Clinical and Biochemical Characterization of Specific GUCY2D Alleles Associated With a Rare Form of Night Blindness

Abstract

Purpose: To clinically and biochemically characterize a rare autosomal recessive rod-cone dysfunction, with electroretinographic similarities to some forms of stationary night blindness (SNB), associated with biallelic variants in GUCY2D.

Methods: Six patients from five families with a history of longstanding night blindness, no fundus features suggestive of retinitis pigmentosa, and an unusual electroretinographic phenotype were ascertained. Clinical examination and genotyping were performed. Selected GUCY2D variants were tested for binding to and activation by guanylate cyclase-activating proteins (GCAPs) in HEK293 cells.

Results: The visual acuity was normal or moderately reduced (20/20–20/60) with three patients having a tritan defect on color vision testing. Retinal imaging showed central macular hypopigmentation with temporal vascular attenuation. Rod photoreceptor-mediated electroretinogram (ERG) components were undetectable or severely reduced in all but one case, and cone-mediated responses were variable. A high degree of ERG stability was demonstrated in three cases. Molecular analyses revealed biallelic variants of GUCY2D in all patients, four of which are clinically and biochemically characterized for the first time, to our knowledge. These allelic variants encoded retinal guanylyl/guanylate cyclase 1 (RetGC1) mutants whose enzymatic activities were significantly diminished due to drastically reduced affinity of RetGC1 for GCAPs.

Conclusions: The apparent lack of retinal degeneration, clinical features, predominant and severe rod photoreceptor involvement, and relatively high degree of ERG stability are similar to rare forms of SNB. Biallelic disease-causing variants in GUCY2D are usually associated with Leber's congenital amaurosis (LCA); however, this study illustrates the phenotypic variability of GUCY2D retinopathies in association with variants not biochemically dissimilar to those causing LCA and highlights the complexity of RetGC1 regulation in rod and cone photoreceptor function.

Inherited retinal disorders (IRDs) are a genetically and phenotypically heterogeneous group of disorders that constitute a major cause of visual disability in the developed world.^1,2 The most severe form of IRDs, Leber's congenital amaurosis (LCA), manifesting as severe visual loss from birth or early infancy, is associated with variants in 24 genes identified so far.³ Biallelic variants in the human retinal guanylyl/guanylate cyclase 1 (RetGC1) coding gene (GUCY2D, MIM: 600179) account for 10% to 20% of autosomal recessive LCA (LCA1, MIM: 204000). Specific variants in GUCY2D can cause autosomal dominant cone–rod dystrophy (CORD, MIM: 601777).⁴ More recently, a rare form of autosomal recessive congenital (stationary) night blindness has been associated with GUCY2D variants (SNB, MIM: 618555).⁵ 

GUCY2D encodes RetGC1, the main guanylate cyclase of the photoreceptors, which is an important element in modulating the photoresponse.⁶ Phototransduction starts from photoisomerization of the photopigment in the photoreceptor outer segments (OSs). This activates the photoreceptor G-protein transducin, which in turn activates the photoreceptor phosphodiesterase (PDE6), causing hydrolysis of cyclic guanosine monophosphate (cGMP) and closure of the cyclic nucleotide-gated (CNG) channels at the OS plasma membrane, ultimately resulting in reduction of intracellular cation influx and hyperpolarization of the photoreceptors.⁶ Low intracellular Ca²⁺ concentrations activate RetGC isozymes 1 and 2 through their interaction with Ca²⁺/Mg²⁺ binding proteins (guanylate cyclase-activating proteins [GCAPs]). Activated RetGC, predominantly RetGC1,⁵ replenishes cGMP, accelerates reopening of the CNG channels, and restores the partial depolarization state of the photoreceptor membrane potential.⁷ 

In vitro analysis of 45 recessive LCA-associated GUCY2D variants showed that they reduce or abolish cGMP synthesis, which would likely keep most of the CNG channels in the rods and cones closed.⁸ However, detailed phenotyping of LCA1 patients with detectable visual function showed cone-mediated vision to be more affected than that driven by the rods.^9,10 In some patients with undetectable cone function, the rod-mediated function had an apparent lack of saturation even under high photopic levels, which was attributed to the absence of cone-mediated rod inhibition.⁹ Conversely, CORD-associated dominant GUCY2D variants are gain-of-function alleles exerting their action through shifting Ca²⁺-sensitivity, thus keeping the CNG channels open even under light-adapted conditions, eventually leading to photoreceptor degeneration.⁸ 

Recently, six patients with childhood-onset night blindness and rod more than cone dysfunction were reported to harbor biallelic GUCY2D variants.^5,11 The associated genotypes did not cluster in a specific domain of RetGC1, despite the strong predilection to affect rod function. Because of the relatively mild nature of this disorder, the authors suggested that the associated missense changes are likely to be hypomorphic alleles.⁵ 

In this study, we determined the clinical and molecular features of a further six patients with predominant night blindness associated with GUCY2D alleles, identified through next-generation sequencing. This study also performed in vitro functional assessment of four of the identified GUCY2D variants and compared them to LCA1-associated alleles. 

