Ethics statement

All procedures for the care and housing of mice conform to the U.S. Public Health Service Policy on the Humane Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Iowa State University.

Tissue preparation

Whole eyes were removed and fixed in 4% paraformaldehyde/phosphate buffered saline (PBS) at 4°C for 1 hour. Following fixation, eyes were rinsed 3 times in PBS at 4°C for 10 minutes. The sclera was dissected away from the intact lens and retina. Retinas (with lens intact) were post-fixed in 4% paraformaldehyde/PBS at 4°C for 15 minutes and then rinsed 3 times with PBS. The lens was removed and the retina taken for further processing depending upon the particular assay.

Whole-mount immunohistochemistry

Dissected retinas were equilibrated at 4°C in sucrose solutions of increasing concentrations in a step-wise manner, 10%, 20%, 30% sucrose (w/v) in PBS, for at least 30 minutes or until the retina had settled to the bottom of the tube. While the retinas were in 30% sucrose/PBS, they were snap-frozen on dry ice and subjected to three freeze/thaw cycles. Retinas were stored at −80°C in the 30% sucrose/PBS solution or used for immunostaining immediately. Following freeze/thaw, retinas were rinsed 3 times in PBS for 30 minutes and blocked for 2 hours at room temperature (RT) in blocking solution [3% goat serum/1% bovine serum albumin (BSA)/0.1% Triton-X100/0.02% sodium dodecyl sulfate (SDS) in PBS]. Retinas were then incubated in primary antibody in blocking solution overnight at 4°C on a rocking platform. The next day, retinas were rinsed 3 times in PBS for 30 minutes and placed in secondary antibody in blocking solution overnight at 4°C on a rocking platform. Retinas were rinsed 3 times in PBS for 30 minutes and then flattened between two coverslips for confocal imaging on a Leica SP5 X MP confocal microscope.

Tissue sectioning

Dissected retinas were equilibrated in 30% sucrose/PBS at 4°C until they sank to the bottom of the tube. They were then frozen in an equilibrated solution of 50% (30% sucrose/PBS): 50% optimal cutting temperature compound (OCT, Tissue-Tek; Sakura Finetek, Torrance, CA) and stored at −80°C. Cryosections were cut at 20 µm, collected on Superfrost Plus slides (Fisher Scientific), and air-dried for 30 minutes at RT before storage at −80°C.

Section antibody staining

Sections were incubated in blocking solution (1% BSA/0.1 Triton-X100 in PBS) for 1 hour at RT and then in primary antibody in blocking solution overnight at 4°C. Slides were rinsed in blocking solution 3 times and incubated with secondary antibody in blocking solution for 2 hours at RT. Slides were rinsed with PBS, mounted in Fluoromount-G and visualized on a Leica SP5 X MP confocal microscope.

Primary antibodies used were anti-Calbindin28K (1∶2000) (Swant, Switzerland), anti-Chx10 (1∶1000) [30], anti-Glutamine Synthetase (1∶10,000) (Sigma), anti-Rhodopsin (1∶100) [31], anti-GFAP (1∶100) (Abcam), and anti-HNF-6/OC1 (1∶200) (Santa Cruz Biotechnology). The anti-Pax6 (1∶50) and anti-Islet1 (1∶50) antibodies were obtained from the Developmental Studies Hybridoma Bank (DHSB), developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biology, Iowa City.

Quantitative immunohistochemical analysis

For analysis, identically-sized fields of immunostained retinal cryosections were counted by hand in Adobe Photoshop, blinded to genotype. After counts were made for WT and KO fields (n = 3 biological replicates for both - except in the case of Calb28k, in which 4 fields of both amacrine cell staining and horizontal cell staining were counted), t-tests were performed to compare counts of the WT and KO fields. Error bars indicate standard error.

