A novel promoter element, photoreceptor conserved element II, directs photoreceptor-specific expression of nocturnin in Xenopus laevis.

Nocturnin is a vertebrate circadian clock-regulated gene, and in Xenopus laevis its mRNA is specifically expressed in retinal photoreceptor cells. We have investigated the transcriptional regulatory mechanism that drives this precise spatial expression pattern of the nocturnin gene. A deletion series of the nocturnin 5'-flanking sequence driving the green fluorescence protein (GFP) reporter was used to generate transgenic Xenopus tadpoles. We found that a construct containing 2.6 kilobase pairs of 5'-flanking sequence targeted high level GFP reporter expression specifically to photoreceptor cells, in a pattern identical to endogenous nocturnin. This photoreceptor-specific expression pattern was maintained with several further deletions of 5'-upstream sequence, including a short 59-base pair fragment. Within this region of 59 base pairs, three perfect repeats of a novel protein binding site were identified by electrophoretic mobility shift assay. Competitions using varying oligonucleotide sequences demonstrated that the sequence required for protein binding is CAGACAGGCTATA, designated photoreceptor-conserved element II (PCE II). The protein complex that binds to this element is enriched in retinal extracts, and mutations of PCE II which fail to bind the protein complex also fail to direct GFP reporter expression to photoreceptors. These results indicate that the PCE II in the proximal promoter of the nocturnin gene is sufficient for driving the photoreceptor-specific expression of nocturnin.

Precise spatial patterns of gene expression are critical for the proper function of all organisms. Within most of the central nervous system of vertebrates, the study of spatial regulation of transcription is difficult because of the vast heterogeneity of this tissue. The retina is a part of the central nervous system which is more amenable to these types of study because the cells are organized in morphologically distinct layers, and the various cell types have been well characterized (1-3).
Many genes have been shown to be expressed specifically in photoreceptor cells within the retina, including those known to be involved in the phototransduction cascade or other photoreceptor processes. Biochemical studies, such as DNase I foot-printing and electrophoretic mobility shift assays (EMSAs), 1 have identified a number of protein binding sites within the promoters/enhancers of several retina-specific genes, including opsin, interphotoreceptor retinoid-binding protein (IRBP), and rod arrestin (4 -7). DNA elements named Ret1 and PCE I were defined in the rat opsin promoter and the arrestin promoter, respectively, but these two elements are quite similar to each other (8,9). Further analysis of the rat opsin gene revealed two additional protein-binding elements named Ret2 and Ret3 (10). A positive acting opsin regulatory element Ret4 was identified using an in vitro transcription system from bovine retinal nuclear extracts (11). It has been demonstrated that a member of the OTD/OTX family, CRX (cone rod homeobox), binds in vitro to the Ret4 and Ret1 sites and to an element named BAT-1, resulting in the stimulation of the transcriptional activities of opsin, IRBP, and several photoreceptor cell-specific gene promoters (5,6). Also within the opsin upstream region, more protein binding sites were found: a sequence similar to the Drosophila glass binding site (12), a putative site E opsin-1 for the basic helix-loop-helix transcription factor Mash-1 (13), BAT-1 and XRS-1 sites (14), and Nrl response element (an extended AP-1-like sequence) or AP-1 sites (15)(16)(17)(18). Despite the identification of these protein-binding elements, the transcriptional regulatory mechanism that governs the photoreceptorspecific expression of these genes is still not known. In many cases the examination of these elements have only been done in vitro.
Nocturnin is a retina-specific gene first identified in Xenopus laevis by a differential display screen seeking circadian clockregulated genes (19,20). In situ hybridization demonstrated that nocturnin was specifically expressed in the photoreceptor layer within the retina (21). In both cyclic light and constant darkness, nocturnin mRNA rises in early night (or subjective night) and returns to low levels around midnight. The mRNA rhythms are controlled by the retinal circadian clock at the level of transcription (21). The precise spatial and temporal pattern of gene transcription makes the nocturnin promoter an interesting candidate for studies of transcriptional regulation.
The transgenic technique in Xenopus provides a unique experimental method to address questions of specific gene expression in vivo (22,23). The transgenic method results in animals that are not mosaic and can therefore be analyzed directly without breeding (24). Previous work has demonstrated that a GFP reporter can be faithfully targeted to specific tissues using tissue-specific promoter/enhancers (22,25,26). To investigate the regulatory mechanism of the nocturnin gene, we analyzed the in vivo expression patterns of GFP reporters driven by various portions of the nocturnin 5Ј-flanking sequence in transgenic Xenopus. The in vivo experiments, coupled with in vitro binding studies, resulted in the identification of a novel element that is sufficient to drive reporter gene expression specifically in retinal photoreceptors. We have named this novel element PCE II.