Methods

This study adhered to the tenets of the Declaration of Helsinki and was approved by the Northwest Research and the Moorfields Eye Hospital ethics committees. Informed consent was obtained from all patients. The animal procedures for this study were approved by the Salus University Institutional Animal Care and Use Committee in accordance with the National Institutes of Health guidelines for animal use in research. 

Patient 1 was evaluated as part of a wider project to explore the genetic causes of severe rod dysfunction in patients without (1) marked loss of cone function based on full-field electroretinogram (ERG), (2) typical features of retinitis pigmentosa, or (3) ON bipolar cell dysfunction.¹² Patients 2 to 5 were subsequently identified from genome sequencing data in the 100,000 Genomes Project, based on similarities of phenotype and genotype.¹³ Finally, patient 6 was ascertained from the Oculome Panel test project, with revision of the phenotype from LCA to SNB through deep phenotyping.¹⁴ 

Clinical Examination

All patients underwent full ophthalmic examination including best-corrected Snellen visual acuity (VA), color vision using the Ishihara pseudoisochromatic test or the Hardy–Rand–Ritter (HRR) color test. Retinal color or pseudocolor images were obtained using Topcon (Tokyo, Japan) or Optos (Dunfermline, United Kingdom) fundus imaging devices, respectively. Fundus autofluorescence (FAF) images were obtained with the Optos device, and spectral-domain optical coherence tomography (SD-OCT) scans were obtained using the SPECTRALIS HRA+OCT device (Heidelberg Engineering, Heidelberg, Germany). Full-field and pattern electroretinogram (PERG) testing was performed to incorporate the standards of the International Society for Clinical Electrophysiology of Vision (ISCEV).^15,16 Photopic On–Off ERGs and S-cone ERGs were also performed.^17,18 

Molecular Genetics

Leukocyte DNA was extracted from all participants. Patient 1 underwent whole exome sequencing (WES) as described previously.¹² The phase of variants was determined by Sanger sequencing of parental DNA. Patients 2 to 5 and their parents underwent whole genome sequencing through the NHS-England Genomic Medical Service (GMS) as described previously.¹³ Patient 6 underwent, as part of the GMS, custom clinical exome enrichment and sequencing as described previously¹⁴; parental testing was not performed. 

Biochemical Methods

GUCY2D \({\rm{Mutagenesis}}\)

GUCY2D mutations were introduced in human RetGC1 cDNA in a modified pRc/CMV-based construct^19,20 using the “splicing by overlap extension” technique²¹ utilizing the following pairs of overlapping mutation-coding primers: 

• • p.(Tyr555Ser)—5′-AACATTGGTGTCTcTGAGGGAGACAGGGTTTG and 5′-CAAACCCTGTCTCCCTCAgAGACACCAATGTT
• • p.(Arg624Gln)—5′-CAAGCTCCAGGAGgTCCGGCATGAG and 5′- CTCATGCCGGAcCTCCTGGAGCTTG
• • p.(Leu587Val)—5′-TCAGAGCACTGCACGCagGGCTCTCTTCAG and 5′- CTGAAGAGAGCCctGCGTGCAGTGCTCTGA
• • p.(Gly653Glu)—5′- GCTGGACCTTATCAAGGaAATAAGGTACCTGC and 5′-GCAGGTACCTTATTtCCTTGATAAGGTCCAGC

The fragments for each mutant were PCR-amplified using a Thermo Scientific Phusion Flash DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA). They were then inserted into the AgeI/DraIII restriction endonucleases sites of the modified mOrange–GUCY2D cDNA construct,²⁰ and then a fragment coding for the extracellular domain of the cyclase amplified from wild-type GUCY2D cDNA was inserted into the HindIII/AgeI sites of each construct, using the 5′-ACCCAAGCTTGATATCGAATTCCGGCCCA and 5′-TGAAGTAACCGGTGCCTCACATAATGG primers, to replace the mOrange tag with the full-size extracellular domain of GUCY2D/RetGC1 in the expressed protein. All constructs used in the study were verified by Sanger sequencing on both strands. 

GUCY2D Expression

Wild-type and mutant GUCY2D gene products were expressed in human embryonic kidney 293 (HEK293) cells from a modified Invitrogen (San Diego, CA, USA) pRc/CMV vector using the calcium phosphate DNA precipitation transfection protocol, and the membrane fraction containing the recombinant RetGC1 was isolated as detailed previously.²² The activity of the cyclase was assayed using [α-³²P] guanosine-5'-triphosphate (GTP) (PerkinElmer, Waltham, MA, USA) as a substrate, and the [³²P]cGMP product was quantified using thin-layer chromatography (TLC) as described previously.²² Briefly, the assay mixture (25 µL) incubated at 30°C contained 30-mM 3-(N-morpholino)propanesulfonic acid (MOPS)–potassium hydroxide (KOH) (pH 7.2), 60-mM potassium chloride (KCl), 4-mM sodium chloride (NaCl), 1-mM dithiothreitol (DTT), 2-mM Ca²⁺/EGTA buffer, 1-mM or 6-mM free Mg²⁺ as indicated in the text, 0.3-mM adenosine triphosphate (ATP), 4-mM cGMP, 1-mM GTP, and 1 µCi of [α-³²P]GTP. The resultant [³²P]cGMP product was analyzed by TLC using fluorescently backed polyethyleneimine cellulose plates (Merck, Rahway, NJ, USA) developed in 0.2-M lithium chloride (LiCl) and eluted with 2-M LiCl, followed by liquid scintillation counting of the radioactivity. 

