In the human, mutations of OTX2 (Orthodenticle homeobox 2 transcription factor) translate into eye malformations of variable expressivity (even between the two eyes of the same individual) and incomplete penetrance, suggesting the existence of subtle thresholds in OTX2 activity. We have addressed this issue by analyzing retinal structure and function in six mutant mice with graded Otx2 activity: Otx2+/+, Otx2+/AA, Otx2+/GFP, Otx2AA/AA, Otx2AA/GFP and Otx2GFP/GFP. Null mice (Otx2GFP/GFP) fail to develop the head and are embryonic lethal, and compound heterozygous Otx2AA/GFP mice show a truncated head and die at birth. All other genotypes develop until adulthood. We analyzed eye structure and visual physiology in the genotypes that develop until adulthood and report that phenotype severity parallels Otx2 activity. Otx2+/AA are only mildly affected whereas Otx2+/GFP are more affected than Otx2+/AA but less than Otx2AA/AA mice. Otx2AA/AA mice later manifest the most severe defects, with variable expressivity. Electrophysiological and histological analyses of the mouse retina revealed progressive death of bipolar cells and cone photoreceptors that is both Otx2 activity- and age-dependent with the same ranking of phenotypic severity. This study demonstrates the importance of gene dosage in the development of age-dependent pathologies and underscores the fact that small gene dosage differences can cause significant pathological states.
Mutations in human homeoprotein (HP) transcription factor gene, OTX2, cause a remarkable diversity of phenotypes and pathologies such as otocephaly-dysgnathia, microphthalmia, anophthalmia, coloboma, hippocampal defects and growth hormone deficiency with associated diminished stature (1–3). In addition to their wide range, these phenotypes often have highly variable expression and incomplete penetrance, as illustrated by the possibility of inheriting the OTX2 mutation and phenotype from an apparently unaffected parent (3). The frequent morphological and functional eye defects suggest that the development of human vision is sensitive to OTX2 dosage (1,4–6). In the mouse, complete deletion of Otx2 is lethal with the embryos failing to develop the anterior neuroectoderm (7,8). Mice in which one allele of Otx2 has been deleted are viable and fertile and often present no gross abnormalities in brain and eyes (9–11), even though cerebral and retinal functions might be affected, depending on the genetic background of the strain (11–13).
In the visual system, Otx2 is expressed in the mouse optic vesicle as early as embryonic day 9 (E9) (9). Two days later, Otx2 is expressed in the retinal pigment epithelium (RPE) and post-mitotic retinal neurons (14). In the retina, Otx2 controls the terminal differentiation of photoreceptors and bipolar cells via transactivation of cone-rod homeobox (Crx) and protein kinase C α (PKCα), respectively (10,15). Recent studies also suggest that Otx2 regulates retinal synaptogenesis through the expression of Pikachurin, an autocrine protein that binds ß-dystroglycan in the presynaptic region of photoreceptor ribbon synapses (16,17). In addition, at the time of eye opening, Otx2 is imported by retinal ganglion cells (RGCs) and acts non-cell autonomously in the opening of a critical period of plasticity in the binocular visual cortex (11). On the basis of the translocation properties shared by many HP transcription factors, Otx2 was used as a therapeutic protein to promote the survival of damaged adult RGCs (18).
The persistence of Otx2 expression in adulthood raises the question of its role after development. In adult mice, the conditional deletion of Otx2 causes the degeneration of photoreceptors in the retina and the disruption of RPE function (19,20), whereas exogenous Otx2 can stimulate the survival of stressed retinal neurons (18). In this study, we have addressed the important issue of the dose-dependent activity of Otx2 in eye morphology, visual physiology and retinal structure by generating mice with different levels of Otx2 activity.
Gross phenotypic characterization of the different mutants
Otx2 null mice show a complete deletion of the early Otx2 expression domain and are thus headless (7,8). In contrast, heterozygous Otx2 mice (with a GFP insertion in one Otx2 allele, thus Otx2+/GFP mice) are normal at a gross anatomical level, survive and reproduce. In the context of another study on the role of Otx2 non-cell autonomous activity in the plasticity of the visual cortex (21), we reported that an arginine–lysine (RK) doublet, located just before the DNA-binding domain (homeodomain) of Otx2 was required for the specific recognition of target cells (parvalbumin PV-cells), via its interaction with cell surface glycosaminoglycans (GAGs). The mutation of the RK doublet into two alanine (A) residues (the mutated protein is called Otx2-AA, Fig. 1A) decreases its specific capture by the PV-cells in the visual cortex. The Otx2-AA protein is still capable of transferring between cells, but transfer specificity is impaired (21).
