Usher syndrome: clinical features, molecular genetics and advancing therapeutics

Usher syndrome has three subtypes, each being clinically and genetically heterogeneous characterised by sensorineural hearing loss and retinitis pigmentosa (RP), with or without vestibular dysfunction. It is the most common cause of deaf–blindness worldwide with a prevalence of between 4 and 17 in 100 000. To date, 10 causative genes have been identified for Usher syndrome, with MYO7A accounting for >50% of type 1 and USH2A contributing to approximately 80% of type 2 Usher syndrome. Variants in these genes can also cause non-syndromic RP and deafness. Genotype–phenotype correlations have been described for several of the Usher genes. Hearing loss is managed with hearing aids and cochlear implants, which has made a significant improvement in quality of life for patients. While there is currently no available approved treatment for the RP, various therapeutic strategies are in development or in clinical trials for Usher syndrome, including gene replacement, gene editing, antisense oligonucleotides and small molecule drugs.


Introduction
Usher syndrome encompasses a group of inher ited disorders characterised by dual sensory impairment of the auditory and visual systems, with a variable presentation of vestibular dysfunc tion in a proportion of cases. It is the most com mon cause of combined sight and hearing loss, accounting for more than half of deaf-blindness cases. 1,2 It has an estimated prevalence of between 4 and 17 in 100 000 people worldwide. 2,3 Further more, it has been estimated to represent 5% of all congenital deafness and 18% of all retinitis pig mentosa (RP) cases. 2,4 Usher syndrome is both clinically and genetically heterogeneous and is divided into three distinct clinical subtypes, asso ciated with a number of genetic loci. The Usher genes encode a variety of proteins that are expressed in the inner ear and retina where they perform essential functions in sensory hair cell development and function, and photoreceptor maintenance. While many promising treatments are under investigation, there is no approved treatment for this disease to date.

