RIT2, a neuron-specific small guanosine triphosphatase, is expressed in retinal neuronal cells and its promoter is modulated by the POU4 transcription factors

Plasmid construction

To obtain human RIT2 genomic clones, we screened a human genomic P1 library with RIT2 cDNA probes following standard procedures [36]. Briefly, a P1 library that consisted of eight Southern membranes containing a high density array of DNAs (Genome Systems, St Louis, MO) was hybridized with ³²P-labeled probes at 65 °C overnight, washed several times, and exposed to X-ray films. The films with positive spots were laid on the original membrane to identify the location of positive signals, and then positive clones were purchased (Genome Systems). RIT2 promoter-luciferase reporter vectors were constructed with two human RIT2 5′-upstream fragments, −1290 to +76 bp (RIT2-1290/luciferase) and −374 to +76 bp (RIT2-374/luciferase), that were amplified by PCR using a P1 clone as the template and the primers listed in Appendix 1. Both fragments were ligated into the SmaI site of pGL2-Basic, which contains the firefly luciferase gene (Promega, Madison, WI), and confirmed by sequencing.

To construct human POU4 expression vectors, full-coding cDNAs of POU4F1, POU4F2, and POU4F3 were generated by reverse transcription (RT)-PCR using total RNA from SK-N-MC human neuroblastoma cells by oligo(dT) priming and PCR amplification (Appendix 1). Forward primers for all POU4 factors included an EcoRI site. Reverse primers included a HindIII site for POU4F1 and POU4F2, and a BamHI site for POU4F3. The cDNA fragments were inserted into the EcoRI/HindIII or EcoRI/BamHI sites in the pcDNA3.1/Myc-His(-) B vector (Invitrogen, Carlsbad, CA), and confirmed by sequencing. A human ISL1 expression vector was constructed in the same manner except that the ISL1 cDNA was made using retinal RNA from human donor eyes (National Disease Research Interchange, NDRI). The forward primer included an EcoRI site, and the reverse primer included a HindIII site (Appendix 1).

Cell culture and reverse transcription polymerase chain reaction

Human neuroblastoma cell lines, SK-N-MC [37], SK-N-AS [38], and SK-N-DZ [39], and the human embryonic kidney line HEK293 [40] (American Type Culture Collection [ATCC], Manassas, VA), were cultured in Dulbecco's Modified Eagle Medium (Invitrogen) containing 10% fetal bovine serum supplemented with penicillin and streptomycin (Invitrogen) at 37 °C with 5% CO₂ and 95% air. Total RNAs were extracted using TRIzol (Invitrogen) from subconfluent cell cultures or human retina (NDRI), and human brain RNA was purchased (Clontech, Mountain View, CA). First-strand cDNAs were synthesized from 1 μg of total RNAs with oligo(dT) primer and SuperScript III reverse transcriptase (Invitrogen). A 381 bp RIT2 fragment was amplified by 35 cycles of PCR using a forward primer in exon 4 and a reverse primer in exon 5, and ribosomal protein S16 was used as the control (Appendix 1).

Laser capture microdissection and reverse transcription quantitative polymerase chain reaction

All mice were treated in accordance with the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act as well as the guidelines of the Institutional Animal Care and Use Committee at Johns Hopkins University. Tissues of individual retinal layers of adult C57BL/6J mice were collected by laser capture microdissection (LCM) as previously described [41-43]. Briefly, after the cornea and lens were removed, the eyecups were immersed in 10%, 15%, and 25% sucrose for 30 min each, and frozen in optimal cutting temperature compound (OCT) with 25% sucrose. Sections were cut at 7 μm, mounted onto slides with charged PEN-foil membranes (Leica Microsystems, Deerfield, IL), fixed in 70% ice-cold ethanol for 30 s, stained with hematoxylin for 10 s, and dehydrated in 70% and 100% ethanol for 1 min each. Tissues of the outer nuclear layer (ONL), inner nuclear layer (INL), and ganglion cell layer (GCL) were isolated separately using an LMD6000 laser capture microdissection microscope (Leica Microsystems). Total RNAs were extracted from each layer using an RNeasy Micro kit (Qiagen, Valencia, CA), and first-strand cDNAs were synthesized by random hexamer priming with SuperScript III reverse transcriptase (Invitrogen). Quantitative PCR (qPCR) analyses were performed with primers listed in Appendix 1 using iQ SYBR Green SuperMix on an iQ5 Real-Time PCR Detection System (Bio-Rad, Hercules, CA). A relative amount of cDNA of each gene was normalized to that of control Gapdh.

