isolation of a mesenchymal cell population from murine ...the faseb journal • research...

The FASEB Journal • Research Communication

Isolation of a mesenchymal cell population frommurine dermis that contains progenitors of multiplecell lineages

Lauren Crigler,* Amita Kazhanie,* Tae-Jin Yoon,† Julia Zakhari,* Joanna Anders,*Barbara Taylor,* and Victoria M. Virador*,1

*Laboratory of Cellular Carcinogenesis and Tumor Promotion, Center for Cancer Research, NationalCancer Institute, NIH, Bethesda, Maryland, USA; and †Department of Dermatology, College ofMedicine, Gyeongsang National University, Jinju, Korea

ABSTRACT The skin contains two known subpopula-tions of stem cells/epidermal progenitors: a basal kera-tinocyte population found in the interfollicular epithe-lium and cells residing in the bulge region of the hairfollicle. The major role of the interfollicular basal kera-tinocyte population may be epidermal renewal, whereasthe bulge population may only be activated and recruitedto form a cutaneous epithelium in case of trauma. Using3-dimensional cultures of murine skin under stress condi-tions in which only reserve epithelial cells would beexpected to survive and expand, we demonstrate that amesenchymal population resident in neonatal murinedermis has the unique potential to develop an epidermisin vitro. In monolayer culture, this dermal subpopulationhas long-term survival capabilities in restricted serum andan inducible capacity to evolve into multiple cell lineages,both epithelial and mesenchymal, depending on cultureconditions. When grafted subcutaneously, this dermalsubpopulation gave rise to fusiform structures, reminis-cent of disorganized muscle, that stained positive forsmooth muscle actin and desmin; on typical epidermalgrafts, abundant melanocytes appeared throughout thedermis that were not associated with hair follicles. Themultipotential cells can be repeatedly isolated from neo-natal murine dermis by a sequence of differential centrif-ugation and selective culture conditions. These resultssuggest that progenitors capable of epidermal differenti-ation exist in the mesenchymal compartment of an abun-dant tissue source and may have a function in mesenchy-mal-epithelial transition upon insult. Moreover, these cellscould be available in sufficient quantities for lineagedetermination or tissue engineering applications.—Crig-ler, L., Kazhanie, A., Yoon, T-J., Zakhari, J., Anders, J.,Taylor, B., Virador, V. M. Isolation of a mesenchymal cellpopulation from murine dermis that contains progenitorsof multiple cell lineages. FASEB J. 21, 2050–2063 (2007)

Key Words: epidermal precursors � murine organotypic culture� keratinocytes

Skin bioengineering has advanced the productionof in vitro human skin reconstructs for trauma patients

with damaged areas too large to repair by the naturalwound-healing process (1). However, the growth ofmouse skin reconstructs in vitro has not been as success-ful. Three-dimensional murine skin cultures would beuseful to specifically study epidermal-mesenchymal cellsignaling. Through the power of mouse genetics andusing primary cells derived from genetically alteredmouse skins, the underlying biology of a number ofdiseases would be studied. In vitro skin regenerationwould be a useful model for the study of geneticdeterminants of wound healing. Human and mouseskin differ in epidermal thickness and appendage con-tent, and thus it is possible that the niche responsiblefor epidermal renewal during wound healing may bedifferent in the mouse and human (2–4).

Many adult tissues contain populations that haverenewal ability under circumstances such as trauma,disease, or aging (1, 5). There is also ample evidencethat a population of skin-specific precursors capable ofrestoring skin integrity upon insult resides in a nicheidentified as the bulge region of the hair follicle (6–8).Another such population, most likely responsible fornormal epidermal homeostasis, is known to reside inthe interfollicular basal layer of the epidermis (9, 10).

Epidermal-mesenchymal interactions are necessaryto provide the context for tissue development (11),differentiation (12), and tissue renewal on injury (13,14). In skin, the mesenchyme immediately adjacent tothe hair follicle epithelium has an inductive effect onfollicle development. Specialized mesenchymal cells,called dermal papilla cells, are embedded in the lowerportion of the hair follicle and are surrounded byundifferentiated epithelial cells. These fibroblast-likedermal papilla cells produce and respond to the signalsthat regulate hair cycles (15, 16). Hair could be re-grown when hair follicles were grafted onto nude micewith dermal papilla cells and not in their absence (17).Furthermore, characteristics of the evolving hair weredetermined by the species of origin of the dermal

1 Correspondence: 9000 Rockville Pike, Bldg. 10, Rm. 2A33,Bethesda, MD 20892, USA. E-mail: [email protected]

doi: 10.1096/fj.06-5880com

2050 0892-6638/07/0021-2050 © FASEB

papilla cell preparations. These findings could be ex-panded and explained with an in vitro model formurine epidermis.

In establishing such a culture system, we were unsuc-cessful in adapting methods reported for human epi-dermal cells in human skin reconstructs. However,selected epidermal subpopulations had the ability tosurvive while cultured in cell-to-cell contact with adermal component at the air-liquid interface underreported culture conditions (18). During the course ofthese studies, we identified a dermal subpopulationthat could by itself expand in culture and expressepidermal markers while cultured in the absence ofcontact with a dermal component at the air-liquidinterface under epidermal growth conditions. Thissubpopulation had the ability to survive in minimalmedium conditions over time, a property that might beexpected of a progenitor population. For the purposesof these studies, a progenitor cell has the capacity tocreate progeny that are more differentiated than itselfand the ability to survive and propagate under minimalculture conditions. Further analysis of the properties ofour isolated subpopulation indicated characteristicsconsistent with a multipotent mesenchymal progenitorpopulation.

MATERIALS AND METHODS

Tissue culture media and other chemicals

Minimum essential medium (S-MEM) without calcium (LifeTechnologies, Inc., Rockville, MD, USA) was used to preparethe two main types of culture media. Standard medium wasS-MEM containing 8% fetal bovine serum (Gemini Bio-Products, West Sacramento, CA, USA) treated with Chelexresin (Bio-Rad Laboratories, Hercules, CA, USA). High cal-cium medium contained 1.4 mM Ca2�. This medium wastypically used to plate and attach keratinocytes or hair folliclebuds and to culture some of the dermal-derived populations.Low calcium medium contained 0.05 mM Ca2�, typically usedwhen cultures were grown under submerged conditions toallow for basal keratinocyte proliferation. In some experi-ments, cells were cultured in medium containing 2% serum.All other chemicals were from Sigma Chemical Company (St.Louis, MO, USA) unless stated otherwise. KGF (R&D Systems,Minneapolis, MN, USA) was used at 8.4 ng/ml in the 3-di-mensional cultures.

Skin fractionation

Cellular subfractionation from both epidermis and dermis ofneonatal BALB/c mice is summarized in Fig. 1A and ex-plained in detail in supplemental Methods.

Epidermal fraction (Epi) was prepared as described (19).Dermal fractions were made following a previously describedmethod of preparing fibroblasts from the skin (20). Briefly,dermises were separated from the epidermis by overnighttrypsin incubation and chopped coarsely in 2 ml of 0.35%collagenase (Worthington, Lakewood, NJ, USA) per dermis.Collagenase suspension was incubated with agitation in awater bath at 37°C for 30 min, and DNase I (Worthington;20,000 U/ml in PBS) was added at 125 �l/ml of collagenasesuspension and incubated for 10 min. This suspension was

filtered through a coarse 100 � Nytex filter and diluted to 100ml per five dermises. Tubes were centrifuged at 800 rpm for5 min, and the supernatant was placed in new tubes to becentrifuged at 1400 rpm for 5 min. The pellets obtained atthe end of this spin were pooled and designated fraction D.

The 800 rpm pellet was dissociated and centrifuged at 300rpm for 5 min, yielding dermal-derived hair follicles (DHF).Supernatant from the 300 rpm pellet was centrifuged twice at1000 rpm for 5 min, then both pellets were pooled anddesignated fraction E.

