Gene expression screening using single retinal cells
In an initial effort to identify candidate novel molecular markers enriched in bipolar cells, oligonucleotide microarrays were used to characterize gene expression in single bipolar cells from the mouse retina. Candidate genes identified in this manner were subsequently validated as being enriched in bipolar cells by RNA in situ hybridization (see further below). Because bipolar cells comprise only a small fraction of total retinal cells (∼10%) compared with rod photoreceptor cells (>70%; Young,1985), individual bipolar cells, instead of whole retinas, were utilized to enrich for genes of interest. Single bipolar cells were picked from enzymatically dissociated mouse retinas based on expression of a GFP reporter construct driven by the Cabp5 gene promoter transfected into retinas at P0 and harvested after 8 days of culture. This Cabp5 promoter has been shown previously to be active in rod bipolar cells and a limited set of cone bipolar cells (Matsuda and Cepko,2004).
By using a previously described, sensitive RT-PCR-based strategy (Trimarchi et al.,2007; see also Materials and Methods), cDNA from four individual bipolar cells was amplified and then hybridized to microarrays (Table 2; Cabp5 Bipolar A3, A4, B1, and B2). To generate additional expression profiles for comparison, randomly chosen single cells from freshly dissected retinas were picked at P5 and from adults and identified retrospectively as bipolar cells (Table 2; P5 Bipolar A3), rod photoreceptor cells (Adult Rod 1 and 2), or Müller glial cells (Adult Müller Glia 1 and 15) based on expression of known markers. For bipolar cells, these known genes included Cabp5 (Haeseleer et al.,2000) and C. elegans ceh-10 homeodomain-containing homolog (Chx10; Liu et al.,1994; Burmeister et al.,1996). For rod photoreceptor cells, these previously characterized molecular markers were cGMP-specific phosphodiesterase 6b (Pde6b; Hurwitz et al.,1985; Baehr et al.,1991; Chang et al.,2007b), cyclic nucleotide gated channel α1 (Cnga1; Kaupp et al.,1989), and rhodopsin (Rho; Jan and Revel,1974; Molday and MacKenzie,1983). For Müller glial cells, these known genes included retinaldehyde binding protein 1 (Rlbp1; Bunt-Milam and Saari,1983) and glutamate-ammonia ligase (Glul; Riepe and Norenburg,1977).
Table 2. Scaled Fluorescence Signal Intensity Levels from Microarrays of Single Mouse Retinal Cells1
|(a)||(b)||(c)||Fluorescence signal intensities||(m)||(n)|
|Probe set ID||Unigene||Gene name||P5 bipolar A3||Cabp5 bipolar A3||Cabp5 bipolar A4||Cabp5 bipolar B1||Cabp5 bipolar B2||Adult rod 1||Adult rod 2||Adult Müller glia 1||Adult Müller glia 15||Ratio of signal intensities of bipolar cells to rod photoreceptor + Müller glial cells||P value|
|1451826_at||Mm.103669||Calcium binding protein 5 (Cabp5)||7,874||21,964||18,135||40,043||37,690||395||23||155||73||155.6||0.0042|
|1424944_at||Mm.41456||Purkinje cell protein 2 (Pcp2)||13||7,342||4,364||259||4,527||41||21||224||12||44.3||0.0406|
|1450945_at||Mm.222178||Protein kinase C α (Prkca)||143||3,273||7,272||9,010||2,643||28||200||184||26||40.9||0.0243|
|1419628_at||Mm.4405||C. elegans ceh-10 homeodomain-containing homolog (Chx10)||4,718||24,762||22,820||53,259||73,768||168||24||12||18,707||7.6||0.0343|
|1438782_at||Mm.321683||Contactin 4 (Cntn4)||3||12,582||6,377||26,568||23,591||527||141||14||4||80.6||0.0240|
|1427482_a_at||Mm.119320||Carbonic anhydrase 8 (Car8)||96||6,773||18,800||24,431||25,708||620||48||244||50||63.0||0.0174|
|1453008_at||Mm.151594||RIKEN cDNA 2300002D11 gene (2300002D11Rik)||14,214||25,518||23,537||16,469||13,327||1,082||60||386||54||47.0||0.0002|
|1457946_at||Mm.134360||Og9 homeobox gene (Og9x)||191||6,663||9,945||17,860||18334||168||634||190||40||41.0||0.0165|
|1424547_at||Mm.342160||Carbonic anhydrase 10 (Car10)||106||915||1,109||3,104||658||111||33||24||66||20.2||0.0471|
