Phototransduction machinery in vertebrate photoreceptors is contained within the membrane discs of outer segments. Daily renewal of 10% of photoreceptor outer segments requires stringent control of gene expression. Rhodopsin constitutes over 90% of the protein in rod discs, and its altered expression or transport is associated with photoreceptor dysfunction and/or death. Two cis-regulatory sequences have been identified upstream of the rhodopsin transcription start site. While the proximal promoter binds to specific transcription factors, including NRL and CRX, the rhodopsin enhancer region (RER) reportedly contributes to precise and high-level expression of rhodopsin in vivo. Here, we report the identification of RER-bound proteins by mass spectrometry. We validate the binding of NonO (p54nrb), a protein implicated in coupling transcription to splicing, and three NonO-interacting proteins—hnRNP M, Ywhaz and Ppp1ca. NonO and its interactors can activate rhodopsin promoter in HEK293 cells and function synergistically with NRL and CRX. DNA-binding domain of NonO is critical for rhodopsin promoter activation. Chromatin immunoprecipitation followed by deep sequencing (ChIP-seq) analysis demonstrates high occupancy of NonO at rhodopsin and a subset of phototransduction genes. Furthermore, shRNA knockdown of NonO in mouse retina leads to loss of rhodopsin expression and rod cell death, which can be partially rescued by a C-terminal NonO construct. RNA-seq analysis of the NonO shRNA-treated retina revealed splicing defects and altered expression of genes, specifically those associated with phototransduction. Our studies identify an important contribution of NonO and its interacting modulator proteins in enhancing rod-specific gene expression and controlling rod homeostasis.
Development and homeostasis require quantitatively precise expression of genes in distinct spatiotemporal patterns. The regulatory information necessary for the transcription of a gene is largely confined to the proximal promoter region, upstream of the transcription start site (TSS); however, distal regulatory elements (such as enhancers) are frequently needed for accurate and cell-type-specific expression (1,2). Enhancer sequences can exert their influence over a long genomic distance (3) and associate with transcriptional co-activators to augment RNA polymerase II-mediated gene expression (4). The combinatorial interaction of specific proteins that bind to promoter and/or enhancer elements determines the activation or repression of a gene (5). The evolution of enhancers in developmentally regulated genes seems to exert a major drive for tissue specification during vertebrate development (6,7). Notably, a vast majority of variations associated with complex traits and common diseases are identified in non-coding, intronic or intergenic regions, and many of these may be present within potential enhancer sequences (8).
The distinctive architecture and functional organization, together with easier accessibility, of the mammalian retina make it an ideal prototype for dissecting gene regulatory networks underlying neuronal development and homeostasis. The rod and cone photoreceptors constitute over 70% of cells in mature retina and are responsible for vision in dim and bright light, respectively. The visual process is initiated in the photoreceptor outer segment discs that provide high density of opsin visual pigment and other phototransduction components for maximal photon capture (9). Approximately 10% of outer segment discs in the mammalian photoreceptors are shed daily at light onset (10–13). The renewal that accompanies shedding of membrane discs (14,15) requires precise yet high synthesis and transport of phototransduction proteins, specifically rhodopsin, which constitutes >90% of the protein in rod outer segments (16). Whether rhodopsin transcripts exhibit a light dependent or circadian-associated pattern of expression in mammalian rods is debatable (17–19). Nonetheless, abnormal expression and/or trafficking of rhodopsin have been associated with the death of rod photoreceptors (20–23).
The expression of rhodopsin is primarily regulated at the level of transcription. Two distinct sequence elements have been defined upstream of rhodopsin TSS: rhodopsin proximal promoter region (RPPR) and rhodopsin enhancer region (RER) (24). RPPR harbors binding sites for the basic motif neural retina leucine zipper protein NRL and cone rod homeobox CRX and is shown to direct expression (though somewhat leaky) of a reporter gene to rod photoreceptors in transgenic mice (25). NRL and CRX are two key transcription factors that work synergistically to activate rhodopsin expression (26). Loss of Nrl in mice results in a cone-only retina with no rods and no rhodopsin expression (27), whereas ectopic Nrl expression can lead to rhodopsin expression in cones (28). The loss of Crx results in abnormal rods with no outer segments and minimal rhodopsin expression (29).
