Translational regulation of angiotensin type 1a receptor expression and signaling by upstream AUGs in the 5' leader sequence.

Rat angiotensin type 1a receptor (AT(1a)R) is regulated by four upstream AUGs present in the 5' leader sequence (5'-LS). Disruption of all four upstream AUGs (QM) results in 2-3-fold higher levels of angiotensin type 1 receptor (AT(1)R) densities in transiently transfected rat aortic smooth muscle cells (A10 cells) and stably transfected Chinese hamster ovary cells. Cells expressing QM have 5-fold higher levels of angiotensin II-induced inositol phosphate production than wild type (WT). Polysome analysis showed that QM mRNA is present in heavier fractions than the WT transcript, and 5.7-fold more AT(1)R protein is produced by in vitro translation from QM transcripts compared with WT transcripts. The AT(1a)R comprises 3 exons. Exon 3 (E3) encodes the entire open reading frame and 3'-untranslated region. Exons 1 and 2 (E1 and E2) and 52 nucleotides of E3 encode the 5'-LS. The AUGs in both exons contribute to the inhibitory effect on AT(1)R expression but not to the same degree. Disruption of the AUGs in exon 2 (DM2) relieves half of the inhibition, whereas disruption of the AUGs in exon 1 (DM1) is without effect. Disruption of the AUGs in exon 2 results in levels of receptor expression and translation that are indistinguishable from the alternative splice variant E1,3, which we previously showed was more efficiently translated than the E1,2,3 transcript. Individual mutations revealed that only the fourth AUG increased AT(1)R translation. In conclusion, all four AUGs present in the 5'-LS function cumulatively to suppress AT(1a)R expression and signaling by inhibiting translation. These data also show that both AUGs in E2 contribute to the inhibitory cis element present in this alternatively spliced exon.

In the majority of eukaryotic mRNAs, the first AUG downstream from the 5Ј cap site is the start of translation (1). In contrast, mRNAs that code for key regulatory proteins, such as transcription factors, protooncogenes, and key signaling molecules, commonly possess AUGs upstream of the AUG start codon. Upstream AUGs can play a critical role in the control of gene expression by causing ribosomal pausing or by forming a translation-competent ribosome that can initiate, terminate, and reinitiate. Both of these mechanisms can lead to reduced translation of the downstream open reading frame. Alternatively, an N-terminal extended protein can be synthesized from initiation at the upstream AUG, thereby competing with translation at the downstream open reading frame (2).
We have been studying the post-transcriptional regulation of the angiotensin type 1 receptor (AT 1 R) 1 (3), which is a G protein-coupled receptor that plays a critical role in regulating blood pressure and fluid homeostasis. Antagonists of this receptor are widely used to control hypertension and reduce the rate of progression of cardiovascular and renal disease (4). The rat AT 1a R comprises three exons (see Fig. 1). E3 harbors the entire open reading frame and the 3Ј-untranslated region, whereas the 5Ј-LS comprises exon 1 (E1), exon 2 (E2), and 52 nucleotides of E3. There are four upstream AUGs present in the 5Ј-LS, two in E1 and two in E2. E2 is alternatively spliced in a tissue-specific manner and contains an unidentified cis element that is inhibitory to receptor expression (5). In this study, we investigated the function of the upstream AUGs present in the 5Ј-LS on AT 1 R expression and signal transduction in transiently transfected rat aortic smooth muscle (A10) cells and in stably transfected Chinese hamster ovary (CHO) cells. We also studied the role of upstream AUGs on receptor translation by polysome analysis in transfected cells and by in vitro translation (IVT) assays.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-The AUGs in the 5Ј-LS of the AT 1a R cloned into the pcDNA5/FRT vector (Invitrogen) were subjected to sitedirected mutagenesis using the QuikChange site-directed mutagenesis system (Stratagene).
CHO Cell Culture and Stable Transfections-CHO cells were cultured in Ham's F-12K with 1.5 g/liter sodium bicarbonate, 2 mM Lglutamine, 10% fetal bovine serum, and antibiotics (as above). When cells were 60 -75% confluent, 20 g of plasmid DNA per 100 mm dish was transfected by the calcium phosphate method (Invitrogen). Individual clones were cultivated as described previously (6).