Organotypic (3-dimensional) cultures

Cells were allowed to adhere overnight to the upper side of aninsert membrane while suspended in 1.4 mM Ca2� medium.Insert membranes used were Millipore� (Multiwell), un-coated, 12 mm diameter, abbreviated Millipore uncoatedmembrane (MU) (0.4 or 3.0 �m pore size used), andcollagen-coated Transwell�-COL Inserts (Corning Inc., Ac-ton, MA, USA), 12 mm diameter, abbreviated Costar-coatedmembrane (CC) (0.4 or 3.0 �m pore size). These membraneswere coated with an equimolar mixture of collagen (types Iand III) derived from bovine placentas. The medium waschanged to 0.05 mM Ca2� and the cultures were submergedfor 72 h. Then the upper side of the membrane was airexposed for 72 h to promote differentiation while a dermis incontact with the membrane was submerged in 0.25 mM Ca2�

medium containing 40 �g/ml ascorbic acid (18, 21). Cultureswere fixed on day 6 in 3% formalin for 5 min and stored in70% ethanol before histological processing. Dermatologicalpunches and histological cassettes and sponges (all fromDaigger, Vernon Hills, IL, USA) were used. Viability andmorphology of the cultures were assessed by H&E staining ofrandom sections. In some experiments, cells from fractions B,D, and E were attached on MU membranes and submerged inlow Ca2� medium containing KGF at 8.4 ng/ml. After 7 days,membranes were air exposed in high Ca2� medium contain-ing 40 �g/ml ascorbic acid. This exposure lasted 48 h topromote differentiation, followed by one or more rounds ofsubmerged conditions to mimic the wound-healing environ-ment. Cells were then fixed at different time points.

Cell sorting and analysis by flow cytometry

Cells were stained with antibodies to CD49f and CD90 (FITC,PharMingen, San Jose, CA, USA), CD34 and CD45 (PE,PharMingen), CD11b (PE) and CD117 (CyChrome, Caltag,Burlingame, CA, USA), and CD105 (Chemicon, Temecula,CA, USA), using standard procedures. Analysis was carriedout on an FACSCalibur flow cytometer (BD Biosciences, SanJose, CA, USA). Cells were stained for the side population(SP) phenotype (22) by incubating cells (1�106 cells/ml inDMEM with 2% FBS) with 3 �g/ml Hoechst 33342 (Molecu-lar Probes, Eugene, OR, USA) for 90 min (37°C, 5% CO2,100% relative humidity). The cells were washed once in coldDMEM with 2% FBS and resuspended for sorting. Cell sortingwas performed on an FACSVantage SE with DiVa option (BDBiosciences). Percents were compared from the unfraction-ated epidermis and dermis and individual fractions.

Cell survival and colony formation assays

Unsorted cells, SP, and CD49f-positive cells from fractionswere placed in a monolayer culture using media containingdifferent levels of calcium and serum. After �4 wk, MTTanalysis was performed to stain mitochondria to locate anyliving cells (Roche Diagnostics, Indianapolis, IN, USA), ac-cording to the manufacturer’s instructions. Prior to solubili-

2051MULTIPOTENTIAL MURINE MESENCHYMAL CELLS

zation, bright-field pictures were taken of representative fieldsin all cultures. To study proliferative capacity, equal numbersof cells (enough to produce confluent monolayers) wereplated in duplicate wells of 24-well plates. Epi were culturedin low calcium medium and dermal-derived fractions werecultured in high calcium medium. Cells were serially passageduntil viable cells were no longer observed in the wells. Tostudy whether some contaminating epidermal progenitorscould affect the proliferative capacity of the dermal fractions,DNase was omitted from the dermal preparations, thus in-creasing keratinocye cross-contamination of the fractions.

Immunoblotting and immunostaining

Freshly isolated cell pellets were lysed with 0.25 M Tris HClbuffer, pH 6.8, containing 3% �-mercapto ethanol and 5%SDS; cells that migrated through a membrane and adhered totissue culture plates were lysed in situ. Electrophoresis was runon NuPage gels (Invitrogen, Carlsbad, CA, USA) and proteinswere electrophoretically transferred onto PVDF membranes(Immobilon-P, Millipore, Bedford, MA, USA). Antibodies forimmunoblotting included anti-keratins 1, 5, 10, 14 (Covance,Princeton, NJ, USA), all at 1:2000 dilution; and actin mono-clonal (C4, Chemicon), rabbit polyclonal to �-tubulin and toPOU3F2 (Abcam, Canbridge, MA, USA) at 1:10000 and1:1000 dilution, respectively. Secondary antibodies were anti-rabbit or anti-mouse IgG horseradish peroxidase linked (Am-ersham, Piscataway, NJ, USA). Chemiluminescent detectionwas done with West Pico (Pierce, Rockford IL, USA).

Immunostaining of paraffin-embedded sections was per-formed according to standard protocols using deparafinationwith Histochoice (Amresco, Solon, OH, USA) and antigenretrieval with 0.01% trypsin or postfixing frozen sections with4% paraformaldehyde. Immunostaining of freshly isolatedcells was performed according to standard protocols onflash-frozen cell spreads postfixed with 4% paraformaldehydefor 10 min. Primary antibodies were anti-keratins K14-FITC,K-10-FITC (Covance) at 1:200 dilution, anti-involucrin (a giftfrom Dr. Richard Eckert) at 1:500 dilution, anti-K15 (BD-PharMingen) at 1:100 dilution, anti-MRP8 (a gift from Dr.Joannes Roth) at 1:200 dilution, and �pep8/DCT (a gift fromDr. Vincent Hearing) at 1:500 dilution. Anti-smooth muscleactin (1:300) dilution was from Biocare Medical (Concord,CA, USA). Antidesmin was from DAKO (Carpenteria, CA,USA), used at 1:500 dilution. Secondary antibodies wereeither Alexa-488 or Alexa-594 conjugates (Molecular Probes).For confocal images, fluorescent cells were examined with aZeiss LSM 510 Confocal Microscope (Carl Zeiss Inc., Thorn-wood, NY, USA) using a 40 � 1.3 NA Plan Neofluar objective.Images were collected sequentially where the FITC, TX Red,and DAPI signals were collected with a BP 505–550 filter,LP560 filter, and BP 385–470 filter after excitation with 488nm, 543 nm, and 364 nm laser lines, respectively.

Multipotential differentiation assays

Differentiation assays for the osteogenic, adipogenic, myo-genic, and chondrogenic lineages were performed according

Figure 1. A) Simplified schematic of fractionpreparation (for more detailed schematic, seeSupplemental Fig. 1). B) Flow cytometry analysisof SP in epidermal subfraction B and dermalfractions D and E. C) Contour plots of CD34and CD49f-positive populations containedwithin the CD117� population of epidermal(subfraction A) and dermal fraction E. PercentCD117� cells were 2.7 and 6.2, respectively. D)MRP8 staining of Epi, DHF, D, and E showingpercent positive cells from 5 independent fields;DHF and D were negative. Top right panel:MRP8-stained neonatal skin; arrow points atarea shown in inset. For description of epider-mal subfractions A and B, see SupplementalMaterials and Methods.

2052 Vol. 21 July 2007 CRIGLER ET AL.The FASEB Journal

to refs. 23, 24. Before induction of differentiation, culturescontaining epidermal subfraction B were grown to conflu-ence in 0.05 mM Ca2� and cultures containing fractions D orE were grown to confluence in 1.4 mM Ca2�-containingmedium.

Osteogenic differentiation was induced with DMEM con-taining 10% FBS, 1% penicillin/streptomycin, nonessentialamino acids, 50 �g/ml ascorbic acid, 10 mM �-glycerolphosphate, and 10 nM dexamethasone for up to 3 wk (24).Mineralization was visualized by staining with Alizarin Red S(2% w/v Alizarin Red S adjusted to pH 4.1 using ammoniumhydroxide) for 5 min at room temperature, followed by awash with water. Osteoblasts were shown by the formation ofcalcium-rich hydroxyapatite in the extracellular space, whichstains orange-red with Alizarin red S. Adipogenic differentia-tion was induced by treating the cultures with DMEM con-taining 10% FBS, 1% penicillin/streptomycin, nonessentialamino acids, 10% rabbit serum, 20 �M ETYA, 10 nM dexa-methasone, and 25 �g/ml insulin (24). For subsequentfeedings, the medium lacked dexamethasone, and rabbitserum concentration was increased to 15%. Adipocytes werevisualized by Oil Red O stain. Chondrogenic differentiationwas induced by plating 5 million cells per well of a round-bottom 96-well plate. Cells were fed with 1.4 mM Ca2�

medium containing 50 �g/ml ascorbic acid, 100 nM dexa-methasone, and 1 ng/ml TGF-� (24). After 3 wk in culture,cell clumps were harvested with a wide bore pipette tip,spread onto poly-l-lysine-coated slides, flash frozen, andpostfixed with 4% paraformaldehyde, followed by stainingwith 0.1% Safranin O. Myogenic differentiation of freshlyisolated fraction E cells plated at confluency was observedafter infection with adenovirally expressed GFP (25). Infec-tion occurred in serum-free medium containing 2.5 �g/mlpolybrene (Sigma) at 10–30 mulitiplicity of infection for 30min at room temperature. Fresh complete medium wasadded thereafter. Cells were observed and photographeddaily. Movie was acquired using an LSM 510 confocal micro-scope.

