Tmem117 in AVP neurons regulates the counterregulatory response to hypoglycemia

Abstract

The counterregulatory response to hypoglycemia (CRR), which ensures a sufficient glucose supply to the brain, is an essential survival function. It is orchestrated by incompletely characterized glucose‐sensing neurons, which trigger a coordinated autonomous and hormonal response that restores normoglycemia. Here, we investigate the role of hypothalamic Tmem117, identified in a genetic screen as a regulator of CRR. We show that Tmem117 is expressed in vasopressin magnocellular neurons of the hypothalamus. Tmem117 inactivation in these neurons increases hypoglycemia‐induced vasopressin secretion leading to higher glucagon secretion in male mice, and this effect is estrus cycle phase dependent in female mice. Ex vivo electrophysiological analysis, in situ hybridization, and in vivo calcium imaging reveal that Tmem117 inactivation does not affect the glucose‐sensing properties of vasopressin neurons but increases ER stress, ROS production, and intracellular calcium levels accompanied by increased vasopressin production and secretion. Thus, Tmem117 in vasopressin neurons is a physiological regulator of glucagon secretion, which highlights the role of these neurons in the coordinated response to hypoglycemia.

This study identifies Tmem117 as a marker of AVP magnocellular neurons that regulates AVP secretion by interfering with ER stress, intracellular calcium, and ROS production.

Introduction

The brain relies almost exclusively on glucose for metabolic energy production. Therefore, maintaining blood glucose levels no lower than ~5 mM is essential for survival (Marty et al, ²⁰⁰⁷). When glycemic levels drop below this threshold, a brain‐orchestrated neuroendocrine reflex triggers a counterregulatory response (CRR). This is characterized by the secretion of multiple hormones that act on peripheral target organs to stimulate endogenous glucose production, suppress insulin secretion, and minimize glucose utilization to restore normoglycemia (Tesfaye & Seaquist, 2010). A key aspect of the CRR is the induction of glucagon (GCG) secretion from pancreatic alpha cells, which triggers enhanced glucose production from the liver (Ramnanan et al, ²⁰¹¹; Thorens, ²⁰²²). Patients with type 1 or advanced type 2 diabetes display loss of hypoglycemia‐induced GCG secretion (Cryer, 2012; Siafarikas et al, ²⁰¹²; Bisgaard Bengtsen & Møller, ²⁰²¹), but the underlying mechanism is poorly understood. In addition to deregulated alpha cell autonomous and intraislet paracrine interactions (Gaisano et al, ²⁰¹²), evidence suggests that defects in the central nervous system (CNS) control of GCG secretion contribute to this blunted response (Beall et al, ²⁰¹²; Stanley et al, ²⁰¹⁹).

In the CNS, changes in glucose concentrations are detected by neurons located in several regions but have been studied most extensively in the hypothalamus and brainstem. Several hypothalamic areas are equipped with glucose‐sensing neurons, including the ventromedial (VMH), dorsomedial (DMH), lateral, arcuate (ARC), paraventricular (PVN), and supraoptic (SON) nuclei (Stanley et al, ²⁰¹⁹). Glucose responsive neurons are activated by either a rise (glucose excited/GE neurons) or a fall (glucose inhibited/GI neurons) in glucose levels. The response to hypoglycemia involves activation by GI neurons of both the parasympathetic and sympathetic branches of the autonomic nervous system, which stimulate GCG secretion and hepatic glucose production, the activation of the hypothalamus–pituitary–adrenal (HPA) axis to stimulate epinephrine and glucocorticoid secretion (Tesfaye & Seaquist, 2010), and the activation of hypothalamic vasopressin (AVP) neurons to stimulate GCG secretion through the actions of AVP on pancreatic alpha cell AVP V1b receptors (Gao et al, ¹⁹⁹²; Yibchok‐anun et al, ²⁰⁰⁴; Kim et al, ²⁰²¹).

Multiple studies have provided evidence in support for the role of hypothalamic and brainstem glucose sensing neurons in CRR. For instance, cell‐specific activation of VMH neuronal subpopulations expressing glucokinase (Gck) (Meek et al, ²⁰¹⁶) or steroidogenic factor 1 (Sf1) (Stanley et al, ²⁰¹⁶) increases GCG secretion and glycemia, while their inhibition blocks CRR. Interestingly, genetic inactivation of Gck in Sf1 neurons impairs GCG secretion in a sex‐dependent manner (Steinbusch et al, ²⁰¹⁶). Glucose responsive neurons located in other nuclei have also been shown to control GCG secretion. This is the case of cholecystokinin‐expressing neurons of the parabrachial nucleus that control VMH neurons (Garfield et al, ²⁰¹⁴), of fibroblast growth factor 15‐expressing neurons of the DMH (Picard et al, ^{2016, 2021}), and of Glut2 GI neurons of the nucleus tractus solitarius, which activate vagal nerve firing to stimulate GCG secretion (Lamy et al, ²⁰¹⁴).

To identify in an unbiased manner novel hypothalamic regulators of CRR induced by insulin‐induced hypoglycemia, we performed a genetic screen using a panel of recombinant inbred BXD mouse lines, derived from the cross of C57Bl/6 and DBA/2 mice (Peirce et al, ²⁰⁰⁴). This screen (Picard et al, ²⁰²²) led to the identification of Agpat5, encoding a lipid biosynthesis enzyme, and the characterization of its essential role in the activation by hypoglycemia of AgRP neurons and GCG secretion (Strembitska et al, ²⁰²²). We further identified Tmem117, located in a QTL on chromosome 15, as another candidate regulator of GCG secretion. Tmem117 encodes an eight transmembrane‐containing membrane protein (Bürgi et al, ²⁰¹⁶) that has been reported to negatively control ER stress and ROS production (Tamaki et al, ²⁰¹⁷).

Here, we show that Tmem117 is expressed in AVP magnocellular neurons, that AVP neurons are in large part GI neurons, and that hypoglycemia induces copeptin (CPP; an AVP surrogate) and GCG secretion. Genetic inactivation of Tmem117 in AVP neurons increases, in a sex‐dependent manner, hypoglycemia‐induced CPP and GCG secretion. Tmem117 inactivation does not affect the glucose sensing properties of AVP neurons but induces ER stress and increases intracellular ROS and Ca²⁺ levels as well as AVP mRNA expression.

Results

Tmem117 is expressed in AVP magnocellular neurons

A genetic screen of a panel of 36 recombinant inbred BXD mice for hypoglycemia‐induced GCG secretion identified a clinical quantitative trait locus (cQTL) on chromosome 8 and one on chromosome 15 (Picard et al, ²⁰²²). The candidate gene on chromosome 8 encodes the lipid‐modifying enzyme Agpat5, which is required for hypoglycemia sensing by AgRP neurons, vagal nerve activation, and GCG secretion (Strembitska et al, ²⁰²²). On chromosome 15, two genes showed strong negative correlation with the GCG trait, Irak4 and Tmem117 (Fig EV1A–D). The role of Irak4 in controlling hypothalamic Il‐1β signaling and GCG secretion has been reported (Picard et al, ²⁰²²). Here, we performed an expression QTL analysis (eQTL) for the level of Tmem117 mRNA in the hypothalamus of the BXD mouse lines. We identified an eQTL on chromosome 15 at the same position of the cQTL (Fig EV1E). Thus, indicating that this genomic locus controls both GCG secretion and Tmem117 expression.

Figure EV1. Hypothalamic Tmem117 is a candidate gene for the control of insulin‐induced glucagon secretion.

1. Clinical quantitative trait locus (cQTL) analysis revealed two genomic regions controlling insulin‐induced GCG secretion, on chromosomes 8 and 15.
2. The peak on chromosome 15, between markers c15.loc87 and rs13482723, contains 42 genes.
3. The genes showing the highest correlation between their hypothalamic mRNA expression and plasma GCG are Irak4 and Tmem117.
4. Correlation between the level of Tmem117 expression in the hypothalamus of the 36 BXD mouse lines and insulin‐induced plasma GCG levels.
5. Expression quantitative trait locus (eQTL) analysis for Tmem117 expression in the hypothalamus of the 36 BXD mouse strains. A cis eQTL was identified on chromosome 15 with a LRS of 43.3.

