Introduction

Wolfram Syndrome (OMIM 222300) is a rare autosomal recessive disease characterized by several symptoms, including diabetes mellitus, optic nerve atrophy, diabetes insipidus, hearing loss, and neurological dysfunction. Two types of Wolfram syndrome have been identified: type 1 and type 2, caused by pathogenic variants of WFS1 and CISD2, respectively. Most Wolfram syndrome cases are classified as type 1, so we will refer to them as “Wolfram syndrome” throughout this review. Since the discovery of WFS1 as the causative gene of Wolfram syndrome, several animal and cellular models have been established. These disease models are indispensable tools in the research on Wolfram syndrome. However, a comprehensive summary of the characteristics and phenotypes of these experimental models is lacking. Therefore, this review aims to fill this gap by providing an overview of the animal and cellular models of Wolfram syndrome established to date. In this review, we first discuss (1) the clinical symptoms of Wolfram syndrome, (2) the function of WFS1 and the pathophysiology caused by its abnormalities, and (3) the characteristics of disease models of Wolfram syndrome.

Clinical features of Wolfram syndrome and the discovery of WFS1

This section describes the clinical features of Wolfram syndrome and the disease concept associated with the causative gene WFS1.

Major symptoms of Wolfram syndrome

The first cases of Wolfram syndrome were reported by Wolfram and Wagener (1938). The report described four siblings who developed juvenile-onset diabetes mellitus and optic nerve atrophy. Subsequently, these patients also developed hearing loss and neurogenic bladder (Paley and Tunbridge 1956). In 1956, Paley and Tunbridge reported two additional cases of diabetes and optic nerve atrophy, suggesting a hereditary component (Paley and Tunbridge 1956). In 1977, a comprehensive review of 91 cases by Cremers et al. (1977) led to the official designation of this condition as “Wolfram syndrome.” Since then, numerous reports on the clinical symptoms of patients have emerged. The first nationwide survey was conducted and reported by Barrett et al. (1995), revealing a prevalence of Wolfram syndrome at 1 in 770,000. Typically, patients with Wolfram syndrome develop diabetes mellitus at a median age of 6 years and optic atrophy around 11 years. Diabetes insipidus and sensorineural hearing loss typically manifest in the second decade of life, while renal tract abnormalities and neurological complications such as cerebellar ataxia appear by the fourth decade. Neurological and urinary tract manifestations are the most common causes of morbidity and mortality (Kinsley et al. 1995). Wolfram syndrome is characterized by a reduced size of intracranial volume, particularly in the brainstem and cerebellum, leading to balance impairment (Hershey et al. 2012; Pickett et al. 2012). However, the severity of these symptoms can vary among individuals with Wolfram syndrome (De Franco et al. 2017).

Discovery of WFS1

Wolfram syndrome was previously believed to be a mitochondrial disease due to the similarities in symptoms and the presence of mitochondrial DNA deletions in some patients (Bu and Rotter 1993; Bundey et al. 1993; Rötig et al. 1993; Vora and Lilleyman 1993). However, subsequent cases demonstrated the absence of mitochondrial DNA abnormalities, and the inheritance pattern suggested an abnormality in the nuclear genome of patients with Wolfram syndrome. Linkage analysis suggested the presence of a causative gene for Wolfram syndrome on chromosome 4p16 (Polymeropoulos et al. 1994). In 1998, Inoue et al., followed soon after by Strom et al., identified the WFS1 gene (Inoue et al. 1998; Strom et al. 1998) (Fig. 1). Since the discovery of WFS1, the understanding of the pathogenesis of Wolfram syndrome has remarkably advanced. Approximately 200 WFS1 variants associated with Wolfram syndrome have been reported, mostly found in exon 8 of WFS1 (Smith et al. 2004). WFS1 encodes the protein WFS1, also known as wolframin, which is an endoplasmic reticulum (ER)-membrane protein with multiple transmembrane domains comprising 890 amino acids (Strom et al. 1998; Takeda et al. 2001). WFS1 is expressed ubiquitously in various tissues, with the highest levels observed in the pancreas, heart, and brain, particularly in the hippocampus, amygdaloid area, and olfactory tubercle (Hofmann et al. 2003; Takeda et al. 2001). Recent studies have reported predominant localization of WFS1 in the ER membrane, particularly in the mitochondria-associated ER membranes, which serve as contact sites between the ER and mitochondria (Angebault et al. 2018; La Morgia et al. 2020).

