Adipsic hypernatremia without hypothalamic lesions accompanied by autoantibodies to subfornical organ

Abstract

Adipsic (or essential) hypernatremia is a rare hypernatremia caused by a deficiency in thirst regulation and vasopressin release. In 2010, we reported a case in which autoantibodies targeting the sensory circumventricular organs (sCVOs) caused adipsic hypernatremia without hypothalamic structural lesions demonstrable by magnetic resonance imaging (MRI); sCVOs include the subfornical organ (SFO) and organum vasculosum of the lamina terminalis (OVLT), which are centers for the monitoring of body‐fluid conditions and the control of water and salt intakes, and harbor neurons innervating hypothalamic nuclei for vasopressin release. We herein report three newly identified patients (3‐ to 8‐year‐old girls on the first visit) with similar symptoms. The common features of the patients were extensive hypernatremia without any sensation of thirst and defects in vasopressin response to serum hypertonicity. Despite these features, we could not detect any hypothalamic structural lesions by MRI. Immunohistochemical analyses using the sera of the three patients revealed that antibodies specifically reactive to the mouse SFO were present in the sera of all cases; in one case, the antibodies also reacted with the mouse OVLT. The immunoglobulin (Ig) fraction of serum obtained from one patient was intravenously injected into wild‐type mice to determine whether the mice developed similar symptoms. Mice injected with a patient's Ig showed abnormalities in water/salt intake, vasopressin release, and diuresis, which resultantly developed hypernatremia. Prominent cell death and infiltration of reactive microglia was observed in the SFO of these mice. Thus, autoimmune destruction of the SFO may be the cause of the adipsic hypernatremia. This study provides a possible explanation for the pathogenesis of adipsic hypernatremia without demonstrable hypothalamus‐pituitary lesions.

Introduction

Sodium (Na) and water homeostasis are critical to sustain life, and Na⁺ levels in body fluids are strictly controlled by the oral intake and urinary excretion of salt/water 9. Na⁺ levels and the osmolality of body fluids are constantly monitored by specific sensors in the brain to control thirst sensation, appetite for salt, and antidiuretic hormone (ADH; vasopressin) secretion 1. Impairments in these mechanisms result in hypernatremia, which is defined as an increase in plasma Na⁺ concentrations to a value exceeding physiological levels (135–145 mM) 24.

Adipsic hypernatremia is clinically characterized by deficiency in the vasopressin release and thirst perception, leading to persistent hypernatremia with a euvolemic state 2. Most cases are accompanied by structural abnormalities in the hypothalamic‐pituitary area, mainly because of congenital diseases, tumors, or inflammation. However, some cases with no structural lesions have been reported since the 1970s and the pathogenetic mechanism remained to be elucidated 3, 4, 5, 7, 11, 26.

We previously described a 6‐year‐old girl who developed adipsic hypernatremia without demonstrable hypothalamic structural lesions 14. The patient did not have any thirst despite severe hypernatremia (199 mM on admission). Normal increases in the secretion of vasopressin in response to serum hyperosmolality were absent. We did not detect any structural anomaly within the brain of the patient by magnetic resonance imaging (MRI), with a special focus on the hypothalamus and pituitary gland. Subsequently, we found autoantibodies to Na_x, a Na channel, in her serum. Na_x is expressed in glial cells (ependymal cells and astrocytes) in the subfornical organ (SFO) and organum vasculosum of the lamina terminalis (OVLT) in the brain 21, 22, 29, 30. Consistently, the immunostaining of mouse brain sections showed that the patient's serum contained antibodies that were specifically reactive to the two regions 14.

SFO and OVLT belong to sCVOs, brain regions exceptionally devoid of the blood–brain barrier (BBB), and are involved in monitoring body‐fluid conditions. Among them, the SFO is especially the center for monitoring Na⁺ levels in body fluids to control salt‐intake behavior 15 and have projections to the supraoptic nucleus (SON) and paraventricular nucleus (PVN), which are responsible for the production of vasopressin 19, 20, 23. This suggested us a pathogenetic role of the autoantibodies in the development of the hypernatremia. We intravenously injected the immunoglobulin (Ig) fraction obtained from the patient's serum into wild‐type mice, and found that it induced abnormalities in water/salt intake, vasopressin release, and diuresis, leading to hypernatremia. This may have been due to complement‐mediated cell death in sCVOs, because deposits of the injected IgG proteins and C3 component of the complement system were specifically observed there.

