Pituitary stem cells: past, present, and future perspectives

Abstract

Pituitary cells that express the transcription factor SOX2 are stem cells because they can self-renew and differentiate into multiple pituitary-hormone-producing cell types as organoids. Wounding and physiological challenges can activate pituitary stem cells, but cell numbers are not fully restored, and the ability to mobilize stem cells decreases with increasing age. The basis for these limitations is still unknown. The regulation of stem cell quiescence and activation involves many different signalling pathways, including those mediated by Wnt, Hippo and several cytokines; more research is needed to understand the interactions between these pathways. Pituitary organoids can be formed from human or mouse embryonic stem cells, or from human induced pluripotent stem cells. Human pituitary organoid transplantation is sufficient to induce corticosterone release in hypophysectomized mice, raising the possibility of therapeutic applications. Today, pituitary organoids have the potential for assessing the role of individual genes and genetic variants on hormone production ex vivo, providing an important tool for advancing the exciting frontiers of pituitary stem cell biology and pituitary organogenesis. In this article, we provide an overview of the notable discoveries in pituitary stem cell function and highlight important areas for future research.

Introduction

The pituitary gland is a neuroendocrine gland situated at the base of the brain; it responds to input from the hypothalamus and regulates the function of many peripheral organs by producing and releasing hormones that influence growth, lactation, reproduction, metabolism and the stress response. Accordingly, congenital and acquired pituitary hormone deficiencies (hypopituitarism) can cause a range of clinical features, including growth retardation, and, in severe untreated cases, death from adrenal crisis or severe hypoglycaemia¹. The pituitary gland comprises a posterior lobe, which contains pituicytes (glial cells) together with the axon terminals of neurons that secrete oxytocin and vasopressin, and an anterior lobe, which contains specialized cells that produce growth hormone (somatotropes), prolactin (lactotropes), thyroid-stimulating hormone (TSH; thyrotropes), adrenocorticotropin (corticotropes) and the gonadotropins, luteinizing hormone (LH) and follicle-stimulating hormone (FSH) (gonadotropes) (Fig. 1). Rodent pituitary glands have a distinct intermediate lobe that contains the cells that process pro-opiomelanocortin to produce melanocyte-stimulating hormone and β-endorphin (melanotropes), but this is less distinct in humans.

Figure 1. Anatomy of the rodent pituitary gland, base of the hypothalamus, and supporting vasculature.

a, The pituitary gland is located at the base of the hypothalamus (shown in sagittal plane with rostral toward the left and caudal to the right). b, The axon terminals containing vasopressin and oxytocin secretory granules project to the posterior lobe of the pituitary gland. The intermediate and anterior lobes contain specialized cells that produce hormones and release them in response to neural inputs from the paraventricular and arcuate nuclei of the hypothalamus. c, Hypothalamic regulatory factors are released into the hypophyseal portal circulation and are transported to the anterior pituitary by portal veins. Pituitary hormones are transported to the peripheral organs, and homeostasis is maintained by feedback regulation to the hypothalamus and pituitary gland by the hypothalamic artery and superior hypophyseal artery, respectively.

The pituitary gland derives from several embryonic tissues. An invagination of oral ectoderm forms the pituitary primordium, Rathke’s pouch, which develops into the anterior and intermediate lobes. An evagination of neural ectoderm produces the pituitary stalk and posterior lobe (Fig. 2). (See Masterclass in Pituitary Development link 1².) Head mesenchyme and neural crest contribute to the pituitary gland and the vasculature^3–6. Rathke’s pouch is evident at embryonic day (E) 10.5 in mice and 5 weeks gestation in humans⁷. (See Human Pituitary Development link 2 ⁸). Stem cells delaminate from the residual cleft of Rathke’s pouch and develop into anterior pituitary hormone-producing cells⁹. In rodents, by late gestation, each of the different pituitary hormone-producing cell types is detectable, but the organ continues to grow and develop postnatally, and the proportions of the different cell types change substantially¹⁰. Two hypothalamic releasing hormones, growth hormone-releasing hormone (GHRH) and thyrotropin-releasing hormone (TRH), stimulate the expansion of hormone-producing cells in the postnatal period^11,12. In humans, the pituitary gland has developed morphologically by 13 weeks, and it continues to grow until 21 weeks⁷.

Figure 2. Pituitary development and stem cells.

a, Mid-sagittal sections from embryonic (E) day E10.5–14.5 mouse foetuses (rostral is to the left; caudal, right) and coronal sections from E17.5 were stained with eosin and haemotoxylin. The neural ectoderm (NE) evaginates to form the infundibulum (INF), which develops into the posterior lobe (PL). Rathke’s pouch (RP) develops from the oral ectoderm (OE) and forms the intermediate (IL) and anterior lobes (AL). III = third ventricle of the brain. b, Stem cells are marked by immunostaining for SOX2, and at E12.5 and E13.5, most of the cells in Rathke’s pouch are stem cells. The differentiating cells (D) are ventrally located. Some co-express SOX2 and PROP1, and at E13.5 most of the cells expressing only PROP1 are transitioning (T) to differentiation. At postnatal (P) day P0, some SOX2 expressing cells reside in the marginal zone (MZ) and others are clustered in the parenchyma (P). Images courtesy of M. L. Brinkmeier, Department of Human Genetics, University of Michigan, USA.

Signalling pathways are important for the formation and growth of Rathke’s pouch. Sonic hedgehog (SHH) expression is evident in the oral ectoderm except for the placode that thickens to produce Rathke’s pouch¹³. SHH signalling induces the expression of the LIM homeobox transcription factor LHX3, indicating commitment to pituitary fate. The growth of Rathke’s pouch is dependent on signalling induced by bone morphogenetic protein (BMP) and fibroblast growth factor (FGF) derived from the ventral diencephalon and developing posterior lobe¹⁴. Canonical WNT signalling via β-catenin is also necessary for the growth of Rathke’s pouch, as well as to stimulate evagination of the neural ectoderm³. Notch signalling regulates cell cycle exit¹⁵, and Hippo signalling regulates the balance of stem cell proliferation versus differentiation¹⁶.

Stem cells are defined as cells that can proliferate to make more undifferentiated stem cells while maintaining multi-potency (self-renewal) and can also differentiate into multiple cell fates. Embryonic stem cells (ESCs) are derived from the inner cell mass of the embryo at the blastocyst stage, while tissue stem cells reside in niches within an organ (See Box 1). Both embryonic and tissue stem cells can be propagated in culture as a non-adherent three-dimensional mass and, with the appropriate signalling factors (a mass of organ-specific differentiated cells grown from stem cells in culture that self-organize to resemble the organ). In this Review, we provide a historical perspective of progress that has been made in pituitary stem cell biology^17–19. Methods for generating pituitary organoids are presented, and the evidence for cancer stem cells in the pituitary gland is discussed. Finally, future directions for exploring pituitary stem cell regulation are outlined.

BOX 1 |. Signalling in pituitary stem cell niches.

In general, stem cell ‘niches’ are areas that contain stem cells, committed progenitors, differentiated progeny, and a variety of supporting cells that interact through cell contact and the extracellular matrix. Cell signalling within the niche is important for regulating the activity of stem cells (that is, determining whether they are. quiescent or proliferative), and the differentiation processes. In the postnatal mouse pituitary, the marginal zone and the clusters of stem cells in the parenchyma of the anterior lobe are both considered stem cell niches. The stem cells secrete WNTs to promote the proliferation of neighbouring cells into hormone-producing cells¹⁰³. The Hippo signalling cascade is active in stem cells, as evidenced by expression of the kinase LATS1 and its effectors YAP and TAZ¹⁶. There is evidence for the involvement of Notch, bone morphogenetic proteins, ephrins, cytokines and chemokines in stem cell niches postnatally^{51,91,93 111}, but additional work needs to be done to clarify the roles and interactions between these pathways (reviewed in^17,19).

