Materials and methods

Animals

Female FVB/N, v-Ha-ras transgenic Tg.AC, and BALB-c mice were obtained at 4–6 wk of age from Taconic Farms (Germantown, NY), CD-1 mice (6 wk of age) from Charles River (Raleigh, NC), and C57Bl/6 mice were obtained at 6 wk of age from NCI (Frederick, Maryland). Mice were held in our animal facility under a 12 h light-dark cycle and given food and water ad libitum, in accordance with NIH Guidelines for Humane Care and Use of Laboratory Animals.

Immunohistochemistry

Formalin-fixed mouse skin was deparaffinized, rehydrated, and then subjected to antigen retrieval by steam heating for 30 min in 0.01 M citrate buffer (pH 6). Tissues were incubated with either rat antimouse CD34 or rat IgG isotype control (BD Pharmingen, San Diego, CA) at 1 : 50 for 1 h at room temperature. Signal was detected using the Vectastain Elite Rat IgG detection kit (Vector Laboratories, Burlingame, CA), following the manufacturer's protocol, with diaminobenzidine (Dako, Carpinteria, CA) as the chromagen. Tissues were immersed in 1% CuSO₄ to enhance membrane staining, and then tissues were counterstained in Harris hematoxylin (Sigma-Aldrich, St. Louis, MO). K15 staining was performed in Bouin's fixed mouse skin, using chicken antihuman K15 (developed by George Cotsarelis) at 1 : 100 overnight at 4°C. Positive staining was detected with the ABC reagent from the Vectastain standard detection kit (Vector Laboratories) after incubating 30 min with biotinylated goat antichicken IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) at 1: 100. Tissues were counterstained with Harris hematoxylin following incubation with diaminobenzidine.

Fluorescence staining

Dorsal skin strips 2 cm by 0.5 cm from Tg.AC mice were rolled up, epidermis side out, placed in cryovials, flash frozen in liquid nitrogen, and then embedded in Cryogel (Instrumedics, Hackensack, NJ). Following acetone fixation and blocking of frozen sections, tissues were incubated for 1 h at room temperature with either primary antibody [fluorescein isothiocyanate (FITC) conjugated rat antihuman 6 integrin (CD49f) at 1: 300, OR FITC-conjugated rat antimouse CD34 at 1: 250], or FITC-conjugated isotype controls (BD Pharmingen). Tissues were washed three times in 1 phosphate-buffered saline and coverslipped with Vectashield with DAPI (Vector Laboratories) prior to confocal microscopy.

Confocal microscopy

A laser scanning confocal microscope (LSM 410 mounted on an Axiovert 135 microscope; Carl Zeiss) was used to obtain fluorescent images. Image stacks of roughly 15 m through the z dimension, at steps 1.0 m apart, were obtained sequentially for each fluorescent channel using a Zeiss C-Apo 40x W 1.2 DIC as the objective lens. For FITC fluorescence, the 488 nm laser line of an Omnichrome argon-krypton ion laser was used for excitation, and a 515–565 nm bandpass filter was used for the emission. For DAPI fluorescence, the 351 nm and 364 nm laser lines of a Coherent Enterprise argon ion laser were used for excitation, and a 410–505 nm bandpass filter was used for the emission. Later, maximum intensity "through-focus" projections of the image stacks were generated in postprocessing, and contrast-stretching was applied uniformly across images from each channel. The software used for acquisition was Zeiss LSM version 3.98 for Windows, and for post-acquisition three-dimensional projections, Zeiss LSM Image Examiner (licensed) version 2.8 for Windows NT.

Keratinocyte isolation and flow cytometry

Keratinocytes were harvested from the dorsal skin of five mice per experiment as previously described (^{Morris et al, 1990}), using Tg.AC, FVB/N, C57Bl/6, BALB-c, and CD-1 mice at 7–8 wk of age. Viable cell counts were determined using 0.4% Trypan Blue. Keratinocytes were incubated for 30 min in the dark at room temperature with FITC-conjugated rat antihuman 6 integrin at 20 l per 10⁶ cells and biotin-conjugated rat antimouse CD34 at 1 g per 10⁶ (BD Pharmingen). Streptavidin Cy-Chrome Conjugate (BD Pharmingen) was added at 20 l per 10⁶ cells for indirect immunofluorescent staining of biotin-CD34 labeled cells. Non-fixed, intact cells were examined by flow cytometry using a Becton Dickinson FACSVantage SE, with propidium iodide (PI; 10 g per ml) added to each sample to identify cells with loss of membrane integrity. Cells were excited with a 488 nm argon laser with an initial gate set on a forward scatter versus PI (575/26 nm) dot plot to exclude dead cells, with 10,000 cells from the initial gate examined for each sample. 6 integrin and CD34-Cy fluorescence were detected at 530/30 nm and 660/20 nm, respectively, and analyzed using Cell Quest software.

Immunocytochemistry

Cytospin (Shandon, Pittsburgh, PA) slides were prepared from flow sorted keratinocyte populations (5000 cells per slide). Cells were fixed in acetone at -20°C, and endogenous biotin blocked using the Avidin/Biotin blocking kit (Vector Laboratories). For all antibodies except vimentin, nonspecific protein was blocked for 1 h using the Vectastain Rabbit IgG kit (Vector Laboratories). Antibodies used included rabbit anticow cytokeratin (Dako), vimentin (Dako), rabbit antimouse K6, K14, K1, and loricrin (all from Covance, Richmond, CA). Negative control was rabbit IgG (Vector Laboratories) at the same protein concentration as primary antibody. Vimentin staining was accomplished using Dako EPOS vimentin/HRP clone 3B4 with negative control immunoglobulins conjugated to horseradish peroxidase serving to demonstrate staining specificity. Antibody concentrations are given in Table I. Frozen sections of normal mouse skin served as the positive control for all antibodies examined.

