Introduction

Aquaporins (AQPs) are a family of evolutionarily conserved membrane proteins that facilitate the transport mainly of water but also of some other small molecules across the cell membrane1,2 and between organelles.3,4 Multiple (patho)physiologic functions have been attributed to AQPs. These include the urine-concentrating mechanism, epithelial fluid secretion, cell migration, mediation of cerebral edema, and neural signal transduction (reviewed in ref. 5). AQPs have six transmembrane domains, the NH2- and COOH-terminal domains being in the cytoplasm. The channel pore is made of two highly conserved short hydrophobic regions with a typical asparagine–proline–alanine (NPA) motif.6,7 To date, 13 AQP family members (AQP0–12) have been identified in various mammalian tissues.2 AQPs are classified, according to their sequence homology and permeability, into three subfamilies: (i) the water-specific “classical” AQPs, (ii) the aquaglyceroporins, and (iii) the unorthodox AQPs.4,8,9,10,11 The aquaglyceroporin subfamily transports glycerol in addition to water because of the presence of an aspartic residue near the second NPA box, resulting in expansion of the pore to accept a larger molecule such as glycerol. The four subtypes of the aquaglyceroporin subfamily are: AQP3, AQP7, AQP9, and AQP10 (reviewed in refs. 9,10). Unorthodox AQPs are highly similar to the other AQP subclasses, but additional three-dimensional structure analyses are required for a better understanding of the pore structure of this subclass.7

Aquaporin 7 (AQP7; MIM602974) was previously referred to as “aquaporin adipose” because it was originally identified in human adipose tissue.12 Subsequently, other researchers found AQP7 expression in testis and renal proximal tubule cells.13,14 The various AQP7 mouse models that were developed (for a phenotype overview, see Supplementary Table S1 online) showed a role for AQP7 in controlling fat and in glucose metabolism.15,16,17 AQP7-knockout mice were shown to develop obesity and insulin resistance, with excess glycerol concentrations in the adipocytes. In addition, it was shown that these mice have reduced water permeability in the renal proximal straight tubule brush border membrane and a marked elevation of glycerol in the urine, possibly indicating the presence of a novel glycerol reabsorption pathway in the proximal straight tubules.18,19

In humans, however, the pathophysiologic role of AQP7 has been less obvious. In a Japanese cohort, three AQP7 missense mutations (R12C, V59L, and G264V) were detected that did not show any association with obesity or with type 2 diabetes.20 Of note, the G264V (rs62542743) AQP7 mutation was shown (in Xenopus oocytes) to inhibit transport of water and glycerol. The only subject who was homozygous for G264V in the Japanese study had no obesity, no diabetes, and no loss of fertility. Another study found the G264V AQP7 heterozygous variant in 8% of a Caucasian population (in contrast to 3.75% in the Japanese population), but again no association was found with obesity or diabetes.21 This later study showed that AQP7 was downregulated in subcutaneous adipose tissue of women with severe obesity but without type 2 diabetes. It was recently suggested that AQP3 and AQP9 represent additional pathways for transport of glycerol in human adipocytes and that these AQPs are therefore potentially responsible for AQP7 redundancy.22

Here we present the cases of three children from unrelated families. All three children had psychomotor retardation of unknown etiology, pronounced renal glycerol loss (hyperglyceroluria), and a subclinical, platelet-dense granule secretion defect. The homozygous AQP7 G264V mutation was detected in these children. The relevance of the mutation in relation to their clinical phenotype, and the role and subcellular localization of AQP7 in platelets were studied.

