Comparison of refractive development and retinal dopamine in OFF pathway mutant and C57BL/6J wild-type mice

Animals and experimental design

We examined the refractive development of Vsx1−/− mice with a genetically induced OFF visual pathway defect [23]. In the mouse, Vsx1 expression is detected as early as postnatal day 5 in the bipolar cell region of the inner nuclear layer (INL) [24]. An in-house breeding colony with male and female heterozygous Vsx1 mutants (a generous gift from Dr. Robert L. Chow, University of Victoria, Victoria, Canada) was maintained at the Atlanta Department of Veterans Affairs Medical Center. Mice were kept in 12 h:12 h light-dark cycles (lights on at 06:00 with about 17 lux during the light phase) with mouse chow and water available ad libitum.

Age-matched male and female wild-type (Vsx1+/+) and Vsx1−/− mice on a 129S1/Sv background were subjected to two different experimental paradigms: refractive development (RD) and form deprivation (FD). For the RD experiments (Vsx1+/+, n=8; Vsx1−/−, n=7), refractive error and axial length measurements were obtained every 2 weeks from 4 to 12 weeks of age under non-manipulated visual conditions. For the FD experiments, baseline refractive measurements for the Vsx1+/+ (goggled, n=7; naïve controls, n=8) and Vsx1−/− (goggled, n=8; naïve controls, n=9) mice were obtained at 4 weeks of age, and then the mice were subjected to monocular form deprivation in the right eye using head-mounted goggling apparatus as described previously [31]. Goggles were applied for a period of 8 weeks (i.e., up to 12 weeks of age). Refractive development in the 129S1/Sv Vsx1+/+ mice was also compared with the commonly used C57BL/6J mice under normal (n=8) and FD (goggled, n=8; naïve controls, n=8) visual paradigms. Ocular compliance with goggling was monitored daily, and animals that showed poor compliance or ocular health were excluded from further measurements. All procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the local Institutional Animal Care and Use Committee.

Refractive state, corneal curvature, and ocular biometry

For the RD and FD paradigms, refractive error and ocular biometry were acquired, as described previously [22,32,33]. Eyes were dilated with 1% tropicamide, and the refractive state was obtained using an automated infrared photorefractor [34] by gently restraining the mouse by the tail in a dark room. A small subset of mice (n=13 for both RD and FD) had axial length measurements acquired before anesthesia using a custom-made 845 nm time-domain partial coherence interferometer (PCI) [33,35]. Mice were stabilized in a conical tube and the head pedestal stabilized with a clip [32]. Following the awake measurements, the mice were anesthetized with ketamine (80 mg/kg) and xylazine (16 mg/kg). The refractive error measurements were repeated under anesthesia to obtain a more stable measurement with standard deviations of less than 0.5 diopter (D) [22].

Subsequently, the corneal radius of curvature was measured using an automated keratometer [36,37]. Upon proper alignment of the eye, the device used infrared light-emitting diodes (LEDs) to create eight Purkinje images on the corneal surface, which was recorded with an infrared light-sensitive video camera to calculate the overall corneal radius of curvature [37].

Axial length measurements were obtained using 1310 nm spectral-domain optical coherence tomography (SD-OCT; Bioptigen, Durham, NC), as previously described [32]. Axial length was measured from the anterior cornea to the RPE border using the OCT calipers (calibrated at a refractive index of 1.433) [38]. After the measurements were completed, the mice were given yohimbine (2.1 mg/kg) to reverse the effects of anesthesia and reduce the development of corneal lesions [39], and placed on a warming pad to recover.

