A Germline Variant in the PANX1 Gene Has Reduced Channel Function and Is Associated with Multisystem Dysfunction*♦

Pannexin1 (PANX1) is probably best understood as an ATP release channel involved in paracrine signaling. Given its ubiquitous expression, PANX1 pathogenic variants would be expected to lead to disorders involving multiple organ systems. Using whole exome sequencing, we discovered the first patient with a homozygous PANX1 variant (c.650G→A) resulting in an arginine to histidine substitution at position 217 (p.Arg217His). The 17-year-old female has intellectual disability, sensorineural hearing loss requiring bilateral cochlear implants, skeletal defects, including kyphoscoliosis, and primary ovarian failure. Her consanguineous parents are each heterozygous for this variant but are not affected by the multiorgan syndromes noted in the proband. Expression of the p.Arg217His mutant in HeLa, N2A, HEK293T, and Ad293 cells revealed normal PANX1 glycosylation and cell surface trafficking. Dye uptake, ATP release, and electrophysiological measurements revealed p.Arg217His to be a loss-of-function variant. Co-expression of the mutant with wild-type PANX1 suggested the mutant was not dominant-negative to PANX1 channel function. Collectively, we demonstrate a PANX1 missense change associated with human disease in the first report of a “PANX1-related disorder.”

Pannexins are a new class of large-pore channels that were discovered early in the new millennium (1,2). Members of the gene family (PANX1, PANX2, and PANX3) are expressed in numerous organs, tissues, and cells with PANX1 being the most prevalent (3). Rodent Panx1 is an ϳ41-48-kDa protein with its broad range in size due to the fact that it is post-translationally modified in what is now referred to as Gly0, Gly1, and Gly2 species to reflect the degree of glycosylation (4 -6). PANX1 oligomerizes into a hexamer that contains a large pore functioning at the cell surface to allow the passage of small molecules below 1000 daltons in size (7,8). Although the scope of molecules that pass through PANX1 pores is likely broad (9,10), the functional consequence of ATP release via these channels is best understood (11). For instance, PANX1 channels have been shown to release ATP in apoptotic immune cells as "find me" signals for the clearing of dying cells (12).
In the last decade, PANX1 channels have become intimately linked to disease because of the fact that they are expressed in the vast majority of human cell types (13). Until this study, the link to disease has been associated with basal or elevated functional levels of PANX1, but the mechanisms involved remain poorly understood (13). In the first reported association with disease, PANX1 was linked to neuronal cell death in models of ischemia and stroke followed later by clear linkages to seizure severity and duration (14 -16). The abundant expression of PANX1 in enteric neurons led to the discovery that these channels played vital roles in inflammatory bowel diseases, including ulcerative colitis and Crohn's disease (17). Surprisingly, PANX1 channels can also be hijacked by viruses to facilitate infection as documented for HIV-1 (18). Furthermore, in mouse models, high levels of Panx1 in melanomas have been shown to facilitate disease progression, although Panx1 overexpression has been shown to be tumor suppressive in glioblastomas, suggesting that pannexins are likely to have tumor-specific effects in cancer (19 -21). The list of connections between PANX1 and disease is extensive and continues to grow as there are elegant studies supporting a link between PANX1 and epilepsy (22,23), glaucoma (24), migraines (25), Alzheimer disease (26), and diabetes (27).
Although no disease-linked germline PANX1 variants have been identified prior to this study, Kwak and co-workers (28), including a member from our team, discovered through sequencing of 96 healthy patients that a single nucleotide polymorphism (400A3 C) existed with a frequency of approximately one-third 400A allele and two-thirds 400C allele. Although none exhibited overt disease, those homozygous for the 400C allele exhibited greater collagen-induced platelet aggregation, suggesting the possibility that there may be some variability in platelet reactivity among healthy individuals (28).
In this study we report on the first patient with a PANX1 homozygous germline variant resulting in an arginine at position 217 being replaced with a histidine (p.Arg217His). This young female patient clinically presents with extensive disease that includes intellectual disabilities, severe hearing loss, and multiple other multisystem defects. Her unaffected parents and sibling are heterozygous for the c.650G3 A PANX1 variant. Generation and characterization of the R217H mutant revealed that it is a loss-of-function variant as assessed by ATP release, dye uptake, and electrophysiological evaluation of the channel properties. Although functional levels of PANX1 have been correlated to the onset and/or progression of over 10 diseases (13), disease-linked germline variants in the PANX1 gene have not been previously reported. This study represents the first report of a patient harboring a disease-associated PANX1 variant.

