Regulation of the Guanylyl Cyclase-B Receptor by Alternative Splicing*

Guanylyl cyclase-B (GC-B) is a single transmembrane receptor that binds C-type natriuretic peptide (CNP). The ligand/receptor appears critical in the regulation of cell proliferation and differentiation where it acts as an adversary of mitogenic signaling pathways. We have isolated three guanylyl cyclase-B isoforms generated from a single gene by alternative splicing and termed them GC-B1, GC-B2, and GC-B3. GC-B1 is full-length and re-sponds maximally to CNP, GC-B2 contains a 25-amino acid deletion in the protein kinase homology domain, and GC-B3 only retains a part of the extracellular li-gand-binding domain. GC-B2 binds CNP, but the ligand fails to activate the cyclase, while GC-B3 fails to bind ligand. When GC-B2 or GC-B3 is expressed coincident with GC-B1, they act as dominant negative isoforms by virtue of blocking formation of active GC-B1 ho-modimers. Relative expression levels of GC-B1, GC-B2, and GC-B3 vary across tissues and as a function of in vitro culture; the relative amount of GC-B2 to GC-B1 is repressed in cultured smooth muscle cells relative to endogenous ratios in the medial layer cells of the aorta. Thus, GC-B isoform levels can be independently regulated. Given that the splice

The guanylyl cyclase (GC) 1 family of enzymes synthesize cGMP from Mg⅐GTP (1). Two subfamilies of the enzyme are found in mammals: single transmembrane GCs and soluble GCs (2). Single transmembrane GCs (seven in mammals labeled GC-A through GC-G) exist as homodimers, are located on the plasma membrane, and share a common domain structure: an extracellular ligand-binding domain (ECD), a single trans-membrane segment (TM), a protein kinase homology domain (KHD), and a cyclase catalytic (CYC) domain (2). The KHD is homologous with the catalytic domain of protein kinases, and although no protein kinase activity has been reported, the KHD acts as a critical regulatory domain (3). The single transmembrane GCs are receptors for extracellular signaling molecules, although ligands have been found for only three. GC-A is a receptor for atrial natriuretic peptide and brain natriuretic peptide, GC-B is a receptor for C-type natriuretic peptide (CNP), and GC-C is a receptor for heat-stable enterotoxins and endogenous peptides of the guanylin family (4 -11). The soluble GC family exists as heterodimers that bind heme, the mediator of nitric oxide stimulation (2,12).
CNP/GC-B has been suggested as a strong adversary of various mitogens in that elevations of intracellular cGMP induced by CNP block proliferation of cell types such as fibroblasts and myofibroblastic hepatic stellate cells and the brainderived neuronal growth factor-or nerve growth factor-induced proliferation of neuronal precursors (13)(14)(15). Many of these effects seem to be mediated by inhibition of the mitogen-activated protein kinase cascade. CNP may also inhibit proliferation through the suppression of arginine vasopressin-induced increases in intracellular Ca 2ϩ levels (16). Additionally CNP at high concentrations inhibits progression of angioplasty-induced balloon injuries (17) and induces cultured aortic smooth muscle cells to change from the synthetic to the contractile phenotype (18). CNP/GC-B also stimulates the differentiation of proliferating chondrocytes into hypertrophic chondrocytes through the activation of GC-B signaling, which is critical for endochondral ossification and longitudinal bone growth (19 -21). On the other side of the adversarial relationship, arginine vasopressin, platelet-derived growth factor, basic fibroblast growth factor, serum, phorbol 12-myristate 13-acetate, and lysophosphatidic acid rapidly and effectively desensitize GC-B to CNP in fibroblasts and other cells (13,16,22,23).
In this report, we isolated and characterized the murine GC-B (Npr2) gene and found that three GC-B isoforms (GC-B1, GC-B2, and GC-B3) are generated from a single Npr2 gene and that GC-B2 and GC-B3 function as dominant negative isoforms.
