SOX2 is required independently in both stem and differentiated cells for pituitary tumorigenesis in p27-null mice

Significance Tumor development can depend on cell intrinsic dysfunction, but, in some cases, extrinsic factors are important drivers. Here, we established a genetically tractable model, demonstrating that the same gene is relevant both cell autonomously and noncell autonomously for tumorigenesis. Deletion of p27, down-regulated in many tumors, predominantly leads to development of murine pituitary tumors. SOX2, transcriptionally derepressed in absence of P27, is important for tumorigenesis in this and other models, but little is known about its interaction. Using loss-of-function and lineage tracing approaches, we establish its regulatory interaction in vivo and show that SOX2 is required independently, both in endocrine and stem cells, to orchestrate tumorigenesis in absence of P27, establishing a powerful model to investigate mechanisms of tumor development.

Humaine, Montpellier) at 1:300, rat anti-CD31 (47) 1:50, p27kip C19 (SantaCruz) 1:300, rabbit anti-ki67 (Abcam) 1:1000. Sections were washed in PBST then incubated for 1 hr at room temperature with the corresponding anti-rat, anti-goat or anti-rabbit secondary antibody conjugated to Alexa-Fluor 488, 555, 594 or 647 in 10% blocking solution with 1 µM 4',6-diamino-2-phenylindole (DAPI). Cell proliferation was analysed following a one-hour EdU pulse in E18.5 embryos (30μg/g body weight), or two intraperitoneal daily injections for 3 consecutive days in adult. Adult pituitaries were harvested at day 4. EdU incorporation was detected using the Click-iT EdU imaging kit following the manufacturer instructions (Thermo Fisher Scientific). Sections were washed with PBST and mounted using Aqua-Poly/Mount (Polysciences, Inc., Warrington, PA, USA). Images were acquired using a Leica SPE confocal microscope or Olympus VS120 Slide Scanner. Settings were established during the initial acquisition. All images taken from the Leica SPE confocal microscope were pre-processed using ImageJ (maximum z-projection) and the ones taken from VS120 Slide Scanner were processed using Qu-path.

Immunohistochemistry
Pituitary glands were fixed in 10% buffered formaldehyde for 16 hours and embedded in paraffin. For staining, 4μm sections were de-paraffinized using xylene and rehydrated through a graded series of ethanol. Antigen retrieval was performed for 20 min at high temperature in either 0.01M citrate buffer (pH6) or Tris-EDTA (10mM Tris base, 1mM EDTA solution, pH9), depending on the antibody. The following antibodies were used: Sox2 (AF2018, R&D), 1:100, on sections treated with antigen retrieval buffer (Ventana) for 48 minutes, followed by a one hour primary antibody incubation. IHC was performed on the Discovery Ultra Ventana platform (Roche). P-ERK1/2 (4370, Cell Signalling Technology) 1:100 O/N at 4ºC, antigen retrieval was performed in the microwave for 23 minutes with 0.01M citrate buffer pH6. Goat anti-rabbit secondary antibody (BA-1000, Vector) was incubated 1:250 for 45 minutes at RT and then signal amplification and HRP detection were performed using the ABC kit (Vectorlabs) for 30 minutes at RT. This IHC was performed manually. Samples were blocked using 1% BSA and incubated overnight at 4°C with the desired antibody, or in blocking buffer for controls. Finally, slides were incubated with the secondary antibody for one hour and washed three times with PBS. For colorimetric staining with diaminobenzidine (DAB) slides were incubated with peroxidase substrate and mounted.

Cell Countings
For proliferation assays quantification, three different fields were chosen on different sections (sections always include IL, SCs flanking the cleft and AL). Within these, the numbers of SOX2-positive, SOX9-positive, EdU-positive, and SOX2;EdU;SOX9 triplepositive cells, representing proliferative stem cells, or SOX9 negative, SOX2;EdU double-positive cells, representing proliferative melanotrophs, were counted blindly and manually. In EdU injected Pomc-CreERT2; Sox2 fl/+ ;p27 -/-;Rosa26 ReYFP/+ mutants, a total of around 1000 PAX7 positive cells was counted from fields randomly chosen and encompassing the whole IL. The number of EdU;PAX7-double-positive or EdU;PAX7;eYFP-triple-positive cells was counted blindly and manually, and the IL surface area was determined using the Fiji software. The Fiji software was used to pseudocolor the different channels in unprocessed micrographs. The Qupath software (48) was also used for image analysis. Briefly, areas of interest were defined after POMC immunostaining for the intermediate lobe, or SOX2;SOX9 for the SC layer. Within the areas of interest, the centre of cell nuclei was then identified as maxima in a filtered DAPI image. Nuclear boundaries were assigned by a propagation algorithm, and then expanded by ~1 micron to define sampling areas. The following data were then recorded: (i) average pixel intensities for each data channel over each sampling area, representing one cell, (ii) the size of the sampling areas; and (iii) the number of positive cells for each data channel over each sampling area.

