SEARCH

SEARCH BY CITATION

Keywords:

  • Keratinocyte;
  • Stem cell;
  • Collagen;
  • Epidermis;
  • Colony forming

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

A prevalent belief in epidermal biology is that stem cells are highly clonogenic; that is, they have the ability to produce many large colonies in vitro. However, it has been well-established in hematology, and recently suggested in epithelial biology, that short-term in vitro clonogenic assays may not be reliable predictors of long-term in vivo repopulating ability. Numerous groups have shown that rapid adhesion to collagen selects for highly clonogenic keratinocytes, but it has not been demonstrated whether this subpopulation is enriched in stem cells as defined by long-term repopulating ability in vivo. We found that although rapid adhesion to collagen (within 5 minutes) selected for cells with increased short-term colony forming ability in vitro, these cells were not enriched in long-term proliferative ability in vitro or in repopulating ability in vivo after 9 weeks. Conversely, keratinocytes that did not adhere to collagen (after 20 minutes) were less clonogenic in short-term assays but possessed equivalent long-term proliferative ability in vitro and superior long-term repopulating ability in vivo. Both the rapidly adherent cell and not rapidly adherent cell populations contained small, noncomplex basaloid cells, expressed integrin α2 (a collagen IV receptor), and expressed the putative epidermal stem cell phenotype integrin α6hiCD71lo. Our results indicate that the superior short-term colony forming ability of collagen-adherent murine keratinocytes does not correlate with long-term repopulating ability in vitro or in vivo and that proliferation in vitro is not a reliable surrogate for stem cell behavior in vivo.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

In mammals, epidermal stem cells (EpSCs) are responsible for tissue homeostasis and repair and have clinical importance in burn therapy, chronic ulcers, blindness, and hair loss. A commonly used assessment of the isolation and enrichment of EpSCs is the clonogenicity, or colony-forming efficiency (CFE) in vitro, of a given cell population [1, [2], [3], [4]5]. However, the power of in vitro findings to predict and describe the in vivo behavior of EpSCs has been questioned recently [6, [7], [8]9]. These studies have cautioned against defining stem cells based solely on their proliferative behavior in culture because of observations that indicate that stem cell populations may lack the appropriate stimuli for growth in vitro [7, 8]. For example, analysis of a stem cell population of murine keratinocytes, isolated on the basis of their ability to efflux Hoechst dye, revealed that although these cells proliferate relatively poorly in vitro, they are capable of sustained tissue regeneration in vivo [8, 10]. Such studies suggest that surrogate in vitro analyses to assess putative stem cell properties, such as clonogenicity, long-term proliferative output, and short-term tissue reconstitution, may not accurately distinguish between epidermal stem and progenitor cells [7].

Numerous findings indicate that a large number of keratinocytes capable of forming colonies in vitro are not EpSCs. The CFE of unsorted murine keratinocytes (using various media supplements, presence or absence of feeder layers, and different strains of mice) ranges from 2.5% to 8.4% [11, [12], [13], [14]15], whereas for human keratinocytes, depending on their derivation and culture conditions, CFE values range from 12% to 25% [16, 17]. In comparison, the estimates of EpSC frequency in vivo indicate that they make up between 0.01% and 10% of the basal layer, depending on the methodology used [7]. These data suggest that EpSCs may be a subset of the colony-forming cells or a different set of cells altogether. Thus, to address methodological discrepancies and accurately identify and describe the true nature of EpSCs, a direct comparison between cell behavior in vitro and long-term repopulating ability in vivo is necessary.

As collagens are a major component of the basement membrane, where EpSCs reside, rapid adhesion to these proteins has long been used as a selection step [11, 17, [18], [19], [20], [21], [22]23]. Previous work has shown that murine keratinocytes that retain DNA labels such as 5-bromo-2′-deoxyuridine (indicating a slow division rate characteristic of stem cells) rapidly adhere to many extracellular matrix proteins, including collagen IV [11]. These rapidly adherent and label-retaining cells form large colonies and produce 1000-fold more cells in vitro compared with the total basal keratinocyte population [11]. In human keratinocytes, rapid adhesion to collagen selects for a clonogenic population [16, 24] that forms a thicker skin equivalent in vitro compared with slowly adherent or nonadherent cells [18, 20]. Furthermore, keratinocytes that express high levels of a laminin receptor, integrin α6, combined with low expression of the proliferation marker CD71, (α6hiCD71lo) have been shown to be slow-cycling [25, [26]27] and highly proliferative in vitro [24, 28]. However, whereas one study showed that rapid collagen adhesion can be used to isolate α6hiCD71lo cells [20], another study found that rapidly adherent keratinocytes express a range of phenotypes, including α6hiCD71lo, α6hiCD71hi, and α6lo [24]. Thus, these studies do not clarify whether rapidly adhering clonogenic keratinocytes are enriched in functional stem cells as defined by long-term repopulating ability in vivo.

