Regulation of Na(+)-coupled glucose transport in LLC-PK1 cells. Message stabilization induced by cyclic AMP elevation is accompanied by binding of a M(r) = 48,000 protein to a uridine-rich domain in the 3'-untranslated region.

In an exploration of the molecular basis of cyclic AMP-induced stabilization of Na+/glucose cotransporter mRNA (SGLT1 isoform) accompanying cell differentiation in the pig kidney cell line LLC-PK1, we have identified a 48-kDa cytoplasmic protein factor, designated SG-URBP, which specifically binds a 120-nucleotide sequence within the 3′-untranslated region of the SGLT1 message. A 46-nucleotide uridine-rich element within this region appears necessary for specific binding, and the presence of the 3′-untranslated region is necessary for message stabilization by cyclic AMP. The binding activity of SG-URBP is up-regulated after cyclic AMP elevation and protein kinase A activation, whereas protein dephosphorylation either in vivo or in vitro is associated with loss of binding activity. The increase in SG-URBP binding activity correlates with an increase in the half-life of the SGLT1 message, suggesting a cause and effect relationship.

In an exploration of the molecular basis of cyclic AMPinduced stabilization of Na ؉ /glucose cotransporter mRNA (SGLT1 isoform) accompanying cell differentiation in the pig kidney cell line LLC-PK 1 , we have identified a 48-kDa cytoplasmic protein factor, designated SG-URBP, which specifically binds a 120-nucleotide sequence within the 3-untranslated region of the SGLT1 message. A 46-nucleotide uridine-rich element within this region appears necessary for specific binding, and the presence of the 3-untranslated region is necessary for message stabilization by cyclic AMP. The binding activity of SG-URBP is up-regulated after cyclic AMP elevation and protein kinase A activation, whereas protein dephosphorylation either in vivo or in vitro is associated with loss of binding activity. The increase in SG-URBP binding activity correlates with an increase in the half-life of the SGLT1 message, suggesting a cause and effect relationship.
Glucose transport by mammalian cells is a highly regulated process. Two structurally, functionally, and genetically distinct classes of glucose transporters have been identified, the GLUT family of facilitative glucose transporters (Gould and Bell, 1990) and the SGLT family which mediates Na ϩ -coupled secondary active transport of glucose. SGLT1 transporters are restricted in tissue expression to the apical membranes of small intestine and renal proximal tubule and have been mapped to human chromosome 22 (Wright, 1993). SGLT1 also transports galactose and has been identified as the locus of the inherited defect glucose-galactose malabsorption (Turk et al., 1991).
Much of what is currently known about regulation of the expression of the SGLT type of glucose transporter has been derived from studies using the cell line LLC-PK 1 , derived from porcine renal proximal tubule (Amsler and Cook, 1982;Peng and Lever, 1993). This cell line permits the reconstitution of a fully functional polarized epithelium in vitro. The development of a confluent monolayer is accompanied by the appearance of a number of differentiated functions characteristic of this epithelium including the formation of tight junctions and microvilli, vasopressin responsiveness, transepithelial salt and water transport, several brush-border marker enzyme activities, and apical membrane Na ϩ -coupled glucose transport activity (reviewed in Lever (1989)). Elevation of intracellular levels of cyclic AMP or treatment with the differentiation inducer hexamethylene bisacetamide (HMBA), 1 greatly increased both the levels of expression and rate of appearance of several of these differentiated functions including Na ϩ -coupled glucose transport (Amsler and Cook, 1982;Yoneyama and Lever, 1984;Peng and Lever, 1993;Yet et al., 1994). These agents appear to act by independent and synergistic mechanisms (Peng and Lever, 1993). SGLT1 transcripts of 2.2 and 3.9 kb, differing in the length of the 3Ј-untranslated region (3Ј-UTR) are observed in LLC-PK 1 cells as a result of alternative cleavage and polyadenylation (Ohta et al., 1990). We have recently demonstrated that the half-life of the larger transcript is increased by more than 8-fold after cyclic AMP elevation, indicating that post-transcriptional regulation exerts a major influence on SGLT1 expression (Peng and Lever, 1995). The observation that the 3Ј-UTR is necessary to observe this effect suggested that sequences within this region may regulate message stability in response to cyclic AMP.