Transgenic mouse studies using murine or bovine RPPR have revealed the requirement of a longer upstream sequence for rod photoreceptor-specific expression of rhodopsin in vivo (30). Footprinting of bovine rhodopsin promoter had previously identified a highly conserved RER, ∼2 kb upstream of TSS (24). Though rod-specific interaction of RER with RPPR and its relevance to rhodopsin expression has been inferred through long-range chromosomal looping (31), RER-binding proteins have not been identified yet, and molecular mechanisms that control quantitatively precise expression of rhodopsin are poorly understood.
In this report, we have identified RER-binding proteins by mass spectrometry; the most abundant protein is non-POU domain-containing Octamer-binding protein (NonO/p54nrb), which has been implicated in coupling of transcription to splicing (32,33). In addition, three previously reported NonO-interacting proteins were detected among RER-bound proteins. In HEK293 cells, NonO and its interactors activated rhodopsin promoter activity synergistically with NRL and CRX. To test the hypothesis that NonO facilitates high-level transcription of rhodopsin and other phototransduction genes, we performed NonO-ChIP-seq and NonO knockdown in mouse retina in vivo. In addition, we did RNA-seq analysis of the retina after NonO-knockdown in vivo to evaluate the role of NonO in splicing. Our studies demonstrate a significant contribution of NonO and its interacting proteins in modulating rod-specific gene expression and splicing.
Identification of NonO as a major RER-binding protein
Phylogenetic analysis of bovine RER sequence using MacVector (version 11.11.1) revealed several evolutionarily conserved elements (Fig. 1A). To identify RER-binding proteins (strategy shown in Fig. 1B), a biotin-tagged RER oligonucleotide (nucleotides −2155 to −2027 upstream of the bovine rhodopsin TSS), encompassing the conserved regions, was incubated with bovine retinal nuclear extract. RER-bound proteins were separated by Streptavidin-tagged magnetic beads and eluted in high salt buffer after several washings. The eluted proteins were resolved on SDS–PAGE and visualized by silver staining (Fig. 1C). Five protein bands detectable in the RER-bound fraction (shown by asterisks in Fig. 1C), but absent in the control, were excised from the gel for mass spectrometry. The analysis of identified peptides revealed a number of proteins that reportedly participate in splicing, transcription or signal transduction (data not shown). In our mass spectrometry analysis, the maximum number of unique peptides was obtained for non-POU domain-containing Octamer-binding protein, NonO/p54nrb. In addition, we identified three known NonO-interacting proteins—hnRNP M, protein phosphatase 1 (Ppp1ca) and Ywhaz among the RER-bound proteins, but not in the control. The presence of NonO, hnRNP M and Ywhaz in RER-bound fraction was validated by immunoblot analysis using specific antibodies (Fig. 1D). We could not independently confirm the presence of Ppp1ca since the available antibody did not cross-react with the cognate bovine protein. However, the interaction of Ppp1ca with NonO has been reported earlier (34).
NonO augments rhodopsin promoter activity
To examine the functional relevance of NonO binding to RER, we performed reporter assays in HEK293 cells using 2.2 kb bovine rhodopsin promoter (including both RPPR and RER) driving a firefly luciferase gene (2.2 kb bRho-Luc; Fig. 2A). A plasmid containing CMV promoter driving renilla luciferase was used as an internal control. Transfection of NonO expression construct produced a relatively small dose-dependent increase in rhodopsin promoter activity, which was significantly enhanced (almost 80-fold) by co-transfection with two known transcriptional activators of rhodopsin—NRL and CRX (Fig. 2A). A synergistic augmentation in rhodopsin promoter activity was also evident when Ppp1ca or Ywhaz was co-transfected with NRL, CRX and/or NonO (Fig. 2B). Notably, when transfected alone, Ppp1ca and Ywhaz did not exhibit a significant increase in rhodopsin promoter activity. NonO also enhanced the activity of 130 bp rhodopsin promoter, which lacks RER but includes RPPR, though to a much lesser extent (Supplementary Material, Fig. S1), presumably because of its direct effect on RNA Pol II (33,35) or additional NonO binding near the TSS in RPPR (see NonO ChIP-seq data later).