Immunohistochemistry of retinal flat mounts and eye sections

Protein expression of RIT2 and POU4F2 was analyzed by immunohistochemistry of retinal flat mounts and eye sections using 7- to 9-week-old C57BL/6J mice as well as mouse eye sections at embryonic day 15.5 (E15.5) and postnatal day 0 (P0) following published protocols [35,44]. Briefly, after the cornea and lens were removed, the eyecups were fixed in ice-cold 4% paraformaldehyde in phosphate-buffered saline (PBS; 10 mM Na₂HPO₄, 2 mM KH₂PO₄, 137 mM NaCl, 2.7 mM KCl, pH 7.4) for 1 h, and washed in 0.3% Triton X-100 in PBS (0.3% PBST). The retinas were dissected, cut from the periphery toward the optic disc, and placed on a slide with the GCL facing up. The retinal flat mounts were blocked with 10% normal donkey serum in 0.1% PBST for 1 h, and then incubated with a primary antibody, anti-RIT2 antibody (1:200; PA1–25559, rabbit polyclonal, Thermo Scientific, Rockford, IL) or anti-POU4F2 antibody (1:500; sc-6026, goat polyclonal, Santa Cruz Biotechnology, Santa Cruz, CA), in 5% normal donkey serum in 0.1% PBST at 4 °C for 2 days. After several washes with 0.1% PBST, the flat mounts were incubated with a secondary antibody, Alexa Fluor 488 donkey anti-rabbit immunoglobulin (IgG; 1:1,000; A-21206, Invitrogen) for RIT2 or Alexa Fluor 647 donkey anti-goat IgG (1:1,000; A-21447, Invitrogen) for POU4F2, for 2 h at room temperature. The flat mounts were washed with 0.1% PBST, mounted in VECTASHIELD HardSet Mounting Medium with 4', 6-diamidino-2-phenylindole (H-1500, Vector Laboratories, Burlingame, CA), and examined on an LSM 510 inverted laser scanning confocal microscope (Carl Zeiss, Thornwood, NY). For the initial analyses of RIT2 expression by immunohistochemistry of the retinal sections, the mouse eyes were fixed in ice-cold 4% paraformaldehyde in PBS for 1 h, washed in 0.3% PBST, cryoprotected in 20% sucrose in PBS at 4 °C overnight, embedded in OCT Tissue-Tek (Ted Pella, Redding, CA), and cut at 12 μm on a cryostat. The eye sections were incubated with the same primary and secondary antibodies as used for the flat mounts, and examined in the same manner as described above.

For immunohistochemistry to further analyze cell-type-specific RIT2 expression in the INL, eye tissues were processed using shorter fixation as previously described [43,45]. Briefly, after the anterior segment of the eye was removed, the eyecups were fixed for 25 min in cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, and subjected to an increasing gradient of sucrose (6.75%, 12.5%, and 25%) in phosphate buffer. The eyecups were then immersed for 1 h in a 2:1 ratio of 25% sucrose in 0.1 M phosphate buffer and OCT Tissue-Tek (Ted Pella), and snap frozen on dry ice in isopentane. Cryosections were cut at 8–10 μm, simultaneously blocked and permeabilized in 2% normal donkey serum in 0.1% PBST, and incubated overnight with a primary antibody in the blocking solution. Primary antibodies used were anti-RIT2 (1:200; PA1–25559, Thermo Scientific), anti-PAX6 (1:500; Developmental Studies Hybridoma Bank, DSHB, mouse monoclonal, University of Iowa, Iowa City, IA), anti-calbindin D28 (1:1,000; CL300, mouse monoclonal, Sigma, St. Louis, MO), anti-vimentin (1:100; DSHB), anti-LIM1 (1:50; DSHB), and anti-visual system homeobox 2 (VSX2; 1:1,000; Ab9016, sheep polyclonal, Millipore, Billerica, MA). Secondary antibodies were anti-mouse, anti-sheep, and anti-rabbit IgG (heavy and light chains) coupled to Alexa Fluor 488 or 647 (1:1,000; Invitrogen). Hoechst 3342 (Molecular Probes, Invitrogen) was used at 10 μg/ml to visualize cell nuclei. Serial sections processed similarly but without a primary antibody were used as the control for the background. Images of retinal sections were acquired with Zeiss laser scanning confocal microscopes LSM 510 and LSM 710, and images were adjusted for brightness and contrast using ImageJ (NIH, Bethesda, MD) or Photoshop (Adobe, San Jose, CA).

