Auditory neuropathy (AN) is a type of auditory disorder mainly characterized by impaired speech comprehension ability, which manifests as a group of specific syndromes in audiology, also known as the auditory neuropathy spectrum disorder (ANSD) (Starr et al., 1996; El-Badry and McFadden, 2009; Talaat et al., 2009).

The pathophysiology of AN indicates that the outer hair cells show normal function, whereas the inner hair cells and auditory nerves as well as their synapses are dysfunctional. The clinical manifestations include: (1) the degree of impaired speech recognition rate is inconsistent with the degree of hearing impairment, and the word comprehension ability and speech discrimination ability are poor, especially in noisy environment (Elrharchi et al., 2020); (2) the audiological examination is characterized by normal cochlear microphonic (CM) and meaningful otoacoustic emission (OAEs), but the auditory brainstem response (ABR) waveform is either missing or illegible in the patient (Starr et al., 1996). Studies showed that the prevalence rate of AN in neonates is 0.009%, and about 6.5% of the congenital sensorineural hearing loss is caused by AN (Boudewyns et al., 2016). AN is also one of the refractory diseases leading to hearing and speech communication disorders in both infants and adolescents, accounting for 10% of permanent hearing loss in children (Chinese Multi-center Research Collaborative Group on Clinical Diagnosis and Intervention of Auditory Neuropathy, 2022).

According to the age of onset, cases of AN are divided into two groups, i.e., congenital AN and late-onset AN. Studies have shown that the etiology of AN is complex, with about 48% as idiopathic, 10% related to various factors, such as ototoxic drugs and metabolic, immune, and infectious factors, and about 42% as genetic (Moser and Starr, 2016). There are three types of inheritance pattern for AN: autosomal recessive, autosomal dominant, and X-linked recessive (Wang et al., 2020). To date, among the over 20 genes associated with AN, apoptosis inducing factor mitochondrion associated 1 (AIFM1) is considered as the most prevalent gene associated with the late-onset AN and X-linked recessive inheritance mode (Zong et al., 2015; Wang et al., 2020).

First cloned and named in 1999, AIFM1 is located in the Xq25-q26 region of human chromosomes (Susin et al., 1999). The AIFM1 gene is 36.471 kb in length, containing a total of 16 exons and encoding a 613-amino acid apoptosis-inducing factor (AIF). The AIF is a flavin protein with oxidoreductase activity, composed of two flavin adenine dinucleotide (FAD) domains, a nicotinamide adenine dinucleotide (NADH) binding domain, a C-terminal domain with oxidoreductase activity, and a mitochondrial localization signal (MLS) at the N-terminal (Yuste et al., 2011). The AIFM1 has two main functions. First, it is involved in the caspase-independent death effect when mitochondria are transferred to nuclei under apoptosis stimulation, resulting in caspase-independent programmed cell death (Joza et al., 2001). Second, as a NADH oxidoreductase that is dependent on FAD, AIFM1 is crucial for oxidative phosphorylation, redox control, and respiratory chain activity in the organism (Ghezzi et al., 2010; Itai et al., 2011).

To date, a total of 19 AIFM1 variants have been identified as pathogenic factors for AN (https://www.hgmd.cf.ac.uk/; accessed on 03 October 2022). It is conceivable that there are other AIFM1 variants still remaining uncovered responsible for pathogenesis of AN. Therefore, in this study, we attempted to detect the pathogenic variants of AIFM1 in a Chinese family with AN enrolled in our hospital and to further investigate the effects of this variant on AN development.

As a specific type of hearing loss, AN is characterized by impaired language comprehension. In most cases, AIFM1 gene is responsible for the cases of late-onset AN (Li et al., 2021), which contributes to the maintenance of normal mitochondrial morphology and physiology as well as caspase-independent apoptosis (Rance and Starr, 2015). In the present study, the possible genetic cause and molecular mechanisms underlying the development of AN were explored in a Chinese family with AN.