The synergy between NonO, NRL and CRX in activating rhodopsin promoter activity suggested an interaction among these regulatory proteins. Consistent with this prediction, NonO protein could be co-immunoprecipitated with anti-NRL, anti-CRX and anti-RNA Pol II (positive control (33)) antibodies, but not with anti-Brn3b (official name, Pou4f2), a retinal ganglion cell-specific transcription factor, or IgG (both negative controls) (Fig. 2C). Notably, NonO could be immunoprecipitated with anti-NRL or anti-CRX antibodies only if the bovine retina was cross-linked before preparing the nuclear extracts.
DNA-binding domain of NonO is essential for rhodopsin promoter activation
The NonO protein of 473 amino acid residues includes two RNA-recognition motifs (RRM-1 and RRM-2) and a DNA-binding region with no on or off transient A/paraspeckle (NOPS) and charged coiled coil domains (36,37) (Fig. 2D). Co-transfection of 2.2 kb bRho-Luc construct with NRL, CRX and different NonO deletion mutants in HEK293 cells demonstrated the requirement of DNA-binding region for activating rhodopsin promoter (Fig. 2D). Although RRMs could not activate reporter gene expression on their own, only partial NonO-mediated transactivation was observed in their absence. We also noted that full-length NonO and ΔC225-NonO proteins are exclusively localized in the nucleus, while the other truncated proteins were detected in both the nucleus and cytosol to varying extent (Supplementary Material, Fig. S2).
ChIP-seq analysis reveals binding of NonO to rhodopsin and a subset of phototransduction genes
To further assess the role of NonO in photoreceptor homeostasis, we performed ChIP-seq analysis to map genome-wide NonO occupancy in P28 mouse retina (see www.nei.nih.gov/intramural/nnrl/nonochipseq/ for complete sequence reads). We identified 3989 NonO-binding sites (hereafter referred to as ‘peaks’); of these, 19% of NonO peaks were detected within the promoter regions (Fig. 3A), and almost 30% overlapped with the reported Nrl and Crx ChIP-seq peaks (38,39) (Fig. 3B and C). NonO peaks were enriched around TSS (Fig. 3D).
Integrated analysis of NonO ChIP-seq peaks with RNA-seq data from the adult mouse retina (40) revealed a positive correlation between NonO-binding and gene expression levels (Fig. 3E). Among the highly expressed retinal genes, photoreceptor-specific functions were more enriched in NonO-associated genes compared with non-NonO occupied genes (Supplementary Material, Tables S1 and S2). NonO ChIP-seq peaks were present near most photoreceptor genes, including Nrl and Crx, and their targets such as rhodopsin (Rho), orphan nuclear receptor Nr2e3 and estrogen-related receptor β (Esrrβ) (Fig. 4A and B). An independent qPCR analysis of NonO ChIP-DNA from mouse retina further validated several distinct NonO ChIP-seq peaks in the rhodopsin gene, including one near TSS and another in RER (Fig. 4B). Thus, our data demonstrate specific binding of NonO to rhodopsin and other phototransduction genes in the mature retina in vivo.
NonO is required for rod photoreceptor survival
To directly investigate the role of NonO in rhodopsin gene regulation, we used shRNA to knockdown NonO in postnatal day (P)0 mouse retina by in vivo electroporation. The efficacy of NonO shRNA was first assessed in HEK293 cells (Supplementary Material, Fig. S3). P0 mouse retina was transfected in vivo with two constructs—2.2 kb bovine rhodopsin promoter driving DsRed reporter (to mark the transfected rods) and either control or NonO shRNA-GFP plasmid carrying a GFP reporter under the control of a ubiquitous promoter (to mark all transfected cells). At P5, almost all of the cells expressing the GFP transgene in the outer nuclear layer (ONL) were also positive for the DsRed reporter in both NonO shRNA-GFP and control-shRNA-GFP transfected retina. However, at P10, only 40% of the GFP+ cells in ONL revealed DsRed reporter expression in NonO shRNA-GFP transfected retina compared with the control (Fig. 5A and B), and only few co-labeled cells remained at P21. These results can be explained by reduced rhodopsin promoter activity and/or by progressive rod death in NonO-knockdown retina; the latter was validated by a strong immunostaining of active caspase-3 in P7 NonO shRNA-GFP transfected retina (data not shown). No change was detected in transfected cells between P10 and P21 retina when control-shRNA-GFP was electroporated. Co-transfection of NonO shRNA-GFP with ΔN224-NonO expression construct (see Fig. 2B) partially rescued the rod cell death (Fig. 5C and D), confirming the specificity of NonO knockdown.
