markers [32,, 33]. More recently, hematopoietic early progenitor cells have been identified as SP cells; these are cells that exclude Hoechst 33342 dye and lack lineage markers [24,, 25,, 28]. These pluripotent hematopoietic cells express the three basic stem cell characteristics of self-renewal, unlimited proliferative capacity, and the ability to differentiate into various mature circulating blood cells (lymphocytes, macrophages, myelomonocytes, granulocytes, erythrocytes, etc.) [24,, 34-, 37]. In the past decade, it has been shown that pluripotentiality is not limited to hematopoietic stem cells. Stem cells located in intestinal crypts have been shown to provide all four cell types of the small intestine (paneth, enteroendocrine, goblet, and intestinal epithelial cells) [38]. In the eye, both epithelial and goblet cells are derived from conjunctival stem cells, whereas corneal cells are derived from limbal stem cells [39-, 42]. Recently, the term pluripotent has been expanded to include multipotent cells, i.e., cells that can transdifferentiate into multiple tissues. For example, skeletal muscle cells were shown to have the capacity to differentiate into hematopoietic cells [28], and adipose cells differentiated into chondrogenic, myogenic, and osteogenic cells in the presence of lineage-specific induction factors [43]. Pluripotent mesenchymal cells were also reported [44]. In a more complicated study, adult neural stem cells injected into an irradiated host produced blood cells, including myeloid, lymphoid, and early hematopoietic cells [45], and hematopoietic stem cells injected into the brain expressed neuronal phenotypes [46,, 47]. In these studies, neural-crest-derived stem cells produced differentiated cells that were embryologically derived from the mesoderm, and mesodermally derived stem cells produced neural-crest-derived cells. Very recently, Toma et al. reported that multipotent stem cells were present in the dermis of mammalian skin [48,, 49]. They showed that under certain culture conditions, these cells preferentially grew as floating spheres of cells that expressed both fibronectin and nestin, markers for neuronal stem cells. If they changed the culture conditions, these dermal stem cells produced markers for smooth muscle and adipocytes. They did not show that dermal progenitors could differentiate into epidermal cells. Our study differs from their in vitro study in that we show multipotency of epidermal stem cells in vivo, and we show that epidermal stem cells retain the capacity to produce cells from all three germ layers. All of these studies, including our study reported here, suggest that altering the environment of somatic progenitor cells allows them to show a much broader potential than expected. Furthermore, there is some evidence from the Drosophila field that this may be a general phenomenon evolutionarily inherited [49].

One of the main problems in determining whether somatic stem cells retain a pluripotent capability is isolating a homogenous population to test. Last year, Clarke et al. cultured homogenous neural cells, then injected them as adherent neurospheres into chick amniotic cavities, or dissociated them and injected the neural cells directly into mouse blastocysts [50]. The culture conditions ensured that these cells were homogenous. The somatic neural stem cells retained a very broad developmental capacity and the ability to contribute to the formation of a variety of organs in both mouse and chick embryos. They examined stage 8 chick and E11 mouse embryos and found a mosaic pattern of cells derived from the neural stem cells; they did not examine the adult animals and they did not test any other cell population. Our data suggest that epidermal stem cells may also have this total pluripotent ability. At least our data suggest that they have the ability to be reprogrammed by the developmental environment. Our method to ensure isolation of a homogenous population differs from that of Clark. We use a laser cell sorting method, modified from a previously published method [24], with gates designed around the position of LRCs [2] to ensure that we isolated a population of homogenous epidermal stem and TA cells. Even though our sorting method differs from the Goodell method [24] in that we also employed a size differential and do not achieve an SP pattern of cells, the dye exclusion property of the epidermal stem cells appears to be similar to that of the hematopoietic SP cells. Both exclude the Hoescht 33342 dye in a verapamil-dependent manner. We have previously examined the quality and integrity of the sorted epidermal stem and TA cell populations and determined that they were homogenous in that they all expressed K14 and not K1 or other cell markers, but only the stem population could be grown clonally or could maintain an epidermis long term [23]. We should also note that the SP cells isolated from the bone marrow, which showed high engraftability, were not homogenous [24], most likely because the bone marrow population to be sorted contained stromal, adipose, and hematopoietic cells, and thus was not initially homogenous, whereas our initial population contained only epidermal basal cells. Thus, our study differs from previous studies in that we were able to test homogenous populations of both the stem and the TA epidermal cells for pluripotency. We found that only the somatic epidermal stem cells could contribute to a variety of tissues during the development of a mouse; the TA cells could not, thereby demonstrating a fundamental difference between these two homogenous cell populations. We also found that the epidermal stem cells altered their phenotype to express a variety of tissue-specific markers, including some normally found in the hematopoietic system, which differs from the neural stem cells that had no contribution to the hematopoietic system [50]. However, had Clarke et al. tried nested PCR, as we did, they may also have found that the neural stem cells contributed to the hematopoietic system at a low level. Such studies suggest that several somatic tissues contain a population of stem or progenitor cells that can be reprogrammed by changing their in vitro environment or by placing them in a developmental environment in vivo.

We also found that these “reprogrammed” cells persisted into adulthood, a trait long proposed for stem cells [5,, 13,, 51]. Very recently, it was reported that if hematopoietic stem cells were significantly enriched by recovering cells that home to the bone marrow within 48 hours, the resultant cell population could incorporate and function in a variety of tissues, including stomach, kidney, lung, and skin [52]. Further, these cells persisted for 11 months after grafting; they were identified as male cells in a female host. Such long-term persistence of grafted male cells in females had been previously reported [53], as well as the persistence for 27 years of male cells in the blood of females who had given birth to male offspring [54]. Such data suggest that stem cells may exist in all tissues, that these stem cells may have the capacity to become any tissue [55-, 57], and that stem cells have a very long life. It has been proposed that individual stem cells might persist throughout the lifetime of an organism [5,, 51]. Cells that retain a nuclear tritiated thymidine label have been shown to persist for at least 6 months in the interfollicular epidermis [2,, 58] and for 14 months in the hair follicle [6]. Although life-long persistence of stem cells has not been proven, it is an attractive hypothesis and well worth further study.


Our data show that epidermal stem cells survive injection into blastocysts, proliferate during development, produce tissues from all three germ layers in embryonic mice, and that the resultant chimeric tissues persist into adulthood. Thus, the skin contains a population of cells in the epidermis that retains the ability to be reprogrammed if transplanted into an embryonic environment. Whether this will hold true for transplantation into an adult environment remains to be determined. These data suggest that the skin might be a source of easily obtainable stem cells that retain the ability to be reprogrammed. If true, the advantages could be tremendous. Not only is the skin the largest organ in the body and potentially contains the greatest number of stem cells, but it could provide an alternative to using embryonic stem cells or other somatic stem cells that are more difficult to retrieve as therapeutic treatment for disease.


We thank the staff of the UI Gene Targeting Core for its technical expertise in injecting the isolated epidermal stem cells into 3.5-day blastocysts, and Dr. Baoli Yang, Associate Director, for his excellent advice. We also thank Justin Fishbaugh and Gene Hess in the UI Flow Cytometry Core for their excellent assistance in sorting the epidermal stem cells. A special thank you to Jason Marley for his help with the antibody staining and to the other members of the Bickenbach lab for their helpful advice. This project was supported by a grant from The National Institute for Arthritis and Musculoskeletal and Skin Diseases at The National Institutes of Health (RO1 AR45259, J.R.B.).