In this study, a large cohort of 1,004 patients with ALS was enrolled from Xiangya Hospital, Central South University, in either an inpatient or outpatient setting. Each patient got a thorough examination and was diagnosed with ALS by at least two experienced neurologists based on the current Gold Coast criteria (Shefner et al., 2020). Detailed demographic data and clinical information were collected when enrolled and updated at follow-up visits. Participants having known pathogenic mutations in the established ALS causative genes were excluded first. In total, 1,258 neurological disease-free individuals of Chinese ancestry matched by geography were recruited as healthy controls (HCs). This study was approved by the Ethics Committee and the Expert Committee of Xiangya Hospital, Central South University. All participants completed written informed consent in compliance with the Helsinki Declaration.

Genomic DNA was prepared from peripheral blood leukocytes of each subject via standard extraction procedures. The purity and quantity of extracted DNA were detected with a NanoDrop spectrophotometer 1,000 (Thermo Scientific). WES was performed on all participants using a previously published method (Zeng et al., 2019). The resulting data were mapped to the reference genome (GRCh37/hg19). Variants with a depth of coverage <10, allele balance <0.25, or Phred quality score <20 were removed. Proceed to annotate variations with the ANNOVAR software (Wang et al., 2010). After the quality control, totaling 985 patients and 1,224 healthy controls were included for further analysis. The data generation processes have already been described in detail (Liu et al., 2021).

We screened for rare, putative pathogenic variants of TP73 in the Chinese ALS population. Inclusion criteria for the rare, putative pathogenic variants were: (1). the frequency was lower than 0.001 in our in-house and the following public database: the 1,000 Genome Project-East Asian (1000G_EAS), the NHLBI Exome Sequencing Project-East Asian (ESP6500s_EAS), the Exome Aggregation Consortium-East Asian (ExAC_EAS), and the Genome Aggregation Database-East Asian (GnomAD_EAS); (2). absent from HCs (3). located in the protein-coding regions; (4). present in a heterozygous state; (5). annotated as missense, stop gain/loss, frameshift, or splice-site variants; and (6). rare missense variants predicted to be pathogenic by at least five of 11 in silico (Quadri et al., 2018). It’s worth emphasizing that, considering the relatively small number of our control group, we only judged the pathogenicity of a rare variant to be robust if it was absent from HCs.

The common and rare variants were defined according to the minor allele frequency (MAF) in all participants in our cohort (common variants: MAF ≥ 0.01; rare variants: MAF < 0.01).

To evaluate the association of TP73 variants with ALS risk in the Chinese population, we conducted the single common variant association test between the ALS and control groups using the Mixed Linear Model (MLM). The extracted top three principal components of population stratification in the principal component analysis (PCA) were included as covariates in our cohort. The possible effect of common variants inside exon-intron boundaries on the TP73 gene expression was assessed by Expression Quantitative Trait Locus (eQTL) analysis (Zhu et al., 2016), which was widely utilized to compare the gene expression levels among individuals with different genotypes. All data were available from the Genotype-Tissue Expression (GTEx) database.

For the gene-based rare variants association test, the cumulative burden of rare variants across defined genomic regions of TP73 between cases and controls was evaluated using the optimal sequence kernel association test (SKAT-O) implemented in the R package SKAT (Lee et al., 2012). The SKAT-O test maximized the test power of detecting the target gene by unifying the advantages of both the burden test and SKAT (Lee et al., 2012). Gender, age, and WES coverage were considered as possible covariates for adjustment before computing the p value.

Descriptive statistics are reported as mean ± SD or median ± SD (standard deviation, SD) for continuous variables and percentages for categorical variables. The comparison of continuous variables was assessed by Student’s t-test. For categorical variables, the Fisher’s exact test or Chi-square test was used to verify the significant differences between the two compared groups. Statistical analyses were carried out in SPSS (version 26.0) software. All tests were two-tailed, and significance was set at p < 0.05.

In all, 985 patients with ALS and 1,224 healthy controls were analyzed in our study. The 985 ALS patients were on average 54.2 years old at the time of onset, and 56.2 years old at the time of sampling; the geographically matched controls were older (mean age 68.47 years). 75.3% (740/985) of our ALS cohort had limb onset, while 19.39% (191/985) had bulbar onset. Detailed demographic information for all individuals is shown in Supplementary Table S1.