Materials and Methods

Glycerol measurements in blood plasma and urine samples

Blood samples were collected from the three patients and from 21 age-matched healthy controls, EDTA-anticoagulated, processed, and stored, as described earlier.23 Urine samples were taken at the same time and immediately stored at −80 °C. Glycerol concentrations were measured using a three-step enzymatic reaction with glycerol kinase, as previously described.23

Genetic screening for AQP3, AQP7, and AQP9

Genomic DNA was isolated from leukocytes by a salting-out method. Platelet RNA was extracted with Trizol (Life Technologies, Merelbeke, Belgium). Approximately 1 μg of total RNA was used in the presence of RNAseI inhibitor (Promega, Leiden, The Netherlands) for oligo(dT)-primed first-strand complementary DNA synthesis with Moloney murine leukemia virus reverse transcriptase (Life Technologies). AQP3 (NG_007476) and AQP9 (NG_011975) were PCR-amplified from the patients’ complementary DNA, using primers AQP3-1 (forward) and AQP3-2 (reverse), and AQP9-1 (forward) and AQP9-2 (reverse), respectively (see Supplementary Table S2 online for primer sequences) and sequenced on an ABI310 automatic sequencer (Applied Biosystems, London, UK).

The AQP7 gene (NG_027764) was screened using genomic DNA from the index cases. The entire exonic sequence with exon/intron boundaries was amplified as five fragments, using specific primer sets (see Supplementary Table S2 online for primer sequences) to avoid coamplification of the AQP7 pseudogene (NR_002817).20

Hematologic counts and functional platelet studies

EDTA-anticoagulated blood was analyzed on an automated cell counter (Cell-Dyn 1300; Abbott Laboratories, Abott Park, IL) to determine blood cell counts and mean platelet volume. Platelet-rich plasma was prepared by centrifugation (15 min at 150g) of whole blood anticoagulated with 3.8% trisodium citrate (9:1). The platelet-rich plasma was used for functional platelet studies and electron microscopy studies, as described previously.24 In short, aggregation studies were carried out by adding Horm collagen (0.5, 1, and 2 μg/ml), epinephrine (1.25 and 2.5 μmol/l), ristocetin (0.5 and 1.2 mg/ml), TRAP6 (thrombin receptor activating peptide) (18 μmol/l), arachidonic acid (1 mmol/l), or adenosine diphosphate (ADP) (5 and 10 μmol/l). ATP secretion tests were performed after stimulation of platelets with Horm collagen (2 μg/ml) and ADP (10 μmol/l).

The effects of the polyclonal rabbit anti-AQP7 antibody (sc-28625; Santa Cruz Biotechnology, Santa Cruz, CA) and a control polyclonal rabbit anti-FOG1 antibody25 on platelet ATP secretion were evaluated using a modified method of the one described by Cho et al. for AQP1.26 First, the monoclonal Fc receptor–blocking antibody IV.3 (purified from hybridoma ATCC-HB217; American Type Culture Collection, Molsheim, France; 10 μg/ml) was added to platelet-rich plasma for 5 min at 37 °C, after which the platelets were permeabilized with 50 μg/ml saponin (Merck, Darmstadt, Germany) for 10 min at 37 °C together with the anti-AQP7 or control anti-FOG1 antibody (at a final concentration of 5 μg/ml). Platelet ATP secretion was measured after stimulation with 20 μmol/l ADP.

Immunoblot analysis

The platelet releasate, microvesicle fraction, and remaining insoluble pellet after stimulation of washed platelets with strong agonists (TRAP6 with A23187 or thrombin) were obtained as described previously.24,27 Equal amounts of protein extracts (10 μg) were resolved by SDS–polyacrylamide gel electrophoresis on 10% gels and transferred onto a Hybond ECL-nitrocellulose membrane (GE Healthcare, Diegem, Belgium). After blocking in 5% milk powder in Tris-buffered saline–Tween 20 (10 mmol/l Tris-HCl pH 8.0, 150 mmol/l sodium chloride, 0.1% Tween 20) for 1 h at room temperature, the membranes were incubated overnight at 4 °C with the anti-AQP7, anti-integrin β3 antibody (no. 4702, Cell Signaling Technology, Leiden, The Netherlands), or anti-β–actin (no. 4970, Cell Signaling Technology) antibody. The membranes were then incubated for 3 h at room temperature with an horseradish peroxidase-conjugated secondary antibody (at 1:1,000). The subsequent staining was performed with western blotting-enhanced chemiluminescence detection reagent (GE Healthcare).

