Paracrine FGF1 signaling directs pituitary architecture and size

Abstract

Organ architecture is established during development through intricate cell-cell communication mechanisms, yet the specific signals mediating these communications often remain elusive. Here, we used the anterior pituitary gland that harbors different interdigitated hormone-secreting homotypic cell networks to dissect cell-cell communication mechanisms operating during late development. We show that blocking differentiation of corticotrope cells leads to pituitary hypoplasia with a major effect on somatotrope cells that directly contact corticotropes. Gene knockout of the corticotrope-restricted transcription factor Tpit results in fewer somatotropes, with less secretory granules and a loss of cell polarity, resulting in systemic growth retardation. Single-cell transcriptomic analyses identified FGF1 as a corticotrope-specific Tpit dosage-dependent target gene responsible for these phenotypes. Consistently, genetic ablation of FGF1 in mice phenocopies pituitary hypoplasia and growth impairment observed in Tpit-deficient mice. These findings reveal FGF1 produced by the corticotrope cell network as an essential paracrine signaling molecule participating in pituitary architecture and size.

Keywords: FGF1; GH deficit; cell networks; cell–cell interactions; organogenesis.

Fig. 1.

Smaller anterior pituitary lobes (AL) in Tpit^−/− mutant mice. (A) Hematoxylin and eosin staining of sections from adult (3-mo-old) WT and Tpit-KO mouse pituitaries. (Scale bar: 500 µm.) IL, intermediate lobe. PL, posterior lobe. (B) AL genomic DNA contents (means ± SEM, n = 3 to 7 ALs/genotype) from Tpit-heterozygous (HT: blue), Tpit-knockout (KO: red), and POMC-KO (purple) mice relative to their WT littermate controls. Data are normalized for body weights. (C and D) Postnatal pituitary growth. Pituitary genomic DNA contents of WT (black), Tpit-HT (blue), and Tpit-KO (red) male (C) and female (D) mice. The right Y axis reports total pituitary genomic DNA content (means ± SEM, n = 3 to 28 mice/group), while the left Y axis corresponds to the deduced total number of cells per pituitary. The X axis represents the age in days. (E and F) Ki67 immunostaining and quantitation (means ± SEM, n = 2 to 8 mice/group) of P5 and 3-wk-old WT and Tpit-HT pituitary sections from males and females. (Scale bar: 200 µm.) Demarcations between pituitary lobes (anterior: AL, intermediate: IL, posterior: PL) are indicated by white dashed lines. All graphs were done using GraphPad Prism 9, and statistical significance was determined using bilateral Student’s t test. *P < 0.05, **P < 0.005, and ***P < 0.0005 vs. WT.

Fig. 2.

GH deficiency in Tpit mutant mice. (A−F) Intimate contacts between corticotrope and somatotrope cells in the adult anterior pituitary. (A) Visualization of corticotrope and somatotrope contacts using eGFP for visualizing corticotropes (POMC cells, green) or immunofluorescence against GH (magenta). (B−D) Homotypic corticotrope (green) interactions do not show strong β-Catenin staining (B, arrow 1), whereas tight contacts (<0.5 µm) marked by β-Catenin staining are observed between somatotropes (C, arrow 2) or between somatotropes (gray) and corticotropes (D, magenta, arrow 3). (Scale bar: 20 µm.) (E and F) Surfacing of eGFP/ACTH (green), β-Catenin (yellow), and GH (purple) signals illustrating (E) and quantifying (F) intimate corticotrope/β-Catenin/somatotrope contacts. (F) Quantification of colocalization of β-Catenin signal with either POMC or GH, and colocalization of β-Catenin within POMC cells in close contact with GH cells [15 z-stacks of images; 4 animals]. (G−I) GH–related defects in Tpit mutant mice. (G) Real-time qPCR (RT-qPCR) assessment of GH and POMC mRNA levels in 3-mo-old normal (WT), heterozygous (HT), and Tpit-deficient (KO) males (Left) and females (Right). Relative mRNA levels were normalized to TBP mRNA and represented as a fraction of WT levels. Data represent averages (±SEM) of 5 to 6 mice per genotype. (H) Tpit dosage-dependent reduction of pituitary GH content. GH contents in 3-mo-old mice were measured as described (42). Data represent averages (±SEM) of 5 mice per genotype. (I) RT-qPCR analyses of GH-dependent liver IGF1 mRNA levels (normalized relative to TBP mRNA) in males (Left) and females (Right) of indicated Tpit genotypes. Data represent averages ± SEM of 5 to 6 mice per genotype. (J) GH immunostaining on 3-wk-old WT and Tpit-HT pituitary sections. (Scale bar: 150 µm.) (K) Percentage of GH-positive cells (means ± SEM, n = 2 to 8 mice/group) in WT and Tpit-HT male and female pituitaries. (L and M) Mouse length measurements (body and tail means ± SEM, n = 7 to 15 mice/group) indicate shorter size in Tpit-KO males and females. Statistical significance was determined using RM one-way ANOVA in Prism (GraphPap) or bilateral Student’s t test in Microsoft Excel. *P < 0.05, **P < 0.005, ***P < 0.0005, and ****P < 0.0001 vs. WT.

