Truncating Variants Contribute to Hearing Loss and Severe Retinopathy in USH2A-Associated Retinitis Pigmentosa in Japanese Patients

Round 1

Reviewer 1 Report

The authors present a manuscript entitled: “Truncating variants contribute to hearing loss and 3 severe retinopathy in USH2A-associated Retinitis 4 Pigmentosa in Japanese patients.”

This work represents the study of a Japanese population with presumed inherited retinal disease that underwent genetic testing. They report the contribution of USH2A variants as 12.5% in this patient population. The also report an association between truncating mutations, which are presumably more deleterious, with syndromic patients (severe RP and hearing loss), whereas non-truncating mutations were associated with milder forms of RP without hearing loss.

They point out that some studies have correlated truncating mutations with syndromic disease, while other studies have disputed any such association. They report that this study in Japanese patients may help clarify the issue and shed light on race-specific differences in USH2A-disease and genotype-phenotype correlations. Overall the project is well designed, the data appear to be well handled and represented. The conclusions drawn in the discussion are supported by the data.

Significant Criticisms:

1. This type of retrospective study suffers from highly variable periods of follow up data between patients. One could include patients who had a minimal period of follow up (for example 3 years minimum). However, with rare diseases like these it is difficult to exclude patient data, and I understand to temptation to include all data from all patients, no matter how incomplete.
2. The target capture panel was variable between 39 and 50 genes for various patients. This represents a limitation of the study, both in the fewness of genes assayed and the depth in which each gene was surveyed.
3. There is no effort to ‘validate’ the degree of Japanese ancestry for each patient. In fact, the concept of nationality and race is simply a poor surrogate for genetics. The concept of race is quite limiting and has limited value as a scientific concept, especially in a genetic study such as this one.
4. What is the reason, even if speculation, that retinal cells (presumably photoreceptors) are have cilia that are more dependent on USH2A in comparison to hair cells of the ear that can tolerate missense mutations in USH2A? Are there compensatory relatives of USH2A that are expressed in hair cells that are not expressed in photoreceptors? What is known from human data? What is known from animal models of USH2A in various species? Please briefly discuss what is known from all species (I realize there are significant species differences) to help understand why photoreceptors are more dependent on USH2A in their cilia (less tolerant of mutations), while hair cells of the ear are more tolerant of mild mutations and therefore less dependent on USH2A in their cilia? Since the mutations in syndromic and non-syndromic USH2A-RP can occur all over the gene, it’s likely not domain-specific allelic changes that cause syndromic vs non-syndromic features. Instead, it is likely dosage of functional USH2A that determines photoreceptor involvement vs photoreceptor plus hair cell involvement. Why are hair cells less sensitive than photoreceptors to levels of functional USH2A?

Minor criticisms:

1. In Figure 4, please make this figure gray scale. There is no value to adding color. More importantly, for 4B, please use boxplots to more accurately represent the data.
2. In Figure 5, please add approximate Snellen equivalent visual acuity next to the logMAR visual acuities on the Y-axis for easier clinical interpretation.
3. Please add the units of age in Figure 5 on the X-axis, presumably ‘years.’
4. The whole manuscript needs professional editing by a native speaker of the English language. For example, the sentence starting on line 160 states, “Additionally, time-dependent changes in the same patients were attempted to understand.” This sentence needs editing. Similar examples are throughout the manuscript and need improvement.

Author Response

Significant Criticisms:

1. This type of retrospective study suffers from highly variable periods of follow up data between patients. One could include patients who had a minimal period of follow up (for example 3 years minimum). However, with rare diseases like these it is difficult to exclude patient data, and I understand to temptation to include all data from all patients, no matter how incomplete.

Reply:  We appreciate the comments.

2. The target capture panel was variable between 39 and 50 genes for various patients. This represents a limitation of the study, both in the fewness of genes assayed and the depth in which each gene was surveyed.

Reply:  In response to this comment, the current manuscript reports prevalence of USH2A-RP (6.9%) in our cohort. Figure 1a was replaced.

Full length of the USH2A gene was examined by 71 probe sets in both 39 and 50 panels. We confirmed the entire USH2A was analyzed and the average read-depth was ~900 (minimum ~350, maximum ~1300) for all 525 cases.

