Recent advances in nemaline myopathy

The nemaline myopathies constitute a large proportion of the congenital or structural myopathies. Common to all patients is muscle weakness and the presence in the muscle biopsy of nemaline rods. The causative genes are at least twelve, encoding structural or regulatory proteins of the thin filament, and the clinical picture as well as the histological appearance on muscle biopsy vary widely. Here, we suggest a renewed clinical classification to replace the original one, summarise what is known about the pathogenesis from mutations in each causative gene to the forms of nemaline myopathy described to date, and provide perspectives on pathogenetic mechanisms possibly open to therapeutic modalities.


Introduction
Nemaline myopathy (NM; [1 , 2] ) is one of the most common of the congenital myopathies, ranging in severity from severe forms, which may cause early lethality to milder muscle disorders with onset in childhood, sometimes presenting as late as in adulthood. It is a genetically and clinically heterogeneous group of disorders, characterised by usually non-progressive or slowly progressive generalised muscle weakness. Pathologically NM is a non-dystrophic myopathy characterised by the presence of cytoplasmic inclusions called nemaline bodies or rods.
Mutations in at least 12 different genes cause NM. What these genes have in common is that they all encode proteins associated with the structure or regulation of the thin filament of the skeletal muscle sarcomere. Most commonly causative are mutations in NEB [3] , encoding nebulin, closely followed by mutations in ACTA1 (skeletal muscle alpha actin, [4] ). Less look for similarities and differences between the forms of NMs caused by different genes, here, we are viewing the known NM genes divided into two groups, the genes that encode integral structural proteins of the thin filament, and the genes encoding regulatory proteins.
Like many other neuromuscular disorders, NM has moved from being regarded as one well-defined disorder, caused by mutations in one gene, to becoming split into several entities because of novel clinical or histological features being described, or due to causative mutations being identified in previously unknown genes. Lately, again like many others, the NMs have been re-lumped, now allowing for the inclusion within the term "NM and related disorders" of a wide range of clinical presentations and intriguing mixes of histopathological findings ( Fig. 1 A). The causative genes currently number at least twelve, in some cases with specific clinical or histological hallmarks providing clues to the "culprit" gene. The original definition of NM as a disorder characterised by muscle weakness and the presence of nemaline bodies (rods) in the muscle fibres, in the absence of other known disorders sometimes associated with rods, still applies, at least roughly [15] . It has, however, become evident that muscle weakness caused by mutations in NM-causing genes may not always be accompanied by nemaline bodies on biopsy, alternatively that the muscle biopsy may show other distinctive features in addition to nemaline bodies, such as cores or caps. Furthermore, while a majority of patients show a predominance of small, slow muscle fibres, this is not consistent [16] .
The clinical pictures of NMs constitute a broad spectrum. Severity varies even between individuals with the same mutations in the same gene. Onset of severe weakness in utero may lead to the foetal akinesia sequence, with arthrogryposis, but presentation at birth with generalised muscle weakness and no contractures is more common. In the typical form, muscle weakness is usually generalised, with weakness of the neck flexors, the face and the proximal muscles of the limbs, often with a later, additional distal involvement.
Milder forms may present later in childhood with delayed motor milestones or other signs of muscle weakness [17] . Patients presenting in adulthood with muscle weakness or respiratory failure may, on examination, be found to have myopathic facies, high-arched palate, small tongue, small mouth, thin ribs and other signs of early muscle weakness, making it apparent that the disorder was, in fact, congenital. If not, a severe autoimmune disease, sporadic late-onset nemaline myopathy (SLONM [18] ), and other non-genetic causes should be sought for.
Current numbers extracted from The International Nemaline Myopathy Consortium Database in Helsinki, now with more than 500 patients included, indicate that what is called the typical congenital form, characterised by congenital onset of muscle weakness, spontaneous respiration at birth and motor milestones delayed but reached, holds up to its name by still being the most common form of NM. However, the incidence of the neonatally severe form characterised at birth by inability to move or breathe, or by the presence of contractures or fractures at birth, roughly equals that of the typical form.