Fig. 3.

Somatotrope ultrastructural abnormalities in Tpit mutant mice. (A–I) Electron micrographs of somatotropes. (A–C) WT somatotropes exhibiting secretory granule polarity toward blood vessels (Cap), as shown by the blue arrow in (A). (D–F) Tpit^+/− somatotropes showing less granule polarity with an example (F) of secretory granules margination. (G–I) Tpit^−/− somatotropes showing loss of secretory granule polarity toward capillaries, while (I) shows two somatotropes with granule margination. (Scale bar: 2 µm.) S = somatotrope, Cap = capillary. (J–L) Quantification of cytoplasmic area (J) granule area (K) and granule diameter (L) in Tpit WT, HT, and KO somatotropes. Data are means ± SEM (n = 4 for all groups) *P < 0.05 and **P < 0.01 vs. WT.

Fig. 4.

ScRNAseq analyses of WT, Tpit-HT, and Tpit-KO pituitaries. (A) UMAP representation of different pituitary cell clusters. Detailed identification of clusters is reported in SI Appendix, Fig. S3 A and B. (B) Cell type proportions (%) per genotype determined in scRNAseq. (C) Number of DEG for each cell cluster of Tpit-HT compared to WT cells. Bars represent the number of genes down-regulated (yellow) or up-regulated (blue) in Tpit-HT vs. WT cells using the following criteria: 1 ≤ log₂FC ≤ −1, −log₁₀P > 4. (D) Volcano plot showing DEG analysis between WT and Tpit-HT corticotrope cells. Examples of genes encoding ligands/receptors up- (Left) or down- (Right) regulated in HT corticotropes are shown on the graph. 41 out of 68 down-regulated genes correspond to corticotrope-signature genes as determined by Seurat (Dataset S2). (E–G) Deregulation of the MAPK/Ras and PI3K-Akt pathways in Tpit-HT somatotropes. (E) Volcano plot showing DEG analysis of WT and Tpit-HT somatotrope cells (cluster #2). Examples of genes of the Ras/MAPK pathway are shown on the graph by arrows. (F) Gene ontology analysis of DEG in Tpit-HT somatotropes (cluster #2) showing deregulation of the MAPK/Ras and PI3K pathways. The size and color of the dots are proportional to the size of the category. (G) Dot plot showing expression of select genes of the MAPK/Ras and PI3K pathways in WT, Tpit-HT, and Tpit-KO somatotropes (WT and HT cluster #2, KO cluster #5). The dot diameters reflect the number of cells expressing a given gene, while the intensity of its color is proportional to the mean expression level of the gene within the cluster.

Fig. 5.

FGF1 is a cortico/melanotrope-specific, Tpit-dependent gene for corticotrope–somatotrope communication. (A) Most significant ligand–receptor interactions between different cell types identified in WT (Left) and Tpit-HT (Right) scRNAseq samples using CellPhoneDB (44). The mean expression and p values of different ligand–receptor pairs are indicated by circle color and size, respectively. Interacting molecules down-regulated in Tpit-HT are shown in red at Left, while those up-regulated in Tpit-HT are in blue at Right. The gray vertical boxes show interaction pairs between corticotropes and somatotropes. Horizontal boxes indicate interactions lost (FGF1, in red) or gained (BDNF, in blue) in Tpit-HT. Other homotypic (Ptprz1) or heterotypic (Plxnb2) corticotrope interactions that are lost in Tpit-HT are shown by red arrows. (B–D) FGF1 expression in pituitary cells. (B) Visualization of FGF1 and TGFBR3 expression in WT pituitary cells. While FGF1 is expressed in the two POMC lineages (corticotropes and melanotropes) and PSC, TGFBR3 is broadly expressed in the pituitary, with the strongest expression in gonadotropes and somatotropes. The numbers in parentheses indicate the normalized expression levels for each cluster. (C and D) UMAP aggregation of WT and Tpit-HT corticotrope clusters showing spatial segregation of the two clusters. This segregation reflects transcriptomic changes, as exemplified by expression of FGF1 (C) that is mostly expressed on the upper (WT) part of the overlay cluster. (E) P-ERK1/2 staining on WT and Tpit-HT 3-wk-old pituitaries and quantifications showing greater labeling in WT pituitaries. White circles show two examples of p-ERK1/2-positive cells next to ACTH-positive cells (corticotropes) in WT pituitary. Quantitations of these occurrences in WT and HT pituitaries is reported below micrographs. Pituitary overviews of these stainings are presented in SI Appendix, Fig. S5D. (Scale bar: 20 µm.) (F) FGF1 is a direct Tpit target gene. RNAseq (Top) and ATACseq (Bottom) profiles at the FGF1 locus in FACS-sorted cortico- and gonadotrope mouse cells (45). ChIPseq profiles (Middle) in AtT-20 corticotrope cells (45, 46) showing three Tpit recruitment peaks (#1-3) exhibiting active enhancer marks, namely bimodal H3K4me1, P300, and ATAC peaks in FGF1 intron B. A palindromic Tpit response element sequence (46, 47) is present under peak #1.

