Hippo Signaling Is Essential for the Maintenance of Zona Glomerulosa Cell Fate in the Murine Adrenal Cortex

Abstract

Cells of the zona glomerulosa (zG), the outermost zone of the adrenal cortex, secrete aldosterone and transdifferentiate into glucocorticoid-producing cells of the zona fasciculata (zF) during adrenal homeostasis. However, our understanding of the signaling pathways mediating zG cell maintenance or their transdifferentiation into zF cells is incomplete. Hippo is a major pathway that regulates cell proliferation/differentiation during embryogenesis and postnatal tissue homeostasis. Hypothesizing that Hippo signaling could be involved in zG cell maintenance or transdifferentiation, we generated a mouse model in which the two main kinases of the Hippo signaling cascade large tumor suppressor homolog kinases 1/2 (Lats1 and Lats2) are specifically inactivated in zG cells. Here we show that loss of function of Lats1 and Lats2 impairs zG steroidogenesis and leads to zG cell transdifferentiation into cells sharing characteristics with chondroblasts/osteoblasts rather than zF cells. Furthermore, we demonstrate that this phenotype can be rescued by the concomitant inactivation of the transcriptional coactivators Yes-associated protein (Yap) and transcriptional coactivator with PDZ-binding motif (Taz) with Lats1 and Lats2. Finally, we show that expression of a constitutively active form of YAP (YAP5SA) in zG cells does not alter their fate as severely as the loss of Lats1 and Lats2 but leads to adrenal hyperplasia. Together, these findings highlight the critical role of Hippo signaling in maintaining zG cell fate and function and provide key insights into broader mechanisms underlying cellular differentiation.

The cortex of the adrenal gland is organized into distinct concentric zones, each responsible for producing and secreting specific steroid hormones essential for maintaining body homeostasis. In mice, the adrenal cortex is formed of 2 zones: the outermost zona glomerulosa (zG), which regulates sodium/potassium balance through mineralocorticoid synthesis, and the inner zona fasciculata (zF), which mediates stress response through glucocorticoid synthesis (1). Maintenance of proper zonation is, therefore, crucial throughout life.

Maintenance of the adrenal cortex involves the centripetal migration of capsular and subcapsular stem and progenitor cells, which first differentiate into zG cells. The zG cells then lose their capacity to synthesize aldosterone as they migrate inward and transdifferentiate into glucocorticoid-producing cells of the zF (2, 3). Transdifferentiation of zG cells into zF cells is mainly regulated by the antagonistic actions of WNT/β-catenin (CTNNB1) and protein kinase A (PKA) signaling, with WNT/CTNNB1 promoting the maintenance of zG cell identity and PKA driving their differentiation into zF cells (4-6). Although these pathways are critical for zG cell maintenance or transdifferentiation, it was recently demonstrated that SUMOylation (7, 8), epigenic programming (9), and Gq signaling (10) also contribute to these processes, suggesting that the mechanisms regulating the balance between zG cell fate maintenance and their transdifferentiation into zF cells are not fully understood.

The Hippo signaling pathway regulates cell proliferation and differentiation via a kinase-signaling cascade. Upon activation by intracellular or extracellular signals, the kinases STE20-like protein kinases 1 and 2 (MST1 and MST2) (11) are first phosphorylated and activated. Following their activation, they phosphorylate and activate the large tumor suppressor homolog kinases 1 and 2 (LATS1 and LATS2), which, in turn, phosphorylate and inactivate the redundant transcriptional coactivators Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ). Upon cascade inactivation, YAP/TAZ accumulate in the nucleus, where they interact with transcription factors to regulate the transcription of target genes (12, 13). While most studies to date have either focused on the role of Hippo signaling in cell fate determination, proliferation, and differentiation during embryogenesis or in tumor formation (12-15), recent studies have demonstrated that the Hippo signaling pathway is also essential for postnatal cell fate maintenance of several cell types including resting cardiac fibroblasts (16), granulosa cells (17), and hepatocytes (18).

Using the Nr5a1-cre model to inactivate components of the Hippo signaling pathway in adrenocortical cells, we previously showed that the inactivation of Yap/Taz in mouse adrenocortical cells leads to the degeneration of the adrenal cortex in males, potentially associated with the depletion of the subcapsular progenitor cell population (19). Furthermore, we showed that while the concomitant loss of Mst1/2 in adrenocortical cells had only a marginal impact on the transcriptional activity of zG cells (20), the concomitant inactivation of Lats1/2 induced the transdifferentiation of perinatal adrenocortical cells into myofibroblast-like cells (21). In the latter model, since transdifferentiation of the adrenocortical cells occurs before the establishment of zonation, the role of Lats1/2 and Hippo signaling could not be assessed in zG cells. To overcome this limitation, in this study, we inactivated Lats1/2 in aldosterone-producing zG cells. Our data reveal that the loss of Lats1/2 and the resulting increase in the transcriptional activity of YAP/TAZ in zG cells alter their identity, affect steroid synthesis, and block their transdifferentiation into zF cells. Furthermore, our study demonstrates that the expression of a stabilized form of human YAP in zG cells does not alter their fate as severely as the loss of Lats1/2 but leads to adrenal hyperplasia.

Material and Methods

Mice

All animal procedures were approved by the Comité d'Éthique de l'Utilisation des Animaux of the Université de Montréal (protocol numbers Rech-1739/Rech-1909) and conformed to the guidelines of the Canadian Council on Animal Care. All mouse lines AS^cre/+ (3), Lats1^flox/flox and Lats2^flox/flox (22, 23), Yap^flox/flox and Taz^flox/flox (24, 25), Rosa26^YAP5SA (26), Rosa^mT/mG (27) have been described previously and were maintained on a C57BL6/J background except Rosa26^YAP5SA, which were maintained on a mixed FVB/Swiss Webster background. Rosa26^YAP5SA/+; AS^cre/+ mice were therefore on a mixed FVB/Swiss Webster/C57BL6/J mixed background as F1 mice were used for this model. Genotype analyses were done on tail biopsies by PCR as previously described (21).