Immunostaining of platelets and CHRF cells

Human megakaryocytic CHRF-288-11 cells (ATCC-CRL10107) were grown as previously described.28 Platelets from the platelet-rich plasma were washed in the presence of Apyrase 1 U/ml and 10 mol/l prostaglandin E1 (PGE1), and resuspended in modified Tyrode-HEPES buffer so as to obtain a concentration of 2 × 1010 platelets/l. CHRF cells were spread for 1 h and platelets for 30 or 45 min at 37 °C on glass cover slips coated with human fibrinogen (100 μg/ml). Thirty minutes after being spread, the platelets were further stimulated with 20 mol/l TRAP6 for 15 minutes at 37 °C. Adherent cells were fixed with 4% paraformaldehyde in cytoskeleton buffer (0.1 mol/l PIPES, 2 mol/l glycerol, 1 mmol/l EDTA, 1 mmol/l MgCl2, pH 6.9), and permeabilized for 15 min with 0.2% triton X-100 (Roche, Mannheim, Germany) at room temperature. After blocking with 1% bovine serum albumin (Albumax; Life Technologies) for 30 min at room temperature, the cells were incubated with a 1:50 anti-AQP7, anti-von Willebrand factor, or anti-CD63 (BD biosciences PharMingen, San Diego, CA) antibody overnight at 4 °C. After five washing steps with phosphate-buffered saline, the cells were incubated with a specific secondary antibody diluted 1:200 (Alexa Fluor 488, or 568, or 647 conjugated; Life Technologies) together with phalloidin rhodamine (Sigma-Aldrich, Poole, UK) for F-actin staining, for 45 min at 37 °C. Analysis was done using a Zeiss Axiovert 100M confocal microscope (Carl Zeiss, Gottingen, Germany). Fluorescence intensities were quantified using the Java image processing program ImageJ 1.34 g (National Institutes of Health image software).

Statistical analysis

Statistical differences between the two groups were evaluated using the Student’s t-test (two-tailed). For evaluating the effect of the anti-AQP7 antibody on platelet secretion, the nonparametric Mann–Whitney U-test was used because of the low number of comparisons. Differences were considered significant for values of P < 0.05.

Results

Clinical description

We describe three children (from unrelated families) with psychomotor retardation of unknown etiology. The pedigrees of the patients are shown in Figure 1a and the detailed clinical descriptions of the index patients and relatives are listed in Table 1 .

Figure 1
figure 1

Clinical and genetic characterization of patients. (a) Pedigrees of the three unrelated families, indicating the AQP7 G264V genotype and phenotype characterized by the absence or presence of hyperglyceroluria, platelet ultrastructural abnormalities as detected by electron microscopy, and neurodevelopmental problems. The index patients are indicated by arrows. Lack of data is indicated by a question mark in the box. (b) Genetic analysis of AQP7. At nucleotide position 963, a homozygous G-to-T substitution was found in the index subjects (right panel), causing a glycine-to-valine substitution at codon 264 (G264V). (c) Evolutionary conservation of the affected AQP7 codon 264 in various species and AQP subclasses. The highly conserved residue is underlined. AQP7, aquaporin 7.

Full size image
Table 1 Genotype, clinical characteristics, and metabolic screen in index patients and their relatives
Full size table

The first child (index subject 3) was the only child of healthy, unrelated parents. He presented to the outpatient clinic at the age of 14 months because of recurrent vomiting, and subsequently developed gross motor delay (walking commenced only at the age of 17 months) and autistic characteristics, and had an IQ of 108. At the age of 2 years, an episode of seizures was reported, with normal electroencephalogram. The child’s facial features were not dysmorphic and he was not overweight (body mass index 14.8 kg/m2 at the age of 5 years, −0.6 SD score, ref. 29). He was taking no medications.

