Molecular Pathogenesis
CNGB3, CNGA3, PDE6C, GNAT2, and PDE6H are all expressed in the cone photoreceptor and are crucial for cone phototransduction:
Light-excited cone visual pigment molecules induce the exchange of GDP to GTP on the transducin alpha subunit (GNAT2) and its release from the inhibitory beta/gamma subunits.
The activated GTP-transducin then binds and activates the alpha' subunit of the retinal cone photoreceptor phosphodiesterase (PDE6C) by retracting the inhibitory gamma subunit (PDE6H).
Retinal cone photoreceptor PDE6C hydrolyzes cGMP, reducing its intracellular concentration and causing closure of the heterotetrameric cGMP-gated cation channels (
CNGA3 and
CNGB3) and, subsequently, membrane hyperpolarization [
Lamb & Pugh 2006].
Transducin thus mediates the first step, the phosphodiesterase the intermediate, and the cGMP-gated channel represents the final step in the phototransduction cascade.
In contrast, the most recently identified ACHM-related gene, ATF6, encodes a ubiquitously expressed transmembrane transcription factor known for its function in the ATF6 unfolded protein response pathway [Walter & Ron 2011, Wang & Kaufman 2012, Kohl et al 2015]. How and why pathogenic variants in this ubiquitously expressed gene result solely in cone dysfunction is to date unknown.
ATF6
Gene structure.
ATF consists of 16 coding exons. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. Thirteen different pathogenic variants have been reported in 15 families [Ansar et al 2015, Kohl et al 2015, Xu et al 2015, Carss et al 2017, Skorczyk-Werner et al 2017]. Nine of the 13 are nonsense variants, splice site variants, small insertions, and deletions. Only four missense variants have been reported.
Six individuals (from 2 unrelated families of Irish/British descent) who are homozygous for c.970C>T have been identified; the families were shown to have a common haplotype of 0.7 Mb suggestive of a founder variant [Kohl et al 2015]. Another pathogenic variant, c.1533+1G>C, was observed recurrently in four French Canadian families, also suggesting a founder variant in this population [Kohl et al 2015, Xu et al 2015].
Table 4.
ATF6 Pathogenic Variants Discussed in This GeneReview
View in own window
Variants listed in the table have been provided by the authors. GeneReviews staff have not independently verified the classification of variants.
GeneReviews follows the standard naming conventions of the Human Genome Variation Society (varnomen.hgvs.org). See Quick Reference for an explanation of nomenclature.
- 1.
Normal gene product.
ATF6 encodes a 670-amino-acid, ubiquitously expressed 90-kd ER stress-regulated transmembrane transcription factor known for its function in one of three unfolded protein response pathways (i.e., ATF6 pathway). It is required for ER stress response and transcriptional induction from ER stress-response elements (ERSEs). On induction of ER stress, the cytosolic ~400-residue N-terminal portion of ATF6 (N-ATF6) is released. N-ATF6 possesses the transcriptional activation domain, the bZIP domain, the DNA-binding domain, and nuclear localization signals. It translocates to the nucleus, where it interacts with several other proteins to form an ERSE-binding complex that is responsible for the induction of ER stress genes (ERSGs) [Walter & Ron 2011, Wang & Kaufman 2012].
Abnormal gene product. Most pathogenic missense variants result in loss of protein function. One missense variant, p.Arg324Cys, localizes to the basic region of the bZIP domain, affecting an arginine residue that is conserved not only among transcription factors of the ATF family but also in those of the AP-1 family, severely impairing ATF6 transcriptional activity [Kohl et al 2015]. The p.Arg324Cys variant was functionally characterized in detail and shown to impair transcriptional activity [Kohl et al 2015, Chiang et al 2017].