Array analysis

Focused arrays (GEArray S series Mouse Stem Cell Array,SuperArray, Bethesda, MD, USA) were used to establish thesignature of epidermal and dermal subpopulations after 1 dayin culture. The arrays contain 258 known genes that encodemarkers expressed by stem and differentiating cells as well asvaried growth factors and other regulators of stem cellbehavior. Cells from epidermal subfraction B, from dermalfractions D, E, and from a combination of D and E (termedFb) were isolated and plated overnight in high Ca2�-contain-ing medium. Total RNA was prepared with Trizol (Invitrogen,Carlsbad, CA, USA) according to the manufacturer’s instruc-tions. Arrays were hybridized according to manufacturer’sprotocols for the nonradioactive format. For analysis, themanufacturer’s recommended method was used (www.

superarray.com), which involved densitometric analysis of the30 s exposures to X-ray film and superimposing figures with agrid pattern for quantification of raw pixels. From raw pixeldata, most of row 17 (which contains no printed cDNAs) wasused to subtract background. Once individual pixel valueswere obtained, the results were processed with Excel software.Rpl13a was selected as the housekeeping gene to normalizedata, as Rpl13a was the least variable housekeeping geneamong all four fractions. Then individual values were dividedby the averaged value for Rpl13a in each fraction. A numberof genes that were marginally up-regulated in only one or twofractions were removed from the analysis by setting up thethreshold at 20% expression.

Statistics

Numerical values are expressed as mean � se. For a compar-ison among all groups in the study, multiple comparisons ofmore than two groups were done by 1-way analysis of variance(ANOVA). Unpaired Student’s t test was used for compari-sons between two groups; P values are reported.

RESULTS

Isolation and characterization of epidermal anddermal subpopulations

In the search for murine skin subpopulations thatwould provide a good model for epidermal formationin vitro, we adapted sedimentation and differentialadhesion protocols already used in our laboratory (26)to isolate cellular subpopulations from neonatal mouseskin. Flow charts of the protocols and characterizationof all the fractions are presented in supplemental Fig. 1and supplemental Table 1. By immunoblot, unplated,freshly prepared fractions contained varying amountsof basal and differentiated keratins (supplemental Ta-ble 1 and supplemental Fig. 1). Keratin 1 or 10 and K14were detected mostly in epidermal and hair follicle-derived fractions. In the dermal-derived fractions, K1and K10 were present in fibroblast (D and E) fractions,which could indicate a small keratinocyte contamina-tion to these fractions; however, we noticed that kera-tinocyte-positive populations expanded from E uponculture in high or low calcium (supplemental Table 1and Fig. 4) and in minimal medium conditions (Fig. 5).CD49f (integrin �6) and CD34 were used as markers ofearly keratinocyte precursors (27–29). CD117 (c-kit)was used as a potential bulge marker (30), as were K15

Figure 2. Dermal fraction E has the potentialfor differentiation into epidermis in a 3-dimen-sional culture. A) Epidermal subfraction B wascultured under submerged conditions in 0.05mM Ca2�, then air-exposed on a fresh dermison an MU insert (0.4 �m) in 1.4 mM Ca2� for3 days. B) A mixture of fraction B and E wascultured as in panel A. For a description ofepidermal subfraction B, refer to supplementalMaterials and Methods.


(31) and p63 (32, 33). All epidermal fractions hadequivalent, nearly 100% expression of CD49f. Dermalfractions derived from hair follicles (26) had loweramounts of CD49F-positive cells (�30%). The dermalfibroblast fractions D and E contained the highestamount of CD34� (�45%) cells and the lowest amountof CD49f� (�10%). CD117� cells had low abundanceoverall and were slightly higher in dermal fractions Dand E, although not statistically significant. TheseCD117� cells might be dermal melanocyte precursors(34). Moreover, in epidermal fractions, CD117-positivecells were CD49 intermediate bright, whereas in dermalfraction E the CD117-positive cells were distinct fromCD49f and CD34, thus arguing that the CD117� pop-ulation would not come from epidermal contaminationof the dermal fractions (Fig. 1C).

The side population (35) has been associated withmesenchymal stem cells, although skin stem cells maynot share this marker (36). SP-positive cells were muchmore abundant in epidermal fractions and in hairfollicle buds from the dermis. Dermal fractions D and Epresented the most reproducible data on SP; withinthem, fraction E had a higher percentage, albeit notsignificantly different from D (Fig. 1B). These datasuggested itwas possible to isolate early precursors but

not keratinocyte stem cells by SP analysis of total skin(Fig. 1A). In fractions D and E, CD34� and CD49f�represented distinct subpopulations within the SP.There were slightly more SP cells in E than in D, and10% more CD34/CD49f double negative cells in frac-tion E; thus, we concluded that this fraction maycontain a subset of precursors that is absent fromdermal fraction D.

We used immunofluorescence of freshly isolated cellspreads to investigate K15, a marker of hair folliclebulge region, and p63 (supplemental Table 1 and Fig.6E). The percent of K15� cells was higher in Epi andDHF and very low in fractions D and E (P�0.0001 Epivs. E). Likewise, p63 distinguished Epi and DHF fromfractons D and E (Epi vs. E, P0.0051; D and E werenot significantly different). MRP8, an inflammation-related molecule found in endothelial cells and kera-tinocytes (37), was found in low amounts in Epi; withinthis fraction, it was present in fraction C. Most cells indermal fraction E were positive (Fig. 1D), and a smallamount was detected in dermal subfraction G (seesupplemental Table 1). The mesenchymal markerCD105 (38, 39) was significantly higher in the hairfollicle-containing populations Epi and DHF (Epi vs. E,P0.0129; D vs. E, nonsignificant). CD90 (40) had a

Figure 3. Survival and expansion of isolatedfractions in monolayer culture. A) Isolatedand/or SP-sorted fractions were plated in 0.05mM Ca2�/2% serum containing SMEM me-dium for 1 month in triplicate wells, then anMTT assay was carried out. Cell survival wascalculated from abs 490–650 (left panel) or abs490–650/cell input (right panel). In a separateexperiment, equal numbers of SP cells wereplated and results followed the same pattern.*Significant difference in the survival of indi-vidual fractions compared with fraction E. Non-sorted fractions, P 0.0018 B vs. E, P 0.0244C vs. E. Sorted fractions P 0.0036 B vs. E, P �0.0001 for both F and G vs. E. B) Fractions B, D,and E cells sorted based on CD49f were placedin 24-well plates at 10,000 cells/well and cul-tured with 2% serum and either 0.05 mM or 1.4mM Ca2� for 1 month, then an MTT assay wascarried out. In CD49f-positive fractions, P 0.00642 B vs. E. Right set of panels showMTT-stained cells. C) SP-sorted dermal sub-populations were plated in duplicate wells of a24-well plate at 10,000 cells per well in minimalmedium containing 1.4 mM Ca2�. In SP-sortedfractions, P � 0.0001 D vs. E. Right panel:MTT-stained cells. These experiments were re-peated at least twice; one representative set ofresults is shown. For a description of epidermaland dermal subfractions, refer to supplementalMaterials and Methods.


higher percent in the dermal fractions and slightlyhigher in E than D, albeit not significant (CD90 Epi vs.E, P0.0032, D vs. E, nonsignificant). As MRP8 mightbe associated with immune infiltrates, we confirmed byCD11b that populations containing granulocytes, mac-rophages, and monocytes were not significantly differ-ent among the fractions. The sum of these findingssuggested that our protocol selectively enriches forepidermal precursors associated with either the hair folli-cle bulge or dermal papillae, or with a yet uncharacterizedmesenchymal progenitor cells present in the dermis.

