Functional and clinical studies reveal pathophysiological complexity of CLCN4-related neurodevelopmental condition

Missense and truncating variants in the X-chromosome-linked CLCN4 gene, resulting in reduced or complete loss-of-function (LOF) of the encoded chloride/proton exchanger ClC-4, were recently demonstrated to cause a neurocognitive phenotype in both males and females. Through international clinical matchmaking and interrogation of public variant databases we assembled a database of 90 rare CLCN4 missense variants in 90 families: 41 unique and 18 recurrent variants in 49 families. For 43 families, including 22 males and 33 females, we collated detailed clinical and segregation data. To confirm causality of variants and to obtain insight into disease mechanisms, we investigated the effect on electrophysiological properties of 59 of the variants in Xenopus oocytes using extended voltage and pH ranges. Detailed analyses revealed new pathophysiological mechanisms: 25% (15/59) of variants demonstrated LOF, characterized by a “shift” of the voltage-dependent activation to more positive voltages, and nine variants resulted in a toxic gain-of-function, associated with a disrupted gate allowing inward transport at negative voltages. Functional results were not always in line with in silico pathogenicity scores, highlighting the complexity of pathogenicity assessment for accurate genetic counselling. The complex neurocognitive and psychiatric manifestations of this condition, and hitherto under-recognized impacts on growth, gastrointestinal function, and motor control are discussed. Including published cases, we summarize features in 122 individuals from 67 families with CLCN4-related neurodevelopmental condition and suggest future research directions with the aim of improving the integrated care for individuals with this diagnosis.


INTRODUCTION
CLCN4 encodes the intracellularly located chloride/proton ionexchanger ClC-4, and is located on the human X chromosome at Xp22.2. Rare inherited or de novo missense and truncating variants are identified in a growing number of males and females with a range of neurodevelopmental and psychiatric complications. However, the establishment of the pathogenicity of previously unreported rare missense variants remains challenging. As of 22nd May 2022, from the 153 missense CLCN4 variants listed in the publicly available database ClinVar, 73% (111) were classified to be of uncertain significance. Without clear establishment of pathogenicity, families remain on a diagnostic odyssey, cannot make fully informed reproductive choices, or benefit from advances in condition-specific management guidelines or targeted therapies.
The first CLCN4 variant was reported in an infant male with developmental and epileptic encephalopathy and suggested CLCN4 as a novel candidate disease gene [1]. Three years later, as part of an X chromosomal exome sequencing study, our group demonstrated that truncating and missense variants were associated with a neurocognitive phenotype in males in five unrelated families [2]. Two families had linkage intervals including Xp22: A two generation French family with five affected males with severe to profound intellectual disability (ID) and variable behavioral difficulties was reported by Raynaud et al., in 1996 [3] and a Belgian family with five males spanning two generations with ID, challenging behaviors and autistic features described by Claes et al. [4]. Heterozygous females in those families were neurotypical or had a mild neurocognitive/psychiatric phenotype. Therefore, a phenotypic entity of X-linked recessive ID (Raynaud-Claes Syndrome) was proposed (MIM *300114).
Subsequently, we reported 10 additional families consisting of 29 hemizygous males and 23 heterozygous females [5]. We clarified that all males had a core phenotype of mild to severe ID, with considerable intrafamilial heterogeneity. For the first time, we reported the phenotype in females with de novo variants, which overlapped in severity with that of males. Other common clinical features included epilepsy, subtle white matter changes on neuroimaging, autism spectrum disorder, challenging behaviors, and mental health complications including bipolar disorder, depression, and anxiety. More recently, an additional six males with CLCN4-related neurodevelopmental condition were reported confirming the core feature of ID and common comorbidities of epilepsy and challenging behaviors [2,6,7]. Xu et al., reported on a female with ID, autistic features and brain abnormalities, with a maternally inherited CLCN4 missense variant where the mother had mild ID [8]. We recently summarized the published genotypic and phenotypic spectrum [9], noting that, to date, all CLCN4 variants studied in the Xenopus expression system demonstrated partial or complete loss-of-function (LOF) [1,10,11].