The second patient (index subject 6) was a girl with obvious psychomotor retardation (skill level of a 2-year-old child at a calendar age of 4 years), hypotonia, and discrete dysmorphic facial features (large mouth with small widely spaced teeth and dysplastic left ear). The brain magnetic resonance imaging result was normal. Her weight and length were at the third percentile (body mass index 15.5 kg/m2 at the age of 5 years, +0.0 SD score). One older sister (subject 7) had learning difficulties at school, and a younger brother (subject 8) had a normal phenotype at 18 months of age. The parents are healthy and unrelated.

The third child in our study (index subject 14) was a boy, the fourth child of consanguineous Turkish parents. He presented with psychomotor retardation and profound hypotonia. From the age of 6 months he had developed generalized seizures, and had been treated with valproic acid and levetiracetam. The brain magnetic resonance imaging was normal. His weight and length were at percentiles 25 and 50, respectively (body mass index 15.9 kg/m2 at the age of 5 years, +0.3 SD score). Clinical examination showed internal strabismus in addition to dolichocephaly and small testes. An older brother (subject 11) had shown a similar constellation of symptoms (psychomotor retardation, hypotonia, and epilepsy) but had died at the age of 8 months. No further data or DNA were available relating to this subject. One sister (subject 13) had learning difficulties at school, attributed to inadequate mastering of language skills. Another older sister (subject 12) and both parents were asymptomatic.

Extensive diagnostic testing was performed in these index patients but failed to pinpoint a particular neurodevelopmental diagnosis ( Supplementary Table S3 online for overview of all tests).

Glycerol measurements and other metabolic parameters

Metabolic screening revealed normoglycerolemic hyperglyceroluria in the three index patients ( Table 1 ). Plasma glycerol levels were similar in the patients (mean 91 μmol/l, range 46–162 μmol/l) and the controls (mean 79 μmol/l, range 34–229 μmol/l), ruling out prerenal causes of severe glyceroluria (495–2,806 mmol/mol creatinine in the patients as compared with <1 mmol/mol creatinine levels in the controls). Urine glycerol excretion was within normal ranges in the parents of the index cases, except for the mother (subject 9) of index patient 14, who had a urine creatinine level of 681 mmol/mol. Of note, urine glycerol levels were elevated in the clinically normal siblings of the index patients (mean 1,024 mmol/mol creatinine, range 4111,432). The patients’ levels of blood glucose, serum cholesterol, and triglycerides were within normal ranges. Elevated levels of thyroid-stimulating hormone but normal thyroid hormone levels were also noted in the patients, except for index patient 6, who also had a low free thyroxine level (0.82 ng/dl, normal 0.93–1.70 ng/dl). She was started on L-thyroxine therapy.

Genetic screening for AQP3, AQP7, and AQP9

Based on the finding of pronounced urine glycerol loss, the AQP3, AQP7, and AQP9 genes were screened for mutations.30,31 No mutations were detected in AQP3 and AQP9. In the AQP7 gene, a homozygous nucleotide substitution (G to T in exon 8)was detected in all three index cases, converting glycine into valine at codon 264 (G264V) in the sixth transmembrane domain of the protein ( Figure 1b ). The parents were heterozygous carriers of the mutation, except for the mother of the third patient (subject 9) who was homozygous ( Figure 1a and Table 1 ). All siblings of the index patients were also homozygous for this mutation. The mutated residue is well conserved in AQP7 of different species but also among the other AQP subfamily members (classical AQPs, aquaglyceroporins and unorthodox AQPs; Figure 1c ), suggesting that it is important for AQP functioning. It is obvious that the homozygous G264V variant cosegregates with the hyperglyceroluria in the three families. The G264V AQP7 variant had previously been screened in Japanese and Caucasian subjects with obesity and/or type 2 diabetes, but no association between the genotype and these phenotypes could be demonstrated.20,21 These studies did not include measurement of the urine glycerol levels, but normal plasma glycerol levels were reported for the homozygous carrier in the Japanese study,20 in line with the findings in our study.

