| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Cutaneous Biology Research Center, Massachusetts General Hospital and Harvard Medical School, Charlestown, Massachusetts, USA; and
Department of Molecular Genetics and Microbiology, University of Florida College of Medicine, Gainesville, Florida, USA
1Correspondence: Cutaneous Biology Research Center, Massachusetts General Hospital, 149 13th St., Charlestown, MA 02129, USA. E-mail: bruce.morgan{at}cbrc2.mgh.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: infundibulum • epidermis • follicular bulge • wound healing
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. This model posits two distinct populations of keratinocytes in the basal layer of the epidermis. The first is an epidermal stem cell population with extensive capacity for self-renewal. These stem cells generate a "transient amplifying" population with more limited proliferative capacity. The TA cells can regenerate themselves and also give rise to the more differentiated keratinocytes that are displaced from the basement membrane and form the superficial layers of the epidermis. The lineage unit of a stem cell, surrounding TA cells in the basal layer, and differentiated cells derived from them in the superficial layers is the EPU [reviewed in (4
5
6)
]. Populations with the different proliferating potentials predicted by this model have been isolated from skin and studied extensively in vitro (7
, 8)
. Other evidence also supports the existence of a comparatively quiescent stem cell population distinct from the bulk population of basal cells (9
10
11)
. However, reliable markers that distinguish the postulated epidermal stem cell and the TA cell in intact skin have not been identified and the existence of distinct stem and TA cells in the epidermis has not been substantiated by direct empirical evidence. As such, the defining characteristic of these cells in vivo that remains is their different life spans in the epidermis. Even this distinction is a relative one. It is thought that stem cells persist for the life of the organism while TA cells endure a more limited life span and must be replaced periodically. However, the life span of either remains to be empirically defined.
In contrast to the continuous regeneration of the epidermis, the hair follicle undergoes cycles of active growth, partial degeneration, and quiescence before being reactivated to start a new growth cycle. The follicular epithelium is regenerated from a well-characterized population of stem cells resident in the "bulge" region of the permanent portion of the follicle (12
, 13)
. While it was once thought that these cells might be the stem cells of all epithelial components of the skin and appendages (14)
, recent experiments demonstrated that follicular stem cells are a distinct population from that which maintains the epidermis. This was revealed by experiments that used cre recombinase to activate a reporter gene in the keratinocytes of the follicle. As the reporter remains expressed in all of their progeny, this noninvasive method provides reliable lineage analysis in intact tissue. A mouse line expressing cre recombinase from the Sonic hedgehog locus (Shh) was used to activate the cre-dependent beta galactosidase gene of the R26R reporter system (15
, 16)
. This labeled the keratinocytes in the epidermal placode at the onset of follicle formation and revealed a lineage distinction between follicular and interfollicular fates (17)
. All of the keratinocytes of the pilosebaceous unit, which is composed of the hair follicle and associated sebaceous gland and includes the stem cells of the follicular bulge, were labeled by this approach. Although the follicular stem cells were labeled, the epidermis remained unlabeled for the life of the animal in the absence of trauma (17)
. A similar conclusion that the stem cells of the follicular bulge do not normally contribute to the epidermis was reached when recombinase activity was targeted more specifically to the adult follicular bulge population using a k15;crePR strain (13)
.
When the epidermis is damaged, keratinocytes from the epidermis and follicles surrounding the wound are mobilized to regenerate an epidermal barrier. Keratinocytes invade the wound surface and generate a thickened and hyperproliferative epithelium that then gradually reverts to a more normally organized stratified epidermis (reviewed in (18)
. The participation of follicular keratinocytes in wound healing has been well documented (13
, 14
, 19
, 20)
. Cells normally confined to the follicle are integrated into the regenerating epidermis and express epidermal differentiation markers. However, recent experiments suggest that as the hypertrophic wound epidermis acquires a more normal epidermal structure, cells derived from the follicle are lost (13)
. In the parlance of the EPU model, it was concluded that follicular cells may act as transient amplifying cells but do not acquire the epidermal stem cell character.
