IMPDH1-associated autosomal dominant retinitis pigmentosa: natural history of novel variant Lys314Gln and a comprehensive literature search

ABSTRACT Background Inosine monophosphate dehydrogenase (IMPDH) is a key regulatory enzyme in the de novo synthesis of the purine base guanine. Mutations in the inosine monophosphate dehydrogenase 1 gene (IMPDH1) are causative for RP10 autosomal dominant retinitis pigmentosa (adRP). This study reports a novel variant in a family with IMPDH1-associated retinopathy. We also performed a comprehensive review of all reported IMPDH1 disease causing variants with their associated phenotype. Materials and Methods Multimodal imaging and functional studies documented the phenotype including best-corrected visual acuity (BCVA), fundus photograph, fundus autofluorescence (FAF), full field electroretinogram (ffERG), optical coherence tomography (OCT) and visual field (VF) data were collected. A literature search was performed in the PubMed and LOVD repositories. Results We report 3 cases from a 2-generation family with a novel heterozygous likely pathogenic variant p. (Lys314Gln) (exon 10). The ophthalmic phenotype showed diffuse outer retinal atrophy with mild pigmentary changes with sparse pigmentary changes. FAF showed early macular involvement with macular hyperautofluorescence (hyperAF) surrounded by hypoAF. Foveal ellipsoid zone island can be found in the youngest patient but not in the older ones. The literature review identified a further 56 heterozygous, 1 compound heterozygous, and 2 homozygous variant. The heterozygous group included 43 missense, 3 in-frame, 1 nonsense, 2 frameshift, 1 synonymous, and 6 intronic variants. Exon 10 was noted as a hotspot harboring 18 variants. Conclusions We report a novel IMPDH1 variant. IMPDH1-associated retinopathy presents most frequently in the first decade of life with early macular involvement.

IMPDH forms filamentous ultrastructures which plays a role in allosteric regulation in response to high demand for guanine nucleotides (8).In biochemistry, allosteric regulation is the regulation of an enzyme by binding an effector molecule at a site other than the enzyme's active site.The site to which the effector binds is termed the allosteric site or regulatory site (9).In humans, there are 2 isoforms (IMPDH1 and IMPDH2) that share 84% sequence identity, and both are widely expressed in human tissues (8).IMPDH1 has a house-keeping function and is found more abundantly in the retina (10,11).IMPDH2 is up-regulated in proliferating cells and has a critical role in the immune system (8,12).IMPDH1 encodes the IMPDH1 protein (514 amino acids) that consists of 2 regulatory/allosteric domains, namely cystathionine Beta Synthase (CBS) 1 and 2 domain and 3 Inosine monophosphate (IMP) binding domains/catalytic domain (Figure 1b).The CBS 1 and 2 domains assemble the Bateman domain (13).The Bateman domain is a regulatory protein module and is present in all organisms.In IMPDH1, the Bateman domain is the binding target for adenosine and guanine nucleotides (14,15).
The retina is one of the most energy demanding tissues in the body, requiring large amounts of ATP/GTP to be created (16).The IMPDH1 protein is responsible for the immense GTP supply to the photoreceptor cells (7,17,18).IMPDH1 regulates the GTP synthesis in the retina depending on the change of illumination (13).Varying illumination leads to alteration of IMPDH1 phosphorylation which plays an important role in cyclic nucleoside metabolism within photoreceptors (5,19).The light exposureinduced phosphorylation occurs at the Bateman domain leading to the accumulation of IMPDH1 aggregates at the rod outer segment layer.With dark adaptation, IMPDH1 returns to the unphosphorylated disaggregated state (13,14).Furthermore, IMPDH1 immunostaining signal is considerably stronger in the photoreceptor layers than in the inner retinal layers (13).IMPDH1 has also been found to be associated with polyribosomes containing rhodopsin mRNA (20).Rhodopsin constitutes 95% of the protein in rod disks and thus any disruption may cause apoptosis of rod photoreceptors (20,21).
Protein misfolding and aggregation, rather than reduced enzyme activity, are the proposed disease mechanism (17).Considering the rod sensitivity in preclinical studies and the energy making responsibility of this gene, IMPDH1 pathogenic variant may cause initial rod disruption with early cone involvement.
This study reports 2 generation family with a novel IMPDH1 variant and reviews the published IMPDH1associated retinopathy variants to help understand the genotype-phenotype correlation of this specific gene.

