Functional analysis of the rod photoreceptor cGMP phosphodiesterase alpha subunit gene promoter: Nrl and Crx are required for full transcriptional activity

To understand the factors controlling expression of the cGMP phosphodiesterase type 6 (PDE6) genes, we have characterized the promoter of the human PDE6A gene that encodes the catalytic alpha-subunit. In vivo DNase I hypersensitivity assays revealed two sites immediately upstream of the PDE6A core promoter region. Transient transfection assay in Y79 cells of constructs containing varying lengths of the promoter region showed a decrease in promoter activity with increasing length. The most active segment contained a 177-bp upstream sequence including apparent Crx and Nrl transcription factor binding sites. Both Crx and Nrl transactivated the PDE6A promoter in HEK293 cells and showed a >100-fold increase when coexpressed. Coexpression of a dominant negative inhibitor of Nrl abolished Nrl transactivation but had no effect on Crx. DNase I footprinting assays identified three potential Crx binding sites within a 55-bp segment beginning 29 bp upstream of the transcription start point. Mutation of two of these sites reduced reporter gene activity by as much as 69%. Gel shifts showed that all three Crx sites required a TAAT sequence for efficient binding. Consistent with a requirement for Crx and Nrl in Pde6a promoter activity, Pde6a mRNA is reduced by 87% in the retina of Crx(-/-) mice and is undetectable in Nrl(-/-) mice at postnatal day 10. These results establish that both Nrl and Crx are required for full transcriptional activity of the PDE6A gene.


INTRODUCTION
The type 6 cGMP phosphodiesterase (PDE6) 1 in rod photoreceptors is a heterotrimeric enzyme that is essential for phototransduction functioning to lower cytoplasmic cGMP levels in response to light activation of the receptor rhodopsin (1)(2)(3). The αand β-subunits comprise the catalytic core of the enzyme and the γ-subunit partially inhibits activity to maintain basal levels in the dark. We have focused our studies on the PDE6A gene, which encodes the α-subunit of PDE6. Defects in the coding region of this gene account for a subset of autosomal recessive RP (4). Because all three genes that encode PDE6 subunits are required for subunit assembly (4)(5)(6), the regulation of gene expression for each subunit could be essential to maximize enzyme production. Dissection of promoter architecture may also be important to develop safe gene therapy vectors with controlled targeted expression in specific cell types. Our initial characterization of the PDE6 α-subunit gene promoter indicated that it shows very weak activity compared to other photoreceptor gene promoters including the PDE6B gene promoter (7). We suggest that the relatively weak expression of this promoter could be useful where over-expression with the well-characterized, most commonly used rhodopsin promoter could induce, rather than reverse retinal degeneration (8;9).
We previously reported that the PDE6A gene promoter is contained within an ~4 kb region upstream of the transcription start point, contains several potential cis-elements and can bind tissuespecific factors in a retinal nuclear extract (10). A DNaseI footprinting region was identified within the first 100 upstream nucleotides that was confirmed by the demonstration of a retina specific mobility shift. The promoter region was delimited to within a 177 bp upstream region that is active both in vitro (11) and in vivo in transgenic mice (12). Contained within this region are several potential cis elements that may bind retina specific transcription factors including CRX sites (CBE) and a non-consensus NRL-like site (NRE). CRX binds to the consensus sequence "(C or T)TAATC(C or A)" (13;14), and is critical for photoreceptor maturation (15;16) and expression of many photoreceptor genes in adult retina (14)(15)(16)(17)(18). NRL is required for normal maturation of rod photoreceptors as indicated by an Nrl knockout mouse that is devoid of functional rods (19), and patients with Cone-rod dystrophy (20). Nrl binds to the consensus sequence "GATGCNTCAGCC" and can transactivate rhodopsin promoter activity synergistically with Crx (14) by physical interaction with Crx (21). In this paper, we show that the PDE6A promoter is contained within a 100 bp segment of the upstream region and requires Nrl and Crx for full transcriptional activity.
Our data and that from studies of the PDE6B gene promoter (7) that shows greater activity, but contains essentially the same binding sites, indicate that promoter strength is at least in part due to the specific sequence and the combinatorial array of cis-elements within a given promoter.