The onset ages of patients with late-onset AN are mostly in the range of 5–20 years old with diverse clinical manifestations (Wang et al., 2020). Studies have shown that about 92.5% (360/389) of AN patients are presented with bilateral hearing loss (Bing et al., 2019) with audiogram mainly of low-frequency ascending type and diverse pure tone hearing manifestations which are identified as normal or mild to severe hearing loss (Wang et al., 2003; Wang et al., 2006; Zong et al., 2015; Wang and Starr, 2018), accompanied by significant decline in speech recognition ability (Starr et al., 1996). The results of our study were consistent with the above characteristics reported in the literature, showing that the proband of the family in this study was presented with typical symptoms of late-onset AN, i.e., the onset age of the proband was 15 years old, the audiogram showed low-frequency ascending type (Figure 1C), the PTA showed low-frequency mild hearing loss in both left and right ears with a disproportionate decrease in SDS (Figure 1C), the ABR waveform was illegible (Figure 1B), and the DPOAE was normal (Figure 1E). It was worth noting that although low-frequency ascending type was the most common type of hearing pattern of AN patients, including the patients involved in this study, there are still great differences in hearing patterns among AN patients. For example, a follow-up study found that some AN patients initially showed a low-frequency ascending hearing pattern, but with the progress of the disease, their high-frequency hearing was decreased rapidly to show a descending hearing pattern (Wang et al., 2020). Furthermore, some AN patients present with a descending hearing pattern involving the full frequency were accompanied by severe motor development impairment and mental retardation (Wang et al., 2019). Furthermore, consistent with previous studies (Wang and Starr, 2018), the threshold of ASSR results in our study was also significantly higher than that of PTA (Figure 1D). Moreover, another important clinical manifestation revealed in our study was that the proband complained that his speech recognition ability was worsened in the noisy environment, which was in accordance with the results of a controlled study of AN patients (Lan et al., 2019), suggesting that the speech recognition ability of AN patients with sound speech recognition rate in quiet environment was significantly decreased in the noisy environment. Meanwhile, the proband also showed tinnitus and aggravation under noise environment, which was consistent with the previous study (Xie et al., 2019).

Since the genetic factors account for 42% of the etiology of AN (Moser and Starr, 2016), the genomic DNA of the proband was extracted for NGS analysis. Combined with Sanger sequencing technology, the results showed that the c.1367A > G variant of AIFM1 carried by the proband came from the mother, and this variant was co-segregated with the phenotype of the family and was not identified in a total of 200 healthy controls. Base on the literature and publicly available databases, the variant c. 1367A > G has not been reported, indicating that this variant of AIFM1 gene identified in this study was a novel one, expanding the pool of the mutations of AIFM1 gene. Meanwhile, this variant was predicted as deleterious by REVEL, SIFT, Polyphen-2, and Mutation Taster (Table 1), suggesting the potentially pathogenic property of this variant. To date, a total of 19 pathogenic variants of AIFM1 related to AN have been reported in literature (Elrharchi et al., 2020; Wang et al., 2020) and databases (Figure 6). The missense mutation (c.1367A > G) revealed in our study provided additional powerful support for molecular diagnosis of AN in clinical settings. These results evidently showed that a novel pathogenic AIFM1 variant, i.e., a missense mutation (c.1367A > G) in our study was first identified.

The variant of c.1367A > G was located in exon 13 of AIFM1 gene, causing the amino acid change of p. D456G in the second FAD domain of AIFM1 protein. Asp is an acidic polar aliphatic amino acid (Agius et al., 2018), which could be used as one of excitation-related neurotransmitter receptors in mammals (Kondoh et al., 2010), while Gly is a kind of non-polar aliphatic amino acid and an inhibitory neurotransmitter in the central nervous system (Roepke et al., 2008). As reported in the previous study (Xu et al., 2017), substitutions of amino acids of different polarities could largely affect the protein structures and functions of target protein. The change from polar Asp to non-polar Gly in this study could largely affect the structure, stability, and activity of the target protein. The 3D protein structure models (Figure 3C) further supported these findings, showing that the secondary and tertiary structures of the protein were largely changed. Studies showed that a total of 19 pathogenic variants previously identified mainly located in the NADH and second FAD domains of AIFM1 (Elrharchi et al., 2020; Wang et al., 2020), indicating that the NADH and FAD domains play an essential role for the normal function of AIFM1 protein. The novel variant identified in this study was also located in the second FAD domain (Figure 6), which is required for FAD-dependent NADH oxidoreductase (Zong et al., 2015), suggesting that the novel variant could probably change the redox activity of AIFM1 to cause abnormal functions of the target protein.