|1425041_at||Mm.15655||LIM homeobox protein 3 (Lhx3)||10,281||56||42||71||7||411||93||386||116||8.3||0.2271|
|1419740_at||Mm.1372||Phosphodiesterase 6B, cGMP, rod receptor, β polypeptide (Pde6b)||21||7,981||20||38||193||514,284||167,133||28||22||0.010||0.0786|
|1451763_at||Mm.23793||Cyclic nucleotide gated channel α1 (Cnga1)||22||1,979||7||14||15||194,172||159,622||28||2,985||0.005||0.0443|
|1418310_a_at||Mm.41653||Retinaldehyde binding protein 1 (Rlbp1)||3||4,521||26||28||13,236||34||7||168,251||109,338||0.051||0.0586|
|1426236_a_at||Mm.210745||Glutamate-ammonia ligase (Glul)||41||3,798||4||13||282||434||296||221,720||89,714||0.011||0.0684|
The fluorescence signal intensity levels from the microarrays of five bipolar cells were compared with those from two rod photoreceptor cells and two Müller glial cells. This comparison was done to enrich for genes expressed relatively specifically in bipolar cells and to filter out more widely expressed genes in at least some other retinal cell types. Table 2 shows the intensity levels for selected gene sequences from the single cells. The signal intensities for several previously characterized bipolar cell genes, including Cabp5, Purkinje cell protein 2 (Pcp2; Berrebi et al.,1991), protein kinase C α (Prkca; Greferath et al.,1990), and Chx10, were enriched between 8- and 156-fold when the average signal intensities for bipolar cells were compared with those for rod photoreceptor and Müller glial cells, confirming the utility of this method in identifying genes enriched in their expression in bipolar cells.
For each of these genes, the average signal intensity for the bipolar cells was significantly greater than the average signal intensity for the rod photoreceptor and Müller glial cells (P < 0.05, one-tailed Student's t-test, Table 2). Table 2 also includes signal intensities for several selected genes that were significantly enriched in the single bipolar cells and whose retinal expression has not been previously characterized. These genes, which exhibited some of the highest signal intensity ratios when comparing bipolar cells with other cells in the filtered results, were of particular interest because the possibility that they could be new bipolar cell-enriched molecular markers was substantial given that known bipolar cell genes were similarly enriched. The novel candidate bipolar cell genes were enriched between 8- and 81-fold and included contactin 4 (Cntn4), carbonic anhydrase 8 (Car8), RIKEN cDNA 2300002D11 gene (2300002D11Rik), Og9 homeobox gene (Og9x), carbonic anhydrase 10 (Car10), and neurofascin (Nfasc). For each of these genes, the average signal intensity for the bipolar cells was also significantly greater than the average signal intensity for the rod photoreceptor and Müller glial cells (P < 0.05, one-tailed Student's t-test, Table 2). For comparison, signal intensities for a subset of known rod photoreceptor (Pde6b, Cnga1, Rho) and Müller glial cell genes (Rlbp1, Glul) are shown in Table 2.
As expected for these rod photoreceptor and Müller glial cell genes, the signal intensity ratios for the bipolar cells to the rod photoreceptor and Müller glial cells were low, ranging from 0.004 to 0.051. These six candidate bipolar cell genes were demonstrated to be expressed in an enriched manner in bipolar cells, as assessed by RNA in situ hybridization (see further below). This limited set of single cells, while insufficient to provide the statistical power for identification of all bipolar cell-enriched genes, or to compare different cell types quantitatively, was sufficient for the purpose of screening for candidate novel bipolar cell molecular markers. Other genes exhibited high signal intensities in single bipolar cells and low signal intensities in rod photoreceptor and Müller glial cells, but these genes were not expressed in an enriched manner in bipolar cells as assessed by RNA in situ hybridization and so were not pursued further (data not shown).