Fig. 1
figure 1

The structure of WFS1. WFS1 consists of eight exons, with the largest one being exon 8. The location of reported variants is depicted in different colors. The image has been adapted from Clinvar (https://www.ncbi.nlm.nih.gov/clinvar/), and the information was accessed on July 23, 2023

Full size image

Wolfram syndrome and WFS1-related disorders

Wolfram syndrome, a rare disorder caused by recessive WFS1 variants, is now recognized to exhibit a phenotypic spectrum. Recent findings suggest that its frequency may be higher in certain races and populations than previously assumed (Bansal et al. 2018; De Franco et al. 2017; Li et al. 2020; Marchand et al. 2021). Moreover, there are related disorders caused by a dominant inheritance form of WFS1, known as “WFS1-related disorders” or “Wolfram-like syndrome.” These disorders manifest with one or more symptoms observed in Wolfram syndrome. The severity and symptoms of WFS1-related disorders can vary, ranging from mild diabetes mellitus, hearing loss, congenital cataracts, and optic atrophy developing independently to severe neonatal cases that exhibit all these symptoms (Kobayashi et al. 2018; Mets et al. 2010; Morikawa et al. 2017). It has also been reported that specific heterozygous WFS1 variants can lead to nonsyndromic low-frequency sensorineural hearing loss (Bespalova et al. 2001; Cryns et al. 2002; Young et al. 2001).

WFS1 functions and their anomalies

Some WFS1 variants have also been implicated in other types of diabetes. Several large-scale genome-wide association studies have reported that several single nucleotide polymorphisms (SNPs) in WFS1 are associated with an increased genetic risk for type 2 diabetes mellitus (T2DM), irrespective of racial differences (Cheurfa et al. 2011; Elek et al. 2015; Heni et al. 2010; Lee et al. 2008; Long et al. 2012; Lyssenko et al. 2008; Sandhu et al. 2007; Sparsø et al. 2008; van Hoek et al. 2008; Westermark et al. 2010). However, a meta-analysis has shown that two specific WFS1 SNPs, rs734312 and rs10010131, have a significant protective effect against the risk of developing T2DM (Cheng et al. 2013). Accordingly, understanding the pathogenesis of Wolfram syndrome and WFS1-related disorders may provide valuable insights into the pathogenesis of other types of diabetes, including T2DM. In this section, we describe the known physiological functions of WFS1 and the pathological conditions that arise due to its abnormalities.

Interaction between WFS1 and Ca2+ channels or pumps

The ER serves as an intracellular Ca2+ store, and WFS1 is expressed on the ER membrane. Therefore, numerous studies have explored the relationship between WFS1 and intracellular Ca2+ homeostasis (La Morgia et al. 2020; Nguyen et al. 2020; Osman et al. 2003). The first study to demonstrate a connection between WFS1 and intracellular Ca2+ was reported by Osman et al. (2003), who showed elevated levels of intracellular Ca2+ in Xenopus oocytes overexpressing WFS1.

Takei et al. first revealed the relationship between Ca2+ concentration in the ER and the expression level of WFS1 (Takei et al. 2006). They found that WFS1 regulates ER Ca2+ storage and cytosolic Ca2+ homeostasis by increasing the Ca2+ uptake and store-operated Ca2+ entry (SOCE). This regulation involves the modulation of Ca2+ pumps and channels, such as sarcoendoplasmic reticulum ATPase (SERCA) and inositol 1,4,5-trisphosphate receptor (IP3R), localized on the ER membrane (Kunnappallil and Hasan 2022). WFS1 deficiency results in upregulated SERCA expression, leading to increased Ca2+ pumping into the ER (Zatyka et al. 2015) (Fig. 2A). Additionally, WFS1 forms a complex with neuronal calcium sensor 1 (NCS1) and IP3R to facilitate Ca2+ transfer between the ER and mitochondria. WFS1 deficiency reduces the expression level of NCS1, followed by impaired ER-mitochondrial contact and subsequent Ca2+ uptake into the mitochondria (Angebault et al. 2018) (Fig. 2B). Therefore, therapeutic approaches targeting intracellular Ca2+ signaling have been explored as potential treatments for Wolfram syndrome (Abreu et al. 2021; Akiyama et al. 2009; Clark et al. 2017; Crouzier et al. 2022a; Lu et al. 2014; Nguyen et al. 2020). Apart from its interactions with Ca2+ pumps and channels, WFS1 also interacts with the Na+/K+ ATPase beta-1 subunit, the V1A subunit of the H1 ATPase, and the voltage-dependent anion channel isoform 1 (VDAC1) (Gharanei et al. 2013; Zatyka et al. 2008; Zatyka et al. 2023). In WFS1-depleted cells, the H1 ATPase V1A subunit is degraded more rapidly (Gharanei et al. 2013).