Previous studies showed that changes in key amino acids could affect the conformation of the skeleton and thus disturb the local structure of the protein, even interfere with the inter- and intra-molecular interactions of proteins (Sapra et al., 2008), while AIFM1 usually interacts with a variety of proteins such as coiled-coil-helix-coiled-coil-helix domain containing 4 (CHCHD4) and Poly (ADP-ribose) polymerase 1 (PARP1). Previous coevolutionary analysis and structural modeling (Hangen et al., 2015; Salscheider et al., 2022) indicated that the N-terminal region of CHCHD4 may form a β-hairpin to bind the C-terminal motif of AIFM1. Furthermore, studies showed that AIFM1 knockdown affected the interaction between CHCHD4 and its substrate and then the biogenesis of electron transport chain, suggesting the essential nature of the interaction between AIFM1 and CDCDH4 in this process (Salscheider et al., 2022). As shown in Figure 3C, the c.1367A > G variant altered the secondary structure of the C-terminus of AIFM1, which may affect its binding to the N-terminus of CDCDH4. PARP1 is a nuclear protein involved in DNA damage, while PAR polymers synthesized by PARP1 bind to AIFM1 on the mitochondrial membrane with damaged DNA, then AIFM1 is released to cytoplasm and translocated to the nucleus, resulting in large-scale DNA fragmentation (Wang et al., 2011; Mashimo et al., 2021). Therefore, changes in AIFM1 protein structure could affect its binding to PAR polymer and thus its ability to induce apoptosis. Furthermore, Asp at position 456 was evolutionarily highly conserved in eight vertebrate species, indicating that amino acid changes may affect protein structure and ultimately protein function. These results indicated that the c.1367A > G variant appeared to be critical for the structures and functions of AIFM1.

As an apoptosis-inducing factor, the AIFM1 is a flavin protein localized in the inner membrane space of mitochondria and is transported from mitochondria to the nucleus to induce cell apoptosis upon the occurrence of the apoptotic injury (Joza et al., 2009). The immunofluorescence analysis showed that the subcellular localization of p. D456G AIFM1 was consistent with that of the wild-type AIFM1, probably due to the MLS (Yuste et al., 2011) and nuclear localization signal (NLS) (Wang et al., 2016) of its N terminus. In a study of AIFM1 mutation causing Cowchock Syndrome, the mutant AIFM1 showed more AIFM1-positive inclusions detected by immunofluorescence, indicating a higher propensity to translocate to the nucleus (Rinaldi et al., 2012). However, our study showed that there was no significant difference between the mutant group and the wild type group, and the subsequent TUNEL assay showed that p. D456G AIFM1 and WT AIFM1 revealed the same TUNEL-positive apoptotic bodies. This difference was probably due to the fact that AIFM1 could cause a variety of diseases, such as childhood cerebellar ataxia (Heimer et al., 2018), mitochondrial encephalomyopathy (Ghezzi et al., 2010), motor neuropathy (Diodato et al., 2016; Hu et al., 2017; Sancho et al., 2017), Cowchock syndrome (Rinaldi et al., 2012) and auditory neuropathy (Zong et al., 2015; Wang et al., 2020), with different severity and potential pathogenesis.

Furthermore, the immunofluorescence analysis also showed that the fluorescence intensity of p. D456G AIFM1 was weaker than that of WT AIFM1, indicating that the variant of p. D456G changed the expression level of AIFM1. To further investigate the effect of p. D456G mutation on AIFM1 protein expression, western blot analysis was performed. The results were consistent with those of the immunofluorescence analysis, showing that the protein expression of p. D456G AIFM1 was significantly lower than that of wild-type AIFM1. As predicted by our 3D protein model, the p. D456G mutation affected the secondary/tertiary structure of the protein, resulting in degradation or low expression of the protein, which was further supported by the results of western blot. Previously, many scholars have studied the effect of AIFM1 mutation on protein expression with varied results. For example, patients with p. G338E (Diodato et al., 2016) and p. F210S (Sancho et al., 2017) variants all showed motor nerve lesions, and the expression level of AIFM1 was decreased. These results were consistent with the findings revealed in our study. Interestingly, in a study on axonal polyneuropathy caused by AIFM1 p. F210L variant (Hu et al., 2017), no significant change was detected in the AIFM1 protein expression, independent of the changes of different amino acids at the same site. Furthermore, in a study of mitochondrial encephalomyopathy caused by R201 del variant in AIFM1, the protein expression was not significantly altered (Ghezzi et al., 2010).