Growth of the fractions in three dimensions

Epidermal populations containing keratinocytes andhair follicle buds grew occasionally and not reproduc-ibly in 3-dimensional cultures. Reproducibility im-proved when we selected hair follicle buds (epidermalsubfraction A) and when we used membranes in con-tact with fresh dermis (18) (Supplemental Fig. 2). Wecompared the growth of epidermal cells (subfraction

B) with dermal fractions D and E. We consistentlyobserved that fraction E but not D was able to formmonolayers that stained H&E positive and appearedviable. Epidermal cells alone did not survive (Fig. 2A),but a mixture of B and E did (Fig. 2B). Taken together,these observations suggested that fraction E containsfunctional epidermal precursors.

To understand whether a small population of con-taminating follicles could be responsible for the prolif-erative capacity of dermal fraction E, we looked at theproliferative capacity of Epi, DHF, D, and E in mono-layer culture. When DNase was included in the dermalpreparations, detectable levels of keratin 10 and 14decreased, suggesting that the use of DNase substan-tially decreases cross contamination of fractions (sup-plemental Fig. 3). When we plated an equal number ofcells from a preparation that did not include DNase, wewere able to carry Epi in low calcium medium for threepassages. Dermal-derived fractions (DHF, D, and E),cultured in high calcium medium, failed to survive pastpassage 7. When DNase was added to the preparation,

Figure 4. Dermal fraction E does not requireunderlying dermis to survive in 3-dimensional cul-tures under epidermal conditions. Fraction E wasplated in a 3 �m MU insert in high calcium-containing medium to favor attachment. The nextday the medium was changed to low calcium-containing KGF and cultured under submergedconditions for 1 wk, followed by air exposure asdetailed below. A) Air exposure for 2 days inmedium containing 0.24 mM Ca2�. B) Air expo-sure for 6 days. C) Air exposure for 9 days.D) Higher magnification of the 2 days culture. E)K14-FITC stain of 2 days culture. F) Double stainK14-FITC and K10 with a Texas Red anti-rabbitsecondary; some colocalization is observed as yel-low signal (arrowhead). G) Fraction E was cultured,submerged, and air-exposed as in panel A, thensubmerged again for 4 days and air-exposed for anadditional 7 days. H) K14-FITC stain of sample inpanel G. I) Double stain of sample in panel G withK14-FITC and K10 with a Texas Red anti-rabbitsecondary; yellow shows increased areas of colocal-ization of the Texas Red and FITC signals com-pared with panel F (arrowhead). J) Double stain ofsample in panel G with K14-FITC and involucrinwith a Texas Red anti-rabbit secondary. K) Frac-tions D and E were on MU inserts (3 � pore size).Cells that migrated to the plastic well below werecultured in submerged conditions for 8 days witheither 0.05 mM (Lo) or 1.4 mM (Hi) Ca2� me-dium. Fraction B was cultured in 0.05 mM Ca2� for3 days as a control. Lysates for one set of wells oftwo replicates are shown. -actin is shown as aloading control. Quantitation of averages for bothreplicates is shown.


Figure 5. Differentiation assays. A) Osteogenesis differentiation, Alizarin Red stain. B) Adipogenic differentiation, Oil Red Ostain. Fraction B did not survive these conditions. C) Chondrogenesis differentiation, Safranin O stain. D) Myogenicdifferentiation of adenovirally infected (GFP) confluent cultures grown in complete MEM medium for 3 wk; some cells withinfraction E had myocyte morphology and were contractile (see supplemental video). E) Staining of cells cultured in MEMmedium containing 0.05 mM Ca2� and 2% FBS for a month. K5 denotes an active proliferating population expressing thebasal keratinocyte marker K5. Tyrp2 denotes the melanoblast/early melanocyte marker dopachrometautomerase.


Epi failed to survive past passage 2; DHF failed tosurvive past passage 9 and dermal-derived fractions Dand E survived to passage 18. At this point, only fractionE had viable cells, which had largely senescent mor-phology and contained some adipocyte-looking cells inthe culture. In the same experiment, we maintainedcells that remained in the well after passage 1 for amonth with only one medium change. In the epidermalfraction, no cells survived. In DHF, cells had lifted anddied by 3 wk; however, a monolayer of cells from bothfractions D and E was still present after a month, similarto what we report in Fig. 3.

Survival and differentiation of monolayer cultures inreduced serum

We hypothesized that epidermal precursors in thedermis would mobilize to restore damaged epidermis,and we could mimic this process in vitro. We tested

whether we could functionally identify which fractionwould respond to wound-healing cues. Those cells, wehypothesized, would be able to survive harsh conditionsthrough quiescence prior to responding to wound-healing signals. To test the former, we isolated and/orsorted these fractions and placed them in monolayerculture for 1 month in medium containing 2% serumand either 0.05 mM or 1.4 mM Ca2�. We then assayedcell survival by MTT. The unsorted epidermal subfrac-tions B and C had increased survival; likewise, SP-sortedcells from hair follicle subfractions (B, F, and G)showed the greatest ability to survive in minimal mediacontaining 0.05 mM Ca2� (Fig. 3A). Epidermal frac-tions sorted based on the integrin marker CD49f sur-vived only under low calcium conditions (Fig. 3B) Themoderately increased survival of a CD49f population inhigh calcium suggested that expansion of an epider-mal-like population under high calcium and air expo-sure (see Figs. 2 and 4) may originate in a population

Figure 6. Additional characterization of thefractions. A) One min exposure of membranearrays (for easier identification of genes). Someexamples are highlighted: red indicates geneswith a positive value in fraction E, blue indi-cates genes with positive values in more thanone fraction and not in fraction E. Gene inten-sity values were calculated from a 30 s exposedfilm in which housekeeping genes (bottomrow) were not saturating. B) Not all housekeep-ing genes in the array were evenly present in allfractions, thus Rpl13a was chosen for normal-ization. C) Bar graph showing percent contri-bution of each fraction to total expressionlevels; most genes that were negatively regu-lated belong to fraction E. D) Bar graph show-ing percent contribution of each fraction to thetotal expression levels in up-regulated genes;only genes in which fraction E had a positivevalue are shown. E) Immunostaining of freshlyisolated flash-frozen fractions with markers p63and K15. Quantitation of positive stained cellsfrom five random fields is shown below; *signif-icant differences between Epi and E (p63,P0.0051; K15, P�0.0001). F) Immunoblots ofproteins detected in fraction E in the array(K14, Fig. 6B) or below detection in E (Pou3f2,Fig. 6C). Left panel shows quantitation of theirrelative levels by NIH image.


distinct from CD49f�. In SP-sorted fractions, a popu-lation of small, vigorously proliferating cells expandedfrom fraction E, and not from D or epidermal subfrac-tion B, under high calcium conditions (Fig. 3C). Selec-tion by c-Kit (CD117) showed no differences betweendermal fractions D and E (data not shown). Together,these observations suggested that a subpopulation of SP� dermal cells, distinct from the CD49� population,contains epidermal precursors capable of expanding inhigher calcium conditions. Therefore, the effects wehad observed when fractions B and E were mixed andcultured in three dimensions (i.e., air-exposed in highcalcium, Fig. 2) could be explained by the ability ofselected dermal cells to form epidermis in vitro.

Survival and differentiation of dermal fraction E in3-dimensional cultures

Based on the results in monolayer cultures, we rea-soned that dermal subpopulations survived morereadily than others because they contained early pro-genitor cells. Upon wounding, healing would arisefrom those populations and provide rapidly amplifyingcells for epidermal formation. Accordingly, we simu-lated conditions that favor epidermal cell expansion. In3-dimensional cultures, these conditions consisted oflow Ca2�-containing medium with KGF when sub-merged and high Ca2� and ascorbic acid-containingmedium (21) when air exposed. Fraction E was platedin 1.4 mM Ca2� to favor attachment to the membrane,and 24 h later the medium was changed to 0.05 mMCa2� medium containing KGF. Cultures were grownsubmerged for a week and air-exposed in mediumcontaining 1.4 mM Ca2� for 2 days, 6 days, and 9 days(Fig. 4A–C). Fraction E was grown submerged andair-exposed as in panel A, then submerged again for 4days and air-exposed for an additional 7 days (Fig. 4G).The early cultures (Fig. 4D) stained positive for CD34(data not shown).

