The microphthalmia-associated transcription factor (Mitf) gene and its role in regulating eye function

Mutations in the microphthalmia-associated transcription factor (Mitf) gene can cause retinal pigment epithelium (RPE) and retinal dysfunction and degeneration. We examined retinal and RPE structure and function in 3 month old mice homo- or heterozygous or compound heterozygous for different Mitf mutations (Mitfmi-vga9/+, Mitfmi-enu22(398)/Mitfmi-enu22(398), MitfMi-Wh/+ and MitfMi-Wh/Mitfmi) which all have normal eye size with apparently normal eye pigmentation. Here we show that their vision and retinal structures are differentially affected. Hypopigmentation was evident in all the mutants while bright-field fundus images showed yellow spots with non-pigmented areas in the Mitfmi-vga9/+ mice. MitfMi-Wh/+ and MitfMi-Wh/Mitfmi mice showed large non-pigmented areas. Fluorescent angiography (FA) of all mutants except Mitfmi-vga9/+ mice showed hyperfluorescent areas, whereas FA from both Mitf-Mi-Wh/+ and MitfMi-Wh/Mitfmi mice showed reduced capillary network as well as hyperfluorescent areas. Electroretinogram (ERG) recordings show that MitfMi-Wh/+ and MitfMi-Wh/Mitfmi mice are severely impaired functionally whereas the scotopic and photopic ERG responses of Mitfmi-vga9/+ and Mitfmi-enu22(398)/Mitfmi-enu22(398) mice were not significantly different from wild type mice. Histological sections demonstrated that the outer retinal layers were absent from the MitfMi-Wh/+ and MitfMi-Wh/Mitfmi blind mutants. Our results show that Mitf mutations affect eye function, even in the heterozygous condition and that the alleles studied can be arranged in an allelic series in this respect.


Results
Mice used in the study. The Mitf mutations studied here have all been described previously 32 and are listed in Table 1. Heterozygotes carrying Mitf mi-vga9/+ and Mitf -Mi-Wh/+ mutations have normal eye size: although the former has normal coat pigmentation, the latter have grey diluted coat colour ( Fig. 1; Table 1) 32,47 . Mice homozygous for the Mitf mi-enu22(398) mutation have a black coat with a white belly and unpigmented spots over the rest of the body, and dark ruby eyes of normal size. The Mitf Mi-Wh /Mitf mi mice have a white coat whereas eye size is normal (Fig. 1). These particular mutations, in heterozygous, homozygous or compound heterozygous conditions, were selected for study as they all have apparently normal eye development. After application of the mydriatic tropicamide, the irises of all mutants fully dilate, except the Mitf Mi-Wh /Mitf mi mice, which show an iris that is less pigmented than that of wild type mice, and which dilates only partially with tropicamide, possibly due to hypoplasia (Fig. 1).

Varying hypopigmentation in Mitf mutant eyes. Representative bright field and fluorescent fundus
images obtained with a rodent fundus camera from the eyes of wild type (n = 6) and mutant mice are shown in Fig. 2. As demonstrated by the bright field images (Fig. 2, upper panel) the optic disk areas were of normal size in all mutants, further indicating that the eyes of these mutant mice develop to a normal size in all cases. All the mutant mice show some hypopigmentation in their fundi, varying in degree. The fundi of the Mitf mi-vga9/+ mice (n = 6) show discrete yellow lesions, with only minor hypopigmentation overall, but discreet pigment mottling scattered throughout the entire fundus. The bright field fundus images from Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6) reveal large unpigmented lesions in both eyes, usually in the superior half of the fundus, with irregular borders. The fundi of the Mitf -Mi-Wh/+ mice (n = 6) also show hypopigmentation and comparable large non-pigmented www.nature.com/scientificreports www.nature.com/scientificreports/ areas, although more extensive, but no pigment mottling. The fundus images of the Mitf Mi-Wh /Mitf mi mice (n = 6) show more widespread lack of pigmentation and larger apparent RPE lesions, without any pigment mottling.
The fluorescent angiography (FA) images from wild type (n = 6) and Mitf mi-vga9/+ mice (n = 6) showed normal retinal vasculature with no hyperfluorescent signals, and extensive capillaries emanating from the major vessels (Fig. 2, lower panel). The FA of the Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6) showed hyperfluorescent areas where there are hypopigmented lesions in the superior half of the fundus, but with capillaries present. However, FA of the Mitf Mi-Wh/+ mice show hyperfluorescent regions across the entire fundus, and a clear reduction in the capillary network. Mitf Mi-Wh /Mitf mi mice showed hardly any fluorescence in their retinal vasculature in FA, but extensive hyperfluorescent signals that appear to emanate from the choroidal vasculature over nearly the entire fundus. These hyperfluorescent signals correspond in area to the hypopigmented lesions evident in the bright field fundus images.
