Targeting of GFP to newborn rods by Nrl promoter and temporal expression profiling of flow-sorted photoreceptors

Abstract

The Maf-family transcription factor Nrl is a key regulator of photoreceptor differentiation in mammals. Ablation of the Nrl gene in mice leads to functional cones at the expense of rods. We show that a 2.5-kb Nrl promoter segment directs the expression of enhanced GFP specifically to rod photoreceptors and the pineal gland of transgenic mice. GFP is detected shortly after terminal cell division, corresponding to the timing of rod genesis revealed by birthdating studies. In Nrl-/- retinas, the GFP+ photoreceptors express S-opsin, consistent with the transformation of rod precursors into cones. We report the gene profiles of freshly isolated flow-sorted GFP+ photoreceptors from wild-type and Nrl-/- retinas at five distinct developmental stages. Our results provide a framework for establishing gene regulatory networks that lead to mature functional photoreceptors from postmitotic precursors. Differentially expressed rod and cone genes are excellent candidates for retinopathies.

Fig. 1.

Nrl promoter directs GFP expression to rods and pineal gland in transgenic mice. (a) Nrl-L-EGFP construct. The upstream Nrl segment contains four sequence regions I–IV that are conserved between mouse and human. E1 represents exon 1. (b) Immunoblot of tissue extracts (as indicated) using anti-GFP antibody, showing retina-specific expression of GFP in the Nrl-L-EGFP mouse. Transgenic mice generated with smaller constructs lacking one or more conserved promoter regions revealed aberrant or no expression of GFP (data not shown). (c) GFP expression in the pineal gland of Nrl-L-EGFP transgenic mice. (d) GFP expression in outer nuclear layer (ONL) of entire adult retina with (e) some nonfluorescent cells in the outer part of the ONL. (f–h) Immunostaining with rhodopsin antibody (red) showing a complete overlap with GFP (green) expression. (i–k) Cells positive for the cone-specific marker peanut agglutinin (red) do not overlap with GFP (green)-expressing cells. (l–n) Immunostaining with cone arrestin (red) reveals no overlap with GFP (green). Arrowheads indicate cone photoreceptor cells. As shown, GFP specifically labels the rod population in the retina. RPE, retinal pigment epithelium; OS, photoreceptor outer segments; IS, inner segments; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. [Scale bar, 100 μm (c), 500 μm (d), and 25 μm (e–n).]

Fig. 2.

The time course of GFP expression corresponds to rod cell birth in developing mouse retina. (a) RT-PCR analysis showing the expression of Nrl and Rho transcripts in developing and adult mouse retina, compared to an Hprt control. E and P indicate embryonic and postnatal day, respectively. W and M represent age in weeks and months, respectively. (b) GFP expression is first observed at E12 in a few cells with longer exposure (b′). (c and c′) Short and long exposures at E14, respectively. (d–g) Progressive increase in the intensity and number of GFP-expressing cells from E16 to P4. (h) Low-magnification view at E16 showing a dorsoventral gradient of GFP expression. (i) Timeline of rod photoreceptor birthdates (green area), major developmental events, and the kinetics of Nrl and rhodopsin (Rho) gene expression. VZ, ventricular zone; NBL, neuroblastic layer. [Scale bars, 25 μm (b–g) and 500 μm (h).]

Fig. 3.

GFP is expressed shortly after cell cycle exit. (a–c) E16 retinas from the wt-Gfp mice immunostained with antiphosphohistone H3 (pH3) (red) and anti-GFP (green) antibody. There is no colocalization, indicating that GFP+ cells are not in M-phase. (d–l) BrdUrd labeling experiments. (d–f) One hour after BrdUrd injection, no GFP+ cells (green, arrowheads) were labeled with BrdUrd (red), demonstrating that GFP+ cells are not in S-phase. (g–i) After 4 h, a small number of colabeled cells (arrows) were observed, indicating that GFP expression starts ≈4 h after the end of S-phase. (j–l) The number of colabeled cells increased 6 h after BrdUrd injection. VZ, ventricular zone; RPE, retinal pigment epithelium. (Scale bars, 10 μm.)