Histopathology, Immunohistochemistry, and Detection of Fluorescence

Isolated adrenal glands for light microscopy histopathologic analysis were fixed in 10% neutral buffered formalin overnight. Tissues were embedded in paraffin, sectioned (5 μm), and stained with hematoxylin and eosin (H&E), Masson's trichrome or Alizarin red. Immunohistochemistry was done using ImmPRESS horseradish perioxidase horse anti-rabbit (RRID: AB_3148616) and horse anti-mouse IgG polymer detection kit (RRID:AB_3148615) (Vector Lab) as directed by the manufacturer. Sections were probed with primary antibodies against cleaved caspase-3 (1:50, Cell Signaling, RRID: AB_234188), CTNNB1 (1:50, Cell Signaling, RRID:AB_11127855), cytochrome P450, family 11, subfamily B, polypeptide 1 (CYP11B1) (1:300, RRID:AB_2687896), disabled homologue 2 (DAB2) (1:50, BD Biosciences, RRID:AB_397837), KI67 (1:1500, Abcam, RRID:AB_443209), nuclear receptor subfamily 2 group F member 2 (NR2F2) (1:200, Perseus Proteomics, RRID:AB_2314222), platelet and endothelial cell adhesion molecule 1 (1:100, Cell Signaling, RRID:AB_2722705), macrophage scavenger receptor 1 (MSR1/SCARA1/CD204) (1:200, Thermofisher, RRID:AB_2785556), sex determining region Y-box 9 (SOX9) (1:200, Cell Signaling, RRID:AB_2665492), Wilms tumor 1 (1:300, Cell Signaling, RRID:AB_2800020), YAP (1:100, Cell Signaling, RRID:AB_2650491), TAZ (1:300, Cell Signaling, RRID:AB_2904134), and vimentin (1:100, Cell Signaling, RRID:AB_10695459). Slides were visualized using an axio imager M1 (Zeiss). Natural fluorescence of reporter proteins was visualized using a modified version of a previously described protocol (28). Briefly, adrenals were fixed in prechilled 95% ethanol for 20 hours in the dark at 4 °C with agitation. Adrenals were then dehydrated into 2 prechilled 100% ethanol for 1 hour each in the dark at 4 °C. Ethanol was then cleared from the adrenals with 2 washes in prechilled toluene for 45 minutes each in the dark at 4 °C. Adrenals were then permeated in paraffin baths for 50 minutes each at 56 °C. Blocks of paraffin tissue were stored in the dark at 4 °C until use. Five μm sections were deparaffinized using subsequent washes of prechilled toluene (1 minute, 2×), 95% ethanol (1 minute, 3×), and PBS (1 minute, 3×) and mounted with 4',6-diamidino-2-phenylindole (Vectashield, Vector Lab) in the dark. Fluorescence was visualized using an IX81 confocal microscope (Olympus Life Science).

LacZ Staining

Isolated adrenal glands were fixed in 4% paraformaldehyde for 1 hour at 4 °C with agitation, washed 3 times for 20 minutes in rinse solution [0.1% sodium deoxycholate, 0.2% IGEPAM, 2 mM MgCl2, 0.1 M phosphate buffer (pH 7.3)], and incubated in the dark at 37 °C overnight in staining solution [1 mg/mL X-Gal (Promega), 5 mM potassium ferricyanide (Sigma), 5 mM potassium ferrocyanide (Sigma)], diluted in rinse solution. Adrenal glands were fixed in formalin overnight and then transferred into 70% ethanol solution overnight. Tissues were embedded in paraffin, sectioned (5 μm), and stained with eosin.

RNA In Situ Hybridization

Adrenal glands were fixed in formalin overnight and transferred into 70% ethanol before embedding. Double-labeled RNAish was performed using a precustom or custom-designed probe spanning the deleted exons of Lats1 and Lats2 (RNAscope® Probe—Mm-Lats2-O1-C1, RNAscope® Probe—Mm-Lats1-O1-C2) and the RNAscope® 2.5 HD Duplex Reagent Kit (Advanced Cell Diagnostics) according to the manufacturer's instructions.

RT-qPCR

Total RNA from adrenal glands was extracted using the RNeasy mini kit (Qiagen). RNA was reverse transcribed using 100 ng of RNA and the SuperScriptVilo™ cDNA synthesis kit (Thermo Fisher Scientific). Real-time PCR reactions were run on a CFX96 Touch instrument (Bio-Rad) using Supergreen Advanced qPCR MasterMix (Wisent). To quantify relative gene expression, the cycle threshold (Ct) of genes of interest was compared to the Ct of ribosomal protein L19 (Rpl19), according to the ratio R = [E^{Ct Rpl19}/E^{Ct target}] where E is the amplification efficiency for each primer pair. Rpl19 Ct values did not change significantly between tissues, and Rpl19 was therefore deemed suitable as an internal reference gene. The specific primer sequences used are listed in Supplementary Table S1 (29).

Cell Quantification

Quantification of total cells/surface and Ki67+ cells/total cells was performed on scanned whole adrenals using an Aperio AT2 microscope slide scanner (Leica) and QuPath software version 0.3.2 (Bioimage Analysis, RRID:SCR_018257). Annotations were made of whole adrenals, adrenal cortex, and capsule, and the positive cell detection feature was used to identify Ki67-positive cells with a threshold set to avoid quantification of background.

Hormone Measurements

Blood samples for plasma collection were obtained from isoflurane-anesthetized mice. Samples were collected by cardiac puncture prior to euthanasia between 9:30 and 10:00 Am in 2K-EDTA microvette tubes (Sarstedt) and centrifuged at 2000g for 15 minutes at 4 °C. Plasma samples were transferred to polypropylene tubes and stored at −80 °C until analysis. Corticosterone and ACTH analyses were performed in singlet by the Center for Research in Reproduction at the Ligand Assay and Analysis Core Laboratory of the University of Virginia. Corticosterone levels were determined by radioimmunoassay (MP Bio, Cat# 07120102, RRID:AB_2783720); ACTH levels in the plasma were determined by immulite 2000 using a dead volume of 300 μL (Siemens Healthineers Global, RRID:AB_2783635). Aldosterone and renin analysis were performed in singlet in house using the ab136933-Aldosterone ELISA kit (Abcam, RRID:AB 2895004) and EA100739 mouse Renin 1 Elisa Kit (Origene, RRID:AB_3564402). Plasma volume needed for each assay was 20 μL for corticosterone (run at a 1:1 dilution), 45 μL (run at a 1:10 dilution) for ACTH, 12.5 μL (run at a 1:8 dilution) for aldosterone, and 6.25 μL (run at a 1:16 dilution) for renin.

