Models for IGHMBP2-associated diseases: an overview and a roadmap for the future

Models are practical tools with which to establish the basic aspects of a diseases. They allow systematic research into the significance of mutations, of cellular and molecular pathomechanisms, of therapeutic options and of functions of diseases associated proteins. Thus, disease models are an integral part of the study of enigmatic proteins such as immunoglobulin mu-binding protein 2 (IGHMBP2). IGHMBP2 has been well defined as a helicase, however there is little known about its role in cellular processes. Notably, it is unclear why changes in such an abundant protein lead to specific neuronal disorders including spinal muscular atrophy with respiratory distress type 1 (SMARD1) and Charcot-Marie-Tooth type 2S (CMT2S). SMARD1 is caused by a loss of motor neurons in the spinal cord that results in muscle atrophy and is accompanied by rapid respiratory failure. In contrast, CMT2S manifests as a severe neuropathy, but typically without critical breathing problems. Here, we present the clinical manifestation of IGHMBP2 mutations, function of protein and models that may be used for the study of IGHMBP2-associated disorders. We highlight the strengths and weaknesses of specific models and discuss the orthologs of IGHMBP2 that are found in different systems with regard to their similarity to human IGHMBP2.


Introduction
Human disease models are critical for better understanding diseases and their progression. They help to identify the role of mutations found in patients, to identify the pathological cellular processes that lead to a disease outcome, to find new targets for therapies, to explore new concepts and to test therapeutic options. Crucially they provide data that cannot be directly obtained from a patient or from tissue samples taken from patients. Disease models are vital for the successful translation of bench research to the clinic. For example, Spinraza (nusinersen), the first approved therapy for spinal muscular atrophy (SMA) was initially proven in a mouse disease model [1] . The initial discovery that a reduction, by genetic means, of the levels of the Dnm2 protein may be beneficial for X-linked centronuclear myopathies was * Corresponding author.
E-mail address: wrzepnikowska@imdik.pan.pl (W. Rzepnikowska). made in a mouse model [2] . This was followed by proof-ofconcept testing [3 , 4] and resulted in the initiation of clinical trials in centronuclear myopathies patients (NCT04033159 and NCT04743557). Thus, disease models are inherent to both research and the development of therapies.
Mutations in the IGHMBP2 (immunoglobulin mu-binding protein 2) gene lead to various phenotypes that are characterized by different clinical manifestations. Firstly there is a rare but devastating form of SMA, referred to as spinal muscular atrophy with respiratory distress type 1 (SMARD1). The second disorder is a severe disorder of the peripheral nervous system known as Charcot-Marie-Tooth type 2S (CMT2S). Finally, there is a relatively wide spectrum of rare atypical phenotypes associated with IGHMBP2 mutations (for details see Table 1 ).
The onset of SMARD1 typically occurs during early infancy. Selective loss of alpha motor neurons in the ventral horns of the spinal cord and subsequent muscle denervation are the cause of muscle atrophy [5][6][7]  CMT diagnosed at 5 months, at 27 months signs of scoliosis, X-ray: acute pneumonia without diaphragmatic paralysis, tube feeding at 29 months, respiratory decline and urine retention at the age of 36 months [108] component of SMARD1 is respiratory failure due to paralysis of the diaphragm, this typically manifests within the first few months of life [6][7][8] and without ventilation support can lead to premature death [5 , 8] . In contrast to SMARD1, CMT2Saffected patients are wheelchair-bound but do not experience the acute respiratory difficulties that often result in sudden death [9 , 10] . The cellular role of IGHMBP2 is puzzling. Several studies suggest the involvement of this abundant helicase in transcription, translation and in RNA metabolism (see Section 3) but, in fact, none of these roles have been explored comprehensively. Scant data relating to processes involving IGHMBP2, a poor characterization of the molecular defects that accompany depletion of this protein and the abundance of the protein limit our understanding of the pathomechanism of IGHMBP2 -associated diseases. This, in turn, translates into complications with identifying potential therapeutic targets and developing specific therapies. Further studies of the fundamental roles of IGHMBP2 are needed and they require appropriate research systems. Unfortunately, many well-known and well characterized laboratory models remain underused for the investigation of IGHMBP2 -associated diseases. Here, in addition to clinical picture associated with IGHMBP2 mutations and function of IGHMBP2 protein, we present the different platforms available for the study of IGHMBP2 and associated diseases; from unicellular organisms, such as the yeast ( Saccharomyces cerevisiae ), to complex animal models, like the mouse ( Mus musculus ). We discuss the strengths and weaknesses of each of the systems available.