3. There is no effort to ‘validate’ the degree of Japanese ancestry for each patient. In fact, the concept of nationality and race is simply a poor surrogate for genetics. The concept of race is quite limiting and has limited value as a scientific concept, especially in a genetic study such as this one.

Reply:  We agree with the comments. We understand the concept of nationality and race is simply a poor surrogate, but we believe that this study could provide additional piece of evidence about differences between Japanese/East Asians, Caucasians and others. Our cohort includes only East Asians as a race as we described in the manuscript.

4. What is the reason, even if speculation, that retinal cells (presumably photoreceptors) are have cilia that are more dependent on USH2A in comparison to hair cells of the ear that can tolerate missense mutations in USH2A? Are there compensatory relatives of USH2A that are expressed in hair cells that are not expressed in photoreceptors? What is known from human data? What is known from animal models of USH2A in various species? Please briefly discuss what is known from all species (I realize there are significant species differences) to help understand why photoreceptors are more dependent on USH2A in their cilia (less tolerant of mutations), while hair cells of the ear are more tolerant of mild mutations and therefore less dependent on USH2A in their cilia? Since the mutations in syndromic and non-syndromic USH2A-RP can occur all over the gene, it’s likely not domain-specific allelic changes that cause syndromic vs non-syndromic features. Instead, it is likely dosage of functional USH2A that determines photoreceptor involvement vs photoreceptor plus hair cell involvement. Why are hair cells less sensitive than photoreceptors to levels of functional USH2A?

Reply:  We appreciate this comments and “Discussion” section of the current manuscript includes description as follows:

“It is not fully understood why USH2A variants lead to a wide range of phenotypes and severity of the diseases. Ush2a-knockout in mice and in zebrafish recapitulated a phenotype of human Usher syndrome with retinal degeneration and hearing problems [16,34]. These models could support our observation that all USH2A-RP patients with two truncating variants (n=6) developed RP and early onset hearing loss. Two protein isoforms, a long isoform and a short N-terminal isoform, are spliced from the USH2A gene [11]. Expression of the N-terminal isoform was only detected in the inner ear, in contrast that the long isoform was localized in photoreceptors and ears [16]. Interestingly, supplementation of a shortened form of Ush2a that lacks exon 12 rescued hearing loss in Ush2a-knockout mice [35]. Additionally, roles of a partner protein PDZD7 could be different in photoreceptors and inner ears [36]. Remarkably, PDZD7 variants are only responsible for congenital hearing loss, but not for RP [36]. These facts and accumulation of additional evidence could contribute to better understanding pathogenesis of USH2A-associated diseases.”

Minor criticisms:

1. In Figure 4, please make this figure gray scale. There is no value to adding color. More importantly, for 4B, please use boxplots to more accurately represent the data.

Reply:  Figure 4 was corrected as suggested.

2. In Figure 5, please add approximate Snellen equivalent visual acuity next to the logMAR visual acuities on the Y-axis for easier clinical interpretation.

Reply:  Figure 5 was updated as suggested.

3. Please add the units of age in Figure 5 on the X-axis, presumably ‘years.’

Reply:  The X-axis of the current Figure 5 is now labeled as “Age (years)”.

4. The whole manuscript needs professional editing by a native speaker of the English language. For example, the sentence starting on line 160 states, “Additionally, time-dependent changes in the same patients were attempted to understand.” This sentence needs editing. Similar examples are throughout the manuscript and need improvement.

Reply:  The entire manuscript was checked by a native speaker of the English language.

Reviewer 2 Report

General remark: be careful when using a percentage if the number analyzed is less than 100. It would be better to indicate the proportions or to indicate for example 11/36 versus 25/36.

Table 1: it would be better to indicate patients’ genotype with nucleotide change and protein change.

For example, for P1 patient, you can write: [c.490G>T; c.13633_13634del] in the nucleotide change column and [p.(Val164Phe)];[p.Pro4545Serfs*16)] in the protein change column. This additional column of nucleotide change provides information about the nature of the variation.

The authors write for example Pro4545Ser fs*16. There is no space. Check for all the frameshift variants in the different table and in the text.

