α1 Soluble Guanylyl Cyclase (sGC) Splice Forms as Potential Regulators of Human sGC Activity*

Soluble guanylyl cyclase (sGC), a key protein in the NO/cGMP signaling pathway, is an obligatory heterodimeric protein composed of one α- and one β-subunit. The α1/β1 sGC heterodimer is the predominant form expressed in various tissues and is regarded as the major isoform mediating NO-dependent effects such as vasodilation. We have identified three new α1 sGC protein variants generated by alternative splicing. The 363 residue N1-α1 sGC splice variant contains the regulatory domain, but lacks the catalytic domain. The shorter N2-α1 sGC maintains 126 N-terminal residues and gains an additional 17 unique residues. The C-α1 sGC variant lacks 240 N-terminal amino acids, but maintains a part of the regulatory domain and the entire catalytic domain. Q-PCR of N1-α1, N2-α1 sGC mRNA levels together with RT-PCR analysis for C-α1 sGC demonstrated that the expression of the α1 sGC splice forms vary in different human tissues indicative of tissue-specific regulation. Functional analysis of the N1-α1 sGC demonstrated that this protein has a dominant-negative effect on the activity of sGC when coexpressed with the α1/β1 heterodimer. The C-α1 sGC variant heterodimerizes with the β1 subunit and produces a fully functional NO- and BAY41-2272-sensitive enzyme. We also found that despite identical susceptibility to inhibition by ODQ, intracellular levels of the 54-kDa C-α1 band did not change in response to ODQ treatments, while the level of 83 kDa α1 band was significantly affected by ODQ. These studies suggest that modulation of the level and diversity of splice forms may represent novel mechanisms modulating the function of sGC in different human tissues.

Since the studies of the late 1970s and early 1980s, which underlined the obligatory role of the endothelium in mediating acetylcholine-induced vasodilatation, nitric oxide (NO) has been recognized as an endogenous nitrovasodilator that medi-ates the local regulation of basal arterial tone (1)(2)(3)(4). Many of the physiological functions of NO in the cardiovascular, neuronal, gastrointestinal, and other systems are mediated through its primary receptor, soluble guanylyl cyclase (sGC). 3 The hemecontaining sGC heterodimer converts guanosine triphosphate into the secondary messenger guanosine 3Ј:5Ј-cyclic monophosphate (cGMP). The sGC activity increases more than 200fold in response to NO (5,6). High concentrations of cGMP produced by activated sGC modulate functions of numerous enzymes, such as cyclic nucleotide phosphodiesterases, cGMP-gated ion channels, and cGMP-dependent protein kinases (PGKs). Recently, the vital importance of sGC for mammalian physiology was directly confirmed by generation of sGC knock-out mice (7)(8)(9). The absence of sGC protein resulted in a significant increase in blood pressure, complete loss of NO-dependent aortic relaxation, and inhibition of platelet aggregation in knock-out animals, which died prematurely at the age of 4 weeks because of severe gastrointestinal disorders (7).
Four sGC isoforms, products of four genes, have been identified so far: ␣ 1 , ␣ 2 , ␤ 1 , and ␤ 2 . Only ␣ 1 /␤ 1 and ␣ 2 /␤ 1 heterodimers are activated by NO (10). The ␣ 1 /␤ 1 sGC is the most abundant isoform and is distributed ubiquitously in mammalian tissues with the highest levels of mRNA in brain, lung, heart, kidney, spleen, and muscle (11). Vascular smooth muscle and endothelial cells express predominantly ␣ 1 -and ␤ 1 -subunits (12). The functional importance of ␣ 1 /␤ 1 sGC was demonstrated by the significantly decreased relaxing effects of major vasodilators (acetylcholine, NO, YC-1, and BAY41-2272) in the ␣ 1 sGC knock-out mice of both genders (9). sGC function is affected not only by NO, but also by regulation of the expression of sGC subunits at transcriptional and post-transcriptional levels. The steady state mRNA levels of ␣ 1and ␤ 1 -subunits decrease with hypertension, aging and vary during embryonic development (13). The expression of sGC subunits is regulated by estrogen (14), cAMP-elevating compounds (15,16), cytokines (NGF, LPS, IL-1␤) (17) and NO donors (18). Subcellular localization of sGC, and its activity can also be affected in proliferating tissue (19) by protein interactions and phosphorylation (13). In mammals, the alternative splicing for the ␣ 2 -subunit generates a dominant-negative variant (20). Splice forms for ␤ 1 -and ␤ 2 -subunits have been also demonstrated (21)(22)(23). Recently, a shortened ␣ 1 sGC transcript, which lacks the predicted translation site in exon 4, has been found, and its expression was correlated with lower sGC activity in several cell lines (24). However, splice variants of ␣ 1 sGC have not been described previously.