Clinical characteristics
Usher syndrome involves a combination of bilat eral sensorineural hearing loss with progressive retinal degeneration in the form of RP. It is cat egorised into three major clinical subtypes according to severity and onset of hearing loss and whether vestibular dysfunction is present 5 (Table 1). However, there is clinical variability within each Usher subtype, with overlapping and atypical presentations described.
Usher syndrome type 1 (Usher 1) is the most severe subtype in which patients exhibit severe to pro found bilateral congenital sensorineural hearing loss (Figure 1), most frequently nonprogressive, with vestibular areflexia. It accounts for approxi mately 25-44% of all Usher syndrome cases. 11 Infants are detected through the newborn hearing screen (Figure 2), and where not undertaken/ available, the diagnosis is often suspected in infancy. Vestibular areflexia is reflected in delayed motor development and children usually do not walk independently before the age of 18 months. When older, they compensate for their vestibular areflexia using vision, until the onset of RP, although they often have higher accidental falls and difficulty in performing activities, which require balance, for example, riding a bicycle. Due to the profound nature of deafness, children with Usher 1 derive limited or no benefit from hearing aids and most patients with Usher 1 would be sign language users if the hearing loss is not treated effectively. Timely use of cochlear  13,14 The standard approach is to offer bilat eral cochlear implants to Usher 1 patients within the first 2 years of life.
Usher syndrome type 2 (Usher 2) is the most common form of the disorder, representing over half of all cases. 11 The sensorineural hearing loss is typically described as sloping, mild to moder ate in the low frequencies and severe to profound in the high frequencies ( Figure 1). 15 Hearing loss is congenital and infants are detected through the newborn hearing screen, however, if unavailable detection can be overlooked till the end of the first decade of life due to the high frequency configuration and degree of hearing loss. 6 Although thought to be nonprogressive, there is evidence to indicate progression of hearing loss over the years, particularly in Usher type 2A. 16 In all types of Usher syndrome, audiological findings are detected/present prior to ophthalmo logical signs and symptoms. Children and adults with Usher syndrome do not have dysmorphism, and the disorder is commonly mistaken for a nonsyndromic isolated sensorineural hearing loss. They will often have the initial aetiologic investigations for hearing loss, which include cytomegalovirus (CMV) testing and magnetic resonance imaging (MRI) scan, genetic testing for GJB2 and mitochondrial m.1555A > G muta tions and electrocardiography (ECG) in some patients, which is expected to be normal. 12,24 Initial investigations for bilateral sensorineural hearing loss also include ophthalmological assess ment, which in the early years can be normal depending on the Usher subtype. Diagnosis of Usher 1 should be suspected in any infant with bilateral profound sensorineural hearing loss and delayed motor milestones, even if the initial oph thalmology screen is normal. Usher 2 should be considered in patients with typical sloping con figuration of hearing loss. Diagnosis of Usher 2 or 3 is made after visual symptoms or signs are detected through routine examination or electro retinography (ERG). A metaanalysis of next generation sequencing (NGS) data in the United States indicates that 7.5% of patients with seem ingly 'isolated deafness' have mutations in the Usher genes and may be at high risk of develop ing RP. 25 With the availability of NGS, Usher genes are included in the 'deafness' and 'retinal' gene panels to help with early genetic diagnosis.
Audiological rehabilitation in all forms of Usher syndrome is started soon after diagnosis by fitting of bilateral hearing aids. Hearing aid fitting in patients with Usher syndrome may need special considerations, especially with visual impair ment. 19 The onset of RP has a significant impact on communication, as patients will have difficulty in lip reading and understanding gesture and sign, and on balance as visual compensation is compro mised. Patients with Usher syndrome and bilat eral vestibular areflexia should be advised about the risk of disorientation and potential drowning with underwater swimming, due to poor availabil ity of visual and proprioceptive inputs. 26,27 RP develops in all three Usher subtypes but with variable onset; Usher 1 is most commonly preadolescent, with Usher 2 within the first two decades of life, and Usher 3 patients typically being postpubertal. 8,28 Visual prognosis also dif fers between the clinical types; Usher 1 patients generally show a more severe visual decline with age, reaching legal blindness on average 15 years earlier than patients with Usher 2. 6,9 Typically, the first presenting symptom is night blindness (nyctalopia) with progressive visual field loss beginning in the midperiphery caused by rod photoreceptor degeneration. It eventually pro gresses to involve cone photoreceptors, resulting in central and colour vision loss. Patients are often registered severely sight impaired but there can be significant intra and interfamilial pheno typic variability. Fundus features include variable amounts of bone spicule pigmentation, retinal pigment epithelium (RPE) atrophy or depigmen tation, retinal arteriolar attenuation and optic disc pallor (Figures 3(a) and 4(a)). A significant pro portion of Usher patients may also develop cata racts and/or cystoid macular oedema. 29 Retinal imaging using fundus autofluorescence shows a ring of hyperautofluorescence in the mac ula (Figures 3(b) and 4(b)), and spectral domain optical coherence tomography (SDOCT) reveals loss of the outer retinal structure (Figures 3(c) and 4(c)), sparing the fovea until late in disease ± cys toid macular oedema. A prospective study using optical coherence tomography angiography (OCTA) of patients with MYO7A and USH2A mutations showed reduced vessel density in the retinal circulation with changes in the superficial capillary plexus (SCP) and deep capillary plexus (DCP) in all patients compared with healthy age matched controls. 30 However, peripheral defects were detected in the choriocapillaris (CC) earlier and more severely in MYO7A patients compared with the USH2A group. It was suggested that this is because the MYO7A protein (myosin VIIa) is mainly expressed in the RPE, thus affecting the CC directly. In the same study, patients were tested for macular sensitivity (MS) using microp erimetry with the Macular Integrity Assessment, and this detected decreased mesopic mean MS in all patients, especially in the periphery. Patients with MYO7A mutations had a slightly lower mean MS than patients with USH2A mutations; how ever, the differences were found not to be statisti cally significant (p = 0.66). Static and dynamic perimetry detects midperipheral visual field loss with progression to residual small central islands, with a small temporal peripheral field preservation in the most advanced stages of the disease. ERG  measurement can show reduction and delay of amplitudes in the early stages of the disease, and is a useful test to perform in an infant born with pro found deafness to determine the likelihood of underlying Usher 1, even before visual dysfunction is otherwise noted. Later, the full field ERG is often nonrecordable.
Dual sensory clinics are now being established to improve the clinical pathways and experience of children with hearing and sight impairment. These clinics will provide access to the relevant multidisciplinary in one visit, hence reducing stress and the burden associated with numerous, separate medical appointments. Children with Usher syndrome have been reported to develop mental and behavioural disorders, including autism, conduct disorder, schizophrenia and learning difficulty. This could be multifactorial, due to sensory deprivation, stress, difficulty in diagnosis and a possible unproven genetic asso ciation. 31,32 Dual sensory clinics will promote faster accurate diagnosis through more extensive genetic testing and detection of visual symptoms and mental health issues at an earlier stage.