In the present study, we extend these observations by identifying a cytoplasmic M r ϭ 48,000 mRNA-binding protein, designated SG-URBP, which binds a cognate sequence containing a uridine-rich element (URE) within the SGLT1 mRNA 3Ј-UTR. SG-URBP binding activity is greatly increased in extracts from LLC-PK 1 cell cultures after treatment with cyclic AMP-elevating agents and is dependent on protein phosphorylation. SG-URBP appears to be a new member in the inventory of RNA-binding proteins which have in common the specific recognition of uridine-rich stability-determining elements in mRNA (Belasco and Brawerman, 1993;Chen and Shyu, 1995) and the only one identified thus far in this category which exhibits regulation by cyclic AMP-dependent protein phosphorylation.
Construction of Plasmids Containing SGLT1 3Ј-UTR Domains-A renal SGLT1 cDNA clone (pPSGT-B1) kindly provided by Dr. David Rhoads (Massachusetts General Hospital, Boston, MA) contains a partial coding region (1818 bp) and the entire 3Ј-UTR (1831 bp) of pig kidney SGLT1 cDNA inserted in the pBluescript II SK(Ϫ) plasmid cloning vector (Stratagene) at the EcoRI site (Ohta et al., 1990). A 1824-base pair (bp) SmaI/NsiI fragment that contains 18 bp of the plasmid sequence and 1806 bp of the renal SGLT1 coding region was excised from pPSGT-B1, trimmed with T4 DNA polymerase, and religated by blunt-end ligation. The resulting plasmid, referred as pB1-3UTR1, contains the entire 1831-bp 3Ј-UTR of renal SGLT1 cDNA (nucleotides 1818 -3649) plus 12 bp of the coding sequence (1807-1818). In the following constructions, any 5Ј or 3Ј protruding ends, when necessary, were treated with either the Klenow enzyme or T4 DNA polymerase in preparation for blunt-end ligation (Maniatis et al., 1989). To produce plasmid pB1-3UTR2, a 1317-bp EcoRV/HindIII fragment (nucleotides 2263-3579) was excised from pPSGT-B1 and subcloned into the pBluescript II SK(Ϫ) vector between the SmaI and HindIII sites. Plasmid pB1-3UTR3, containing a 952-bp segment, (nucleotides 2698 -3649) was produced from pPSGT-B1 by deleting a 2725-bp SmaI/ StuI fragment containing 2712 bp of the SGLT1 sequence and 13 bp of the vector; pB1-3UTR4, containing a 553-bp segment from nucleotides 3097-3649, was generated from pPSGT-B1 by deleting a 3129-bp SmaI/ NheI fragment containing 3116 bp of the SGLT1 sequence and 13 bp of the vector. All constructs were verified using both restriction enzyme mapping and DNA sequencing.
Sense-strand RNA transcripts were synthesized using T3 RNA polymerase (Stratagene) with the exception of probes AU80, T3/T7 ␣-19⌬60, and GAPDH486 which were transcribed using T7 RNA polymerase (Stratagene). After transcription, RNase-free DNase (RQ1 DNase, Promega) was added and mixtures were incubated for an additional 30 min at 37°C to remove template DNA. Then, 20 g of yeast tRNA was added to each tube as carrier, followed by dilution with diethylpropylcarbonate-treated water to a final volume of 100 l. Aliquots of 1 l of each reaction mixture were removed and further diluted with 99 l of diethylpropylcarbonate-treated water for trichloroacetic acid precipitation (Maniatis et al., 1989) to calculate the amount of cRNA produced. Labeled transcripts were extracted with phenol/chloroform and precipitated twice with 2 M ammonium acetate in 2.5 volumes of ice-cold ethanol at Ϫ80°C for at least 1 h. The final pellet was washed with 80% ice-cold ethanol, air-dried, dissolved in diethylpropylcarbonate-treated water, stored at Ϫ20°C, and used as soon as possible. The integrity of the transcripts was verified by agarose electrophoresis.