Fig. 2
figure 2

Multiple functions of WFS1. A WFS1 regulates Ca2+ uptake in the ER by modulating SERCA activity. WFS1 dysfunction increases ER Ca2+ levels and SOCE. Increased Ca2+ concentration activates Calpain-2, leading to cell death. B WFS1 activates IP3R through NCS1 and stimulates Ca2+ release from the ER. WFS1 dysfunction impairs IP3R activity, followed by decreased Ca2+ release from the ER and Ca2+ uptake into the mitochondria. C WFS1 stabilizes HRD1, an E3 ubiquitin ligase, and degrades ATF6. WFS1 abnormality induces the hyperactivation of ATF6. D WFS1 plays a vital role in maintaining the pH within secretory granules. WFS1 dysfunction leads to impaired granule acidification and insulin exocytosis. E The C-terminal of WFS1 binds to vesicular cargo proteins. The N-terminal of WFS1 is recognized by the protein transport protein SEC24, a component of coat protein complex II (COPII). WFS1 dysfunction disrupts the generation of mature COPII vesicles and hinders intercellular trafficking from the endoplasmic reticulum (ER) to the Golgi complex. ATF6 activating transcription factor 6, HRD1 HMG-CoA reductase degradation 1 homolog, SERCA sarcoendoplasmic reticulum ATPase, SOCE store-operated Ca2+ entry, IP3R 1,4,5-trisphosphate receptor, NCS1 neuronal calcium sensor 1, GRP75 glucose-regulated protein 75, VDAC1 voltage-dependent anion channel 1, MCU mitochondrial Ca2+ uniporter. (Created with BioRender.com)

Full size image

WFS1 as a component of unfolded protein response

The link between WFS1 and ER stress was first described by Ueda et al. (2005). They demonstrated that the WFS1 expression level increased in response to ER stress. The gene expression levels of Wfs1 are considered ER stress markers (Lipson et al. 2006) due to the presence of a conserved sequence in the promoter region similar to the ER stress response element (Kakiuchi et al. 2006). ER stress-induced WFS1 upregulation requires the activation of inositol requiring 1 and PKR-like endoplasmic reticulum kinase, which are key regulators of the unfolded protein response (UPR) (Fonseca et al. 2005). Another UPR regulator, activating transcription factor (ATF) 6β, binds to the Wfs1 gene promoter and induces both gene and protein expression (Odisho et al. 2015).

WFS1 has additional roles in the regulation of ER stress and protein degradation. It stabilizes the E3 ubiquitin ligase HRD1 and regulates the degradation of ATF6α by facilitating ATF6 transport to the proteasome (Fonseca et al. 2010) (Fig. 2C). This means WFS1 acts as a UPR regulator; its deficiency in pancreatic β-cells induces pathogenic ER stress, leading to impaired cell cycle and accelerated cell apoptosis (Yamada et al. 2006). However, considering their close interaction, both ER stress and intracellular Ca2+ homeostasis may be involved in the pathogenesis of Wolfram syndrome. In fact, WFS1 deficiency induces ER stress, resulting in IP3R dysfunction and disturbed cytosolic Ca2+ homeostasis, which subsequently affects mitochondrial dynamics (Blackstone et al. 2016). This leads to inhibited mitochondrial fusion and trafficking, as well as augmented mitophagy, contributing to delayed neuronal development. ER stress is recognized for disrupting intracellular Ca2+ homeostasis and inducing cellular inflammation. The concept of “sterilized inflammation,” which refers to non-infectious inflammation triggered by pathogenic ER stress (Lerner et al. 2012; Oslowski et al. 2012), has recently emerged. In line with these observations, ER stress-induced sterilized inflammation caused by WFS1 dysfunction accelerates disease progression in Wolfram syndrome (Morikawa et al. 2022; Panfili et al. 2021).