At that time, we did not know whether it was an exceptional case or the autoimmunity to the sCVOs generally cause adipsic hypernatremia. In this study, we report three newly identified pediatric patients who were diagnosed with adipsic hypernatremia without demonstrable hypothalamic structural lesions in common. We found that their sera all contained antibodies reactive to the mouse SFO. The antibodies in one of the three cases also reacted with the OVLT. Wild‐type mice injected with the Ig fraction of a patient showed abnormalities in water/salt intake, vasopressin release, and diuresis, leading to hypernatremia, as observed in our previous case mentioned above 14. In the brains of these mice, cell death was observed in the SFO. These results provide supporting evidence for our claim that autoimmunity to the SFO generally causes adipsic hypernatremia and that autoimmunity to the OVLT is not essential for the development of the hypernatremia.

Patient and Methods

Authorization for human subject research

Informed consent was obtained from all patients and/or their parents. The study protocol was approved by the Ethics Committee for Human Research of National Institute for Basic Biology, the Human Studies Ethical Committee for Epidemiology of Hiroshima University, the Institutional Review Board of Kawasaki Municipal Hospital, and the Ethical Committee of Hirosaki University Graduate School of Medicine.

Experimental animals

All experiments with animals were performed according to the guidelines of the National Institute for Basic Biology and Hiroshima University. Male mice (C57BL/6J, CLEA Japan, Tokyo, Japan) were maintained in plastic cages under a constant room temperature (23°C) with a 12‐h light/dark cycle with ad libitum access to water and food, and were used in experiments at 8–16 weeks of age.

Fluid deprivation test and vasopressin loading test

For the fluid deprivation test, the patient was instructed not to eat after dinner (from 20:00). On the next day, urine and plasma samples were obtained hourly from 9:00 to14:00. For the vasopressin loading test, 4 μg of l‐deamino‐8‐d‐arginine‐vasopressin (dDAVP, Sanofi, Paris, France) was administered percutaneously. Urine and plasma samples were obtained hourly for 3 h after the dDAVP administration.

Immunohistochemistry

The procedure for immunohistochemistry of mice brain was performed as previously reported 14. The coronal mouse brain sections were incubated with sera from the patients or a healthy control subject (1:500), or rabbit anti‐P2Y₁₂ antibody (1:1000) 12 in the blocking buffer for 2 days at 4°C. The sections were washed and incubated with Alexa488‐conjugated anti‐human IgG antibody (1:500, Thermo Fisher Scientific, Waltham, MA, USA) or Alexa488‐conjugated anti‐rabbit IgG antibody (1:500, Thermo Fisher Scientific), overnight at 4°C.

Western blot analysis

We tested the antibody binding to Na_x, TRPV1 and TRPV4 as previously reported 14. The C6 cell line (C6M16 and C6H17), in which the expression of mouse and human Na_x is, respectively, inducible under the control of the tetracycline‐responsive element, was reported previously 14, 28. To prepare human TRPV1‐ or TRPV4‐expressing cells, an expression vector containing cDNA for human TRPV1 (provided by Dr. Yasuo Mori, Kyoto University), or for TRPV4 18 was introduced into HEK293 cells. Preparation of cellular membrane proteins from the C6 or HEK293 cells was performed as previously described 28. The blotted membrane was probed with sera of the patients or a healthy control subject (1:500), followed with horseradish peroxidase (HRP)‐conjugated sheep antibody to human IgG (1:5000, GE Healthcare, Little Chalfont, UK). As controls, we also performed Western blotting with rabbit anti‐Na_x antibody (1:5000; 15, rabbit anti‐TRPV1 antibody (1:200, Alomone Labs, Jerusalem, Israel), or goat anti‐TRPV4 antibody (1:200, Santa Cruz Biotechnology, Dallas, TX, USA) as primary antibodies, followed by detection with corresponding HRP‐conjugated secondary antibodies.