To synthesize unlabeled transcripts in quantity for competition experiments, transcription reactions were performed by the same procedure described above except [ 32 P]UTP was replaced by 0.5 mM UTP, and the total volume of each reaction mixture was increased to 100 l.
Preparation of Cell Lysates for RNA-Protein Binding Assay-LLC-PK 1 monolayers grown in 6-cm culture dishes were washed twice with ice-cold phosphate-buffered saline and aspirated. Monolayers were lysed at 4°C by direct addition to each dish of 1 ml of lysis/extraction buffer containing 10 mM HEPES, pH 7.4, 3 mM MgCl 2 , 40 mM KCl, 2 mM dithiothreitol, 5% glycerol, 0.5% Nonidet P-40, 8 g/ml aprotinin, and 100 g/ml phenylmethysulfonyl fluoride for 15 min, as described by You et al. (1992). Nuclei and cell debris were removed by centrifugation at 10,000 ϫ g for 5 min at 4°C, and cytoplasmic lysates were aliquoted and stored at Ϫ80°C until use. Protein was determined using a microwave version of the bicinchoninic (BCA) method (Atkins and Tuen, 1992) modified for use with 96-well microtiter plates.
UV-cross-linking Assay of RNA-Protein Interaction-The procedure described by You et al. (1992) was used with minor modifications. LLC-PK 1 cytoplasmic lysates (10 -40 g of protein) and 32 P-labeled RNA probe were incubated in a 10-or 20-l reaction volume at room temperature for 15 min in a buffer containing 10 mM HEPES (pH 7.6), 3 mM MgCl 2 , 40 mM KCl, 2 mM dithiothreitol, 10% glycerol, and 0.5% Nonidet P-40. Heparin (10 mg/ml, final concentration) and yeast tRNA (200 g/ml, final concentration) were added to reduce nonspecific binding. Then, unbound RNA was digested by 5-10 units of RNase T1 for 20 to 60 min at either room temperature or 30°C, depending on the probe used. The RNA-protein mixtures containing equal amounts of cellular proteins were immediately transferred into wells of a 96-well microplate on ice and exposed to short-wave (254 nm) radiation in a UV Stratalinker 1800 (Stratagene) for 5 min on the automatic setting. The UV-cross-linked samples were then transferred into 1.5-ml microtubes, 1/3 volume of 4 ϫ SDS sample buffer was added, and tubes were heated in boiling water for 3 min. Samples were resolved by SDS-PAGE (10% acrylamide). The dried gel was exposed to x-ray film at Ϫ80°C with double intensifying screens.

RESULTS
Cyclic AMP Elevation Is Accompanied by Formation of a 48-kDa RNA-Protein Complex-Our previous findings that the 3.9-kb SGLT1 transcript, which contains a 1.8-kb 3Ј-UTR, was stabilized by more than 8-fold after cyclic AMP elevation, whereas the half-life of the 2.2-kb transcript, which lacks the 3Ј-UTR, was unaffected (Peng and Lever, 1995), suggested that the 3Ј-UTR contains sequences which regulate the stability of this message in response to cyclic AMP. In order to identify putative trans-acting factors that may stabilize this transcript after cyclic AMP elevation by specifically interacting with ciselements in the 3Ј-UTR, we produced by in vitro transcription four non-overlapping contiguous sense RNA probes (named 3UTR1, 3UTR2, 3UTR3, and 3UTR4) which encompassed the entire 3Ј-UTR of SGLT1 mRNA (Fig. 1). These probes were designed to segregate some of the sequence elements found in the SGLT1 3Ј-UTR which have been shown to regulate stability in other mRNAs (Belasco and Brawerman, 1993;Sachs, 1993). These include a 46-nt uridine-rich region (URE), which contains 68% uridine residues, present within 3UTR2, and three AUUUA motifs contained in probes 3UTR3 and 3UTR4, diagrammed in Fig. 1. RNA "band shift" assays using nondenaturing electrophoresis of the products of an RNA binding assay containing extracts from Nonidet P-40-solubilized LLC-PK 1 cells and the 32 P-labeled sense RNA UTR probes revealed a unique band shift pattern for each probe with a signal strength proportional to the amount of cellular extract added and abolished by proteinase K, indicating that the observed bands were due to interactions of proteins with the labeled RNAs (data not shown). However, only in the case of probe 3UTR2 was a difference in band shift pattern and intensity noted when extracts from control and IBMX-induced cells were compared (not shown).