Retinal dopamine quantification

To determine the levels of retinal DA and DOPAC (the primary metabolic by-product of DA) [40] during the critical period of normal refractive development in mice [30], the retinas of 4-week-old Vsx1−/− (n=9), Vsx1+/+ (n=10), and C57BL/6J (n=10) mice were harvested between 4 and 6 h after light onset. For the FD experiments, the retinal DA and DOPAC levels were measured after the final endpoint at 8 weeks for the Vsx1+/+ (control: n=3, goggled: n=6) and Vsx1−/− (control: n=4, goggled, n=7) mice and at 4 weeks for the C57BL6/J mice (control: n=6, goggled: n=4). Experimental mouse retinas were collected 48 h after the final measurements to eliminate any effect of anesthesia. Harvested retinas were immediately frozen on dry ice and stored at −80 °C. The frozen retinas were processed as previously described [41]. Briefly, the retinas were homogenized in 0.1 N HClO₄ solution (0.01% sodium metabisulfite and 50 ng/ml internal standard 3, 4-dihydroxybenzylamine hydrobromide) and centrifuged. Supernatant fractions were separated with high-performance liquid chromatography (HPLC) using a mobile phase of 0.1 M sodium phosphate, 0.1 mM EDTA, 0.35 mM sodium octyl-sulfate, and 6% acetonitrile (pH 2.7) mobile phase to quantify the DA and DOPAC levels with coulometric detection. The DA and DOPAC levels were calculated using the internal standard method with a standard curve generated with 0.1–1 ng DA and DOPAC. All FD cohort retinas were homogenized individually; the retinas from the right and left eyes of the RD cohort were pooled together for analysis.

Data analysis

Statistical analyses were performed using commercial software (SigmaStat 3.5, Aspire Software International, Ashburn, VA). Changes in refractive error, keratometry, and axial length between the Vsx1+/+, Vsx1−/−, and C57BL/6J animals with age under normal non-manipulated (RD cohort) and form-deprived (FD cohort) conditions were analyzed with repeated-measures two-way ANOVA (ANOVA) and Holm-Sidak post-hoc tests for statistical significance. To minimize the influence of body size (mean bodyweight at 4 weeks; Vsx1+/+: 15.94±0.91 g, Vsx1−/−: 13.6±0.44 g, and C57BL/6J: 13.19±0.34 g, one-way ANOVA on Rank, H=7.07, p=0.029) on eye size [42], the axial length measurements were normalized to the respective bodyweight across age. For the RD cohort, where the right and left eyes received the same treatment, the refraction and biometric measurements from the two eyes were averaged to represent a single value from each mouse. For animals for which PCI and SD-OCT measurements were available, the axial length values were averaged since the techniques have good agreement [32]. For the FD cohort experiments, refractive errors are presented as “myopic shift,” the difference between goggled and opposite eyes or right minus left eyes for naïve controls. To eliminate any inter-subject variability, the axial length measurements of the FD cohort were normalized to 4-week-old values (i.e., baseline values). After the values were normalized to baseline, “axial shift” was determined by calculating the difference in eye length between the right eye and the left eye. The DA and DOPAC measurements from the 4-week-old RD cohorts were analyzed using one-way ANOVA with Holm-Sidak post-hoc comparisons, or when normality failed, a one-way ANOVA on Rank with Dunn’s method post-hoc comparisons. Additionally, the DA and DOPAC values from the FD cohorts are presented as a ratio of the right eye to the left eye (the OD/OS ratio) to illustrate the changes in the retinal DA levels between the goggled and control eyes, where a ratio of 1 would indicate similar DA levels in the goggled (right) and opposite (left) eyes. DA data from the FD cohorts were analyzed using two-way ANOVA with Holm-Sidak post-hoc comparisons.

Refractive development altered by strain, but not the OFF pathway defect

Under a normal laboratory visual environment, the 129S1/Sv (Vsx1−/− and Vsx1+/+) and C57BL/6J mouse strains showed a significant increase in hyperopic refraction with increasing age from 4 to 12 weeks (Figure 1A; two-way repeated-measures ANOVA main effect of age, F(8,195)=76.88, p<0.001). However, there were significant differences in the pattern of refractive development for these strains across age (two-way repeated-measures ANOVA interaction, F(16,195)=6.69, p<0.001). At 4 weeks of age, the Vsx1−/− and Vsx1+/+ mice were relatively myopic (mean refractive error; mean ± standard error of the mean (SEM), Vsx1+/+: −4.73±0.98 D, Vsx1−/−: −5.28±0.75 D) and became significantly hyperopic by the end of 12 weeks (Vsx1+/+: 5.33±0.81 D, Vsx1−/−: 3.28±0.82 D). Throughout the course of refractive development, refractive errors were not significantly different between the Vsx1−/− and Vsx1+/+ mice (Holm-Sidak multiple comparisons, p<0.05), suggesting that the Vsx1 mutation had no significant effect on normal refractive development of the eye. In comparison, the C57BL/6J mice had relatively hyperopic refractive errors at younger ages (mean refraction at 4 weeks, 3.40±0.43 D), and developed more hyperopia until about 7 weeks of age (8.07±0.55 D) before stabilizing.