Table I - Staining Properties of Flow Sorted Keratinocytes.

Full table

Cell cycle analysis

6⁺CD34⁺, 6⁺CD34^–, and unseparated cells were isolated by flow cytometry, pelleted, resuspended in 5 ml of cold 70% ethanol, and stored at 4°C. Ethanol-fixed cells were pelleted, washed once in phosphate-buffered saline, and then stained with a PI solution (20 g per ml PI per 1000 units of RNase ONE) (Promega, Madison, WI) for 20 min at room temperature. PI stained cells were examined using a Becton Dickinson FACSort, with an initial gate set on a PI area versus width dot plot for doublet discrimination. 7500 cells were examined for each sample. Margins were set around the visible G₂/M peak and analyzed using Cell Quest software.

Colony forming assay

Keratinocytes were harvested from 7-wk-old C57Bl/6 mice, and then 6⁺CD34⁺, 6⁺CD34^–, and the unseparated mixture ("all sorted"; mixture of CD34⁺ and CD34^– keratinocytes) were isolated by flow cytometry. One thousand cells per population were seeded onto irradiated Swiss 3T3 cells in 60 mm dishes and grown at 32°C/5% CO₂ in William's E medium (Life Technologies) containing 20% fetal bovine serum, gentamicin, and supplements. Cultures were grown for 2 wk, fixed in 10% NBF, (neutral buffered formalin) and then stained with 1% Rhodamine B to visualize colony growth. The remaining cultures were grown for an additional 2 wk (total of 4 wk in culture) and then fixed and stained as described above. Colony size was measured using two dishes per population per experiment, over a total of three separate experiments.

Reverse transcription polymerase chain reaction (RT-PCR) detection of CD34 expression

Keratinocytes from the skins of five 12-O-tetradecanoylphorbol-13-acetate (TPA) treated and five untreated female Tg.AC mice were harvested and subjected to flow cytometric analysis as described above. The shaved dorsal skin surface was dosed with 5 g TPA (Sigma Aldrich, St. Loius, MO) in 200 l acetone twice weekly for 2 wk, and then keratinocytes were harvested 24 h after the last treatment. Cells were pelleted and flash frozen in liquid nitrogen, and total RNA was extracted using the MicroRNA Isolation kit (Stratagene, La Jolla, CA); then cell populations were analyzed for CD34 expression. Following reverse transcription (42°C 15 min, 99°C 5 min, and 4°C 5 min), samples were amplified (35 cycles: 94°C 30 s, 55°C 30 s, and 72°C 1 min). Primers were as follows: antisense 5'-GCTCTCTGCCTGATGAGTCTG and sense 5'-CCCTTAATGGCAC-TCGGAGC, generating a 455 bp product. RNA integrity was assessed using expression of mouse 2 microglobulin as previously described (^{Battalora et al, 2001}).

Statistical analysis

Comparisons of relative 6-FITC fluorescence between 6⁺CD34⁺ and 6⁺CD34^– keratinocytes were carried out by paired Student's t tests, using fluorescence data from six separate experiments. Keratinocytes were harvested in all six experiments from 7-wk-old C57Bl/6 mice. For comparisons of colony size between keratinocyte populations, analysis of variance was used to assess the significance of colony size differences, replicate variability, and the interaction between the two factors. The variance-stabilizing logarithmic transformation was used in these analyses.

Results

CD34 expression in the follicular bulge region

Immunostaining of formalin-fixed, paraffin-embedded skin taken from normal, untreated Tg.AC mice revealed intense membrane localization of CD34 in the hair follicle bulge (Figure 1a). The area of CD34 staining coincided with the location of LRCs, which mark slowly cycling KSCs in vivo (^{Cotsarelis et al, 1990;Morris and Potten, 1999}) (Figure 1b), and with K15 expression, which is preferentially expressed in the hair follicle bulge in human and mouse skin (^{Lyle et al, 1998;Porter et al, 2000}) (Figure 1c). A similar pattern of CD34 expression was found in skin from FVB/N, C57Bl/6, DBA/2, CD-1, BALB-c, and SENCAR mice (data not shown). Expression of CD34 is prominent during telogen and is also evident in anagen hair follicles (Figure 1d, telogen; Figure 1e, anagen). We did not detect expression in other hair follicle structures such as the bulb, or in the interfollicular epidermis.

Figure 1.

Localization of CD34 expression, LRCs, and K15 expression in murine hair follicles. (A) CD34 expression in formalin-fixed, paraffin-embedded dorsal skin from an untreated Tg.AC mouse, showing membrane localization in the hair follicle. Arrow indicates CD34-positive cells. (B) Formalin-fixed skin from a CD-1 mouse injected with tritiated thymidine followed by a 10 wk chase. Arrow indicates LRCs. (C) K15 expression in Bouin's fixed, paraffin-embedded skin from an untreated Tg.AC mouse. Arrow indicates K15-positive bulge keratinocytes. (D) CD34 expression in a murine telogen hair follicle. (E) CD34 expression in a murine anagen hair follicle. Arrow indicates CD34-positive keratinocytes in panels D and E. Scale bars: 100 m.