Mitf has an impact on visual function. ERG recordings were obtained from wild type and mutant mice.
Dark-adapted ERG responses evoked by the 1.87 log cd sec/m 2 stimulus are presented in Fig. 3a-c. Two of the Mitf mutant mice, Mitf -Mi-Wh/+ and Mitf Mi-Wh /Mitf mi showed no detectable dark-adapted ERG responses above baseline (Fig. 3a), suggesting that they have severely impaired retinal function. The dark-adapted ERG responses of these mice were all less than could be detected above baseline regardless of stimulus luminance (data not shown), while the other two mutant mice (Mitf mi-vga9/+ and Mitf mi-enu22(398) /Mitf mi-enu22(398) ) showed dark-adapted ERG a-and b-waves that were not significantly different in amplitude (Fig. 3d,f) or implicit time (Fig. 3e,g) from wild type at any of the levels of the stimulus luminance tested (P > 0.05). The mean amplitude of the dark-adapted a-wave of the ERG in response to the 1.87 log cd sec/m 2 stimulus was 189.1 ± 30.9 µV in wild type (n = 6), 183.4 ± 22.4 µV in Figure 1. Phenotype of the mice used in this study. Coat appearance of mice examined in this study (upper panel). Representative images of the anterior part of the eye after pupil dilatation (n = 6) (lower panel). All eyes are of normal size. The overall appearance of the eyes of all mutants, except the Mitf mi-vga9/+ mouse, is lighter than that of the wild type. Mitf mi-vga9/+ mice (n = 6), and 184.4 ± 27.8 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6). The mean amplitude of the dark-adapted b-wave of the ERG in response to the same stimulus was 383.3 ± 26.0 µV in wild type mice, while it was 357.6 ± 11.5 µV in Mitf mi-vga9/+ mice (n = 6) and 352.4 ± 35.6 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6). The a-and b-wave implicit times were normal in those mutants that showed dark-adapted ERG responses at any of the levels of the stimulus luminance tested (P>0.05). The mean of the a-wave implicit time in response to the 1.87 log cd sec/m 2 stimulus was 8.9 ± 0.8 µV in wild type, while it was 8.0 ± 1.1 µV in Mitf mi-vga9/+ mice, and 7.9 ± 0.5 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice. The mean of the b-wave implicit time in response to the 1.87 log cd sec/m 2 stimulus in normal mice was 33.7 ± 2.0 µV, whereas it was 31.8 ± 2.6 µV in Mitf mi-vga9/+ mice, and 27.1 ± 1.8 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice.
The dark-adapted oscillatory potentials, evoked by the 1.87 log cd sec/m 2 stimulus but isolated with a 100-500 Hz digital filter are presented in Fig. 4a-c. These wavelets riding on the dark-adapted b-wave, reflecting inner retinal activity, were not significantly different between Mitf mi-vga9/+ , Mitf mi-enu22(398) /Mitf mi-enu22(398) and wild type mice at any of the levels of the stimulus luminance tested (P > 0.05) (Fig. 4d). The mean amplitude of the dark-adapted OPs in response to the 1.87 log cd sec/m 2 stimulus was 43.4 ± 15.5 µV in normal mice, 39.0 ± 6.5 µV in Mitf mi-vga9/+ mice, and 44.1 ± 6.4 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice in response to the same stimulus.