Section in situ hybridization

Microarray experiments

qPCR

RNA was isolated from individual mouse retinas at various ages using TRI-reagent (Sigma). cDNA was generated from 400 ng of retinal RNA using random primers and SuperScript III (Life Technologies) according to the manufacturer’s instructions. qPCR was performed using SYBRGreen MasterMix (Thermo) and a BioRad CFX96 Real Time System with BioRad C1000 Thermal Cycler using the following program: 15′ at 95°C and 40 cycles of 15 sec at 95°C, 30 sec at 56°C, and 30 sec at 72°C. Each sample was normalized to β-Actin primers. The analysis of the qPCR data was performed exactly as in Livak and Schmittgen (Livak and Schmittgen, 2001). Briefly, the difference in average ΔΔC(t) values and the difference plus and minus the standard error of the difference were computed on the C(t) scale. The base-2 antilogs of these three values were computed to obtain values with error bars on the fold change scale. Results were plotted on a logarithmic scale.

Mouse genotyping

OC1 and OC2 deficient mice were genotyped with primer pairs for the WT and null alleles. OC1-WT: 5′-CAGCACCTCACGCCCACCTC-3′, 5′-CAGCCACTTCCACATCCTCCG-3′; OC1-KO: 5′-CTGTGCTCGACGTTGTCACTG-3′, 5′-GATCCCCTCAGAAGAACTCGT-3′. OC2-WT: 5′-GCCACGCCGCTGGGCAAC-3′, 5′-CAGCTGCCCGGACGTGGC-3′; OC2-KO: 5′-GACCGAGTACAAGCCCACG-3′; 5′-GTCCGCGACCCACACCTT-3′. Reactions containing all four primer pairs for each mouse strain were amplified by 35 cycles of PCR, with the same program for both strains (94°C for 30 seconds, 60°C for 30 seconds, 72°C for 30 seconds, with a 3 minute initial denaturation at 94°C). OC1 PCR reactions included WT primers at 0.2 µM each and null allele primers at 1.6 µM each. OC2 PCR reactions included all WT and null allele primers at 0.8 µM, as well as 1X MasterAmp (Epicentre). OC1 genomic DNA was isolated by adding 200 ul 50 mM NaOH to tissue samples (ear clips for live mice, confirmed by tail clips upon harvesting) and incubating at 95°C for 30 minutes, followed by the addition of 50 ul 1 M Tris pH 8 and centrifugation for six minutes at 13,300 rpm at 4°C. OC2 genomic DNA preparation required phenol-chloroform (1∶1) extraction followed by ethanol precipitation.

Retinal phenotypes in the OC1-deficient mouse

Previous studies have demonstrated that OC1 is a critical transcription factor in liver development (Clotman, et al., 2002). Due to this important function, most OC1-KO mice die either at birth or in the immediate postnatal period. Despite the high mortality rate resulting from complete loss of OC1, we were able to obtain a few mature, postnatal day (P)21 OC1-KO mice. Since the effects of OC1 deficiency on retinal development had not previously been studied in this particular mouse line, we examined whether mature retinal cell populations were present in the absence of OC1. To specifically determine any deficits in populations of retinal cells, P21 OC1-KO retinas were surveyed using immunohistochemistry with antibodies that mark each subset of retinal cells (Figure 2). Compared to wildtype littermate control animals (Figure 2A–E), populations of rod photoreceptors (Figure 2A’), amacrine cells (Figure 2B’), bipolar cells (Figure 2C’,D’), and Muller glia (Figure 2E’) were unchanged in adult OC1-KO mice. OC1 has previously been shown to play an important role in the development of horizontal cells using a conditional knockout mouse model [29]. Consistent with this report, a significantly decreased horizontal cell population was readily apparent in our P21 OC1-deficient mice (Figure 2F, F’).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Immunohistochemistry of Adult OC1 Knockout Retinas.