Construction of GFP Reporter Plasmids
The transcription start site of nocturnin has been identified previously using both primer extension and 5Ј-rapid amplification of cDNA ends (21). Abbreviations of upstream sequences are based on the sequence position relative to the transcription start site. For example, XNP(Ϫ2.6kb/ϩ20bp)-GFP refers to the construct of GFP driven by the Xenopus nocturnin promoter from 2.6 kb upstream to 20 bp downstream of the transcription start site.
XNP(Ϫ2.6kb/ϩ20bp)-GFP-A Xenopus nocturnin genomic clone was cut with SacI and BsmBI to generate a fragment extending from ϳ2.6 kb upstream of the transcription start site to 20 bp downstream. This fragment was cloned into the SacI and SmaI sites of the pGEM7 vector and named p42B2. This ϳ2.6-kb fragment was then excised with SacI and HindIII and cloned into the SacI/HindIII sites of the pEGFP vector (CLONTECH).
XNP(Ϫ398/ϩ20bp)-GFP-The p42B2 clone was digested with SpeI and HindIII and then cloned into the NheI and HindIII sites of the pGL2-basic vector (Promega), XNP(Ϫ398/ϩ20bp)-Luc. This fragment was then excised with SacI and HindIII and transferred into the pEGFP vector.
The two complementary XNP(Ϫ77/Ϫ18bp) oligonucleotides were annealed as described in the EMSA method (below) and cloned into the SacI and BglII sites of the pGL2-basic vector. It was then excised by XhoI and HindIII and transferred into the pEGFP vector. The two complementary Mut-PCE oligonucleotides were annealed and then directly subcloned into the XhoI and HindIII sites of the pEGFP vector.

Microscopy
Tadpoles (10 -14 days old) were fixed overnight in 4% paraformaldehyde in phosphate buffered saline. The samples were equilibrated in 30% sucrose in phosphate-buffered saline, embedded in O.C.T. compound (Ted Pella, Inc.), and frozen sections (12-m thickness) were prepared. Images were obtained on an Olympus IX70 inverted microscope and digitized by OlymPix camera system. A fluorescein isothiocyanate/enhanced GFP long pass filter cube (41012, Chroma) was used for fluorescent images. Some images were generated using confocal microscopy (Nikon Eclipse TE200, KECK Center for Cellular Imaging, University of Virginia). Green fluorescent images were generated using 488 nm blue argon. Three-dimensional images were obtained using serial optical sections with the C.Imaging software (Compix, Inc.). Each image of the Z-series composite is 20 m thick.

Isolation of Retinal Nuclear Extracts
Adult frogs were maintained on 12-h light:12-h dark cycles. Adult eyes were dissected at Zeitgeber time (ZT) 2 and ZT 11.5, in which ZT 0 is defined as the time of light onset (dawn) and ZT 12 as dark onset (dusk). Brain, muscle, liver, and heart were dissected at ZT 11.5. Nuclear extracts for EMSA were prepared from adult Xenopus retinas and other tissues using the method modified from (27). Tissues were homogenized and pelleted in phosphate-buffered saline and then quickly resuspended in hypotonic buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol) with 10% Igepal (Fisher) at 4°C for 15 min. After centrifugation at 15,000 rpm, the pellet was dissolved in high salt buffer (20 mM HEPES, pH 7.9, 25% glycerol, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol). The nuclei were extracted for 20 -30 min on a shaking platform at 4°C and then centrifuged at 15,000 rpm. The supernatant was quickly frozen in aliquots and then stored at Ϫ70°C until use. Protein concentrations were tested using the standard method of D c protein assay (Bio-Rad).