Identification of bipolar cell candidate genes by using SAGE
In an additional effort to identify candidate novel bipolar cell molecular markers, retinal SAGE data were also examined (Blackshaw et al.,2004). A previous analysis of gene expression during mouse development was carried out with SAGE by using 10 retinal cDNA libraries generated at various stages between embryonic day (E)12.5 and adulthood. Bipolar cell-enriched molecular markers were previously identified based on large-scale RNA in situ hybridization studies of genes shown to have dynamic temporal expression patterns within the SAGE data (Blackshaw et al.,2004). To mine the SAGE data further for additional genes enriched in bipolar cell expression, a nearest neighbor method was applied. Because bipolar cells are born in a discrete developmental period and differentiate relatively late (Young,1985), genes with late onset of expression might be novel bipolar cell genes. A search was carried out by using nearest neighbor analysis and known bipolar cell genes that have late onset of expression. Candidate genes with temporal gene expression patterns that displayed minimal Euclidean distances to patterns for known bipolar cell genes were selected for further analysis by RNA in situ hybridization (see further below).
The temporal expression patterns for the known bipolar cell gene Cabp5 (Haeseleer et al.,2000) and eight genes with similar profiles are shown in Figure 1. The onset of expression for these genes was from P0 to P6.5, and the relative expression levels peaked between P6.5 and adulthood. This set included genes previously shown to be expressed in bipolar cells, such as metabotropic glutamate receptor 6 (Grm6; Nakajima et al.,1993; Vardi and Morigiwa,1997), γ-aminobutyric acid (GABA)-C receptor rho subunit 1 (Gabrr1; Koulen et al.,1998), visual system homeobox 1 homolog (Vsx1; Chow et al.,2001,2004; Ohtoshi et al.,2001,2004), and G protein β3 (Gnb3; Huang et al.,2003), confirming the utility of this method in identifying genes enriched in their expression in bipolar cells. Also included was 2300002D11Rik, a gene identified as a candidate bipolar cell molecular marker from the microarray screening.
Figure 1. Temporal expression patterns of mouse bipolar cell and rod photoreceptor cell genes. Relative expression levels of selected genes are plotted as normalized SAGE tag units over developmental time between E12.5 and P10 and in adulthood. Patterns for Cabp5 (thick blue line) and neighboring bipolar cell genes (thin blue lines; Grm6, Gabrr1, Vsx1, Gnb3, 2300002D11Rik, Scgn, Trpm1, 6330514A18Rik). Patterns for rhodopsin (thick orange line) and neighboring rod photoreceptor cell-enriched genes (thin orange lines; Guca1a, Aipl1, Gnat1, Rom1, Gngt1, Guca1b, Grk1, Pde6g) are shown.
Download figure to PowerPoint
Figure 1 also shows the temporal expression pattern of three additional genes with relatively late onset of expression: secretagogin (Scgn), transient receptor potential action channel subfamily M member 1 (Trpm1), and RIKEN cDNA 6330514A18Rik gene (6330514A18Rik). The Euclidean distance between the average temporal expression pattern of these eight known and candidate bipolar cell genes and the pattern of Cabp5 was 4.01 ± 0.42 normalized tag units (average ± SEM). For comparison, this distance was significantly smaller than the distance between the Cabp5 pattern and the average profile for nine genes expressed in differentiating rod photoreceptor cells (5.05 ± 0.37 normalized tag units; P = 0.0425 by Student's one-tailed t-test), suggesting that at least a subset of bipolar cell-enriched genes have temporal expression patterns that cluster in a distinct window apart from patterns for a subset of rod photoreceptor cell-enriched genes.