EXPERIMENTAL PROCEDURES
Cell lines and cell culture--Media and antibiotics for tissue culture were purchased from Life Technologies. Fetal bovine serum (FBS) was purchased from Hyclone. HEK293 and Y79 cells were purchased from American Type Culture Collection and maintained in DMEM supplemented with 5% FBS, and 100 units/ml of penicillin and streptomycin, or RPMI 1640 supplemented with 10% FBS.
DNaseI hypersensitivity assay-Y79 or HEK293 cell nuclei were isolated at 4°C using a nuclei isolation kit (Sigma) following the manufacturers' protocol. The nuclei were pelleted at 2K rpm in an IEC clinical centrifuge and resuspended in 1x PBS. Approximately 5x10 6 nuclei were digested with DNaseI at varying concentrations ranging from 0-100 µg/ml. The reaction was incubated 10' on ice, and reactions were quenched by the addition of EDTA to a final concentration of 25 mM. Genomic DNA was extracted using the DNeasy Tissue Kit (Qiagen). For Southern blot analysis, ~15 µg of DNA was co-digested with AflII & XhoI at 37°C for 4 hours with one addition of each enzyme after 2 hours, and the digestion products were electrophoresed in a 0.7% agarose gel. DNA was transferred to a Nylon membrane, and hybridized with an α[ 32 P]-dCTP labeled 3'end probe. The results were visualized by autoradiography after a 5 day exposure.
These eight constructs are representative of different lengths of the upstream region of the PDE6A gene. The pGL3-260 and pGL3-260R constructs were prepared by amplifying with Pfu polymerase (Stratagene) using U-Sal and U-Xba primers with the pBS-322 vector as template. The PCR product was then cut with StuI and the 260-bp fragment obtained was cloned into the SmaI site of pGL3-Basic. Forward and reverse orientation was obtained and verified by PCR and restriction digests. The pGL3-300 construct was generated by restriction digest of the pBS-322 plasmid with SalI and NcoI and was inserted in to the XhoI/NcoI site of pGL3-Basic vector. The pGL3-1060 construct was generated by amplifying a ~1279-bp fragment from pBS-4.1 using primers designed to amplify that portion of the upstream region (HAU6-R; 5'-GTGCTGGGATTACAGGCGTGAG-3' and HARE-3; 5'-CTGTAAATTCTCCTGAAAGTCCCGCAGGAG-3'). The product was partially digested with NcoI to obtain a 1060-bp fragment, which was cloned into the SmaI and NcoI sites of pGL3-Basic. The pGL3-3800 was prepared by digesting pBS-4.1 with XbaI and NcoI and cloning the ~3.3-kb fragment into the NheI and NcoI sites of pGL3-Basic vector. A ~500-bp NcoI to NcoI fragment was then cloned into this intermediate construct to generate the pGL3-3800 construct. The pGL3-593 was made by digesting pGL3-3800 with SacI and ApaI and purifying the vector with the insert using a Qiagen (Valencia, CA) gel extraction kit following the manufacturer's protocol. The ends of the vector were filled in using T4 DNA polymerase and blunt-end ligated together. The pGL3-1542 plasmid was generated by digesting the pGL3-3800 construct with NheI and MluI, followed by purifying the vector with insert using a Qiagen gel extraction kit. The ends of the vector were filled in using T4 DNA polymerase and blunt-end ligated together. Plasmid preps were performed using Qiafilter Endofree Maxi or Mega kits (Qiagen) following manufacturer's recommended protocol. All plasmids were sequenced to verify integrity.
Adenofection of Y79 cells-Y79 cells were transfected using the adenofection method previously described for this cell line (11). Cells were harvested 48 h after infection and lysed in 100 µl of 1 X reporter lysis buffer (Promega) and light emission was measured immediately in a Turner Designs 20/20 luminometer (Promega). Values were normalized to viral β-gal expression.
All experiments were performed in triplicate a minimum of three times (n≥9).