Materials and methods
This paper was conducted in 2 parts.First, the ophthalmic biomarkers of 3 patients with a novel likely pathogenic IMPDH1 variant.The second part of the study was to review all publications reporting IMPDH1 variants with an ophthalmic phenotype.The methods are described below.

Novel IMPDH1 variant
Retrospective review of 3 patients from a two-generation family with a confirmed likely pathogenic variant in IMPDH1 gene recruited from the Save Sight Institute, Sydney, NSW, Australia.Genomic analysis was performed at the molecular genetics laboratory Casey Eye Institute (Molecular Vision Laboratory), USA.Genomic variant classification was performed according to the American College of Medical Genetic (ACMG) guidelines (22).The study adhered to the Declaration of Helsinki and was approved by the ethical board of Sydney Southeast Health District committee.

BCVA
Best corrected visual acuity (BCVA) was measured using a logarithm of minimum angle of resolution (logMAR) at each visit.Patients with VA (logMAR) worse than 1.0 (Snellen equivalent of 6/60) were examined using Sheridan Gardiner single letter and if failed continued to finger counting, hand movement, and perception of light which were converted to logMAR equivalent values as described by Lange et al. (23,24).

Spectral Domain-Optical Coherence Tomography (SD-OCT)
SD-OCT were obtained using the Heidelberg Spectralis (Heidelberg Engineering, Heidelberg, Germany) and Zeiss Cirrus (Carl-Zeiss Meditec, Dublin, CA, USA).The central subfield thickness (CST) was derived from the central subfield of the ETDRS grid (26,27).The Bruch's membrane (BM) marker was adjusted manually to correct segmentation errors.The residual thickness conversion from Cirrus to Spectralis was calculated using the formula from Sun, et al. (28).Hyperreflective foci (HF) are defined as discrete, focal, and hyperreflective lesions discovered on SD-OCT and recorded if present.HF can be found in several retinal disorders including age-related macular degeneration, diabetic retinopathy and inherited retinal dystrophy (IRD) (29,30).Epiretinal membranes (ERM) were classified according to the previous study by Govetto,et al. (31).
We selected published subjects that reported a variant classification of "Pathogenic," "Likely pathogenic," or "variant of uncertain significance (VOUS)." Published genetic variants were analyzed using Alamut Visual Plus (v1.6.1,SOPHiA Genetics, USA) software and classified according to functional studies (8) where possible.To date, not all variants have had functional studies to enable accurate pathogenicity classification.
A subanalysis of selected individual cases that reported at least one of the following ophthalmic clinical outcome measures: age of onset, visual acuity (VA), fundus description, fundus autofluorescence (FAF), optical coherence tomography (OCT), kinetic visual field (VF), and electroretinography (ERG) were documented in our form and were compared to our cohort data.VA data was converted to logMAR.The visual decline in IRD's is acknowledged to be a non-linear progression.In addition, where there is longitudinal follow-up analysis is impacted by differences in age, disease stage and genetic variant.
VF was reported as square degree (deg 2 ).This included the automated calculation from perimeters or manual calculation from Goldman kinetic perimetry.In those cases with horizontal visual field results only presented in degrees, an assumption was made due to the nature of the retinal dystrophy that the residual VF was circular.For comparison, the horizontal degrees were converted to square degree (deg 2 ) with formula of πr 2 (π = 3.14).