Full figure and legend (193K)

To characterize further CD34 expression in mouse skin, we used an FITC-conjugated form of the antibody on frozen sections for direct labeling (Figure 2e). With confocal microscopy, we show similar membrane localization in the hair follicle (Figure 2e). Figure 2(b) represents merged images of CD34-FITC (green) and DAPI (blue) staining to highlight the nuclei and orient CD34 expression in the follicle (E, epidermis; D, dermis; HF, hair follicle, B, bulge; HS, hair shaft). We also examined 6 integrin, using an FITC-conjugated antibody, to demonstrate that 6 integrin is expressed only on basal keratinocytes (Figure 2e, e represents the merged images of 6-FITC staining and DAPI). As described below, this characteristic 6 integrin expression was used in flow cytometric analysis of keratinocytes, to exclude suprabasal cells as well as nonkeratinocytes. Figure 2(c), (f) shows FITC-conjugated isotype control (rat IgG), the negative control for CD34-FITC, and 6-FITC staining, respectively. In panels C and F, the dotted line marks the epidermal-dermal junction and the hair shafts are labeled (HS), highlighting the lack of staining in the sections using isotype controls.

Figure 2.

Fluorescence detection of CD34 expression and 6 integrin in mouse skin. Frozen dorsal skin sections were stained with either FITC-conjugated CD34 or FITC-conjugated 6 integrin, coverslipped with mounting medium containing DAPI, and subjected to confocal microscopy for imaging. (A) CD34-FITC staining in hair follicle. (B) CD34-FITC staining merged with DAPI to highlight epidermal and follicular nuclei. E, epidermis; D, dermis; HF, hair follicle; B, bulge. (C) Isotype control (rat IgG-FITC) for CD34. HS, hair shaft. (D) 6 integrin-FITC, with signal localized to the basement membrane aspect of basal keratinocytes. (E) 6-FITC staining merged with DAPI. (F) Isotype control (rat IgG-FITC) as negative control for 6 integrin staining. Scale bars: 50 m.

Full figure and legend (84K)

Isolation and characterization of 6⁺CD34⁺ cells isolated from mouse keratinocytes

When harvested keratinocytes were stained with antibodies to 6 integrin (FITC conjugate) and CD34 (biotin conjugate, with Streptavidin-CY secondary) and then subjected to FACS analysis, a distinct and reproducible 6⁺CD34⁺ population was found, which comprised 7%–8% of 6⁺ keratinocytes and is shown in the upper right quadrant of Figure 3(a). Most of the cells were 6⁺CD34^– (approximately 64%), as shown in the lower right quadrant of the plots. A small population (approximately 2%–3%) of 6^–CD34⁺ cells was also found, as shown in the upper left quadrant of the plots (Figure 3e). This population was not consistently observed when the cells were analyzed using CD34-FITC and 6 integrin directly conjugated to CY-5 (BD Pharmingen). No change was observed with respect to the 6⁺CD34⁺ population, however, when the fluors were switched (data not shown), demonstrating the validity and reproducibility of this population of cells and therefore strengthening our data.

Figure 3.

Analysis of 6/CD34 expression in various mouse strains by flow cytometry. (A) Keratinocytes were isolated from 7- to 8-wk-old Tg.AC, C57Bl/6, FVB/N, BALB/c, and CD-1 mice, stained with 6 integrin-FITC and CD34-Cy and examined by flow cytometry. Negative controls were prepared using isotype control antibodies to determine staining specificity (not shown). 6⁺CD34⁺ cells are shown in the upper right-hand quadrant of each contour plot, whereas 6⁺CD34^– cells are shown in the lower right-hand quadrant, with percentage of each sorted cell population for all five strains examined listed. (B) Analysis of CD34 expression by RT-PCR (upper panel, 455 bp product). Lanes 1, 4, unseparated cells; lanes 2, 5, CD34^– cells; lanes 3, 6, CD34⁺ cells (from TPA-treated and untreated skin, respectively); lane 7, positive control (spleen). Lower panel shows mouse 2 microglobulin (212 bp) expression to assess RNA integrity.

Full figure and legend (67K)

Because monoclonal antibodies may cross-react with proteins unrelated to their original targets (^{Kramer et al, 1997;Weinberg and Yuspa, 1997;Lyle et al, 1998}), and because CD34 expression in keratinocytes has not been previously reported, we confirmed the specificity of the CD34 antibody used in these experiments. Because steady-state expression of CD34 message is correlated to the presence of cell surface CD34 (^{Krause et al, 1994}), we analyzed the sorted cell populations for expression of the CD34 message in cells harvested from both TPA-treated and nontreated mouse skin using RT-PCR (Figure 3e). We examined cells from treated and untreated mice in order to assess whether the CD34 mRNA expression pattern would be affected by the proliferative status of the skin. CD34 expression was detected in the unseparated mixture (Figure 3e, upper panel, lanes 1, 4) and in CD34⁺ cells (lanes 3, 6). Only faint product was evident in CD34^– cells (lanes 2, 5), probably reflecting the presence of contaminating CD34⁺ cells and highlighting the sensitivity of the PCR amplification. RNA integrity was assessed with expression of mouse 2-microglobulin (Figure 3e, lower panel). Therefore, we confirmed the presence of CD34 mRNA in the sorted population in both untreated and TPA-treated skin, and demonstrated the specificity of the CD34 antibody used in flow cytometry experiments. Furthermore, we found that the compartmentalization of CD34 mRNA expression in CD34⁺ keratinocytes did not change in TPA-treated skin.

