ORIGINAL ARTICLE

Angela Webb . Amy Li . Pritinder Kaur

Location and phenotype of human adult keratinocyte stem cellsof the skin

Received September 8, 2004; accepted in revised form September 17, 2004

Abstract The location and identity of interfollicularepidermal stem cells of adult human skin remain unde-fined. Based on our previous work in both adult murineand neonatal human foreskin, we demonstrate that cellsurface levels of the a6 integrin and the transferrin re-ceptor (CD71) are valid markers for resolving a putativestem cell, transit amplifying and differentiating com-partment in adult human skin by flow cytometry. Spe-cifically, epidermal cells expressing high levels of a6integrin and low levels of the transferrin receptor CD71(phenotype a6

briCD71dim) exhibit several stem cell char-acteristics, comprising a minor population (2%–5%) ofthe K14bri fraction, enriched for quiescent and smallblast-like cells with high clonogenic capacity, lackingthe differentiation marker K10. Conversely, the major-ity of K14bri K10neg epidermal cells express high levelsof CD71 (phenotype a6

briCD71bri), and represent theactively cycling fraction of keratinocytes displayinggreater cell size due to an increase in cytoplasmic area,consistent with their being transient amplifying cells.The a6

briCD71bri population exhibited intermediateclonogenic capacity. A third population of K14dim butK10 positive epidermal cells could be identified by theirlow levels of a6 integrin expression (i.e. a6

dim cells), rep-resenting the differentiation compartment; predictably,this subpopulation exhibited poor clonogenic efficiency.Flow cytometric analysis for the hair follicle bulgeregion (stem cell) marker K15 revealed preferential ex-pression of this keratin in a6

bri cells (i.e., both stem andtransient amplifying fractions), but not the a6

dim popu-lation. Given that K15 positive cells could only be de-

tected in the deep rete ridges of adult skin in situ, weconclude that stem and transient amplifying cells residein this location, while differentiating (K15 negative)cells are found in the shallow rete ridges.

Key words adult skin � stem cells � K15 � rete ridges �flow cytometry

Introduction

The epidermis of the skin forms the outer protectivebarrier of the body and is a rapidly renewingtissue comprised of a proliferative basal layer anddifferentiating suprabasal layers that undergo a series ofmorphologic and biochemical changes resulting in theproduction of squames. Cell replacement in this tissueis dependent on proliferative cells in the basal layerconsisting of a minor population of long-lived stemcells that cycle slowly, and a large pool of activelydividing, but short-lived transient amplifying cells. Thecombined activities of these two compartmentsensure homeostatic cell production throughout the life-time of the organism although we lack a clear under-standing of the role of stem versus progenitor cells inperturbed conditions (such as wound healing and can-cer), largely due to the paucity of markers that distin-guish these populations within the basal layer.Consequently, the precise spatial location of these cellsis unclear and is the subject of much controversy anddebate.

In murine skin, detailed cell kinetic analyses led tothe identification of stem cells as slowly cycling cellsthat could be identified in situ as DNA label-retainingcells (Potten and Bullock, 1983; Mackenzie and Bic-kenbach, 1985; Morris et al., 1985). Thus, epidermalstem cells could be visualized in interfollicular epidermisas single cells within the basal layer. However, thehair follicle organized its stem cells as clusters of1These two authors made equal contributions.

Angela Webb1 � Amy Li1 � Pritinder Kaur ( .*)Epithelial Stem Cell Biology LaboratoryPeter MacCallum Cancer CentreSt. Andrew’s PlaceMelbourne, Victoria 3002, AustraliaFax : 61 3 9656 3738E-mail: pritinder.kaur@petermac.org

U.S. Copyright Clearance Center Code Statement: 0301–4681/2004/7208–387 $ 15.00/0

Differentiation (2004) 72:387–395 r International Society of Differentiation 2004

slow-cycling cells in a macroscopic niche termed thebulge region (Cotsarelis et al., 1990). More recently,gene marking studies in situ (Ghazizadeh and Taich-man, 2001), in transgenics (Morris et al., 2004; Tumbaret al., 2004) or in whole epidermal mounts of BrDU-labeled mice (Braun et al., 2003), elegantly confirm thepresence of clusters of stem cells in the hair follicle bulgeregion and single stem cells in the interfollicularepidermis.

