Subjects were recruited from either an ongoing study on vitamin D status in Danish albino patients or from the national low-vision clinic. Patients were identified among individuals with a medical record at the former National Eye Clinic for the Visually Impaired (NEC). This clinic has maintained an updated and compulsory register on children with reduced visual acuity (VA) = 6/18 (0.3) or lower. The diagnostic criteria for albinism included nystagmus, reduced visual acuity, iris translucency, fundus hypopigmentation, and foveal hypoplasia. Crossed asymmetry of the retinocortical projections (misrouting) was used as an additional sign if present. Most patients had one or more eye examinations at NEC including Goldmann perimetry, fundus photography, and visual evoked potential (VEP) recording. For VEP recordings, we applied both a stimulator (PS 22C; Telefactor-Grass Instruments Co., Quincy, MA) with a flash lamp (PST2100) for flash stimulation and a video pattern stimulator (VPS 10S; both from Braintronics, Almere, The Netherlands) for measurements of cortical responses to monocular and binocular on/off checkerboard size 4 stimuli subtending an angle of 0.86° at 100 cm distance from a monitor. Two active silver disc electrodes (Oxford Instruments, New York, NY) were placed over the occipital lobes 3 cm to the right and left to an electrode placed at the inion. The reference was a gold ear-clip electrode (Viayss Healthcare, Madison, WI). For recordings, an averager (Viking IV Nicolet Corp., Madison, WI) was used. Crossed asymmetry was considered present when subtraction of recordings from the right and left occipital lobe of right and left eye stimuli, respectively, displayed significant phase differences. Information on skin and hair pigmentation was obtained either by a standardized examination by a dermatologist or patients were asked to send in color photographs with a questionnaire on pigmentation and tanning. In addition, the patient files at NEC were scrutinized. 

We compiled data on six skin and hair parameters (skin color, pigmentation pattern [Fitzpatrick types], color of scalp hair, eyebrows, and eye lashes, freckles, and birthmarks) and eight eye and vision parameters (photophobia, visual acuity with best spectacle correction, refractive value, nystagmus, iris translucency, fundus pigmentation, foveal hypoplasia, and misrouting of the optical pathways). However, all data were not attainable in all patients. We therefore decided to use only semiquantitative phenotype analyses to classify patients as either OCA or AROA. X-linked OA in males were excluded based on either OA1 mutation analysis or the presence of maternal retinal carrier signs or both. 

The project was performed according to the Declaration of Helsinki and approved by the Regional Ethics Committee. Written and verbal consent were obtained from all participants before inclusion. 

DNA was extracted from blood using a standard salting out method. ¹⁹ Genomic DNA was used as a template for PCR amplification of coding exons plus 20 bp of adjacent intron sequences as well as 20 bp of 5′UTR and 3′UTR sequences. Primer sequences are listed in Supplementary Table S1, and PCR conditions are available on request. Mutational analysis was performed by denaturing high-performance liquid chromatography (dHPLC, Varian Inc., Palo Alto, CA), at two temperatures. Fragments with aberrant chromatograms were sequenced using the same primers as for dHPLC analysis and dye terminator chemistry (BigDye v3.1; Applied Biosystems, Inc. [ABI], Foster City, CA) with analysis (ABI3100 prism; ABI), according to the manufacturers’ directions. All single nucleotide changes were confirmed on a new PCR product. The presence of novel missense mutations were tested in a group (n = 50) of healthy ethnically matched control subjects. Heterozygous patients were systematically resequenced, to ensure that the dHPLC screening had not missed mutations. 

Seventeen patients had only one mutation identified, and 11 of those agreed to have a second blood sample taken for establishing a cell line. Nested PCR was performed to test for biallelic expression in patients in whom only one mutation was identified. Total RNA was extracted from Epstein Barr Virus-transformed lymphocytes, with a mini kit (RNeasy; Qiagen, Hilden, Germany), according to the manufacturer’s instructions. Approximately 10 μg RNA was treated with 10 units amplification grade DNaseI (Invitrogen, Carlsbad, CA), according to the manufacturer’s instructions. One microgram of total RNA was reverse transcribed with random hexamer primers and reverse transcriptase (SuperscriptIII; Invitrogen), according to the manufacturer’s instructions. Two microliters cDNA were used as a template for the first-round PCR (AmpliTaq gold enzyme; ABI). One microliter of the first-round PCR was used as the template for a second PCR. Primers and PCR conditions are available on request. 

A total of 218 patients, 108 males and 110 females with OCA or AROA, born from 1961 to 2005 were identified (Table 1) . During the same period 3,018,248 live children were born. The corresponding prevalence of any kind of autosomal recessive albinism in the Danish population was approximately 1 in 14,000 live births. The categorization as either OCA (55%) or AROA (45%) was nearly equal in the material as a whole and in both sexes (Table 1B) . 