Immunoaffinity isolation of antigens

SFO tissues were dissected from coronal slices of mouse brains under stereoscopic microscope. The tissues were lysed with lysis buffer (1% NP40 and 150 mM NaCl in 10 mM Tris‐HCl, pH 7.4) containing proteinase inhibitors (Roche Diagnostics, Basel Switzerland), and the supernatant was collected by centrifugation. Immunoglobulins (20 µg of proteins) prepared from human sera by ammonium sulfate precipitation procedure were mixed with protein‐G‐conjugated magnet beads suspension (100 µL, Immunoprecipitation kit Dynabeads Protein G, Thermo Fisher Scientific), and binding antibodies were covalently cross‐linked using disuccinimidyl suberate (DSS, Thermo Fisher Scientific), a homobifunctional reagent. The beads (20 µL of the original volume of the beads suspension) were incubated with 1 mg protein of SFO extracts overnight at 4°C. Once the beads were washed, the bound proteins were eluted with the elution buffer provided by the manufacture (Thermo Fisher Scientific). The eluted proteins were separated on a 15% polyacrylamide gel by tricine‐SDS‐PAGE, followed by silver staining.

Passive transfer of the Ig fraction into mice and food preference test

The Ig fraction was collected from the serum by precipitation with 33% saturated ammonium sulfate followed by dialysis against normal saline. The serum from the patient (Case 1) or a healthy control subject was intravenously injected into mice 2.5 mg Ig in 200 μL per 30 g body weight. Ten mice were used for each experiment. At one month after the injection day, the water‐restricted condition (0.5 mL/day) was started for 3 days. The food weight was measured at 5:00 p.m. every day. We could not examine the Ig from the other cases, because we could not obtain their agreements to use serum for animal experiments.

Measurement of plasma Na and vasopressin levels of mice

Na concentration in mouse blood was measured with a blood analyzer Fuji Drychem 7000 (Fujifilm, Tokyo, Japan) or i‐STAT (Abbott Point of Care , Princeton, NJ, USA). Plasma vasopressin concentrations of mice were determined by radioimmunoassay (RIA) after extraction with acetone and diethyl ether, according to the manufacturer's instructions (AVP RIA, Peninsula Laboratories, San Carlos, CA, USA).

Measurement of urine volume of mice

Urine volume was measured as previously described 14 with metabolic cages (KN‐645, Natsume Seisakusho, Tokyo, Japan).

Terminal deoxynucleotidyl transferase‐mediated dUTP nick‐end labeling

Terminal deoxynucleotidyl transferase‐mediated dUTP nick‐end labeling (TUNEL) was carried out using an ApopTag Fluorescein In Situ Apoptosis Detection Kit (Millipore, Massachusetts, MA, USA) as follows. Coronal brain sections (16‐μm thick) were prepared as described above.

Statistical analyses

All values are expressed as the mean ± SE. Data were tested for significance with KyPlot software (KyensLab, Tokyo, Japan). The statistical significance was estimated by the two‐tailed t‐test. A value of P < 0.05 was considered to be significant.

Results

Patients with hypernatremia

Case 1: The patient was a 9‐year‐old Asian girl (second‐born child) with hypernatremia who had no family history of hormonal or immunological diseases, and had a birth weight of 2590 g (−1.0 SD) at 38 weeks of gestation. At age 3, she was admitted to Hiroshima University Hospital for further examinations for adipsic hypernatremia and obesity, which had been progressing from the age of 1 year and 7 months old. She began to become obese shortly after she had influenza A. A blood examination showed hypernatremia (Na, 155 mM) and high serum osmolality without a sense of thirst. Her blood pressure was slightly elevated, but unremarkable (116/52 mmHg). An endocrinological examination revealed the low response of vasopressin to high serum osmolality (Figure 1A and Supporting Information Table S1).

