There are 2 class 9 myosin proteins expressed in humans; MYO9A (chromosome 15, Figure 1.13) and MYO9B (chromosome 19). Both myosins follow the same basic structural composition as other unconventional myosins, containing a force
generating motor domain in the head region which binds actin and hydrolyses ATP. This is preceded by an extension that has similarities to a Ras-association domain, however it has been shown not to bind Ras in both MYO9A and MYO9B (Chieregatti et al., 1998, Kalhammer et al., 1997). Unique to this subset of myosins is the
presence of a large 100-200 amino acid insertion in the head domain, termed loop 2. This extension contains an IQ motif for binding calmodulin and has been shown to bind actin, allowing this single-headed motor protein to move processively along actin filaments (Elfrink et al., 2014, Saczko-Brack et al., 2016). A further 6 IQ motifs for MYO9A and 4 motifs for MYO9B are present in the neck domain, which can also bind calmodulin. Activity of MYO9B has been shown to be responsive to calcium due to the conformational changes induced by a switch from calcium-free calmodulin to calcium-calmodulin, which inhibit its ATPase activity (Liao et al., 2010) and recently, the ability of MYO9A to bind calcium-calmodulin has also been demonstrated (Li et al., 2017). Both class IX myosins have an extended tail region that contains an atypical C1 domain for which the function is not yet known (typical C1 domains bind diacylglycerol) and a Rho-GTPase activating protein (GAP) domain (Gorman et al., 1999, Bahler et al., 2011). The tail region of MYO9A is longer than MYO9B,
containing a sequence predicted to adopt a coiled-coil structure and two alternatively spliced regions (Chieregatti et al., 1998), resulting in a 293 kDa protein for MYO9A as opposed to 243 kDa for MYO9B. The Rho-GAP domains were found to stimulate the GTPase activity of Rho but not Rac (Post et al., 1998, Chieregatti et al., 1998). MYO9A is predicted to possess 20 different protein isoforms that differ in 5’ and 3’ truncations, as well as absence of 16 cassette exons. Of these splice variants only 4
include the tail region bearing the patient mutations, and 2 of these contain all of the identified domains (Thierry-Mieg and Thierry-Mieg, 2006). Tissue specific expression of these different isoforms may provide an explanation for the NMJ-specific
phenotype of the patients.
Figure 1.12. Structure of MYO9A. Diagram of MYO9A protein domains identified,
including the head region Ras-associating domain and myosin motor domain that contains an extension at loop 2 with an IQ motif. The neck region consists of 6 IQ motifs for binding of calmodulin. The tail region of MYO9A has 3 sequences
predicted to adopt a coiled-coil conformation, as well as an atypical C1 domain and Rho-Gap domain. The location of the 3 MYO9A-CMS patient mutations are noted: patient 1 in purple, compound heterozygous, patients 2 and 3 in green, homozygous. The location of reported mutations in an arthrogryposis patient (Bayram et al., 2016) are also included (blue, compound heterozygous), and homozygous mutations in a foetal akinesia patient shown in red (Maddirevula et al., 2019). Modified from: O’Connor et al (2016).
MYO9A RNA is expressed ubiquitously, with highest levels present in the testis and peripheral nerves, and to a lesser extent in the muscle and brain (GTEx,
https://gtexportal.org/home/gene/MYO9A, last accessed 26.03.19, Chieregatti et al., 1998). As there are a number of different isoforms, expression of specific exons that contain patient mutations were also analysed, revealing that exons 25 (p.R1517H) and 26 (p.D1698G) are predominantly expressed in the testis, nerve, oesophagus muscle and arteries. Exon 40, in which the p.R2283H variant resides, is expressed in the brain, muscles and many other tissues. Considering protein expression levels using the Human Protein Atlas resource, the level of expression is similar across all aforementioned tissues, with the RNA enrichment in the testis absent at the protein level (Uhlen et al., 2015). MYO9B on the other hand is expressed to a higher extent in endocrine tissues, immune cells, muscle, testis and many other tissues at both the protein and RNA level (GTEx, https://gtexportal.org/home/gene/MYO9B, last
The human class IX myosin canonical protein sequences are only 57% similar, which includes only 71% coverage of MYO9A, thus highlighting the likely differing roles of the two proteins within this class. Human MYO9A compared to the mouse homolog displays an 88% similarity, covering 100% of the sequence. Zebrafish also display a high degree of homology with human MYO9A, with Myo9aa 63% similar covering 100% of the sequence, and Myo9ab 62% similar covering 82% of the sequence (https://blast.ncbi.nlm.nih.gov/Blast.cgi, last accessed 26.03.19). Overall MYO9A is well conserved, including within the species relevant to this project.
