Posttranscriptional regulation of soluble guanylylcyclase expression in rat aorta

We investigated the molecular mechanism of cyclic GMP-induced downregulation of soluble guanylylcyclase expression in rat aorta. 3-(5’-hydroxymethyl-2’-furyl)-1-benzyl indazole (YC-1)1, an allosteric activator of this enzyme, decreased the expression of soluble guanylylcyclase α 1 subunit mRNA and protein. This effect was blocked by the enzyme inhibitior 4 H -8-bromo-1,2,4-oxadiazolo(3,4-d)benz(b)(1,4)oxazin-1-one (NS 2028)2, and by actinomycin D. Guanylylcyclase α 1 mRNA-degrading activity was increased in protein extracts from YC-1-exposed aorta and was attenuated by pretreatment with actinomycin D and NS 2028. Gelshift- and supershift-analysis using an AU3-rich ribonucleotide from the 3’-untranslated region of the α 1 mRNA and a monoclonal antibody directed against the mRNA stabilizing protein HuR4 revealed HuR mRNA binding activity in aortic extracts, which was absent in extracts from YC-1-stimulated aortas. YC-1 decreased the expression of HuR, and this decrease was prevented by NS 2028. Similarly, downregulation of HuR by RNA interference in cultured rat aortic smooth muscle cells decreased α 1 mRNA and protein expression. We conclude that HuR protects the guanylylcyclase α 1 mRNA by binding to the 3’-untranslated region. Activation of guanylylcyclase decreases HuR expression, inducing a rapid degradation of guanylylcyclase α 1 mRNA and lowering α 1 subunit expression as a negative feedback response.

In addition to an acute activation of sGC the output of the NO-cGMP pathway can also be controlled at the level of sGC expression. Thus, a reduced vasodilator response to exogenous NO consistent with a downregulation of sGC has been observed in aortic tissue of aged spontaneously hypertensive rats (4). On the other hand, an upregulation of sGC expression was found in aortic tissue from nitroglycerin-tolerant rats (5) and from rats suffering from chronic heart failure (6), despite diminished vasodilator responses to NO. This apparently discrepant finding indicates that altered sGC expression does not necessarily translate into predictable changes in cGMP-dependent functional responses, but that other mechanisms, such as altered NO bio-availability, may overrun the influence of altered sGC expression.
These findings exemplify the need for understanding the molecular mechanisms accounting for regulation of sGC expression. There is evidence that expression of sGC is controlled by second messenger cyclic nucleotides via a post-transcriptional mechanism: in various cells cyclic AMP eliciting agonists decrease the expression of sGC mRNA and ________________________ 5 sGC: soluble guanylylcyclase: GTP pyrophosphate-lyase (cyclizing), EC 4.6.1.2 6 NO: nitric oxide 7 cGMP: guanosine 3':5'-cyclic monophosphate protein (7,8) by a destabilization of the sGC mRNA. This effect is mimicked by activation of the cGMP-signaling pathway, e.g. application of NO donors, stimulation of particulate guanylate cyclase by atrial natriuretic factor, and stimulation of cGMP-dependent protein kinase by the stable cGMP-analogue 8-chlorophenylthio-cGMP (9).
The objective of the present investigation was to characterize the mechanism accounting for sGCα 1 mRNA destabilization induced by increased cGMP formation in isolated rat aorta. We observed that the elav family protein HuR (10) stabilizes the sGCα 1 mRNA by binding to AU-rich elements (ARE) 8 in its 3-untranslated region (UTR) 9 , and that an increase in intracellular cGMP strongly decreases HuR expression and sGCα 1 mRNA binding activity, leading to accelerated mRNA degradation. ________________________ 8 ARE: AU-rich element 9 3'-UTR: 3'-untranslated region

MATERIALS AND METHODS
The polyclonal chicken antibody directed against the α 1 -and β 1 -subunit of the rat lung sGC was produced by BioGenes GmbH (Berlin, Germany), which also provided the rabbit-antichicken antibody. For some experiments a sGCα 1 specific peptide antibody was obtained from Dr. Stasch, Bayer AG, Leverkusen. The oligonucleotides for RT-PCR 10 , in-vitrotranscription and gelshift analysis were synthesized by BioSpring GmbH and MWG-Biotech.