Figure 1.

Patients showed impaired vasopressin release without demonstrable hypothalamic structural lesions. A. Relationship between serum osmolality and plasma vasopressin levels. The green area indicates the normal range (40). B. Sagittal T1‐weighted MRI.

In the water restriction test, urine osmolality remained abnormally low (under 150 mOsm) after 2.5 h, while blood osmolality was greater than 300 mOsm, which was consistent with the defect in vasopressin release (data not shown). The percutaneous administration of desmopressin (dDAVP; 4 µg/dose) increased urinary osmolality from 150 to 620 mOsm in 2.5 h, suggesting normal renal sensitivity to dDAVP. We administered dDAVP as a treatment for hypernatremia. Abdominal X‐ray computed tomography (CT) did not show any abnormal structural findings. It is important to note that no structural lesions were found in the hypothalamus or pituitary gland by MRI (Figure 1B).

Case 2: The patient was a 16‐year‐old Asian girl (first‐born child) with no family history of hormonal or immunological diseases and a birth weight of 2496 g (−1.2 SD) at 38 weeks of gestation. She had medical histories of nystagmus and truncal ataxia associated with a high fever, from which she recovered with the intravenous administrations of glucocorticoid drugs at the ages of 1 year and 3 months, 1 year and 10 months, 2 years and 2 months, and 5 years. She was diagnosed with opsoclonus myoclonus syndrome (OMS). In children, this syndrome is often associated with neuroblastic tumors and assumed to be of an autoimmune origin 13. However, no neoplasia was detected.

She visited the Kawasaki Municipal Hospital for an examination at age 4. Blood examinations at the first visit (Supporting Information Table S1, Case 2) revealed hypernatremia (Na, 165 mM) and high osmolality without a sense of thirst, and the low response of vasopressin to serum osmolality (Figure 1A). Correspondingly, plasma renin activity (PRA) levels were high.

When she was 7 years and 10 months old, paralysis of the lower extremities developed for one month. An electrophysiological study suggested polyneuritis; however, no structural lesions were found in the hypothalamo‐pituitary region by MRI (Figure 1B). DDAVP was administered (5 μg/day) as a treatment for hypernatremia.

Case 3: The patient was an 8‐year‐old girl (second‐born child) who had been morbidly obese since the age of 4. Nocturnal enuresis began at age 6, which was ameliorated with the oral use of dDAVP (60 µg/day). Pubic hair appeared when she was 7 years and 8 months old. She visited Hirosaki University Hospital at the age of 8 years and 4 months for further examinations. Hypernatremia (161 mM) and plasma hyperosmolality (329 mOsm/kgH₂O) without a sense of thirst were noted at the first visit (Supporting Information Table S1).

Despite the continuation of dDAVP, her serum Na levels remained extremely high, ranging from 154 to 169 mM, without any sense of thirst. Her plasma vasopressin level was low (2.6 pg/mL under the oral administration of dDAVP 60 μg/day) in light of severe hyperosmolality (Figure 1A). Her therapeutic measures included oral dDAVP titration along with mandatory water intake (1.5 L/day) to achieve normonatremia, MRI revealed no structural abnormality in the hypothalamus or pituitary gland (Figure 1B).

Patients shared common features

A normal increase in vasopressin secretion in response to serum hyperosmolality was commonly deficient in all cases (Figure 1A). The impaired secretion of vasopressin associated with adipsia was considered to be a direct cause of persistent hypernatremia, which led us to diagnose patients with adipsic hypernatremia. Most of the previously reported cases of adipsic hypernatremia had organic intracranial defects that may have caused hypothalamic dysfunction; however, our cases were all free of those lesions. The posterior pituitary gland was clearly identifiable on MRI, with the local presence of secretory granules 10, suggesting that vasopressin synthesis was preserved in our three cases (Figure 1B). Of note, the urine chemical tests revealed that urine‐concentrating capacities were intact in all cases (Supporting Information Table S1).