Here we report the isolation and characterization of three new ␣ 1 sGC splice forms encoding N-and C-terminal-truncated proteins. We demonstrate that the N-terminal-truncated C-␣ 1 splice form heterodimerizes with the ␤ 1 sGC subunit to create an active NO-sensitive enzyme both in Sf9 and human neuroblastoma BE2 cells. Moreover, this splice variant is more resistant to ODQ-induced protein degradation than the wildtype sGC. N1-␣ 1 sGC splice form lacking the C-terminal catalytic domain has a dominant-negative effect when co-expressed with ␣ 1 /␤ 1 sGC in Sf9 or BE2 cells. The functional role of N2-␣ 1 sGC splice variant is yet to be determined. Q-PCR and semiquantitative RT-PCR analyses of different human tissues demonstrate tissue-specific expression of the identified splice forms. Together our data suggest that alternative splicing of the ␣ 1 sGC subunit may be a novel mechanism that regulates sGC function and activation in some human tissues.
Primers, RNA, and RT-PCR-All primers were custom synthesized by Integrated DNA Technologies (Coralville, IA). To subclone N1-␣ 1 sGC we used the upstream primer PR1, 5Ј-318 CAACACCATGTTCTGCACGAAGC-3Ј and the downstream primer PR2, 5Ј-1411 GCTTTCATATTCAAGATAG-TATTATG-3Ј (numbering according to sequence with Gen-Bank TM no. CR618242). To subclone the N2-␣ 1 sGC, we used the upstream primer PR3, 5Ј-192 CAACACCATGTTCTGCA-CGAAGC-3Ј and the downstream primer PR4, 5Ј-611 GTATC-ACTCTCTTTGTGTAATCC-3Ј (GenBank TM no. BC012627). To detect the deletion of exon 4 in C-␣ 1 sGC we used the upstream primer PR5, 5Ј-120 GCTAGAGATCCGGAAGC-ACA-3Ј and the downstream primer 5Ј-317 TTGCAAATACT-CTCTGCCAAA-3Ј (GenBank TM no. AK226125). To detect the deletion of in the exon 7 of C*-␣ 1 sGC we used the upstream primer PR7, 5Ј-561 GAA CGG CTG AAT GTT GCA CTT GAG-3Ј and the downstream primer PR8, 5Ј-922 GTA GGG CTG ATT CAC AAA CTC G-3Ј (GenBank TM no. BX649180). Total RNA from BE2 cells was isolated using the RiboPure kit (Ambion, TX). The panel of total RNA from human tissues was purchased from Ambion (FirstChoice Human Total RNA Survey Panel, Lot 08608142). 5 g of total RNA was used for the RT reactions, which was performed with a mixture of oligo(dT) and Random Hexamer primers using SuperScript III RT (Invitrogen) according to the manufacturer's protocol. PCR reactions with PfuUltra DNA polymerase (Stratagene) were performed for 35 cycles at the T a of 55°C. PCR products were separated on an agarose gel, purified using the QIAEX II gel extraction kit (Qiagen), and sequenced using PCR primers. All sequencing was performed by the Nucleic Acid Core Facility at the Medical School of University of Texas in Houston.
Real-Time Quantitative RT-PCR-Real-time quantitative RT-PCR (RT-qPCR) was performed utilizing the 7700 or 7900 Sequence Detector instrument (Applied Biosystems, Foster City, CA) (25,26). Specific quantitative assays for ␣ 1 sGC, N1-␣ 1 sGC, and N2-␣ 1 sGC were developed using the Primer Express software version 1.0 for Macintosh (Applied Biosystems), Beacon Designer (Premier Biosoft), or RealTimeDesign (Biosearch Technologies) based on sequences from Gen-Bank TM . The assays are listed in supplemental Table S3. cDNA was synthesized in 10 l (96-well plate) or 5 l (348-well plate) total volume by the addition of 6 l or 3 l/well of RT master mix consisting of: 400 nM assay-specific reverse primer, 500 M deoxynucleotides, Superscript II buffer, and 10 units of Superscript II reverse transcriptase (Invitrogen) to a 96-(ISC Bioexpress, Kaysville, UT) or 384-well plate (Applied Biosystems) and followed by a 4-l or 2-l volume of sample (25 ng/l), respectively. Each sample was determined in triplicate, plus a control without reverse transcriptase to access DNA contamination levels. Each plate also contained an assay-specific sDNA (synthetic amplicon oligo) standard spanning a 5-log template concentration range and a no template control. Each plate was covered with Biofilm A (Bio-Rad) and incubated in a PTC-100 (96) or DYAD (384) thermocycler (Bio-Rad) for 30 min at 50°C, followed by 72°C for 10 min. Subsequently, 40 l or 20 l of a PCR master mix (400 nM forward and reverse primers (IDT, Coralville, IA), 100 nM fluorogenic probe (Biosearch Technologies, Novato, CA), 5 mM MgCl 2 , 200 M deoxynucleotides, PCR buffer, 150 nM SuperROX dye (Biosearch Technologies, Novato, CA), and 1.25 units of Taq polymerase (Invitrogen) were added directly to each well of the cDNA plate. RT master mixes and all RNA samples were pipetted by a Tecan Genesis RSP 100 robotic work station (Tecan US, Research Triangle Park, NC); PCR master mixes were pipetted utilizing a Biomek 2000 robotic work station (Beckman, Fullerton, CA). Each assembled plate was then covered with optically clear film (Applied Biosystems) and run in the 7700 or 7900 real-time instrument using the following cycling conditions: 95°C, 1 min; followed by 40 cycles of 95°C, 12 s, and 60°C, 30 s. The resulting data were analyzed using SDS 1.9.1 (7700) or SDS 2.3 (7900) software (Applied Biosystems) with ROX as the reference dye.