WFS1 function in intracellular trafficking

WFS1 localization extends beyond the ER to include secretory granules (Hatanaka et al. 2011). In WFS1-deficient pancreatic β-cells, insulin secretory granule acidification becomes impaired, leading to impaired insulin exocytosis (Hatanaka et al. 2011) (Fig. 2D). The acidification of insulin granule is necessary for efficient proinsulin processing because endopeptidase PC1/3 and PC2 activate with a low pH optimum. While the precise interaction between WFS1 expressed on the insulin granule membrane and impaired insulin granule acidification has not been elucidated, it is speculated that WFS1 regulates the activities of V-type H+-ATPase and CLC-3Cl− channels expressed on the insulin granule membrane (Hatanaka et al. 2011). More recently, it was demonstrated that WFS1 serves as a vesicular cargo protein involved in intracellular trafficking (Wang et al. 2021). Specifically, the C-terminus of WFS1 directly binds to transported proteins within the ER lumen, while the N-terminus of WFS1 interacts with SEC24, a subunit of the coat protein complex II (COPII) located in the cytoplasm. This interaction allows for vesicular transport from the ER to the Golgi (Fig. 2E). This function of WFS1 has been shown to be necessary for the transportation of proteins such as proinsulin from the ER to the Golgi in pancreatic β-cells (Wang et al. 2021).

Experimental models of Wolfram syndrome

Experimental models are indispensable tools for conducting research to understand the pathogenesis of Wolfram syndrome and to develop therapeutic strategies. Over the years, several disease models have been established to study this condition. In the 2000s, Wfs1 knockout mice were generated as an early animal model. Subsequently, in the 2010s, disease models were extended to include rats, flies, and zebrafish. Recently, a Wfs1 pathogenic variant knock-in mouse model has also been reported. In this section, we summarize the phenotypic characteristics observed in these Wolfram syndrome animal models and also include information on induced pluripotent stem cell (iPSC)-derived cellular models (Table 1).

Table 1 Animal models of Wolfram syndrome
Full size table

Animal models of Wolfram syndrome

The first animal model, Wolfram syndrome, was established by Ishihara et al. (2004). The whole-body Wfs1-deficient B6 background mice exhibited decreased pancreatic insulin content at 2 weeks old, impaired glucose homeostasis, and decreased insulin secretion around 17 weeks old. The diabetic phenotype in these mice resulted from ER stress-induced pancreatic β-cell death and was accelerated by crossing with agouti lethal yellow mice, a model that develops obesity, insulin resistance, and pancreatic β-cell hyperplasia (Akiyama et al. 2009). Similar to Wolfram syndrome patients, these mice also exhibited symptoms of central diabetes insipidus due to arginine vasopressin (AVP) deficiency (Kurimoto et al. 2021), impaired retinal functions (Bonnet Wersinger et al. 2014), and psychiatric symptoms observed in behavioral studies (Kato et al. 2008). In addition to the B6 strain, another whole-body Wfs1 knockout mouse model on the 129S6 background was developed (Kõks et al. 2009; Luuk et al. 2008). These 129S6 mice, in which LacZ replaces exon 8 of Wfs1, showed an earlier onset of diabetes compared to the B6 strain (Abreu et al. 2020). Interestingly, males were found to be at higher risk of developing diabetes than females (Noormets et al. 2011). The phenotype of this 129S6-background mouse model is well studied, and besides the diabetic phenotype, this mouse model recapitulates many of the symptoms observed in patients with Wolfram syndrome (Blackstone et al. 2016; Ivask et al. 2021; Luuk et al. 2009; Noormets et al. 2009, 2014; Richard et al. 2023; Visnapuu et al. 2013a, b; Waszczykowska et al. 2020). The only Wfs1-conditional knockout mouse model was reported by Riggs et al. These B6 background mice, with specifically knocked out Wfs1 in pancreatic β-cells, develop glucose intolerance and insulin deficiency by 12 weeks of age (Riggs et al. 2005). Most recently, a Wfs1 E864K knock-in mouse model was reported (Richard et al. 2023). These Wfs1 E864K knock-in mice develop an early onset of severe vestibular dysfunction besides hearing loss. Human patients with Wolfram syndrome carrying the WFS1 p.E864K variant develop low-frequency sensorineural hearing loss, optic atrophy, and impaired glucose tolerance. However, no vestibular neuropathy has been reported. These facts suggest that the role of WFS1 in the human vestibular system may not be as crucial as in rodents.