The discovery of pituitary stem cells in rodents

Evidence for the activity of a progenitor cell population in the pituitary gland

The mature pituitary gland displays low cell turnover and slow regenerative capability. For example, in adult rats, the prevalence of mitotic and apoptotic cells is only 0.066% and 0.03%, respectively²⁰. Drug-induced near-total ablation of growth-hormone (GH)-producing cells during embryogenesis is associated with dwarfism at 4 weeks and profound pituitary hypoplasia²¹. This system uses synthetic nucleosides, like FIAU (1-(2-deoxy-2-fluoro-beta-delta-arabinofuranosyl)-5-iodouracil) to block DNA synthesis and kill cells expressing a herpes simplex virus thymidine kinase transgene. In mice treated with FIAU from postnatal day 2 through to 8 weeks, the rate of GH-producing cell regeneration was slow; 6 weeks after drug withdrawal, new GH-producing cells were evident, indicative of partial regeneration. Vankelecom and colleagues ablated GH-producing cells in 8–12 week-old adult mice by treating transgenic mice expressing an inducible diphtheria toxin receptor with diphtheria toxin²². They demonstrated restoration of approximately 50% of GH-producing cells after 4–5 months, consistent with a slow recovery rate. The researchers extended the previous work by showing that the pituitary stem cell population increased during GH-producing cell regeneration. Although regeneration was not complete, complete regeneration might not be necessary because approximately 40–70% of GH-producing cells can be ablated in mice during embryogenesis without causing notable growth insufficiency postnatally, suggesting that full pituitary-hormone-producing capacity is not needed for normal growth²³.

Pituitary cells that are resistant to histological stains are referred to as chromophobes and were thought to be stem cells^24,25. They were first described as folliculo-stellate cells based on their appearance in electron micrographs of rat pituitary gland^26,27. They constitute a network of agranular cells adjacent to granular secretory cells, with cytoplasmic processes organized around cavities or follicles. These folliculo-stellate cells increased in number after hypophysectomy²⁴. Their multipotency was demonstrated by transplanting them into hypophysectomized rats and observing their differentiation into acidophils (GH-producing and prolactin-producing cells) and basophils (corticotropes, thyrotropes and gonadotropes)^24,25. The folliculo-stellate cells had the highest mitotic indices of pituitary cells in young rats, and the presence of dividing cells decreased with age²⁸. These cells could differentiate into new endocrine cells but rarely proliferated in adult rats; their proliferation in adulthood could be associated with tumour formation. For many years, these chromophobes had no molecular markers and were classified by histological means, hindering deeper dissection of their biology in vivo.

Identification of a stem cell population in the pituitary gland

In 2005, Thomas and colleagues identified a colony-forming cell population that accounted for 0.2% of total adult mouse pituitary cells²⁹. These pituitary colony-forming cells had high proliferation rates and consisted of two morphologically distinct cell types: folliculo-stellate cells, as described previously^26,27 and round, refractile somatotrope-like cells. Folliculo-stellate cells stain positive for S100β and glial fibrillary acidic protein (GFAP) and take up the fluorescent dipeptide β-Ala-Lys-N-epsilon-7-amino-4-methylcoumarin-3-acetic acid (AMCA)³⁰. AMCA-positive cells enriched by fluorescence-activated cell sorting were able to form colonies and produce GH, while AMCA-negative cells could not. Thus, pituitary colony-forming cells constitute a subpopulation of folliculo-stellate cells. This notion was supported by the observation that 3.3% of AMCA-positive, GH-negative cells injected into mice could differentiate and express GH after 6 weeks in vivo³¹. These studies confirmed the progenitor potential of a subpopulation of folliculo-stellate cells, based on their ability to form colonies in vitro and to differentiate in vivo.

Studies in many organ systems have revealed that normal stem cells and cancer stem cells differ from the main cell populations by exhibiting rapid efflux of Hoescht 33342 dye. Vankelecom and colleagues used fluorescence-activated cell sorting to identify a ‘side population’ of pituitary cells with this characteristic³²; approximately 1.7% of adult mouse anterior pituitary cells were in this side population. These cells expressed markers generally associated with stem cells, including SOX2 and SOX9, as well as components of the Notch, SHH and Wnt signalling pathways, which are involved in stem cell renewal and fate determination, and could form non-adherent spheres in culture^22,33,34.

SOX2 expression in pituitary stem cells

In 2008, Fauquier and colleagues characterized a small population of SOX2-expressing cells in the adult mouse pituitary gland³⁵. The SOX2-positive cells also expressed SOX9, nestin, and the folliculo-stellate cell marker S100β (although not all S100β−expressing cells express SOX2). The SOX2-expressing cells could form pituispheres (three-dimensional organoids grown in culture that contain hormone-producing cells), self-renew into secondary pituispheres, and differentiate into each of the pituitary lineages, whereas SOX2-negative cells could not. SOX2, OCT4 (POU5F1), KLF4 and c-MYC are sufficient to induce pluripotency in different kinds of differentiated cells, and they are among the factors regulated by leukaemia inhibitory factor (LIF) to maintain pluripotency during normal development³⁶. This broad functional role for SOX2 as a pluripotency-inducing factor supports the idea that it has a key functional role in pituitary stem-cell multipotency, rather than simply being a marker of the stem-cell reservoir.

In the adult mouse pituitary gland, SOX2⁺ cells account for 3–5% of the cells in the anterior lobe. These cells are in small clusters scattered throughout the parenchyma of the anterior lobe and at the edge of the residual Rathke’s cleft, known as the marginal zone (Fig. 2). The marginal zone is at the boundary between the intermediate and anterior lobes of the pituitary gland, which are derived from the rostral and caudal aspects of Rathke’s pouch. SOX2⁺ cells have high proliferation rates early in embryogenesis and low rates in adults. An analysis of rat pituitary cell gene expression using single cell RNA sequencing revealed that 15% of pituitary cells expressed SOX2; these cells included, but were not limited to, folliculo-stellate cells³⁷.

Several studies have demonstrated the utility of SOX2 as a pituitary stem cell marker. In 2012, Vankelecom and colleagues showed that ablation of adult mouse GH-producing cells using an inducible diphtheria toxin receptor transgene was associated with a subsequent increase in the number of SOX2⁺ cells in the marginal zone and the regeneration of GH-producing cells over a 5-month period²². This ‘wounding’ paradigm demonstrated the capacity of adult pituitary stem cells to regenerate ~40% of somatotropes; the findings were similar for the regeneration of lactotropes following their ablation³⁴. Regeneration was still incomplete after an extended period, however, and the ability to form colonies declined with age^38,39. The failure of aged pituitary stem cells to respond to wounding might be caused by the presence of a pro-inflammatory state in the aged gland, like that observed for muscle stem cells^40–42.

Lineage tracing demonstrates that SOX2 is a bona fide pituitary stem cell marker

In 2013, Andoniadou and Martinez Barbera used genetic lineage-tracing in mice to show that SOX2⁺ stem cells from embryonic and adult mouse pituitary glands could self-renew and differentiate into each hormone-producing cell type⁴³. They also showed that lineage-traced cells gave rise to differentiated cells more frequently in mice approaching adulthood (4–6 weeks) than in older mice (8–14 months). Rizzoti and colleagues used similar approaches to demonstrate that SOX2⁺ cells and SOX9⁺ cells could form pituispheres and differentiate into hormone-producing cells but the SOX2^– cells could not⁴⁴. Adrenalectomy, gonadectomy or both induced an increase in pituisphere formation in vitro and demonstrated that stem cells were mobilized to produce new corticotropes and gonadotropes, although the number of new differentiated cells was modest⁴⁴.

Vankelecom’s group showed that ablation of 80% of SOX2⁺ cells in adult mice did not result in any regeneration of stem cells, even after 6 months⁴⁵. Moreover, this extent of SOX2 ablation did not decrease the hormone-producing cell number or affect pituitary exerted homeostasis. Thus, a robust stem cell pool is not required for the maintenance of normal pituitary function in rodents.

Additional pituitary stem cell markers in rodents and humans

In 2016, Pérez Millán et al. discovered that PROP1 was involved in the differentiation of the SOX2⁺ stem cell population in mice⁴⁶. PROP1 directs an epithelial-to-mesenchymal (EMT)-like transition process in pituitary stem cells (see later), which is necessary for differentiation. During mouse embryogenesis, Prop1 is co-expressed with Sox2. Window labelling experiments demonstrated that Prop1 is required for stem cells to move out of their non-endocrine cell niche in the pituitary into the parenchyma of the gland⁹. It is not known whether this is an active or passive process; however, this event is associated with upregulated expression of the transcription factor Zeb2 and N-cadherin (Cdh2), which is typical of EMT in many tissues. Classic EMT requires modification of the adhesion molecules expressed by the cells to allow the cells to adopt a migratory and invasive behaviour. The transcription factor SNAI2 (SLUG) is known to suppress the expression of E-cadherin (Cdh1). Prop1 is required for appropriate Snai2 expression in the developing mouse anterior pituitary at E14.5⁴⁷.