AT 1 R Radioligand Binding-A10 and CHO cell membranes were used in radioligand binding assays using [ 125 I-Sar 1 ,Ile 8 ]Ang II and a Brandel cell harvester as described (7). K d and B max values from Scat-* This research was supported by American Heart Association Beginning Grant-in-aid 0060205U, a National Kidney Foundation grant-inaid (to H. J.), and National Institutes of Health Grant HL57502 (to K. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Ribonuclease Protection Assay-Total RNA was isolated, and E1,3 and E1,2,3 mRNA were measured by ribonuclease protection assay as described previously (3) using a probe based on the coding region and thus common to both transcripts (see Fig. 1). In brief, the cDNA encoding the rat AT 1a R coding region in the pBluescript II vector (Stratagene) was linearized with EcoRI and transcribed in vitro with T7 RNA polymerase to yield a 380-bp protected cRNA fragment after hybridization with 3 g of total RNA followed by ribonuclease digestion according to the Ribonuclease Protection Assay III protocol (Ambion). The probe for ␤-actin was generated from pTRI-␤-actin-mouse cDNA (Ambion) with T7 RNA polymerase and yielded a 245-bp cRNA fragment. Radioactive signals were detected by a phosphoimaging device after electrophoresis on a 5% acrylamide gel.
Inositol Phosphate Assay-CHO cells stably expressing WT and QM were cultured to 70% confluence in 24-well plates before being treated for 16 h with Dulbecco's modified Eagle's medium containing 3 Ci/ml myo-[ 3 H]inositol (Amersham Biosciences). Ang II-induced IP production was assayed as described (8).
In Vitro Translation-WT and mutated cDNAs were subcloned into the pcDNA5/FRT expression vector (Invitrogen). Plasmid DNAs were linearized by XhoI digestion and in vitro transcribed into capped RNA (T7 mMessage mMachine, Ambion). One g RNA was translated in wheat germ extracts (Promega) in the presence of [ 35 S]methionine (Amersham Biosciences) as described (9).
Polynomal Distribution Analysis-Polysome analysis was performed as described (3) based on the principle that the largest polysomes are denser and therefore will sediment through a sucrose gradient faster than monosomes or free ribosomal subunits not bound to mRNA (10). The amount of WT or QM mRNA in each fraction was determined by ribonuclease protection assay as described (3). The same probe was used in the ribonuclease protection assay for WT and QM because it bound to an identical region in both transcripts. Sample variation in cytoplasmic levels of AT 1 R mRNA was controlled by normalizing the AT 1 R mRNA recovered in each fraction to the total amount of AT 1 R mRNA recovered from the entire fractionation.
Statistics-Data are expressed as means Ϯ S.E. Statistical significance (p Ͻ 0.05) of the differences between groups was determined by Student's t test; p Ͼ 0.05 was defined as not significant.

Effect of Disrupting All Four Upstream
AUGs on AT 1 R Densities-There are four upstream AUGs in the 5Ј-LS of the E1,2,3 AT 1a R mRNA transcript (WT) (Fig. 1). All four were disrupted by site-directed mutagenesis to create QM (Fig. 1).
To further characterize the QM mutant, QM and WT plasmids were stably transfected into CHO cells. Three independent and randomly selected clones were isolated for each plasmid. Saturation isotherms (Fig. 3A) and Scatchard plots ( Fig.  3B) were performed on each clone. All three individual QM clones expressed higher AT 1 R densities than all three WT clones (Fig. 3C). When the B max values from all three WT and QM clones were averaged (Fig. 2D), QM exhibited 2.1-fold higher AT 1 R densities compared with WT (B max (fmol/mg): WT, 54 Ϯ 1.2 versus QM, 112 Ϯ 12, n ϭ 3, p Ͻ 0.02) (Fig. 2D). No differences in AT 1 R mRNA levels were detected by real time PCR (Fig. 3E). When WT and QM B max values were normalized to AT 1 R mRNA levels, QM expressed 1.8-fold higher AT 1 R densities than WT-expressing cells (Fig. 3F).
Effect of Disrupting All Four Upstream AUGs on IP Signaling-To determine whether the differences in receptor densities affected Ang II signal transduction pathways, Ang IIinduced IP production was measured in the CHO cells stably expressing WT and QM. An Ang II dose-response curve on CHO cells expressing WT (Fig. 4A, inset) and QM (Fig. 4A) showed that 100 nM Ang II resulted in maximum accumulation of IP after 20 min. All three QM clones produced significantly more IP in response to 100 nM Ang II stimulation than all three WT clones (Fig. 4B); QM-expressing cells produced 5-fold higher levels of IP compared with the WT clones (IP (cpm): WT, 322 Ϯ 47 versus QM, 1642 Ϯ 37, n ϭ 3, p Ͻ 0.0001) (Fig. 4C). When normalized to AT 1 R B max , QM-expressing cells produced 2.2-fold higher levels of IP than WT-expressing cells (Fig. 4D).
