The impact of modifier genes on cone-rod dystrophy heterogeneity: An explorative familial pilot study and a hypothesis on neurotransmission impairment

Abstract

Cone-rod dystrophies (CORDs) are a heterogeneous group of inherited retinopathies (IRDs) with more than 30 already known disease-causing genes. Uncertain phenotypes and extended range of intra- and interfamilial heterogenicity make still difficult to determine a precise genotype-phenotype correlation. Here, we used a next-generation sequencing approach to study a Sicilian family with a suspected form of CORD. Affected family members underwent ophthalmological examinations and a proband, blind from 50 years, underwent whole genome and exome sequencing. Variant analysis was enriched by pathway analysis and relevant variants were, then, investigated in other family members and in 100 healthy controls from Messina. CORD diagnosis with an intricate pattern of symptoms was confirmed by ophthalmological examinations. A total of about 50,000 variants were identified in both proband's genome and exome. All affected family members presented specific genotypes mainly determined by mutated GUCY2D gene, and different phenotypical traits, mainly related to focus and color perception. Thus, we looked for possible modifier genes. According to relationship with GUCY2D, predicted functional effects, eye localization, and ocular disease affinity, only 9 variants, carried by 6 genes (CACNG8, PAX2, RXRG, CCDC175, PDE4DIP and LTF), survived the filtering. These genes encode key proteins involved in cone development and survival, and retina neurotransmission. Among analyzed variants, CACNG8c.*6819A>T and the new CCDC175 c.76C>T showed extremely low frequency in the control group, suggesting a key role on disease phenotypes. Such discovery could enforce the role of modifier genes into CORD onset/progression, contributing to improve diagnostic test towards a better personalized medicine.

Fig 1. Pedigree of the Sicilian family affected by an undefined form of CORD.

The affected (black fill) and unaffected (no fill) members are shown. Arrow: proband; circle: female; square: male; double line: blood relatives; question mark: uncertain phenotype.

Fig 2. Proband’s fundus, optical coherence tomography (OCT) and electroretinogram (ERG).

Fundus photo of left eye is shown upside down. Fundus examination demonstrated mild changes typical of RP in the left (a) and the right (e) eyes, including slightly pale optic discs with almost normal retinal vessels, generalized pigmentary granularity, and moderate bone-like spicules in peripheral areas. The OCT scans of the left eye (b) and the right eye (f) revealed decreased macular thickness, as well as a severe loss of EZ line and outer segments. Scotopic (c, g) and photopic (d, h) ERGs revealed a generalized cone dysfunction with rod involvement. ONL = Outer Nuclear Layer. IS-OS Junction = Inner Segment-Outer Segment Junction. RPE Complex = Retinal Pigment Epithelium Complex.

Fig 3. Graphical workflow of data analyses.

Data analyses involved several filtering levels, to obtain final candidate variants.

Fig 4. Sanger validation of candidate GUCY2D causative variants.

Both GUCY2D identified variants (c.2513G>C on the left, and c.3044-7G>T on the right) were validated by Sanger sequencing in all family members. Black text = Healthy family member; Red text = Affected family member. Orange text = Uncertain phenotype in this family member.

Fig 5. Alternative splicing predictive analysis for GUCY2D c.3044-7G>T.

The red box highlights the differences between codon usage of GUCY2D wild-type (at the top) versus GUCY2D carrying the splice variant c.3044-7G>T (at the bottom), suggesting the possible alternative splicing event of mutated gene. More details through the text.

Fig 6. Functional groups in ClueGO overview.

The pie charts present functional groups for analyzed genes, based on GO, KEGG, Reactome and WikiPathways annotation by ClueGO Cytoscape plug-in, then clustered into STRING action enriched (a), combined score (b), database score (c), and textmining (d) categories. The name of the group is given by the group leading term (the most significant term in the group). The group sections correlate with the number of the terms included in the group. Network structure of functional groups is calculated on kappa Score, which shows the relationships between the terms based on their overlapping genes.

Fig 7. Sanger validation of final modifier gene candidate variants.

All nine final selected variants carried by candidate modifier genes were validated by Sanger sequencing, and electropherograms are reported in figure. a–h, variants validated in proband; i validated in one proband’s son.

Fig 8. Symptom patterns and variants.

a) By considering an additive model, for each symptom we calculated difference between average genotype for affected vs unaffected patients. This value was then assigned a score according to effect size (mild = 0.5, promising> = 1). Score is shown on a heatmap that reflects the strength (mild or strong) and putative causative (reds), or protective (blues) role of examined variants. White rows arise when all patients show a given symptom, thus making comparison of genotype distribution impossible. b) Stem plot shows total score obtained for a given variant when considering absolute value of effect sizes gathered across all symptoms; this measure represents the overall relevance of that variant in explaining symptom patterns. Lines highlight median (black), more one (blue) and two (red) standard deviations with respect to score distribution.

Fig 9. Proposed regulatory pattern related to CORD pathogenic mechanisms.

Our hypothesis is based on primary cone compromised activity, originated from GUCY2D defects, might worsened by altered retrograde signaling from the inner retinal cells due to variants carried by candidate modifier genes. CACNG8 c.*6819A>T could decrease the expression of TARP γ-8 physiologically interacting with DACs and RGCs AMPARs, reducing the retrograde and anterograde signal transmission determining, respectively, an impairment of optic nerve activity and co death. Additionally, the presence of the RXRG c.875+6A>G and PAX2 c.-375C>A may have already compromised cones and their differentiation, probably involving astrocytes and other neural supporting cells, especially in RPE. This scenario could also promote rods death, due to the lack of the fundamental anti-inflammatory activity of lactotransferrin (failed because of LTF c.85G>A and c.140A>G and, probably, of CCDC175 c.-79C>G and c.76 C>T), and due to the reduced expression of PDE4DIP (caused by the two 3’ UTR variants c.*39G>A and c.*105 G>A), which could prevent the proper transport of phototransduction proteins between the inner and the outer photoreceptor segments. More details in the text.