Data information: Panels A–C are adapted from Picard et al (2022). LRS: likelihood ratio statistic, TPM: transcripts per million.

To establish the sites of Tmem117 expression in the hypothalamus we performed immunofluorescence microscopy analysis. We found strong immunolabeling for Tmem117 in the SON and PVN (Fig 1A) and in the posterior pituitary (Fig 1B). The specificity of the immunostaining was confirmed by the lack of signal in brains of mice with Tmem117 gene inactivation in AVP neurons (Tmem117 ^fl/fl ;AVP‐IRES‐Cre‐D ^tg/+ mice) (Fig 1C and D). Costaining for neuropeptides revealed that Tmem117 was expressed in AVP magnocellular cells (Fig 1E–G). Quantification in three mice and two consecutive hypothalamic sections for each mouse (bregma −0.7 and −0.8) showed that ~90% of AVP neurons in the PVN and ~97% in the SON were Tmem117 positive (Fig 1H and I). Super‐resolution microscopy revealed a punctuated intracellular distribution of Tmem117, which colocalized in part with AVP granules in both the soma and axons of magnocellular neurons (Fig 1J).

Figure 1. Tmem117 is expressed in AVP magnocellular neurons.

Immunofluorescence microscopy detection of Tmem117 in the mouse hypothalamus and pituitary.

• A, B
  Positive immunostaining was observed in the PVN, the SON, and the posterior pituitary.
• C, D
  No immunostaining was detected in the hypothalamus and pituitary of mice with constitutive inactivation of Tmem117 in AVP neurons.
• E–G
  Costaining for Tmem117 and AVP in the (E) PVN, (F) SON, and (G) posterior pituitary.
• H, I
  Quantification of AVP and Tmem117 double positive cells in the PVN and SON, respectively.
• J
  Higher resolution depiction of an AVP and Tmem117 double positive neuron in the PVN captured with structured illumination microscopy. The white arrows highlight representative positions of signal colocalization.

Data information: 3V: third ventricle, AP: anterior pituitary, MZ: medial zone, opt: optic tract, PP: posterior pituitary, PVN: paraventricular nucleus, SCh: suprachiasmatic nucleus, SON: supraoptic nucleus. Scale bar = 100 μm for panels A–G, =10 μm for panel J.

Source data are available online for this figure.

Tmem117 inactivation in AVP neurons increases insulin‐induced copeptin and glucagon secretion

To inactivate Tmem117 specifically in AVP neurons, we generated Tmem117 ^fl mice that allowed Cre‐dependent excision of exon 3, which encodes the third transmembrane domain (amino acids 93–136) of the predicted structure of Tmem117 (Jumper et al, ²⁰²¹; Varadi et al, ²⁰²²) (Fig EV2A and B). To induce Tmem117 ^fl recombination in AVP neurons, we used a viral‐mediated approach for expression of Cre recombinase. This was preferred over the use of AVP‐IRES‐Cre‐D ^tg/+ mice because these display reduced endogenous AVP levels (Cheng et al, ²⁰¹⁹). We, thus, constructed an AAV plasmid containing the full‐length AVP promoter placed upstream of a codon‐improved Cre recombinase (iCre) sequence (Ponzio et al, ²⁰¹²) (Fig EV2C). The resulting plasmid was packaged in serotype 6 adeno‐associated viruses (AAV6‐AVP‐iCre), which infect magnocellular neurons with high efficacy (Ponzio et al, ²⁰¹²) and have a high retrograde transport capacity (Salegio et al, ²⁰¹³). Injection of AAV6‐AVP‐iCre in the posterior pituitary of adult Tmem117 ^fl/fl mice triggered efficient recombination in the PVN and SON as detected by PCR analysis of genomic DNA (Fig EV2D) and by the expression of tdTomato when the AAV6‐AVP‐iCre was injected in the posterior pituitary of Rosa26 ^tdTomato mice (Fig EV2E).

Figure EV2. Tmem117 ^fl mice and viral construct generation.

1. Predicted structure and transmembrane localization of the mouse Tmem117 (adapted from AlphaFold). The pink amino acids highlighted by the pink arrows are the first (aa 93) and the last (aa 136) encoded by the exon 3 sequence that corresponds to the whole 3^rd transmembrane helix.
2. Schematic representation of the Tmem117 wild‐type locus (+), the Tmem117 ^fl locus (fl), and the recombined Tmem117 locus (KO). Primer binding sites and expected amplicon sizes are indicated.
3. Generation of AAV‐AVP‐iCre plasmid. The pCDH‐CB‐iCre plasmid was used as a donor for the iCre sequence located between the NotI and AgeI restriction sites. The p2.0VPI.EGFP plasmid contained the full‐length 2 kb AVP promoter in an AAV vector. The GFP‐containing NotI‐AgeI fragment of this plasmid was replaced with the iCre sequence to generate the AAV‐AVP‐iCre plasmid.
4. PCR analysis of Tmem117 recombination. DNA extracted from microdissected PVN and SON of mice injected in the posterior pituitary with the AAV6‐AVP‐iCre showed a 225 bp fragment indicative of Tmem117 ^fl recombination. SON DNA from Tmem117 ^fl mice crossed with AVP‐IRES2‐Cre‐D ^tg/+ mice was used as positive control.
5. Injection of AAV6‐AVP‐iCre in the posterior pituitary of Rosa26 ^tdTomato/+ mice led to efficient expression of tdTomato in AVP magnocellular neurons of the PVN and SON.

Data information: AVP: vasopressin, PP: posterior pituitary, PVN: paraventricular nucleus, SON: supraoptic nucleus. Scale bar = 100 μm.

To assess the role of Tmem117 in GCG secretion, Tmem117 ^fl/fl and Tmem117 ^+/+ mice were injected with the AAV6‐AVP‐iCre in the posterior pituitary to generate AVP ^TM117KO and AVP ^TM117WT mice, respectively (Fig 2A). Two weeks later, the mice received an intraperitoneal (i.p.) injection of saline and blood was collected 1 h later for plasma CPP and GCG measurements. The same experiment was repeated 1 week later with injection of insulin instead of saline to induce hypoglycemia. Insulin induced the same hypoglycemic levels in AVP ^TM117WT and AVP ^TM117KO male mice (Fig 2B); CPP secretion was increased in both groups of mice, but significantly more in AVP ^TM117KO mice than in AVP ^TM117WT mice (Fig 2C); the same pattern was observed for the secretion of GCG (Fig 2D).

Figure 2. Tmem117 inactivation in AVP neurons enhances hypoglycemia‐induced copeptin and glucagon secretion.

• A
  Experimental scheme. AAV6‐AVP‐iCre was injected in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice.
• B
  Glycemia of male mice 1 h after saline (black) or insulin (red) i.p. injection (n = 15–16 mice per group).
• C
  CPP plasma levels of male mice 1 h after saline or insulin injection (n = 13 mice per group; mean ± SEM for INS: WT 89 ± 5 vs. KO 121 ± 13 pg/ml).
• D
  GCG plasma levels of male mice 1 h after saline or insulin injection (n = 15–16 mice per group; mean ± SEM for INS: WT 124 ± 19 vs. KO 186 ± 21 pg/ml).
• E
  Glycemia of female mice 1 h after saline (black) or insulin (red) i.p. injection (n = 18–22 mice per group).
• F
  CPP plasma levels of female mice 1 h after saline or insulin injection (n = 18–19 mice per group; mean ± SEM for INS: WT 78 ± 6 vs. KO 92 ± 8 pg/ml).
• G
  GCG plasma levels of female mice 1 h after saline or insulin injection (n = 17–20 mice per group; mean ± SEM for INS: WT 242 ± 14 vs. KO 277 ± 18 pg/ml).
• H–J
  Analysis of glycemic levels and of CPP and GCG secretion following insulin injection in female mice at each stage of the estrus cycle (n = 3–7 mice per group). (H) Glycemic levels. (I) CPP plasma levels. (J) GCG plasma levels.