Western blotting

Protein lysates were prepared from the mouse retinas at P2, P5, P15, and adult (8 weeks old) and used for western blot analysis. Briefly, the retina was harvested from two eyecups at each time point, rinsed with ice-cold PBS, and lysed by sonication in 200 μl radio immunoprecipitation assay (RIPA) lysis buffer (R0278, Sigma) on ice. The lysed retinal tissues were mixed gently, kept on ice for 15 min, and centrifuged at approximately 14,000 ×g for 15 min to pellet the debris. The supernatant was transferred to a fresh tube, and the protein concentration was measured using DC Protein Assay (500–0113, 500–0114, and 500–0115, Bio-Rad). After heating at 99 °C for 5 min, 30 μg of protein lysates for each sample were resolved with sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. The blotted membrane was then blocked in TBST (10 mM Tris-Cl [pH 7.5], 100 mM NaCl, 0.1% Tween-20) containing 5% non-fat dry milk at room temperature for 1 h and probed with anti-RIT2 antibody PA1–25559 (1:1,000; Thermo Scientific) or anti-α tubulin antibody (1:500; T9026, mouse monoclonal, Sigma) in TBST containing 5% non-fat dry milk at 4 °C overnight. After washing with TBST three times, the membrane was incubated with horseradish peroxidase–conjugated anti-rabbit IgG and anti-mouse IgG antibodies (1:10,000; 7074 and 7076, Cell Signaling Technology, Danvers, MA) for detecting RIT2 and α tubulin, respectively, at room temperature for 1 h. After washing with TBST, protein signals were visualized with chemiluminescence using SuperSignal West Femto Maximum Sensitivity Substrate (PI34095, Thermo Scientific) with BioMax MR X-ray films (8941114, Kodak, Rochester, NY), and the signal intensity was analyzed with ImageJ (NIH).

Bioinformatic analysis of RIT2 upstream region

MatInspector [46] was used in combination with the Matrix Family Library database (Genomatix, Munich, Germany), and more recently with the MatBase database (Genomatix), to analyze the upstream region of RIT2. The matrix-based pattern matching program Patser [47,48] (Regulatory Sequence Analysis Tools) [49] was used with two published position-specific scoring matrices, Matrix 1 for POU4F1/F2 [50] and Matrix 2 for POU4F2 [27], to scan for the presence of potential binding sites for POU4 proteins in the RIT2 upstream region from −1290 to +76 bp.