ClC-4 is one of the nine members of the CLC gene family encoding anion-transporting membrane proteins [12]. CLC proteins are divided into two groups: four members (ClC-1, ClC-2, ClC-Ka, and ClC-Kb) are Cl − channels localized in the plasma membrane, while the remaining CLCs (ClC-3 to -7) are secondary active Cl − /H + antiporters physiologically localized in intracellular endo-/lysosomal membranes; the latter are also called vesicular CLCs (vCLCs). Among the vCLCs, ClC-3 to -5 are highly homologous and are localized to endosomes, while the more distantly related ClC-6 and ClC-7 are localized to late endosomes and lysosomes, respectively [12]. The vesicular Cl − /H + antiport activity is important for ionic homeostasis of endo-/lysosomes by assisting in vesicular acidification and increasing luminal Cl − concentration. The function of ClC-4 critically depends on the highly related ClC-3 transporter, with which it forms heterodimers [13,14]. While most CLCs are physiologically homodimeric, ClC-4 appears to preferentially associate with ClC-3, whereas ClC-4 homodimers are biochemically relatively unstable [13,14].
ClC-4, and other members of this protein family, ClC-3, ClC-6, ClC-7, and Ostm1, an obligatory subunit of ClC-7, are implicated in neurological disorders [2,5,12,15,16]. This could be postulated to be related to the postmitotic nature of neurons and their heavy reliance on vesicular trafficking. For example, mice lacking late endosomal ClC-6 transporters show signs of lipofuscin accumulation [17], and lacking lysosomal ClC-7 exhibit a severe lysosomal storage phenotype, respectively [18]. Recently a recurrent gain-offunction (GOF) variant reported in CLCN6 caused the severe neurodegenerative disease CONRIBA (Neurodegeneration, childhood-onset, hypotonia, respiratory insufficiency and brain imaging abnormalities CONRIBA; MIM 619173) [15] while a variant found in a patient with clinical features of late-onset neuronal ceroid lipofuscinosis [17] was found to have greatly reduced functional activity [19]. LOF of ClC-3 in mice leads to neurodegeneration [20] and both GOF and LOF CLCN3 variants in humans cause severe global developmental delay [16]. Conversely, knock-out mouse models of ClC-4 have no overt phenotype [21], implying a complex causative mechanism that requires further exploration to understand the pathophysiological basis of CLCN4-related neurodevelopmental condition.
Understanding the pathogenicity of missense variation in CLCN4 both clinically and functionally is therefore the next step [6]. We firstly undertook a collaborative study aiming to further characterize the genotypic and phenotypic spectrum of CLCN4related neurodevelopmental condition in both males and females. Secondly, we studied the functional impact of novel and previously reported missense variants in heterologously expressing Xenopus oocytes by employing electrophysiological measurements using extended voltage-protocols.

SUBJECTS AND METHODS Subjects
We collected de-identified detailed clinical data on 55 individuals from 43 previously unreported families with (presumed) CLCN4-related neurodevelopmental condition, including individuals from three families where the proband had a blended clinical phenotype with a second genetic diagnosis. Data were obtained through an international collaborative process wherein clinicians and diagnostic laboratories with variants identified in CLCN4 contacted our team, and we also contacted the laboratory or clinician who had deposited variants in CLCN4 in the public databases DECIPHER, ClinVar, and LOVD [22][23][24]. In each participating center, written informed consent was obtained from the individual's legal guardians before genetic testing as approved by relevant local ethical committees. Clinical information was obtained by review of medical records and examination of affected individuals. Written informed consent for the publication of clinical data and photographs was also obtained from the participants' legal guardians.

Expression construct
The human ClC-4 cDNA was cloned in the pTLN expression vector [25], in which the disease-associated variants were introduced using standard restriction-free mutagenesis. All constructs were verified by Sanger sequencing.

Expression in oocytes
RNA was transcribed using the SP6 mMessageMachine kit (Thermofisher, Milan, Italy) after linearization with MluI. Xenopus laevis oocytes were injected with~6 ng of RNA and incubated at 18°C for 2-5 days prior to measurements as described previously [26].