Synthetic DNA oligos used as standards (sDNA) encompassed the entire 5Ј-3Ј amplicon for the assay (Invitrogen). Each oligo standard was diluted in 100 ng/l yeast or Escherichia coli tRNA-H 2 O (Invitrogen or Roche Applied Sciences) and span a 5-log range in 10-fold decrements starting at 0.8 pg/reaction. It has been shown for several assays that in vitro transcribed RNA amplicon standards (sRNA) and sDNA standards have the same PCR efficiency when the reactions are performed as described above. 4 Because of the inherent inaccuracies in quantifying total RNA by absorbance, the amount of RNA added to an RT-PCR 4 G. L. Shipley, personal communication.
from each sample was more accurately determined by measuring a housekeeping transcript level in each sample. The final data were normalized to 36B4 (500-fold dilution of sample in tRNA-H 2 O).
Cell Culture-BE2 human neuroblastoma cell line (American Type Culture Collection) was cultured in 1:1 mixture of Dulbecco's modified Eagle's medium/F12K media supplemented with 10% fetal bovine serum, 0.1 mM MEM nonessential amino acids, penicillin-streptomycin mixture (50 units/ml and 50 g/ml), 10 mM Hepes (pH 7.4), 1 mM sodium pyruvate, 2 mM L-glutamine (all from Invitrogen), and maintained at 37°C and 5% CO 2 . For in vivo ODQ treatments 80% confluent neuroblastoma cell cultures were treated with 20 M ODQ for up to 24 h. To prepare lysates, the cells were collected by trypsinolysis, washed twice with phosphate-buffered saline, resuspended in 40 mM TEA (pH 7.4) containing protease inhibitor mixture (Roche Applied Science), and disrupted by sonication. The lysates were centrifuged at 15,000 ϫ g for 30 min to prepare the cleared supernatant fractions, which were used for Western blotting, immunoprecipitation, or activity measurements.
Generation of BE2 Stable Transfectant Lines-Coding sequences of N1-, N2-, and C-type ␣ 1 sGC variants obtained from BE2 total RNA as described above were first cloned into pCR-Blunt vector (Invitrogen). For N1-and N2-type ␣ 1 forms, the coding sequence of the FLAG peptide was inserted by PCR in front of the stop codon and recloned into pCR-Blunt. The coding regions of N1-, N2-, and C-type ␣ 1 sGC variants were then isolated by restriction with NsiI/XbaI enzymes and subcloned into PstI/XbaI sites under the control of the CMV promoter of the mammalian expression vector pMGH2 (Invitrogen). These pMG-C␣F, pMG-N1␣F, and pMG-N2␣F plasmids were transfected into BE2 cells by the Lipofectamine reagent (Invitrogen) according to the manufacturer's protocol. 48-h post-transfection, the cells were plated on 96-well plates at 2500 cells/ml density and selected by 380 g/ml hygromycin (Sigma). Two weeks later, individual hygromycin-resistant colonies were collected and expanded on 100-mm 2 tissue culture dishes. The clones stably transfected by N1-or N2-type ␣ 1 sGC were identified by anti-FLAG Western blotting of lysates prepared from the hygromycin-resistant cultures. The clones stably expressing C-type ␣ 1 sGC were identified using anti-␣ 1 sGC Western blotting.
Co-immunoprecipitation-BE2, BE-C␣F, and BE-N1␣F cells collected from confluent 10-cm culture dishes, were washed twice with phosphate-buffered saline, resuspended in 500 l of phosphate-buffered saline containing protease inhibitor mixture, disrupted by sonication, spun down at 15,000 rpm for 30 min at 4°C, and supernatants were collected. Then polyclonal anti-␤ 1 -sGC antibodies or prewashed anti-FLAG M2 affinity resin (Sigma) were added and tumbled for 1.5 h or overnight, respectively, at 4°C. The lysate mixture with anti-␤ 1 -sGC antibodies was then combined with 100 l of prewashed protein A-agarose beads (Upstate) and further incubated for 1.5 h. The protein A-agarose beads were washed three times with 40 mM TEA, 200 mM NaCl, 1% Nonidet P-40, pH 7.4, and bound proteins eluted by boiling in 100 l of Laemmli buffer. The anti-FLAG M2 affinity resin was washed three times with ice-cold TBS buffer, and bound proteins were eluted with high salt buffer or with buffer containing the FLAG peptide, according to the manufacturer's instructions. The Western blot was probed for the ␣ 1 -and ␤ 1 -subunits of sGC using polyclonal anti-␣ 1 -sGC and monoclonal anti-␤ 1 -sGC antibodies.