Experimental Procedures
Sequencing-Genomic DNA was extracted from whole blood from the proband and her parents. Exome sequencing was performed on exon targets captured using the Agilent SureSelect Human All Exon V4 (50 Mb) kit (Agilent Technologies, Santa Clara, CA). The sequencing methodology and variant interpretation protocol has been previously described (29). Briefly, whole exome sequencing (WES) 4 produced 18.2 Gb of sequencing data for the proband. Mean coverage of captured regions was ϳ251ϫ for the proband's sample, with ϳ99.17% covered with at least 10ϫ coverage, an average of 89.82% of base call quality of Q30 or greater, and an overall average mean quality score of Q35. Filtering of common single nucleotide polymorphisms (Ͼ10% frequency present in the 1000 Genomes database) resulted in ϳ4187 variants in the proband sample. After automated filtering of variants with a minor allele frequency of Ͼ10%, manual curation was performed to filter less common variants with a minor allele frequency of 1-10% and single variants in genes inherited from unaffected parents and to evaluate predicted effects of rare variants and associated human conditions. Cells and Reagents-Normal rat kidney (NRK), mouse Neuro2A (N2A), and human embryonic kidney (HEK293T) cells were obtained from the American Type Culture Collection (Manassas, VA). Ad293 cells (derivative of human embryonic kidney 293 cells with improved cell adherence) were obtained from Agilent Technologies (Palo Alto, CA). All cells were grown in high glucose (4500 mg of glucose/liter) DMEM (Invitrogen) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. Cells were cultured within a humidified environment that maintained 5% CO 2 and a temperature of 37°C. Probenecid (catalog no. P36400) was purchased from Molecular Probes, and ATP (catalog no. PV3227) was obtained from Thermo-Fisher Scientific. Carbenoxolone (catalog no. C4790) and Dulbecco's phosphate-buffered saline (catalog no. D1283) were obtained from Sigma.
Mutant DNA Constructs-In humans, the gene encoding pannexin 1 is referred to as PANX1. Panx1 is used when referring to the equivalent mouse gene, and the encoded proteins are not italicized. An expression vector encoding human PANX1 was purchased from InvivoGen (pUNO1-hPanx1). The R217H PANX1 mutant encoding expression vector was generated by special order to Norclone Biotech Industries (London, Ontario, Canada). The fidelity of the PANX1-and R217H-encoding vectors was confirmed by DNA sequencing at the Robarts Research Institute DNA Sequencing Facility (London, Ontario, Canada) using an Applied Biosystems (Foster City, CA) 3730 analyzer.
Transfections and Immunofluorescent Labeling-Transient transfections were performed using Lipofectamine2000 (Life Technologies, Inc.) according to the manufacturer's instructions. For some experiments, a GFP-encoding vector was cotransfected with the PANX1/R217H-encoding vector. For stable transfections, cells were kept under G418/blasticidin selection pressure.
HEK293T cells, maintained as described above but with no antibiotics, were plated onto 100-mm dishes and transfected when reaching 60 -80% confluence. Transfected cells were subsequently resuspended and seeded onto 35-mm dishes for electrophysiological recordings. The remaining sister cells were used for immunoblotting, as described below.
For immunolabeling studies, cells were fixed in 80% methanol, 20% acetone for 20 min at 4°C and blocked for 30 min with 2% BSA. For PANX1, we used an affinity-purified custom-made rabbit anti-human PANX1 polyclonal antibody (PANX1 CT-412, 0.5 g/ml), generated by Genemed Synthesis (San Francisco) against the C-terminal sequence of human PANX1 ( 412 NGEKNARQRLLDSSC 426 ) (30). An anti-Cx43 monoclonal antibody (1:50 dilution; P4G9 from Dr. Paul D. Lampe, Fred Hutchinson Cancer Research Center, Seattle) was also used. Primary antibody labeling was followed by fluorescently tagged secondary antibodies Alexa Fluor 555 and Alexa Fluor 488 (Life Technologies, Inc.; diluted 1:500) for 1 h at room temperature. Cell nuclei were stained for 5 min with TO-PRO-3 Iodide (642/661) (Life Technologies, Inc.) and then rinsed in distilled H 2 O before mounting. All labeling was visualized with an LSM 510 META inverted confocal microscope equipped with a 63ϫ oil objective (Carl Zeiss, Jena, Germany).
Cell Surface Biotinylation-Ad293 or N2A cells were transiently transfected with the cDNAs encoding wild-type PANX1 or the R217H mutant. Forty eight hours after transfection, cells were treated for 40 min with either DMEM ϩ 5% fetal bovine serum (FBS) (control) or DMEM/FBS containing 140 mM K ϩ Glu and 100 mM K ϩ Cl Ϫ . Cells were then placed on ice, washed in ice-cold Hanks' balanced salt solution, and cell surface proteins labeled with 1.5 mg/ml EZLink Sulfo-NHS-SSbiotin (ThermoFisher catalog no. 21331) for 1 h. Biotin was subsequently quenched with 100 mM glycine for 30 min, and cells were lysed. Cleared supernatant containing 150 -250 g of protein was incubated with 25 l of NeutrAvidin affinity beads (ThermoFisher catalog no. 29200) overnight, rotating at 4°C to precipitate biotin-labeled proteins. Following incubation, the beads were washed twice with PBS, and precipitated proteins were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and immunoblotted to identify biotinylated PANX1 proteins using a primary rabbit anti-human PANX1 antibody (1:1000 dilution). Total PANX1 and GAPDH levels were determined by immunoblotting 20 g of protein from each cell lysate used for biotinylation and the input.
