In this study, we combined MRI and WES to reveal and broaden the phenotypic and genotypic spectrums in 122 Chinese CCDDs patients from 96 families.
KIF21A and TUBB3 were the common pathogenic genes of Chinese CFEOM
The mutation detection rate of CFEOM was the highest (63.6%, 42/66), which was much higher than other disease groups (DRS, MBS and HGPPS). And the mutation detection rate of familial CFEOM was 100%. The common pathogenic genes of CFEOM were KIF21A (73.8%, 31/42) and TUBB3 (26.2%,11/42) genes. Whitman et al. summarized the results of previous researches and found that the variants of KIF21A, TUBB3, PHOX2A, TUBA1A, COL25A1 and TUBB2B were identified in 55%, 35%, 10%, < 1%, < 1% and < 1% of the CFEOM patients, respectively [22]. These results are consistent with our current research findings.
KIF21A belongs to a family of kinesin motor protein, which involved in cargo transport along microtubules [5]. Pathogenic variants in KIF21A gene may cause CFEOM by modifying the autoinhibitory interaction between the motor domain and a regulatory region in the stalk [23]. The majority of previously reported variants located in coiled coil domains of KIF21A, meanwhile, the CFEOM patients are almost all present with isolated phenotype [5]. However, the novel variant F355S identified in our study was located in conserved N-terminal motor domain, and the patient showed rare syndromic phenotype. To our knowledge, only three variants located in the motor region of KIF21A have been reported, they were C28W, D352E, M356T [5, 24, 25]. The patients harboring C28W and M356T variants showed isolated CFEOM, while the patients harboring D352E variant present with syndromic phenotype. Notably, patient harboring F355S in this study and the patient harboring D352E had the similar phenotypes, including CFEOM, facial weakness, frontal bony prominence, exposure keratitis and delayed developmental milestones. The difference is that the F355S patient also had the hypoplasia of corpus callosum in MRI. We speculate that the F355S may attenuate autoinhibition to a greater degree, or induce disease via different pathways [26].
TUBB3 gene encodes the neuron-specific beta-tubulin isotype 3 which plays a critical role in axon guidance and neuronal migration. Pathogenic variants in TUBB3 may cause CFEOM by altering dynamic instability of microtubules or disrupting the interaction of microtubules with kinesin motors [27].
We found a novel variant TUBB3-S78T in a CFEOM pedigree, the proband has normal developmental history (C9-2). The S78T is located in the N-terminal domain of TUBB3, so far, several variants have been reported in the region of S78. Whitman et al. [28] reported the patients harboring G71R and G98S variations presented with syndromic phenotypes, manifested as CFEOM, nystagmus, torticollis, developmental delay, intellectual and social disabilities. MRI showed the hypoplasia of the corpus callosum and anterior commissure, malformations of hippocampi, thalami, basal ganglia and cerebella, and hypoplasia of brainstem and cranial nerve. The patients harboring the S78T variant in this study and the patient harboring R62Q variant reported by Max A. Tischfield in 2010 presented with isolated CFEOM [6]. Although the S78T patient (C9-2) underwent MRI exam of the ocular motor nerves, and had the normal corpus callosum configuration, however the high resolution brain MRI data was absent, so we cannot rule out the possibility that he has subtle cortical dysplasia. In addition, we noted that G82R have been reported to cause MCD without CFEOM [29].
Compared to patients harboring KIF21A variants, the CFEOM patients with TUBB3 variants tend to show syndromic phenotypes. In the past, researchers named the disease caused by the TUBB3 variants as the “TUBB3 Syndromes” [6]. In recent years, researchers found that different variants of TUBB3 usually produce different syndromic phenotypes [6]. In our cohort, the patients harboring E410K and R262H variants of TUBB3 gene exhibited CFEOM and accompanied by different multiple congenital malformations, and their clinical and MRI findings were highly similar to previous reports [4, 30], which confirmed that the correlations between specific variants and phenotypes were strong. Therefore, researchers preferred to name these syndromic phenotypes after specific variants. Up to date, “TUBB3 E410K syndrome”, “TUBB3 R262H syndrome” and “TUBB3 M323V syndrome” were used to name the diseases caused by E410K, R262H and M323V variants, respectively [4, 31, 32]. Furthermore, the patients with “TUBB3 E410K syndrome” and “TUBB3 R262H syndrome” were easily misdiagnosed as MBS owing to similar clinical manifestations (facial weakness and limitation of eye movements etc.) [33]. MRI, WES and specific general examinations based on the good phenotype-genotype correlations can contribute to accurate diagnosis and follow-up.