Two electrode voltage clamp recordings
Recording pipettes were filled with 3 M KCl (resistance about 0.6 MOhm) and currents were recorded using a TEC03 two electrode voltage clamp amplifier (npi electronics, Tamm, Germany). Ground electrodes were connected to the bath via agar bridges. The standard extracellular solution contained 100 mM NaCl, 5 mM MgSO 4 , 10 mM HEPES (pH 7.3). For solutions at pH 6.3 and 5.3, HEPES buffer was replaced by MES (2-(nmorpholino) ethanesulfonic acid) buffer. pH was adjusted with NaOH. Currents were acquired using the custom GePulse acquisition program and an itc-16 interface (Instrutech, Colorado, USA), filtering at 5 kHz and sampling at 50 kHz. Two types of stimulation protocols were applied from a holding potential of −30 mV. The first consisted of 10 ms pulses to voltages ranging from +160 to −120 mV (in 20 mV steps) without leaksubtraction. The second protocol consisted of steps ranging from +170 to −10 mV (in 10 mV steps), applying linear leak and capacity subtraction using a 'P/4' leak subtraction protocol from the holding potential −30 mV. For this procedure 4 pulses of ¼ of the regular amplitude were applied towards negative voltages, their response was averaged, adequately scaled, and subtracted. This procedure approximately eliminates linear capacitive currents and 'leak', assuming that ClC-4 is inactive at negative voltages. from > =6 non-injected oocytes from the same batch were subtracted. For the average IV curves, currents were normalized to the current measured for WT from the same batch at 170 mV, and data from at least four injections for each construct were averaged. For the average ratios of mutant versus WT currents at a given voltage, data, currents were normalized to the respective current measured for WT. This procedure highlights possible alterations of the voltage-dependence. A voltageindependent reduction (or increase) in current size would result in a voltage-independent ratio. We interpret alterations of the voltagedependent rectification as a change of a gating process that depends on both subunits. Such a gating process is clearly present in ClC-6 and ClC-7 transporters [19,27] and similarly most likely underlies the extreme rectification of ClC-3, ClC-4, and ClC-5 [28]. In agreement with this hypothesis, practically all variants found here that lead to an apparent shift of the voltage-dependence to more positive voltages are located close to the dimer interface.
For data analysis of currents measured at various external pH values, the following leak-subtraction was performed. For each oocyte, currents measured at pH 7.3 were fitted in the range −120 mV <= V <= 0 mV with a straight line. The line was extrapolated to all voltages and subtracted from the current-voltage relationships (IVs) measured in the various conditions, and normalized to the current at pH 7.3, 160 mV. This is because for WT ClC-4 and for most variants, at pH 7.3, currents recorded at voltages V <= 0 mV are very small and indistinguishable from currents in uninjected oocytes and represent a mixture of leak and endogenous currents. Similar to the "voltage-shifted" variants, we interpret the emergence of inward currents at acidic as a partial disruption of the gating process that in WT keeps the transporter inactive at negative voltages, similar to what described for CLCN3 variants [16]. Error bars in all figures represent SEM. Statistical significance was assessed by Student's unpaired two-tailed t-test.
Variance is similar between all groups because the same batches of oocytes were utilized for WT and variant measurements.