ATP Release Assay-N2A cells were transfected to express wild-type PANX1 or the R217H mutant and subcultured into 24-well dishes. Forty eight hours post-transfection, the cell medium was exchanged to medium with 5% FBS (heat-inactive) plus 20 M ARL67156 (Santa Cruz Biotechnology, Santa Cruz, CA). ATP release was stimulated with 140 mM potassium gluconate and 100 mM KCl ("High K ϩ " solution) for 20 min. In addition, some cells were pre-treated with 70 M carbenoxolone (CBX, Sigma) to block the opening of PANX1 channels (31,32). Samples were taken and processed for ATP content using a bioluminescence ATP determination kit (Life Technologies, Inc.). Briefly, 10 l of the cell supernatants were transferred to 96-well dishes with 90 l of a luciferase standard reaction solution, according to the manufacturer's protocol. ATP released into the medium was quantitatively measured by luminometry. Although cells were plated at equal numbers, to account for any unexpected differences in cell numbers, luminescence values were normalized to the total amount of protein for each well (as determined by a BCA assay, Pierce), and ATP concentrations were then determined from an ATP standard curve. Cell-free solutions were tested in the presence of different concentrations of ATP as a control. Plotted values are means Ϯ S.E. * ϭ Ͻ0.05, ** ϭ Ͻ0.01, and *** ϭ Ͻ0.001 as determined by one-way ANOVA analysis followed by a Bonferroni corrected post hoc test (n ϭ 3). All statistical analyses were performed using GraphPad version 4.1.
Dye Uptake Assay-Ad293 cells were transiently and/or stably engineered to express GFP together with wild-type PANX1 or the R217H mutant. Briefly, cells were sparsely subcultured on 35-mm glass-bottom dishes (MatTek Corp., Ashland, MA) for 1 day prior to the medium being replaced with 5% FBS (heat-inactive), DMEM containing 140 mM potassium gluconate, 100 mM KCl, and 10 M ethidium bromide (EtBr) (7,33,34). To block PANX1 channels, cells were treated with 1 mM probenecid (35). Rapid time-lapse imaging was used to detect EtBr uptake and GFP fluorescence every 5 min for up to 40 min using a Zeiss LSM 510 META imaging system. Mean fluorescence intensity was measured using Zen lite (Zeiss). For each treatment, 10 -50 cells were analyzed for mean fluorescence intensity at 0, 5, 30, and 40 min after exposure to the dye. Data are presented as means Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 PANX-expressing cells compared with wild-type or R217H mutant-expressing cells as determined by a one-way ANOVA followed by Tukey's Multiple Comparison Test. Statistical analysis was performed using GraphPad version 4.1.
Electrophysiology-Tight-seal whole-cell recordings were conducted at room temperature 24 -72 h after transfection using an Axon MultiClamp 700A patch clamp amplifier and Digidata 1322A data acquisition system (Molecular Devices). On the day of experiments, recordings were performed using an interleaved design from 293T cells expressing wild-type PANX1, R217H, or both prepared in parallel. Transfected cells were visually selected for recording on the basis of GFP fluorescence. Using a gravity-driven multi-barreled perfusion system, cells were continuously superfused with bath solution (extracellular fluid) containing 140 mM NaCl, 5.4 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 33 mM D-glucose, and 25 mM HEPES, adjusted to pH 7.4 (with NaOH), and osmolarity between 310 and 320 mosM/liter. Patch pipettes were pulled from borosilicate glass (World Precision Instruments, Inc., Sarasota, FL) and had a resistance of 4 -5 megohms when filled with a pipette solution (intracellular fluid) containing 150 mM cesium gluconate, 2 mM MgCl 2 , and 10 mM HEPES, adjusted to pH 7.3 (with CsOH), and osmolarity between 290 and 300 mosM/liter. For the high potassium treatment, a modified bath solution was applied containing 90 mM NaCl, 50 mM potassium gluconate, 5.4 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 33 mM D-glucose, and 25 mM HEPES, adjusted to pH 7.4 (with NaOH), and osmolarity between 310 and 320 mosM/liter.
In cells voltage clamped at Ϫ60 mV, PANX1 and R217H currents were recorded in response to 500-ms voltage ramps (Ϯ100 mV, 1/10 s). PANX1 and R217H ramp currents were first recorded in normal bath solution (or high potassium solution) for 5 min and then inhibited by a bath solution (normal or high potassium) containing 100 M carbenoxolone. Signals, filtered at 2 kHz and sampled at 10 kHz, were collected and analyzed using Clampex 9.2 and Clampfit 9.2 software (Molecular Devices). Data are presented as means Ϯ S.E. Statistical analyses were conducted using GraphPad Prism 5.0 software (GraphPad Software, Inc., La Jolla, CA). Comparisons between PANX1 and R217H groups were made using two-way ANOVA analyses with Bonferroni corrected post hoc test. Comparisons between PANX1, R217H, and the co-expression groups were made using one-way ANOVA analyses with Bonferroni corrected post hoc test. A difference was considered significant at *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.