In this study, we examined the relationship between clonogenicity and long-term repopulating ability using an in vivo regeneration assay [29, [30]31] alongside both short- and long-term in vitro experiments. We found that rapidly adherent murine keratinocytes were enriched in short-term colony-forming cells but not in long-term repopulating cells. Furthermore, keratinocytes that did not adhere rapidly produced large proliferative colonies [32], were highly proliferative in vitro over a 9-week time period, contained small, basaloid cells expressing the α6hiCD71lo phenotype, and displayed superior long-term repopulating ability in vivo. We conclude that a population of keratinocytes not enriched in short-term clonogenicity in vitro can still possess significant long-term repopulating ability in vivo.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Animals

C57BL/6-TgN(ACTbEGFP)1Osb mice (Jackson Laboratory, Bar Harbor, ME, http://www.jax.org) were used as the source of green fluorescent protein (GFP)-positive keratinocytes. These mice ubiquitously express GFP under control of the chicken β-actin promoter. GFP-negative siblings were used as a source of GFP-negative keratinocytes and dermal fibroblasts. Nonobese diabetic severe combined immunodeficiency (NODSCID) mice (Jackson Laboratory) served as the silicone chamber transplant recipients. Mice were housed in our animal facility and given food and water ad libitum, and all experiments were performed in accordance with an approved Veterans Affairs Institutional Animal Care and Use Committee protocol.

Keratinocyte and Fibroblast Isolation

Primary neonatal murine keratinocytes and dermal fibroblasts were isolated from day 4 C57BL/6 GFP-positive and GFP-negative pups as follows. Briefly, dorsal skin (approximately 2 cm2 in area) was excised and the subcutaneous fat was removed. Skins were incubated in dispase (25 U/ml in Hanks' balanced saline solution; BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) for 24 hours at 4°C or 2.5 hours at 37°C. Following dispase treatment, the dermis was examined histologically to ensure that both the interfollicular and follicular epidermis had been removed. To isolate keratinocytes, the epidermis was removed with forceps and placed in prewarmed (37°C) 0.05% trypsin-EDTA (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 15 minutes and then neutralized with chelexed fetal bovine serum (FBS) (Atlas Biologicals, Boulder, CO). Keratinocytes were separated from the stratum corneum and supernatant by gentle centrifugation at 500 rpm for 10 minutes. We routinely recovered 3.5–4 million cells per dorsal skin, and on average, we used 4–8 pups per experiment. Primary fibroblasts were isolated by further incubating the dermis in 0.25% mg/ml collagenase type IA (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) at 37°C for 1 hour. Cell viability was assessed by trypan blue exclusion.

Collagen IV Adhesion

Tissue culture dishes (100 mm) were incubated with 10 μg/ml collagen IV (in distilled H2O; American Type Culture Collection, Manassas, VA, http://www.atcc.org) overnight at 4°C. The collagen solution was removed, and the dishes were washed once with cold phosphate-buffered saline (PBS) before being sterilized by UV radiation for 20 minutes. Keratinocytes (3.5 × 106 per 100-mm dish) were added to prewarmed dishes and incubated at 37°C for the indicated time points. Cells were removed from the dish by gentle scraping to avoid possible cell surface antigen stripping due to trypsinization. Use of this method also avoids growth factors present in the serum-containing trypsin neutralizing buffer that could conceivably give cells being trypsinized a growth advantage over nonadhered cells. After scraping, a single-cell suspension was achieved by centrifugation and gentle resuspension in CNT07. On average, 90%–95% of the cells were recovered and viable. Rapidly adherent cells (RACs) were defined as keratinocytes that adhered within 5 minutes; not rapidly adherent cells (NACs) were defined as keratinocytes that did not adhere after 20 minutes.

Colony-Forming Efficiency Assays

Freshly isolated keratinocytes were seeded at clonal density (100–500 cells per cm2) into tissue culture dishes (either 60-mm dishes or 35-mm six-well plates) coated with Matrigel (0.5 mg/ml; BD Biosciences), a basement membrane extract, and cultured for 2 or 4 weeks in Progenitor Cell Technology Epidermal Keratinocyte Medium Complete (CNT07 with supplements; Chemicon, Temecula, CA, http://www.chemicon.com) at 37°C with 5% CO2. We found Matrigel-coated tissue culture plastic to be a good substitute for irradiated fibroblast layers that were not able to grow in the CNT07 medium. The medium was changed every 2–3 days until the cells were fixed with 10% neutral buffered formalin (Sigma-Aldrich) and stained with toluidine blue (Sigma-Aldrich). CFE was expressed as the percentage of colonies 2 mm2 or more divided by the number of cells seeded.