To address further the specificity of the antibody used in these studies, we stained formalin-fixed sections of mouse skin with an antibody against CD34 from a different source (rat antimouse CD34, catalog number ab8158-100, Abcam, Cambridge, U.K.). We found similar localization of expression to bulge keratinocytes in the hair follicle (data not shown). Therefore, based on the expression of the CD34 in CD34⁺ sorted cells, and similar location to bulge cells using a different CD34 antibody, we conclude that the antibody used in immunohistochemistry and flow cytometry is specifically recognizing the CD34 epitope on harvested keratinocytes.

Stem and progenitor cells would be expected to express higher levels of integrins to facilitate attachment to the basement membrane. Previous investigators have identified a subset of keratinocytes expressing relatively high levels of 1 integrin that are thought to be stem cells (^{Jones and Watt, 1993}). This has not been observed with 6 integrin (^{Li et al, 1998}). To determine if a difference in 6 integrin expression was present in our CD34⁺ cells, we examined the relative mean 6-FITC fluorescence of CD34⁺ and CD34^– keratinocytes using data collected from six individual experiments, with cells collected from 7-wk-old C57Bl/6 mice (Figure 4). Examination of keratinocytes stained with 6 and CD34 did show a detectable increase in 6 levels in the CD34⁺ population (Figure 4e). Quantitation of 6⁺CD34⁺or 6⁺CD34^– cells from six different experiments showed that CD34⁺ cells had a statistically significant (p<0.05) increase in the level of 6 fluorescence compared to CD34^– cells (Figure 4e). Although variability in the degree of increased 6 fluorescence was observed in any given experimental pair, an overall increase in the level of 6 expression was always observed between the CD34⁺ and CD34^– cells. This trend was also apparent in cells harvested from other strains as well (data not shown). The consistency of this observation over six separate experiments using C57Bl/6 mice from different litters strengthens the finding. In spite of the fact that the difference in mean channel fluorescence is small, it was statistically significant (p<0.05). These data indicate that 6⁺CD34⁺ cells comprise a subset of the 6-integrin-positive population that have high levels of 6 integrin expression.

Figure 4.

Analysis of 6 integrin fluorescence in CD34⁺ and CD34^– keratinocytes. Keratinocytes were isolated from C57B1/6 Mice, stained with 6 integrin-FITC and CD34-Cy, and examined by flow cytometry. 6⁺CD34⁺ and 6⁺CD34^– were gated as described in Figure 3. (a) Contour plot demonstrating relative increase in 6 fluorescence in CD34⁺ cells compared to CD34^– cells. (B) Quantitation of 6 fluorescence in CD34⁺ and CD34^– cells. The mean 6-FITC fluorescence for each individually gated population of cells was determined using CellQuest software.

Full figure and legend (55K)

Because CD34 is expressed on other cell populations in the skin, such as endothelial and dendritic cells (^{Nickoloff, 1991}), we confirmed that our cell sorting strategy was preferentially isolating keratinocytes by staining cytospin preparations with antibodies to various cytokeratins, vimentin, and CD31 (Table I). All flow-sorted subpopulations were positive for a broad spectrum cytokeratin antibody as well as K14 (Table I,Table II), specific for basal cells (^{Fuchs and Green, 1980}). In addition, the cells were positive for K6, although 6⁺CD34⁺ were enriched compared to 6⁺CD34^– cells (Table I,Table II), demonstrating that CD34⁺ cells were follicular in origin (^{Roop et al, 1984}). All of the cell populations were negative for K1, an early differentiation marker (^{Roop et al, 1983;Regnier et al, 1986}), for loricrin, expressed only in terminally differentiated stratum granulosum cells (^{Mehrel et al, 1990;Fuchs and Byrne, 1994}), and for vimentin (^{Langa et al, 2000}), expressed in cells of mesenchymal origin. To assess further the incidence of K6 and K14 staining in the flow-sorted cell populations, antibody-stained slides from three separate experiments were examined for the numbers of keratin-positive cells (Table II). Three hundred to five hundred consecutive cells were counted using a Zeiss Axioplan microscope (Carl Zeiss, Thornwood, NY) at 100magnification and the percentage of keratin-positive cells was calculated (Table II). Virtually all cells from all three populations were K14 positive, as expected of cells derived from the basal keratinocyte compartment. In contrast, only 7%1% and 12%2% of the CD34^– and all sort cells, respectively, were K6 positive, whereas 98%1% of the CD34⁺ cells were found to express K6, demonstrating the enrichment for follicular cells using CD34 and 6 integrin antibodies.

Table II - Keratin 6 and 14 Staining Incidence in Flow Sorted Keratinocytes.

Full table

Finally, cells were assessed for expression of the endothelial cell determinant CD31 (^{Albeda et al, 1991}). All sorted cells were negative for CD31 expression (Table I). Hence, double immunostaining with 6 and CD34 in combination with FACS enabled us to isolate a minor subpopulation of basal keratinocytes derived from the hair follicle bulge.

6⁺CD34⁺ cells are quiescent and form large colonies in culture

In the skin, stem cells are thought to be normally slowly cycling to provide for long-term tissue maintenance as well as to persist over the lifetime of the animal (^{Lavker et al, 1993;Lavker and Sun, 2000;Fuchs and Raghavan, 2002}). Because CD34 expression localized to a subset of cells in the hair follicle bulge region and appeared to identify LRCs, we were interested in characterizing the cell cycle properties of these cells. Figure 5 summarizes the cell cycle analysis of 6, CD34 sorted cells from three different mouse strains. 6⁺CD34⁺ cells were enriched for cells in G₀/G₁ relative to 6⁺CD34^– cells, and had few cells in S phase and G₂/M, consistent with quiescence (Figure 5). In addition, CD34⁺ cells were smaller than CD34^– cells (data not shown), suggesting a more primitive morphology. Hence, our results demonstrate that 6⁺CD34⁺ cells are small, slowly cycling cells, properties of stem and progenitor cells.