The late-expressed, previously characterized rod photoreceptor cell-enriched genes included rhodopsin (Rho; Molday and MacKenzie,1983; Jan and Revel,1974), guanylate cyclase activator 1a (Guca1a; Subbaraya et al.,1994), aryl hydrocarbon receptor-interacting protein-like 1 (Aipl1; van der Spuy et al.,2002), G protein α1 (Gnat1; Lerea et al.,1986), rod outer segment membrane protein 1 (Rom1; Bascom et al.,1992), G protein γ1 (Gngt1; Peng et al.,1992), guanylate cyclase activator 1b (Guca1b; Howes et al.,1998), G protein-coupled receptor kinase 1 (Grk1; Zhao et al.,1998), and cGMP-specific phosphodiesterase 6G (Pde6g; Tuteja and Farber,1988). These three additional candidate bipolar cell genes identified from SAGE libraries were demonstrated to be expressed in a highly enriched manner in bipolar cells as assessed by RNA in situ hybridization (see further below). Other genes exhibited temporal expression pattern similar to the profile of Cabp5, but these genes were not expressed in an enriched manner in bipolar cells as assessed by RNA in situ hybridization and were not pursued further (data not shown).
RNA in situ hybridization analysis
RNA in situ hybridization analysis in P21 mouse retinal sections was carried out to evaluate expression of the candidate bipolar cell genes identified as described above. This validation process revealed that five of the candidate bipolar cell genes appeared to have highly enriched expression in bipolar cells, whereas five candidate bipolar cell genes exhibited enriched expression in bipolar cells and some expression in additional retinal cell types, as detailed below. For reference, the pattern for Chx10, a transcription factor gene expressed in all bipolar cells and a subset of Müller glial cells, is shown (Liu et al.,1994; Burmeister et al.,1996; Rowan and Cepko,2004). Chx10 expression was observed in cells on the outer (scleral) side of the INL where bipolar neuron cell bodies are located (Fig. 2A). Expression of the cell adhesion gene Cntn4 was found in a subset of bipolar cells, when compared with the Chx10 expression pattern, and weakly in a subset of amacrine cells, found in the inner (vitreal) part of the INL (Fig. 2B). Car8 expression was seen in a subset of bipolar cells (Fig. 2C). The uncharacterized cDNA 2300002D11Rik was expressed strongly in bipolar cells and weakly in the ganglion cell layer (Fig. 2D). The homeobox transcription factor gene Og9x was expressed in a subset of bipolar cells (Fig. 2E).
Figure 2. Expression patterns of novel bipolar cell-enriched gene candidates. RNA in situ hybridization patterns from representative sections of P21 mouse retinas. A: Chx10. B: Cntn4. C: Car8. D: 2300002D11Rik. E: Og9x. F: Car10. G: Nfasc. H: Scgn. I: Trpm1. J: 6330514A18Rik. K: Lhx3. ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Scale bar = 100 μm in K (applies to A–K).
Download figure to PowerPoint
Car10 expression was observed in a subset of bipolar cells and weakly in amacrine cells and the ganglion cell layer (Fig. 2F). Expression of the cell adhesion molecule Nfasc was found in a subset of bipolar cells and weakly in amacrine cells, photoreceptor cells, and the ganglion cell layer (Fig. 2G). The calcium-binding protein gene Scgn was expressed strongly in a subset of bipolar cells (Fig. 2H). Trpm1 expression was seen in a subset of bipolar cells (Fig. 2I). The uncharacterized cDNA with homology to POLO-like kinase genes, 6330514A18Rik, was expressed strongly in a subset of bipolar cells and weakly in the ONL (Fig. 2J). Lhx3 was enriched in its expression eightfold in bipolar cells compared with rod photoreceptor and Müller glial cells in the microarray analysis, but the difference in expression levels was not significant (Table 2), perhaps because Lhx3 is only expressed in a limited subset of bipolar cell types represented in the microarray data set. Consistent with this notion, expression of Lhx3 was observed in a small subset of bipolar cells (Fig. 2K). Thus, the RNA in situ hybridization study results using retinal sections validated the utility of the microarray and SAGE screening. These screening methods were effective in identifying genes expressed in an enriched manner in bipolar cells.