Our Shhcre-based analysis of the fate of follicular keratinocytes during wounding was consistent with previous demonstrations that follicular keratinocytes are recruited to regenerate the epidermis after wounding. However, the persistence of labeled cells in the wound epithelium for extended periods seemed inconsistent with the conclusion that follicular cells could not be reprogrammed to become epidermal stem cells in the wound environment. In the work reported here, we evaluate the lineage label technique used in this analysis and confirm that it remains a reliable indicator of follicular origin in the context of wound healing experiments. Neither the Shh gene nor cre recombinase expression is activated in epidermis or wound epithelium during the wound-healing response. We further demonstrate that cells of follicular origin contribute to a long term-repopulating keratinocyte pool resident in the basal layer of the regenerated epidermis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1 cm2 of the epidermis. The wound was disinfected and let to heal in open air. In situ hybridization, ß-galactosidase detection, and immunostaining were performed as described previously (17)
.
ShhGFPcre/R26R wound analysis and scoring
A total of 10 wounds were analyzed at 1 wk (n=1), 2 wk (n=3), 9 wk (n=3), and 16 wk (n=3) in Shhcre experiments. At the end of the healing period, ShhGFPcre/R26R mice were sacrificed and the skins, including the healed wound, were processed for ß-galactosidase activity as described previously (17)
. In brief, dorsal skin was dissected into cold PBS solution and subcoutaneous fat and connecting tissue were removed. Following 20 min fixation in 0.2% gluteraldehyde in PBS at 4 degrees the skin pieces were washed and incubated in ß -galactosidase buffer for 48 h at room temperature. Skins were fixed in 4% paraformaldehyde in PBS, dehydrated in graded sucrose, and mounted into OCT blocks. Sections (7µM) were subjected to immunostaining with keratin 1 (Covance MK1 PRB-165P, 1:2000 dilution; Covance, Berkeley, CA, USA) and keratin 14 antibodies (Covance MK14, PRB-155P, 1:10000 dilution) and 3,3'-diaminobenzidine detection (Vector SK-41007; Vector Laboratories, Burlingame, CA, USA). A total area of 300–400 µM2 of each wound area was analyzed. An average of 6 sections per wound were scored for the percentage of basal blue cells residing within the wound area. The epidermal cells in direct contact with the dermis and flanked by two follicles marking the edge of the wound region were scored. Each section scored was positioned at least 35 µM apart from the neighboring one to ensure random cell counts. Serial sections were scored to evaluate whether groups of labeled cells were separated from follicles on all sides by unlabeled cells.
ShcreERT2/R26R wounding and induction
A total of 19 wounds was analyzed in ShhcreERT2 experiments. Full thickness excisional wounds (4 mm diameter) were generated in ShhcreERT2;R26R skin in the telogen stage of the hair cycle. Depilation was performed on a 1 cm2 region elsewhere on the mouse to generate follicles in the anagen stage of the hair cycle during the period when tamoxifen was injected. Overlapping series of daily i.p injections of 1–2 mg of 4-hydroxytamoxifen (H#7904; Sigma, St. Louis, MO, USA) were performed. These include injections from d1-d5 (n=3), d5-d6 (n=1), d5-d10 (n=4), and d10-d15 (n=2) as well as d2-d11, d4-d11, d7–16, and d9–16 (n=1 each). Additional mice received injections every other day from d1–5 (n=2) and from d6–10 (n=2). Samples were harvested between 2 and 3 wk after wounding. The healed skin was processed for ß-galactosidase activity and scored for labeled keratinocytes in regenerated wound epidermis and in epidermis surrounding it. For all wounds in this sample set, labeling in the matrix region of hair follicles in the anagen phase confirmed the effective activation of cre recombinase by OHT.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 21
, 22)
and data not shown]. In adult skin, Shh expression is confined to keratinocytes in the cycling portion of the hair follicle (Fig. 1
and data not shown). It is not observed in the permanent portion of the hair follicle, including the follicular bulge stem cells, or in interfollicular epidermis. This restriction was the basis of previous lineage analysis that employed cre recombinase inserted into the Shh gene to catalyze the conversion of a reporter gene to an active state in cells that express Shh and their progeny. The transient expression of Shh and cre recombinase in the progenitors of the hair follicle was sufficient to activate a cre-dependent lineage reporter in follicular keratinocytes (17)
. The absence of label in the epidermis of these mice not only demonstrated that the stem cells of the follicular bulge do not repopulate the epidermis, they also confirmed that the Shh locus and the cre recombinase sequences embedded in the gene are not expressed in interfollicular epidermis at levels sufficient to activate the reporter gene in the absence of trauma. However, it remained possible that Shh is induced in epidermal keratinocytes during the wound healing process. This would preclude analysis of the fate of follicular cells during wound healing by the ShhGFPcre;R26R method. To evaluate this possibility, Shh transcripts were examined during wound healing by in situ hybridization. Shh transcripts were not detected in either wound epithelium (Fig. 1A
), or in the interfollicular areas surrounding the biopsy wound (Fig. 1B
). The detection of Shh expression in anagen follicles adjacent to the wound served as a positive control for the sensitivity of the detection reaction (Fig. 1A
, arrow).