The authors describe for non-syndromic USH2A-RP cases, four frameshift variants and one nonsense variant. I have check three nonsense variants (p.(Tyr3776*), p.(Trp4922*) and p.(Trp2133*)) and two frameshift variants (p.(Gly2799Valfs*31) and p.(Tyr3938Argfs*8)).

For the P34 patient, it would be interesting to check if mutations are on the same allele. P.Cys870Phe and p.Ile4102Asn could be only responsible for the disease.

For the patient P36, the splice variation c.4758+3A>G is predicted benign in ClinVar and no notable effect is predicted for this variant by using NNsplice, Splice site prediction, etc…

Page 6 line 126-127: there is an error in the sentence. Syndromic RP patients were aware of symptoms significantly earlier than non-syndromic patients.

It is very difficult to conclude when you compare only one patient versus five.

Page 9 line 181: Add mean to the sentence. The mean age that our patients….

Page 9 line 225-226: To my knowledge, mutations in USH2A were responsible for Usher syndrome type 2 and for isolated RP, but not for isolated deafness.

Author Response

General remark: be careful when using a percentage if the number analyzed is less than 100. It would be better to indicate the proportions or to indicate for example 11/36 versus 25/36.

Reply:  Sample numbers are now clearly indicated in the manuscript to avoid confusion. 

1. Table 1: it would be better to indicate patients’ genotype with nucleotide change and protein change.

For example, for P1 patient, you can write: [c.490G>T; c.13633_13634del] in the nucleotide change column and [p.(Val164Phe)];[p.Pro4545Serfs*16)] in the protein change column. This additional column of nucleotide change provides information about the nature of the variation.

Reply:  Table 1 in the current manuscript includes both nucleotide and protein change columns.

2. The authors write for example Pro4545Ser fs*16. There is no space. Check for all the frameshift variants in the different table and in the text.

Reply:  All the frameshift variants were checked and corrected including update of Table 4.

3. The authors describe for non-syndromic USH2A-RP cases, four frameshift variants and one nonsense variant. I have check three nonsense variants (p.(Tyr3776*), p.(Trp4922*) and p.(Trp2133*)) and two frameshift variants (p.(Gly2799Valfs*31) and p.(Tyr3938Argfs*8)).

For the P34 patient, it would be interesting to check if mutations are on the same allele. P.Cys870Phe and p.Ile4102Asn could be only responsible for the disease.

Reply:  We fixed Figure 2 and all incorrect descriptions in our manuscript.

For the P34, we agree with the comments. We discussed a need for segregation analysis with the patient for further evaluation as described in the footnote of Table 1. Obtained data will be uploaded to the public data bank. We added an explanation to Table 1 footnote as follows: “Pathogenicity of each variant needs further evaluation in the patients with more than three variants detected.”

4. For the patient P36, the splice variation c.4758+3A>G is predicted benign in ClinVar and no notable effect is predicted for this variant by using NNsplice, Splice site prediction, etc…

Reply:  We confirmed that c.4758+3A>G is benign/likely benign in the latest ClinVar, and therefore this variant was removed from the manuscript. Table 3 and Figure 3 were updated.

5. Page 6 line 126-127: there is an error in the sentence. Syndromic RP patients were aware of symptoms significantly earlier than non-syndromic patients.

Reply:  This sentence was corrected as follows: “Comparing the onset age of RP in 11 syndromic and 25 non-syndromic USH2A-RP patients, syndromic RP patients were aware of symptoms significantly earlier than non-syndromic patients.”

6. It is very difficult to conclude when you compare only one patient versus five.

Reply:  We fully agree with the comment. We do not intend that the data from these 6 cases are enough for drawing a conclusion. Hopefully these data can be utilized in the future such as a source for meta-analyses.

7. Page 9 line 181: Add mean to the sentence. The mean age that our patients….

Reply:  This sentence was updated as follows: “The mean age that our patients noticed symptoms, such as night blindness and visual constriction, ~”

8. Page 9 line 225-226: To my knowledge, mutations in USH2Awere responsible for Usher syndrome type 2 and for isolated RP, but not for isolated deafness.

Reply:  We rechecked the reference and confirmed that mutations in USH2A are also responsible for isolated deafness. We hope this manuscript could provide useful information.