Characterisation of a mural cell network in the murine pituitary gland

The anterior and intermediate lobes of the pituitary are composed of endocrine cells, as well as vasculature and supporting cells, such as folliculostellate cells. Folliculostellate cells form a network with several postulated roles in the pituitary, including production of paracrine signalling molecules and cytokines, coordination of endocrine cell hormone release, phagocytosis, and structural support. Folliculostellate cells in rats are characterised by expression of S100B protein, and in humans by glial fibrillary acid protein. However, there is evidence for another network of supporting cells in the anterior pituitary that has properties of mural cells, such as vascular smooth muscle cells and pericytes. The present study aims to characterise the distribution of cells that express the mural cell marker platelet derived growth factor receptor beta (PDGFRβ) in the mouse pituitary and establish whether these cells are folliculostellate. By immunohistochemical localisation, we determine that approximately 80% of PDGFRβ+ cells in the mouse pituitary have a non‐perivascular location and 20% are pericytes. Investigation of gene expression in a magnetic cell sorted population of PDGFRβ+ cells shows that, despite a mostly non‐perivascular location, this population is enriched for mural cell markers but not enriched for rat or human folliculostellate cell markers. This is confirmed by immunohistochemistry. The present study concludes that a mural cell network is present throughout the anterior pituitary of the mouse and that this population does not express well‐characterised human or rat folliculostellate cell markers.


| INTRODUC TI ON
The pituitary is an endocrine gland located underneath the brain that produces and releases circulating hormones in response to signals from the hypothalamus. It is formed of three lobes. Neurones in the posterior lobe (pars nervosa) connect to the hypothalamus and store oxytocin and vasopressin. The intermediate lobe (pars intermedia) is formed of melanotrophs that produce melanocyte-stimulating hormone and the anterior lobe (pars distalis) is formed of heterogeneous specialised endocrine cells that each make a particular hormone type: adrenocorticotrophic hormone is made by corticotrophs, growth hormone by somatotrophs, prolactin by lactotrophs, thyroid-stimulating hormone by thyrotrophs, and the gonadotropihns follicle-stimulating hormone and luteinising hormone by gonadotrophs. The folliculostellate cell is a further non-endocrine cell type in the anterior pituitary. Folliculostellate cells are named after their appearance in electron micrographs: they have a distinctive stellate shape and surround follicular structures. 1 Their long cytoplasmic processes connect to form a three-dimensional network 2 that is considered to perform multiple roles in the pituitary, including production of paracrine signalling molecules and cytokines, 1 coordination of endocrine cell hormone release, 3 phagocytosis, 4 and structural support. 5 There is evidence that folliculostellate cells in rats are composed of multiple subpopulations that may arise from different precursors. 6 Mural cells such as vascular smooth muscle cells and pericytes are required for the formation and stability of the vasculature and the support of endothelial cells. They are characterised by the expression of the markers chondroitin sulfate proteoglycan 4 (more commonly known as nerve/glial antigen 2 or NG2), alpha smooth muscle actin (SMA) and platelet derived growth factor receptor beta (PDGFRβ). 7,8 In rat anterior pituitaries, they have also been shown to express desmin. 9 A recent study reported that mouse folliculostellate cells express NG2 and that inactivation of retinoblastoma protein in NG2+ cells resulted in adenohypophysial tumours with immunohistochemical and ultrastructural features resembling those of aggressive Pit1-lineage tumours in humans. 10 Furthermore, PDGFRβ has been shown to be expressed in the mouse TtT/GF folliculostellate cell line. 11 Taken together, this evidence suggests that

| Immunohistochemistry
NG2 staining was only possible with cryopreserved sections. To co-stain for NG2 and PDGFRβ, Pituitaries were frozen on dry ice and transferred to a −80°C freezer. Prior to sectioning, pituitaries were encompassed in Tissue-Tek OCT compound (Sakura, Osaka, Japan), mounted on a cryostat chuck, sectioned at 5 µm and mounted on poly lysine coated slides (PO425-72EA; Sigma-Aldrich, St Louis, MO, USA). Sections were dried at room temperature for 12 hours, then sections fixed under a droplet of 4% buffered formaldehyde for 5 minutes. Slides were then stained with a double fluorescent tyramide method as described previously 12,13 using antibodies to NG2 (dilution 1:8000; catalogue no. AB5320; Millipore, Billericia, MA, USA; RRID:AB_91789) and PDGFRβ (dilution 1:8000; catalogue no. ab32570; Abcam, Cambridge, MA, USA; RRID:AB_777165). A non-pressurised heating by microwave in pH6 0.01 mol L -1 sodium citrate buffer was performed after the first tyramide reaction to denature remaining antibodies from the first set and avoid non-specific reactions with the second set of antibodies. 14 Double staining of either NG2 first or PDGFRβ first, both with and without primary antibodies, was performed to confirm this (see Supporting information, Data S1).
All other antibodies could be used on formaldehyde-fixed paraffin-embedded sections. Pituitaries were fixed for 24 hours in 4% buffered formaldehyde. Fixed tissues were processed and embedded in paraffin wax and sections (3 µm) were used for immunohistochemical analysis.
Fluorescence immunostaining was performed either by a single or double antibody tyramide fluorescence immunostaining method, as described previously. 12,13 Antibodies used were PDGFRβ (dilu- ab92494; Abcam; RRID:AB_10585428) optimised for the method and conditions used. A non-pressurised heating by microwave in 0.01 mol L -1 sodium citrate buffer (pH 6) was performed after the first tyramide reaction to denature remaining antibodies from the first set and avoid non-specific reactions with the second set of antibodies. 14 On resin-perfused sections, staining was performed using a colorimetric method. Briefly, sections were deparaffinised and rehydrated, and high-pressure antigen retrieval was performed in citrate buffer. Slides were incubated with normal goat serum/Tris-buffered saline/bovine serum albumin (NGS/TBS/BSA) to block non-specific antibody binding, before being incubated overnight with rab-

