To dissect nonsense mutations to be treated with tRNAs, variations within eight genes causing primary hyperammonemia were evaluated. Overall, 156 patient mutation and 60 variation sites were reported in 140 publications and in two genomic databases of 114,878 participants, respectively. Counting the frequency revealed that CGA>TGA was by far the hottest spot, accounting for 41% of patient mutations and 50% of nonsense alleles, which were underestimated in preceding database searches but match the observed high CpG mutation rate. Over 20 variations were linked to patient mutations in the same population at both CGA and non-CGA sites, which will be informative for therapy. While values in the database could predict incidence rate of autosomal enzyme deficiencies, most of them were lower than the equilibrium value expected in autosomal recessive genes, which was calculated from germline CpG mutation rate and selection coefficients estimated here. Toxicity inherent to nonsense variations, if expressed during reproduction, explains this observation, although the details need to be elucidated. The presence, maintenance, and pathogenesis of nonsense variations will thus be diverse. Pertinent clinical information including personal genomic data will enrich our understanding of their characteristics.
Article
Human nonsense mutations in primary hyperammonemia – Analysis of publicized patient mutations and variations in general populations in eight disease-causing genes
https://doi.org/10.21203/rs.3.rs-3090477/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Among the diverse human disease-causing mutations, nonsense mutation occupies a distinct position owing to its prevalence and pathogenicity [1]. It is defined as one of the three types of single nucleotide substitutions in the coding sequence (CDS), which produces a premature termination codon (PTC) [2]. Roughly 10% of human genetic diseases are caused by nonsense mutations [3]. Pathogenesis is derived from the loss-of-function phenotype evoked by PTC. Toxic effects of truncated proteins are expected to be eliminated by nonsense-mediated mRNA decay (NMD) [4], which is also evoked by PTC.
Transfer RNAs (tRNAs) with mutation on their anticodon, which are referred to differently among researchers as suppressor tRNAs or anticodon-engineered tRNAs (ACE-tRNA) [5], were shown to suppress nonsense mutations in mouse models of Duchenne-type muscular dystrophy [6] and Hurler syndrome [7] or in human cells with cystic fibrosis transmembrane conductance regulator gene [5]. Hence, the prospects for using these tRNAs for human diseases are increasing. We are interested in which type of engineered tRNA should be prepared with priority and have decided to evaluate the pathogenesis and frequency of disease-causing nonsense mutations reported in patients and databases.
In this study, primary hyperammonemia [8], or urea cycle disorders [9,10] were chosen as a model. The results confirmed previous database searches that CGA>TGA and CAG>TAG are the two prevalent patterns. Evaluation of mutation occurrences and codon usage bias revealed that CGA was the hottest spot. Symptoms caused by enzymatic deficiencies were severe in most cases. The estimated selection coefficient was over 0.7 in five deficiencies, suggesting that the frequency of nonsense alleles can be low. Allele frequencies (AFs) of nonsense variation in two population databases, however, were even lower than the value expected in autosomal recessive disorders from the observed mutation rate at CpG sites, suggesting that most nonsense alleles cannot expand or are lost through the generations. Factors affecting AF values are discussed.
1. Patient and variant data acquisition
Patients diagnosed with nonsense mutation(s) between 1989 and 2022 were collected from publications. Nonsense variations listed in the jMorp [11] and gnomAD v3.1.2 [12] databases (populations described in Table S1) were obtained using the search terms “pLOF” and “stop gained.” Mutation sites, affected cases, and variation sites are listed as Datasets 1, 2, and 3, respectively. Details for the selection of patients and references are described in the Supplementary Information.
2. Genetic data
Gene names are described with the HGNC gene symbol. GRCh38 was used for genomic coordinates. Enzymes and then transporters are presented in the order of potency at detoxifying ammonium:
(1) N-acetyl glutamate synthase (NAGS, EC 2.3.1.1),
(2) carbamoyl phosphate synthetase 1 (CPS1/CPSI, EC 6.3.4.16),
(3) ornithine transcarbamylase (OTC, EC 2.1.3.3),
(4) argininosuccinate synthetase 1 (ASS1, EC 6.3.4.5),
(5) argininosuccinate lyase (ASL, EC 4.3.2.1),
(6) arginase-1 (ARG1, EC 3.5.3.1),
(7) SLC25A15 (ornithine transporter 1/ORNT1), and
(8) SLC25A13 (aspartate/glutamate carrier or citrin).
Chromosomal locations and CDS lengths of MANE (matched annotation between NCBI and EBI) select sequences are shown in Fig. S1. Codon frequency is shown in Table S2.
The sequence of the PTC “TAA,” “TAG,” or “TGA” is described as “_ochre,” “_amber,” or “_opal,” respectively, when necessary, using the underscore [13] to indicate a type of mutation within the “range” of PTC.
3. Evaluation of patient data
Assignments and the number of assigned patients are described in Dataset 2 and Table S3, respectively. Inheritance was classified according to the genetic diagnosis of the parent(s) and the presence of a family history.
Onsets were classified into the following four categories: neonatal (up to 28 days), late (29 days–2 years), childhood (2–19 years), and adult (20 years and above). Screening and recently reported female cases of OTC deficiency [14] were classified according to the authors’ description.