Kato and colleagues identified the tetraspanin CD9 as a novel stem cell marker that is co-expressed with S100β and SOX2 in cells in the marginal zone and parenchymal stem cell clusters of the anterior lobe of the rat pituitary⁴⁸. As well as being able to differentiate into hormone-producing cells, these CD9⁺/S100β⁺/SOX2⁺ cells can differentiate into endothelial cells, thus contributing to neovascularization in a prolactinoma model^49,50.

Kato and colleagues also demonstrated that ephrin B3 (Efnb3) and Eph B2 receptor (Ephb2) are both expressed in SOX2⁺ pituitary stem cells in the marginal cell layer and in the dense clusters of stem cells in the parenchyma of the rat pituitary^50,51. This cis interaction is postulated to be involved in the regulation and maintenance of these two stem cell niches. A trans interaction between Ephb2 in stem cells and various Efns expressed in differentiated cells may also be important⁵⁰. Tight-junction-related molecules such as zonula occludens and occludin have also been identified in pituitary stem cell clusters and hypothesized to have a role in pituitary gland architecture and remodelling⁵².

In mammals, Hippo signalling has important roles in stem cell regulation, lineage specification, and cell differentiation in many organ systems. High levels of Hippo signalling cause LATS kinases to phosphorylate the transcriptional co-activators YAP and TAZ, resulting in their cytoplasmic retention and degradation, and thereby maintaining growth restriction. By contrast, low levels of Hippo signalling promote the movement of YAP and TAZ into the nucleus, where they interact with transcription factors such as TEAD and Trithorax family members to induce the expression of genes involved in stemness, proliferation and anti-apoptosis. In this context, Andoniadou and colleagues demonstrated that LATS–YAP–TAZ signalling is essential for pituitary stem cell regulation in normal physiology, and its dysregulation can drive tumorigenesis¹⁶. Restricting YAP and TAZ activation during development is essential for the establishment of normal pituitary gland size and the specification of hormone-producing cells from SOX2⁺ pituitary stem cells. Postnatal deletion of LATS kinases and subsequent upregulation of YAP and TAZ leads to uncontrolled clonal expansion of SOX2⁺ stem cells and disrupts their differentiation, causing the formation of non-secreting, aggressive pituitary tumours. Sustained expression of YAP alone results in the expansion of SOX2⁺ pituitary stem cells that are capable of differentiation and devoid of tumorigenic potential, possibly owing to differences in the intensity of YAP activation between the two models.

The cluster of differentiation family of genes is expressed in juvenile and adult pituitary stem cells (Cd9 and Cd24)^48,49,53,54. These proteins are localized at the cell surface, interact with integrins, and can be used in the purification of stem cells. The SOX2, S100β-expressing cells in the rat pituitary gland express Cd9, and these cells can differentiate into endothelial-like cells in vitro^48,49. Lineage-tracing experiments are needed to determine whether these cells can give rise to hormone-producing cells in vitro and to the vasculature in vivo.

Advances in single-cell RNA sequencing technologies have facilitated the discovery of novel molecular markers of individual pituitary populations^{37,42,53–61}. (See single cell mouse pituitary sequence data from the postnatal period link 3 and adults link 4^53,54.) As expected, there are more proliferating stem cells in the neonatal pituitary gland than in the adult pituitary gland. Several different stem cell populations can be distinguished based on the expression levels of genes encoding keratin, the transcription factors SIX1 and SOX2, and numerous proliferation markers. The functional significance of these differences is not known. Although single-cell RNA sequencing has a limited capacity to detect genes with low levels of expression, it has detected transcripts for proteins predicted to be localized at the cell membrane. For example, transcripts for folate receptor 1 (Folr1) can be detected in both juvenile and adult pituitary stem cells^53,54. Folate (vitamin B9) is an important maternal nutrient during embryonic development, during which it plays a vital role in driving neural tube closure, but its role in the pituitary gland is largely unknown^62–64. Future studies are necessary to determine whether activation of FOLR1 or other receptors on stem cells affects pituitary stem cell proliferation and/or differentiation.

The human pituitary gland contains cells that express many of the same stem cell markers as the murine pituitary gland, such as SOX2, suggesting they are the equivalent cell type. The development of human and rodent pituitary glands is similar in terms of the signalling pathways involved and the stem cell markers expressed. Single-cell RNA sequencing experiments from human pituitaries at different embryonic ages revealed the human pituitary stem cell transcriptome to be similar to those of mouse and rat, including the expression of SOX2, PROP1 and genes involved in tight junctions and extracellular matrix interactions⁵⁹.

Regulation of the stem cell niche in vivo

EMT during pituitary organogenesis

During embryonic development, cells in many organs can dynamically transition from epithelial to mesenchymal states in a process known as EMT. EMT-like events are required for pituitary gland development^9,47. Identifying candidate genes involved in the regulation of the stem cell niche and the orchestration of EMT in pituitary stem cells is an important challenge [reviewed in¹⁹]. Pituitary stem cells in the marginal zone of Rathke’s pouch undergo an EMT-like process during their differentiation into all major hormone-producing cell types^9,46,65. This transition requires three major events. First, the expression of Cdh1, which encodes E-cadherin, is reduced at the ventral zone of Rathke’s cleft, whereas that of Cdh2, which encodes N-cadherin, is induced. Second, the stem cells acquire specific cellular markers that indicate the switch from a polarized, planar, shape to a rounded cell morphology. And third, these cells move into the parenchyma of the anterior lobe causing the organ to expand⁹.

SNAIL and SLUG, officially known as SNAI1 and SNAI2 respectively, belong to the Snail family of zinc-finger transcription factors that are involved in downregulating human and mouse Cdh1 and upregulating the expression of genes that encode metalloproteinases, chemokines and proteins characteristic of mesenchymal cells^66–68. A subpopulation of pituitary S100β⁺ cells with characteristics of neural crest cells expresses Snai2 in P3–P60 rats, with highest expression at P10⁶⁹. Snai2 can induce the expression of matrix metalloproteases and chemokines and bring about a change in cell morphology and proliferation.

ZEB1 and ZEB2 are members of the two-handed zinc-finger homeodomain transcription factors that are key regulators of EMT through the transcriptional repression of Cdh1 in several tissues⁷⁰. In mouse pituitary stem cells, Zeb2 expression is upregulated by Prop1. RNA sequencing experiments and reverse transcription quantitative real-time PCR analysis performed on stem cell colonies derived from Prop1-deficient mouse pituitaries showed a downregulation of Zeb2 and an upregulation of Cdh1 compared with wild-type colonies^46,71. Furthermore, silencing Zeb2 in vitro increases Cdh1 expression, and stem cell colonies exhibit an abnormal morphology as they fail to initiate EMT-like processes⁴⁶.

The highly conserved Twist genes (Twist1 and Twist2) encode basic helix–loop–helix transcription factors regulate morphogenesis in some systems⁷². Although the exact role of Twist in pituitary development remains unclear, the expression of both Twist1 and Twist2 is enriched in adult mouse pituitary stem cells⁷³. Consistent with this observation, Twist2 is expressed in the GHF-T1 cell line, which also expresses Pou1f1 but no hormones, but is not detectable in the TαT1 cell line, which expresses both Pou1f1 and TSH and resembles differentiated thyrotropes in the secretion of TSH in response to TRH, retinoids and diurnal cues^74–77. This observation suggests that Twist genes could be involved in EMT events in the pituitary.

The pituitary-specific transcription factor PROP1 promotes stem cell proliferation, the EMT-like process of cell migration, and the induction of Pou1f1 expression. POU1F1 is a transcription factor required for the differentiation of stem cells into somatotropes, lactotropes and thyrotropes (producing GH, PRL and TSH). The differentiation of corticotropes and gonadotropes, however, does not require Prop1. Mice lacking functional Prop1 (Prop1^df/df, S83P) have a smaller functional pool of stem cells than their wild-type counterparts, which correlates with the reduced expression of cyclin D1 and D2, persistent expression of SOX2, and the failure to activate Notch signalling^9,47,78. Prop1 mutants have a highly dysmorphic stem cell niche due to the failure of cell migration, and the anterior pituitary gland is hypoplastic owing to the failed differentiation of the Pou1f1 lineage⁴⁶. Unsuccessful EMT in Prop1 mutants is associated with the elevated expression of E-cadherin, claudins and keratins, and the reduced expression of Zeb2 and matrix metalloproteinases⁴⁶.

