Tandem Organization of Medaka Fish Soluble Guanylyl Cyclase a1 and b1 Subunit Genes IMPLICATIONS FOR COORDINATED TRANSCRIPTION OF TWO SUBUNIT GENES*

We determined the complete nucleotide sequences of the a1 subunit gene (OlGCS-a1) and the b1 subunit gene (OlGCS-b1) of medaka fish soluble guanylyl cyclase. In the genome, OlGCS-a1 and OlGCS-b1 are organized in tandem. The two genes are only 986 base pairs apart and span approximately 34 kilobase pairs in the order of OlGCS-a1 and OlGCS-b1. The nucleotide sequence of a large part of the 5*-upstream region of OlGCS-a1 is complimentarily conserved in that of OlGCS-b1. To analyze the promoter activity of each gene, a fusion gene construct in which the 5*-upstream region was fused with the green fluorescent protein gene was injected into medaka fish 2-cell embryos. When the fusion gene containing the OlGCS-a1 upstream region was injected, green fluorescent protein fluorescence was detected in the embryonic brain. The 5*-upstream region of OlGCS-b1 alone was insufficient for the reporter gene expression in the embryos. When the OlGCS-a1 upstream region was located upstream of the OlGCS-b1green fluorescence protein fusion gene, the reporter gene was expressed in the brain and trunk region of the embryos. These results suggest that the 5*-upstream region of OlGCS-a1 can affect the expression of OlGCS-b1. It is therefore possible that the expression of OlGCS-a1 and OlGCS-b1 is coordinated.

The nitric oxide (NO) 1 /cGMP signaling pathway plays a critical role in various physiological phenomena (1,2). The second messenger, cGMP, is synthesized by guanylyl cyclase (GC), which is present in two forms, a membrane form and a soluble form (3). The soluble form of GC is a heme-containing heterodimer composed of ␣ and ␤ subunits (4) and is activated by NO and CO (1). Both subunits possess a catalytic domain in the C-terminal position; the primary structure of which is conserved in the membrane form of GC and adenylyl cyclase (5). However, it is known that each subunit of soluble GC has no catalytic activity by itself (3). To date, two isoforms for each subunit of soluble GC have been reported in mammals (6). GCS-␣ 1 and GCS-␤ 1 cDNA have been characterized in bovines, rats, and humans (5,(7)(8)(9)(10), and GCS-␣ 2 and GCS-␤ 2 have been isolated by homology screening from a human fetal brain cDNA library and a rat kidney cDNA library, respectively (11,12). Coexpression of the ␣ 1 subunit with the ␤ 1 subunit in COS cells generates an active heterodimer (5,13,14), whereas coexpression of the ␣ 2 subunit with the ␤ 1 subunit generates a less active enzyme (11). Recently, it has been demonstrated that the ␤ 2 subunit inhibits activation of the ␣ 1 /␤ 1 heterodimer by NO (15).
In vasodilation, soluble GC activated by NO derived from the endothelium induces relaxation of vascular smooth muscle through cGMP-dependent protein kinase I (1,16). As a neuronal messenger, NO affects synaptic plasticity via generation of cGMP in the hippocampus and olfactory bulb in mammals (17)(18)(19). Soluble GC localized in the inner segments of photoreceptor cells is activated by NO (20) and modulates synapses between cone and horizontal cells (21). The NO/cGMP signaling pathway is also expected to participate in synaptogenesis (22) and synaptic suppression in a neuromuscular junction (23).
Despite the rapid accumulation of information concerning the functional importance of the NO/cGMP signaling pathway, there have been only a few reports related to the regulation of the soluble GC gene expression. It has been demonstrated that cAMP causes decreases in the level of mRNA for ␤ 1 subunits of soluble GC in the rat fetal lung fibroblast (24) and for both subunits of soluble GC in the rat aortic smooth muscle cell (25).
The ␣ 1 and ␤ 1 subunit genes of soluble GC are colocalized in human and rat chromosomes (26,27). This and the fact that the coexpression of both subunits is essential for enzyme activity imply that the expression of both genes is coordinated. However, there has been no report on the genomic structure and transcriptional regulation of soluble GC. Recently, we have isolated cDNA clones encoding the ␣ 1 subunit (OlGCS-␣ 1 ) and ␤ 1 subunit (OlGCS-␤ 1 ) of soluble GC of the medaka fish, Oryzias latipes (28). As a first step in investigating whether the expression of both genes is coordinated, we determined the genomic structure of OlGCS-␣ 1 and OlGCS-␤ 1 and report here that OlGCS-␣ 1 and OlGCS-␤ 1 are organized in tandem in the medaka fish genome. We analyzed promoter function of these genes by introducing promoter-green fluorescent protein (GFP) fusion constructs into embryos. The results suggest that the 5Ј-upstream region of OlGCS-␣ 1 controls expression of both OlGCS-␣ 1 and OlGCS-␤ 1 genes.