Sixty-two unrelated individuals participated in the mutation screening. The patients were mainly of Danish origin with some exceptions: Four individuals were from the Middle East, one was from Bosnia, and four were from the Faroe Islands. Two patients had Polish/Danish and Greenlandic Inuit/Danish parents. The age at inclusion in the study was from 1.5 months to 74 years (median, 19 years). Forty-one individuals with X-linked OA born 1965 to 2005 were excluded. 

Visual acuity (VA) ranged from 0.8 to 0.05 (median, 0.25) in 59 of 62 individuals (Table 2) . Refractive values expressed as spherical equivalent values of the right eye had a median value of +4.00 (range, +7.00 to −11.25; 0.1-fractile, +0.50; and 0.9-fractile, −3.00). Astigmatisms were prevalent among the participants (median, −2.50; range, 0.00 to −5.00) mainly in horizontal axes (Table 2) . Nystagmus was present in 58 of 62 probands. VA in the four individuals without nystagmus was only moderately reduced (between 0.8 and 0.5). All four had conspicuous (n = 3) or moderate (n = 1) iris translucency, and in three examined cases misrouting of the retinocortical pathways was verified. Iris translucency was severe in 34 participants and moderate in 21 (Table 2) . Six of the 62 cases had no iris translucency, among which 3 had severe and 2 showed moderate photophobia, whereas one proband experienced no such problems. In one case, information was missing. In 56 individuals, the retina had an albinoid appearance comprising a central pigmented zone and a distinct transition into a sparsely pigmented periphery with exposure of the choroidal vessels. Clinical information on the foveal region was missing in 26 cases. Among the remaining 36, only 2 had normal foveal reflexes, whereas the remaining 34 showed foveal hypoplasia. Most of the probands exhibited small and slightly dysmorphic optic nerve heads. 

Based on clinical characteristics of mainly hair color and complexion, the patients in the mutation study were classified as either OCA (n = 41) or AROA (n = 21). Among patients with OCA, 33 were sporadic cases. Also cases of AROA occurred mostly sporadically, but five patients (33, 34, 35, 36, and 38) came from families with direct vertical transmission—assumedly pseudodominant inheritance. Ocular symptoms showed no obvious difference between the OCA and the AROA groups except for iris translucency, which was missing in four patients with AROA but in only two with OCA. In another three patients with AROA, the ocular fundus showed normal pigmentation, whereas normal fundus pigmentation was observed in two with OCA. 

Results of the mutation analyses of TYR and OCA2 are shown in Table 3 . Sixteen patients were found to be homozygous or compound heterozygous for mutations in the TYR gene. Seven changes were novel. Three of the patients had a mutation in OCA2, in addition to the two changes in TYR. Nine patients were homozygous or compound heterozygous for mutations in OCA2; one of these also had a mutation in TYR. Seven of the OCA2 mutations were novel. In 10 and 7 patients, only one mutation was identified in TYR and OCA2, respectively. A single patient had one mutation in TYR and one mutation in OCA2. Two patients had mutations in MATP; these two patients were of Iraqi and Danish origin, respectively. One was homozygous for a novel missense mutation (c.1147G>A, p.G416E), and the other was compound heterozygous for two splice site mutations, c.386-1G>A and c.386-2A>G, the latter being novel. No alterations were found in TYRP1. In 17 patients, no mutations were identified. Fifteen of those were screened for mutations in SLC24A5, which revealed one sequence variation, c.1079-4A>T, in one patient. 

The majority of the mutations were missense. A total of 44 mutations were found in TYR; of those 34 (77%) were missense mutations, whereas 10 (23%) were nonsense, splice site mutations, deletions, frameshift mutations, or in the start codon. For OCA2 a total of 29 mutations were found, of which 26 (90%) were missense mutations and 3 (10%) were frameshift mutations resulting in truncation of the protein. 

Five families showed direct vertical transmission, and mutations were identified in three of these (patients 33, 34, and 38; Fig. 1 ). In one family (index patient 34) two mutations were detected in TYR and one mutation in OCA2. In the affected son we found one mutation in TYR and one mutation in OCA2 in addition to two polymorphisms in OCA2; however, we could not fully explain the OCA in the son. In another family, the index patient 33 was heterozygous and expressed only the mutant allele, and thus a second unidentified mutation preventing expression is thought to be present on the second allele. Both children of patient 33 were affected and were shown to have inherited the c.1327G>A mutation; however, sequencing of OCA2 in the daughter revealed no further mutation. We did not have the opportunity to perform allele expression analysis in the daughter. Patient 38 was homozygous for a mutation in OCA2 (c.1243T>C), but no family members were available for further testing. We thus have no proof of true dominant inheritance in families with vertical transmission, nor was pseudodominant inheritance convincingly established. 