GCAP Expression

Human GCAP1 (Ser6) was expressed from a Novagen/Calbiochem (Darmstadt, Germany) pET-11d vector in a BLR(DE3) Escherichia coli strain harboring yeast N-myristoyl transferase and purified by hydrophobic interaction chromatography and size-exclusion chromatography to ∼95% purity by SDS-PAGE as previously described.²³ For co-transfection experiments in HEK293 cells, bovine GCAP1 was expressed tagged at the C-terminus with SuperGlo (Clontech, Mountain View, CA, USA) enhanced green fluorescent protein (GFP) using pQBIfN3 vector (Clontech) as previously described.^23,24 Human GCAP2 cDNA was PCR-amplified from a human retinal cDNA library (StrataGene, La Jolla, CA, USA), inserted into the NcoI/BamHI sites of pET-11d, and, after verification by sequencing, expressed in E. coli and purified as previously described for bovine and mouse GCAP2.^25,26 Human GCAP3 cDNA, main splice variant,²⁷ was chemically synthesized (Genewiz, South Plainfield, NJ, USA), inserted into the NcoI/BamHI sites of the pET-11d vector, and expressed in E. coli following protocol for the isolation of myristoylated GCAP2. The purity of the recombinant GCAP2 and GCAP3 expressed in E. coli was >90% by SDS-PAGE/Coomassie Blue stain. 

Ca²⁺/EGTA Buffers

Ca²⁺/EGTA buffers maintaining defined free Ca²⁺ and Mg²⁺ concentrations in the RetGC1 assay were calculated, prepared, and verified by fluorescent Ca²⁺ indicator dyes as previously described in detail.²⁸ 

Co-Transfection Experiments

The mOrange-tagged RetGC1 constructs were co-expressed with the SuperGlo GFP-tagged bGCAP1 in HEK293 cells at the cyclase-to-GCAP1 plasmid ratio of ∼100:1 as described in detail previously,²⁰ using ∼3 µL/µg FuGENE HD Transfection Reagent (Promega, Madison, WI, USA). Confocal images were taken utilizing an FV1000 confocal laser scanning biological microscope (OLYMPUS, Tokyo, Japan) using 543-nm and 488-nm excitation for the red and the green fluorochromes, respectively, collected in a sequential mode. No changes to the original images were made except for minor gamma correction applied to the whole image. Fluorescence intensities were compared by scanning non-corrected images. 

Retinal Tissue

Mouse retinae were harvested from dark-adapted GCAPs^–/– mice²⁹ under infrared illumination, and the tissue was processed for guanylyl cyclase assay as previously described.^26,30 Assays using dark-adapted mouse retinae were conducted under infrared illumination using infrared goggles following a procedure previously described³¹; the reaction mixture in the assay was supplemented with creatine kinase and phosphocreatine.³² 

Results

Clinical and Electrophysiological Features

The clinical and ERG characteristics of this cohort are summarized in Table 1. The age at the last clinical assessment ranged between 13 and 42 years. All patients had childhood-onset nyctalopia with normal to moderately reduced VA (20/20–20/60). When available, color vision screening showed normal red–green discrimination and a tritan defect. Anterior segment examination was unremarkable in all patients, and there was no nystagmus. Except for patient 3, all patients had temporal retinal vascular attenuation (Fig. 1), and some had central macular hypopigmentation (Supplementary Fig. S1). The FAF images (Fig. 1) showed central macular hyperautofluorescence corresponding to variable degrees of ellipsoid zone disruption on OCT (Fig. 2). All patients had undetectable rod-specific dark-adapted (DA) 0.01 ERGs with cone system dysfunction (Figs. 3, 4). The clinical histories and specific retinal findings are given below. 

Table 1.

 

Demographic, Clinical, and Electroretinographic Features of Rod–Cone Dysfunction in Six Patients With Biallelic Variants in GUCY2D

 

Figure 1.

 

Ultra-widefield retinal pseudocolor and autofluorescence (AF) images for patients 1 to 6. Except for patient 3, all patients had temporal retinal vascular attenuation. The left fundus image of patient 1 showed an inferonasal alteration, better delineated on AF as a patch of reduced AF outlined by an increased signal. All patients had a hyperautofluorescent spot at the macular center.

Figure 2.

 

OCT scan of the right and left eyes of patients 1 to 6. All patients had variable degrees of disruption of the foveal ellipsoid–interdigitation zones. Patients 2 and 4 had a small vitelliform-like lesion at the fovea.

Figure 3.

 

ISCEV standard full-field ERGs. On-Off ERGs, S-cone ERGs, and PERG from one eye each of patients 1 (P1, 26 years), 2 (P2, 32 years), 3 (P3 at 17 years and 24 years, respectively), and representative normal (N) examples for comparison. Standard ERGs are to xenon flashes of light, and all responses were recorded using gold foil corneal electrodes. Recordings showed a high degree of interocular symmetry and are shown from one eye only, with patient traces superimposed to demonstrate reproducibility.

Figure 4.