Pituitary Dissociation for FACS, sc-RNAseq and Sphere Assay
Pituitaries were harvested. The anterior lobe was separated from the intermediate and posterior lobes by cutting through the cleft. The posterior lobe was then removed and the anterior and intermediate lobes incubated separately in a solution of papain (10108014001, Sigma-Aldrich, 1mg/ml in HBSS) for 15 min at 37ºC in presence of 10µg/ml of both DNAse (10104159001, Thermo Scientific) and Rock Inhibitor (M1817, Abmole Bioscience). The papain solution was then removed, and mechanical dissociation was performed on ice in pituisphere medium (18). Pituispheres were derived as previously described (18) and spheres counted manually and blindly after 1 week in culture.

mRNA extraction and reverse transcriptase-PCR (rt-qPCR)
Total mRNA was extracted from dissected IL using the RNeasy Micro kit (Qiagen) according to the manufacturer's protocol. The extracted RNA was reverse-transcribed into cDNA using the Superscript VILO cDNA synthesis kit (Thermo Fisher Scientific, Figure  2) or SMART-Seq v4 Ultra Low Input RNA Kit (Takara Bio USA, Figure 5) according to the manufacturer's protocol.

Quantitative real-time PCR (RT-qPCR)
Each sample was assayed in technical duplicate with each tube containing diluted template cDNA, 250 nM primers and 1xAbsoluteSybrGreen ROX mix (Thermo Fisher Scientific). Each sample was assayed for the genes of interest together with the reference housekeeping gene Gapdh (Supplementary Table 3). Relative expression of the genes of interest was calculated by normalisation of the detected expression value to the geometric mean of the reference genes using the ΔΔCt method (49). Data is shown as mean±SEM with the number of biological samples indicated in each figure. Sidak's ( Figure 1I) and Tukey's multiple comparison tests ( Figure 2H and 5K) were used to assess significance of the data.

Radioimmunoassay (RIA)
Pituitaries were homogenized in phosphate-buffered saline and hormonal contents measured by RIA (50) using National Hormone and Pituitary Program reagents kindly provided by A.L. Parlow for ACTH and GH and using alpha MSH RIA kit (RB303-Invitech).

In-situ hybridization (RNAscope)
Fluorescent in situ hybridisations were performed manually using RNAscope (Multiplex Fluorescent Reagent Kit v2, 320293) on cryosections according to the manufacturer's protocol. Images were analysed and quantified using the Fiji package.

Statistical analyses
Statistical analyses were performed using Prism v.8.0c (GraphPad Software, USA). To examine significance of the data, tests were selected according to the experiment analysed. For survival curves generated using the Kaplan Meier method, log-rank tests were applied (Fig.2B). When comparing two groups of values with normal distribution, unpaired t-test was performed ( Figure . All results are represented as means ± standard deviation (SD) for raw data and means ± standard error of the mean (SEM) for graphs. Standard significance levels were used: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

RNA sequencing sample preparation
Bulk RNA sequencing. RNA was extracted from dissected male and female IL or FAC sorted cells using the RNeasy Micro kit (Qiagen) according to the manufacturer's protocol. RNA quality was assessed using the Agilent RNA 6000 Pico Kit (Agilent Technologies). cDNA was generated using Ovation RNA-seq System V2 (Tecan, 7102-A01), libraries were constructed using Ovation Ultralow System V2 (Tecan, 0344NB-A01) according to the manufacturer's instructions. Libraries were quantified using the TapeStation (Agilent) and pooled in equimolar proportions. Library were sequenced on an Hiseq4000 (Illumina), to achieve an average of 25 million reads per sample. Single-cell RNA sequencing. Male and female IL cells were dissociated as described above. Libraries were generated using Chromium Single Cell 3' kit v3 (10x genomics, 100092) according to the manufacturer's instructions. Both cDNA and libraries were quantified using the TapeStation (Agilent) and sequenced on an Hiseq4000 (Illumina), to achieve an average of 50,000 reads per cell. Detailed aims and animal information is available in Supplementary Table 5.