Animals-Mouse F1 hybrids between 129/SvEvTac and C57BL/6N strains were obtained from Taconic and used for tissue collection to isolate genomic DNA, RNA, protein, and cultured aortic smooth muscle cells (SMCs). All experiments with the mice were conducted as approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center at Dallas.
Isolation and Characterization of Murine Guanylyl Cyclase-B Gene-Approximately 1 ϫ 10 6 clones of a 129/SvEvTac mouse genomic library (Stratagene) were screened with a rat GC-B cDNA probe fragment (EcoRI-XbaI, 2.0 kb) (5). Phage plaques were transferred to Hybond-N as recommended by the supplier and were then used for screening. The probe was labeled with [␣-32 P]dCTP by the random prime method (RediPrime II random prime labeling kit, Amersham Biosciences) to give a specific activity of about 1 ϫ 10 9 dpm/g of DNA. Hybridization was with a Rapid-hyb hybridization solution supplemented with 0.1 g/ml salmon sperm DNA (sonicated) at 65°C for 16 h. The membranes were washed once in 2ϫ SSC, 0.1% SDS at ambient temperature for 10 min, once in 2ϫ SSC, 0.1% SDS at 65°C for 30 min, and twice in 0.1ϫ SSC, 0.1% SDS at 65°C for 30 min. The membranes were then exposed to Hyperfilm at ambient temperature for 24 h. Five positive clones were isolated, and three clones (nos. 7, 11, and 12) (Fig. 1a) were characterized. Fragments of the phage clones were subcloned into pBluescript II KS(ϩ) (Stratagene) and characterized by restriction enzyme mapping, Southern blot analyses, and DNA sequencing. Exon-intron and intronexon boundaries were determined by a comparison of nucleotide sequences of cDNA and the genomic DNA. The copy number was determined by Southern blot analysis of genomic DNA isolated from murine tails. DNA was digested with KpnI, SpeI, SacI, or XbaI; electrophoresed on a 0.8% agarose, Tris-acetate-EDTA gel; transferred to Hybond-Nϩ; and then probed with the 0.3-kb cDNA fragment encompassing exons 4 -7 (from the KpnI site to the location of primer RT-AS, Fig. 1a). The probe labeling, hybridization, and washing were as described for the library screening. The membranes were then exposed to Hyperfilm at Ϫ80°C with intensifying screens for 12 h.
RNA Isolation-RNA was extracted from various organs and primary cultures of aortic SMCs with TRIzol reagent as recommended by the supplier. Pituitary glands, adrenal glands, aortas, and ovaries were pooled (n ϭ 3-5) prior to RNA isolation.
Reverse Transcription and Polymerase Chain Reaction-Five micrograms each of total RNA was reverse-transcribed by a Superscript II reverse transcriptase with a random primer (Invitrogen) following the supplier's instruction. PCR was with a recombinant Taq DNA polymerase (Invitrogen) using 2 l each of the 20-l reverse transcription (RT) reactions as the template unless otherwise specified. To isolate cDNA fragments containing full-length coding sequences, PCR of 40 cycles was accomplished with the cerebellar RT reaction with an Ex-Taq DNA polymerase (Takara) and primers listed in Table I (first set). Amplified cDNA fragments were subcloned into a pGEM-Teasy TA cloning vector (Promega) and used for further analyses. The expression of GC-B1 and GC-B2 mRNA and the expression of GC-B1 and GC-B3 mRNA were detected by PCR with primers listed in Table I (second and  third sets). Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) mRNA was detected by PCR using 1 l of the RT reaction and a mouse glyceraldehyde-3-phosphate dehydrogenase Control Amplimer Set (Clontech) as internal control. The cycle number in each PCR was determined as amplification is in the exponential phase: 35 cycles for GC-B1 and GC-B2, 27 cycles for GC-B1 and GC-B3, and 25 cycles for Gapdh. Transcripts of smooth muscle myosin heavy chain isoforms (SM1 and SM2) (24) were detected by PCR with the primers listed in Table I (fourth set); the cycle number of the PCR was 35. The PCR products were electrophoresed on 1.2% agarose, Tris-acetate-EDTA gels, detected by ethidium bromide staining, and quantified by a Multi Image Light Cabinet (Alpha Innotech). For some experiments, the gels were transferred to Hybond-Nϩ and detected by 32 P-labeled oligonucleotide probes located between the primers without containing the primer sequences: 5Ј-TGCAGTTTGGCAACTCGGATCGC-3Ј for GC-B1 and GC-B2, 5Ј-GCAGAGAAGCAGATTTGGTGGACAG-3Ј for GC-B1 and GC-B3, and 5Ј-GCCTTGACTGTGCCGTTGAATTTGCCGTGA-3Ј for Gapdh; the hybridization and wash were at 42°C. The 0.7-kb cDNA fragment corresponding to the C-terminal half of the ECD and the TM of GC-B, which was used as the probe for Northern blot analyses, was amplified for 40 cycles with the primers for GC-B1 and GC-B3 (Table I) and subcloned in a pGEM-Teasy vector.