Genetics of Usher syndrome
Usher genes All Usher syndrome types are inherited in an auto somal recessive manner. To date, at least 10 causa tive genes have been identified for the disease, which include six Usher 1 genes, three Usher 2 genes and one Usher 3 gene. Historically, traditional Sanger sequencing of all Usher gene exons was found to provide a genetic diagnosis for more than 80% of Usher families, 33,34 but this is timeconsuming and costly, particularly for large patient cohorts. Microarraybased testing provided ~33% detection for Usher patients but can only screen for known mutations. 35,36 NGS, including whole exome and genome sequencing, is now the method of choice with high efficiency offering the advantage of detect ing a range of known and novel mutations, includ ing large genomic DNA rearrangements. Targeted exome gene panel testing can reach diagnostic rates of around 70-80% in Usher families. 29,[37][38][39] Currently, there are nine loci (USH1BJ) known to be involved in Usher 1. The genes identified for six of these loci are as follows: 45 and CIB2 (USH1J). 46 Of these genes, MYO7A is the most frequent cause of Usher 1, accounting for more than half of cases. 34 The USH1E, USH1H and USH1K loci have been mapped to chromo somes 21q21, 15q22-23 and 10p11.21-q21.1, respectively, 47-49 but the genes are yet to be identi fied. It is worth noting that CIB2 biallelic loss of function variants has been reported in patients with nonsyndromic recessive hearing loss (DFNB48) but with no retinal symptoms. 50 Three genes underlying Usher 2 have been identi fied as USH2A (USH2A), 51 ADGRV1 (USH2C) 52 and WHRN (USH2D). 53 65,66 Myosin VIIA is also an essen tial RPE protein, [67][68][69] and evidence suggests that clarin1 is restricted to the retinal Müller glia. 70 Various studies have indicated the involvement of Usher proteins in a range of processes, including cohesion, mechanotransduction, synaptic matu ration, and protein and organelle transport. USH2A has a diverse mutation spectrum, which includes nonsense, frameshift, missense and splice affecting mutations, as well as deletions and dupli cations. The most common mutation found in USH2A is a single base pair (bp) deletion in exon 13, c.2299delG p.(Glu767Serfs*21), 34,92 which has been shown to be associated with exon splicing. 93 This variant is predicted to produce a severely trun cated protein and/or be subject to nonsensemedi ated decay; however, transcript analysis showed that it caused skipping of exon 13 or exons 12 and 13. 93 Mutations of USH2A are associated with up to 23% of nonsyndromic RP cases, 94 and specific mutant alleles are more frequent among such patients and families, the most common being missense variant c.2276G > T p. (Cys759Phe). 95 Several USH2A phenotypegenotype studies have been carried out to date. A survey conducted by Lenassi and colleagues of patients with USH2A associated RP reported several 'retinal disease specific' alleles that were rarely found in Usher 2 families, mostly missense variants that were likely to be less deleterious, while the Usherassociated variants mostly included those that were pre dicted to produce no viable protein (e.g. those causing premature truncation). 96 They proposed an allelic hierarchy model in which the presence of at least one retinal diseasespecific allele in a patient with USH2Arelated retinopathy resulted in the preservation of hearing. While this has not been supported by subsequent studies, 6,98,99 the same analysis on a different cohort 100 combined with two large external cohorts 98,101 found that the allelic hierarchy model was valid in 86% of individuals with nonsyndromic USH2ARP. 100 In addition, it has been reported that Usher 2 patients with one copy of the p.(Cys759Phe) allele showed a later onset of RP and milder hear ing loss compared with the general Usher 2 popu lation, 6 and the presence of the p.(Cys759Phe) variant in a homozygous state or in combination with other USH2A missense mutations has been associated with isolated RP or RP with late onset hearing loss. 99 In contrast, the p.(Glu767Serfs*21) variant results in a more rapid deterioration and severe hearing threshold, heralding the need for careful audiological monitoring and considera tion of cochlear implants. 100 In general, severe hearing impairment has been associated with truncating variants in USH2A. 6 112,113 although it harbours the risk of insertional mutagenesis. Among the other Usher models, gene delivery using AAV vectors has also produced significant improvements in auditory and vestibular hair cell function in mouse models of USH1C, 114 USH1G, 124 USH2D, 132 and USH3. [135][136][137] This was achieved by viral injection into the inner ear through the round window membrane 114,124,[135][136][137] or poste rior semicircular canal 132 in postnatal neonatal mice.