To examine whether the RK-to-AA mutation of Otx2 modifies its DNA-binding properties, we produced Otx2-AA recombinant protein in bacteria and first compared its ability to bind the Rbp3 promoter DNA sequence with that of the wild-type Otx2 protein (22). Both Otx2 and Otx2-AA proteins are able to bind the DNA probe, with the binding by Otx2-AA seemingly greater than Otx2 (Fig. 1B). We also assessed the transcriptional activity of Otx2-AA in cultured mammalian cells expressing secreted alkaline phosphatase under the control of the Rbp3 promoter normally recognized by Otx2 (23,24). Otx2-AA is a less active trans-activator than Otx2 at the Rbp3 promoter (Fig. 1C). In other experiments using PKCa or Crx promoters (10,15) to drive the expression of luciferase, the mutated protein provided ∼71 and 60% of the transcriptional activity of Otx2, respectively (data not shown).
The latter results suggest that Otx2-AA cell autonomous activity, in addition to transfer specificity, is perturbed. This precludes the precise discrimination of the two aspects of Otx2 activity (autonomous and non-autonomous), and hereafter, we use the term ‘activity’ to include both cell autonomous and non-cell autonomous activities. We thus generated a knock-in mutant, the Otx2-AA mouse line, in which the RK doublet was replaced by an AA doublet (Fig. 1A), and investigated the influence of the hypomorphic mutation of Otx2. The altered activity of Otx2-AA was confirmed at the genetic level by comparing Otx2+/GFP mice (with one wild-type allele) with Otx2AA/GFP mice (with only one AA allele). As shown in Figure 2A, the former mice have a grossly normal morphology, whereas the latter ones have a truncated head and die at birth. Otx2AA/GFP mice develop a hindbrain but only rudiments of the forebrain and midbrain (data not shown), suggesting that one AA allele is insufficient for normal brain development. In contrast, as for Otx2+/GFPmice, the heterozygous Otx2+/AA and homozygous Otx2AA/AA mice are viable, reproduce and show apparently normal head development (Fig. 2A). With regard to eye development, Otx2+/AA and Otx2+/GFP mice were without obvious defects, whereas some Otx2AA/AA mice exhibited microphthalmia (Fig. 2C) and anophthalmia (Fig. 2D), with incomplete penetrance. These phenotypes were observed in 15% of Otx2AA/AA mice, the others showing apparently normal eyes (Fig. 2B). Of the Otx2AA/AA mice showing anophthalmia or microphthalmia, two mice out of >30 were affected bilaterally. Therefore, the expression of these phenotypes was mostly asymmetric, suggesting a lack of robustness in eye development. The absence of such major phenotypes in Otx2+/AA and Otx2+/GFP mice indicated that a single wild-type allele is sufficient to generate a normal eye structure at a gross level.
Therefore, still at a gross level, the severity of the anatomical phenotype increases with decreasing Otx2 activity: Otx2+/+ < Otx2+/AA < Otx2+/GFP < Otx2AA/AA < Otx2AA/GFP < Otx2GFP/GFP. We then asked whether there were functional and/or structural defects and, if so, how these defects vary according to Otx2 activity. In the rest of this study, Otx2GFP/GFP and Otx2AA/GFP were not tested because of embryonic lethality and the absence of head/eyes. In addition, we included for analysis only Otx2AA/AA mice with externally normal eyes.
Otx2 hypomorphs have reduced visual acuity
Visual acuity of post-natal day 60 (P60) mice was evaluated using the optomotor test (25) with optotypes consisting of 100% contrast black and white gratings of different spatial frequencies. The number of head turns is a measure of the ability of a mouse placed within the drum to distinguish the rotating vertical black and white pattern. As the spatial frequency of the stripes increases, the ability to separate them decreases with a parallel decrease in the number of head turns. No significant difference in visual acuity was measured between Otx2+/AA and Otx2+/+ mice at all spatial frequencies tested between 0.125 and 0.75 cycles per degree (cpd) (Fig. 3). In contrast, Otx2+/GFP mice made significantly fewer head turns than their wild-type littermates at all spatial frequencies (principally 0.67 ± 0.49 versus 3.71 ± 0.97 at 0.375 cdp and 0.40 ± 0.40 versus 3.00 ± 0.50 at 0.25 cdp, respectively). Even worse, Otx2AA/AA mice failed to respond to any spatial frequencies tested (Fig. 3).