RESULTS
Detailed clinical data were analyzed on 55 previously unreported individuals, 22 hemizygous males and 33 female heterozygotes, from 43 previously unreported families, as well as updated clinical information on one previously reported female who was now recognized to have a recurrent variant [5]. The 44 families were divided into five groups (A-E). This includes families with missense variants, who were divided into groups A-D based on the functional results obtained in the Xenopus laevis oocyte model for the CLCN4 missense variants as described below, as well as three additional patients with novel truncating variants (Group E). Demographic details of these affected individuals, the CLCN4 variants, their frequency in the gnomAD database, in silico pathogenicity predictors, and results of in vitro functional studies in Xenopus oocytes are presented in Table 1. Table 1 also includes the details of variants from the public database ClinVar which we investigated with in vitro functional studies in Xenopus oocytes but where we were unable to obtain consent to publish clinical data, as well as variants from ClinVar (as of 25 th May 2022) and publications which were recurrent with our investigated variants. New ClinVar accession numbers were obtained for any variant with functional data not already listed in ClinVar and added to Table 1. Figure 1 shows the pedigrees of the unreported families with a novel inherited CLCN4 variant previously unpublished in the peer-reviewed medical literature. Figure 2 shows clinical photographs and MRI brain images. Figure 3A is a schematic drawing of the CLNC4 gene and ClC-4 protein with all variants of clearly affected males and females with clinical information available (this study and published). More details on the clinical presentation of these previously unreported families are detailed in Supplementary Table 1  To test for possible impact of variants on the electrophysiological properties of the ClC-4 Cl − /H + antiporter, missense changes as present in the affected individuals were introduced in ClC-4 expression constructs and studied by the 2-electrode voltageclamp recording method in Xenopus oocytes. Example recordings are shown in Fig. 4A and all results in Table 1 and Supplementary  Fig. 1. WT ClC-4 shows typical outwardly rectifying currents as described [29,30]. An overview of all variants investigated functionally in this study are shown in a topology model and in a three-dimensional model in Fig. 3B, C.
Group A consists of 21 previously unreported families with detailed clinical data (see Supplementry Table 1) whose functional studies are part of this study and demonstrated a LOF, reduced function or shift of voltage dependence, as detailed below. These families included 13 with a male proband and eight with a female proband. In Table 1, Group A also includes seven families that we previously reported on with loss or reduced function [5]. It also includes data from 23 families whose variant was recurrent with one in our cohort, for whom only limited data was available in public databases (ClinVar, DECIPHER, LOVD) as of the 25th May 2022, or from publications from other groups [7,8,11,31]. We included these families as further evidence of the pathogenicity of these recurrent variants. Some variants, for example p.(Val92Met), showed current levels that were barely above those seen in un-injected oocytes ( Fig. 4A-C). A similar near complete loss or reduced function was observed for other variants e.g., p.(Lys62Arg), p.(Ser278Arg), p.(Gly342Glu), and p.(Gly484Arg) (for full list see Table 1, Supplementary Fig. 1). As the variant p.Gly731Val affected the last amino acid of exon 12, we analyzed if this variant impacted splicing ( Supplementary Fig. 2), but this could not be demonstrated. Little mechanistic insight can be obtained from these LOF variants as we did not analyze for example if protein stability was affected.
Other variants, for example p.(Asn309Ser), showed a reduced expression level, but no sign of altered voltage-dependence ( Fig. 4A-C). The lack of altered voltage-dependence is highlighted in Fig. 4C, which shows that the ratio of currents mediated by variant p.(Asn309Ser) and currents of WT ClC-4 has a practically voltage-independent value of~0.25. A similar, partially reduced function was observed for variants p.(Ile374Thr) and p.(Gln489Lys) ( Table 1, Supplementary Fig. 1).
In contrast to the voltage-independent reduction seen in the variants described above, several other variants, including p.(Leu276Phe), showed a "right-shifted" voltage dependence. This is difficult to appreciate by just comparing the raw current traces (Fig. 4A) or the average current-voltage relationship ( Fig. 4B) but is clear in Fig. 4C. For p.(Leu276Phe), the ratio of currents compared to WT is small for V~20 mV but progressively enlarged at more positive voltages. The reduction of currents at "physiological" voltages is overcome by sufficiently large positive voltages. This essential LOF phenotype likely reflects an effect on the gating process of ClC-4, as detailed in Subjects and methods. Similar LOF by apparently right-shifted gating was observed to various degrees for variants p.  Table 1, Supplementary Fig. 1).