Wide spectrum of phenotypes associated with IGHMBP2 mutations
In 1974 Mellins and colleagues reported two atypical cases of SMA beginning with respiratory failure, resulting from bilateral diaphragmatic palsy, and evolving to global weakness and areflexia [11] . In 1999, Grohmann and colleagues reported nine patients from different ethnic backgrounds (Lebanese, German and Italian) who presented spinal muscular atrophy with respiratory distress (SMARD) that was not linked to the classical SMA locus 5q11.2 -q13.3 ( SMN1 gene), but instead mapped to chromosome 11q13-q21 [5] . Two years later, in six SMARD type1 (SMARD1) families, the causative mutations (three recessive missense, two nonsense, one frameshift deletion and one splice donorsite mutation) were identified in the IGHMBP2 gene [12] . In 2014, the phenotype of diseases resulting from mutations in IGHMBP2 was widened to include the hereditary motor and sensory neuropathy with recessive inheritance, known as CMT2S. Contrary to patients suffering from SMARD1, CMT2S-affected patients do not present with acute respiratory distress and the phenotype is limited to a slowly progressive weakness accompanied by wasting of the distal muscles of the upper and lower limbs. At least some CMT2S-affected patients also exhibit signs of autonomic neuropathies (bladder and gastrointestinal dysfunction, achalasia) [9 , 13] . Similarly, SMARD1 patients also exhibit autonomic dysfunction including: excessive sweating; tachycardia; constipation; retention of urine and attacks of anxiety [14] . Nomura and colleagues even described a catastrophic autonomic crisis in a girl affected with SMARD1, most likely aggravated by psychological stress [15] . In turn, Tomaselli and coworkers reported enteral autonomic dysfunction in the other patient, who developed gastrointestinal symptoms during the first decade of life (abdominal bloating, constipation). At age 27, a percutaneous endoscopic gastrostomy (PEG) was performed. At age 30, the patient started on total parenteral nutrition [16] .
The distinction between SMARD and CMT2S phenotypes appears not be as sharp as initially thought. Kulshrestha and colleagues reported on a CMT2S-affected boy who developed diaphragmatic weakness at 9 years old, suggesting the existence of an intermediate phenotype between CMT2S and SMARD1 [17] . In general, the spectrum of phenotypes associated with IGHMBP2 mutations is much broader than the early studies suggested ( Table 1 ). In clinical terms, the most variable component of the phenotype concerns disruption of the autonomic nervous system. In some studies the severity of the SMARD1/CMT2S phenotype was correlated with levels of IGHMBP2 [9] , however not all studies support this finding. For example, Pedurupillay and coworkers described identical compound heterozygous mutations, segregated within the same family, as SMARD1 or CMT2S in siblings, suggesting the presence of genetic modifiers [18] . In addition, Guenther and colleagues, reported a decrease in IGHMBP2 levels in SMARD1 affected patients but did not find a significant difference between residual protein levels in a patient with infantile onset (patient no. 2) and one with juvenile onset (Px1) [19] . Thus, despite a general association between IGHMBP2 levels and the severity of the symptoms observed, it appears that this does not always correlate with the onset of disease and phenotype. Without a doubt, further studies of IGHMBP2 -associated diseases are required; studies that utilize the various models of CMT2S/SMARD1 disease. These models may serve to improve our understanding of the pathogenic effects of IGHMBP2 mutations and also to screen potential therapies.