RNA Sequencing Analyses

Total RNA was extracted from adrenal glands (n = 3 from mutants and control littermates) using the RNeasy mini kit (Qiagen), and RNA sequencing (RNA-seq) was performed by the genomics core facility of the Institute for Research in Immunology and Cancer (Montreal, Quebec, Canada) as previously described (30). BAM files were generated from the sequencing platform using STAR, and reads were sequenced and aligned to the reference mouse genome version GRCm38 (gene annotation from Gencode version M23, based on Ensembl 98) using the SeqMonk NGS visualization and analysis tool (v1.48.1, Babraham Institute). Sequencing for 1 of the Y5SA control samples had lower complexity than the other samples (38% fewer read counts) and was therefore not used for the subsequent analyses. However, this sample is included in the RNA-seq data deposited in the GEO database (accession number GSE227353). Gene expression was obtained using the RNA-seq quantitation pipeline in SeqMonk and was normalized into reads per kilobase of transcripts per million mapped reads values and log2 transformed. DeSeq2 version 1.22.2 was then used to detect differential expression. Thresholds used for P-value and fold change were .05 and ±1.5. RNA-seq data were further analyzed using the Metascape gene annotation and analysis resource (RRID:SCR_016620) to evaluate the biological processes regulated by the up- or downregulated genes (1.5-fold or more) and DAVID 2021 (RRID:SCR_001881) (31, 32) annotation and analysis resources and by conducting Gene Set Enrichment Analysis (GSEA) using GSEA 4.3.2. (RRID:SCR_003199). Permutations were set to 1000 and performed on gene sets, and GSEA output was displayed as enrichment plots. Heatmaps were generated by using the online open access Morpheus matrix visualization tool (Broadinstitute, RRID:SCR_017386). Genes used for heatmap, GSEA, or gene ontology (GO) term analyses were curated from the literature (33) and the GSEA analysis database for inflammation, macrophage, and phagocytosis associated genes; from the GSEA analysis database for mesenchymal cells and osteoblast/chondroblast; and from the GSEA analysis database and curated from the literature (34) for Sertoli cells [Supplementary Table S2 (29)]. Finally, Hippo signaling associated genes were identified by comparing RNA-seq/microarray data from 4 different studies either inactivating Lats1/2 (35, 36) or overexpressing YAP or TAZ (37, 38) for Hippo signaling associated genes. For this latter category, genes were selected if they were overexpressed in at least 3 out of the 4 models analyzed. The pairwise Venn diagrams were created in R using the venneuler package.

Statistical Analyses

All statistical analyses were performed with Prism software version 7 (GraphPad Software Inc., RRID: SCR_002798). A two-tailed Student's t-test was performed to compare 2 groups unless an F-test indicated that the groups had different variances. In such cases, a two-tailed Welch t-test was performed. For the comparison of 3 groups, a one-way ANOVA followed by Tukey's post hoc test was performed for normally distributed samples. For nonnormally distributed samples, a Kruskal–Wallis test was used along with planned comparisons using Dunn's test. Means were considered significantly different when P-values were <.05. All data are presented as means ± SEM.

Results

Inactivation of Lats1/2 in Steroidogenic zG Cells Leads to Adrenal Dysplasia

To investigate the role of Hippo signaling in zG cells, we first crossed mice bearing floxed alleles for Lats1/2, the core kinases of Hippo signaling, with the AS^cre/+ strain, which targets the aldosterone-producing cells of the zG to generate Lats1^flox/flox; Lats2^flox/flox; AS^cre/+ (referred herein as L1;L2;AS) mice. Adrenals from L1;L2;AS mice appeared normal at the gross level at all ages evaluated [Supplementary Fig. S1A (29)], and L1;L2;AS mice had normal adrenal somatic indices (adrenal gland weight/body weight) at all ages assessed except in 6 month-old females, in which a small decrease was observed [Supplementary Fig. S1B and S1C (29)].

Contrary to what was observed at the gross level, an abnormal phenotype was observed at the histological level in the adrenal cortex of L1;L2;AS mice [Fig. 1; Supplementary Fig. S1D-S1G (29), Supplementary Fig. S2 (29)]. In males, adrenal cortexes from L1;L2;AS mice were initially indistinguishable from their age-matched controls (Fig. 1). However, by 2 weeks of age, a thin extracellular matrix (ECM) layer could be observed inside the zG of mutant animals (Fig. 1, arrow). By 4 weeks of age, the zG of L1;L2;AS mice was disorganized [Fig. 1; Supplementary Fig. S1D (29)], and an increase in ECM accumulation was observed in the zG and the outer zF [Fig. 1, arrow; Supplementary Fig. S1D (29)]. ECM accumulation was also observed in the inner zF of the most affected animals [Supplementary Fig. S1E, arrow (29)]. Furthermore, a thickening of the capsule [Fig. 1; Supplementary Fig. S1D (29)] and a disorganization of the vascularization [Fig. 1; Supplementary Fig. S1E (29)] were also observed in L1;L2;AS mice, as highlighted by platelet-endothelial cell adhesion molecule 1 expression pattern [Supplementary Fig. S1F (29)]. The phenotype observed in older L1;L2;AS males was similar to the 1 observed in 4-week-old mice. However, the zF was less affected in most adrenals, and some ECM-rich regions were often found deeper in the zF (Fig. 1, arrow), suggesting that these regions were displaced by centripetal migration. A phenotype similar to the 1 observed in L1;L2;AS males was also observed in L1;L2;AS females [Supplementary Fig. S2A (29)]. However, adrenal cortexes were usually less affected from 14 weeks of age onward [Supplementary Fig. S2A (29)]. Nonetheless, small ECM regions were frequently observed at the cortico-medullary junction of older mice [Supplementary Fig. S2B (29)]. As adrenal cortexes were less affected in older females, males were used for subsequent experiments unless otherwise stated.

