Retinitis pigmentosa (RP) denotes a group of hereditary retinal dystrophies with a worldwide prevalence of approximately 1 in 4000. The disease is clinically and genetically very heterogeneous.¹ The phenotype of the juvenile form of RP shows overlap with Leber’s congenital amaurosis (LCA), an autosomal recessive retinal dystrophy with a worldwide prevalence of 1 in 35 000.² Patients with juvenile RP are considered to have good central vision during the first decade of life, while the age of onset of severe visual impairment in LCA ranges from birth to the first year of life.^3,4 Retinitis pigmentosa is characterised by night blindness, progressive constriction of the visual field, pale discs, retinal vascular narrowing, pigmentary changes in the retina, and reduced electroretinogram (ERG) amplitudes.⁵ LCA is characterised by congenital nystagmus, a decreased pupillary response, pigmentary changes in the retina (although the fundoscopic appearance is initially normal), and reduced ERG amplitudes.^2,6 Various intermediate phenotypes between LCA and RP are known and are sometimes described as “early onset severe rod-cone dystrophy” or “early onset retinal degeneration.”³ The clinical distinction between the subtypes of RP is not always clear, and the diagnostic criteria are frequently not agreed upon between ophthalmologists.⁷ In 29% of patients, RP is part of a syndrome, such as Usher’s or Bardet-Biedl’s.¹ Of the non-syndromal RP cases in the Netherlands, 37% is isolated (IRP); of the remainder, the transmission mode is autosomal recessive in 30%, autosomal dominant in 22%, and X linked in 10%.⁸ In addition, several unusual inheritance patterns—such as digenic inheritance⁹ and uniparental isodisomy,¹⁰ in which a child inherits two copies of a single parental chromosome—have been observed. LCA is most often inherited in autosomal recessive mode.²

The molecular genetic diversity within the different phenotypes of RP is also well documented.^3,4,11 At least 17 genes are currently known to be involved in autosomal recessive RP (ARRP), while five additional loci have been identified.¹² So far, mutations can only be identified in approximately 50% of ARRP and LCA patients. Eight LCA genes have been cloned so far: AIPL1,¹³CRB1,¹⁴CRX,¹⁵GUCY2D,¹⁶RDH12,¹⁷RPE65,¹⁸RPGRIP1,¹⁹ and TULP1.^20,21 The genetic overlap between ARRP and LCA is illustrated by the fact that all but three of these genes (AIPL1, CRX, and RPGRIP1) were previously implicated in ARRP (CRB1,^14,22GUCY2D,^16,23RDH12,^24,25RPE65,^4,18 and TULP1^26–28).

To illustrate the diversity and complexity of genotype–phenotype correlations in RP and LCA, two examples are mentioned here. Previously, a correlation was suggested between mutations in the RPE65 gene and a phenotype consisting of severe visual impairment within the first year of life, with a measurable visual acuity at the age of six to 10 years, and congenital nystagmus in three of four patients. However, no such correlation was observed in another study.^3,4 Similarly, two parents diagnosed with RP had a daughter who carried a homozygous RPE65 mutation and who was reported to have LCA (absent or severely diminished vision within the first year of life).⁴

In summary, there appears to be significant overlap between juvenile ARRP and LCA in both a clinical and a genetic sense. This makes the clinical distinction between juvenile ARRP and LCA difficult if not impossible.^7,12 We hypothesised that mutations in the five most common LCA genes could also be responsible for juvenile ARRP.^7,12,28,29 We therefore tested our ARRP, IRP, and LCA patients for mutations in the following “LCA” disease genes: AIPL1, CRB1, GUCY2D, RPE65, and RPGRIP1.

METHODS

Patients

This study was performed in agreement with the declaration of Helsinki. Patients included in the study or their legal representatives gave permission for blood to be drawn for mutation detection. All patients diagnosed with ARRP and IRP with a disease onset before age 20 years, as well as all patients with an LCA diagnosis, and presenting to the Netherlands Ophthalmic Research Institute (NORI) were included in the study. Most patients were Dutch, four were Belgian, one originated from Turkey, one had a Mexican and a Dutch parent, and one had an Indonesian and a Dutch parent.

The diagnoses of RP and LCA were based on a combination of the following criteria.

For RP, the criteria were: (subjective) night blindness, a pale optic disc, narrow arterioles, pigmentary changes, ERG amplitude reduction (>50%), and visual field abnormalities, compatible with RP.¹ Patients were considered to have juvenile RP when their first reported symptoms occurred before the age of 20.