We compared the expansion of K10/K14 stainingcells in the cultures. Figure 4E, H demonstrates K14staining of the membranes shown in panels D, G. Figure4F, I shows stronger staining for K10 of the panel Gsample, suggesting that a second phase of submergeculture, followed by air exposure, directs differentia-tion of keratinocytes in vitro. Colocalization of a laterkeratinocyte marker involucrin with K14 further sup-ported this differentiation (Fig. 4J). This experimentwas subsequently repeated with epidermal fraction B,but the cells were nonviable. Immunoblots of dermalsubpopulations that migrated through 0.3 � mem-branes adhered to the plastic underneath and werecultured in two individual wells with 0.05 mM or 1.4mM Ca2�, suggested that fraction E more than fractionD contained cells able to migrate through membranepores and expand keratin-positive cells under bothcalcium conditions; the differences, however, were notstatistically significant (Fig. 4K).

Differentiation assays

To test whether fraction E had multilineage potential,we conducted the following assays (24). Fractions B, ortotal epidermal preparation, and D were used as com-parison controls. Fraction E was positive for osteogen-esis, chondrogenesis, and adipogenesis (Fig. 5A, B).Under starvation and 0.05 mM Ca2� conditions, frac-tion E and Epi but not D expanded a K5-positivepopulation whereas fraction D more than E expandeda melanocyte-positive population (as marked by theearly melanocyte marker Tyrp8/Dct) (Fig. 5E). DCT,an early melanoblast marker, has been proposed to bea marker of the bulge (16).

In addition to these findings, both fractions D and E,but most prominently fraction E, gave rise in culture tocells with contractile behavior (Fig. 5D and supplemen-tal movie). We observed this phenotype on confluentmonolayers and on air-exposed 3-dimensional culturesas well (data not shown).

Stem cell and differentiation markers weredifferentially expressed in skin subpopulations

Epidermal cells (subfraction B) and dermal fractions Dand E were plated in complete high calcium medium.Total RNAs were prepared from the overnight-adheredcells and hybridization to mouse stem cell arrays wascarried out. These arrays confirmed functional obser-vations of differences between epidermal and dermalfractions and within dermal subfractions D and E.Differentiation and stem cell markers were differen-tially expressed in those populations (Fig. 6A). Asexamples, genes circled in red were up-regulated infraction E; blue circles mark genes that were belowdetection limits in fraction E while being up-regulatedin the others. Figure 6C shows that many genes wereexclusively below the limits of detection in E butup-regulated in other fractions, suggesting that nega-tive selection might enrich for stem cells in fraction E(24, 41). Specifically, genes involved in cellular adhe-sion such as integrin �4 and ICAM 5 were belowdetection in E and highest in D. Figure 6D representsthe genes detected in E, as well as in other subfractions,and their percent of distribution. Changes in theexpression of K14, 10, and of the transcription factorPou3f2 were confirmed at the protein level by immu-nofluorescence and immunoblotting (Fig. 6F).

In vivo engraftment and differentiation

Cells from Epi, DHF, D, and E fractions were engraftedinto skin of immunodeficient mice in two ways: theywere either subcutaneously injected or grafted under asilicon dome covering the graft bed as typical epider-mal grafts.

Isolated subpopulations were injected at definedsubcutaneous sites on the back of immunodeficientmice. Mice were sacrificed at 1 and 2 wk postinjection.In the 1 wk histology sections, clusters of cells in the


injected sites of fractions Epi and DHF could not befound. Fraction D formed a fibrotic mass that hadabundant vascularity, consistent with previous reports(42). In the fraction E group, 1 wk after subcutaneousinjection some pigmented aggregates were observedwithin host muscle (Fig. 7A, Fontana Masson stain). Inaddition, both fraction D and E, but most prominentlyfraction E, displayed fusiform structures that stainedpositive for desmin and smooth muscle actin in theperiphery and positive for smooth muscle actin towardthe center (Fig. 7A, 1 wk). These structures were not asprominent in the 2 wk histology sections. These stain-ing patterns suggested early differentiation of the mes-enchymal cells into the myogenic lineage. In the 2 wkpostinjection set, we were able to find the injected Epiand DHF cells, which formed abundant cysts that werelarger in DHF (Fig. 7A, 2 wk), but fraction D cells couldnot be found. Fraction E showed a few small cysts,numerous melanocytes surrounding these and inter-spersed in the dermal tissue, as well as some nonmela-nized, fusiform structures (Fig. 7A, arrowhead).

For the epidermal grafts, 5 days postgrafting thesilicon domes were removed to facilitate wound heal-ing. At that time animals grafted with fraction E had asignificantly bloodier wound, which nonetheless healedfaster, as shown in Fig. 7B. All four subpopulations hadgraft take as evidenced by the presence of dermalmelanocytes not associated with hair follicles, consis-tent with previous reports (43). A very dark healedwound was appreciated in DHF, which contained nu-merous hair follicles and cysts as observed in thehistology (Fig. 7B). Animals grafted with fraction Edisplayed a dark ring around the edges of the woundsite (Fig. 7B, arrow). On histology, it could be observedthat there were pigmented nests deep within the host

dermis, where a few contaminating hair follicles accu-mulated. The environment of the wound edge has beenreported to be more favorable to hair follicle develop-ment (43, 44). In the graft central areas, away fromwound edges, very few melanocytes were observed inEpi and D fractions; in DHF they were concentrated inareas surrounding preexisting follicles and cysts,whereas in fraction E melanocytes could be found inthe center of the graft. Melanocytes appeared vacuo-lated in Epi and were lighter in color than those in theE fraction (Fig. 7B).

DISCUSSION

Murine skin bioengineering lags behind progress madein the field of human skin bioengineering (45, 46). Acomplete epidermis may be reformed after the growthof mouse keratinocytes on nonviable, de-epidermizeddermis (10, 47), and some studies have shown thatmixing human and murine skin components may re-construct an epithelial layer (48). Collectively, thesestudies suggest there may be different requirements forthe in vitro formation of mouse and human skin, suchas growth factors and cellular selection. In vivo, anepidermis can be formed by grafting epidermal cellswith a dermal component onto nude mice (49). Thisdermal component was made up of a combination offractions D and E, described in this paper. Weinberg etal. were able to demonstrate hair growth when hairfollicles were grafted with dermal papilla and not intheir absence (17). Our work aimed at expanding thesefindings in vitro. We could not consistently replicate thein vitro formation of an epidermal layer when we usedneonatal mouse epidermal cells cultured on artificial

Figure 7. Engraftment and differentiation offraction E in vivo. Isolated subpopulations wereinjected at defined subcutaneous sites on theback of immunodeficient mice. Mice were sac-rificed 1 and 2 wk postinjection. A) In thefraction E group, pigmented aggregates areseen within host muscle (arrow and upper rightinset). Arrowhead and lower left inset showpositive Fontana Masson fusiform structure,which stains positive for desmin and smoothmuscle actin in the periphery and positive forsmooth muscle actin toward the center (stainedsections are from the area in the lower leftinset). Clusters of cells in the injected sites offractions Epi and DHF could not be found inthe animals. At 2 wk, injected Epi and DHF cellsformed abundant cysts, which were larger inDHF. Fraction E had some small cysts, numer-ous melanocytes surrounding these and inter-spersed in the dermal tissue, as well as somenonmelanized fusiform structures (arrowhead).Fraction D cells could not be found in theanimals. B) Week 3 postgrafting with siliconchambers. Arrows point in graft area in all four

panels. A dark healed wound was seen in DHF, which contained numerous hair follicles and cysts, as observed in the FontanaMasson panel below. Animals grafted with fraction E displayed a dark ring around the edges of the wound site; pigmented nestswere deep within the host dermis, as seen in the Fontana Masson panel below. In the graft central areas, away from wound edges,very few melanocytes are observed in Epi and many more appear in fraction E.