In monkey palm epidermis, Lavker and Sun (1982)have shown that DNA label-retaining cells are found inthe deep rete ridges postulating that this site may pro-vide protection for the long-lived stem cell populationfrom harmful environmental agents. However, the lo-cation of stem cells in human skin is equivocal given ourinability to mark slow-cycling cells in situ, and thelack of assays that permit us to ascertain whetherspecific markers uniquely identify epidermal stemcells. In neonatal human foreskin and adult breast skintissue, the location of the stem cells has been inferredbased on the differential expression of cell surfacemarkers thought to identify epidermal stem cells (Jensenet al., 1999; Legg et al., 2003). Specifically, it has beenproposed that clusters of epidermal cells expressinghigh levels of b1 integrin (b1

bri) also identifiable as rap-idly adhering keratinocytes, found in the shallow reteridges, represent stem cells due to the absence of K10staining in this sub-population and lack of BrDU in-corporation indicating their slow-cycling status (Jensenet al., 1999).

b1bri human foreskin keratinocytes isolated by flow

cytometry also demonstrate greater short-term colony-forming efficiency when plated in culture (Jones andWatt, 1993; Jones et al., 1995), although this may not bea rigorous assay for stem cell activity given the vastproliferative capacity of most basal keratinocytes inneonatal foreskin (Li et al., 2004). Interestingly, a recentstudy utilizing adult human palm, sole, and breast skintissue demonstrated that rapidly adhering keratinocytes(b1

bri cells) expressing lower levels of desmosomal pro-teins are enriched for cells with greater clonogeniccapacity and proliferative output (Wan et al., 2003).Thus, Dsg-3dim cells represent the proliferative com-partment of adult epidermis at various anatomical sites.Surprisingly, these b1

bri cells with low to negative levelsof desmosomal proteins (desmoplakin and desmogleins)were localized to the basal cells at the tips of the reteridges in contrast to earlier data (Jensen et al., 1999;Legg et al., 2003). Here, we present data utilizing cellsurface markers previously shown by us to enrich forepidermal stem cells of neonatal human foreskin andmurine skin (Li et al., 1998; Tani et al., 2000), andan intracellular marker K15 to demonstrate thatthe proliferative compartment of adult skin epidermiscomprising both stem and progenitor cells residesin the tips of rete ridges as proposed by Wan et al.(2003).

Methods

Isolation of human adult keratinocytes

Human adult skin specimens from elective operations were collect-ed with patient consent and Institutional Ethics approval andprocessed within 4 hr of harvesting. Epithelial sheets were obtainedafter overnight incubation in Dispase II at 4mg/ml at 41C. Basalkeratinocytes were isolated by trypsinization for 5min and resus-pended in keratinocyte growth medium (KGM) (Clonetics, SanDiego, CA) in preparation for antibody staining.

Antibodies

Anti-cytokeratin hybridoma supernatants, K10 (LHP2, IgG1; 1:10),K14 (LL001, IgG2b; 1:1000), and K15 (LHK15, IgG2a; 1:10) werekind gifts from Dr. I. Leigh, Royal London Hospital. Ki67(DAKO) was used at 1:100. Mab 4F10 to the a6-integrin (IgG2b;Serotec Ltd., Oxford, UK) was used at 10 mg/ml. Mab 10G7 (IgG2a)against the human transferrin receptor CD71 was used as neat hy-bridoma supernatant. Isotype-matched negative control Mab1A6.11 (IgG2b), 1D4.5 (IgG2a) were used as neat hybridoma supe-rnatants. IgG1-negative control Mab WM59 (Becton DickinsonPharmingen, San Jose, CA) was used at 1:100.Biotinylated or fluorochrome (FITC and PE)-conjugated iso-

type-specific goat anti-mouse antibodies (Caltag Laboratories Inc.,Burlingame, CA) were used to detect binding of the primary an-tibodies. In some experiments, biotinylated mouse anti-humanCD71 Mab (Becton Dickinson) was used and detected withStreptavidin APC (Becton Dickinson).