Data information: d: day, di: diestrus, est: estrus, INS: insulin, met: metestrus, PP: posterior pituitary, pro: proestrus, SAL: saline. For panels B–G, bars correspond to the mean value per group. For panels H–J, lines correspond to the mean value per group and error bars represent ± SEM. B–G: two‐way ANOVA RM with Bonferroni post hoc test; H–J: unpaired t test; *P < 0.05, **P < 0.01, ***P < 0.001.

Source data are available online for this figure.

Female AVP ^TM117WT mice and AVP ^TM117KO mice were then tested in the same conditions. Insulin induced the same level of hypoglycemia (Fig 2E) and triggered comparable secretion of CPP and GCG in both groups of mice (Fig 2F and G). However, when the hormone levels were analyzed separately for each phase of the estrus cycle, we found that while the insulin‐induced hypoglycemic levels were comparable across the estrus cycle between genotypes (Fig 2H), the level of secreted CPP and GCG were significantly higher in AVP ^TM117KO mice as compared to AVP ^TM117WT mice during the proestrus phase (Fig 2I and J).

Glucose responsiveness of SON AVP neurons is not affected by Tmem117 inactivation

To assess whether AVP neurons were activated by insulin‐induced hypoglycemia, C57Bl/6N male mice were injected with a saline solution or with insulin and their brains were collected 2 h later. Immunofluorescence microscopy analysis revealed that hypoglycemia (<3.9 mmol/l) robustly increased c‐Fos expression in both PVN and SON (Fig 3A–E) with an approximately threefold increase in the number of c‐Fos⁺ AVP neurons compared to the saline injected controls (Fig 3C and F). Furthermore, analysis of microdissected PVN and SON tissue 1 h after insulin injection revealed downregulation of Tmem117 mRNA in the SON (Fig 3G–I).

Figure 3. Insulin‐induced hypoglycemia triggers activation of AVP magnocellular neurons and reduces Tmem117 mRNA levels in the SON.

• A–F
  C57BL6/N male mice were injected i.p. with saline or insulin and c‐Fos expression (green) in AVP (magenta) neurons was quantitated by immunofluorescence microscopy 2 h later. (A) c‐Fos expression in the PVN after saline injection. (B) c‐Fos expression in the PVN after insulin injection. (C) Quantitation of c‐Fos‐positive AVP cells in the PVN after saline and insulin injections (n = 4 mice per group). (D) c‐Fos expression in the SON after saline injection. (E) c‐Fos expression in the SON after insulin injection. (F) Quantitation of c‐Fos‐positive AVP cells in the SON after saline and insulin injections (n = 4 mice per group).
• G–I
  C57Bl/6N male mice were injected i.p. with saline or insulin and brain tissue was collected 1 h later. PVN and SON were microdissected for quantification of Tmem117 mRNA levels by RT‐PCR. (H) Tmem117 mRNA levels in the PVN (n = 4 mice per group). (I)Tmem117 mRNA levels in the SON (n = 4 mice per group).

Data information: 3V: third ventricle, AVP: vasopressin, INS: insulin, opt: optic tract, PVN: paraventricular nucleus, SAL: saline, SON: supraoptic nucleus. Scale bar = 100 μm. Lines correspond to the mean value per group and error bars represent ± SEM. Unpaired t test; ns: P > 0.05, *P < 0.05, **P < 0.01.

Source data are available online for this figure.

To determine whether Tmem117 inactivation would modify insulin‐induced c‐Fos expression in AVP neurons, we prepared AVP ^TM117KO mice and AVP ^TM117WT control mice by injection of AAV6‐AVP‐iCre in the posterior pituitary of Tmem117 ^fl/fl and Tmem117 ^+/+ male mice (Fig 4A). Two weeks later, the mice were treated as described earlier for the C57Bl/6N. Induction of c‐Fos expression in AVP neurons of the SON was comparable between the two mouse groups (Fig 4B and C).

Figure 4. Glucose responsiveness of Tmem117 KO AVP neurons.

1. Experimental scheme. AAV6‐AVP‐iCre was injected in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice.
2. c‐Fos expression in the SON.
3. Quantitation of c‐Fos‐positive AVP cells in the SON after saline and insulin injections (n = 4–5 mice per group).
4. Experimental scheme. AAV6‐AVP‐iCre was coinjected with AAV8‐DIO‐mcherry in the SON of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice.
5. Proportion of GI and NR AVP neurons in the SON (n = 16–18 cells per group).
6. Representative traces of GI and NR AVP neurons in each group.
7. Membrane potential was increased comparably in response to 0.1 mM glucose in both groups (n = 7–9 cells per group).
8. No change in membrane potential in NR neurons of both genotypes (n = 9 cells per group).
9. Membrane resistance was decreased in six of nine AVP ^TM117WT and six of seven AVP ^TM117KO cells in response to 0.1 mM glucose (n = 7–9 cells per group).
10. Membrane resistance in NR neurons was stable for both genotypes (n = 9 cells per group).

Data information: AVP: vasopressin, d: day, GI: glucose inhibited, INS: insulin, NR: nonresponding, opt: optic tract, SAL: saline, w: week. Scale bar = 100 μm. Lines correspond to the mean value per group and error bars represent ± SEM. C: Two‐way ANOVA with Tukey's post hoc test; E: Fisher's exact test; G–J: paired t test for all graphs except Δ membrane potential (unpaired); ns P > 0.05, *P < 0.05, ***P > 0.001.

Source data are available online for this figure.

To determine whether AVP neurons were sensitive to the decreased glucose availability and whether Tmem117 would modify this sensitivity, we performed patch clamp recordings of AVP neurons in the presence of 2.5 and 0.1 mM glucose (Labouèbe et al, ²⁰¹⁸). Tmem117 ^+/+ and Tmem117 ^fl/fl adult male mice were coinjected in the SON with AAV6‐AVP‐iCre and an AAV8‐DIO‐mCherry to allow fluorescence visualization of the AVP neurons (Fig 4D). SON was preferred over PVN due to its homogeneous magnocellular AVP population. Direct injection was used instead of retrograde labeling by injection in the posterior pituitary due to its higher labeling efficiency. One to two weeks after surgery, we performed patch clamp analysis of AVP neurons. In brain slices of AVP ^TM117WT mice, we found that approximately half of the recorded AVP neurons were activated by low glucose (GI neurons) and that the other half were glucose nonresponder (NR) neurons; no glucose excited (GE) neurons were observed (Fig 4E–H; black). When the same measurements were performed in cells with Tmem117 gene inactivation, the proportion of GI and NR cells and their membrane potential responses were similar to that of AVP ^TM117WT mice (Fig 4E–H; red). Thus, AVP neurons were in large part GI neurons and Tmem117 inactivation did not alter their glucose responsiveness. Furthermore, exposure of AVP neurons to low glucose concentration decreased the membrane resistance of six of seven AVP ^TM117KO GI cells and six of nine AVP ^TM117WT GI cells (Fig 4I), with no effect in NR cells of both genotypes (Fig 4J), suggesting a cell autonomous response to hypoglycemia at least in a subpopulation of AVP GI neurons. However, given that the overall comparison did not reach statistical significance, this is just a suggestive observation.

Tmem117 inactivation increases ER stress, ROS production, intracellular Ca²⁺ levels, and AVP mRNA expression

Tmem117 has been identified in a siRNA screen of the HTC116 colon cancer cell line as a negative regulator of ER stress and ROS production (Tamaki et al, ²⁰¹⁷). To verify its contribution to ER stress, we used the colon adenocarcinoma cell line SW480, which endogenously expresses Tmem117. SW480 cells were transfected with a siRNA targeting Tmem117 or a siRNA control and 48 h later the cells were collected for RNA and protein extraction. A small reduction in Tmem117 mRNA levels (log₂fold approximately −0.5) robustly increased expression of the ER stress markers sXbp1 and BiP (Fig EV3A) and calnexin (Fig EV3B). Tmem117 silencing in this cell line did not increase ROS production (Fig EV3C), but Tmem117 overexpression significantly reduced ROS production (Fig EV3D). These data confirm a link between Tmem117 expression and ER stress and ROS production in carcinoma cell lines.

Figure EV3. Tmem117 is modulating ER stress and ROS production in SW480 cells.