Light-adapted ERG responses were recorded after 10 minutes of light adaptation with a steady white background light of 1.7 log cd/m 2 , which was kept present during the recordings, while evoking ERG responses with light flashes of increasing intensity.  showed a-and b-waves. One-way ANOVA followed by Bonferroni's post-hoc comparisons test (wild type set as control) was used. Data are presented as means ± SEM, n = 6, P > 0.05. type and Mitf mutant mice, evoked by the 1.87 log cd sec/m 2 flash stimulus. As shown in Fig. 5a, the ERG recorded from normal mice showed a prominent b-wave whereas the Mitf -Mi-Wh/+ and Mitf Mi-Wh /Mitf mi mice showed a light-adapted ERG response that could not be detected clearly above baseline, suggesting complete lack of cone vision. In contrast, Mitf mi-vga9/+ and Mitf mi-enu22(398) /Mitf mi-enu22(398) mice showed light-adapted ERG b-waves evoked by the 1.87 log cd sec/m 2 flash stimulus that were comparable to the b-wave recorded from wild type mice. The mean amplitudes of the cone b-waves were not significantly different between those recorded from wild type mice (n = 6), Mitf mi-vga9/+ and Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6) at any of the levels of stimulus luminance tested (P>0.05) (Fig. 5d). The amplitude of the light-adapted b-wave increased in a comparable manner with increasing stimulus luminance in wild type (n = 6), Mitf mi-vga9/+ (n = 6) and Mitf mi-enu22(398) /Mitf mi-enu22(398) mice (n = 6). The mean amplitude of the dark-adapted b-wave of the ERG in response to the same stimulus in wild type mice was 206.9 ± 17.7 µV, while it was 196.0 ± 9.2 µV in Mitf mi-vga9/+ mice, and 192.3 ± 8.994 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice. No significant differences were observed in the b-wave implicit time in those mutants at any of the levels of stimulus luminance tested (P>0.05) (Fig. 5e). The mean amplitude of the b-wave implicit time in response to the 1.87 log cd sec/m 2 flash stimulus in normal mice was 25.3 ± 3.1 µV, 20.7 ± 1.6 µV in Mitf mi-vga9/+ mice, and 18.1 ± 1.2 µV in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice.

Discussion
The different mutations at the mouse Mitf locus have different effects on the phenotype, ranging from no visible effects to deaf animals lacking melanocytes with severe microphthalmia and osteopetrosis (reviewed in 32 ). Here we study eye function and morphology in mice carrying alleles which result in mild effects on adult eye size and apparent external structures; all have normal eye size. Our analysis shows that Mitf is not only important for normal adult eye size and external structure but is also critical for normal retinal function at 3 months of age. We showed that although eye size is normal in these mutants, pigmentation of the fundus is altered in all the mutants. The alterations in the fundus vary from a widespread lack of pigmentation to various degrees of hypopigmentation to yellow spots with non-pigmented areas ( Table 2). We found that despite some depigmentation of the RPE, two of the mutants, Mitf mi-vga9/+ and Mitf mi-enu22(398) /Mitf mi-enu22(398) not only have normal adult eye size and external structure but also have functioning retinas. The histology shown in Fig. 6a shows pigment in the RPE layer of these two mutants, although reduced, while choroidal melanocytes are near absent from Mitf mi-enu22(398) /Mitf mi-enu22(398) mice but present in Mitf mi-vga9/+ mice. The pigmentation observed by fundus photography is dependent on the presence of melanin in both RPE cells and choroidal melanocytes, both of which are affected by Mitf mutations. Hypopigmentation observed in the fundus of these mice is not necessarily an indication of RPE dysfunction or retinal degeneration. However, it is of course possible that despite lack of morphological changes, these mutations affect the expression of important downstream target genes affecting photoreceptor function or integrity at older ages. For example, homozygous Mitf mi-vga9 mice develop photoreceptor degeneration as they age due to effects on PEDF expression 48 Table 2). Thus it appears that Mitf mutant mice with intact RPE layers have normal retinal function and structure at 3 months of age, even though there is a severe reduction of choroidal melanocytes. We have yet to determine the exact time course of these retinal degenerations, which may be either progressive and rapid, or present at birth and thus directly due to the effects of Mitf during eye development.