Other missense changes have been studied and divided into class I, 2, or 3:
Class 1. Disease-associated
ATF6 missense variants that result in impaired ER-to-Golgi trafficking and diminished regulated intramembrane proteolysis and transcriptional activity
Class 2. Disease-associated
ATF6 missense variants that retain the entire ATF6 cytosolic
domain with fully intact transcriptional activity and constitutive induction of downstream target genes, even in the absence of ER stress
Class 3. ATF6
missense variants with complete loss of transcriptional activity because of absent or defective bZIP domains
Primary fibroblasts from patients with class 1 or class 3 ATF6 pathogenic variants show increased cell death in response to ER stress [Chiang et al 2017].
Of note, the Atf6 knockout mouse model does not recapitulate the human achromatopsia phenotype [Kohl et al 2015].
CNGA3
Gene structure.
CNGA3 consists of eight coding exons [Wissinger et al 2001]. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. More than 150 different pathogenic variants have been associated with ACHM [Kohl et al 1998, Wissinger et al 2001, Johnson et al 2004, Tränkner et al 2004, Nishiguchi et al 2005, Varsányi et al 2005, Ahuja et al 2008, Koeppen et al 2008, Reuter et al 2008, Koeppen et al 2010, Thiadens et al 2010, Genead et al 2011, Vincent et al 2011]. The vast majority of pathogenic variants are missense (~80%). Only a few nonsense variants, insertions, and deletions have been observed.
Normal gene product.
CNGA3 encodes the cyclic nucleotide-gated cation channel alpha 3 (the alpha subunit of the cone photoreceptor cGMP-gated cation channel [CNG]). CNGA3 has 694 amino acids and a predicted weight of 78.8 kd. An alternatively spliced exon that extends the open reading frame by an additional 55 amino acids has been reported [Wissinger et al 2001]. Alpha subunits on CNG channels are able to form functional homo-oligomeric channels, yet their biophysical properties differ from those of heteromeric native CNG channels consisting of three alpha subunits and one beta subunit.
Abnormal gene product. Functional analysis has shown that in many cases channel function is strongly impaired or completely absent. The pathogenic missense variants mostly affect amino acid residues that are highly conserved among the members of the CNG channel family, and cluster at structural and functional domains including the cGMP-binding domain [Wissinger et al 2001, Faillace et al 2004, Patel et al 2005, Koeppen et al 2008, Reuter et al 2008].
Some pathogenic variants in the pore region and the cGMP binding domain are associated with incomplete achromatopsia. These abnormal proteins can form functional channels, but with grossly altered properties, including altered affinity for cGMP and/or cAMP, and changes in the gating properties of the cone CNG channels, like Ca2+ blockage and permeation [Tränkner et al 2004, Liu & Varnum 2005, Reuter et al 2008, Koeppen et al 2010].
Animal models have helped to clarify the underlying pathogenic mechanisms:
Mouse.
Cnga3(-/-) mice show absence of cone function, a decrease in the number of cones in the retina, and morphologic abnormalities of the remaining cones.
Cnga3(-/-) cones fail to transport opsin into the outer segment and downregulate various proteins of the phototransduction cascade. Apoptotic cell death is induced; however, loss of
Cnga3 does not appear to affect the transcription of other cone-specific genes [
Biel et al 1999,
Michalakis et al 2005]. Gene therapy has been successfully tested in these mouse models and shown to restore cone-mediated vision [
Michalakis et al 2012].
CNGB3
Gene structure.
CNGB3 consists of 18 coding exons [Kohl et al 2000]. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. More than 125 different pathogenic variants have been reported [Kohl et al 2000, Sundin et al 2000, Rojas et al 2002, Johnson et al 2004, Michaelides et al 2004, Okada et al 2004, Kohl et al 2005, Nishiguchi et al 2005, Varsányi et al 2005, Khan et al 2007, Wiszniewski et al 2007, Thiadens et al 2009b, Azam et al 2010, Mayer et al 2017]. The vast majority are pathogenic nonsense variants, frameshift deletions and insertions, and putative splice site variants. Only a few pathogenic missense variants (~10%) have been observed.