Effect of Disrupting All Four Upstream AUGs on Translation-Polysome distribution profiles of AT 1 R mRNA were performed on WT-and QM-expressing CHO cells to compare WT and QM translational efficiencies in cells. The majority of the AT 1 R mRNA was located in fractions 2 and 3 in the WTexpressing cells, whereas the AT 1 R mRNA was shifted to the denser and more actively translated polysome fractions (fractions 1 and 2) in the QM-expressing cells (Fig. 5A). To determine whether QM was also translated in vitro more rapidly than WT, capped WT and QM RNAs were IVT in wheat germ extracts (Fig. 5B, inset). These IVT assays showed that 5.7-fold more AT 1 R protein was synthesized by QM compared with WT (Fig. 5B).
Relative Contribution of Upstream AUGs in E1 and E2 on Inhibition of AT 1 R Densities and IVT-We used site-directed mutagenesis to create DM1, in which the two AUGs in E1 were disrupted and DM2, in which the two AUGs in E2 were disrupted as well (Fig. 1). Radioligand binding analysis of transfected A10 cells revealed that AT 1 R densities were 1.4-fold higher in DM2-compared with WT-transfected cells, whereas no differences in AT 1 R densities were observed between DM1and WT-transfected cells (Fig. 6A). AT 1 R densities in DM2transfected cells were not as high as in QM-transfected cells; transfection of A10 cells with QM resulted in 1.4-fold higher AT 1 R densities than transfection with DM2 and 2.0-fold higher densities than transfection with WT cDNA.
To determine how upstream AUGs contribute to inhibitory RNA cis elements within the 5ЈLS, IVT assays were performed on QM, DM1, and DM2 RNA transcripts. No differences in AT 1 R translational efficiency were observed between the DM1 and WT transcripts, although DM2 resulted in a 2.0-fold increase in AT 1 R protein levels (Fig. 6B). The level of IVT in DM2-expressing cells was not as high as QM; QM resulted in 2.3-fold higher levels than DM2 and 4.5-fold higher levels than WT.
To further dissect the contribution of individual upstream AUGs in inhibiting AT 1 R expression, the two AUGs in E2 were individually mutated by site-directed mutagenesis to create M1 and M2 (Fig. 1). Radioligand binding studies showed an incremental increase in AT 1 R densities for M1 (1.1-fold) and a 1.3-fold increase for M2 (Fig. 7A). IVT assays showed a similar trend; M2 resulted in a 1.6-fold increase in IVT, whereas IVT of M1 was indistinguishable from WT (Fig. 7B). Because relief from translational repression was only observed in M2 and only the fourth AUG was in optimal Kozak consensus sequence (the ϩ4 position is G and the Ϫ3 position is A (1)), we investigated whether initiation could occur at this fourth AUG by disrupting the in-frame stop codon (TAA-TAT), which is also in-frame with the downstream open reading frame encoding the AT 1 R. A larger protein encompassing the extra 44 amino acids, however, was not detectable by IVT (data not shown).

FIG. 4. Comparison of WT and QM AT 1 R signaling.
A, dose-response curves for Ang II-stimulated IP accumulation in CHO cells transiently transfected with QM and WT (inset). The data are expressed as the amount of Ang II-stimulated IP accumulation, which is defined as the total levels of IP accumulated in the presence of Ang II minus basal levels over a 20-min period as a function of Ang II concentration. The data are representative of three experiments, each performed in triplicate. B, Ang II-stimulated IP accumulation in WT-1, WT-2, WT-3, QM-1, QM-2, and QM-3. The levels of Ang II-stimulated IP accumulation are shown after a 20-min incubation with 100 nM Ang II. The data are averaged from three experiments that were performed in quadruplicate on each clone. C, Ang II-stimulated IP production averaged from WT-1, WT-2, and WT-3 and from QM-1, QM-2, and QM-3. D, Ang II-stimulated IP production values normalized to AT 1 R B max levels WT-and QM-transfected cells. The data are calculated from the IP data in Fig. 4B and the B max data in Fig. 3C. AU, arbitrary unit.

