Immunophenotyping of the human bulge region: the quest to define useful in situ markers for human epithelial hair follicle stem cells and their niche

Tissue collection

Human temporal and occipital uninflamed scalp skin was obtained from a total of 13 patients with informed consent and following Helsinki guidelines during routine face-lift surgery. Per antigen, scalp skin cryosections derived from at least five different individuals were examined (3–10 cryosections per patient). The specimens were embedded in Shandon Cryomatrix (Pittsburgh, PA, USA), snap-frozen in liquid nitrogen and stored at −80°C until use. The samples were cut into 6–8 μm thick cryosections and then processed for immunohistochemistry.

Immunohistochemistry (ABC-AP)

The cryosections were first air dried for 10 min and then fixed in acetone at −20°C for another 10 min. After air drying for 10 min the slides were washed three times for 5 min in Tris-buffered saline (TBS) and were then preincubated with 10% goat normal serum in TBS. The application of the different primary antibodies (Table 2) followed in their respective dilution in TBS overnight at 4°C. After washing three times for 5 min in TBS, sections were stained either with goat anti-mouse, goat anti-rabbit (1:200 in TBS, Jackson ImmunoResearch, Cambridgeshire, UK) or goat anti-rat (1:200 in TBS, Beckman Coulter, Marseille, France) biotinylated antibodies for 45 min. After washing three times for 5 min in TBS, a 30 min application of avidin–biotin complex, alkaline phosphatase conjugated solution (ABC-AP, Vector Laboratories, Burlingame, CA, USA; 30 min) followed, both at room temperature (RT). Finally, the slides were labelled with Fast Red (DAKO, Glostrup, Denmark) and counterstained with haematoxylin with washing steps in between.

EnVision®-AP

The fixation, the preincubation and the incubation of the primary antibody (Table 2) were performed as described above using standard immunohistochemistry. Then the cryosections were incubated with EnVision® solution (alkaline phosphatase, DAKO, rabbit/mouse) for 30 min. After that, Fast Red AP-chromogen was employed for visualizing the immunosignals and finally, sections were counterstained with haematoxylin.

Immunofluorescence (FITC, rhodamine)

To identify the IR in the bulge region of the human HFs in human skin via immunofluorescence staining we used primary antibodies (Table 2) and the secondary antibodies goat anti-mouse, anti-rat or anti-rabbit IgG conjugated with fluorescein isothiocyanate (FITC) or rhodamine (1:200 in TBS, Jackson ImmunoResearch, Cambridgeshire, UK) for 45 min at RT. The cryosections were counterstained with DAPI (4′,6-diamidin-2′-phenylindol-dihydrochlorid, Boehringer Mannheim, Mannheim, Germany) for 1 min and mounted with Fluoromount-G (Southern Biotechnologies, Birmingham, AL, USA).

Tyramide signal amplification

After standard fixation the cryosections were washed three times for 5 min using TNT (Tris-HCL NaCl Tween) buffer (0.1 mol/l Tris–HCl, pH 7.5; containing 0.15 mol/l NaCl and 0.05% Tween 20). Then horseradish peroxidase was blocked by washing with 3% H₂O₂ in phosphate-buffered saline (PBS) for 15 min. Preincubation was performed with the incubation of avidin and biotin for 15 min and 5% goat normal serum in TNT for 30 min with washing steps in between. Primary antibodies (Table 2) were diluted in TNT and incubated overnight at 4°C followed by a biotinylated secondary antibody goat anti-mouse (1:200 in TNT) for 45 min at RT. Next, streptavidin horseradish peroxidase (TSA kit; Perkin-Elmer, Boston, MA, USA) was administrated (1:100 in TNT) for 30 min at RT. The reaction was amplified by tetramethylrhodamine- or FITC-tyramide amplification reagent at RT for 5 min (1:50 in amplification diluent provided with the kit). The cryosections were counterstained with DAPI for 1 min and mounted with Fluoromount-G.

For all immunostaining assays, primary antibodies were omitted as a negative control. As routine internal positive controls, reproduction of published follicular IR patterns was chosen. Only those specific IR patterns that were well reproducible between at least five different individuals and were clearly above background were photodocumented and are reported here (Table 3).