Loss of Lats1/2 in zG Cells Leads to Their Transdifferentiation into ECM-producing Cells and Impairs the Zonation Process

As adrenal cortex renewal normally occurs via centripetal migration and zonal transdifferentiation (2, 3), we then decided to determine the fate of the zG Lats1/2-negative cells and the impact of the loss of Lats1/2 on zonation. First, to evaluate the fate of the recombined cells, we performed a lineage-tracing experiment. To do this, L1;L2;AS mice were crossed with Rosa^mTmG/mTmG;Lats1^flox/floxLats2^flox/flox mice to generate Rosa^mTmG/+;Lats1^flox/flox;Lats2^flox/flox;AS^cre/+ animals (referred herein as R;L1;L2;AS) in which tdTomato fluorescence is replaced by EGFP fluorescence in recombined cells and their descendants. In 4-week-old males, a mixture of recombined (green) and unrecombined (red) cells was observed in the zG (Fig. 2), while cells present in the ECM observed in the zG and upper zF were also recombined (Fig. 2, arrow). Finally, a few recombined cells could also be seen at the cortico-medullary junction (Fig. 2, arrowhead). Similar results were obtained in 14-week-old males and 28-week-old males, although globally the recombined cells surrounded by the ECM were found deeper in the cortex (Fig. 2, arrow), as was observed by H&E staining (Fig. 1). These results contrast with what is observed in Rosa^mTmG/+;AS^cre/+ control mice (referred herein as R;AS), in which recombined cells progressively replaced unrecombined cells by centripetal migration (Fig. 2), and suggest that Lats1/2-negative zG cells are unable to transdifferentiate into zF cells and instead transdifferentiate into ECM-producing cells. As adrenocortical renewal is sexually dimorphic and occurs faster in females, we then also performed lineage-tracing experiments in 14-week-old females, an age at which complete adrenal renewal should have occurred (2) (Fig. 2). As was observed in males, recombined cells were mainly present in the upper adrenal cortex (Fig. 2) and in cells present in the ECM (Fig. 2, arrow). Finally, RNAscope analyses also confirmed the H&E and lineage-tracing experiments, as loss of Lats1/2 was observed in rare cells of the zG in 2-week-old mutant mice [Supplementary Fig. S3A (29)], whereas Lats1/2-negative cells were only observed in the zG and cells surrounded by the ECM in 14-week-old males and females [Supplementary Fig. S3A and S3B (29)]. Taken together, lineage-tracing experiments and RNAscope analyses suggest that Lats1/2-negative zG cells transdifferentiate into ECM-producing cells rather than into zF cells.

To evaluate how the abnormal transdifferentiation of zG cells affected the zonation of the adrenal cortex, we then performed immunohistochemistry for the zF marker CYP11B1 (Fig. 3A) and 2 zG markers, CTNNB1 and DAB2 (Fig. 3B and 3C). In L1;L2;AS mice, CYP11B1 was, as expected, expressed in the zF (Fig. 3A). Cells expressing CYP11B1 were also observed between the zG and the ECM regions (Fig. 3A, red arrow), suggesting that either some newly differentiated zG cells were able to transdifferentiate to zF cells or that some progenitor/stem cells could directly differentiate into zF cells. CTNNB1 expression was limited to the zG, although the zone defined by its expression was slightly wider and its expression was reduced in L1;L2;AS mice (Fig. 3B), again suggesting that loss of Lats1/2 affects the identity of zG cells. However, contrary to CTNNB1, DAB2 expression remained elevated in the zG of L1;L2;AS mice, and DAB2+ cells were also found in some cells surrounded by ECM (Fig. 3C, arrowhead). Furthermore, the zone expressing DAB2 was wider, and DAB2 expression was also detected in some zF cells in 14-week-old L1;L2;AS mice (Fig. 3C, arrow).

To determine if the impairment of the zonation process also affected the capsular stem cell population and the subcapsular progenitor cell population, the expression of markers of these cell populations was first evaluated by RT-qPCR. Interestingly, an increase in the expression markers of the capsular stem cells (Gli1, Rspo3, Wnt2b) but not of the subcapsular progenitor cells (Axin2, Shh) was observed in L1;L2;AS mice [Supplementary Fig. S4A (29)], suggesting that capsular stem cells could be trying to replenish the adrenal cortex in L1;L2;AS mice. Immunohistochemistry for NR2F2 further revealed that cells positive for the capsular stem cell marker NR2F2 were found both in the capsule and in the subcapsular region in L1;L2;AS mice [Supplementary Fig. S4B (29)]. Finally, an increase in the proliferation index (determined by counting Ki67-positive cells) was also observed in the capsule of L1;L2;AS mice [Supplementary Fig. S5 (29)]. However, an increase in the proliferation index was also observed in the adrenal cortex of L1;L2;AS mice [Supplementary Fig. S5 (29)], suggesting that cell populations other than the capsular stem cell population may also be involved in adrenal replenishment and zF maintenance in these mice. Taken together, expression of capsule, zG and zF markers, and adrenal proliferation indices all suggest that adrenocortical renewal is affected in the adrenal cortex of L1;L2;AS mice.

Adrenal Steroidogenesis Is Affected in L1;L2;AS Mice

To determine how the alterations observed in L1;L2;AS mice affected the function of the adrenal cortex, the expression of steroidogenic genes and levels of adrenal hormones were assessed in 4- and 14-week-old males. Steroidogenic genes specific to the zG were downregulated in mutants at both ages, while genes expressed in both the zG and zF were solely downregulated in 4-week-old mice and genes specific to the zF were not affected at either age (Fig. 4A and 4B), suggesting that the loss of Lats1/2 primarily affects zG cells. However, circulating aldosterone levels were not reduced in L1;L2;AS males compared to L1;L2 and AS control mice (Fig. 4C). As this result was unexpected, renin levels were then evaluated to determine if potential defects in aldosterone secretion could be mitigated in the mutant model by an increase in renin-angiotensin activity. Circulating renin levels were similar in 4-week-old mutant and control mice but were increased in 14-week-old animals (Fig. 4D), confirming the dysfunction of zG cells in mutant mice. Although the expression of steroidogenic zF specific genes was not affected at either age, the ACTH/corticosterone feedback loop was compromised in 4-week-old mutant mice but not in older animals (Fig. 4E and 4F). This result is consistent with the observation that the zF appears more affected in younger mice [Fig. 1, Supplementary Fig. S1 (29)].

