TMEM151A variants associated with paroxysmal kinesigenic dyskinesia

TMEM151A, located at 11q13.2 and encoding transmembrane protein 151A, was recently reported as causative for autosomal dominant paroxysmal kinesigenic dyskinesia (PKD). Here, through comprehensive analysis of sporadic and familial cases, we expand the clinical and mutation spectrum of PKD. In doing so, we clarify the clinical and genetic features of Chinese PKD patients harboring TMEM151A variants and further explore the relationship between TMEM151A mutations and PKD. Whole exome sequencing was performed on 26 sporadic PKD patients and nine familial PKD pedigrees without PRRT2 variants. Quantitative real-time PCR was used to assess the gene expression of frameshift mutant TMEM151A in a PKD patient. TMEM151A variants reported to date were reviewed. Four TMEM151A variants were detected in four unrelated families with 12 individuals, including a frameshift mutation [c.606_607insA (p.Val203fs)], two missense mutations [c.166G > A (p.Gly56Arg) and c.791T > C (p.Val264Ala)], and a non-pathogenic variant [c.994G > A (p.Gly332Arg)]. The monoallelic frameshift mutation [c.606_607insA (p.Val203fs)] may cause TMEM151A mRNA decay, suggesting a potential pathogenic mechanism of haploinsufficiency. Patients with TMEM151A variants had short-duration attacks and presented with dystonia. Our study provides a detailed clinical description of PKD patients with TMEM151A mutations and reports a new disease-causing mutation, expanding the known phenotypes caused by TMEM151A mutations and providing further detail about the pathoetiology of PKD.


Introduction
Paroxysmal kinesigenic dyskinesia (PKD, MIM 128200) is characterized by episodic involuntary movements including dystonic postures, chorea, athetosis, or ballism affecting the limbs and/or face. PKD usually starts in late childhood or early adulthood, at which time affected individuals attempt to initiate movements after a period of being static and experience overflow muscle activation worsened by voluntary action. Primary PKD can be sporadic, although most reported cases are familial and inherited in an autosomal dominant fashion with incomplete penetrance (Tomita et al. 1999;van Vliet et al. 2012).
Next-generation sequencing techniques have expedited the discovery of several novel autosomal dominant PKD genes including PRRT2, the main gene implicated in familial PKD and infantile convulsions and choreoathetosis (ICCA) (Ebrahimi-Fakhari et al. 2015;Heron et al. 2012;Lee et al. 2012). Other genes related to the pathogenesis of PKD such as PNKD, SLC2A1, SCN8A, KCNMA1, KCNA19, and DEPDC5 have also been reported (Erro et al. 2014;Gardella et al. 2016;Gardiner et al. 2015;Huang et al. 2020;Tian et al. 2018;Wang et al. 2016;Yin et al. 2018). These efforts have provided an increasing proportion of previously undiagnosed PKD patients with a molecular diagnosis. Nevertheless, approximately half of patients with primary PKD do not harbor mutations in any of these genes, suggesting that other disease-causing genes exist. TMEM151A, located at 11q13.2 and encoding transmembrane protein 151A, was first reported as causative for PKD in 2021 . Subsequent studies have shown that patients with Hua lin Huang, Qing xia Zhang, and Fei Huang are co-first authors.
1 3 TMEM151A variants are phenotypically different to patients with PRRT2 variants Ma et al. 2022;Tian et al. 2022;Wang et al. 2022;Wirth et al. 2022), the former more likely to be sporadic, have shorter attacks, and present with dystonia . Like the PRRT2 protein, TMEM151A is also highly expressed in the brain, markedly increasing in expression postnatally, peaking at postnatal day 14, and then declining in adulthood. Biologically, loss-of-function TMEM151A variants might play an important role in regulating synaptic transmission and neuronal excitability .
However, the mechanistic basis of TMEM151A-related PKD is unclear, and the phenotypes and inheritance patterns of different TMEM151A mutations need further clarification. Here we studied 26 sporadic patients and nine familial pedigrees meeting clinical criteria for PKD but without PRRT2 mutations (Cao et al. 2021). To further explore the relationship between TMEM151A mutations and PKD, we screened our cohort for TMEM151A mutations and analyzed the clinical manifestations and genetic features.