To evaluate which types of bipolar cells express the newly identified genes, double-label expression studies were carried out by using fluorescent riboprobes hybridized to dissociated cells from P14 retinas. Probes for novel bipolar cell-enriched markers were labeled with a red fluorophore, and co-expression with a rod bipolar cell-enriched gene, Pcp2 (Berrebi et al.,1991), an ON bipolar cell-specific gene, Grm6 (Nakajima et al.,1993; Vardi and Morigiwa,1997), or a pan-bipolar cell gene, Chx10 (Liu et al.,1994; Burmeister et al.,1996), each labeled by a green fluorescent probe, was assessed. Examples of double labeling are shown in Figure 3, and quantitation of the results by using computer-based image analysis is shown in Table 3. Bipolar cells are a relatively rare class of neurons in the mouse retina (∼10%) compared with rod photoreceptor cells (>70%; Young,1985), and so the majority of dissociated cells were unlabeled by the known bipolar cell markers. Nonetheless, Og9x-positive cells were mostly also positive for Pcp2, Grm6, and Chx10, suggesting that Og9x is a rod bipolar cell-specific gene (Fig. 3A–C). The percentages of Og9x-positive cells that were also positive for Pcp2, Grm6, and Chx10 were 69.8%, 83.7%, and 98.8%, respectively (Table 3; Og9x row; columns i, j, k). Rod bipolar cells are only a fraction of ON bipolar cells and bipolar cells in general (Euler and Wässle,1995; Ueda et al.,1997), and consistent with the suggestion that Og9x is a rod bipolar cell-specific gene, only a fraction of Grm6-positive cells and Chx10-positive cells were also Og9x positive (Table 3; Og9x row; columns g, h).
Figure 3. Characterization of novel bipolar cell-enriched genes by double-labeling with markers of bipolar cell subtypes and other retinal cells. Fluorescent RNA in situ hybridization and antibody staining images from representative fields of dissociated P14 mouse retinal cells. A–F: Og9x hybridization signal is shown in red. G–L: Scgn hybridization signal is shown in red. A,G: Pcp2 hybridization signal is shown in green. B,H: Grm6 signal is shown in green. C,I: Chx10 signal is shown in green. D,J: Glul antibody staining signal is shown in green. E, K, Pax6 signal is shown in green. F,L–R: Calb antibody staining signal is shown in green. M: Cntn4 hybridization signal is shown in red. N: Car8 hybridization signal is shown in red. O: 2300002D11Rik hybridization signal is shown in red. P: Nfasc hybridization signal is shown in red. Q: Trpm1 hybridization signal is shown in red. R: 6330514A18Rik hybridization signal is shown in red. Arrows, double-positive cells. Nuclei were stained blue with DAPI. Scale bar = 10 μm in A–R.
Download figure to PowerPoint
Table 3. Quantitation of Dissociated P14 Mouse Retinal Cells Double-Labeled with Fluorescent Riboprobes for Novel and Known Bipolar Cell-Enriched Markers1
|Experimental probe||Experimental||Pcp2||Grm6||Chx10||Doubly labeled with experimental and Pcp2/Pcp2||Doubly labeled with experimental and Grm6/Grm6||Doubly labeled with experimental and Chx10/Chx10||Doubly labeled with experimental and Pcp2/experimental||Doubly labeled with experimental and Grm6/experimental||Doubly labeled with experimental and Chx10/experimental|