Expression in Sf9 Cells-To express the ␣ 1 splice forms in Sf9 cells, the coding regions of N1, N2, or C-␣ 1 sGC were subcloned into the pVL1392 transfer vector. The baculoviruses producing the ␣ 1 splice forms were generated by recombination with BaculoGold DNA using the manufacturer's protocol (BD Biosciences). To obtain sGC enzyme containing the splice form ␣ 1 -subunits, Sf9 cells at 1.8 ϫ 10 6 cells/ml were infected with the baculoviruses expressing full-length ␤ 1 sGC subunit and the corresponding splice form at a multiplicity of infection of 2.
Assay of sGC Activity-Soluble guanylyl cyclase activity in lysates of BE2 or Sf9 cells were assayed by formation of [ 32 P]cGMP from [␣-32 P]GTP at 37°C as described previously (27). The concentration of DMSO used as a vehicle for BAY41-2272 did not exceed 0.1% and alone had no effect on sGC activity.
Statistical Analysis-All data are presented as means Ϯ S.E. or S.D. Statistical comparisons between groups were performed by Student's t test. Nonlinear regression and calculations of EC 50 and IC 50 were performed using Graph Pad Prism 3.0 software (GraphPad Software).

Identification of ␣ 1 sGC Alternative Splice Variants in the NCBI Data
Base-A previous report demonstrated the existence of alternative splicing for the ␣ 1 sGC in human tissues (24). We used the ␣ 1 sGC cDNA (GenBank TM accession number Y15723) to screen a human RNA data base available on the Human Genome web page (NCBI) for additional splice variants of sGC. This in silico analysis identified twelve unique ␣ 1 sGC cDNA sequences cloned from different human tissues (supplemental Table S1). Their comparison with human genome sequences revealed that they are all generated by alternative splicing from the ␣ 1 sGC gene.
Seven of these sequences encode the full size ␣ 1 sGC protein, while five of them encode truncated ␣ 1 proteins (supplemental Table S1). With alternative splicing, one of the identified ␣ 1 sGC mRNAs (GenBank TM accession numbers CR618242, CR614534) lost the non-coding exon 2 and the coding exons 8 through 10, but acquired additional 131 bp at the end of exon 7. This mRNA encodes a protein that maintained 363 N-terminal amino acids of ␣ 1 sGC, but lost the catalytic domain because of a splice-generated frameshift (Fig. 1, A and B). Three new splice-specific amino acid residues and a premature stop codon were acquired (supplemental Table S2). We named this splice variant N1-␣ 1 sGC (Fig. 1).
In N2-␣ 1 sGC mRNA (GenBank TM accession number BC012627), splicing eliminates exons 7 and 8 and introduces an additional 54 base pairs and a premature stop codon in exon 9. The N2-␣ 1 protein retains only the first 126 residues of the ␣1 sequence, but acquires an additional 17 amino acids at the C terminus (Fig. 1, A and B and supplemental Table S2).
The third identified ␣ 1 sGC splice variant, termed C-␣ 1 sGC, lost 240 N-terminal amino acids, but maintained part of the regulatory and the complete catalytic domains (Fig. 1B). Interestingly, the same C-␣ 1 protein is encoded by two differentially spliced species of mRNA (Fig. 1A). In one sequence (Gen-Bank TM accession number AK226125, C-␣ 1 in Fig. 1A), the alternative splice acceptor in intron 3 generates a 179-bp deletion in exon 4, eliminating the translation start site. The open reading frame (ORF) starts at the alternative methionine located in exon 7. In another splice variant (GenBank TM accession number BX649180, C*-␣ 1 in Fig. 1A), the alternative acceptor site results in a 140-bp deletion and a premature stop codon in exon 7. The alternative start codon in exon 7 restores the ORF. Thus, both C-␣ 1 andC*-␣ 1 splice mRNA species encode the same C-␣ 1 sGC protein.
Alternative Splicing of ␣ 1 sGC in Human BE2 Neuroblastoma-We have previously demonstrated that the human BE2 neuroblastoma cell line expresses high levels of functional sGC enzyme (28). Using the information on the structure of ␣ 1 spliced mRNA, we designed pairs of primers to amplify the fragments specific for N1-␣ 1 , N2-␣ 1 , C-␣ 1 , and C*-␣ 1 splice forms (Fig. 1A). As depicted in supplemental Fig. S1, RT-PCR analysis showed that BE2 cells express all found ␣ 1 splice mRNA except C*-␣ 1 . Each of the splice-specific fragments was purified, and the identity of the splice form confirmed by sequencing.

Expression of ␣ 1 sGC Alternative Splice Variants in Different
Human Tissues-We next investigated the tissue-specific expression of the ␣ 1 sGC splice variants. We designed a Q-PCR assay for N1-and N2-␣ 1 sGC splice forms based on their unique sequences inserted by alternative splicing (supplemental Table  S3). We quantified the abundance of the full-length and spliced mRNA in various human tissues. As expected, full-length ␣ 1 sGC mRNA was detected in RNA of all tested tissues. N1-␣ 1 sGC was observed at detectable levels in all human organs, except bladder, testis, thyroid, placenta, and skeletal muscle (Fig. 2). N2-␣ 1 sGC was present in all tissues, but at significantly lower levels than ␣ 1 sGC.