Patient outcomes were classified into four categories in the following order of severity:
De: dead, for those who died of hyperammonemia;
Sy: symptomatic, for those who needed support for their daily activities;
Al: alive, for those who were alive and stable at the time of publication; and
Cu: cured, for those who had received liver transplantation.
In each gene, s was determined using the following equation:
s = (nNDe + nNSy)/ (nNDe + nNSy + nNAl + nNCu) (1)
in which the number of patients who died, for example, is described as NDe, and n is the number of nonsense alleles in the patient (n =1 or 2). That of OTC was determined from male patients.
4. Epidemiological data
The incidence rate of inborn errors of metabolism was cited from published results of screening [15]. Incidence rates of cases with a nonsense mutation were determined from Japanese nationwide studies [16,17]. Birth numbers in Japan were calculated from statistical surveys conducted by the Ministry of Health, Labour, and Welfare.
5. Evaluation of AF
AF of disease-causing variations was determined using preceding calculations applied to a monogenic disease gene [18,19], with a minor modification. All variants classified as “Pathogenic/Likely Pathogenic,” “Pathogenic,” or “Likely Pathogenic” in ClinVar were collected. In addition, regardless of the description in ClinVar, all nonsense and frameshift variants in MANE select CDS (CDS of matched annotation between NCBI and EBI select transcripts) were included. In this study, when there were variants with classifications that conflicted between two databases, they were excluded. For instance, two missense variations in the ASL gene, classified as ‘conflicting_interpretations_of_pathogenicity’ in the jMorp database, ClinVar variants 92359 and 555574, were excluded. Further information regarding the collected variants is available upon request.
AF of disease-causing variations in the specific gene, Qdisease-causing, was determined using the following equation:
in which the total number of sorted variation sites is n, and AF of the variation at site i is qi(pathogenic). Similarly, AF of all nonsense variations, Qnonsense, was determined by summing up AF values, qi (stop gained), which were obtained by the search term “stop gained.”
The statistical significance of differences of AF among variants was evaluated using Kruskal–Wallis test, followed by Dunn’s multiple comparison tests.
6. Penetrance of nonsense variants
Penetrance of variants was considered under the assumption that the subject received no treatment. Unless specified, penetrance of nonsense variations is hypothesized to be complete. Penetrance in the SLC25A13 gene is supposed to be incomplete because the estimated carrier frequency of variants from the jMorp3.8K genomic database was reported to be much higher than that from the incident rate [19]. Using reported values of carrier frequency (1/19,000 [19]) and incidence rate (1/96,000 [15]), the proportion of individuals carrying the variation who develop the disorder is calculated to be 19.8%.
Primary hyperammonemia as a model
We selected a model from the following two requirements [A] and [B], considering condition [C].
[A] Sample number: Substantial numbers of nonsense mutations and/or variations are present.
[B] Severity of the disease: Either of the natural history or the current prognosis of the disease is poor. These diseases will be candidates for site-specific treatment targeting nonsense mutations.
[C] Conditions of data: Accuracy, transparency, or sustainability of the data. Conditions are satisfied by the usage of publicized data.
AF of disease-causing alleles of monogenic diseases differs among populations. Table 1 lists the epidemiological incidence rates of PAH (phenylalanine hydroxylase), ASS1, ASL, and citrin/SLC25A13 deficiencies as revealed by mass screening in Japan and Germany[15] and AF in the corresponding populations available in two databases.
Values in the row for PAH indicate that the AF of nonsense alleles is not always correlated with that of all disease-causing alleles. Since no gene will universally satisfy requirement [A], we selected primary hyperammonemia, which will without doubt satisfy requirement [B] and is associated with eight different genes involved in the urea cycle (Fig. S1). The prevalence of urea cycle disorders is generally recognized as high. Even if the prevalence of one gene deficiency is low in one specific population, other deficiencies with high prevalence can cover the number of samples. In addition, as the row for SLC25A13 shows, the prevalence of citrin deficiency, one form of primary hyperammonemia caused by SLC25A13 deficiency [20], is high in Japan. AF value from 38KJPN (approximately 38,000 Japanese population in the jMorp database) exceeded that of PAH, satisfying requirement [A].
Overview of patient mutations and clinical symptoms
In total, 156 sites have been reported as 383 independent events in 140 references (Fig. S2 and Table S4). The inheritance of 383 independent patient mutations, which is described as “occurrences” in this work (see the column of “occurrence” in Dataset 2), was classified according to the parent(s)’ diagnosis and family history (Fig.1A and Table S3-1). Out of the 289 occurrences of the seven autosomal recessive gene deficiencies, all 25 diagnosed occurrences were inherited from their respective parents. No de novo inheritance was diagnosed. In contrast, out of 94 occurrences in an X-linked dominant OTC deficiency, 10 were diagnosed as de novo, while 12 were inherited. Thus, the cohort collected with nonsense mutations retains a characteristic of Mendelian inheritance.
Onset, outcome and a selection coefficient (s) of the cases in Dataset 2 were evaluated (Fig. 1B and 1C, Table S3-2 and S3-3) in seven gene deficiencies in which penetrance is hypothesized to be complete. The value of s (0≤ s ≤1), calculated from equation (1), indicates the efficiency of elimination of the disease allele to the next generation and is 1 when there is no transmission at all to the next generation [21,22].