Structure and function of IGHMBP2
The IGHMBP2 gene contains 15 exons and encodes a protein of 993 amino acids (aas) [12 , 20] . It is ubiquitously expressed with moderate expression in fibroblasts and lymphoblastoid cell lines [9 , 21] . Following birth, the expression of IGHMBP2 increases in the cerebellar cortex, while decreasing slightly in other regions of the brain. In adults, the highest IGHMBP2 expression level was observed in the cerebellum [9] .
IGHMBP2, also called SMUBP2 (or S μBP2), cardiac transcription factor 1 (CATF1) and glial factor 1 (GF-1) is an ATP-dependent 5 → 3 helicase for both DNA and RNA. IGHMBP2 belongs to superfamily 1 (SF1) of the helicases, in particular to the Upf1-like family [22 , 23] . Generally, helicases have been grouped into six superfamilies (SF1-SF6) based on their sequence, structure and functionality. SF1 and SF2 are a large groups of non-ring forming RNA and DNA helicases [24] . The helicase core of all helicases of SFs 1 and 2 is formed by two RecA-like domains arranged in tandem and contains several characteristic sequence motifs [24 , 25] . Some motifs are present in all SF1 and SF2 helicases, additional motifs define the classification into families [24] . It seems that IGHMBP2 forms tetramers in vitro [19] and consists of a helicase domain, a single-stranded (ss) RNA and DNA binding R3H domain and an AN1-type zinc finger motif (zf-AN1; Fig. 1 A and 1 B) [26][27][28] . The helicase domain, similarly to all members of the UPF1-like family, possesses two RecA-like domains (1A and 2A) and two domains (1B and 1C) inserted into the first RecA domain 1A. The R3H domain specifically recognizes the phosphorylated 5 -ends of ssDNA or ssRNA [29] and enhances both RNA binding and the ATPase activity of the IGHMBP2 helicase domain [27] . The R3H domain alone is not sufficient for highaffinity RNA binding [21] . The helicase core of IGHMBP2 is not able to unwind dsDNA on its own, it requires the support of its C-terminal domains [28] . The majority of the mutations observed in SMARD1 and CMT2S are located within or near to the helicase domain and appear to affect the enzymatic activities of the protein. Thus, biochemical defects of IGHMBP2 mutants may be the underlying cause of SMARD1 [8 , 23 , 27] .
The role of IGHMBP2 in cells remains elusive; in particular it is not known why mutations of the gene encoding this abundant protein are mainly deleterious to neuronal cells. The IGHMBP2 cDNA was initially isolated as a cDNA that encoded a protein that bound to the single-stranded DNA containing sequence related to the immunoglobulin mu-chain switch (S μ) region [21 , 30] . Early reports showed that IGHMBP2 primarily localizes to the nucleus and suggested that it may be involved in transcription and pre-mRNA splicing [21 , 30-36 ]. However, subsequent approaches indicated that IGHMBP2 is predominantly cytoplasmic and, in mouse embryo-derived motor neurons, is principally localized to the perinuclear region and spreads to growth cones and axons [9 , 23 , 37 , 38] . There are several pieces of strong evidence that IGHMBP2 is involved, directly and/or indirectly, in translation ( Fig. 2 ). Co-localization of mouse Ighmbp2 with the eukaryotic translation initiation factor eIF4G2 and with ribosomal RNA (rRNA) was observed [23] . What is more, Ighmbp2 specifically associates with ribosomal subunits and 80S ribosomes [23] . Also, IGHMBP2 interacts with ribosomes [23] and with proteins involved in the processing of pre-rRNA (activator of basal transcription 1; ABT1), and in ribosome biogenesis (Pontin and Reptin) [39] . IGHMBP2 may also influence tRNA metabolism. It has been shown that IGHMBP2 interacts with tRNAs, particularly tRNATyr and with the 220 kDa subunit of transcription factor IIIC (TFIIIC220), which is responsible for tRNA transcription Fig. 2. Contribution of IGHMBP2 to the translation process. IGHMBP2 protein may be involved in translation in several ways. Firstly, it may play a role in ribosome biogenesis and/or function. IGHMBP2 binds to Pontin and Reptin, proteins required for U3 snoRNP (small nucleolar ribonucleoprotein) biogenesis in human cells, and to ABT1 (activator of basal transcription 