1. Expression levels of ER stress markers in SW480 cells. qRT‐PCR quantification (n = 8–9 samples/wells per group).
2. Protein levels of calnexin in SW480 cells upon Tmem117 silencing. Western blot quantification (n = 6 samples/wells per group).
3. Intracellular ROS measurement in SW480 cells. DHE fluorescent signal in SW480 cells (n = 16 samples/wells per group).
4. Intracellular ROS measurement in SW480 cells upon overexpression of Tmem117. DHE fluorescent signal in SW480 cells (n = 16 samples/wells per group).

Data information: DHE: dihydroethidium, RFU: relative fluorescent units. Bars correspond to the mean value per group and error bars represent ±SEM. Unpaired t test; ns: P > 0.05, **P < 0.01, ***P > 0.001.

To test whether inactivation of Tmem117 in AVP magnocellular neurons in vivo would also affect ER stress and ROS production, we prepared AVP ^TM117KO and AVP ^TM117WT mice by injecting AAV6‐AVP‐iCre in the posterior pituitary of Tmem117 ^fl/fl and Tmem117 ^+/+ mice. Two weeks later, we collected their brains and measured the level of Bip mRNA by in situ hybridization (RNAScope). Quantitative analysis at the single‐cell level revealed higher expression of Bip mRNA in AVP neurons of AVP ^TM117KO mice as compared to those of AVP ^TM117WT mice; non‐AVP cells in the SON showed no difference in Bip mRNA expression between the two groups of mice (Fig 5A–C). Furthermore, we found markedly elevated levels of AVP mRNA in AVP ^TM117KO mice as compared to control mice (Fig 5C and D). AVP transcription has been reported to be induced by ROS (St‐Louis et al, ^{2012, 2014}). We, thus, assessed the ROS levels in the SON by injecting i.p. the fluorescent ROS indicator dihydroethidium (DHE) in AVP ^TM117KO and AVP ^TM117WT mice 24 h before tissue collection (Fig 5E). Quantification of the DHE fluorescence intensity revealed higher ROS production in the SON of AVP ^TM117KO mice (Fig 5F and G).

Figure 5. Tmem117 inactivation in AVP neurons increases AVP and BiP mRNA, ROS levels, and intracellular calcium in vivo .

1. Quantification of BiP mRNA per cell in AVP‐positive cells of the SON (n = 483–704 cells per group; each group consisted of four mice).
2. Quantification of BiP mRNA per cell in AVP‐negative cells of the SON (n = 347–584 cells per group; each group consisted of four mice).
3. Fluorescence microscopy detection of AVP mRNA (green) and BiP mRNA (magenta) in the SON.
4. Quantification of AVP mRNA per cell (n = 483–704 cells per group; each group consisted of four mice).
5. Experimental scheme. AAV6‐AVP‐iCre was injected in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice. DHE was injected i.p. (50 mg/kg) 24 h before tissue collection.
6. Fluorescence microscopy detection of DHE‐derived fluorescence (red) in the SON.
7. Quantification of DHE‐derived fluorescence in the SON (n = 8 SON per group; each group consisted of four mice).
8. Experimental scheme. AAV6‐AVP‐iCre was injected in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice. An optic fiber was implanted in the same area for signal analysis in freely moving mice.
9. Basal calcium levels measured in AVP magnocellular terminals of AVP ^TM117WT and AVP ^TM117KO mice. The gray dashed line represents the time at which the optic fiber cable was plugged onto the implanted cannula (n = 5–6 mice per group).
10. Calcium recordings in AVP magnocellular terminals 10 min before and 1 h after the i.p injection of insulin (red and orange) or saline (black). The vertical arrow corresponds to the i.p. injection (time = 0).
11. Quantification of mean signal intensity in 10‐min time bins (n = 6–8 mice per group).
12. Stereoscope image of the pituitary gland depicting the optic fiber tract in the posterior pituitary.
13. Fluorescence microscopy detection of GCaMP (green) in posterior pituitary. Immunofluorescent labeling was used to localize the AVP terminals (magenta).

Data information: AUC: area under the curve, AVP: vasopressin, d: day, DHE: dihydroethidium, INS: insulin, opt: optic tract, PP: posterior pituitary, SAL: saline, SON: supraoptic nucleus. Scale bar = 20 μm. For panels G and K, data are represented as mean ± SEM. For the violin plots of panels A–C, dashed lines correspond to the median value and dotted lines to the quartile values. A–C,G: unpaired t test; I: two‐way ANOVA RM with Bonferroni post hoc test; K: one‐way ANOVA with Tukey's post hoc test; *P < 0.05, ***P > 0.001.

Source data are available online for this figure.

Both ER stress and ROS can increase intracellular Ca²⁺ concentrations (Giorgi et al, ²⁰¹⁸), which would enhance neurosecretion. We, thus, measured the intracellular Ca²⁺ levels in AVP neuronal terminals at the posterior pituitary, the site of AVP exocytosis, by in vivo fiber photometry. AVP ^TM117KO and AVP ^TM117WT mice were prepared by coinjection of the AAV6‐AVP‐iCre virus and an AAV9‐hsyn‐FLEX‐jGCaMP7 in the posterior pituitary of Tmem117 ^fl/fl and Tmem117 ^+/+ mice. During the same surgery an optic fiber cannula was implanted in the posterior pituitary for fluorescence monitoring (Fig 5H). Two weeks later, the baseline jGCaMP7 fluorescence signal was recorded in both control and knockout mice. This showed that inactivation of Tmem117 induced a markedly higher baseline Ca²⁺ signal (Fig 5I). We then performed Ca²⁺ recordings after insulin‐induced hypoglycemia. The first measures were performed at days 14–16 after viral injection. A second session of recording was performed over days 19–24 in AVP ^TM117WT mice injected with saline and these data were used as control for the first set of experiments. For both sessions, the Ca²⁺ signal was recorded during a 30‐min baseline period and for 1 h following i.p. insulin or saline injections (Fig 5J). Each signal was normalized to the mean intensity of the baseline (corresponding to the last 10 min preceding the i.p. injection) and the area under the curve (AUC) was quantified for the mean trace of each group over 10‐min period (Fig 5K). In AVP ^TM117WT mice, insulin injection increased the Ca²⁺ signal as compared to saline injection, a difference that became significant 20 min after the injection. In AVP ^TM117KO mice, insulin induced a significantly higher Ca²⁺ signal than in AVP ^TM117WT mice already 10 min after injection (Fig 5J and K). The correct placement of the optic fiber (Fig 5L) was assessed at the end of each experiment and the expression of jGCaMP7 in AVP magnocellular terminals was verified by fluorescence microscopy (Fig 5M).

Together, the above experiments indicate that Tmem117 inactivation in AVP magnocellular neurons triggered ER stress, increased ROS production, intracellular Ca²⁺ levels, and AVP mRNA levels.

Tmem117 inactivation progressively triggers cell death

Permanently elevated ER stress and increased levels of intracellular ROS and Ca²⁺ may trigger cell death (Zhivotovsky & Orrenius, 2011; Iurlaro & Muñoz‐Pinedo, ²⁰¹⁶; Giorgi et al, ²⁰¹⁸). Therefore, we investigated the fate of AVP neurons 1 month after Tmem117 inactivation in Tmem117 ^fl/fl mice coinjected in the SON with AAV6‐AVP‐iCre and AAV9‐hsyn‐DIO‐EGFP (AVPT ^M117KO mice). Two groups of control mice were prepared. One consisted of Tmem117 ^+/+ mice injected with both viruses (AVP ^TM117WT mice); these allowed to assess by immunofluorescence microscopy the total number of Tmem117 ⁺ cells and the total number of infected, EGFP‐expressing cells. The second group consisted of Tmem117 ^fl/fl mice injected only with AAV9‐hsyn‐DIO‐EGFP (AVP ^TM117FL mice), which were used as control for the effect of the genotype on the total number of Tmem117 ⁺ cells in the SON (Fig EV4A). Fluorescence microscopy images of EGFP and Tmem117 and of DAPI staining for the three groups are presented in Fig EV4B.