In other experiments on the day of recordings, 293T cells expressing wild-type PANX1 or R217H were first pretreated with brefeldin A (BFA) (5 g/ml; Tocris Bioscience, UK)-containing medium for 6 h at 37°C, a treatment regimen shown previously to prevent ion channel forward trafficking (36). Later, the culture medium was removed and replaced by a bath solution containing 5 g/ml BFA right before the recording.

Results
Clinical Presentation-We report on a 17-year-old female who initially presented for clinical genetics evaluation at 15 years of age with multiple concerns, including primary ovarian failure, intellectual disability, sensorineural hearing loss, and kyphosis. She had significant speech delay that led to the diagnosis of severe sensorineural hearing loss at 15 months of age, requiring bilateral cochlear implants at age 6. Despite the implants, her receptive and expressive language remains delayed with nearly absent speech and the ability to use only basic signs. She is unable to read or to count past 30. Formal psychological testing at 16 years of age revealed both verbal comprehension index and perceptual reasoning indices of 45 (less than 2nd percentile) on the Wechsler Intelligence Scale for Children (WISC-IV), commensurate with a child who is 6 years old, and academic skills at a first grade level, although it was noted that this test is not specifically designed for assessing students who are deaf or hard-of-hearing. She is mainly inde-pendent with basic activities of daily living but does require some help with bathing, cooking, oral care, managing medications, and, occasionally, toileting. Primary ovarian failure was diagnosed via elevated follicle-stimulating hormone (108.6 mIU/ml), luteinizing hormone (23.7 mIU/ml), and low estradiol (Ͻ 5 pg/ml), in the setting of primary amenorrhea with normal pelvic ultrasound and secondary sexual characteristics. Further endocrine evaluation revealed that she had a delayed bone age, hypothyroidism, and ultrasound evidence of fatty liver disease associated with mild transaminitis of unclear etiology. Kyphosis was first noticed in early adolescence and improved, but did not resolve, with physical therapy. Other investigations included a normal echocardiogram, renal ultrasound, and ophthalmic exam, although her family reports some suspected issues with night vision. Physical examination at the age of 16 years revealed mild short stature with a height of 150.2 cm (3rd percentile) and weight of 46.8 kg (12th percentile). She was not facially dysmorphic and generally resembled her immediate family, although she did have kyphosis and bilateral 2-3 toe syndactyly. The family declined the inclusion of patient photos in this report.
Family History and Gene Sequencing-Family history was notable for one sister, 2 years her elder, with mild short stature (height 150 cm) and hypothyroidism, her mother with mild short stature (height 155 cm), and her father having hypothyroidism and fatty liver disease (Fig. 1). The proband's parents are known to be first cousins, originally from Turkey. The family is not aware of any other consanguineous relationships further back in the pedigree. Genetic evaluation of the proband demonstrated a normal chromosome analysis. Chromosome microarray revealed no microdeletions or duplications but did show 139.4 Mb of homozygosity spread over seven chromosomes, consistent with the known consanguinity. Given her complex phenotype, the possibility of a new genetic syndrome Full shading represents the homozygous R217H variants, and half-shading represents heterozygous R217H PANX1 variants. Arrow indicates proband; "ϩ" indicates wild-type allele.

Disease-associated Loss-of-function Panx1 Variant
was considered, and whole exome sequencing of the patient and her parents was performed. A homozygous missense variant in PANX1, c.650G3 A (R217H) was identified, with each parent and her sister found to be heterozygous. These findings were confirmed by Sanger sequencing. This variant is predicted to be deleterious by PhyloP, PolyPhen2, SIFT, CADD, and MutationTaster and is present in the Exome Aggregation Consortium (ExAC) Browser with an allele frequency of Ͻ1/10,000 (6.65e-05). The PANX1 gene is within one of her regions of homozygosity, on chromosome 11.
Expression and Localization of the R217H Mutant-Given that arginine at position 217 is conserved in vertebrates, and in silico analysis predicts the motif containing this residue is likely of structural and/or functional importance, we engineered and sequence-confirmed this mutation into the pUNO1-hPANX1 expression vector. Expression of the R217H mutant in mouse neuroblastoma (N2A) and NRK cells revealed that the mutant trafficked to the cell surface (as revealed by the location of the Cx43 gap junction protein) with seemingly identical efficacy as wild-type PANX1 (Fig. 2A). The R217H mutant was glycosylated to the well documented Gly1 and Gly2 species (4, 30, 37) of PANX1 when expressed in N2A, 293T, and Ad293-human embryonic kidney cells (Fig. 2B). This was not unexpected given that the topological location of the mutation is within the intracellular loop of PANX1 (Fig. 2C), whereas glycosylation occurs on the second extracellular loop (4,30). Thus, through the use of several cell lines to eliminate any potential cell type differences, these studies strongly suggest that the R217H mutant is trafficking-competent and appropriately glycosylated.