Transfection of HEK293 cells with calcium phosphate-HEK293 cells were transfected
with calcium phosphate using a standard protocol. Briefly, ~6 x 10 5 HEK293 cells were plated in a 60 mm Petri dish in DMEM supplemented with 5% FBS and 1% penicillin/streptomycin and grown for 24 to 36 hours at 37°C to achieve 50 to 75% confluency. The medium was changed once 3 hours prior to transfection. Up to 16 µg DNA in 300 µl 250 µM CaCl 2 was combined dropwise with 300 µl 2X HBS and incubated 30 min. The DNA consisted of 1 µg β-gal internal control, 4-5 mg test plasmid, and 2-5 µg of Crx or Nrl expression construct. The DNA mixture was then added dropwise to the cells, incubated for 48 hours with one media change after 12 hours incubation.
Luciferase and β-Galactosidase assays--Cells were harvested and lysed in 350 µl of Galacto-Star reporter lysis buffer (Tropix, Inc.), cell debris was pelleted at 12K xg for 2 min at 4°C. For luciferase assays, 20 µl of cytosolic extract was diluted to 300 µl and light emission was measured immediately in the luminometer using a luciferase activity kit (Promega) following the suggested protocol. β-galactosidase activity was assayed with the Galacto-Star activity kit (Applied Biosystems) using 20 µl of the same cytosolic extract diluted to 300 µl. The luminometer was set to 30% (for HEK293 cells) or 40% sensitivity (for Y79 cells) with a delay time of 2 s and an integration time of 10 s. All transfections were done in triplicate a minimum of 3 times.
DNaseI footprinting of Crx-The plasmid pGL3-300 was used as a template for PCR separately with either sense or antisense radiolabeled primers to generate DNaseI footprinting probes. The reaction was performed in binding buffer containing 12.5 mM Hepes, pH 7.6, 100 mM KCl, 5 mM ZnSO 4 , 0.5 mM DTT, 2% polyvinyl alcohol, and 10% glycerol. Binding reactions were performed on ice for 20 min. in a total volume of 50 ul including labeled probe, 1x binding buffer with or without 100 ng purified bovine CrxHD-GST fusion protein (14). For DNaseI digestion, MgCl 2 and CaCl 2 were added to final concentrations of 5 and 2.5 mM, respectively, followed by 0.005, 0.01, or 0.02U DNaseI for 1 min. at RT. The reactions were terminated with 90 uL of 1x stop buffer (20 mM EDTA, pH 8.0, 1% SDS, 0.2 M NaCl, and 250 ug/mL glycogen). Ten µL of 2.5 mg/mL proteinase K was added and incubated at RT for 5 min. followed by phenol chloroform extraction, precipitation, and resuspension in 1x DNA sequencing gel loading buffer. DNA sequencing reactions were run along side in a 6% denaturing acrylamide gel.
Electrophoretic mobility shift assay-Gel shift assays were performed essentially as described (10) with minor modification. KCl optimum was determined to be 75 mM, and 5 % glycerol was included in the reactions. Incubations with and without label were done for 15 min. The binding reaction was run on a non-denaturing 6% acrylamide gel in 0.5x TBE for two hours at 380 volts.
CrxHD-GST protein was purified on a glutathione affinity column as previously described (14) Real-time quantitative PCR--Crx −/− mice were obtained from Dr. Connie Cepko, Dept. of Genetics, Harvard Medical School. The Crx −/− mouse was outcrossed to a congenic C57Bl/6 x SJL hybrid strain that is +/+ at the rd1 locus (6). The mice were bred to homozygosity for the Crx knockout allele and maintained for greater than six generations. Total RNA was isolated from eyes of 10-day old Crx −/− and Nrl −/− mice as well as background matched controls using RNAqueous RNA isolation kit (Ambion, Inc.) or TRIZOL (Invitrogen). Reverse transcription reactions (23) were performed with 2 µg total RNA using MMLV (Ambion) or Superscript II (Invitrogen) reverse transcriptase following manufacturer's guidelines. The reaction was diluted to 100 µl with water.
One µL of the diluted cDNA was used in each PCR reaction using either the Dynamo SYBR green qPCR kit (MJ Research) or Platinum Taq polymerase (Invitrogen) with SYBR green I (Molecular Beacons). Primer pairs were used that amplify portions of Pde6a-F, 5'- The reactions were run in the real time PCR Opticon II system (MJ Research) or iCycler (BioRad) and analyzed using the comparative C t method (∆∆C t ) with Gapdh as the normalizer (http://dorakmt.tripod.com/genetics/realtime.html). Error values are expressed as standard error of the mean (n=8).