Northern Blot Analysis-Total RNA (20 g unless otherwise specified) was denatured by the formamide method; separated on a 1.0% agarose, MOPS gel; and transferred to Hybond-N. The 0.7-kb murine GC-B cDNA probe was labeled, and the hybridization and wash were as described for the library screening. The membranes were then exposed to Hyperfilm at Ϫ80°C with intensifying screens for 48 h.
Expression Vector Construction-Complementary DNA clones of each murine GC-B isoform (GC-B1, GC-B2, and GC-B3) were distinguished by restriction enzyme mapping and DNA sequencing. The 3.4-kb insert was isolated from pGEM/mGC-B3 after EcoRI digestion and ligated into the EcoRI site of a pCMV5 expression vector, generated from a pCMV1 vector through deletion of the HpaI-EcoRI fragment from the SV40 origin region (25), to make pCMV5/mGC-B3. About 2.2-kb SmaI-EcoRI fragments of the inserts of pGEM/mGC-B1 and pGEM/mGC-B2, in which three GC-B isoforms are different from each other, were ligated between the SmaI and EcoRI sites of pBluescript II SK(ϩ) vectors (Stratagene). The SmaI-EcoRV fragments of resultant plasmids were used to replace the 2.4-kb SmaI-EcoRV fragment of the pCMV5/mGC-B3, which generated pCMV5/mGC-B1 and -B2; the EcoRV site existed outside the EcoRI cloning sites.
Aortic SMCs were isolated from the thoracic and abdominal portion of the aorta of a male mouse at 5 weeks of age as described previously (26). At passage 3, the cells were fixed in 4% paraformaldehyde, 0.1 M phosphate buffer (pH 7.4) and permeabilized with Dulbecco's phosphate-buffered saline (Ca 2ϩ -and Mg 2ϩ -free) containing 0.5% Triton X-100 and 1% normal horse serum. Immunocytochemistry was with the anti-␣-smooth muscle actin mouse monoclonal antibody (1:400) and anti-von Willebrand factor rabbit polyclonal antibody (1:200) using a Vector M.O.M. immunodetection kit (Vector Laboratories) and a VEC-TASTAIN Elite ABC kit (Vector Laboratories), respectively. Greater than 95% of the cells were positive for ␣-smooth muscle actin, and no cells positive for von Willebrand factor could be detected, indicating that the cells were predominantly SMCs with little contamination by endothelial cells.
Estimation of Guanylyl Cyclase Activity-COS-7 cells transfected with GC-B isoform expression vectors were incubated with CNP at various concentrations or vehicle, and intracellular cGMP concentrations were estimated as described previously (13).