One, resulting in the p.Ser435Phe mutated protein, causes "Pingelapese blindness" in achromats originating from the island of Pingelap in Micronesia [Kohl et al 2000, Sundin et al 2000].
The recurrent single base-pair deletion c.1148delC is the most common pathogenic variant underlying achromatopsia worldwide, accounting for approximately 70% of all CNGB3 disease-causing alleles and approximately 40% of all achromatopsia-associated alleles. The c.1148delC deletion results from a founder effect [Wiszniewski et al 2007].
Table 5.
CNGB3 Pathogenic Variants Discussed in This GeneReview
View in own window
| DNA Nucleotide Change | Predicted Protein Change | Reference Sequences |
|---|
| c.1148delC | p.Thr383IlefsTer13 |
NM_019098.3
NP_061971.3
|
| c.1304C>T | p.Ser435Phe |
Variants listed in the table have been provided by the authors. GeneReviews staff have not independently verified the classification of variants.
GeneReviews follows the standard naming conventions of the Human Genome Variation Society (varnomen.hgvs.org). See Quick Reference for an explanation of nomenclature.
Normal gene product.
CNGB3 encodes for cyclic nucleotide-gated cation channel beta 3 (the beta subunit of the cone photoreceptor cGMP-gated cation channel). CNGB3 is 809 amino acids long. The beta subunits are not able to form functional homo-oligomeric channels; they are therefore thought to be modulatory subunits. Functional cone CNG channels consist of three alpha subunits and one beta subunit.
Abnormal gene product. Functional analysis has shown that in many cases channel function is strongly impaired or completely absent [Peng et al 2003, Okada et al 2004, Bright et al 2005]. However, certain disease-associated CNGB3 variants in the subunit are apparent gain-of-function variants [Okada et al 2004, Bright et al 2005]. Expression of human wild type CNGA3 and mutated CNGB3 containing the Pingelapese blindness-associated p.Ser435Phe variant generated functional heteromeric channels that exhibited an increase in apparent affinity for both cAMP and cGMP and changes in the pore properties of the channel compared with wild type heteromeric channels.
Animal models have helped to clarify the underlying pathogenic mechanisms. Two naturally occurring CNGB3-null canine models, Alaskan malamute and German shorthaired pointer, have been identified [Sidjanin et al 2002]. In the Alaskan malamute, cone-degenerate pups develop day blindness and photophobia. Cone function, detectable on electroretinogram in very young affected pups, begins to fail at a few weeks' age and is undetectable in mature affected dogs. Adult affected retinas lack all cones. The first gene therapy studies in these animals showed restoration of cone-mediated vision, but the success was dependent on the age of intervention [Komáromy et al 2010].
GNAT2
Gene structure.
GNAT2 consists of eight coding exons. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. Sixteen different disease-associated variants (nonsense variant, deletions and/or insertions, one large deletion of exon 4, and a variant c.461+24G>A activating a cryptic splice site and resulting in frameshift and PTC) have been described to date [Aligianis et al 2002, Kohl et al 2002, Michaelides et al 2003, Piña et al 2004, Rosenberg et al 2004, Ouechtati et al 2011, Langlo et al 2016, Bryant et al 2017, Carss et al 2017, Taylor et al 2017, Ueno et al 2017].
Table 6.
GNAT2 Pathogenic Variants Discussed in This GeneReview
View in own window
| DNA Nucleotide Change | Predicted Protein Change | Reference Sequences |
|---|
| c.461+24G>A | -- |
NM_001377295.2
|
Variants listed in the table have been provided by the authors. GeneReviews staff have not independently verified the classification of variants.
GeneReviews follows the standard naming conventions of the Human Genome Variation Society (varnomen.hgvs.org). See Quick Reference for an explanation of nomenclature.
Normal gene product.
GNAT2 encodes for guanine nucleotide-binding protein G(t), alpha-2 subunit (the cone-specific alpha subunit of transducin), a heterotrimeric G protein that couples to the cone photopigments. The protein is 354 amino acids long.
