Propagation of human prostate tissue from induced pluripotent stem cells

Abstract Primary culture of human prostate organoids and patient‐derived xenografts is inefficient and has limited access to clinical tissues. This hampers their use for translational study to identify new treatments. To overcome this, we established a complementary approach where rapidly proliferating and easily handled induced pluripotent stem cells enabled the generation of human prostate tissue in vivo and in vitro. By using a coculture technique with inductive urogenital sinus mesenchyme, we comprehensively recapitulated in situ 3D prostate histology, and overcame limitations in the primary culture of human prostate stem, luminal and neuroendocrine cells, as well as the stromal microenvironment. This model now unlocks new opportunities to undertake translational studies of benign and malignant prostate disease.

patient-specific drug testing to more accurately guide outcomes are lacking. 1 Encouragingly, the capacity to generate in vitro 3D organoid cultures is transforming the study of human diseases. 2 These structures faithfully mimic in vivo epithelial architecture and present novel opportunities for preclinical studies. 3,4 However, the widespread adoption of organoid culture in prostate studies is hampered by inherent shortcomings, including limited access to patient samples and the inefficient establishment of cancer organoid cultures. These issues also apply to the other established approach of patient-derived xenografts (PDXs). 5 Previously, successful organoid cultures were solely restricted to advanced metastatic tumors 3 ; however, recent advances have included the addition of stromal coculture to sustain organoids derived from localized cancers. 6 In cases where longer-term cultures are established, an emerging understanding of the substantial genotypic and phenotypic drift that occurs through in vitro culture adaptation restricts their translational value. 7 Approaches that allow robust isogenic models of cancer are required 8 and the generation of tissue from pluripotent stem cells appears to be a suitable alternative.
Human embryonic stem cells (ESCs) through coengraftment with rodent urogenital sinus mesenchyme (UGM) can generate prostate tissue in vivo. 9 However, current in vitro human prostate organoid approaches, from either tissue-derived cells or ESCs, do not fully recapitulate the full breadth of in situ prostate differentiation as they do not contain neuroendocrine (NE) cells. 4,10 Of note, emerging data show that NE differentiation drives treatment-resistant prostate cancer. 11 Furthermore, alternatives to ESCs would avoid significant ethical and regulatory restrictions and also enable greater access to organoid generation to groups worldwide. The use of induced pluripotent stem cells (iPSCs) is becoming increasingly established in generating tissues from many organs for translational study, 12 but surprisingly, for the study of the most common male cancer, prostate cancer, the development of such tools remains lagging. We had previously shown the ability to reprogram human prostate cells to provide an easy-to-handle and rapidly proliferating source of cells delivering a solution to the problems of limited input from primary biopsies, restrictive ethics and lack of access to patient biopsies. 13 Thereafter, iPSC lines have also been generated from human fetal prostate fibroblasts, prostate cancer-associated fibroblasts, and basal prostatic epithelial cells, providing further useful tools to study normal prostate development and prostate disease. [14][15][16][17] Herein, we demonstrate for the first time that tissue recombinants comprising human iPSCs and rat UGM generated both in vivo xenografts and in vitro prostate organoids that recreated the full breadth in situ prostate epithelial differentiation, including NE cells, as well as the stromal compartment. Details of patients from whom iPSC lines were generated are described in Table S1.

| Human iPSC generation
iPSC lines were generated from three patients (Table S1). The reprogramming efficiency was 0.02%. For each patient, seven clones were characterized and validated for characteristic ESC marker expression and functional pluripotency in generating all three germ-layer lineage (see Figure S1 for example of such characterization and validation for patient 13671 clone 1). Human prostate cell culture, characterizations by realtime polymerase chain reaction (PCR), DNA fingerprinting, karyotyping, immunofluorescence, alkaline phosphatase staining, and assays of pluripotency (embryoid body formation and teratoma formation) were described previously. 13 Three representative clones from these patients were taken forward and subsequent data were generated.

Significance statement
Growing cells from prostate cancer biopsies in the laboratory to study mechanisms of disease and to discover new treatments is fraught with difficulties and often not possible.
This work establishes a new means to grow "mini 3D prostates" in the laboratory. It shows proof of concept that genetic modifications are possible in this innovative model, which lays the foundations for new preclinical approaches to personalized care previously considered too challenging.
Specifically, in future work, one can develop genetically engineered prostate cancers in a dish, tailored to the specific genetic profiles of individual patients, and determine their best response to a range of drug treatments.
(Corning, New York) coated plates in mTeSR1 medium (STEMCELL Technologies, Vancouver, Canada). The medium was changed every 48 hours.