This first test enabled us to rank the four genotypes according to their performance in visual function from normal to most severe: Otx2+/+ = Otx2+/AA < Otx2+/GFP < Otx2AA/AA. It is noteworthy that the product of two AA alleles (Otx2AA/AA) is not equivalent to the product of one wild-type allele (Otx2+/GFP) in terms of visual acuity. We conclude that Otx2-AA protein activity in vivo is <50% of the wild-type protein.
Otx2 hypomorphs have functional deficits in the retina
Retinal physiological function was assessed by electroretinogram (ERG) measurements under dark adapted (scotopic) and light adapted (photopic) conditions at P30, P60 and P120. In the following description of our data, all changes observed in mutants are in comparison with wild-type mice from the same litter.
Photoreceptor function in the outer nuclear layer (ONL) can be monitored by recording the ERG to measure a-wave amplitude and latency, which are two indexes of photoreceptor number and physiological activity (Fig. 4A and B). Under scotopic conditions, the Otx2+/GFP mice showed a statistically significant decrease in the a-wave amplitude at P30 and P60 when stimulated with an intense flash (Fig. 4A). The decrease in a-wave amplitude became statistically significant for the Otx2AA/AA mice at P60. In contrast, the a-wave amplitudes measured for the Otx2+/AA mice remained indistinguishable from those of their wild-type littermates until P120 when the difference became statistically significant (Fig. 4A). When considering a-wave latencies, their differences were not statistically significant at P30, whereas the values increased significantly at P60 for the Otx2+/GFP and Otx2AA/AA mice (Fig. 4B). A similar increase in latency was observed for Otx2+/AA mice but only at P120 (Fig. 4B). These changes in the latencies are likely to be related to the photoreceptor dysfunction indicated by the a-wave amplitude decreases. However, the change in b-wave (mentioned later) could also induce such effects. These results are consistent with graded alterations in photoreceptor function for these various genotypes with a greater effect in Otx2+/GFP and Otx2AA/AA mice.
To investigate inner retinal layer (INL) function, b-wave amplitude was measured for all genotypes (Fig. 4C and D). The b-wave is often attributed to ON bipolar cell activity because it is suppressed when blocking mGluR6 receptors specific to these cells. Similarly as for the a-wave amplitudes, b-wave amplitudes were strongly decreased in Otx2+/GFP and Otx2AA/AA mice as early as P30, the differences being even greater at P60 for both genotypes (Fig. 4C). At P120, the Otx2+/AA mice also exhibited diminished b-wave amplitudes (Fig. 4C). These decreases in b-wave amplitudes were relatively greater than the decrease in a-wave amplitude leading to a further decrease in the b/a amplitude ratio for the different genotypes considered. Therefore, these ERGs indicated a major dysfunction of the inner retina and most likely of bipolar cells. When considering the b-wave latencies, they were similar in all genotypes at P30 but increased significantly at P60 for the Otx2+/GFP and Otx2AA/AA mice in agreement with bipolar cell dysfunction (Fig. 4D). A small increase in b-wave latency was also seen in Otx2+/AA mice at P120 (Fig. 4D).
To examine further the cone pathway, we recorded photopic ERGs by saturating rods with a background light (Fig. 5). We found that Otx2AA/AA mice had a non-measurable photopic ERG as soon as P30 whereas Otx2+/GFP mice displayed a statistically significant decrease in photopic ERG amplitude as soon as P30 (Fig. 5A). In contrast, Otx2+/AA mice showed normal photopic ERG responses. These data suggested a strong deficit of the cone circuit in Otx2AA/AA mice and a milder dysfunction of this cone pathway in Otx2+/GFP mice.
The latter functional measurements demonstrate a graded retinal physiological dysfunction in the different genotypes with a ranking similar to that obtained in eye development analysis and in the optomotor test. In addition, they also strongly suggest that a decrease in Otx2 dosage/activity leads to a progressive decrease in the number and/or activity of cone pathway neurons, such as cone photoreceptors and cone-bipolar cells, correlated with a diminished physiological performance.
Otx2 hypomorphs have morphological defects in the retina
Hematoxylin and eosin (H&E) staining of retinal sections at P14, P30, P60 and P120 showed no major disorganization or deformation of the eye and retinal structures in Otx2+/+, Otx2+/AA and Otx2+/GFP animals whereas Otx2AA/AA mice with externally normal eyes could present iris defects, rosette-like structures in the retina, festooned RPE and a disorganized optic nerve with variable expressivity: 46% for iris defects, 23% for rosettes, 46% for festooned RPE and 31% for optic nerve defects (data not shown). The images of retinas analyzed later did not contain gross deformities, and the following results are always based on comparisons with wild-type littermates.