EMSA
Oligonucleotides used as probes in EMSA are listed in Fig. 8B (Life Technologies, Inc. and MWG Biotech, Inc.). Two complementary oligonucleotides were annealed in the polynucleotide kinase buffer by heating to 65°C for 10 min and then cooling slowly to room temperature. Double-stranded oligonucleotides were end labeled with [␥-32 P]ATP (NEN Life Science Products) by T4 polynucleotide kinase (New England Biolabs) and purified by NucTrap columns (Stratagene). Binding reaction mixtures included the following: 2-4-g nuclear extracts (for ret-inas it was a mixture of extracts isolated at ZT 2 and ZT 11.5), 8 mM HEPES, 60 mM KCl, 2 mM EDTA, 4 mM spermidine, 1 g of salmon sperm, 1 g of poly(dI-dC), 0.1 g/l bovine serum albumin, 0.03% Nonidet P-40, 0.5% Ficoll, 10% glycerol, 1 mM dithiothreitol, and ϳ1 ng of radiolabeled probe. Binding reactions (15 l) were incubated for 0.5 h at room temperature. For competition assays, 50-and 200-fold unlabeled oligonucleotides were added and incubated for 20 -30 min at room temperature before the addition of radiolabeled probe. A 6% polyacrylamide gel (29:1 acrylamide:bisacrylamide) was preelectrophoresed in 0.5 ϫ TBE (44.5 mM Tris, 44.5 mM boric acid, 1.25 mM EDTA, pH 8.3) for at least 1 h at 4°C. Samples were electrophoresed at a constant voltage of 210 V at 4°C for 2 h. Dry gels were exposed to film for 1-12 h at room temperature.

RESULTS
In the studies presented here, our goal was to define the portion of the nocturnin promoter/enhancer responsible for correct spatial expression of the nocturnin gene. Previous analysis of the endogenous nocturnin mRNA by Northern blot showed that nocturnin mRNA can be detected only in the retina and is not evident in brain, heart, kidney, skeletal muscle, or liver (19). In situ hybridization demonstrated that within the Xenopus retina the expression of the nocturnin mRNA is spatially restricted to the retinal photoreceptor cells (19).
A portion of a Xenopus nocturnin genomic clone containing ϳ2.6 kb of 5Ј-flanking sequence and 20 bp of transcribed sequence was isolated. A portion of the most proximal (Ϫ398/ ϩ20bp) of this sequence is shown in Fig. 1A. There are several putative protein binding sites in this region, including an Ebox-like element, CRX-like elements, and three perfect repeats of a previously unidentified element (labeled PCE II). Serial deletions (constructs 1-5) and a mutant (construct 6) of the 5Ј-flanking sequence were made as shown in Fig. 1B. GFP reporters driven by different lengths of the nocturnin 5Ј-flanking sequence were used to generate transgenic tadpoles for examination of the resulting GFP expression patterns.
Analysis of Transgenic-positive Tadpoles by PCR-All resulting tadpoles were genotyped by PCR, using genomic DNA isolated from clipped tails. Endogenous nocturnin was amplified (ϳ250-bp band) to monitor the quality of genomic DNA, and GFP coding sequence was amplified (ϳ480-bp band) to examine whether XNP-GFP constructs had been integrated into the tadpole genome. Fig. 2 shows the PCR results from the various samples of transgenic-positive tadpoles (tadpoles 2-7), and a nontransgenic tadpole (tadpole 1) is exhibited in Figs. 3 to 7. Endogenous nocturnin bands were seen in all of the tadpoles, whereas the GFP bands were observed only in the transgenicpositive tadpoles.
The probability of survival after transgenesis in our hands varies from 3 to 15% of total eggs injected. Our PCR results confirmed that 40 -80% of the living tadpoles resulting from Photoreceptors-We first produced transgenic tadpoles using the construct containing 2.6 kb of 5Ј-flanking sequence, XNP(Ϫ2.6kb/ϩ20bp)-GFP (Fig. 3). Careful examination of the living tadpoles generally revealed no expression in any nonocular tissue. In one tadpole (out of more than 20 examined), low level GFP expression was observed in pineal gland and olfactory epithelium (data not shown), possibly because of the position effect on the inserted transgene. We sectioned tadpoles carrying this transgene and examined retinal expression. In the phase-contrast images at low magnification (Fig. 3A), the sections of the tadpoles show the structures of the eyes as well as some other body tissues. In the corresponding fluorescence image, GFP expression is observed only in the outer lamina of the retina, with no detectable GFP expression in other retinal cell layers, lens, or nonocular tissues. With higher magnification, it is clear that GFP is expressed in the cell bodies of the photoreceptor cells but not in outer segments of photoreceptor cells or other retinal cells (Fig.  3A). An image of the retina from a sibling tadpole that did not have the XNP(Ϫ2.6kb/ϩ20bp) construct (the PCR-negative tadpole 1, Fig. 2) exhibits only a low level of background autofluorescence at the bottom of the outer segments of the photoreceptor cells, and no detectable GFP is observed in the cell bodies (Fig. 3A). The tadpole produced using linearized GFP vector with no promoter also does not exhibit any GFP expression in any of the examined tissues (Fig. 3B) and is identical to the PCR-negative tadpole in Fig. 3A. This same pattern of very low background autofluorescence right at the inner and outer segment junction is also observed in wild-type embryos (data not shown) and is clearly distinguishable from the GFP fluorescence observed in the transgenic retinas.