Patient sera contained antibodies reactive to the SFO

Since the parents and siblings of our patients had no histories of hypernatremia, it was unlikely that the cause was genetic. We investigated whether sCVOs were recognized by patient sera. When mouse brain sections were used for immunostaining, the sera of the three patients commonly and specifically stained the SFO (Figure 2). The OVLT was also stained with the serum of Case 3. No signals were observed in the other brain loci, including the PVN, SON, median eminence, and pituitary gland, which are also relevant to body‐fluid homeostasis. Control experiments using the serum of a healthy subject exhibited no detectable staining (Figure 2, Control; see also ref. 14).

Figure 2.

Sera of patients contained antibodies specifically reactive to mouse sCVOs. Immunostaining of the coronal brain sections of mice with the sera of patients and a control healthy subject. All sera examined were reactive to the SFO. The serum of Case 3 was also reactive to the OVLT. SFO, subfornical organ; OVLT, organum vasculosum laminae terminalis; PVN, paraventricular nucleus; SON, supraoptic nucleus; ME, median eminence. Scale bars, 100 μm.

Mice injected with the Ig fraction of a patient showed symptoms of hypernatremia

We examined the pathophysiological effects of autoantibodies to the SFO by an intravenous injection of the Ig fraction obtained from Case 1 into mice. We did not use the Ig from the other patients because we were unable to obtain agreements to use their sera for animal experiments. We examined the effects of the patient's Ig on salt‐intake behavior 30 days after the injection. Treated mice were allowed free access to Na‐depleted and Na‐repleted (normal) food under the condition that the amount of drinking water was restricted to 0.5 mL/day (see Figure 3A,B).

Figure 3.

Mice injected with patient Ig showed an abnormal NaCl preference under water‐restricted conditions. A. Strategy of the food preference test. Mice were injected with Ig obtained from a patient (Case 1) or healthy control subjects 30 days before the test (day −30) and housed under usual conditions (day −30 to day 0). Na‐depleted and Na‐repleted food were served from days 0 to 5, and mice had ad libitum access. The amount of drinking water was restricted to 0.5 mL/day from days 0 to 3. B. Preference ratio for Na‐repleted food. **P < 0.01 by the two‐tailed t‐test (n = 10 for each group). C. Plasma Na⁺ levels ([Na⁺]) were monitored at the indicated time points, as shown in A. *P < 0.05 and **P < 0.01 by the two‐tailed t‐test (n = 10 for each group). D–F. Water intake volumes during the dark period (12 h) (D), plasma vasopressin levels (E), and urine volumes (F) of mice that received the passive transfer of the patient or control Ig. *P < 0.05 and **P < 0.01 by the two‐tailed t‐test (n = 10 for each group).

Prior to this water restriction, mice did not show a specific preference for normal food or Na‐depleted food (data not shown), and plasma Na⁺ levels in both groups were similar (Figure 3C, i). During water restriction, mice injected with control Ig showed an aversion to Na‐repleted food (Figure 3B, Control Ig), whereas those injected the patient's Ig did not show such an aversive response (Figure 3B, Patient Ig). After water had been restricted for 3 days, the plasma Na⁺ levels of mice injected with the patient's Ig was more than 160 mM, which was significantly higher than that in mice injected with control Ig (Figure 3C, ii). On repletion with a sufficient amount of water, the aversion to Na‐repleted food of mice injected with control Ig disappeared and the plasma Na⁺ levels recovered to the normal level within 2 days (Figure 3B, days 4 and 5). In contrast, Na⁺ levels in the plasma of mice injected with the patient's Ig were maintained at significantly higher levels (Figure 3C, iii). Furthermore, mice injected with the patient's Ig consumed significantly less water than those injected with control Ig under non‐feeding conditions (Figure 3D).