Except for female OTC deficiency, most cases were neonatal or late onsets. Cases diagnosed after the neonatal period are listed in Table S5. Onset after the neonatal period was not associated with a good prognosis. For example, the outcome of death or exhibiting symptoms was reported in p.Trp265*_amber OTC deficiency diagnosed at 4.3y [23] or p.Arg1262*_opal CPS1 at 13y [24], respectively. Mutations were in the middle of the OTC or CPS1 gene (Fig. 2A). These alleles are expected to severely impair enzymatic function and lead to a poor prognosis. Thus, the natural clinical courses of patients with nonsense mutations are generally severe (Fig. 1B and 1C). However, it should be noted that an early intervention followed by liver transplantation drastically increases cured cases [16,25–28] (Fig. 1C, see Supplementary Results for detailed clinical evaluation). Successful treatment decreases the value of s, which in turn increases the allele frequency (AF) of disease-causing variations. AF of nonsense variations in general populations is evaluated using s (see Discussion).
Patient mutations (Fig. 2A)
In Fig. 2A, the number of occurrences is shown along with affected genes. Mutations forming TAA, TAG, and TGA were observed at 40 (26%), 61 (39%), and 55 (35%) sites, respectively. 50 sites (32%) were in the X-linked dominant OTC gene. Autosomal CPS1, SLC25A13, ASL, and ASS1 genes accommodated 31, 26, 19, and 13 mutation sites, which corresponds to 20%, 17%, 12%, and 8.3% of the total, respectively. Five or six sites (3% or 4%) were reported in NAGS, or ARG1 and SLC25A15 genes, respectively.
All possible 23 nucleotide changes were reported within eight genes. The ratio of base transition to transversion was 1.29. CAG/Gln>TAG and CGA/Arg>TGA were the two most preponderant patterns, which accounted for 31 (20%) and 27 (17%) sites, respectively. Three other changes caused by base transition were also prominent. CAA/Gln>TAA, TGG/Trp>TAG, and TGG/Trp>TGA accounted for 10 (6%), 9 (6%), and 11 (7%) sites, respectively. In base transversion, change patterns starting from codons for glutamate were the two most prominent. GAA>TAA and GAG>TAG accounted for 11 (7%) and 9 (6%) sites, respectively. Because evaluation of 23 different values in a single table is extremely complicated, we developed a graphical representation (Fig. 3A), adopting the octagon representation employed by Garen for the description of nonsense triplets [29]. The overall tendency was similar to that in preceding database searches [30,31] (Fig. 3E and 3F), but CGA>TGA, previously reported as the most frequent (Fig. 3E and 3F) was next to CAG>TAG in hyperammonemia (Fig. 3A). This could be due to the limited number of CGA codons within the eight genes because, in contrast to 30 CGA codons, 132 CAG codons are present (Fig. S3). We therefore counted an independent mutation event as an “occurrence” (see Dataset 2).
With this information, CGA>TGA and CAG>TAG have been reported as numbering 156 (41%) and 69 (18%), respectively (Fig. 3B). Thus, CGA>TGA was the hottest spot for the mutations causing primary hyperammonemia, which reported 2.28 times more frequently than CAG>TAG mutation.
Most of the 156 sites were sporadic. Overall, 107 (69%) and 20 (13%) sites were reported only once and twice, respectively. Meanwhile, 29 sites were recurrent, or reported thrice or more, among which 20 were CGA>TGA (Fig. S3 marked with an atsign: @). The sum of occurrences at these CpG sites accounted for 146 independent mutation events (38.1% of 383 occurrences overall and 93.6% of 156 CpG occurrences), making these sites the major source of nonsense mutations causing hyperammonemia.
The mutation with the highest occurrence per site, however, was not at CpG sites. In the nine non-CGA sites reported more than twice, two prominent sites were derived from a single ethnic population. These were CAG>TAG (ASL p.Gln*354_amber) [32] and TCG>TAG (SLC25A13 p.Ser225*_amber) [17,20], mutations both having occurrences of 30 in Saudi Arabian and Japanese populations, respectively (Fig.2A and Dataset 1). AF of the latter in 38KJPN was 0.000426. Thus, while CGA>TGA is a hot spot, some nonsense mutations derived from non-CGA codons can become more prevalent than CGA>TGA.
Nonsense variations in general populations (Fig. 2B)
The number of variation sites was limited to 60 within the MANE select CDS (CDS of MANE select transcripts, Table S6-1). Because two variations, p.Glu271*_ochre and p.Glu273*_amber, in the OTC gene did not pass the filter, no nonsense variations were present in the X-linked OTC. Among 23 possible patterns, 16 patterns were observed. CGA>TGA and CAG>TAG were again two preponderant patterns, numbering 19 (32%) and 9 (15%) in total, respectively (Fig. 3C).
The count of alleles in the seven autosomal genes was 217 (Table S6-2). Nearly half were in the form of CGA>TGA (Fig. 3D). Nucleotide change and codon usage are two known biases that affect nucleotide replacement [33]. To confirm that CGA is indeed a hotspot for nucleotide changes, we adjusted the effects of codon usage, by dividing the obtained frequency values (Tables S4-2 and S6-2) by the codon frequency (Table S2-2). The usage of CGA codons in all eight genes was never high (1.9% of 23 patterns, Table S2-2). After this adjustment, the occupancy of CGA among all eight genes reached 57.4% and 58.1%, as shown in Tables S4-4 and S6-4, respectively, indicating that CGA is clearly the hottest spot of nonsense nucleotide change. However, the highest contributor was once again p.Ser225*_amber, because the codon usage of UCG/Ser was lower than that of CGA/Arg in the SLC25A13 gene. Because this variation has been solely identified from Japanese [34,35], inspection for variants by gene and by population is important.