Electrophoretic mobility shift assays

Electrophoretic mobility shift assays (EMSAs) were performed as previously described [51,52]. Briefly, proteins to be tested, POU4F1-s (short), POU4F1-l (long), POU4F2-s, POU4F2-l, and POU4F3, as well as control luciferase protein, were generated in vitro using the TnT T7 Quick Coupled Transcription-Translation System (Promega). Synthesis of the appropriate protein was confirmed in a separate reaction by ³⁵S-methionine labeling followed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis separation and autoradiography. Six oligonucleotide probes (Probes 1-6), which were 24 bp long and contained each predicted POU4 binding site in the middle, were labeled by fill-in reaction as previously described [53]. Briefly, oligonucleotide pairs complementary to each other were annealed to create overhangs, mixed with [α-³²P] deoxycytidine triphosphate and a mixture of non-radioactive deoxyadenosine triphosphate, deoxythymidine triphosphate, and deoxyguanosine triphosphate, and the recessed ends were extended using Klenow fragment. Mutated probes were made for Sites 1, 2, and 6 (Probes 1m, 2 m, and 6 m, respectively) by introducing a mutation (TTAA to GGCC). The labeled probe (about 30,000 cpm) was incubated with 3 μl of in vitro generated protein in binding solution (10 mM Tris-Cl, pH 7.9, 100 mM KCl, 5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 5% glycerol) containing 1 μg poly(dI-dC) on ice for 30 min and analyzed on a 5% polyacrylamide gel. For cold oligomer competition, labeled Probe 2 was mixed with 1-, 10-, 100-, and 1,000-fold molar excess of unlabeled competitor oligonucleotides, and then added to the binding solution containing the POU4F1-s protein.

Transient transfection

In all experiments, cells were transfected 24–48 h after plating using Lipofectamine LTX (Invitrogen). For transfection with a RIT2 promoter-luciferase construct alone, neuroblastoma cells in a 60-mm dish were transfected with 3 μg of a luciferase construct (RIT2-1290/luciferase, RIT2-374/luciferase, or empty pGL2-Basic) and 0.5 μg of pCMV-lacZ as the internal control. Cell lysates were prepared 48–60 h after transfection using Reporter Lysis Buffer (Promega), and luciferase and β-galactosidase activities were measured as previously described [53]. Luciferase activity was normalized with β-galactosidase activity, and relative luciferase activity was calculated as the ratio of the normalized luciferase activity with RIT2 promoter constructs to that with pGL2-Basic. For cotransfection with a RIT2 promoter-luciferase construct and POU4 and/or ISL1 expression vectors, SK-N-AS cells in 12-well plates were transfected with 0.2 μg of a firefly luciferase construct, 0.4 μg of each pcDNA3.1 expression vector (either with cDNA or no insert), and 0.1 ng of pRL-CMV containing the Renilla luciferase gene (Promega). The total amount of the expression vectors was adjusted to 0.8 μg with empty pcDNA3.1 in all experiments. Cell lysates in Passive Lysis Buffer (Promega) were analyzed using the Dual-Luciferase Reporter System (Promega). Firefly luciferase activity was normalized with Renilla luciferase activity, and relative luciferase activity was calculated as the ratio of the normalized luciferase activity with pcDNA3.1 containing cDNA to that with empty pcDNA3.1. Three to four independent transfections were performed in duplicate each time. Statistical analysis was performed using the Student t test to compare relative luciferase activity with a POU4 factor and ISL1 to that with a POU4 factor alone.

The 5′-upstream region of human RIT2 drives expression in neuronal cells

As an initial step for identifying the RIT2 promoter, we searched genomic databases at the National Center for Biotechnology Information (NCBI) and the University of California Santa Cruz (UCSC) genome browser. These analyses revealed that the human and mouse genes encoding RIT2 are unusually large; they contain only five exons, with the coding region totaling 651 bp, but span 300 kb due to their extensive introns. In contrast, the genes encoding RIT1, a homolog of RIT2, are small and extend only a little over 10 kb (Appendix 2). Although the presence of such large introns in RIT2 raises the possibility that intronic sequences could contribute to its neuron-specific expression pattern, we decided initially to focus on RIT2’s immediate upstream region. We first determined the transcription start site for RIT2 by primer extension using total RNA from human retinas. The strongest signal was seen at 179 bp upstream of the initiation ATG (data not shown). This position is 5 bp upstream of the messenger RNA (mRNA) 5′-end annotated in RefSeq (GenBank: NM_002930). Sequence alignment of the 5′-flanking region of human and mouse Rit2 revealed many blocks of conserved sequences within the 1 kb upstream from the transcription start site and in the 5′-untranslated region. Based on the concept that the conserved regions are more likely to contain functional regulatory elements, we chose the 1.3 kb upstream containing the conserved 1 kb region as a tentative promoter, and constructed two human RIT2 promoter-luciferase reporter vectors, RIT2 −1290 to +76 bp (RIT2-1290/luciferase) and −374 to +76 bp (RIT2-374/luciferase), by PCR. As the template for PCR, we used a P1 genomic clone for human RIT2, instead of human genomic DNA, because RIT2 genomic clones had already been obtained earlier by screening a human genomic P1 library, and therefore were available in the laboratory.