Comparison of the Effects of DM2 and the Splice Variant E1,3 on AT 1 R Expression and IVT-
The two AT 1a R splice variants (E1,2,3 and E1,3) differ only in the length of their 5Ј-LS. Thus, both transcripts code for identical proteins. We recently showed that E2 contains an inhibitory RNA cis element, which results in reduced AT 1 R expression and signaling (5). As found previously, radioligand binding assays showed that AT 1 R densities in E1,3-expressing cells were 1.4-fold higher than WTexpressing cells, which were indistinguishable from DM2 (Fig.  6A). IVT assays showed a similar result; the E1,3 transcript was translated 2.0-fold more efficiently than WT, and no differences were observed between the levels of IVT for DM2 and E1,3 (Fig. 6B).

DISCUSSION
In this paper, we show that disruption of all four AUGs leads to marked increases in AT 1 R binding in both transiently transfected and stably transfected cells. ␤-galactosidase was equivalently expressed in WT and QM co-transfection experiments, and at least three independent and randomly selected stable clones of QM expressed higher levels of AT 1 R binding when compared with three independent and randomly selected WT clones; these findings rule out the likelihood that differences in transfection efficiencies or different sites of integration account for these results. Scatchard analysis of radioligand binding studies indicates that the increase in AT 1 R binding in cells expressing QM compared with cells expressing WT arises from an increase in receptor density rather than increased receptor affinity. This is consistent with the fact that WT and QM code for identical proteins. It is unlikely that the increase in AT 1 R densities in QM-expressing cells is due to increased QM mRNA levels, because AT 1 R densities were still significantly higher after normalization to AT 1 R mRNA levels in both transiently and stably expressing cells.
Ang II-stimulated IP production is markedly higher in QMcompared with WT-expressing cells, which illustrates the functional significance of the 2-fold increase in AT 1a R densities in QM-expressing cells. These data also support studies showing a close correlation between AT 1 R density and signal transduction in vascular smooth muscle cells (11). In response to Ang II stimulation, QM-expressing cells produce more IP than WTexpressing cells when normalized to AT 1 R B max , and this is consistent with the expected amplification of signaling that occurs in signal transduction cascades.
The observation that QM mRNA was associated with heavier polysome fractions (and thus with more actively translated mRNAs) than WT mRNA suggests that upstream AUGs inhibit translational efficiency in cells. This is further supported by IVT assays, in which QM is translated in vitro with greater efficiency than WT mRNA.
Mutagenesis studies show that disruption of both AUGs in E1 offers no liberation from the 5Ј-LS inhibition of AT 1 R expression. The data showing that AT 1 R densities in DM2-transfected cells were half the levels present in QM-transfected cells indicate that the upstream AUGs in E1 at least partially contribute to the RNA inhibitory cis elements within the 5ЈLS. It is possible that part of the effect of QM involves a conformational change in the 5Ј-LS that only occurs when all four AUGs are disrupted. Thus, even though disruption of the AUGs in E1 had no effect on AT 1 R binding or translation, their disruption contributed to more efficient translation of QM.
Disruption of the two AUGs in E2 completely relieved the effects of the E2 inhibitory cis element on receptor expression and signaling (5). This finding suggests that the third and fourth AUGs comprise a major component of the inhibitory cis element within E2 and thus may be the RNA cis elements that contribute to control of AT 1 R regulation by alternative splicing. By controlling the degree of alternative splicing of E1,2,3, an additional level of control is available by which the cell can tightly regulate the expression and function of the AT 1 R.
We recently showed by Northern blot analysis that all three exons are expressed in rat tissue and that both splice variants (E1,3 and E1,2,3) are expressed in all rat tissues studied thus far, indicating that the splice variants are not the result of cDNA errors due to incompletely spliced introns or reverse transcription-PCR errors (5). Furthermore, we found that splicing is regulated in a tissue-specific manner and that the splice ratio of E1,3 to E1,2,3 tightly correlates with tissue specific differences in AT 1 R expression, suggesting that regulation of alternative splicing contributes to differential tissue-specific expression of the AT 1a R (5). Thus, in addition to regulation by other mechanisms (such as transcription, mRNA stability, RNA binding proteins, receptor desensitization, ligand-mediated receptor internalization, and receptor recycling), alternative splicing of the AT 1 R is one more mechanism by which AT 1 R expression can be controlled (3,(12)(13)(14).
