hESCs (KhES-1) were provided by RIKEN BioResource Center and were used in accordance with the hESC research guidelines of the Japanese government. All experimental protocols and procedures were approved by the Ethics Committee of Nagoya University Graduate School of Medicine (approval ES-001). Maintenance and differentiation culture of hESCs was performed as described (6–8, 12) with modifications (9) ( Supplementary Figure 1 ). POs harvested 100-200d after differentiation were used.

All animal experiments were approved by the Animal Experimentation Committee of the Nagoya University Graduate School of Medicine and were performed in accordance with institutional guidelines for animal care and use. Severe combined immunodeficient (SCID) male mice aged 8-9wk (C.B-17/Icr-Hsd-Prkdc^scid, Japan SLC, Shizuoka, Japan) underwent transaural hypophyseal ablation (13). Mice were anaesthetized with intraperitoneal (i.p.) injection of a mixture of 3 agents (medetomidine 0.75mg/kg, midazolam 4mg/kg, butorphanol 5mg/kg) (14) and pituitary tissue was aspirated from the sella turcica via the auditory meatus using a needle (KN-390, Natsume Seisakusyo, Tokyo, Japan) fitted to a 1ml syringe containing 0.2 ml saline. After the procedure, mice were injected with the medetomidine antagonist atipamezole (0.75mg/kg i.p.). Since hypopituitary SCID mice are sickly, they were bred in a clean environment in the P2A laboratory, a newly constructed animal facility at our university. Their cages were changed twice a week. Complete hypophysectomy was confirmed post-mortem in all mice used ( Supplementary Figure 2 ).

Blood collection and ACTH determination

ACTH levels were assessed as a biomarker of pituitary function. Blood samples were collected by tail transection between 1300 and 1700, with sampling before and 1h after administration of human CRH (2μg/kg, i.p.; Tanabe, Osaka, Japan). Plasma was separated from blood samples by centrifugation at 1,000 x g for 15min, 4°C. Plasma ACTH assay used an ACTH ELISA kit (MD Bioproducts, Oakdale, MN) reactive against human and mouse ACTH, with solution absorbance (ACTH concentration) read using Cytation 5 (Biotek, Winooski, VT). CRH loading tests were conducted 1wk after hypophysectomy and 1/~4wk after PO transplantation, including in sham-operated mice, until 6mo later ( Figure 1A ). Blood samples were collected repeatedly from the same mice. To prevent adrenal crisis, all mice received intramuscular dexamethasone, 0.2mg/0.61ml, after each blood collection. Mice with plasma ACTH levels < 10pg/ml after CRH stimulation were classed as hypopituitary and used as subjects.

Five POs were incubated at 37°C for 72h in 2.5ml cell culture medium (Iscove’s modified Dulbecco’s medium, Sigma-Aldrich, St. Louis, MO; Ham’s F12, Thermo Fisher Scientific, Waltham, MA (1:1); 1% GlutaMAX, Thermo Fisher Scientific; 1% Chemically Defined Lipid Concentrate, Thermo Fisher Scientific; 450 µM 1-thioglycerol, Sigma-Aldrich; and 20% KnockOut Serum Replacement, Thermo Fisher Scientific). Culture supernatants were collected. ACTH concentrations in supernatants were determined using an electrochemiluminescence immunoassay (ECLIA) kit (SRL, Tokyo, Japan) employed clinically in Japan.

Mice were anesthetized with isoflurane and placed supine. In inguinal subcutaneous white adipose tissue (ISWAT) transplantation, the left inguinal area was shaved. A 4mm vertical skin incision was made and a pocket in subcutaneous adipose tissue was created, with placement of 5 POs harvested from cell culture medium into the pocket using a wide-bore tip under microscopy. Nylon-suture closure over the POs was followed by skin closure ( Figures 1B–G ). In the sham-operated group, a vertical skin incision was made in the left inguinal region, a small pocket was created in the subcutaneous fat, and surgical wound closure was performed without PO transplantation. In the avascular region (AR) transplantation group, a vertical skin incision was made in the left dorsal skin, with placement of 5 POs. After these manipulations, mice received intramuscular dexamethasone (0.2mg/0.61ml).

Weight was followed and rate of weight loss was evaluated. Activity testing used a running wheel device (ENV-044; Med Associates, Georgia, VT).