Changes in mature retinal cell populations resulting from the loss of OC1 were examined using immunohistochemistry. Most retinal populations were unchanged between wildtype and OC1-KO littermates, including rod photoreceptors (A,A’), amacrine cells (B,B’,C,C’), bipolar cells (D,D'), and Muller glia (E,E’). However, anti-Calb28k-staining shows a decrease in horizontal cells in OC1-KO retinas as compared to a wildtype littermate (F,F'). Arrowheads indicate horizontal cell nuclei. Scale bars represent 100 µm.

https://doi.org/10.1371/journal.pone.0110194.g002

Retinal phenotypes in the OC2 deficient mouse

We observed that OC2 is expressed at developmental timepoints and in cell populations reminiscent of its family member, OC1 (Figure 1). Therefore, we next explored the consequences resulting from the loss of OC2 during retinal development. To determine whether the development of retinal cell populations was disrupted by loss of OC2, adult (P21) retinal cell populations of OC2-deficient mice were surveyed by immunohistochemistry using a similar panel of antibodies as in the OC1-deficient mouse retinas (Figure 3). No changes were observed in populations of rod photoreceptors (as marked by αRho4d2, Figure 3A, A’), amacrine interneurons (αPax6, Figure 3B, B’), bipolar cells (αIsl1, Figure 3C, C’; αChx10, Figure 3D, D’), or Muller glia (αGS, Figure 3E, E’). However, horizontal cell populations were significantly decreased in OC2-KO, to a similar degree as observed in the OC1-KO (Figure 3F, F’). To ensure there were not any small, but significant differences in these cell types, cell counts were performed on sections from n = 3 different wildtype and OC2-KO mice. No significant differences were observed for any cell types except the horizontal cells (Figure 3G). To further confirm the results obtained through immunohistochemical analysis, we performed in situ hybridizations on cryosections from P21 OC2-KO retinas. We did not observe any significant differences between OC2-KO animals and their wildtype littermates in either the rod or cone photoreceptor populations (Figure S1 A–C’), bipolar cells (Figure S1D,D’), GABAergic (GAD1– Figure S1E,E’) and glycinergic (SLC6A9– Figure S1F,F’) amacrine interneurons (Figure S1G,G’,H,H’), ganglion cells (Figure S1I,I’), and Muller glia (Figure S1J,J’). However, ISH did reveal a significant decrease in horizontal cells (Figure S1K,K’,L,L’).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Immunohistochemistry of Adult OC2 Knockout Retinas.

Changes in mature retinal cell populations resulting from the loss of OC2 were examined using immunohistochemistry. Rod photoreceptors (A,A'), amacrine cells (B,B',C,C'), bipolar cells (D,D'), and Muller glia (E,E') were unchanged, whereas the horizontal cell population was greatly decreased (F,F'; arrowheads indicate horizontal cell bodies). Scale bars represent 100 µm. (G) The results of the immunohistochemical analyses were quantified for identically-sized fields of cryosectioned retinal tissue. Although staining for Brn3b, Chx10, Pax6, and the Calb28k that marks amacrine cell bodies were unchanged, horizontal cell bodies marked by Calb28k were significantly decreased in OC2-KO retinas (n = 3, p<0.005).

https://doi.org/10.1371/journal.pone.0110194.g003

To gain perspective on the overall patterning of horizontal cell loss, whole, flat-mounted retinas were stained using an antibody against the horizontal cell marker Calbindin-28k (Calb28k) (Figure 4). Whole mount staining showed a significant loss of the horizontal cell population across the entire retinas of both the OC1 (Figure 4A, A’) and OC2-deficient mice (Figure 4B, B’). In P21 OC2-deficient retinas, horizontal cells were decreased to about 20% of the levels seen in wildtype littermates (n = 3, p = 2e⁻⁷). Given the difficulty of obtaining OC1 mice at maturity we were only able to quantify one animal, but the horizontal cells were decreased by 35%. These observations demonstrate show that loss of either Onecut1 or Onecut2 interferes with either the generation or maintenance of horizontal cells in the murine retina.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Assessment of horizontal cells by flat mount retina staining.