Functional Analysis of the R217H Mutant-Because PANX1 channels are large-pore channels (7,9,32) suitable for the uptake of small fluorescent dyes (38,39), we assessed whether channels formed from the R217H mutant were functionally compromised. Ad293 cells expressing wild-type PANX1 or the R217H mutant (Fig. 3A) were assessed for ethidium bromide (EtBr) dye uptake (9) after channels were induced open in high potassium medium (Fig. 3B) (7,34,40). Quantitative assessment of EtBr uptake revealed that cells expressing the R217H mutant exhibited significantly reduced dye uptake over a time course of 5-40 min (Fig. 3B). Confirming the involvement of PANX1-based channels, the pannexin channel blocker probenecid (35) was found to block dye uptake in cells expressing PANX1. Moreover, in untransfected Ad293 cells, high potassium treatment did not stimulate dye uptake. These studies strongly suggest that channels formed from the R217H mutant had impaired ability to uptake a small fluorescent dye.
As PANX1 channels are probably best known as ATP release channels (7, 12, 41), we next determined whether channels assembled from the R217H mutant had reduced ability to release ATP. To eliminate any confounding problems that might be due to ATP release from connexin-based hemichannels (42), we expressed the R217H mutant and wild-type PANX1 in connexin-and pannexin-deficient N2A cells (43) and assessed channel activity under physiological levels of calcium. As predicted, upon exposure to high potassium, there was a surge of ATP release in N2A cells expressing wild-type PANX1 but not when cells expressed the R217H mutant (Fig.  3C). To confirm that cell surface channels were indeed responsible for ATP release, the channel blocker CBX (12,44) eliminated the surge of ATP release induced by high potassium. Variability in ATP release in control cells with and without PANX1 or mutant may represent depletion of intracellular stores of ATP due to baseline activity of PANX1 channels. Nevertheless, these studies support the dye uptake data and strongly suggest that the R217H variant greatly attenuates channel function.
To determine whether the R217H variant would alter ionic current flow through the channel, we performed whole-cell voltage clamp recordings. For these experiments, we made use of 293T cells, devoid of endogenously expressed CBX-sensitive currents (45,46), and we assessed pannexin function by recording membrane currents generated by voltage ramps from Ϫ100 to ϩ100 mV. In 293T cells expressing R217H, ramp currents recorded at ϩ100 mV were reduced by ϳ50% compared with wild-type PANX1 (Fig. 4, A and B). The reversal potentials of R217H (Ϫ71.77 Ϯ 4.415, n ϭ 6) and PANX1 currents (Ϫ61.83 Ϯ 3.508, n ϭ 6) were comparable, suggesting that reduced currents in R217H-expressing cells could not be attributed to a change in ionic permeability. Consistent with previous reports (7,34,40), PANX1-mediated currents were augmented by treatment with high potassium, most notably at negative holding potentials (Fig. 4C). High potassium augmented ramp currents in R217H-expressing cells were reduced by ϳ50% when compared with wild-type PANX1 (Fig. 4C). Ramp currents recorded from R217H-and PANX1-expressing cells, in control or high potassium-containing solutions, were suppressed by CBX (Fig. 4, A-C). The absence of comparable CBXsensitive currents in mock-transfected 293T cells (Fig. 4B) confirms the specific involvement of pannexin channels under our recording conditions. To exclude the possibility that the lossof-channel function was due to a dramatic reduction in the expression of the R217H mutant, PANX1 immunoblotting confirmed that both the mutant and wild-type PANX1 were expressed at equal levels in 293T cells prepared for electrophysiological measurements (Fig. 4D).
To further assess whether the R217H mutant was dominant to the functional channel properties of PANX1, we co-expressed equal quantities of plasmids encoding both wild-type and mutant PANX1. The ramp currents recorded at ϩ100 mV were reduced by ϳ50% for the mutant compared with wild-type PANX1, although currents were not significantly reduced when both mutant and wild-type PANX1 were co-expressed (Fig. 5, A  and B). Similar findings were observed when channel function was augmented by high potassium (Fig. 5C) suggesting that the R217H mutant was not dominantly inhibiting the function of PANX1. As before, carbenoxolone was effective in blocking PANX1 channel function as well as the residual mutant channel function (Fig. 5C).