Under normal circumstances, AIFM1 is first transcribed and translated into a 67-kD precursor molecule (Liu et al., 2018), which is then transported to mitochondria through the N-terminal domain MLS sequence. The mitochondrial processing peptidase (MPP) cleaves AIFM1 in mitochondria into 62-kD mature AIFM1, which acts as the FAD-dependent NADH oxidoreductase to contribute to the stabilization and maturation of mitochondrial oxidation of respiratory chain complex I as well as the removal of peroxide from the mitochondria (Susin et al., 1999). When cells are damaged by apoptosis, mature AIFM1 is cleaved into soluble apoptotic precursor protein, i.e., the truncated AIF (tAIF) of about 57 kD (Susin et al., 1999; Joza et al., 2001). The tAIF is released from mitochondria into cytosol and nucleus to induce two typical caspase-independent apoptosis phenomena: chromatin condensation and fragmentation of large DNA fragments of approximately 50 kD (Yuste et al., 2011).

A study of AIFM1 p. P488L variants in auditory neuropathy and peripheral neuropathy showed that AIFM1 p. P488L variants caused a mild increase in the rate of caspase-independent apoptosis in cells (Wang et al., 2019). Because the immunofluorescence and WB assays showed that the p. D456G variant reduced AIFM1 protein expression, the Tunel assay was performed to evaluate its adverse effect on cell survival. The results showed that the ability of AIFM1 to induce apoptosis was not significantly affected by the p. D456G mutation, showing the same effect on cell survival as that of WT AIFM, indicating that the p. D456G variant did not affect cell survival. This may also be the reason for the generation of mild symptoms of AN patients. Previous studies showed that the AIFM1 R201del variant caused severe mitochondrial encephalomyopathy and significantly increased parthanatos-linked cell death (Ghezzi et al., 2010). In the study of Cowchock Syndrome caused by AIFM1 p. E493V variant, Tunel assay of muscle biopsy showed a large number of apoptotic cells in the samples of the mutated individuals, while little or no staining was detected in the muscles of healthy controls (Rinaldi et al., 2012). However, in patients with distal motor neuropathy carrying the p. F210S mutation, the mutation did not enhance the apoptosis (Sancho et al., 2017), which was consistent with our study. Similarly, as mentioned above, the p. F210S mutation also reduced the protein expression level with mitochondrial fragmentation also observed in fibroblasts of this patient, suggesting that the development of distal motor neuropathy in this patient may be related to the defective mitochondrial respiration (Sancho et al., 2017). Therefore, we hypothesized that this mutation might affect the role of AIFM1 in oxidative phosphorylation, redox control, and respiratory chain activity. Study showed that knockdown of AIFM1 could attenuate mitochondrial respiration and ATP production, ultimately affecting cell functions (Zong et al., 2020). Furthermore, some studies showed that in the auditory pathway, including inner hair cells, glial cells in neural pathways, and spiral ganglion cell (SGC), the normal energy metabolism was the key to maintain its physiological activity (Yang et al., 2015). Therefore, as suggested in the previous studies (Zong et al., 2015), the variant revealed in our study may cause AN by affecting the function of AIFM1 protein in mitochondria, ultimately affecting the mitochondrial respiration and ATP synthesis in inner ear. Moreover, as mentioned above, the 3D prediction results showed that the p. D456G mutation affected both the secondary and tertiary structures of the C-terminal, while the binding of the N-terminal of CHCHD4 and the C-terminal of AIFM1 played an important role in the biogenesis of mitochondrial respiratory chain complex Ⅰ. The decreased expression level of AIFM1 could affect this interaction, and then affect the function of the mitochondrial respiratory chain. These functional experiments further demonstrated that the novel mutation identified in this study could cause damage to AIFM1 protein, which could be the underlying genetic etiology of this family with AN, and further experiments of mitochondrial respiratory chain activity and oxidative phosphorylation are necessary to explore the pathological mechanism of this mutation.

In summary, we identified a novel c.1367A > G AIFM1 variant in a Chinese family with AN by targeted capture sequencing, expanding the AN-related mutation spectrum of AIFM1. Bioinformatics prediction and functional analysis showed that AIFM1 c.1367A > G was a pathogenic mutation and may be the genetic cause of AN in this Chinese family, providing additional molecular and clinical evidence to support the establishment of a strong genotype-phenotype correlation for AN.