Long-Term In Vitro Proliferation

Freshly isolated keratinocytes were cultured in six-well tissue culture plates coated with Matrigel (0.5 mg/ml; BD Biosciences) and incubated at 37°C in 5% CO2 in CNT07 medium (Chemicon). Cultures were initiated with selected populations of keratinocytes plated at 500 cells per cm2 (clonal density) and 3,000 cells per cm2 (high density). The medium was changed every 2–3 days, and the cultures were passaged when confluent. Keratinocytes were detached with 0.05% trypsin-EDTA (Invitrogen), counted, replated, and continually passaged. Cumulative cell outputs were calculated on the basis of the assumption that all cells from previous passages had been replated. For each condition, experiments were performed in duplicate.

Chamber Implantation, Imaging, and Limiting Dilution Analysis

Chambers were implanted onto the backs of 6–8-week-old host NODSCID mice as previously described [29]. Briefly, after the mice were anesthetized, a 0.3-cm2 area of skin on the back was excised down to the fascia. The chamber (6-mm internal diameter; Renner GmbH, Germany, http://www.renner-gmbh.de), shaped like a top hat with the top cut off, was implanted and sutured into place. Coban (3M, St. Paul, MN, http://www.3m.com) wrap was placed around the chamber and body of the mouse, the cells were added (fibroblasts were added first and 5–10 minutes later the keratinocytes were added), and a breathable dressing, Tegaderm (3M), was placed over the chamber for 2–3 weeks. Images were taken of the chambers at 3, 5, 7, and 9 weeks post-chamber implantation using an epifluorescent stereomicroscope (Stemi SV; Carl Zeiss, Inc., Thornwood, NY, http://www.micro@zeiss.com). Bright-field images were captured at 200 ms and fluorescent images at 900 ms with a UV 488 nm filter (Carl Zeiss, Inc.). For limiting dilution analysis, the chamber was scored as positive if at least one GFP-positive epidermal cell cluster was detected.

Fluorescence-Activated Cell Sorting Analysis

Monoclonal antibodies against integrin α6 (CD49f, fluorescein isothiocyanate [FITC]-conjugated, GoH3 clone), CD71 (phycoerythrin [PE]-conjugated, C2F2 clone), integrin α2 (CD49b, FITC-conjugated, Ha1/29 clone), and isotype control antibodies were purchased from BD Pharmingen (San Diego, http://www.bdbiosciences.com/index_us.shtml). Unsorted, RAC, and NAC subpopulations were isolated and resuspended at a concentration of 1 × 106 cells per milliliter in PBS with 2% FBS (Atlas Biologicals). Cells (1 × 106 per sample) were incubated on ice with either FITC-α2 integrin or a combination of FITC-α6 integrin and PE-CD71 for 30 minutes. After being washed two times, cells were resuspended in PBS with 2% FBS and 0.25 μg of 7-aminoactinomycin D (BD Biosciences) to exclude nonviable cells and then analyzed on a FACScan system (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) using CellQuest software. All experiments were performed in triplicate using the appropriate isotype controls.

As one intent of this study was to extend prior findings, we chose to define keratinocyte subsets using previously reported fluorescence-activated cell sorting (FACS) characteristics as follows. Basal cells, which make up 45% of total murine keratinocytes [29], are cells with the smallest cell size (low forward-scatter height) and least cellular complexity (low side-scatter height) [17]. A gate (P1) consistent with this distribution was first placed on a population of unsorted keratinocytes and then used to evaluate the relative distributions of basal cells and large, differentiating cells within the RAC and NAC subpopulations (Fig. 4). Differences in integrin α6 and CD71 antigen expression in basal cells of the RAC and NAC subpopulations were assessed as described in previous reports [8, 20]. In these previous studies, EpSCs were defined as the 7%–10% of α6 integrinhi cells with the lowest CD71 expression (α6 integrinhiCD71lo), and transit-amplifying (TA) cells were defined as the 7%–10% of α6 integrinhi cells with the highest CD71 expression (α6 integrinhiCD71hi). Basal cell subsets from both the RAC and NAC populations were then evaluated using these established gates. Finally, differences in expression of integrin α2 between the RAC and NAC subpopulations were measured by FITC mean fluorescent intensity.