We further evaluated the expression of endogenous rhodopsin, by immunocytochemistry, in dissociated cells from P7 retina transfected at P0 with NonO or control-shRNA-GFP. As predicted, NonO shRNA-GFP-expressing cells had reduced or no rhodopsin immunostaining, whereas strong rhodopsin expression was detected in control-shRNA transfected cells (Fig. 6A and B).
Knockdown of NonO in developing retina leads to aberrant splicing in a subset of phototransduction genes
To investigate whether NonO can alter splicing efficiency in vivo, we knocked down NonO expression in P1 mouse retina and performed RNA-seq analysis (Supplementary Material, Table S3). As complete loss of NonO initiates cell death (see Fig. 5) by P10, we reduced the amount of NonO shRNA-GFP for electroporation and analyzed transfected area of the retinas at P8. We estimate that ∼25% of cells received the transgene based on the reported electroporation efficiency and rod cell numbers in the developing retina. As predicted, the knockdown of NonO under these conditions led to splicing defects and altered gene expression (Supplementary Material, Table S4). Gene ontology analysis revealed that phototransduction genes are highly enriched among those with splicing defects (Fig. 7A). In particular, increased intron inclusion and slight increase in total transcript levels were apparent in rhodopsin (Rho) and other phototransduction genes such as Gnb1 and Cnga1 (Fig. 7B and Supplementary Material, Table S4). However, we did not observe such defects in ubiquitously expressed genes (e.g. Hprt) or those expressed specifically in other retinal neurons (such as Grm6, a bipolar marker, or Pou4f, a ganglion cell marker) (Fig. 7C). Our data demonstrate an important role of NonO in regulation of splicing in vivo of highly expressed rod-specific genes such as rhodopsin.
Daily turnover of 10% of photoreceptor outer segments requires a homeostatic balance of many structural and functional components that include the major disc protein rhodopsin. The expression and transport of rhodopsin must be controlled stringently; thus, multiple levels of regulation are expected. While the fundamental role of key transcription factors (including NRL, CRX and NR2E3) in controlling rod gene expression and rod differentiation is becoming clearer (38,39,41,42), the regulatory mechanisms of temporal and quantitatively precise expression of rhodopsin and other phototransduction genes are poorly understood. We now report the identification of NonO and three NonO-interacting proteins that bind to RER and modulate the expression of rhodopsin. ChIP-seq data strongly argues for the involvement of NonO and probably its interactors in controlling the expression of many rod-specific genes. Analysis of RNA-seq data also revealed aberrant splicing of a subset of phototransduction genes when NonO is knocked down in the retina in vivo. Our studies thus define additional transcriptional control components and would help in constructing gene regulatory network underlying development and homeostasis in mammalian rod photoreceptors.
NonO is a widely expressed multifunctional protein of Drosophila behavior/human splicing family, implicated in transcriptional regulation and splicing (33,37,43,44). Functional integration of transcription to splicing provides a greater control in regulatory networks (32,45,46). RNA polymerase II and transcription factors can physically or functionally interact with splicing factors and control cell-type-specific gene expression patterns (47–50). Combined analyses of ChIP-seq and gene profiling data from knockout mouse retina have demonstrated that NRL and CRX together activate the expression of most, if not all, rod photoreceptor genes (38); thus, we propose that the binding of NonO to enhancer elements (such as RER in rhodopsin) in rod-expressed genes would augment NRL- and CRX-mediated transcriptional activation (represented schematically in Fig. 8). NonO occupancy on highly expressed phototransduction genes, such as rhodopsin, in the retina indicates its role in augmenting expression levels that might be needed at the time of outer segment renewal.