This difference was more clearly apparent when RNA-protein incubations were UV-cross-linked after the RNase T1 digestion step to generate covalent bonds between the 32 P-labeled sense RNA probe and associated proteins. The products were analyzed by SDS-PAGE in a label transfer experiment (Fig. 2). A 28-kDa RNA-protein complex was observed in samples derived from control cultures (lane 2) whereas samples derived from IBMX-induced cultures (lane 5) exhibited a 48-kDa RNA-protein complex accompanied by a loss of the 28-kDa complex. The presence of a 250ϫ molar excess of specific competitor RNA (unlabeled 3UTR2 RNA) effectively blocked the appearance of these complexes (lanes 3 and 6), while addition of the same amount of nonspecific competitor RNA (a 529-bp unlabeled antisense glyceraldehyde-3-phosphate dehydrogenase RNA) had much less effect (lanes 4 and 7), indicating that these binding activities are specific. The formation of the 48-kDa complex was proportional to the amount of cellular extract added to the binding reaction and abolished if the binding reaction mixture was treated with proteinase K (Fig. 3). These results demonstrate the presence of a trans-acting protein factor in LLC-PK 1 cells which specifically interacts with the 3UTR2 mRNA fragment in response to IBMX treatment.
The formation of the 48-kDa complex was also induced by other cAMP-elevating agents such as 8-bromo-cAMP, dibutyryl cAMP, and forskolin (Fig. 4, lanes 4, 5, and 6). The IBMXinduced complex formation was reduced substantially when LLC-PK 1 cells were exposed to H-89 (lane 3), a specific protein kinase A inhibitor (Chijiwa et al., 1990). Addition of the protein kinase C inhibitor H-7 (Hidaka et al., 1984) to cells produced a marginal inhibitory effect on the IBMX-induced 48-kDa complex (lane 8). These results strongly suggest that protein kinase A is involved in the regulation of the formation of the 48-kDa RNA-protein complex.
The differentiation inducer HMBA also promoted an increase in the half-life of the 3.9-kb SGLT1 message (Peng and Lever, 1995). HMBA, 5 mM, also induced an increase in the 48-kDa RNA-protein cross-linked complex (Fig. 4, lane 7) which was diminished by H-89 but not by H-7 (not shown), suggesting a possible role of cyclic AMP-protein kinase A in mediating this response.
A Phosphoprotein Is Required for Formation of the 48-kDa Complex-The possibility that the binding affinity of these protein factors for their cognate sites on the mRNA might be regulated by a reversible phosphorylation/dephosphorylation mechanism was tested. Cytoplasmic extracts from either IBMX-treated or untreated control cells were treated with potato acid phosphatase for 5 min to remove O-linked phosphates prior to binding assay. As shown in Fig. 5, phosphatase treatment resulted in a decrease in formation of both the 48-kDa and 28-kDa RNA-protein complexes. The presence of microcystin LR, a specific, potent inhibitor of phosphatase types I and 2A (Mackintosh et al., 1990), partially protected the RNAprotein complexes from acid phosphatase attack (Fig. 5, lanes  6 -8) indicating that inhibition of complex formation was not FIG. 1. Restriction map of SGLT1 3-UTR cDNA. The region encompassed by plasmid pB1-3UTR1 is shown. Regions encoding the four sense RNA probes used in this study are indicated (3UTR1-4). The thin lines represent the sense SGLT1 RNA probes produced by in vitro transcription from plasmid constructs. A uridine-rich region (URE) and three AUUUA motifs are indicated in the diagram.