Microscopy

For fluorescence microscopy we used a Zeiss Axiovert 200 M microscope using a green fluorescent protein (GFP) filter set (AHF Analysentechnik AG, Tübingen, Germany). The slides were photographed using a Zeiss AxioCam MRm Rev.3 Fire Wire (D) and the Zeiss AxioVision Rel. 4.5 software.

For light microscopy we used an Olympus BH-2 microscope (Olympus Optical Co., Hamburg, Germany). Photos were taken using a ColorView12 camera from Olympus and the analySIS® software (Soft Imaging System GmbH, Münster, Germany).

Quantitative immunohistochemistry

The immunostaining intensity levels for the selected examined antigens were compared by quantitative immunohistochemistry as previously described (13,17,43), using NIH image software (NIH, Bethesda, MD, USA). Two reference areas were defined (1, insertion point of the APM; 2, middle of the isthmus) and then analysis was performed on sections deriving from three to seven different individuals.

Statistical analysis

For evaluating statistical significance, the measurements were pooled and the mean and the SEM were calculated. With the statistical analysis software SPSS (SPSS Inc., Chicago, IL, USA) P-values (*P < 0.05) were assessed using the Mann–Whitney U-test for unpaired samples.

Cytokeratin 15, cytokeratin 19 and CD200 are upregulated in the bulge region of human HFs, but their expression is not restricted to it

CK15 is upregulated in human and murine bulge cells which are able to reconstitute all components of the cutaneous epithelium (2,30). Staining for CK15, by either EnVision®- or TSA-immunohistochemistry showed homogeneous IR in the outermost layer of the ORS (1, 5 ) – not only in the bulge region, but also in the proximal part of the isthmus of human HFs – and is significantly upregulated (Fig. 4). As a morphological marker for the bulge, the ‘follicular trochanter’ also showed positive IR [not shown, for documentation see (11)]. Importantly, CK15 IR was also found in the outermost layer of the proximal ORS (1, 5) and in human sweat glands (not shown).

Immunohistochemistry and -fluorescence methods contrast with the CK15 negativity of the upper and lower ORS in human anagen HFs (44). Although a previous report of infundibular CK15 IR (45) supports our finding that CK15 IR in human HFs is not restricted to the bulge, we did not observe CK15 expression in the peri-infundibular ORS (see Table 3 for summary of our own IR results, which highlights differences of our findings to previously published IR patterns of human HFs). Given its expression in the mitotically active basal cell layers of the HF, CK15 expression appears to regulate an early stage in the pathway of keratinocyte differentiation that precedes the decision of a cell to become epidermal or hair-like, which is provided by the bulge region as a reservoir for eHFSCs.

CK19, which should commit stem cells to an epidermal cell fate and differentiation in vivo and in vitro (46,47), shows upregulated IR in the outermost layer of the ORS in the bulge region of normal human anagen VI scalp HFs, including the ‘follicular trochanter’ (11), but was again not restricted to it (1, 4, 5). The detected IR pattern for CK19 in the bulge and distal to it (i.e. in the proximal part of the isthmus) was heterogeneous, with some cells displaying stronger CK19 IR than their neighbours, and sometimes only isolated cells in the ORS being highly CK19 positive (1, 5). This observation goes in line with previously reported murine CK19 positive cells, which were identified as [³H]thymidine-label-retaining cells (48). Occasionally, isolated CK19 positive cells were also identified in the outermost layer of the proximal ORS, at a large distance of the bulge. These isolated CK19 positive HF cells may represent Merkel cells (48–52).

Our findings confirm an upregulation of CK19 expression (Fig. 4) in the human bulge, but contrast with the previously reported restriction of CK19 expression to this ORS region (48,53–55) and with several other published IR patterns (3,44,56,57) [(58) cave: lanugo HF examined!] [(59) cave: vellus follicles examined!] (see Table 3).

