Mice used in this study were generated from heterozygote breeding of Manf ^+/− and Cdnf ^+/− mice (16, 18) (Lindahl et al., submitted manuscript). Tissues from Manf^−/− and Cdnf^−/− animals were used as a negative control in immunohistochemistry and ELISA. All mice were maintained on ICR background and housed on a 12-h dark/light cycle, ad libitum fed and with free access to water. Finnish Animal Ethics Committee of the State Provincial Office of Southern Finland approved all experimental procedures involving mice.

MANF (Manf_D06, C57BL/6N-Manf^{tm1a(KOMP)Wtsi}) (and CDNF (Cdnf_CO1, C57BL/6N-Cdnf^{tm1a(KOMP)Wtsi}) targeted embryonic stem cell clones were generated by the trans-National Institutes of Health (NIH) Knockout Mouse Project and obtained from the KOMP Repository (http://www.komp.org).

Islet isolation was performed as described previously (16). In brief, islets from 8 weeks old mice were isolated by standard collagenase (Collagenase P; Roche Diagnostic) digestion followed by handpicking under a stereomicroscope.

Mouse tissues were collected and fast frozen in liquid nitrogen. RNA isolation, reverse transcription, and RT-qPCR were performed as described previously (16). Total RNA was isolated from mouse tissues or isolated islets by homogenizing them in TriReagent® (Molecular Research Center) according to standard protocol procedures. The RNA amount and purity was measured by NanoDrop ND-1000 spectrophotometer (Thermo Scientific). Glycogen (RNA grade, Thermo Scientific) was used during the RNA precipitation procedure to visualize the pellet. cDNA synthesis was done by reverse transcription reactions with RevertAid™ Premium Reverse Transcriptase (Fermentas UAB, Thermo Fisher Scientific Inc.) according to the manufacturer's protocol. RT-qPCR was performed using LightCycler® 480 SYBR Green I Master (Roche Diagnostics GmbH) and Roche LightCycler® 480 Real-Time PCR System. The expression levels were normalized to the levels of β-actin in the same samples. Primer sequences for Gh, Prl, Pit1, Tshβ, mPomc, Fsh, Lhb, Grp78, Chop, sXbp1, tXbp1, Atf4, Atf6α, Atf6β, Manf genes used in this study were from (5, 16, 41). Primer sequences for Cdnf gene were F 5′-AGC TGC TCA ACT TTT GCT CA-3′, R 5′-TAG GAT CTT GGT GGC TGC AT-3′.

Mouse tissue samples were homogenized in ice-cold lysis buffer (100 mM Tris-HCl, pH 7.5, 300 mM NaCl, 4 mM EDTA, 0.2% Triton X-100, 2 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM sodium orthovanadate, protease inhibitor cocktail (Complete, Mini EDTA free, Boehringer Mannheim) and incubated on ice for 30 min. Thereafter the homogenates were centrifuged at 13,000 rpm for 30 min at +4°C and supernatants were collected and stored at −80°C for MANF and CDNF ELISAs.

The concentrations of endogenous MANF in mouse tissue lysates were analyzed by in-lab developed sandwich ELISA. The development and Validation of the mouse (m)MANF ELISA was described in detail elsewhere (42). The sensitivity of mMANF ELISA is 29 pg/ml, and the dynamic range is from 62.5 to 1 000 pg/ml. The ELISA gave no signal from MANF knockout mouse samples, indicating that it was specific for MANF. Briefly, a 96-well plate (MaxiSorp, Nunc) was coated with goat antibodies to MANF (AF3748, R&D⁺ systems) in carbonate coating buffer (pH 9.6) for overnight at 4°C. Next day, after washing with PBS-0.05% Tween (PBST), the wells were blocked with 1% casein in PBST to prevent unspecific binding. Mouse tissue samples and recombinant MANF standard were diluted in blocking buffer and incubated on the plate for overnight at +4°C. On the third day, the wells were washed with PBST, and rabbit anti-MANF antibody (LS-B2688, LSBio) was diluted in blocking buffer and incubated in the wells for 3 h at 37°C. For the detection of bound MANF, a secondary antibody conjugated to horseradish peroxidase (HRP; NA9340V, GE Healthcare) was applied to the plate and incubated for 2 h at room temperature. HRP signal was detected with 3,3′,5,5′-tetramethylbenzidine substrate (DuoSet ELISA Development System, R&D Systems) and absorbance measurement using Victor³ reader (Perkin Elmer) at 450 nm and 540 nm (for wavelength correction).