Collectively, our functional assays monitoring dye, ATP, and ion flux support that R217H represents a loss-of-channel-function variant. Loss-of-function was evident in cells maintained under basal (i.e. reduced current amplitude in "normal" potassium) and high potassium-stimulated conditions. Loss-offunction with variants targeting channels that operate at the FIGURE 3. R217H mutant exhibits defective dye uptake and ATP release. A, untransfected Ad293 cells or cells expressing wild-type PANX1 or the R217H mutant were immunolabeled for PANX1 (red), and all cells were counterstained for TO-PROா-3 (blue). Bar, 10 m. B, untransfected (UNTR) Ad293 cells or cells expressing wild-type PANX1 or R217H, or expressing wild-type PANX1 and treated with probenecid were subjected to ethidium bromide (EtBr) dye uptake over a period of 40 min. Mean fluorescent measurements revealed that wild-type PANX1-expressing cells were capable of dye uptake, whereas R217H-expressing, untreated cells, or cells expressing wild-type PANX1 and treated with probenecid exhibited significantly reduced dye uptake. *, p Ͻ 0.05; ***, p Ͻ 0.001, n ϭ 3. C, untransfected (UNTR) N2A cells or N2A cells expressing wild-type PANX1 or the R217H mutant were assayed for ATP release upon treatment with a high potassium (K ϩ ) medium containing or lacking the channel blocker CBX. ns, not significant. **, p Ͻ 0.01, n ϭ 3.
cell surface can result from changes in channel function and/or defects in folding and cell surface expression. Evidence demonstrating that R217H expression, glycosylation, and cell surface trafficking is unaltered suggests that reduced current amplitudes observed in the absence of high potassium stimulation can be attributed to reduced channel function. However, with extended high potassium stimulation an additional mechanism could be recruited, namely reduced stimulated delivery of pannexin channels to the cell surface. Although high potassium is reported to allosterically augment pannexin function in a rapidly reversible manner (7,47), a parallel increase in pannexin surface expression via increased forward trafficking could also contribute to augmented pannexin function. In this context, reduced R217H function could be attributed in part to a deficit in stimulated forward trafficking despite unaltered steady state surface expression. To address this possibility, we used cell surface biotinylation to first examine whether high potassium treatment altered the surface expression of R217H relative to that of PANX1. As might be expected, high potassium did not increase the amount of wild-type PANX1 or the R217H mutant at the cell surface of Ad293 or N2A cells (Fig. 6A). Furthermore, we undertook a series of whole-cell recordings from 293T cells expressing PANX1 or R217H, which were pretreated with BFA, to block protein transport from the endoplasmic reticulum to the cell surface. Although current amplitude in cells expressing either R217H or wild-type PANX1 channels was reduced by BFA treatment, likely reflecting reduced surface expression, R217H functional deficits at rest and in the presence of high potassium were maintained (Fig. 6, B and C).

Discussion
The pannexin family of channel-forming glycoproteins has grown in importance given their documented roles in ischemia, stroke, overactive bladder, HIV infections, Crohn's disease, platelet aggregation, and over a half-dozen other diseases (13,28,45,48,49). In most of these cases, PANX1 was identified as being the key pannexin linked to the disease, but the causal mechanism of how PANX1 large-pore channels are associated  R217H mutant (bottom). B, summary of ramp currents recorded at ϩ100 mV from wild-type PANX1 and R217H mutant expressing or mock-transfected (UNTR) HEK293T cells. C, ramp currents at Ϫ60 mV recorded in HEK293T cells expressing wild-type PANX1 and the R217H mutant, before (control) and after treatment with high potassium (High K ϩ ) or high potassium and CBX. D, representative Western blot reveals comparable cellular expression of wild-type PANX1 and R217H channels in whole-cell lysates prepared from sister HEK293T cells used in electrophysiological recordings (position of molecular mass standards are shown). For all panels, the number of cells recorded is indicated in parentheses. p values were calculated in comparison with PANX1 groups using two-way ANOVA analyses with Bonferroni post-tests (B and C), ***, p Ͻ 0.001.
with disease susceptibility remains poorly understood (13). Given the breadth of diseases linked to PANX1 and its ubiquitous distribution, we predicted that loss-or gain-of-function pathogenic variants in the PANX1 gene would likely be causally related to disease, and the disease may involve many organs. To the best of our knowledge, this report is the first and only germline variant identified in any of the three pannexin-encoding genes that has been associated with inherited human disabilities and disease. As anticipated, the proband has extensive and severe multi-organ involvement that has affected her cognition, hearing, skeleton, and reproductive organs. These clinical presentations are consistent with the current literature on Panx1 expression and function in animal studies (3). One of the organs with the highest expression of Panx1 is the brain (21,50). Neurons and glial cells of the central nervous system have been shown to express Panx1 at early stages of development (51,52), and therefore, the expression of a functionally impaired channel may have an effect on neuronal development and differentiation. Similarly, Panx1 expression in the cochlea, primarily in epithelial cells of the organ of Corti, and neurons of the spiral ganglia (49,53), would predict a potentially important function for these channels in hearing. Consistent with severe hearing loss found in the proband, hearing loss was recently reported in a Panx1 conditional knock-out (KO) mouse from the cochlea (54,55). At present, there is evidence of Panx1 expression in osteoblasts (56), but because Panx3 has been the most studied pannexin in bone and cartilage (57), our knowledge of the role of Panx1 in skeletal development remains limited. Likewise, although Panx1 has been reported in the ovaries and male reproductive organs, its functional importance is poorly understood (50,58). However, based on the proband in this study, we suggest that PANX1 plays a role in both the skeleton and reproductive organs, although it likely is not accountable for the proband's short stature as her mother and sister have short stature as well. Also, in the absence of additional patients harboring homozygous PANX1 gene mutations, we cannot definitively assign causal relationships to all the clinical presentations of the proband to PANX1 as this awaits further confirmation.