In most patients, weakness is accompanied by muscle hypotonia, but rare patients present with muscle hypertonia, stiffness or rigidity ( [19], unpublished observations), caused at least partly by slow relaxation [20] . Cardiac involvement is rare (unless untreated respiratory insufficiency leads to cor pulmonale), but has been diagnosed in a few patients with mutations in ACTA1 or MYPN, or with a contiguous deletion of TNNT1 and TNNI3 [7 , 21-23] .
Ophthalmoplegia is uncommon also, but has been observed in patients with mutation of KLHL40 or LMOD3, or in exceptional cases of NEB [10 , 14 , 24] .
Signs of dystrophy are usually absent and serum concentrations of creatine kinase are mostly normal or only slightly elevated.
In all forms of NM and related disorders, weakness of the respiratory muscles should be regarded as an important risk for the patient, and regular monitoring, with early treatment where needed, should be instituted according to established international guidelines [25 , 26] . Hypoventilation may have an insidious onset and the cause of death is most commonly respiratory. In the absence of expert monitoring of respiratory function, and adequate ventilator support where needed, respiratory insufficiency may ensue, either slowly, or even suddenly and without any preceding symptoms [17] . If untreated, respiratory insufficiency leads to cor pulmonale.

Original and renewed classification
At an ENMC workshop in 1999, a clinical classification was defined to facilitate gene discovery [27] . During the time that has elapsed since then, the data accumulated show that genotype-phenotype correlations are often weak, if at all present. Thus, we have proposed a revised, simplified classification, based on current knowledge of the variation between patients identified as having NM ( Table 1 ; previous version in [16] ), with the purpose of making the classification more user-friendly and helpful for at least a rough estimate of the prognosis. Amongst the categories in the original classification, the term "the intermediate form" was coined because of its more severe course than that of the typical (mainstream) form of NM. The intermediate form was found to be an unpractical category for early clinical classification, as the more severe course only was possible to discern in late childhood. Moreover, we now know that the most commonly causative genes, NEB and ACTA1, may both cause this form, and no specific "intermediate" genes have been identified. Another category in the original classification that has turned out to be largely redundant is the adult-onset form. To our knowledge, there are few, if any, cases where onset in adulthood and a genetic cause have been clearly proven. The acquired, rapidly progressive adult-onset form, sporadic lateonset NM, is often associated with monoclonal gammopathy or HIV, and early treatment is important (see [18] ). showing variable distribution and number of rods including peripheral clusters, scattered rods throughout some fibres and rods in lines; b) Small type 1 fibres and core-like areas in a 17-year-old patient with a variant in ACTA1 stained for NADH-TR; c) Uniform fibre typing and cores in a 6-year-old patient with heterozygous variants in NEB stained for NADH-TR; d) Pronounced increase in connective tissue and adipose tissue in a 6-year-old patient with heterozygous variants in TNNT1 stained with haematoxylin and eosin; e) Very small fibres labelled with an antibody to foetal myosin (these are often seen in congenital myopathies) in a 6-year-old patient with heterozygous variants in NEB; f) Immunolabelling with an antibody to fast 2A and 2X myosin in the same patient showing uneven distribution of fibre types, small positive and negative fibres (with slow myosin) and hybrid fibres with intermediate labelling intensity; g) Labelling of actin in rods with phalloidin in a patient with a variant in ACTA1; h) Immunolabelling of α-actinin in rods in a patient with a variant in ACTA1 ; i) Immunolabelling of myotilin in rods in a patient with a variant in ACTA1; j) Electron micrograph of rods sectioned longitudinally and transversely and with myofibrils attached in a patient with a variant in ACTA1; electron micrograph of actin accumulation and very small rods in a patient with a homozygous variant in CFL2 ( actin accumulation can also be a feature of ACTA1); l) Rectangular shaped rods with a fringe of myofibrils in a patient with variants in KLHL40. These are also a feature of patients with variants in LMOD3 . Image courtesy of Prof Caroline Sewry, electron micrographs courtesy of Cathy Timson.