Patients
Twenty-six sporadic PKD probands and nine familial PKD pedigrees without PRRT2 variants were recruited from the Department of Neurology, Second & Third Xiangya Hospital affiliated to Central South University, China. A neurologist reviewed all clinical information on patients with a putative diagnosis of PKD. Patients were included in the study only if they fulfilled the criteria for PKD defined by Cao et al. (2021). For patients with a TMEM151A mutation, clinical data were retrospectively extracted from their medical records including age of onset, family history, type of trigger, attack duration and frequency, phenotype, laterality, aura, and response to treatment. The Ethics Committee of the Third Xiangya of Central South University approved the study, and all participants (or their parents/guardians) provided written informed consent.

Mutation screening of TMEM151A
TMEM151A variants were screened by whole exome sequencing (WES) of 26 sporadic PKD probands and 27 individuals in nine familial PKD pedigrees without PRRT2 variants. Five ml of peripheral blood was collected from patients and their relatives. Genomic DNA was extracted from peripheral blood samples using DNA extraction kits (Tiangen Biotech, Beijing, China). Agilent Sure Select Whole Exome Capture (Agilent, Santa Clara, CA) and Illumina sequencing technology (Illumina, San Diego, CA) were used for exome capture and sequencing. Data were analyzed using the Illumina Bioinformatics Analysis Pipeline. Basic bioinformatics analysis involved mapping the original FASTQ file to a human reference sequence (UCSC Hg19). Using an internal next-generation sequencing analysis platform and a common database, we screened for single nucleotide variants and short insertions and deletions based on functional annotations. All suspected pathogenic gene mutations were searched in the Human Gene Mutation Database (HGMD) (http:// www. hgmd. cf. ac. uk) and ClinVar (https:// www. ncbi. nlm. nih. gov/ clinv ar). The pathogenicity of all the detected variants detected via Sanger sequencing was predicted using SIFT, Poly-phen2, Mutation Taster, FATHMM-MKL, PROVEAN, and ExAC (http:// exac. broad insti tute. org/). If the mutation frequency in three public databases (1000 Genomes variant database, the NHLBI Exome Sequencing Project, and ExAC) was less than 0.1%, it was considered a candidate mutation. To evaluate the functional effects of candidate variants, changes in protein structure were predicted using Alphafold2 based on the Q8N4L1 PDB template (Jumper et al. 2021;Varadi et al. 2022). The impact of missense variants on three-dimensional (3D) structures was visualized using PyMOL v1.7.4. The MetaDome server was used to identify the landscape of variants tolerance (Wiel et al. 2019).

Expression analysis
Total RNA was extracted from the blood sample of a patient harboring the c.606_607insA mutation, her parents, and a normal control using TRIzol reagent (Invitrogen, Carlsbad, CA), which was reverse transcribed into cDNA using the HiScript III 1st Strand cDNA Synthesis Kit (+ gDNA wiper) (Vazyme, NanJing, China) according to the manufacturer's instructions. To evaluate the relative expression of human TMEM151A, two sets of primers targeting the middle (F1) and downstream (F2) regions of TMEM151A mRNA were designed for quantitative realtime PCR (qRT-PCR) analysis (Supplementary Table S1). The experiments were repeated three times. Expression was compared between groups using Student's t-test in GraphPad Prism 9.0 (***p < 0.001; GraphPad Software, La Jolla, CA).

Literature review
Articles describing cases with TMEM151A gene mutations were identified in the PubMed database using the search term of "TMEM151A". The literature search included publications from database inception to February 18, 2023.

