Transcription factor Creb3l1 regulates the synthesis of prohormone convertase enzyme PC1/3 in endocrine cells

Abstract Transcription factor cAMP responsive element‐binding protein 3 like 1 (Creb3l1) is a non‐classical endoplasmic reticulum stress molecule that is emerging as an important component for cellular homeostasis, particularly within cell types with high peptide secretory capabilities. We have previously shown that Creb3l1 serves an important role in body fluid homeostasis through its transcriptional control of the gene coding for antidiuretic hormone arginine vasopressin in the neuropeptide‐rich magnocellular neurones of the supraoptic nucleus. In response to osmotic stimuli such as dehydration, vasopressin magnocellular neurones undergo remarkable transcriptome changes, including increased Creb3l1 expression, to ensure that the supply of vasopressin meets demand. To determine where else Creb3l1 fits into the secretory cell supply chain, we performed RNA‐sequencing of Creb3l1 knockdown anterior pituitary mouse corticotroph cell line AtT20. The target chosen for further investigation was Pcsk1, which encodes proprotein convertase enzyme 1 (PC1/3). PC1/3 is crucial for processing of neuropeptides and peptide hormones such as pro‐opiomelanocortin (POMC), proinsulin, proglucagon, vasopressin and oxytocin. Viral manipulations in supraoptic nuclei by over‐expression of Creb3l1 increased Pcsk1, whereas Creb3l1 knockdown decreased Pcsk1 expression. In vitro promoter activity and binding studies showed that Creb3l1 was a transcription factor of the Pcsk1 gene binding directly to a G‐box motif in the promoter. In the dehydrated rat anterior pituitary, Creb3l1 and Pcsk1 expression decreased in parallel compared to control, supporting our findings from manipulations in AtT20 cells and the supraoptic nucleus. No relationship was observed between Creb3l1 and Pcsk1 expression in the neurointermediate lobe of the pituitary, indicating a different mechanism of PC1/3 synthesis by these POMC‐synthesising cells. Therefore, Creb3l1, by regulating the expression of Pcsk1, does not control the processing of POMC peptides in the intermediate lobe.


| INTRODUC TI ON
cAMP responsive element-binding protein 3 like 1 (Creb3l1), also known as OASIS, is a transcription factor in the CREB/ATF family.
Because of the structural similarity to the endoplasmic reticulum (ER) stress inducer ATF6, early studies on Creb3l1 focused on its role in ER stress pathways. 1 We now know that the actions of Creb3l1 are much more wide-ranging than simply ER stress. Indeed, recent reports suggest that Creb3l1 is involved in cellular processes such as secretion, hormone synthesis, the formation of the extracellular matrix and cellular proliferation. [2][3][4][5] This is backed up by studies showing that Creb3l1 expression can be regulated by transforming growth factor beta, glucocorticoids and progesterone. [6][7][8] These studies suggest that ER stress is one of many mechanisms through which Creb3l1 protein can be activated.
To act as a transcription factor, Creb3l1 is cleaved in the Golgi to liberate the transcriptionally active N-terminal fragment, which then enters the nucleus to activate the transcription of target genes. 9 Creb3l1 is expressed in a range of tissues, mostly secretory organs/ cells, such as the pancreas, placenta, prostate gland, thyroid gland, gastrointestinal tract, osteocytes and neuroendocrine cells of the hypothalamus. 10 We have previously reported increased Creb3l1 expression in the supraoptic nucleus (SON) of the rat and mouse is response to hyperosmotic stress, 11 and identified Creb3l1 as a transcription factor for vasopressin (AVP) gene expression. 12 The expression profile of Creb3l1, together with recent studies in endocrine/secretory tissue, 3,13 suggests that Creb3l1 may also regulate expression of genes involved in hormone secretion.
To identify transcriptional targets of Creb3l1, we performed high throughput transcriptomic RNA sequencing to catalogue genes altered in expression in AtT20 cells by stable Creb3l1 knockdown.
AtT20 is a well-characterised secretory cell line, derived from mouse anterior pituitary corticotroph cells. These cells express high levels of the pro-opiomelanocortin (POMC) hormone precursor, which is processed by proteolytic cleavage into several mature biologically active peptides that are subsequently secreted. 14 To direct our candidate search towards secretory cells, we compared the AtT20 cell gene list with previously published transcriptomic data from the dehydrated rat and mouse SON. 11 Creb3l1 is increased in the SON by dehydration. 8,12,15,16 Thus, we considered genes that increased in the SON in response to dehydration and decreased in AtT20 cells following stable knockdown of Creb3l1. We theorised that this cross-analysis would help to identify transcriptional targets of Creb3l1 in the hormone synthesis and secretory pathway.
From our transcriptomic comparisons, the Pcsk1 gene, which encodes the proprotein convertase enzyme 1 (PC1/3), was chosen for further investigation. PC1/3 is predominantly expressed in neural and endocrine tissues [17][18][19][20][21] and is crucial for processing of neuropeptides and peptide hormones such as POMC, proinsulin, proglucagon, AVP and oxytocin. 22 However, to date, the knowledge on transcriptional regulation of Pcsk1 gene in these systems remains little understood.