Since 1999, descriptions have appeared of clinical forms not fitting into the original classification. One such novel form, caused by mutations in the gene TNNT1 , was first described in the Amish. It is characterised by early tremor, clonus, muscle weakness and atrophy with muscle stiffness, axial and thoracic rigidity, and severe, progressive contractures [8 , 28] . The biopsy mostly shows small fibres of both fibre types, larger type II fibres and early endomysial fibrosis. The EMG pattern may be "myopathic" [8] or sometimes "neurogenic" [21] . Subsequently, several patients of non-Amish origin with similar clinical and histological features have also been described [21 , 29-32] , rendering the term "Amish NM" outdated. Thus, we suggest replacing it with a term based on the genetic cause, TNNT1 myopathy.
Another form not fitting into the original classification of NM is the one with an unusual slowness of movements due to slow relaxation kinetics, and with cores as well as nemaline bodies, caused by mutation of KBTBD13 [12 , 33-35] .
In addition to these two novel forms with clinical features distinctly rare in NM, the literature published after the original classification holds descriptions of patients with unusual distributions of muscle weakness, such as scapuloperoneal weakness or distal weakness only. These might be seen to warrant classification as separate entities. However, in view of the wide individual variation in the distribution of weakness, and occurrence of contractures early or late in the course of NM, it would be difficult to delineate such novel entities in a way that would clearly distinguish them from the forms previously described.
However, because, in clinical practice, distal presentations render the patient falling into the diagnostic category of distal myopathies, it is important to point out that the genes causing NM have also been found to cause distal myopathy with nemaline bodies as well as distal myopathy without nemaline bodies [36][37][38] . Although these entities make up one end of the spectrum ranging from the typical form with a distal involvement to forms remaining at least principally distal, it may be practical to reserve a category of distal myopathies within the renewed classification, so that these genes are taken into account in the molecular genetic differential diagnosis. The same argument might apply to patients presenting distinctly with distal arthrogryposis only [9 , 39] .
Thus, we propose a re-classification of NM ( Table 1 ). This somewhat simplified classification offers some guidance for prognostication. In the severe form, the prognosis is often grave. There have, however, been exceptions, with patients experiencing improvement over time [40] . The course of the typical form is often static or only slowly progressive, some patients even showing improvement, sometimes related to active physical training. Onset in childhood or at juvenile age often implies a mild course. The distal forms, likewise, although not always remaining entirely distal, usually follow a mild course.
In TNNT1 (Amish) NM, the course is usually relentlessly progressive, with death due to thoracic immobility and restrictive lung disease often following in early childhood [8 , 21 , 28-32] . However, a dominantly inherited TNNT1 mutation causing a different clinical picture has been reported, with similarities to the childhood-onset form caused by mutations in other genes [41] . Moreover, recently, a likewise milder disorder, with presentation in adulthood including unusual clinical findings and multi-minicores as well as nemaline bodies, was described in a patient with a homozygous frameshift mutation in exon 13 of TNNT1 [42] . Similarly, mutations in LMOD3 may sometimes cause milder disease [43 , 44] .
The dominantly inherited form with slowness caused by mutations in KBTBD13 usually follows a milder course [34 , 35] .

Uncertain clinical entities
A severe clinical picture has been described in the only NM patient hitherto identified with mutation of TNNT3 , who had contractures, hip dislocation (unusual in NM) and ventilator dependence until death at the age of 8 months [9] .
The presence of nemaline bodies in a muscle biopsy is not pathognomonic for NM. Patients with MYO18B variants have had, in addition to muscle weakness and nemaline bodies on biopsy, Klippel-Feil anomaly and dysmorphic features, and some short stature [45][46][47] , placing these patients outside the original definition of NM. The same can be said for patients with TPM2 mutations and multiple pterygium or Escobar syndrome [48] .
