Genotype–phenotype correlates in Joubert syndrome: A review

Abstract Joubert syndrome (JS) is a genetically heterogeneous primary ciliopathy characterized by a pathognomonic cerebellar and brainstem malformation, the “molar tooth sign,” and variable organ involvement. Over 40 causative genes have been identified to date, explaining up to 94% of cases. To date, gene‐phenotype correlates have been delineated only for a handful of genes, directly translating into improved counseling and clinical care. For instance, JS individuals harboring pathogenic variants in TMEM67 have a significantly higher risk of liver fibrosis, while pathogenic variants in NPHP1, RPGRIP1L, and TMEM237 are frequently associated to JS with renal involvement, requiring a closer monitoring of liver parameters, or renal functioning. On the other hand, individuals with causal variants in the CEP290 or AHI1 need a closer surveillance for retinal dystrophy and, in case of CEP290, also for chronic kidney disease. These examples highlight how an accurate description of the range of clinical symptoms associated with defects in each causative gene, including the rare ones, would better address prognosis and help guiding a personalized management. This review proposes to address this issue by assessing the available literature, to confirm known, as well as to propose rare gene‐phenotype correlates in JS.


| INTRODUCTION
Joubert syndrome (JS) is a rare congenital neurodevelopmental primary ciliopathy with a population-based prevalence reaching 1.7 per 100,000 in the age range 0-19 years . First described by Dr Marie Joubert about 50 years ago (Joubert, Eisenring, Robb, & Andermann, 1969), JS is now diagnosed upon recognition of a pathognomonic malformation of the midbrain-hindbrain junction which results in the brain imaging finding "molar tooth sign" (MTS).
This malformation, found in all patients, consists of cerebellar hypoplasia with vermian dysplasia, thick and horizontally oriented superior cerebellar peduncles, and an abnormally deep interpeduncular fossa (Maria et al., 1997). A spectrum of severity of the MTS has been reported (Poretti, Huisman, Scheer, & Boltshauser, 2011), and mild MTS presentations may be difficult to assess. Conversely, other cerebellar and brainstem malformations are sometimes wrongly interpreted as a mild MTS, leading to misdiagnosis (Aldinger et al., 2014;D'Abrusco et al., 2021;Powell et al., 2021).
Even though some facial dysmorphisms are often observed, facial features do not strongly support the clinical diagnosis, as in many patients they can be unremarkable (Braddock, Henley, & Maria, 2007).
A variable involvement of other organs (such as eye, kidney, liver, and skeleton) is present in two-thirds of individuals with JS, and can manifest at different ages and with variable severity . This complex presentation makes JS a multisystem condition, and some clinical issues may be progressive, complicating medical management.
Based on the presence of associated extra-CNS features, JS can be classified into clinical subgroups Romani et al., 2013): • Purely neurological JS (pure JS) • JS with ocular involvement (JS-O) • JS with renal involvement (JS-R) • JS with oculorenal involvement (JS-OR) • JS with hepatic involvement (JS-H, or COACH syndrome) • JS with orofaciodigital involvement (JS-OFD, or OFDVI syndrome) • JS with acrocallosal features • JS with Jeune asphyxiating thoracic dystrophy Occasional features observed in all subgroups include polydactyly, which can be pre-, meso-, or post-axial and variably involve hands and feet, and other CNS abnormalities, such as corpus callosum abnormalities, hydrocephalus, encephalocele, or polymicrogyria. While these complex phenotypes had initially been termed "Joubert syndrome related disorders" Satran, Pierpont, & Dobyns, 1999), nowadays the unifying term "JS" applies to all patients with the MTS, including those with and without any extra-neurological involvement (Romani et al., 2013).
Like other syndromic ciliopathies, JS is characterized by extreme genetic heterogeneity with more than 40 causative genes, all of which encoding proteins responsible for the formation or functioning of the primary cilium, a subcellular organelle playing essential roles in developing and adult tissues (Reiter & Leroux, 2017). Pathogenic variants in known genes overall account for~62-94% of affected individuals depending on the cohort, sequencing method, and criteria for defining pathogenicity of identified variants (Bachmann-Gagescu, Parisi, 2019;Phelps et al., 2018;Shaheen et al., 2016). In the largest cohort reported to date, five major genes (CPLANE1, CC2D2A, AHI1, CEP290, and TMEM67) accounted each for~6-9% of JS cases, three additional genes (CSPP1, TMEM216, and INPP5E) accounted each for~3%, and six more genes accounted each for~1-2%; the remaining genes were mutated only in few families .
JS is mainly inherited in an autosomal recessive fashion, with the exception of an X-linked recessive form due to pathogenic variants in the OFD1 gene (Coene et al., 2009) and a recently reported autosomal dominant form due to truncating or splice-site variants in the SUFU gene . Of note, almost all JS genes have also been implicated in other ciliopathies, such as Meckel syndrome (MKS), isolated nephronophthisis (NPHP), Leber congenital amaurosis (LCA), oralfacial-digital syndromes (OFDS), Bardet-Biedl syndrome (BBS), and others. While the JS diagnosis is usually made early in life in the setting of a neuropediatrics or genetics clinic (and can even be suspected prenatally), the possible impairment of other systems requires patients to enter a diagnostic workflow and regular follow-up examinations, with referral to distinct specialists according to the organs which are involved. In this light, the establishment of gene-phenotype correlations would enable reliable prognostic predictions and ensure the optimal assessment and management of disease complications. Yet, despite certain JS gene variants appear to be highly associated with specific features, such correlations are still challenging, due to the extreme genetic heterogeneity of this ciliopathy. Indeed, even in the largest published cohorts, the number of patients with pathogenic variants in the same gene is small, limiting statistical power for correlations.
Here, we aim to provide an overview of the more consistent and reliable correlations between phenotype and genetic cause in JS, supporting the notion that these correlates may help prognostic definition and the development of a personalized management; in addition, we mention possible associations between gene and phenotype, that still require confirmation in larger cohorts ( Figure 1). Finally, in a concluding paragraph, we underline the clinical and genetic overlap of JS with other ciliopathies, in particular MKS. JS and MKS were originally described as distinct and clinically recognizable entities; however, to date, at least 16 genes have been found to cause both conditions, and the more cases are reported, the more blurred their clinical distinction becomes. Indeed, these syndromes show a variable combination of overlapping clinical findings and likely represent different expressions of the same disease spectrum.