Novel IMPDH1 variant
Index female patient (II.3) presented with nyctalopia, leading to a diagnosis of retinitis pigmentosa at age 7 years.The probands affected sons developed nyctalopia and/or fundus abnormalities at an average age of 3.5 years of age, with the younger child also presenting with exotropia (Figure 3).
A second single heterozygous variant in USH2A gene, c.14753C>T (p.Thr4918Met) was also found in the mother (II.3) and her younger son (III.2) (Supplementary Table S2) but not in III.1.Segregation in the mother of II.3, (I.2), has identified this single variant without any ocular phenotype.

BCVA
The average of BCVA follow-up was 11.4 (SD 4) years with non-linear progression.The visual decline becomes apparent from age 5 years (Figure 4a).BCVA comparison between our cohort and the published studies was attempted.The most common variant reported Asp311Asn comprised 61 unique patients with 10 of them having 2 time points.Visual acuity changes with age for this variant compared to our novel variant is presented in Supplementary Figure S1.

Retinal imaging
Peripheral retinal atrophy with sparse pigmentary changes was identified early in the disease.The FAF showed a consistent pattern with a central area of macular hyperAF continued by ring of hypoAF outside vascular arcade there was blotchy areas of hypoAF which became more confluent with time (Figure 5).

SD-OCT
Ellipsoid zone (EZ) loss or disruption was found in all patients.The EZ was replaced by HF.Increased signal hypertransmission into the choroid was also found in all patients (Figure 5).The central subfield thickness (CST) plotted against age is shown on Figure 4b.Epiretinal membranes (ERM) were found in all patients (Figure 5).

Comprehensive literature search
The literature search identified 55 studies (1,3,5,19,32, reporting IMPDH1 variants in ophthalmic patients.These reports consisted of : 22 RP/RCD/Leber Congenital Amaurosis (LCA); 17 IRDs; 9 IMPDH1-associated retinopathies; 1 sectoral RP; 1 cone-rod dystrophy; 1 hereditary eye disease; 1 IRD and optic nerve disease; 1 pseudogene, 1 visual impairment; and 1 ARHGEF18retinopathy studies.As many as 36 studies reported IMPDH1 variants in their main text and 19 studies reported the variants in their supplementary materials.Comparison of the genomic data obtained from the literature search was challenging due to some  of the previous studies using older or alternative gene transcript references (i.e.NM_ number) during variant annotation.As part of our analysis, we added the variant annotation according to the NM_000883.4for the previously reported variants (Supplementary Table S1).
Including our variant, 56 heterozygous variants, 1 compound heterozygous, and 2 homozygous variant were reported (Supplemental Table S1) (1,5,40,19,32,37,38,(43)(44)(45)(46)(47)(49)(50)(51)(52)(53)(54)(55)59,61,62,73,78,82).In the heterozygous group there were 43 Missense, 3 in-frame, 1 nonsense, 2 Frameshift, 1 synonymous, and 6 intronic variants, respectively.For the compound heterozygous group, 1 allele was a missense variant while the other allele reported was a nonsense variant.For the homozygous group there were two missense variants reported.Exon 10 contained the largest number of variants with 18 listed, followed by exon 8 and 9 both containing 7 variants each.Variant p.Asp311Asn was the commonest reported featuring in 21 unique publications (Supplementary Table S1).We note that the Kennan et al. and Jordan et al. (51,52) studies were counted as one, as the former reported the variant while the latter described the clinical data (Supplementary Table S1).A summary variant map of IMPDH1 is presented in Figure 6.Six (6) studies reported additional variants in other IRD disease genes (Supplementary Table S2).Twenty-two (22) of 55 studies reported clinical data, and we compared the published studies to our cases (Table 2).Characteristics of these 22 studies are presented in Table 3.There were 99 patients with clinical data available from these 22 studies.Kennan et al. and Jordan et al. (51,52) report 16 patients' data descriptively without specifically pointing at each patient.Thus, their data were reported as 1 patient.Table 3 highlights that 89% (88) of the 99 patients have their variant in exon 10.This leaves too few patients with different variants in the other exons to undertake a reliable statistical analysis.