Detection of TEME151A gene variants
TEME151A ((NM_153266) variants were screened for in all sporadic patients and probands of familial pedigrees. One frameshift mutation [c.606_607insA (p.Val203fs)] was detected in a sporadic patient (Fig. 1A) and two heterozygous missense mutations [c.166G > A (p.Gly56Arg), c.791T > C (p.Val264Ala)] were detected in two sporadic patients (Fig. 1B). Further testing for mutations in both parents of the three TEME151A mutation-positive patients with sporadic PKD showed that the three mutations occurred de novo. In addition, a missense variant [c.994G > A (p.Gly332Arg)] was detected in five patients (I-2, II-2, II-4, III-1 and III-2) in Family 1 (Fig. 1C) but not in the other affected patients (III-3 and III-4). Of note, an unaffected family member (I-1) also harbored this variant. No TEME151A mutations were identified in the remaining PKD patients.

Clinical characteristics of PKD patients with TEME151A variants
Three sporadic PKD probands and five individuals in a PKD pedigree were found to have mutations in TMEM151A. The clinical features of PKD patients with TMEM151A variants are summarized in Table 1.
Patient 1 (Fig. 1A) was a 26-year-old woman. At age 13, she experienced recurrent and brief attacks of dystonia in her limbs and face bilaterally 5-10 times every day. Frequently, but not always, these attacks were precipitated by sudden or voluntary movements and by the intention to move. The attacks were mostly similar but sometimes affected the left and sometimes the right side of her body, although the left was more commonly affected. The attacks lasted less than 10 s and were completely relieved by carbamazepine. Attacks were not associated with pain nor alterations in consciousness. Serum calcium, phosphate, and parathyroid hormone levels were normal, as were other laboratory tests including liver and renal function, serum electrolytes, growth and thyroid hormones, and cortisol levels. Brain magnetic resonance imaging (MRI) and electroencephalography (EEG) were normal.
Patient 2 ( Fig. 1B) was an 18-year-old male with no family history of neurological disorders. He had a five-year history of dyskinetic movements involving the left limbs and trunk, often triggered by sudden voluntary actions such as sudden standing or starting to run and exacerbated by emotional stress. He knew when the attacks were about to occur. Neurological examination, laboratory tests (including a metabolic screen), and MRI were normal. Patient 3 (Fig. 1B) was a 24-year-old man with a normal birth and developmental history. He developed dystonic movements at eight years of age. Attacks lasted less than ten seconds and occurred approximately ten times a day. They were often triggered by sudden movement. Most events occurred during the day but were also occasionally present at night. There was no associated loss of consciousness, and he was often able to sense the start of an event. There was no family history of epilepsy nor involuntary movements, including PKD. Neurological examination (both immediately after and between episodes), MRI, and metabolic studies were normal. An EEG documenting the attacks did not reveal any abnormalities. The attacks stopped completely after two days of carbamazepine therapy (100 mg twice a day).

In silico prediction of TMEM151A variant pathogenicity
The TEME151A mutations identified in this study were predicted to be damaging by multiple in silico tools ( Table 2). The mutation c.791T > C (p.Val264Ala) was predicted to be causative according to SIFT, Polyphen2, Mutation Taster, FATHMM-MKL, and PROVEAN. The mutation c.166G > A (p.Gly56Arg) was predicted to be deleterious according to SIFT, Polyphen2, and FATHMM-MKL, while c.994G > A (p.Gly332Arg) was also predicted to damaging (possibly damaging) according to Polyphen2, Mutation Taster, and FATHMM-MKL. None of the predictive tools derived scores for the c.606_607insA (p.Val203fs) mutation.
The identified TMEM151A variants were classified according to American College of Medical Genetics and American College of Pathologists (ACMG/AMP) criteria (Richards et al. 2015). Three TMEM151A variants c.606_607insA (p.Val203fs), c.166G > A (p.Gly56Arg), and c.994G > A (p.Gly332Arg) were classified as of "uncertain significance". The missense variant c.791T > C (p.Val264Ala) was classified as "pathogenic". The amino acid sequence alignment of the four variants in different species showed that all four residues were highly conserved across vertebrates (Fig. 1D). The MetaDome server was used to identify the variant tolerance landscape (Wiel et al. 2019), and the variants corresponded to intolerant residues except for p.Gly332Arg (Fig. 1D).