From 4 to 12 weeks of age, there was significant flattening in corneal curvature for each group (mean flattening, 0.17±0.001, 0.18±0.001, and 0.15±0.007 mm for Vsx1−/−, Vsx1+/+, and C57BL/6J, respectively; Figure 1B; two-way repeated-measures ANOVA main effect of age, F(8,171)=281.95, p<0.001). However, the corneal curvatures were significantly flatter for the 129S1/Sv (Vsx1−/− and Vsx1+/+) mice (Holm-Sidak multiple comparisons at all ages, p>0.05) compared to the C57BL/6J mice across development (two-way repeated-measures ANOVA main effect of strain, F(2,171)=6.47, p=0.008).

Consistent with the increase in bodyweight with age, there was a significant decrease in the axial length normalized to bodyweight across all groups (mean change in axial length between 4 and 12 weeks of age; Vsx1−/−: −0.08±0.003, Vsx1+/+: −0.06±0.005, C57BL/6 J: −0.09±0.002 mm; Figure 1C; two-way repeated-measures ANOVA main effect of age, F(8,199)=169.83, p<0.001). Except for a small difference in axial length between the Vsx1−/− and Vsx1+/+ mice at 4 weeks of age, the changes in axial length were not significantly different between either the genotype or the strain throughout the developmental period (two-way repeated-measures ANOVA main effect of strain, F(2,199)=2.50, p=0.107). The standard deviation for raw axial length measurements (about 50 µm) was greater than the inter-user measurement variability of the instrument (21 µm) [32] and the predicted axial length difference based on eye modeling (1 D of refractive change=to about 5 µm) [36]. Therefore, the SD-OCT measurements may not have been able to detect small changes in axial length between strains despite the significant differences in refractive errors (Figure 1A).

Form deprivation myopia response absence in Vsx1−/− and Vsx1+/+ mice

To examine the effects of the altered visual environment on ocular development, mice were form deprived from 4 to 8 weeks of age, and the magnitude of the myopic shift in response to goggling was compared between the genotypes (Figure 2). The Vsx1+/+ (Figure 2A) and Vsx1−/− (Figure 2B) mice showed no significant myopic shift associated with goggling at any measured time point compared to their age-matched non-goggled naïve controls (Vsx1+/+: myopic shift at 8 weeks; goggled animals: 0.48±1.25 D, controls: 0.02±0.74 D, two-way repeated-measures ANOVA treatment by time interaction, F(4,65)=0.131, p=0.97; Vsx−/−: mean myopic shift at 8 weeks; goggled animals: −0.32±0.57 D, controls: −1.81±1.2 D, two-way repeated-measures ANOVA interaction effect, F(4,65)=0.609, p=0.658), even after an extended follow-up for up to 12 weeks (data not shown). In comparison, the C57BL/6J mice exhibited a significant myopic shift in the goggled animals (−4.29±0.42 D) compared to the naïve controls (0.09±0.69 D) at 4 weeks after the FD treatment (two-way repeated-measures ANOVA, F(4,72)=3.085, p=0.024, Figure 2C). FD did not cause significant changes in the axial length or corneal curvature of the goggled mice compared to the untreated naïve controls for any genotype or strain (two-way repeated-measures ANOVA, p>0.05; data not shown).

We observed a considerable degree of tear film instability in the 129S1/Sv mice, which may have induced some variability in ocular response to FD between animals. Therefore, to ensure minimal effects of tear film on measurements, repeatability of measurements was closely observed, and saline eye drops were frequently applied to retain corneal hydration during the OCT measurements.