For LCA, the criteria were: signs of blindness or severe visual impairment from birth or within the first year of life, an ERG reduction of more than 50%, and congenital nystagmus.¹

Patients were diagnosed with autosomal recessive inheritance when they had two unaffected parents and one or more affected siblings or consanguineous parents. Patients were diagnosed with IRP when no affected relatives were known, and there was no parental consanguinity, or no inheritance pattern could be established. Patients were excluded when their family tree showed evidence of autosomal dominant or X linked inheritance,¹ or when their RP or LCA was part of a syndrome.

Mutation detection

The DNA isolated from peripheral blood was stored at 4°C until use.³⁰ The following primers were used for the amplification of AIPL1: exon 1 forward (1F) cctggtcccctgtcttcttt, exon 1 reverse (1R) tgttgaaagctgctgtgggg, 2F ggccttgaacagtgtgtcta, 2R gagcccagaaaagactagtc, 3F ggccttttatggcccaccta, 3R ctgtccctctccagtgctgg, 4F tggggtccctgcctcactga, 4R ccccagagtcagcgccactt, 5F aagtggcgctgactctgggg, 5R tgtctccgtggccctgggct, 6F cctctgaggctgggaaggga, 6R gaccaggccacttgctccct. Primers used for CRB1, GUCY2D, RPE65, and RPGRIP1 along with their characteristics are referred to in table 1.

The length of the polymerase chain reaction (PCR) product was checked on a 2% agarose gel, after which it was mixed with an equal volume of PCR product from a healthy control person in order to detect homozygous sequence changes. This was followed by a heteroduplex formation step before samples were analysed for mutations using denaturing high performance liquid chromatography analysis (DHPLC) (Transgenomic, Santa Clara, California, USA) at two or three different temperatures.

Exons showing a DHPLC pattern that differed from a healthy control, and the healthy control itself, were reamplified and purified using a QIA-quick PCR purification kit (Qiagen, Valencia, California, USA) for direct sequencing. A sequence reaction (25 cycles consisting of 10 seconds at 96°C, five seconds at 50°C, and four minutes at 60°C) was carried out using forward and reverse primers separately (table 1). Sequence products were precipitated with 75% isopropanol and dissolved in template suppression reagent, after which the products were denatured at 94°C for two minutes. Samples were subsequently analysed using the ABI-310 (Applied Biosystems, Foster City, California, USA). We described mutations using the following Genbank reference cDNA sequences³²: AIPL1, gi6716706; CRB1, NM_012076.2; GUCY2D, NM_000180.1; RPGRIP1, NM_020366.2; RPE65, gi3077820; the first A of the ATG was chosen as nucleotide +1, and the ATG as codon +1.³³

Definition of pathogenicity

We considered a sequence variation to be pathogenic when we found it exclusively in patients and not in at least 120 chromosomes from 60 control persons. In addition, we made a literature based prediction of the effect of a new mutation, using the description of mutations in or the function of similar domains. We considered sequence variations polymorphic when their allele frequency was not significantly higher in patients than in control persons, or when they did not co-segregate with the disease phenotype in a family. Our patient group consisted of 35 patients and our control group of at least 60 healthy persons (120 chromosomes).

RESULTS

Patients

Our mutation screening panel consisted of 17 ARRP, nine LCA, and nine IRP patients. Table 2 shows the clinical characteristics of the 12 patients in whom we found mutations.

Visual field changes corresponding to RP were seen in ARRP, IRP, and LCA patients. As expected, an ERG reduction of at least 50% was reported in nearly all RP and LCA patients (data were missing for two patients). Nystagmus, corresponding to an LCA phenotype, was seen in all LCA patients and in one IRP patient. The RP and LCA patients, in whom no mutations were found, showed similar overlapping clinical features (data not shown).

Mutations

We identified nine different mutations in 12 patients (34%), in the CRB1 (11%), GUCY2D (11%), RPE65 (6%), and RPGRIP1 (6%) genes. In three of these patients two mutations were found (9%)—one patient had a homozygous mutation and two had a compound heterozygous mutation. In nine patients we detected a mutation in a single allele only. We found a mutation in three ARRP patients (17%), in four LCA patients (44%), and in five IRP patients (56%). Two of the mutations identified in this study are new, and we describe a new combination of two mutations. All mutations are presented in table 3.