Brains and lungs were isolated from mice at 5 weeks of age and homogenized with a Teflon homogenizer in 9 volumes (g wet weight/ml) of homogenization buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM Na 2 EDTA, 1 mM Na 3 VO 4 , 50 mM NaF, 10 g/ml leupeptin, 10 g/ml pepstatin, 10 g/ml aprotinin, 10% glycerol). The membrane fraction was collected as a pellet after centrifugation at 100,000 ϫ g for 30 min at 4°C. The pellet was suspended in the homogenization buffer of the original volume, and the protein concentration was estimated by the BCA method (BCA Protein Assay kit, Pierce). GC activities in the membrane fraction of 40 -80 g of protein were estimated as described previously (13).
Binding Assay-[Tyr 0 ]CNP was iodinated by the chloramine-T method to give a specific activity of 800 Ci/mmol. COS-7 cells transfected with GC-B isoform expression vectors were prepared in 12-well plates as described above. The binding assay was with 125 I-[Tyr 0 ]CNP at 0.1-1 nM with or without 0.5 M CNP to define nonspecific binding (27).
For immunoprecipitation, the cells were solubilized in 150 l/plate solubilization buffer (150 mM NaCl, 20 mM HEPES, pH 7.4, 1% Triton X-100, 0.1% SDS) by passing the solution 20 times through a 25-guage needle followed by incubation on ice for 1 h. The solution was then cleared by centrifugation at 16,000 ϫ g for 30 min at 4°C. The super-natant fluid of 400 l was then combined with 40 l of prewashed, agarose-conjugated anti-FLAG antibody (M2, Sigma, 1:1 suspension in the solubilization buffer) in the presence or absence of a FLAG peptide, added to block binding to the anti-FLAG antibody. The immunoprecipitation was for 1 h at 4°C. The resultant pellet was washed three times in the solubilization buffer and eluted in the 2ϫ sample buffer by incubation for 5 min at 60°C. The eluate was cleared by centrifugation at 16,000 ϫ g for 1 min and incubated with 100 mM dithiothreitol at 60°C for 5 min. The samples were then subjected to immunoblot analyses as described above.
Statistical Analyses-Statistical analyses of data were performed by GraphPad Prism Version 3.00 for Windows (GraphPad Software). Data of GC activities were analyzed by two-way analysis of variance with Bonferroni's posttest, and dose-response curves were drawn by a nonlinear regression analysis of the sigmoid dose-response type.

Structure of Npr2
Gene-The murine GC-B (Npr2) gene is greater than 20 kb and consists of 22 exons and 21 introns (Fig.  1a). All exon-intron and intron-exon boundaries follow the GT and AG rule (Table II) (28), and only one gene exists based on Southern blot analysis (Fig. 1b). The gene structure is well conserved between murine and human (29) in that genes from both species consist of the same number of exons with introns located at the same positions (Table III). Both species also have a long second intron of about 6.5 kb (Table III). The organization of the closely related GC-A receptor gene of mice, rats, and humans is also highly conserved in structural organization (29 -31) and contains introns at the same locations as in the Npr2 gene.
The coding sequence of the GC-B cDNA is 3,141 bp in length, and the deduced protein contains 1,047 amino acids. The sequence around the putative initiation codon "CCCCATGG" is consistent with a consensus initiation sequence (32); there also is an in-frame termination codon 126 nucleotides upstream from the initiation codon (Fig. 2a). The deduced amino acid sequences of murine, rat, and human GC-B show high identity across the species (e.g. murine and human GC-B (29) are 92 and 98% identical at the nucleotide or amino acid levels, respectively). Hydropathic analysis suggests that cleavage of the signal peptide occurs after amino acid 22 (Fig. 2a) yielding a mature protein of 115 kDa. The ECD, TM, KHD, and the CYC domain are predicted to be encoded by exons 1-6, 7, 8 -15, and 16 -22, respectively (Fig. 1a). The ECD contains six cysteine residues that participate in intramolecular disulfide bridges and seven potential N-linked glycosylation sites (Fig. 1c); this is also the case for rat GC-B (5). The serine and threonine residues (Ser 513 , Thr 516 , Ser 518 , Ser 523 , Ser 526 , and Thr 529 ) that have been identified as phosphorylation sites in rat GC-B (33) and an ATP-binding motif-like sequence (Leu 519 -Xaa-Gly 521digested with KpnI, SpeI, SacI, or XbaI was analyzed with the 0.3-kb cDNA probe corresponding to the region from the KpnI site to primer RT-AS shown in a. Size markers are shown at the left. c, expected structures of the three GC-B isoforms. The deduced amino acid sequence at the border between exons 8 and 9 is shown in three-letter codes with amino acid numbers (the Met of the initiation codon is 1) at the top, the expected structure of the GC-B1 homodimer is shown in the middle, and the expected structures of GC-B2 and GC-B3 monomers are shown at the bottom. Phosphorylated Ser and Thr residues are written in bold italic letters, and the Leu and Gly residues of the ATP-binding motif-like sequence (Leu-Xaa-Gly-Xaa-Xaa-Xaa-Gly) are underlined. ECDs, TMs, KHDs, and CYCs are shown by lines with loops, closed boxes, open ellipses, and closed ellipses, respectively. Regions encoded by exon 9 are shown as hatched boxes. Cys residues that may participate in extracellular disulfide bonds and potential N-glycosylation sites are shown by the letter C and closed triangles, respectively, on the upper GC-B1 molecule. The closed circle at the C terminus of GC-B3 indicates the extra 12-amino acid sequence encoded by intron 5.
Three GC-B Isoforms Generated by Alternative Splicing-Three different GC-B mRNA species are detected in brain. A comparison of the cDNA sequence with the genomic sequence reveals that all three mRNA species are generated by alternative splicing of a single Npr2 gene; one transcript contains all 22 exons and retains no intron sequences (GC-B1), a second (GC-B2) splices out exon 9 (75 bp) but remains in-frame with GC-B1, and a third (GC-B3) retains the 180 bp of intron 5 (Fig.  1, a and c). GC-B2 (112 kDa) lacks three of the six potential phosphorylation sites (Ser 523 , Ser 526 , and Thr 529 ) and the putative ATP-binding motif (Fig. 1c). GC-B3 is predicted to be 44 kDa and contain only a portion of the ECD due to an in-frame termination codon within intron 5 (Fig. 1, a and c). The use of primers over the entire coding region failed to detect other alternatively spliced transcripts.
GC-B1 and GC-B2 are approximately the same size (120 kDa), while GC-B3 is about 60 kDa based on Western blot analysis (Fig. 3). Digestion with peptide:N-glycosidase F significantly reduced the molecular size of each GC-B isoform suggesting N-glycosylation of all three GC-B isoforms.
CNP Binding to GC-B Isoforms-Equal expression of the GC-B isoforms (Western blot analysis, not shown) allowed a determination of relative CNP binding for each isoform. GC-B1 and GC-B2 bound CNP with similar apparent affinity (one-half maximal binding at about 0.4 -0.5 nM), while GC-B3 did not appear to bind CNP (Fig. 4).  GC-B Isoform Activities-CNP-independent and CNP-activated activities of GC-B isoforms were examined using COS-7 cells. Western blot analysis confirmed equal expression of the various isoforms. Intracellular cGMP levels were more than 50-fold higher in cells expressing GC-B1 than in mock-transfected cells and were increased by CNP in a dose-dependent manner (Fig. 5). Almost a 300-fold increase in cGMP occurred with 100 nM CNP. Basal cGMP levels of cells expressing GC-B2 were similar to those of cells expressing GC-B1, but CNP failed to elevate cyclic nucleotide concentrations (Fig. 5). There were no significant differences in cGMP levels between cells expressing GC-B3 and mock-transfected cells and no effects of CNP (Fig. 5).