Transplanted cell aggregates, skin, and fat, from SCID mice were fixed in 10% formalin, dehydrated for paraffin infiltration, and sectioned by sliding microtome. Sections at 5µm were stained with hematoxylin and eosin (H&E) or subjected to immunofluorescence microscopy for various antigens with nuclear 4′,6-diamidino-2-phenylindole counterstaining. Antigen targets included ACTH (mouse, 1:200, 10C-CR1096M1; Fitzgerald), LHX3 (LIM homeobox protein 3, rabbit, 1:3000, AS4002S; RIKEN custom), human nuclei (mouse, 1:1000, MAB4383; Millipore), E-cadherin (rat, 1:50, M108; Takara) and SMA (smooth-muscle actin, mouse, 1:200, M0851; DAKO).

RNA extraction and cDNA synthesis from POs and undifferentiated hESCs

RNA was extracted from POs and undifferentiated hESCs using the RNeasy Mini Kit (Qiagen, Hilden, Germany) following manufacturer’s instructions. RNA quality was evaluated using TapeStation 4150 (Agilent Technologies, Santa Clara, CA). cDNA was synthesized using ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo, Osaka, Japan).

Quantitative PCR (qPCR) of 5 POs and 5 transplanted POs (differentiated from 10,000 ESCs/sample) used a LightCycler 480 system (Roche Diagnostics, Rotkreuz, Switzerland). Data were normalized to those for GAPDH as an endogenous control and determined using standard curve-based relative quantitation. Primers used were: VEGFA, forward 5’-CTGTCAGGGCTGCTTCTTC-3’, reverse 5’-TTGCTGTGCTTTGGGGATTC -3’; VEGFB, forward 5’-TTGACTGTGGAGCTCATGGG-3’, reverse 5’-TGTGTTCTTCCAGGGACATCT-3’; VEFGC, forward 5’-TGTTTTCCTCGGATGCTGGA-3’,reverse 5’-ACATTGGCTGGGGAAGAGTT-3’; FGF2, forward 5’-AGGAGTGTGTGCTAACCGTT-3’, reverse 5’-CAGTTCGTTTCAGTGCCACA-3’; ANGPT2, forward 5’-TGACTGCCACGGTGAATAAT-3’, reverse 5’-CGTGTAGATGCCATTCGTGG-3’; GAPDH, forward 5’-CATCACTGCCACCCAGAAGACTG-3’, reverse 5’-ATGCCAGTGAGCTTCCCGTTCAG -3’. These primer sequences do not cross-react between human and mouse genes. qPCR was performed in POs and contralateral ISWAT. As thus assayed, expression of these genes was clearly lower in mouse ISWAT than in POs ( Supplementary Figure 3 ).

All data were analyzed using IBM SPSS statistics software (version 28.0.0.0, IBM, Armonk, NY). Data are expressed as means ± standard error. Comparisons between groups were performed using Student’s t-test. Comparisons among groups were performed by one-way ANOVA with post hoc Tukey’s test. P values of < 0.05 (*), < 0.01 (**), and < 0.001 (***) were considered significant.

Four methods were examined: Pre-vascularization of subcutaneous tissue using temporary placement of either a gelatin hydrogel sustained-release device containing basic fibroblast growth factor (FGF) with heparin (GEL) (15) or of a medically approved vascular access catheter (deviceless, DL) (16); graft siting in ISWAT (17); and graft siting in a relatively AR beneath dorsal skin. We hypothesized that blood supply would determine graft fate. AR siting served as a control for the 3 other options. Adipose tissue is inherently vessel-rich, while the GEL and DL pre-vascularization methods increase blood supply to the graft site. However, those methods require invasive manipulation 2wk or 1mo, respectively, before transplantation, potentially stressing hypopituitary mice. Plasma ACTH levels 1mo after transplantation of hESC-derived POs were highest in ISWAT-cohort mice (ACTH levels in POs culture medium 30100 pg/ml; all transplanted POs from the same lot). ISWAT transplantation thus appeared best ( Figure 1 , Supplementary Figure 4 ).

ISWAT group (n=6), sham-operated group (n=6), and AR group (n=5) mice, as stated above, served as controls for higher-vascularity ISWAT work, with follow-up of plasma ACTH levels for 6mo. Pre-transplant basal plasma ACTH levels and CRH-stimulated plasma ACTH levels did not significantly differ among the 3 groups (ISWAT vs. sham vs. AR, basal; 4.6 ± 1.8 pg/ml vs. 2.3 ± 1.6 pg/ml vs. 2.5 ± 1.1pg/ml, p = 0.27 – 0.92, stimulated; 1.7 ± 0.8 pg/ml vs. 1.7 ± 2.3 pg/ml vs. 7.2 ± 2.1 pg/ml, p = 0.06 – 0.72).