The first new mutation identified in this study was a heterozygous deletion resulting in a frameshift in GUCY2D:c.3283delC, p.Pro1069ArgfsX37 in IRP patient 25556. This male patient had a disease onset at age 1.5 years. At age 13 he had a visual acuity of 0.1, a severely reduced ERG, a limited visual field, and nystagmus. Co-segregation of the deletion in the family could not be determined owing to the unavailability of parental DNA. However, the deletion was not found in our control population. We also identified GUCY2D mutations in ARRP patients 25855 and 8432 (p.Pro701Ser)³⁴ and in LCA patient 26669 (p.Arg768Trp).³⁵

The second new mutation was a deletion causing a frameshift in RPGRIP1:c1614_1623del, p.Glu538Glufs2 in IRP patient 25140. Visual problems in this patient were noted in the second year of life; she had a visual acuity of 0.1 at the age of 15, and experienced night blindness and colour vision impairment. Fundoscopy revealed an atrophic macular area, and the ERG was severely reduced. We did not detect this deletion in our control population. In addition, we found one known RPGRIP1 mutation in IRP patient 25474 (p.Asp1114Gly).³¹

We found a new combination of two mutations in CRB1 in ARRP patient 25378 ([c.2536A→T, p.Cys948Tyr]¹⁴+[c.2978G→A, p.Lys801X]²²). The age of onset was five years; at age 14, the visual acuity was finger counting, and subjective night blindness was reported. Earlier data on visual acuity and night blindness were missing. The ERG was severely reduced and the fundoscopic picture showed narrow arterioles, a pale optic disc, and pigmentary changes in the periphery. CRB1 mutations were also found in LCA patient 22802 and in IRP patients 25718 and 26425 (p.Cys948Tyr).¹⁴

In the RPE65 gene we found mutations in LCA patient 9359 ([p.Tyr368His]+[c.11+5G→A])^35,36 and in LCA patient 24670 (p.Tyr368His homozygously). We confirm the pathogenic nature of the RPE65 c.11+5G→A sequence change. It is located in a donor splice site where the G nucleotide is predicted to be conserved in 84% of primate splice sites.^35–37. In addition, we do not find it in our controls.³⁸

Polymorphic sequence changes

We observed seven new and 15 known polymorphisms in all five genes. In addition, we established or confirmed the polymorphic nature of three sequence changes previously described as pathogenic mutations. Table 4 summarises the polymorphisms along with their allele frequency in our patient population.

We identified seven new polymorphic sequence changes which are described below. Three new intronic polymorphisms were located in the CRB1 gene (c.283-35T→C, c.330+34C→T, and c.330-54C→A), and two new silent polymorphisms were found in GUCY2D (c.814C→T, p.His247His and c.2011C→T, p.Leu646Leu). In RPGRIP1 a new insertion (c.302-16_-15insATA), and a new deletion (c.907-153_-154delGG) was found. Bioinformatic analysis of all intronic sequence changes did not lead to changes in the splice sites. Analysis of the silent polymorphisms did not reveal predicted changes in exonic splicing enhancers.^41,42

We identified three sequence changes of uncertain pathogenic nature. In published reports, these are tentatively described as polymorphisms. The first sequence change was AIPL1:c.286G→A, p.Val96Ile,¹³ with an allele frequency in patients of 1%, and in matched control persons of 7% (n = 122 chromosomes). The second sequence change, CRB1:c.6147T→C, p.Ile205Thr,⁴³ was found only once in our patient population (allele frequency 1%). We did not find it in our controls. However, we found no co-segregation with the disease phenotype in ARRP family 21724 (fig 1). The third sequence change we identified was RPGRIP1:c.1639G→T, p.Ala547Ser⁴⁴ with similar allele frequencies of 14% in patients and 13% in controls. Interestingly, in three patients with that Ala547Ser variation, we detected an additional heterozygous pathogenic mutation: (GUCY2D:p.Pro701Ser (ARRP patient 25855), RPGRIP1:p.Asp1114Gly (IRP patient 25474). IRP patient 25140, as well as her unaffected father, had the RPGRIP1:p.Ala547Ser variation in combination with the RPGRIP1:p.Glu538Glufs2 mutation. The latter indicates that the p.Ala547Ser sequence change in combination with the p.Glu538Glufs2 mutation is not sufficient to lead to an RP phenotype.

DISCUSSION

Both clinically and genetically, there was a considerable overlap between our RP patients and our LCA patients. Clinically, visual field changes, as seen in RP, were present in RP patients and in two LCA patients. Nystagmus, which is frequently seen in LCA, was observed in LCA patients as well as in an IRP patient. In addition, fundoscopic changes were seen in ARRP, IRP, and LCA patients.

In summary, CRB1 and GUCY2D mutations were found in RP patients as well as in LCA patients, RPE65 mutations were found only in LCA patients, and RPGRIP1 mutations, previously only described in LCA, were newly identified in two RP patients.

Mutations: frequency and distribution

In this study we found nine mutations in 34% of our 35 patients. We identified mutations in 56% of our LCA patients, which corresponds well to published reports (50%),^28,29 although in contrast to the current study, most other studies did not screen for all known genes at once. In 27% of our ARRP patients we found one or two mutations, which is less than previously reported.¹² This may reflect the fact that we screened our patients for LCA gene mutations, and not for all the known RP gene mutations. Screening our ARRP patients for additional known RP gene mutations will most probably lead to the detection of several more mutations.