Rat aortic tissue
Male normotensive Wistar Kyoto rats (300 g) were obtained from Möllegaard (Skensved, Denmark) and were kept according to institutional guidelines, in compliance with German laws. The thoracic aorta was isolated from anaesthetised rats (200 mg/kg ketamin (Exalgon), 100 mg/kg xylazin (Rompun), cleaned from fat and connective tissue, and cut into rings of equal length (3 mm). The endothelium was removed by gentle forcing and rolling a glass rod through the lumen. The aortic rings were kept in culture dishes (6 well) in MEM 11 under a carbogen atmosphere (4.5% CO 2 ) at 37°C. The rings were exposed to YC-1, NS2028 and/or actinomycin D for different periods of time and were then snap-frozen in liquid nitrogen and stored at -70°C.

Isolation of total RNA from rat aorta and RT-PCR
________________________ 10 RT-PCR: reverse transcriptase-polymerase chain reaction 11 MEM: Modiefied Eagle's Medium Frozen tissue was ground in liquid nitrogen with porcelain mortar and a pestle. Total RNA was extracted by the modified guanidine isothiocyanate method of Chomczynski and Sacchi (11). The reverse transcriptase-polymerase chain reaction (RT-PCR) for the sGCα 1 mRNA (product size 826 bp 12 ) and elongation factor II (225 bp) was performed exactly as described previously (4).

Poly(A) + RNA (mRNA) isolation from rat lung
Poly(A) + mRNA was purified from total RNA by means of the Messagemaker kit (Life Technologies, GIBCO-BRL). Total RNA (2 mg; 0.55 mg/ml) was denatured for 5 min at 65°C. The salt concentration was adjusted to 0.5 M NaCl. Subsequently the RNA was incubated with the oligo(dT) cellulose suspension and heated for 10 min at 37°C. After filtration the suspension was washed with 20 mM Tris/HCl pH 7.5, 0.5 M NaCl and then with 20 mM Tris/HCl pH 7.5, 0.1 M NaCl. The mRNA was eluted with RNase-free water.

Preparation of sGCα 1 transcripts by in-vitro transcription
Total RNA of rat lungs was used as a template for RT-PCR amplification of the

RNA-protein binding reactions and supershift assays
Electrophoretic mobility shift assays (EMSA) 18 were carried out by a modification of the method of Wang (12). The oligoribonucleotide (3UTRSK2, 50 -200 ng) was incubated with 40 -100 µg native nuclear extract (prepared according to (13)) from endothelium-denuded rat aorta, and a 10x reaction buffer (15 mM Hepes pH 7.9, 600 mM KCl, 10  h and blotted onto a nylon membrane (Biodyne B, Pall) overnight in 10 x SSC. Blocking and detection of biotin-labeled bands was performed as described for northern blots. For supershifts, 4-15 µg of the monoclonal HuR-antibody was incubated with the native nuclear extract for 1 h on ice before the specific riboprobe was added; all subsequent steps were performed as described for native gels.

Immunodetection of the sGCα 1 subunit
Total protein was precipitated (1.5 ml 100 % isopropanol) from the phenol-ethanol supernate (Trizol-method) of the RNA extraction, and the precipitate dissolved in 1 % SDS. The ________________________ 18 EMSA: electrophoretic mobility shift assay 19 TAE: 40 mM Tris pH 8.5, 0.1 % acetic acid, 2 mM EDTA protein (40 µg per lane) was fractionated on Lämmli gels and electro-blotted onto nitrocellulose filters (Protrans; Schleicher & Schuell). The blots were blocked at 4°C over night and then incubated for 2 hours at RT with either a polyclonal chicken antibody (IgY) or a rabbit antibody (IgG) directed against the α 1 -subunits of sGC (1:100 dilution in blocking buffer). The blots were washed and then developed either with a peroxidase A-conjugated anti-chicken IgY (IgG, rabbit, 1:5,000 in blocking buffer), or anti rabbit IgG (goat, 1:10,000).
Immunoreactive peptides were visualized by chemiluminescence and exposure to x-ray film.
The autoradiographs were analyzed by scanning densitometry. Equal protein loading and blotting was verified by α-actin immunostaining.

Design of HuR-specific siRNA
A HuR specific siRNA (HuR-siRNA) was designed by selecting a target region from base position 163 -183 relative to the start codon, which fulfilled the specific sequence requirements: (AA(N 19 )dTdT; N is any nucleotide; 21-nt sense and 21-nt antisense strand; approximately 50% G/C content and a symmetric 2-desoxythymidine 3' overhang. Sense and antisense oligonucleotides were synthesized by Xeragon Oligonucleotides (Xeragon-Qiagen).
The lyophilised siRNA was dissolved in sterile annealing buffer (100 mM potassium acetate, 30 mM Hepes-KOH, 2 mM magnesium acetate, pH 7.4) to obtain a 20 µM solution. Then oligonucleotides were heated to 90°C for 1 min followed by 1 hour at 37°C. In addition, we also used a HuR-siRNA targeting base position 564-584 relative to the start codon (kindly provided by S. Sengupta) for cell transfections. Lyophilised or dissolved siRNAs were stored at -20°C.