inserts with gamma-irradiated dermal cells cultured onthe underside (50). Therefore, we focused on isolatingdifferent cellular subpopulations from neonatal mouseskin, taking advantage of properties such as differencesin buoyant density or differential attachment to extra-cellular matrix components (51), which can selectivelyenrich for hair follicle-derived cells (17, 52). Ourisolation scheme selected seven populations [threederived from the epidermis (A, B, C) and four from thedermis (D, E, F, G)], characterized as shown in supple-mental Table 1. Again, we had limited success withsubpopulations obtained from the epidermal compart-ment on organotypic cultures. Among the dermalfractions, F grew well under submerged conditions inlow calcium but failed to grow in 3-dimensional cul-tures. Cocultures of epidermal cells and dermal frac-tions, designed to enrich mesenchymal cells, grewsuccessfully in 3-dimensional cultures. Moreover, weobserved that de-epidermized dermis controls oftenreformed an epithelium even after gamma irradiation,and thus the dermis may contain radio-resistant pro-genitors. We reasoned that our isolation procedurecould shed cells from the epithelial fractions into themesenchymal component of the dermis. Those precur-sors would be derived from the bulge or the dermalpapilla sections of the hair follicle and would, inconjunction with fibroblasts, favor epidermal formationin vitro. We tried to assess the effects of contaminationof epidermal cells in the dermal preparations. Increas-ing keratinocyte contamination was found when DNasewas left out of our isolation protocol (the use of DNaseis a current practice in our laboratory, as described inref. 15, which does not specify why it was added). Toour knowledge, other papers published that have ex-amined dermal cells, such as ref. 53, which describesDEEP-1 cells, fail to use DNase; we demonstrate that theuse of DNase in preparations of dermal cells signifi-cantly reduces hair follicle contamination.

When plated on membranes, mesenchymal cell prep-arations formed multiple layers and migrated throughpores to the underside. Furthermore, in submergedconditions, keratinocyte monolayers formed that strat-ified at the air-medium interphase and expressed ker-atin 14 and other markers, as seen in Fig. 4.

The mesenchymal cells expressed cell surface mark-ers that distinguished them from keratinocytes. Theyexpressed low levels of CD49f and were rich in CD34;they had detectable CD117 and a higher level of CD90expression. These cells distinctly expressed MRP8(S1000A8), and this was not due to differential contentin cells that express the type 3 complement receptor,CD11b. Changes in the expression of S100 A8 havebeen reported in wound fibroblasts (54). A distinguish-ing characteristic of the mesenchymal population wasthe ability to survive and propagate under minimalculture conditions. This may have parallels in vivo, asreserve or hematopoietic stem cells are reported tosurvive for long periods in a poorly oxygenated niche(55). Thus, the long-term survival of a mesenchymalprecursor population may require that the cells have

enhanced survival capabilities under stressful condi-tions.

The mesenchymal cells (from fraction E) coulddifferentiate into different lineages, thus forming os-teocytes, chondrocytes, myocytes, and adipocytes. Werepeatedly observed that when dermal cells were placedin 3-dimensional cultures under submerged conditions,only isolated cells survived on the membranes, andthose cells were evenly spaced. We hypothesized thatthose cells would differentiate into epidermis if we gavethem an opportunity to expand with appropriategrowth factors; we tested this by simulating the skinwound-healing environment. Our simple model of skinwound healing consisted of the growth of fraction Ealone under alternating submerged/air exposed-sub-merged � KGF/air-exposed conditions. Early cultureswere positive for the stem cell marker CD34 and for thebasal keratinocyte marker K14, whereas cultures thathad undergone more than one round of submerged/air-exposed conditions acquired the early differentia-tion keratinocyte markers K10 and involucrin.

We have not determined the origin of these mesen-chymal cells. It is possible that cells from the hairfollicle bulge area become dislodged and mixed in withthe dermal populations in our isolation protocol. It hasbeen reported that putative human keratinocyte stemcells have the lowest amounts of desmoglein (56) andtherefore could become loose and remain in the der-mal subpopulations separated from the dermal hairfollicle cells. Thus, these precursors could come fromeither the bulge (10) or the dermal papilla, which maycontain early transient amplifying cells (16). However,the significant lower levels of p63 and K15 between E,D, and DHF argue against bulge contamination of ourcultures.

Stem cells and early precursors express lineage-spe-cific genes at low levels before lineage commitment(41). Analysis of gene array data showed that amongthe few genes detected in fraction E were the interme-diate filament keratins 14 and 17. K14 is a basal keratinthat precedes K1 and the later stages of epidermalsquamous differentiation; K14 was relatively moreabundant in fraction E than in D, thus confirming ourimmunoblotting results. K17 is an early epidermalprogenitor marker (57, 58) and a marker for bulge (16,59). Injury to the skin results in an induction of K17concomitant with activation of keratinocytes for reepi-thelialization. K17 was not significantly different infraction E. Only Acta2 was more prominent in fractionE than in the others. Acta2 encodes the alpha form ofsmooth muscle actin. The contractile isoform alpha-2actin is also called smooth muscle and aorta isoform; itis present in musculoskeletal connective tissue cells andhas been demonstrated in early progenitors (60, 61).The expression of smooth muscle actin suggests thatthe population might contain myofibroblast precursors(62). The bone morphogenic proteins BMP3 andBMP6, signals required for stem cells, the stem cellmarker CD34, GJB1, a member of the gap junctionconnexin family, and nerve growth factor were ex-


pressed about equally in all fractions analyzed and thuswere not specific to fraction E. However, Cnp1, amarker for myelin, was expressed at much lower levelsin E and B than in D and Fb, suggesting that oligoden-drocytes or their precursors could be present in the Dpreparation and absent in fractions E and B. Cnp-positive cells were detected in bulge cells after differ-entiation in culture (63). Some genes whose expressionwas lower in the E fraction were Fzd1 and 8, which weremore prominent in D and B, consistent with recentlypublished patterns of frizzled genes in skin and hairfollicles (64). Many stem cells and differentiation genesfrom the array were below the limits of detection in Ebut up-regulated in the other fractions.

When injected subcutaneously, fraction E gave rise tostructures that stained positive for smooth muscle actinand desmin. In silicon chamber grafts, which are con-ducive to epidermal differentiation, melanocytes notassociated with hair follicles were abundant in thedermis. Smooth muscle actin-positive cells and melano-blasts are both neural crest derivatives, and one or theother may predominate, depending on the microenvi-ronment (65). Collectively, our grafting and subcuta-neous injection results suggest that dermal fraction Ehas the ability to differentiate into muscle-like struc-tures when subcutaneously grafted or into melanocyteswhen grafted in an environment in which the canonical�-catenin pathway predominates.

In summary, we isolated a mesenchymal populationdevoid of hematopoietic cell markers and capable offorming adipocytes, chondrocytes, osteoclasts, func-tional smooth muscle cells, and keratinocytes in vitro.Moreover, this population was able to generate anepidermal layer in a 3-dimensional model of murineskin. By comparison, this population lacks the expres-sion of markers shared by epidermal and other dermalsubpopulations such as ICAM 5 and �-4 integrin; theexpression of a selected set of genes is strikingly differ-ent from that of other epidermal and dermal subfrac-tions. We were able to demonstrate the relative absencefrom dermal fractions D and E of epithelial stem cellmarkers (K15, p63); CD90, a mesenchymal progenitormarker, distinguished D and E from Epi and DHFwhereas Cd105 did not. MRP8 (S100A8) is a distin-guishing marker to this fraction but CD11b did notdistinguish it from the other fractions studied.

This isolation procedure will be useful for the studyof genetically altered mouse strains as models forvarious human skin conditions. Further analysis ofthese cells may provide new insight into how mesenchy-mal-to epithelial transitions play a role in tissueregeneration.

The authors thank Stuart Yuspa, Ulrike Lichti, Julio Valen-cia, and Luowei Li for critical discussions and helpful advice.We also thank Kevin Taylor, Joseph Zakhari, and SusanGarfield for expert technical assistance. This research wassupported in part by the Intramural Research Program of theNational Institutes of Health National Cancer Institute, Cen-ter for Cancer Research.