Figure EV4. Tmem117 inactivation in AVP neurons of the SON overtime leads to cell death.

1. Experimental scheme. AAV6‐AVP‐iCre was coinjected with AAV9‐DIO‐EGFP in the SON of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice. AAV9‐DIO‐EGFP alone was injected in the SON of Tmem117 ^fl/fl (AVP ^TM117FL ) to control for any genotype‐derived effect.
2. Tmem117 (magenta), EGFP (green), and DAPI nuclear staining (white) in the SON. The dashed line is delimiting the SON area.
3. Quantification of Tmem117 positive cells in the SON (n = 3–4 mice per group).
4. Quantitation of Tmem117 and EGFP double positive cells in the SON (n = 3 mice per group).
5. Quantitation of the number of cells (DAPI positive) in the SON (n = 3–4 mice per group).

Data information: opt: optic tract, SON: supraoptic nucleus, w: week. Scale bar = 100 μm. Data are represented as mean ± SEM. C, E: one‐way ANOVA with Tukey's post hoc test; D: unpaired t test; *P < 0.05, **P < 0.01, ***P < 0.001.

Quantitative analysis of the immunofluorescence data showed that the total number of Tmem117⁺ cells was the same in the two control groups but was reduced by ≥ 50% in AVP ^TM117KO mice as compared to the control mice (Fig EV4C). When considering only the EGFP⁺ and Tmem117⁺ double‐labeled cells, the reduction was even more striking, reaching ~75% (Fig EV4D), indicating that most of the Cre‐infected AVP neurons had disappeared during the postinfection period. This was also reflected by a reduction in the total number of neurons in the SON, as measured by DAPI staining (Fig EV4E).

To determine whether the loss of AVP neurons would suppress the oversecretion phenotype, we injected AAV6‐AVP‐iCre in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) and Tmem117 ^+/+ (AVP ^TM117WT ) male mice and, except from the 2‐week time point, we also assessed hormonal secretion upon insulin‐induced hypoglycemia at the later time point of 1 month. Then, at the end of the experiment, brains were collected for histological analysis (Fig 6A). Immunofluorescence microscopy analysis revealed a significant reduction in the number of AVP neurons in the PVN and SON of AVP ^TM117KO mice as compared to AVP ^TM117WT mice (Fig 6B–D). Insulin‐induced hypoglycemia reached the same level in both groups (Fig 6E), and induced a transiently enhanced secretion of CPP and GCG in AVP ^TM117KO mice that was evident only at the early time point (Fig 6F and G). Correlation analysis between plasma CPP and GCG levels for each sample revealed a strong positive correlation (Fig 6H), further supporting their causal relationship. Thus, inactivation of Tmem117 in AVP neurons led, over time, to cell death and the disappearance of the oversecretion phenotype.

Figure 6. Tmem117 inactivation in AVP neurons over time triggers neuronal death.

1. Experimental timeline. AAV6‐AVP‐iCre was injected in the posterior pituitary of Tmem117 ^fl/fl (AVP ^TM117KO ) or Tmem117 ^+/+ (AVP ^TM117WT ) mice.
2. Immunofluorescence microscopy detection of AVP in the PVN and the SON.
3. Quantitation of AVP‐positive cells in the PVN (n = 3–4 mice per group).
4. Quantitation of AVP‐positive cells in the SON (n = 3–4 mice per group).
5. Glycemia 1 h after saline (black) or insulin (red) i.p. injections in AVP ^TM117WT and AVP ^TM117KO male mice (n = 8–9 mice per group).
6. CPP plasma levels 1 h after saline or insulin injection in AVP ^TM117WT and AVP ^TM117KO male mice (n = 8 mice per group).
7. GCG plasma levels 1 h after saline or insulin injection in AVP ^TM117WT and AVP ^TM117KO male mice (n = 8–9 mice per group).
8. Correlation between the levels of plasma CPP and GCG in each sample (n = 24 samples per group; eight mice per group tested at three distinct time points).

Data information: 3V: third ventricle, AVP: vasopressin, d: day, INS: insulin, KO: AVP^TM117KO, SAL: saline, opt: optic tract, PVN: paraventricular nucleus, SON: supraoptic nucleus, WT: AVP^TM117WT. Scale bar = 100 μm. For panels C and D, lines correspond to the mean value per group and error bars represent ± SEM. For panels E–G, bars correspond to the mean value per group and error bars represent ± SEM. C,D: unpaired t test; E–G: two‐way ANOVA for each time point in comparison to SAL baseline with Bonferroni post hoc test; *P < 0.05, **P < 0.01, ***P < 0.001.

Source data are available online for this figure.

Discussion

In this study, we characterized the site of expression and functional role of Tmem117, a genetically controlled, hypothalamic regulator of insulin‐induced GCG secretion. We found that Tmem117 is expressed in AVP magnocellular neurons and Tmem117 inactivation increases hypoglycemia‐induced CPP and GCG secretion. This phenotype is observed in male mice and only in the proestrus phase in female mice. c‐Fos immunodetection showed that AVP neurons are activated by hypoglycemia and patch clamp analysis revealed that about half of them are GI neurons. Inactivation of Tmem117 did not affect the glucose responsiveness of AVP neurons, but instead increased ER stress, elevated intracellular ROS, and Ca²⁺ concentrations, leading to increased AVP production and secretion. These results highlight the physiological role of AVP neurons in the control of GCG secretion and identify Tmem117 as a novel regulator of their function.

Tmem117 is an eight transmembrane domain‐containing protein (Bürgi et al, ²⁰¹⁶). We found it to be present in AVP neurons of the PVN and SON, with a widespread intracellular localization, from the soma to the axons and in their terminals located in the posterior pituitary, the site of AVP secretion in the bloodstream. The level of Tmem117 mRNA expression in the hypothalamus of the BXD mouse lines used for the genetic screen showed strong and negative correlation with hypoglycemia‐induced GCG secretion (Picard et al, ²⁰²²) (Fig EV1D). Our physiological studies indicated that AVP neurons are activated by insulin‐induced hypoglycemia, leading to CPP secretion, a response that was augmented by Tmem117 inactivation and was accompanied by increased GCG secretion. These observations are in agreement with previous studies showing increased secretion of AVP in response to hypoglycemia (Baylis & Robertson, 1980; Baylis et al, ¹⁹⁸¹; Chiodera et al, ¹⁹⁹²) and that AVP stimulates GCG secretion through its binding to the AVP V1b receptor of pancreatic α cells (Dunning et al, ¹⁹⁸⁴; Spruce et al, ¹⁹⁸⁵; Gao et al, ¹⁹⁹²; Yibchok‐anun et al, ²⁰⁰⁴; Kim et al, ²⁰²¹; Liu et al, ²⁰²¹). On the other hand, these results are also in line with our genetic screen that showed a negative correlation between Tmem117 expression and insulin‐induced GCG secretion.

Interestingly, the effect of Tmem117 inactivation on GCG secretion is sex dependent; it is present in male mice, but only during the proestrus phase in female mice. Circulating levels of estradiol peak during proestrus phase and have been linked to adaptations in AVP magnocellular neurons and their afferent connections (Sladek & Somponpun, 2008), as well as to decreased GCG secretion from pancreatic alpha cells (Godsland, 2005; Mårtensson et al, ²⁰⁰⁹). Thus, our data suggest a possible contribution of estradiol to the AVP‐stimulated GCG release in conditions of hypoglycemia. Preceding studies have also reported that the CRR displays sex dimorphism (Steinbusch et al, ²⁰¹⁶; Briski et al, ²⁰¹⁷), though the precise mechanisms involved still need to be fully characterized. Given the complexity and redundancy of the CRR implicated pathways, this sex dimorphism is most probably a result of sex hormones acting at various sites and affecting different aspects of the response. Our data suggest that the action of estrogen on magnocellular AVP neurons is one of these aspects. AVP neurons express both nuclear and plasma membrane estrogen receptors that exert direct effects on intracellular signaling and AVP release (Sladek & Somponpun, 2008). Plasma concentration of AVP is decreased upon estradiol treatment in rodents (Peysner & Forsling, 1990) and during the follicular phase in women (Stachenfeld, 2008), suggesting a suppressing effect of estrogen on AVP secretion. Therefore, one could hypothesize that hypoglycemia‐induced AVP secretion would be decreased during estradiol‐rich phases of the estrus cycle. Our data appear in line with such a hypothesis, since plasma CPP levels during hypoglycemia tend to be lower in WT mice that are in the proestrus phase (Fig 2I, black dots, P = 0.13), but further studies focusing specifically on AVP secretion across the estrus cycle and its effect on CRR are required in order to get a detailed understanding of this interplay. What is clearly pointed out by our results is that inactivation of Tmem117 in AVP neurons of female mice leads to increased AVP secretion during the estradiol‐rich proestrus phase.

