2.1 Cell culture

DP cells were isolated³³ from HF occipital scalp samples obtained from consenting patients who underwent hair transplantation procedures. DP cells were sub-cultured on bovine collagen-coated (Sigma-Aldrich) surfaces in Dulbecco's modified eagle's medium (DMEM, Sigma-Aldrich) with 10% foetal bovine serum (FBS) and 1% antibiotic-antimycotic solution (Gibco). Neonatal foreskin hMel were purchased from Cell Applications (catalog no. 104-05n) and cultured in the recommended HEM complete medium. Unless otherwise stated, the cell densities used to establish the monocultures were 2x10⁴ cells/cm² (DP cells and hMel).

Physoxia cultures were established with 5% O₂ in a hypoxic chamber (Coy O2 Control Glove Box; Coy Laboratory Products). Cells cultured under normoxia were used as controls. Cells were used up to passage 4 (hMel) or passage 7 (DP cells).

To assess the link between ROS and ALP, DP cells were cultured overnight and then treated with hydrogen peroxide (H₂O₂; 300 μΜ, PanReac AppliChem), with N-acetyl cysteine (NAC; 1 mM, Sigma-Aldrich) or with NAC for 2h before H₂O₂.

2.2 Morphology and aggregation analysis

After 3 days of culture, DP cells were fixed with 10%-formalin for 15 minutes at room temperature (RT) and their F-actin filaments stained with Phalloidin-TRITC (0.1 mg/mL, Sigma-Aldrich) for 1 hour at RT. Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI) (1:1000, Biotium) for 15 minutes at RT. Images (six per triplicate) were acquired with an Axio Observer microscope (Zeiss) and used for the quantification of cell area, perimeter and major axis length, with the software module “MeasureObjectSizeShape” of CellProfiler^TM 3.0.0.³⁴

Cell aggregation was analysed after 7 days of culture, after labelling the cells' nuclei with DAPI. Images (ten per triplicate) were acquired (Axio Observer, Zeiss) and analysed with the CellProfiler 3.0.0^TM module “RelabeledNuclei”. Cells were considered adjacent if the distance between their nuclei was below 8 pixels. Groups of 30 or more adjacent cells were counted as one aggregate.

2.3 Senescence-associated-β-Galactosidase assay

DP cells were seeded at a density of 1 x 10⁴ cells/cm² and cultured overnight. Next day, the cells medium was replaced by serum-free DMEM and DP cells were kept in culture for 5 days. The detection of senescence cells was performed using the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology) following manufacturer instructions. Images (12 per each triplicate) used to quantify the percentage of senescent cells were taken using an AxioVert.A1 microscope (Zeiss).

2.4 Quantification of collagenous and non-collagenous proteins

The total amount of collagenous (COL) and non-collagenous (NCOL) proteins were quantified using the Sirius Red/Fast Green Staining Kit (Chondrex) according to the supplier instructions. DNA values were used to normalize data.

2.5 Migration assay

hMel (4 x 10⁴ cells/cm²) were seeded in 8.0 µm pore size inserts (Corning) in non-supplemented HEM medium, while complete medium was added to the bottom well. After 48 hours of culture under normoxia or physoxia, cells that migrated from the insert to the bottom well were fixed with 10% formalin (15 minutes, RT). Images were acquired with an AxioVert.A1 (Zeiss) for quantification (12 per triplicate) or with an Axio Observer (Zeiss) microscope after staining the cells' nuclei with DAPI.

2.6 Tyrosinase activity quantification

hMel were incubated for 5 minutes (ice) with 20 mmol/L Tris (hydroxyethyl) aminomethane (pH 7.5) containing 0.1% Triton X-100 and a protease inhibitor cocktail (1:100, Sigma-Aldrich). Cell lysates were then centrifuged at 14 500 rcf (10 minutes, 4°C) and 70 μL of the supernatant transferred into transparent 96-well plates. As a substrate for tyrosinase, a 0.1% (wt/v) L-Dopa (Sigma-Aldrich) solution was prepared in sodium phosphate buffer (pH 6.8) and 140 μL were added to each well. Plates were incubated for 2 hours at 37°C, and the absorbance measured at 475 nm using a microplate reader (Synergy HT, BioTek). Data are presented as relative tyrosinase activity after normalization with DNA values.