The levels of endogenous mouse CDNF in mouse tissue lysates were analyzed by in-lab developed and validated sandwich ELISA. MaxiSorp (Nunc, Fisher Scientific) 96-well microtiter plates were coated overnight at +4°C with 1 μg/ml goat anti-mCDNF (AF5187, R&D Systems) in 0.05 M carbonate/bicarbonate coating buffer (pH 9.6). The next day, unspecific binding to plates was blocked with 3% BSA in PBS for 2 h at room temperature. After blocking, the study and standard (recombinant mCDNF protein, 5187-CD, R&D Systems) samples were diluted in the blocking buffer and applied on the plate as duplicate measurements. The samples were incubated on the plate overnight at +4°C in 100 rpm agitation.

Following day, a detection antibody rabbit anti-CDNF (6) was applied at 0.2 μg/ml in blocking buffer, and the plate was incubated for 3 h at +37°C in 100 rpm agitation. Finally, horse radish peroxidase (HRP)-linked donkey anti-rabbit (GE Healthcare) secondary antibody was diluted at 1:2,000 in blocking buffer and incubated in the plate for 2 h at room temperature in 100 rpm agitation. Before each step, liquids from the previous step were removed, and the wells were washed with PBS-Tween 20 (0.05%). Blocking and washing were performed in a volume of 200 μl, while antibodies and samples were added at 100 μl volume. The bound HRP-linked antibody was visualized by the addition of 3,3′,5,5′-tetramethylbenzidine (DuoSet ELISA Development System, R&D Systems), and absorbance was read at 450 and 540 nm (Victor³ reader, Perkin Elmer). MANF and CDNF concentrations in tissue lysates were normalized to the total amount of sample protein, which was analyzed by a DC Protein Assay (Bio-Rad).

Mouse CDNF ELISA was validated for specificity, sensitivity, dynamic range and precision. Specificity of the assays was determined by the response to Cdnf ^−/− tissue lysates. Sensitivity was determined by the mean absorbance value of ten blank samples added by 3 standard deviations and calculating resulting concentration from the standard curve. The dynamic range of the assay was determined by the accuracy of the back-calculated concentration values for each standard curve point from six individual assays. Within the assay dynamic range, the individual back-calculated accuracy values were within ±10% relative error (% RE = derived concentration/expected concentration × 100%) and precision max 10% coefficient of variation (% CV = SD/mean × 100%). Within-run precision (i.e., intra-assay variation) was determined by running three samples of varying concentration in replicates of ten on different parts of the same microtiter plate. Between-run precision (i.e., inter-assay variation) of the assays was determined by running three samples in duplicate on six independent assay runs performed on different days. Dilutional linearity and recovery of spiked analyte were analyzed in the study samples.

The sensitivity of the mouse CDNF ELISA was 6 pg/ml and the dynamic range from 15.6 to 500 pg/ml (Supplementary Table 1). Intra- and interassay precision was 10.3% CV (9.0–12.1% CV) and 6.8% CV (4.2–8.7% CV), respectively (Supplementary Table 2). The assay gave values below the ELISA sensitivity level for Cdnf^−/− tissue samples (Supplementary Table 3). The linearity of dilution was within 86–106% RE (n = 5, three sequential dilutions per sample, Supplementary Table 4) and recovery of 100 and 250 pg/ml spikes of recombinant mouse CDNF was within 85–135% RE (n = 3 per spike, Supplementary Table 5).

Embryonic stem cells were cultured on 0.1% gelatin-coated dishes in ES cell medium (Knockout DMEM™ high glucose, Gibco, 10829–018, 15% FBS, 2 mM GlutaMax™-I (Gibco), 1 mM Non-Essential Amino Acids, 1,000 U/ml LIF and 1 μM β-mercaptoethanol. Cells were fixed in 2% PFA for 1 h, washed several times with PBS and stained with LacZ staining as described below. The induction of the Manf gene by reporter gene β-galactosidase expression was analyzed by LacZ staining of embryonic stem cells and whole mount staining of embryonic day (E)7.5 decidua and E9.5 embryos. Immunohistochemistry (IHC) was applied for sectioned E9.5 and E13.5 embryos, and adult mouse tissues.