At P14, P30 and P60, the thickness of the different retinal layers of Otx2+/AA mice appeared similar to that of wild-type siblings. At P120, the INL and ONL were reduced in thickness (76.2 ± 11.5% of wild-type, n = 4, 81.2 ± 7.3% of wild-type, n = 4, respectively) (Fig. 6). The INL in the retina of Otx2+/GFP mice was already thinner at P14 (85.4 ± 1.2%, n = 3), and this progressively worsened at P30, P60 and P120 (Fig. 6). The Otx2AA/AA mice also showed a thinner INL at P14 (66.9 ± 2.0%) that worsened with age (38.1 ± 2.3% at P120, n = 3). The ONL thickness in Otx2AA/AA retina was comparable with wild-type at P14 and was reduced in thickness from P30 (77.6 ± 1.5%, n = 3). In the ganglion cell layer (GCL), the number of cells was comparable between the four genotypes at P14, P30 and P60. At ages of >10 months, the number of cells in the GCL was significantly reduced in Otx2+/AA (77.9 ± 1.5, n = 4) and Otx2AA/AA (74.8 ± 0.6%, n = 4) mice (data not shown).
These results show that retinal morphology is affected in Otx2 dose-dependent and age-dependent manners. Layer thinning occurs earlier in the INL of Otx2+/GFP and Otx2AA/AA mice compared with Otx2+/AA mice, and the GCL is affected at later ages only in the Otx2+/AA and Otx2AA/AA mice.
Otx2 hypomorphs have fewer photoreceptors, bipolar cells and horizontal cells
We next examined specific cell losses by using immunofluorescence for Otx2 [photoreceptors (PRs) and bipolar cells (BPs)], R/G opsin (red/green sensitive cones), Vsx2 (BPs), Calbindin (horizontal cells), Sox2 (Müller glia), Pax6 (Amacrine cells) and Brn3b (retinal ganglion cells) (Fig. 7A). At P30, there was a significant decrease in the number of Otx2-expressing photoreceptors in the ONL of Otx2AA/AA retina (Fig. 7B). In terms of photoreceptor type, red/green sensitive cones were reduced over 70% in the retina of Otx2AA/AA at this age (Fig. 7B). Cone photoreceptors could therefore account for most of the photoreceptor loss in the ONL at P30. This loss of cone cells is consistent with the decrease or disappearance of the photopic ERG response in this genotype.
At P30, the number of Vsx2-positive bipolar cells was reduced by 35.9% in Otx2+/GFP retina and by 62.6% in Otx2AA/AA retina (Fig. 7B). Otx2-positive cells in the INL were reduced to a similar extent (i.e. 37.8% in Otx2+/GFP retina and 62.9% in Otx2AA/AA retina). This loss of bipolar cells is consistent with the reduction in b-wave amplitudes. Calbindin-positive horizontal cells in the INL were reduced by ∼60% in Otx2+/GFP and Otx2AA/AA mice retinas. The number of Sox2- and Pax6-expressing cells did not significantly differ among the four genotypes at P30 (Fig. 7B). At P30, in the GCL, the numbers of Brn3b-positive RGCs were similar in the three Otx2 mutants and in the wild-type mice (Fig. 7B). These results show that, in the ONL, there are fewer cones in Otx2AA/AA mice at P30 and that, in the INL, there are fewer bipolar cells in Otx2+/GFP retina and even less in Otx2AA/AA retina and fewer horizontal cells in both genotypes. Thus, the reduced number of cells in the retina at P30 is most severe in Otx2AA/AA and intermediate in Otx2+/GFP mice, whereas Otx2+/AA are indistinguishable from Otx2+/+ mice at this age (Fig. 7B).
At ages older than P300, compared with wild-type littermates, the number of R/G-opsin-positive cone photoreceptors was reduced not only in Otx2AA/AA but also in Otx2+/GFP mice by 50.7% and in Otx2+/AA mice by 48.4% compared with wild-type littermates (Fig. 8A and B). These results suggest that mice with a higher dose of Otx2 activity (Otx2+/AA mice) lose retinal cells later than mice with a lower dose of Otx2 activity (Otx2AA/AA mice).