In order to investigate the mechanisms by which Tmem117 inactivation regulates CPP secretion in hypoglycemic conditions, we explored the glucose responsiveness of the AVP neurons. First, c‐Fos immunofluorescence analysis showed that insulin‐induced hypoglycemia led to an approximately threefold increase in the activation of AVP neurons in the SON, suggesting that they are glucose responsive. This was confirmed by patch clamp analysis, which revealed that ~50% of them are GI neurons activated by low extracellular glucose concentrations. Thus, AVP neurons can respond to decreased glucose levels. Interestingly, a recent study reported that hypoglycemia can also activate AVP neurons through afferent connections arising from GI neurons of the basolateral medulla (Kim et al, ²⁰²¹). Thus, AVP neurons are part of a brainstem hypothalamus neuronal circuit where hypoglycemia can be sensed by neurons located at multiple sites to activate the secretion of AVP leading to increased secretion of GCG.

Patch clamp and c‐Fos immunofluorescence analysis revealed that inactivation of Tmem117 does not affect the glucose responsiveness of AVP neurons. Instead, it induces ER stress and increases intracellular ROS and Ca²⁺ levels, leading to enhanced AVP production and secretion. These results extend previous observations made in the HCT116 cancer cells showing that Tmem117 silencing increases ER stress and ROS production (Tamaki et al, ²⁰¹⁷) and are in line with the reports that increased intracellular ROS stimulate AVP transcription in magnocellular neurons (St‐Louis et al, ^{2012, 2014}). It is striking that a small reduction in Tmem117 expression in the SW480 cell line (~30%) led to a significantly increased expression of ER stress markers. This suggests that Tmem117 expression level has an important regulatory role. This is compatible with the genetic screen that showed strong correlation across the BXD lines between the level of Tmem117 expression and hypoglycemia‐induced GCG secretion. Thus, fine‐tuning of Tmem117 expression appears as a physiological mean of controlling AVP neuron secretory capacity. This hypothesis is further supported by our data reporting decreased Tmem117 mRNA levels in the SON in response to insulin‐induced hypoglycemia. On the other hand, such a fine‐tuning mechanism could also be implicated in pathological conditions, where a maladaptive upregulation of Tmem117 expression would lead to decreased AVP secretion. In contrast, genetic inactivation of Tmem117, which leads to permanent increase in ER stress, ROS production, and elevated intracellular Ca²⁺ levels, is not compatible with the long‐term survival of these neurons. Indeed, 1 month after induction of Tmem117 recombination we observe neuronal death, probably as a result of apoptosis, as reported in the HCT116 cells with silencing of Tmem117 (Tamaki et al, ²⁰¹⁷).

Collectively, this study highlights the physiological role of the AVP neurons in the CRR and identifies Tmem117 as a genetic determinant of AVP and GCG secretion. Inactivation of Tmem117 did not affect the glucose‐sensing properties of AVP GI neurons, but led to increased CPP secretion associated with increased ER stress, intracellular ROS, and Ca²⁺ levels, and increased AVP mRNA expression. Thus, Tmem117 emerges as a so far uncharacterized regulator of AVP neuron secretory capacity, by increasing AVP mRNA expression and secretion in response to hypoglycemia. Defining how Tmem117 controls these intracellular processes at the molecular level will, however, require future studies. Finally, this study further demonstrates that the central response to hypoglycemia is highly complex, integrating not only brainstem and hypothalamic glucose responsive neurons that activate the HPA axis and both branches of the autonomic nervous system, but also the secretion in the blood of AVP by magnocellular neurons.

Materials and Methods

Experimental models and subject details

Mice

All animal care and experimental procedures were in accordance with the Swiss National Institutional Guidelines of Animal Experimentation (OExA; 455.163) with license approval (VD3363, VD3674) issued by the Vétérinaire Cantonal (Vaud, Switzerland). Mice were housed up to five per cage in a temperature‐controlled room with a 12‐h light/dark cycle and ad libitum access to water and standard laboratory chow (SAFE 150). For all the experiments, 8‐ to 10‐week‐old mice were used.

Generation of Tmem117 ^fl mice

Tmem117 ^fl mice were generated by homologous recombination in embryonic stem cells (genOway, Lyon, France). A reporter cassette for targeting exon 3, preceded by a splice acceptor, was inserted in antisense into the respective intron. Both the cassette and the targeted exon were flanked by loxP sites and mutated loxP sites to enable deletion monitoring using the FLEx approach (Schnütgen et al, ²⁰⁰³) (Fig EV1A). The derived mouse line was developed onto a C57BL/6N background.

Tmem117 ^fl;AVP‐IRES2‐Cre‐D mice

Tmem117 ^fl/fl;AVP‐IRES2‐Cre‐D^tg/+ mice were generated by crossing the Tmem117 ^fl line with the AVP‐IRES2‐Cre‐D (Jackson Laboratory code: 023530) for verifying the specificity of the Tmem117 antibody (Fig 1C and D).

Rosa26^tdTomato (Ai14) mice

Ai14 mice (Jackson Laboratory code: 007914) were used for verifying the specificity of the newly generated AAV6‐AVP‐iCre.

Cell lines

SW480 cells

The human colorectal cancer cell line SW480 (ATCC CCL‐228) was cultured in Leibovitz's L‐15 medium, supplemented with 10% fetal bovine serum (FBS). Cells were incubated at 37°C in a 0.3% CO₂ atmosphere and passaged twice per week up to 30 passages.

Method details

In vivo

Stereotaxic surgery/optic fiber cannula implantation

Mice were anesthetized with a ketamine and xylazine solution (100 mg/kg ketamine, 5 mg/kg xylazine injected i.p.) and were placed on a stereotaxic frame (Stoelting, Chicago, IL, USA). A small incision was made in the skin to reveal the skull. A small opening, to allow the passage of a 33‐gauge needle, was made and the viral solution was injected at a rate of 100 nl/min. The stereotaxic coordinates (relative to the bregma) and total volume injected for each area were: posterior pituitary: AP: −3, ML: 0, DV: −5.8, 0° angle/400 nl; SON: AP: −0.46, ML: ±1.15, DV: −5.45, 0° angle/140 nl per hemisphere. After the viral injection, the incision was closed by suturing and an analgesic solution was provided (0.1 mg/kg buprenorphine s.c.). Ophthalmic ointment was used throughout the procedure to prevent eye dryness. A heat pad was used to maintain the proper body temperature.

For fiber photometry recordings, an optic fiber (Doric Lenses B280‐2408‐6.4) was implanted in the posterior pituitary after the viral injection and the incision was closed with dental cement.

For insulin‐induced hypoglycemia experiments, 400 nL of AA6‐AVP‐iCre (8.1 × 10¹² [viral genomes] vg/ml) were injected in the posterior pituitary. For electrophysiological recordings (Fig 4) and viability assessment (Fig EV4), 140 nl of a 1:1 mix of AAV6‐AVP‐iCre (8.1 × 10¹² vg/ml) with AAV8‐syn‐DIO‐mCherry (2.2 × 10¹³ vg/ml) or AAV9‐syn‐DIO‐EGFP (2.4 × 10¹³ vg/ml) were injected per hemisphere in SON. For fiber photometry, 400 nl of a 1:1 mix of AAV6‐AVP‐iCre (8.1 × 10¹² vg/ml) with AAV9‐syn‐FLEX‐jGaMP7s (2.5 × 10¹³ vg/ml) were injected in the posterior pituitary.