The chemokine (C-X-C motif) receptor 4 (CXCR4) and its ligand CXCL12 are important regulators of EMT-associated cell migration. Both Cxcr4 and Cxcl12 transcripts are enriched in pituitary stem cells^33,79. Treatment of primary rat pituitary cells with a CXCL12 inhibitor or antagonists of CXCR4 inhibits the extension of processes between folliculo-stellate cells, whereas administering CXCL12 to these cells increases their migration and invasion into Matrigel⁸⁰. CXCR4 signalling also activates the transforming growth factor-β (TGF-β) pathway, which inhibits lactotrope growth⁸¹.

Signalling pathways that regulate stem cell growth and differentiation

Signalling pathways mediated by BMPs, FGFs, NOTCH, WNT, SHH and Hippo regulate pituitary stem cell self-renewal, proliferation, and differentiation during organogenesis in rodents. It is not known whether these signalling pathways have different effects on stem cells after birth and/or in adult animals.

BMP

Endogenous BMP signals are important during embryonic development of the pituitary gland by inducing the formation of Rathke’s pouch, as demonstrated by animal models altering BMP signalling^14,82,83. Bmp2 can first be detected in the mesenchyme ventral to Rathke’s pouch at E10.5 but is found throughout the pouch at E12.5⁸². Bmp4 is expressed in the infundibulum from E11.5 until at least E14.5. Studies in homozygous Bmp4 null mouse embryos indicate that BMP4 signalling is necessary for the induction and formation of a pouch rudiment⁸³. The BMP inhibitors Noggin and Chordin refine the domains of BMP activity, influencing the patterning and growth of the pouch^14,84. Ectopic expression of Noggin prevents expansion of the pouch¹⁴. Accordingly, Noggin^–/– embryos show increased BMP4 activity and display several pituitary defects that range from a lack of a morphological Rathke’s pouch to the formation of secondary pituitary tissue⁸⁴.

FGF

Genetic experiments in mice demonstrated that FGF signalling is crucial for pituitary development, as FGF8 and FGF10 promote the initial growth as well as the survival of cells in Rathke’s pouch through FGFR2^83,85,86. In vitro studies using the FGF receptor antagonist SU5402 provided evidence that FGFs derived from the infundibulum are required for later aspects of pituitary development, including proliferation and differentiation of anterior pituitary cells⁸⁷. Expression of β-catenin in the neural ectoderm is necessary for the expression of Bmp4, Fgf8 and Fgf10, evagination of the neural ectoderm, and growth of Rathke’s pouch⁸⁸. Mice lacking both copies of the gene encoding the transcription factor OTX2 in the neural ectoderm show ablated FGF10 signalling in this region; the neural ectoderm fails to evaginate and growth of Rathke’s pouch is reduced⁸⁹. Disruption of OTX2 in patient-derived and control human induced pluripotent stem cells (hiPSCs) showed that hypothalamic OTX2 and FGF10 stimulate the proliferation and differentiation of pituitary stem cells by increasing the expression of LHX3⁹⁰. FGF signals through mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK) pathways, the overactivation of which by mutations in BRAF or K-Ras disrupts the development of hormone-producing cells, causing hypopituitarism in humans and mice^91,92. Mice genetically engineered to similarly activate the ERK/MAPK pathway showed an increase in the proliferation of SOX2⁺ stem cells, a dysmorphic stem cell niche and impaired lineage commitment. Activating somatic mutations in BRAF in humans are also associated with papillary craniopharyngioma⁹².

Notch

Notch signalling is required for the maintenance of an undifferentiated proliferative state during pituitary development, and Notch2 expression is dependent upon Prop1^15,78,93,94. Pituitary-specific conditional knockout of Notch2 leads to the misplacement and progressive loss of pituitary stem cells, indicating that Notch2 is necessary for their maintenance and localization⁹³. Similarly, deletion of RBP-Jκ, a nuclear effector of the Notch intracellular domain (NICD), leads to a marked reduction in the number of SOX2⁺ stem cells in the pituitary gland of the mutant embryos⁹⁴. Conditionally knocking out Notch2 in the pituitary causes decreased expression of the Notch targets Hey1 and Hes1 and the transcription factors Grainyhead-like 2 (GRHL2) and Prop1⁹⁵. Consistent with these observations, mice that lack both Hes1 and Hes5 display severe hypoplasia caused by the accelerated differentiation of stem cells⁹⁶, whereas overactivation of the Notch pathway efficiently blocks cell fate acquisition in a time-dependent manner¹⁵. However, neither Notch overactivation through ectopic expression of NICD nor Notch inactivation through deletion of RBP-Jκ in the POU1F1 lineage affected cell fate specification, whereas cells in the proopiomelanocortin (POMC)-expressing lineage remained sensitive to ectopic Notch activation, suggesting differential sensitivity to NOTCH signalling activity in different cell types.

WNT

WNT signals through both canonical (dependent on the nuclear accumulation of β-catenin) and non-canonical (independent of β-catenin) pathways. Genetically engineered mouse models have demonstrated that both pathways are required for the proper formation of Rathke’s pouch. Disruption of Wnt4 and Wnt5a, but not Wnt6, is associated with pituitary abnormalities in mice^14,97. Whereas mice deficient in Wnt4 have reduced Pou1f1 expression leading to a mild reduction of anterior lobe size, Wnt5a^–/– mice exhibit Rathke’s pouch dysmorphology and show an expanded region of BMP and FGF signalling⁹⁷. β-catenin is necessary for the proper expression in the ventral diencephalon of Bmp4, Fgf8 and Fgf10, which induce the early formation of Rathke’s pouch from the oral ectoderm⁸⁸. Mutant embryos with a specific deletion of β-catenin (Ctnnb) within the ventral diencephalon (Nkx2.1-cre; Ctnnb^fx/fx) have reduced Fgf8 expression and develop a smaller Rathke’s pouch, whereas gain-of-function experiments using Nkx2.1-cre; Ctnnb1^tm1Mmt mice showed that stimulation of canonical WNT signalling expands Fgf8 expression, resulting in a dysmorphic Rathke’s pouch⁹⁸.

In addition to its role during the early stages of pituitary organogenesis, regulated canonical WNT signalling is essential for normal pituitary growth⁹⁹. Transgenically expressing a degradation-resistant form of β-catenin in SOX2⁺ pituitary stem cells leads to the development — in a non-cell-autonomous manner — of large, cystic pituitary tumours that resemble human adamantinomatous craniopharyngioma. Here, overactivation of the WNT–β-catenin pathway acts on neighbouring cells in a paracrine manner through the senescence-activated secretory phenotype pathway¹⁰⁰. By contrast, loss of β-catenin function causes a drastic reduction in the Pou1f1 lineage (somatotropes, thyrotropes and lactotropes)⁹⁸, consistent with the transcriptional regulation of Pou1f1 by the complex of β-catenin and PROP1^101,102. Postnatally, pituitary growth and cell expansion depends on Wnt signalling in all cell lineages — both stem cells as well as endocrine cells^61,103. The Wnt target gene Axin2 is expressed in a subset of all the different pituitary cell types, including SOX2⁺ cells, and fate-mapping experiments show that most new pituitary cells are derived from Axin2⁺ cells.

Wnt ligands bind to Frizzled (Fzd) receptors. The genes for all 10 Fzd receptors are expressed in the postnatal mouse pituitary gland; however, Fzd1, Fzd3, Fzd4, Fzd6 and Fzd7 are the most highly expressed receptors in stem cells. Elucidating which ligand–receptor pairs are critical for Wnt signalling is challenging. Loss of Wnt signalling in pituitary stem cells causes reduced proliferation and anterior lobe hypoplasia. At the same time, secretion of Wnt ligands from pituitary stem cells is required to induce Wnt signalling and proliferation in committed endocrine cells, highlighting how pituitary stem cells are important signalling centres that induce proliferation in other cells (especially in tumour models).