FIG. 2. IBMX treatment induces a 48-kDa RNA binding activity specific to 3UTR2 RNA. Radiolabeled sense RNA probe 3UTR2 was incubated with lysates from control (lanes 2-4) or IBMX-treated (lanes 5-7) LLC-PK 1 cells in the absence (lanes 2 and 5) or presence of a 250ϫ molar excess of either specific competitor (SC, lanes 3 and 6) or nonspecific competitor RNA (NC, lanes 4 and 7). Following RNase T1 digestion, samples were UV-cross-linked and resolved by 10% SDS-PAGE. due to another contaminating enzymatic activity. The contribution of protein phosphorylation was also tested in vivo by addition of the specific phosphatase inhibitor, okadaic acid (Bialojan and Takai, 1988) to cells. Results shown in Fig. 6 indicate that okadaic acid, 2 M, in the absence of IBMX produced the same increase in the formation of the 48-kDa RNAprotein complex as that produced by IBMX, but produced only a small additional increase in IBMX-treated cells. These observations suggest that the protein factor involved in the formation of the 48-kDa complex is regulated by protein phosphorylation and dephosphorylation.
A Uridine-rich Region Is Necessary for Formation of the 48-kDa RNA-Protein Complex-Since the 435-nt 3UTR2 fragment contains a 46-bp uridine-rich region (68% U), at the 3Ј portion (Fig. 1), the possibility that this U-rich element is a binding site for the protein factor forming the 48-kDa complex was examined. Results shown in Fig. 7 revealed that poly(U) (lane 8), but not poly(A), poly(G), or poly(C) (lanes 7, 9, and 10), competitively inhibited the formation of the 48-kDa RNA-protein complex, indicating that complex formation is uridine-dependent. Other U-rich RNAs were effective competitors. A synthetic 80-nt RNA (AU80) derived from a 63-nt plasmid T3/T7 ␣-19 sequence containing four reiterated AUUUA motifs (Malter, 1989) and a wild-type c-fos ARE RNA fragment (AREwt) which contains AUUUA elements with surrounding U-rich sequences (Shyu et al., 1989) were used as unlabeled competitors. As shown in Fig. 7, formation of the 48-kDa complex was inhibited in the presence of a 250-fold excess of either 3UTR2 RNA (lane 3), AU80 RNA (lane 4), or ARE-wt RNA (lane 6), but not by a 64-nt T3/T7 ␣-19 RNA (lane 5) which is same as AU80 except it lacks the four AUUUA motifs. Since 3UTR2 RNA does not contain an AUUUA motif, the specific competition of its protein binding by these AUUUA-containing RNAs would be attributed to their U-rich property.

32
P-labeled 3UTR2 RNA was incubated with increasing amounts of cytoplasmic extracts from either control (lanes 2-5) or IBMX-treated (lanes 6 -9) LLC-PK 1 cells and analyzed by the UV-cross-linking assay. B, cytoplasmic lysates were preincubated with the indicated amounts of proteinase K at 37°C for 30 min prior to the UV-cross-linking assay. To test whether AU80 or c-fos ARE RNA interacts with the same protein factor that forms the 48-kDa complex with SGLT1 3UTR2, cytoplasmic extracts from IBMX-treated LLC-PK 1 cells were incubated with radiolabeled AU80 or c-fos ARE RNA and then a UV-cross-linking experiment was performed. As shown in Fig. 8, in the case of AU80, 3 RNA-protein complexes, with apparent molecular masses of 63, 54, and 40 kDa, respectively, were observed (lane 4). Interestingly, complexes of 63 and 54 kDa were also observed to be cross-linked to SGLT1 probes 3UTR3 and 3UTR4 (not shown), which contain AUUUA motifs (Fig. 1). The c-fos ARE RNA formed a 42-kDa crosslinked complex (lane 5). However, none of these RNA segments formed the 48-kDa complex observed with SGLT1 3UTR2 RNA (lane 3). Although the possibility exists that small differences in apparent molecular mass of each complex may reflect a contribution from different sizes of the residual RNA fragment cross-linked in each case rather than differences in the identity of the protein, this finding suggests that neither AU80 nor c-fos ARE RNAs contain the cognate sequence recognized by protein(s) forming the 48-kDa complex with SGLT1 3UTR2.