RT-PCR analysis was performed in 11 of the 17 patients in which only one mutation was identified, and it showed four patients with exclusive expression of the mutant allele (patients 2, 13, 33, and 63), whereas six patients expressed only the wild-type allele (patients 3, 30, 44, 47, 53, and 59). Patient 15 showed expression of both the mutant and the wild-type allele (Table 3) . 

Segregation analysis showed random parental inheritance in cases with one mutation—that is, there was no difference in paternal versus maternal transmission of the mutation. 

Among patients classified as OCA, one or two mutations were identified in 85% (35/41) as opposed to the AROA group in which one or two mutations were found in only 48% (10/21). In the AROA group, mutations in both alleles were identified in five patients (two TYR and three OCA2) supporting the notion that some of these milder cases may be due to mutations in known OCA genes. Of the 11 patients with AROA with no identified mutation, 5 were female and 6 were male. Despite a tendency to a more severe phenotype among individuals with TYR mutations, considerable overlap of skin color, hair color, and Fitzpatrick tanning pattern were present within the TYR and OCA2-subgroups and the ocular parameters were uninformative with respect to mutational background. The only exception was a clear preponderance of severe photophobia among patients with TYR mutations. 

Most of the reported patients were examined at NEC, indicating a high diagnostic reliability. Our prevalence estimate was based on a compulsory low-vision register at NEC receiving information both from ophthalmologists and teachers in the country. Children with only mild visual impairment may have escaped notice. The completeness of the registration is unknown. However, there was a bias, because the compulsory registration at NEC requires a VA of 0.3 or less. Among 12 participants with VA better than 0.3 (Table 2) , two were recruited from the vitamin D study. The remaining 10 children were referred for diagnostic reasons due to the presence of nystagmus. Accordingly, the estimated birth prevalence of albinism 1:14,000 should be considered as a minimum figure. It is noteworthy that Sjostrom et al. ²⁰ in a cohort study of Swedish schoolchildren with subnormal acuity reported exceedingly high albinism prevalence rates approximately 100 times higher than our estimate. ²⁰ In a similar study among Mexican children the same group reported a somewhat lower but still markedly high prevalence. ²¹  

The high proportion of autosomal recessive OA in the population may be explained by the use of VEP examination in all patients with congenital nystagmus. When misrouting was established and the skin and hair pigmentation did not depart from a normal pattern, a diagnosis of AROA was confirmed. In a Nordic population such as the Danish the categorization as either OCA or AROA is often arbitrary, because of the abundance of fair skin and hair in the general population, and children without any ocular signs of albinism with white hair and some iris translucency are quite common. In cases of doubt, we were guided by information on pigmentation in the sibs or parents when these were at the same age. It cannot be excluded that some overdiagnosis of AROA in OCA cases was made for the above-mentioned reasons. The even sex distribution among cases with OA gives indirect evidence of an effective exclusion of X-linked cases. 

Systematic mutational analysis of the four known OCA genes showed that mutations in TYR are the major cause of OCA/AROA in Denmark, contributing 26%, whereas mutations in OCA2 account for 15%. Mutations in MATP are only a minor cause of OCA in this population, and no mutations were found in TYRP1. We could explain the OCA based on the finding of two mutations in 27 (44%) of 62 patients. It should be noted that some mutations may not have been detected due to the difficulties in detecting homozygous variants using dHPLC. 

Six of 10 mutations found in the AROA group have also been reported for OCA patients, and three patients were compound heterozygous for mutations previously reported to cause OCA. An explanation could be an overdiagnosis of AROA in OCA cases, as previously mentioned, or that other pigmentation genes modify the phenotype. 

Six missense mutations in TYR and six in OCA2 were novel. All mutations were conserved during evolution and were not found in 50 healthy, unrelated control individuals. Furthermore, in silico analysis using SIFT (http://blocks.fhcrc.org/sift/SIFT.html; provided in the public domain by the Fred Hutchinson Cancer Research Center) predicted all mutations except L452V in TYR to be pathogenic. It is difficult to conclude whether the L452V mutation is pathogenic (no family members were available for analysis); however, we would expect the other 11 novel missense mutations to be pathogenic. 

As in previous studies, mutational analysis of the four known OCA genes renders a substantial proportion of patients without identified mutations. In our study, this was especially evident in the AROA group, where 11 of 21 did not have any mutations identified. This indicates that other albino genes exist that may account for a less severe phenotype. A candidate gene is SLC24A5. Mutations in golden, the Zebrafish orthologue of SLC24A5, cause hypopigmentation of skin and retina, and a single-nucleotide polymorphism (SNP) of SLC24A5 (111Ala/Thr) has been shown to be associated with lighter complexion in African Americans and African Caribbeans. ¹⁶ Furthermore, recent studies have shown that mutations in the mouse gene Slc24a5 cause OA. ¹⁸ However, we did not find any pathogenic sequence variations in SLC24A5. 