Interactions among GC-B Isoforms-Compared with intracellular cGMP levels of COS-7 cells transfected with pCMV5/  FLAG-mGC-B1 and pCMV5 (mock), the cGMP levels of cells transfected with pCMV5/FLAG-mGC-B1 and pCMV5/Myc-mGC-B2 were about twice only where treated with vehicle but about 50% where treated with CNP at 10 nM-1 M; the same number of the cells were transfected with the same amount of plasmids, and molar ratios of plasmid pairs were kept at 1:1 (Fig. 6a). The cGMP levels of cells co-transfected with pCMV5/ FLAG-mGC-B1 and pCMV5/HA-mGC-B3 were similar to those found in cells transfected with pCMV5/FLAG-mGC-B1 and pCMV5, but in the presence of CNP, the inclusion of GC-B3 clearly resulted in a much lower response (Fig. 6b). The EC 50 for CNP-induced cGMP elevations was similar between cells expressing FLAG-mGC-B1 alone, cells expressing FLAG-mGC-B1 and Myc-mGC-B2, or cells expressing FLAG-mGC-B1 and HA-mGC-B3 (ϳ30 nM) (Fig. 6, a and b).
When FLAG-mGC-B1 was stimulated by 100 nM CNP, intracellular cGMP levels were decreased by co-expression of Myc-mGC-B2. The magnitude of the suppression was increased up to 75% as the molar ratio of pCMV5/Myc-mGC-B2 to pCMV5/ FLAG-mGC-B1 was increased as high as 2 (Fig. 6c). HA-mGC-B3 had much smaller effects when co-expressed with FLAG-mGC-B1 (Fig. 6c).
Heterodimer formation between GC-B1 and GC-B2 or between GC-B1 and GC-B3 was assessed in co-immunoprecipitation experiments. Myc-mGC-B2 was co-immunoprecipitated with FLAG-mGC-B1 by anti-FLAG-agarose when both FLAG-mGC-B1 and Myc-mGC-B2 were co-expressed, and this coimmunoprecipitation was completely blocked by the FLAG peptide (Fig. 7a). HA-mGC-B3 was also co-immunoprecipitated with FLAG-mGC-B1 when both FLAG-mGC-B1 and HA-mGC-B3 were co-expressed, and again the co-immunoprecipitation was completely blocked by the FLAG peptide (Fig. 7b). Thus, GC-B2 or GC-B3 is able to form heterodimers with GC-B1.
Expression of GC-B Isoform mRNAs in Murine Organs-Npr2 transcripts of about 4 kb were detected in a wide variety of murine tissues (Fig. 8a). The expression levels were high in the cerebrum, cerebellum, brain stem, pituitary gland, adrenal glands, and ovaries; were moderate in lungs, heart, skeletal muscle, eyes, and the growth plate cartilage; and were low in the thymus, aorta, liver, kidneys, and testes. The signal was barely detectable in the spleen, stomach, jejunum, or colon. GC-B2 and GC-B3 mRNA was distinguished from GC-B1 mRNA by RT-PCR since the differences in mRNA size between GC-B isoforms were too small to be distinguished on Northern blots (Fig. 8b). The ratios of GC-B2 to GC-B1 mRNA levels were moderate (30 -50%) in the cerebrum, cerebellum, brain stem, thymus, and spleen; low (10 -30%) in the pituitary gland, heart, aorta, kidneys, adrenal glands, stomach, skeletal muscle, testes, and small intestine; and barely detected relative to GC-B1 (Ͻ10%) in lungs, liver, ovaries, and uterus. Transcript ratios of GC-B3 to GC-B1 were high (Ͼ50%) in the cerebrum, cerebellum, skeletal muscle, testes, ovaries, and uterus; moderate in lungs, aorta, and stomach; and low in the brain stem, pituitary gland, heart, spleen, kidneys, and adrenal glands. GC-B3 was barely detectable in the thymus, liver, or jejunum. In the growth plate cartilage, which showed high Npr2 mRNA expression (Fig. 8a), GC-B2 and GC-B3 transcripts were barely detectable (not shown).