Next, we used transient transfection assays to identify promoter activity of the 5′-upstream region of human RIT2. For these studies, we sought a cell line expressing RIT2, since such a cell line would more likely mimic the in vivo regulatory milieu compared to non-expressing cells. We screened three human neuroblastoma cell lines, SK-N-MC, SK-N-AS, and SK-N-DZ [37-39], and the transformed embryonic kidney line HEK293 [40] for RIT2 expression by RT-PCR using the human brain as the positive control. Morphologically, SK-N-MC (labeled as MC) and SK-N-AS (AS) did not show notable neurite outgrowth, whereas SK-N-DZ (DZ) extended long processes and formed intricate networks of neurites (Figure 1A). Interestingly, given that RIT2 has been implicated in neurite outgrowth [10], the expression levels of endogenous RIT2 generally correlated with the degree of neurite outgrowth, with SK-N-DZ expressing RIT2 at the highest level (Figure 1B). RIT2 expression was undetectable in HEK293. Using the three neuroblastoma and HEK293 cell lines, we performed transient transfection with RIT2-1290/luciferase and RIT2-374/luciferase. Promoter activity of RIT2-1290/luciferase overall correlated with the level of endogenous RIT2 expression, with the highest activity observed in SK-N-DZ (Figure 1C). Promoter activity of RIT2-374/luciferase was approximately 40% of that of RIT2-1290/luciferase in SK-N-DZ, but the difference in activity between the cell lines was not as prominent as seen with the longer −1290 fragment. These results suggest that the RIT2 −1290 to +76 bp region has promoter activity that drives expression preferentially in differentiated neuronal cells, and that at least some of the regulatory elements responsible for neuronal expression are located in the region between –1290 and –374 bp.

Rit2 is expressed in the ganglion cell layer and inner nuclear layer of the mouse retina

Because of our interest in retinal biology, we next aimed to clarify the expression pattern of Rit2 in the retina. Although we had previously shown by in situ hybridization that Rit2 mRNA is present in RGCs [5], we wanted to examine RIT2 expression at the protein level by immunostaining retinal flat mounts and sections with a focus on RGCs. We analyzed the expression of RIT2 and POU4F2 proteins, with the latter serving as an RGC marker. In retinal flat mounts (Figure 2A–D), RIT2 was detected in the cell body with stronger staining in cell membranes, and RIT2-positive cells were more numerous than POU4F2-positive cells. Because approximately 70% of RGCs express POU4F2, and the GCL contains displaced amacrine cells, the staining pattern suggests that RIT2 is expressed in both RGCs and amacrine cells. Immunohistochemistry of retinal sections (Figure 2E–H) showed RIT2 protein in all GCL cells, and this distribution was broader than that of POU4F2, confirming the results of the flat mounts. In addition, RIT2 protein was also detected in the inner plexiform layer (IPL) and in cells located at the inner and outer borders of the INL (Figure 2E; short and long arrowheads, respectively). We also analyzed RIT2 expression in the mouse embryonic retina by immunohistochemistry. Weak staining of the RIT2 protein was observed in the GCL and inner neuroblastic layer (NBL) at E15.5, and RIT2 expression significantly increased from E15.5 to newborn to adults (Appendix 3).

We further analyzed Rit2 retinal expression using LCM. The GCL, INL, and ONL fractions were collected by LCM of mouse retinal sections, and the resulting samples were analyzed by RT-qPCR to measure the mRNA levels of Rit2, Pou4f1, Pou4f2, and Gapdh. Consistent with the immunohistochemistry data, Rit2 was highly expressed in the GCL and INL, but Rit2 mRNA was not detected in the ONL sample (Figure 2I). Pou4f1 and Pou4f2 mRNAs were detected predominantly and exclusively in the GCL, which is consistent with published results [26,27,29,34]. Interestingly, Pou4f1 mRNA was also detected in the INL, albeit at a low level. This observed Pou4f1 expression in the INL is unlikely due to contamination of the samples because Pou4f2 mRNA was not similarly detected in the INL (Figure 2I). We also analyzed RIT2 protein levels in the mouse retina at different ages by western blotting. RIT2 protein levels increased from P2 to adults by more than 15-fold (Figure 2J,K).