Structural alteration in the TMEM151A protein
The TMEM151A protein contains a Pfam domain (aa 26-373) and other domains, as predicted by the SMART web tool (Schultz et al. 2000). All four variants were located in the Pfam domain (Figs. 1D, 2A). The p.Val203fs mutation was predicted to yield a truncated protein (Fig. 2B). Both original residue Gly56 and replaced residue Arg56 formed hydrogen bonds with Ala60 and Ile51 (Fig. 2C). Original residue Val264, same as the replaced residue Ala264, formed hydrogen bonds with Phe266 (Fig. 2C). When glycine was replaced with arginine at Gly332, one hydrogen bond with Pro330 appeared (Fig. 2C).

TEME151A mRNA expression analysis
Only the normal TMEM151A c.606_607 site was detected in the PKD patient with c.606_607insA by Sanger sequencing. qRT-PCR revealed about 50% TMEM151A transcript levels in the PKD patient with the c.606_607insA mutation compared with a normal control (Fig. 1A). Therefore, this heterozygous frameshift mutation may cause mRNA decay of TEME151A from the mutant allele.
Fifty-three patients were prescribed anti-epileptic drugs, most commonly carbamazepine (90.6%, 48/53) followed by oxcarbazepine (5.7%, 3/53). Only two patients were prescribed other anti-epileptic drugs (phenytoin and lamotrigine). Of patients taking medicine, 73.6% (39/53) experienced complete remission and 26.4% (14/53) experienced incomplete remission. Seven mutations were reported several times and the other forty-eight reported mutations were novel. The detailed clinical and genetic features of TMEM151A mutations of all currently reported PKD cases are summarized in Supplementary Table S2, and the TMEM151A mutations reported previously are summarized in Supplementary  Table S3.