Flow cytometry

Freshly isolated epidermal cells were labeled simultaneously for thea6-integrin and either keratins (K14, K10, K15), CD71 or Ki67,and were detected by using fluorochrome-conjugated secondaryantibodies. For the cell surface markers a6-integrin and CD71, vi-able cells were labeled as described previously (Li et al., 2004). Forthe detection of intracellular markers in combination with a6-in-tegrin cells were permeabilized by treatment with cold (� 201C)ethanol for 10min prior to staining. Cells were washed twice withHank’s balanced salt solution with 5% FCS between each antibodylayer. Labeled cells were resuspended in KGM for fluorescence ac-tivated cell sorting. For analysis, labeled keratinocytes were resus-pended in 200–300 ml of FACS fix (PBS containing 0.1%v/vformalin and 2%w/v glucose) and stored at 41C until analysis.

Cell cycle analysis

Fractions of primary basal keratinocytes were collected after labe-ling with a6-integrin and CD71 by FACS as described previously(Li et al., 1998, 2004). Cells were washed twice in PBS before fix-ation with cold 70% ethanol, and stored at � 201C until analysis.After thawing, cells were washed twice with PBS and incubated inPBS containing 40mg/ml propidium iodide and RNAse (40 mg), for30–60min at 371C. The DNA content of each fraction was analyzedby flow cytometry using a FACScan (Becton Dickinson), to deter-mine the percentage of cells in various phases of the cell cycle,particularly S1G2M.

Morphological and clonogenic capacity of fractionated adult skinkeratinocytes

Forward and side scatter parameters of specific sub-populationswere determined by analyzing data accumulated in Listmodefiles from FACS experiments used to collect the a6

briCD71dim

388

a6briCD71bri and a6

dim fractions for various experiments. For mor-phological analysis, cytospins of fractionated keratinocytes wereprepared and stained with Giemsa, for microscopic analysis. Im-ages were collected using a digital camera at �20 magnification.To determine the relative proliferative capacity of fractionated

keratinocytes, the a6briCD71dim a6

briCD71bri, and a6dim fractions

isolated from primary adult breast skin tissue were placed inco-culture with irradiated J2 Swiss 3T3 feeder layers as describedpreviously (Li et al., 1998). 40,000 cells from each fraction wereplated in triplicate in 12-well plates and cultured in a 2:1 ratio ofDMEM: Ham’s F12 (Gibco BRL, Melbourne, Australia, Cat.21700-075), containing 10% FCS, 5 mg/ml insulin (Sigma ChemicalCompany), 0.18mM adenine (Sigma), 0.4 mg/ml hydrocortisone(Sigma), 0.1 nM cholera toxin (Sigma), 2 nM tri-idothyronine (Si-gma), and 4 mM glutamine (Pellegrini et al., 1999). The fractionswere maintained in culture for 10 days at 371C in a 5% CO2 in-cubator. The cells were then fixed in 4% formaldehyde in PBS andstained with 1% Rhodamine B. The density of colonies was deter-mined by scoring the number of colonies present in three randomlyselected fields per well, in three replicate wells and expressed asmean � SEM.

Immunofluorescence staining

Skin tissue was embedded in optimal cutting temperature withoutprior fixation and stored at � 701C until cryosectioning. Frozensections (4 mm) were labeled with K15 antibodies followed byFITC-conjugated secondary antibody at room temperature in ahumidified box for 1–11

2hr each; sections were washed three times

with PBS–0.05% Tween between each antibody layer.

Results

The proliferative compartment of adult humanepidermis is characterized by high levels of a6integrin expression

We have previously demonstrated that the proliferativecompartment of the epidermis in human neonatal fore-skin can be distinguished from keratinocytes that haveembarked on their differentiation program by theirpreferentially high expression of the integrin a6 (Liet al., 1998; Kaur & Li, 2000). Specifically, primaryhuman keratinocytes with the phenotype a6

bri expressedhigh levels of the basal marker K14 and did not expressthe differentiation markers K10 or involucrin; this pop-ulation was also enriched for keratinocytes with thegreatest short-term colony-forming efficiency and long-term proliferative output. In contrast, primary keratin-ocytes expressing low levels of a6 integrin (phenotypea6dim) were shown to down-regulate K14 expression, and

were positive for the K10 and involucrin proteins, in-dicating their commitment to differentiation consistentwith their decreased growth capacity in vitro (Li et al.,1998). We therefore set out to establish whether primarykeratinocytes derived from adult human skin from var-ious sources could similarly be separated into the pro-liferative versus differentiating compartments based ondifferential cell surface expression of a6 integrin. Fresh-ly isolated primary human adult keratinocytes were