<<<Figure 1>>>
The  Figure 1 is the portion of the PDE6A core promoter region analyzed in this study. This region contains three potential Crx binding elements (CBE1-3), one site resembling an Nrl/AP1 binding element (NRE/AP1) and one additional segment that is conserved in the upstream regions of both human and mouse (10). <<<Figure 2>>> To assess the chromatin status of the region around the PDE6A promoter region in vivo, a DNaseI hypersensitivity assay was performed with Y79 cells, that transcribe PDE6A mRNA (26). Treatment of intact Y79 nuclei with DNaseI revealed two fragments in close proximity (Fig. 2B, lanes 6, 7) within a 6.4 kb AflII/XhoI genomic fragment spanning the promoter region ( Fig. 2A). No cleavage products were observed using HEK293 nuclei, which do not express PDE6A (not shown). The estimated size of these fragments places the hypersensitive sites immediately upstream of the predicted core promoter region. This result is consistent with an active promoter within the region previously identified in vitro (10).

A 177 bp segment of the upstream region yields the highest reporter gene activity in vitro-
To delimit the core promoter region, luciferase reporter gene constructs containing varying lengths of the upstream region were assayed by transient transfection in Y79 Rb cells. Due to the relatively low promoter activity a modified adenofection method was devised to increase the efficiency of gene transfer several fold over previously reported methods (11). As shown in Figure 3, the luciferase activity was highest with the shortest fragments tested, pGL-300 and, pGL-260. Both of these constructs contain 177 bp of sequence extending upstream from the transcription start point, and either the entire 5'-untranslated region (pGL-300) or a downstream truncation of 40 bp (pGL-260), respectively. Reversing the orientation of the pGL3-260 construct (pGL3-260r) reduced activity equivalent to the promoterless control (pGL3-pl). The remaining four constructs tested (pGL3-593, -1060, -1542, -3800) all contain upstream repetitive elements (10) that appear to attenuate promoter activity.
Crx and Nrl transactivate the PDE6A promoter-To identify cis-elements necessary for transcriptional activity, we first tested transactivation activity of the photoreceptor-specific transcription factors, Crx and Nrl. Transient transactivation assays were performed in HEK293 cells with and without Crx and Nrl expression constructs. As shown in Figure 4, cotransfection of the pGL3-300 PDE6A promoter construct in the absence of Crx or Nrl expression is below detectable limits. In contrast, the inclusion of the Crx or NRL expression construct increased luciferase activity at least 15-20 fold above background levels. Inclusion in the assay of both Crx and Nrl expression constructs further stimulated activity (>100 fold over background) similar to the synergistic activation of rhodopsin (21).

<<<Figure 4>>>
Crx binds within a 55 bp segment containing three potential binding sites-To identify the precise sequence within the PDE6A promoter where Crx binds, DNaseI footprinting was done with a purified recombinant Crx fragment containing the DNA binding domain (14). An extended binding region was observed on both strands of the DNA in the region from -29 to -84 ( Figure 5, cf. lanes 2, 3 to lanes 4, 5). This region consists of three potential Crx binding sites all containing a TAAT core consensus Crx binding sequence [see Fig. 1; (13;14)].

<<<Figure 6>>>
Crx binds more tightly to CBE-1-The relative binding strength of each potential Crx binding site was determined by EMSA. Eleven double stranded segments of varying lengths were generated containing various combinations of mutated and intact CBE sites (Fig. 6A). The shorter oligomers (Fig. 6B, lanes 4-11) showed comparable labeling to the longer oligomers (lanes 1-3) when electrophoresed for shorter periods of time (not shown). For each oligomer, the AAT of the core TAAT Crx binding site was mutated to CCG. Mutation of CBE-1 alone (Fig. 6B, lane 11

Crx and Nrl are required for full promoter activity in transfected cells-The contribution of Crx
and Nrl to promoter activity was assessed by site-directed mutagenesis of the TAAT core sequence recognized by Crx, and with an Nrl dominant negative expression construct shown to abolish Nrl activity (22). The Crx core binding sequence TAAT was mutated to CCGT at sites CBE-2 and CBE-3 to generate constructs m13 and m16, respectively. Transient transfection analysis with the normal or mutated sequences in the absence of Crx (300-Crx, m13-Crx, m16-Crx, Fig. 7A) showed very low background activity. Mutation of the CBE-2 site (m13+Crx) showed a 50 % reduction in activity, and mutation of CBE-3 (M16+Crx) reduced activity by 69 %. Inclusion of the CBE-2 mutant with Crx and Nrl expression constructs (m13+Crx+Nrl) caused a 61 % reduction in reporter gene expression, and the CBE-3 mutant with Nrl and Crx (m16+Crx+Nrl) caused a 79 % reduction compared to the activity seen with the 300 construct, Crx and Nrl (300+Crx+Nrl). The CBE-3 mutant construct cotransfected with Nrl and Crx reduced reporter gene activity close to the level seen with Nrl alone (300+Nrl) suggesting that the synergism between Crx and Nrl was abolished.
Nrl transactivation of the promoter mutants in the absence of Crx was not significantly different than that observed with the intact promoter (not shown).