Figure 5.

Cell cycle analysis of 6 integrin, CD34 stained mouse keratinocytes. Cells were isolated from 7-wk-old C57Bl/5, BALB-c, and CD-1 mice and examined for changes in DNA content. (A) Representative histograms derived from 6⁺CD34⁺, 6⁺CD34^–, and "all sorted" cells from 7-wk-old C57Bl/6 mice. (B) Dot plot analysis of the three cell populations (G₂/M, G₀/G₁, and S phase) derived from three mouse strains (C57Bl/6, BALB-c, and CD-1).

Full figure and legend (47K)

When sorted keratinocytes were seeded at clonal density, both CD34⁺ and CD34^– cells formed colonies, as did the unseparated (all sorted) cells (Figure 6). After 2 wk in culture, there was little if any difference in colony size among the three populations. After 4 wk, however, there was a clear enrichment for larger colonies from CD34⁺ keratinocytes (Figure 6; Table III), indicating a higher proliferative potential for CD34⁺ keratinocytes. Analysis of variance, using data from all three experiments, revealed a significant (p<0.05) difference in colony size between the CD34⁺ and CD34^– populations. Although this difference was evident in all three experiments, the actual magnitude of the colony size differences varied significantly (p<0.05) among the three experiments (Table III). Despite the small sample size, however, there was a statistically significant difference in colony size between the two populations, indicating that CD34⁺ keratinocytes consistently form larger colonies in clonogenic culture than CD34^– cells. Taken together, we conclude that CD34 identifies a slowly cycling population of hair follicle bulge keratinocytes that form large colonies in cultures, evidence that CD34 is a determinant for keratinocytes with stem and progenitor cell characteristics.

Figure 6.

Colony growth of 6 integrin, CD34 stained keratinocytes. Keratinocytes were harvested from 7-wk-old female C57Bl/6 mice and sorted using FACS under sterile conditions. CD34⁺, CD34^–, and the unseparated mixture ("all sorted") were grown for 4 wk on irradiated Swiss 3T3 cells, fixed in 10% NBF, and stained with Rhodamine B to visualize colony growth.

Full figure and legend (169K)

Table III - Colony size of flow sorted keratinocytes grown in clonogenic culture.

Full table

Discussion

We report here that the hematopoietic stem and progenitor cell determinant, CD34, localizes to the bulge region in mouse skin, and in combination with 6 integrin identifies a minor population of slowly cycling keratinocytes.

The bulge is a heterogeneous collection of cells in the middle permanent region of the hair follicle, and it is widely considered to be the location of KSCs (^{Cotsarelis et al, 1990;Morris and Potten, 1994;Lavker and Sun, 2000}). Besides being protected from wounding or environmental damage, the bulge creates an area isolated from the changes that occur to the follicle as it cycles from anagen (active growth) to telogen (resting), providing a sequestered microenvironment for stem cells. Moreover,^{Oshima et al (2001)} elegantly demonstrated that bulge keratinocytes of vibrissa follicles can produce the hair follicle, sebaceous gland, and the interfollicular epidermis.

It is generally accepted that LRCs mark KSCs in vivo (^{Cotsarelis et al, 1990;Morris and Potten, 1999;Lavker and Sun, 2000}). The skin of neonatal mice receiving repeated injections of ³H-thymidine followed by a long chase period retains the label in cells sequestered in the bulge region, with very few labeled cells found in the interfollicular epidermis. This supports the hypothesis that these cells are slowly cycling and most likely represent a stem cell population. Moreover, an ability to respond to proliferative stimuli supports the role stem cells play in maintaining tissue homeostasis and regeneration over the lifetime of the animal (^{Cotsarelis et al, 1999}). This was demonstrated by^{Morris and Potten (1994)}, who showed that LRCs in vivo are clonogenic in vitro (^{Morris and Potten, 1994}). In contrast, the TA cells identified by pulse labeling do not form large colonies in culture. Therefore, LRCs in the hair follicle bulge represent slowly cycling cells with a high proliferative capacity, both well recognized hallmarks of stem cells.

Bulge cells express both K19 (^{Lane et al, 1991;Michel et al, 1996}) and K15 (^{Lyle et al, 1998}). Nevertheless, K19 expression is not unique to the bulge because, during hair growth, expression extends down the outer root sheath.^{Lyle et al (1998)} observed that K15-expressing cells remain localized to the bulge region in human skin regardless of the growth stage, and are a subset of K19 cells in the follicle. In resting follicles, K15 and K19 are colocalized, suggesting K15 cells in the bulge give rise to K19 cells (^{Lyle et al, 1998}), but further studies are necessary to determine a progenitor-progeny relationship. In addition, K15-expressing bulge keratinocytes were found to correspond to LRCs and express high levels of 1 integrin, suggesting that this is a population of stem and progenitor cells (^{Lyle et al, 1998}).