Expression of GC-B Isoforms in Aortic Smooth Muscle Cells of Different
Phenotypes-GC-B isoform mRNA expression was compared between the medial layer of the aorta, which contains mostly SMCs, and cultured aortic SMCs (Fig. 9). Both SM1 and SM2 isoform mRNA of smooth muscle myosin heavy chain was detected in the aorta, while only SM1 mRNA was detected in cultured aortic SMCs (Fig. 9); the aorta contains SMCs of the contractile phenotype, and cultured aortic SMCs switch to the synthetic phenotype (35,36). GC-B1 mRNA expression in cultured aortic SMCs was increased at passages 1-3 compared with that of aorta (Fig. 9). The ratio of GC-B2 to GC-B1 in mRNA levels was about 20% in the aorta but less than 10% in the cultured aortic SMCs at passage 1; GC-B2 was barely detectable in cultured aortic SMCs at passages 2 and 3 ( Fig. 9). GC-B3 mRNA expression levels remained similar in aorta and cultured aortic SMCs (Fig. 9). DISCUSSION This study shows that three GC-B mRNA isoforms (GC-B1, GC-B2, and GC-B3) are generated from the Npr2 gene by alternative splicing (Fig. 1, a and c). In the rat, a GC-B2 isoform also appears to exist (37).
Murine GC-B1 retains some basal GC activity independent of CNP but upon binding of ligand is markedly activated in a concentration-dependent manner. Thus, GC-B1 is a signaling receptor for CNP. GC-B2 retains basal GC activity and CNP binding similar to GC-B1 but is not activated by CNP. This is not surprising since deletion of the entire KHD from rat GC-A eliminates atrial natriuretic peptide-dependent increases of GC activity, and previous work has suggested that the KHD is essential for ligand-induced activation of various GC receptors (3). Furthermore site-directed mutagenesis (S523A or S526A) reduces CNP-stimulated GC-B activity by greater than 70% (33). GC-B2 lacks Ser 523 and Ser 526 . It also has been reported that disruption of the putative ATP-binding motif of GC-B eliminates CNP-induced activation of GC-B (34), and GC-B2 lacks Gly 521 found within this motif. Therefore, it is concluded that the region encoded by exon 9 is indispensable for CNP- induced activation of GC-B, although this region is not critical for either basal guanylyl cyclase activity or ligand binding.
GC-B3 does not possess either guanylyl cyclase activity or the ability to bind CNP. The failure of CNP to bind to GC-B3 could have been due to secretion of the protein into the medium as a soluble receptor; however, we detected GC-B3 in the membrane preparation (Fig. 3a)  after secretion into the extracellular space, or there may be failure of GC-B3 trafficking from the endoplasmic reticulum to the plasma membrane, resulting in no CNP binding.
In addition to these isoforms, previous work in the human has suggested the presence of a fourth isoform (NPR-Bi); it contains a 71-bp insertion (intron 19) resulting in a frameshift that would disrupt the CYC domain (38). This isoform, nevertheless, has been suggested as a signaling receptor for CNP; in this case activation of a protein tyrosine kinase pathway is suggested to inhibit a K ϩ conductance in the apical membrane of human proximal tubule cells (39). We did not detect this isoform in the mouse. It should also be noted that no human expressed sequence tags encoding GC-B transcripts lacking exon 9 have been reported. The possibility that GC-B2 or GC-B3 utilizes a signaling pathway other than cGMP, as seen with NPR-Bi, remains to be determined.
Npr2 mRNA expression was detected in a wide variety of tissues by Northern blot analysis, and RT-PCR suggests that GC-B isoform expression levels and the ratios among the isoforms vary from tissue to tissue (Fig. 8b). Thus, an analysis of the expression level of only GC-B1 is probably not sufficient information for estimation of CNP-sensitive guanylyl cyclase activity in tissues or cells. Npr2 mRNA expression, estimated by Northern blot analysis, was high in brain tissues (Fig. 8a), but GC-B2 or GC-B3 ratios relative to GC-B1 are also relatively high in the cerebrum and cerebellum. Thus, GC-B2 and GC-B3, given the dominant negative effects of the isoforms, would be expected to play an important role in adjusting the magnitude of a CNP signal in brain, and in fact, CNP marginally increased guanylyl cyclase activity in homogenates of brain while significantly increasing cyclase activity in homogenates of lungs where GC-B2 mRNA was barely detectable (Fig. 8b).