Our model is consistent with the suggestion that RER comes in contact of RPPR by chromosome looping in rod photoreceptors but not in other cell types which do not express rhodopsin (31). In addition, the binding of NonO to transcriptional complexes (including RNA polymerase II) may accelerate accurate and concurrent splicing of rhodopsin and other NonO-bound transcripts in photoreceptors. In concordance, NonO knockdown in the developing retina in vivo revealed an enrichment of rod phototransduction genes among those showing aberrant splicing patterns by RNA-seq analysis. As an example, rhodopsin transcripts showed enhanced inclusion of intron 2 in NonO shRNA-treated retina compared with controls.
Identification of three NonO-interacting proteins, Ywhaz, Ppp1ca and hnRNP M (34,51,52), among the RER-bound proteins further argues in support of NonO's role in linking transcriptional complex(es) to splicing factors to accelerate rhodopsin expression when higher synthesis of phototransduction proteins might be required. Our data show that DNA-binding domain of NonO is essential for rhodopsin promoter transactivation, but RRMs that facilitate splicing (34) are required for complete augmentation. RRMs contain consensus binding domains for Ppp1ca that modulates distinct functions of NonO in RNA transcription and splicing (34). Activation of the rhodopsin promoter activity by Ppp1ca in the presence of NRL, CRX and NonO in HEK293 cells provides evidence in support of the hypothesis that NonO shuttles from the splicing complex to the transcriptional complex upon hypophosphorylation by Ppp1ca. Further enhancement in rhodopsin promoter activity by Ywhaz indicates its potential role in preventing dephosphorylation of NonO and other factors. The importance of splicing in photoreceptor homeostasis is illustrated by the fact that mutations in splicing factors are associated with photoreceptor degeneration (53).
Our ChIP-seq data show a number of NonO target genes that belong to circadian feedback regulatory loop, in agreement with the recent findings of abnormal circadian and neurological changes in NonO gene-trap mutant mice (54). These results are consistent with a possible role of NonO in augmenting gene expression at the time of outer segment renewal. Since loss of NonO is reported to lead to enhanced cell proliferation and decreased senescence (54,55), shRNA knockdown of NonO in neonatal retinal progenitors may hamper cell cycle exit and consequently the birth of rod photoreceptors. More significantly, altered expression and splicing of rhodopsin (and other rod genes), resulting from NonO knockdown, would likely lead to accumulation of aberrantly spliced transcripts and trigger death of rods that are already born.
In summary, we have identified NonO as a novel enhancer-binding protein that activates rhodopsin expression synergistically with two key photoreceptor transcription factors, NRL and CRX. We hypothesize that NonO and its three RER-bound interactors (Ppp1ca, Ywhaz and HnRNP M) facilitate long-range chromosome looping between RER and RPPR to enhance rhodopsin expression. In vivo occupancy of NonO on rhodopsin and selected rod phototransduction genes suggests its role in fine-tuning the expression of rod genes that are associated with outer segment morphogenesis.
MATERIALS AND METHODS
Oligonucleotides were procured from Integrated DNA Technologies (Coralville, IA, USA). Bovine RER fragment (−2155 to −2027) was generated by PCR using the following oligonucleotides: GTCCCTCTGTCTGGCCACCAG (sense, biotin-labeled) and GGCAGGTGTGGCGGGTGGGTG (antisense). PCR-amplified RER was purified by extraction after gel electrophoresis and sequence verified. A biotin-labeled 39-mer scrambled oligonucleotide (GCAACTTCCGCTTGCAACTTCCGCTTGCAACTTCCGC TT) was used as control.
Isolation of RER-bound protein complexes
Bovine retinal nuclear fraction was prepared by differential centrifugation. Nuclear extract (300 μg) was incubated with equimolar amount of biotin-labeled RER or control oligonucleotide in binding buffer (final conc. 12 mm HEPES, pH 7.9, 60 mm KCl, 4 mm MgCl2, 1 mm EDTA, 1 mm DTT, 12% glycerol and protease inhibitor cocktail) containing poly dI-dC and salmon sperm DNA for 16 h at 4°C. Thirty microliters of mixture (50:50) of Dynabeads MyOne™ Streptavidin C1 and Dynabeads MyOne™ Streptavidin T1 (Invitrogen, Carlsbad, CA, USA) was added to the reaction for 2 h. After 10× washing of beads with binding buffer, bound proteins were eluted with 1 m KCl, separated on SDS–PAGE, and visualized by silver staining. Proteins detected in RER but not in control reaction were excised and subjected to mass spectra analysis.