REFERENCES

1. Medalie, D. A., Eming, S. A., Tompkins, R. G., Yarmush, M. L.,Krueger, G. G., and Morgan, J. R. (1996) Evaluation of humanskin reconstituted from composite grafts of cultured keratino-cytes and human acellular dermis transplanted to athymic mice.J. Invest. Dermatol. 107, 121–127

2. Spradling, A. C. (2004) Stem cells: more like a man. Nature 428,133–134

3. Petiot, A., Conti, F. J., Grose, R., Revest, J. M., Hodivala-Dilke,K. M., and Dickson, C. (2003) A crucial role for Fgfr2-IIIbsignalling in epidermal development and hair follicle pattern-ing. Development 130, 5493–5501

4. Webb, A., Li, A., and Kaur, P. (2004) Location and phenotype ofhuman adult keratinocyte stem cells of the skin. Differentiation72, 387–395

5. Asahara, T., Murohara, T., Sullivan, A., Silver, M., van der Zee,R., Li, T., Witzenbichler, B., Schatteman, G., and Isner, J. M.(1997) Isolation of putative progenitor endothelial cells forangiogenesis. Science 275, 964–967

6. Potten, C. S., and Bullock, J. C. (1983) Cell kinetic studies in theepidermis of the mouse. I. Changes. in. labeling index. with.time after. tritiated. thymidine administration. Experientia 39,1125–1129

7. Watt, F. M. (1998) Epidermal stem cells: markers, patterningand the control of stem cell fate. Philos. Trans. R. Soc. Lond. B.Biol. Sci. 353, 831–837

8. Lavker, R. M., and Sun, T. T. (2000) Epidermal stem cells:properties, markers, and location. Proc. Natl. Acad. Sci. U. S. A.97, 13473–13475

9. 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

10. Taylor, G., Lehrer, M. S., Jensen, P. J., Sun, T. T., and Lavker,R. M. (2000) Involvement of follicular stem cells in forming notonly the follicle but also the epidermis. Cell 102, 451–461

11. Kalcheim, C., and Ben-Yair, R. (2005) Cell rearrangementsduring development of the somite and its derivatives. Curr. Opin.Genet. Dev. 15, 371–380

12. Masson-Gadais, B., Fugere, C., Paquet, C., Leclerc, S., Lefort,N. R., Germain, L., and Guerin, S. L. (2005) The Feederlayer-mediated extended lifetime of cultured human skin kera-tinocytes is associated with altered levels of the transcriptionfactors Sp1 and Sp3. J. Cell Physiol. 206, 831–842

13. Zeisberg, M., Shah, A. A., and Kalluri, R. (2005) Bone morpho-genic protein-7 induces mesenchymal to epithelial transition inadult renal fibroblasts and facilitates regeneration of injuredkidney. J. Biol. Chem. 280, 8094–8100

14. Bickenbach, J. R., and Grinnell, K. L. (2004) Epidermal stemcells: interactions in developmental environments. Differentiation72, 371–380

15. Rogers, G., Martinet, N., Steinert, P., Wynn, P., Roop, D.,Kilkenny, A., Morgan, D., and Yuspa, S. H. (1987) Cultivation ofmurine hair follicles as organoids in a collagen matrix. J. Invest.Dermatol. 89, 369–379

16. Fernandes, K. J., McKenzie, I. A., Mill, P., Smith, K. M., Akhavan,M., Barnabe-Heider, F., Biernaskie, J., Junek, A., Kobayashi,N. R., Toma, J. G., et al. (2004) A dermal niche for multipotentadult skin-derived precursor cells. Nat. Cell Biol. 6, 1082–1093

17. Weinberg, W. C., Goodman, L. V., George, C., Morgan, D. L.,Ledbetter, S., Yuspa, S. H., and Lichti, U. (1993) Reconstitutionof hair follicle development in vivo: determination of follicleformation, hair growth, and hair quality by dermal cells. J. Invest.Dermatol. 100, 229–236

18. Madison, K. C., Swartzendruber, D. C., Wertz, P. W., andDowning, D. T. (1990) Sphingolipid metabolism in organo-typic mouse keratinocyte cultures. J. Invest. Dermatol. 95,657– 664

19. Dlugosz, A. A., Glick, A. B., Tennenbaum, T., Weinberg, W. C.,and Yuspa, S. H. (1995) Isolation and utilization of epidermalkeratinocytes for oncogene research. Methods Enzymol. 254, 3–20

20. Lichti, U., Weinberg, W. C., and Yuspa, S. H. (1991) Propertiesof murine pelage hair follicles in monolayer and collagenmatrix cultures. J. Invest. Dermatol. 96, 81S-82S

21. Pasonen-Seppanen, S., Suhonen, T. M., Kirjavainen, M., Suihko,E., Urtti, A., Miettinen, M., Hyttinen, M., Tammi, M., and


Tammi, R. (2001) Vitamin C enhances differentiation of acontinuous keratinocyte cell line (REK) into epidermis withnormal stratum corneum ultrastructure and functional perme-ability barrier. Histochem. Cell Biol. 116, 287–297

22. Goodell, M. A., Brose, K., Paradis, G., Conner, A. S., andMulligan, R. C. (1996) Isolation and functional properties ofmurine hematopoietic stem cells that are replicating in vivo. J.Exp. Med. 183, 1797–1806

23. Rombouts, W. J., and Ploemacher, R. E. (2003) Primary murineMSC show highly efficient homing to the bone marrow but losehoming ability following culture. Leukemia 17, 160–170

24. Baddoo, M., Hill, K., Wilkinson, R., Gaupp, D., Hughes, C.,Kopen, G. C., and Phinney, D. G. (2003) Characterization ofmesenchymal stem cells isolated from murine bone marrow bynegative selection. J. Cell Biochem. 89, 1235–1249

25. Li, L., Lorenzo, P. S., Bogi, K., Blumberg, P. M., and Yuspa, S. H.(1999) Protein kinase Cdelta targets mitochondria, alters mito-chondrial membrane potential, and induces apoptosis in nor-mal and neoplastic keratinocytes when overexpressed by anadenoviral vector. Mol. Cell Biol. 19, 8547–8558

26. Yuspa, S. H., Wang, Q., Weinberg, W. C., Goodman, L., Ledbet-ter, S., Dooley, T., and Lichti, U. (1993) Regulation of hairfollicle development: an in vitro model for hair follicle invasionof dermis and associated connective tissue remodeling. J. Invest.Dermatol. 101, 27S-32S

27. Li, A., Simmons, P. J., and Kaur, P. (1998) Identification andisolation of candidate human keratinocyte stem cells based oncell surface phenotype. Proc. Natl. Acad. Sci. U. S. A. 95, 3902–3907

28. Tani, H., Morris, R. J., and Kaur, P. (2000) Enrichment formurine keratinocyte stem cells based on cell surface phenotype.Proc. Natl. Acad. Sci. U. S. A. 97, 10960–10965

29. Trempus, C. S., Morris, R. J., Bortner, C. D., Cotsarelis, G.,Faircloth, R. S., Reece, J. M., and Tennant, R. W. (2003)Enrichment for living murine keratinocytes from the hairfollicle bulge with the cell surface marker CD34. J. Invest.Dermatol. 120, 501–511

30. Nishimura, E. K., Jordan, S. A., Oshima, H., Yoshida, H., Osawa,M., Moriyama, M., Jackson, I. J., Barrandon, Y., Miyachi, Y., andNishikawa, S. (2002) Dominant role of the niche in melanocytestem-cell fate determination. Nature 416, 854–860

31. Lyle, S., Christofidou-Solomidou, M., Liu, Y., Elder, D. E.,Albelda, S., and Cotsarelis, G. (1998) The C8/144B mono-clonal antibody recognizes cytokeratin 15 and defines thelocation of human hair follicle stem cells. J. Cell Sci. 111,3179 –3188

32. Mills, A. A., Zheng, B., Wang, X. J., Vogel, H., Roop, D. R., andBradley, A. (1999) p63 is a p53 homologue required for limband epidermal morphogenesis. Nature 398, 708–713

33. Pellegrini, G., Dellambra, E., Golisano, O., Martinelli, E., Fan-tozzi, I., Bondanza, S., Ponzin, D., McKeon, F., and De Luca, M.(2001) p63 identifies keratinocyte stem cells. Proc. Natl. Acad.Sci. U. S. A. 98, 3156–3161