We screened all TP73 exon regions and their surrounding sequences in each ALS patient. A total of six rare, heterozygous putative pathogenic mutations that fulfilled pathogenicity criteria were discovered among six unrelated sALS patients but not in HCs. The overall frequency of patients with rare pathogenic TP73 mutations was 0.6%. All these six variations were protein-altering missense mutations. And the majority of them substituted amino acids that were highly conserved across species. The six variants were c.187G > A (p.A63T), c.1226C > T (p.P409L), c.1613G > A (p.R538H), c.1628G > A (p.R543Q), c.1679 T > C (p.L560P), and c.1736G > A (p.R579H). The variant, c.187G > A (p.A63T), was novel and absent from all databases. Three of these six variations (50%), namely c.187G > A (p.A63T), c.1226C > T (p.P409L), and c.1613G > A (p.R538H) have never been reported in association with the ALS-FTD spectrum. Interestingly, when investigating the distribution of these six potential pathogenic loci present in TP73, we discovered that a significant portion of them (4/6, 66.7%) were located inside exon 14, which differed from earlier studies. Consequently, we hypothesize that exon 14 of TP73 might be a mutant hotspot in our ALS series. The location and pathogenicity information of these six variants identified in this study were summarized in Figure 1 and Supplementary Tables S2, S3 (Liu et al., 2015).

For the six patients carrying potential pathogenic TP73 variants described above, the age at disease onset ranged from 32 to 69 years old. We observed a completely masculine involvement since all six patients were men. In addition, we screened for other pathogenic mutations in the known ALS genes in these patients (Supplementary Table S4). There were no reported pathogenic mutations in the causative genes for ALS in each of these six individuals. Notably, two of the six patients (patient A0048 and patient S7918) had multiple variants in ALS-related genes with uncertain pathogenicity (patient A0048: KIF5A c.A86G: p.K29R and TP73 c.1613G > A: p.R538H; patient S7918: CCNF c.2199delC: p.D733fs and TP73 c.187G > A: p.A63T), and the remaining four patients only had TP73 mutations. Considering the burden of multiple variants in ALS-causing genes on the disease expression, including the age at onset, progression rate, and survival, we separated these six patients into two subgroups for further clinical phenotype analysis (one subgroup with only TP73 mutations, the other with multiple mutations in both TP73 and other ALS-related genes) (Cady et al., 2015; Pang et al., 2017; Naruse et al., 2019). When the clinical phenotype analysis was restricted to patients with ALS with only TP73 mutations, a distinct clinical profile emerges, with a considerably higher mean age at onset (61.5 ± 7.72 years), a shorter survival time (19.25 ± 10.01 months), and invariably limb onset. In the other subgroup, in contrast to patients carrying only TP73 mutations, the age at onset in patients with ALS with multiple mutations is accelerated (patient A0048: 32-year-old, patient S7918:38-year-old). Furthermore, one of the six patients, S0423, suffered from cognitive impairment and behavioral problems and was eventually diagnosed with ALS-FTD. Most of these affected individuals exhibited typical symptoms and signs of the simultaneous impairment of upper and lower motor neurons. The clinical features of the six sALS patients were listed in Table 1.

The variant, c.1226C > T (p.P409L), was identified in patient S0423, diagnosed with ALS-FTD. He was a farmer without specific environmental exposure. At the age of 65, he presented with muscle weakness in his upper limbs initiated from bilateral hands, which quickly extended to all four extremities in only 1 month. He then developed dysarthria and dysphagia 2 months later. Concurrently, his primary care providers complained of his language difficulties, behavioral changes, and cognitive impairment, which included trouble retrieving words, effortful speech restricted to short, simple sentences, hyperphagia, irritability, attacks on others, decreased memory, dropped comprehension, inability to find out things, and failure to understand some instructions. At the time of the first observation after 8 months of onset (at 66 years old), he was unable to walk or talk, and was reliant on family members for his daily requirements. On comprehensive neurological evaluation, he showed signs of upper motor neuron (UMN) damage and lower motor neuron (LMN) depletion features, such as positive palmomental reflex, hyperreflexia, and muscle atrophy of all four extremities. Electromyogram (EMG) revealed abundant and diffuse ongoing denervation (spontaneous potentials) and chronic reinnervation changes in all four segments (bulbar, cervical, thoracic, and lumbar). Brain magnetic resonance imaging (MRI) scan indicated mild age-related brain atrophy (Supplementary Figure S1). A battery of neuropsychological tests, including the Edinburgh Cognitive and Behavior ALS Screen (ECAS), the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and the Frontal Assessment Battery (FAB) was not completed due to his inability to communicate. He eventually died of respiratory failure 13 months after disease onset.

Another two male patients, S3513 and S4096, carrying variants c.1736G > A (p.R579H) and c.1628G > A (p.R543Q) separately, had similar clinical manifestations. They all began with muscle weakness in the right hand, at the ages of 61 and 69, respectively. They both developed muscle weakness and atrophy in all four limbs over the next year, along with extensive fasciculation, and dysarthria. Patient S3513 subsequently reported simultaneous involvement of the contralateral arm, and bilateral legs, accompanied by fasciculation, 5 months after disease onset. He developed dysarthria in the seventh month of onset, at which point the weakness of four limbs aggravated: he was unable to lift heavy objects with his upper limbs and had difficulty walking independently. For patient S4096 with a right-hand onset, his left upper limb and bulbar (choking) were reported as the second symptomatic sites. Following that, at the fifth month of the disease course, the weakness continuously progressed to both lower limbs, and dysarthria appeared. Their bilateral deep tendon reflexes were brisk, the Hoffman sign and palmomental reflex were present. EMGs both showed abundant and diffuse ongoing denervation as well as chronic reinnervation alterations at four segments. However, brain MRI revealed that patient S4096 with the c.1628G > A (p.R543Q) variant had mild brain atrophy (Supplementary Figure S1), whereas patient S3513 with the c.1736G > A (p.R579H) had no obvious abnormalities. ECAS score of patient S4096 was 56/136, and each subgroup score of this scale suggested that the decline was driven from executive dysfunction and memory impairments, rather than behavioral dysfunction. In the end, patient S3513 died at 13 months after onset and 17 months for patient S4096.