Total RNA was isolated from 1-month-old C57Bl/6J mouse retina using QIAGEN kit (Qiagen, Valencia, CA, USA), and cDNA was prepared using Superscript III First-Strand Synthesis System (Invitrogen). Coding region of NonO was PCR-amplified using ATGCGGATCCATGCAGAGCAATAAAGCCTT (sense) and ATGCGCGGCCGCCTAATATCGGCGGCGTTTATTT (antisense) primers. Amplified product was gel purified and cloned into pGemT-Easy vector (Promega, Madison, WI, USA). NonO cDNA was then subcloned into pcDNA4C, a mammalian expression vector (Invitrogen). NonO deletion fragments were amplified using the primer pairs listed below and cloned in pcDNA4C.
ΔN323-NonO: ATGCGGATCCATGAGGCAGGATTTGATGAG (sense) and ATGCGCGGCCGCCTAATATCGGCGGCGTTTATTT (antisense); ΔN224-NonO: ATGCGGATCCGTGACTGTGGAGCCTATGGA (sense) and ATGCGCGGCCGCCTAATATCGGCGGCGTTTATTT (antisense); ΔC225-NonO: ATGCGGATCCATGCAGAGCAATAAAGC-CTT (sense) and ATGCGGATCCCACAGGCCGAGGAAATGTAG (antisense); DBD-NonO: TGCGGATCCGTGACTGTGGAGCCTATGGA (sense) and TTCTTGACGTCTCATCAAATCC (antisense); ΔM-NonO was constructed by cloning ΔC225-NonO fragment into ΔN323-NonO construct. Ppp1ca and Ywhaz were amplified using the primer pairs listed below and cloned in pcDNA4C.
Ppp1ca: ATGCGGATCCATGTCCGACAGCGAGAAGCT (sense) and ATGCGCGGCCGCCTATTTCTTGGCTTTGGC (antisense); Ywhaz: ATGCGGATCCATGGATAAAAATGAGCTGGT (sense) and ATGCGCGGCCGCTTAATTTTCCCCTCCTTC (antisense).
Bovine rhodopsin promoter (25) was cloned into pGL3 basic vector (Promega). GFP-expressing NonO shRNA vector (NonO shRNA-GFP) against NonO exon 2 target sequence (GACCTTTACACAGCGTAGC) and control-shRNA-GFP vector by scrambling the target sequence (AGATAACTGCGCCGTACTC) were generated in pG-Super vector (56) using the following oligonuclotides: NonO shRNA-GFP: GATCCCCGACCTTTACACAGCGTAGCTTCAAGAGAGCTACGCTGTGTAAAGGTCTTTTTGGAAA (sense) and AGCTTTTCCAAAAAGACCTTTACACAGCGTAGCTCTCTTGAAGCTACGCTGTGTAAAGGTCGGG (antisense).
Control-shRNA-GFP: GATCCCCAGATAACTGCGCCGTACTCTTCAAGAGAGAGTACGGCGCAGTTATCTTTTTTGGAAA (sense) and AGCTTTTCCAAAAAAGATAACTGCGCCGTACTCTCTCTTGAAGAGTACGGCGCAGTTATCTGGG (antisense).
All constructs were verified by sequencing.
Luciferase reporter assays
Dual luciferase reporter assays (Promega) were performed using HEK293 cells as described (57). The data from three-independent experiments was analyzed using Student's t-test, and P < 0.05 was considered significant.
Bovine retinas were cross-linked with 5 mm dimethyl 3,3′-dithiobis propionimidate *2HC (Thermo Scientific Pierce, Rockford, IL, USA) on ice for 1 h. The reaction was stopped by adding 1 m Tris (pH 7.5) for 20 min. Retinas were washed with PBS and homogenized, and nuclear fraction was prepared by differential centrifugation in the presence of protease and phosphatase inhibitors (Roche Applied Bioscience, Indianapolis, IN, USA). For immunoprecipitation, nuclear fraction was incubated with a specific antibody or IgG (control) at 4°C. After overnight incubation, Protein A Dynabeads (30 μl) were added for 2 h to allow the binding of protein complexes. Beads were washed five times, and bound protein complexes were eluted in 0.2 m glycine (pH 2.5) and neutralized by 1 m Tris (pH 9.0). Immunoblot analysis of eluted proteins was performed as described (57). The following antibodies were used: rabbit anti-NonO (Protein Tech, Chicago, IL, USA), anti-hnRNPM (Santa Cruz Biotechnologies, Santa Cruz, CA, USA), RNA Pol II (Abcam, Cambridge, MA, USA). Anti-Pou4f2 antibody was a gift from Tudor Badea.