ADSSL -caused myopathy is reportedly the most common form of NM in Japan [49] , but the basic mechanism may be one of energy metabolism. It is characterised histologically by lipid droplets and myofibrillar disruption in addition to the nemaline bodies, and radiologically by fatty infiltration at the periphery of muscles of the lower limbs. One fourth of the patients have hypertrophic cardiomyopathy [49] .
One family with a male patient and his mother presenting in adulthood with upper limb distal myopathy and MRI findings as in myofibrillar myopathy, showing, for the first time, nemaline bodies and ring fibres on biopsy, were found to have a heterozygous mutation of filamin C ( FLNC ) [50] .

Nemaline myopathies are caused by mutations in genes encoding proteins associated with the structure or regulation of the thin filament
The twelve known genes causing NM encode proteins of the sarcomere ( Fig. 2 , Table 2 ). Muscle weakness in NM has been found to be associated with decreased thin filament length, reduced acto-myosin interaction, or altered calcium sensitivity. Although we do not yet have the complete picture of how any given mutation in one of the known NM genes leads to NM, the various existing animal models and functional studies are shedding light on the pathogenetic mechanisms, thus providing insights into the aberrant muscle function in NM, and highlighting potential novel targets for future therapies.

Integral structural proteins
Nebulin is an enormous protein spanning the entire length of the thin filament, and responsible for its integrity [52] . It anchors the thin filament into the intermyofibrillar network, and into the Z disc through its Cterminus. In a mouse model lacking the C-terminal SRC homology 3 (SH3) domain of nebulin ( Neb SH3 ), most of the sarcomeric function was preserved [53] . Despite having slightly altered force-frequency relationship in vitro , these mice had no visible pathological phenotype in vivo. However, deleting a larger fragment, including both C-terminal domains (SH3 and the serine rich region, SRR) in the Neb 163-165 mouse model, resulted in a moderate NM phenotype [54] , suggesting that both domains are needed for the normal function of the sarcomere. Being a filamentous protein, nebulin stiffens the thin filament and contributes to thin filament activation and myosin cross-bridge recruitment [55 , 56] . Furthermore, being moderately elastic, nebulin can provide considerable compressive force when extended, stabilising the thin filaments during contraction [57] .
More than 250 disease-causing variants have been published in NEB , making NEB mutations the most common cause of NM [51] . The vast majority of the variants are recessive, and the patients are often compound heterozygous. The pathogenic variants identified are scattered along the entire length of the gene, and an individual variant is seldom shared by more than a few families [24] . The most common types of mutations are splice-site (34%), frameshift (32%) and nonsense mutations (23%). Missense variants in NEB account for the majority (63%) of the variants in the coding region, splice sites and untranslated regions [24] . Nebulin seems to tolerate missense changes well, and the interpretation of their pathogenicity is extremely difficult, especially if the change is located outside the conserved actin-or tropomyosin-binding sites [24] . Two loss-of-function variants in constitutive exons 5 of exon 180 have not been identified in any patient, suggesting that complete loss of nebulin is not compatible with human life [24] . Neb -KO models have demonstrated that in the absence of nebulin, actin filaments are assembled, but are not stable in muscle contraction [58] , leading to early death of the mice.
Recently, the first dominantly inherited disease-causing variant was found in NEB , causing distal nemaline/cap myopathy in a three-generation family [59] . This large ( ∼100 kb) in-frame deletion encompassing NEB exons 14-89 results in the expression of a substantially smaller nebulin, expected to act in a dominant-negative manner. Subsequently, an even larger dominant deletion, removing exons 11-107 was found in mosaic form in a patient presenting with a predominantly distal myopathy and asymmetric muscle NEB , ACTA1 , TPM3 , TPM2 , TNNT1 and TNNT3 and MYPN encode integral structural proteins of the thin filament (nebulin, α-actin, α-tropomyosin, βtropomyosin, slow troponin T, fast troponin T and myopalladin, respectively). Actin interacts with myosin during muscle contraction, while the filamentous proteins tropomyosin and nebulin cover the actin filament. The myosin-binding sites on actin are revealed through a conformational change in tropomyosin, triggered by a signal from the troponin complex. Nebulin is an enormous protein spanning the entire length of the filament, anchoring the thin filament into the Z disc, via interaction with myopalladin and other mechanisms. Cofilin-2 and leiomodin-3 ( CFL2 and LMOD3 , respectively) have opposing functions in regulating the actin filament dynamics (F-actin/G-actin). KBTBD13 , KLHL40 and KLHL41 encode proteins involved in the regulation, stability or turnover of the thin filament proteins. KLHL40 and KLHL41 interact with leiomodin-3 and nebulin, and potentially have similar localisation patterns as leiomodin-3 along the A-band region, with additional localisation at the I-band region of the filament. KBTBD13 is involved in a pathway of protein ubiquitination, and is known to bind to actin along the thin filament, with further localisation at the Z-disc where the adjacent filaments overlap.