The electroretinographic findings indicate that mice carrying the Mitf mi-vga9/+ and Mitf mi-enu22(398) /Mitf mi-enu22(398) mutations have normal dark-adapted and light-adapted ERG responses, while the other two, Mitf -Mi-Wh/+ and Mitf Mi-Wh /Mitf mi mice, show no ERG responses to any stimuli under either condition ( Table 2). The compound heterozygote  (Table 1) 32,47 . In the homozygous condition, the Mitf mi mutation leads to white, microphthalmic mice which die at 3 weeks of age due to osteopetrosis; heterozygotes are normal apart from a small belly spot 23,31 . The Mitf Mi-Wh/ Mitf mi compound heterozygote is of particular interest since it shows interallelic complementation as its phenotype is more normal with respect to eye development than the phenotype of each homozygote alone 31 . Interestingly, we observed that Mitf Mi-Wh/+ and Mitf Mi-Wh/ Mitf mi mice have no ERG response and the retina is abnormal at 3 months of age. The normal development of a full size eye, as induced by the interallelic complementation, is clearly not sufficient to prevent severe and early degeneration. Our results show that the Mitf Mi-Wh/+ mutant is clearly semidominant with respect to eye function. Interestingly, these mice show a profound hearing dysfunction due to an absence of melanocytes from the post natal cochlea 34 suggesting semidominant action also in the ear. The hearing abnormality present in these mice reinforces the notion of their similarity to individuals with Waardenburg and Tietz syndromes. Waardenburg syndrome, of which there are four subtypes with different penetrance of the clinical features, is a congenital, dominantly inherited pigmentation anomaly characterized by deafness, hypopigmentation of the skin, hair and eyes, with a midline white forelock of hair, and in some cases facial anomalities 37,49,50 . Ocular symptoms Light microscopy analysis of histological sections of the eyes of the mutants revealed that the RPE layer in the Mitf mi-enu22(398) /Mitf mi-enu22(398) mice shows localized thinning of melanin-containing areas. This is probably due to the localized degeneration of choroidal melanocytes in these animals. An equivalent light microscopy analysis of the eyes of Mitf mi-vga9/+ mice shows a normal RPE layer and apparent choroidal melanocytes, although fundus images reveal fine pigment mottling as white spots across the fundus. The Mitf -Mi-Wh/+ mice showed a near complete absence of the melanin-containing areas of the RPE layer, and Mitf Mi-Wh /Mitf mi mice showed a thin monolayer of RPE cells but absence of apparent choroidal melanocytes or a Bruch's membrane. Retinal thickness analysis showed that the POS and ONL layers are absent from Mitf -Mi-Wh/+ and Mitf Mi-Wh /Mitf mi eyes; the IPL is thinner compared to wild type. In addition, the Mitf Mi-Wh /Mitf mi mice showed thin INL compared to normal mice. In both cases, the major phenotypic effects can be attributed to the Mitf -Mi-Wh allele which clearly has severe effects on the eye. Despite substantial thinning of the melanin-containing areas of the RPE layer in Mitf mi-enu22(398) /Mitf mi-enu22(398) mice, due to absence of choroidal melanocytes, and some hypopigmentation and pigment mottling in the fundi of Mitf mi-vga9/+ mice, no significant differences were observed in the thickness of the ONL, INL and IPL in these mice as compared to wild type mice; the total retina is thinner in the Mitf mi-enu22(398) /Mitf mi-enu22(398) mice than in either wild type or Mitf mi-vga9/+ mice.
This study provides more evidence that a functional RPE is important for normal photoreceptor function. Thus, mutations in genes expressed in the RPE such as Mitf and Otx2, can have profound effects on retinal structure and function, as a consequence of alterations in the RPE 17,18,55 . RPE-specific ablation of Otx2 in the adult mouse retina leads to RPE dysfunction, which in turn leads to photoreceptor degeneration comparable to those that occur in Mitf Mi-Wh/+ and Mitf Mi-Wh /Mitf mi mice 17,18 . Mutations in the human OTX2 gene on chromosome 14q22.3, which codes for a transcription factor that shares 100% amino acid identity with its mouse ortholog (Otx2), cause pattern dystrophy of the RPE, and slow photoreceptor degeneration in patients 19 , similar to what occurs in mice with Otx2 ablation 17,18 . However, unlike Mitf, Otx2 is also expressed in the bipolar and photoreceptor cells of the retina, in addition to its role in regulating visual cycle function in the RPE 17,20 . Thus, it can be stated with certainty that retinal dysfunction or degeneration in Mitf mutant mice found here are due to mutations in a gene which in the eye is only expressed in the RPE 28 , while those seen in Otx2 mutant mice, and in patients with OTX2 mutations, are more complex.
