Ten newborn (0 days old) C57BL/6J mice and 18 nude male mice (Balb/c, nu/nu; 4–6 weeks old) were obtained from the Experimental Animal Centre of Southern Medical University (Guangzhou, China). Sixteen newborn (0 days old) red fluorescent protein (RFP) mice were obtained from Beijing Vitalstar Biotechnology (Beijing, China). Ten newborn (0 days old) and 24 adult (6–8 weeks old) female green fluorescent protein (GFP) mice were obtained from the Institute of Animal Models of Nanjing University (Nanjing, China). All animal experiments were performed under the approval of the Southern Medical University Animal Care and Use Committee.

Full thicknesses of dorsal skin were derived from newborn RFP mice at natal day 0. The dermis and epidermis were separated using dispase (Sigma, St. Louis, MO, USA) by incubation at 4°C overnight. The piece of skin was rinsed three times with phosphate-buffered saline (PBS, Gibco, Grand Island, NY, USA), then the skin piece was split into epidermis and dermis with forceps. Each component was minced. The dermis was digested in 0.2% collagenase (Sigma, St. Louis, MO, USA) at 37°C for 1 h. After digestion, an equal volume of Dulbecco’s modified Eagle’s medium (DMEM, Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, Grand Island, NY, USA) was added, and the cell suspension was filtered sequentially through 100 μm and 40 μm mesh cell strainers. The cell suspension was centrifuged at 230 g for 5 min, then the cell pellet was resuspended in DMEM. The epidermis was digested in 0.25% trypsin-EDTA (Gibco, Grand Island, NY, USA) at 37°C for 10 min to obtain freshly isolated epidermal cells, as previously reported [15]. The preparation of cells from GFP newborn mice is the same as previously described.

Ninety milliters of water was pipetted into a 250 ml beaker, then 10 ml of solution 1from Cell-in-a-Box kit (Sigma, St. Louis, MO, USA) was added. The pipette was rinsed in the hardening bath. The mixture was stirred for 10 min. For encapsulation, the speed of the stir bar was reduced to the lowest practical speed. The cells were washed twice in PBS and counted. Dermal cells (1.4×10⁶) from GFP newborn mice and epidermal cells (0.7×10⁶) from RFP mice were placed in a sterile 1.5 ml microcentrifuge tube. The cells were centrifuged at 200g for 5 min and the supernatant was discarded. One milliliter of solution 1 was added to the cell pellet and the pellet was resuspended by pipetting up and down until the cells were uniformly dispersed. The formation of air bubbles was avoided. A red plastic filling needle (G18½, blunt end) was added to a 1 ml Luer lock syringe and the cell suspension was drawn up. The filling needle was replaced with a green plastic droplet needle (G34, blunt end), taking especial care to assure that the needle was screwed firmly in place. Air bubbles were eliminated from the syringe. The needle was held vertically, 2–3 cm above the hardening bath. Droplets were dispensed at a moderate rate of 1–2 drops per second while maintaining the same drop height. The needle was moved around slightly to prevent droplets from landing at the same spot in the bath. We continued to make as many capsules as required, but did not dispense droplets after 1 min. After dispensing the last droplet, the capsules were stirred for 5 min. The stirrer speed was adjusted to ensure that the capsules were moving continuously in the bath. The stirrer was stopped and the capsules were allowed to settle. Fifty milliliters of the bath solution were discarded using a serological pipette, then 100 ml of sterile PBS was dispensed into the beaker. Stirring was restarted to wash the capsules for 10 min. One hundred milliliters of the bath solution was discarded and 100 ml of fresh sterile PBS was added into the beaker. The bath solution was washed again for 5 min. The remaining PBS was discarded, leaving just enough liquid to cover all of the capsules. The bath solution was washed three times with 30 ml of PBS, then three additional times with 30 ml of cell culture medium. A rinse cycle was performed by the addition of 30 ml of liquid, which was then removed by pipette. The preparation of blank capsules was the same as above.