<<<Figure 7>>>
The expression construct DD10 generates a naturally occurring truncated Nrl protein that inhibits Nrl activity (22). Use of this construct with the 300 construct and an Nrl expression construct abolished Nrl mediated transactivation (+Nrl+DD10, Fig. 7B). Inclusion of all three expression constructs (+Crx+Nrl+DD10) reduced promoter activity to a level similar to that seen with Crx alone (+Crx) thereby removing the synergistic activation of the promoter.

<<<Figure 8>>>
Crx and Nrl are required for full transcriptional activity in vivo: knockout mouse studies-To assess the requirement of Nrl and Crx for PDE6A transcription we determined the relative levels of transcript in Crx (15) and Nrl knockout mouse retinas. Total RNA isolated from whole eyes of post natal day 10 (PN10) Crx −/− , and Nrl −/− mice, and from adult Nrl −/− mice, was converted to cDNA, amplified with primer pairs that amplify short segments (300-500 bp) of the coding region of Pde6a, and visualized in real time. Pde6b, Rho were used as positive controls and Gapdh was used for normalization. In the Crx −/− mouse Pde6b and Rho transcript levels were reported to be at 75 and 8 %, respectively of the levels observed in Crx +/+ littermates (15). As shown in Fig. 8 19). It seems most likely that the Pde6a signal detected in Nrl null mice is from inner retinal cells consistent with our previous report showing in situ hybridization labeling in the inner nuclear layer of adult mouse retina (12).
Comparison of the anatomy of the Pde6a and Pde6b promoters reveals a striking similarity.
Both core promoters are contained within 105 bp upstream of the tsp, both initiate transcription at one major and one minor site, and both consist of a similar array of cis elements including a GC box, an E-box, Crx and Nrl binding sites, yet there is a 5-8 fold lower activity of the Pde6a promoter (7;11). The primary differences are the spatial arrangement of the cis elements, the sequence of the NRE/AP1 site, and the absence of repetitive elements upstream of the Pde6b promoter. We have found that the difference in promoter activities can be accounted for by the repetitive elements and the NRE/AP1 sequence. Conversion of the Pde6a AP1 sequence to that of Pde6b and removal of the repetitive elements raises the activity of Pde6a to the level of Pde6b 3 . We also found that the position of the cis binding sequence is critical for promoter activity. Moving the Pde6b NRE/AP1 site further upstream did not elevate Pde6a activity. These results and the results reported here indicate that there is a direct correlation between the spatial array of cis elements and promoter strength. Thus, it may be possible to develop a series of promoters with defined strengths simply by rearranging the same cis-elements; a hypothesis that is worth testing for the development of safe predictable vehicles for gene therapy.
Our results show that of three CBE binding sites in the Pde6a promoter, a site that is overlapping with the NRE/AP1 like site (CBE-1) binds Crx very tightly in isolation, however DNaseI footprinting, transient transfection assays and gel shift assays indicate that CBE-2 and CBE-3 sites together can also show strong binding. The close proximity of Nrl and Crx could promote protein/protein interaction that has been demonstrated with these factors (21). The gel shift assays result where CBE-1 and CBE-2 show weak binding in isolation (Fig. 6A, lanes 9, 10), but binding at each site is relatively strong when the sites are combined (lanes 3, 6, 7), is seemingly contradictory to the other binding data with CBE-1. The data is consistent, however with a requirement for defined DNA bending, and possible cooperativity of Crx binding. Also consistent with this idea is our previous finding using human retina nuclear extracts in gel shift assays, where a retina specific shift was only observed when the entire promoter region was included in the probe (10). The rhodopsin proximal promoter region is a comprised of a very similar array of cis-elements with an NRE flanked on both sides by Crx binding sites (29), and this promoter also shows weak activity in the absence of an upstream enhancer (30). Overall, the data indicate that binding of Crx to all three CBE sites, Nrl binding to the NRE/AP1 site, and presumably the basal transcription machinery are required for full activation of the Pde6a promoter.