Genotype-phenotype correlations
The mouse inner ear is immature at birth and continues to mature postnatally. The acquisition of hearing (measured by the onset of startle response) occurs 12 days postnatally in mice, pro viding a window of opportunity for effective inter vention with gene therapy. 145 Comparatively, hearing in humans is fully mature at birth (onset of startle response at 19 weeks gestation). The hearing loss in Usher 1 is established at birth and it is not clear whether the hair cells in the human inner ear are a viable therapeutic target. In order to be effective, human intervention should be considered within the foetal stage, before estab lishment of hearing (at ~18 weeks of gestation). 145 Hence, where therapeutic response in mice is suc cessful when given soon after birth, it is question able whether the same effect will be seen in patients with postnatal treatment. Overall, further studies in nonhuman primates will be useful in addressing some of these issues. 66 One alternative approach to gene replacement is gene editing, which involves cutting around genetic mutations through the use of nuclease enzymes and correcting the DNA error by homol ogous recombination with a DNA template con taining the wildtype sequence. 146 This can be used to correct point mutations, small indels and splice site mutations, and is suitable for any gene size. Early investigations into the use of this strat egy for Usherdirected treatment employed the use of two zinc finger nucleases to correct an USH1C point mutation (c.91C > T p.[Arg31*]) and induce fulllength harmonin expression in cultured cells. 115 In recent years, the CRISPR/ Cas9 system has become highly popular for gene editing due to its efficiency and ease of use. This technique has been used for successful in vitro mutation repair in USH2A patient fibroblasts

USH1C
Gene replacement  126 However, the efficiency of mutation correction was only 2.5% in fibro blasts 125 and up to 3% in iPSCs, 126 although the second study reported an 80% editing efficiency in the small number of homozygous USH2A clones that survived. Encouragingly, neither study reported offtarget effects, which are unwanted mutations induced at DNA locations that show homology to the guide sequence. CRISPR/Cas9based editing shows huge prom ise for the treatment of IRDs caused by a range of mutations; however, ensuring the absence of offtarget effects and a high level of editing effi ciency in retinal cells will be essential for future investigations.

Round window membrane injection of AAV vectors
Small moleculebased methods for treatment of Usher syndrome have included the use of transla tional readthroughinducing drugs (TRIDs), which bind to the translational machinery and are able to induce insertion of an amino acid at the site of premature stop codons, allowing readthrough of nonsense mutations. These small molecule drugs include ataluren (PTC124) and designer aminoglycosides (NB compounds such as NB54). Several TRIDs have been used to suppress Usher 1associated PCDH15 and USH1C nonsense mutations in vitro, in cell cultures and in retinal exp lants. 116,117,[121][122][123] Furthermore, in vivo administra tion of NB54 and PTC124 was able to restore expression of fulllength harmonin in mouse reti nas transfected with Ush1c reporter constructs. 117 For Usher 2 investigation, PTC124 was adminis tered to a human embryonic kidney (HEK) cell model expressing a cDNA fragment of USH2A containing the c.11864G > A p.(Trp3955*)

Round window membrane injection or posterior semicircular canal injection of AAV vectors
Whirler mice Isgrig and colleagues, 132 Yasuda and colleagues 133 Usher 3

CLRN1
Gene replacement

Subretinal or intravitreal injection of AAV vectors
Wild-type mice Dinculescu and colleagues 134