Insulin‐induced hypoglycemia; blood collection

Mice were food deprived for 6 h (8 am–2 pm). At 12 pm, they were placed in individual cages and glycemia was measured. At 1 pm, glycemia was measured again and the mice were injected i.p. with either saline (baseline control) or insulin (0.8 U/kg). At 2 pm, glycemia was measured again and 100‐120 μl of blood were collected under isoflurane‐induced general anesthesia by submandibular vein incision.

Fiber photometry

Starting 5 days after surgery, mice were habituated to the recording cage and placement of the optic fiber in 30‐min sessions at least twice per week for 2 weeks. On the day of recording, the mice were placed in the recording cage for 30 min and signal recording was started. The optic fiber was then connected to capture the baseline and stimulated signal intensity. The fiber photometry system was from Doric Lenses using LED light sources for GCaMP (465 nm) and isosbestic (405 nm) measurements; signal was recorded and analyzed using the Doric Neuroscience Studio software. The GCaMP7 signal was first normalized over the isosbestic signal (i = i ₄₆₅−i ₄₀₅) and then over the mean signal for the whole trace (ΔF/F0). Each point was then normalized to the baseline signal to obtain the corresponding z score calculated using the formula: z _(i) = [ΔF/F0_(i) – median (ΔF/F0 _baseline)/MAD_baseline]. For baseline intensity traces, z‐score generation was calculated based on a 1‐s recording period before the cable was plugged onto the optic cannula. For insulin‐induced hypoglycemia traces, z‐score generation was calculated based on the baseline signal recorded over 10 min before the i.p. injection.

Estrus cycle monitoring

The estrus cycle phase was identified based on vaginal cytology samples after violet crystal staining (McLean et al, ²⁰¹²).

Tissue collection

For immunofluorescence microscopy analysis, mice were transcardially perfused with 10 ml ice‐cold phosphate‐buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄) followed by 40 ml ice‐cold paraformaldehyde (PFA, 4%) in PBS. Brain and pituitary were postfixed in 4% PFA O/N at 4°C and then kept for 24 h in 30% sucrose in PBS at 4°C. Tissues were frozen and stored at −80°C.

For SON and PVN DNA extraction, fresh brains were collected and kept in RNAlater solution (Thermo Fisher Scientific) at 4°C for 1 week. Then, 250‐μm‐thick vibratome sections were prepared and PVN and SON nuclei were dissected under a stereomicroscope. Microdissected tissue was frozen with dry ice and stored at −80°C, before DNA extraction.

Ex vivo

Electrophysiological recordings

Tmem117 ^fl/fl and Tmem117 ^+/+ mice were injected in the SON with an AAV6‐AVP‐iCre and an AAV8‐DIO‐mCherry. One to two weeks later, they were deeply anesthetized with isoflurane, decapitated, and their brain taken out and immediately placed in an ice‐cold high‐glucose artificial cerebrospinal fluid (ACSF) containing (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 1 MgCl₂, 2 CaCl₂, 26 NaHCO₃, and 10 glucose (300 ± 5 mOsm) equilibrated with 95% O₂/5% CO₂. Coronal sections (250 μm) containing SON were prepared using a vibratome (VT1000S, Leica) and immediately transferred to an oxygenated ACSF solution containing 2.5 mM glucose and maintained at 32°C for at least 1 h before starting the recordings. Experiments were performed using an upright epifluorescence microscope (BX51WI, Olympus, Japan) mounted on a motorized stage coupled to a micromanipulator (MPC‐325, Sutter Instrument, USA) and equipped with a mercury lamp and an Evolve EMCCD camera (Teledyne Photometrics Technology, USA) and appropriate filters allowing the visualization of mCherry‐expressing neurons.

Glucose responsiveness of AVP neurons was then assessed in the whole‐cell configuration using a MultiClamp 700B amplifier associated with a 1440A Digidata digitizer (Molecular Devices). Borosilicate glass pipettes (resistance: 2–5 MΩ.; GC150F‐7.5, Harvard Apparatus, USA) were prepared with a P‐97 horizontal micropipette puller (Sutter Instrument, USA). The patch pipettes were filled with an intrapipette solution containing (in mM): 130 K‐gluconate, 5 NaCl, 1 MgCl₂, 10 NaPhosphocreatinine, 10 HEPES, 0.2 EGTA, 4 MgATP, 0.5 Na₂GTP (pH 7.2–7.4; 280 ± 5 mOsm). Membrane potential and membrane resistance were monitored in current‐clamp mode in the presence of 2.5 mM or 0.1 mM glucose after a 10‐ to 15‐min baseline. Neurons with an access resistance > 25 MΩ or changing by > 20% during the recording were excluded from the analysis. Signals were filtered at 2 kHz, digitized at 10 kHz, and collected online with a pClamp 10 data acquisition system (Molecular Devices).

Cell culture

SW480 cells were plated in six‐well tissue culture plate at a density of 300,000 cells/well. After 48 h cells were transfected with siRNA against Tmem117 or siRNA control (30 pmol/well) with lipofectamine RNAiMAX (Thermo Fisher Scientific, Cat# 13778100). Forty‐eight hours after transfection, cells were trypsinized and pelleted by mild centrifugation (300 g for 3 min). Supernatant was removed and cell pellet was used for RNA and protein extraction using Qiagen RNeasy Mini kit.

Sample preparation/analysis

Immunofluorescence microscopy

Tissue sections (25‐μm thick) were washed with PBS for 5 min at RT, incubated with blocking solution (2% normal goat serum +0.3% Triton X‐100 in PBS) for 1 h at RT and then incubated O/N with primary antibodies (Guinea Pig anti‐(Arg8)‐Vasopressin, BMA Biomedicals, Cat# T‐5048; Rabbit anti‐Tmem117, Novus, Cat# NBP1‐94078) in blocking solution at 4°C. Next, sections were washed with PBS 3 times for 10 min, incubated with secondary antibodies (Goat anti‐Rabbit Alexa Fluor 488, Molecular Probes, Cat# A‐11078; Goat anti‐Guinea pig Cy5, Abcam, Cat# ab102372) in blocking solution for 2 h at RT, washed again with PBS 3 times for 10 min, incubated with DAPI (1:10,000 in PBS) for 20 min at RT, washed with PBS 3 times for 10 min, allowed to dry, and finally mounted using fluoromount (Sigma F4680).

For c‐Fos immunodetection tissue sections were washed with PBS for 5 min at RT, incubated with blocking solution (4% normal goat serum +0.3% Triton X‐100 in PBS) for 1 h at RT and then incubated O/N with primary antibody for c‐Fos (Rabbit anti‐cFos, Cell Signaling, Cat# 2250) in blocking solution at RT. Next, sections were washed with PBS 3 times for 10 min, incubated with secondary antibody in blocking solution for 3 h at RT, washed again with PBS 3 times for 10 min, and incubated O/N with primary antibody for AVP in blocking solution at 4°C. Next, sections were washed with PBS 3 times for 10 min, incubated with secondary antibody in blocking solution for 3 h at RT, washed again with PBS 3 times for 10 min, incubated with DAPI (1:10,000 in PBS) for 20 min at RT, washed with PBS 3 times for 10 min, allowed to dry, and finally mounted using fluoromount.

For all secondary antibodies the final concentration in the working solution was 1:500. For primary antibodies the concentrations used were: anti‐Tmem117 1:250, anti‐AVP 1:250, anti‐cFos 1:1,000.

In situ hybridization (RNAscope)

Tissue sections (25‐μm thick) containing the SON were costained by in situ hybridization for AVP mRNA (Cat# 401391) and BiP mRNA (Cat# 438831‐C3) using RNAscope probes and RNAscope Fluorescent Multiplex Detection Reagents (Advanced Cell Diagnostics, Newark, CA, USA) following manufacturer's instructions.

DHE fluorescent labeling

Tissue sections were washed with PBS for 5 min at RT, allowed to dry, and mounted using fluoromount (Sigma F4680).