2.7 Co-cultures

DP cells resuspended in DMEM with 10% FBS were seeded at a density of 2 x 10⁴ cell/cm² and cultured overnight before establishing the co-culture with hMel, seeded at 2 x 10⁴ cell/cm² in HEM in 0.4 µm pore size inserts (Corning). Monocultures of each cell type were prepared as controls, either by culturing cells in their regular medium (DMEM or HEM), or in the medium used in to establish the co-culture, DMEM with HEM at a 1:1 ratio (DMEM:HEM), to control possible medium effects. Direct co-cultures were established by seeding 3 x 10³ DP cells in round bottom ultra-low attachment 96 wells (Corning) in 50 µL of DMEM with 10% FBS for 24 hours before the addition of 1.5 x 10³ hMel resuspended in 25 µL of HEM medium. Both co-cultures were performed for 3 days.

2.8 DNA, active alkaline phosphatase and ROS quantification

Cells were lysed in water with 0.01% sodium dodecyl sulphate. A 1 hour incubation at 37°C was followed by freezing at −80°C. DNA content was quantified using the Quant-iT™ PicoGreen® dsDNA kit (Thermo Fisher Scientific), and ROS levels were measured using the OxiSelect™ In vitro ROS/RNS assay kit (Cell Biolabs Inc). Cell lysates were also used to quantity active ALP levels in DP cells, using the Alkaline Phosphatase Detection Kit (Sigma-Aldrich). For DNA and ALP quantification in cell aggregates, a 5s sonication step (ice) was first performed to ensure the complete disintegration of the 3D aggregates prior quantification. All commercial kits were used according to the manufacturer instructions. DNA values were used to normalize ROS and ALP results.

2.9 Alkaline phosphatase staining

The detection of active ALP was performed by incubating formalin-fixed DP cells (15 minutes, RT) for 20 minutes with a solution prepared with 5 μL of p-nitroblue tetrazolium chloride (NBT) and 3.75 μL of 5-bromo-4-chloro-3-indolyl phosphate (BCIP, Roche) in 1 mL of staining buffer [100 mmol/L NaCl, 100 mmol/L Tris-HCl (pH 9.5) and 50 mmol/L MgCl₂ in water]. Representative images were acquired with an AxioVert.A1 microscope (Zeiss).

2.10 ELISA

After co-culture, inserts containing the hMel were removed and DP cells were cultured in serum-free DMEM for 24 hours. The supernatant of the cells was then collected, centrifuged (1000 g, 10 minutes) and single-use aliquots were stored at −80°C. Human vascular endothelial growth factor (VEGF) ELISA Development Kit (PeproTech) and human bone morphogenetic protein-2 (BMP2) Standard ELISA Development Kit (Petrotech) were then used following the manufacturer instructions to determine VEGF and BMP2 levels. DNA values were used to normalize data.

2.11 Immunofluorescence staining

DP-hMel aggregates were fixed in 10%-formalin (overnight, 4°C), embedded in HistoGel^TM (Thermo Scientific) and processed for paraffin inclusion. 4-µm paraffin-embedded sections were then dewaxed and heat-mediated antigen retrieval was performed with sodium citrate buffer (pH 6.0). Primary antibodies (Table S1) were detected with Alexa Fluor®488/594 (1:500, Molecular Probes) secondary antibodies and nuclear counterstain was performed with DAPI. Images were acquired using an Axio Imager Z1m microscope (Zeiss).

Haematoxylin and eosin (H&E) staining was performed according to standard protocols and representative images acquired with a DM750 microscope (Leica). Images were used to count cell nuclei and determine the DP cells/hMel ratio present within the cell aggregates for normalization of tyrosinase and ALP activity using the DNA amount of the corresponding cells.

2.12 Statistical analysis

Statistical analysis and data visualization were performed using the GraphPad Prism 7.03. The D’Agostino & Pearson normality test was used to determine whether data followed a Gaussian distribution. Non-parametric data were analysed with a two-tailed Mann-Whitney test (two groups, unpaired) or with a Friedman (paired) or Kruskal-Wallis test (unpaired) when more than 2 groups were compared. Parametric data were analysed with a two-tailed Student's t test (two groups, paired or unpaired). The comparison of more than two groups was performed with an ordinary (unpaired) or RM (paired) one-way ANOVA (one independent variable) or two-way ANOVA (two independent variables). Data are presented as mean ± standard error of the mean (SEM). For data displayed as dot plots, black dots represent data points and red bars indicate the mean. Differences with p-values <.05 were considered significant.