Discussion
IMPDH1 is ubiquitously expressed in many tissues of the human body, however pathogenic IMPDH1 variants are most commonly associated with isolated autosomal dominant RP.IMPDH1 is an essential enzyme in GTP synthesis (Figure 1a) and has higher expression in the retina than in other tissues (8).The process of phototransduction is highly energy dependent with specific demands for purine nucleotides (12,88,89).Thus, disruption of energy metabolism is likely to have serious consequence on retinal function.However, the exact disease mechanism remains unknown (8).
The novel p.Lys314Gln variant identified in our family was positioned in the CBS 2 domain of exon 10, the most commonly reported IMPDH1 location.To date, there are 17 reported variants located in exon 10 suggesting this is an IMPDH1 mutation hotspot (Supplementary Table S1 and Figure S1b).This exon 10 is also the location of 3 Class I IMPDH1 variants (p.Arg309Pro, p.Asp311Asn, p. Lys323Glu) which have been shown to result in defective response to GTP inhibition (Figure 1b) (8).The CBS domains are hypothesized to function as sensors of intracellular metabolites, where the disruption of this process involving IMPDH1 has been associated with ADRP (90,91).The CBS 1 and 2 domains assemble the Bateman domain, which is the binding target for adenosine and guanine nucleotides (13)(14)(15).IMPDH1-related retinal disease is potentially caused by the disruption of this regulation (13).Taken together, our variant may cause interference of GTP regulation affecting rod function.
There have been three reports in the literature of biallelic IMPDH1 disease.The first is a homozygous IMPDH1 p. Asp311Asn variant in a large consanguineous Pakistani family, where interestingly the heterozygous family members were reported to have had a normal phenotype (92).The p. Asp311Asn is the most common pathogenic variant reported in the literature, so the reason for this potential heterozygous rescue remains unknown and was not addressed by the authors (92).The second family described with biallelic IMPHD1 disease had compound heterozygous variants (60).Segregation and the phenotype of the parents were not available for this family.The third is homozygous p.Lys314Asn (65).However, the pathogenicity was not provided by the author.
The proband in our study is presented with the onset of disease within the first decade of life.This age of onset is similar to that described in the majority of published studies (1,32,44,50).In the published studies reporting the age of onset, the mean age was 15.6 yrs (SD = 14.2, n = 31) (Table 3).In the other 29 patients of previous studies, only descriptive age ranges were reported.In this group of patients, the onset was more commonly reported in the first decade.
BCVAs in our cohort had worsening trend from 5 years (yrs) of age (Figure 3).Older IMPDH1 patients do not always have poor BCVA.BCVA of 6/6 or 0.0 logMAR unit was reported in a 58-year-old male (53), while light perception (LP) was reported in a 65-year-old male (40).Wada et al. reported BCVA (logMAR) of 0.0 in a 41-year-old patient and 2.0 in a 34-year-old patient with the same common variant, p. Asp311Asn (32).On the other hand, Bennet et al. reported a p. Asp311Asn patient with a stable BCVA over 10 years of followup (37).These findings suggest that phenotypic variability exists among IMPDH1 patients even with the same variant.
The phenotypic variability in IMPDH1-associated retinopathy may be elucidated by the finding of 2 different classes in IMPDH1 variants (8).Class I has a defective GTP regulation, while Class II has GTP regulation similar to wild type.1), reflecting phenotypic variation.These Class II variants are yet to be shown to have an effect on the biochemical activity of IMPDH1, thus it is hypothesized that their effect on filament assembly may be their disease-causing pathway (8).
By further investigating the macular phenotype we identified a central area of hyperAF.This was associated with midperipheral hypoAF.This appearance was consistent across our family members and regularly reported in other IMPDH1associated retinopathy cases (37,49,53).These FAF findings are similar to those found in MERTK-retinopathy which has disruption in retinal pigment epithelium (RPE) function, resulting in both rod and cone photoreceptors degeneration in a relatively similar time course (94).The disruption of purine synthesis in IMPDH1-retinopathy may affect rod and cone photoreceptor over a comparable time course (8).The IMPDH1 FAF pattern differs from the classic macula ring of           hyperAF pattern found in RP which represents the border of damaged retina outside and healthier retina inside (95)(96)(97)(98).However, this hyperAF ring is also common in IMPDH1retinopathy cases (37) (Table 3) suggesting the phenotypic variability.Patients with worse BCVA can be found with foveal hypoAF indicating more advance macular damage (45,49,53) (Table 2).
Our SD-OCT examination found EZ progression from foveal EZ island to the total absence of EZ towards increasing age which reflects earlier rod involvement and in concordance to previous studies (37,44,49).Hyper-reflective foci found in our study were similar to Bennet et al. (37).Higher HF number is found to be correlated with higher aqueous flare, smaller visual field and thinner outer retinal layer in RP (99).HF also reflects vascular and metabolic alterations in this disease (100).Given the importance of HF in RP natural history and the development of HF counting automation (101) suggest a potential biomarker to be observed in RP cases.
Increased signal hypertransmission found in our study suggested RPE damage and is also reported in other IRDs and other conditions such as RPE detachment and choroidal neovascularization (102)(103)(104)(105). ERM was found in all our patients and has been linked to RP, suggesting disease etiology is contributed to by an inflammatory process (106).The effect of IMPDH1 variants on immune function has also been superficially studied using complete blood counts with differential platelet counts, however these results did not reveal any noticeable difference between carrier and non-carrier (wild type) family members (19).The effect of IMPDH1 variation in the immune system is another avenue for future investigation.However, it is likely that more specific biomarkers will be needed.
Severe constriction of the kinetic VFs to the level of legal blindness (<20deg equivalent to 314deg 2 ) was identified in all three patients.This was in concordance with previous IMPDH1 reports (40,43,44,50).Wada et al. cohort of p. Asp311Asn seems to have a relatively preserved average VF, 2586.4 (SD 2684.2) deg 2 from 21 patients (7-57 yrs) (32).However, their patients with early onset had a relatively poor VF (<500deg 2 ) while patients with later onset or asymptomatic had larger VF (>2000deg 2 ).On the other hand, Grover et al. cohort of p.Arg309Pro reported more severe VF with an average of 28 (SD 22.3) deg 2 from 7 patients (7-40 yrs) with <19.63deg 2 can be found in their 2 youngest patients (7 and 9 yrs).However, the oldest patient in that cohort (40 yrs) had better VF with 78.5deg 2 (50).The diverse phenotypic variability within IMPDH1-associated retinopathy is well described in the literature and our cases further add to this diversity.This diversity is also seen for variants within the same exon, including the two most common (p.Asp311Asn and p.Arg309Pro) and our variant all within exon 10.
The ffERG performed at diagnosis in our cases identified both rod and cone disturbances.Initially the rod system was more severely affected with early cone involvement.These findings were consistent with previous studies (19,37,43,44).Furthermore, our pERG also indicated the early macular dysfunction.The initial rod involvement might be explained by the mouse model ERG study which shows that the inhibition of IMPDH1 significantly delayed rod response recovery while not affecting the cone (13).However, photoreceptor energy metabolism disturbance in IMPDH1-associated retinopathy may explain the early cone involvement (8).
Treatment options are actively being studied for IMPDH1.A gene replacement approach using recombinant adenoassociated viral (rAAV) vectors prevents photoreceptor damage and preserves synaptic connectivity in a mouse model (107).Another approach being trialed is the use of ansamycin antibiotic which binds to heat shock protein Hsp90 and helps protect the photoreceptor against degeneration (108).
Given the retrospective nature of this study, we do not have more time points to reflect the natural history of our variant in the age gap between the mother (proband) and her children.Not all previous studies had the same multimodal imaging as our cohort, thus restricting the ability for accurate comparisons.In conclusion, we report 3 patients with IMPDH1-associated retinopathy from a family carrying a novel variant.On review of the literature, we identified 55 studies with 60 IMPDH1 variants in 99 patients and compared their phenotype with our cohort.Our work expands the genotype-phenotype understanding of IMPDH1 in relation to eye disease.Exon 10 is a hotspot of IMPDH1 gene and might be a suitable target for future therapeutics.IMPDH1-associated retinopathy frequently presents in the first decade of life with variable progression and phenotype.We highlight that IMPDH1-associated retinopathy has initial rod photoreceptor disturbance with early macular or cone involvement.Our patients with variant p.Lys314Gln reached legal blindness within the first 2 decades of life and consequently had an early window for therapy.We suggest that biomarkers are tightly correlated with patients' age and duration of disease, leading to an understanding for the optimum timing for therapy.