34. Tosney, K. W. (2004) Long-distance cue from emerging dermisstimulates neural crest melanoblast migration. Dev. Dyn. 229,99–108

35. Zhou, S., Schuetz, J. D., Bunting, K. D., Colapietro, A. M.,Sampath, J., Morris, J. J., Lagutina, I., Grosveld, G. C., Osawa,M., Nakauchi, H., and Sorrentino, B. P. (2001) The ABCtransporter Bcrp1/ABCG2 is expressed in a wide variety of stemcells and is a molecular determinant of the side-populationphenotype. Nat. Med. 7, 1028–1034

36. Terunuma, A., Jackson, K. L., Kapoor, V., Telford, W. G., andVogel, J. C. (2003) Side population keratinocytes resemblingbone marrow side population stem cells are distinct fromlabel-retaining keratinocyte stem cells. J. Invest. Dermatol. 121,1095–1103

37. Thorey, I. S., Roth, J., Regenbogen, J., Halle, J. P., Bittner, M.,Vogl, T., Kaesler, S., Bugnon, P., Reitmaier, B., Durka, S., Graf,A., et al. (2001) The Ca2�-binding proteins S100A8 and S100A9are encoded by novel injury-regulated genes. J. Biol. Chem. 276,35818–35825

38. Rokhlin, O. W., and Cohen, M. B. (1995) Differential sensitivityof human prostatic cancer cell lines to the effects of proteinkinase and phosphatase inhibitors. Cancer Lett. 98, 103–110

39. Lee, R. H., Kim, B., Choi, I., Kim, H., Choi, H. S., Suh, K.,Bae, Y. C., and Jung, J. S. (2004) Characterization and

expression analysis of mesenchymal stem cells from humanbone marrow and adipose tissue. Cell Physiol. Biochem. 14,311–324

40. Miyahara, Y., Nagaya, N., Kataoka, M., Yanagawa, B., Tanaka, K.,Hao, H., Ishino, K., Ishida, H., Shimizu, T., Kangawa, K., et al.(2006) Monolayered mesenchymal stem cells repair scarredmyocardium after myocardial infarction. Nat. Med. 12, 459–465

41. Eckfeldt, C. E., Mendenhall, E. M., and Verfaillie, C. M. (2005)The molecular repertoire of the ‘almighty’ stem cell. Nat. Rev.Mol. Cell Biol. 6, 726–737

42. Zheng, Y., Du, X., Wang, W., Boucher, M., Parimoo, S., andStenn, K. (2005) Organogenesis from dissociated cells: genera-tion of mature cycling hair follicles from skin-derived cells.J. Invest. Dermatol. 124, 867–876

43. Lichti, U., Weinberg, W. C., Goodman, L., Ledbetter, S., Dooley,T., Morgan, D., and Yuspa, S. H. (1993) In vivo regulation ofmurine hair growth: insights from grafting defined cell popula-tions onto nude mice. J. Invest. Dermatol. 101, 124S-129S

44. Lichti, U., Scandurro, A. B., Kartasova, T., Rubin, J. S., LaRo-chelle, W., and Yuspa, S. H. (1995) Hair follicle developmentand hair growth from defined cell populations grafted ontonude mice. J. Invest. Dermatol. 104, 43S-44S

45. Arosarena, O. (2005) Tissue engineering. Curr. Opin. Otolaryn-gol. Head Neck Surg. 13, 233–241

46. Andriani, F., Garfield, J., Fusenig, N. E., and Garlick, J. A. (2004)Basement membrane proteins promote progression of intraepi-thelial neoplasia in 3-dimensional models of human stratifiedepithelium. Int. J. Cancer 108, 348–357

47. Carroll, J. M., and Moles, J. P. (2000) A three-dimensional skinculture model for mouse keratinocytes: application to trans-genic mouse keratinocytes. Exp. Dermatol. 9, 20–24

48. Maas-Szabowski, N., Szabowski, A., Stark, H. J., Andrecht, S.,Kolbus, A., Schorpp-Kistner, M., Angel, P., and Fusenig, N. E.(2001) Organotypic cocultures with genetically modified mousefibroblasts as a tool to dissect molecular mechanisms regulatingkeratinocyte growth and differentiation. J. Invest. Dermatol. 116,816–820

49. Kartasova, T., Scandurro, A. B., Denning, M. F., Wirth, P. J.,Yuspa, S. H., and Lichti, U. (1995) Factors mediating theinteractions between epidermal and dermal cells in skin graftsthat might be important for hair follicle development. J. Invest.Dermatol. 104, 21S-22S

50. Maas-Szabowski, N., Starker, A., and Fusenig, N. E. (2003)Epidermal tissue regeneration and stromal interaction in Ha-CaT cells is initiated by TGF-alpha. J. Cell Sci. 116, 2937–2948

51. Ramos, C. A., Venezia, T. A., Camargo, F. A., and Goodell, M. A.(2003) Techniques for the study of adult stem cells: be fruitfuland multiply. BioTechniques 34, 572–578, 580–574, 586–591

52. Morris, R. J., Fischer, S. M., Klein-Szanto, A. J., and Slaga, T. J.(1990) Subpopulations of primary adult murine epidermalbasal cells sedimented on density gradients. Cell Tissue Kinet. 23,587–602

53. Medina, R., Kataoka, K., Takaishi, M., Miyazaki, M., and Huh, N.(2006) Isolation of epithelial stem cells from dermis by athree-dimensional culture system. J. Cell. Biochem. 98, 174–184

54. Rahimi, F., Hsu, K., Endoh, Y., and Geczy, C. (2005) FGF-2, IL-1beta and TGF-beta regulate fibroblast expression of S100A8.FEBS J. 272, 2811–2827

55. Ceradini, D. J., and Gurtner, G. C. (2005) Homing to hypoxia:HIF-1 as a mediator of progenitor cell recruitment to injuredtissue. Trends Cardiovasc. Med. 15, 57–63

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

57. Troy, T. C., and Turksen, K. (2005) Commitment of embryonicstem cells to an epidermal cell fate and differentiation in vitro.Dev. Dyn. 232, 293–300

58. McGowan, K. M., and Coulombe, P. A. (1998) Onset of keratin17 expression coincides with the definition of major epitheliallineages during skin development. J. Cell Biol. 143, 469–486

59. McGowan, K. M., Tong, X., Colucci-Guyon, E., Langa, F.,Babinet, C., and Coulombe, P. A. (2002) Keratin 17 null miceexhibit age- and strain-dependent alopecia. Genes Dev. 16,1412–1422


60. Ng, W. A., Doetschman, T., Robbins, J., and Lessard, J. L. (1997)Muscle isoactin expression during in vitro differentiation ofmurine embryonic stem cells. Pediatr. Res. 41, 285–292

61. Wang, D., Park, J. S., Chu, J. S., Krakowski, A., Luo, K., Chen,D. J., and Li, S. (2004) Proteomic profiling of bone marrowmesenchymal stem cells upon transforming growth factor {beta}1stimulation. J. Biol. Chem. 279, 43725–43734

62. Mori, L., Bellini, A., Stacey, M. A., Schmidt, M., and Mattoli, S.(2005) Fibrocytes contribute to the myofibroblast population inwounded skin and originate from the bone marrow. Exp. CellRes. 304, 81–90

63. Amoh, Y., Li, L., Katsuoka, K., Penman, S., and Hoffman, R. M.(2005) Multipotent nestin-positive, keratin-negative hair-follicle

bulge stem cells can form neurons. Proc. Natl. Acad. Sci. U. S. A.102, 5530–5534

64. Reddy, S. T., Andl, T., Lu, M. M., Morrisey, E. E., and Millar,S. E. (2004) Expression of Frizzled genes in developing andpostnatal hair follicles. J. Invest. Dermatol. 123, 275–282

65. Dunn, K. J., Brady, M., Ochsenbauer-Jambor, C., Snyder, S.,Incao, A., and Pavan, W. J. (2005) WNT1 and WNT3a promoteexpansion of melanocytes through distinct modes of action.Pigment Cell Res. 18, 167–180

Received for publication February 1, 2006.Accepted for publication February 6, 2007.


isolation of a mesenchymal cell population from murine ...the faseb journal • research...

Documents