| Definitive endoderm induction
To differentiate iPSCs into definitive endoderm (DE) cells, we modified the previously reported DE induction protocol, 18 Figure S4).

| Athymic nude mouse host xenografting
Male athymic nude mice (Hsd:Athymic Nude-Foxn1nu; Charles River Laboratories, Wilmington, Massachusetts) aged 10 weeks were used for subrenal capsule grafts. Following castration, a 1-cm skin incision along the dorsal midline was made, followed by another incision (~6-8 mm) of the body wall along the line of fat which runs parallel to the spine immediately above the kidney area. Then the kidney was exteriorized and a capsulotomy was made to prepare the subcapsular space for the grafts. Grafts were then placed underneath the renal capsule and maneuvered into various locations along the kidney. Two grafts were placed into each kidney (upper and lower poles), which was then reintroduced back into the mouse. Surgical incisions were closed with suture (body wall) and staples (skin). A testosterone pellet (25 mg) was inserted s.c. into the scruff of the neck. 21

| Xenograft harvest and processing
Hosts were sacrificed 6 weeks after grafting by anesthetic (Penthobarbital) overdose followed by cervical dislocation. Grafts were harvested, and kidneys removed en bloc. Whole kidneys were placed in 10% neutral buffered formalin (Sigma-Aldrich) for 24 hours.
After fixation, kidneys were processed, paraffin was embedded, and sections were cut at 5 μm for Haemotoxylin and Eosin (H&E) staining and immunohistochemistry (IHC). 21  Ham containing 2% ITS and 10 nM DHT as previously used for successful in vitro culture of UGS. 22 The media was further supplemented with 1 μg/mL ROCK inhibitor Y-27632 (STEMCELL Technologies) and replaced every 48 hours. From 2 weeks, the media was changed to prostate organoid medium. 4 Wells were harvested 6 weeks onward for histology and RNA extraction. Wells for histology were removed as a Matrigel plug, fixed in 10% formalin overnight, and processed before embedding into paraffin. For RNA extraction, Matrigel was digested by incubation with dispase (STEMCELL Technologies) at 37 C until the gel was completely dissolved. The mixture was gently pipetted to further break up the Matrigel, and transferred to an Eppendorf for centrifugation at 2000 rpm for 5 minutes. The supernatant was removed and the pellet snap frozen in isopentane (Radnor, Pennsylvania) and stored at −80 C.

| Coculture of human iPSCs with rat UGM cells
The efficiency of prostate organoid generation is 100% using an input of 1 × 10 4 DE cells (with 1:3.5 UGM cells). We were unable to generate prostate organoids using 5 × 10 3 or less DE cells. Given the inability to utilize <5 × 10 3 DE cells but achievement of 100% efficiency at the higher density, we did not proceed with exploring further resolution between these densities. At higher ratios, organoids began to merge (Table S2).

| Immunohistochemistry
IHC was performed on formalin-fixed paraffin-embedded (FFPE) sections (4 μm) that were initially deparaffinized and hydrated. Microwave antigen retrieval was carried out with citrate buffer pH 6 to unmask surface antigens. Endogenous peroxidase activity was removed by blocking with 3% H 2 O 2 (Sigma-Aldrich). Sections were then blocked in horse serum (Vector Laboratories, Burlingame, California) and incubated in primary antibody overnight 4 C. The antibodies used were antihuman mitochondria (1:200, Abcam, Cambridge). This is a human specific antibody used in xenographic model research [23][24][25]

| RNA sequencing analysis
Total RNA was extracted from cells using Ribozol RNA extraction reagent (Amresco, Solon, Ohio) following manufacturer's instructions.

RNA-Seq library construction and sequencing was performed at
Otogenetics Corporation (Atlanta, Georgia) according to standard protocols. The resulting RNA-seq fastq reads were aligned to Hg19 (GRCh37) using STAR 26 and mapped to genes using HTSeq counts (http://htseq.readthedocs.io/en/master/count.html). Normalized count and differential expression analysis data were generated using DESeq2. 27 Gene Set Enrichment Analysis (GSEA) 28 3 | RESULTS

| Generation of human iPSC-derived prostate tissue in vivo
First, as the tissue of origin used to generate iPSCs can affect subsequent differentiation, 30 we used a modified integration-free Sendai virus approach to reprogram human prostate cells 13 ( Figure S1). which is driven by inductive UGM, we undertook subrenal capsule coengraftment of iPSCs with UGM in nude mice ( Figure S4). 31 This resulted in formation of prostatic tissue by 12 weeks (Figure 1), as previously also shown with ESCs. 9 Grafts comprehensively recreated typical human prostate tissue histology, consisting mainly of glandular structures surrounded by myofibroblasts ( Figure 1A,B). The human origin of the epithelial cells was verified by immunolocalization with antihuman mitochondria detection ( Figure 1C) and expression of cytokeratins CK8/CK18 on the cell surface and nuclear p63 demonstrated stratification of epithelium into characteristic prostate luminal and basal cells, respectively ( Figure 1D,E). AR is an essential driver of prostate differentiation and both nuclear and cytoplasmic expression of the receptor was demonstrated ( Figure 1F). Terminal differentiation was confirmed by nuclear localization of prostate specific NKX3.1 32 and secretory PSA in luminal epithelial cells-again confirming the human nature of the tissue ( Figure 1G,H). Critically,