After transplantation, basal plasma ACTH levels (“basal”, Figure 2A ) consistently were higher in the ISWAT group than in the sham-operated group. At 2, 4, 8, and 17wk after transplantation, ACTH values differed significantly between the groups (p < 0.001 – 0.05, Figure 2A ). CRH-stimulated plasma ACTH levels (“stimulated”, Figure 2A ) also were higher in the ISWAT group than in the sham-operated group, with statistically significant differences between the groups at 2, 4, 8, 21, and 26wk (p < 0.001 – 0.005, Figure 2A ).

Pre-transplant ACTH secretion in vitro, as assessed by ACTH levels in POs culture medium, did not differ significantly between the ISWAT and AR groups (ISWAT vs. AR, 44416 ± 8435 pg/ml vs. 45800 ± 17291 pg/ml, p=0.876); i.e., ACTH secretory capacity of transplanted POs was similar between groups. Basal plasma ACTH levels in intact mice were 182 ± 40.7 pg/ml and CRH-stimulated plasma ACTH levels were 278.6 ± 43.2 pg/ml (n=6, Figure 2B ). After transplantation, plasma basal ACTH levels and CRH-stimulated ACTH levels were higher in the AR group than in the sham-operated group and lower than in the ISWAT group, with ISWAT plasma basal ACTH levels significantly higher than AR levels at 2wk (p = 0.009) and ISWAT plasma CRH-stimulated ACTH levels significantly higher than AR levels at 2, 8, and 21wk (p < 0.001 – 0.013). These results indicated that POs in subcutaneous tissues released ACTH more efficiently when implanted in well-vascularized sites such as adipose tissue than in non-vascularized sites.

Whilst running wheel testing found greater activity in the ISWAT group than in the sham-operated group, the ISWAT group was slightly less active than the intact group ( Figure 2C ). The rate of weight loss in the ISWAT group was modest by comparison with that in the sham-operated group, but still > 10% ( Figure 2D ).

On macroscopy 4wk after ISWAT transplantation, graft neovascularization was apparent ( Figure 3A ). At 21wk, no adipose tissue was observed macroscopically, but the grafts appeared intact ( Figure 3B ). On microscopy of the skin and subjacent grafts harvested en bloc at 21wk, transplanted cell aggregates lay within subcutaneous tissue ( Figures 3C, G ). Fluorescence immunomicroscopy revealed that the grafts expressed ACTH; E-cadherin, a marker of oral ectoderm; and LHX3, a pituitary-progenitor marker ( Figures 3D, E, H, I ). Simultaneous expression of human nuclei indicated that the cells in question were transplanted hESC-derived pituitary cells ( Figure 3I ). Furthermore, SMA, a vessel-wall marker, was expressed as clusters around and within the grafts ( Figures 3F, J ), indicating neovascularization. These observations indicate that the hESC-derived POs engraft and function in vivo. To confirm by microscopy that POs were present was not possible in AR-group mice that secreted ACTH poorly, whereas in an AR-group mouse with relatively good ACTH secretion tissue with engrafted POs was found on H&E staining. Fluorescence immunomicroscopy also demonstrated reactivity, although fewer cells marked than in ISWAT material ( Supplementary Figure 5 ).

Acknowledging that hESC-derived POs function after subcutaneous transplantation, the question arose of how vascularization affects engraftment of POs transplanted subcutaneously. Vascular endothelial growth factor (VEGF), FGF2, and angiopoietin 2 (ANGPT2) are related to early-stage angiogenesis (18–20). We evaluated by qPCR whether the POs expressed these genes. Pre-transplant POs, POs harvested 12h after transplantation, and undifferentiated hESCs as controls were analyzed. Expression levels of VEGFA, VEGFB, VEGFC, and ANGPT2 in the POs were significantly higher than those in undifferentiated hESCs. Expression levels of VEGFC and ANGPT2 in particular rose after transplantation ( Figures 4A–E ). These results suggested that the POs themselves express angiogenic factors, which might promote engraftment into subcutaneous tissue.

Finally, to assess further the importance of angiogenesis in survival of subcutaneous POs grafts, we examined POs function in vivo by administering bevacizumab, a VEGF inhibitor (2mg/kg, i.p., 2x/wk for 2wk (21); Selleck, Houston, TX). Hypopituitary SCID mice were divided after ISWAT transplantation into a bevacizumab administration group (n=9) and a vehicle administration group (n=9). Whilst ACTH levels in the culture medium of POs used in grafting did not differ significantly between the groups (bevacizumab vs. vehicle, 19445 ± 6774 pg/ml vs. 20295 ± 6610 pg/ml, p=0.93), plasma ACTH levels after CRH stimulation were significantly lower in the bevacizumab group than in the vehicle group (4.6 ± 3.0 pg/ml vs. 34.5 ± 11.7 pg/ml, p=0.035). We infer that angiogenesis is important for engrafted-POs functionality in ISWAT ( Figure 4F ).