DM2 and E1,3 result in the same levels of AT 1 R expression, but this finding does not rule out the presence of additional inhibitory cis elements within E2; it is possible that inhibitory cis elements exist which are distinct from the two AUGs but are disrupted by forming DM2. In this regard, we have recently reported that deletion of the loop in a putative hairpin in E2 markedly relieved the translational repression of E2 (5).
In most mammalian mRNAs, the small (40 S) ribosomal subunit complexes with the 5Ј-end of the mRNA and then begins to scan linearly along the mRNA until it reaches the first AUG. At this point, the anticodon in Met-tRNA i base pairs with the AUG codon, the large 60 S ribosomal subunit joins the complex, and translation ensues. This "first AUG rule" of translation initiation can be escaped by several mechanisms (15). Context-dependent leaky scanning allows the ribosome to pass by the first AUG until it finds an AUG in optimal Kozak context (GCCACCAUGG; the AUG is italicized and the most crucial flanking sequences for optimal context are indicated by boldface type). The first three AUGs in the AT 1a R 5Ј-LS are not in optimal context and thus could be bypassed as initiation codons by the mechanism of leaky scanning. Studies suggest that AUGs close to the 5Ј-end of a mRNA are poorly recognized by ribosomes (16); Thus, it is likely that the two AUGs in exon 1 are too close to the 5Ј-end to sustain ribosomal initiation even if they were in optimal context. A third mechanism of escape is "reinitiation." Reinitiation at a downstream AUG in optimal context can occur after initiation at an upstream AUG when the upstream AUG is followed in-frame by a stop codon. All of the four AUGs in the AT 1a R 5Ј-LS are followed in-frame by a termination codon (Fig. 1).
A longer AT 1 R was not detected by IVT when the in-frame stop codon in E2 was disrupted, which suggests that the fourth upstream AUG inhibits AT 1a R expression by ribosomal pausing rather than by initiation, termination, and reinitiation or by another potential escape mechanism, internal ribosomal entry (17). In this regard, the rat AT 1a R appears distinct from the human AT 1 R, in which initiation can occur at an upstream AUG in the E1,3,4 splice variant, resulting in a longer form of the receptor that is functionally distinct from the short form (18). M2 caused a significant increase in IVT, whereas M1 or the two AUGs in E1 (DM1) did not, which suggests that ribosomal pausing is greatest at the fourth AUG and is consistent with the observation that the fourth AUG ( Fig. 1) is the only one in optimal Kozak consensus sequence (1). However, these studies do not rule out the possibility that initiation occurs at the fourth AUG but at a level that is undetectable under the assay conditions.
There is accumulating evidence that an unfavorable 5Ј-LS can serve a physiological purpose. It is well documented that upstream open reading frames inhibit translation of the major open reading frame in many key regulatory proteins such as growth factors, protooncogenes, transcription factors, and key signaling molecules (19,20). There are also several examples of how mutations affecting mRNA translational efficiency can contribute to pathological conditions. For example, a mutation in the 5Ј-LS that results in an out-of-frame AUG codon leads to reduced translation of the P16 tumor suppressor; accordingly, this mutation results in a predisposition toward melanoma (21). In another example, a splicing mutation that eliminates the upstream AUGs in the 5Ј-LS of thrombopoietin mRNA results in efficient translation of this protein and subsequent thrombocythemia (22). It is thus possible that aberrant splicing of the AT 1a R under certain pathological conditions could have profound repercussions on the renin angiotensin system and its control of blood pressure and fluid homeostasis.
In summary, these data indicate that upstream AUGs in both E1 and E2 in the 5Ј-LS of the AT 1a R act cumulatively to repress receptor expression and signaling by inhibiting trans-lation. The upstream AUGs in E1 require the presence of the upstream AUGs in E2 to be inhibitory, whereas the upstream AUGs in E2 are inhibitory in and of themselves. However, the inhibitory effects of AUGs in E2 are amplified by the presence of the upstream AUGs in E1, suggesting that 5Ј-LS secondary structure is also important to the translational repression. Translational repression by these upstream AUGs is most likely due to ribosomal pausing rather than reinitiation because a longer form of the AT 1 R was not detected when the in-frame stop codon in E2 was disrupted. In addition, these data suggest that the third and fourth upstream AUGs are part of the inhibitory cis element present in E2 and therefore may contribute to regulation of AT 1a R expression by alternative splicing.