Discussion
Mutations in TMEM151A have recently been reported as causative in PKD. Data on TMEM151A mutations in patients with PKD support the notion that TMEM151A mutations are responsible for more sporadic than familial PKD cases ). Here we screened for TMEM151A mutations in a Chinese cohort of PRRT2-negative PKD patients. Three novel TMEM151A variants were detected in three sporadic probands, including a frameshift mutation [c.606_607ins A (p.Val203fs)] and two missense mutations [c.166G > A (p.Gly56Arg), c.791T > C (p.Val264Ala)], which accounted for 11.5% of PRRT2-negative sporadic PKD cases (3/26) and no familial PKD cases. The overall frequency of TMEM151A variants in PKD was 8.6% (3/35) in our cohort, slightly higher than the 4.8% reported in Tian et al. (2022), most likely reflecting the relatively small sample size.
TMEM151A is a relatively newly reported gene located at 11q13.2 with as yet unknown functions. It contains two exons and encodes a 468-amino acid protein predicted to contain two transmembrane domains . TMEM151A is highly expressed in the central nervous system, especially the brain, with a higher content in the cerebral cortex and hippocampus. TMEM151A localizes to the endoplasmic reticulum (ER), suggesting that TMEM151A may act as an ER-associated Ca 2+ channel and play a vital role in intracellular Ca 2+ dynamics ). TMEM151A expression is relatively low during the embryonic period, markedly increases postnatally, and then declines in adulthood, similar to the expression pattern of PRRT2 . Thus, it is reasonable to Fig. 1 TMEM151A variants in paroxysmal kinesigenic dyskinesia. A The TMEM151A mutation [c.606_607insA (p.Val203fs)] was detected in the DNA of a patient with sporadic PKD and was associated with half the mRNA expression level of TMEM151A than normal. Squares denote males, circles females, and solid symbols PKD patients; blue column denotes patients and red column normal controls (***p < 0.001). B Sequencing profiles of TMEM151A mutations [c.166G > A (p.Gly56Arg),c.791T > C (p.Val264Ala)] identified in sporadic patients. C Pedigree analyses of Family 1 and sequencing profiles of a TMEM151A variant [c.994G > A (p.Gly332Arg)]. The double line denotes a consanguineous marriage, and a slash symbolizes divorce. The arrow indicates the index patient (II-2). The upper sequencing profile represents the mutant sequence in family members I-1, I-2, II-2, II-4, III-1, and III-2, and the lower sequencing profile represents the normal sequence in family members II-1, III-3, and III-4. D Conservation analysis of the amino acid sequences of the four variants among different species and a domain diagram of TMEM151A protein and the location of the four variants ◂  hypothesize that the spontaneous remission reported in adult PKD patients might be related to the decrease in expression of these proteins observed in adult mice. However, the precise role played by TMEM151A mutations in PKD remains unknown.
We also reviewed the clinical manifestations of all currently reported PKD cases with TMEM151A mutations. Li et al. (2021) identified three TMEM151A variants in three PKD pedigrees and eight variants in eight isolated patients, including four truncated variants, three missense variants, and a non-frameshift deletion. Tian et al. (2022) enrolled a large sample of 521 individuals, in whom they detected variants in 29 patients (from 25 families) and identified 24 heterozygous variants in TMEM151A in 25 probands (4.8%, 25/521). These included 18 missense and six nonsense mutations, 23 variants of which had not previously been reported.  completed an exome-wide rare variant burden analysis in 86 PRRT2-negative PKD probands and detected six variants, including three novel, rare protein-altering TMEM151A variants in ten unrelated probands accounting for 11.6% (10/86) of PRRT2-negative cases. Chen et al. (2022) identified seven TMEM151A variants (6.9%, 17/245) in a cohort of 131 PKD probands (108 without PRRT2 variants and 23 newly recruited), five of which were novel and one of which occurred de novo. Wirth et al. (2022) reported a de novo mutation in TMEM151A in a patient with PKD. Ma et al. (2022) reported two novel TMEM151A variants from two families, while Wang et al. (2022) also identified a novel variant in four individuals from one pedigree. They described the typical manifestations of PKD in a three-generation family and suggested that the TMEM151A gene may be associated with the disease spectrum of PKD-PKD/IC-benign familial infantile convulsions (BFIC). Mounir Alaoui et al. (2023) described three sporadic PKD patients, including two novel variants. Notably, one of the patients also had autism spectrum disorder. To date, 48 TMEM151A variants have been reported to be associated with PKD. PRRT2 mutations account for 77% to BIS benign infantile seizures, BFIC benign familial infantile convulsions, ET essential tremor, NA not available, C choreoathetosis, D dystonia, CBZ carbamazepine, OXC oxcarbazepine, IR incomplete remission, CR complete remission 93% of familial cases and 21% to 45% of sporadic cases in different cohorts Huang et al. 2020;Liu et al. 2022;Méneret et al. 2012), while TMEM151A mutations account for less than 10% of PKD patients according to our literature review. About half of patients with primary PKD have unclear genetic causes.