Lats1/2 Inactivation Affects the Identity of zG Cells

To determine the global impact of the loss of Lats1/2 expression in the adrenal cortex, we then conducted bulk RNA-seq analyses on the adrenal glands of 4-week-old L1;L2;AS and L1;L2 males, followed by GO analyses using Metascape. Adrenals from 4-week-old mice were selected, as sexual dimorphism is not a factor at this age and the phenotype was similar in males and females. Four hundred ninety-six genes were downregulated, and 1837 genes were upregulated by 1.5-fold or more in the adrenal cortex of mutant mice. GO analyses indicated that the downregulated genes were associated with regulation of the metabolism of lipids and steroid hormones [Supplementary Fig. S6A (29)], which is consistent with our steroidogenesis analysis. Interestingly, only a subset of zG-specific genes and genes know to be overexpressed in WNT/Cy11b2-high expressing cells (39) were downregulated in mutant mice [Supplementary Fig. S6B and S6C (29)].

As our lineage-tracing experiments showed that zG cells transdifferentiate into ECM-producing cells in the L1;L2;AS model, we then performed GO analyses on the upregulated genes to try to determine the identity of these cells. Consistent with the phenotype observed in mutant mice, sets of genes associated with ECM organization were among the most upregulated gene sets [Supplementary Fig. S7 (29), Supplementary Fig. S8A and S8B (29)]. Numerous sets of transcripts were also involved in cell fate commitment (ie, regulation of epithelial to mesenchymal transition, mesenchyme development, cell fate commitment, negative regulation of cell differentiation) and tissue development (ie, heart development, ossification/regulation of bone remodeling/skeletal system development, lung development, reproductive tissue development) [Supplementary Fig. S7 (29)]. We therefore decided to evaluate some of these gene sets in greater detail. First, we decided to evaluate if cells of the zG gain characteristics of mesenchymal cells, as several studies have demonstrated that inactivation of Lats1/2 in various cells, including in the developing adrenal cortex (21), leads these cells to maintain or acquire characteristics of mesenchymal cells (40, 41) that can further transdifferentiate in other cell types (17, 21, 36, 42, 43). Consistent with what was observed in other tissues, GSEA analyses demonstrated a marked enrichment of genes associated with mesenchymal cells in the adrenal gland of L1;L2;AS mice [Fig. 5A, Supplementary Fig. S9 (29)]. Furthermore, a marked increase in the expression of vimentin, a marker of mesenchymal cells (44), was also observed in the zG of L1;L2;AS mice (Fig. 5B), confirming that Lats1/2-negative zG cells acquire characteristics of mesenchymal cells.

Because inactivation of Lats1/2 in ovarian granulosa cells, which share a common developmental origin with the adrenal cortex, leads to their transdifferentiation into osteoblast and Sertoli cells (17), we then decided to further evaluate whether zG cells could also transdifferentiate into these cells. GSEA analyses demonstrated that a subset of genes involved in the differentiation of chondroblasts and osteoblasts, both common [Fig. 6A, 6C, and 6D, Supplementary Fig. S9 (29)] and specific [Supplementary Fig. S9 (29)] to each cell type, were increased in mutant mice. However, the ECM was rich in collagen but did not calcify (Fig. 6B). Only a few genes expressed in Sertoli cells were upregulated in the adrenal cortex of mutant mice (Fig. 6C and 6D). However, 2 of these genes, Dhh and Dmrt1, are normally almost exclusively expressed in the testis (45, 46). Finally, SOX9, a gene involved in early stages of Sertoli cell (47) and chondroblast/osteoblast differentiation (48), was also expressed in some ECM-producing cells (Fig. 6E). Taken together, these results suggest that Lats1/2-negative zG cells gain some characteristics of other cell lineages but cannot fully transdifferentiate into a specific cell type.

Although numerous gene sets were involved in ECM formation and tissue development, GO analyses also showed that some of the most enriched gene sets were associated with an inflammatory/immune response [Fig. 7A and 7B; Supplementary Fig. S7 (29); Supplementary Fig. S10 (29)]. An increase in the number of cells positive for SCARA1/CD204, a macrophage surface receptor involved in the clearance of apoptotic cells, could also be observed in the zG and surrounding the ECM in the adrenal cortex of mutant animals (Fig. 7C, arrow). However, surprisingly, only rare apoptotic cells were observed at the cortico-medullary junction in 4-week-old mutant animals [Supplementary Fig. S11 (29)], despite the presence of phagocytic macrophages. Nonetheless, these results suggest that macrophages might be attempting to clear abnormal transdifferentiating zG cells.

Elevated YAP/TAZ Activity Causes the Phenotype Observed in L1;L2;AS Mice

As previously mentioned, inactivation of the Hippo signaling pathway leads to an increase in the transcriptional activity of YAP/TAZ. To determine if this was the case in the adrenal cortex of L1;L2;AS mice, the expression of known downstream transcriptional targets of YAP/TAZ was first evaluated. As expected, the expression of the majority of these targets was upregulated in L1;L2;AS mice (Fig. 8A and 8B). To further determine if the activation of YAP/TAZ activity was responsible for the phenotype observed in the mutant animals, we then generated a quadruple knockout mice model: Yap^flox/flox; Taz^flox/flox; Lats1^flox/flox; Lats2^flox/flox; AS^cre/+ (referred herein as Y;T;L1;L2;AS). Histopathological analysis shows that zG no longer transdifferentiates into ECM-producing cells (Fig. 8C) following the concomitant loss of Yap and Taz (Fig. 8D) with Lats1/2 (Fig. 8E). Immunohistochemistry for CTNNB1, DAB2, and CYP11B1 further suggested that zonation was normal in these animals [Supplementary Fig. S12 (29)]. Together, these results confirmed that an increase in YAP/TAZ activity was directly responsible for the abnormal transdifferentiation of the zG cells observed in L1;L2;AS mice.