Notable variations coincident with patient mutations
Overall, 30 of the 60 variants were located at the reported patient mutation sites. To delineate the relationship between a patient mutation and frequency of nonsense alleles in a population, 60 variants were dissected according to the population from which the variant was derived (Table S7). AF values in a population were then grouped according to the recurrence of a mutation (Fig. 4A). Two variants near the 3’ end of CDS, CPS1-Gln1413_EAS and SLC25A13_Trp606_JPN, were classified into group “0” (not-occurring). This can be because the penetrance of nonsense mutation becomes incomplete near the carboxyl-terminus in some genes especially when the degree of truncation is small for the function of the polypeptide, although the details of this need to be clarified. Some variants derived from an African (AFR) or American (AMR) population were outliers in the group “≤2” or “0,” which could be attributed to a low opportunity of genetic diagnosis in these populations. We focused on the two largest populations, 38KJPN and NFE (Fig. 4B). AFs were significantly different according to whether the mutation is recurrent, sporadic, or not occurring (P <0.0001 between “≥3” and “0” and P =0.0087 between “≥3” and “≤2”). The significance was further confirmed by whether a single mutation occurred or not (Fig. S4A, P =0.00060). Thus, the value of AF is a factor determining the recurrence of the mutation.
From the viewpoint of patient mutations, among 289 occurrences in the seven autosomal genes, 119 (41%, Fig. 4C, rightmost columns) were coincident with the 21 ethnicity-matched variation sites (Fig. 2B, leftmost column in the table, population name in the figure underlined). Overall, 14 were recurrent sites (Fig. 4B, a group of “≥3”), at which 109 occurrences were reported (92% of 119). The calculated correlation coefficient (r) between occurrence number and AF was 0.776, indicating a positive correlation. One of the most prominent sites among these 14 notable variations is Arg179 in the SLC25A15 gene, at which eight independent occurrences from five different ethnicities, three from Japanese and two from Chinese, were reported. In accordance with occurrences, 28 nonsense alleles were reported from five different populations, indicating that multiple populations can be involved in one site.
We analyzed the location and frequency of nonsense mutations reported in patients and variations listed in public databases. Two important factors, mutation at CpG site and variants present in the population, contributed to approximately 40% of independent occurrences (Fig. 4C). Because positions of these sites are already known, these sites will be good targets for therapy using engineered [5,7,36] and natural [37] tRNAs reported to date.
With an inspection of eight genes, we also observed how these values were influenced by several factors, including: codon frequency of CGA and other codons involved in nonsense mutations (Fig. S3), whether and how homozygotic occurrences were involved (Fig. 4C), how intensively genomes of affected populations were investigated, and whether and how the genotype of each patient was intensively studied and reported. The search will need to be extended to other genes and continued periodically. Nationwide studies like those cited here [16,17,23] will be an important data source to ensure transparency and accessibility of clinical data. They will also reduce publication bias overlooking mutations that have already been reported.
Half of the 60 variation sites in general populations did not coincide with known mutation sites. Identified patient mutation sites will thus continue to increase in number. In contrast, all 27 CGA/Arg codons in the seven genes were known mutation sites and contributed to over 40% of occurrences, which is an estimation 1.9 to 2.3 times higher than that in preceding database searches (Fig. 3). Counting the frequency is necessary to exactly evaluate nonsense mutations, which, to the best of our knowledge, is comprehensively performed by us for the first time. In sections below, we are going to explain the origin of this frequency information, AF in a genomic database. Then, we are going to relate AF to the mutation rate, the single nucleotide substitution of which has recently been measured in trios analysis [38–40].
Origin of the frequent variants
In this study, four factors influencing the AF of disease-causing nonsense variants were identified. In decreasing order of observed frequency, they were 1) regional bias, 2) mutation at CpG methylation site, 3) codon usage bias, and 4) likelihood of mutation involving base transition rather than base transversion. These factors can overlap in some variants. The origins of 1) and 2) are discussed below.
Regional bias, which can be one outcome of genetic drift, was prominently observed in small populations (Fig. S4B). Genetic bottlenecks or founder effects associated with a history of a limited population even within a country can be the source of these variations [22]. The rate of consanguinity can be involved, as this has been reported to be tightly linked to the founder mutation Gln354*_amber in the ASL gene [32] in the central province of Saudi Arabia [41] Similar factors would have worked for Japanese p.Ser225*_amber variation, although the details need to be elucidated.
Nucleotide change and codon usage are two known biases [33]. Among these biases, the most striking was for the CpG site, which, upon adjustment for codon usage, stood for 38.6% and 82.8% of nonsense variations in the JPN and NFE populations (data not shown), respectively. This former rate was low, because the Ser225*_amber variant specific to the Japanese population constituted 46.9% of the total. Mutation at a CpG site would originate from unrepaired TG mismatch in germline cells after the spontaneous deamination of 5-methylcytosine at CpG methylation [42].