It is difficult to determine whether all cells in the photoreceptor layer are expressing GFP using traditional microscopy because the cells within the photoreceptor layer are organized in a somewhat staggered pattern. Therefore, only a subset of the cells is in focus at once. To examine more closely the photoreceptor cells that express GFP, confocal microscopy was used to generate a Z-series of optical sections (see "Experimental Procedures"). Confocal images through cryosections of fixed eyes from transgenic-positive tadpoles again showed that GFP expression is limited to photoreceptor cells. Within the photoreceptor layer, the combined Z-series show that both rod and cone cells express GFP (Fig. 4). These results are consistent with the expression patterns of the endogenous nocturnin, suggesting that the XNP(Ϫ2.6kb/ϩ20bp) is sufficient to target the GFP reporter to photoreceptor cells appropriately.
Deletion of the Nocturnin Promoter to XNP(Ϫ108/ϩ20bp)-Further deletions were made from the 2.6-kb fragment to narrow the region of the nocturnin promoter which is responsible for proper spatial expression. The GFP expression pattern did not change with the deletion of the nocturnin promoter from XNP(Ϫ2.6kb/ϩ20bp) to XNP(Ϫ398/ϩ20bp), although the GFP signals seem to be generally weaker in the latter case (Fig. 5). Therefore, further deletions of the nocturnin promoter were performed to detect whether the spatial expression pattern of the GFP reporter would change. In tadpoles produced using both XNP(Ϫ161/ϩ20bp)-GFP and XNP(Ϫ108/ϩ20bp)-GFP, GFP can again only be observed in the photoreceptor layer, not in other retinal cells or other tissues (Fig. 5). These results indicate that elements sufficient for photoreceptor-specific expression must be contained within the Ϫ108/ϩ20bp sequence.

GFP Reporter Is Only Expressed in Fully Differentiated
Photoreceptor Cells-The ciliary marginal zone (CMZ) is the region at the peripheral edge of the retina and contains undifferentiated retinal progenitor cells. The youngest cells and stem cells are closest to the periphery, the proliferative retinoblasts in the middle, and the cells that have stopped dividing at the central stage (28). In amphibians, the retina grows throughout life by adding new cells of all types from the CMZ. Fig. 6 shows that the GFP reporter directed by the XNP(Ϫ108/ϩ20bp) is targeted only to the mature photoreceptor cells. Those cells in the CMZ which have not differentiated to photoreceptors do not exhibit any detectable GFP (Fig. 6). The same pattern has been observed in the other GFP constructs directed by longer nocturnin promoters (data not shown). These results indicate that the transcription of nocturnin is a feature of the fully differentiated photoreceptor cells (see "Discussion").
The GFP Expression Pattern Is Maintained in the Transgenic Tadpoles Produced by XNP(Ϫ77/Ϫ18bp)-Within the Ϫ108/ ϩ20bp sequence, we noticed several elements of interest. One was a sequence with high similarity to an E-box, the other was three nearly perfect repeats of a novel sequence (Fig. 1A). To determine whether the E-box-like element was involved in directing spatial expression of nocturnin, we deleted this element to generate a XNP(Ϫ77/Ϫ18bp) fragment which only contains the three repeated elements. The GFP expression levels from this shorter promoter fragment are low in some of the photoreceptor cells (Fig. 7A), but from the confocal image it appears that the GFP is still expressed in both rod and cone cells (Fig. 7B). No cells were found within the photoreceptor layer that did not express GFP, although the levels of expression were variable. PCE II Was Identified by EMSA-To determine the protein binding sequences within XNP(Ϫ77/Ϫ18bp), EMSAs were performed using nuclear extracts isolated from Xenopus retina (Fig. 8). As mentioned previously, we had observed three perfect repeats of 5Ј-CAGACAGGCTTATA-3Ј within the proximal promoter of the nocturnin gene (Fig. 8A), suggesting that the repeated sequence may be the protein binding site. One prom-   Fig. 8B). The addition of excess unlabeled oligonucleotides identical to the radiolabeled probe resulted in a dosage-dependent inhibition of the DNA binding activity (lanes 2-4), whereas the addition of the nonspecific competitor poly(dI-dC) (lanes 26 -28) had no effect, suggesting that the binding activity is specific. The binding was not inhibited by competition with oligonucleotides