Round window membrane injection of AAV vectors
Clrn -/-(KO-TgAC1) Previous studies [135][136][137] Round window membrane injection of AAV vectors AAV, adeno-associated virus; ASO, antisense oligonucleotides; BMSC, bone marrow-derived stem cells; CNTF, ciliary neurotrophic factor; HDR, homology-directed repair; HEK, human embryonic kidney; iPSC, induced pluripotent stem cell.  127 Further studies demonstrated PTC124 efficacy in restoring USH2A protein expression and primary ciliogenesis capability in USH2A patientderived fibroblasts with the c.9424G > T p.(Gly3142*) mutation. 128 Overall, TRIDs show promise as a safe and effective strategy to treat a range of Usher related nonsense mutations; however, these par ticular variants cause ~16% of USH2Arelated RP and 13% of all IRD cases. 128 An additional small molecule that has been of interest for Usher syndrome treatment, known as BioFocus 844 (BF844), was identified through cellbased high throughput screening as capable for stabilising the defective Clarin1 protein pro duced by the common CLRN1 missense variant p.(Asn48Lys). 138 BF844 was shown to protect against progressive hearing loss when adminis tered intraperitoneally to an Usher 3 knockin mouse model.
A further therapeutic option is the use of ASOs, which are short synthetic modified nucleic acids that bind RNA through complementary base pairing. They can be designed to bind premRNA at splice enhancer or silencer target sites, prevent ing or stimulating binding of the spliceosome thereby modulating premRNA splicing. ASOs have been used to rescue both the hearing and vestibular defects in Ush1c knockin mice, which possess a cryptic splice site mutation that results in truncated harmonin protein. [118][119][120] Initially, peritoneal injection of ASOs in neonatal mice was shown to partially correct defective premRNA splicing of mutant Ush1c transcripts; the same group delivered ASOs to Ush1c knockin foetal mice in utero via transuterine injection into the amniotic cavity and observed partial correction of vestibular function and hearing in the mice post natally, 119 while most recently transuterine injec tion directly into the developing inner ear produced more substantial improvements in both hearing and vestibular function that sustained into adulthood. 120 ASOs have also been used to correct a splicing defect caused by a deep intronic mutation in the USH2A gene (c.7595 2144A > G) which leads to insertion of a pseu doexon (PE40), in both patientderived fibroblasts and a minigene splice assay. 129 Overall, there are a number of promising thera peutic strategies in the development for the Usher subtypes. The use of patientderived retinal organoids, which have already been generated for Usher syndrome caused by USH2A muta tions, 147,148 will further aid in the testing of novel treatments by providing the opportunity to dem onstrate therapeutic potential in retinalspecific cells in vitro.

Identifying outcomes
Although several treatment strategies are already under development for Usher syndrome and other IRDs, identifying metrics that display detectable changes within relatively short time periods (e.g. 1-2 years) in otherwise slowly pro gressive conditions will aid the assessment of therapeutic efficacy in clinical trials. This is espe cially important when the treatments are admin istered systemically, such as orally, and the untreated eye cannot be used as a control. , in addition to the use of artificial intelligencebased methods will further aid in identifying suit able outcome metrics for clinical trials for Usher syndrome. This will likely be tailored to the mechanism (gain of function or slowing of disease progression) and target of treatment, that is, ret inawide or central retina. If gain of function is anticipated, trials could be relatively short (between 12 and 18 months) to arrive at an esti mate of potential longevity.