Fig. 6.

FGF1 gene inactivation phenocopies pituitary hypoplasia and growth retardation observed in Tpit-KO mice. (A) Western blot analysis of kidney protein extracts (200 µg) from two WT and two FGF1-KO mutant mice revealed an anti-FGF1 monoclonal antibody. The arrow indicates the expected size of FGF1 migration, the asterisk shows a nonspecific band used as a loading control. (B and C) Smaller anterior pituitary lobes (AL) in FGF1^−/− mutant mice. (B) Hematoxylin and eosin staining of sections from adult (3-mo-old) WT and FGF1-KO mouse pituitaries. (Scale bar: 500 µm.) IL, intermediate lobe. PL, posterior lobe. (C) AL genomic DNA contents (means ± SEM, n = 4 to 12 ALs/genotype) from FGF1-KO (light red) and Tpit-KO (red) mice relative to their WT littermate controls. Data were normalized for body weights. (D and E) Reduced proliferation rate in FGF1-KO pituitaries at 3 wk. (D) Ki67 immunostaining of 3-wk-old WT and FGF1-KO pituitary sections. (Scale bar: 150 µm.) Demarcations between pituitary lobes (anterior: AL, intermediate: IL, posterior: PL) are indicated by white dashed lines. (E) Percentages (means ± SEM, n = 3 to 4 mice/group) of Ki67-positive cells in WT and FGF1-KO male and female pituitaries. (F–I) Reduced GH signaling in FGF1-KO mice. (F) GH immunostaining on 3-wk-old WT and Tpit-HT pituitary sections. (Scale bar: 50 µm.) (G) Percentage of GH-positive cells (means ± SEM, n = 3 to 4 mice/group) in WT (gray) and FGF1-KO (light red) male and female pituitaries. (H) RT-qPCR analyses of GH-dependent liver IGF1 mRNA levels (normalized relative to TBP mRNA) in FGF1-KO (light red) mice compared to WT controls (gray). Data represent averages ± SEM of 5 to 6 mice per genotype. (I) Mouse length measurements (tail means ± SEM, n = 12 to 20 mice/group) indicating shorter tail size in FGF1-KO mice (light red) compared to WT controls (gray). All graphs were done using GraphPad Prism 9 and statistical significance was determined using bilateral Student’s t test. *P < 0.05, **P < 0.005, and ***P < 0.0005 vs. WT.

Fig. 7.

Tpit and FGF1 coordinate late postnatal pituitary organogenesis. (A) Schematic representation of corticotrope–somatotrope–vasculature unit in WT and Tpit-KO pituitaries. There are intimate contacts between corticotropes and somatotropes and between somatotropes and capillaries, while corticotropes are interacting with the capillaries via cytonemes (29). GH secretory granules in somatotropes accumulate mainly next to capillaries. In Tpit-KO mice, somatotropes are smaller, with fewer and smaller GH granules, that lose polarity. (B) Proposed model on the role of corticotrope secreted FGF1 in postnatal pituitary development. Ras/MAPK/PI3K/Akt mediated actions of FGF1 act on somatotrope proliferation (mitogenic effect), growth (cell size), GH granule size, and localization through transcriptional regulation of genes involved in corresponding pathways. FGF1 could also directly act on the cytoskeleton of somatotropes via the PLCγ pathway.