As previously reported by Ohyama et al. (2), CD200-positive cells obtained from a human HF suspensions demonstrated high colony-forming efficiency in clonogenic assays, indicating successful enrichment of living human bulge stem cells. By using different immunostaining techniques, we show that prominent CD200 IR can be found in the outermost layer of the ORS between the insertion of the APM and that of the sebaceous gland duct [note that this region is classically defined as isthmus (19), while only its proximal end includes the bulge region]. In addition, by extrasensitive TSA immunofluorescence, we found that the dermal papilla (DP) including its blood vessels, the APM and the sweat gland mesenchyme and epithelium of normal human scalp skin show very prominent CD200 positive IR, and stains the companion layer of the human HF homogeneously (1, 5). The ‘follicular trochanter’ was also prominently CD200 positive [not shown, see (11)]. Therefore, even though CD200 is indeed upregulated in the human bulge, on the protein and gene expression level (2) as confirmed by quantitative immunohistochemistry (Fig. 4), once again is not exclusively expressed in the region and therefore CD200 alone does not suffice being a specific heHFSC marker.

Lhx2 is not a useful bulge marker in human anagen scalp HFs

Here, we also report the first IR pattern for Lhx2 in human HFs. Lhx2 has a role for maintaining the growth and undifferentiated properties of murine HF progenitors (31). Unexpectedly, the ‘follicular trochanter’, a proposed morphological bulge marker structure (11), was predominantly negative for Lhx2 IR (1, 5). Only relatively weak Lhx2 IR was found in the innermost cell layer of the ORS in the proximal isthmus and bulge region (1, 5).

Instead, the companion layer located between the ORS and the IRS, which may be an integral component of the latter (60) and is not considered to harbour heHFSCs (5,6), was most prominently positive for Lhx2 (1, 5). In addition, isolated Lhx2 positive cells were also seen in the proximal ORS (1, 5) and in the suprainfundibular ORS (1, 5), i.e. distant from the bulge in both distal and proximal directions.

These immunostaining results render it unlikely that Lhx2 can be a useful bulge marker in the HF, but raise intriguing new questions about the functions of this transcription factor in human hair biology. Based on the previously reported expression patterns, including our staining results of the differentiated epithelial compartments, Lhx2 maintains epithelial stem cells in an undifferentiated state in murine skin/hair epithelium (31), but it is unlikely that Lhx2 exerts direct effects on human eHFSCs and more likely represents, among others, a regulator of the stem cell niche quiescence (61).

Tenascin-C is upregulated in the bulge connective tissue sheath of human HFs

Next, tenascin-C was examined, since it is a key extracellular matrix protein that has been discussed as a functionally important component of stem cell niches (30,62) and is co-expressed with β1 integrin and fibronectin in human HFs (63). Interestingly, tenascin-C is significantly upregulated in the bulge region of human scalp HFs (Fig. 4), even though there is also extensive homogeneous IR for this antigen along the connective tissue sheath (CTS) of human scalp HFs (1, 5).

Tenascin-C IR is present both throughout the CTS and in the BM at the levels of the proximal part of the isthmus, the bulge and the hair bulb, with the strongest IR signal seen in the bulge. In addition, the APM is also positive (1, 5). This confirms our previous report that the CTS around the ‘follicular trochanter’ and the APM show tenascin-C IR (11) and is in line with the less detailed expression analyses of Shikata et al. (64–66) and Schalkwijk et al. (67). However, these authors had not noted an upregulation of tenascin-C IR in the bulge mesenchyme, which is clearly documented here (Fig. 4).

α6 integrin, β1 integrin, fibronectin, nidogen, LTBP-1 and fibrillin-1 are not upregulated in the bulge epithelium and/or mesenchyme

α6 integrin is expressed throughout the outermost layer of the ORS and the BM of the whole HF and demonstrates gene upregulation in murine bulge cells compared with basal keratinocytes (12). This IR intensity can also be seen in the DP and in the ‘follicular trochanter’, whereas only weak IR was shown in the APM (2, 5). Previously reported differential expression patterns in human HFs (56,68–70) (see Table 3 for details) could not be exactly reproduced by our group. Quantitative immunohistochemistry did not reveal significant differences in the IR intensity between the bulge region and other parts of the ORS (data not shown).