Cognate Sequence Recognized by mRNA-binding Proteins-A series of deletions from either the 3Ј or the 5Ј end were constructed within the 435-nt region encompassed by the 3UTR2 probe in order to identify specific sequences involved in protein binding. 32 P-labeled sense RNA transcripts were produced from these constructs and used for UV-cross-linking assay. The results shown in Figs. 9 and 10 indicate that transcripts 3UTR2⌬2, -⌬3, and -⌬4, which lack the U-rich element (URE) (Fig. 10), did not form either the 48-kDa or the 28-kDa complex (Fig. 9, lanes 4, 5, and 6). This indicates that the binding sites for both the 28-kDa and the 48-kDa protein factors must be in a region between the NspI site and the StuI site. A 5Ј deletion up to the EcoO109I site, represented by probe 3UTR2⌬5, retained the same binding activity for both the 28-kDa and the 48-kDa complexes (Fig. 9, lanes 7 and 11) as the full-length 3UTR2 probe (lane 2). Probe 3UTR2⌬7, a 120-nt AvaII/StuI fragment, which contains the entire 46-nt URE, exhibits full ability to form the 28-kDa and 48-kDa complexes (Fig. 9, lanes  9 and 13); therefore, the binding site for both complexes must reside within this 120-nt region. 3UTR2⌬7 was cut at a unique HincII site within the U-rich sequence to produce probes 3UTR2⌬6 and 3UTR2⌬8 (Fig. 10). Probe 3UTR2⌬6, which contains the region upstream of the HincII site, conferred a very weak ability to form either the 28-kDa or 48-kDa complex (Fig.  9, lanes 8 and 12), while probe 3UTR2⌬8, which contains the region downstream of the HincII site, could not form the 28-kDa complex and exhibited substantially reduced formation of the 48-kDa complex (lanes 10 and 14). Taken together, these data strongly indicate that the entire U-rich element (URE) is necessary for effective formation of both the 28-kDa and the 48-kDa complexes with SGLT1 mRNA, and binding is abolished if it is cut at the HincII site. The observation that increased 48-kDa binding activity was consistently accompanied by a reduced 28-kDa binding activity, yet both apparently map to the same 120-nt sequence (shown in Fig. 10C), suggests that both binding activities may recognize either the same site or FIG. 6. In vivo inhibition of phosphatases stimulates the formation of the 48-kDa complex in uninduced control cell extracts. Cytoplasmic lysates from control or IBMX-treated cells as indicated were prepared at the indicated times after addition of 2 M okadaic acid (OA) and analyzed by UV-cross-linking using radiolabeled 3UTR2 (1 ng). Lane 1, free probe (FP).
FIG. 7. Effect of uridine-rich RNA competitors on the formation of the 48-kDa complex. A cytoplasmic lysate (30 g) from IBMXtreated LLC-PK 1 cells was preincubated for 10 min with a 250ϫ molar excess of the indicated unlabeled RNAs (lanes 3-10) prior to the addition of radiolabeled 3UTR2 RNA and UV-cross-linking assay. Lane 1, free 32 P-labeled 3UTR2 probe without lysate; lane 2, no unlabeled RNA added. SC, specific competitor RNA (unlabeled 3UTR2 RNA).
FIG. 8. Specificity of the 48-kDa RNA binding activity. Cytoplasmic lysates (30 g) from IBMX-treated cells were incubated with 32 Plabeled 3UTR2, AU80, or c-fos ARE sense RNA probes (1 ng of each) as shown and analyzed by the UV-cross-linking assay. Lane 1, free 32 Plabeled 3UTR2 probe without lysate; lane 2, 32 P-labeled 3UTR2 RNA incubated with control lysate. different, partially overlapping sites, and their binding activities are reciprocally regulated by protein kinase A activation.

DISCUSSION
The mechanisms responsible for maintenance of the differentiated phenotype in renal epithelial cells are poorly understood. Expression in cell culture of the high affinity Na ϩ /glucose cotransporter (SGLT1), a marker for late proximal tubule (S3), is regulated by cell density, differentiation inducers, as well as a number of agonists of intracellular signaling pathways. We have recently demonstrated that post-transcriptional stabilization of the SGLT1 message plays a prominent role in differentiation-associated expression of this transporter induced by either cyclic AMP or the differentiation inducer HMBA (Peng and Lever, 1995). Message stabilization required the presence of the 3Ј-UTR, suggesting that cis-acting sequences in this region were involved. This mechanism would permit the cell to increase rapidly its commitment to synthesize a specialized end product of differentiation in response to various extracellular signals.