Frequency needs to be discussed for the disease-causing variations, or variants that reside in CDSs that are indispensable for life. AF of Gln312+*_amber in the SLC25A13 gene is as high as 0.000191 in the NFE population (Dataset 3). This can be because this variant does not reside in MANE select CDS and the function of the gene will not be compromised. Even when variants reside in a MANE select CDS, the AF of some variants is as high as 0.525967 or 0.000517 in the ACTN3 [43] or PCSK9 [44] gene, respectively. This is because, nonsense variations of these genes are not disease-causing but can be beneficial in some cases.
Evaluation of AF values
According to Hardy–Weinberg and mutation–selection equilibria [22], AF of disease-causing variation of Mendelian autosomal recessive disorders, or q, can be written, using the mutation rate, µ, as:
where s is the selection coefficient determined in Fig. 1C.
The values of µ for CpG sites evaluated from pseudogenes [45] or in trios analyses [38] were 1.6 × 10 –7 and 1.12 × 10 –7, respectively. In our analysis, s was close to 1, without intervention (Fig. 1C). To estimate the minimal value of q, let s be 1, the equilibrium value of q in CpG sites is 0.000334 at minimum.
Among the populations in which the participant number exceeded 10,000, Japanese p.S225*_amber in SLC25A15 was the sole variation in which AF exceeded this value (Fig. 4B). In most cases, nonsense AF values were less than this threshold, indicating that CGA>TGA variation alleles were unable to reach the equilibrium level. Possible reasons are discussed below. Wild-type and nonsense alleles are described as A and a, respectively. The following four factors are evaluated.
- Lack of feasibility of AF values in genomic databases: because AF in a genomic database is experimentally measured, the value includes errors. Feasibility of data is evaluated using incidence rate of gene deficiency (Table 2).
- Bias in enrolment onto a population study: if carriers, or unaffected parents of diseased children, refuse enrolment, observed q becomes low.
- Frequency of other disease-causing variant(s) a’ in the same gene: if prevalent, the frequency of nonsense allele a is reduced by forming a compound heterozygote aa’.
- Toxicity of nonsense allele: because CDS with a nonsense mutation encodes a shortened polypeptide, which can be toxic if expressed, it will be worthwhile considering the biological background.
(i) Evaluation of feasibility of AF values: Incidence rates of Mendelian recessive disorder is expressed using following equations [22]:
Incidence ratedisease-causing = Qdisease-causing2 (4)
Incidence ratenonsense = 2 × Qdisease-causing × Qnonsense (5)
in which Qdisease-causing and Qnonsense is obtained from equation (2) for the expected incidence rates of the entire disease and disease evoked by nonsense mutations, respectively. The term Qnonsense2 in equation (5) was neglected for approximation [22]. Values calculated from equation (5) are shown in the first column of Table 2.
The observed incidence rate was independently determined from Japanese nationwide studies. The square root of the ratio to this observed incidence rate was used for evaluation (Table 2). Values of CPS1 matched and those of SLC25A13 differed by 3.37, most likely reflecting incomplete penetrance. In ASS1, no nonsense alleles were reported in 38KJPN (Table 1). If one nonsense allele had been listed, the expected occurrence would have been 1:23,204,870, with a square root value of 0.90-fold. The value in ASL was 1.47, which indicates that one more patient should have been diagnosed. The incidence rate of ASL diagnosed by screening was very low in Japan [15] and no prevalent pathogenic variant was present (see below for factor(iii)). Under these conditions, formation of the case with nonsense mutation would be difficult, although more searches are necessary. These findings imply that the overall architecture of AF values and annotations for the clinical assignment of the pathogenicity of variants are feasible to predict observed incidence rate, although gene-specific consideration is necessary. In contrast, the addition of an equilibrium value 0.000334 to the term Qnonsense in equation (5) drastically increases the estimated incidence rate. One explanation for this is that nonsense mutations in a population are produced more than the current AF level, but most disappear quickly without being recognized to have existed.
(ii) Evaluation of enrolment bias: Because the occurrence of recessive disorder is expressed by equation (4), if a carrier parent refuses to be enrolled in a population study, the design of which has been described [46,47], the gap between the observed value Q and real value Qreal is expressed by the following equation:
Q = Qreal – Qreal2. (6)
Let x be Qreal, then x2 – x + Q = 0. Because x and Q have small values, x should be equal to the smaller solution,
Q is a small number, thus using a stringent approximation, 
= 1 – 2Q – 2Q2,
Qreal = Q + Q2. (8)
If both parents refuse to be enrolled, equation (7) can be written as Q = Qreal – 2Qreal2. This will give a final solution Qreal = Q + 2Q2, instead of solution (8). Factor (ii) will be negligible.
(iii) Evaluation of other disease-causing variants: Details of this factor will be described elsewhere (manuscript in preparation). Briefly, in the 38KJPN population, there were no prevalent pathogenic variations in ASL. In CPS1, ASS1, and SLC25A13, prevalent pathogenic variations of a type other than nonsense variation were present and numbered 1 or 3 (see Supplementary Discussion). In ARG1 and SLC25A15, notable nonsense variations at Trp122 and Arg179, respectively, were the most prevalent. In ASS1 and SLC25A13, AF of prevalent variations was much higher than the value calculated from equation (3). Therefore, in some genes, variants other than nonsense variants were contributing as factor (iii), suggesting the need to adjust the threshold value downwards using AFs of other disease-causing variants shown as Qdisease-causing in Table 1. Except for one variant CPS1 p.Arg1262*_opal and variants in the SLC25A13 gene, in which penetrance is incomplete, AF of nonsense variants was even lower than the adjusted threshold value (based on an unpublished observation). Thus, the question arises of whether there is a biological mechanism that reduces AF of nonsense variations.