weakness [60] . Additionally, large recessive pathogenic CNVs in the triplicate (TRI) region in the central part of NEB are estimated to be present in 10-15% of NEB -NM patients [61] . A gain or loss of one TRI copy is considered as benign variation but gains of two or more copies have been interpreted to be pathogenic.
The most common NEB mutation is a deletion removing the entire exon 55 [62 , 63] . Although the reading frame is not disrupted, the modular super-repeat structure of nebulin is perturbed. Interestingly, the phenotype a homozygous mouse model with the exon 55 deletion ( Neb del55 ) was considerably more severe than the human phenotype, and the mice died within the first week [64] . The protein and RNA levels in the mouse model were considerably low, suggesting rapid removal of the defective products. Indeed, it has been suggested that in some patients, the level of nebulin expression may correlate with the severity of disease [65] . Many of the conditional Neb -cKO mice survived to adulthood with very low quantities of nebulin [66] . While the mice carried no pathogenic variants causing absence of expression, they recapitulated some of the hallmark features of NM, including muscle weakness, nemaline bodies and slower fibre types compared with the control mice.
Modelling a milder, or typical NM phenotype has appeared difficult in the case of NEB . Two compound heterozygous mouse models were recently published, one with a combination of missense and nonsense mutations, Neb Y2303H,Y935X [67] , and another with a missense variant combined with the exon 55 deletion, Neb S6366I/ Exon55 [68] . Both did show some of the hallmark features of NM, but only the latter presented a detectable phenotype also in vivo , resembling typical NM in patients. In both cases the compound heterozygous genotype lead to a more severe phenotype than in the parental strain with the missense mutation in a homozygous form [67 , 68] . In humans, homozygous missense mutations in NEB cause distal myopathy, while the same missense mutation paired with a more disruptive mutation leads to NM [24] .
Actin polymerises into long filaments to form the core of the thin filament, while the filamentous proteins tropomyosin and nebulin cover the actin filaments [33] . The main function of actin is to interact with myosin during muscle contraction. Pathogenic variants in ACTA1 are most commonly de novo dominant missense mutations (90%), and predominantly lead to severe NM by a dominant-negative effect. Autosomal recessive variants (10%), including splice-site, nonsense, frameshift and some missense mutations, result in null alleles [51] . In the transgenic mouse models harbouring known missense variants found in NM patients, the variants strengthened the actin subunit interactions. This changed the intra-filament distances and stiffened the filament, leading to disruption of the proper attachment of myosin heads to actin [69 , 70] . Promising results have been obtained in restoring the force-generating capacity by manipulating myosin expression [71] . Actin function was also restored in the early lethal  [72] .
The myosin-binding sites on actin are revealed by a conformational change in tropomyosin in the presence of calcium [33] . Actin filaments coated with tropomyosin show resistance to bundling, severing, branching, and disassembly by actin-binding proteins [73] . The pathogenic variants identified in TPM3 and TPM2 , accounting for less than 10% of NM cases, affect the coiled-coil structure of tropomyosin, changing the dimer-forming and actin-binding properties of the proteins [19 , 51] . Pathogenic variants in both genes are most commonly dominant missense mutations or in-frame deletions removing one amino acid. Recessive variants leading to severe NM are more common in TPM3 than in TPM2 . Pathogenic variants in TPM2 usually lead to a milder NM phenotype [51] . In TPM3 , a large homozygous deletion, removing the promoter and the first two exons has been found causing a severe form of NM [74] .