Statistical Analysis

For comparisons of CFE, cumulative cell output and FACS analysis in different subpopulations, a Student t test was used. L-CALC (Stemsoft, Vancouver, BC, Canada, http://www.stemsoft.com) was used to determine the frequency of repopulating units in each subpopulation at all time points. The limiting dilution analysis was then subjected to a χ2 test. Lack of a significant p value in the χ2 test was used to demonstrate internal consistency in the distribution of results for each dilution.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The Difference in Clonogenic Ability Between Rapidly Adherent and Not Rapidly Adherent Murine Keratinocytes Decreases over Time

A fraction of primary neonatal murine keratinocytes rapidly adhere to collagen IV [11, 16, 24]. We aimed to select the most adhesive keratinocytes by using a low concentration of collagen (10 μg/ml) and short incubation time (5 minutes). After 2 minutes, 10.5% (± 0.7%) of the cells plated had adhered. Between 0 and 5 minutes, 17.5% (± 0.5%) of the cells adhered (RACs), and between 5 and 20 minutes, 29.2% (± 1.2%) of the cells adhered (Fig. 1A, 1B). After 20 minutes, 45.8% (± 2.2%) of the cells were still not adherent (NACs). These results were highly reproducible in more than 50 experiments and are similar to previously published results [11].

thumbnail image

Figure Figure 1.. The difference in clonogenic ability between RACs and NACs decreases over time. (A): Keratinocyte adherence to collagen IV-coated dishes (10 μg/ml) over time. After 5 minutes, 17.5% (± 0.5%) of the cells adhered, and after 20 minutes, 46.7% (± 1.2%) of the cells adhered. After 20 minutes, 45.8% (± 2.2%) of the cells were not adherent. (n = 17; on average, 93% of the cells were recovered in each experiment.) (B): Relative concentration of RACs (17.5%) and NACs (45.8%) isolated from the total keratinocyte population. (C): Relative CFE of RACs and NACs after 2 weeks in culture as compared with the unsorted population (defined as = 1.0). RACs had a fivefold greater CFE compared with NACs at 2 weeks (p < .001; n = 4). (D): Decreased difference in CFE between RACs and NACs evident at 4 weeks, at which time RACs had only a twofold greater CFE compared with NACs (p < .05; n = 3). Error bars show the SEM. Abbreviation: CFE, colony-forming efficiency.

Download figure to PowerPoint

Previous studies of clonogenic potential in murine keratinocytes have compared rapidly adherent cells using a variety of extracellular matrices or compared rapidly adherent cells to the unsorted population [11, 17]. Here, we compared the clonogenic potential of the rapidly adherent population (RACs) to the less adherent population (NACs). As shown in Figure 1C, RACs had a significantly greater (fivefold) CFE (p < .001) than NACs when cultured for 2 weeks. However, this degree of enrichment decreased when the cells were cultured for longer time periods. After 4 weeks in culture, RACs had only a twofold greater CFE compared with NACs (p < .05; Fig. 1D). Colony forming assays performed for time periods less than 4 weeks missed slowly developing but highly proliferative colonies, primarily in the NAC population.

The Long-Term Proliferative Ability of Rapidly Adherent and Not Rapidly Adherent Murine Keratinocytes Is Similar In Vitro

Despite the differences in colony number, after 4 weeks both RACs and NACs formed large proliferative colonies [32], as well as small differentiating colonies, and they did so in similar proportions (Fig. 2A, 2B). It is noteworthy that NACs consistently produced very densely cellular colonies that were more darkly stained than colonies of similar diameter in the unsorted and RAC dishes that appeared to have fewer cells. Furthermore, when cultured at either low or high density and passaged at confluence, the cumulative cell output of NACs was similar to that of RACs over short time periods (2 and 4 weeks), as well as over a longer time period (9 weeks) (p > .05; Fig. 2C). This effect of culture time on CFE enrichment has been noted before in murine keratinocytes [4]. In an additional attempt to reveal any differences in long-term in vitro growth potential between RACs and NACs, we performed experiments in which cultures were allowed to become confluent and the medium was continually changed but the cultures were not passaged. Although this method is limited because it may reflect a combination of cell survival, senescence, and proliferation, both RAC and NAC cultures remained confluent for more than 6 months (data not shown). Thus, despite differences between the RAC and NAC populations in colony forming ability in short-term in vitro experiments, the NAC population contains significant proliferative potential in long-term in vitro experiments.

thumbnail image

Figure Figure 2.. Long-term proliferative ability of RACs and NACs is similar in vitro. (A): Image of colonies produced by unsorted, RAC, and NAC populations after 4 wks from one representative experiment. (B): Scatter plot showing the distribution of colony sizes produced by unsorted, RAC, and NAC populations after 4 wks. Each triangle represents a single colony from one representative experiment. Colony sizes were binned into five groups: 2–3 mm2, 3–4 mm2, 4–6 mm2, 6–8 mm2, and >8 mm2. (C): The cumulative cell output of unsorted, RAC, and NAC populations at 2, 4, and 9 wks. Cumulative cell output is expressed as the total number of cells produced (at that time point) relative to that of the unsorted population. (n = 3; error bars show SEM). Abbreviations: NAC, not rapidly adherent cell; RAC, rapidly adherent cell; wks, weeks.