1), which is involved in early pre-rRNA processing and may associate with U3 snoRNP and the 5 ETS (external transcribed spacer) of pre-ribosomal RNA (pre-rRNA). Additionally, IGHMBP2 associates with ribosomes and ribosomal subunits. Secondly, IGHMBP2 seems to participate in regulation of tRNA synthesis and/or maturation. It interacts with the 220 kDa subunit of TFIIIC (transcription factor IIIC), an essential factor for tRNA transcription and with the tRNAs themselves. [39] . Interestingly, a genetic modification on chromosome 13 that specifically halted motor neuron degeneration and rescued the neuropathological features and the clinical outcomes of a mouse model for SMARD1 ( nmd mouse; see below) [20] occurred in a region that contains five genes encoding five tRNATyrs and a gene encoding ABT1. The corresponding region in the human genome also contains four tRNATyr genes and a gene encoding the activator of basal transcription 1 (ABT1) [40] . Finally, Ighmbp2 deficiency in cultured nmd mouse primary motor neurons leads to a translational delay of β-actin mRNA and a selective reduction of protein levels in axonal growth cones. This is despite the fact that Ighmbp2 does not bind directly to β-actin mRNA and that no abnormalities were detected in the amount of the corresponding mRNA or the total protein level [41] . These findings reinforce the idea that IGHMBP2 is important for translation processes in the cell ( Fig. 2 ).

Therapeutic approaches to IGHMBP2 -associated diseases
To date, there are no approved therapies for SMARD1 or CMT2S that can cure the diseases or even modify their progression. The clinical approaches for treatment of SMARD1 or CMT2S patients support vital functions and mitigate symptoms. The evaluation for respiratory insufficiency is one of the most critical issue allowing to decide whether to intubate and mechanically ventilate the patient. Pulmonary care includes also the use of airway clearance strategies, such us cough augmentation devices. Other vital aspects of treatment include continuous physical and occupational therapy to delay the progression of muscle weakness and onset of orthopedic problems. Further disease management focus on the specific symptoms present in the affected individual. Patients with urinary retention require catheterization, recurrent airway infections are treated with antibiotics and difficulty in feeding due to muscle weakness and gastrointestinal dysfunction needs nutrition therapy [14 , 42] . Management of affected children falls on families and requires huge efforts [8] . For patients and their relatives genetic counseling is recommended. As the survival rate for SMARD1 is so low, practically all therapies evaluated for IGHMBP2 -associated diseases so far have been concentrated on this disorder. However it should be noted that such therapies may also be relevant to CMT2S patients, as both diseases originate from mutations in the same gene. Treatment and disease management strategies are not the main focus of this review and have been described in detail elsewhere [42][43][44] . Below we will only briefly outline the therapeutic strategies tested for SMARD1 treatment.
One possible candidate for future therapy is neurotrophic factor, insulin-like factor 1 (IGF1). IGF1 is a hormone characterized by multiple functions including muscle and neuron survival or differentiation and axonal growth during development. It has been demonstrated that IGF1 is reduced in nmd mice, suggesting that it may cause some of the neuropathological symptoms of SMARD1. IGF1 conjugated to polyethylene glycol (PEG-IGF1) was found to restore IGF1 serum levels and improve some key features of the disease, but without significantly increasing motor neuron survival [45] . A monoclonal antibody, Mab2256, an agonist of tyrosine kinase receptor C (TrkC), which is involved in the regulation of neuronal plasticity and synaptic strength was tested in nmd mice. The researchers reported an initial beneficial effect on muscular strength and function, which was confirmed by electrophysiological studies, along with a slowing of disease progression. These effects were however only transient and did not prolong survival [46] .