Morphologic and functional platelet studies

Hematologic analysis showed a normal platelet count with a mildly increased mean platelet volume (9.7 ± 1.0 fl vs. 7.8 ± 1.0 fl, P = 0.001) in all homozygous carriers of G264V AQP7 carriers relative to heterozygous carriers ( Table 2 ). Electron microscopy revealed enlarged and especially more rounded platelets with centrally localized granules in some of the platelets and a pronounced open canalicular system ( Figures 1a and 2a ). Subject 10, a heterozygous carrier who was the father of index patient 14, had structurally normal platelets. We also detected a mild subclinical platelet dense granule secretion defect in the homozygous carriers, as highlighted by the absence of a secondary aggregation response to epinephrine stimulation ( Figure 2b and Table 2 ). This type of aggregation defect is typically present in patients with reduced platelet dense granule release, as we have earlier described with respect to patients with storage pool disease.24 The use of a stronger agonist such as Horm collagen showed no defects in aggregation response in the patients because this type of platelet stimulation is less dependent on secretion.24 In line with this finding, the standard platelet aggregation tests with different agonists (ristocetin, arachidonic acid, U46619, ADP, and Horm collagen) that are generally used to detect a platelet-related clinical bleeding problem were all normal for subject 6. In addition, a reduction in ATP secretion was found in response to stimulation with ADP, the response being less pronounced with Horm collagen in the patients ( Figure 2c and Table 2 ). These findings further supported the likelihood that the mild dense granule secretion defect was present in these subjects. Although these homozygous G264V carriers presented with a platelet secretion defect, none of them had any obvious clinical bleeding problems.

Table 2 Platelet studies in homozygous G264V subjects
Full size table
Figure 2
figure 2

Platelet morphology and functioning. (a) Electron microscopy studies of platelets from individuals who were homozygous for AQP7 G264V showed enlarged, more rounded platelets as compared with the typical discoid control platelets. Some platelets also presented with centralized granules and prominent open canalicular systems, as indicated by arrows. Bars represent 1 μm. (b) Absence of the secondary aggregation wave after platelet stimulation with epinephrine (1.25 μmol/l) in index patient 3 as compared with a control individual. This curve is representative of those obtained for the other homozygous G264V carriers. (c) ADP (10 μmol/l) and collagen (2 μg/ml) induced aggregation (upper curves) and ATP secretion (lower curves) in platelets of index patient 6. These curves are representative of those obtained for the other homozygous G264V carriers. ADP, adenosine diphosphate; AQP7, aquaporin 7; ATP, adenosine triphosphate.

Full size image

Subcellular localization of AQP7 in human platelets

Given that the AQP7 G264V variant affects platelet morphology and functioning, we further studied the subcellular localization of the AQP7 protein in normal human platelets. After full stimulation of platelets with strong agonists (A23187/TRAP6 or thrombin), AQP7 was found mainly in the platelet releasate, using immunoblot analysis ( Figure 3a ). Integrin β3 is a typical platelet cell membrane protein that was, in line with expectations, detected mainly in the pellet and microvesicle membrane fractions. Actin is bound to the membranes but, after stimulation, can also be detected in the platelet releasate, as described by other proteomic studies.27 This is probably because the actin that is bound to the platelet granule-membrane is partially released during stimulation; however, the underlying mechanism for this is not known. We hypothesize that AQP7 as a membrane protein is released through a similar mechanism. Indeed, evidence exists of the unexpected presence of cytoskeletal and actin-binding proteins in the platelet releasate.24,27 In addition, immunostaining of platelets spread on fibrinogen showed a reduced expression of AQP7 for the duration of platelet spreading, and an even more pronounced reduction after TRAP6 stimulation ( Figure 3b ). These experiments again indicate that AQP7 is partially released during spreading, initiating mild platelet activation and secretion, and that it is further released when stimulated by TRAP6, which induces full activation and granule release. To further specify the exact subcellular location of AQP7, immunostaining was performed. It showed colocalization of AQP7 with CD63 (a marker for platelet dense granules) but not with von Willebrand factor (a marker for platelet alpha granules) ( Figure 3c ). The colocalization of AQP7 with CD63 could also be demonstrated in the human megakaryocytic cell line CHRF.