Our present results show that the 3Ј-UTR of porcine renal Na ϩ /glucose cotransporter SGLT1 mRNA contains a U-rich region (URE) that exhibits a specific, cyclic AMP-dependent and protein phosphorylation-dependent interaction with a cytoplasmic protein of apparent M r ϭ 48,000 which we have named SG-URBP. The simplest hypothesis to explain our results is that SG-URBP is a phosphoprotein in its activated state, and its binding affinity to SGLT1 mRNA is directly activated by protein kinase A-mediated phosphorylation. Alternately, protein kinase A may be indirectly involved via a kinase cascade nor can we rule out the possibility that phosphorylation of another protein may be involved in activation of SG-URBP. The active phosphorylated form of SG-URBP would be hydrolyzed to an inactive (or low affinity) dephosphorylated form by an unidentified protein phosphatase. Our UV-crosslinking assay is only capable of assaying SG-URBP in its activated, RNA binding state, and, in the absence of specific antibodies, the putative inactive form cannot be measured. Direct demonstration of the mechanism will require purification of SG-URBP.
We have identified a 120-base region within the 3Ј-UTR of the renal SGLT1 message (Fig. 10C) which specifically interacts with SG-URBP to form the 48-kDa RNA-protein complex. This region contains a 46-nt U-rich element (68% U), and binding activity is effectively competed only by poly (U), unlabeled 3UTR2 RNA, and other U-rich RNA sequences, but not by poly(A), poly(G), and poly(C), or irrelevant RNA sequences which contain no U-rich element. Furthermore, cleavage of this segment within the U-rich region abolished the ability to form the complex. These data suggest that the U-rich element may be involved directly in RNA-protein interaction. Sequence data for this region of the 3Ј-UTR of human, rabbit, or rat SGLT1 are not available for comparison with the pig sequence in order to determine whether this URE motif is conserved across species.
Uridine-rich elements involved in destabilization of several nuclear transcription factor, cytokine, or lymphokine mRNAs have been divided into two broad categories (Chen and Shyu, 1994): those containing 1-3 copies of AUUUA motifs within the U-rich region, in some cases forming overlapping copies of the FIG. 9. Mapping of the protein binding sites in the 3-UTR region of SGLT1 3UTR2 RNA. A series of 32 P-labeled sense RNA probes representing different regions of 3UTR2 (see Fig. 10) were transcribed and incubated with cytoplasmic lysates. Lane 1, free 32 P-labeled 3UTR2 probe without lysate (FP); lanes 2-6, a set of 3Ј-deleted 3UTR2 probes incubated with lysates from IBMX-induced cells; lanes 7-14, a set of 5Ј-deleted 3UTR2 probes incubated with lysates from uninduced control (lanes 7-10) or IBMX-induced (lanes 11-14) cells. UUAUUUA(U/A)(U/A) nonamer, and those which lack the AUUUA motif. The SGLT1 URE falls within the latter category since it lacks an AUUUA motif within the U-rich element. Although three AUUUA motifs are present further downstream (Fig. 1), they do not participate in SG-URBP binding.