The expression pattern and colocalization of CD34 with K15 and LRCs in the hair follicle bulge suggested that this cell surface marker might be a selectable determinant for living follicular keratinocytes (neither K15 protein nor LRCs enable this selection). Cell populations with stem-cell-like properties can be isolated from primary keratinocytes using cell surface markers in combination with flow cytometry, including 1 integrin (^{Jones and Watt, 1993}), and a negative selection strategy using 6 integrin in combination with CD71 (^{Tani et al, 2000}). CD34, used with 1 integrin, identifies a subset of 1 integrin⁺CD34⁺ cells (data not shown); however, the use of 6 integrin with CD34 enables positive selection for follicular bulge cells and enriches this population. Furthermore, our analysis provides evidence of a subset of 6 integrin bright cells not apparent in cells stained with 6 alone (^{Li et al, 1998}). KSCs may express more integrin in order to maintain tight adherence to the basement membrane, as evidenced by 1 integrin expression (^{Jones and Watt, 1993}). Therefore, our finding that CD34 identifies an 6 integrin bright subset supports the hypothesis that this population of bulge keratinocytes has characteristics in common with stem and progenitor cells.

It has been proposed that very early TA cells in the hair follicle are undifferentiated enough to still retain a high proliferative potential (^{Taylor et al, 2000}). One measure of that capacity is the use of in vitro assays for colony formation (^{Rheinwald and Green, 1975;Barrandon and Green, 1987;Mathor et al, 1996}). Because the basal epidermis is composed of a continuum of cells with varying degrees of proliferative capacity (^{Oshima et al, 2001;Taylor et al, 2000}), many basal keratinocytes will form colonies after 2 wk in culture. The size of the colonies after 4 wk, however, reflects the proliferative capacity of the cell of origin. It has been demonstrated that LRCs are clonogenic in vitro, whereas TA cells identified by pulse labeling do not form colonies in culture (^{Morris and Potten, 1994}). Based on our observations that CD34⁺ cells form large colonies in culture relative to those formed by CD34^–, we conclude that CD34⁺ keratinocytes have a higher proliferative potential than CD34^– keratinocytes. In addition, the increased 6 integrin expression in CD34⁺ cells may also indicate a higher proliferative potential, based on previous evidence that the highest colony forming activity was distributed in keratinocytes expressing the highest levels of 1 integrin (^{Jones and Watt, 1993}). Finally, the 6⁺CD34⁺ cells, based on the distribution of CD71 expression, most probably represent a subset of 6^briCD71^dim cells, which were shown to be slowly cycling LRCs (^{Tani et al, 2000}).

A recent paper by^{Albert et al (2001)} suggested that CD34 is not expressed in murine keratinocytes, in direct opposition to our study conclusions. The apparently conflicting data probably reflect differences in methodology. Using their fixation method, we confirmed their results (data not shown), suggesting that at least one fundamental difference lies in our use of nonfixed, intact cells.

The ability to harvest keratinocyte stem cells from the skin has highly significant implications with respect to gene therapy applications. In order to accomplish long-term changes in gene expression, it is necessary to target slowly cycling stem cells because TA cells are rapidly removed from the proliferative pool through terminal differentiation. Therefore, targeting proliferating cells results in only transient expression. To date, there have been no cell surface determinants capable of targeting the specific localization of keratinocyte stem cells. Our work with CD34 not only demonstrates specific localization to the hair follicle bulge, but also suggests that these cells behave like stem and progenitor cells in that they are quiescent, form large colonies in culture, and express relatively more 6 integrin compared to CD34^– cells. In addition, CD34 expression remains localized to the hair follicle, with no evidence of expression in the interfollicular epidermis, making it a tool for isolating and collecting this specific population of cells, offering the potential of precise targeting of the cell population necessary for the development of ex vivo gene therapy applications.

In summary, our findings demonstrate that the cell surface marker CD34 can be used for the physical enrichment of a population of follicular cells with stem or progenitor cell characteristics, such as quiescence, high proliferative capacity, and localization in a protected microenvironment. Although the role of CD34 in keratinocytes of the hair follicle bulge is unknown, further investigation of the functional properties of this cell surface marker may well provide insights into stem cell biology as it pertains to the hair follicle.

We have shown that CD34 expression is localized to the same region of the hair follicle as LRCs using immunohistochemical and autoradiographic techniques. We have recently isolated CD34⁺ and CD34^- keratinocytes by FACS from 10-week-old CD-1 mice injected with tritiated thymidine as 3- to 6-day-old pups. From this, we have determined that 65% of the LRC cells are found in the CD34⁺ population, while 7% LRC are in the CD34^- population, directly confirming that CD34⁺ keratinocytes are enriched for label retaining cells.