1.5.4.2 Protein function
Consistent with the identification of an active RhoGAP domain in MYO9A, analysis of the MYO9A interactome as generated by the STRING database reveals associations of this unconventional myosin with a range of Rho family proteins (Figure 1.13). This includes RhoA, with which MYO9A has been shown to interact with in numerous studies by negatively regulating its activity (Abouhamed et al., 2009, Omelchenko and Hall, 2012).
Figure 1.13. String association network for MYO9A. Human MYO9A and
associated proteins are shown, linked with lines representing confidence of
interaction. Interactions do not necessitate binding but suggest the two proteins work towards the same function, based on experimental evidence, databases, text mining, co-expression, neighbourhood, gene fusion and co-occurrence data, performed using the STRING database (Szklarczyk et al., 2019).
Human MYO9A was originally cloned by Gorman et al (1999), performed because it was in a region identified to contain a gene causing the disease Bardet-Biedl
syndrome. Despite this, MYO9A variants have never been identified in patients with this disease. The previous year, Myr 7, the rat MYO9A homolog, was identified as a protein similar in structure to Myr 5 (MYO9B). Initial analysis revealed Myr 7/MYO9A could activate RhoA, but not Rac1 and only partially CDC42 GTPase activities in vitro (Chieregatti et al., 1998). These first observations, coupled with expression studies identifying the protein within the brain, lead to the hypothesis that MYO9A may be important for regulating Rho activity in neurons. It was not until 10 years later that further research emerged suggesting a role for this unconventional myosin protein in brain ependymal epithelial cells that line the ventricles, due to its absence causing hydrocephalus in a mouse model (Table 1-5). This was attributed to altered epithelial cell morphology and cell junctions, but could be attenuated by blocking the Rho- associated protein kinase (ROCK) pathway, a downstream target for RhoA
(Abouhamed et al., 2009). Therefore, this demonstrated for the first time in vivo that MYO9A activity has functional consequences for the RhoA pathway.
Mouse
genotype Tissue affected Phenotype PMID
Myo9a-/- Kidney Bilateral renal disease, polyuria and low molecular
weight proteinuria 26136556
Myo9a-/- Brain
(ventricles) Severe hydrocephalus with stenosis, closure of the ventral caudal 3rd ventricle and the aqueduct 19828736 Myo9a+/- Brain
(hippocampus) Impairment of long term potentiation, synaptic transmission and cognitive function 26834556 Myo9a-/- Gross
phenotype Abnormal gait and small size 19828736
Table 1-5. Phenotypes associated with depletion of MYO9A in mice. Table
shows genotype of mice (homozygous or heterozygous knockout) along with associated phenotypes and main tissue investigated. Pubmed IDs are provided (PMID).
The downstream network of RhoA includes a number of proteins with diverse
functions. However, one of the main roles is in modulation of the actin cytoskeleton, as well as some microtubule and intermediate filament interactions (Jaffe and Hall, 2005). Through these functions, Rho GTPases such as MYO9A are particularly important for cell migration, polarity, movement and junction formation. This is exemplified by reports highlighting that an absence of MYO9A from bronchial epithelial cells leads to cell scattering based on a loss of local actin cytoskeleton organisation and inability of cells to form new adhesions. The binding of actin filaments could be reinstated by expression of the motor domain in the head,
whereas rescue of cell junction formation required the RhoGAP domain (Omelchenko and Hall, 2012). In further support for a role for MYO9A in regulating RhoA-
dependent processes, it has been reported that MYO9A contains a PDZ-binding motif at the c-terminus (p.2540-2550), allowing binding of a viral PDZ-containing protein that brings MYO9A and RhoA into close association allowing MYO9A-
mediated inactivation of RhoA. This leads to altered actin cytoskeleton dynamics and allows viral spread between cells (Handa et al., 2013). In two studies performed to identify genes that may contribute to desirable features in farmed animals there was a suggested contribution of MYO9A to abdominal fatness in chickens and muscle mass in pigs. However, no functional work was performed to support these findings and thus mechanisms of action are unclear (Zhang et al., 2012, Liu et al., 2018). As described, actin dynamics and other unconventional myosins have been shown to play important roles in vesicle secretion from cells. Corroborating this, MYO9A in mouse was shown to localise to cortical actin in kidney tubule cells, where its loss affected endocytosis of ligand bound receptors from the cell surface (Thelen et al., 2015). A previously unreported effect on diaphanous-related formin 1 was also identified, a Rho effector protein that modulates the cytoskeleton and thus is likely a downstream target of the MYO9A/RhoA pathway.