|
Despite these observations, levels of Shh expression below the level of detection of in situ hybridization experiments might activate the reporter system in nonfollicular cells. This would compromise the use of this reporter as a guarantor of follicular origin. Mice harboring an inducible form of cre recombinase (creERT2) inserted into the Shh locus (ShhcreERT2) were used to evaluate this possibility (15)
. After a full thickness punch biopsy wound was performed, ShhcreET2;R26R mice were injected with Tamoxifen in overlapping periods spanning 16 d after wound healing to activate the cre recombinase fusion protein. During this period the keratinocytes covered the wound area and the wound epidermis begins the return to a normal epidermal structure. Neither the healed epidermis nor the epidermis surrounding the wound exhibited labeled cells (Fig. 1C, D
). Effective activation of the creERT2 fusion protein was demonstrated by the detection of labeled cells in hair follicles in the anagen stage elsewhere in the skin of the same mouse (Fig. 1E
). Thus, the Shh locus is not activated in interfollicular or wound epidermis during wound healing at levels sufficient to activate the lineage reporter used in these studies. The failure to detect Shh expression or ShhcreERT2 activity in the epidermis or wound epithelium confirms that ß-galactosidase expression is a reliable marker of derivation from follicular keratinocytes in the skin of ShhGFPcre;R26R mice irrespective of whether the skin has been wounded or not.
Follicular cells participate in regeneration of the epidermis in wounded skin
The contribution of keratinocytes derived from the follicle to the wound epithelium was evaluated in ShhGFPcre;R26R mice. In this wound-healing model, keratinocytes from the tissue surrounding the wound are mobilized shortly after wounding and cover the wound surface in the course of 2–3 d. Labeled cells can be observed migrating into the wound as soon as a day after wounding (Fig. 2
A, arrowheads). After 2 d, trails of blue cells leading from the follicles to the wound are observed (Fig. 2B
). Within a week, the wound is completely covered with keratinocytes of both epidermal and follicular origin. By 2 weeks a thickened epidermis covers the entire wound, composed of hyperproliferative cells in both the basal and the thickened suprabasal layers. After whole mount detection of ß-galactosidase at this stage, a surface view of the healed wounds revealed extensive labeling (Fig. 2C
). The healed epidermis consisted of both labeled cells derived from the follicle and unlabeled keratinocytes. These latter cells could be derived either from interfollicular epidermis or from the comparatively rare unlabeled cells of the hair follicle. Although the surface view of the wounds indicated over 50% of the superficial skin is labeled, cross sections through the wounds revealed that the extent of labeling is reduced in the deeper layers underlying the stratum corneum (Fig. 2D
). In the basal layer of the wound epidermis, 23 ± 12.5% of the cells was positive for ß-galactosidase activity. A single sample analyzed 1 week after harvest showed a similar representation and distribution of labeled cells. Blue basal cells were phenotypically indistinguishable from adjacent unlabeled basal cells. The distribution of labeled cells within the basal layer appeared to be random, lacking an apparent recurring pattern or spacing. In some areas of the wound, the labeled basal cells were continuous with the infundibulim of an adjacent hair follicle. At this stage in wound healing, labeled basal cells are associated with overlying blue cells in the suprabasal layers, but not all labeled cells in the suprabasal layers are readily traceable to labeled basal cells through a continuous column of labeled cells. Control experiments with k14cre;R26R mice in which the reporter is expressed in all keratinocytes of the skin confirmed the ability to detect ß-galactosidase in all cells of the wound epidermis at this and all stages analyzed (data not shown).