The severity of developmental abnormalities is in striking contrast to the lack of overt phenotype noted in homozygous FIGURE 5. R217H mutant is not dominant-negative to co-expressed PANX1. A, representative current-voltage relationships recorded in HEK293T cells expressing PANX1, R217H, or both. B, summary of ramp currents recorded at ϩ100 mV from wild-type PANX1, R217H, or PANX1/R217H co-expressing cells. C, ramp currents at Ϫ60 mV recorded in HEK293T cells expressing wild-type PANX1, the R217H mutant, or both, before (control) and after treatment with high potassium (High K ϩ ) or high potassium and CBX. For all panels, the number of cells recorded is indicated in parentheses. *, p Ͻ 0.05; **, p Ͻ 0.01. ns, not significant.

Disease-associated Loss-of-function Panx1 Variant
and heterozygous global Panx1 KO mice (3,59) suggesting that reduced expression by 50% or more does not phenocopy the human R217H variant. However, the lack of overt phenotypes in global Panx1 KO mice may be due, at least in part, to compensation by other pannexins. For example, when Panx1 was constitutively deleted in a Panx1 KO mouse, there was an increase in Panx3 expression in dorsal skin (60). A similar upregulation of Panx3 expression was observed in the wall of thoracodorsal arteries of Panx1 KO mice compared with C57Bl/6 controls (61). In another case, both Panx1 and Panx2 needed to be deleted to observe neuroprotective effects in a mouse model of ischemia because it appeared that one pannexin compensated for the other (62). It should also be noted that the expression of the loss-of-function PANX1 variant that reduces chan-nel function does not equate to the loss of PANX1 expression. For example, we have observed in the connexin channel field that genetically modified mice heterozygous for the GJA1 (Cx43) gene do not exhibit disease, whereas heterozygous mice harboring the I130T mutant, reducing overall Cx43 function to 50% of controls, phenocopy the human disease known as oculodentodigital dysplasia (63)(64)(65).
Our multidimensional molecular analysis of the R217H mutant revealed that PANX1 channels formed from the mutant have an overall functional capacity of only 50% compared with controls. In this context, it is relevant to consider whether the developmental abnormalities are strictly attributed to loss-offunction or whether, alternatively, a gain-of-toxic-function (irrespective of reduced channel flux) could contribute. Of FIGURE 6. A, cell surface biotinylation of wild-type PANX1 and the R217H mutant in the presence and absence of high potassium. Cells were not transfected (NT) or transiently transfected with cDNA encoding either wild-type PANX1 or the R217H mutant. Forty eight hours after transfection, cells were untreated (Ϫ) or treated (ϩ) with 140 mM K ϩ Glu in DMEM for 40 min. Cells were placed on ice, washed in ice-cold Hanks' balanced salt solution, and cell surface proteins biotinylated. Biotinylated proteins were precipitated using neutravidin-conjugated beads, subjected to SDS-PAGE, transferred to nitrocellulose, and immunoblotted (IB) for PANX1 or GAPDH. Note the abundance of cell surface biotinylated PANX1 and the R217H mutant, although the lack of biotinylated GAPDH indicates that only cell surface proteins were biotinylated. Position of molecular mass standards are shown. B, R217H mutant showed a dramatic reduction in channel function when expressed in HEK293T cells pretreated with BFA for 6 h while CBX effectively blocked channel function. C, after pre-treatment with BFA for 6 h, ramp currents at Ϫ60 mV were recorded in HEK293T cells expressing wild-type PANX1 or the R217H mutant, before (control) and after treatment with high potassium (High K ϩ ) or high potassium and CBX. n ϭ 6 for each panel. ***, p Ͻ 0.001. note, expression of the R217H mutant in five distinct cell lines was well tolerated with no apparent detrimental effect on cell viability or morphology, suggesting that this latter scenario is unlikely. Thus, we propose that the R217H mutant disrupts some essential function of pannexin channels critical for development in a variety of tissue types. Because the proband's parents, as well as her sister, are heterozygous for R217H and exhibit no symptoms that can be linked to the variant, it would suggest that the disease is autosomal recessive in nature. Although the allele frequency of the R217H variant is not specifically known in the Turkish population, it has been reported in 3/8596 European alleles in the Exome Sequencing Project and 7/66,102 European alleles and 1/16,412 South Asian alleles in the Exome Aggregation Consortium. Consistent with the proband's parent and sister not exhibiting the clinical symptoms of the proband, we found that the mutant was not dominant to the function of PANX1 when co-expressed in cultured cells.