While patients with PRRT2 and TMEM151A variants share common features of PKD (movement triggers, episodic involuntary movements, no loss of consciousness, presence of aura, spontaneous remission), there is also clinical heterogeneity. According to comparative cohorts , five variable features include duration of attack, age of onset, presence of benign infantile epilepsy, predominant character of the hyperkinetic movement, and response to treatment with anticonvulsants. In our study, patients with TMEM151A variants had a shorter duration of attacks and all presented with dystonia, similar to previously . All responded to carbamazepine. In addition, TMEM151A variants were more common in sporadic PKD patients (100%, 3/3), consistent with very recent data . Thus, the occurrence of TMEM151A variants in sporadic cases might also be attributed to a lower penetrance compared with PRRT2 variants or de novo mutagenesis. The basis for the clinical heterogeneity seen in the presentation of the paroxysmal disorders due to TMEM151A mutations remains unknown. Future research will likely further expand the clinical phenotype of TMEM151A mutations.
We identified three novel TMEM151A mutations [c.606_607insA (p.Val203fs), c.166G > A (p.Gly56Arg), and c.791T > C (p.Val264Ala)] not previously reported. Further testing for mutations in both parents of the three TEME151A mutation-positive patients with sporadic PKD showed that the three mutations occurred de novo. According to ACMG rules for classifying sequence variants, the missense variant c.791T > C (p.Val264Ala) was classified as "pathogenic", as also reported by Tian et al. (2022). The c.166G > A (p.Gly56Arg) and c.606_607insA (p.Val203fs) mutations were classified as of "uncertain significance" but were respectively co-segregated in their families, so we consider these two mutations to be pathogenic. Overall, the causative mutations in TMEM151A [c.606_607insA (p.Val203fs), c.166G > A (p.Gly56Arg) and c.791T > C (p.Val264Ala)] accounted for 11.5% of PRRT2-negative sporadic PKD cases (3/26) and no familial PKD case.
Interestingly, Sanger sequencing detected the c.994G > A (p.Gly332Arg) mutation in five patients in Family 1 (I-2, II-2, II-4, III-1 and III-2) and an unaffected family member (I-1) but not in the other affected patients (III-3 and III-4). Therefore, the c.994G > A (p.Gly332Arg) mutation did not co-segregate with the clinical phenotypes in this family, indicating that this mutation was unlikely to be causative. Correspondingly, it was classified as of "uncertain significance" according to ACMG criteria. Tian et al. (2022) used a WES database containing 1500 non-paroxysmal disorder controls (WESctrl_1500) to compare differences in candidate gene frequency between PKD and other diseases. In doing so, they detected the c.994G > A (p.Gly332Arg) variant in WESctrl_1500, with the minor allele frequency outweighing the incidence of autosomal dominant inherited diseases (0.0007). Non-segregation in Family 1 was considered strong evidence that c.994G > A (p.Gly332Arg) was a benign variant. This missense change is similar to that reported in 1500 normal controls in Tian et al. (2022) and consistent with its low pathogenicity.
Among our three novel TMEM151A mutants, c.606_607insA is a frameshift mutation leading to the appearance of premature termination codon (PTC), which can result in the production of truncated proteins or the degradation of messenger RNAs by nonsense-mediated mRNA decay (NMD). NMD is a quality control pathway that degrades mRNAs containing PTCs due to nonsense or frameshift mutations (Holbrook et al. 2004;Khajavi et al. 2006). Transcript analysis of the patient with the p.Val203fs variant revealed a clear ~ 50% reduction in TMEM151A mRNA by semiquantitative analysis, i.e., heterozygous NMD of TMEM151A (p.Val203fs), which may lead to a TEME151A heterozygous knockout-like phenotype (i.e., PKD caused by TMEM151A mutation may be due to haploinsufficiency). Interestingly, Chen et al. (2011) also found that truncated PPRT2 resulted in altered cellular localization and perhaps complete loss of transmembrane function, which was rescued by applying an NMD pathway inhibitor. Among the identified PRRT2 mutations, most are truncating mutations (Chen et al. 2011;Ebrahimi-Fakhari et al. 2015;Méneret et al. 2013). Therefore, NMD of truncating mutations in TMEM151A and PRRT2 may be a common mechanism leading to PKD.
In conclusion, together with previous studies, this study further supports the hypothesis that TMEM151A is causative for PKD with possible incomplete penetrance in Chinese people. Despite the small sample size, these detected mutations in TMEM151A expand the clinical phenotype and mutation spectrum of the disease and provide further details about its pathoetiology. NMD of TMEM151A to produce haploinsufficiency maybe the pathogenic mechanism underlying PKD. Further functional studies are now needed to understand the underlying pathogenic mechanisms associated with TMEM151A mutations.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s00439-023-02535-3. assisted in the preparation of the figures. FH carried out the experimental design and data analysis later. QXZ, CYM collected the clinical data. XYL, BX and GLL supervised the study. FH wrote and revised the final version of the paper. HLH, ZS and DL revised the manuscript and gave final approval of the version to be published. All authors contributed to revising the manuscript and read through and approved the submitted version. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Human and animal rights This study involving human participants was reviewed and approved by the Institutional Review Board of the Third Xiangya Hospital of Central South University (NO.22254).
Informed consent All patients or their participants' legal guardian/next of kin have signed an informed consent form.