To better understand the distribution of horizontal cell loss in the OC1 and OC2-KO retinas, immunohistochemistry was performed using an anti-Calbindin 28k antibody on age-matched WT (A,B), OC1-KO (A’), and OC2-KO (B’) retinas. Scale bars represent 100 µm.

https://doi.org/10.1371/journal.pone.0110194.g004

In a previous study, Onecut1 deficient retinas were found to exhibit degeneration of the photoreceptor layer (ONL) by 8 months of age [29]. Therefore, we examined retinas of OC2-deficient mice for possible ONL degeneration at two different ages (Figure S2). Staining for Calbindin-28k indicated that the decrease in horizontal cells in OC2-KO mice compared to WT mice was maintained well into maturity (Figure S2). However, antibody staining for rod photoreceptors in the OC2-KO appeared grossly normal at 12 months of age when compared to a wildtype littermate (Figure S2A,A’). Conversely, a 16-month old OC2-deficient mouse displayed a thinner layer of rhodopsin staining compared to a WT age-matched littermate (Figure S2B,B’). These results indicate that although the horizontal cells are similarly decreased in both Onecut1 and Onecut2 deficient mice, the photoreceptor degeneration is much slower and less pronounced in the OC2-KO mouse compared to the OC1-KO.

We have observed that mature mouse retinas are either unable to generate or maintain wildtype levels of horizontal cell populations in the absence of either OC1 or OC2. To better understand the developmental stage when this phenotype becomes apparent, we investigated OC2 deficient retinas at various stages of maturity, starting with P5–10 (Figure 5). At these stages, nearly all retinal cells, including horizontal cells, are postmitotic [13]. Immunohistochemical analysis showed that at P5, the development of other retinal cells, including the late-developing rod photoreceptors (αRho, Figure 5A, A’) and bipolar cells (αChx10, Figure 5B, B’) appear normal. However, postnatal retinas lacking Onecut2 already show a marked decrease in horizontal cells both at P5 (Figure 5C, C’) and at P10 (Figure 5D, D’). These results suggest that the loss of horizontal cells in OC2-deficient retinas may not be due to a lack of horizontal cell maintenance, at least in the postnatal stages of development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Examination of early postnatal stages in OC2-KO retinas.

To determine whether this decrease in horizontal cells was confined to the fully mature retina, or was present earlier in development, immunohistochemistry was performed on the early postnatal retina. Age-matched animals were processed for either retinal sections (P5) or flatmounts (P10) and stained with an anti-Rhodopsin antibody (A,A’), an anti-Chx10 antibody (B,B’), or an anti-Calbindin 28k antibody (C,C’,D,D’). Scale bars represent 100 µm.

https://doi.org/10.1371/journal.pone.0110194.g005

It is apparent that horizontal cell populations are disrupted before retinal maturation is completed in OC2-KO retinas. However, it is unclear whether the progenitor cells that would normally become horizontal cells undergo a fate change and become other retinal cell types, or if they differentiate normally into horizontal cells only to die in the absence of OC2. Horizontal cells are among the earliest born cell types in the retina, so a cell-fate change might be predicted to affect the population sizes of other early-born neurons. To examine whether a cell fate change was apparent in early retinal differentiation, ISH was performed at E16.5 (Figure S3). A high percentage of early-born retinal cells including cone photoreceptors, some amacrine interneurons, and ganglion cells were generated by this late embryonic stage. There were no significant differences in progenitor cells at E18.5 as visualized by Chx10 (Figure S3A,A’), photoreceptor precursor cells (Prdm1, Figure S3B,B’; Otx2, Figure S3C,C’; Gnb3, Figure S3D,D’), amacrine cell precursors (Tcfap2b, Figure S3E,E’), and developing ganglion cells (Sncg, Figure S3F,F’) in the absence of OC2.