|Og9x||5.6 ± 0.5||9.5 ± 2.1||9.9 ± 2.1||13.9 ± 1.6||54.4 ± 4.0||33.5 ± 4.8||34.3 ± 6.1||69.8 ± 6.2||83.7 ± 2.3||98.8 ± 0.6|
|Scgn||5.2 ± 0.7||9.5 ± 1.7||10.0 ± 1.9||13.7 ± 4.7||10.5 ± 3.1||19.4 ± 4.5||17.1 ± 5.9||13.4 ± 1.6||34.4 ± 5.7||67.1 ± 20.3|
|Cntn4||8.8 ± 2.4||6.4 ± 0.0||10.2 ± 5.2||14.0 ± 5.3||56.1 ± 14.5||33.3 ± 3.6||47.1 ± 7.2||44.7 ± 9.0||46.5 ± 13.9||59.2 ± 21.0|
|Car8||7.0 ± 2.7||9.5 ± 2.1||9.9 ± 3.1||14.0 ± 2.7||62.8 ± 20.9||29.4 ± 14.8||47.6 ± 10.4||70.3 ± 5.8||75.1 ± 19.6||97.2 ± 2.8|
|2300002D11Rik||7.7 ± 1.5||9.5 ± 2.3||9.9 ± 3.6||14.1 ± 3.7||42.6 ± 10.9||31.7 ± 6.4||48.3 ± 11.8||60.0 ± 9.0||71.7 ± 16.7||60.9 ± 18.8|
|Nfasc||8.0 ± 2.2||9.5 ± 2.3||10.5 ± 3.7||14.0 ± 4.4||29.4 ± 1.3||41.8 ± 1.7||33.9 ± 14.6||67.1 ± 15.5||33.6 ± 8.8||82.2 ± 9.5|
|Trpm1||11.6 ± 3.1||9.5 ± 2.0||10.0 ± 1.3||14.1 ± 1.1||25.0 ± 5.2||76.3 ± 10.1||37.2 ± 5.8||37.7 ± 4.4||38.8 ± 8.6||83.3 ± 2.8|
|6330514A18Rik||18.5 ± 5.7||7.5 ± 2.3||10.0 ± 3.0||13.5 ± 1.5||41.2 ± 18.8||58.7 ± 19.8||24.9 ± 8.1||17.8 ± 6.9||38.5 ± 20.8||46.5 ± 12.5|
In contrast, Scgn-positive cells were almost all Pcp2 negative, even though the majority was Chx10 positive, suggesting that Scgn is expressed in cone bipolar cells (Fig. 3G,I; Table 3; Scgn row; columns i, k). Scgn-positive cells were both Grm6 positive and Grm6 negative (Fig. 3H; Table 3; Scgn row; column j), suggesting that Scgn is expressed in both ON and OFF cone bipolar cells. Data for Car10 and Lhx3 are not shown because the fluorescent in situ hybridization signals for these genes were too weak to detect in dissociated cells. Cntn4 is expressed in bipolar cells and amacrine cells, and from the double-labeling experiments, it was found in at least some Pcp2-positive cells and Grm6-positive cells. The majority of Car8-positive cells were also positive for Pcp2, Grm6, and Chx10, suggesting that Car8 is a rod bipolar cell-specific gene. 2300002D11Rik is expressed in bipolar cells and the ganglion cell layer, and from the double-labeling studies, it was observed in at least some Pcp2-positive cells and Grm6-positive cells.
Nfasc is expressed in bipolar cells and the amacrine and the ganglion cell layer, and it was detected in dissociated cells in at least some Pcp2-positive cells and a small fraction of Grm6-positive cells. Trpm1-positive cells were almost all labeled with the Chx10 probe, but only a minority of cells was positive for Pcp2 or Grm6, suggesting that Trpm1 is expressed in cone OFF bipolar cells and in a small number of cone ON bipolar cells and rod bipolar cells. 6330514A18Rik is expressed in bipolar cells and photoreceptor cells, and from the double-labeling experiments, it was found in a small minority of Pcp2-positive cells and in at least some Grm6-positive cells. Thus, the novel bipolar cell-enriched genes were expressed in a variety of different subtypes of bipolar cells.