Increase in YAP Activity Causes Adrenal Cortex Hyperplasia

As YAP and TAZ are often considered to be redundant, we decided to determine if increasing the transcriptional activity of YAP was sufficient to recapitulate the phenotype observed in L1;L2;AS mice. To do this, we generated the mouse model Rosa26^YAP5SA/+; AS^cre/+ (referred herein as Y5SA;AS). The Rosa26^YAP5SA allele consists of a constitutively active form of YAP1, the human ortholog of YAP (in which the 5 canonical LATS phosphorylation sites are mutated to prevent inhibition/degradation), followed by a C-terminal IRES-nuclear LacZ under the control of a CAGGS promoter targeted to the Rosa26 locus.

In contrast to what was observed in L1;L2;AS mice, adrenal glands from Y5SA;AS male and female mice were larger than their control counterparts from 8 weeks of age onward [Fig. 9A; Supplementary Fig. S13A-S13D (29)]. Histopathologic evaluation of the adrenal cortex of Y5SA;AS further demonstrated a progressive disorganization of the zG (Fig. 9B and 9C). Similar to what was observed in L1;L2;AS mice, ECM accumulation was also present in the adrenal cortex of Y5SA;AS mice (Fig. 9D). However, these ECM regions were usually fewer and smaller than the ones found in L1;L2;AS mice.

To determine if the size increase of the adrenal gland was caused by cellular hypertrophy or hyperplasia, histomorphometric analysis was performed on Y5SA;AS males. This analysis demonstrated that the cellular density (number of nuclei per surface area) was similar in the cortex of Y5SA;AS mice and their control counterparts [Supplementary Fig. S14A (29)], indicating that the expansion of the adrenal cortex in these mutant mice resulted from hyperplasia rather than hypertrophy. A small but significant increase in the adrenocortical proliferation index was also observed in mutant mice [Supplementary Fig. S14B (29)]. However, unlike what was observed in L1;L2;AS mice, the proliferation index of capsular cells did not increase in Y5SA;AS mice [Supplementary Fig. S14B (29)]. Consistent with this finding, the expression of most capsular markers did not increase in Y5SA;AS mice, although the expression of Wnt2b significantly increased in 14-week-old animals [Supplementary Fig. S14C and S14D (29)]. A small but significant increase in Axin2 and Shh expression was also observed in 4-week-old mutant mice and in 14-week-old animals for Axin2 [Supplementary Fig. S14C and S14D (29)], suggesting that progenitor cells might contribute to the increase in proliferation observed in the adrenal cortex of Y5SA;AS mice. However, the increase in KI67+ cells was observed throughout the cortex [Supplementary Fig. S14B (29)], suggesting that other cortical cells proliferate and contribute to the observed hyperplasia. Finally, unlike what was observed in L1;L2;AS mice, aldosterone/renin and corticosterone/ACTH levels did not differ between Y5SA;AS and control mice [Supplementary Fig. S15A-S15D (29)]. Nonetheless, the expression of zG-specific steroidogenic genes also decreased in Y5SA;AS mice [Supplementary Fig. S15E and S15F (29)], although this decrease was less pronounced in younger animals [Supplementary Fig. S15E and S15F (29)].

To further characterize Y5SA;AS mice, b-galactosidase (LacZ) staining was performed to identify the cells in which recombination occurred. Similar to what was observed in L1;L2;AS mice, LacZ expression was already detectable in the zG, the outer half of the zF, and in some cells at the cortico-medullary region by 4 weeks of age in Y5SA;AS males (Fig. 10A), a pattern that was maintained in 14-week-old mutant males (Fig. 10A). At 30 weeks of age, fewer LacZ+ cells were present in the adrenal cortex of mutant animals (although these cells could be seen throughout the cortex in some regions) (Fig. 10A), suggesting a greater heterogeneity of recombined and nonrecombined zG cells (and their zF descendants) at this age (Fig. 10A). Despite the fact that a similar phenotype was observed in male and female Y5SA;AS mice [Supplementary Fig. S16A (29)], recombination appeared to be less efficient in females [Supplementary Fig. S16B (29)].

To evaluate adrenocortical zonation and YAP activity in Y5SA;AS mice, immunohistochemistry for CTNNB1, DAB2, and CYP11B1 was performed. Similar to what was observed in L1;L2;AS mice, CTNNB1 expression was limited to the zG and appeared slightly reduced (Fig. 10B), while the zone delimited by DAB2+ cells expanded more deeply in the cortex (Fig. 10B). CYP11B1 was expressed in zF cells; however its expression was heterogeneous (Fig. 10B). To determine if the YAP5SA allele was specifically expressed in CYP11B1− cells, CYP11B1 staining was then performed on b-galactosidase-stained tissues. In these analyses, LacZ− cells were found to express CYP11B1 [Supplementary Fig. S17, red arrow (29)], whereas the majority of LacZ+ cells, including cells present in the ECM, did not [Supplementary Fig. S17, green arrow (29)]. Nonetheless, a few LacZ+ cells expressed CYP11B1 [Supplementary Fig. S17, black arrow (29)], suggesting that some cells expressing YAP5SA can still transdifferentiate into zF cells.

Constitutive Expression of YAP Induces an Inflammatory/Immune Response

To better determine the differences and similarities observed between the Y5SA;AS and the L1;L2;AS models, we decided to perform RNA-seq analyses on the adrenal glands of 4-week-old Y5SA;AS and Y5SA mice and then compared these data to the RNA-seq analyses performed on L1;L2;AS mice. One hundred nineteen genes (vs 496 in L1;L2;AS mice) were downregulated and 962 genes (vs 1837 in L1;L2;AS mice) were upregulated by 1.5-fold or more in the adrenal cortex of Y5SA;AS animals (Fig. 11A and 11B). Out of the 119 downregulated genes, 23% (27/119) were also downregulated in L1;L2;AS mice (Fig. 11A). Although the regulation of hormone levels was among the functions regulated by these genes [Supplementary Fig. S18A (29)], only rare zG-related genes were downregulated by 1.5-fold or more in 4-week-old Y5SA;AS mice [Supplementary Fig. S18B (29)], suggesting that, at this age, constitutive expression of YAP did not alter zG cell function as severely as Lats1/2 loss. However, qPCR analyses performed on older animals suggest that loss of zG cell identity might only be delayed in Y5SA;AS mice compared to L1;L2;AS mice [Supplementary Fig. S18C and S18D (29)].