| Animals
Male Sprague-Dawley rats weighing 200-300 g were used in the present study. Rats were housed under a 14:10 hour light/dark cycle (lights on 5.00 am) at a constant temperature of 22°C and a relative humidity of 50%-60%. Rats had free access to food and tap water for at least 1 week prior to experimentation. Animal experiments were performed between 9.00 am and 2.00 pm Experiments were performed under a Home Office UK license held under, and in strict accordance with, the provision of the UK Animals (Scientific Procedures) Act (1986); they had also been approved by the University of Bristol Animal Welfare and Ethical Review Board.

| Hyperosmotic experiments
To induce acute hyperosmotic stress, a single intraperitoneal injection (i.p) of 1.5 mL 100 g -1 body weight of 1.5 mol L -1 NaCl solution was performed. Rats were randomly allocated into one of six groups: control (0), 10 minutes, 30 minutes, 1, 2 and 4 hours after administration of hypertonic saline. After injection, rats were placed back in their home cages, and water, but not food, was removed for the duration of the experiment. The control group had access to food and water ad lib. throughout the experimental period. For chronic hyperosmotic stimulation, drinking water was removed for 3 days (dehydration) or replaced with 2% (w/v) NaCl solution for 7 days (salt loading). For RNA and protein samples, rats were killed by striking of the cranium.
Tissues were removed and immediately frozen using powdered dry ice and stored at −80°C until used. The pituitary gland was collected whole or separated into anterior and neurointermediate lobe (NIL) using sterile scalpel blades. For immunofluorescence staining, rats were anaesthetised using pentobarbital and perfused transcardially with phosphate-buffered saline (PBS) followed by 4% (w/v) paraformaldehyde/PBS. Brains and pituitaries were removed and post fixed in 4% (w/v) paraformaldehyde/PBS overnight at 4°C, then cryoprotected in 30% (w/v) sucrose/PBS for approximately 72 hours before being frozen over liquid nitrogen and stored at −80°C.

| Introduction of viral vectors into the SON
The Creb3l1 and non-targeting short hairpin RNAs (shRNAs) (see Supporting information, Table S1) were cloned into pGFP-A-shAAV K E Y W O R D S dehydration, pituitary, POMC, prohormone processing, supraoptic nucleus, transcription (OriGene, Rockland, MD, USA). Adeno-associated viral particles (AAV1/2) were produced using a helper free packaging system (Cell Biolabs, San Diego, CA, USA) to a titer of 6 × 10 12 genome copies mL -1 as described previously. 23 The production of the constitutively active (CA) Creb3l1 and green fluorescent protein (GFP) lentiviral vectors has been described previously. 12 For SON injections, rats were anaesthetised by i.p administration of a medetomidine and ketamine mix and placed in a stereotaxic frame in the flat skull position. A 2-cm rostral-caudal incision was made to expose the surface of the skull. Two 1-mm holes were drilled at co-ordinates 1.3 mm posterior to bregma and 1.8 mm lateral to midline. A 5-µL pulled glass pipette was positioned −8.8 mm ventral to the surface of the brain and 1 µL of virus was delivered into nuclei over 10 minutes. The glass pipette was fixed in position for a further 5 minutes to minimise back tracking of the virus. Following injections, the incision was closed and atipamezole was administered intramuscularly.
After surgery, animals were individually housed in standard laboratory cages for 2-3 weeks.
Creb3l1 knockdown cell lines were established using lentivirus expressing shRNAs as described previously. 8 The sequences of all oligonucleotides and primers used in this study are provided in the Supporting information (Table S1). ). Paired-end library reads of greater than 30-40 million were generated for each individual library. The data were then processed using rta and casava (Illumina Inc.), thus providing sets of compressed FASTQ files per library. All raw reads were pre-processed for quality assessment, adaptor removal, quality trimming and size selection using the fastqc toolkit 24 to generate quality plots for all read libraries. We adopted a phred30 quality cut-off (99.9% base call accuracy).