Mouse retinal bipolar cell bodies appear intermingled in sections in the INL with horizontal cell and Müller glial cell bodies. Double-labeling experiments in dissociated retinal cells were conducted by using probes against novel bipolar cell-enriched genes and antibodies against calbindin (Calb), a protein expressed abundantly in horizontal cells and weakly in a subset of amacrine cells (Hamano et al.,1990; Elshatory et al.,2007), and Glul, a Müller glial cell marker (Riepe and Norenburg,1977), to determine whether the novel molecular markers are found in these cell types. Additionally, the possibility that novel bipolar cell-enriched genes are expressed in amacrine cells was addressed by double-labeling dissociated cells using a Pax6 riboprobe, which marks amacrine, horizontal, and ganglion cells (de Melo et al.,2003). Cells positive for Calb expression were negative for Og9x (0/11), Scgn (0/15), Cntn4 (0/15), Car8 (0/15), 2300002D11Rik (0/14), and Trpm1 (0/16) expression (Fig. 3, Supplementary Fig. 1). However, Nfasc- and 6330514A18Rik-positive cells overlapped with strongly labeled Calb-positive cells (15/16 and 6/6, respectively), suggesting that these two genes are also found in horizontal cells or amacrine cells. Horizontal cells are so rare in the retina (<1%; Jeon et al.,1998) that quantitation of single-labeling results was not carried out. Fields were selected for examination based on presence of a Calb-positive cell, and then double-labeling was assessed. Only a small minority of Glul-positive cells was positive with probes for Og9x, Scgn, Cntn4, Car8, 2300002D11Rik, Nfasc, or Trpm1, suggesting that these genes are not found in Müller glial cells at high frequency (Fig. 3, Supplementary Fig. 1, Supplementary Table 1).
Approximately one-third of Glul-positive cells were labeled with the probe for 6300514A18Rik, indicating that a moderate number of Müller glial cells express this gene. Only a small minority of Pax6-positive cells was positive with probes for Og9x, Scgn, Car8, 2300002D11Rik, or Trpm1, suggesting that these genes are not found in amacrine, ganglion, or horizontal cells at high frequency (Fig. 3, Supplementary Fig. 1, Supplementary Table 2). Among Pax6-positive cells, ∼30–40% were labeled by probes for Cntn4, Nfasc, or 6330514A18Rik, indicating that a moderate number of amacrine, ganglion, and/or horizontal cells express these genes.
Bipolar cell gene expression in Bhlhb4−/− retinas
To assess further the identity of bipolar cells in which the novel markers are expressed, in situ hybridization studies were conducted in retinas from P21 WT and Bhlhb4-deficient mice in which rod bipolar cells have previously been shown to die from P8 to P12 (Bramblett et al.,2004). A null mutation at the Bhlhb4 gene locus was introduced by gene targeting and Cre/loxP-mediated recombination in ES cells, as shown in Figure 4. Bhlhb4-positive cells were completely absent in retinas of homozygous null mice as assessed by in situ hybridization (Fig. 4D). The specific loss of rod bipolar cells in Bhlhb4 mutants was confirmed by examining Prkca expression, a known rod bipolar cell gene (Greferath et al.,1990). Whereas WT retinas displayed robust expression of Prkca in bipolar cell bodies closely apposed to the OPL and in a subset of amacrine cells, bipolar cells expressing Prkca were completely absent from the Bhlhb4-null retina; amacrine cell expression was maintained, however (Fig. 5A,B).
Figure 4. Mutagenesis of the mouse Bhlhb4 gene. A: Gene targeting strategy showing partial restriction map of WT Bhlhb4 allele, the targeting vector, the targeted ES cell allele, and the Bhlhb4 null allele. The Bhlhb4 gene (which is a single exon gene), the PGK-neomycin cassette, and the PGK-diptheria toxin cassette are represented by rectangles; the arrows represent open reading frames, and the triangles represent loxP sites. Thin lines show the positions of 5′ and 3′ probes used in Southern blotting analysis. BsmI restriction sites (B), used for screening for integration by homologous recombination from the 5′ side of the gene, and NheI restriction sites (N), used for screening from the 3′ side, are indicated. B: Southern blot analysis of ES cells. Genomic DNA was digested with either BsmI or NheI, and Southern blots were analyzed by using either the 5′ or the 3′ probe, respectively. Fragment sizes for WT (+/+) and targeted (−/+) DNA are indicated. C: PCR genotyping from mouse tail DNA from WT (+/+), heterozygous (−/+), and Bhlhb4-null (−/−) animals. WT allele, 216 bp; Bhlhb4-null allele, 299 bp. D: RNA in situ hybridization for Bhlhb4 RNA. Retinal sections from P21 WT (+/+) and Bhlhb4-null (−/−) mice are shown. Scale bar = 100 μm in D.