SHH

SHH is essential to induce proliferation in, and to determine the cell type of, pituitary stem cells^13,104,105. In mice, SHH is expressed in the ventral diencephalon and in the oral ectoderm, except in the placode that will become Rathke’s pouch. Overexpression of HIP, an inhibitor of HH, in the oral ectoderm and Rathke’s pouch in mouse embryos (Pitx1HS-Hip) results in pituitary hypoplasia and a failure to express BMP2 and, subsequently, the transcription factor GATA2¹⁰⁴. Conversely, overactivation of SHH signalling results in an expansion of thyrotrope and gonadotrope cells. Gain-of-function and loss-of-function experiments have provided evidence supporting an additional role for SHH in the specification of LHX3⁺, LHX4⁺ cells in Rathke’s pouch in vivo¹³. Loss of SHH expression in the anterior hypothalamus in mouse embryos (Hesx1^Cre/+;Shh^fl/−) causes developmental arrest and failure to activate LHX3 and LHX4 expression in the pouch; by contrast, overactivation of SHH leads to increased proliferation of LHX3⁺,LHX4⁺ cells, severe hyperplasia and enlargement of the SOX2⁺ stem cell compartment. Activation of SHH signalling in adult mouse pituitary stem cells induces their proliferation and differentiation, and it enhances hormone production in already differentiated endocrine cells¹³

Hippo

In 2016, the Hippo cascade was demonstrated to participate in pituitary gland development. Results from RNAscope in situ hybridization and immunofluorescence studies in mouse embryos revealed that components of the Hippo pathway (MST1/MST2 kinases, LATS1 kinase and Yap1/Taz effectors) are active in the stem-cell population throughout development as well as at postnatal stages, consistent with their role in promoting the stem-cell state¹⁰⁶. Indeed, the regulation of this molecular cascade is necessary for establishing normal pituitary size and specification of SOX2⁺ pituitary stem cells¹⁶. Deletion of LATS and the consequent nuclear localization of YAP and TAZ disrupts the ability of SOX2⁺ pituitary stem cells to differentiate, turning these cells into a tumorigenic state. By contrast, sustained YAP activity and its cytoplasmic localization promotes SOX2⁺ cell differentiation into the endocrine lineages (see¹⁰⁷).

Activating quiescent pituitary stem cells

The postnatal pituitary gland demonstrates a remarkable capacity to meet increased demand for pituitary hormones. This response can be accomplished by several means: hypertrophy and increased activity of existing differentiated cells; proliferation of existing differentiated cells; transdifferentiation; or differentiation of new cells from stem cells¹⁸. In adult 8-week-old mice, pituitary stem cells are largely inactive and only ~1–2% of Sox2⁺ cells proliferate^22,44 to replace cells that are naturally turning over²⁰.

Pathological activation

Extreme challenges in mice, such as pituitary wounding, adrenalectomy or gonadectomy, create an imbalance in the associated hypothalamus–pituitary axis and stimulate quiescent pituitary stem cells to proliferate and differentiate to generate nascent endocrine cells in an attempt to maintain homeostasis, although proliferation and hypertrophy of existing differentiated cells also make important contributions to the increase in demand^44,108,109. In cases of reduced levels of adrenal and gonadal hormones, the production and secretion of pituitary adrenocorticotropic hormone (ACTH) and LH/FSH, respectively, increase due to the loss of negative feedback. Surgical removal of the adrenal glands or gonads in mice was shown to stimulate an increase in cell proliferation in male rat pituitary glands¹¹⁰. The proliferating cells did not express ACTH or LH, and adrenalectomy plus gonadectomy did not cause an additive effect, indicating that a single non-endocrine population responded mitotically to the increased demand for pituitary hormones. This is presumably the stem cell population.

The pituitary can also be induced to produce more hormones by conditionally ablating cells through the inducible activation of the diphtheria toxin receptor, as outlined previously. After three days of conditional ablation of somatotropes in adult mice (8–12 weeks old) using the diphtheria toxin, the number of proliferative Sox2⁺ cells increased; ethynyl-deoxyuridine (EdU) labelling showed that these cells differentiated into somatotropes²². Although the proliferative index of pituitary stem cells increased, they regenerated only 50% of lost somatotropes after 5 months and extending the recovery period up to 19 months did not increase this regeneration any further³⁹, indicating that pituitary stem cells have limited in vivo regenerative capacity. Somatotrope ablation in older (8-month-old) mice failed to stimulate pituitary stem cell regeneration even after an 8-month recovery, suggesting that pituitary stem cells gradually lose regenerative capacity with age. Notably, pituitaries from 8-month-old mice contain fewer Sox2-expressing cells, which form smaller pituispheres, than pituitaries from 8-week-old mice. Conditional ablation of lactotropes in adult mice also caused an increase in the number of proliferative Sox2-expressing cells, although existing lactotrope proliferation and somatotrope transdifferentiation also contributed to the regeneration³⁴.

In adult mice, interleukin 6 (IL-6) is upregulated during the replenishment of hormone cell populations in the diphtheria-ablation model, and it can promote the proliferation of pituitary stem cells in vivo from younger (2–3 months) but not older mice (10–15 months)⁴². The pituitary glands of IL-6-knockout mice appear normal physiologically, indicating that IL-6 is not essential for normal pituitary growth and development¹¹¹. However, IL-6-deficient pituitary stem cells collected from neonatal mice form fewer organoids in culture, and IL-6 supplementation can improve stem cell proliferation and viability during passaging. It is possible that multiple cytokines are involved in pituitary stem cell regulation.

Physiological activation

Physiological demand for LH and FSH increases significantly during puberty, and post-pubertal mice possess 1.7x more gonadotropes than do prepubertal mice¹¹². After puberty, the gonadotropes are closely associated with the vasculature, potentially optimizing their secretory output. It is unknown whether Sox2⁺ stem cells contribute to the increase in gonadotrope number at puberty.

High serum levels of prolactin are typical of late pregnancy and lactation. Some studies have suggested that the lactotrope population expands in number to meet this physiological need; however, these studies failed to account for changes in cell size. Indeed, cell-lineage studies have shown that lactotrope population expansion does not occur during pregnancy in mice, and that high serum levels of prolactin during lactation are instead driven by lactotrope hypertrophy¹⁰⁹. In mice, the first pregnancy induces permanent remodeling of lactotrope connectivity to coordinate prolactin secretion in subsequent pregnancies¹¹³.

The regenerative capacity of human pituitary stem cells is unclear, although cases of congenital or acquired pituitary hormone deficiencies generally do not resolve over time if left untreated¹¹⁴. However, this lack of resolution could be because congenital cases can be caused by mutations that prevent differentiation of endocrine cells from stem cells, while acquired cases occur in later life, at which point human pituitary stem cells might have lost their regenerative capacity. An improved understanding of the molecular mechanisms that drive and limit pituitary stem cell regeneration in vivo might enable the development of therapies directed at the activation of pituitary stem cells to combat pituitary hormone deficiencies.

Differentiating embryonic and pluripotent stem cells into pituitary cells

Several approaches exist for generating differentiated pituitary cells from stem cells in vitro. Mouse and human embryonic stem cells (ESCs) and hiPSCs can be induced to differentiate into pituitary hormone-producing cells in culture^115–119. In addition, mouse pituitary tissue stem cells can be differentiated in culture or propagated in culture as stem cell spheres before being placed under the kidney capsule for further differentiation^35,61,120.

Mouse ESCs

Sasai pioneered methods to differentiate ESCs into a variety of tissues in culture, including the eye, neocortex and several other neuronal cell types. In 2011, Sasai and colleagues established a three-dimensional (3D) culture technique for differentiating pituitary hormone-producing cells from mouse ESCs (mESCs)¹¹⁵. mESCs cultured in a chemically defined medium could be differentiated into each of the hormone-producing cell types depending on which pathways were activated or blocked exogenously. Moreover, these 3D cultures could be implanted ectopically into hypophysectomized mice to rescue hormone production, establishing the concept of pituitary stem cell therapy.