3.1 Physoxia reduces the negative impact of in vitro culture conditions on DP cells

DP cell cultures are typically established under normoxia, rapidly losing their native phenotype and intrinsic properties,^{35, 36} including their key self-aggregation capacity.^{37, 38} Moreover, they also have a short lifespan ³⁹ in culture, which is accompanied by morphological changes such as shifting from a small polygonal morphology to a spindle-like shape,^{40, 41} before acquiring an enlarged morphology.²³ Therefore, we investigated whether those changes also occurred under physoxia to understand how the O₂ level impacts DP cells’ phenotype in culture. DP cells under physoxia depicted a polygonal and less spindle-like shape and higher nuclei-to-cytoplasm ratio, as demonstrated by the significant decrease in the cells’ area, perimeter and major axis length in comparison with normoxia (Figure 1A-D). Physoxia also significantly decreased the percentage of senescent DP cells in culture (Figure 1E) and improved their aggregative capacity (Figure 1F). Moreover, it enhanced cell proliferation, albeit the DNA amount at day 3 was similar in normoxia and physoxia (Figure 1G). Interestingly, an opposite effect was observed regarding COL (Figure 1H) and NCOL (Figure 1I) protein secretion under physoxia, which was only beneficial after 3 days in culture. Altogether, these results suggest that physoxia promotes a healthier state in cultured DP cells, which featured characteristics typically associated with low passage cells.

3.2 Physoxia enhances hMel migration, tyrosinase activity and proliferation within short culture times

Although hMel are normally cultured under normoxia, there are indications that their proliferation and tyrosinase activity are favoured under lower oxygen tensions.²⁸ We found that physoxia significantly increased both hMel migration (Figure 2A) and tyrosinase activity (Figure 2B), although this last effect was not sustained along with the culture. Similarly, significantly higher DNA levels were observed for hMel cultured under physoxia at day 3 of culture (Figure 2C), suggesting an improved proliferative capacity. This effect was lost with the culture time, despite the high number of Ki67-positive cells (Figure 2D). Physoxia did not seem to affect the morphology of hMel (Figure 2E). Collectively, these results indicate that physoxia supports hMel functional features better than normoxia but only for short culture periods.

3.3 DP cell and hMel response to physoxia depends on their type of interaction

Although residing in close vicinity in the hair bulb and having their functions coupled to anagen,^{42, 43} little is known about how human DP cells and hMel interact and potentially affect each other's functionality. Knowing that physoxia individually improved hMel and DP functional features after 3 days in culture, we then explored its effect when these cells were indirectly co-cultured (Figure 3A). The co-culture with DP cells did not add to the increased hMel proliferation induced by physoxia (Figure 3B), in opposition to normoxia. Like for proliferation, co-culture with DP cells under physoxia did not affect tyrosinase activity in hMel, contrarily to normoxia, which promoted a recovery from the negative effect of the co-culture medium (Figure 3C).

Regarding DP cells, they proliferated significantly more under physoxia than in normoxia but only when cultured in their conventional medium. Thus, the higher DNA amount detected in co-culture might be due to the medium used. This is also sustained by the results similar to the control condition, both under physoxia and normoxia (Figure 3D). Active ALP levels in co-cultured DP cells were not affected by physoxia but, as for proliferation, the medium used led to a significant increase of this inductive marker. In the presence of hMel, a significant decrease of DP cells’ active ALP was observed under physoxia, but not in normoxia (Figure 3E).

In the HF, hMel and DP cells are separated only by a thin and permeable basal lamina.² Therefore, we sought to investigate whether physoxia effects were different from those observed in the indirect co-cultures, assuming a direct interaction between hMel and DP cells. When hMel were cultured with DP spheroids, they organized themselves around the spheroid, displaying a polarized position over one-half of the DP spheroid, independently of the oxygen level (Figure 3F). Highly stable aggregates with similar DNA content were obtained (Figure 3G). Interestingly, the phenotype of both cell types was improved under physoxia, as demonstrated by a significant increase of tyrosinase activity in hMel (Figure 3H) and by the higher amount of active ALP in DP cells (Figure 3I), their main functional markers, respectively. Moreover, the production of COL and NCOL proteins by aggregates cultured in physoxia was significantly higher than in normoxia (Figure 3J). Physoxia benefits hMel and DP cells functionality when both cell types are directly contacting. Interestingly, hMel response to physoxia does not seem to be indirectly affected by DP cells, while hMel signalling appears to have an impact on DP cells’ ALP activity.