Retinal dopamine turnover significantly elevated in 129S1/Sv mice compared to C57BL/6J mice

DA and its metabolite, DOPAC, were measured in the 129S1/Sv and C57BL/6J retinas during the critical period of visual development (Figure 3) and after FD (Figure 4) to investigate the role of DA in refractive development. When the mice were 4 weeks old, the retinal DOPAC levels in the Vsx1−/− mice (564±67 pg/mg) were significantly greater than in the Vsx1+/+ (364±23 pg/mg) mice (one-way ANOVA on Rank, H=16.78, p<0.001, Dunn’s method multiple comparisons, p<0.05; Figure 3A). When we compared the strains, we found no significant differences in retinal DOPAC levels between the Vsx1+/+ and C57BL/6J (284±21 pg/mg) mice. Although the DA levels were similar between the two genotypes on the 129S1/Sv background (mean DA levels; Vsx1+/+: 1673±169 pg/mg; Vsx1−/−: 1955±95 pg/mg), the levels were significantly elevated in the C57BL/6J mice (2238±159 pg/mg) compared with the Vsx1+/+ strain (one-way ANOVA, F(2,28)=3.76, p<0.05, Holm-Sidak multiple comparisons, p<0.05; Figure 3B). The DOPAC/DA ratios were similar between the 129S1/Sv Vsx1+/+ (0.23±0.02) and Vsx1−/− (0.30±0.05) genotypes (Figure 3C). However, the ratios were significantly lower for the C57BL/6J mice (0.13±0.01) compared to the Vsx1+/+ mice (one-way ANOVA on Ranks, H=16.66, p<0.001, Dunn’s method multiple comparisons, p<0.05), suggesting considerable strain differences in DA turnover during the critical phase of refractive development.

Under FD conditions, the ratio of the right eye to the left eye (the OD/OS ratio) for retinal DOPAC was not significantly different between the goggled and control animals for any strain or genotype (Figure 4A), indicating goggling had no significant effects on DA metabolism between the different genotypes. Similar results were found for retinal DA and the DOPAC/DA ratio (Figure 4B,C).

Normal refractive development in Vsx1−/− mice

In this study, we observed no significant differences in the ocular refractive development curve between the Vsx1−/− mice and their age-matched Vsx1+/+ controls, which suggests the Vsx1 mutation does not have a significant effect on the normal refractive development of mammalian eyes. The Vsx1−/− and Vsx1+/+ mice were substantially myopic at younger ages, and became significantly hyperopic with increasing age (Figure 1A). In this study, similar to previous reports [22,36,43-45], the juvenile C57BL/6J mice were considerably hyperopic (mean refractive error at 4 weeks, 3.4±0.43 D) as opposed to the Vsx1+/+ mice that were significantly myopic (−4.73±2.78D). Given the insufficiency of the axial length (same as the C57BL/6J mice) and corneal curvature (flatter than in the C57BL/6J mice) changes in explaining the myopic refractive errors in the Vsx1+/+ mice, we hypothesize that it could be due to differences in other ocular optical parameters (such as the crystalline lens). Although this is the first report of the refractive state in 129S1/Sv mice, similar variations in refractive error [34] and ocular parameters [42] between different mice strains have been reported.

Previously, immunolabeling of the axonal termini with cone bipolar cell markers in Vsx1−/− mice has shown that the expression of Vsx1 is restricted to only a small subset of ON (type 7) [28] and OFF (types 2, 3, and 4) [23,24,26,27,46] cone bipolar cells, suggesting that Vsx1 may not be essential for late terminal differentiation of all cone bipolar cell subtypes in the mammalian retina. Similarly, Chow et al. [23] found the overall number of cone bipolar cells was largely similar between Vsx1−/−, Vsx1+/−, and Vsx1+/+ mice, and concluded that the majority of ON and OFF cone bipolar cells do not require Vsx1 to retain their gross morphology and viability. Given the normal morphology and function of Vsx1 independent bipolar pathways and other retinal cell types in Vsx1−/− mice, we hypothesize that normal refractive development in Vsx1−/− mice may be due to normal processing of visual information through other functional non-Vsx1 containing cone bipolar interneurons in the retina.

Although the genetic defect in the OFF visual pathway caused by the Vsx1 mutation does not appear to influence normal ocular refractive development, a mouse model of a non-functional ON pathway [17,18] with a null mutation in Nyx [18] has been shown to cause a small, but significant, myopic shift during adolescence [22]. This may be due to differences in designated rod and cone signaling through the ON and OFF pathways or differences in retinal DA levels in phenotypes expressing ON and OFF pathway defects (discussed below). A defect in the ON pathway would affect the primary rod signaling through ON rod bipolar cells to retinal ganglion cells via AII amacrine cells (although gap-junctions would retain some alternative rod signaling) [47], as well as cone signaling through ON cone bipolar cells. However, the OFF pathway defect would involve only hyperpolarizing OFF cone bipolar cells [23] with no significant effect on ON and OFF rod signaling or on cone signaling through the ON pathway. Additionally, the Vsx1 mutation affects only a small subtype of OFF cone bipolar cells [23,24,48] leaving a subset of Vsx1 independent OFF bipolar cells functional for normal visual transmission. However, the nob [22] and Vsx1−/− mice were on different background strains (i.e., C57BL/6J and 129S1/Sv, respectively), which could influence the refractive development and retinal DA profile observed between the two mouse strains. Future studies with mouse mutants that express complete loss of ON and OFF pathways on the same background strain are required to closely compare the effects of ON and OFF pathway deficits on normal refractive development of the eye.