| RNA sequencing
RNA sequencing alignment and data analysis were all performed in house using our high-performance computer; "Hydra".
Our pipeline makes use of bash and python scripting to accept RNA sequencing post-trimmed data as input, before ultimately producing output tables of differentially expressed transcripts. Pairedend (2 × 75-bp) raw input data is initially aligned with star 25 to the thirty-eighth iteration of the Mus musculus reference genome (GRCm38.p6). featurecounts 25 is used to generate read counts, using the ENSEMBL Mus.musculus.GRCm38.97 annotation for reference. 26 Our pipeline then uses deseq2 (version 1.22.2) 27 from the r bioconductor package (https://www.bioco nduct or.org) to call differential gene expression. All P values were adjusted for multiple testing using the procedure of Benjamini and Hochberg. The data has been deposited in NCBI's Gene Expression Omnibus and is accessible through GEO Series accession number GSE147978 (https://www.ncbi.nlm.nih.gov/geo/query /acc.cgi?acc=GSE14 7978).

| Quantitative polymerase chain reaction (PCR)
SON samples were collected and total RNA extracted as described previously. 12 For pituitary samples, 200 µL of TRI reagent (Sigma) was added to frozen pituitaries and tissue was homogenised in 1.5-mL Biomasher tubes (Takara, Kusatsu, Japan). For cells, 400 000 cells were seeded into 12-well plates. At the time of collection, culture media was removed and 350 µL of TRI reagent was added to the well and incubated at room temperature for 5 min-

| Western blotting
Cells were seeded into six-well plates (800 000 cells well -1 ). After 24 hours, cells were washed with PBS and harvested by scraping in 500 μL of radioimmunoprecipitation assay (RIPA) buffer supplemented with protease inhibitor cocktail (P8340; Sigma). Lysate was incubated on ice for 15 minutes with vortexing every 5 minutes.
Debris was removed by centrifugation at 10 000 g for 10 minutes.  ab171870; Abcam). The purified DNA was used for a quantitative PCR using mouse Pcsk1 ChIP primers.

| Statistical analysis
Replicates in all experiments are biological replicates. Statistical differences between two experimental groups were evaluated using independent-sample unpaired Student's t tests. One-way ANOVA with Tukey's or Dunnett's post-hoc tests was used to determine the difference between more than two samples with only a single influencing factor. Two-way ANOVA with a Bonferroni post-hoc test was used to determine interactions between two independent variables on the dependent variable. Data are presented as the mean ± SEM P < 0.05 was considered statistically significant.
F I G U R E 1 RNA sequencing of Creb3l1 knockdown AtT20 cells. RNA sequencing was performed using RNA extracted from Creb3l1 knockdown AtT20 cells induced by stable expression of Creb3l1-shRNA1 (n = 5). A non-targeting short hairpin RNA (shRNA) was used as a control (n = 5). The data were compared with previously published gene lists from microarrays of the rat and mouse supraoptic nucleus (SON) from control and dehydration (1.5-fold cut off for the microarray data). 11 A, Venn diagram showing the number of overlapping genes that changed in these data sets. B, List of genes that changes in all data sets. Genes in red change in the direction that implies Creb3l1 is a positive regulator (down in Creb3l1-knockdown). C, Quantitative polymerase chain reaction (PCR) validation of RNA sequencing data for Creb3l1, Scg2, Rasd1, Nr4a1 and Pcsk1. D, Quantitative PCR and western blotting analysis of Creb3l1 and Pcsk1 in Creb3l1-knockdown AtT20 cells using two different Creb3l1-shRNAs. NT, non-targeting control; FL, full length; PC1/3, proprotein convertase enzyme 1. *P < 0.05; **P < 0.01; ***P < 0.001