Transfection of cultured vascular smooth muscle cells (VSMC)
Cultured smooth muscle cells from rat aortas were grown in minimal essential medium (MEM, Gibco) containing 10% FCS and 1% penicillin and streptomycin at 37 °C and 5% CO 2 . SMCs were trypsinized, mixed with fresh MEM (+ antibiotics) without FCS and seeded on

Influence of YC-1 on the expression of sGCα 1 mRNA and protein in rat aorta
To assess the effect of increased cGMP formation on the sGCα 1 subunit expression in rat aorta freshly isolated endothelium-denuded aortic rings from WKY rats were kept under organ culture conditions, either in the absence (control +/-0.2 %DMSO 20 ) or presence of the sGC activator molecule YC-1 (10 µM), and the specific sGC inhibitor NS 2028 (10 µM).
After 24 h the vascular tissue was snap-frozen and homogenized in liquid nitrogen, then further processed for sGCα 1 subunit mRNA and protein expression by RT-PCR ( Fig 1A) and western blot (Fig 1B), respectively. According to densitometric analysis of the RT-PCR product (Fig. 1A) and the immunoreactive protein (α 1 = 82 kD, Fig. 1B) the abundance of sGCα 1 subunit mRNA and protein was markedly lower in YC-1 exposed aorta compared to controls (α 1 -mRNA = 93% lower, α 1 -protein = 56% lower vs. controls; bar graphs; Fig. 1A and B), while the levels of elongation factor II mRNA (Fig. 1A) and α-actin protein (Fig. 1B) were not affected by YC-1. In the presence of NS2028 the ability of YC-1 to decrease sGC subunit gene expression was almost completely blocked ( Fig. 1A and B). These findings show that long-lasting activation of sGC in the rat aorta decreases sGCα 1 subunit expression at the mRNA and protein level. Act D prevented the YC-1-induced decrease of sGCα 1 mRNA abundance ( Fig.2A). The mRNA levels of elongation factor II remained stable for up to 9 h and were not affected by Act D and YC-1 (Fig. 2A). These results suggest that YC-1 decreases the stability of sGCα 1 mRNA by a mechanism requiring transcriptional activation of an unknown factor.
The time course of YC-1-induced sGCα 1 mRNA decay was mirrored by a quite similar time course of sGCα 1 protein expression, as assessed by western blots analysis (Fig. 2B). In contrast, the expression of α-actin was constant for the same period of time (Fig. 2.B).

YC-1-induced sGCα 1 poly(A) + RNA-destabilizing activity in the native protein extract from rat aorta
To further corroborate our finding of a YC-1-induced destabilization/accelerated degradation of sGCα 1 mRNA in the rat aorta, we assessed the effect of a protein extract from YC-1exposed rat aorta on the rate of sGCα 1 mRNA degradation. Therefore, total native protein was isolated from a part of the aortic rings used in the previous experiments ( Figs. 1 and 2) and 20 µg protein was incubated at 37°C with 1 µg enriched poly(A) + RNA isolated from rat lung (cf. methods). After different periods of time (10 -50 min (Fig. 3A) or 15 -45 min (Fig.   3B) an aliquot of the incubation mixture was probed for sGCα 1 and elongation factor II mRNA by northern blotting (see methods). In the absence of aortic protein (control) the sGCα 1 mRNA was stable for up to 50 min under these assay conditions (Fig. 3A). In the presence of protein from aortas exposed to 0,2 % DMSO (solvent control) a moderate time-dependent decrease in sGCα 1 mRNA abundance was observed ( Fig. 3A and B, lanes "DMSO"). The rate of sGCα 1 mRNA decay was considerably accelerated by protein isolated from YC-1exposed aortas (Fig. 3A,B, lanes "YC-1"). In contrast, elongation factor II mRNA was quite stable even in the presence of protein from YC-1-exposed aorta ( Fig. 3A and B, lower autoradiographs), indicating that YC-1 specifically induced factors which led to accelerated decay of sGCα 1 mRNA. The formation of these factors was apparently prevented by a preincubation of the aortas with Act. D (Fig. 3A) or NS2028 (Fig. 3B), since under these conditions the aortic protein extract exhibited markedly less sGCα 1 mRNA degrading activity.