The capsules were picked up with a 25 ml serologic pipette, and placed in culture dishes. An appropriate volume of cell culture medium was added (a mixture of DMEM containing 10% FBS and keratinocyte serum-free medium (K-SFM, Sigma, St. Louis, MO, USA) at a ratio of 2:1). Then, the dishes were incubated at 37°C and 5% CO₂ in air. The medium was changed at regular intervals 2–3 times a week. The capsules were observed regularly. Cell spheres were removed from some capsules 20 days later, which were cultured in vitro. Some of the cell spheres were reseeded in culture dishes and the migration of these cells was observed under a microscope. Other cell spheres were harvested for paraffin sectioning and hematoxylin-eosin (HE) staining. The remaining capsules were cultured for an additional 10 days.

GFP and nude mice were cleaned with Betadine and put under anesthesia by intraperitoneal injection of 10 g/l pentobarbital sodium (0.4 ml/100 g). The capsules that contain dermal cells from GFP newborn mice and epidermal cells from RFP newborn mice were then transplanted subcutaneously into nude mice. The blank capsules were also transplanted as the control group. Each group contained six mice and each mouse had one site where 10 capsules were transplanted as a whole. Cell mixtures from RFP mice (dermal cells, 5×10⁶; epidermal cells, 5×10⁶) in a total volume of 100 μl, were injected under the panniculus carnosus of the GFP mouse skin. Each mouse was injected at six points. In addition, dermal and epidermal cells were also injected as a control. Each group contained 6 mice as well. After recovery from anesthesia, mice were caged individually.

Skin color changes at the injected site were monitored every day. Fourteen days later, the mice were killed by an overdose injection of anesthetic, the skin of the injected site was harvested and the sample was observed with stereoscopic microscopy. Paraffin sections were prepared and the histologic sections were stained with HE. Frozen sections and 4′,6-diamidino-2-phenylindole (DAPI, Sigma, St. Louis, MO, USA) staining were made. The specific distribution of the fluorescent cells in the reconstructed hair follicle was observed in the frozen sections. Some of the graft sites where the capsules were transplanted were harvested 10 d later, and observed under a stereoscope and upright microscope. Cell spheres were removed, some of which were harvested for paraffin sections and HE staining, others were transplanted into nude mice. The hair follicle formation was monitored. Other capsules were retained in vivo and were harvested and assessed 40 d later.

The above steps were repeated with dermis and epidermis cells from newborn RFP mice. Capsules were transplanted into GFP mice. Ten days after transplantation, cell spheres inside the capsules were harvested, then transplanted into GFP mice. Similarly, the hair follicle reconstruction was observed.

To certify whether or not the host cells were involved in the hair follicle regeneration process, we transplanted combined cell mixtures, as well as single cells, into mice to form hair follicles. Dermal and epidermal cells from RFP mice were injected into GFP mice at a ratio of 2:1, and blebs were observed. The injected site became gray 4 d later and turned black 8 d later. Fourteen days later, the injected sites were harvested for detection. Abundant hair follicles were observed in sites injected with mixed cells (Fig 1A); however, when injected alone, epidermal and dermal cells failed to form hair follicles (Fig 1B and 1C). The completely developed structure of hair follicles was observed under the panniculus carnosus, and the panniculus carnosus was continuous and complete (Fig 1D). Frozen sections of the grafts were prepared and examined under a fluorescence microscope. The results indicated that newly formed follicles were mostly composed of red fluorescent cells, as well as a small number of green fluorescent cells (Fig 1E–1H), conclusively demonstrating that the host cells were involved in the process of hair follicle regeneration.