| Creb3l1 knockdown in AtT20 cells
To identify new transcriptional targets of Creb3l1, we performed RNA sequencing on our previously reported Creb3l1 knockdown AtT20 stable cell line. 8 The result showed that 5706 genes were significantly changed (P adjusted < 0.01) in expression by Creb3l1 knockdown (see Supporting information, Table S2). To select targets for further investigation, we compared these data with previously published microarray datasets from dehydrated mouse and rat SON where Creb3l1 increases in expression. 11 Comparisons of mouse and rat microarray datasets revealed 67 genes that were changed (increased or decreased; >1.5 fold) in the SON of both rodents by dehydration ( Figure 1A). Of these, 24 genes had differing expression in our RNA sequencing data from AtT20 cells ( Figure 1B). Of these, 15 decreased in expression in Creb3l1 knockdown AtT20 cells ( Figure 1B, red) and 14 increased in the rodent SON following dehydration, suggesting that Creb3l1 could be a transcriptional regulator for these genes.
To validate these data from Creb3l1 knockdown cells

| Effect of Creb3l1 on Pcsk1 expression in the SON
We have previously reported increased Creb3l1 expression in the rat SON following 3 days of dehydration, 7 days of salt loading and during an acute time course following hypertonic saline injection. 8 Here, we show that Pcsk1 abundance also increases (chronic F 2,15 = 85.63, P < 0.0001; acute F 5,24 = 9.18, P < 0.0001) in the SON in the same experimental models (Figure 2A). We also looked at the closely related proprotein convertase Pcsk2 ( Figure 2B), which is expressed in the SON. 18 The abundance of Pcsk2 mRNA increased ( To investigate the relationship between Creb3l1 and Pcsk1 expression in vivo, we used viral-mediated gene transfer to either overexpress or to knockdown Creb3l1 in the SON of control rats.

Validation of Creb3l1 overexpression by quantitative PCR has been
confirmed alongside concomitantly increased AVP mRNA expression in these samples. 12 Here, we show that overexpression of  Figure 2D).

| Creb3l1 positively regulates transcription of Pcsk1 by binding to a G-box on its promoter
The relationship between Creb3l1 and Pcsk1 expression in different models suggested that Creb3l1 may directly regulate transcription of the Pcsk1 gene. We have previously shown that Creb3l1 mRNA and protein expression increases in response to increasing intracellular cAMP levels in AtT20 cells 8 and, in the present study, we show in- the Pcsk1 promoter had no effect on Creb3l1-mediated luciferase activity, suggesting that a binding site might be located close to the transcriptional start site. Alignment of nucleotides −81 to +205 bp from human, mouse and rat Pcsk1 genes showed that the proximal promoter was highly conserved ( Figure 3E). The direct interactions of Creb3l1 with the rat AVP promoter are mediated by interactions with G-box sequence (GCCCACGTGTGT). 12 Here, we have identified a similar core G-box motif GACGTG within the Pcsk1 promoter. This sequence has recently been validated as a CrebA (Creb3l1-orthologue)-binding site in Drosophila. 13 We subsequently made luciferase reporter constructs containing the

| Knockdown of Creb3l1 decreases POMC expression in AtT20 cells
To investigate downstream effects of a Creb3l1 regulated Pcsk1 pathway, we first looked at POMC expression in our Creb3l1 knockdown cell lines. AtT20 cells, being derived from anterior pituitary corticotroph cells, endogenously express and process POMC. 33 There was no change in Pomc mRNA abundance following Creb3l1 knockdown ( Figure 4A), although western analysis showed that the 31-kDa POMC prohormone was reduced by Creb3l1 knockdown  Figure 4B).