3.4 ROS generation due to hMel and DP cells interaction does not directly correlate with DP cells functionality

During hair growth and pigmentation, the bulb is a ROS-enriched environment³¹; therefore in addition to the functionality of hMel and DP cells, we addressed the involvement of ROS in their response. The production of ROS by hMel was significantly lower under physoxia, although this effect was significant only for the co-cultures (Figure 4A). Moreover, hMel in co-culture produced significantly more ROS than in the control conditions, regardless of the oxygen level. Physoxia also led to a reduction of ROS levels in DP cells in comparison with normoxia, independent of the culture conditions (Figure 4B). The indirect co-culture with hMel under physoxia also resulted in significantly higher amounts of ROS than in control conditions (Figure 4B). Surprisingly, when cells were directly cultured, ROS production in physoxia was significantly higher than in normoxia (Figure 4C), the opposite of what was observed in indirect co-cultures.

Considering that in indirect (Figures 3E and 4B) or direct (Figures 3I and 4C) co-cultures the effect of physoxia over the amount of active ALP and ROS followed a common trend, we then investigated whether there was a correlation between these responses. For that, DP cells were treated with H₂O₂ (exogenous ROS) to increased oxidative stress, with the ROS inhibitor NAC or both. Treatment with H₂O₂ led to a significant decrease in the amount of DNA (Figure 4D) but surprisingly, it did not affect ROS intracellular levels in DP cells (Figure 4E) while a significant decrease of active ALP was observed (Figure 4F). Moreover, NAC pre-treatment before H₂O₂ addition further enhanced ROS production in comparison with H₂O₂ alone, but it also reduced the H₂O₂ effect on the amount of active ALP. Although it is not clear the mechanism by which H₂O₂ decreases ALP activity in DP cells, these results suggest that it is not directly correlated with an increase in ROS levels.

3.5 Physoxia and 3D co-culture supports DP cells and hMel phenotype

To further explore both physoxia and hMel influence on DP cells hair regenerative potential, we looked at the production of a known promoter^{44, 45} or inhibitor factor⁴⁶ of hair induction, namely VEGF and BMP2, respectively. In physoxia, the amount of VEGF released by DP cells was significantly higher than in normoxia independently of the culture condition (Figure 5A). The co-culture medium negatively impacted VEGF release by DP cells, which was not overcome with the co-culture with hMel. In opposition, the amount of BMP2 in physoxia was significantly lower than in normoxia, for both co-cultures and conventional DP culture medium (Figure 5B). As this effect was not seen in the control established with the co-culture medium, it suggests that hMel presence was essential for the observed result.

Considering that hMel-DP cell aggregates better resemble the HF bulb than plain 2D cultures, and the observed advantages regarding their functionality under physoxia, we then investigated the rescue of cell's native-like phenotype under these conditions. The expression profile of different DP and hMel markers in the 3D aggregates (Figure 5C) showed that the cells display a phenotype similar to the native HF (Figure 5C, upper panel). The identification of specific markers of hMel—tyrosinase, PMEL, MelanA—and markers expressed also by DP cells—vimentin and S100—showed clear compartmentalization between the cell types. Moreover, Ki67 immunolabeling confirmed a low number of proliferative cells within both cellular compartments (Figure 5C). Noteworthy, tyrosinase expression in the cellular aggregates was higher than in conventionally cultured hMel (Figure S1). Moreover, the in vivo predominant V2-isoform of versican,⁴⁷ a DP inductive marker typically absent in 2D-cultured cells (Figure S1) was weakly expressed in the cellular aggregates, confirming what was previously described for DP spheroids.³⁵

Overall, these results indicate that under physoxia DP cells’ inductive secretome is promoted which, in conjunction with 3D-culture conditions, allows an improved recovery of hMel and DP cell functional markers.