EXPERIMENTAL PROCEDURES
Animals and Embryos-Mature adults of the orange-red variety of medaka O. latipes were purchased from a dealer and kept in indoor tanks as described previously (29). Naturally spawned and fertilized eggs were collected and cultured in distilled water containing 0.00006% methylene blue at 27°C. The developmental stage is expressed in days, and the day of fertilization is referred to as Day 0. Hatching usually occurred at Day 10.
Isolation of Genomic Clones for OlGCS-␣ 1 and OlGCS-␤ 1 -A genomic library of medaka fish (white strain) constructed in the Lambda Fix II vector was purchased from Stratagene and used for the isolation of genomic clones for OlGCS-␣ 1 and OlGCS-␤ 1 . The cDNA fragments (nucleotide positions 1290 to 2346 for OlGCS-␣ 1 cDNA and nucleotide positions 241 to 1229 and 1833 to 2620 for OlGCS-␤ 1 cDNA) amplified by polymerase chain reaction (PCR) were labeled with digoxigenin-dUTP using the digoxigenin-High Prime (Roche Molecular Biochemicals) and used for screening as probes. Finally, four positive clones were obtained from approximately 1.35 ϫ 10 6 recombinant phages. Phage DNA was purified by using the QIAGEN Lambda kit (QIAGEN), and the insert DNA was subcloned into pBluescript II KS(-) (Stratagene). The sequence of the insert DNA was determined by the dideoxynucleotide chain termination procedure (30) with an Applied Biosystems 373A or PRISM 377 DNA sequencer and analyzed on DNASIS software  (Hitachi Software Engineering Co.). 3Ј-Rapid Amplification of cDNA Ends (3Ј-RACE)-Total RNA was extracted from Day 9 embryos according to the acid guanidinium thiocyanate-phenol-chloroform extraction method (31). The first strand synthesis and first PCR were performed using the 3Ј-Full RACE Core Set (Takara Shuzo Co., Ltd.). The specific primers for OlGCS-␣ 1 used for amplification were as follows: LF-1 for first PCR, 5Ј-GTGCAACTACT-TGTATGTTTC-3Ј (identical to nucleotides 2267-2287); LF-2 for second PCR, 5Ј-TTATTGATGTCTGACAGCCTA-3Ј (identical to nucleotides 2304 -2324); and LF-4 for second PCR, 5Ј-GTGTGGGTTGTGGATA-AAACT-3Ј (identical to nucleotides 2522-2542). Second PCR was performed using a 1/50 volume of the first PCR products as a template. The second PCR products were subcloned into pBluescript II KS(-) (Stratagene), and the sequence of the insert DNA was determined as described above.
Primer Extension Analysis-Total RNA was extracted from the adult medaka fish brain as described above. Poly(A) ϩ RNA was isolated using Oligotex-dT30 Super (Roche) according to the manufacturer's protocol. The oligonucleotides used for the primer extension experiments were as follows: PE-L1 for OlGCS-␣ 1 , 5Ј-AAGACAGATGCGCTCGAG-3Ј (complementary to nucleotides 97-114) and PE-S3 for OlGCS-␤ 1 , 5Ј-ATGCT-GAGATTGTCGGTGTT-3Ј (complementary to nucleotides 2-21). The oligonucleotides were hybridized with 2 g of the brain poly(A) ϩ RNA and extended by 200U SUPERSCRIPT II Reverse Transcriptase (Life Technologies, Inc.) in 45 mM Tris-HCl (pH 8.3) containing 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, 0.5 mM dNTP, 1.85 MBq of [␣-32 P]dCTP at 42°C for 1 h. The sequence reaction was carried out by using Sequenase version 2.0 for the labeled dCTP kit (USB) and [␣-32 P]dCTP. Primer-extended products were treated with RNaseA and then separated on a 7 M urea, 6% polyacrylamide gel with sequence reaction products. The radioactive signals were analyzed using a FUJIX Bio-Imaging Analyzer BAS2000 (Fuji Photo Film).
Southern Blot Hybridization-Genomic DNA was isolated from the adult medaka brain as described (28). The genomic DNA (10 g) was digested overnight with BamHI, EcoRV, and HindIII (Takara Shuzo Co., Ltd.) and then separated on a 0.7% agarose gel. The DNA was transferred to a nylon membrane (Hybond-N, Amersham Pharmacia Biotech) using 0.4 M NaOH as the transferring solution and hybridized with a 32 P-labeled cDNA probe amplified by PCR (nucleotide positions 1290 to 1518 for OlGCS-␣ 1 cDNA) as described previously (28).