Although changes in early-born retinal populations are not apparent at E16.5 in OC2-KO mice, we were also interested in visualizing the cell types generated in the early stages of retinogenesis in OC1-KO mice. We performed in situ hybridization at E16.5, a timepoint when horizontal cells, ganglion cells, cone photoreceptors, and amacrine cells are being actively produced (Figure S4). As in the OC2-KOs, no changes were identified during mid-embryonic stages in the developing photoreceptors as visualized using probes to Prdm1, Gnat2, and Gnb3, (Figure S4A,A’–C,C’). Additionally, no significant changes were noted in amacrine cells (Tcfap2b, Figure S4D,D’) or ganglion cells (Sncg, Figure S4E,E’). Horizontal cells are an extremely small population of neurons that are generated over multiple days during murine retinogenesis [15] and are, therefore, difficult to track over the span of their generation. We have not observed any changes in other retina cell populations or indications of increased retinal cell death. However, the effect of the loss of cells or fate-switch in such a small population may not generate statistically significant changes in cell death or cell-fate change at any single point over the developmental timeline.

Global gene expression changes in Onecut2-deficient mouse retina

To assess global changes in gene expression that were not immediately apparent from our visual inspection of retinal cell populations, we performed microarray analysis on adult OC2 retinas (data available on NCBI GEO: reference number GSE57918). Retinas were dissected from OC2-KO animals and their wildtype littermates and RNA was isolated, then reverse-transcribed to cDNA and hybridized to Affymetrix microarrays. We first examined genes that were downregulated in adult OC2-KO retinas (Table S1). Our prediction was that genes with lower expression would point to any additional cellular deficiencies in the OC2-KO mice. Consistent with our previous observations regarding a significant deficit in horizontal cells in these mice, the genes Calb1 and Septin4 [32] were significantly downregulated (Table S1). Surprisingly, the signals for two bipolar cell genes, Pcp2 and Car8 [33], were also significantly downregulated, pointing to a possible subtle bipolar cell phenotype in the OC2-KO mice.

The LIM1 class transcription factor Lhx1 is specifically expressed in differentiating and mature horizontal cells [34]. We examined our microarray data and found that while the expression of Lhx1 was decreased, it did not reach significance (p = 0.159). To ascertain whether Lhx1 was in fact downregulated in OC2-KO retinas, we performed qPCR and found that Lhx1 expression was significantly decreased in the OC2-KO adult retina (Figure 6A). To further pinpoint when this downregulation occurs, we performed qPCR for Lhx1 on WT and OC2-KO retinas at E16.5. At this timepoint, we also observed that Lhx1 expression was significantly lower in the OC2-KO than in the WT (Figure 6A), suggesting that horizontal cells are decreased in these mice from very early on in development. Interestingly, our data suggests that this regulation of Lhx1 by the Onecut factors may be dosage dependent. Removal of just a single copy of either OC1 or OC2 (data not shown) reduced the amount of Lhx1 present. This result was consistent, but not high enough to reach statistical significance. Together, these microarray and qPCR results point to an early defect in horizontal cells and a possible bipolar cell phenotype in the OC2 deficient mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Unbiased search for changes in the OC2-deficient retina.

(A) Quantitative PCR (qPCR) analyses determining the relative amounts of Lhx1 in OC1-KO, and OC2-KO retinas at E16.5 or adult compared to WT littermate retinas. (B) qPCR determining the relative amounts of various genes identified in models of retinal damage or disease expressed in adult OC2 retinas. Each marker known to be involved in retinal stress is significantly upregulated in OC2 compared to their WT littermates, including GFAP, Edn2, C4b, and Cebpd. (C) WT and OC2-KO retinas were stained with an anti-GFAP antibody. Scale bars represent 100 µm. (D) qPCR determining the relative amounts of Opn4 in E14.5 OC1-KO retinas compared to WT littermate retinas. * indicates p<0.05, ** indicates p<0.01, *** indicates p<0.005. Scale bars represent 100 µm. All quantitative results are plotted on logarithmic scale.