Bioinformatics analysis
Bulk RNA sequencing. The sequencing was performed on biological duplicates or triplicates for each data point. The RSEM package (version 1.3.30) (51) was used in conjunction with the STAR alignment algorithm (version 2.5.2a) (52) for the mapping and subsequent gene-level counting of the sequenced reads with respect to Ensembl mouse GRCm.38.89 version transcriptome. All parameters for RSEM were run as default except "-forward-prob" which was set to 0.5. Normalisation of raw count data and differential expression analysis was performed with the DESeq2 package (version 1.18.1) (53) within the R programming environment (version 3.4.3) (54). Differentially expressed genes were defined as those showing statistically significant differences between pairwise groups if the adjusted P value was less than 0.05 (FDR < 0.05). Differentially expressed genes were taken forward and their pathway and process enrichments were analysed using Metacore (https://portal.genego.com). A hypergeometric test was used to determine statistical enriched pathways and processes and the associated P-value was corrected using the Benjamini-Hochberg method. Single-cell RNA sequencing. 10x CellRanger (version.3.0.2) was used to generate single cell count data for each genotype (Sox9 iCreERT2/+ ;Rosa26 ReYFP/+ , Sox9 iCreERT2/+ ;p27 -/-Rosa26 ReYFP/+ and Sox9 iCreERT2/+ ;Sox2 fl/+; p27 -/-Rosa26 ReYFP/+ ) using a transcriptome built from the Ensembl mouse GRCm38 release 89. All subsequent analyses were performed in R v.3.6.0 using the Seurat (v3) package (55). Primary filtering was performed on each dataset by removing from consideration: cells expressing fewer than 50 genes and cells for which mitochondrial genes made up greater than 10% of all expressed genes. Each dataset was normalised using the' LogNormalize' function, with a scale factor of 10,000. The top 2000 highly variable genes were found using the 'FindVariableGenes' function and the data centred and scaled using the 'ScaleData' function. PCA decomposition was performed and after consideration of the eigenvalue 'elbow-plots, the first 20 components were used to construct Uniform Manifold Approximation Projection (UMAP) plots. Clusters relating to Melanotrophs and Stem Cells were identified, using the expression of known markers [melanotroph markers : Pomc, Pax7 and Pcsk2; stem cell markers : Sox2 and Sox9], and these clusters were integrated across the three genotypes (Sox9 iCreERT2/+ ;Rosa26 ReYFP/+ , Sox9 iCreERT2/+ ;p27 -/-Rosa26 ReYFP/+ and Sox9 iCreERT2/+ ;Sox2 fl/+; p27 -/-Rosa26 ReYFP/+ ) using Seurat 3's standard integration workflow.
Differentially expressed genes between clusters across genotypes were determined using the 'FindMarkers" function.

Data availability
The RNA-sequencing datasets have been deposited in the Gene Expression Omnibus with the grouped accession number GSE152010. The bulk RNAseq datasets are GSE152007 (p27 -/-IL compared to wild-type, two-and seven-month old) and GSE152008 (p27 -/-IL compared to p27 -/-;Sox2 +/-and wild-type, two-to three-month old); the single-cell RNAsequencing datasets have been deposited with accession number GSE152009. Quantification after RNAscope show that in p27 -/-, SCs flanking the cleft express Sox2 at higher levels (28±6) than and wt (11.4±8, ***p=0.0002). The same is observed in melanotrophs: p27 -/-(12±2) and wt (3.5±2.5, *p=0.0134), (n=5-4 in each group). C) Principal Component Analysis (PCA) plots of bulk RNA-seq data of 2-and 7-month old wildtype and p27 -/-IL samples. There is a clear segregation of samples according to genotype, even before tumorigenesis (2 month-old). Furthermore while wild-type samples segregate according to age, p27 -/-samples appear more similar, independently of the age of the animal. D) GH content were measured by RIA in 2-month old male pituitaries. p27 -/-pituitary glands contain higher GH levels (*p=0.0340, n=4 in each group) which is consistent with gigantism affecting these mutants.