analyzed by flow cytometry for their simultaneous ex-pression of a6 versus K14 or K10. The results obtainedshown in Figs. 1A, 1B illustrate that the majority ofcells isolated from adult epidermal sheets express K14,confirming their identity as keratinocytes. These K14positive cells could be further sub-divided into twopopulations based on differential expression of the a6integrin, i.e., an a6

bri versus an a6dim fraction. While the

a6bri cells uniformly expressed K14, the a6

dim cells exhib-ited more heterogeneous K14 levels (Fig. 1B). Dualstaining for a6 and K10 shown in Figures 1C, 1D re-vealed that the a6

bri population was largely negative forK10, indicating their undifferentiated state, while thea6dim fraction were all positive for K10, and could be

fractionated further into three subsets displaying in-creasing levels of this differentiation marker, most ev-ident when displayed as a histogram (Fig. 1D). Thus,the a6

dim fraction most likely comprises a mixture ofbasal and suprabasal populations of keratinocytes thatdecrease K14 expression while up-regulating K10, andthe a6

bri cells represent the proliferative basal cells. Thisinterpretation is also consistent with the Ki67 expres-sion pattern observed in these populations shown inFigures 1E, 1F showing enrichment for Ki67 positivecells in the a6

bri population, while revealing heterogene-ity for this proliferative marker in the a6

dim fraction.Presumably, Ki67 expression is lost gradually withcommitment to differentiation, with a concomitant in-crease in K10 expression.

The proliferative compartment of adult epidermis canbe further resolved into a putative stem (KSC) andtransient amplifying (TA) compartment based on CD71expression

We next sought to determine whether the proliferativecompartment of adult epidermis, i.e., the a6

bri/K14bri/K10neg population, could be further resolved into pu-tative stem cells and transient amplifying cells using thecell surface marker CD71, given our previous demon-stration that actively cycling epidermal cells expresshigh levels of this protein, while the CD71dim fractioncomprises blast-like cells enriched for either label-re-taining cells in murine adult epidermis (Tani et al.,2000) or human neonatal foreskin cells with the greatestlong-term proliferative output in vitro (Li et al., 1998).Dual labeling of primary human adult keratinocytesfrom various anatomical sites examined, stained fora6 and CD71, exhibited a similar phenotypic patternas reported by us previously for murine and neonatalhuman skin (Figs. 2A, 2B). Thus the a6

bri fractioncould be fractionated into a minor population ofCD71dim cells (termed a6

briCD71dim) with an incidenceof about 5.36% � 2.34% on average compared with7.71% � 1.6% in neonatal skin (Figs. 2A–2C). Themajority of a6

bri keratinocytes in adult skin expressed

389

high levels of CD71 as shown in Fig. 2A, 2B (termeda6briCD71bri), just as observed previously for neonatal

human (Li et al., 1998) and adult murine epidermis(Tani et al., 2000). Cell cycle analysis of the threephenotypically distinct fractions of epidermal cells fromadult breast skin epidermis was performed sub-sequently. The data obtained demonstrated that thea6briCD71dim cells were enriched for quiescent keratin-

ocytes (Fig. 3A) whereas a6briCD71bri were enriched for

cycling cells as determined by the total number of cellsin each fraction that were in the S and G2M phase, from

3 separate sorting experiments. Notably, the a6dim frac-

tion also contained a significant number of cycling cellsconsistent with the presence of Ki67 positive cells foundin this population earlier (Figs. 1E, 1F). We also ex-amined the size and cytoplasmic complexity of thesepopulations of keratinocytes and found that the quies-cent a6

briCD71dim cells were immature blast-like cells(Fig. 3B) with low forward scatter (FSC) indicatingtheir small size, and low side scatter (SSC) character-istics, i.e., lacking in cytoplasmic complexity (Fig. 3D).In contrast, a6

briCD71bri cells appeared to be larger in

Fig. 1 Phenotype of adult epidermal keratin-ocytes fractionated on the basis of a6 integrinexpression. Basal keratinocytes isolated fromthe facial skin of a 72-year-old male werestained with anti-a6 integrin versus K14 (A,B); K10 (C, D); or Ki67 (E, F) for flow cyto-metric analysis. Data are displayed as dotplots of dual-stained samples (A, C, E) orhistograms of a6

bri (purple line) or a6dim (or-

ange line) cells for each marker (B, D, F),compared with the profile obtained with iso-type-matched control antibodies (mAb1D4.5 and 1B5) indicated by dotted lines.The result is representative of four individualadult skin samples examined. Panel A: circlemarks K14 negative cells; panel E: rectanglemarks a6

dim Ki67neg cells.