In spite of the recessive nature of the disease, a second disease causing mutation was not found in nine of 12 patients. The design of our DHPLC analysis (sensitivity close to 99%) makes it unlikely that a second exonic mutation in one of the screened genes was left undetected.^47,48 Obviously, it remains possible that promoter or regulatory sequences of the screened genes contain as yet unidentified mutations.

The spectrum of mutations in our study (CRB1 (11%), GUCY2D (11%), RPE65 (6%), and RPGRIP1 (6%)) shows only slight differences compared to those reported in the literature. This indicates that the gene distribution in our cohort is similar to juvenile RP and LCA populations worldwide.^{13,14,16,19,18}

New mutations

Two new mutations were identified in this study. The first is the first the mutation reported in the intracellular C-terminal domain of GUCY2D:c.3283delC, p.Pro1069ArgfsX37, in IRP patient 25556. The deletion creates a stop codon at position 1106 of the cDNA, which results in a larger, most likely unstable, mRNA or a dysfunctional protein. GUCY2D is a well conserved protein, containing an extracellular ANF receptor domain, a transmembrane domain, an intracellular protein kinase homology domain, and an intracellular guanylate cyclase domain.⁴⁹ Mutations have been described in all these domains,^16,34 except for the intracellular C-terminal domain. The function of the latter domain is unknown. GUCY2D is one of the enzymes indirectly responsible for the regulation of phototransduction in the photoreceptor cells through regulation of cGMP levels.⁵⁰ An altered GUCY2D protein could disturb this process, leading to an RP phenotype, as seen in this IRP patient. Patient 25556 had an age of onset and visual field changes compatible with an RP phenotype. However, the lack of fundoscopic changes and the presence of nystagmus indicate a phenotypic overlap with LCA.

The second new mutation identified in this study is RPGRIP1:c.1614_1623del, p.Glu538Glufs2 in IRP patient 25140. A stop codon is introduced at position 540 of the cDNA, which will prevent the transcription of the protein kinase C conserved region 2 (C2) and the RPGR binding domain.⁵¹ The RPGRIP1 protein is localised in the connecting cilia of human cone and rod photoreceptors that connect inner and outer segments, where it binds RPGR to the cilium.⁵¹ RPGR plays a role in the maintenance of photoreceptor viability.⁵²RPGR mutations were previously found in X linked RP, and RPGRIP1 mutations are known to cause LCA.^19,31 Therefore we consider this RPGRIP1 deletion likely to be involved in RP, even though a second mutation was not found in this patient (25140). Patient 25140 had a constricted visual field, a visual acuity at age 15 of 0.1, and no nystagmus, which goes against the diagnosis of LCA. However, the absence of fundoscopic changes, the relatively early onset (1.5 years), and the presence of colour vision defects are more typical of an LCA related phenotype. This patient is a typical example of the phenotypic overlap that exists between juvenile RP and LCA.⁷

Polymorphic sequence changes

We established or confirmed the polymorphic nature of three sequence changes previously reported as pathogenic mutations. The first (AIPL1:c.286G→A, p.Val96Ile) was previously described as a disease causing mutation.¹³ We consider it to be a rare neutral polymorphic change because of its high allele frequency (7%) in our control subjects. The second sequence change we report to be polymorphic (CRB1:c.6147T→C, p.Ile205Thr) was initially described as a disease causing mutation in a Spanish ARRP family.⁴³ More recently it was reported in a Dutch study as a polymorphic sequence change.⁵³ We find it heterozygously in a single family where is does not co-segregate with the disease phenotype (fig 1). Moreover, the change is absent in 192 control chromosomes, which lead us to suggest that Ile205Thr is a rare neutral sequence change. The pathogenic nature of the third sequence change (RPGRIP1:c.1639G→T, p.Ala547Ser) is unclear in published reports. It was described as a polymorphism,¹⁹ and a disease causing mutation.⁴⁴ The high frequency of this sequence change in our control group (13%) suggests RPGRIP1:p.Ala547Ser is indeed a neutral polymorphism. We found it heterozygously along with a homozygous disease causing RPE65 mutation in LCA patient 9359. Furthermore, the presence of Ala547Ser in the unaffected parents of LCA patient 9359 and IRP patient 25140, along with other heterozygous mutations, suggests it does not cause a disease phenotype.

Conclusions

We observed a large overlap in the mutation spectrum and the clinical signs in the juvenile ARRP, LCA, and juvenile IRP patients, both in our study and in the literature.^{4,7,14,22,34,36,54} Our clinical data, combined with the mutation spectrum found in our patients, do not provide hard evidence that a clear distinction between juvenile RP and LCA can be made. We expect that eventually more clarity will come from molecular genetic analysis of the underlying gene defects.