Another patient (S5941), carrying c.1679 T > C (p.L560P) variants, was a male, who complained of progressive muscle weakness and atrophy in his right arm at the 51-year-old. The weakness gradually extended to his bilateral lower limbs, making stair climbing difficult. Besides, he developed autonomic dysfunction, manifesting as sometimes excessive sweating during the day and heat intolerance, with no nocturnal sweating or cold intolerance, as revealed by the Scale for Outcomes in Parkinson’s disease (PD) for Autonomic Symptoms (SCOPA-AUT), a self-reported questionnaire widely used for the assessment of autonomic function in neurodegenerative disease (Visser et al., 2004; Damon-Perriere et al., 2012; Del et al., 2020; Assante et al., 2022). Neurological examination revealed no evidence of upper motor neuron damage, and EMG demonstrated abundant and diffuse ongoing denervation and chronic reinnervation changes in three segments (cervical, thoracic, and lumbar). No symptoms of dysphagia or dysarthria were found at the most recent follow-up visit after 34 months of onset. Based on all the available information thus far, he was clinically diagnosed with progressive muscular atrophy (PMA).

The remaining two patients, both harbored multiple variants in TP73 and other ALS-related genes. Patient A0048 who carried the c.1613G > A (p.R538H) variant, and patient S7918 who had the c.187G > A (p.A63T) variant, were both young men with no family history of the condition, starting with ALS phenotype at 32 and 38-year-old, respectively. Patient A0048 first displayed weakness in his left lower extremity, and then spread fast in the early disease course. By the fourth month from symptom onward, he had weakness in all four limbs; he then gradually experienced bulbar muscle weakness, presenting with dysarthria, dysphagia, and bucking during the next 2 years. Patient S7918 showed bulbar onset, initially with dysarthria and dysphagia, then developed asymmetric weakness and atrophy in bilateral lower limbs and the right arm within 3 years, accompanied by extensive fasciculation. Neurological evaluation and EMGs on them both indicated evident signs of involvement of both upper and lower motor neurons. According to ECAS, patient A0048 had executive function deficits and memory impairments (total score: 63/136, subscore of executive function:14/60; subscore of memory: 3/24), whereas patient S7918 had no cognitive changes.

We performed the single-variant analysis of each common variant discovered in the exon and near exon-intron boundary regions to ascertain the association between common variations of TP73 and ALS. As shown in Figure 2, 36 common TP73 variants were identified in our cohort (Figure 2, Supplementary Table S5). Then, utilizing PLINK 1.90, we obtained linkage disequilibrium (LD) statistics for these 36 common variants to plot linkage blocks (Supplementary Table S6). As 24 of the 36 common variants were mapped to 6 separate blocks, the estimated number of independent tests was 18, and the corresponding assumed Bonferroni-corrected significance threshold of p was 0.0028 (0.05/18). The resulting data revealed that none of the common variants inside the exon regions displayed significant univariate association with ALS. While two common variants rs2181486 and rs2146657 in the same block at exon-intron boundary, attained Bonferroni-corrected statistical significance (Table 2). Taking into consideration these two common variants in nearly perfect LD (the R² of these two variants is 0.9850), we next chose the rs2181486 mutant locus to search the publicly accessible GTEx database to find if these two common variations impact the expression of TP73. Of note, based on the calculation of the eQTL algorithm, the rs2181486 variants might give rise to a decreased TP73 protein level in the cerebral cortex, spinal cord, as well as skeletal muscles (Table 3). These results provided suggestive evidence for association of ALS and common TP73 variants in the near-exon intron boundaries.

We applied the SKAT-O test to compare the aggregate burden of rare variants residing in the coding regions, untranslated regions, and intron-exon boundaries of TP73 between ALS cases and HCs, separately. Compared with HCs, the frequency of carriers of rare TP73 variants in the coding regions and intron-exon boundaries was not significantly higher in the ALS group. We observed an enrichment of rare TP73 variations in the UTRs among ALS cases, with a p value of 0.046 for the cumulative burden of rare TP73 variations in the UTRs between the two compared groups, indicating that rare TP73 UTRs variants were significantly associated with ALS (Supplementary Table S7).