TUBB3 R380C variant was identified in one CFEOM patient (C6) in our cohort, and this variant has been previously reported only once in a CFEOM patient [6]. The clinical and MRI findings of these two patients were highly similar, both of them manifested CFEOM with intellectual impairment, multiple congenital craniocerebral malformations which involved in commissural fiber, brainstem, basal ganglia, thalamus, cerebral ventricle and cerebellum etc. However, “bilateral chest wall malformation” and impaired gait were observed only in our R380C patient. Previous research indicated sensorimotor polyneuropathy occurred in individuals harboring R262H, E410K and D417H/N [6]. We speculated that the R380C patient in this study may have similar sensorimotor polyneuropathy, and electromyography results are needed to confirm the suspected diagnosis.
Furthermore, we also noted that TUBB3 R380C variant had been reported in patients diagnosed with “neurodevelopmental disabilities” and “cerebellar dysplasia (CD)” [34, 35], however the ocular manifestations of the patients were not described clearly by authors.
We noted that Chew et al. and Whitman et al. reported CFEOM patients harboring E410K and R262H variants were accompanied by Kallmann syndrome (anosmia with hypogonadal hypogonadism) [4, 31]. Furthermore, Yasuko Nakamura et al. also reported E410K patients presented with Kallmann syndrome [33].
In present study, three patients with TUBB3 E410K and R262H variants (C21, C32 and C52) fitted the diagnostic criteria of Kallmann syndrome, confirmed the previous report results and suggested that TUBB3 gene may involve in the pathogenesis of Kallmann syndrome [4].
Genetic screening results of sporadic DRS, MBS and HGPPS cases were poor
DRS is a congenital motility disorder which characterized by restricted abduction or adduction and accompanied by globe retraction and narrowing of the palpebral fissure [36]. CHN1 gene was a common causative gene for familial non-syndromic DRS, which was identified in approximately 35% of familial DRS cases [11]. Functional analysis revealed that CHN1 variants caused CN6 growth stalled, leading to secondary dysinnervation of the LR by the CN3 [37]. To date, variant detection rate was poor in sporadic isolated DRS patients, so genetic screening is suggestive for those individuals with positive family history [38]. In our DRS cohort (2 familial and 6 sporadic), we identified one novel variant H217R in CHN1 gene in a large DRS family, no causative genes were identified in all sporadic DRS cases and one DRS family. These results were similar to the previous study [38].
MBS is a rare disease with minimum diagnostic criteria of congenital limitation of ocular abduction and facial weakness [3]. In 2015, Laura Tomas-Roca et al. reported that PLXND1 and REV3L may be pathogenic genes for MBS, and animal researches indicated PLXND1 and REV3L may cause a defect in the facial branchiomotor neuron migration [17]. This was the first report of pathogenic gene for MBS, however, these two genes have not been reported again in other MBS cases to date. In addition to genetic factors, intrauterine environmental factors were also reported be involved in MBS [39]. In our study, no causative gene was identified in all MBS patients, and more than half of the patients had intrauterine exposure to adverse factors. These results suggested that multifactorial pathogenic mechanism maybe exist in MBS.
HGPPS is a rare autosomal recessive disorder characterized by severe restriction of conjugate horizontal eye movements and progressive scoliosis [16]. Reviewed the published reports about HGPPS, it could be found that ROBO3 was the pathogenic gene and the HGPPS patients harboring ROBO3 variants often existed parental consanguinity [40, 41]. ROBO3 plays an important role in axonal guidance which mediate midline crossing of neurons during embryogenesis of the spinal cord [16]. Our study enrolled five sporadic HGPPS patients who were from non-consanguineous family. No homozygous variants of ROBO3 gene were found, and the possibility of compound heterozygous mutation could not be ruled out. More than a dozen heterozygous variants of ROBO3 gene were identified in five HGPPS patients, however, no variants were classified as pathogenic or likely pathogenic according to ACMG standards, so the pathogenicity of these variants cannot be determined.
The mutation detection rates in sporadic isolated DRS, MBS and non-consanguineous families with HGPPS were poor in this study. The reasons could be multiple, such as epigenetic factors, environmental or other nongenetic factors may be involved in the pathogenesis of diseases.
The phenotypes of CCDDs are diverse and complex, often involving multiple tissues and organs, many new MRI and clinical phenotypes have been reported and discovered. Face to face communication with radiologist is necessary. At the same time, the participation of clinical multidisciplinary doctors is required to carry out professional examinations, diagnosis and treatment. Considering these facts, future efforts need to focus on mining, refining and quantifying extraocular manifestations to improve our understanding of clinical and genetic basis of CCDDs. Meanwhile, the novel variants need future functional studies to determine its pathogenicity.