 

ISCEV standard full-field ERGs and PERG from one eye each of patients 1 (P1, 39 years), 2 (P2, 35 years), 4 (P4, 6 years), 5 (P5, 6 years), 6 (P6, 8 years), and representative normal (N) examples for comparison. Standard ERGs were recorded with LED flashes of white light and were recorded with gold foil corneal electrodes, although patient 4 (S4) was tested with silver thread electrodes. Recordings showed a high degree of interocular symmetry and are shown from one eye only, with patient traces superimposed to demonstrate reproducibility.

Patient 1 (GC25541)

A 27-year-old man of European ancestry had reduced night vision noticed at the age of 11 years, resulting in a diagnosis of retinitis pigmentosa. He struggled under flood illumination and dusk-like light levels and had difficulty distinguishing between shades of green and blue. He had difficulty tracking small fast-moving objects such as cricket or golf balls. He was otherwise in good health. On his most recent examination (at age 42), color contrast thresholds were normal along the protan and deutan axes but showed elevation along the tritan axis.³³ The retinal images and features are given in Figure 1, Supplementary Figure S1, and Table 1. Notably, there was a small region of outer retinal alteration in the inferonasal midperiphery of the left eye. 

Dark-adapted and light-adapted (LA) full-field ERGs were performed at the age of 26 years (Fig. 3). The dim flash rod (DA 0.01) ERG was almost undetectable. The strong flash (DA 10) ERG showed a markedly subnormal a-wave and subnormal b:a ratio, with a b-wave of short peak time, similar to the DA red flash ERG waveform and consistent with a predominant contribution from the DA cone system. LA flicker (30 Hz) ERG was severely delayed (by 14 ms) with relatively mild amplitude reduction, and single flash cone (LA 3) ERGs were also delayed and of subnormal amplitude. Photopic On-Off ERG revealed greater delay in the b- than d-waves, and a subnormal b:a ratio. The S-cone ERG was undetectable. The PERG P50 was reduced, consistent with mild macular dysfunction (Fig. 3). Repeat testing (at the age of 32 years) after prolonged (overnight) dark adaptation of one eye had no effect on ERGs. Further ERG and PERG were performed (age 39 years), revealing no evidence of worsening retinal or macular function over more than 13 years (Figs. 4, 5). 

Figure 5.

 

Graphs showing stability of the main full-field ERG and PERG P50 component amplitudes in patients 1 (P1), 2 (P2), and 3 (P3). All amplitudes are plotted as a percentage of the age-matched lower 5th percentile to allow direct comparison of DA and LA responses and recordings to xenon and LED-based stimuli (LED stimuli used to obtain final follow-up ERGs in patients 1 and 2). Note that baseline DA 0.01 ERGs in P2 were undetectable, but likely partly due to technical factors (responses at the three follow-up visits were consistently relatively preserved in comparison).

Patient 2 (GC22985)

A 35-year-old woman of European ancestry complained of a mild decrease in the VA of both eyes, from 20/15 to 20/30. She had lifelong nyctalopia and difficulty in dim light. Upon direct questioning, after dark-adapting for 45 minutes, the patient noted that she was able to see the surrounding objects. The medical history was significant for kyphosis, bronchial asthma, and vitamin D deficiency for which she was receiving supplements, with no history of malabsorption. Her serum vitamin A level was reported to be normal. There was no family history of night blindness. Color vision testing showed normal red–green color discrimination on the HRR test, with a mild defect in the tritan axis. The retinal changes depicted in Figure 1 remained unchanged over a 4-year follow-up period. 

Baseline DA 10 ERG at the age of 32 years showed subnormal a-waves and a b-wave of short peak time, of similar timing to the preserved DA red flash ERG x-wave, in keeping with a predominant DA cone system contribution (Fig. 3). LA 30-Hz ERG was mildly delayed and subnormal, and the LA 3 ERG was subnormal. On-Off ERG b- and d-waves were of normal timing, but the b:a ratio was subnormal. S-cone ERGs were just within the reference range. PERG P50 was reduced and stable over a 17-month period, consistent with macular dysfunction. Overnight dark adaptation of one eye had no effect on ERGs. Further ERGs performed nearly 4 years after initial testing revealed no evidence of worsening retinal function (Figs. 4, 5). 

Patient 3 (GC18451)

A 17-year-old Caucasian British woman had long-standing difficulty with night vision, first noticed at the age of 10 years. The retinal features depicted in Figure 1 were stable over 13 years of follow-up. At the age of 17 years, DA 0.01 ERGs were residual, and responses to stronger flashes (DA 10 ERGs) had a subnormal a-wave, subnormal b:a ratio and b-wave peak of short peak time, similar to that of the DA red flash ERG x-wave (Fig. 3). LA 30-Hz and LA 3 ERGs were markedly delayed and of normal amplitude. On-Off ERG b- and d-waves were delayed and there was a low b:a ratio bilaterally. The S-cone ERG components were subnormal bilaterally. pERG P50, performed on the left eye at the final visit, was normal, revealing no evidence of macular dysfunction. Overnight dark adaptation of one eye failed to reveal a significant interocular ERG asymmetry. Repeat testing more than 7 years after baseline ERGs (at age of 24 years) revealed no evidence of worsening rod or cone system function (Figs. 2D, 4C). 