| METHODOLOGY
We performed a literature review, searching for English written publications in the National Center for Biotechnology Information's PubMed database (https://www.ncbi.nlm.nih.gov/pubmed), using the following search terms: "Joubert syndrome" AND "gene" OR "genetics." We selected all relevant articles from 2004, when the first gene implicated in JS, AHI1, was identified (Ferland et al., 2004), to present.
In reporting the results, emphasis has been given to larger studies and more consistent associations, while studies reporting single or few observations have been referenced but not discussed in detail.

| Brain imaging
While essential to establish the diagnosis of JS, the MTS is not usually helpful to establish correlations with different genotypes. A possible exception is represented by the detection of thinner superior cerebellar peduncles and less severe vermis hypoplasia seen in patients with NPHP1 homozygous deletion (Castori et al., 2005;Parisi et al., 2004). In addition, a "mild MTS" consisting in vermis hypoplasia, superior cerebellar folial dysplasia, and subtle-to-mild abnormalities of the superior cerebellar peduncles (which variably appeared long, thick, and horizontal), has been reported in patients with SUFU heterozygous loss of function variants as well as in some patients carrying biallelic variants in CBY1, CPLANE1, and FAM149B1 (Enokizono et al., 2017;Epting et al., 2020;Serpieri et al., 2021;Shaheen et al., 2019).
Aside from the MTS, a wide range of additional brain anomalies have been described in JS individuals, such as occipital encephalocele, ventriculomegaly, dysgenesis of the corpus callosum, neural migration defects, and hypothalamic hamartoma (Bachmann-Gagescu, Poretti et al., 2017). All these features have been associated with variants in several genes, lacking specific gene-phenotype correlates.
Poretti et al. reported neuroimaging findings of 110 JS patients, of whom 89 (81%) with a genetic diagnosis. In this series, dysgenesis of the corpus callosum was identified in patients with pathogenic variants in CC2D2A, CPLANE1, CSPP1, and CELSR2 .
Notably, the only patient of the cohort with KIF7-related JS had a normal corpus callosum, although pathogenic variants in KIF7 are known to cause both JS and acrocallosal syndrome (ACLS), which is characterized by corpus callosum agenesis, distal anomalies of limbs, minor craniofacial anomalies, and intellectual disability (Dafinger et al., 2011;Putoux et al., 2011). In other studies, corpus callosum anomalies have been occasionally detected in association to KIF7 as well as several other genes (Akizu et al., 2014;Bader et al., 2016;Ben-Salem et al., 2014;Edvardson et al., 2010;Shen et al., 2020;Stephen et al., 2017;Wentzensen et al., 2016). F I G U R E 1 Genotype-phenotype correlations in Joubert syndrome. Top (dark gray oval): genes definitively associated with the feature (statistically significant associations as detected in large studies). Middle (medium gray oval): genes probably associated with the feature (associations reported in three or more papers or in at least 10 distinct families). Bottom (light gray oval): genes possibly associated with the feature (associations reported in less than three studies). JS: Joubert syndrome