Download figure to PowerPoint

Limiting Dilution Analysis Demonstrates That Keratinocytes That Do Not Rapidly Adhere to Collagen Have Significant Long-Term Repopulating Ability In Vivo

To directly compare short- and long-term clonogenic and proliferative ability in vitro with long-term repopulating ability in vivo, we used a quantitative epidermal regeneration model. This method involved the use of dissociated keratinocytes allowed to re-form a cornified, stratified epithelium on top of dermal fibroblasts seeded onto the subcutaneous fascia of immunodeficient mice. By seeding progressively lower numbers of GFP-positive keratinocytes in this repopulating assay, limiting dilution analysis quantifies the frequency of cells with long-term repopulating ability in a given population [29]. As shown in Figure 3 and Table 1, implantation of GFP-positive keratinocytes (7,500–240,000) along with 1 million GFP-negative keratinocytes permitted the visualization of GFP-positive repopulation units within an epidermis of GFP-negative keratinocytes. Whereas GFP-positive units presumably derived from transit-amplifying cells disappeared after 5 to 7 weeks (Fig. 3A, 3C), GFP-positive repopulation units derived from stem cells (as defined by long-term repopulating ability) persisted for the duration of the study (9 to 30 weeks) (Fig. 3A, 3D). The regenerated epidermis displayed a normal structure when viewed either from above (Fig. 3C, 3D) or in cross-section (Fig. 3E).

thumbnail image

Figure Figure 3.. Limiting dilution analysis demonstrates that NACs have significant long-term repopulating ability in vivo. (A): Schematic of epidermal layers in which either GFP-positive units derived from TA cells disappeared after 5 to 7 wks (top row) or GFP-positive cells derived from stem cells persisted for the duration of the study (9 to 30 wks) (bottom row). (B): Diagram of the in vivo experimental model using a 6-mm chamber into which, first, dermal fibroblasts were seeded and then a mixture of GFP-positive and GFP-negative keratinocytes was added. (C): Representative example of the loss of GFP-positive repopulating units by 7 wks. Top panels show bright-field images of regenerated skin grown from rapidly adherent keratinocytes, and bottom panels show epifluorescent images taken at the same time as the corresponding bright-field images above them. (D): Representative example of the maintenance of GFP-positive repopulating units up to 14 wks. Top panels show bright-field images of regenerated skin grown from not rapidly adherent keratinocytes, middle panels show epifluorescent images taken at the same time as the corresponding bright-field images above them, and bottom panels show an overlay of the GFP-positive repopulating units on the regenerated skin. Orientation, magnification (×1), and exposure time were the same in each image; the chamber was 6 mm in diameter. (E): Hematoxylin and eosin staining of cross-section of regenerated skin. Magnification, ×10 (upper panel) and ×20 (lower panel). (F): Frequency of repopulating cells in the unsorted, RAC, and NAC populations as determined by limiting dilution analysis at 3, 5, 7, and 9 wks. The frequency of repopulating cells in all populations decreased over time and reached a plateau by 7 wks. *, The frequency of long-term repopulating cells in the RAC population was significantly lower than that in the unsorted population at all time points (p < .05). The frequency of long-term repopulating cells in the NAC population was not statistically different from that of the unsorted population at all time points. Error bars show the SE. Abbreviations: d, dermis; e, epidermis; GFP, green fluorescent protein; hf, hair follicle; NAC, not rapidly adherent cell; RAC, rapidly adherent cell; TA, transit-amplifying; wks, weeks.

Download figure to PowerPoint

Table Table 1.. Frequency of repopulating units in RACs, NACs, and unsorted murine keratinocytes over time
Thumbnail image of