Patients 4 and 5 (GC31207)

Patients 4 and 5 were female siblings of Southeast Asian ancestry, 18 years and 13 years old, respectively. Patient 4 had longstanding difficulties with seeing in the dark, with more recent problems with central vision. She had mild vitiligo but was otherwise fit and well. The retinal features (Fig. 1) are described in Table 1. When patient 4 was 18 years old, ERGs were performed with corneal electrodes according to the ISCEV standard. The DA 0.01 ERGs were undetectable, and there was marked DA 10 ERG a-wave reduction; the b:a ratio was normal, but b-waves had a short peak time, similar to that of the LA 3 ERGs (Fig. 4). The DA red flash ERG was undetectable. LA 30-Hz ERGs were delayed and severely reduced, and LA 3 ERGs were severely subnormal. The S-cone ERGs were detectable bilaterally. PERG P50 was delayed and markedly subnormal (P50 peak time 65 ms, amplitude 0.5 µV). The findings were consistent with severe generalized rod and cone dysfunction with marked macular involvement. 

Patient 5 was a younger sibling of patient 4 with similar symptoms of poor vision under dim illumination. The clinical features are given in Table 1 and shown in Figure 1. At the age of 6 years, her ERG phenotype was very similar to that seen in her older sibling, showing evidence of severe generalized rod and cone photoreceptor dysfunction with delayed and markedly subnormal PERGs, consistent with severe macular involvement (Fig. 4D). 

Patient 6 (GC31196)

Patient 6 was a 16-year-old Sri Lankan female who had childhood-onset difficulties with seeing under dim illumination and required additional support with navigation in the dark. She also had experienced difficulties with color discrimination since early childhood but denied photo-aversion. She had no other medical issues. Her retinal features are depicted in Figure 1 and summarized in Table 1. Color vision testing using HRR plates showed normal red–green color discrimination and a tritan defect. When she was 8 years old, DA 10 ERG a-waves showed marked reduction with b-wave peak times that were short and of similar timing to DA red flash ERG x-waves, suggesting a cone system origin. LA 30-Hz and LA 3 ERGs showed marked delay with moderately reduced amplitudes. PERG P50 was subnormal, in keeping with macular involvement. 

Molecular Characteristics

The pedigrees and GUCY2D (NM_000180) variants detected by WES (patient 1) or whole-genome sequencing (patients 2–5) are shown in Figure 6 and summarized in Table 2. Patients 1 to 3 were compound heterozygotes for missense changes, and patients 4 to 6 were homozygotes for missense changes. Patients 1 and 3 had the known GUCY2D LCA variant: c.2302C>T, p.(Arg768Trp), likely in trans with previously unpublished missense changes. The variants in this cohort are in the following GUCY2D domains: juxtamembrane domain, kinase homology domain, and cyclase catalytic domain (Table 3). Except for the p.(Arg768Trp) pathogenic allele, all of the other variants have very low allele frequencies in the Genome Aggregation Database (gnomAD) and are classified as variants of uncertain significance according to American College of Medical Genetics and Genomics (ACMG) criteria (Table 3). 

Figure 6.

 

Pedigrees of all six patients. The affected family members are shown as solid symbols. Parental phasing is also shown.

Table 2.

 

Combinations of GUCY2D (NM_000180) Variants Associated With Stationary Night Blindness in This Case Series and Previous Studies

 

Table 3.

 

Characteristics of the GUCY2D (NM_000180) Variants Associated With Stationary Night Blindness in This Case Series and Previous Studies

 

Functional Variant Testing

We conducted in vitro studies to test four variants seen in this cohort: p.(Tyr555Ser), p.(Leu587Val), p.(Arg624Gln), and p.(Gly653Glu). Functional testing of the variant p.(Arg768Trp) has been undertaken in the past,^34,35 and it has been found to impair RetGC1 activation by GCAP1 and/or completely suppress its catalytic activity. 

The activity of RetGC1 and its SNB mutants identified in this study was assayed in vitro in the presence of GCAP1 (Figs. 7–9). We initially confirmed the presence of RetGC1 in the transfected cells by immunoblot and confirmed that the cells expressed similar levels of mutant RetGC1 variants and wild type (Fig. 7A). We then tested RetGC1 activation by GCAP1 and found that the four SNB mutants displayed either much lower than wild-type RetGC1 activity (Tyr555Ser, Leu587Val, and Arg624Gln) or no detectable activity at all (Gly653Glu), even at high concentrations of Mg²⁺ GCAP1 (Fig. 7B). Suppression of RetGC1 activity in the presence of high concentrations of GCAP2 and GCAP3 was also observed for all SNB mutants (Supplementary Fig. S2). This severe lack of activity was similar to that seen in null alleles associated with the GUCY2D LCA phenotype.³⁵ The residual RetGC1 activity for the p.(Tyr555Ser) and p.(Leu587Val) mutants detectable in the presence of GCAP1 showed calcium sensitivity similar to that of the wild type, whereas calcium sensitivity for the p.(Arg624Gln) mutant was drastically elevated (Figs. 7C, 8, 9B). In that regard, Ca²⁺-sensitivity of the p.(Arg624Gln) mutant was shifted in the opposite way from that of the CORD6-linked variant, p.(Arg838Ser) (Fig. 8).^36,37 This indicates that it would be even more difficult for GCAP to activate p.(Arg624Gln) RetGC1 when Ca²⁺ concentrations have declined after illumination, and this would even further restrain the already reduced enzymatic activity of the RetGC1/GCAP complex in rods and cones. 