|
Follicular cells can transform into epidermal stem cells as a consequence of trauma
Wound epidermis was evaluated 9 and 16 wk after wounding when the epidermis has reverted to a more normal structure. At 9 wk 26 ± 10.2% of the basal cells were labeled. Columns of labeled suprabasal cells overlay groups of labeled basal cells in an arrangement suggestive of epidermal proliferative units (Fig. 3
A–C). This organization in columnar units persisted in the regenerated epidermis for extended periods. Wounds evaluated at 4 mo after wounding showed a similar pattern of labeling (Fig. 3D, E
). However, the relative proportion of labeled cells in the basal layer increased to 37 ± 16%. At all time points evaluated, isolated groups of labeled basal cells could be observed separated from the hair follicles located at the edge of the wound by unlabeled cells on all sides. However, in contrast to the early wound epidermis where labeled and unlabeled cells were extensively interspersed (Fig. 2D
), the contiguous patches of labeled or unlabeled cells became larger and more homogenous over the time course analyzed. This was most pronounced at the final time point analyzed (Fig. 3)
. The persistence of follicular descendants in the basal layer of the regenerated epidermis and its differentiated progeny in suprabasal layers demonstrate that in wounding conditions follicular cells can undergo reprogramming into long-term repopulating progenitors of the epidermis.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Here we show that Shh expression in the skin remains restricted to its normal pattern in the hair follicle matrix even after epidermal injury. In situ hybridization analysis and postinjury induced lineage tracing demonstrated that neither the epidermal cells nor regions of the follicle other than those that normally express Shh induce expression of this gene in response to injury. Moreover, keratinocytes participating in wound repair do not express Shh. During follicle development, Shh promotes proliferation of follicular epithelium, either directly or indirectly (21)
. However, expression of Shh in wound epidermis does not appear to play a significant role in the hyperproliferation of keratinocytes in response to wounding. As induced Shh expression is not directly involved in the injury response, the Shhcre;R26R reporter system can be used as a reliable in vivo assay to detect changes in follicular cell fate. As such, these results reveal that the follicular epithelium makes a substantial contribution to the initial resurfacing of the wound. In addition, cells derived from follicular epithelium remain resident in the basal layer of the epidermis and ultimately become the long-term repopulating cells that maintain the regenerated epidermis in the wound area. Thus, an effective lineage conversion is achieved as cells that would normally not give rise to epidermis effectively adopt a new fate to maintain this compartment. By morphological and marker gene analysis, this fate conversion is complete as the cells derived from the follicle and those derived from the original epidermal compartment are indistinguishable by these criteria. The apparent trend toward segregation of cells derived from the follicle and epidermis might be indicative of some remaining distinction between these two populations that encourages sorting between the two groups. However, the irregular borders of the labeled sectors that remain even after extended periods would seem inconsistent with a strong sorting mechanism. Alternative explanations for this phenomenon are discussed below. Irrespective of this point, these results clearly demonstrate that cells from the pilosebaceous unit, cells that would normally not contribute to the epidermis, can be converted to the long-term repopulating cells resident in the epidermis as a consequences of wounding.