| Magnetic activated cell sorting (MACS)
Pituitaries were removed during dissection and the anterior lobes were dissected and used for MACS. Anterior pituitaries from approximately 40 mice on a mixed C57Bl/6 background were used in each MACS experiment, and MACS was repeated three times. The pool of anterior pituitaries was digested into a single cell suspension in accordance with a previously published protocol. 15 MACS was performed using protocols and reagents from Miltenyi Biotec (Bergisch Gladbach, Germany). Briefly, the single cell suspension was incubated first with 1:500 PDGFRβ antibody (catalogue no.

| Quantitative reverse transcriptase-polymerase chain reaction (qPCR)
Multiplex qPCR was performed on cDNA for the genes listed in Table 1 using an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) and the Roche Universal Probe library (Roche, Welwyn, UK). The expression of all genes was related to an internal housekeeping gene assay for Actb (Roche) as described previously. 13 Actb Ct values were similar between samples when normalised to the number of cells that had been RNA extracted. The resulting data were analysed using the ΔΔCt method.

| Resin perfusion
Resin perfused pituitaries were obtained from Dr Diane Rebourcet (University of Newcastle, Callghan, NSW, Australia), prepared in accordance with previously published protocols. 16

| Immunogold electron microscopy
Pituitaries were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 mol L -1 phosphate buffer (pH 7.2) at room temperature for 3 hours, then transferred to 0.25% glutaraldehyde and 0.2% paraformaldehyde in 0.1 mol L -1 phosphate buffer (pH 7.2) and then kept at 4°C. Pituitaries were prepared for immunogold electron microscopy as described previously. 17

| PDGFRβ cells of the mouse anterior pituitary did not express rat or human folliculostellate cell markers
We then investigated whether pituitary PDGFRβ cells express the same markers as rat and human folliculostellate cells. These Localisation of S100 calcium-binding protein B (S100B) is a characteristic of folliculostellate cells in rats. 22 S100b transcript was not This was confirmed by immunohistochemistry for pan-S100 protein in whole mouse pituitary sections. S100 staining was not seen in the parenchyma of the anterior lobe ( Figure 4B) but was seen in  Figure 4G). As a positive control to ensure that the antibody is functional, staining could be seen in astrocytes in the mouse brain ( Figure 4H). These results indicate that the PDGFRβ+ cell population in the anterior pituitary is not enriched for rat or human folliculostellate cell markers.

| SOX2+ cells in the adult pituitary gland did not express PDGFRβ
To clarify whether PDGFRβ+ cells in the pituitary express the previ-

| A network of mural cells in the mouse pituitary has a spatial distribution similar to folliculostellate cells
In the present study, we have shown that staining for PDGFRβ

| The novel mural cell network does not express folliculostellate cell markers
The folliculostellate cells of the pituitary also form a network that is commonly characterised by expression of S100B or GFAP markers 0.0 S100 Nuclei S100 Nuclei S100 Nuclei S100 Nuclei GFAP Nuclei GFAP Nuclei shows that it is localised to cells with ultrastructural characteristics of both folliculostellate cells and pericytes. Previous studies (including those using the same antibody as that employed in the present study) have shown S100 staining in the pituitary, 28-34 although these S100-positive cleft cells are often defined as 'folliculostellate cells' but do not show staining throughout the parenchyma. Interestingly, Allaerts et al 35 note that they do not observe S100 staining in the mouse pituitary. The S100+ cells found in the cleft are also likely to be SOX2+ as a result of their location and previous reports of co-staining of SOX2 and S100 in the rat. 36 A further 54 papers have been published using the TtT/GF cell line, which is an immortal cell line generated from a mouse thyrotrophic pituitary tumour, 37 classified as 'folliculostellate-like' based on strong immuno-positive staining for GFAP and weak staining for S100 protein. 37 The TtT/GF cell line was isolated from a mouse thyrotrophic tumour that had been induced by radiothyroidectomy and so it is possible that the transformation of these cells and their culture ex vivo has changed the characteristics of the cell line from that of the originating cell in vivo. However, it is also interesting to note a recent study reporting that the TtT/GF cell line has 'pericyte characteristics' including the expression of NG2. 38 It is likely that this is a folliculostellate cell line but has either switched on or increased expression of S100 as well as mural cell markers, or that it has been derived from a sub-population of folliculostellate cells, 6 potentially the from the SOX2+ S100+ population of cleft cells.
Our evidence suggests that the mural cell network in the mouse anterior pituitary does not comprise folliculostellate cells, although we could not establish a folliculostellate marker to unambiguously determine this, and so there is still the possibility that these two networks overlap.

| CON CLUS IONS
The anterior pituitary is composed of well-characterised networks of endocrine and folliculostellate cells. 42 Here, we present evidence

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jne.12903.