The RPE plays a vital role in the visual cycle 56 . This process is a key element of the interactions between the RPE cells and photoreceptor outer segments. In the RPE, Mitf is known to regulate the expression of two visual cycle genes, Rlbp1 which encodes retinaldehyde binding protein-1 (RLBP1), and Rdh5, which encodes retinol dehydrogenase-5 (RDH5) 55 . Thus, Mitf controls expression of visual cycle genes and consequently the reconstitution of functional rhodopsin, although not the same processes of the visual cycle as Otx2 in the RPE. Wen et al. 55 also found that postnatal injection of 9-cis-retinal can partially rescue the retina in homozygous Mitf mi-vga9 mutant mice with respect to gene expression, structure and function. This means that the lack of Mitf in the RPE may have such profound developmental effects that the adjacent retina might be permanently damaged 55 . It is likely that these processes of the visual cycle play a role in inducing the degeneration found in the eyes of Mitf Mi-Wh/+ and Mitf Mi-Wh /Mitf mi mice. We have shown here that mutations in a transcription factor expressed in the RPE, and not in the neuroretina, have profound effects on the appearance of the fundus of the eye, in particular its pigmentation, and can lead to changes in neuroretinal function, and to severe retinal degeneration in some cases. The study further demonstrates that the RPE is essential for vision, and that transcription factors whose expression in the eye is restricted to the RPE play a role that is critical for normal retinal structure and function.  fundus photography. Live fundus photographs were obtained from anaesthetized mice with a Micron IV rodent fundus imaging system (Phoenix Research Labs Inc., Pleasanton, CA). Prior to fundus examination the mouse was given an intraperitoneal injection of a mixture of 40 mg/kg −1 Ketamine (Pfizer, Denmark) and 4 mg/ kg −1 Xylazine (Chanelle Pharmaceuticals, Ireland), and the anaesthesia was maintained as needed with half of that dose. Pupils were dilated with topical application of Mydriacyl (1% tropicamide) (Alcon Inc., U.S.A) eyedrops, and Alcaine (Alcon, Inc., U.S.A) drops were applied as a local corneal anaesthetic. A photograph of the front of the anterior chamber of the eye was then taken with the fundus camera to assess both eye size and pupil dilation. A 1% methylcellulose gel was applied to the cornea of both eyes and a mini contact lens (Ocuscience LLC, Henderson, NV) 2.5 mm in diameter placed on the cornea, to reduce the risk of cataract formation 57 . The contact lens from one eye was removed before fundus photography and 1% methylcellulose applied to the eye again. A bright field image of the anterior segment was first obtained, and then bright field images of the fundus with the optic nerve in the centre of the field. Fluorescent angiography fundus images were obtained after intraperitoneal injection of fluorescein sodium salt (0.1%, 10 µl/kg). Recording procedure. Prior to each ERG experiment the mouse was given an intraperitoneal injection of 40 mg/ kg −1 Ketamine and 4 mg/kg −1 Xylazine and then the anaesthesia was maintained with half of that dose. The electrode always contacted the cornea through a layer of 1% methylcellulose 58 . The pupil was dilated with topical application of Mydriacyl (1% tropicamide) (Alcon Inc., U.S.A) and Alcaine (Alcon, Inc., U.S.A) drops were used as a local corneal anesthetic. Saline solution drops were systematically applied to the eye as a tear supplementation, in order to prevent dehydration and cataract formation, and to provide adequate electrical conduction for the corneal electrode. In order to further prevent cataract formation, a mini contact lens (Ocuscience LLC, USA) completely covered the cornea through a layer of 1% methyl cellulose gel during dark adaptation 57 , which was removed under dim red illumination before placing the corneal electrode. The mouse was placed on a heating pad with a rectal thermal sensor beneath it, both connected to a thermal regulator. The head of the mouse was positioned in a straight line with the body, by a short string threaded behind its upper front teeth, to maintain normal breathing. The mice were dark adapted for 30 minutes before recording the scotopic ERG responses. At least one minute elapsed before stimulus presentations in order to allow the photoreceptors to recover from any photopigment bleaching and to maintain the dark adaptation state between flashes 59 . Afterwards the mouse was light adapted with a steady background light of 1.7 log cd/m 2 for ten minutes. The ERG recording protocol was the same after light adaptation as the one following dark adaptation. At the end of the experiment the anaesthetized animal was sacrificed via cervical dislocation and then bled by femoral artery sectioning.

electroretinogram (eRG
ERG data analysis. The amplitudes of the ERG and its components are presented as means ± standard error of the mean, in microvolts (µV). The a-wave amplitude was measured from baseline to the lowest base of the cornea-negative deflection. The b-wave was either measured from baseline or lowest point of the a-wave (if present), to its peak. Implicit times of the a-, b-waves were measured from flash onset and until their apex 59,60 . Oscillatory potentials (OPs) were extracted with a band-pass filter (100-500 Hz). The peaks of the first four OPs were measured and summed to generate the total OP amplitude as previously described 61,62 . The amplitude of each OP was measured individually from the preceding trough to the peak of the OP considered, except for OP1 that was measured from the preceding baseline to its peak.
Histology and measurement of the retinal layers. Mice were euthanized by cervical dislocation. Eyes were enucleated and fixed in 4% paraformaldehyde over night at 4 °C. Subsequently, the eyes were embedded in paraffin and sectioned at 5 µm onto glass slides. Retinal tissue sections were deparaffinised and then stained with hematoxylin and eosin (H&E). Microscopic examination was performed using an Olympus IX71 microscope (Tokyo, Japan). The distance from the retinal surface to the end of the photoreceptor outer segments was measured as the retinal thickness. The thicknesses of the ONL, INL and IPL in the central retina were measured by light microscopy at the point where the ratio of the distance from the optic disc to the point to the distance from