Since the patient showed an abnormal reduction in the vasopressin response (Figure 1A), plasma vasopressin levels in the treated mice were examined under hydrated (water‐satiated) and dehydrated conditions. No significant differences were observed under the hydrated conditions (Figure 3E, Hydrated). After 24 h of dehydration, vasopressin levels in mice injected with the patient's Ig were only slightly elevated, but were still markedly lower than those of mice injected with control Ig (Figure 3E, Dehydrated). The amount of urine in mice injected with the patient's Ig was significantly larger than that of mice injected with control Ig during the dehydration period (Figure 3F).

Injection of patient's Ig‐induced cell death and the infiltration of inflammatory cells in the SFO

Considering the half‐life (a few days) of Ig, some irreversible destruction appears to underlie the defects observed in mice injected with the patient's Ig 30 days after the injection. Therefore, we determined whether the antibodies of Case 1 induced cell death and subsequent inflammatory responses, as described in the previous case 14. We performed terminal deoxynucleotidyl transferase‐mediated dUTP nick‐end labeling (TUNEL) assays, which label cells undergoing apoptosis and some forms of non‐apoptotic cell death 6. TUNEL‐positive cells were already detected in the SFO of mice 3 days after being injected with the patient's Ig, but not in the controls (Figure 4A,B). We also detected immunopositive signals for P2Y₁₂ receptors, a specific marker for reactive microglia 25, on the same day in the SFO, but not in the OVLT, PVN, SON, or pituitary gland (Figure 4C). No signals were observed in mice injected with control Ig.

Figure 4.

Infiltrates of reactive microglia and cell death were observed in the SFO of mice that received the passive transfer of a patient's Ig. A. TUNEL staining of tissue sections of the SFO, OVLT, PVN, SON, and pituitary gland from mice, 3 days after the injection of control (uppers) or patient Ig (lowers). Magnified view of the squared region in the SFO is shown below. Arrow heads indicate TUNEL‐positive cells. Scale bars, 100 μm. B. Summary of TUNEL assays. **P < 0.01 (significantly different from control Ig in the same brain region), by a two‐tailed t‐test (n = 10 for each group). C. Anti‐P2Y₁₂ immunostaining of the SFO, OVLT, PVN, SON, and pituitary gland of mice 3 days after the injection of control Ig (upper panels) or patient Ig (lower panels). Arrow heads indicate P2Y₁₂‐positive microglia. Scale bars, 100 μm.

Discussion

We herein report three pediatric patients with adipsic hypernatremia and impaired vasopressin release. The binding of their serum Ig to the mouse SFO (and OVLT in Case 3) suggested that autoantibodies cross‐reacted with corresponding antigen(s) in mice. The passive transfer of Ig obtained from the patient (Case 1) to mice led to the excessive ingestion of salt, even under water‐restricted conditions, along with the insufficient intake of water. In these mice, plasma Na⁺ levels became significantly higher than in those injected with control Ig. More importantly, the vasopressin response was defective. Taken together with the presence of cell death and infiltration of reactive microglia in the SFO, complement‐mediated cell death might occur in the SFO. Thus, the autoimmune destruction of the SFO may be a common entity of the pathogenesis. Since the SFO is the center for water/salt intake and has projections to the nuclei responsible for vasopressin production 14, 15, 19, the autoimmune destruction of the SFO may have caused adipsia, impaired salt‐intake control, and low vasopressin release, thereby leading to hypernatremia.

No obvious structural anomalies were detected by MRI in the hypothalamus in any of our cases. This was similar to our previously reported case 14. Therefore we diagnosed our patients with adipsic hypernatremia without demonstrable hypothalamic structural lesions 11. Patients reported so far with this disorder 3, 4, 5, 7, 11, 26 including ours 14 were all less than 13 years of age. Of note, Case 2 showed ataxia since the age of 1, and was diagnosed with OMS. In a few rare patients, the syndrome appeared to be caused by infections with identifiable pathogens such as Mycoplasma pneumoniae or hepatitis C virus 8, 17. This may have triggered the first autoimmunity response to the cerebellum and induced nystagmus and truncal ataxia in Case 2. In this case, we were unable to determine the exact onset of hypernatremia of this patient. We speculate that autoantibodies targeting the SFO were produced secondarily and antigen molecule(s) differed from those of OMS.