LacZ staining on whole mount embryos was performed on deciduae containing E7.5 embryos using X-gal solution. First, the removed deciduae were fixed in 2% PFA for 1 h on ice and washed several times in PBS at RT for 2 h. Thereafter deciduae were incubated in staining solution [PBS supplemented with 2 mM MgCl₂, 5 mM potassium ferrocyanide (K₄Fe(CN)6-3H₂0], 5 mM potassium ferricyanide [K3Fe(CN)6, 0.01% NP-40, 0.01% Sodium deoxycholate] containing X-Gal reagent (5-Bromo-4-chloro-3-indolyl beta-D-galactopyranoside) at final concentration of 1 mg/ml for several hours at 37°C in dark. After staining, deciduae were dehydrated and embedded in paraffin followed by sectioning of the samples and counterstaining with nuclear fast red.

For IHC the animals were euthanized by carbon dioxide followed by cervical dislocation. Mouse tissues or embryos were collected and fixed in 4% paraformaldehyde (PFA) for 72 h following embedding to paraffin. Deparaffinised 5 μm thick sections were boiled in citrate buffer pH 6.0 for retrieval of the antigen, rinsed in Tris-buffered saline (TBS) (pH 7.4) and kept in 0.3% H₂O₂ solution for 30 min to suppress endogenous peroxidase activity. Next, the sections were washed in TBS with 0.1% Tween (TBST) and blocked with 5% goat or 5% donkey serum for 1–2 h at room temperature. Samples were incubated with primary antibodies (Table 1) in the appropriate blocking buffer overnight at 4°C. For light microscopy peroxidase staining, sections were incubated with biotinylated secondary antibodies for 1 h at room temperature, washed in TBST and incubated with avidin-biotinylated enzyme complex (ABC) according to manufacturer's instructions. Finally, after the washing procedure in TBST, samples were developed with 3, 3'-Diaminobenzidine (DAB), counterstained with hematoxylin and mounted with Depex. For immunofluorescence staining, sections were subjected to antigen retrieval and incubated with primary antibody o/n, followed by incubation with appropriate Alexa-conjugated secondary antibodies, for 4 h at room temperature, washed in TBST and mounted with mounting medium containing DAPI (Vectashield, Vector Laboratories, Burlingame, CA, USA) to visualize nuclear staining.

Importantly, anti-MANF antibodies have been validated by using Manf^−/− tissues. No background was detected in the tissues of Manf^−/− mice (Supplementary Figures 1H–J, 3A–M).

All data are presented as mean ± SEM. Statistical significance was assessed via Student's t-test between 2 groups as mentioned in the figure texts using GraphPad Prism 6 software (GraphPad Software, Inc., La Jolla, CA).

Manf mRNA expression has previously been analyzed by in situ hybridizations in E12.5 mouse embryos (39). Here we analyzed Manf gene expression in early embryonic development using staining for LacZ enzymatic activity of Manf^+/− embryos containing a β-galactosidase reporter gene (lacZ), which is controlled by the endogenous MANF promoter (16). We found LacZ staining indicating MANF expression in targeted Manf^+/− embryonic stem cells (ESC) (Supplementary Figures 1A,B). During early embryogenesis, LacZ staining was detected in E7.5 Manf^+/− embryos. LacZ negative decidual cells embedding the embryo are originating from the Manf^+/+ mother (Figures 1A–D). Positive LacZ staining was detected in most cells of all three germ layers—ectoderm, endoderm and mesoderm in the E7.5 embryo (Figures 1A–D). Notably, no LacZ staining was detected in extraembryonic ectoderm, which contributes to the formation of the placenta (Figures 1A–D).

In order to analyze MANF expression pattern in mouse tissues, we used highly specific MANF antibodies, which we validated on Manf knockout tissue as a control for a negative signal. No background was detected in the tissues from Manf^−/− mice (Figures 1F,H,J, Supplementary Figures 1C–H, 3A–M).

Next, we studied the expression of MANF in E9.5 Manf^+/+ and Manf^−/− embryos by IHC. Within the neural tube, MANF expression was observed in the regions forming the forebrain, midbrain, and hindbrain (Figure 1E). Outside the nervous system, MANF immunoreactivity cells were detected in somites giving rise to the dermis, skeletal muscles, and vertebrae (Figures 1E,G). MANF expression was also found in the developing heart, mandibular component of the first branchial arch and in the origin of second and third branchial arches in the E9.5 embryo (Figure 1E). The same positive signal was observed by LacZ in E9.5 heterozygous Manf^+/− mouse embryo (Supplementary Figures 1C,D). No staining was detected in Manf^−/− embryos (Figure 1F).