Vsx1+/+ and Vsx1−/− mice are unresponsive to form deprivation

To investigate the influence of the Vsx1 mutation on refractive development with an altered visual environment, we compared the susceptibility to form deprivation myopia between Vsx1+/+ and Vsx1−/− mice. For the Vsx1+/+ and Vsx1−/− mice (on the 129S1/Sv background), application of the diffuser goggle even for a protracted period of 8 weeks did not produce a significant myopic shift compared to the control animals. Interestingly, in agreement with previous reports [22,34,38,49,50], the C57BL/6J mice showed a significant myopic shift of about 4 D in response to 4 weeks of FD treatment.

These variations in susceptibility to FD myopia could be due to differences in endogenous retinal DA levels and/or metabolism between strains. We found the retinal DOPAC and DA levels at 4 weeks in the C57BL/6J mice were lower or higher, respectively, than in the Vsx1+/+ mice. Importantly, the DOPAC/DA ratios, a reflection of dopamine turnover, were also significantly higher in the Vsx1+/+ mice compared to the C57BL/6J mice. The absence of response to FD could be due to a higher DA metabolism, storage, and/or availability in the 129S1/Sv retinas, making the eye resistant to imposed optical manipulation. Inhibitory effects of higher retinal dopamine levels on form-deprivation myopia have been shown in chickens [51,52], guinea pigs [53,54], and primates [55]. In addition, previous studies have indicated strain differences in response to form deprivation in mice [34], perhaps due to differences in ocular components or retinal signaling.

Following 8 weeks of form deprivation, the ratio of the retinal DA and DOPAC levels in the goggled eye (right eye) to the contralateral eye (left eye) were not significantly different between the Vsx1+/+ and Vsx1−/− mice (which was consistent with the absence of a form deprivation response in both mouse strains). However, we also found no difference in the retinal DA and DOPAC levels in the C57Bl/6J mice with a significant myopic shift. These findings suggest that baseline retinal DA metabolism and/or dopamine turnover during the critical period of visual development (Figure 3) are more essential for determining the predisposition of the mouse eye to form deprivation myopia than changes in DA with FD. In view of recent evidence suggesting DA acts on different dopamine receptors to influence eye growth in mice [56] and synthesis of DA from other retinal sources (such as the RPE) [57], it appears that the role of DA in ocular response to visual deprivation is not simple; instead, DA participates in a complex pathway that connects the visual experience to postnatal refractive development of the eye.

Normal corneal curvature in Vsx1−/− mice

Missense mutations in Vsx1 have been associated with corneal dystrophies such as posterior polymorphous dystrophy and keratoconus in human eyes [29]. Since both corneal disorders exhibit abnormally steep corneas [58,59], we sought to examine the differences in corneal curvature between the Vsx1+/+ and Vsx1−/− animals. In this study, no significant differences in corneal curvatures were observed between the two genotypes during their developmental period. In agreement with this observation, Chow et al. [23] also found no corneal abnormalities in Vsx1−/− mice on gross and histological examinations.

In conclusion, our results suggest that selective impairment of retinal OFF visual pathway signaling due to the Vsx1 mutation does not significantly alter the normal refractive development of the mouse eye, potentially due to normal visual transmission through other Vsx1 independent ON and OFF retinal bipolar cells. Elevated retinal DA turnover in the 129S1/Sv background strain early in refractive development may prevent FD myopia in Vsx1+/+ and Vsx1−/− mice. Finally, our study indicates that the severity of refractive deficits caused by the ON pathway defect [22] is more severe than those caused by the OFF pathway, perhaps due to altered retinal DA levels in these phenotypes.