The troponin complex is responsible for calcium sensing by its subunit, troponin C, and for transmitting the signal to the tropomyosin covering the myosin-binding sites on actin by the subunits T and I [75] . The subunit found defective in NM is troponin T, binding to actin and tropomyosin. Most of the pathogenic variants found in the gene encoding slow troponin T1, TNNT1 , have been recessive, leading to a severe, progressive NM [8] . In Amish patients with NM, a homozygous nonsense mutation was found in the TNNT1 gene, resulting in a stop codon in exon 11. The mutant protein was rapidly degraded at the protein level as a protective mechanism by the muscle cells [76] . The first dominant variant, a heterozygous missense mutation in TNNT1 , was recently described causing mild NM [41] . The mutant protein was expressed and was thought to act via a dominant-negative effect.
The mouse models representing NM caused by mutation of slow troponin T1 presented with characteristic atrophy and loss of type I fibres, and compensatory hypertrophic growth of fast fibres [75] . The defects resulted in impaired fatigue tolerance of the diaphragm muscle, and hypotrophy of the diaphragm, potentially mimicking the respiratory insufficiency in patients. Surprisingly, complete loss of slow troponin T did not affect the normal lifespan of the KO model, but resulted in loss of slow-twitch extra-fusal muscle fibres and reduced the resistance of its muscle to fatigue [75] . However, this was not sufficient to fully explain the severe phenotype, especially the neurological abnormalities such as tremors, interestingly noted in heterozygotes also [8 , 28] . Recent studies demonstrated that the loss of troponin T and a compensatory increase of cardiac troponin T in nuclear bag fibres increase myofilament calcium sensitivity and tension, thus affecting spindle activities, providing a possible explanation for the very unusual pathophysiology of TNNT1-related NM [75] . In an ovine model of TNNT1 myopathy, the pathogenic variant did not reduce troponin T1 protein levels or affect its localisation, but impaired its ability to modulate muscle contraction in response to changes in calcium concentrations [77] .
Hitherto, only one NM-causing pathogenic variant, a homozygous splice-site mutation, has been identified in the gene encoding fast troponin T3, TNNT3 [9] . This mutation leads to deficiency of the fast troponin T protein with secondary loss of fast troponin I.
Myopalladin recruits proteins at the Z disc during myofibrillogenesis and connects the thin filaments to the Z disc via interaction with nebulin's C terminus [7] . Myopalladin plays a critical role in the control of skeletal muscle growth through its effect on actin dynamics, and, consequently, the serum response factor pathway [78] . In addition, myopalladin is important for the maintenance of Z-line integrity during exercise and ageing. All NM-causing MYPN variants are recessive loss-of-function mutations, leading to total lack of the protein, or to very low levels of myopalladin, and resulting in a relatively mild NM with slowly progressive muscle weakness [7] . In a Mypn -KO model, reduced myofibre cross-sectional area (CSA) was found to be responsible for the observed muscle weakness, and the disease phenotype was aggravated by exercise [78] .

Regulatory proteins
Cofilin-2 and leiomodin-3 have opposing functions in regulating actin filament dynamics (F-actin/G-actin) [13 , 14] . Leiomodin-3 is an actin-nucleation protein, whereas cofilin-2 is responsible for severing the actin filament at the pointed end of the thin filament. Leiomodin-3 also binds to nebulin and is located along the entire A-band region of the thin filament, with its major localisation at the pointed end. Pathogenic variants in CFL2 have been recessive missense mutations (two families) or small deletions (one family), leading to loss of function [13 , 79 , 80] . Deficiency of cofilin-2 results in uncontrolled actin filament growth in the sarcomeres [13 , 81] . Progressive disruption of sarcomeric architecture and accumulation of F-actin lead to severe muscle weakness and early lethality in the KO-mouse model for CFL2 NM [82] . A novel KI-mouse model harbouring the first human missense mutation linking the gene to NM mimicked the phenotype of the KO mice. The KI model was normal at birth, but rapidly deteriorated and died by postnatal day nine [83] .