To verify the function of the involved host cells in the process of hair follicle reconstruction, fresh epidermal cells from RFP newborn mice and dermal cells from GFP newborn mice were encapsulated with the Cell-in-a-Box kit. We generated approximately 100 capsules every time, each capsule contained 2.1×10⁴ cells. Capsule diameter was 3.13±0.16 mm (mean±standard deviation). With the use of capsules, we isolated the injected cell mixtures from the host cells. We found that cells in the capsules gradually aggregated into small multicellular aggregates in vitro (Fig 2A and 2B), which are highly motile. Dermal and epidermal cells came from different fluorescent protein transgenic mice, they were observed under an inverted fluorescence microscope (Fig 2C and 2D). Four days later, multicellular aggregates aggregated into a hybrid spheroid (Fig 2E). Seven days later, the morphology of hybrid spheroids became stable, spheroids were almost uniform in size, the diameter was 281±7.5 um (Fig 2F), but there were no significant changes in the subsequent culture (Fig 2G). Twenty days later, cell spheroids were taken out of some capsules after in vitro cultivation (Fig 2H); no apparent differentiation was observed through the HE section of hybrid spheroids (Fig 2I). To eliminate this possibility that the lack of nutrients resulted in the death of encapsulated cells, some cell spheroids were then reseeded in culture dishes with the medium (DMEM containing 10% FBS and K-SFM [mixed 2:1]). The spheroids then attached to the dishes and the cells migrated from the spheroids (Fig 2J–2L). The remaining capsules were cultured in vitro for another 10 d, but no apparent change was observed (Fig 2M). With the help of laser confocal microscopy, we discovered that dermal cells were preferentially located in the center and epidermal cells were sorted to the surface (Fig 2N–2Q). The spontaneously formed layered structure of hybrid spheroids was similar to the natural three-dimensional organization of the hair bulb (a shell of keratinocytes surrounding the center of aggregated dermal papilla cells). The above results showed that the environment in capsules can support the survival of cells. The epidermal and dermal cells in the capsules aggregated gradually, which is similar to the state in the three-dimensional environment created by the hanging drop culture or the matrigel [16–18]. Nevertheless, the hybrid spheroid appeared to be the final outcome, and failed to continue the differentiation toward the structure of the hair follicle.

Whether or not the lack of some growth factors or host cells lead to the failure of cell mixtures developing into hair follicles in vitro is unknown. To explore the possibility that cells in capsules could reconstruct hair follicles in vivo, we used capsules and the cells taken from the capsules for transplantation. Intact capsules, which contain epidermal and dermal cells, were transplanted into nude mice. No adverse effects were observed following capsule transplantation. Such being the case, we mimicked the microenvironment in vivo; transplanted cells could accept all the nutrition from the host, but were also insulated from host cells. Blank capsules were transplanted as a control. The wound healed well. Ten days later, some transplanted sites were harvested (Fig 3A). Transplanted capsules were intact, inside of which the cells formed spheroids, and similar to in vitro cultures (Fig 3B and 3C). There was nothing in the blank capsules (Fig 3D). HE sections indicated that the cells formed concentric circles, whereas no hair follicles or obvious hair shafts were found (Fig 3E). Thus, cells in capsules differentiated toward hair follicles in vivo. To verify whether or not the immature hair follicle-like structures could mature, the other transplanted sites were harvested 40 d later, yet there were still concentric circles without any mature follicles. In conclusion, even if supplied with sufficient nutrients from the host, the encapsulated cells could not form mature hair follicles when insulated from host cells. Other cell spheroids harvested 10 d after in vivo transplantation were transplanted into nude mice. As expected, 30 d after transplantation, approximately 70%–80% of spheroids developed mature hair follicles. We then repeated the above steps with dermal and epidermal cells from RFP mice, and capsules were transplanted into GFP mice. Ten days later, cell spheres were harvested and transplanted into GFP mice. Mature hair follicles were also observed (Fig 4A), approximately 70%–80% of spheroids developed mature hair follicles. New reconstructed follicles were mostly constituted of red fluorescent cells, as well as a small number of green fluorescent cells (Fig 4B–4E).