| Neurodevelopmental features
Neurodevelopmental features in JS are related to the underlying brain malformation and, as such, represent constant features of the syndrome, and usually the earliest to manifest. These consist in neonatal hypotonia, abnormal eye movements (mainly ocular motor apraxia, OMA), and breathing dysregulation (apneas and/or hyperpneas usually resolving within few months of life). A delay of developmental milestones becomes evident soon after, followed by ataxia (with broad-based, unsteady gait and difficulties in running or climbing stairs) and intellectual disability of variable severity; nevertheless, a minority of individuals have borderline, or even normal cognitive functions (Bulgheroni et al., 2016;Poretti, Alber, Bürki, Toelle, & Boltshauser, 2009;Summers et al., 2017). Notably, Poretti et al. reported that neuroimaging may predict the neurodevelopmental outcome, as a high degree of vermis hypoplasia was found to correlate with worse prognosis . Of note, expressive speech is more affected than comprehension, also due to concurrent oralmotor apraxia .

| Neurological phenotype
Patients presenting only neurological manifestation of JS without other organ involvement are hystorically classified as "pure JS." Overall, in the UW cohort, only 68 out of 201 exhaustively phenotyped patients (34%) had an exclusively neurological phenotype (Bachmann-Gagescu, , eventually associated with polydactyly and other brain abnormalities (Parisi, 2019;Valente, Dallapiccola, & Bertini, 2013). Similarly, neurological JS accounted for 28% in another series of 100 JS individuals by the National Institutes of Health (NIH) Clinical Center . In contrast with these observations, a neurological phenotype was the commonest subtype (62.7%) in another, distinct cohort of 59 patients, but age at examination was too early to assess the occurrence of other organ abnormalities, hampering gene-phenotype correlates (Radha Rama Devi, Naushad, & Lingappa, 2020).
Although no major genes have been statistically linked to this purely neurological presentation, some genes seem to recur more frequently in this group. For instance, in the UW cohort, JS patients mainly carried pathogenic variants in CPLANE1 (13%), CC2D2A (13%), AHI1 (12%), and CSPP1 (9%) (Bachmann-Gagescu, . In the NIH cohort, almost 29% of JS patients carried CPLANE1 defects, while 18% and 14% had pathogenic variants in CC2D2A and KIAA0586, respectively (Brooks et al., 2018;Vilboux, Doherty, et al., 2017). An association of CPLANE1 defects with JS was also detected in a smaller Northern European cohort of 51 JS patients, in which five out of six patients with causative variants in CPLANE1 showed a purely neurological phenotype . More significantly, in a large cohort of 313 JS individuals mainly of Italian origin, 27 out of 28 patients carrying biallelic variants in CPLANE1 manifested an exclusively neurological phenotype, although polydactyly was a common associated feature . Of note, we recently reported a novel association of a relatively mild neurological phenotype with haploinsufficiency of the SUFU gene, which to date represents the only genetic cause of JS with dominant inheritance and reduced penetrance .