The data on the loss of specific subtypes in the three Otx2 mutants suggest that the ERG perturbation observed in the Otx2 hypomorphic mice might be caused by the loss of bipolar cells and cone photoreceptors. More generally, it can be argued that the cone pathway, composed of cone photoreceptors, cone-bipolar cells and horizontal cells, is most severely affected by the decrease in Otx2 activity. The less Otx2 activity there is, the earlier the loss of retinal cells and changes in retinal physiological activity (Fig. 8C). Pooling the results from all experiments (Fig. 8C) thus confirms the severity ranking Otx2+/+< Otx2+/AA < Otx2+/GFP < Otx2AA/AA and establishes that the observed phenotypes are both Otx2 activity dependent and age dependent.
The product of the Otx2 gene is essential for patterning the anterior neuroectoderm during early embryonic development, and later for the differentiation and maintenance of retinal neurons. Otx2 is also expressed in the post-natal brain where it plays an important role in regulating synaptic plasticity (11). Based on our findings that exogenous Otx2 can stimulate the survival of stressed adult retinal neurons in vitro and in vivo (18), we expected that reduced levels of Otx2 activity would lead to retinal degeneration. Here we examined, in functional and morphological detail, four Otx2 genotypes: Otx2+/+, two knock-in mice with hypomorphic Otx2-AA alleles (Otx2+/AA and Otx2AA/AA) and a GFP knock-in mouse (Otx2+/GFP). The data combined with an analysis of wild-type animals and with the Otx2 null phenotypes allow us to compare mice with six different levels of Otx2 activity.
Previously we showed that a sequence at the N-terminus of the Otx2 homeodomain, particularly an arginine–lysine (RK) doublet at positions 36/37, is required for glycosaminoglycan binding and that this binding confers cell-specific accumulation of the HP in the neocortex with functional consequences (21,26). This led us to develop a mouse line in which the RK doublet is mutated to AA. According to the specific cell surface recognition role of the RK doublet, the Otx2-AA protein is modified in its extracellular binding to cell surface GAGs and shows a decrease in the specificity of Otx2 transfer (21). Although this mouse was intended to study Otx2 non-cell autonomous activity, we could not exclude cell autonomous defects. In fact, we show here that Otx2-AA retains near normal DNA-binding activity but that its transcriptional activity differs from that of the wild-type protein. Thus, our results do not allow us to formally distinguish between cell autonomous and non-cell autonomous activities, leading us to combine the two aspects under the generic term of ‘activity’. However, based on the comparison between Otx2+/GFP and Otx2AA/AA phenotypes, we argue that two active Otx2-AA loci do not add up to one active wild-type locus, thus the in vivo Otx2-AA activity is reduced by >50% overall. Given that Otx2-AA transcription is not grossly perturbed (Fig. 1C), we cannot preclude a non-cell autonomous component. We are currently developing mice to specifically abolish Otx2 intercellular transfer, which will allow us to more precisely investigate the distinction between cell autonomous and non-cell autonomous activities.
Otx2+/AA and Otx2AA/AA mice were crossed with Otx2+/GFP mice to produce Otx2AA/GFP. The analysis of Otx2+/+, Otx2+/AA, Otx2+/GFP, Otx2AA/AA and Otx2AA/GFP mice in terms of visual behavior and retinal function, and anatomical structure of the eyes, demonstrates that the severity of deficits closely follows this order. Phenotype comparison strongly suggests that Otx2AA/AAmice are more affected than Otx2+/GFP mice and thus that the activity of the product of two AA alleles is less than the activity present in the Otx2+/GFP genotype (50% of the wild-type). Although the following numbers are to be taken with caution, we propose that the strengths of the genotypes go from 0% (null) to 100% (wild-type) with intermediate values of ∼20% (Otx2AA/GFP), 40% (Otx2AA/AA), 50% (Otx2+/GFP) and 70% (Otx2+/AA). This gave us an unprecedented possibility for the present gene dosage study.
Mice with no functional allele of Otx2 (i.e. Otx2GFP/GFP) fail to develop the anterior part of the head and are embryonic lethal. Having a single AA allele (Otx2AA/GFP) results in mice born with severe anterior deformities that die at birth. Two AA alleles (Otx2AA/AA) are sufficient for viability and fertility; however, eye development is affected with variable expression. In addition, in these mice, there are important deficits in visual acuity, retinal physiology and retinal maintenance. Previous studies have reported variable craniofacial malformation phenotypes that depended on genetic background. Here we observed that a single wild-type allele (Otx2+/GFP) on a B6D2 background causes no obvious craniofacial phenotype and allows for some visual functions associated with less severe physiological and anatomical deficits. A wild-type allele with an AA allele (Otx2+/AA) provides sufficient Otx2 activity for the mice to be indistinguishable from wild-type mice on a number of measurements. It is important to note that at long survival times, even Otx2+/AA mice manifest functional and structural deficits meaning that full Otx2 activity is necessary for normal development and aging. With regard to younger ages, we observed a normal number of R/G opsin cones in P14 Otx2+/GFP, fewer R/G opsin cones in P14 Otx2AA/AA and fewer BPs in retinas from both Otx2+/GFP and Otx2AA/AA mice (not shown).