MANF was widely expressed in Manf^+/+ embryos at E13.5 when many of the CNS and peripheral nervous system (PNS) structures, as well as other peripheral tissues, begin to differentiate (Figure 1I, Supplementary Figures 1E–G). In agreement with our previous in situ hybridization results, within the CNS, MANF immunostaining was observed in the neocortex, striatum, midbrain, hypothalamus and spinal cord (Figures 1I,K–P). High levels of MANF immunoreactivity was detected in the choroid plexus, the lumen of Rathke's pouch (Figure 1Q), an ectodermal structure forming the anterior pituitary gland, and vestibulocochlear ganglion (Figure 1R). In the PNS, MANF was highly expressed by neurons in trigeminal (Figure 1S) and dorsal root ganglia (Figure 1T). MANF immunoreactive cells were also observed in the developing peripheral tissues such as heart, lung, liver, intestine and developing pancreas (Figures 1U–Y) (5).

Our previous studies have demonstrated the wide distribution of Manf mRNA expression in the adult mouse brain by in situ hybridization, as well as protein expression in the adult cortex, hippocampus, cerebellum, striatum, and substantia nigra and non-neuronal tissues (39).

In this study, we analyzed and compared the levels of MANF and CDNF mRNA and protein expression in a broad panel of different tissues from adult mice using RT-qPCR and ELISA. Highest levels of Manf mRNA were observed in pituitary glands, endocrine islets of Langerhans, in the pancreas exocrine tissue separated from islets using collagenase digestion, and testis (Figure 2A) (5). Moderate levels of Manf mRNA expression was detected in ovary, heart, liver, and parts of intestine including duodenum and jejunum compared with low levels in the adrenal gland, brain, thymus, lung, salivary gland, kidney, spleen, ileum, colon, and muscle (Figure 2A).

Next, we analyzed the MANF protein expression by mouse MANF ELISA developed in our laboratory (42). In accordance with Manf mRNA levels, the highest concentrations of MANF protein was observed in pancreas and testis (Figure 2B). The MANF concentration in the pancreas was 894 ± 154 ng/mg total protein (n = 3), and in the testis 1,170 ± 168 ng/mg total protein (n = 4), respectively, indicating that MANF corresponds to ~1 promille of all proteins in these tissues. MANF concentration in the pituitary gland, liver, and salivary gland was between 300 and 400 ng/mg total protein, which is ~3 times lower compared to the MANF levels in testis and pancreas.

In the brain, brown adipose tissue, spleen, kidney, ovary, and thymus MANF protein concentration varied between 100 and 250 ng/mg total protein. The lowest MANF protein concentrations were found in skeletal muscle (82 ± 6 ng/mg total protein, n = 4), white adipose tissue (67 ± 4 ng/mg total protein, n = 3) and heart (38 ± 2 ng/mg total protein, n = 4). Protein concentration was almost 3-fold in brown adipose tissue (199 ± 14 ng/mg total protein, n = 4) compared to white adipose tissue.

Highest levels of Cdnf mRNA were found in the muscle, heart and exocrine tissue of pancreas by RT-qPCR (Figure 2C). Moderate levels of Cdnf mRNA expression were detected in pituitary gland and testis (Figure 2C). Low levels of Cdnf mRNA was observed in the other tissues studied (Figure 2C). Thus, the expression levels and tissue distribution of Manf mRNA differs from Cdnf mRNA in mouse tissues.

Consistent with Cdnf mRNA levels, highest levels of CDNF protein were observed in the skeletal muscle (19.2 ± 1.5 ng/mg of total protein, n = 4), heart (18.8 ± 1.4 ng/mg total protein, n = 4) and testis (10.8 ± 1.4 ng/mg total protein, n = 4) (Figure 2D). Interestingly, we observed high levels of CDNF protein also in brown adipose tissue (11.9 ± 0.4, ng/mg of total protein, n = 4). However, the median difference in CDNF and MANF concentration was ~125-fold in favor of MANF. The highest difference between MANF and CDNF concentrations was found in the pancreas (MANF vs. CDNF concentration 10 341x), liver (3 015x) and salivary gland (1 153x). In these tissues CDNF concentration was lowest measured (i.e., pancreas 0.09 ± 0.02 ng/mg total protein, n = 2; liver 0.12 ± 0.02 ng/mg total protein, n = 4; and salivary gland 0.28 ± 0.04 ng/mg total protein, n = 4, respectively). The smallest difference between MANF and CDNF concentrations was found in the heart (MANF vs. CDNF concentration 2.0 x), skeletal muscle (4.3x) and thyroid gland (4.5x). In these tissues, CDNF protein concentration was highest measured, while MANF concentration was at the lower end.