The absence of sequences specific only for C-␣ 1 or C*-␣ 1 mRNA and the insignificant difference in size from the fulllength sGC transcript (supplemental Table S1) precluded us from using Q-PCR or Northern blotting to estimate their levels and tissue distribution. Therefore, we used a semi-quantitative RT-PCR method using primers designed to detect the deletions specific for both C-and C*-␣ 1 sGC RNA species. The identity of amplified fragments was confirmed by sequencing. The full size ␣ 1 sGC transcript was detected in all tissues together with C-␣ 1 sGC, albeit at different ratios (supplemental Fig. S2). However, C*-␣ 1 sGC mRNA was not detected in esophagus, heart, kidney, liver, lung, bladder, brain, cervix, and colon (supplemental Fig. S3).
Effect of Expression of ␣ 1 sGC Alternative Splice Variants on sGC Activity-Next we tested whether ␣ 1 sGC splice variants possess any catalytic activity or whether they affect the function of full-length ␣ 1 /␤ 1 sGC in BE2 cells. We selected several stable clones expressing N1-␣ 1 or N2-␣ 1 sGC, which were tagged with the FLAG epitope at the C-terminal end. The stable clone expressing 41-kDa N1-␣ 1 sGC protein was identified using anti-FLAG antibodies (BE2-N1␣F in Fig. 3A). We measured the rate of cGMP production in the lysates of these cells in response to the NO donor, DEA-NO, or DEA-NO in the presence of the heme-dependent allosteric regulator BAY41-2272. As demon-  MAY Fig. 3B, the BE2-N1␣F clone showed a significant decrease of sGC-dependent cGMP synthesis in comparison with parental BE2 cells. This lower sGC activity cannot be attributed to a decreased expression of sGC, because Western blotting showed no significant decrease in the levels of endogenous full-length ␣ 1 -and ␤ 1 -subunits (Fig. 3B). The lower cGMP synthesis was observed in all tested BE-N1␣F clones (results not shown), suggesting that the N1-␣ 1 sGC splice form acts as an inhibitor of sGC function. This conclusion was confirmed when the N1-␣ 1 splice form was co-expressed together with the full-length ␣ 1 -and ␤ 1 -subunits in Sf9 cells. The addition of increasing amounts of virus producing N1-␣ 1 sGC protein correlated with the decrease of DEA-NO, BAY41-2272-, and DEA-NO/BAY41-dependent ␣ 1 /␤ 1 sGC activity (supplemental Fig. S4). The lysates from stable clones overexpressing N2-␣ 1 sGC did not show significant changes of the NO and/or BAY41-2272-dependent cGMP synthesis. We did not observe significant changes of the NO/BAY41-dependent cGMP pro-duction in lysates of the lines overexpressing the N2-␣ 1 sGC splice form (results not shown).

Novel Splice Forms of Soluble Guanylyl Cyclase
We also selected BE2 stable lines expressing the C-␣ 1 sGC splice variant tagged with the FLAG epitope. BE2-C␣F line was selected using antibodies raised against a sequence at the C-terminal end of the human ␣ 1 sGC subunit, which recognizes both the short C-type ␣ 1 sGC (54 kDa) and the full-length (83 kDa) ␣ 1 -subunit (Fig. 3A). We found that the DEA-NO-or DEA-NO/BAY41-2272-stimulated cGMP synthesis in the lysates of the BE2-C␣F clone was not significantly different from parental BE2 cells. On the other hand, the amount of ␣ 1 -sGC protein in BE2-C␣F was lower than in BE2 cells (Fig. 3A). It appears that the expression of C-␣ 1 variant can compensate for decreased levels of full size ␣ 1 -sGC subunit through a formation of active heterodimer.
Interaction of the C-␣ 1 Splice Variant with the ␤ 1 -Subunit-To confirm the interaction of the C-␣ 1 sGC with the ␤ 1 sGC subunit we tested for co-immunoprecipitation of the C-␣ 1 sGC with the ␤ 1 sGC subunit from the BE2-C␣F clone. As shown in Fig. 5A, polyclonal antibodies raised against the C terminus of the ␤ 1 -subunit precipitated both ␣ 1 and C-␣ 1 variants. This immunoprecipitation correlates with the depletion of the C-␣ 1 and ␣ 1 signals from the post-immunoprecipitate lysates (Fig.  5A). Moreover, the similar intensity of the co-precipitated signals suggests that both ␣ 1 and C-␣ 1 heterodimerize with ␤ 1 FIGURE 2. Expression of ␣ 1 sGC, N1-, and N2-␣ 1 sGC mRNA in human tissues. Expression was quantified by qPCR analysis and normalized to the levels of ribosomal housekeeping gene 36b4. Human total RNA Survey Panel (Ambion) containing mRNA from 20 different tissues pooled from three healthy donors was used in this comparison. equally well. Alternatively, immunoprecipitation of C-␣ 1 sGC with anti-FLAG M2 affinity gel precipitated the ␤ 1 sGC subunit from the lysate of BE2-C␣F stable clones, but not from BE2 lysates (Fig. 5A). The amount of ␤ 1 sGC was higher in the eluates treated with the FLAG peptide, consistent with a FLAGspecific elution.