Similar to previous investigators, we also found a high fraction of heterozygotes, both for TYR and OCA2, which cannot be explained by the carrier frequency alone. ^{22 23} Recently, deletions in the 3′ region of TYR including exon 4 and 5 have been disclosed. ²⁴ This mutation would not have been detected by our mutational screening strategy. Other explanations for missing mutations could be mutations abolishing normal splicing, mutations in the regulatory regions, or large deletions in the gene. RT-PCR analysis indicated a second unidentified mutation, possibly in the regulatory regions, abolishing expression of the apparently normal allele in four patients. Six patients showed exclusive expression of the wild-type allele. One of those has a frameshift mutation introducing a premature stop codon. Therefore, this allele is most likely degraded by nonsense-mediated decay (NMD), which is a cellular mechanism that specifically degrades mRNAs containing premature stop codons. ²⁵ The remaining five heterozygous patients with only wild-type allele expression harbor missense mutations, and expression of the mutant allele was expected. However, recent studies show that nucleotide substitutions originally predicted to cause missense mutations or synonymous substitutions may affect splicing if they are located in exon splicing enhancer (ESE) or exon splice silencer (ESS) sequences. ^{26 27} In silico analyses using the ESE finder (http://rulai.cshl.edu/cgi-bin/tools/ESE/ provided in the public domain by the Zhang Lab, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) and RESCUE-ESE (http://genes.mit.edu/burgelab/rescue-ese/ provided in the public domain by the Burge Laboratory, Massachusetts Institute of Technology, Cambridge, MA), ²⁸ show that the missense mutations create/abolish ESE sites. Even though the results are not in total agreement, it is possible that some of the mutations cause splicing defects leading to NMD and consequently to no expression of the mutant allele. However, this theory is purely speculative, and RT-PCR analysis would have to be performed to validate it. 

Of note, one mutation in OCA2, c.1465A>G, was found in two patients; in one patient only the wild-type product could be detected while both the wild-type and the mutant allele were detected in the other patient. The observed discrepancy in these two patients could be due to the presence of other sequence variations. ²⁹  

Another striking observation is that three of the four patients with exclusive expression of the mutant allele had AROA, whereas all six patients with exclusive expression of the wild-type allele had OCA. An explanation for this observation could be residual enzyme activity of the mutant protein causing the milder OA phenotype, whereas the more severe OCA phenotype might be unrelated to the heterozygous mutation, but due to mutations in an as yet unknown OCA gene. However, since no enzyme activity has been measured of the mutant proteins, this has not been proven. 

Another possible explanation for the high proportion of heterozygotes is epigenetic effects. To explore whether epigenetic effects could be involved in the pathogenesis of OCA, we investigated from which parent the heterozygous patients had inherited the mutation. However, the inheritance was random, thus excluding epigenetic effects due to parent-of-origin effects. 

In a recent study, Hutton and Spritz ³⁰ investigated 36 patients with AROA and found 9 (25%), 25 (69.5%), and 2 (5.5%) with no, one, and two mutations, respectively, in TYR, OCA2, or TYRP1, further supporting the notion that autosomal recessive OA can be caused by mutations in known OCA genes. Our study confirmed this finding but in a different distribution of mutations, with 11 (52%), 5 (24%), and 5 (24%) patients with no, one, and two mutations, respectively. Furthermore, Hutton and Spritz ³⁰ report that all 20 patients heterozygous for a TYR mutation also had the common polymorphism c.1205G>A. In our study, three patients with AROA were heterozygous for a TYR mutation, and one of those had the c.1205G>A polymorphism. In a HapMap European population, the distribution of A/A, A/G, and G/G genotypes was found to be 0.050, 0.333, and 0.617, respectively (dbSNP available at http://www.ncbi.nlm.nih.gov/SNP/ rs1126809; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD), and the distribution in our AROA group is 0.048, 0.381, and 0.571 and thus not different from the normal population. Hence, we do not think that the polymorphism contributes to the albino phenotype. 

In conclusion, we have shown that mutations in TYR and OCA2 can be the genetic cause of both OCA and AROA. Furthermore, our study shows that a mutational analysis of all four known OCA genes can explain only a fraction of the albino phenotypes supporting the hypothesis that there may be other not yet identified genes that cause AROA and probably also OCA. 

 

Supplementary Table S1 - (PDF) 

The authors thank the Danish Albino Society and the families for their participation and Donna Arbuckle, Bente Tøt, and Anette Riis Madsen for technical support and sample collection.