Group B. This group includes nine missense CLCN4 variants from 17 independent families: 14 were previously unreported, one, a female with the de novo variant p.(Ala555Val), was previously reported by our group [5], and four families with the same variants or amino acid mutated, that were included in public databases but for whom we could not obtain detailed clinical data. These variants were grouped, as they showed compelling clinical evidence for pathogenicity (rarity, de novo status, matching clinical phenotype, and recurrence across unrelated families), without gross effect at the regular recording conditions at pH 7.3. However, p.(Ile549Asn) and some other variants exhibited a characteristic alteration (Fig. 4C): the ratio of currents compared to WT became progressively larger towards more negative voltages. This behavior is reminiscent of a GOF effect described for CLCN3 variants [16]. Indeed, closer inspection of this and other variants revealed a dramatic GOF that is apparent particularly at acidic

GROUP E: Truncating variants
Data presented include genomic coordinates, reporting laboratory assessment of pathogenicity using ACMG criteria as reported in ClinVar or determined by the authors using VARSOME prior to functional studies were conducted, demographic details, inheritance, recurrence within families, recurrence across families, including public databases as of 25th May 2022, selected in silico pathogenicity scores, frequency in gnomAD database, functional impact in Xenopus oocyte model, and if the individual has more than one genetic diagnosis (blended phenotype). Data which are supportive of pathogenicity is color-coded orange (with darker orange for most supportive data), data which are not supportive of pathogenicity are coded green. ACMG American College of Medical Genetics and Genomics, CBS cystathionine β-synthase, NA not applicable, NR not reported; N-term N terminus, GOF gain of function, LOF loss-of-function, LOVD Leiden Open Variation Database, MPS massively parallel sequencing, WT wild type. In silico scores include PolyPhen, CADD, REVEL and SpliceAI. extracellular pH (Fig. 4E-G). While outward currents of WT ClC-4 were slightly inhibited at acidic pH and inward currents remained undetectable, comparably large inward currents became visible at pH 6.3 and 5.3 for variant p.(Ile549Asn) (Fig. 4D, E). A quantitative analysis revealed that similarly large inward currents were seen for variants p.(Phe268Leu) and p.(Ala555Val) (Fig. 4B, C). Smaller, but highly significant inward currents were also detected for variants p.  (Fig. 4F, G and Supplementary  Fig. 1). Evidently, these variants partially disrupted the gating process of ClC-4 that normally prevents inward currents even at very acidic pH. For variants p.(Phe268Leu) and p.(Ile549Asn) inward currents were large enough to estimate reversal potentials at pH 6.3 and pH 5.3. The fact that the reversal potential in these conditions differed by about 12.5 mV for both variants (Fig. 4H) demonstrates that the inward currents carried by these variants are at least partially mediated by H + transport. However, the difference falls short of the expected value of~20 mV for a coupled 2Cl − /1H + antiporter [32], suggesting that currents mediated by the variants are at least partially uncoupled. More detailed studies will however be needed to determine precise transport stoichiometry for these variants as well as for WT ClC-4.
Group C consisted of families with variants p.(Asp34Asn), p.(Ser69Leu) and p.(Pro369Leu). Although these variants all showed a functional LOF similar to those in Group A (Supplementary Fig. 1), the affected individuals had more complex clinical presentations (Supplementary Fig. 3) and an additional genetic condition was proven or strongly suspected, consistent with a blended phenotype. Consequently, they were separated from the other groups, and not included in the clinical summary in Table 2.  Table 1, Supplementary Fig. 1). It is plausible that these variants are pathogenic by a mechanism not modelled in our cellular system, but given the lack of evidence on their pathogenicity, the families with Group D variants were not included in the summary Table 2.
Group E consisted of three individuals with a frameshift or nonsense variant in CLCN4 for whom detailed clinical data were available.
This study thus brings the total number of individuals with (likely) pathogenic variants in CLCN4 to a total of 122: 58 males and 64 females. For 20 of the females, parental studies demonstrated the variant to be de novo, while the other 44 females were identified as being heterozygous for a CLCN4 variant only after a relative (usually a son, but on two occasions a daughter) was identified in their family to have CLCN4-related condition [1,2,5,7,11,31]. The clinical features of this expanded cohort are summarized in Table 2.