Despite a clear inhibitory effect of the N1-␣ 1 splice form ( Fig.  3B and supplemental Fig. S4), we were not able to observe the co-precipitation of the N1-␣ 1 sGC with the ␤ 1 -subunit (Fig. S5).
Level of C-␣ 1 Splice Form Is Not Affected by ODQ Treatment-A recent report indicated that heme-deficient sGC or sGC with an oxidized heme prosthetic group is more prone to degradation, which may contribute to endothelial dysfunction (29). In those studies, treatment of intact cells with the sGC inhibitor ODQ decreased the level of sGC protein because of ubiquitin-dependent degradation. Thus, we tested whether the active C-␣ 1 /␤ 1 heterodimer shows a similar response to ODQ. Indeed, as demonstrated in Fig. 6A, the amount of full-length ␣ 1 band decreased after BE2-C␣F cells were treated for 24 h with 20 M ODQ. The C-␣ 1 splice band, however, was not affected by the same treatment. A time course study showed that the intensity of the ␣ 1 band decreased shortly after ODQ administration in all tested cells, while the level of C-␣ 1 even slightly increased with time (Fig.  6B). These data suggest that the C-␣ 1 subunit protein is more stable to intracellular processes occurring after the oxidation of sGC heme.

DISCUSSION
Alternative splicing frequently occurs in eukaryotic genes and provides an important mechanism for tissue-specific and developmental regulation of gene expression. About 15% of mammalian gene mutations associated with pathological conditions affect RNA splicing signals (30).
A number of previous reports demonstrate that alternative splicing is an essential mechanism regulating the function of several members of the cGMP signaling pathway. For example, a splice-dependent deletion of 90 -100 N-terminal residues of cGMP-dependent protein kinase I (PKGI) produces two splice variant isoforms, I␣ and I␤, which have different tissue distribution and target specificity (31). Recent studies also demonstrated that PKGI splice isoforms respond differently to activation by hydrogen peroxide (32). Guanylyl cyclase B (GC-B) is another example with two truncated splice forms. It was proposed that these splice forms regulate the function of the full-length subunit and show different tissue distributions (33).
Several reports suggested that splicing may also be a method of sGC regulation. For example, the ␣ 2i sGC splice variant, carrying an insertion in the catalytic domain was detected in sev- Arrows indicate the protein band of appropriate size in Sf9 cells and the BE2-N1F clone. B and C, identification of stable lines expressing C-␣ 1 variant by Western blotting with anti-␣ 1 sGC antibody and protein levels of endogenous ␣ 1 and ␤ 1 sGC in stable lines and parental BE2 cells. Arrows indicate the protein bands corresponding to ␣ 1 and C-␣ 1 sGC in BE2-C␣F. B, representative Western blot; C, densitometry analysis of ␣ 1 and ␤ 1 levels normalized to ␤-actin and expressed as percent of expression in control BE2 cells taken as 100%. The value for BE-N1␣ and BE-C␣F clones are presented as mean Ϯ S.D. calculated from three independent lysates. D, rate of cGMP production in lysates of BE2 cells, BE2-N1␣F and BE2-C␣F in response to DEA-NO (200 M) alone or in combination with BAY41-2272 (2 M). Data from six (for BE2 cells) or three (for BE2-N1␣F and BE2-C␣F clones) independent lysates measured in triplicates are presented as mean Ϯ S.E. #, difference with BE2 cells is statistically significant (p Ͻ 0.05); *, difference is not significant (p Ͼ 0.05).
eral human tissues (20). The ␣ 2i sGC splice variant has a dominant-negative function. In addition, the expression of two mRNA species of human ␣ 1 sGC with deletions in the exon 4 correlates with decreased sGC activity in immortalized B-lymphocyte cell lines (24). These ␣ 1 sGC splice forms, however, were never isolated or characterized.
In this report, we characterized several newly identified splice forms of the ␣ 1 -subunit of human soluble guanylyl cyclase. Analyzing the abundant data available in the NCBI data base, we identified a large number of individual ␣ 1 sGC transcripts. It is notable that the diversity of uncovered ␣ 1 sGC transcripts (12 cDNAs, see supplemental Table S1) is much larger than for the ␤ 1 sGC (5 cDNAs). This diversity does not appear to be unique for humans. Analysis of the mouse genome data base also identified the N1-␣ 1 splice variant (GenBank TM number AK031305). Conservation of this particular splice form points to a functional importance of the N1-␣1 variant in mammals. Most of these transcripts are produced by sequence rearrangements in 5Ј-and 3Ј-untranslated regions, which are usually associated with altered post-transcriptional regulation of gene expression. For example, binding of the RNA-stabilizing protein HuR to the 3Ј-UTR of ␣ 1 sGC plays an important role in post-transcriptional regulation of sGC expression in response to cyclic nucleotides and in aging in rats (15,34,35). Thus, the diversity of the 5Ј-and 3Ј-untranslated region sequences most likely reflects multiple selective mechanism(s) regulating the stability of ␣ 1 sGC transcripts in response to various extracellular and intracellular stimuli in different cell types or at different developmental stages.