The PCR product containing the 5Ј-upstream regions of OlGCS-␣ 1 was subcloned into the pEBFP vector (CLONTECH), and the resultant fusion gene construct was named AB-1. A DNA fragment containing the 5Ј-upstream regions of OlGCS-␣ 1 and the blue fluorescent protein (BFP)-coding region was excised from AB-1 by digesting with HindIII and StuI and cloned into the corresponding site of BG-1. The resultant fusion gene construct was named ABBG-1. To construct a fusion gene that lacked a part of the 5Ј-upstream region of OlGCS-␤ 1 containing the putative TATA box, PCR was performed with oligonucleotide primers 5Ј-CCGACAATCTCAGCATCTGT-3Ј (identical to nucleotides ϩ156 to ϩ175 of OlGCS-␤ 1 ) and 5Ј-GTTCACAGACCACAGTCGAT-3Ј (complimentary to nucleotides Ϫ213 to Ϫ232 of OlGCS-␤ 1 ). Both ends of the PCR products were blunted by T4 polymerase and then self-ligated. The resultant fusion gene construct was named ABBG-2.
Microinjection of Fusion Gene Constructs and Detection of GFP-To prevent hardening of the chorion after fertilization, medaka fish fertilized eggs were incubated in 20 mM CAPS-NaOH (pH 10.5) containing 2 mM glutathione (Wako Pure Chemicals Co., Ltd.) at 4°C for several hours (32). Micropipettes were made by the horizontal puller (PN-3, Narishige) from a siliconized and sterilized 1 ϫ 90-mm fiber-filled glass capillary tube (GD-1, Narishige). The DNA solution (10 ng/l) was injected into the cytoplasm of both blastomeres of a medaka fish 2-cell stage embryo using a micropipette. The injected embryo was cultured as described above. GFP fluorescence was detected under a fluorescence microscope (IX70, Olympus).

Isolation and Characterization of Medaka Fish OlGCS-␣ 1
and OlGCS-␤ 1 Genes-A medaka fish genomic library was screened with cDNA fragments of OlGCS-␣ 1 or OlGCS-␤ 1 . After repeated screening, four different clones (A2, A11, B5, B16) were obtained. Restriction enzyme mapping of these clones demonstrated that they overlap each other (Fig. 1). The nucleotide sequence of the insert DNA of each clone was determined and compared with that of the respective cDNA. As shown in Fig. 1 and Table I, OlGCS-␣ 1 consists of 9 exons and OlGCS-␤ 1 consists of 13 exons. The GT-AG rule was conserved for all splice sites except exon/intron 8 in OlGCS-␣ 1 (Table I). In the medaka fish genome, OlGCS-␣ 1 and OlGCS-␤ 1 are 986 bp apart and organized in tandem. The two genes span approximately 34 kilobase pairs in the order of OlGCS-␣ 1 and OlGCS-␤ 1 .
In a previous study, the nucleotide sequence of a part of the 3Ј-noncoding region of the cDNA for OlGCS-␣ 1 , which should contain a polyadenylation signal sequence, has not been determined (28). To determine the nucleotide sequence of the 3Ј-end of OlGCS-␣ 1 , 3Ј-RACE was performed using total RNA from the Day 9 embryos. The 3Ј-RACE product with LF-2 primer contained the nucleotide sequence corresponding to that of intron 8 in OlGCS-␣ 1 . The 3Ј-RACE with LF-4 primer, which was designed to cross the insert site of the intron 8, produced a 344-bp cDNA fragment. Each 3Ј-RACE product contained a putative polyadenylation signal sequence, AATAAA, 24 -18 bp upstream of poly(A).