In these studies, the shortest amount of time that chambers were kept was 9 weeks; most chambers were assessed up to 14 weeks and some as long as 30 weeks. In accordance with previous findings [29], the frequency of long-term repopulating cells in the unsorted population at 9 weeks was 1 in 33,000 (95% confidence limits, 1 in 14,000 to 1 in 75,000), and positive repopulating units that remained after 7 weeks persisted indefinitely (Fig. 3F; Table 1). At 3 weeks, the frequency of repopulating cells in the NAC population, 1 in 21,000 (95% confidence limits, 1 in 12,000 to 1 in 37,000), was approximately twice that of the RAC population, 1 in 48,000 (95% confidence limits, 1 in 30,000 to 1 in 76,000) (Fig. 3F; Table 1). The number of GFP-positive repopulating units in each population declined at 5 and 7 weeks as GFP-positive units were lost, presumably due to differentiation of more short-term repopulating units. Loss of GFP-positive repopulating units was never seen after 7 weeks, even in chambers assessed up to 30 weeks. At no point was the value of the χ2 test statistically significant (Table 1), demonstrating internal consistency in our assay. By 9 weeks, the frequency of repopulating cells in the NAC population, 1 in 36,000 (95% confidence limits, 1 in 22,000 to 1 in 60,000), was more than twice that of the RAC population, 1 in 82,000 (95% confidence limits, 1 in 51,000 to 1 in 131,000) (Fig. 3E; Table 1). Thus, the NAC population contained significantly more (p < .05) repopulating cells than the more clonogenic RAC population at all time points (Table 1).

Interestingly, although keratinocytes that adhered to collagen between 5 and 20 minutes (slowly adherent cells [SACs]) had an intermediate CFE between that of RACs and that of NACs, initial studies showed that these cells were significantly depleted of long-term repopulating ability (data not shown), and thus they were not included in further analyses. To confirm that our separation procedure did not alter repopulating ability, we determined that the frequency of the recombined subpopulations (20% RACs + 30% SACs + 50% NACs) was 1 in 43,000 (95% confidence limits, 1 in 21,000–1 in 89,000) after 9 weeks and not significantly different (p > .05) from unsorted (data not shown). These data show that long-term repopulating cells are less common than short-term repopulating cells. Also, keratinocytes that do not rapidly adhere to collagen contain significant long-term repopulating ability and are capable of sustained tissue regeneration over extended time periods (at least 6 months). Furthermore, rapid adhesion to collagen actually selects against long-term repopulating ability.

The Rapidly Adherent and Not Rapidly Adherent Populations Both Contain Keratinocytes with an Epidermal Stem Cell Phenotype

To determine whether the NAC population contained cells with stem-like phenotypes, we first performed FACS analysis to measure the relative frequencies of basal and differentiated cells in the RAC and NAC keratinocyte subpopulations. Figure 4A demonstrates the gate (P1) defining the smallest, least complex 45% of unsorted keratinocytes [29]. Both RACs and NACs have similar frequencies of small, basaloid cells and large, differentiating cells (p > .05; n = 5) (Fig. 4B). On the basis of previous studies [8, 20], unsorted basaloid cells were used to define gates around the α6 integrinhiCD71hi keratinocytes (gate P2) and α6 integrinhiCD71lo keratinocytes (gate P3) (Fig. 4C). Although the RAC subpopulation contains significantly more α6 integrinhiCD71lo keratinocytes (p < .05; n = 3) than the NAC subpopulation, both subpopulations have a small fraction (RACs, 6.3%; NACs, 4.5%) of cells with this phenotype (Fig. 4D). Interestingly, we did not detect a difference in the expression of integrin α2, a collagen IV receptor, between the RAC and NAC subpopulations as measured by mean fluorescence intensity (p > .05; Fig. 4E), indicating that murine keratinocytes must use a diverse repertoire of collagen receptors.

thumbnail image

Figure Figure 4.. The RAC and NAC populations both contain cells with an epidermal stem cell phenotype. (A): Cell size and complexity of unsorted keratinocytes with gate P1 differentiating the 45% of small noncomplex cells characteristic of basal cells from the remaining large differentiating cells. (B): Percentages of small, basaloid and large, differentiating cells in the unsorted, RAC, and NAC populations; the percentages were similar in all three groups (p > .05; n = 5). (C): Surface expression of integrin α6 and CD71 on unsorted basaloid keratinocytes, with gates P3 and P2 containing the putative epidermal stem cell (α6 integrinhiCD71lo) and putative transit-amplifying (α6 integrinhiCD71hi) populations, respectively. (D): Prevalence of α6 integrinhiCD71lo and α6 integrinhiCD71hi in unsorted, RAC, and NAC basaloid cells demonstrating that, although it is not enriched over the unsorted population, the RAC population has approximately 30% more α6 integrinhiCD71lo expressing cells than the NAC population (p < .05). (E): Integrin α2 on RAC and NAC cells as measured by mean fluorescence intensity was similar (p > .05; n = 3). Abbreviations: FSC, forward scatter; NAC, not rapidly adherent cell; RAC, rapidly adherent cell; SSC, side scatter.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

We have developed and used a long-term repopulating assay to test for the most rigorous definition of an epidermal stem cell, sustained tissue regeneration and maintenance in vivo. Using this in vivo assay combined with short- and long-term in vitro experiments, we have shown that keratinocytes that are less clonogenic in short-term assays and nonadherent to collagen after 20 minutes contain substantial proliferative potential in vitro, as well as significant long-term repopulating ability in vivo, over extended time periods (9 to 30 weeks). Although rapidly adherent cells (RACs) are enriched in short-term colony-forming cells in vitro, this population contains significantly fewer long-term repopulating cells than the NAC population in vivo. Furthermore, both RAC and NAC populations have similar long-term proliferative potential in vitro. Thus, these studies indicate that short-term clonogenic experiments are not a reliable predictor of long-term expansion potential in vitro or repopulating ability in vivo.