Absence of leiomodin-3 in the KO-mouse model led to severe muscle weakness and postnatal growth retardation [84] . Most of the disease-causing variants in patients are recessive loss-of-function mutations (nonsense or frameshift), leading to complete absence of the leiomodin-3 protein, causing severe, often lethal forms of NM [14 , 85] . Loss of the protein resulted in shortening and disorganisation of the thin filaments in the muscles of LMOD3 -NM patients.
The Kelch proteins, KLHL40, KLHL41 and KBTBD13, are involved in the stability or turnover of the thin filament proteins, a critical process that is required for normal functioning of skeletal muscle [86][87][88] . KLHL40 and KLHL41 regulate leiomodin-3 and nebulin, and potentially have similar localisation patterns to leiomodin-3 along the A-band region, with an additional localisation at the Iband region of the filament. According to the current model, KLHL40 serves to increase nebulin and leiomodin-3 abundance and prevents degradation of leiomodin-3 by the proteasome, through blocking its ubiquitination [86] . In NM patients, loss of KLHL40 leads to a severe lethal phenotype, associated with destabilisation of thin filament proteins [10 , 86] . In the Klhl40 -KO mouse model, deficiency of KLHL40 resulted in almost complete absence of leiomodin-3, and a 50% reduction of nebulin quantity [86] .
Recessive mutations in KLHL41 cause clinically diverse forms of nemaline myopathy, ranging from mild to severe phenotypes [11] . Defects in KLHL41-mediated ubiquitination of sarcomeric proteins contribute to structural and functional deficits in skeletal muscle [87] . Loss of KLHL41 activity in the KO mouse model leads to nebulin aggregation, as well as to down-regulation of nebulin and other sarcomeric proteins and increase in NRAP [87 , 89] . Pathologically increased quantities of NRAP lead to myopathy in a transgenic NRAPoverexpressing zebrafish model [87] . In the klhl41-deficient zebrafish recapitulating NM pathology, downregulation of NRAP resulted in a significant improvement in skeletal muscle pathophysiology [11 , 87] . Consequently, it has been proposed that regulating NRAP could act as a potential target for therapy for NM patients with KLHL41 mutations.
Pathogenic variants in KBTBD13 have been found in autosomal dominant NM with cores, and characteristic slowness of movements [12] . KBTBD13 is involved in a pathway leading to ubiquitination of proteins destined for degradation [90] , and acts as a regulator of skeletal muscle relaxation through its interaction with actin [88] . Defects in KBTBD13 impair muscle-relaxation kinetics by causing structural changes in the thin filament, possibly at least in part explaining the slowness of movement in these patients [88] .

Concluding remarks and future prospects
Since the first NM-causing TPM3 mutations were published in 1995 [5] , at least eleven new NM genes have been identified, some encoding structural proteins of the sarcomere, others proteins with mainly regulatory roles. This dual composition of proteins aberrant in NM has complicated the search for common denominators in the pathogenetic process leading to nemaline body formation and muscle weakness. Now that a growing number of animal models for NM has been established, in fact representing all causative genes except TPM2 and TNNT3 , our understanding of the pathogenesis is steadily building up. Not surprisingly, there are multiple pathways leading to the same outcome.
Is it thus possible to discern differences between the NMs caused by mutations altering structural proteins and those causing aberrations in regulatory protein function? The small number of patients with mutations in the regulatory genes hampers drawing definitive conclusions, but we highlight some aspects.
It is clear that a conformational change in any protein in a highly organised structure, such as the sarcomere, may cause changes in the positioning of the binding sites, thus disrupting or altering protein interactions. While mutations in the genes encoding integral structural proteins may affect thin filament interactions, the stability and turnover of these proteins by the regulatory proteins appears equally important for thin filament function.