The MANF protein expression pattern in the brain was studied using MANF antibody staining. Our results are consistent with our previous in situ Manf mRNA hybridization studies on P10 and adult mouse brain (39, 40). In this study, we analyzed in detail the expression of MANF in the juvenile brain of 2 week-old mice using MANF IHC. We observed MANF expression in all brain regions in the P14 mouse brain (Figures 3A–D, Supplementary Figures 2A–BB, Supplementary Text).

The hypothalamus is a key area of the brain known for the regulation of food intake, energy balance, and the coordination of the peripheral endocrine system through the pituitary gland (43, 44). Consistent with previous studies (5, 33, 39, 40), abundant expression of MANF was detected in different hypothalamic regions responsible for the hormonal regulation via adenohypophysis of the pituitary gland such as medial preoptic area (MPO) (Figure 3C) (release of gonadotrophin-releasing hormones), paraventricular hypothalamic nuclei (PVN) (Figure 3D) (secretion of somatostatin, growth-hormone-inhibiting hormone, corticotropin-releasing hormone, thyrotropin-releasing hormone), arcuate hypothalamic nucleus (ARC) (Figure 3G) (responsible for the secretion of growth hormone-releasing hormone and dopamine-mediated prolactin inhibition) and median eminence (functional link between the hypothalamus and pituitary gland) (Figure 3G). Magnocellular neurons of the PVN and supraoptic nuclei (SON) (Figure 3E) parts of hypothalamus are also responsible for the production of oxytocin and vasopressin that are released to the bloodstream via neurohypophysis of the pituitary gland. In this study we identified that MANF is co-expressed with oxytocin and vasopressin positive neurosecretory cells in the PVN and SON nuclei of the hypothalamus (Figures 3S–X), suggesting that proper amount of MANF expression may be needed for cells regulating processes such as sexual behavior, social recognition, stress response, depression, aging, and water balance of the body.

We also detected high MANF-expression in hypothalamic nuclei that are important for appetite, feeding and energy expenditure including ARC (contains appetite-stimulating (orexigenic neuropeptide Y and agouti-related peptide-expressing neurons) and appetite-suppressing neurons [anorexigenic pro-opiomelanocortin (POMC)-expressing neurons] (45), PVN, DMH (Figure 3F), LH and VMH (Figure 3H).

Another key brain structure involved in the regulation of food intake is the brainstem. We identified high levels of MANF expression in the locus coeruleus (LC) nucleus (Figure 3M), dorsal nucleus of the vagus nerve (DMNX) (Figure 3R), rostral ventrolateral medulla (RVLM) (Figure 3O) and low levels in parabrachial nucleus (PBN) (Figure 3N), area postrema (AP) (Figure 3P), sensory nucleus of the solitary tract (NTS) (Figure 3Q). Additionally, MANF expression was detected in circumventricular organs, such as the subcommissural organ (SCO) (Supplementary Figure 2L). Importantly, we detected MANF expression in choline acetyltransferase positive motor neurons of the dorsal nucleus of the vagus nerve (Figures 3Y–AA), which are known for their role in the regulation of gastrointestinal tract secretion, motility and in mediating pancreatic secretion (46).

In addition, the brain reward system is implicated in the control of hedonic feeding regulated by the mesolimbic and mesocortical dopaminergic pathways. Moderate expression levels of MANF was observed in the medial prefrontal cortex (mPFC) (Figure 3I), nucleus accumbens (NAc) (Figure 3J), paraventricular nucleus of the thalamus (PVT) (Figure 3K), and ventral tegmental area (VTA) (Figure 3L). We identified that MANF positive cells are co-expressed with tyrosine hydroxylase (TH)-positive dopamine neurons in the VTA (Figures 3BB–DD).