Fig. 4.

GFP colocalizes with S-opsin in photoreceptors of the Nrl-ko-Gfp retina. (a) wt-Gfp and Nrl-ko-Gfp retinas (at P6) were immunostained with anti-S-opsin antibody. GFP and S-opsin are colocalized in the Nrl-ko-Gfp but not in the wt-Gfp mouse retina. (b) Dissociated cells from the P10 Nrl-ko-Gfp mouse retina were immunolabeled with S-opsin antibody. Bisbenzimide labels the nuclei. All GFP+ cells express S-opsin. However, ≈40% of S-opsin+ cones do not express GFP. This may reflect the loss of GFP during dissociation and immunostaining; decreased GFP expression in the absence of Nrl, which can activate its own promoter in mature rods (unpublished data); and/or contributions from the cohort of normal cones. Thus, GFP+ cells from the wt-Gfp and Nrl-ko-Gfp retina represent pure populations of rods and cones, respectively. [Scale bars, 50 μm (a) and 10 μm (b).]

Fig. 5.

Gene profiles of FACS-purified GFP+ photoreceptors reveal unique differentially expressed genes and significant advantages over whole retina analysis. (a) Bitmap for gene expressions. The 45,101 probesets were determined as present (black) or absent (white) at each of five developmental stages; all genes were assigned to one of the 2⁵ = 32 possible expression clusters, which are represented by black/white patterns and correspond to 32 rows in the bitmap. The bitmap of gene expression profiles for wild-type developing rods is shown, with the number of genes in each cluster indicated. The boxed clusters represent molecular signatures for each developmental stage. A similar bitmap was generated for developing cones from the Nrl-ko-Gfp retina (not shown). (b) Comparison of gene profiling data from FACS-purified photoreceptors (reported here) with those from the whole retina (26). The two data sets (red from this report and green from ref. 26) were analyzed by using FDR-CI with 2-fold maximum acceptable difference (MAD) constraint. The horizontal axis represents the sorted gene index according to FDR P values, and the vertical axis represents FDR P values. At similar FDR P values, >10 times more differentially expressed genes are extracted in the profiling data reported here compared to Yoshida et al. (26), thereby allowing for much stronger discovery power. (c) SOM clustering of selected wt (wt-Gfp) gene expression profiles. Clusters of top 1,000 differentially expressed genes over five developmental stages were projected onto a 2D 2 × 4 grid. These were selected empirically, for maximal changes in expression levels over time, to capture biologically nonredundant patterns of interest. Within each image, expression levels are shown on y axis and the five developmental stages (in a) are shown on x axis from left to right (from earliest to latest). The middle curve (blue) is the mean expression profile of genes in that cluster, and the upper/lower curves (red) show the standard deviation (±). The cluster index (c#) and the number of genes in each cluster are indicated. The cluster containing rhodopsin (highlighted in yellow) includes genes whose expression increases progressively as photoreceptors mature, from P6 to adult. (d) SOM clustering of selected Nrl^−/− (Nrl-ko-Gfp) gene expression profiles. The details are essentially the same as in c. The cluster containing S-cone opsin and genes involved in cone maturation is highlighted in yellow.

Fig. 6.

Cluster analysis of differentially expressed genes. (a) Hierarchical clustering of top 1,000 differentially expressed genes across wt, Nrl-ko, and five developmental stages, selected by two-stage filtering. Bright blue boxes indicate lowest signal with increasing values indicated by darkening color toward bright yellow, representing peak signal. (b) Cluster I includes genes that exhibit increased expression during cone development and show dramatically increased expression in the Nrl^−/− photoreceptors, such as Opn1sw (S-cone opsin), Gnb3 (cone transducin), and Elovl2 (long-chain fatty acid synthase). (c) Cluster II includes genes that exhibit increased expression during rod development and show dramatically reduced expression in the cones, such as Rho (rhodopsin), Nr2e3 (nuclear receptor, mutated in rd7 mice), Pde6b (rod GMP phosphodiesterase 6B, mutated in rd1 mice), and Nrl.