| Abnormal ocular movements
Abnormal ocular movements in JS patients, which tend to improve over time, include OMA, nystagmus, and strabismus. OMA manifests as the abolition of the vestibulo-ocular reflex, decreased smooth pursuit, and the inability to visually follow objects, which is compensated by turning head movements. In a recent comprehensive review of 254 JS individuals with different eye phenotypes, OMA was the commonest ocularmotor abnormality in JS, associated with variants in most genes (Wang et al., 2018). Of note, patients with heterozygous SUFU variants may present with a spectrum of neurodevelopmental phenotypes encompassing congenital OMA and mild JS (Schröder et al., 2020;Serpieri et al., 2021).
Horizontal nystagmus at birth, improving with age, is also a frequent feature; besides this, torsional and pendular rotatory nystagmus have been occasionally observed. AHI1 molecular defects have been frequently correlated with this issue, but causative variants in several other genes have also been reported (Edvardson et al., 2010;Niceta et al., 2020;Wang et al., 2018).
Strabismus is also commonly observed in JS as in many other conditions associated with nonprogressive cerebral or cerebellar abnormalities (Salman & Chodirker, 2015). In the comprehensive review by Wang et al., TMEM237 was the commonest altered gene in patients with this complication but, as for other ocular movement abnormalities, strabismus was also associated to several other gene variants (Huppke et al., 2015;Niceta et al., 2020;Wang et al., 2018).

| Other neurological features
Behavioral disturbances, when present, include temper tantrums, selfinjury, autism, depression, anxiety, and auditory hallucinations Seizures have been observed in more than 10% of JS individuals but no seizure type nor genetic cause appear to be prevalent (Bachmann-Gagescu, . In the large UW cohort, none of the 55 individuals with seizures presented causative variants in CEP290 (despite this gene represented the third commonest genetic cause of JS), suggesting a negative association (Bachmann-Gagescu, . However, other studies have occasionally reported seizures in CEP290-mutated patients (Helou et al., 2007). Conversely, biallelic variants in CC2D2A seem to be more

| Eyes and vision
The ocular findings in JS range from mild to severe, often depending on the underlying genetic cause; sometimes, variability can be noted even within the same genotype. Ocular involvement can be either degenerative (e.g., retinal dystrophy) or developmental (e.g., coloboma), even if rarely co-occurring in the same individual (Brooks et al., 2018;Wang et al., 2018).
The spectrum of ophthalmological features in JS has been prospectively ascertained in the UW and NIH cohorts, while two other studies have reported a detailed analysis of the published literature (Bachmann-Gagescu, Brooks et al., 2018;Vilboux, Doherty, et al., 2017;Wang et al., 2018).  (Brancati et al., 2007;Valente et al., 2006).
Moreover, these two genes seem more likely associated with more severe retinal degeneration than others, such as INPP5E, MKS1, and the "morning glory disc anomaly" has been observed in an Austrian family with biallelic TMEM237 pathogenic variants (Huang et al., 2011).
Interestingly, in the UW series, a negative correlation was observed between TMEM67 pathogenic variants and retinal disease (OR 0.1, CI 0.01-0.8; p = .006), indicating that TMEM67-mutated patients are less likely to be diagnosed with the retinal disease (Bachmann-Gagescu, . Similarly, in the NIH cohort, no retinal degeneration was observed in patients with TMEM67 defects, as well as in those carrying CPLANE1 and KIAA0586 variants (Vilboux, Doherty, et al., 2017), and we also reported these negative associations in our cohort (Iannicelli et al., 2010;Romani et al., 2015;Roosing et al., 2015).
Pigmentary irregularities in the peripheral or mid-peripheral retina, or in the entire retina, which are early signs of retinal dysfunction, have been reported in 4.5% of JS patients. About 72% of them carried AHI1 (five patients) or CEP290 (three patients) molecular defects, but causative variants had also been detected in CPLANE1 and CC2D2A