Almost 20 years after initial predictions that MYO9A would play a role in neurons, 2 studies reported effects of MYO9A in the CNS and PNS. The first of these was by Folci et al (2016), demonstrating an association of MYO9A with the postsynaptic density of hippocampal neurons from rat, where it binds to the GluA2 subunit of the alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR). Heterozygous knockout of MYO9A from mice impacted on long term potentiation and performance in cognitive tasks as outlined in Table 1-5. While mechanisms for such an effect were not investigated, it was hypothesised that MYO9A affects AMPAR endocytosis and trafficking through both RhoA-mediated and myosin motor-mediated pathways. As a result, its loss leads to a surface build-up of AMPARs and affects synaptic plasticity. The second such study to implicate a role for MYO9A in neurons was based on our observations that 3 patients with CMS harboured variants in MYO9A that segregated with the disease, and found MYO9A to localise at the NMJ in mouse, and loss of the protein to impact on NMJ development in zebrafish
(O'Connor et al., 2016). Another disease, arthrogryposis, has also been associated with MYO9A mutations, in which compound heterozygous missense variants were
identified in one patient (Bayram et al., 2016). Association of a gene with both arthrogryposis and CMS has occurred previously, as there are overlapping phenotypes. For example, reduced foetal movements due to prenatal NMJ
transmission defects sometimes present in CMS and can lead to joint contracture (arthrogryposis) formation (Brownlow et al., 2001, Vajsar et al., 1995). A homozygous nonsense mutation in MYO9A was recently reported in an autozygosity study for a foetal akinesia patient, linking in with the report of arthrogryposis, and also with the hydrocephalus phenotype observed in MYO9A KO mice (Maddirevula et al., 2019). The mutation was positioned in the motor domain, the same region in which one of the arthrogryposis variants was located, and thus may explain the severity of the patient’s phenotype due to the likely truncation of the protein near the n-terminus. Reflecting this observation, the motor domain of MYO9A has been shown to play an important novel role delineated in vitro, in which MYO9A crosslinks actin filaments to form polar structures that may provide support for localised RhoGAP activity, as well as involvement in cell polarity and migration (Saczko-Brack et al., 2016). Further defects due to a role of this myosin in cytoskeleton regulation has been highlighted by the identification of MYO9A gene dysregulation in monosomy blastocysts (in which a chromosome is absent, chromosome 15 in this case). Decreased expression of MYO9A was hypothesised to affect implantation of monosomy blastocysts in pregnancy due to roles in cytoskeleton modulation and cell junction formation, although no functional data was obtained (McCallie et al., 2016).
With regards to clinical effects of RhoA pathway modulation due to MYO9A, the presence of ANXA2-MYO9A gene fusions were reported to significantly worsen the prognosis of diffuse gastric cancer patients. The gene fusion product maintained the RhoGAP domain of MYO9A and caused its overexpression, subsequently reducing the activity of active RhoA in cells (Yang et al., 2018). Hum-7 is the C.elegans
homolog of MYO9A/B. Although it only has approximately 20% protein homology with its human counterparts, it contains the main domains. Depletion of Hum-7 increased embryonic lethality, corroborating previous studies suggesting roles for MYO9A in early development (Bayram et al., 2016, Maddirevula et al., 2019, McCallie et al., 2016, O'Connor et al., 2016). Further supporting the role of MYO9A in nervous development, Hum-7, was also found to affect postembryonic axon guidance, causing axon migration defects when knocked down (Wallace et al., 2018).
Overall, a number of recent studies have built on the suggested roles for MYO9A in neuronal function but also broadly in cytoskeletal structure and RhoA pathway modulation. However, the precise role of MYO9A at the NMJ remains to be established. Furthermore, whether such a role is due to interactions with the
cytoskeleton and RhoA pathway in particular is not known, or if there is involvement of other pathways. A mechanism by which dysfunction of MYO9A could disrupt the safety factor of the NMJ and lead to impaired signalling resulting in CMS also requires investigation. These questions will be addressed within this thesis, with a particular focus on the roles of MYO9A in morphology and function of the NMJ.