390

size with a greater cytoplasmic area (Fig. 3C) that ex-hibited more complexity than the a6

briCD71dim fraction(Fig. 3E).

These distinct morphological features of thea6briCD71dim and a6

briCD71bri keratinocytes, combinedwith their incidence in the epidermis and cycling status,are consistent with these two populations representingthe putative stem cell and transient amplifying popula-tion, respectively. Despite the presence of cycling activ-ity in the a6

dim fraction, its expression of K10 clearlyindicates the onset of differentiation. We then set outto determine the comparative clonogenic ability of

these fractions to establish whether these phenotypicallydistinct compartments could also be distinguished interms of their ability to regenerate keratinocytes. Thedata obtained revealed that the putative stem cell pop-ulation i.e. a6

briCD71dim was enriched for clonogeniccells (Fig. 4). The a6

briCD71bri fraction (putative TA)had intermediate clonogenic capacity, whereas the a6

dim

fraction exhibited the poorest clonogenic capacity con-sistent with its more differentiated state (Fig. 4). Weconclude that the total a6

bri fraction contains themajority of clonogenic cells of adult skin epidermisand that CD71 permits further fractionation of this

Fig. 2 Fractionation ofa6bri keratinocytes on the

basis of CD71 expressionin adult epidermis. Basalkeratinocytes were isolat-ed from the skin of var-ious body sites obtainedfrom healthy adult hu-mans and stained simul-taneously with anti-a6(mAb GoH3) and CD71(mAb 10G7) for flowcytometric analysis. (A)Groin skin from a 34-year-old female. (B) Fa-cial skin from a 72-year-old male. Basal keratin-ocytes from these tissuesexhibited three distinctphenotypic populationsof cells as observed inneonatal foreskin. Thegating demarcates thea6briCD71dim putative

KSC (R2), a6briCD71bri

putative TA (R3), anda6dim differentiating (R4)

cells. Percentage of theputative KSC compart-ment (a6

briCD71dim) wasdetermined and is indi-cated on each panel. C.Size of a6

briCD71dim pu-tative stem cell compart-ment in neonatal andadult skin specimens.

391

proliferative compartment into putative stem and TAfractions.

K15 is a marker of stem and transient amplifyingepidermal cells revealing their location in thedeep rete ridges

It has been shown that the keratin 15 promoter is activein the bulge region of the murine hair follicle and that itmay therefore be a marker of the hair follicle stem cells(Morris et al., 2004). Given that the location of stemcells of the interfollicular epidermis remains controver-sial, we set out to establish whether K15 could be usedto identify these cells in situ. Interestingly, immuno-staining for K15 on sections of neonatal foreskin

showed that all basal cells expressed this keratin ho-mogeneously (Fig. 5B). Although variations in stainingintensity were observed within the basal layer, no con-sistent pattern was detectable across a number of sam-ples. In contrast, adult skin tissue exhibited both K15positive and negative cells within the basal layer; nota-bly, the cells at the tips of deep rete ridges of adult skinconsistently expressed high levels of K15 (Fig. 5A),while those at the tops of the shallow rete ridges werenegative. In order to determine whether the K15 pos-itive cells expressed high or low levels of a6 integrin, weperformed dual staining for this cell surface markerwith K15 in freshly isolated populations of primaryadult keratinocytes. Figures 5C, 5D show that the a6

bri

population expressed high levels of K15, and that its

Fig. 3 Cycling activity and cellular morphol-ogy of primary adult keratinocytes fraction-ated on the basis of a6 and CD71 expression.(A) Cell cycle analysis of total keratinocytes(R1), a6

briCD71dim (putative KSC), a6bri

CD71bri (putative TA), and a6dim (PMD)

fractions reveals enrichment for non-cyclingcells in the a6

briCD71dim fraction. (B, C)

Giemsa staining of cytospins of thea6briCD71dim (B) and a6

briCD71bri (C) frac-tions revealing blast-like morphology of thequiescent putative KSC fraction (B) com-pared with the larger cells found in the pu-tative TA (a6

briCD71bri) fraction (C). (D, E)FSC versus SSC properties of a6

briCD71dim

(D) and a6briCD71bri (E) fractions revealing

smaller cell size of the putative stem cellpopulation (D) compared with the larger andmore complex putative TA population (E).