(iv) Evaluation of biological background: Equation (3) is based on the autosomal recessive phenotype of Mendelian disorders that allele a is solely lost after the formation of individuals with aa. Discrepancy between expected and observed AF suggests that nonsense alleles in the genes for primary hyperammonemia cannot expand, as discussed for factor (iii), or are lost faster than predicted from the Mendelian inheritance of recessive traits. This faster disappearance can be explained by the toxic phenotype appearing on gamete (a) or germ and embryo (Aa), although the molecular mechanisms behind this are yet to be elucidated.
A decrease of lethal alleles in the form of heterozygotes was reported in a systematic analysis of consanguineous marriages in human populations [48]. In the case of nonsense variants, because nonsense mutations can occur in any organisms with translational machinery, questions related to the decrease of AF can be addressed using model organisms. In fact, the involvement of NMD [4] in the hypothetical toxicity during reproduction will be best investigated in mice, because the presence of NMD during reproduction has been reported in this organism [49]. The caveat for all these discussions is that irrespective of the molecular mechanism involved, the toxicity is not an absolute factor preventing reproduction, because cases with nonsense homozygotes or a compound heterozygote with two different nonsense alleles occur (Fig. 4C).
Results and discussion described here imply that nonsense variations in human disease-causing genes are good target for research as well as therapy.
Acknowledgments
We thank laboratory members for helpful discussions and support. NK is grateful to Drs. Tadashi Kaname and Shu Tadaka for communicating about trios analysis and filtering of jMorp database, respectively. Computations were partially performed on the NIG supercomputer at ROIS National Institute of Genetics.
This work was supported by JSPS KAKENHI Grant Number JP19K07403 to NK.
Disclosures
All authors declare no competing financial interests.
- Antonarakis SE, Cooper DN. Human Gene Mutation in Inherited Disease: Molecular Mechanisms and Clinical Consequences. In: Pyeritz RE, Korf BR, Grody WW (eds). Emery and Rimoin’s Principles and Practice of Medical Genetics, 7th edn. Academic Press: London, 2019: 125–200.
- Goldberg ML, Fischer JA, Hood L, Hartwell L. Chapter 9 Gene Expression: The Flow of Information from DNA to RNA to Protein. In: Genetics: From Genes to Genomes. 7th edn. McGrawHill: New York, 2021: 257–302.
- Nussbaum R, McInnes R, Willard H. Chapter 4 Human Genetic Diversity: Mutation and Polymorphism. In Thom & Thom Genetics in medicine. 8th edn. Elsevier: Philadelphia, PA, USA, 2016: 43–56.
- Kurosaki T, Popp MW, Maquat LE. Quality and quantity control of gene expression by nonsense-mediated mRNA decay. Nat Rev Mol Cell Biol 2019;20:406–420.
- Lueck JD, Yoon JS, Perales-Puchalt A, Mackey AL, Infield DT, Behlke MA, et al. Engineered transfer RNAs for suppression of premature termination codons. Nat Commun 2019;10:822.
- Buvoli M, Buvoli A, Leinwand LA. Suppression of nonsense mutations in cell culture and mice by multimerized suppressor tRNA genes. Mol Cell Biol 2000;20:3116–3124.
- Wang J, Zhang Y, Mendonca CA, Yukselen O, Muneeruddin K, Ren L, et al. AAV-delivered suppressor tRNA overcomes a nonsense mutation in mice. Nature 2022;604:343–348.
- Häberle J. Clinical practice: the management of hyperammonemia. Eur J Pediatr 2011;170:21–34.
- Batshaw ML, Tuchman M, Summar M, Seminara J, Members of the Urea Cycle Disorders Consortium. A longitudinal study of urea cycle disorders. Mol Genet Metab 2014;113:127–130.
- Matsumoto S, Häberle J, Kido J, Mitsubuchi H, Endo F, Nakamura K. Urea cycle disorders-update. J Hum Genet 2019;64:833–847.
- Tadaka S, Hishinuma E, Komaki S, Motoike IN, Kawashima J, Saigusa D, et al. jMorp updates in 2020: large enhancement of multi-omics data resources on the general Japanese population. Nucleic Acids Res 2021;49:D536–544.
- Karczewski KJ, Francioli LC, Tiao G, Cummings BB, Alföldi J, Wang Q, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature 2020;581:434–443.
- den Dunnen JT, Dalgleish R, Maglott DR, Hart RK, Greenblatt MS, McGowan-Jordan J, et al. HGVS Recommendations for the Description of Sequence Variants: 2016 Update. Hum Mutat 2016;37:564–569.
- Gobin-Limballe S, Ottolenghi C, Reyal F, Arnoux J-B, Magen M, Simon M, et al. OTC deficiency in females: Phenotype-genotype correlation based on a 130-family cohort. J Inherit Metab Dis 2021;44:1235–1247.