Contrary to the downregulated genes, 74% (714/962) of the upregulated genes observed in Y5SA;AS mice were also upregulated in L1;L2;AS mice (Fig. 11B). Looking more specifically at Hippo signaling related genes, 24/37 genes were upregulated in both models (Fig. 11C and 11D). A subset of genes regulating the inflammatory/immune response in L1;L2;AS mice were also increased in Y5SA;AS mice [Supplementary Fig. S19 (29); Supplementary Fig. S20 (29)], which was corroborated by the increase in the number of SCARA1+ cells (Fig. 11E). Finally, a subset of genes associated to the ECM, mesenchymal cells, development of chondroblasts/osteoblasts, and Sertoli cells upregulated in the adrenal of L1;L2;AS mice was also upregulated in Y5SA;AS mice [Supplementary Fig. S21 (29); Supplementary Fig. S22 (29)]. Taken together, these results confirm that YAP is a key effector acting downstream of Lats1/2 but that the overactivation of its transcriptional activity alone is insufficient to fully recapitulate the phenotype observed in L1;L2;AS mice.

Discussion

The adrenal cortex is maintained throughout postnatal life by pools of capsular stem cells and subcapsular progenitor cells that adopt different functional steroidogenic states as they migrate centripetally through the cortex. While the essential roles of signaling pathways such as WNT and PKA signaling in the progressive proliferation/transdifferentiation of these cells have been clearly demonstrated (5, 49), other pathways such as Gq signaling (10) can also contribute to the maintenance of proper zonal steroidogenic cell differentiation. In this report, we have shown that the Hippo signaling pathway is essential to maintain the steroidogenic capacity and identity of zG cells and their subsequent transdifferentiation into zF cells by inhibiting their ectopic transdifferentiation into cells that share characteristics with cell lineages of mesenchymal origin like chondroblasts/osteoblasts and the activation of an inflammatory/immune response.

Two transgenic mouse models were generated to evaluate the effect of Hippo signaling inactivation in the zG; L1;L2;AS mice in which the main kinases of Hippo signaling are inactivated and Y5SA;AS mice, which are characterized by the expression of a constitutively active form of YAP, the main Hippo signaling effector. Interestingly, expression of YAP5SA led to adrenal hyperplasia and did not fully recapitulate the phenotype observed following the loss of Lats1/2 even though genes involved in the inflammatory response; several of the main Hippo signaling transcriptional targets; and several genes associated with mesenchyme development, chondroblasts, and osteoblasts were regulated in a similar fashion in both models. Several hypotheses might explain why expression of YAP5SA was unable to fully recapitulate the phenotype observed in L1;L2;AS mice. The difference in the genetic background between L1;L2;AS and Y5SA;AS mice and the fact that the overexpressed allele is the human form of YAP (which could have different affinities with YAP DNA-binding partners than the mouse allele) could both contribute to the differences between models. However, a more likely explanation is that the inactivation of Lats1/2 leads to an increase of the transcriptional activity of YAP and TAZ while YAP5SA only increases the transcriptional activity of YAP. Although these 2 Hippo effectors are redundant in several cell types, overlapping of YAP and TAZ binding sites have been shown to vary between 98% and 50% depending on the model (50-52), suggesting that YAP and TAZ might only have partial redundant functions in the adrenal cortex. In accordance with this hypothesis, concomitant inactivation of Yap and Taz in Lats1/2-null cells rescue the phenotype observed in L1;L2;AS mice.

One of the observations made while characterizing the L1;L2;AS mice (and to a lesser extent the Y5SA;AS) is that some zG cells appear to gain some characteristics of chondroblast/osteoblast and, to a lesser extent, Sertoli cells (among other cell types) following Lats1/2 loss. To the best of our knowledge, no study has ever demonstrated the transdifferentiation of zG cells into chondroblast/osteoblast-like cells. However, a population of multipotent mesenchymal stem cells able to transdifferentiate into osteoblast has been isolated from human adrenal cortex (53). One study has also suggested that adrenocortical cells can transdifferentiate into cells that share characteristics with Sertoli cells, as eliminating NR5A1 SUMOylation in mouse adrenocortical cells led to the appearance of SOX9+ cells and the expression of its Sertoli cell downstream targets in their adrenal cortex (54). However, the potential role of Hippo signaling was not evaluated in these models. Our GO analyses indicate that epithelial-to-mesenchymal transition and mesenchyme development were also among the biological processes controlled by upregulated genes. This suggests that inactivation of Lats1/2 in zG cells may lead them to acquire characteristics of pluripotent mesenchymal precursors, which could facilitate their transdifferentiation into other cell types. This is consistent with the observations in granulosa cells where inactivation of Lats1/2 resulted in epithelial-to-mesenchymal transition and subsequently transdifferentiation into Sertoli cells and osteoblasts (17). This is also consistent with several other studies that have demonstrated that inactivation of Lats1/2 in various cell types, including developing adrenocortical cells (21), causes them to maintain or acquire characteristics of mesenchymal cells (40, 41) and, in some cases, further transdifferentiate into myofibroblasts (21, 36, 42). Here, Last1/2 inactivation appears to initiate the ectopic transdifferentiation of zG cells into multiple lineages but does not fully commit to any of them. This is different to what is observed in the ovary, in which differentiated Sertoli cells and osteoblasts are observed (17), but reminiscent to what is observed in hepatocytes in which Last1/2 inactivation induces parts of the genetic programs of various cell types (18).