Analysis of the 5Ј-Upstream Regions of OlGCS-␣ 1 and Ol-GCS-␤ 1 -To determine the transcription initiation sites of Ol-GCS-␣ 1 and OlGCS-␤ 1 , primer extension experiments were performed using medaka fish brain poly(A) ϩ RNA (2 g). Using a specific primer for OlGCS-␣ 1 , one major and two minor bands were detected. The nucleotide corresponding to the major band, which is located most upstream of the three bands, was assigned to the transcription initiation site for OlGCS-␣ 1 (Fig. 2). In the same way, the nucleotide corresponding to one detected band was assigned to the transcription initiation site for Ol-GCS-␤ 1 . A putative TATA box, TATAGAA, is present 30 -25 bp upstream of the transcription initiation site of OlGCS-␤ 1 (Fig.  3). There is no TATA box in the corresponding region of OlGCS-␣ 1 , although a TATA box consensus sequence, TATAATA, is present at 141-136 bp upstream of the transcription initiation site for OlGCS-␣ 1 (Fig. 2). A number of E-boxes are present in the 5Ј-upstream regions of each gene (Fig. 4). Other known cis-regulatory elements, AP1, MEF-2, GATA, Sp1, CREB, and C/EBP binding sequences, are found in the 5Ј-upstream region of OlGCS-␣ 1 . On the other hand, there is one GATA, AP1, and two C/EBP binding sequences in the 5Ј-upstream region of OlGCS-␤ 1 .
Genomic Southern Hybridization-In a previous study we performed a Southern hybridization analysis using medaka fish genomic DNA to examine whether OlGCS-␣ 1 and Ol-GCS-␤ 1 exist as a single copy (28). However, the results were not clear because the cDNA fragments used as probes crossed over many introns. Therefore, in this study we performed a genomic Southern analysis using a different probe containing a single exon (exon 5 of OlGCS-␣ 1 ). As shown in Fig. 6, only one major band was detected in each of the three lanes. The size of the band in each lane is consistent with that of the DNA fragments obtained from the digestion of genomic clones by the respective restriction enzymes. These results suggest that Ol-GCS-␣ 1 is a single copy gene.
Promoter Analysis of OlGCS-␣ 1 and OlGCS-␤ 1 -To examine whether the 5Ј-upstream region of OlGCS-␣ 1 and the intervening region between OlGCS-␣ 1 and OlGCS-␤ 1 have promoter activity, a fusion gene containing the 5Ј-upstream region of each gene and the GFP gene was constructed and named AG-1 or BG-1, respectively (Fig. 4). The fusion gene was injected into the cytoplasm of both blastomeres of medaka fish 2-cell stage embryos. At 7 days after fertilization, the number of live embryos and embryos with GFP fluorescence was counted (Table  II). GFP fluorescence was detected in the brain and somite of the Day 4 embryos injected with AG-1 (Fig. 7A). At 7 days after fertilization more than half (57.1%) of the injected embryos were alive, and GFP fluorescence in these embryos, which tended to increase as development proceeds, was detected in 14.8% of the live embryos (Fig. 7, E and F). In the embryos injected with BG-1, 74.3% of the embryos were alive at 7 days after fertilization. But no GFP fluorescence was detected in any injected embryos at this stage (Fig. 7C), suggesting that the intervening region between OlGCS-␣ 1 and OlGCS-␤ 1 alone is not sufficient for detectable promoter activity.
Roles of the 5Ј-upstream region of OlGCS-␣ 1 and the intervening region between the two subunit genes in OlGCS-␤ 1 expression were investigated by using a GFP-fusion construct (ABBG-1) in which the OlGCS-␣ 1 upstream region connected to a BFP-coding sequence were inserted upstream of the intervening region of BG-1 (Fig. 4). GFP fluorescence was detected in the brain and trunk region of embryos injected with ABBG-1 (Table II). On the other hand, the reporter gene expression was not observed when a short upstream region (from Ϫ226 to ϩ154) of OlGCS-␤ 1 containing a putative TATA box and a transcription initiation site was removed from ABBG-1 (Table  II, ABBG-2). These results suggest that the 5Ј-upstream region of OlGCS-␣ 1 can activate transcription of OlGCS-␤ 1 in conjunction with the intervening region between OlGCS-␣ 1 and OlGCS-␤ 1 .