Figure 7.

 

The effects of the GUCY2D variants on guanylyl cyclase activity and regulation by GCAP1. (A) Immunoblotting of the human GUCY2D expressed in HEK293 cells. The HEK293 cells were transfected with GUCY2D cDNA using calcium phosphate precipitation method and the membrane fractions from the transfected cultures were prepared as described in Methods. After the transfer from 7% SDS, the blot was probed by antibody against RetGC1 catalytic domain⁵³ using a Thermo Scientific SuperSignal West Femto Chemiluminescence Substrate. The amount of the GUCY2D in subsequent biochemical analyses was normalized in all cases by the level of the chemiluminescence signal relative to the wild type. (B) Dose dependence of GUCY2D activation by GCAP1. The membrane fractions from HEK293 cells expressing wild-type, Y555S, L587V, R624Q, or G653E GUCY2D were reconstituted with purified recombinant human GCAP1 (hGCAP1) and assayed at saturating 10-mM MgCl₂ in the presence of 2-mM EGTA. The data (mean ± SD) from three independent experiments were fitted (except for G653E, which had no detectable activity) using the function A = (A_max)/(1 + ([GCAP]_1/2/[GCAP])^h), where A is the activity of the cyclase, A_max is the maximal activity at saturation with GCAP, [GCAP]_1/2 is a GCAP concentration producing half-maximal stimulation, and h is a Hill coefficient. The respective A_max values (mean ± SD) from the fit for the wild-type, Y555S, L587V, and R624Q GUCY2D were 39.4 ± 0.4, 6.1 ± 0.35, 4.6 ± 0.2, and 12.4 ± 0.7 nmol/min/mL (activity in all mutants was different from the wild type (P <0.001, t-test). The respective [GCAP]_1/2 values were 0.98 ± 0.03, 4.4 ± 0.43 (P = 0.005), 2.4 ± 0.2 (P = 0.005), and 4.1 ± 0.5 (P = 0.01). (C) The Ca²⁺ sensitivity of GCAP-stimulated activation for wild-type, Y555S, L587V, and R624Q GUCY2D. The cyclase activity in membranes containing recombinant GUCY2D was assayed in the presence of 14-µM GCAP1 and Ca²⁺/EGTA buffers, varying free Ca²⁺ concentrations in the presence of 0.9-mM free Mg²⁺ and normalized as a percentage of the maximal activity of each preparation. The data (mean ± SD, n = 3) were fitted assuming a Hill function, such that A% = 100/(1+([Ca]_f/[Ca]_1/2)^h), were A% is the fractional activity of the cyclase, [Ca]_f is a free Ca²⁺ concentration in the assay, [Ca]_1/2 is a free Ca²⁺ concentration producing half-maximal inhibition of the cyclase activity, and h is a Hill coefficient. The [Ca]_1/2 values for wild-type, Y555S, L587V, and R624Q GUCY2D were 176 ± 6, 139 ± 8 (P = 0.042), 114 ± 7 (P < 0.001), and 57 ± 7 (P < 0.001), respectively.

Figure 8.

 

SNB and CORD6-linked mutations in GUCY2D alter the Ca²⁺ sensitivity of cyclase regulation in the opposite manner. Ca²⁺ sensitivity of the wild-type, R624Q, and R838S GUCY2D (the R838S data are from Dizhoor et al.⁵⁴).

Figure 9.

 

The effects of the GUCY2D mutations on the cyclase ability to bind GCAP1 in cyto. (A, B) Mutations found in SNB patients affect regulation of the mOrange-tagged GUCY2D cyclase activity similarly to the untagged GUCY2D, in both dose dependence of activation by GCAP1 (A) and Ca²⁺ sensitivity (B). mOrangeGUCY2D (wild-type, Y555S, L587V, R624Q, or G653E) with the fluorescent tag replacing a portion of the extracellular domain of the cyclase was expressed in HEK293 cells with bovine GCAP1 (bGCAP1). (C) Confocal images of the mOrange–GUCY2D (red fluorescence) and bGCAP1GFP (green fluorescence) co-expressed in HEK293 cells. The HEK293 cells were co-transfected with the bGCAP1GFP (left to right: wild-type, R624Q, and G653E) and mOrangeGUCY2D cDNAs at a molar ratio of ∼1:100 using FuGene reagent. (D) Fluorescence profile of the mOrangeGUCY2D (red trace) and bGCAP1GFP (green trace) across the cells marked in panel C. Note the evidence of co-localization with GCAP1 in wild-type and R624Q GUCY2D and a complete lack of co-localization in G653E. The Y555S and L587V mOrangeGUCY2D displayed co-localization patterns similar to that of the R624Q (not shown).