Although our evaluations focused on the effect of Lats1/2 inactivation on the transdifferentiation of zG cells, their inactivation can also affect the zF of juvenile L1;L2;AS mice. This effect, however, appears to be mostly indirect and caused by the ectopic transdifferentiation of zG cells as tracing experiments, and RNAscope analyses demonstrate that zF cells are not recombined. There are 3 potential explanations for this. First, it is possible that the capsular stem cell and subcapsular progenitor cell populations (in which recombination does not occur) directly transdifferentiate into zF cells. Second, the transdifferentiation of stem/progenitor cells into zG cells into zF cells (and centripetal migration) may be accelerated in mutant animals, leading to the transdifferentiation of some zG cells into zF cells before recombination occurs. Finally, it is also possible that other stem cell populations (55) or zF cell self-duplication could also be involved in zF maintenance as suggested by the fact that proliferating cells are seen throughout the cortex in L1;L2;AS mice. Interestingly, maintenance of the zF independently of the zG has also been observed in other mouse models using the AS^cre strain (3, 7, 56).

Even if at first glance, the zF of L1;L2;AS (and Y5SA;AS) mice appears mostly normal, DAB2+ cells (but not CTNNB1+ cells) were observed within the upper zF. It was previously demonstrated that the activation of Gq DREADD receptors in zG cells also leads to the expression of DAB2 in zF cells without altering the expression of CTNNB1 (10). As Gq pathway activation has been shown to increase YAP activity (57, 58) and DAB2 is a known downstream target of YAP/TAZ (37, 59), it would be interesting to determine if YAP act downstream of Gq signaling in this model to maintain elevated expression of DAB2 in zF cells. However, DAB2 expression was not lost in the Y;T;L1;L2;AS model, suggesting that YAP/TAZ are not essential to maintain DAB2 expression in the zG of mice.

Aside from the ectopic transdifferentiation of zG cells, GO analyses also demonstrated that the inactivation of Lats1/2 and expression of YAP5SA lead to an increase in the expression of numerous genes regulating the inflammation/immune response. Recent reports have demonstrated that Hippo signaling inactivation (Lats1/2 inactivation or YAP activation) induces an important inflammatory/immune response in several models (18, 60-63) although different outcomes were observed in these models. For example, activation of YAP induces the secretion of cytokines by hepatocytes and the subsequent recruitment of macrophages that permit transformed hepatocytes to evade immune clearance and form tumors (61), while Lats1/2-negative acinar cells produce cytokines that activate pancreatic stellate cells and recruit immune cells to the pancreas, leading to fibrosis (62). Interestingly, mesenchymal cells also exert immunomodulatory functions (64, 65), highlighting a potential correlation between the transdifferentiation of the zG cells and their ability to activate an immune response.

It has previously been demonstrated that YAP1 is overexpressed in Cushing's adrenal adenoma (66) and that YAP1 mRNA expression is a marker of poor prognosis in pediatric adrenocortical tumors (67) and adrenal cortex carcinoma (ACC) (68). Despite showing hyperplasia, our Y5SA;AS model does not develop tumors. Such discrepancies between the expression/mutation of genes involved in ACC formation and the phenotype observed following their overexpression/inactivation in transgenic mouse models was also observed for other genes such as TP53 and CTNNB1 (4, 56, 69-71). It is noteworthy that it was recently demonstrated that the inactivation of Znrf3, a gene mutated in approximatively 20% of ACC (71, 72), leads to adrenocortical hyperplasia in the adrenal cortex of young mice (6) that regresses in males but eventually progresses to tumor in older females (33). Hyperplasia regression in males was associated with an important androgen-dependent innate immune response, suggesting that this response inhibits tumor progression in these animals. As a robust immune response is observed in juvenile Y5SA;AS (and L1;L2;AS) mice, it is possible that tumor formation is blocked in our mouse models. Experiments are underway to determine if older Y5SA;AS males and/or females will develop tumors. However, we do not expect it will be the case as, contrary to the Znrf3 model in which recombination occurs in the whole adrenal cortex (including the subcapsular progenitor cells and zF cells), YAP overexpression is mostly limited to zG cells and recombination efficiency declined with age. Furthermore, recombination was also less efficient in females than in males, making it difficult to evaluate the effect of sexual dimorphism.

In summary, by generating distinct and complementary transgenic mouse models, we demonstrated that Hippo signaling, through the regulation of YAP and TAZ transcriptional activity, is essential for maintaining adrenal zG cell identity and their proper transdifferentiation into zF cells. Thus, we established that the Hippo signaling pathway is not only a critical mediator of adrenal cortex development but also essential for postnatal adrenal homeostasis.

Acknowledgments

The authors would like to thank Dr. Celso Gomez-Sanchez (University of Mississippi, Medical Center, Jackson, MS) and Dr. Randy L. Johnson (M.D. Anderson Cancer Center, Houston, TX) for generously providing the CYP11B1 antibody and Lats1/2 floxed mice, respectively.

Funding

This work was supported by Discovery Grants to A.B. (RGPIN-2020-05230) and to G.Z. (RGPIN-2018-06470) from the Natural Sciences and Engineering Research Council of Canada and by a research project grant to D.T.B. from the National Institutes of Health (R01DK123694). The University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core is supported by the National Institutes of Health/Eunice Kennedy Shriver National Institute of Child Health and Human Development (Grant P50-HD28934). N.A.N. is a recipient of a PhD scholarship from the Fonds de Recherche du Québec—Nature et technologies.

Author Contributions

A.B. and N.A.N. conceived the study. N.A.N. performed most experiments and analyses and interpreted the data. N.J. performed some of the tracing experiments. M.C.M. performed the RNAscope experiment. L.B. and L.C. performed immunohistochemistry for platelet-endothelial cell adhesion molecule 1 and vimentin. J.B. trained N.A.N. in computational analysis of transcriptomic data. M.P. and G.S.J. trained N.A.N. in the use of QuPath. J.M. generated the Y5SA mouse model. D.T.B. generated the AS mouse model. D.B., G.Z., J.B.D.A., and A.B. supervised N.A.N., N.J., L.B., and L.C. D.P. supervised M.C.M. N.A.N. and A.B. wrote the first draft of the manuscript. A.B. wrote the final draft. D.B., D.T.B., and G.Z. revised the manuscript.

Disclosures

The authors have nothing to disclose.

Data Availability

All relevant data can be found within the article, cited references, and its supplementary information (29). RNA-seq data are deposited in the GEO database, accession number GSE227353.

References