392

expression was down-regulated as cells decreased theirsurface levels of this integrin. Notably, most a6

dim cells inadult epidermis were K15 negative, whereas in neonatalepidermis, a significant number of these continued toexpress K15 (Fig. 5D) consistent with its expressionthroughout the foreskin basal layer (Fig. 5B). Thus inadult epidermis K15 is a good intracellular marker forkeratinocytes expressing high levels of a6 integrin i.e.a6bri cells. Given that the a6

bri fraction comprises theproliferative compartment of the epidermis harboringboth the stem and TA populations, we conclude thatthese cells are located in the deep rete ridges in adultepidermis, given this preferred location of K15 positivecells. It remains difficult, however, to ascertain wherethese cells are located in neonatal epidermis.

Discussion

While there has been considerable progress made indefining the precise location and phenotype of murinekeratinocyte stem cells of the skin, the same cannot besaid for human keratinocyte stem cells. It is difficult toperform the kind of kinetic studies in humans that haveproven invaluable in defining murine epidermal stemcells due to ethical considerations, making it necessaryto resort to surrogate assays to assign stem cell status onpopulations of cells isolated on the basis of markers. Wehave previously shown that it is possible to discern ki-netically distinct classes of both neonatal human andadult murine epidermal cells based on their compositecell surface phenotype in terms of relative expression of

the cell surface proteins a6 integrin and CD71 (Li et al.,1998; Tani et al., 2000). In this study, we have been ableto demonstrate that these two markers are also useful infractionating adult epidermis into three distinct com-partments: (a) a6

briCD71dim keratinocytes that representa minor population of the basal layer enriched for slow-cycling cells with the morphological and clonogenicproperties expected of a putative stem cell population;(b) a6

briCD71bri keratinocytes comprising the majorityof K14 positive, undifferentiated cells with greater cellsize and complexity, but lower clonogenic capacity; and(c) a6

dim keratinocytes representing a mixture of basaland suprabasal cells expressing various levels of K10and exhibiting diminished clonogenic capacity in vitro.

These data suggest that these cell surface markers,particularly CD71, may be useful in distinguishingrelatively slow-cycling cells within the proliferativecompartment of various epithelia. Interestingly, neona-tal human KSCs and TA populations separated withthe same markers did not differ as markedly in theircolony-forming ability and we speculate that neonatalTA cells may have a greater capacity for growth thanadult TA cells. We also observed a modest decrease inthe incidence of the a6

briCD71dim population of adultskin of various body sites compared with that of neo-natal skin, although there was no clear correlation withincreasing age. This remains for the time being an in-teresting observation although more mundane explana-tions may be due to anatomical variations at differentbody sites. Clearly, it would be valuable to accumulatedata from a greater number of samples than the limitednumber analyzed here (n5 11).

Although K15 has been reported as a stem cellmarker of the murine hair follicle bulge region, our datasuggest that it is widely expressed throughout the basallayer in neonatal human skin. Given that K15 promot-er-driven GFP or b-galactosidase expression is restrict-ed to the hair follicle bulge region (Morris et al., 2004),it is likely that K15 mRNA is transcribed exclusively instem cells, but the K15 protein persists in the progeny ofstem cells as demonstrated by our data. Variations inpatterns of immunostaining reported by independentlaboratories may also be achieved by using differentantibodies to distinct epitopes, which may not exhibitsimilar affinity for binding to their epitope, or throughthe use of low antibody concentrations. Nevertheless,the K15 promoter is valuable for targeting expression ofgenes to hair follicle bulge cells in transgenic animals,whereas the K15 protein is a marker for the prolifer-ative compartment of adult human epidermis. Giventhe segregation of a6

bri/K14bri/K10neg cells into K15positive cells, and a6

dim/K14 dim/K10 pos cells into K15negative cells, we were able to use K15 as a surrogatemarker for a6

bri cells to determine the location of theseproliferative keratinocytes in situ.