DISCUSSION
Our study addresses the interpretation of novel missense variants, a common clinical conundrum across clinical genetic practice [33]. We robustly demonstrate a much wider range of functional impacts of CLCN4 variants in the Xenopus oocyte model than had been previously demonstrated. In addition, we provide new insights into the common clinical features of CLCN4-related neurodevelopmental condition, which have enabled us to provide updated clinical management advice to clinicians [9] and improved patient and family education via the patient advocacy group CureCLCN4.
We confirm that cognitive disability is the most common clinical feature in males, most commonly in the moderate or severe/ profound range (Table 2 and Supplementary Table 1). For the first time, however, we report a male with a verbal IQ in the normal range. This 12-year-old male (Family A21; p.(Phe319Ser)mat; functional studies: LOF by shifted voltage dependence) had a verbal IQ of 90 on formal psychometric testing (WISC-II-NL) but did have a lower performance IQ (61: within the mild ID range) and significant comorbidities with delayed language acquisition, articulation difficulties, severe treatment-resistant epilepsy, autistic features, and hyperactivity. We also report the first observation of a male with CLCN4-related condition and mild ID who has had a family (Family A3; p.(Val92Met); functional studies: almost complete LOF). He has two daughters who are obligate heterozygotes, one with mild ID and the other with specific learning disabilities.
Phenotypic prediction of cognitive function in females with CLCN4-related condition is very difficult with a wide spectrum of severity of neurodevelopmental and medical issues, including about half of heterozygous female carriers being apparently completely unaffected (Table 2). In general, females with a de novo variant had a more severe phenotype than those with an inherited variant. However, this observation is far from absolute, as evidenced by several female individuals in the cohort. For example, in families A10 and A24, the mother of a severely affected male had a de novo variant and yet was completely unaffected. On the other hand, we report females with inherited variants who have severe phenotypes: for example, the proband in Family A40 had moderate ID and a missense variant (p.(Pro635Arg); functional studies: LOF), which she inherited from a mother with mild ID. The proband had no additional genetic condition identified by WGS, and there was no evidence of mosaicism in the unaffected mother. We have previously reported that X-inactivation status does not correspond to clinical severity [5], and, as demonstrated across this and previous studies [8,31], female-to-female inheritance from a very mildly, or even apparently non-affected mother does not ensure a mild phenotype in the daughter. Clinically, were a de novo missense CLCN4 variant to be detected on a prenatal exome in a female embryo, there would remain a degree of uncertainty whether there would be a neurodevelopmental phenotype postnatally. From our previous study, it was apparent that females with a CLCN4 frameshift or nonsense variant or a small intragenic chromosomal deletion of CLCN4 are typically unaffected [5]. This was confirmed in the current study: both female carriers in the two families with inherited truncating variants in Group E were unaffected (Family E1 and E2). This observation may signify that the impact of missense variants in females could lead to a toxic gain-of function or a LOF that could be at least partially imparted also on ClC-3/ClC-4 heterodimers.