Another set of tests was conducted to evaluate the effects of GUCY2D mutations on the cyclase ability to bind GCAP1 in cyto, using co-localization of GFP-tagged GCAP1 with mOrange-tagged RetGC1 co-expressed in HEK293 cells (Fig. 9). When assayed in vitro, the mOrange-tagged RetGC1 and its mutants displayed biochemical characteristics of regulation by GCAP1 similar to those of the respective untagged cyclase variants (Fig. 9A vs. Fig. 7B), indicating the decreased affinity for GCAP1 in SNB mutants. Co-localization of RetGC1 mutants with GCAP1 observed in cyto using confocal imaging was consistent with the change in their apparent affinity for the activator protein observed in biochemical assays in vitro. Co-localization of the GFP-tagged GCAP1 and mOrange-tagged RetGC1 in the endoplasmic reticulum membranes was evident in the case of wild-type RetGC1, demonstrating a clearly defined co-distribution of both fluorochromes across the cells, but it was either poorly defined (Arg624Gln) or completely abolished (Gly653Glu) for the SNB mutant (Figs. 9C, 9D). 

Discussion

This study provides insights on the genetics and presentation of rare GUCY2D-associated night blindness based on detailed clinical, retinal imaging, and electrophysiological assessments. Distinctive structural and functional features were determined, including serial data that highlight a largely stable functional phenotype, at least until the fifth decade of life. Five previously unreported variants were identified, and functional testing revealed that four of these variants were not functionally different in vitro from those associated with GUCY2D-related LCA. 

The phenotype of our patients largely matches that described by Rodilla et al.¹¹ and Stunkel et al.⁵; however, some differences in our cohort are notable: (1) the retinal periphery in 10 of the 12 eyes showed temporal vascular attenuation, without bone-spicule–like pigmentation; and (2) all 12 eyes had abnormalities at the level of the foveal cone outer segments. Some of the patients described by the other groups were older than those presented here; hence, it is possible that pigmentary changes develop over time. Due to the retrospective nature of this clinical study, refraction data were not available. The combination of visual acuity of counting fingers or less and hypermetropia higher than +7 diopters has been reported as a feature of GUCY2D-related LCA. Characterizing the axial length and refractive error of SNB patients may provide insight into the role of RetGC1 in ocular development.³⁸ 

The DA ERG phenotype was remarkably similar within our cohort of six patients with evidence of a complete or near-complete absence of rod photoreceptor function in all but one. Dim flash rod (DA 0.01) ERGs (selective for the rod system) were undetectable or residual in all but patient 2, and all six showed evidence that DA 10 (strong flash) ERGs were largely or entirely arising in the DA cone system; DA 10 ERG b-waves were of short peak time with similarities to cone-mediated DA red flash ERG x-waves and/or LA 3 ERGs, and some cases had a reduced DA 10 b:a ratio, likely representing cone-mediated activity (the “photopic hill” phenomenon manifesting under DA conditions in the absence of rod function).^36,37 LA 30-Hz ERGs were delayed by 5 to 14 ms (median delay, 9.5 ms), and amplitudes were normal (n = 1), mild-moderately subnormal (n = 3), or severely subnormal (n = 2). There was PERG evidence of macular dysfunction in all but patient 3 (with a borderline/preserved P50 component). Patient 2 reported improved scotopic vision after extended periods in the dark, prompting further ERGs after overnight dark adaptation, but there was no ERG evidence of improved rod system function in the three patients tested. The same three patients had the mildest LA ERG reductions and underwent serial ERG testing over periods of 46 months (patient 2), 87 months (patient 3), or 157 months (patient 1), showing no evidence of worsening retinal or macular function in either eye. The ERGs and functional phenotypes in patients 1, 2, 3, and 6 are broadly consistent with those described in the six previously published SNB cases.^5,11 A notable difference is that the previous reports described ERG worsening in two children(ages 12 and 9 years) who underwent repeat assessment at 17 and 15 years, respectively, including one with concomitant visual field constriction.⁵ 

All three patients whose color vision was tested using the HRR plates had a tritan defect, with patients 1 and 3 showing subnormal S-cone ERGs. Of the remaining three, only one was tested with Ishihara plates, which do not detect tritan defects. Tritan defects have been shown to be the earliest manifestation of GUCY2D-related CORD in asymptomatic patients, suggesting that S-cones are more vulnerable to abnormal RetGC1 function.³⁹ 

Within our cohort, the alleles were located at the juxta-membrane, kinase homology, and cyclase catalytic domains of RetGC1, regions typically associated with null or very low/basal RetGC1 activity, due to an ineffective interaction with GCAPs.^9,10,40 The variants p.(Arg768Trp), p.(Gly653Glu), and p.(Arg666Trp) were the previously reported alleles within our cohort. p.(Arg768Trp) is one of the most common GUCY2D LCA variants, and it is thought to represent a northwest European founder mutation.^41–43 It affects a highly conserved residue and has been found to cause a severe reduction in GUCY2D activation by GCAP1; it has been reported in LCA patients in homozygous and compound heterozygous states.⁹ One individual with LCA was found to be a carrier of p.(Gly653Glu) (with no second allele found),⁴³ and p.(Arg666Trp) has been reported in a patient with the night blindness phenotype in trans with a frameshifting allele.⁵ 