| Expression of Creb3l1, PC1/3 and POMC in the pituitary in response to dehydration
Stress-induced plasma ACTH secretion is attenuated in dehydrated rats, which is not the result of altered Pomc mRNA expression suggesting altered processing of POMC. 34

| D ISCUSS I ON
The present study reveals Creb3l1 to be a transcription factor for the We performed RNA sequencing of Creb3l1 knockdown AtT20 cells to identify genes regulated by this transcription factor. With such a large number of gene changes, we needed a method to identify promising target genes for further investigation. We reasoned that because Creb3l1 is increased in the SON by dehydration, target genes would also be increased in this model, and so we made comparisons with transcriptome catalogues from the dehydrated rat and mouse SON.
A number of common genes were identified and validated, including Nr4a1, which was previously identified by us as a transcription factor regulating Creb3l1 expression, 16 Scg2, a potential sorting receptor that targets proteins to secretory granules, and Rasd1, a small G protein that we recently identified in AVP neurones of the hypothalamus. 39 These genes are highly expressed in neuroendocrine cells of the hypothalamus, including the SON and paraventricular nucleus (PVN), although they were not subjected to further investigation at this time.
Rather, in the present study, we focused on the Pcsk1 gene that encodes the hormone-processing enzyme PC1/3.
We have previously reported up-regulation of the Creb3l1 gene in the hypothalamus following dehydration, salt loading and hyperosmotic stress 12 and, in the present study, we demonstrate increased Pcsk1 expression under these same conditions. During dehydration, AVP biosynthesis increases, 40,41 and an additional demand for this peptide necessitates changes in cell components necessary for processing and secretion. 42 Because the AVP prohormone can be processed by PC1/3 in vitro 31 and in vivo, 43  There is a considerable literature that describes the role of PC1/3 in prohormone processing and its tissue distribution. However, the transcriptional regulation of the Psck1 gene is less well understood.
Creb3l1 and PC1/3 expression are both upregulated by increased cellular cAMP levels. A cAMP-responsive element has been identified on the Pcsk1 promoter and CREB and activating transcription factor 1 have been proposed to activate Pcsk1 promoter activity. 44,45 Here, we identify Creb3l1 as a transcriptional regulator of the Pcsk1 gene binding at a consensus Creb3l1-binding site (G-box) in AtT20 cells. The Creb3l1 binding site (G-box) was first described by computational analysis of human transcription factor binding, 46 and we showed that Creb3l1 binds to a similar G-box sequence within the AVP promoter to regulate its transcription. 12 The identification of this core consensus G-box sequence for Creb3l1 may lead the way for finding further gene targets for Creb3l1 in secretory cells.  be compensated for PC2. 58 In PC1/3 knockout mice, α-MSH production is unchanged consistent with PC2 compensation but can also be decreased consistent with impaired POMC processing in the pituitary. Thus, our data imply that α-MSH formation should decrease during dehydration as a result of a decline in the synthesis of both PC enzymes and POMC itself. A decrease in Pomc expression has previously been described in the rat NIL following salt loading, 63 supporting our findings of the present study with respect to dehydration. Any physiological significance of these modifications to prohormone processing machinery in the intermediate lobe in a state of dehydration remains to be determined.
In the SON, Creb3l1 increases Pcsk1 expression to cope with the increased biosynthesis of AVP being released from the posterior pituitary. We report a complete contrast of events regarding the PC systems in the intermediate lobe and anterior portions of the pituitary where processing of prohormones would appear to be blunted in dehydration. Thus, to maintain hydromineral balance in chronic dehydration, the activity of the hypothalamo-neurohypophyseal system increases whereas the HPA axis becomes less responsive to stress. We propose that changes to the HPA axis act to enhance the survival capabilities of dehydrated animals by reducing their response to stressors and promoting social behaviours. It has been proposed that such a mechanism may have evolved to suppress fear and anxiety in animals approaching a communal water source where predators may be encountered to enable drinking of fluids to restore body water content. 64 Therefore, both these mechanism act to restore the physiological needs of the animal.
To summarise, we have identified Creb3l1 as a transcription factor of the Pcsk1 gene in both the hypothalamus and corticotroph cells, providing new understanding about how Pcsk1 gene expression is controlled. This information may be useful for future applications in diseases related to altered Pcsk1 expression. It is important to note that Pcsk1 expression does not appear to be regulated by Creb3l1 in pituitary melanotrophs. Therefore, this newly identified transcriptional pathway may not be a universal mechanism regulating Pcsk1 expression, and as such, should be assessed on a cell-type-specific basis.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.