Because SMARD1 disease is caused by a single gene mutation, it may be an appropriate candidate for gene therapy. Indeed, IGHMBP2 gene transfer using adeno-associated virus serotype 9 (AAV9) as a vector results in a highly efficacious rescue of both survival outcome and pathological phenotypes in the nmd mouse [47][48][49][50] . The positive effects of stem cell transplantation in the nmd mouse model have been demonstrated. A spinal cord neural stem cell population, isolated on the basis of aldehyde dehydrogenase (ALDH) activity (ALDH hi SSC lo ), transplanted in nmd animals resulted in delayed disease progression, a promotion of motor neuron and ventral root axon survival and increased lifespan [51] . In addition, human-induced pluripotent stem cells (iPSCs)derived neuronal stem cells were correctly localized to the anterior horn of the mouse lumbar spinal cord. This treatment improved both the neuromuscular function and the lifespan of the mice [52 , 53] .
Despite promising pre-clinical studies, none of the proposed therapies have entered into the clinical trial phase. Today, supportive therapy is still the main approach for treatment of SMARD1. However current advances in the treatment of inherited disorders, such SMA, allow us to be optimistic about SMARD1 cures in the future.

Models for the study of human IGHMBP2 -associated disorders
A patient's tissue is not usually available for a direct study of a given disorder. For neurological diseases, the ability to examine neurons, which are principally affected, is especially limited. Thus, in order to advance our understanding of the pathology and physiology of diseases and for pre-clinical studies for potential therapies, the modeling of human diseases is essential. In this section we address the best known research models and discuss their potential for studying the molecular and physiological mechanisms of IGHMBP1 -associated diseases. A summary of their strengths and weaknesses is presented in Table 2 .

Baker's yeast, Saccharomyces cerevisiae -when low cost and speed are important
The budding yeast Saccharomyces cerevisiae is an important model organism, as many molecular mechanisms, such as cell cycle control, signal transduction, DNA replication, transcription, translation and DNA repair are similar to those of higher eukaryotes. Yeast has advantages over other more complex models in terms of its low cost, short generation time, simple cultivation and genetic tractability. Yeast is widely used to study human diseases, including neurodegenerative diseases, such as Alzheimer's and Parkinson's diseases, Chorea-acantocytosis and CMT [54][55][56][57] and has also been used as the basis for highthroughput screens to find genetic or chemical factors that reverse the unfavorable phenotypes of mutations found in patients [58][59][60][61] .
The S. cerevisiae orthologue of IGHMBP2, Hcs1 (dna HeliCaSe), is an ATP-dependent 5 → 3 DNA helicase [62,63] belonging to the helicase SF1 Upf1-like family [24 , 64] . It is truncated (683 aas) and lacks the C-terminal domains when compared to the human protein ( Fig. 1 B), however it is well conserved. The full-length proteins share 23% of identity and 38% similarity. When Hcs1 is compared with only the N-terminal region of IGHMBP2 (aas 1-647) this increases to 35% of identity and 57% of similarity (EMBOSS Needle [65] ) Hcs1 protein was first identified as a helicase associated with a high molecular weight, multiprotein complex of DNA polymerase α [66] , however, like IGHMBP2, it is predominantly located in the cytoplasm (Yeast RGB database; https:// shmoo.weizmann.ac.il/ elevy/ YeastRGB/ HTML/YeastRGB.html). The hexameric form of Hcs1 appears to be the active form of the protein [62] . Similarly to human IGHMBP2 [22 , 23] , it hydrolyzes ATP in a strictly DNA-dependent manner [63 , 66] . In vitro , Hcs1 was found to exhibit a preference for pyrimidine-rich DNA sequences and for polynucleotides; oligonucleotides shorter than 20 bp did not support Hcs1 ATP hydrolysis [62] .