We concentrated our studies on four RNA splice variants which encode truncated ␣ 1 sGC, with deletions at the N or C termini of the protein (supplemental Table S1 and Fig. 1). We found that human neuroblastoma BE2 cells express three (N1-␣ 1 , N2-␣ 1 , C-␣ 1 ) out of four alternatively spliced RNAs (supplemental Fig. S1). Quantitative PCR analysis of the N1-␣ 1 and N2-␣ 1 splice mRNAs demonstrates that the majority of human tissues express more than one splice variant, but the levels and the relative ratio of the ␣ 1 and the N1-␣ 1 and N2-␣ 1 splice The rates of cGMP production in Sf9 cell lysates expressing C-␣ 1 /␤ 1 or ␣ 1 /␤ 1 sGC in response to various concentrations of DEA-NO was determined as described under "Materials and Methods." B, C-␣ 1 /␤ 1 sGC and ␣ 1 ␤ 1 proteins show similar responses to BAY41-2272. The rates of cGMP production in Sf9 cell lysates expressing C-␣ 1 /␤ 1 or ␣ 1 /␤ 1 sGC in response to various concentrations of BAY41-2272 were determined as described under "Materials and Methods." C, C-␣ 1 /␤ 1 sGC and ␣ 1 ␤ 1 proteins have the same sensitivity to ODQ. Sf9 lysates with ␣ 1 /␤ 1 or C-␣ 1 /␤ 1 heterodimers were mixed with different amounts of ODQ, and the rate of cGMP production induced by 50 M DEA-NO was measured. Data from four independent experiments are presented as mean Ϯ S.D. forms is tissue-specific ( Fig. 2 and supplemental Table S4). For example, N2-␣ 1 is found in all tested tissues, while N1-␣ 1 sGC is not detected in esophagus, bladder, testes, thyroid, placenta, and skeletal muscle (Fig. 2). Similarly, C-␣ 1 sGC is detectable in all tested human tissues, although the relative ratio was tissuespecific (supplemental Fig. S2), while C*-␣ 1 sGC was detected only in some tissues (supplemental Fig. S3). Taken together, these data indicate that the expression of ␣ 1 sGC splice forms is independently regulated. It should be noted, though, that the levels of N1-, N2-, C-, and C*-␣ 1 splice mRNAs are on average lower than the full size ␣ 1 sGC. Interestingly, the semi-quantitative PCR analysis suggests that in human adipose tissue the level C-␣ 1 and C*-␣ 1 mRNA may be collectively comparable with the level of ␣ 1 sGC mRNA. Because the comparative analysis reported here is based on RNAs extracted from entire organs, it might not represent the true abundance of individual spliced sGC transcript at the cellular level. Cellular and subcellular localizations of individual ␣ 1 sGC splice variants in human tissues remain to be determined. Localization may significantly alter the functions of these splice forms beyond the properties reported here for BE2 neuroblastoma cells.
In this report, we identified and characterized N1-␣ 1 and N2-␣ 1 sGC splice proteins lacking the catalytic domain (Fig. 1). Although by themselves or in combination with the ␤ 1 -subunit these truncated splice ␣ 1 proteins do not have any catalytic activity (data not shown), one of them displays dominant-negative properties. BE2 cells overexpressing N1-␣ 1 sGC have a significantly decreased NO-and NO/BAY42-2272-induced cGMP production despite the unchanged level of endogenous ␣ 1 /␤ 1 sGC (Fig. 3). Moreover, when co-expressed in Sf9 cells with ␣ 1 /␤ 1 heterodimer, N1-␣ 1 reduced NO-and BAY41-2272dependent sGC activity in direct correlation with the expression level of the N1-␣ 1 protein (supplemental Fig. S4). We were not able to detect any direct heterodimerization of the N1-␣ 1 protein with the ␤ 1 -subunit, even when both proteins were coexpressed in large quantities in Sf9 cells (supplemental Fig. S5). This suggests that the dominant-negative effect of N1-␣ 1 is not due to a direct competition with the full-length ␣ 1 for the binding with ␤ 1 -subunit, but is rather indirect. Recent deletion mapping showed that the segment spanning residues 363-372 of the ␣ 1 -subunit is important for the formation of the ␣ 1 /␤ 1 heterodimer (36). Because the N1-␣ 1 splice protein constitutes the first 363 residues of the ␣ 1 -subunit, perhaps, it interferes with the heterodimerization function of the adjacent 363-372 segment of the full-length ␣ 1 -subunit.
Direct regulation of guanylyl cyclase by dominant-negative splice variants has been proposed before. For example, the ␣ 2i sGC splice variant blocks the formation of a functional ␣ 1 ␤ 1 sGC heterodimer (20), while the truncated GC-B splice variants hinder the formation of active full size GC-B homodimers (33). Considering the wide distribution of the N-␣ 1 splice form in different human tissues ( Fig. 2 and supplemental Table S1), modulation of N1-␣ 1 sGC protein expression may also be a regulatory mechanism controlling the amount of active sGC heterodimer.