- Shibata N, Hasegawa Y, Yamada K, Kobayashi H, Purevsuren J, Yang Y, et al. Diversity in the incidence and spectrum of organic acidemias, fatty acid oxidation disorders, and amino acid disorders in Asian countries: Selective screening vs. expanded newborn screening. Mol Genet Metab Rep 2018;16:5–10.
- Kido J, Matsumoto S, Sugawara K, Sawada T, Nakamura K. Variants associated with urea cycle disorders in Japanese patients: Nationwide study and literature review. Am J Med Genet A 2021;185:2026–2036.
- Kido J, Häberle J, Sugawara K, Tanaka T, Nagao M, Sawada T, et al. Clinical manifestation and long-term outcome of citrin deficiency: Report from a nationwide study in Japan. J Inherit Metab Dis 2022;45:431–444.
- Wright CF, West B, Tuke M, Jones SE, Patel K, Laver TW, et al. Assessing the Pathogenicity, Penetrance, and Expressivity of Putative Disease-Causing Variants in a Population Setting. Am J Hum Genet 2019;104:275–286.
- Yamaguchi-Kabata Y, Yasuda J, Uruno A, Shimokawa K, Koshiba S, Suzuki Y, et al. Estimating carrier frequencies of newborn screening disorders using a whole-genome reference panel of 3552 Japanese individuals. Hum Genet 2019;138:389–409.
- Kobayashi K, Sinasac DS, Iijima M, Boright AP, Begum L, Lee JR, et al. The gene mutated in adult-onset type II citrullinaemia encodes a putative mitochondrial carrier protein. Nat Genet 1999;22:159–163.
- Johnston HR, Keats BJB, Sherman SL. Population genetics. In: Pyeritz RE, Korf BR, Grody WW (eds). Emery and Rimoin’s Principles and Practice of Medical Genetics, 7th edn. Academic Press: London, 2019: 359–374.
- Strachan T, Read A. Chapter 12: Human Population Genetics. In Hum Mol Genet. 5th edn. CRC Press: Boca Raton, FL, USA, 2019: 397–417.
- Lu D, Han F, Qiu W, Zhang H, Ye J, Liang L, et al. Clinical and molecular characteristics of 69 Chinese patients with ornithine transcarbamylase deficiency. Orphanet J Rare Dis 2020;15:340.
- Wakutani Y, Nakayasu H, Takeshima T, Adachi M, Kawataki M, Kihira K, et al. Mutational analysis of carbamoylphosphate synthetase I deficiency in three Japanese patients. J Inherit Metab Dis 2004;27:787–788.
- Caldovic L, Morizono H, Panglao MG, Cheng SF, Packman S, Tuchman M. Null mutations in the N-acetylglutamate synthase gene associated with acute neonatal disease and hyperammonemia. Hum Genet 2003;112:364–368.
- Funghini S, Thusberg J, Spada M, Gasperini S, Parini R, Ventura L, et al. Carbamoyl phosphate synthetase 1 deficiency in Italy: clinical and genetic findings in a heterogeneous cohort. Gene 2012;493:228–234.
- Feldmann D, Rozet JM, Pelet A, Hentzen D, Briand P, Hubert P, et al Site specific screening for point mutations in ornithine transcarbamylase deficiency. J Med Genet 1992;29:471–475.
- Yokoi K, Nakajima Y, Yasui T, Yoshino M, Yoshikawa T, Kurahashi H, et al. Novel ARG1 variants identified in a patient with arginase 1 deficiency. Hum Genome Var 2021;8:8.
- Garen A. Sense and Nonsense in the Genetic Code. Science 1968;160:149–159.
- Mort M, Ivanov D, Cooper DN, Chuzhanova NA. A meta-analysis of nonsense mutations causing human genetic disease. Hum Mutat 2008;29:1037–1047.
- Yamaguchi-Kabata Y, Shimada MK, Hayakawa Y, Minoshima S, Chakraborty R, Gojobori T, et al. Distribution and effects of nonsense polymorphisms in human genes. PLoS One 2008;3:e3393.
- Al-Sayed M, Alahmed S, Alsmadi O, Khalil H, Rashed MS, Imtiaz F, et al. Identification of a common novel mutation in Saudi patients with argininosuccinic aciduria. J Inherit Metab Dis 2005;28:877–883.
- Nei M. Molecular population genetics and evolution. Front Biol 1975;40:1–288.
- Kikuchi A, Arai-Ichinoi N, Sakamoto O, Matsubara Y, Saheki T, Kobayashi K, et al. Simple and rapid genetic testing for citrin deficiency by screening 11 prevalent mutations in SLC25A13. Mol Genet Metab 2012;105:553-558.
- Wongkittichote P, Sukasem C, Kikuchi A, Aekplakorn W, Jensen LT, Kure S, et al. Screening of SLC25A13 mutation in the Thai population. World J Gastroenterol 2013;19:7735–7742.
- Kamoshita N, Nomoto A, RajBhandary UL. Translation initiation from the ribosomal A site or the P site, dependent on the conformation of RNA pseudoknot I in dicistrovirus RNAs. Mol Cell 2009;35:181–190.
- Kuchino Y, Muramatsu T. Nonsense suppression in mammalian cells. Biochimie 1996;78:1007–1015.