The value of an in vivo assessment of EpSC function is well recognized [6, 9, 28]. Kolodka et al. showed that only clonogenic human keratinocytes that have long-term growth potential in vitro could persist when transplanted in vivo [6]. Realizing that the colony-forming cell (CFC) may not truly represent the stem cell, there has been much recent effort to use alternative enrichment procedures to select for EpSCs. To determine the most effective stem cell enrichment strategy, it is necessary to know the relative frequency of stem cells in the enriched population compared with the frequency in the total (unsorted) population. For example, the relative enrichment of a sorted population (marker-positive) compared with the remaining cells (marker-negative) may be 100-fold, but only 3-fold over the absolute frequency of the unsorted population. Since the remaining cell population is variable, depending on each isolation strategy, comparison among strategies is not possible without first establishing a baseline standard frequency. Thus, using the quantitative in vivo regeneration assay described in this study, we have established a foundation from which to compare the effectiveness of multiple EpSC enrichment strategies.

In the epidermis, Schneider et al. first showed quantitatively that 1 murine keratinocyte in 30,000 is a long-term (more than 6 weeks) repopulating stem cell in vivo [29], and in the present study, the finding that 1 unsorted murine keratinocyte in 33,000 is a long-term repopulating cell validates the baseline frequency in this model. Consistent with this previous work, we have also confirmed that the amount of time necessary for short- and intermediate-term repopulating units in mouse skin to be depleted is 6 to 7 weeks, by which time the frequency of repopulating units has declined to a plateau and no longer decreases. It should be noted that by applying limiting dilution analysis at serial time points, transit-amplifying markers could be differentiated from stem cell markers. However, as this assay selects for cells that engraft best, it is not known whether these cells behave similarly to stem cells in normal tissue. Furthermore, it is possible that not all stem cells engraft in this system. Despite these limitations, the use of quantitative in vivo assays has become de rigueur in other systems, such as hematology and mammary epithelial biology, to establish a primitive cell hierarchy [33, 34]. Furthermore, quantitative in vivo assays have also been used to assess stem cell differentiation and regulation [35, [36]37] and for determining the phenotype of cancer stem cells [38, [39]40].

Variations in keratinocyte origin and methodology in in vitro studies produce a range of results that make interpretation of colony forming assays difficult and establishment of a standard CFE not feasible. Collagen adhesion has often been used to select for cell populations with a high CFE. However, rapidly adherent keratinocytes, slowly adherent keratinocytes, and nonadherent keratinocytes have been selected for adherence to different concentrations of collagen (10 to 100 μg/ml), at various temperatures (37°C, 25°C, and 4°C), over a wide range of time points (5 minutes to 24 hours), and with keratinocytes from different origins (human and mouse). In this study, we used a very low concentration of collagen to select for highly adhesive murine keratinocytes, and hence, our NAC population may be comparable to what others have called slowly adherent cells (SACs). In addition, after selection on collagen the keratinocytes were cultured on Matrigel for the in vitro analyses. A previous study of murine epidermal stem cells has shown that label-retaining cells adhere well to Matrigel [11], and relative CFE values were identical in our pilot study regardless of the culture system (irradiated fibroblast feeder layers vs. Matrigel coating). However, the possibility that variations in the culture system could select different populations of keratinocytes cannot be ruled out. It is also interesting to note that the experimental time frame appears to be highly relevant. The fivefold difference in colony-forming ability between our RAC and NAC populations decreased to twofold after 2 additional weeks. Similarly, Trempus et al. have shown differences in the CFE between two murine keratinocyte populations (CD34+ and CD34−) at 2 versus 4 weeks [4]. These results reaffirm previous findings in hematopoietic stem cell biology. A longstanding premise in murine hematopoiesis is that early hematopoietic progenitors, such as high-proliferative potential colony-forming cells [41] and burst-forming units-erythroid [42], cycle less rapidly than later progenitors, such as granulocyte monocyte progenitors and colony-forming units-erythroid, but have a greater proliferative and self-renewal capacity than the latter [43]. As a consequence, colonies from these earlier progenitors take longer to form (10–14 days as compared with 4–7 days), but the colonies are much larger than those formed by late progenitors. In addition, although previous work demonstrated that murine RACs could be subcultured up to 12 times, whereas total basal cells could only be subcultured three times [11], we were able to subculture unsorted murine keratinocytes, RACs, and NACs up to 30 times, reflecting significant differences between culture systems. Thus, colony forming assays produce variable results depending on both the timeframe and culture conditions used in the experiment.