Chromogranin A (ChrA) expression identified infrequent NE cells indi-
cating this methodology recapitulated the full breadth of prostate epithelial differentiation ( Figure 1I).

| In vitro human iPSC-derived prostate organoids recapitulated the full breadth in situ prostate epithelial differentiation
Because UGM induced prostate differentiation from iPSCs in vivo, we hypothesized it may also direct prostate differentiation in vitro and proceeded to employ a similar 3D coculture methodology (Figure 2A).
During embryogenesis, the prostate gland is derived from the endo-  Figure 2A,B). These cells were identified as human by antihuman mitochondria and human specific PSA staining ( Figure 2C,G). Expression of CK8/18 and p63 appropriately localized to luminal and basal located epithelial cells, respectively ( Figure 2D,E, Figure S7A).
Furthermore, luminal epithelial cells expressed nuclear AR and the terminally differentiated nature of the organoids was confirmed by secretory PSA (Figure 2F,G). Previously, NKX3.1, along with AR, were shown to be the essential master regulators of prostate specific differentiation in mice 34 and we confirmed these expressions in our human prostate organoids ( Figure 2H). Sporadic NE cells were again identified by ChrA ( Figure 2I) and additionally by Enolase 2 and Synaptophysin expression ( Figure S7B). These data demonstrate that similar to the in vivo grafts, the in vitro organoids also faithfully recreated the full breadth of in situ prostate epithelial differentiation and represents a major breakthrough in establishing an easily accessible preclinical model. Of interest, co-staining of ChrA with Ki67 to determine the proliferative status of NE cells did not reveal coexpression ( Figure S8).
Additionally, the somatic stem cell enrichment marker DLK1, known to mark basal cells essential for long term maintenance of prostate epithelium, was expressed within a subpopulation of cells (3.0 ± 1.3%, n = 650 cells, n = 3 organoid cultures) ( Figure 2J). 35 The summary statistics are presented, from a total of 183 organoids (n = 3 clones and n = 3 separate experiments), confirming reproducible and appropriate spatially restricted expression of basal and luminal specific markers ( Figure 2K). Also, following cellular disaggregation for passage beyond 12 weeks, 3D culture led to branching ductal structures ( Figure 2L), therefore fully recapitulating the human prostate histology.  (Tables S3-S5).

|
GSEA confirmed the mature organoids shared the terminally differentiated transcriptomic identities of the benchmark primary adult cells ( Figure 3A,B). Mature organoid formation was also characterized by enrichment of NE marker expression ( Figure 3C). Additionally, pathway ontology analyses revealed new insights into the mechanisms of differentiation, such as p53, inflammation and Myc related pathways, known to be central in cancer biology and provides focus for future studies ( Figure 3D). 37,38 3.4 | Prostate iPSCs generated a self-maintaining stromal compartment in mature prostate organoids In the iPSC-derived prostate tissue xenografts, glands were surrounded by "nests" of cells with classical stromal morphology ( Figure 1A). These were demonstrated to be mesenchymal cells as shown by α-SMA expression and also of human origin as evidenced by expression analyses of the human specific antimitochondria marker ( Figure 4A). These data show that human iPSCs also generated the stromal compartment. In the initial stages of differentiation, the UGM cells, whose inductive properties are transient, 39

| DISCUSSION
In this report, we show for the first time that prostate iPSCs enabled generation of human prostate tissue both in vivo and in vitro. Using UGM-directed differentiation, human iPSC-derived prostate tissue comprehensively recapitulated in situ prostate histology and the full breadth of prostate specific epithelial differentiation-including NE cells. The archetypal histology of the human prostate is described by extensive acinar-tubular branching. Our work has shown that iPSC-derived organoid cultures also recapitulate this characteristic branching morphology, which is likely to be related to the stromal compartment (derived from the iPSCs) that is known to support branching. 6 The impact of NE differentiation on prostate cancer progression and driving treatment resistance has recently gained significant attention. 11

CONFLICT OF INTEREST
The authors declared no potential conflicts of interest.

AUTHOR CONTRIBUTIONS
A

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.