Neurons of hypothalamus make projections onto neurons within the medulla that in turn project via the spinal cord through intermediolateral column (IML) to connect with the autonomic nervous system (ANS), thereby regulating heart rate, respiration, brown adipose tissues (BAT)/thermogenesis, and white adipose tissue (WAT) to modulate energy state. MANF expression was detected in neuronal bodies in the dorsal and ventral horns of the adult mouse spinal cord at P56 (Figures 4A–E). We also detected high MANF immunoreactivity in the IML nucleus (Figure 4C) of the spinal cord. MANF positive staining was observed in glial cells of the gray matter (Figure 4D) and in the myelinated axons of the white matter (Figure 4E). Very strong expression of MANF was detected in motor neurons and their dendrites immuno-labeled with choline acetyltransferase (Figures 4F–H) and neuronal and neuroendocrine protein gene product 9.5 (PGP9.5) markers (Figures 4I–K).

In the adult PNS, strong expression of MANF was observed within the sensory neuronal bodies of adult dorsal root ganglia (DRG) (Figures 4L–N) and in sympathetic celiac ganglia (CLG) (Figures 4U–W). BAT thermogenesis is regulated by a negative feedback loop via the DRG sympathetic nervous system stimulation (47). CLG is a part of the ANS which innervates most of the digestive tract, liver, and pancreas, thereby regulating their functions. Similarly to the spinal cord, only neuronal bodies were positive for MANF and not nerve fibers in DRG and CLG (Figures 4O–T,X–Z,AA–CC).

The endocrine system consists of the hypothalamus, pituitary gland, pineal body, thyroid and parathyroid glands, adrenal glands, and the reproductive organs (ovaries and testes). These glands secrete hormones to the blood circulation, which control cellular and organ growth, development, metabolism, reproduction, sleep, and response to stress stimuli among others. Incorrect amount of hormone production leads to medical conditions such as insulin-deficiency in Type 1 diabetes, growth hormone-deficiency in postnatal growth defects caused by hypopituitarism and hormonal dysregulation in metabolic syndrome (43, 48, 49). Our qPCR and ELISA results for MANF expression showed that MANF is highly expressed in mouse endocrine tissues such as the pituitary gland, the thyroid gland, pancreatic islets and testis as well as exocrine tissues such as exocrine pancreas, salivary glands and other tissues containing cells with secretory function (Supplementary Text and Supplementary Figure 4) compared to the rest of the tissues (Supplementary Text and Supplementary Figures 5, 6). Therefore, we studied the cellular distribution of MANF expression by IHC in different juvenile and adult mouse endocrine organs (Figure 5). In the adult mouse pituitary gland, strong MANF immunoreactivity was observed in most cells of the endocrine anterior adenohypophysis (Figures 5A,B) and the intermediate lobe (Figures 5A,C) as well as in cells of the neurohypophysis (Figures 5A,D). Additionally, we found strong MANF expression in the thyroid follicular cells producing thyroxine (T4) and triiodothyronine (T3) (Figures 5E,F). MANF was also co-expressed with calcitonin in parafollicular C-cells, responsible for reducing calcium levels in the blood by inhibiting osteoclast activity in bones and inhibiting renal reabsorption (Figures 5H–J). MANF immunoreactivity was strong also in parathyroid glands (Figures 5E,G), that is responsible for the production of parathyroid hormone and thereby regulation of calcium and phosphate levels in the blood. In the adrenal gland, in contrast to moderate and low levels of MANF mRNA and protein detected by RT-qPCR and ELISA (Figures 2A,C), intense MANF staining was detected in the cortex and less in the medulla (Figure 5K). Furthermore, MANF was co-expressed with TH in the medullar chromaffin cells, responsible for the synthesis and release of catecholamine, adrenaline, and noradrenaline, in response to stress (Figures 5L–N).

In the mouse reproductive system, we observed MANF expression in the mouse ovaries and ampullas of the Fallopian tubes (Figures 5O–Q). MANF expression was detected in granulosa or cumulus cells, which surround and nourish the oocytes in the ovary (Figure 5P). Additionally, we detected MANF expression in male seminiferous tubules of the mouse testis at P14 (Figure 5R).

The phenotype of Manf^−/− mice is characterized by the progressive loss of circulating insulin leading to diabetes and a severe growth defect (16). We have previously demonstrated that the dramatic growth retardation in conventional Manf^−/− mice is not dependent on the diabetic phenotype as the growth defect is not recapitulated in the pancreas-specific Pdx-1Cre::Manf^fl/fl mice, that develop diabetes but no growth retardation (16). In addition, neuron-specific NestinCre^+/−::Manf^fl/fl mice do not develop an equally severe growth defect compared to global Manf^−/− mice [(16); Pakarinen et al. submitted manuscript]. Short stature or dwarfism is usually caused by a pituitary growth-hormone deficiency in humans (50). Thus, the short stature of Manf^−/− mice and the high MANF expression in the pituitary gland prompted us to study the pituitary gland in more detail in the Manf^−/− mice.