| Kidney
Up to 25-30% of JS individuals develop renal disease mainly presenting as juvenile NPHP, which may remain asymptomatic for several years (Fleming et al., 2017;. It usually presents, in Of the 29 individuals with kidney disease in the NIH cohort (out of 97), 31% had NPHP, 35% presented an overlapping phenotype with PKD/NPHP, 10% had a unilateral multicystic dysplastic kidney, and 24% had indeterminate-type cystic kidney disease (Fleming et al., 2017;Vilboux, Doherty, et al., 2017). In this cohort, the gene most commonly observed to cause renal disease was CEP290,

| Liver
Approximately 10-15% of JS patients present with liver involvement, typically manifesting as congenital hepatic fibrosis (CHF) (Bachmann- As already discussed, hepatic fibrosis is often associated with ocular colobomas and sometimes with kidney disease as well. The distinctive syndromic combination of colobomas, cognitive impairment ("oligophrenia"), ataxia, cerebellar vermis hypoplasia, and hepatic fibrosis was previously referred to with the acronym "COACH" . Indeed, in the UW cohort, 50 out of 362 patients (14%) had liver fibrosis, which was strongly associated to coloboma (Bachmann-Gagescu, . Thus, the likelihood of having liver fibrosis in individuals with coloboma resulted in 6.5 times the likelihood of having liver fibrosis in individuals without coloboma. Our research group was the first to report a strong association between liver disease and the TMEM67 gene, detecting biallelic variants of TMEM67 in eight of 14 (57%) JS families with congenital liver fibrosis (Brancati et al., 2009). This significant association has been subsequently confirmed (Iannicelli et al., 2010). In the NIH cohort,
Other skeletal anomalies in JS are definitely rare. Among these, cited as examples, camptodactyly and bowing of long bones have been observed respectively in four and one patients out of 20 with TMEM216-related JS . Moreover, camptodactyly of digits III and V was also present in both hands of a child with MKS1related JS (Bader et al., 2016). Other skeletal anomalies were reported in single studies, such as tibial and fibular mesomelic dysplasia in one patient with B9D2-related JS (Bachmann-Gagescu,  or abnormal cone-shaped epiphyses of hands and feet (not associated with JATD) in one patient with biallelic variants in CELSR2 (Vilboux, Doherty, et al., 2017;Vilboux, Malicdan, et al., 2017).
Finally, club foot has been reported in a child with RIPGRIP1L-related JS, even though this complication may also be present in all severe cases presenting fetal hypokinesia (Brancati et al., 2008).
3.10 | Oral-facial-digital syndrome type 6 The term OFD syndromes describe a group of disorders mainly characterized by distinguishing facial features, oral abnormalities, and polydactyly. At least 18 clinical subtypes have been described, and the whole spectrum of findings tends to overlap with MKS, short-rib thoracic dystrophies, and JS (Bruel et al., 2017).
Among these syndromes, OFD6 (also known as Varadi-Papp syndrome) has been regarded as a rare phenotypic subtype of JS. The diagnosis of OFD6 requires the MTS as well as one or more of the following features: a distinctive preaxial or mesaxial polydactyly with Yshaped metacarps (but the less specific postaxial polydactyly has also been observed), syndactyly and/or bifid toe, a bifid or lobulated tongue due to soft-tissue nodules or multiple hamartomas, multiple oral frenulae, palate, and/or lip clefting, craniofacial features that include hypertelorism and upper lip notch, and hypothalamic hamartoma sometimes with absent pituitary gland (Bonnard et al., 2018;Poretti et al., 2012).
There has been some debate in the literature whether CPLANE1 may represent the most relevant gene associated to OFD6. A first survey identified pathogenic variants in the CPLANE1 gene in nine out of 11 fetuses with OFD6 features (82%), suggesting that this could indeed represent the major causative gene for OFD6 . Other groups subsequently confirmed that CPLANE1 is one of the causative genes for this condition (Bayram et al., 2015;Wentzensen et al., 2015;Wentzensen et al., 2016). However, we identified CPLANE1 pathogenic variants in only two of 17 living individuals with classical OFD6 phenotype, questioning whether CPLANE1 could be regarded as the main OFD6-causative gene . In two further studies, novel CPLANE1 recessive variants were reported in seven subjects with pure OFD6 (from five unrelated families) (Bonnard et al., 2018), as well as in four JS Chinese families with mild neurological and neuroradiological features . Taken together, these observations define a spectrum of phenotypes associated to CPLANE1 variants, ranging from an exclusively neurological phenotype to the full blown OFD6 presentation. In general, features of preaxial and/or mesaxial polydactyly and hypothalamic hamartoma seem more commonly related to CPLANE1 pathogenic variants, whereas tongue hamartomas and lingual frenula are less frequently associated with molecular defects in this gene Romani et al., 2015).