Besides the mouse model, the Wfs1-deficient rat is a well-established animal model for Wolfram syndrome (Plaas et al. 2017). This model involves whole-body Wfs1 knockout in Sprague–Dawley background rats, which show impaired glucose homeostasis starting at 3 months of age. At 15 months of age, these rats exhibit retinal gliosis, optic nerve atrophy, and reduced medullary volume. Although the onset age of each symptom is older than that in the mouse models, the larger body size of rats makes it advantageous for conducting therapeutic interventions and sample collection. Recently, the drosophila and zebrafish models with WFS1 homolog knockout have been reported as additional models for Wolfram syndrome. Optical transparency of the zebrafish embryo enables the observation of neural development in vivo. Drosophila has a short generation time and a small body size, making it possible to test many potential therapeutic molecules. Using these Wolfram syndrome zebrafish and Drosophila models, phenotypes caused by WFS1 deficiency, including neurodegeneration, visual dysfunction, and mitochondrial dysfunction, have been well studied (Cairns et al. 2021; Crouzier et al. 2022b; Kunnappallil and Hasan 2022; Sakakibara et al. 2018).

The pursuit of novel therapeutic approaches to combat Wolfram syndrome is actively underway. One promising strategy involves repurposing existing diabetic medications for the treatment of Wolfram syndrome, such as GLP1 receptor agonists (Panfili et al. 2023). Notably, exenatide and dulaglutide have been reported to improve glucose tolerance in Wolfram syndrome mouse models (Gorgogietas et al. 2023; Kondo et al. 2018; Sedman et al. 2016). Similarly, another GLP-1 receptor agonist, liraglutide, was found to prevent the development of glucose intolerance, reduce neuroinflammation, protect against retinal ganglion cell death, and delay the progression of hearing and vision loss in Wfs1 mutant rat models (Jagomäe et al. 2021; Seppa et al. 2019, 2021; Toots et al. 2018).

iPSC-derived models of Wolfram syndrome

The first successfully developed insulin-producing cells from patients with Wolfram syndrome were reported by Shang et al. (2014). In their study, the authors generated insulin-producing cells from skin fibroblast-derived iPSCs of Wolfram syndrome patients. These patient-derived insulin-producing cells showed lower insulin content and reduced tolerance to ER stress compared to controls. This cellular model provided crucial insights into ER stress-related diseases derived from human patients. Maxwell et al. improved the differential protocol for iPSC-derived β-cells (SC-β-cells) and used CRISPR/Cas9 techniques to explore the potential of gene therapy and personalized cell therapy for Wolfram syndrome (Maxwell et al. 2020). In their study, they successfully restored insulin secretion in patient-derived SC-β-cells through the correction of WFS1. Transplanting the gene-edited SC-β-cells into diabetic mice resulted in improved blood glucose profile. In addition, they found that SC-β-cells differentiated from Wolfram syndrome patients with mild WFS1 variants retained their insulin secretory capacity (Kitamura et al. 2022). This indicates that these SC-β-cell models exhibit phenotypes consistent with the symptoms observed in patients. Using these SC-β-cell models, a combination treatment of 4-phenylbutyric acid and tauroursodeoxycholic acid has been demonstrated to effectively address the diabetic phenotype in Wolfram syndrome patients with the WFS1 p.R558C variant (Kitamura et al. 2022). Besides pancreatic β-cells, neural cells have also been developed as a disease model from Wolfram syndrome patient iPSCs. Impaired neurite outgrowth was observed in neurons derived from Wolfram syndrome patients, and this morphological change was prevented by treatment with valproic acid (Pourtoy-Brasselet et al. 2021). Valproic acid has been reported to reduce ER stress and act as a histone deacetylase inhibitor for neurons. It is not clear how valproic acid promotes neurite outgrowth in the neuronal cell model of Wolfram syndrome. However, valproic acid may exert neuroprotective effects by reducing ER stress and through pleiotropic mechanisms. Zatyka et al. demonstrated that the interaction between WFS1 and VDAC1 is essential for mitochondrial function and dynamics in iPSC-derived neural cells (Zatyka et al. 2023). These models could prove useful for studying the pathogenesis of brainstem atrophy and neurodegeneration observed in Wolfram syndrome.

Future expectations

Numerous clinical and basic studies have significantly contributed to understanding the function of WFS1 and advancing toward a cure for Wolfram syndrome. In future, further research is expected to uncover the unknown pathogenesis, particularly regarding the development of optic nerve atrophy and neurological symptoms associated with WFS1 dysfunction. Furthermore, the development of therapeutic agents against Wolfram syndrome necessitates the use of patient-derived disease models. In addition to pancreatic β-cells and neural cells, the development of retinal ganglion, inner ear, and AVP-producing neural cells is important. These patient-derived cellular models can serve as invaluable tools for developing preclinical and personalized therapeutic reagents against visual impairment, hearing loss, and central diabetes insipidus in Wolfram syndrome. Research on Wolfram syndrome, including the development of various disease models, continues to serve as a prototype for studying ER stress-related diseases.