https://doi.org/10.1371/journal.pone.0110194.g006

Next, we examined genes that were upregulated in the absence of OC2 (Table S2). Curiously, OC2 was one of the most highly upregulated genes in OC2-KO retinas as compared to their WT littermates. To understand this conundrum, we investigated the location of the sequences used on the microarray and ISH-probes and found that they were located at the 3′ end of the Onecut2 gene. To see if this upregulation was consistent throughout the entire sequence, we targeted the 5′ coding sequence of the Onecut2 gene by qPCR and found it to be virtually undetectable (data not shown). Additionally, OC2 protein as visualized through immunohistochemistry was not observed in the OC2-KO retinas (data not shown). These observations suggest that the remaining leftover OC2 sequence in the OC2-KO mouse is upregulated at the RNA level, but that no functional protein is produced.

Another highly upregulated gene observed in OC2-KO retinas was glial fibrillary acidic protein (GFAP), a marker of activated gliosis [35]. To confirm the upregulation of GFAP, qPCR was performed on OC2-deficient retinas. A twofold upregulation of GFAP was observed (n = 7 WT, n = 9 KO animals; p = 0.002; Figure 6B). Antibody staining for GFAP revealed that the processes of Muller glia have upregulated the protein in OC2-KO retinas when compared with the WT littermates (Figure 6C). To get a greater sense of other changes present in the KO retinas, especially those related to a possible stress response, we explored changes in other genes identified to be upregulated in models of retinal disease. Indeed, further investigation of markers of retinal stress indicated that adult OC2 retinas overexpress numerous genes implicated in stress or injury-related contexts ([36]–[39], such as Edn2, C4b, and Cebpd, in both microarray and follow-up qPCR (Figure 6B). In addition, CD44, a Muller glia expressed gene that is increased upon retinal degeneration [40], is upregulated in OC2-KO retina microarrays (Table S2). Despite the normal rhodopsin staining observed in OC2-KO retinas, the upregulation of these stress response and gliosis-associated genes suggest that retina health is compromised in the absence of OC2 and/or horizontal cells.

To ascertain any developmental phenotypes in the OC1-KO retinas, microarrays were performed on retinas isolated from E14.5 mice (data available on NCBI GEO: reference number GSE57917). Among the genes that were consistently upregulated, two of them stood out. Gephyrin and Sox4 were significantly upregulated in the E14.5 OC1-KO retinas (Table S3). Given the expression of these genes in ganglion and amacrine cells [41], [42], their upregulation may point to a possible subtle phenotype in these cells. One interesting gene that was observed to have a higher, but not quite statistically significant, expression in the E14.5 OC1-KO versus WT was melanopsin (Opn4). Opn4 encodes a photopigment protein that is expressed in a small subset of intrinsically photosensitive retinal ganglion cells [43]. While much research has been done on the function of these cells, very little is known about how these cells are generated and how the Opn4 gene is regulated. Since the differential expression of Opn4 only approached significance (p = 0.08), we examined the expression of Opn4 by qPCR in WT and OC1-KO retinas at E14.5. By this assay we observed a statistically significant upregulation of Opn4 (Figure 6D). Surprisingly, when we examined OC1-KO retinas at slightly later timepoints (E16.5 and E17.5) we no longer observed this upregulation of Opn4 (data not shown); this suggests that other factors may be able to compensate for the loss of OC1. These data identify the Onecut1 transcription factor as a possible repressor of Opn4 expression.

We also examined the genes that were downregulated in E14.5 OC1-KO retinas when compared to WT littermates (Table S4). While most of the genes that were downregulated have unknown retinal functions, two genes, Irx4 and Scratch2, stood out. These two genes have been associated with developing retinal ganglion cells [44], [45], suggesting that there may be subtle defects in retinal ganglion cell development or differentiation in the OC1-KO retinas that warrant further investigation.