RIT2 protein is detected in all neuronal cell types in the inner nuclear layer

During the initial analyses of RIT2 expression with a focus on RGCs, we also detected RIT2 expression in cells of the INL by immunohistochemistry of the retinal sections. To identify the cell types that express RIT2 in the INL, we performed double-labeling immunohistochemistry (Figure 3) with markers preferentially expressed in horizontal cells (LIM1), bipolar cells (VSX2), and Müller glia (vimentin, labeled as VIM) as well as two markers that stain horizontal, amacrine, and ganglion cells (calbindin and PAX6). We observed colocalization between RIT2 and each of the neuronal cell markers, but failed to see overlap with vimentin. Nuclear localized LIM1 expression had a near perfect overlap with RIT2 in cells located at the outer aspect of the INL where horizontal cells are present (Figure 3A,F,K). Similar colocalization was observed between VSX2 and RIT2 (Figure 3B,G,L). Antibodies against calbindin D28 labeled horizontal cells, including their dendrites, as well as a subset of amacrine cells and cells in the GCL. Although calbindin D28 generally has a stronger and more discrete signal and a better separation of sublaminae of the IPL, RIT2 and calbindin D28 exhibited a high degree of overlap, particularly regarding horizontal cell bodies and their dendritic arbors (Figure 3C,H,M). PAX6, although often used as a marker for amacrine and ganglion cells, is also expressed in horizontal cells; however, the spatial separation of these cells in the retina makes it easy to distinguish them (Figure 3D,I,N). Therefore, we confirmed the identity of these cells by the association of RIT2 and PAX6. The only marker that failed to show a close association with RIT2 was the Müller radial glial marker vimentin (Figure 3E,J,O). The thin thread-like projections spanning the retina were intensely labeled with vimentin but not with RIT2. Taken together, it appears that all neuronal cell types in the INL express RIT2, while Müller glia does not. In addition to cells in the INL, we also observed RIT2 positive staining, although weaker, in the inner segments of photoreceptors. This observation conflicts with the LCM RT-qPCR results, which showed no detectable Rit2 expression in the ONL (Figure 2I), and remains to be resolved.

Bioinformatic prediction of POU4 protein binding sites

To identify candidate regulatory elements within the RIT2 1.3 kb promoter region, we repeatedly searched for the presence of predicted transcription factor binding sites using MatInspector and found a number of consensus binding sites, including multiple AP1, four E-box, multiple POU domain factor including seven POU4F, a CREB1, and an SP1 sites. Based on the essential role of POU4 factors in the development and survival of sensory ganglion neurons that also express RIT2, we hypothesized that POU4 factors might regulate RIT2 in RGCs. We therefore searched possible binding sites for POU4 proteins in the RIT2 upstream region in further detail using a different program, Patser [49], with the published position-specific scoring matrices for POU4F1/F2 (Matrix 1) [50] and POU4F2 (Matrix 2) [27], of which the consensus elements are gcATAATTAAT and a/gCTCATTAAt/c, respectively. Although the former consensus site was shown to have 1,000-fold higher affinity than the latter consensus site [50], we used both matrices for comparison in the case of our native RIT2 promoter elements. Patser predicted 15 and 17 sites using Matrices 1 and 2, with scores ranging from 4.97 to 0.07 (highest possible score 12.195) and 8.25 to 0.05 (highest possible score 12.542), respectively. Using a semiarbitrary cut-off score of 3.0, four sites were predicted with each Matrix, and Sites 2 and 6 were predicted with both matrices, resulting in a total of six predicted sites (Sites 1–6; Figure 4 and Appendix 4).