Figure 1 .
Figure 1.Purine biosynthesis pathway and IMPDH1 gene mapping with parallel protein domain.(a) IMPDH mechanism in the de novo pathway of generating GTP and ATP from glucose.IMPDH exists in 2 isoforms (IMPDH1 and IMPDH2).IMPDH1 is the isoform that is most prevalent in the retina.IMP will be bound by the catalytic domain (green) while the end products, ATP and GDP, will be bound by the regulatory domain (blue).(b) Variant p.Lys314Gln was found in our study and are written in italic and bold (black arrow).Variants located in regulatory domain/Bateman domain are written in blue and variants located in catalytic domain/TIM barrel/IMP binding domain are written in green.Burrell et al. classified IMPDH1 variants into 2 classes.Class I with GTP regulation defect and class II with GTP regulation similar to wild type (8). a Class I IMPDH1 mutation b Class II IMPDH1 mutation ADP=adenosisne diphosophosphate, AMP=adenosine monophosphate, ATP=adenosine triphosphate, CBS=cystathionine beta synthase, IMP=inosine-monophosphate, IMPDH=inosine monophosphate dehydrogenase, GDP=guanosine diphosphate, GMP=guanosine monophosphate, GTP=guanosine triphosphate, PRPP=phosphoribosylpyrophosphate, sAMP=succinyl AMP, TIM=triose-phosphate isomerase, XMP=xanthosine monophosphate.