Download figure to PowerPoint
Figure 5. Bipolar cell gene expression in the Bhlhb4-deficient retina. RNA in situ hybridization patterns from representative sections of P21 mouse retinas. A,C,E,G,I,K,M,O,Q,S,U,W,Y,A′:, WT retinal sections. B,D,F,H,J,L,N,P,R,T,V,X,Z,B′:, Bhlhb4-deficient retinal sections. A,B: Prkca. C,D: Pcp2. E,F: Grm6. G,H: Chx10. I,J: Og9x. K,L: Car8. M,N: Nfasc. O,P: Cntn4. Q,R: 2300002D11Rik. S,T: Trpm1. U,V: Scgn. W,X: 6330514A18Rik. Y,Z: Car10. A′,B′: Lhx3. Scale bar = 100 μm in M (applies to A–Z,A′,B′).
Download figure to PowerPoint
In the mutant retina, bipolar cells strongly expressing Pcp2 and closely apposed to the OPL were lost, but weakly expressing bipolar cells located slightly closer to the center of the INL were retained, indicating that Pcp2 is expressed in rod bipolar cells and a subset of cone bipolar cells (Fig. 5C,D). Fewer Grm6- and Chx10-expressing bipolar cells were evident in the mutant retina, reflecting the fact that these genes are expressed not only in rod bipolar cells but also in some cone bipolar cells, which remain (Fig. 5E–H). Similar analysis using the novel bipolar cell markers confirmed that Og9x (Fig. 5I,J) and Car8 (Fig. 5K,L) are rod bipolar cell-specific genes because cells expressing these genes were absent from the mutant retina. Among bipolar cells, Nfasc also appears to be expressed specifically in a subset of rod bipolar cells (Fig. 5M,N). Slight to moderate reductions in bipolar cells expressing Cntn4 (Fig. 5O,P), 2300002D11Rik (Fig. 5Q,R), and Trpm1 (Fig. 5S,T) were observed in mutant retinas, indicating that these genes are found in both rod bipolar cells and cone bipolar cells. Scgn expression was unchanged, consistent with the idea that Scgn is expressed in a subset of cone bipolar cells (Fig. 5U,V).
Additionally, bipolar cells expressing 6330514A18Rik (Fig. 5W,X), Car10 (Fig. 5Y,Z), and Lhx3 (Fig. 5A′,B′) appeared unchanged in mutant retinas, indicating that these genes are expressed predominantly in cone bipolar cells. Supplementary Figure 2 shows the results of sense riboprobe hybridizations as negative controls. Table 4 lists the identities of the cells in which novel bipolar cell markers are expressed. Many of the known and novel bipolar cell molecular markers are also expressed outside the eye. These genes are represented in cDNA libraries from different tissue types in the Unigene database (http://www.ncbi.nlm.nih.gov/sites/entrez?db=unigene; Supplementary Table 3), and some genes are found to be expressed in various brain regions (Allen Brain Atlas, Lein et al.,2007; Supplementary Table 3).
Table 4. Summary of Expression of Novel Bipolar Cell-Enriched Markers
|Gene name||Bipolar cell expression||Other retinal cell expression|
|2300002D11Rik||Rod bipolar cells, subset of cone ON and OFF bipolar cells||Ganglion cell layer|
|6330514A18Rik||Subset of cone ON and OFF bipolar cells||Photoreceptor cells, horizontal cells, Müller glial cells, amacrine cells|
|Car8||Rod bipolar cells|| |
|Car10||Subset of cone bipolar cells||Amacrine cells, ganglion cell layer|
|Cntn4||Rod bipolar cells, subset of cone ON and OFF bipolar cells||Amacrine cells, ganglion cell layer|
|Lhx3||Subset of cone bipolar cells|| |
|Nfasc||Subset of rod bipolar cells||Photoreceptor cells, horizontal cells, amacrine cells, ganglion cell layer|
|Og9x||Rod bipolar cells|| |
|Scgn||Subset of cone ON and OFF bipolar cells|| |
|Trpm1||Subset of rod bipolar cells, subset of cone ON and OFF bipolar cells|| |