A complete loss of function of the integral structural proteins is rare, with the exception of the troponins and myopalladin, while in NM with aberrations of the regulatory proteins, function is often lost ( KLHL40 , KLHL41 , LMOD3 ). In some such cases, a compensatory mechanism may rescue the phenotype in part. An example is provided by the case of null mutations of a structural protein; rare patients devoid of slow skeletal muscle actin were partly compensated for the absence by spontaneous upregulation of ACTAC -encoded cardiac actin. Indeed, upregulating cardiac actin rescued the phenotype in the early lethal Acta1 -KO model [72] . However, in most patients, ACTA1 mutations act in a dominant-negative manner. The effects of ACTA1 alterations have been shown in the Tg. ACTA1 D286G mouse model to be dose-dependant, suggesting that changing this ratio in patient muscle could offer a target for therapy for patients with dominant ACTA1 disease [91] .
Altered levels of the Kelch proteins regulating the major players in the thin filament may result in secondary effects on the regulation of other sarcomeric proteins [86] . As the roles of the proteins associated with protein turnover and stability are becoming known, the defects in their function may reveal another level of pathogenetic mechanisms in NM, offering novel targets for therapy. Recently, for example, it has been proposed that regulating NRAP could act as a potential target for therapy for NM patients with KLHL41 mutations [87] .
Clearly, there is some overlap to be found in the molecular mechanisms leading to NM from mutations in structural and regulatory genes. For example, ACTA1 mutations may affect the distances between the actin monomers within the thin filament, leading to ineffective binding of the myosin heads [69 , 70] . A similar effect has recently been revealed in the case of KBTBD13 mutations [88] . It is possible that in nebulin also, a change in conformation could affect all down-stream binding sites, leading to altered interactions along the length of the filament [92] . Furthermore, mutations affecting the function of troponin T and tropomyosin have had similar effects in the activation of thin filaments during muscle contraction [75 , 93] . Indeed, while the primary defect is in the thin filament, the interaction between actin and myosin is often perturbed [33] . For this reason, a fair proportion of current research has focused on compounds enhancing myosin activation and contraction.
In the Acta1 His40Tyr mouse model, in which effective actomyosin interaction has been found to be impaired, promising results have been obtained in restoring the force-generating capacity by manipulating myosin expression [71] . In this model, the lower number of strongly bound cross-bridges was underlying the force depression observed [71] . The same mechanism was observed in the Neb Y2303H,Y935X model [67] , suggesting that a similar therapeutic modality might be functional in cases of different causative genes.
A number of pharmacological compounds targeting thick and thin filament interactions and thus promoting force generation could potentially be harnessed for improving actomyosin function [94] . Enhanced acto-myosin interaction may also be achieved by calcium sensitizers, compensating for the impaired response of the contractile proteins to calcium binding [95] .
Exon skipping has been used for restoring the reading frame or suppressing a stop codon in genes encoding large muscle proteins, such as dystrophin [94] . This method could potentially be applied to NEB mutations as well, but attempts have been greatly hindered by the complex structure of the nebulin protein, i.e. its modularity and periodical actinand tropomyosin-binding sites along the protein [52] . These indicate that skipping of a larger fragment is likely to be required, further complicating the matter. Moreover, in the case of NEB , very few patients share the same mutation, and the 250 mutations identified to date are dispersed across the entire length of the gene [51] . It is also currently uncertain whether the efficacy of the method is high enough to restore the quantity of functional protein sufficiently to achieve clinically relevant improvement [94] .
For future clinical trials of potentially effective and safe therapeutic modalities, in addition to a thorough understanding of the underlying mechanisms, comprehensive natural history data will be required. It remains to be seen whether some types of therapy will be found to be beneficial in all forms of NM, while more specific ones will be effective in the case of certain genes or mutations. Most likely, both types of therapy will need to be combined for improving NM patients' state of health. The growing understanding of the pathogenetic mechanisms has led us closer to completing the picture.

Declarations of Competing Interest
None.