Given that current studies point to the involvement of human IGHMBP2 in basic processes (transcription, translation and/or rRNA and tRNA metabolism), the sequence similarity between the human and yeast protein and the fact that the most human disease-associated mutations are located in the helicase domain, it seems quite rational to use S. cerevisiae to study IGHMBP2 -associated diseases. This simple model can be exploited to better understand the significance of rare sequence variants found in patients, to shed light on the molecular mechanisms underlying the pathogenicity of mutations and to search for therapies and therapeutic targets based on screens of both small molecules and genetic libraries. Studies using yeast as the model system cannot be used to probe and explain the tissue-specific effects of IGHMBP2 mutations. But yeast can be used to identify abnormalities in conserved processes that lead to disease phenotypes.

Worm, Caenorhabditis elegans -a simple but problematic model
Caenorhabditis elegans represents a free-living, nonparasitic nematode, with a short developmental cycle (about 3 days) and an overall lifespan of 2-3 weeks. C. elegans has many of the organ systems present in more complex organisms and exhibits complex behavior. Together with its easy cultivation, small size, well-known biology and completely sequenced genome C. elegans is one of the most versatile and powerful model organisms. C. elegans has emerged as an excellent experimental platform with which to study the molecular and cellular aspects of human diseases in vivo including Alzheimer's disease, Parkinson's disease and polyglutamine-expansion disorders [67][68][69][70][71] .
The gene encoding the C. elegans ortholog of IGHMBP2, ERI-7 ( E nhanced R NAi; RNA interference) was first identified as a negative regulator of exogenous RNAi in a screen for mutants displaying an enhanced response to exogenous double-stranded RNA (dsRNA) [72] . Immediately, a close relationship between eri-7 and eri-6 (also identified in the same screen) was discovered. In fact, divergently transcribed eri-6 and eri-7 RNAs are assembled into one mRNA in C. elegans strain N2, resulting in the translation of a single ERI-6/7 protein of 925 aas. In addition, in the nematode species C. briggsae and in four of 27 other natural isolates of C. elegans, eri-6 and eri-7 constitute a single, contiguous gene [72] . The ERI-6/7 protein is a SF1 helicase that negatively regulates exogenous RNAi and functions as part of an endogenous RNAi pathway [72 , 73] . The ERI-6/7 protein is produced in the ASK amphid neuron and in the somatic gonad and, similarly to the human protein, is predominantly cytoplasmatic [72] . It shares 16% identity and 27% similarity with its human ortholog (EMBOSS Needle [65] ). In the most similar region (IGHMBP2 aas 196-611 and ERI-6/7 aas 310-740) identity and similarity are 28% and 43% respectively (EMBOSS Matcher [65] ). ERI-6/7 participates in the generation and/or stability of two classes of siRNAs [74] . Finally, ERI-6/7 is required to silence retrotransposons and integrated viral genes [75] .
C. elegans , like yeast, may be used to dissect toxicity mechanisms, to determine in vivo (also in high-throughput screens) factors that ameliorate or cure pathological conditions caused by changes in IGHMBP2 and for rapid and inexpensive drug evaluation . Furthermore, the effect of a single mutation may be established in a particular tissue, system or organ. Importantly though, the functioning of ERI-6/7 as a IGHMBP2 homolog is in doubt. There is no evidence that IGHMBP2 also serves in RNAi processing and an Ighmbp2 deficiency only slightly alters the level of mRNAs and newly synthesized proteins in mouse primary cultured motor neurons [41] . This rather excludes IGHMBP2 from having a role in mRNA decay. Thus, despite the sequence similarity, it seems likely that ERI-6/7 and IGHMBP2 have differentiated towards other functions and participate in different cellular processes.

Zebrafish Danio rerio -a low cost, high throughput option
The zebrafish, Danio rerio, possesses a number of advantages: it can be grown in a cost-efficient manner, has relatively short generation times and its embryos are malleable to genetic manipulation. Helpfully, D. rerio embryos are optically transparent, which permits real-time imaging and allows for the visualization of individual genes (fluorescently labelled or dyed) throughout the developmental process using non-invasive imaging techniques. As an additional bonus, and uniquely among vertebrate models, it is amenable to high throughput drug screening [76][77][78] . It is therefore should not be surprising that zebrafish has become a laboratory tool used to improve our understanding of the molecular bases of human diseases [79][80][81] .