Essentially, this was also the case for β1 integrin IR. As documented in 2, 4, 5, there is homogeneous IR in the outermost layer of the ORS and the BM throughout the entire length of human HFs. In addition, the CTS, the DP and the ‘follicular trochanter’ region are positive for β1 integrin. This contrasts with the reported relative upregulation of β1 integrin expression in the bulge (3). However, closer examination of this paper shows that the authors did report prominent β1 integrin IR also well above and below the bulge area of the ORS, with no conclusive evidence provided that β1 integrin IR really is more intense in the bulge zone than in adjacent ORS regions. Therefore, our study discourages the use of exploiting β1 integrin antigen expression as a marker for human bulge-associated eHFSCs.

Fibronectin, nidogen, LTBP-1 and fibrillin-1 were all prominently expressed throughout the HF CTS and in the DP, with fibronectin and nidogen also expressed in the BM (2, 5), without noticeably higher IR in the bulge mesenchyme, thus rendering them unsuitable as immunohistological markers for this region.

We hereby present the first report of fibrillin-1 antigen expression in the human pilosebaceous unit – fibrillin-1 was expressed along the entire follicular CTS uniformly, with no upregulation in the bulge mesenchyme (2, 5).

CD34, connexin43 and nestin IR are useful negative markers for human bulge epithelial HF stem cells

Confronted by this paucity of convincing positive human bulge and heHFSC markers in situ, we advise the incorporation of postulated negative bulge stem cell markers to the positive markers (5–7,11). Since, in contrast to murine eHFSCs, where CD34 is a viable upregulated marker, together with CK15, to isolate bulge cells which possess stem cell characteristics, including multipotency and high proliferative potential (5), CD34 is not detectable in the human bulge (2,5,71). We have re-examined the CD34 IR pattern, using also more sensitive immunohistology methods (Envision®). Indeed, we found that the isthmus, bulge and infundibulum regions of the human ORS are negative for CD34 (3, 5), while the outermost layers of the ORS proximal to the bulge (i.e. lower ORS) exhibits slight, homogeneous IR (3, 5). Heterogeneous IR is also seen in scattered cells of the human CTS (3, 5). Our findings are largely in line with the report by Ohyama et al. (2) that human CD34 in the bulge is downregulated on both the gene and protein level, but contrast with an isolated report that claims CD34 expression is ‘in all human bulge regions’ (72).

A second potential negative heHFSC marker is connexin43, a gap junction protein (73) which is downregulated in epithelial stem cells. This demonstrates the incompetence of heHFSCs for gap junction-mediated cell-to-cell communication (74). Using a different immunohistology protocol, our studies essentially confirm a report (which examined human newborn foreskin) by showing that the entire ORS of adult human scalp HFs proximal and distal to the bulge shows homogeneous IR for connexin43 (53) (3, 5), while it is negative in the bulge itself. Importantly, the ‘follicular trochanter’ (3, 5) is also negative for connexin43. Contrary to a report by Arita et al. (75), no connexin43 IR was detected in any subset of cells in the human bulge or the IRS.

The intermediate filament protein nestin has long been used as a marker for neural stem cells, but is now widely accepted to be a general progenitor cell marker (76–78). While in nestin promoter-driven GFP transgenic mice, cells that are highly GFP positive have been reported in the murine epithelium (79–82). This GFP positive cells produce several cell types, including cells with neuronal, astrocytic, oligodendrocytic, smooth muscle, adipocytic and other phenotypes (42). We had previously not been able to detect nestin IR cells in the human bulge epithelium (83). However, Wang et al. (55) have reported nestin IR in the human bulge epithelium. Therefore, we systematically re-examined nestin IR, using an antibody specifically directed against human nestin as well as highly sensitive TSA immunofluorescence. In all human scalp HFs examined, contrary to what has been reported in mice, we did not detect any prominent nestin IR within the HF epithelium (including the bulge and the rest of the ORS) (3, 5). However, isolated nestin-positive cells were routinely noted in the human HF CTS and DP (3, 5), and in the human sweat gland epithelium (while non-specific nestin IR was prominent in the IRS, data not shown) (3, 5).