that were mutated in any part of the sequence of 5Ј-CAGACAGGC-TATA-3Ј (lanes [11][12][13][14][15][16][17][18][19][20][21][22], suggesting that this sequence is involved in the DNA-protein interaction. However, the competitors with mutations outside of the recognition site still resulted in a dosage-dependent inhibition (lanes 5-10, [23][24][25], which is similar to that seen when using competitor identical to the probe. The flanking sequences of the different competitors of the PCE II probe are similar but not identical, and the results of the competition assays (lanes 5-10 and 23-25) demonstrate that they are probably not critical for the DNA-protein interactions. In addition, when wild-type short oligonucleotides (like M3-M6, only with the wild-type 14-bp core sequence) were used as probes, we observed the same prominent shift as that using the longer PCE II oligonucleotides as probes (data not shown). In fact, no changes in the flanking sequence, outside the 14-bp core, had any detectable effect on protein binding. Therefore, the protein-binding element is defined as CAGA-CAGGCTTATA. Because this sequence appears to be a novel protein binding site involved in photoreceptor-specific expression, we have named it PCE II.
Tissue specificity of PCE II was also tested using nuclear extracts isolated from Xenopus brain, muscle, liver, and heart (lanes 29 -34). The results show that, as before, there is a strong and specific shift with nuclear extracts from retinas. A similar complex is seen with nuclear extracts from brains, but it is much weaker than seen in retinal extracts. There also may be low levels of complex formation in extracts from other tissues, but again the levels are much weaker than seen in retinal extracts. These results confirm that the protein complex that binds to PCE II is enriched in retina.
Mutants of PCE II Failed to Target the GFP Reporter to the Photoreceptor Cells-A mutant PCE II was designed based on the results from the competition assays of EMSAs: CAGACA was mutated to TGAGTC (M3 and M4 in Fig. 8). These base changes should disrupt complex formation. Three repeats of mutant PCE II were subcloned upstream of the GFP reporter to generate the Mut-PCE-GFP construct and was used to produce transgenic tadpoles. Six individual PCR-positive tadpoles are examined, and results from three of them are shown in Fig. 9. One of the six tadpoles exhibited a very low level of GFP in all tissues (data not shown). Within the retina, GFP was seen in a few cells, but there was no evidence of photoreceptor specificity (Fig. 9, tadpole 2). The other five PCR-positive tadpoles did not show any GFP expression in any tissue, including photoreceptor cells. Note the autofluorescence in the inner and outer segment junctions (Fig. 9, tadpoles 1 and 3) as seen before in nontransgenic tadpoles (Fig. 3). Altogether, these results demonstrate that XNP(Ϫ77/Ϫ18bp) contains sufficient information to direct GFP expression to photoreceptor cells in vivo. DISCUSSION We have isolated and characterized a Xenopus nocturnin promoter element capable of directing correct spatial expression. Transgenic experiments showed that a small portion of the nocturnin promoter was capable of driving reporter gene expression specifically to photoreceptors, a pattern identical to endogenous nocturnin expression. This small piece of nocturnin promoter contains three perfectly repeated sequences, and EMSAs defined the protein binding site as CAGACAGGC-TATA. Photoreceptor specificity was abolished when mutations of PCE II were used to direct the GFP reporter in the transgenic experiments. Comparison of this sequence with elements that have been shown to regulate photoreceptor-specific expression of other genes revealed that this was a novel element, and we therefore designated it as PCE II.
Several promoter elements have been reported to be involved in the rod-specific (opsin and rod arrestin) or photoreceptorspecific (IRBP and nocturnin) control of transcription (Table I). These protein binding motifs were identified by EMSAs and DNase I footprinting assays from a number of different laboratories. Comparing these sequences with each other makes it obvious that even the binding motifs with the same name found from different genes or from different species are quite variable. However, PCE II, the novel element described here, has a unique sequence that is not similar to any of the elements described previously.