To stimulate the differentiation of pituitary cells, Sasai and colleagues developed culture conditions that attempted to replicate known in vivo processes that induce Rathke’s pouch and pituitary formation during development (Fig. 3a). The ability of the cells to self-organize in culture was remarkable. mESCs were cultured at high density to allow them to aggregate into spherical bodies, replicating the interface between the layers of neural and oral ectoderm that is crucial for inducing the initial invagination and growth of the oral ectoderm into Rathke’s pouch. Activating SHH signalling using a smoothened receptor agonist (SAG) further induced portions of the oral ectodermal layer to invaginate inwards and express the pituitary-specific transcription factor LHX3. Application of exogenous BMP4 stimulated expression of the pituitary factors Pitx2 and Lhx3, but it inhibited expression of the hypothalamic factor Rax. After the establishment of this anterior progenitor region, treatment with specific chemical regimens stimulated the differentiation of these Lhx3-expressing cells into different endocrine cell-types. Inhibition of Notch signalling by the γ-secretase inhibitor DAPT increased the differentiation of corticotropes, whereas treatment with a WNT-activating molecule, BIO, increased the emergence of cells expressing Pou1f1. Further treatment of BIO-treated, Pou1f1-expressing cells with hydrocortisone and insulin led to the differentiation of somatotropes, whereas oestradiol and insulin stimulated lactotrope differentiation. Gonadotropes and thyrotropes arose only after culturing undifferentiated organoids with conditioned medium from PA6 skull bone marrow mesenchyme stromal cells, although the specific factor required for gonadotrope and thyrotrope differentiation was not identified. Notably, in vitro differentiated corticotropes that expressed ACTH could be transplanted into hypophysectomized mice to rescue adrenal insufficiency, demonstrating their in vivo functionality.

Figure 3. Pituitary organoid differentiation from mouse and human embryonic stem cells requires a variety of growth factors and inhibitors.

a, Mouse embryonic stem cells (ESCs) are maintained in a medium with 1% fetal calf serum, 10% knockout serum replacement media, and leukaemia inhibitory factor, a differentiation inhibitor¹¹⁵. Differentiation to pituitary-hormone producing cells is initiated by high density large cell aggregation (LCA) and culture in a growth-factor-free chemically defined medium containing 1:1 Iscove’s Modified Dulbecco’s Medium (IMDM): Ham’s F12 with a smoothened agonist (SAG), which induced RAX (neural ectoderm marker) and PITX1 (Rathke’s pouch marker) expression by day 6 and pituitary cell fate marker LHX3 by day 10. At 10 days, half of the media is replaced without SAG, and aggregates are cultured in high O₂. Differentiation into corticotropes requires the addition of DAPT, an inhibitor of Notch signalling; by day 21, some cells within the aggregates express ACTH and TBX19. Differentiation into cells that produce GH and prolactin is similar except for the addition of BIO, an activator of WNT signalling, at day 18; additionally, aggregates are transferred into media containing insulin and either hydrocortisone (for somatotropes) or oestradiol (for lactotropes) at day 21. Differentiation into cells that produce LH, FSH or TSH is less efficient, but enhanced by culture in conditioned medium that mimics head mesenchyme. b, Human ESCs are aggregated at a lower density than mouse ESCs; a ROCK inhibitor is added to prevent apoptosis (a ROCK inhibitor was not used in the mouse organoids)¹¹⁷. Supplementation with SAG and BMP4 is sufficient to induce the expression of PITX1 and the neural ectoderm markers RAX and NKX2.1. FGF2 supplementation induces the expression of LHX3. By 70 days in culture, corticotrope differentiation is evident by expression of TBX19 and production of ACTH. Differentiation into the POU1F1 lineage and gonadotrope lineage is stimulated by dexamethasone and the Notch inhibitor DAPT, respectively. c, Schematic representation of the self-organizing cells within pituitary organoids (adapted from^115,157). Cultured ESCs form spheres that differentiate into neural ectoderm and express the hypothalamic markers RAX and NKX2.1 in the inner area of the aggregate. The cells at the surface of the sphere appear columnar, differentiate into oral ectoderm and express the pituitary markers PITX1 and LHX3. As differentiation proceeds, the pituitary cells invaginate, mimicking the structure of Rathke’s pouch and its closure. Cells committed to the pituitary fate appear to be extruded from the surface cell layer of the sphere as they lose their columnar shape and begin to produce pituitary hormones. ACTH, adrenocorticotropic hormone; BMP4, bone morphogenetic protein 4; FGF2, fibroblast growth factor 2; FSH, follicle-stimulating hormone; LH, luteinizing hormone; ROCK, Rho-associated kinase; TSH, thyroid-stimulating hormone.

Human ESCs

In 2013, the Studer group described a technique to differentiate human ESCs (hESCs) into pituitary hormone cells in vitro beginning with a 2D monolayer of adherent cells¹¹⁶. The researchers showed that the BMP inhibitor Noggin shifted hESCs from a predominantly neuroectodermal PAX6⁺ state towards a SIX1⁺ pre-placode, precursor state. The media was then supplemented with SHH, further differentiating placodal precursor cells into pituitary placode-like cells. The expression of Pou1f1-lineage markers was increased in response to DAPT, and cells expressing FSH, ACTH and GH emerged naturally over approximately 30 days. When these in vitro differentiated human cells were injected into mice, human GH, FSH and ACTH could be detected in the serum, clearly demonstrating their in vivo functionality.

In 2016, Suga and colleagues successfully generated pituitary cells from hESCs in 3D culture by implementing several changes to the differentiation protocol they had used for mESC differentiation, including the addition of an initial treatment with a ROCK inhibitor, and subsequent treatment with SAG, BMP4, and FGF2¹¹⁷ (Figure 3b). These organoids differentiated into both RAX⁺ hypothalamic neural ectoderm and Rathke’s pouch like structures expressing PITX1 and LHX3 (Figure 3c). The hESC system takes longer than the mESC system for endocrine cell differentiation (up to 100 days in culture for somatotropes, lactotropes, thyrotropes and gonadotropes). Notably, the in vitro differentiated cells responded to positive and negative cues for hormone secretion and could rescue lethality of hypophysectomized mice.

hiPSCs

Several groups have reported protocols for generating hormone-producing cells from iPSCs^116,118,119. Studer and colleagues used a 2D culture system approach to differentiate hiPSCs into pituitary-hormone producing cells^116,118 (Figure 4). In this approach, adherent iPSCs were cultured with a TGFβ-inhibitor and SHH, which led to cells adopting a pituitary fate, as indicated by the expression of PITX1 and LHX3, with little differentiation into non-pituitary cell types. Differentiation into individual cell types is enhanced with various supplements: FGF8 treatment alone induces the highest proportion of corticotropes, whereas a mix of FGF8 and BMP2 induces a higher proportion of somatotropes and lactotropes than any other cell type. BMP2 alone is effective at giving rise to gonadotropes. It is intriguing that these 2D systems do not remain as monolayers during differentiation: they form embryonic body-like mounds, even though they lack the 3D organoid structure^116,118 (Fig. 4). They have the practical advantage over the 3D method^115,117, in that there is no requirement for 40% O₂ incubation.

Figure 4. Differentiation of human pituitary cells from induced pluripotent stem cells in monolayer cultures.

a, Human induced pluripotent stem cells can differentiate into pituitary hormone producing cells in culture. Somatic cells, such as fibroblasts, are transfected to express factors (OCT4, SOX2, KLF4, and c-MYC) that reprogram them to be pluripotent (hiPSCs). Two protocols have been developed for in vitro differentiation of pituitary cells from hiPSCs: a 2D adherent model¹¹⁸ and a 3D organoid model that requires hyperoxia¹¹⁷. Both use varying combinations of BMPs, FGFs, and SHH or SHH agonist (SAG) to induce pituitary progenitor commitment and subsequent hormone cell differentiation. Controlled addition of growth factors drives differentiation into hormone producing cells in both cases. Cells generated by both protocols are functional in vivo when transplanted into hypopituitary mice. b, Human iPSCs were differentiated using slightly adapted protocols^116,118 and representative images were taken on days 0, 7, 15, 40 and 60. The images were compared with cells in the control state, maintained in E6 medium only, without stimuli (C) Comparison by immunofluorescence of LHX3 protein expression, at days 0 and 40 of the differentiation protocol. Nuclei were stained with DAPI. Note the presence of structures resembling embryoid bodies and cords.

Applications of in vitro stem-cell-derived pituitary cells

ESCs and iPSCs have considerable potential as tools with which to study pituitary development ex vivo. One application, for example, is to establish 3D organoid disease models using patient-derived cells for use in conjunction with gene editing technology to assess the functional significance of genetic variants found in individuals with pituitary hormone deficiency⁹⁰. Another application of differentiating hiPSCs into hormone-producing cells is to assess whether findings in mouse models apply to human pituitary differentiation in organoids.