The pituitary gland plays crucial roles in growth, reproduction, endocrine functioning, and functions to convey signals from the hypothalamus to various organs (43, 51). The endocrine anterior pituitary gland contains five types of cells, each secreting different hormones, where somatotropes secrete growth hormone (GH), lactotropes prolactin (PRL), gonadotropes follicle stimulating hormone (FSH) and luteinizing hormone (LH), corticotropes adrenocorticotropic hormone (ACTH), and thyrotropes thyroid-stimulating hormone (TSH) (43). We stained sections from the pituitary glands with hematoxylin-eosin for morphological analysis (Figures 6A,B). This revealed that the size of the adenohypophysis in Manf^−/− mice was reduced compared to the adenohypophysis from Manf^+/+ mice in relation to body weight at 6 weeks of age (Figure 6E), while the area of the intermediate lobe was not changed between the groups (Supplementary Figure 7A) and the area of neurohypophysis was increased in Manf^−/− mice compared to control (Supplementary Figure 7B). The ratio of granule-filled acidophilic cells (purple) representing somatotropes and lactotropes (Figures 6C,D) compared to basophils (blue) were significantly reduced in the anterior lobe of 6-week-old Manf^−/− mice compared to Manf^+/+ mice (Figure 6F). Double immunofluorescence staining revealed co-localization of MANF with growth hormone (Figures 6G–I) and prolactin (Figures 6J–L) in cells of the anterior pituitary lobe at 6 weeks of age.

Immunoperoxidase staining of the pituitary glands revealed decreased GH (Figure 6N) and PRL (Figure 6P) expression in the adenohypophyses of the Manf^−/− mice compared to Manf^+/+ mice (Figures 6M,O). Quantification of the ratio of GH- and PRL-positive cells in the adenohypophyses confirmed a severely reduced number of GH (Figure 6Q) and PRL (Figure 6R) producing cells in Manf^−/− mice compared to that of Manf^+/+ mice. In line with the reduced size of the adenohypophysis in Manf^−/− mice, we found a clear reduction in the number of proliferative cells (Figure 6S, Supplementary Figures 7C,D) but no increase in the number of dying cells (Supplementary Figure 7E) in the adenohypophyses of Manf^−/− mice.

Consistent with the reduced number of GH and PRL positive cells, mRNA expression levels of Gh and Prl (Figure 6T) genes were significantly decreased in the pituitaries isolated from Manf^−/− mice compared to those of Manf^+/+ mice regardless of gender. The levels of pro-opiomelanocortin (Pomc) mRNA and Thyroid-stimulating hormone (Tsh) mRNA were significantly increased in Manf^−/− pituitaries (Figure 6T), probably reflecting the higher relative number of corticotropes and thyrotropes in the Manf^−/− pituitaries. In accordance with the essential role of the pituitary-specific positive transcription factor 1 (Pit1) for the development of somatotropes and lactotropes, we observed that Pit1 mRNA expression was significantly downregulated in the Manf^−/− mutant pituitary glands (Figure 6T). Both mRNA levels for luteinizing hormone (Lh) and follicle-stimulating hormone (Fsh) were non-significantly downregulated in the pituitaries of 6 weeks old Manf^−/− male mice compared to Manf^+/+ controls (Figures 6U,V), while Fsh mRNA levels were significantly upregulated and Lh mRNA was not affected in pituitaries of Manf^−/− female mice.

As chronic ER stress precedes the reduced pancreatic β-cells proliferation and increased β-cells death leading to diabetes in Manf^−/− mice (16), we investigated whether ablation of MANF from the pituitary leads to ER stress and chronic UPR activation. Importantly, we found significant upregulation of UPR genes representing all three main UPR pathways PERK, IRE1α, and ATF6 in the pituitaries isolated from 6 week-old Manf^−/− mice indicating that chronic ER stress may be a cause of the acidophilic cell loss in the pituitary glands of Manf^−/− mice (Figure 6W).

Hence, these results suggest that MANF expression is needed to support the GH and PRL expressing cells in the anterior pituitary of mice.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.