Although ABBG-1 and ABBG-2 contain the BFP gene driven by the 5Ј-upstream region of OlGCS-␣ 1 , BFP fluorescence was not observed in embryos injected with these constructs (data not shown). This might be because of much weaker fluorescence intensity of BFP than that of the enhanced variant of GFP used in this study. DISCUSSION In this study we demonstrated that the 5Ј-upstream region of OlGCS-␣ 1 is essential for the expression of OlGCS-␣ 1 . On the other hand, the intervening sequence between OlGCS-␣ 1 and OlGCS-␤ 1 seemed not to be sufficient for OlGCS-␤ 1 expression. Considering the tandem organization of OlGCS-␣ 1 and OlGCS-␤ 1 , both genes might be cotranscribed as a single polycistronic mRNA as in the mouse and human upstream of the GDF gene (UOG-1) and the growth/differentiation factor-1 gene (GDF-1) (33). Alternatively, other regions such as the 5Ј-upstream region and/or intron of OlGCS-␣ 1 would affect the transcription of OlGCS-␤ 1 . Our results support the latter possibility. First, GFP fluorescence was observed in embryos injected with the ABBG-2 construct in which the 5Ј-upstream region of Ol-GCS-␣ 1 was located upstream of the OlGCS-␣ 1 /OlGCS-␤ 1 intervening sequence followed by the GFP gene. Second, there is a transcription initiation site for OlGCS-␤ 1 and a TATA box consensus sequence, TATAGAA, 30 -25 bp upstream of the transcription initiation site for OlGCS-␤ 1 . The reporter gene expression was abolished when the TATA box and the transcription initiation site were removed from ABBG-1. Therefore, basal transcription factors probably act on the intervening region between OlGCS-␣ 1 and OlGCS-␤ 1 , and an enhancer in the OlGCS-␣ 1 upstream region can affect the promoter activity. The above could make it possible to temporally and spatially coordinate the transcription of OlGCS-␣ 1 and OlGCS-␤ 1 during  the embryogenesis of the medaka fish. Sequence analysis revealed that a number of E-boxes, which are known to play a critical role in nerve and muscle differentiation (34), are present in the 5Ј-upstream region of each gene. This is consistent with our detection of GFP fluorescence in the brain of Day 4 embryos injected with the 5Ј-upstream region of the OlGCS-␣ 1 -GFP construct. Scholz et al. (35) have demonstrated that NO synthase and NO-sensitive guanylyl cyclase are broadly distributed in the central nervous system of lobsters at hatching. The participation of the NO/cGMP signaling pathway in synaptogenesis has also been reported (22). Detection of GFP fluorescence in the brain of Day 4 embryos suggests a relation between soluble GC and neuronal development during the embryogenesis of the medaka fish. The ratio of embryos with GFP fluorescence to live embryos 7 days after fertilization was relatively low (14.8%). Although a higher concentration of DNA solution was able to increase the number of embryos with GFP fluorescence, it tended to cause morphological abnormalities and/or death (data not shown).
It has previously been demonstrated that the mRNA level of each subunit gene is decreased by cAMP (24,25). In this regard, it should be noted that a cAMP-response element is present in the 5Ј-upstream region of OlGCS-␣ 1 . This element may participate in the regulation of transcription with a cAMPresponse element modulator, which is known to inhibit transcription by binding to the cAMP-response element (36). There is one GATA and one MEF-2 binding sequence in the 5Ј-upstream region of OlGCS-␣ 1 , and it has been suggested that these sequences may participate in the differentiation of vascular smooth muscle cells (37)(38)(39). Considering that soluble GC induces the relaxation of vascular smooth muscle in mammals (1), these elements may regulate the expression of both genes in the medaka fish. In addition, others including Sp1, C/EBP, and AP1 in the 5Ј-upstream region of OlGCS-␣ 1 and OlGCS-␤ 1 may also be involved in the regulation of expression of both genes for soluble GC subunits in the medaka fish (40).
The nucleotide sequences of the 5Ј-upstream region of Ol-GCS-␣ 1 are highly conserved in relation to that of the intervening region between OlGCS-␣ 1 and OlGCS-␤ 1 in six different regions. Some of these regions have cis-regulatory elements in common with each other, suggesting that the highly conserved regions also participate in coordinated transcription.
Adenylyl cyclase, which synthesizes another second messenger, cAMP, also has two catalytic domains punctuated by a membrane-spanning domain (41). The primary structure of each catalytic domain of adenylyl cyclase is conserved in those of the soluble and membrane forms of GCs (5). It has been demonstrated that adenylyl cyclase as well as soluble GC require the two catalytic domains for cyclase activity (41) and that changes in a couple of the amino acids in the catalytic domain of adenylyl cyclase cause a functional change in adenylyl cyclase from cAMP production to cGMP production (42). This suggests that soluble GC is evolutionarily related to adenylyl cyclase, although their forms are quite different. In this study, we determined the complete structure of OlGCS-␣ 1 and OlGCS-␤ 1 , demonstrating that the two genes are tandemly organized like a single gene. A comparison of the genomic structure between soluble GC and adenylyl cyclase may clarify the evolutionary relationship between both enzymes, although the genomic structure of the latter enzyme has not yet been reported.
The soluble form of GC is present as a heterodimer, and the coexpression of both subunits is required for generating enzyme activity (13,14). It has been reported that two functionally related genes such as collagen IV ␣1 and ␣2 chain genes are coordinately expressed by a bidirectional promoter (43). In this regard, our results presented here suggest the possibility of temporally and spatially coordinated transcription of both subunit genes for soluble GC during embryogenesis.