In terms of behavior, visual acuity was greatly reduced in the Otx2+/GFP mice at P60. These mice have significantly fewer bipolar cells in the INL; however, there is not frank loss of photoreceptors in the ONL at this age (not shown). The absence of an optokinetic response in the Otx2AA/AA may be attributable to the severe loss of cone photoreceptors that would be active in the light-adapted condition of the test. Alternatively, we cannot rule out that the lack of the optokinetic response is due, at least in part, to dysfunctional central nuclei such as the abducens or vestibular nuclei. In Otx2+/GFP mice, GFP fluorescence appears to be present in all cells in the ONL confirming that the Otx2 locus is normally active in rods and cones.
The b-wave deficit suggests a defect in INL function and the reduced amplitude points toward fewer bipolar cells in the affected mice. This prediction was confirmed by significantly fewer Vsx2-positive bipolar cells in the INL, with the expected order of severity (Otx2+/+< Otx2+/AA < Otx2+/GFP < Otx2AA/AA). The same pattern was observed for Otx2-expressing cells in the INL, and the close correspondence between the numbers of Otx2- and Vsx2-positive cells suggests that most or all bipolar cells express Otx2. The early (P14) thinning of the INL in Otx2+/GFP and Otx2AA/AA mice suggests developmental defects. However, retinal cell types degenerate to some extent in the three Otx2 mutant mice, suggesting dose- and age-dependent effects of Otx2 are also involved. While we found no differences in the number of Müller glia and amacrine cells between the four genotypes, mice with a single wild-type Otx2 allele or two mutated alleles had less than half of the normal complement of horizontal cells. We are currently testing whether horizontal cells express Otx2 or whether this loss results from a direct (Otx2 transport) or indirect non-cell autonomous activity of Otx2. In the former case, it will be important to verify whether the addition of exogenous Otx2 is neuroprotective in mice with a single wild-type Otx2 allele.
R/G opsin-positive cone loss is progressive and occurs late in age (>P300) in Otx2+/AA and Otx2+/GFP mice, whereas in Otx2AA/AA it is observed already at P30. This very specific sensitivity of cones (compared with rods) is surprising. Although we did not count rod photoreceptors, the greatly diminished ONL thickness in aged Otx2AA/AA mice suggests that rods may be vulnerable in this genotype at older ages. With regard to cones, these cells are not affected at P30 in Otx2+/GFP mice but are almost absent in the Otx2AA/AA mice (∼20% remain). If we accept that the difference is very small in terms of Otx2 activity (50 versus 40%) between the two genotypes, this selective death of the cone photoreceptors may be explained in terms of a threshold effect for a cell-specific function of the Otx2-AA protein. Another distinct possibility is the reduced non-cell autonomous specificity of Otx2-AA protein described previously (21).
The issue of Otx2 non-cell autonomous activity will be the focus of a separate study. However, in this context, it is important to note that Otx2 is captured by RGCs where the Otx2 locus is silent and that blocking Otx2 transfer, probably from bipolar cells, retards the opening of the critical period for binocular vision (11). We also showed that, in a mouse model of glaucoma, Otx2 injected in the eye is internalized by the RGCs and protects them from degeneration (18). It is thus possible that the death of RGCs observed after P300 in mice expressing Otx2-AA (this study) is in part owing to a deficiency in Otx2 transfer and RGC Otx2 content.
This study allowed us to follow at different levels (cell survival, physiology, histology and behavior) the phenotypes of mice with a graded reduction in Otx2 activity. An important point is that small differences in gene dosage can have very important consequences. This indeed can translate into a better understanding of pathologies associated with the deregulation of transcription factors. With regard to human pathologies associated with OTX2 mutations, it is interesting to note the remarkable diversity of phenotypes. These include otocephaly-dysgnathia, microphthalmia, anophthalmia, coloboma, hippocampal defects and growth hormone deficiency with associated diminished stature (1–3).