- Kong A, Frigge ML, Masson G, Besenbacher S, Sulem P, Magnusson G, et al. Rate of de novo mutations and the importance of father’s age to disease risk. Nature 2012;488:471–475.
- Maretty L, Jensen JM, Petersen B, Sibbesen JA, Liu S, Villesen P, et al. Sequencing and de novo assembly of 150 genomes from Denmark as a population reference. Nature 2017;548:87–91.
- Ton ND, Nakagawa H, Ha NH, Duong NT, Nhung VP, Hien LTT, et al. Whole genome sequencing and mutation rate analysis of trios with paternal dioxin exposure. Hum Mutat 2018;39:1384–1392.
- AlTassan R, Bubshait D, Imtiaz F, Rahbeeni Z. A retrospective biochemical, molecular, and neurocognitive review of Saudi patients with argininosuccinic aciduria. Eur J Med Genet 2018;61:307–311.
- Strachan T, Read A. Chapter 10: Gene Regulation and the Epigenome. In Hum Mol Genet. 5th edn. CRC Press: Boca Raton, FL, USA, 2019: 325–57.
- Yngvadottir B, Xue Y, Searle S, Hunt S, Delgado M, Morrison J, et al. A genome-wide survey of the prevalence and evolutionary forces acting on human nonsense SNPs. Am J Hum Genet 2009;84:224–234.
- Cohen JC, Boerwinkle E, Mosley TH, Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease. N Engl J Med 2006;354:1264–1272.
- Nachman MW, Crowell SL. Estimate of the mutation rate per nucleotide in humans. Genetics 2000;156:297–304.
- Kuriyama S, Yaegashi N, Nagami F, Arai T, Kawaguchi Y, Osumi N, et al. The Tohoku Medical Megabank Project: Design and Mission. J Epidemiol 2016;26:493–511.
- Hozawa A, Tanno K, Nakaya N, Nakamura T, Tsuchiya N, Hirata T, et al. Study Profile of the Tohoku Medical Megabank Community-Based Cohort Study. J Epidemiol 2021;31:65–76.
- Morton NE, Crow JF, Muller HJ. An estimate of the mutational damage in man from data on consanguineous marriages. Proc Natl Acad Sci USA 1956;42:855–863.
- Huang L, Wilkinson MF. Regulation of nonsense-mediated mRNA decay. Wiley Interdiscip Rev RNA 2012;3:807–828.
- Stenson PD, Mort M, Ball E V, Chapman M, Evans K, Azevedo L, et al. The Human Gene Mutation Database (HGMD®): optimizing its use in a clinical diagnostic or research setting. Hum Genet 2020;139:1197–1207.
Table 1 Incidence rate and allele frequency (AF) of disease-causing variations for inborn errors of metabolism in Japanese and European populations
|
Gene symbol |
Incidence rate a) |
Qdisease-causing b) [38KJPN] |
Qnonsense b) |
Incidence rate a) |
Qdisease-causing b) |
Qnonsense b) |
|
PHA |
1: 46,000 |
0.005969 |
0.000284 |
1: 5,000 |
0.012716 |
0.000382 |
|
CPS1 |
ND |
0.000751 |
0.000311 |
ND |
0.000515 |
0.000103 |
|
ASS1 |
1: 306,000 |
0.001589 |
0 (no variant) |
1: 60,000 |
0.001455 |
0.000044 |
|
ASL |
1: 1,121,000 |
0.000530 |
0.000104 |
1: 292,000 |
0.004029 |
0.000191 |
|
SLC25A13 |
1: 96,000 |
0.011380 |
0.000853 |
NA (no incidence) |
0.000561 |
0.000176 |
a) Incidence rates in Japan (JP) and Germany (DE) are cited from Shibata et al. [15].
b) Sums for Japanese (38KJPN) and non-Finnish European (NFE) populations are calculated from equation (2) in the text, using observed AF values listed in jMorp and gnomAD v.3.1.2, respectively.
NA, not applicable; ND, no data.
Table 2 Expected and observed incidence rates of cases caused by a nonsense mutation in a Japanese population
|
Gene symbol |
Expected from QQ a) |
Genetically observed b) |
|
|
CPS1 |
1: 2,140,768 |
1: 2,456,221 |
1.07 |
|
ASS1 |
NA (no allele) |
1: 19,649,770 |
0 |
|
ASL |
1: 9,071,118 |
1: 19,649,770 |
1.47 |
|
SLC25A13 |
1: 51,508 |
1: 585,031 |
3.37 |
a) Calculated from equation (5) in the text, using a value of Q (Table 1).
b) Case number in Japan was divided by birth number during the study period. Case numbers were 8, 1, 1, and 36 in a nationwide study of CPS1, ASS1, ASL [16], and SLC25A13 [17], respectively.
NA, not applicable; ND, no data.
(Not answered)
- 3TableS21CodonFrequencies.xlsx
Table S2-1. Codon frequency of the eight genes
- 3TableS5Late230611.xlsx
Table S5 Post-neonatal cases
- 3TableS7230616.xlsx
Table S7 Nonsense variants in populations
- Dataset1.xlsx
Dataset 1
- Dataset2.xlsx
Dataset 2
- Dataset3.xlsx
Dataset 3
- 3SupplementaryInformationI230620.pdf
Supplementary Information_I
- 4SupplementaryInformationIISupplementaryReferences.docx
Supplementary Information II