The variant p.(Leu587Val) has not been previously reported with this phenotype; however p.(Leu587Arg) has been reported to be associated with early-onset retinitis pigmentosa, without further details regarding zygosity and phenotype.⁴⁴ p.(Tyr555Ser) and p.(Tyr909Cys) are also novel variants, with extremely low allele frequencies (Table 3). This mutation most likely directly affects the active site of the enzyme rather than its interaction with GCAP, because it is located in the catalytic domain, whereas GCAPs bind RetGC via the kinase homology domain and dimerization domains of RetGC1.^19,20 The p.(Arg624Gln) has been reported in ClinVar as a Variant of Unknown Significance (Variation ID: 957183), associated with both LCA and CORD phenotypes, and it is present in population databases (rs746020263, counted five times, allele frequency of 0.00002%). This variant is in exon 12, outside of the mutation hot spot for CORD, and no segregation data were provided for the CORD case.⁴⁵ Finally, p.(Gly653Gln) is absent from the literature and only appeared once in population databases (rs139168077, allele frequency of 0.000004%). 

Our functional testing revealed that p.(Tyr555Ser), p.(Leu587Val), p.(Arg624Gln), and p.(Gly653Glu) disabled RetGC1–GCAP/retinal degeneration 3 (RD3) binding to various degrees, similar to LCA alleles.³⁵ As the same alleles have been reported in LCA and night blindness, we cannot assume it is a particular location or conformational change that causes the different phenotypes. It is possible that the interaction between alleles partially neutralizes their effect, keeping a basal activity of RetGC1. If there was a residual rod function, the functioning cones might exert a strong lateral inhibition preventing recovery of rod function on ERG. However, overnight dark-adaptation failed to elicit measurable improvement of the rod function in our patients. 

The role of RetGCs and their interactions with GCAPs and RD3 are of key importance in the phototransduction process; yet, the effects that these proteins have on each other have yet to be fully deciphered. In vitro and in vivo studies have shown that reducing the function of GCAP can leave RetGC without its activation at low Ca²⁺ concentrations in the light, creating cGMP production and restoring the permeability of CNG channels at basal levels only, causing a slower yet with larger amplitude response to light stimulus (resulting in loss of temporal resolution to bright stimuli).^46,47 When RetGC is affected, the effect differs among variants associated with the CORD and LCA phenotypes. The former, located at the mutation hot spot of the dimerization domain (most frequently Arg838), have a dominant negative effect on RetGC1, decreasing its sensitivity to high intracellular calcium, thus causing an abnormal increase in cGMP production in the dark.^48,49,54 LCA variants, on the other hand, prevent proper formation of the RetGC:GCAP complex or/and reduction of RetGC1 catalytic activity^32,35 leading to reduced levels of cGMP, failure to open the proper fraction of the CNG channels, and a consequent reduction of the ERG a-wave.⁸ Rigg's type ERG is a rare form of congenital SNB in which constitutive activation of rod phototransduction leads to constant closure of the CNG channel and failure to produce graded potentials in response to photic stimulation.³⁶ The ERGs reflect a functionally cone-isolated retina; however, in GUCY2D-related SNB, the cones are also partially dysfunctional, potentially suggesting the presence of a compensatory yet sluggish mechanism exclusive to the cones but not necessarily fully restoring their function. 

Some of the difficulties in fully understanding these interactions are likely due to the lack of data regarding RetGC and RetGC–GCAP complex structure.⁶ Besides knowing RetGC1 location at the photoreceptor OS membrane,⁵⁰ that rods have both RetGC1 and RetGC2 and GCAP1 and GCAP2, and that cones express almost solely GCAP1 and RetGC1,^51,55 the nuances of such interactions have yet to be understood. Modeling this type of SNB using patient-derived retinal organoids could potentially contribute to our understanding of the pathophysiology of this condition. For example, it cannot be excluded³³ that, in SNB patients, in contrast to LCA patients, cGMP regulation in L- and M-cones (less frequently in S-cones), but not in rods, for some reason becomes partially compensated by the second isozyme of RetGC, RetGC2 (GUCY2F), which usually presents a minor component in cGMP production.^26,52 Why such compensation happens only in some individuals and results in the SNB phenotype but in more frequent cases it does not occur and results in LCA is difficult to explain at this time. 

In conclusion, this study has detailed the largest cohort of patients with GUCY2D-associated SNB to date, to our knowledge. Variants leading to this rare presentation can affect multiple functional domains in GUCY2D and may also be associated with GUCY2D LCA. Patients often complain of experiencing night-vision issues since the late first decade of life, present mildly subnormal visual acuity and reduced color discrimination, and show subfoveal outer retinal abnormalities on OCT. Electroretinography may reveal a distinctive and relatively stable functional phenotype, characterized by severe loss of rod function with variable cone system involvement. 

Acknowledgments

Supported by grants from the Foundation Fighting Blindness (RB); National Eye Institute, National Institutes of Health (EY011522 and EY034861 to AMD); and Commonwealth Universal Research Enhancement Program (CURE), Pennsylvania Department of Health (AMD). SL is funded by a MRC Clinician Scientist Fellowship. 

Disclosure: R. Ba-Abbad, None; I.V. Peshenko, None; M. Daich Varela, None; S. Lin, None; G. Arno, None; O.A. Mahroo, None; C. Egan, None; A.G. Robson, None; E.V. Olshevskaya, None; A.M. Dizhoor, None; A.R. Webster, None