Clinical trials
Owing to the success of preclinical investigations, there are several completed and ongoing clinical trials for patients with Usherrelated RP. For Usherspecific gene therapy, the first clinical trial evaluated subretinal injection of a recombinant equine infectious anaemia virus (EIAV)-based lentiviral vector for delivery of MYO7A cDNA (UshStat) for treating patients with MYO7A related Usher 1 (NCT01505062). 113 However, this phase I/IIA trial has been terminated by the sponsor Sanofi due to review of clinical develop ment plans and priorities. A second trial is ongoing to assess longterm safety of patients who received UshStat (NCT02065011). A further clinical trial is being prepared using dual hybrid AAV vectors to deliver MYO7A to the retina of USH1B patients (https://cordis.europa.eu/project/id/754848/ it). 109,153 Considering the Food and Drug Administration and European Medicines Agency approval of Spark Therapeutics Luxturna gene therapy for patients with RPE65related retinal dis ease, gene replacement therapy has become a more likely future option for the treatment of several Usher subtypes. However, such therapies are likely to be highly costly, and conventional viral methods are not appropriate for very large genes like USH2A (cDNA length >15 kb).
For ASObased treatments, there is currently a trial sponsored by ProQR for an ASO candidate (QR421a), which has been designed to exclude the whole exon 13 in the USH2A mature mRNA transcript; this has been shown preclinically to result in restoration of functional usherin pro tein. 130 Considering that two of the most common pathogenic USH2A mutations occur in exon 13, 34,92,96 if successful this treatment would be suitable for a large proportion of patients with USH2Arelated Usher 2 and RP. The phase I/II clinical trial is currently ongoing for intravitreal injection of QR421a in patients with USH2A exon 13 variants (NCT03780257).
In addition to Usher genespecific clinical trials, subretinal implantation of capsules containing human NT501 cells that release ciliary neuro trophic factor (CNTF) has been trialled in patients with choroideremia and RP, including some with Usher 2 and Usher 3 (NCT00447980, NCT01530659). 140,141 CNTF has been found to prolong photoreceptor survival in mouse and rat models of retinal degeneration. 142,143 Viral deliv ery of rodderived cone viability factor (RdCVF) is also under investigation for the treatment of RP; RdCVF is a factor naturally secreted by rods to protect cone photoreceptors, 154 and has been found to promote photoreceptor survival in mouse models of RP after viralmediated expres sion in the retina. 155 If such strategies are effective in humans, this could be suitable for the signifi cant number of IRD patients without a confirmed molecular diagnosis.
When choosing the method of treatment for Usher syndrome and other IRDs, the stage of dis ease will be an important consideration. The strategies already described are likely to be only effective at a stage where retinal photoreceptors are still intact. At the advanced stages of retinal degeneration, cell replacement therapies 156 or retinal prosthesis 157 may be the most feasible options. Advances in embryonic stem cell and iPSC technology make cell transplantation an everlikely option for patients with latestage reti nal disease. 156 Bone marrow-derived stem cells (BMSC) have been trialled in five ungenotyped patients with varying subtypes of Usher syndrome as part of the Stem Cell Ophthalmology Treatment Study (SCOTS; NCT01920867 and NCT03011541). 144 Each Usher patient received autologous BMSC through either retrobulbar, subtenons, intravitreal, subretinal or intraoptic nerve injections into both eyes, followed by intra venous injections. The average pretreatment logarithm of the minimum angle of resolution (LogMAR) acuity was 0.635, and the average postoperative change was a gain of 0.18 LogMAR. In the murine retina, it has been found that endogenous BMSC migrate and fuse with Müller glia cells after damage has been inflicted. 158 The journals.sagepub.com/home/oed 13 resulting hybrids were found to contribute to the replacement of damaged neurons, demonstrating the regenerative potential of BMSC in the mam malian retina.
Although there are no ongoing clinical trials for the Usherspecific hearing loss, there was a clini cal trial for a recombinant adenovirus 5 (Ad5) vector containing the human atonal transcription factor (ATOH1) cDNA for administration via intralabyrinthine infusion in patients with severe to profound sensorineural hearing loss (NCT02132130). The results are yet to be pub lished. The successful preclinical work with sev eral Usher mouse mutants makes gene therapy a promising future option; however, as discussed, these studies have involved treatment administra tion in prenatal or neonatal mice when the inner ear is still developing, and the use of similar thera pies in children or adults with Usher syndrome may not be able to achieve reversal of the con genital inner ear defects.

Conclusion
Usher syndrome is a disorder with vast clinical and genetic heterogeneity, typically resulting in significant dual sensory loss causing great impact on patient quality of life. More than ever, the prospect of an available treatment for at least some Usher subtypes looks promising. However, there are still obstacles to overcome in developing safe treatments that work for each gene size and mutation. In addition to the many gene and mutationspecific treatments under investigation, finding universal treatments that use common mechanisms for the treatment of RP should be a priority for the many patients that remain without a molecular diagnosis. Further patient analysis is necessary to determine better genotype-pheno type correlations for each clinical subtype to pre dict prognosis; this will inform genetic counselling, preimplantation diagnosis and the choice of best outcomes for each treatment trial.