References

References

1.	Albeda SM, Muller WA, Buck CA & Newman PJ. Molecular and cellular properties of PECAM-1 (endoCAM/CD31): A novel vascular cell-cell adhesion molecule. J Cell Biol (1991) 114: 1059–1068. | Article | PubMed | ISI | ChemPort |
2.	Albert MR, Foster RA & Vogel JC. Murine epidermal label-retaining cells isolated by flow cytometry do not express the stem cell markers CD34, Sca-1, or Flk-1. J Invest Dermatol (2001) 117: 943–948. | Article | PubMed | ISI | ChemPort |
3.	Barrandon Y & Green H. Three clonal types of keratinocyte with different capacities for multiplication. Proc Natl Acad Sci, USA (1987) 84: 2302–2306. | PubMed | ChemPort |
4.	Battalora MS, Spalding JW & Szczesniak CJ et al. Age-dependent skin tumorigenesis and transgene expression in the Tg. AC (v-Ha-ras) transgenic mouse. Carcinogenesis (2001) 22: 651–659. | Article | PubMed | ISI | ChemPort |
5.	Brown J, Greaves MF & Molgaard HV. The gene encoding the stem cell antigen, CD34, is conserved in mouse and expressed in haematopoietic progenitor cell lines, brain, and embryonic fibroblasts. Int Immunol (1991) 3: 175–184. | PubMed | ISI | ChemPort |
6.	Cotsarelis G, Sun TT & Lavker RM. Label-retaining cells reside in the bulge area of pilosebaceous unit: Implications for follicular stem cells, hair cycle, and skin carcinogenesis. Cell (1990) 61: 1329–1337. | Article | PubMed | ISI | ChemPort |
7.	Cotsarelis G, Kaur P, Dhouailly D, Hengge U & Bickenbach J. Epithelial stem cells in the skin: Definition, markers, localization, and functions. Exp Dermatol (1999) 8: 80–88. | PubMed | ISI | ChemPort |
8.	Fuchs E & Byrne C. The epidermis: Rising to the surface. Current Opinion Genet Dev (1994) 4: 725–736. | ChemPort |
9.	Fuchs E & Green H. Changes in keratin gene expression during terminal differentiation of the keratinocyte. Cell (1980) 19: 1033–1042. | Article | PubMed | ISI | ChemPort |
10.	Fuchs E & Raghavan S. Getting under the skin of hair follicle morphogenesis. Nat Rev Genet (2002) 3: 199–209. | Article | PubMed | ISI | ChemPort |
11.	Jones PH & Watt FM. Separation of human epidermal stem cells from transit amplifying cells on the basis of differences in integrin function and expression. Cell (1993) 73: 713–724. | Article | PubMed | ISI | ChemPort |
12.	Kramer A, Keitel T, Winkler K, Stocklein W, Hohne W & Schneider-Mergener J. Molecular basis for the binding promiscuity of an anti-p24 (HIV-1) monoclonal antibody. Cell (1997) 91: 799–809. | Article | PubMed | ISI | ChemPort |
13.	Krause DS, Ito T, Fackler MJ, Smith OM, Collector MI, Sharkis SJ & May WS. Characterization of murine CD34, a marker for hematopoietic progenitor and stem cells. Blood (1994) 84: 691–701. | PubMed | ISI | ChemPort |
14.	Lane EB, Wilson CA, Hughes BR & Leigh IM. Stem cells in hair follicles. Cytoskeletal studies. Ann N Y Acad Sci (1991) 642: 197–213. | PubMed | ISI | ChemPort |
15.	Langa F, Kress C, Colucci-Guyon E, Khun H, Vandormael-Pournin S, Huerre M & Babinet C. Teratocarcinomas induced by embryonic stem (ES) cells lacking vimentin: An approach to study the role of vimentin in tumorigenesis. J Cell Sci (2000) 113: 3463–3472. | PubMed | ISI | ChemPort |
16.	Lavker RM & Sun T-T. Epidermal stem cells: Properties, markers and location. Proc Natl Acad Sci, USA (2000) 97: 13473–13475. | Article | PubMed | ChemPort |
17.	Lavker RM, Miller S, Wilson C, Cotsarelis G, Wei Z-G, Yang J-S & Sun TT. Hair follicle stem cells: Their location, role in hair cycle, and involvement in skin tumor formation. J Invest Dermatol (1993) 101: 16S–26S. | Article | PubMed | ChemPort |
18.	Lee JY, Qu-Petersen Z & Cao B et al. Clonal isolation of muscle-derived cells capable of enhancing muscle regeneration and bone healing. J Cell Biol (2000) 150: 1085–1100. | Article | PubMed | ISI | ChemPort |
19.	Li A, Simmons PJ & Kaur P. Identification and isolation of candidate human keratinocyte stem cells based on cell surface phenotype. Proc Natl Acad Sci, USA (1998) 95: 3902–3907. | Article | PubMed | ChemPort |
20.	Lyle S, Christofidou-Solomidou M, Liu Y, Elder DE, Albelda S & Cotsarelis G. The C8/144B monoclonal antibody recognizes cytokeratin 15 and defines the location of human hair follicle stem cells. J Cell Sci (1998) 111: 3179–3188. | PubMed | ISI | ChemPort |
21.	Lyle S, Christofidou-Solomidou M, Liu Y, Elder DE, Albelda S & Cotsarelis G. Human hair follicle bulge cells are biochemically distinct and possess an epithelial stem cell phenotype. J Invest Dermatol Symp Proc (1999) 4: 296–301. | ISI | ChemPort |
22.	Mathor MB, Ferrari G, Dellambra E, Cilli M, Mavilio F, De Cancedda R & Luca M. Clonal analysis of stably transduced human epidermal stem cells in culture. Proc Natl Acad Sci, USA (1996) 93: 10371–10376. | Article | PubMed | ChemPort |
23.	Mehrel T, Hohl D & Rothnagel JA et al. Identification of a major keratinocyte cell envelope protein, loricrin. Cell (1990) 61: 1103–1112. | Article | PubMed | ISI | ChemPort |