In contrast to N1-␣ 1 , N2-␣ 1 sGC had no effect on sGC activity in cell lysates when overexpressed in BE2 cells or co-expressed with ␣ 1 /␤ 1 in Sf9 cells. However, it cannot be excluded that this splice variant may serve another function besides directly affecting sGC activity. Our studies also demonstrate that the C-␣ 1 sGC isoform clearly forms a fully functional sGC heterodimer with ␤ 1 sGC subunit. The activity of the recombinant C-␣ 1 /␤ 1 enzyme expressed in Sf9 cells was indistinguishable from the ␣ 1 /␤ 1 sGC, at least in regard to the degree of activation by the NO donor and allosteric activator BAY41-2272 and inhibition by ODQ (Fig. 4). The preservation of the sGC activity in the BE-C␣F stable line (Fig. 3) and co-immunoprecipitation of C-␣ 1 sGC and ␤ 1 sGC subunits (Fig. 5) all support the conclusion that C-␣ 1 -and ␣ 1 -subunits are interchangeable. Also, these results suggest that the lack of 240 N-terminal amino acid residues by the ␣ 1 sGC subunit does not affect the heterodimerization and enzyme activity. This observation does not support previous findings that the region between residues 61 and 128 of ␣ 1 is mandatory for heterodimerization (37). The properties of this naturally occurring splice variant, however, are in agreement with earlier reports, which show that a significant portion of the N-terminal region of the ␣ 1 sGC subunit can be deleted without affecting NO sensitivity and heterodimerization (36,38,39).
The existence of ␣ 1 -positive bands other than the 83-kDa full-length ␣ 1 sGC subunit was mentioned in several previous reports. While screening selected human tissues with anti-␣ 1 antibodies, Zabel et al. (40) observed in cortex, cerebellum, and lungs a band similar in size to the ϳ54 kDa C-␣ 1 described in this report. Interestingly, the antibodies used were raised against the same epitope as in current studies. Moreover, additional bands were also detected in crude lysates from human amygdale (41). Presence of several bands with electrophoretic mobility similar to C-␣ 1 indicates that either C-␣ 1 form may undergo additional tissue-specific processing and/or modifications, or that additional, yet unidentified, ␣ 1 sGC splice forms exist.
Recent studies suggested that some conditions of endothelial dysfunction may be associated with the accumulation of oxidized and heme-free sGC leading to a poor response to NO (29). The same report showed that in ODQ-treated cells sGC is subjected to ubiquitination and subsequent degradation. Although our current studies showed that C-␣ 1 /␤ 1 and ␣ 1 /␤ 1 heterodimers have identical sensitivity to ODQ with a similar IC 50 (Fig. 4), C-␣ 1 and ␣ 1 sGC subunits show an opposite response to ODQ-induced degradation (Fig. 5). Western blotting confirms that the 83-kDa ␣ 1 sGC band disappears after exposure to ODQ, while the level of C-␣ 1 protein does not decrease. These data suggest that C-␣ 1 lacks the structural cues contributing to decreased levels of ␣ 1 after ODQ treatment. The functional studies presented here prove that the C-␣ 1 /␤ 1 heterodimer can fully compensate for sGC activity. In light of reported previously decreased levels of ␣ 1 -subunit in aged or diseased vessels (41)(42)(43)(44), the expression of a more stable C-␣ 1 may be a specific protective adaptation to these conditions. Thus, it is possible that the positive vasodilatory effects of BAY58-2667 in diseased blood vessels observed previously (29) are mediated by the more stable C-␣ 1 /␤ 1 heterodimer. Future studies will show whether the C-␣ 1 splice form may be, under certain circumstances, the main or the sole type of ␣ 1 -subunit.
Although the functional studies suggest that the N-terminal fragment missing in the C-␣ 1 -subunit is not important for the activity of sGC, this region is preserved in evolution in the ␣ 1 -subunits of all vertebrates. It is possible that this region, often referred to as the regulatory region (37,38), is responsible for the integration of NO/cGMP signaling with other regulatory pathways. In this case, despite similar catalytic properties, ␣ 1 /␤ 1 and C-␣ 1/ ␤ 1 heterodimers may be modulated differently by NO-independent mechanisms, e.g. protein modifications or protein-protein interactions.
cGMP-dependent kinase is one of the major effectors of cGMP generated by sGC in response to NO. Increased cGMP levels in neuronal cells leads to a PKGI-dependent phosphorylation of the Splicing Factor 1 (SF1), which functions at early stages of pre-mRNA splicing and splice site recognition (45). Thus, it is appealing that sGC-dependent cGMP production may regulate sGC activity by affecting the spliceosome assembly and/or switching between different splice sites during pre-mRNA processing. This, in turn, may affect the diversity of expressed ␣ 1 sGC splice proteins modulating the function of the sGC heterodimer.
In summary, our present study identifies and characterizes several new splice variants of the human ␣ 1 -subunit of sGC. The splicing of ␣ 1 mRNA seems to be ubiquitous and follows the domain organization of the ␣ 1 sGC subunit. Our findings point to new mechanisms of modulation of sGC function and activity that may have significant effects on nitric oxide and cyclic GMP signaling and perhaps pathophysiology.