Figure 3
figure 3

AQP7 localization and functioning in normal human platelets. (a) AQP7 is present mainly in the platelet releasate (R) after full stimulation of platelets from a healthy control subject. Integrin β3, used as a control membrane protein, was present in the pellet (P) and microvesicle (M) membrane fractions. (b) Left: AQP7 immunostaining in platelets spread for 30 and 45 min without adding an agonist, and at 45 min after costimulation with TRAP6. Bars represent 10 μm. Right: Quantification of AQP7 staining during platelet spreading. Mean intensities are given as percentages of the starting fluorescence intensity. Error bars represent SDs. The data are from two independent experiments. (c) AQP7 immunostaining in spread platelets shows a granular pattern (arrows in left panel). AQP7 presents a strong colocalization with the dense granule marker CD63 in platelets (arrowheads in middle left panel), which was also seen in megakaryocytic CHRF cells (arrowheads in middle right panel). However, only weak colocalization could be seen with the α-granule marker von Willebrand factor (VWF) (right panel). A negative control with the secondary antibody excluded aspecific binding (data not shown). Bars represent 10 μm. (d) ADP-induced ATP secretion after incubation with the polyclonal rabbit anti-AQP7 or anti-FOG1 antibody in permeabilized platelets. Mean values along with SD are depicted. The data are from three independent experiments. ADP, adenosine diphosphate; AQP7, aquaporin 7; ATP, adenosine triphosphate; TRAP6, thrombin receptor activating peptide.

Full size image

To evaluate the role of AQP7 in platelet-dense granule release, permeabilized platelets were incubated with anti-AQP7 or a control antibody before ATP secretion was measured. This anti-AQP7 antibody was raised against the amino acid sequence where the G264V mutation is located (the epitope being amino acids 169–269 of the C-terminus of AQP7). Although not significant (P = 0.1), a trend toward a decreased ATP secretion was found after stimulation with 20 μmol/l ADP for the anti-AQP7 antibody as compared with the aspecific control anti-FOG1 antibody ( Figure 3d ).

Discussion

We describe, for the first time, the cosegregation of the homozygous G264V AQP7 variant with a marked loss in urine glycerol. A pronounced loss of glycerol in the urine coincident with normal plasma concentration values has never been described in humans, although AQP7-deficient mice have been shown to have hyperglyceroluria.19 Renal AQP7 is responsible for reabsorption of glycerol at the renal proximal tubules in coordination with AQP3.30,31 All the homozygous carriers of G264V AQP7 in our study had hyperglyceroluria, whereas no glycerol loss was present in the heterozygous parents. The hypothesis that this mutation has functional relevance was earlier supported by the findings of Kondo et al.,20 who could not show any association between the presence of the mutation and the occurrence of obesity or type 2 diabetes; however, they did find the first homozygous G264V subject with impaired plasma glycerol release from fat tissue after exercise tests, although urine glycerol levels were not determined. AQPs as a family have a conserved structure,8,32 existing in membranes as homotetramers. Each subunit monomer forms a separate permeable pore composed of six α-helix transmembrane domains with inverted symmetry between the first and last three domains, and a pore-forming loop with the signature NPA motif.33 The G264V mutation is located in the highly conserved GxxxG motif of the sixth transmembrane helix in nearly all AQPs ( Figure 1c ).33,34,35 Glycine in the motif can sometimes be replaced by alanine.34 It should be noted that the presence of a mutation in another genetic locus, cosegregating with the aforementioned mutation, still remains a possibility.