It has been shown that GC-B mRNA expression is increased in cultured aortic SMCs relative to normal arterial walls (40). CNP/GC-B signaling in SMCs of the synthetic phenotype serves as a counter-regulatory system to inhibit the migration and proliferation and to induce a phenotypic reversion of cultured aortic SMCs (41,42). GC-B1 mRNA expression is increased, and GC-B2 mRNA expression is almost completely repressed in cultured aortic SMCs compared with aortic tissue, and thus GC-B2 may serve as a new molecular marker for SMCs of the contractile phenotype. We also suggest that CNP/ GC-B signaling is activated in cultured aortic SMCs not only because of increased expression of GC-B1 but also because of repression of the dominant negative isoform, GC-B2.
CNP/GC-B signaling is therefore regulated at two levels with respect to expression: transcription and post-translational modification. GC-B mRNA expression in cultured rat aortic SMCs is attenuated by transforming growth factor-␤ 1 (43). Signaling through GC-B also can be regulated by changes in the mRNA expression of natriuretic peptide receptor-C, which plays an important role in the clearance of natriuretic peptides. Stimulation of the ␤ 2 -adrenoreceptor attenuates natriuretic peptide receptor-C mRNA expression in cultured rat aortic SMCs (44), thus potentially affecting GC-B signaling as well. The dephosphorylation of GC-B desensitizes GC-B to CNP as a post-translational modification (22,45). Serum, platelet-derived growth factor, basic fibroblast growth factor, and phorbol FIG. 8. Expression of Npr2 mRNA in murine tissues. a, Npr2 mRNA expression in murine tissues detected by Northern blot analysis. Twenty micrograms each of total RNA isolated from a series of murine tissues was analyzed unless otherwise specified. Signals of Gapdh are shown at the bottom as internal controls. b, messenger RNA expression of GC-B isoforms in murine tissues detected by RT-PCR. Five micrograms each of total RNA was reverse-transcribed and one-tenth of the RT reaction was subjected to PCR to detect transcripts of GC-B isoforms (GC-B1, GC-B2, and GC-B3). One fiftieth of the RT reaction was used to detect the Gapdh transcript. *, growth plate cartilages of knee and ankle joints from mice at 8 days of age.
FIG. 9. Difference in mRNA expression of GC-B isoforms between medial layer of aorta and cultured aortic smooth muscle cells. Five micrograms each of total RNA isolated from the medial layer of the aorta and cultured aortic SMCs at passages 1, 2, and 3 (p1, p2, and p3) was reverse-transcribed, and one-tenth of the RT reaction was subjected to PCR to detect mRNA of smooth muscle myosin heavy chain isoforms (SM1 and SM2 at the top row) and GC-B isoforms (GC-B1 and GC-B2 at the second row and GC-B1 and GC-B3 at the third row). One-fiftieth of the RT reaction was used in PCR to detect the Gapdh transcript (at the bottom). RT reactions of each sample were with (ϩ) or without (Ϫ) the reverse transcriptase to rule out signals from contaminating genomic DNA, and PCR primers were designed so the size of amplification products would be different among the isoform mRNA or genomic DNA. In the lane indicated as 0.5ϫ, the RTϩ reaction was diluted 2-fold with the RTϪ reaction and then used in PCR.
12-myristate 13-acetate cause dephosphorylation and decrease CNP-induced activation of GC-B in NIH-3T3 fibroblasts overexpressing rat GC-B1 (13,22,45). Here we show yet another means of regulating CNP/GC-B signaling: the regulation of relative levels of the various isoforms. This type of regulation has been previously suggested for soluble GC where ␣ 2i and ␤ 2 isoforms have been proposed as dominant negative isoforms (46,47), and thus the guanylyl cyclase family, in general, may utilize such dominant negative isoforms as regulators.