The zebrafish Ighmbp2 is quite similar to its human ortholog. It possess the same domain structure (helicase domain, R3H and zf-AN1) ( Fig. 1 B) and is 56% identical and 71% similar (EMBOSS Needle [65] ). We have not been able to find any data relating to research into IGHMBP2associated diseases in D. rerio . However a zebrafish model was successfully used to study SMA (for review see [82] ) and to establish the pathogenicity of mutations in other genes that result in the SMARD phenotype, namely LAS1L, a component of the ribosome biogenesis pathway [83] . Thus, D. rerio has been successfully implemented in the study of SMARD1 phenotypes, causes and potential therapies.

Mouse Mus musculus -an invaluable research model
There is no doubt that mouse models of human disease are a powerful system in which to study molecular and cellular pathophysiology. The Ighmbp2 region is conserved in mice; it is located on chromosome 19 instead of chromosome 11 [84] . Mouse Ighmbp2 is very similar to the human protein; it is identical in length (993 aas), exhibits the some domain architecture ( Fig. 1 B) and the two sequences are 77% identical and 85% similar (EMBOSS Needle [65] ). Mouse Ighmbp2 is also abundantly expressed, with the highest levels found in the brain, spinal cord and muscle (at embryonic day 15). In the spinal cord, Ighmbp2 is found at a high-level during embryonic and early postnatal development. Between postnatal days 10 and 21, levels of Ighmbp2 decrease greatly [37] .
The mouse model for SMARD1, called nmd for neuromuscular degeneration ( nmd mouse, particularly B6.BKS Ighmbp nmd-2J /J ) is commonly used to study this devastating disease. This mouse strain possesses a spontaneous homozygous mutation (A-G) in intron 4 of the Ighmbp2 gene, which leads to abnormal splicing in almost 80% of transcripts; correspondingly only about 20% of transcripts are full-length [20] . Thus, levels of the full-length Ighmbp2 are significantly reduced in these mice [19 , 37] . Between postnatal days 14-21, nmd mice develop a phenotype that is very similar to that of SMARD1. Progressive muscle weakness and paralysis resulting from motor neuron loss begins in the hind limbs and then progresses to the forelimbs. The mutant mice usually survive only a few months [37 , 84] . Severe motor neuron degeneration is observed before the first clinical symptoms become apparent [37] . That is a first indication that non cell-autonomous disease mechanisms might participate in motoneuron loss (see also [38] ). The loss of cell bodies in the lumbar spinal cord is followed by axonal degeneration in the corresponding nerves and loss of axon terminals at motor endplates, however the neurotransmission at the endplate is primarily not affected [37 , 85] . Specific muscles are characterized by their different susceptibilities to denervation, with the muscles of the distal appendicular region exhibiting the highest degree of vulnerability [50] . In contrast to motor neurons derived from a mouse model of SMA, cultured embryonic nmd mutant motor neurons do not show any defects in terms of survival, axonal growth or growth cone size [37 , 41] . These findings suggest that the pathomechanism of SMARD1 is different from that of SMA [37 , 41] .
In 2019 Shababi and coworkers created a new SMARD1 mouse model that contains the same intron 4 mutation in Ighmbp2 as found in nmd mice but in an alternative congenic background (FVB, instead of C57BL/6). Similar to nmd mice, the FVBnmd mice develop symptoms by the second week of age but demonstrate a more severe phenotype than the original nmd mice with respect to survival, weight and motor function. This new model was primarily established to monitor therapeutic efficacy without interference from the common congenital conditions found in the original SMARD1 mouse model. These congenital conditions include hydrocephalus and the presence of a modifier variant (allele) on chromosome 13 [86] .