It is worth noting that although the a6dim popu-

lation can be easily identified by flow cytometry, the

Fig. 4 Enrichment of clonogenic keratinocytes of adult epidermisin the a6

briCD71dim fraction. Relative colony-forming ability offractionated primary adult breast keratinocytes in Swiss 3T3 co-culture. Colony-forming density per well of unfractionated kera-tinocytes (R1), a6

briCD71dim, a6briCD71bri, and a6

dim fractions reveal-ing greatest colony formation capacity in the putative KSC fraction(1).

393

differences in the levels of this cell surface proteinamong basal keratinocytes are not detectable by the eyeusing immunohistochemistry. However, the correlationof K15 expression in the a6

bri population enabled us todeduce that these cells resided in the tips of the deep reteridges of adult skin. These data correlate with the workof Wan et al. (2003), who have demonstrated that basalkeratinocytes expressing low levels of the desmosomalprotein desmoglein-3 (Dsg-3dim), which have a greatergrowth capacity in both short-term and long-term pro-liferation assays, reside in the tips of the deep rete ridgesof human adult breast and palm skin. Given that Dsg-3bri keratinocytes gave rise largely to abortive colonies,we suggest that this population most likely representsdifferentiating keratinocytes, whereas the Dsg-3dim frac-tion is equivalent to the a6

bri proliferative compartmentof adult skin comprising both stem and progenitor ker-atinocytes, consistent with the observation that greaternumbers of desmosomes are found in differentiatingkeratinocytes. Thus Dsg-3dim cells and K15pos keratin-ocytes are found exclusively at the tips of the deep reteridges, placing them in a more protected location awayfrom the environment as proposed originally by Lavkerand Sun (1982).

These data bring into question the observations re-ported by Watt and colleagues proposing the locationof stem cells in the shallow rete ridges based on thepresence of clusters of b1

bri cells at this site. One pos-sibility is that the epitope recognized by the antibodyutilized in these in situ localization studies recognizes anepitope of b1 integrin that is more readily accessible

than other regions of the integrin or is highly expressedin differentiated keratinocytes. Notably, the site ofbrightest staining for b1 described by these workers waspredominantly in the lateral intercellular junctions be-tween keratinocytes, and this localization is also ob-served in differentiating keratinocytes in vitro (Carteret al., 1990). The interpretation that the b1

bri brightclusters are differentiated keratinocytes is also support-ed by the report that these cells do not stain for BrDU,although the absence of K10 is difficult to explain. It isworth noting that the work of Watt and colleagues wasperformed with whole sheets of epidermis and it is notdifficult to envisage that accessibility of antibodies tovarious antigens particularly cell surface receptors thatmay be bound to their ligands, and efficient removal ofexcess antibody could be problematic in these wholemounts. Although our data indicate the location ofstem and transient amplifying cells in the tips of deeprete ridges, whether the stem cells exist as single cells orin clusters remains undetermined. Clearly, the identifi-cation of further positive markers that definitively iden-tify keratinocyte stem cells will be essential to furtheradvancing our knowledge. The recent genome profilingof murine bulge hair follicle cells (Morris et al., 2004;Tumbar et al., 2004) may provide some common mark-ers for human KSCs, although protein profiling may beeven more valuable.

Acknowledgments We thank Ralph Rossi and Andrew Fryga forexpert cell sorting expertise and advice. Angela Webb is the recip-ient of the Melville Hughes scholarship. This work is supported by

Fig. 5 Expression of K15 in neonatal andadult human skin. (A, B) In situ staining forK15 in the epidermis of a 72-year–old’s facialskin (A) and neonatal foreskin (B). Note thepresence of K15 bright cells at the tips of thedeep rete ridges of adult epidermis (Panel A).The arrow indicates the decrease in K15 ex-pression along the basal layer in the adultskin. Neonatal foreskin epidermis shows K15expression throughout the basal layer. (C, D)Two color flow cytometric analysis of adultkeratinocytes (C) and neonatal foreskin ker-atinocytes (D) for the simultaneous expres-sion of a6 integrin and K15, demonstratingthat a6

bri cells expressed higher levels of K15compared with the differentiating a6

dim cells.Results shown are typical of several individ-ual skin samples (n45).

394

grants from the National Health & Medical Research Council ofAustralia to P.K.

References

Braun, K.M., Niemann, C., Jensen, U.B., Sundberg, J.P., Silva-Vargas, V. and Watt, F.M. (2003) Manipulation of stem cellproliferation and lineage commitment: visualisation of label-re-taining cells in wholemounts of mouse epidermis. Development130:5241–5255.