Despite the work done investigating photoreceptor-specific transcriptional regulation in a number of species, no clear general mechanism for photoreceptor-specific gene expression has yet emerged. It has been shown that Ret1 can direct gene expression in rod photoreceptors in transgenic rats (29). A 221-bp fragment of the mouse opsin promoter which contains Ret1 and BAT-1 also directs expression specifically to the rod photoreceptors of transgenic mice (4). A reporter gene driven by the 70-bp promoter of the murine IRBP which contains Ret1/ PCE I and CRX binding sites is active in cultures of retinal cells and brain cells (30). A portion of the rod arrestin promoter, including CRX and AP1 binding sites, directs expression of a reporter gene only in rod cells in the transgenic X. laevis (7). In this paper we showed that the XNP(Ϫ77/Ϫ18bp) containing three repeated PCE II is sufficient to target the GFP reporter to the photoreceptors in X. laevis.
In several cases the transcriptional factors that bind to the retina-specific elements have been characterized (Table I) (15,(31)(32)(33). The homeodomain protein RX has been shown to bind to the Ret1/PCE I site and activates the TATA-less arrestin and IRBP promoters (29,34,35). Because the nocturnin PCE II is novel, the photoreceptor protein(s) that bind to this sequence are unknown.
Several transcriptional factors are expressed in the developmental process in Xenopus which produce rod and cone photoreceptors, including genes that encode paired-type homeodomain proteins, such as PAX6, CHX10, RX, and NeuroD (36 -42). PAX6 and CHX10 are expressed in the retinal progenitors and maintained in inner nuclear layer cells, but their expression is excluded from developing photoreceptor cells in zebrafish, mouse, and X. laevis (36,38,43). RX is also expressed in retinal progenitors in Xenopus and subsequently down-regulated in all cells upon differentiation (44). NeuroD is a member of the basic helix-loop-helix family usually expressed during and/or after the terminal mitosis of neuronal precursors in rodent (45). Its expression is maintained in a subset of mature photoreceptors (46). The hierarchical pathway of these transcriptional factors has been examined using double in situ hybridizations on cross-sections of the CMZ in Xenopus (28). However, the targets of these transcriptional factors during development are largely unknown, and most of their known roles are limited to early development. Therefore, it seems unlikely that these proteins control the PCE II-mediated photoreceptor-specific expression pattern because nocturnin expression is a feature of fully differentiated photoreceptor cells.
Another transcription factor that is known to be present in both developing and mature photoreceptor cells is CRX (5,6). This otd/otx gene family encodes paired-like homeodomain proteins that are involved in the regulation of anterior head structure and sensory organ development (47). CRX expression is restricted to developing and mature photoreceptor cells (6). It has been shown that CRX binds the sequence TAATCA/A, resulting in transactivation of transcription. This element is found in the upstream sequence of several photoreceptor-specific genes (5, 6). The nocturnin 5Ј-flanking region has several sequences that are similar to CRX binding sites, but deletion of these elements does not alter spatial expression (Figs. 1A and 5). The Xenopus homolog of CRX has not been cloned. Therefore, the role of CRX in the regulation of nocturnin is unclear.
The transgenic technique in frogs is a powerful tool to investigate the spatial expression pattern and regulation of genes of interest (7,22,25,26). GFP has been commonly used as the reporter because of its stability and ease of detectability for in vivo studies. Compared with making transgenic mice, it is relatively easy and inexpensive to generate these animals, and several hundreds of transgenic embryos can be produced within 1 day. One limitation of the transgenic technique in Xenopus is the difficulty in controlling the levels of GFP because the copy numbers and positions of the construct are randomly determined in the individual tadpoles. Among those PCR-positive transgenic tadpoles in this study, there was variability in the levels of GFP expression (Figs. 5 and 6). This variability makes it difficult to make quantitative assessments of gene expression levels. Although we observed generally lower GFP signals from the XNP(Ϫ398/ϩ20bp)-GFP and XNP(Ϫ77/Ϫ18bp)-GFP constructs, this needs to be confirmed using other methods. These questions can be addressed by using a reporter gene such as luciferase, which is easier to quantitate than GFP, and by raising lines of the transgenic animals to provide a large number of tadpoles containing homogeneous copy numbers and integration sites (24). These more quantitative methods will also be important in the temporal analyses of nocturnin expression because it will be necessary to compare expression levels at different times of day.
Xenopus offers a unique advantage of combining the rapid and powerful transgenic approach with in vitro methods for the study of cis-elements in controlling cell-specific expression in vertebrates. Using these methods, we have demonstrated that the nocturnin promoter PCE II is sufficient to drive photore-ceptor-specific expression in Xenopus. The novel sequence of PCE II suggests that an as yet undescribed transcriptional mechanism is driving photoreceptor-specific expression of nocturnin.