Behavioral and mental health disorders are the next most common clinical features. The four most common conditions were attention deficit hyperactivity disorder (ADHD) or significant hyperactivity, impulsiveness, or restlessness affecting 59% of all males and 46.7% of females with de novo variants; autism spectrum disorder (or autistic behavior) affecting 54.5% of all males and 40% of females with de novo variants; angry outbursts or challenging behaviors, affecting 36.4% of males and 26.7% of females, and lastly anxiety, affecting 27.2% of all males, 53% of females with de novo variants and 10.5% of females with inherited variants or variants with unknown inheritance. The mental health conditions were reported to significantly impact the affected individual's ability to learn and their quality of life. Less frequent mental health disorders included obsessive compulsive disorder and depression/ bipolar disorder, which commonly had onset in late teenage years or early adulthood and caused a significant deterioration in quality of life. This highlights the need for close monitoring of all individuals for psychiatric complications, with appropriate referral to a psychiatrist skilled in the management of individuals with neurodevelopmental conditions. Epilepsy is also confirmed as a significant feature of CLCN4related neurodevelopmental condition, affecting 59% of all males and 20% of females. Most individuals with epilepsy had seizure onset within the first three years of life, although two were diagnosed at age 13, highlighting the need for ongoing seizure surveillance beyond childhood. Seizure semiologies were broad, including generalized absence and tonic-clonic seizures and focal onset seizures, as evidenced by EEG showing focal onset in some and generalized onset in others. Epilepsy can be severe, consistent Fig. 4 Expression of CLCN4 variants in Xenopus oocytes. Panel A shows example recordings of the indicated constructs evoked by the voltage-clamp protocol indicated in the inset and using a "P/4" leak subtraction protocol (see Methods). Scale bars apply to all constructs. B shows average normalized IV relationships of the same variants. Currents are normalized to that of WT at 170 mV as described in Methods. In C, currents are normalized to the current of WT at the same voltage (see Methods). Data points are significantly different at practically all voltages from the value of 1 (i.e., WT) for all indicated variants (p < 0.05). Panel D shows typical current traces recorded without leak subtraction of WT ClC-4 in the presence of neutral and acidic extracellular pH, with outward currents being inhibited and inward currents remaining at a negligible level [29]. E illustrates the pH response of variant p.(Ile549Asn), which shows the activation of relatively large inward currents at acidic pH. F, G quantitative analysis of pH dependence of indicated GOF variants. Currents recorded at pH 5.3 were normalized to values measured at pH 7.3 as described in methods. The GOF effect of variants p.(Val317Ile) and p.(Ser395Arg) becomes apparent in panel G that shows the same data as panel F at a magnified scale. Data points are significantly different at voltages <= −40 mV from WT for all indicated variants (p < 0.05). Panel H shows differences in reversal potential measured at pH 6.3 and pH 5.3 for variants p.(Phe268Leu) and p.(Ile549Asn). The red line indicates the value expected for a stoichiometrically coupled 2 Cl − /1 H + antiporter. This table excludes patients with rare missense variants from Group D, for whom the functional studies were similar to wild type, and patients from Group C that had a more severe phenotype due to an additional monogenic condition.
studies, but a more complex clinical phenotype, which we could attribute to a likely, or confirmed, blended genotype due to two monogenic conditions. For example, the male proband in Family C1 (p.(Asp34Asn)); whose functional studies were consistent with a LOF of ClC-4, has short stature and distinctive skeletal and facial features consistent with a diagnosis of Desbuquois dysplasia (XYLT1-related) that he shares with his sister. However, he has significant ID, epilepsy, and autism spectrum disorder, which are atypical for Desbuquois syndrome, and thus most likely has a blended phenotype of Desbuquois dysplasia and CLCN4-related neurodevelopmental condition. The finding of four patients with a blended phenotype due to suspected or proven multi-locus pathogenic variation in a total cohort of 122 individuals with CLCN4 variants (4/122: 3.3%) is consistent with other studies estimating this phenomenon occurs in about 5% of individuals with an identified diagnosis after unbiased sequencing [39]. It emphasizes that for affected individuals, whose clinical features are not entirely consistent with their diagnosed monogenic condition, broadening the scope of genomic sequencing to an unbiased exome or whole genome sequencing approach may be appropriate to look for additional pathogenic findings In summary, our study considerably expands our knowledge of the range of phenotypic and genotypic variation in CLCN4-related condition and for the first time robustly demonstrates a range of functional impacts, including gain of function. Variant classification still remains a nuanced art, rather than a precise science [40]. Fully informed genetic counselling is required to guide families through the diagnostic limitations and uncertainties inherent in genetic testing for neurodevelopmental conditions [41]. Several research priorities remain. We need to better ascertain the causality of all rare missense variants to elucidate targeted treatments. Establishment of a robust animal model is an urgent priority. This could potentially be a rat model, given that Clcn4 is on the X chromosome in the rat, as opposed to in the mouse where it is autosomal [42]. A high throughput functional and therapeutic assay system, such as neuronal micro-electrode assays, which have been successfully applied in other neurodevelopmental conditions [43], would also be very helpful. With recent inclusion of CLCN4 in the SFARI gene project [44] scientists and clinicians are working together to better understand and manage this condition.