Furthermore, iPSCs are being considered for use in stem cell therapies for human diseases, by replacing cell types that are absent in certain diseases with physiologically functional cells that have been differentiated in vitro. Pituitary hormone deficiencies are prospective targets for stem cell therapies because the majority of known genetic causes of isolated or combined pituitary hormone deficiency arise from the absence of certain endocrine lineages during development¹. Several challenges must be overcome for the translational potential of stem cell treatments for combined pituitary hormone deficiency to be achieved. The growth factors and media used will need to be well-defined and consistent in quality from batch to batch in accordance with good manufacturing practices outlined by regulatory agencies such as the FDA. Other challenges are to generate the appropriate quantity of hormone cell types and to achieve functional integration of the cells to combat an individual patient’s pituitary hormone deficiency.

In vitro differentiation models aim to replicate the developmental processes that give rise to each of the pituitary hormone cell-types, but our current inability to attain appropriate cell proportions indicates that we do not fully understand the molecular mechanisms that regulate pituitary development and the delicate interplay and balance between these, and other, elements that are necessary for homeostasis. Other elements might include environmental and/or dietary factors, feedback from endocrine target organs and novel signalling pathways. For example, retinoic acid, which is derived from dietary vitamin A, has a role in the regulation of Pou1f1-lineage cell types, including differentiation into GH-producing cells¹²¹. Prop1 is required for retinoic acid signalling, and retinoic acid feeds back and stimulates the expression of Prop1^54,122. Mice expressing a dominant-negative form of the retinoic acid receptor during pituitary development showed pituitary dysmorphology and a developmental reduction in Pou1f1-lineage hormone expression, particularly TSH⁵⁴. These observations raise the question of how other environmental or dietary factors might influence pituitary differentiation and/or function.

Optimizing the dosage and timing of administration of many signalling molecules during in vitro stem cell differentiation could increase the yield of hormone-producing cell types. Bioengineering techniques can facilitate this process by using microfluidics to incorporate gradients of factors along multiple axes within differentiation chambers. This approach has been successfully applied to improve the efficiency and predictability of stem cell differentiation into neural tube structures in vitro¹²³. The application of these established optimization techniques to pituitary organoids is an important area of future investigation.

Cancer stem cells

Cancer stem cells (CSCs) comprise a subpopulation of tumour cells that have multipotent differentiation capacity, are capable of self-propagation in vitro (clonogenic potential) and show tumorigenic potential in transplantation experiments. In addition, CSCs are resistant to chemotherapy and radiotherapy. The origin of CSCs is still unknown; they are proposed to be derived either from the oncogenic transformation of normal tissue-specific stem cells^124–126 or from differentiated cells through reprogramming and dedifferentiation^127–129. CSCs were first identified in human leukaemia¹³⁰ and then in different types of solid tumour, including those of the breast, brain and liver^131,132. Both human and canine pituitary tumours were observed to contain small chromophobes lacking hormone granules, which at the time were thought to be stem cells^133–135.

CSCs reside in a microenvironment or niche that is mainly composed of fibroblasts and endothelial, mesenchymal and immune cells; extracellular matrix components; and several cytokines and growth factors that are crucial for the regulation of their self-renewal, activation and differentiation¹³⁶. The signalling pathways that regulate the balance between self-renewal and differentiation of both normal stem cells and CSCs are classically associated with oncogenesis, and include Notch, SHH and the WNT–β-catenin pathways¹³⁷. For example, stem cells of the intestinal crypt become CSCs when the WNT–β‐catenin pathway is overactivated¹²⁵.

Identifying CSCs

During the past decade, efforts have been made to identify CSC markers — a challenging task, mainly because of considerable tumour cell heterogeneity. CSCs isolated from different cancers share common markers, including OCT4, SOX2, Nanog and Nestin¹³⁸. Most of these CSC markers are also expressed in normal stem cells¹³⁹. An analysis of stem cells purified as a side population from pituitary adenomas revealed that they are more active in proliferation than stem cells in wild-type pituitary glands, and express certain cytokines and chemokines at higher levels than controls¹⁴⁰.

Many approaches can be taken to characterize CSCs. The clonogenic assay assesses proliferation potential; the in vitro sphere formation assay is used to assess self-renewal; and the in vivo tumour assay is used to identify cells that can initiate tumour formation. Immunostaining assays and flow cytometry can detect CSC markers, while RT-PCR can be used to assess the expression of CSC marker mRNAs and miRNAs. The best experimental assay for identifying CSCs is serial tumour transplantation at limited dilutions in immunodeficient mice. This approach allows the assessment of self-renewal and multipotency of a putative CSC subpopulation. The main limitation of this technique is the need to isolate CSCs for transplantation, which inevitably disrupts contacts between cells, attachments between cells and the extracellular matrix, and signals from the microenvironment, with possible consequences on the tumour-initiating potential.

CSCs in the pituitary gland?

Most pituitary tumours are benign adenomas, although rare aggressive or atypical adenomas exist; pituitary carcinomas with proven metastases constitute less than 1% of pituitary tumours¹⁴¹. Nevertheless, studies in mice and humans have provided support for the idea that CSCs are involved in benign and malignant pituitary tumorigenesis¹⁴².

In 2009, stem-like cells were isolated from human benign GH-producing and hormone-null pituitary adenomas and demonstrated to express stem cell markers, such as OCT4, CD133, NOTCH4 and nestin¹⁴³. When differentiated in culture, these cells downregulated their expression of stem-cell associated genes and synthesized multiple pituitary hormones in response to hypothalamic hormone stimulation. Finally, when transplanted into immunocompromised mice, these sphere-forming cells generated intracranial tumours that could be serially transplanted and resembled the original human tumour.

Lineage-specific transcription factors, such as Pax7, can be used to classify adenomas, especially those that do not secrete hormones (so-called ‘silent’ tumours)^141,144. However, additional markers would be beneficial in understanding the relationship of stem cells to the adenomas. CD133 is often used as a CSC marker in gliomas and is detected in 26% of human pituitary adenomas¹⁴⁵. However, CD133 positivity in adenomas did not correlate with patient characteristics such as age, gender, tumour size, or postoperative recurrence rate. In another study, human pituitary adenomas were found to contain cells that express CD133 and nestin; these cells could form spheres and self-renew in vitro, differentiate into neuronal and astrocytic lineages, and initiate synaptophysin-positive tumours when transplanted subcutaneously into immune compromised mice¹⁴⁶. However, the nature of these cells is not clear because CD133 and nestin are non-specific markers, and their ability to differentiate into pituitary hormone cell types was not demonstrated.

Human pituitary adenomas have a side population of cells that show a high efflux capacity (conferring resistance to chemotherapeutics), which is thought to be a functional characteristic of CSCs¹⁴². The side population was shown to express stem-cell markers, including CD44, KLF4, SOX2, NESTIN and CXCR4, as well as to overexpress genes related to EMT and angiogenesis. This cell population had clonogenic, sphere-forming potential in vitro, and could be differentiated into all pituitary hormone-producing cells. However, in contrast to previous studies, these human adenoma cells failed to expand when xenografted into immune compromised mice^142,143. The reason for this failure is unknown. A similar side population exists in the mouse pituitary corticotrope tumour cell line AtT20, which expresses SOX2 and CXCR4; these cells can form de novo tumours in xenograft transplantations, suggesting that CSCs formed part of the original tumour used to derive the AtT20 cell line. This observation supports the idea that CSCs contribute to pituitary adenoma formation.

The pro-angiogenic and invasive potential of pituitary CSCs has been successfully demonstrated in vivo using zebrafish embryos^147,148. Stem cell populations were isolated from 56 human pituitary tumours: these cells were CD133⁺, had clonogenic potential, and 68% could form spheres in vitro; they expressed established stem cell markers, including OCT4, SOX2, CXCR4 and nestin, and had the potential to differentiate into hormone-producing pituitary cells¹⁴⁷. These cells were not tumorigenic in mice, in contrast to CD133⁺ tumour cells or AtT20 cells. However, they elicited a pro-angiogenic response and behaved invasively when grafted into zebrafish embryos. These pituitary adenoma stem-like cells express receptors for dopamine (DRD2) and somatostatin (SSTR2, SSTR5), and activation of these receptors as a potential therapeutic strategy using a dopamine–somatostatin chimeric agonist conferred antiproliferative effects¹⁴⁷. In a separate, similar study, CSCs were isolated from 46 non-functioning human pituitary tumours, and 70% of them formed spheres in culture¹⁴⁸. The spheres were able to self-renew in vitro and expressed several markers of stemness, including pluripotent embryonic stem cell markers (SOX2, OCT4, KLF4, and nestin) and the dopamine and somatostatin receptors. These observations suggest the presence of committed precursors in spheres cultured in vitro in stem-cell-permissive medium. Moreover, these cells showed tumorigenic potential in zebrafish model, as demonstrated by their invasive behaviour and their pro-angiogenic activity¹⁴⁸. Together, these data indicate the presence of a hormone-negative cell subpopulation expressing stem cell markers in pituitary tumours.