24.	Michel M, Torok N, Godbout M-J, Lussier M, Gaudreau P, Royal A & Germain L. Keratin 19 as a biochemical marker of skin stem cells in vivo and in vitro: Keratin 19 expressing cells are differentially localized in function of anatomic sites, and their number varies with donor age and culture stage. J Cell Sci (1996) 109: 1017–1028. | PubMed | ISI | ChemPort |
25.	Morris RJ & Potten CS. Slowly cycling (label-retaining) epidermal cells behave like clonogenic stem cells in vitro. Cell Prolif (1994) 27: 279–289. | PubMed | ISI | ChemPort |
26.	Morris RJ & Potten CS. Highly persistent label-retaining cells in the hair follicles of mice and their fate following induction of anagen. J Invest Dermatol (1999) 112: 470–475. | Article | PubMed | ISI | ChemPort |
27.	Morris RJ, Fischer SM, Klein Szanto AJ & Slaga TJ. Subpopulations of primary adult murine epidermal basal cells sedimented on density gradients. Cell Tissue Kinet (1990) 23: 587–602. | PubMed | ISI | ChemPort |
28.	Nickoloff BJ. The human progenitor cell antigen (CD34) is localized on endothelial cells, dermal dendritic cells, and perifollicular cells in formalin-fixed normal skin, and on proliferating endothelial cells and stromal spindle-shaped cell in Kaposi's sarcoma. Arch Dermatol (1991) 127: 523–529. | Article | PubMed | ISI | ChemPort |
29.	Oshima H, Rochat A, Kedzia C, Kobayashi K & Barrandon Y. Morphogenesis and renewal of hair follicles from adult multipotent stem cells. Cell (2001) 104: 233–245. | Article | PubMed | ISI | ChemPort |
30.	Porter RM, Lunny DP & Ogden PH et al. K15 expression implies lateral differentiation within stratifed epithelial basal cells. Lab Invest (2000) 80: 1701–1710. | PubMed | ISI | ChemPort |
31.	Potten CS. Stem Cells: Their Identification and Characterisation (1983) Edinburgh, London, Melbourne, New York: Churchill Livingstone pp 200–232.
32.	Regnier M, Vaigot P, Darmon M & Prunieras M. Onset of epidermal differentiation in rapidly proliferating basal keratinocytes. J Invest Dermatol (1986) 87: 472–476. | Article | PubMed | ISI | ChemPort |
33.	Rheinwald JG & Green H. Serial cultivation of strains of human epidermal keratinocytes: The formation of keratinizing colonies from single cells. Cell (1975) 6: 331–343. | Article | PubMed | ISI | ChemPort |
34.	Roop DR, Hawley-Nelson P, Cheng CK & Yuspa SH. Keratin gene expression in mouse epidermis and cultured epidermal cells. Proc Natl Acad Sci, USA (1983) 80: 716–720. | ChemPort |
35.	Roop DR, Cheng CK, Titterington L, Meyers CA, Stanley JR, Steinert PM & Yuspa SH. Syntheitic peptides corresponding to keratin subunits elicit highly specific antibodies. J Biol Chem (1984) 259: 8037–8040. | PubMed | ISI | ChemPort |
36.	Schwarting R & Stein H. Cluster Report: CD71. In: Knapp W, Dorken B, Gilks WR, Schmidt RE, Von dem Borne AEGK (eds)Leukocyte Typing IV. White Cell Differentiation Antigens (1989) Oxford University Press.
37.	Sonnenberg A, Calafat J & Janssen H et al. Integrin 6/4 complex is located in hemidesmosomes, suggesting a major role in epidermal cell-basement membrane adhesion. J Cell Biol (1991) 113: 907–917. | Article | PubMed | ISI | ChemPort |
38.	Spradling A, Drummond-Barbosa D & Kai T. Stem cells find their niche. Nature (2001) 414: 98–104. | Article | PubMed | ISI | ChemPort |
39.	Tani H, Morris RJ & Kaur P. Enrichment for murine keratinocyte stem cells based on cell surface phenotype. Proc Natl Acad Sci, USA (2000) 97: 10960–10965. | Article | PubMed | ChemPort |
40.	Taylor G, Lehrer MS, Jensen PJ, Sun TT & Lavker RM. Involvement of follicular stem cells in forming not only the follicle but also the epidermis. Cell (2000) 102: 451–461. | Article | PubMed | ISI | ChemPort |
41.	Vanderwinden JM, De Rumessen JJ, Laet MH, Vanderhaeghen JJ & Schiffmann SN. CD34 immunoreactivity and interstitial cells of Cajal in the human and mouse gastrointestinal tract. Cell Tissue Res (2000) 302: 145–153. | Article | PubMed | ISI | ChemPort |
42.	Watt FM. Epidermal stem cells: Markers, patterning and the control of stem cell fate. Phil Trans R Lond B (1998) 353: 831–837. | ISI | ChemPort |
43.	Weinberg W & Yuspa S. An antibody to p53 recognizes soluble keratin in epidermal keratinocyte cultures under differentiating, but not proliferating, conditions. J Invest Dermatol (1997) 109: 611–612. | Article | PubMed | ISI | ChemPort |
44.	Young PE, Baumhueter S & Lasky LA. The sialomucin CD34 is expressed on hematopoietic cells and blood vessels during murine development. Blood (1995) 85: 96–105. | PubMed | ISI | ChemPort |

Acknowledgments

The authors wish to extend their thanks to Norris Flagler and Pat Stockton of the Laboratory of Experimental Pathology at NIEHS for aid in preparing photomicrographs (N.F.) and frozen sections (P.S.) for histology, and Dr. Joseph Haseman (Biostatistics, NIEHS) for statistical analysis. We also wish to thank Drs Jackie Akunda and Anton Jetten for their review of this manuscript. This work was supported in part by NIH grant CA45293 and CA87780 (R.J.M.).