| Miscellaneous features
Although JS is not a typical dysmorphic syndrome, some facial dysmorphisms are often observed, such as prominent forehead, arched eyebrows, ptosis, trapezoid-shaped mouth with lower lip eversion, and prognathia, which tend to change with age (Braddock et al., 2007). Nevertheless, facial features do not support the clinical diagnosis, as in many patients they can be mild or absent.
Structural heart disease is not considered a typical JS feature, even though many ciliopathies are known to have an increased risk of cardiac malformations and their presence is of cardinal importance for the outcome. In the large UW cohort, only seven out 532 JS individuals had congenital heart defects (with a minimum prevalence of 1.3%) (Bachmann-Gagescu, . Thus, no cardiac screening is recommended for JS patients. Yet, although the incidence of heart defects in JS is small, these have been occasionally reported in patients harboring mutations in several genes, including some prevalent ones such as AHI1, CEP290, CPLANE1, KIAA0586, and MKS1 (Bader et al., 2016;Fraser & Davey, 2019;Parisi, 2009;Parisi et al., 2006;Zhang, Qu, et al., 2021). In particular, it is of interest the report of a few CEP290 mutated cases presenting with heart defects, mainly atrio-ventricular septal defect (Alharbi et al., 2018;Karp, Grosse-Wortmann, & Bowdin, 2012;Trevino et al., 2020).
Little is known about mortality in JS. In the UW cohort, 40 of 565 patients with JS were deceased, mainly due to extra-neurological involvement, such as kidney disease or liver fibrosis. The underlying genetic defects had been identified in 80% (32/40) of the deceased cohort. Although the numbers were too small for any statistical analyses, six of nine individuals with biallelic variants in CEP290 had died from complications of kidney disease, while two of the three with molecular defects in TMEM67 had died from liver fibrosis, and all three individuals with TCTN2-related JS died from respiratory complications.
Thus, a close monitoring of these issues should be considered in patients with these additional risk factors (Dempsey et al., 2017).   (Zaghloul & Katsanis, 2010). This oligogenic model, with two or more genes concurring to define the final phenotype, was subsequently proposed in other ciliopathies such as NPHP , and increased in complexity when not only pathogenic variants but also rare variants or even common polymorphisms in other ciliary genes were found to genetically interact with the "main" recessive variants, to modulate the expressivity of the ciliary phenotype. For instance, some common heterozygous variants in the RPGRIP1L and AHI1 genes were found to correlate to a higher occurrence of retinal degeneration in various ciliopathies and to an increased risk of NPHP1-deleted patients to develop a more severe neurological and ophthalmological phenotype, respectively (Khanna et al., 2009;Louie et al., 2010). Thus, while recessive pathogenic variants in a causative gene likely determine much of the phenotype, it was suggested that other variants (including common ones) in modifier genes may explain a substantial portion of the observed phenotypic variability. However, this was not confirmed by an independent, systematic study, questioning the validity of the original hypothesis (Phelps et al., 2018).

| Clinical and genetic overlap with other disorders
The establishment of meaningful genotype-phenotype correlates, which pertains not only to JS and other primary ciliopathies but to the majority of inherited diseases, represents the greatest challenge of genetic research for the years to come; deeper phenotyping of patients, analysis of large cohorts through multicenter collaborations and a better understanding of our genomic structure and variability at the individual level will represent essential assets to successfully accomplish this task.

CONFLICT OF INTEREST
The authors declare to have no competing financial interests.

Simone Gana, and Enza Maria Valente conceived the study; Simone
Gana and Valentina Serpieri wrote the manuscript; Enza Maria Valente revised the manuscript.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.