Carter, W.G., Wayner, E.A., Bouchard, T.S. and Kaur, P. (1990)The role of integrins alpha 2 beta 1 and alpha 3 beta 1 in cell-celland cell-substrate adhesion of human epidermal cells. J Cell Biol.110:1387–1404.

Cotsarelis, G., Sun, T.-T. and Lavker, R.M. (1990) Label-retainingcells reside in the bulge region of the pilosebaceous unit: impli-cations for follicular stem cells, hair cycle, and skin carcinogen-esis. Cell 61:1329–1337.

Ghazizadeh, S. and Taichman, L.B. (2001) Multiple classes of stemcells in cutaneous epithelium: a lineage analysis of adult mouseskin. EMBO J 20:1215–1222.

Jensen, U.B., Lowell, S. and Watt, F.M. (1999) The spatial rela-tionship between stem cells and their progeny in the basal layer ofhuman epidermis: a new view based on whole-mount labellingand lineage analysis. Development 126:2409–2418.

Jones, P.H., Harper, S. and Watt, F.M. (1995) Stem cell patterningand fate in human epidermis. Cell 80:83–93.

Jones, P.H. and Watt, F.M. (1993) Separation of human epidermalstem cells from transit amplifying cells on the basis of differencesin integrin function and expression. Cell 73:713–724.

Kaur, P. and Li, A. (2000) Adhesive properties of human epidermal cells: ananalysis of keratinocyte stem cells, transit amplifying cells, andpost-mitotic differentiating cells. J Invest Dermatol 114:413–420.

Lavker, R.M. and Sun, T.-T. (1982) Heterogeneity in epidermalbasal keratinocytes: morphological and functional correlations.Science 215:1239–1241.

Legg, J., Jensen, U.B., Broad, S., Leigh, I. and Watt, F.M. (2003)Role of melanoma chondroitin sulphate proteoglycan in pat-

terning stem cells in human interfollicular epidermis. Develop-ment 130:6049–6063.

Li, A., Pouliot, N., Redvers, R. and Kaur, P. (2004) Extensivetissue-regenerative capacity of neonatal human keratinocyte stemcells and their progeny. J Clin Invest 113:390–400.

Li, A., Simmons, P.J. and Kaur, P. (1998) Identification andisolation of candidate human keratinocyte stem cells basedon cell surface phenotype. Proc Natl Acad Sci USA 95:3902–3907.

Mackenzie, I.C. and Bickenbach, J.R. (1985) Label-retaining ker-atinocytes and Langerhans cells in mouse epithelia. Cell TissueRes 242:551–556.

Morris, R.J., Fischer, S.M. and Slaga, T.J. (1985) Evidence that thecentrally and peripherally located cells in the murine epidermalproliferative unit are two distinct cell populations. J Invest De-rmatol 84:277–281.

Morris, R.J., Liu, Y., Marles, L., Yang, Z., Trempus, C., Li, S.,Lin, J.S., Sawicki, J.A. and Cotsarelis, G. (2004) Capturingand profiling adult hair follicle stem cells. Nat Biotechnol22:411–417.

Pellegrini, G., Golisano, O., Paterna, P., Lambiase, A., Bonini, S.,Rama, P., De Luca, M. (1999) Location and clonal analysis ofstem cells and their differentiated progeny in the human ocularsurface. J Cell Biol 145:769–782.

Potten, C.S. and Bullock, J.C. (1983) Cell kinetic studies in theepidermis of the mouse. I. Changes in labelling index withtime after tritiated thymidine administration. Experientia39:1125–1129.

Tani, H., Morris, R.J. and Kaur, P. (2000) Enrichment for murinekeratinocyte stem cells based on cell surface phenotype. ProcNatl Acad Sci USA 97:10960.

Tumbar, T., Guasch, G., Greco, V., Blanpain, C., Lowry, W.E.,Rendl, M. and Fuchs, E. (2004) Defining the epithelial stem cellniche in skin. Science 303:359–363.

Wan, H., Stone, M.G., Simpson, C., Reynolds, L.E., Marshall,J.F., Hart, I.R., Hodivala-Dilke, K.M. and Eady, R.A.J. (2003)Desmosomal proteins, including desmoglein 3, serve as novelnegative markers for epidermal stem cell-containing populationof keratinocytes. J Cell Sci 116:4239–4248.

395