While the contribution of follicular cells to the forming wound epithelium detected by this method is consistent with previous reports (13
, 14
, 19
, 20
, 23
, 24)
, the persistence of these cells in the regenerated epidermis for extended periods contrasts with the reported inability of cells derived from the follicular bulge to become stem cells of the regenerated epidermis after wounding (13)
. That conclusion was reached based on analogous experiments using a k15crePR line to preferentially label the stem cells of the follicular bulge (13)
. By contrasting and synthesizing the results of these two sets of experiments, we may learn more about the process by which an epidermal stem cell compartment is formed in wound epidermis. The salient differences between those experiments and the ones reported here are summarized schematically in Fig. 4
. The population labeled in the experiments reported here encompasses the follicular bulge cells and their progeny that were labeled in the k15cre experiments but it is substantially broader (Fig. 4A
). It includes constituents of the pilosebaceous epithelium that are not normally derived from the follicular bulge. Both the infundibulum and the sebaceous gland keratinocytes are part of the original follicular lineage unit and remain segregated from the epidermal compartment in the absence of injury (13
, 17)
. However, lineage analysis suggests that they are maintained by populations distinct from the stem cells of the follicular bulge (13
, 17
, 19)
. Thus, the different outcomes of the experiments may reflect intrinsic differences in the populations from the follicular epithelium that colonize the wound epithelium. The keratinocytes of the infundibulum or sebaceous gland may be more capable of converting to epidermal stem cells than the stem cells of the follicular bulge.
|
It is noteworthy, however, that in the absence of competition from other keratinocyte populations, bulge-derived cells can sustain the epidermis. Bulge stem cells, purified by FACS using the same keratin 15 driver sequences employed in in vivo lineage analysis, can be used to reconstitute skin that persists for extended periods (12)
. Thus any intrinsic difference in the ability to form epidermal stem cells between follicular bulge stem cells and other keratinocyte populations is relative rather than absolute. In that context, two other aspects of the labeled populations that might influence competition in the specification of stem cells in regenerating wound epidermis must be considered. First, as a consequence of the broader lineage label in the follicle (Fig. 4A
), a larger percentage of the initial wound epithelium is labeled in the Shhcre experiments than in the k15crePR work. Thus, a competitive advantage of labeled cells may be attributed to superior representation in the initial wound epidermis population. Second, the cells of the upper follicle, which are mobilized rapidly and may colonize the wound epidermis substantially ahead of those migrating from the follicular bulge, are labeled in the ShhGFPcre experiment (Fig. 4A
). This may confer a competitive advantage based on timing rather than overall representation. The preferential accumulation of labeled cells in the suprabasal layers of the early wound epidermis that is observed in the k15cre experiments may reflect this temporal disadvantage. The comparatively late arrival of follicular bulge stem cells in the wound epidermis may limit their contribution to a basal layer already occupied by keratinocytes from the epidermis and upper follicle (Fig. 4B
).
In considering the potential significance of these competitive advantages in the absence of intrinsic differences, we discuss three models for the establishment of stem cells in the wound epidermis. At one extreme is a framework that emphasizes the stem cell as a unique cell type. Mobilized stem cells, migrating as such from either the epidermis or the follicle, home to a vacant niche in the wound, akin to the homing of a transplanted HSC to its bone marrow niche after myeloablation (25)
. In the context of this model, the epidermal and follicular bulge stem cells are equally represented in both experiments. One must then postulate that a distinct stem cell population resident in the upper follicle becomes the stem cell of the wound epidermis (Fig. 4A
). The persistence and, if anything, increase in the abundance of labeled cells in the wound suggests that this population is at least as effective at maintaining the epidermis as stem cells that originated in the epidermal compartment.
While an intrinsic advantage in different stem cell populations may explain their success in this model, the competitive advantage to stem cells of the upper follicle based on timing may also be relevant in this model. The competitive advantage based on numbers is less likely to apply to this model, while both the timing and numerical advantages pertain to the following models.