In vitro knockdown of NonO in HEK293 cells
We designed an shRNA targeting the exon 2 of the mouse NonO gene as reported earlier (58) and a scrambled control-shRNA. The shRNA constructs contained a GFP reporter driven by signal recognition particle receptor promoter, while the shRNA hairpin was driven by histone H1 promoter. HEK293 cells were transfected with either NonO shRNA-GFP or control shRNA-GFP along with NonO-pcDNA4c expression construct. After 48 h, cells were harvested, washed with PBS and lysed in RIPA buffer. Immunoblots (10 μg of protein) were probed with anti-Xpress or anti-NonO antibodies. Anti-β-actin antibody was used as control.
In vivo electroporation
Neonatal CD-1 mice (Charles River, Wilmington, MA, USA) were used for in vivo transfection by electroporation (59). Equimolar amounts of plasmids (10 μg/μl; 0.2 μl) were used for co-transfection studies. Transfected eyeballs were harvested at different times for immunohistochemical analysis. For some experiments, P0 retinas (electroporated with NonO or control-shRNA-GFP) were collected at P7 and subjected to dissociation protocols using trypsin. An equal number of cells on eight-chamber slides were then stained for endogenous rhodopsin. GFP (green, which marks the shRNA transfected cells) and rhodopsin (red) signals were recorded using confocal microscope, and pixel intensity of green and red signal in each cell was calculated using ImageJ software (rsbweb.nih.gov/ij/). Total numbers of green (shRNA-GFP) and red (endogenous rhodopsin) pixel intensity in each cell were plotted for control and NonO shRNA-GFP electroporated cells.
Cryosections were probed with specific antibodies as described (60) and visualized using Leica SP5 confocal laser scanning unit (Leica Microsystem Inc., Buffalo Grove, IL, USA).
C57Bl/6J mice were housed in 12 h light and 12 h dark cycles. The retinas were dissected at light onset and fixed with formaldehyde. ChIP was performed using anti-NonO antibody, followed by deep sequencing using Illumina GAIIx. ChIP-seq libraries were constructed and sequenced, as previously described (38). Raw sequencing reads from NonO ChIP DNA and input DNA were mapped to the mouse genome (NCBI build 37) using Genomatix Mining Station (GMS), and ChIP-seq peaks were called using MACS (61). NonO ChIP-seq peaks were compared with NRL ChIP-seq (38) and CRX ChIP-seq peak regions (39) using GenomeInspector (Genomatix). NonO ChIP-seq peaks were assigned to nearby genes by proximity (within 5 kb). Enriched biological processes were analyzed using Genomatix software and ranked by P-value.
P1 retinas of C57Bl/6J mice were transfected in vivo with NonO or control-shRNA-GFP (2.5 µg/µl; 0.2 µl) by electroporation, as described above, and harvested at P8. The transfected area of the retina was identified by GFP fluorescence and excised out for RNA-seq analysis. Total RNA was extracted using TRIzol reagent (Life Technologies, Gaithersburg, MD, USA), followed by RNeasy mini columns (Qiagen). Directional RNA-seq libraries were constructed with 30 ng RNA and sequenced on Illumina GAIIx platform, as previously described (62). Sequence reads were aligned to the mouse genome (NCBI build 38) using GMS, and uniquely mapped reads were used for subsequent analysis. Differential expression analysis was carried out using Genomatix Genome Analyzer, and sequence reads over splice junctions were viewed in SeqMonk software.
This work was supported by intramural research program of the National Eye Institute.
We are grateful to Garry G. Borisy for the generous gift of shRNA vector pG-Super. We thank Douglas Forrest, Tiziana Cogliati, Tudor Badea, Chunqiao Liu, Jerome Roger, Shobi Veleri, Radu Cojocaru and James Friedman for constructive advice and comments.