Identification of HuR as a sGCα 1 mRNA-binding protein in rat aorta
The 3-UTR of the rat sGCα 1 mRNA bears several AUUUA-motifs (AU-rich elements, AREs) 8 (Fig. 4A), which target the mRNA for rapid degradation by specific endonucleases HuR-antibody the RNA-protein band was further retarded (supershifted), demonstrating, that HuR forms a complex with the ARE containing sequence of sGCα 1 mRNA (Fig. 5, lanes 6 and 7).

Activation of sGC by YC-1 decreases HuR-ARE-binding activity
In order to clarify whether activation of sGC decreases HuR-like binding activity, endothelium-denuded rat aortic segments were kept for 12 h under organ culture conditions (cf Methods), either in the absence or presence of YC-1 (100 µM), and NS 2028 (100 µM), or the solvent control (0,2 % DMSO). Nuclear protein extracts were prepared from the vascular tissue and the expression of HuR-like ARE-binding activity was assessed by RNA-EMSA using the 3UTRSK2 probe. In the presence of protein (80 µg) from control aortas a similar bandshift as shown in Fig. 5 was observed (Fig. 6, lane 2+3). The protein extract from DMSO-treated aorta induced a quite similar shift (Fig. 6, lanes 4+5). In contrast, with protein from YC-1-exposed aorta the shifted band markedly decreased (Fig. 6, lanes 6+7). This effect of YC-1 was prevented by concomitant exposure of the aorta to the sGC inhibitor NS 2028 (Fig. 6, lanes 8+9). Addition of the monclonal HuR antibody induced a strong supershift, which under these chromatographic conditions unfortunately superimposed with the shift (Fig. 6, lane 10). These findings indicate that YC-1 either induces a reduction of the HuR affinity for the 3'-UTR of GCα 1 mRNA, or that it downregulates HuR expression.

Activation of sGC in rat aorta by YC-1 decreases expression of HuR
By western blot analysis we assessed whether YC-1 affected HuR expression. Protein from rat aorta incubated with YC-1 and NS 2028 as shown in Fig. 6 was loaded on a SDS-PAGE,

HuR gene knockdown by RNA interference decreases expression of sGC± 1
RNA interference allows targeted genes to be easily and efficiently "switched off", using short stretches of double-stranded RNA that contain the same sequence as mRNA transcribed from the target gene (16). We used this approach to assess whether specific gene knockdown of HuR in cultured rat aortic smooth muscle cells (RASMC) 22 affects expression of sGC.

Incubation of RASMC for 24 h with two different HuR siRNA oligonucleotides decreased
HuR expression at the protein (Fig. 8A "siRNA") and mRNA level (Fig. 8B "siRNA"). In the same cell extracts the expression of sGC± 1 protein and mRNA was decreased as well (Fig.   8A,B), compared to controls. Expression of actin protein (Fig. 8A) and elongation factor II mRNA (Fig. 8B) was not affected by HuRsiRNA. This finding clearly shows that specific knockdown of HuR decreases sGC± 1 expression in vascular smooth muscle cells.