The nmd mouse model is essential for studying disease mechanisms and also for the development of therapies or a possible cure. All therapeutic options for SMARD1 were initially established using this model (see above). However, the nmd mouse model does not completely mimic the disease symptoms observed in patients. First of all, the onset of respiratory distress, which is an early and very prominent feature of human SMARD1, only occurs during the later stages of the disease. Surprisingly, it seems that breathing problems in the nmd mice do not result from nerve degeneration, but instead arise from a defect in the diaphragm itself; even at late stages of the disease no significant reduction in the axon numbers of phrenic nerves was observed [37 , 50 , 87] . Furthermore, myopathic changes were observed in nmd mice: in the quadriceps muscle, both with and without expression of Ighmbp2 in the central nervous system [37 , 87] , in the Levator Auris Longus (LAL) muscle [85] , in the diaphragm [37 , 45] and partly in the gastrocnemius muscle [45] . The pathogenesis of nmd mice also involves the death of cardiomyocytes, leading to progressive cardiomyopathy and heart failure [87] . This is not typically observed in SMARD1 patients [8] . It appears that cardiac problems underlie the early lethality of nmd mice. Selective expression of Ighmbp2 in the central nervous system prevents motor neuron degeneration and restores normal axonal morphology but does not increase the life-span of affected mice; all mice eventually develop dilated cardiomyopathy and congestive heart failure [87] . On the other hand, muscle-specific expression of Ighmbp2 prevents cardiomyocyte loss and restores normal cardiac morphology and function and markedly extended life-span, albeit with severely restricted mobility [88] . Thus, despite the mouse model offering great potential and being a powerful tool for further study, some of the results have to be analyzed with caution. Finally, it has not been excluded that some of the symptoms observed in nmd mice, are also presented in SMARD1 patients, but they are not well-established. For example, some SMARD1 patients develop arrhythmia [6 , 8 , 14 , 89] . This finding implies that patients may also need a more detailed examination and that the symptoms of SMARD1 could require reevaluation.

Cellular model -induced pluripotent stem cells (iPSCs)
Induced pluripotent stem cells (iPSCs) exhibit a molecular profile and differentiation potential similar to that of embryonic stem cells (ESCs), yet they can be generated from somatic cells. Among other cellular models, iPSCs have a number of advantages: they have human origins; can differentiate into almost any cell type, including neuronal cells; are easily available, as may be derived from skin fibroblasts or blood cells; and compared to ESCs, there are fewer ethical concerns about their use. Additionally, iPSCs can be generated from individual patients and thus represent a personalized model including all genetic and epigenetic factors influencing the disease progression. This allows for the generation of different disease-relevant cell types and for studying interactions among them [90 , 91] . SMARD1 iPSC derived motor neurons were obtained by Simone and coworkers. SMARD1 iPSC derived motor neurons do not present developmental defects, but in longterm cultures a significantly reduced cell number and axon length were observed compared to those from wildtype iPSCs. Additionally, it was shown that human neural stem cells from iPSCs protected and increased the length of neurites of SMARD1 motor neurons when they are cocultured, confirming the beneficial effect of cellular therapy observed in the mouse model [52] . Thus, iPSCs are an excellent model in which to study pathology and to develop potential therapeutic options.

Summary
Reviewing the available data, serves to make us aware of how many knowledge gaps we still have to fill in terms of the IGHMBP2 gene and the function of protein that it encodes. There are many clues that indicate possible cellular roles and mechanisms for IGHMBP2 but none of these have yet been thoroughly investigated. Instead, they serve as hints and pointers that direct future research efforts. The use of different disease models may help to fill some of these knowledge gaps. Because each model has its own advantages and drawbacks (see Table 1 ), in order to obtain a complete picture of the pathology a combination of data from different models is needed. Detailed studies of IGHMBP2 homologs in most popular model organisms, yeast, worm and fish have yet to be carried out. An exception here is the mouse model, which is now quite well established and has contributed greatly to the understanding of the pathomechanism and also to the development of the first therapies. We hope, that the introduction of other models for IGHMBP2 -associated diseases will help to improve basic knowledge of the diseases and also broaden the therapeutic landscape.

Funding
The research related to this review are supported by National Science Centre Poland Grant No 2020/04/X/NZ2/00377

Declarations of Competing Interest
none