Given that this patient is the first reported individual with biallelic PANX1 gene variants, and that WES in any individual will reveal variants in numerous genes, we assessed the evidence that the PANX1 variants are responsible for this individual's features. A review of the full WES findings reveals variants in three other genes known to be related to a human disease phenotype as follows: a single pathogenic variant in the autosomal recessive RDX gene associated with a form of nonsyndromic hearing loss, inherited from her unaffected mother; a missense variant in FGFR3, an autosomal dominant gene responsible for various forms of skeletal dysplasias, inherited from her unaffected father; and a single missense variant in the autosomal recessive POR gene, associated with disorders of steroidogenesis, inherited from her asymptomatic mother. None of the phenotypes associated with these genes are similar to the proband's presentation, and two of the three genes associated with autosomal recessive inheritance are not located in our patient's regions of loss of heterozygosity, and have only one variant found. In addition, despite the large runs of homozygosity seen on chromosome microarrays, WES only identified three other genes of interest with homozygous variants in addition to PANX1 (R217H): PHF12 (R765W), MRPL49 (R88C), and NRXN2 (L53F). PHF12 encodes a PHD finger protein that has been proposed to be associated with the regulation of intraocular pressure, which is not associated with our proband's phenotype (66). MRPL49 encodes the mitochondrial ribosomal protein L49, which is involved in protein synthesis (67). This gene is associated with glossopharyngeal neuralgia, a condition characterized by repeated episodes of severe pain in the tongue, throat, ear, and tonsils, conditions not experienced by the proband in this study. The NRXN2 gene is expressed in the brain and encodes a synaptic organizing protein that helps to mediate the differentiation of certain synapses. At least one loss-offunction truncating mutation of this gene has been identified in the heterozygous state in a family with autism spectrum disorder, language delay, and a family history of schizophrenia, suggesting autosomal dominant inheritance (68). The autosomal dominant mode of inheritance proposed by this and other studies does not fit in this family as both parents are carriers of the missense change and are unaffected. Furthermore, the NRXN2 proband did not present with the scope of organ involvement noted in the proband of this study. Although this provides a compelling argument to rule out the involvement of NRXN2, protein structure/function studies also provide supportive evidence against a pivotal role for NRXN2. In brief, NRXN2 encodes a type I transmembrane protein with six conserved interacting LNS domains (laminin/neurexin/sex hormone) (69 -71). The LNS domains are key for interactions between neurexin and several postsynaptic binding partners that have been described (e.g. neuroligins, leucine-rich repeat transmembrane protein, and dystroglycan). Whereas LNS6 is the most important domain in terms of binding, the proband's variant is positioned in the first LNS region, with no known interacting partners. Therefore, LNS1 is predicted to not be a functionally relevant domain.
This analysis leaves us with PANX1 as the best gene candidate for the proband's multi-organ involvement, although given the fact that other homozygous and heterozygous variants exist in several genes, we await substantiation of this case study by the identification of other patients that harbor PANX1 variants. A PANX1 gene search of the Online Mendelian Inheritance in Man (OMIM) database did not reveal any PANX1 gene mutations linked to any known disease. In addition, OMIM was searched for conditions that include all of the major and most striking features of the proband's phenotype, namely intellectual disability, primary ovarian insufficiency, and sensorineural hearing loss, to make sure that the genes associated with these other conditions had good coverage on WES and did not contain any variants of uncertain significance. The main conditions that matched the proband's phenotype on this search included the Perrault syndromes and Woodhouse-Sataki syndrome. No suspicious variants were found in any of these genes associated with these conditions on WES. Overall, our findings fit well with the broad spectrum of PANX1 expression in all of the proband's affected organs (13) and to the fact that a new conditional mouse model of Panx1 ablation in the cochlea revealed severe hearing loss (54,55). We suggest that PANX1 be considered as a rare cause of intellectual disability, particularly when accompanied by other systemic dysfunction, especially sensorineural hearing loss and premature ovarian failure. Finally, as this is the first report of a disease-associated PANX1 germline variant, it is possible that a broader contribution of PANX1 to human disease has previously been overlooked. Accordingly, this study may raise awareness of PANX1 as a candidate gene contributing to the genetic basis of disease.
Author Contributions-K. L. and A. S. identified the patient, obtained informed consent, and worked closely with the family to acquire all the clinical information. J. J. and K. L. L. evaluated all the gene sequencing. Q. S., J. K., J. L. E., and S. P. performed the expression, ATP release, biotinylation, and dye uptake studies. R. S. and M. F. J. performed all the electrophysiology studies. D. W. L. oversaw the project and wrote the manuscript with assistance from all authors, especially K. L., J. J., and M. F. J.