Determining how various in vitro assays relate to stem cell behavior in vivo is complex. Early in vitro studies determined that large proliferative and self-renewing colonies derived from human keratinocytes may represent stem cells [32]. More recent studies have suggested, however, that large proliferative colonies derived from murine keratinocytes may represent both the stem and TA pools, as determined by cell surface phenotypes [7]. In agreement, in our experiments, both the RAC and NAC populations formed large proliferative colonies in vitro but differed in their long-term repopulating ability in vivo. Moreover, as Jones et al. [16] and Fortunel et al. [18] have discussed, a process of autoregulation may adjust the frequency of primitive cells independently of their initial frequency in the culture. The possibility that passaging may alter primitive cell frequency in vitro may explain why both the RAC and NAC populations produced equal numbers of cells at all time points and yet produced different numbers of colonies at early time points (by 2 weeks the cells in Fig. 2C had been passaged 1–2 times). As many have noted before, since short-term formation of large colonies may not be a unique stem cell indicator and long-term subculturing assays may “reset” positional information that could affect stem cell behavior, in vitro assays may elucidate different cell behaviors and define qualities different from those of murine and human epidermal stem cells in vivo [6, 7, 44, 45].

Much research is inconsistent with the concept that a CFC is a stem cell. The difference in the frequencies of CFCs versus long-term repopulating (LTR) cells in murine skin suggests that LTR cells are a subset of CFCs or a different set altogether. For murine keratinocytes, the CFE ranges from 2.5% to 8%, suggesting that as many as 1 in 10 cells is a stem cell. However, in vivo evidence has shown that EpSC frequency is much lower and more similar to the frequency of stem cells in other tissues; the frequency of LTR cells in murine bone marrow is 1 in 10,000 [46], and in murine mammary epithelia, it is 1 in 4,900 [47]. Because collagen adhesion is a relatively simple isolation step and has been widely used in the literature, we sought to determine the frequency of LTR cells in the RAC population, as well as to directly compare the in vitro and in vivo behavior of these cells.

Given the superior CFE of RACs in vitro, we were surprised that we did not see a much higher frequency of short-term repopulating cells in vivo compared with NACs or unsorted keratinocytes. If the RAC population is enriched in TA cells (short-term repopulating cells) in vivo, these cells may have already been depleted during the initial weeks when the regenerated epidermis is re-forming itself and presumably is in a hyperproliferative state. In our assay, it took 7 weeks before we were able to discern that a GFP-positive repopulating unit was derived from a stem cell rather than a more short-term repopulating cell. Thus, the intermediate-term repopulating units that exhaust their proliferative potential between 3 and 5 weeks, or between 5 and 7 weeks, as was seen in both these and previous studies [29], raise the interesting possibility that there may be a hierarchy of short-term and intermediate-term repopulating cells in the TA population similar to that established in hematopoietic cells.

The high frequency and highly proliferative nature of CFCs, as well as the lack of correlation with in vivo studies of functional EpSCs, strongly suggests that CFCs from murine epidermis may comprise a mixture of TA and stem cells, or even TA cells alone. This paradigm would fit most closely with knowledge from hematology where it has been shown that short-term colony assays are not adequate for the detection of the most primitive progenitors [48]. On the basis of previous EpSC studies [11, 17], we were surprised to find that the NAC, more than the RAC, population contained significant repopulating ability. However, this finding correlates well with good long-term proliferation in vitro, the presence of significant numbers of small noncomplex cells, and the presence of α6 integrinhiCD71lo cells in this less adherent population. Thus, in this study, the superior colony forming ability of collagen-adherent murine keratinocytes did not translate into long-term repopulating ability in vitro or in vivo and was not an adequate surrogate assay for epidermal stem cell behavior.

Disclosure of Potential Conflicts of Interest

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The authors indicate no potential conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

This work was supported by NIH/National Institute on Arthritis and Musculoskeletal Diseases and Skin Grant R01-AR020786 (to R.G.) and a Department of Veterans Affairs Merit Review Program Award (to R.G.). We thank Jim Boonyaratanakornkit, M.D., for helpful advice in preparing the manuscript and Lili Yue, M.S., for excellent technical assistance.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References