Four independent AQP7-knockout mouse models that are viable and healthy have been described ( Supplementary Table S1 online). Based on these models, AQP7 was found to be crucial in fat metabolism as well as in renal glycerol reabsorption. Except for the fact that AQP7-knockout mice as well as the human patients in our study had profound urine glycerol loss, there is nothing in common between the two phenotypes. This can be explained by the difference between having a complete loss of AQP7 and having only an AQP7 missense mutation with probable partial loss of activity. In addition, significant differences have been noted between the different AQP7-knockout models, probably related to their genetic backgrounds and possibly indicating redundancy by other AQPs. The latter factor could also play a role in humans. To our knowledge, no study exists that describes a role for AQP7 in platelets or links AQP7 deficiency to behavior changes using AQP7-knockout mice.

Platelets are easy to isolate in a nonactive state and are ideal cells in which to evaluate G-protein signal transduction, secretion, and adhesion in hemostatic disorders. In addition, we have shown that these types of functional platelet studies are also very useful in characterizing an unknown disorder that includes only a subclinical platelet defect.36 We therefore performed various platelet tests in these patients after all other diagnostic tests (see Supplementary Table S3 online) were negative. Our study indicates that AQP7 is also important for platelet morphology and functioning. Homozygous carriers have enlarged and more rounded platelets with centrally localized granules. This results in a reduced release of granule content when platelets are stimulated with a weak agonist. Even though AQP7 was detected in the releasate after platelet stimulation, it is not clear whether it is actually secreted, given that the releasate also captures other membrane-bound proteins such as actin.37 The published literature contains extremely limited support for the hypothesis that AQPs have a role in the morphology and functioning of platelets. It was only very recently that AQP6 was shown to be involved in the regulation of platelet volume through a G protein–mediated pathway.38 Although most AQPs reside at the plasma membrane, their presence in intracellular organelles such as secretory granules has been demonstrated in other types of cells, wherein they also play a role in secretion through regulation of the vesicle volume.39,40 Regulated secretion is a feature of specialized cells such as neurons and neuroendocrine cells, but platelets also secrete their granules at sites of vascular damage in response to stimulation. A role for AQPs in platelet granule secretory function has never been reported. AQP7 could interfere with platelet-dense granule formation and functioning, as previously described with respect to AQP1, which was found to be associated with exosomes during reticulocyte maturation.41

A largely variable neurologic phenotype was observed among the homozygous G264V carriers, ranging from severe hypotonia, psychomotor retardation, and/or epilepsy needing medication in the index cases to normal functioning in their healthy siblings. More studies are needed to define whether mental disability is also related to the AQP7 G264V variant, or whether a separate genetic factor is responsible for the neurological phenotype in the affected homozygous carriers. AQP7 was found to be expressed during perinatal development of the mouse brain.42 AQP1 and AQP6 are associated with synaptic vesicles and participate in their swelling,43 implying a possible role for AQPs in neuronal granule secretion. The functions of AQPs in the central nervous system seem to be diverse; studies have shown a role for AQPs in bidirectional transport of water between brain and blood vessels, in cerebrospinal fluid formation, in neural signal transduction, in osmoreception, and in some brain pathologies such as brain edema, brain tumors, and autism (reviewed in refs. 8,44). Given that AQP7 also facilitates glycerol transport, it might have a role in the energy metabolism of neurons, as was previously found with respect to AQP9. The latter showed increased expression in human glioblastoma, and this increased expression was associated with increased energy metabolism of glioma cells.45

In conclusion, this study is the first to show a cosegregation of the G264V defect in AQP7 with hyperglyceroluria, and to imply the involvement of a new category of proteins, the AQPs, in platelet morphology and granule secretion.

Disclosure

The authors declare no conflict of interest.