Figure 2 .
Figure 2. Flowchart of comprehensive literature search.This flowchart shows the literature search in PubMed and LOVD and the selection to get 55 studies.

Figure 3 .
Figure 3. Pedigree of the one family with IMPDH1 associated retinopathy in our cohort.IMPDH1 variant p.Lys314Gln was found de novo in our index patient II.3.Segregation showed that her parents (I.1 & I.2) did not have this variant.

Figure 4 .
Figure 4. BCVA and CST plot per age.(a) BCVA in LogMAR is plotted against age in years (higher is worse).The trend of declining BCVA started at around 5 yrs of age.Improving BCVA before 5 yrs of age shows possible learning curve.(b) CST in micrometer is plotted against age in years.The CST trend is thinning through increasing age.BCVA = best corrected visual acuity, CST = central subfield thickness, VF = visual field.

Figure 6 .
Figure 6.Kinetic VF and ffERG of selected patients.(a) Kinetic VF of all patients.Tunnel vision was principal defect identified in these 3 patients.Thick red and blue lines represent normal VF.(b) ffERG of III.2 and III.1 showed almost extinguished DA and LA ERG with residual cone response.DA = dark adapted, ffERG = full-field electroretinography, LA = light adapted, VF = visual field.

Table 1 .
ACMG classification of IMPDH1 variants found in our cohort.
ACMG=American College of Medical Genetics, CBS = cystathionine beta synthase

Table 2 .
Clinical characteristics of our IMPDH1-associated retinopathy cases compared with published literature.

Table 3 .
younger than onset due to examined as family screening while asymptomatic and normal eye condition.A=annular, AO=age of onset, AP=age at presentation, BE=bull's eye, C=childhood, CC=cystic changes, C-FAF=central fundus autofluorescence, DRA=diffuse retinal atrophy, ↓DA ↓LA=reduced Demographic and patient characteristics from published IMPDH1 studies compared with this cohort.
#Age of presentation