In our previous case 14, the patient had a solid tumor adjacent to the adrenal gland with a pathological diagnosis of ganglioneuroma, which was mainly composed of Na_x‐positive Schwann‐like cells, and this neoplasia likely evoked the production of autoantibodies to Na_x. However, none of the present three patients had any tumors; the exact etiologies of the autoimmune responses in the present three cases currently remain unknown. Case 1 was infected with influenza A when she was a toddler. In Case 2, some occult infection may have resultantly induced OMS because the initial development of nystagmus appeared after a high fever. Even in many patients with OMS with underlying neuroblastoma, infectious episodes, often of a viral origin, have been reported before the onset of symptoms 13. Case 3 may have also had an occult infection that her parents do not remember. Such infectious events at the early infant stages may trigger autoimmunity to some brain regions, especially to sCVOs in which the BBB is absent. This suggests the existence of a novel group of adipsic hypernatremia, caused by autoimmune processes without a paraneoplastic etiology. Although the target proteins of these autoantibodies have not yet been identified, the proteins selectively expressed in the SFO and OVLT may be plausible candidates. Trials to reduce or eliminate autoantibodies in our patients deserve consideration in the future.

We attempted to identify specific antigens of the autoantibodies in the three patients but all attempts failed. We first determined whether patient sera contained antibodies to Na_x by using C6H17 and C6M16 (C6 rat glioblastoma cell lines) 14, 28, in which the expression of human or mouse Na_x channels is inducible under the control of a tetracycline‐responsive element, respectively. The membrane extracts of these cells were subjected to a Western blot analysis; however, Na_x was not detected in any serum samples (Supporting Information Figure S1A). Transient potential receptor vanilloid 1 (TRPV1) and TRPV4, which have been suggested to play roles in the osmosensing mechanisms of body fluids in the brain 18, 27, were also negative (Supporting Information Figures S1B and S1C). We next attempted to identify specific antigens expressed in SFO tissues by Western blotting using crude extracts of the SFO and cortex. However, we did not detect protein bands specific for the SFO (data not shown). However, we cannot completely exclude the possibility that very small amount of antibodies against Na_x, TRPV1, or TRPV4 are present in the respective patient's sera by these experiments. So, we finally prepared affinity beads coated with the antibodies of our patients and healthy subjects, and attempted to identify antigen proteins specific for respective patients in the immunoprecipitate from the SFO extract. However, this attempt was again unsuccessful (data not shown). These results suggest that the content of antigen molecules is not abundant in the SFO.

In summary, we herein report 3 novel cases of adipsic hypernatremia, the sera of which commonly contained autoantibodies targeting the SFO. These autoantibodies may have induced persistent tissue damage within the SFO through complement activation, although only Case 1 was verified experimentally by passive transfer. Some tissue damage in the SFO, which is the center of controlling thirst/salt appetite and have neural efferents to several hypothalamic nuclei including PVN and SON, may have caused adipsic hypernatremia without detectable hypothalamic lesions. These results also provide new evidence for the presence of a distinct group of adipsic hypernatremia without demonstrable hypothalamus‐pituitary lesions, in which autoantibodies binding to the SFO are produced by unknown etiologies other than paraneoplasia. The immunostaining of SFO tissues using patient serum needs to be considered for a diagnosis of this type of adipsic hypernatremia.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher's web‐site.

Figure S1. Sera of patients did not contain antibodies against Na_x, TRPV1, or TRPV4. A. Western blot analyses of membrane extracts of C6 cells with (+) or without (–) the expression of human or mouse Na_x using sera from patients or a control subject. The rightmost panel shows positive control staining using an anti‐Na_x antibody. B, C. Western blot analyses of membrane extracts of HEK293 cells expressing (+) human TRPV1 (B) or TRPV4 (C) and of control cells not expressing (–) using the sera of patients or a control subject. The rightmost panels show positive controls using anti‐TRPV1 (B) and anti‐TRPV4 (C) antibodies, respectively.

Table S1. Values in clinical tests on the first visit to the hospital.