Quantitative microscopy analysis

Fluorescence images were acquired on a ZEISS Axio Imager.M2 microscope, equipped with ApoTome.2, and a Camera Axiocam 702 mono (Zeiss, Germany). Specific filter cubes were used for the visualization of green (Filter set 38 HE eGFP shift free (E) EX BP 470/40, BS FT 495, EM BP 525/50), red (Filter set 43 HE Cy 3 shift free (E) EX BP 550/25, BS FT 570, EM BP 605/70), blue (Filter set 49 DAPI shift free (E) EX G 365, BS FT 395, EM BP 445/50) fluorescence, and far red (Filter set 50 Cy 5 shift free [E] EX BP 640/30, BS FT 660, EM BP 690/50). Different magnifications were selected using a Zeiss 20x objective (Objective Plan‐Apochromat 20x/0.8 M27, FWD = 0.55 mm) and a 40x oil‐immersion objective (Objective C Plan‐Apochromat 40x/1.4 Oil DIC M27 [FWD = 0.13 mm]). For super‐resolution microscopy the ZEISS ELYRA7 SIM² system was used with an 63x oil‐immersion objective.

Cell quantification in PVN and SON was performed on images derived from two consecutive hypothalamic sections for each mouse (bregma −0.7 and −0.8). Fluorescence intensity quantification was performed using the QuPath‐0.3.2 software on images derived from two consecutive hypothalamic sections for each mouse (bregma −0.7 and −0.8). Neuroanatomical landmarks (third ventricle, optic tract, fornix) and DAPI staining were used to define the region of interest and cell area was marked manually by a blinded experimenter.

Western blot

The protein content of the samples was quantified by Bradford assay. Total protein (25 μg) per sample were prepared in SDS‐PAGE sample buffer (2% SDS, 10% glycerol, 50 mM Tris, 0.1% bromophenol blue, 5% β‐mercaptoethanol) and separated on a 10% SDS‐PAGE gel. Proteins were transferred onto a nitrocellulose membrane, incubated with blocking solution (3% milk in PBS‐T [PBS + 0.1% Tween‐20]) for 1 h at RT, washed with PBS‐T 3 times for 10 min, and incubated O/N at 4°C with primary antibody (Rabbit anti‐Calnexin, Abcam, Cat# ab133615) at a dilution of 1:1,000 in PBS‐T. Then membranes were washed with PBS‐T 3 times for 10 min, incubated for 1 h at RT with horseradish peroxidase (HRP)‐conjugated secondary antibody (Anti‐Rabbit HRP, Amersham, Cat# NA934) at a dilution of 1:10,000 in blocking solution, washed again with PBS‐T 3 times for 10 min, incubated for 1 min with enhanced chemiluminescence (ECL, Amersham) buffer, and the HRP‐mediated chemiluminescence signal was detected/imaged using the Fusion FX6 Spectra imaging platform (VILBER). The derived images were analyzed for band intensity using ImageJ. The calnexin signal of each sample was normalized to the total protein (Pierce Reversible Protein Stain Kit). Data are presented as percent of the control.

DNA extraction/PCR

DNA was extracted from microdissected tissue using the Arcturus Picopure DNA extraction kit (Applied Biosystems, Ref # KIT0103). The extracted DNA was then subjected to PCR for detection of recombination using the following primers: 5′‐CTTTCTTCATAAAAAGCCGGAAGGCATTAC‐3′ (212), 5′‐GCCTGAAATATAAATATCGCAAGTGAGTGTGC‐3′ (213), 5’‐CAACTGACCTTGGGCAAGAACATAAAGTG‐3′ (216) (graphical representation Fig EV1A).

RNA extraction/real‐time PCR

Total RNA was extracted from SW480 cell pellets using the Qiagen RNeasy Mini kit based on manufacturer's instructions and from microdissected tissue using the Arcturus Picopure RNA extraction kit (Applied Biosystems, Ref # KIT0204). RNA was then reverse transcribed, and the derived cDNA was quantified using specific primers against Tmem117 (FW:5′‐TGTGATGCAGGACTGGGAAT‐3′; RV:5′‐TTGAACTGCATGTGAGGCGT‐3′), Xbp1 (FW:5′‐TGGCCGGGTCTGCTGAGTCCG‐3′; RV:5′‐ATCCATGGGGAGATGTTCTGG‐3′), usXbp1 (FW:5′‐CAGCACTCAGACTACGTGCA‐3′; RV:5′‐ATCCATGGGGAGATGTTCTGG‐3′), sXbp1 (FW:5′‐CTGAGTCCGAATCAGGTGCAG‐3′; RV:5′‐ATCCATGGGGAGATGTTCTGG‐3′), BiP (FW:5′‐TGTTCAACCAATTATCAGCAAACTC‐3′; RV:5′‐TAGGTGGTCCCCAAGTCGAT‐3′). Expression data were normalized to the housekeeping gene Gusb (FW: 5′‐CCACCAGGGACCATCCAAT‐3′; RV: 5′‐AGTCAAAATATGTGTTCTGGACAAAGTAA‐3′).

Quantification of circulating hormones

CPP and GCG levels in the plasma were quantified by Copeptin ELISA (MyBioSource, Cat# MBS2020621) and Glucagon ELISA (Mercodia, Cat# 10‐1281‐01), respectively, based on the manufacturer's instructions.

Constructs/cloning/AAV packaging

The p2.0VPI.iCre (AVP‐iCre) plasmid was generated by subcloning the iCre sequence (AgeI 2732nt, NotI 3813nt) from pCDH‐CB‐iCre plasmid (Addgene # 72257) in the p2.0VPI.EGFP backbone (AgeI 6237nt, NotI 6972nt; Addgene # 40868) (see Fig EV1). The packaging of the plasmid in the AAV6 was performed by the University of North Carolina (Chapel Hill, NC) vector core.

The TM117‐GFP plasmid was provided by the Van der Goot group (EPFL, Lausanne; Bürgi et al, ²⁰¹⁶).

eQTL analysis

eQTL mapping was performed using the R package R/qtl (Broman et al, ²⁰⁰³) with a genotype map from the BXD panel composed of GeneNetwork genotypes (www.genenetwork.org) merged with available RNA sequencing data from whole hypothalamic tissue (Picard et al, ²⁰¹⁶). eQTL interval mapping was calculated using the expected maximization algorithm, a 5% genotyping error rate, and pseudomarkers were generated every cM. eQTL location was obtained by 6.915 likelihood ratio statistics (LRS) support intervals. Significant eQTLs were determined for the trait using a 5% false discovery rate threshold estimated from 1,000 permutations.

Quantification and statistical analysis

All graphs and statistical analysis were generated using Prism software (GraphPad Prism 9), further details regarding sample size and the statistical analysis used in each case can be found in the corresponding figure legends. In brief, for experiments concerning comparisons between two groups on a single independent variable (Figs 2H–J, 3C, F, H, and I, 4GΔ and IΔ, 5A, B, D, and G, 6C and D, EV3A–D, and EV4D), we used unpaired two‐tailed t test. For experiments concerning comparisons between two dependent measurements (Fig 4G–J), we used paired two‐tailed t test. For those concerning comparisons among three groups on a single independent variable (Figs 5K and EV4C and E), we used one‐way ANOVA with Tukey's post hoc test. For two‐factor designs concerning repeated measures (Figs 2B–G, 5I and 6E–G), we used two‐way ANOVA RM with Bonferroni post hoc test. For two‐factor designs concerning nonrepeated measures (Fig 4C), we used two‐way ANOVA with Tukey's post hoc test. For the comparison of neuronal subpopulation proportion (Fig 4E), we used Fisher's exact test. In all graphs, error bars are depicting ± SEM.

Author contributions

Sevasti Gaspari: Conceptualization; formal analysis; investigation; visualization; methodology; writing – original draft; project administration; writing – review and editing. Gwenaël Labouèbe: Formal analysis; investigation; visualization; methodology; writing – review and editing. Alexandre Picard: Formal analysis; writing – review and editing. Xavier Berney: Formal analysis; methodology. Ana Rodriguez Sanchez‐Archidona: Software; visualization; writing – review and editing. Bernard Thorens: Conceptualization; resources; supervision; funding acquisition; visualization; writing – original draft; project administration; writing – review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

PDF+

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

EMBO reports (2023) 24: e57344

Data availability

No primary datasets have been generated and deposited.