Abnormal gene product. The majority of pathogenic variants result in loss of protein function [Cai et al 2001]. The c.461+24G>A variant results in leaky aberrant splicing, resulting in a milder phenotype described as incomplete achromatopsia or oligo-cone trichromacy [Rosenberg et al 2004].
Animal models have helped to clarify the underlying pathogenic mechanisms.
An achromatopsia mouse model is homozygous for the murine Gnat2 pathogenic variant NM_008141.3:c.598G>A (p.Asp200Asn) in exon 6 (also referred to as the cpfl3 variant) [Chang et al 2006]. Homozygous mice have poor cone-mediated responses on electroretinogram (ERG) at three weeks that become undetectable by nine months. Microscopy of the retina reveals progressive vacuolization of the photoreceptor outer segments. Immunocytochemistry with cone-specific markers shows progressive loss of labeling for Gnat2 protein, but the cone outer segments in the oldest mice examined remain intact [Chang et al 2006].
PDE6C
Gene structure.
PDE6C consists of 22 coding exons [Piriev et al 1995]. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. More than 50 different pathogenic variants in PDE6C have been described; they include missense variants, nonsense variants, small indels, and variants affecting splicing [Chang et al 2009, Thiadens et al 2009b, Grau et al 2011, Huang et al 2013, Weisschuh et al 2018].
Normal gene product.
PDE6C encodes PDE6C, the phosphodiesterase 6C, cGMP-specific, cone, alpha-prime. This alpha' subunit of the cone-specific phosphodiesterase consists of 858 amino acids.
Abnormal gene product. Disease-associated variants result in markedly reduced to completely absent PDE6C enzymatic activity [Chang et al 2009, Grau et al 2011].
Animal models have helped to clarify the underlying pathogenic mechanisms. The cone photoreceptor function loss 1 (cpfl1) mouse mutant is a model for Pde6c-related achromatopsia [Chang et al 2009], which has a 116-bp insertion between exons 4 and 5 (NM_001170959.1:c.864_865ins116) and an additional 1-bp deletion in exon 7 (NM_001170959.1:c.1042delT) in cis (on the same allele). The phenotype can be easily typed by ERG as early as age three weeks. Histology of cpfl1 mouse retinae revealed grossly normal morphology and layering. However, as early as age three weeks, there was vacuolization of a small subset of cells in the photoreceptor layer with subsequent rapid, progressive depletion of cone photoreceptors. Loss of cones progresses, such that very few were detected in retinal sections of five-month-old animals [Chang et al 2009].
PDE6H
Gene structure.
PDE6H consists of only three coding exons [Shimizu-Matsumoto et al 1996]. For a detailed summary of gene and protein information, see Table A, Gene.
Pathogenic variants. Originally the single homozygous pathogenic nonsense variant c.35C>G in PDE6H was described in three affected individuals from two independent families originating from Belgium and the Netherlands [Kohl et al 2012]. Recently, two Pakistani brothers were shown to be homozygous for the same pathogenic variant [Pedurupillay et al 2016].
Table 7.
PDE6H Pathogenic Variants Discussed in This GeneReview
View in own window
Variants listed in the table have been provided by the authors. GeneReviews staff have not independently verified the classification of variants.
GeneReviews follows the standard naming conventions of the Human Genome Variation Society (varnomen.hgvs.org). See Quick Reference for an explanation of nomenclature.
Normal gene product.
PDE6H encodes phosphodiesterase 6H, cGMP-specific, cone, gamma; PDE6H, the inhibitory gamma subunit of the cone photoreceptor phosphodiesterase. PDE6H consists of only 83 amino acids.
Abnormal gene production. The sole PDE6H nonsense variant is predicted to result in complete loss of function of PDE6H either by degradation of the mRNA by nonsense-mediated decay or the truncation of the protein [Kohl et al 2012].
Of note, the Pde6h knockout mouse model does not recapitulate the human achromatopsia phenotype [Brennenstuhl et al 2015].