An intermediate model assumes the epidermis is maintained by a dedicated stem cell-population but that population is regenerated de novo in wound epithelium from among the cells of the basal epidermis in the healing wound. This process is analogous to the initial specification of stem cells in the skin. While specification of stem cells remains poorly understood, lineage analysis suggests that stem cells in the pilosebaceous unit are specified in situ from within the follicular epithelium (17)
. This model is sensitive to the representation of labeled cells in the basal layer at the time at which stem cells are selected. In the work reported here, 23% of the basal cells in the wound epithelium were labeled in samples analyzed at 1 and 2 wk after wounding, although superficial layers are labeled at higher frequency. The representation of labeled cells in the basal layer was not reported in the k15 work, but at 8 d after wounding it appears substantially lower than that observed in the ShhGFPcre experiments (Fig. 4B
). This under-representation in the basal layer could, in principle, be adequate to explain the exclusion of bulge-derived cells from the epidermal stem cell population. To assess the probability that they would not be selected in a random sampling of the basal epidermis, the frequency of stem cells must be known. Estimates of stem cell frequency range from 5 to 15% based on EPU models, down to 0.5 to 2% based on surrogate assays (3
, 13
, 19
, 26
27
28)
. Given the size of the wound epidermis and the frequency of labeled basal cells, exclusion based on random sampling is a viable explanation only if stem cell frequency is on this extreme low end of this range.
A third model posits no difference between epidermal stem cell and TA cell. As discussed above, the empirical distinction between stem and TA cell is based on relative life span in the basal epidermis. Most models assume the TA cell persists for at most a few weeks, while the epidermal stem cell is thought to persist for extended periods, perhaps the life of the organism. However, a maximum life span that clearly distinguishes the TA from the stem cell remains elusive, as does an unequivocal demonstration that these two distinct populations exist. In this alternative model, cells of the basal layer are randomly displaced to the suprabasal layer and replaced by a symmetric division of a neighboring basal cell. A notable feature of this model is that the population dynamics of labeled and unlabeled cells remains sensitive to the frequency of labeled cells in the basal layer well past an establishment phase. In contrast, stem cell-based models become insensitive to the frequency of labeled cells once the stem cell population is established. Like the other models, this one accommodates both the loss of comparatively rare basal cells derived from the follicular bulge and the persistence of the more abundant labeled cells in the Shhcre experiment. However, this model best explains another aspect of labeled cell distribution observed in the Shhcre wound epithelium (Fig. 4C
). Over time there is a progressive increase in the homogeneity of label within a given region. Although the general representation of labeled cells in the epidermis remains similar, larger contiguous sectors of labeled or unlabeled cells are observed. The stem/TA model accommodates such behavior only if the stem cells are extremely rare and TAs are substantially longer-lived than normally thought.
Finally, the formal possibility remains that the recruitment of follicular cells to the wound epidermis during wound healing abrogates a lineage or affinity boundary and allows the subsequent repopulation of the wound epidermis by continued recruitment of keratinocytes from the pilosebaceous unit. These would presumably have to be derived from infundibulum or sebaceous gland, as they would otherwise be expected to appear in both experimental groups. While we cannot completely exclude this possibility, the fact that "islands" of labeled cells are observed, separated on all sides by unlabeled cells from adjacent follicles in the unwounded epidermis, argues against it. Cells migrating from the follicle to the wound epidermis to maintain the labeling of the wound epidermis would have to violate this affinity boundary to regenerate islands of labeled cells.
In conclusion, the work reported here demonstrates that cells of follicular origin can assume the fate of resident epidermal stem cells. By nature of their follicular origin, these are distinct from the epidermal stem cell pool thought to enjoy a competitive advantage over follicular bulge cells in this assay. While these experiments are not adequate to fully address the potential of follicular bulge stem cells, they raise the possibility of no intrinsic difference between follicular bulge stem cells and the other cells of the hair follicle and epidermis with respect to their capacity to become long-term repopulating cells of the wound epidermis. They further suggest that alternatives to the generally accepted stem/TA model of epidermal maintenance are consistent with available data, at least in the context of regenerated epidermis, and merit further critical evaluation.
Received for publication July 19, 2006. Accepted for publication December 14, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. D. Yu, A. Mukhopadhyay, and C. Wong Skin and hair: models for exploring organ regeneration Hum. Mol. Genet., April 15, 2008; 17(R1): R54 - R59. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Fuchs Skin stem cells: rising to the surface J. Cell Biol., January 28, 2008; 180(2): 273 - 284. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. MAES, T. KOBAYASHI, and H. M. KRONENBERG A Novel Transgenic Mouse Model to Study the Osteoblast Lineage in Vivo Ann. N.Y. Acad. Sci., November 1, 2007; 1116(1): 149 - 164. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