In addition to the wide range of human phenotypes, a highly variable expressivity and incomplete penetrance of these phenotypes is noteworthy. This is particularly well illustrated by cases of affected individual inheriting the OTX2 mutation phenotype from an apparently unaffected parent (3). Even more remarkable is the variable expressivity within an individual with an OTX2 mutation, which was also observed in several Otx2AA/AA mice (Fig. 2D). In one cohort, Ragge and colleagues reported three of eight patients with asymmetric phenotypes that ranged from asymmetric clinical anophthalmia, anterior segment dysgenesis and optic nerve size (1). Two of these patients were of parents with normal wild-type OTX2 alleles. In a more recent cohort (27), four of eight cases manifested asymmetric nystagmus, microphthalmia, anophthalmia or coloboma. The highly variable expressivity observed in humans is well represented in the phenotypes of our hypomorphic mice.
In conclusion, this work shows that transcription factor dysfunction is not an all or none issue. Small differences in the range of 10 to 20% can be significant in terms of the phenotype expressivity and/or severity of the pathology as well as in the timing of its appearance, an observation that may shed new light on pathologic states as well as on normal aging.
MATERIAL AND METHODS
Experiments were carried out according to the guidelines of the European Community Council directives of November 24, 1986 (86/609/EEC) and KAIST IACUC-12–110, and steps were taken to minimize the number of animals used and to reduce pain and discomfort.
The targeting molecule was generated by modification of a previous construct through PCR-mediated mutagenesis (28). Homologous recombinant clones (E14Tg2a4 cell line) were identified by PCR and confirmed by Southern blot using the same probes as previously described (28) (data not shown). The Otx2-AA mouse line was generated by C57BL/6 blastocyst injection, and removal of the selectable neo cassette was performed using a Cre-deleter mouse strain. Four different genotypes were analyzed in this study, Otx2+/+, Otx2+/GFP, Otx2+/AA and Otx2AA/AA. For Otx2+/GFP mice, Otx2+/GFP males were crossed with B6D2F1 females. For Otx2+/AA mice, Otx2+/AA males were crossed with B6D2F1 females to obtain heterozygous mice; for Otx2AA/AA mice, Otx2+/AA males were crossed with Otx2+/AA females. For Otx2AA/GFP mice, Oxt2AA/AA males were crossed with Otx2+/GFP females. All animals have a B6D2 genetic background. In all tests performed, there were no differences between the wild-type Otx2+/+ animals from different litters.
Tail biopsies were digested with proteinase K in DirectPCR tail buffer (Viagen) at 55°C for 16 h. After inactivation of the enzyme at 85°C for 45 min and centrifugation, 1 μl of the supernatant was analyzed by PCR. For Otx2+/GFP mice, the primers Mal (5′ ACT CCA GGC GAA TCG AGA CCG TC) and GFP-Q (5′ CTT GAA GAA GTC GTG CTG CTT CA) were used. For Otx2+/AA and Otx2AA/AA mice, the primers AA-Fw (5′ ACT TGC CAG AAT CCA GGG TGC AG) and AA-Rv (5′ CCA GGC TAA AAG ACC CTG GTT C), which give a 200-bp band for wild-type Otx2 and a 290-bp band for Otx2-AA, were used.
Production of recombinant proteins
Recombinant protein with a myc tag and a 6xHis tag in C-terminal position was cloned into pTrac plasmid (Life Tech). Recombinant protein was produced in bacteria BL21 CodonPlus RP (Stratagene) at 37°C overnight in auto-induced medium (MagicMedia, Life Tech). After washing the bacterial mass with PBS, the protein was extracted with buffer A+ (10 mm Tris, 100 mm Na2HPO4, 20 mm imidazole, 6 m guanidine HCl, pH 8). The lysed bacteria were centrifuged for 20 min at 20 000 g and the supernatant passed through an affinity chromatography column charged with 100 mm NiSO4 (HiTrap Chelating HP column, GE Healthcare). The recombinant protein was then eluted with 6 m guanidine HCl, 0.2 m acetic acid and immediately dialyzed against 20 mm phosphate and 0.5 M NaCl. The protein solution was complemented to 10% glycerol and frozen.
Electrophoretic mobility shift assay
EMSA was performed with recombinant Otx2 and Otx2-AA proteins produced in bacteria with the LightShift Chemiluminescent EMSA kit (Pierce) following the manufacturer's instructions using a synthetic biotinylated DNA oligonucleotide corresponding to the Rbp3 promoter (22).