POU4 proteins bind to the predicted sites

To experimentally validate Sites 1–6, we performed EMSAs with proteins made by in vitro translation for all POU4 factors and their isoforms [34,54]. The production of five POU4 (POU4F1-s [short isoform], POU4F1-l [long], POU4F2-s, POU4F2-l, and POU4F3) and control luciferase proteins was confirmed by ³⁵S-methionine labeling, and clear major bands of the expected size were obtained for all (Figure 5A). First, we performed EMSAs using Probes 1–6, corresponding to Sites 1–6 (Figure 4), with unlabeled POU4 and luciferase proteins. Probe 2 yielded strong shifted bands with all POU4 proteins within an hour of autoradiography (Figure 5C), whereas 2–3 days of exposure were required to obtain clear bands with Probes 1 and 6 (Figure 5B, D), and even after exposure for 7 days, Probes 3, 4, and 5 did not show detectable shifts (data not shown). Next, we tested the sequence-specificity of binding using wild-type (Probes 1, 2, and 6) and mutated (TTAA to GGCC; Probes 1m, 2m, and 6m) probes with the POU4 and luciferase proteins. All shifted bands were lost with the mutated probes (Figure 5B–D). To compare the relative binding strength of these sites, we employed cold oligomer competition. Binding of ³²P-labeled Probe 2 to POU4F1-s protein was competed with unlabeled “competitor” oligonucleotides containing Sites 1, 2, 4, 6, and 2m as well as the reported POU4F1 consensus site [50]. The strongest competition, i.e., competition with the lowest amount, was observed with the POU4F1 consensus site (Figure 5E). The competition with Site 2 was 10-fold weaker than that with the consensus site, but 10-fold- to 100-fold stronger than that with Sites 1 and 6. In contrast, Site 4 competed only marginally, and Site 2m hardly competed (Figure 5E). In summary, of the six predicted sites, three were bound by POU4 proteins under the conditions used, and the relative binding strength was approximately 1,000, 100, 1–10, and 1–10 for the POU4F1 consensus site, Sites 2, 6, and 1, respectively.

Different POU4 proteins show distinct activities in modulating the RIT2 promoter

We next performed cotransfection assays with POU4 expression vectors to test whether POU4 factors can modulate the RIT2 promoter. We used SK-N-AS cells as host for these studies because i) the cells express endogenous RIT2 at a detectable level (Figure 1B), indicating that they contain the transcriptional machinery for activating the RIT2 promoter; ii) their endogenous RIT2 expression is low (Figure 1B), suggesting that background activation of the RIT2 promoter would be low; and iii) they express POU4 factors at low or undetectable levels as analyzed by RT-PCR (data not shown), suggesting that these cells might be responsive to increased expression of POU4 factors. The results of the cotransfection studies were somewhat complicated, with different POU4 factors and isoforms having different effects (Figure 6). The effects of POU4 factors were generally stronger with RIT2-1290/luciferase, as compared to RIT2-374/luciferase. With the RIT2-1290/luciferase construct, POU4F1-s, POU4F2-l, and POU4F3 all significantly increased reporter activity to different degrees, whereas POU4F2-s significantly repressed it, with POU4F1-l showing no effect. With the RIT2-374/luciferase construct, POU4F1-s still significantly increased reporter activity; however, the effects of POU4F2-l and POU4F3 were no longer significant. Both POU4F1-l and POU4F2-s significantly decreased reporter activity. In addition to POU4 factors alone, we also tested whether ISL1, a LIM homeodomain transcription factor expressed in RGCs [55,56], has an effect on the RIT2 promoter modulation by POU4 factors. ISL1 by itself had no effect on either construct; however, ISL1 modestly but significantly enhanced the activity of POU4F2-l with both the −1290 and −374 constructs (Figure 6). The effect of ISL1 was also significant on the activity of POU4F1-s with the −374 construct, POU4F1-l with the −1290 construct, and POU4F2-s with both constructs. These results suggest that the POU4 factors can indeed modulate the RIT2 promoter, and that various POU4 factors and their isoforms have distinct activities, which can be further modulated by the activity of ISL1.