Discussion
The heterodimeric hemoprotein and NO receptor sGC is a key component of the NO/cGMP signal transduction pathway in vascular smooth muscle and other tissues. In addition to an acute regulation by positive (NO) or negative (superoxide radical) input signals (1) the activity of this pathway can also be controlled at the level of sGC expression (4). Previous studies have shown a feedback-inhibition of sGC expression by its product cGMP (9,17,18). This finding was related to an accelerated decay of the sGCα 1 and β 1 mRNA (17).
We set out to reveal the mechanism accounting for the downregulation of the sGCα 1 subunit expression in response to sGC activation. Our rationale for studying sGCα 1 was that in preliminary studies we found that this subunit was less expressed in rat vascular tissues than the β 1 subunit and therefore formation of the NO-sensitive α 1 β 1 holoenzyme would be limited by the α 1 subunit. A NO-independent activator of sGC, YC-1 (19), was chosen here to avoid possible interference by cGMP-independent effects of NO on gene expression (20,21).
The 3-UTR of the rat sGCα 1 mRNA bears several AUUUA-motifs (AU-rich elements, AREs), which are targeted by trans-acting factors for regulation of mRNA stability (15, 22).
One of these factors is the ubiquitious 34 kDa protein HuR, which binds to AREs with high affinity and selectivity (10), thereby protecting the respective mRNA from accelerated decay (12). By western blot analysis we were able to show for the first time that HuR is constitutively expressed in the rat aorta. We provide evidence by in vitro mRNA degradation assay and RNA-EMSA that HuR protects the rat sGCα 1 mRNA by binding to ARE present in the 3'-UTR. Furthermore, we could demonstrate that prolonged (12 h) sGC-activation by YC-1 decreases the expression of HuR protein and HuR binding activity for sGCα 1 mRNA.
Consequently, the expression of the sGCα 1 subunit was decreased at the mRNA and protein level. All these effects could be blocked by an inhibitor of YC-1-stimulated sGC activity, NS 2028 (23), indicating that they were caused by an increased sGC activity/cGMP formation. In this regard, sGC is not just another HuR-regulated gene, but also a regulator of HuR expression, linking increased cGMP levels to depression of HuR activity and lower sGC expression. Though we did not investigate whether lowering of resting cGMP-levels will increase HuR expression, our findings suggest the existence of a negative feedback loop formed by sGC and HuR.
To confirm the hypothesis that a decrease in HuR expression induces a decrease in sGC expression we used the RNA knockdown (RNA interference) technique (16). RNA interference is a gene silencing mechanism that uses double-stranded (ds) RNA as a signal to trigger the degradation of the targeted mRNA. 48 h after transfection of cultured rat aortic smooth muscle cells with a 21mer ds RNA homologous to base position 163 -183 relative to the start codon of the HuR message (HuRsiRNA) we observed a strong decrease in HuR as well as in sGC± 1 expression at the mRNA and protein level (Fig. 8). This experiment proves that downregulation of sGC mRNA is a consequence of decreased HuR expression.
The signaling cascade accounting for cGMP-dependent downregulation of HuR could not be revealed in this study. Since concomitant application of Act D during exposure of the rat aorta to YC-1 prevented the downregulation of HuR expression and binding activity it is likely that cGMP induces the transcriptional activation of (unknown) factors which decrease HuR expression. Our preliminary data indicate that a cGMP-activated proteinkinase and the transcription factor AP-1 are involved in downregulation of HuR by sGC activators (S. Kloess, A. Mülsch, unpublished), but an in depth study is required. AP1 sites are present in the mouse HuR promoter region (24). Interestingly, CREB sites were also found in this promoter region (24). Agents which increase intracellular cyclic AMP decrease sGC subunit mRNA levels and cellular cGMP formation in response to NO-donor compounds (7,8). We observed that cyclic AMP-eliciting agonists decrease expression of HuR in rat aortic cmooth muscle cells as well (Kloess and Mülsch, unpublished results), suggesting that HuR also mediates the down-regulation of sGC in response to increased cAMP levels. It appears that HuR can integrate cyclic nucleotide second messenger signaling and translate changes in cAMP and cGMP levels in altered gene expression. This underlines the increasing importance of mRNA stability regulation for gene expression (25), as compared to transcriptional regulation. In addition to sGC, other components of the NO/cGMP pathway are also regulated by altered mRNA stability. In human mesangial cells, which exhibit a smooth muscle celllike phenotype, the expression of the cytokine-inducible NO synthase II is also downregulated by NO and cGMP. Part of this negative modulation is caused by decreased mRNA stability (26). The 3'-UTR of NO synthase II also bears AREs, and HuR was shown to stabilize the NOS III mRNA by binding to several of these AREs. The expression of HuR in cytokine-exposed DLD1 cells (human intestinal epithelium) decreased concomitantly with enhanced NOS III-derived NO formation (27). Furthermore, an increase or decrease in HuR expression brought about by stable transfection with HuR-sense or -antisense vectors increased or decreased NO synthase II expression. Collectively, these examples and our present findings emphasize that several major components of the NO/cGMP pathway are controlled at a post-transcriptional level by HuR in a negative feedback manner. Future studies will have to reveal the relative importance of HuR-regulated mRNA stability versus transcriptional processes for NO/cGMP dependent gene expression.  Rat aortic rings were exposed in organ culture to 10 µM YC-1 and/or 10 µM actinomycin D for 3, 6 and 9 hours, frozen and processed for RT-PCR and western blot analysis of sGCα 1 .
A) The upper fluorographs show ethidium bromide-stained agarose gels containing RT-PCR products of the sGCα 1 mRNA amplified from 2 µg total RNA as well as of elongation factor (ef) II mRNA. The sGCα 1 RT-PCR product intensities were normalized for ef II intensities.
Representative data from 3 rats.
B: Densitometric analysis of the HuR-specific bands , normalized by α-actin-staining.