Transfections and luciferase assay
The secreted alkaline phosphatase reporter construct Rbp3 prom/pSEAP2-Basic vector was a generous gift from Dr. Lamonerie (22). The reporter construct was co-transfected with Otx2 or Otx2-AA expression vector (pCL-Otx2 or pCL-Otx2-AA) into 293T human embryonic kidney cell line by Lipofectamine 2000. Forty-eight hours post-transfection, secreted alkaline phosphatase was analyzed using the Great EscAPe SEAP Chemiluminescence Assay according to the manufacturer's protocols (Clontech).
Visual acuity was assessed by optomotor response as described previously (18,29) with modifications. Briefly, mice adapted to ambient light were placed on a raised grid platform centered in a well-lit motorized drum with 100% contrast black and white vertical stripes. The drum rotated at 2 rpm, with spatial frequencies from 0.75 to 0.125 cpd. The number of head turns was counted during clockwise and counter-clockwise rotation of the drum during 1 min in each direction.
Animals were anesthetized with a mix of xylazine/ketamine (Imalgène 500 Virbac France, 100 mg/kg, Rompun 2% Bayer, 10 mg/kg), the iris dilated with tropicamide (Mydriaticum 0.5% Théa, France) and the cornea anesthetized locally with oxybuprocaine hydrochloride. Gold electrodes were placed in contact with each eye, reference electrodes were placed subcutaneously in the submandibular area, and a ground electrode was placed subcutaneously on the back of the animal. ERG was performed in five animals of each genotype with a mobile apparatus (SIEM Bio-Medicale, France) with LED lamps in a Ganzfeld chamber controlled by the VisioSystem software.
Immunostaining and cell counts
At the indicated ages, mice were sacrificed by cervical dislocation; the eyes were removed and fixed in 4% PFA, 2% ZnCl and 20% isopropyl alcohol for at least 48 h before processing. The eyes were included in paraffin and sectioned at 5 μm around the optic nerve. At least six slides per eye were obtained, and one of them was stained in H&E (Excalibur Pathology Inc., Oklahoma City, USA).
For immunohistochemistry, the deparaffinized slides were blocked with 5% normal donkey serum and 5% normal goat serum in PBS/0.2% Triton X-100 prior to incubating with appropriate primary antibodies at 4°C for 16 h. The following primary antibodies were used in this study: anti-Otx2 (1:400, Millipore, AB-9566), anti-R/G opsin (1:400, Millipore, AB5405), anti-Rhodopsin (1:1000, Millipore, MAB5356), anti-Vsx2 (1:300, Santa cruz, sc365519), anti-Brn3b (1:400, Santa Cruz Biotechnology, sc31989), anti-Pax6 (1:400, covance, PRB-278P), anti-Calbindin (1:500, Sigma-Aldrich, C9848) and anti-Sox2 (1:200, Santa Cruz Biotechnology, sc17320). Immunolabeled cells were detected by fluorescence-conjugated donkey IgG (1:300, Jackson Immunoresearch Laboratories) and subsequently analyzed by Olympus FV1000 confocal microscope.
Statistical analysis has been performed with Prism software (version 5.0a). Pair wise comparison was done using Student t-test. A difference was considered significant at P < 0.05.
Overall phenotype ranking
All data at every age were determined in percentage of wild-type: for one age and one experiment (for instance the optomotor test at P60 or the number of Vsx2-positive cells at P30), the Otx2+/+ mice response was considered to be 100%, and the responses from the mutant mice were calculated in percentage of the wild-type. For each genotype, at every age, we calculated the mean and standard error of all percentages. These means were plotted together on one graph from P30 to P365 (Fig. 8C).
This work was funded by the Agence Nationale pour la Recherche (ANR-11-BLAN-069467 to A.P.), Fovea Pharmaceuticals (to A. P. and K. L. M.), the Global Research Laboratory Program (NRF-2009-00424 to J.W.K. and A.P.), Brain Research Program (NRF-2009-0081465; NRF-2013-056566 to J.W.K.), Drug Target Validation (2010-0029943 to J.W.K.) and Stem Cell Research (NRF-2006-2004289 to J.W.K.) funded by the Korean Ministry of Science, ICT, and Future Planning (MSIP) and the Italian Association for Cancer Research (AIRC) (IG-5499) (to D.A. and A.S.).
We thank Dr. Nicolas Chassain of the Service de Génétique Médicale and the Hôpital Purpan CHU Toulouse for valuable discussions.
Conflict of Interest statement. A. P. and K. L. M. are co-inventors for the use of recombinant Otx2 for the treatment of glaucoma. Portions of this work were financed by a research contract from Fovea Pharmaceuticals.