CD15-expressing cells can initiate pituitary adenoma formation¹⁴⁹. Pituitary adenomas that were enriched for CD15 expression also expressed SOX2 and PAX7. Isolated CD15⁺ cells were able to form spheres in culture and initiate pituitary tumours in mouse xenotransplant experiments. Interestingly, recurrent pituitary adenomas contained a larger population of CD15⁺ cells than did non-recurrent tumours, suggesting that patients with adenomas that express high levels of CD15 could benefit from aggressive surgical interventions, chemotherapy and/or radiotherapy.

Multiple lines of evidence therefore support the existence of pituitary-adenoma associated stem cells, but more studies are needed to understand the role of these cells in the initiation and progression of disease^140,150. Many of the published studies investigated CSCs in non-functional adenomas; little information is available for other types of pituitary tumour. Once the role of adenoma-associated stem cells is better understood, and the mechanisms involved in their regulation are uncovered, new therapeutic strategies might be developed. The identification of new biomarkers could be valuable for predicting pituitary tumour behaviour, response to intervention, and identifying novel therapeutic targets.

In animal models, SOX2⁺ pituitary stem cells can act as tumour-initiating cells (CSCs) and tumour-inducing cells. As mentioned previously, constitutive activation of WNT signalling in pituitary stem cells in mice causes the formation of tumours that resemble human adamatinomatous craniopharyngiomas through the senescence-activated phenotype pathway, which induces tumorigenesis in a paracrine manner^99,100. Although it is not yet clear which cells become tumorigenic in this model, paracrine secretion of Wnt from pituitary stem cells is required for the normal proliferation of committed cells that are adjacent to the stem cells¹⁰³.

In p27-null mice, which develop intermediate lobe tumours arising from melanotropes¹⁵¹, Sox2 expression is required in stem cells to allow tumour formation from melanotropes, indicating that a paracrine factor originates from stem cells to drive proliferation in another cell population¹⁵². On the other hand, stem cell-derived tumours can occur with constitutive activation of the Hippo pathway through deletion of LATS kinases in pituitary stem cells, leading to SOX2-cell-derived carcinoma-like tumours in mice¹⁶.

Conclusions

The existence of pituitary stem cells was proposed a long time ago. SOX2-expressing pituitary stem cells were formally identified in 2008, at which time they were demonstrated to self-renew and differentiate. Since then, we have learnt a great deal about pituitary stem cells. Wnt, BMP, FGF and Hippo signalling, cytokines, chemokines and EMT are promising areas for future investigation¹⁵³. It will be especially important to determine how the signalling pathways interact (for example, WNT and Hippo). When stem cells are activated in response to organ ablation or wounding, it is not known how selective differentiation into the desired cell type is orchestrated. Also unknown is how the control of stem cell proliferation and quiescence changes throughout the lifespan. New technologies, such as single-cell RNA sequencing, enable robust and simple access to the transcriptomes of single cells, thereby facilitating the identification of gene regulatory networks and description of the developmental trajectories, although single-cell/nuclei technologies carry their limitations, such as shallow per cell read depth or genome annotation inaccuracies. Future directions could include study of the transcriptome in situ. Spatial -omics methods are already improving our understanding of human and mouse tissues in research, diagnostic and therapeutic settings. It will be interesting to study pituitary stem cells and functional homotypic and heterotypic cellular networks in their tissue context. Engineering technologies and microfluidics could further improve the efficiency and predictability of stem cell differentiation into hormone-producing cells in culture, thereby providing a mechanism to better study human pituitary development.

Several clinical trials have been conducted in the field of regenerative medicine using iPSCs — in the retina and heart, for example. Pituitary cells derived from iPSCs could have potential for patients with hypopituitarism, although more research needs to be done. The main challenges of stem-cell transplantation include immune rejection, tumorigenesis, location of region, method of transplantation and cost. It is important that pituitary function is regulated appropriately. Advances have already been made in some systems, such as the establishment of vascularized organoids through the incorporation of human umbilical vein endothelial cells, and hiPSC-derived blood vessel organoids have been reported, which could aid in vascularizing pituitary organoids¹⁵⁴. Co-culture systems involving organoids derived from iPSCs giving rise to different tissue cells (such as the adrenal gland, thyroid or liver) present interesting models for studying the hypothalamus–pituitary–target organ axis and inter-organ communication^155,156. Advances in iPSCs and organoids are set to revolutionize developmental biology, disease modelling and drug discovery in pituitary research.

Key points.

• The cells of the anterior pituitary gland have a low turnover rate, and the tissue has a limited regenerative capacity.

• Pituitary stem cells secrete factors like WNT and Hippo that initiate proliferation in adjacent cells; the recipient cells are probably guided to differentiate into specific hormone-producing cell types by hypothalamic input and/or end organ feedback.

• Organoids and 2D cultures derived from embryonic stem cells and induced pluripotent stem cell cultures show promise as tools for studying differentiation ex vivo.

• Stem cells are known to exist in pituitary tumours, but the role of these cells in tumour initiation, progression, recurrence, and resistance to pharmacological therapy needs to be further elucidated.

Table of Contents text:

This Review provides an overview of the notable discoveries in pituitary stem cell function, and highlights important areas for current and future research, including the use of pituitary organoids for advancing pituitary stem cell biology and pituitary organogenesis, as well as in potential therapeutic approaches.

Biographies

María Inés Pérez Millán received her PhD from University of Buenos Aires, Argentina where she studied the role of dopamine signaling in pituitary adenoma formation. Her postdoctoral training was at University of Michigan focusing on the mechanistic function of PROP1. As an Assistant Professor at University of Buenos Aires she has identified novel genes that cause pituitary hormone deficiency in humans.

Leonard Y. M. Cheung received his PhD from University College London, conducted postdoctoral work at the University of Michigan, and is currently an Assistant Professor at SUNY Stony Brook, NY. He uses single cell RNA sequencing in the pituitary gland and is interested in the changes in stem cell functioning throughout the lifespan.

Florencia Mercogliano received her PhD and conducted her postdoctoral training at University of Buenos Aires, Argentina. She is interested in the development of in vitro models to study human genetic disorders to screen potential therapeutic targets, with a special focus on breast cancer stem cells. Studying factors important for organ development will provide fundamental information useful for understanding the pathways involved in tumorigenesis.

Maria Andrea Camilletti received her PhD from University of Buenos Aires, Argentina. She is a postdoctoral fellow from CONICET. She is interested in expanding our knowledge of the genetic and molecular basis of pituitary development and understanding the pathogenic mechanisms that underlie congenital hypopituitarism. She is focusing on generating in vitro models of this disease through reprograming human pluripotent stem cell lines derived from patients with a molecular diagnosis of hypopituitarism.

Gonzalo Tomás Chirino Felker got his Master’s in Biological Science from the University of Buenos Aires, Argentina. He is currently doing his PhD in disease modeling of combined and isolated pituitary deficiencies using patient-derived iPSCs with likely pathogenic variants in novel genes in hypopituitarism. His goal is to develop an optimized method for obtaining hormone producing cells from iPSCs in vitro.

Santiago Miriuka, M.D. is a physician scientist at CONICET in Buenos Aires, Argentina. He is a specialist in stem cell reprogramming from differentiated human cells and differentiation to other lineages. Lately, he developed Artificial Intelligence-Based Omics Data Systems for understanding of complex human genetic diseases.

Lucia Natalia Moro received her PhD from University of Buenos Aires, Argentina. She is a researcher from CONICET specialized in stem cells and genetic disease modelling. She is interested in gene editing by CRISPR and gene therapy development. During her career she has generated several patient derived iPSCs lines with this purpose.

Michelle L. Brinkmeier received her MS from University of Michigan and has 30 years of experience in pituitary development.

Sally A. Camper received her PhD from Michigan State University and postdoctoral training at Fox Chase Cancer Center in Philadelphia and Princeton University in New Jersey. She has been a Professor at University of Michigan since 1988 and has focused on the use of mouse models of human pituitary hormone deficiency to understand disease pathophysiology.