The functional significance of SG-URBP binding to the Urich mRNA sequence was not directly demonstrated in the present study. SG-URBP binding activity was activated by cyclic AMP-elevating agents and by phosphatase inhibitors. The increase in half-life (8.6-fold) of SGLT1 3.9-kb mRNA after cyclic AMP elevation (Peng and Lever, 1995) correlates well with the 5-10-fold increased formation of the SG-URBP complex under parallel conditions (Table I). Our previous results demonstrated that inhibitors of protein synthesis did not block stabilization of the message after cyclic AMP elevation (Peng and Lever, 1995). We propose that the post-transcriptional stabilization of SGLT1 mRNA in response to agents that elevate intracellular cAMP levels is mediated, at least in part, by the interaction between SGLT1 mRNA and its specific binding protein, SG-URBP. The ability of SG-URBP to bind its target site on the mRNA would be activated by protein phosphorylation, directly or indirectly mediated by protein kinase A. LLC-PK 1 cells express two distinct but similar protein kinase A catalytic subunits encoded by different genes and both R I and R II regulatory subunits (Adavani et al., 1987). A protein kinase A-deficient mutant of this cell line was deficient in Na ϩ /glucose cotransporter expression in addition to alterations in other differentiated properties (Amsler et al., 1991). The U-rich sequence (URE) identified in this study encompasses the specific binding site for SG-URBP. This sequence may serve as a component of the degradation machinery and participate in the selective degradation of SGLT1 mRNA. It is downstream from the first polyadenylation site suggesting it may function in the nucleus as well as the cytoplasm. The URE may be a nuclease hypersensitive site or a binding site for a destabilizing factor which directs nucleases to the region. A potential hairpin loop region occurs upstream from the URE but does not appear to participate in SG-URBP binding. Presumably, the binding of phosphorylated SG-URBP would mask the U-rich sequence and protect the message from degradation.
U-rich elements found within the 3Ј-UTRs of some shortlived mRNAs such as c-fos and c-myc play an important role in regulation of turnover of these mRNAs (Belasco and Brawerman, 1993;Chen and Shyu, 1995). Several U-rich element binding activities have been identified. These include a 59-kDa binding activity that interacts with a c-myc ARE in Balb/c3T3 cells (Alberta et al., 1994), a 37-kDa serum-inducible, translation-dependent protein that recognizes the c-fos ARE in NIH 3T3 cells (You et al., 1992), and a human elav-like neuronal protein 1 (Hel-N1) which specifically binds to a U-rich region in the 3Ј-UTR of c-myc mRNA (Levine et al., 1993) and forms a 28-kDa RNA-protein complex. We have observed a 28-kDa binding activity which also interacts with the URE of the SGLT1 3Ј-UTR. Whether this 28-kDa protein is a kidney cell counterpart of Hel-N1 or a different protein remains unknown. We observed that an increase in the 48-kDa SG-URBP binding activity after cyclic AMP elevation was consistently accompanied by a corresponding reduction in the formation of the 28-kDa complex. Both binding activities recognize the same 120base region and are competed by U-rich RNAs. Therefore, these two RNA binding activities are reciprocally regulated by cyclic AMP and may share a common binding site.
Several AU-rich mRNA-binding protein activities can be modulated by reversible phosphorylation/dephosphorylation. The adenosine-uridine binding factor AUBF (Malter 1989) is a phosphoprotein which specifically binds to AUUUA motifs of several lymphokine, cytokine, and oncogene mRNAs (Gillis and Malter, 1991) after activation of protein kinase C by TPA and is proposed to contribute to agonist-induced stabilization of these labile mRNA species (Malter and Hong, 1991;Rajagopalan and Malter, 1994). Two cytoplasmic proteins of 65 and 45 kDa reported to stabilize prostaglandin endoperoxide synthase II (COX II) mRNA in response to IL-1b and TPA in renal mesangial cells specifically bind the 3Ј-UTR in a phosphorylation-dependent manner (Srivastava et al., 1994).
A group of AU-rich element (ARE) mRNA-binding proteins in the M r range of 32,000 -40,000 (Brewer, 1991;Vakalopoulou et al., 1991;Bohjanen et al., 1991Bohjanen et al., , 1992Bickel et al., 1992;Port et al., 1992;Huang et al., 1993) bind to target mRNAs containing both AUUUA pentameric motifs and flanking U-rich domains. Within this category is a 35-kDa protein up-regulated by isoproterenol which has been implicated in agonist-mediated destabilization of ␤-adrenergic receptor mRNA (Huang et al., 1993). However, AUUUA motifs are neither always necessary nor always sufficient to target an mRNA for degradation (Shyu et al., 1989;Kabnick and Housman, 1988;You et al., 1992;Alberta et al., 1994). Comparative analysis has found an obvious correlation between mRNA half-life and uridine content in the immediate vicinity of the AUUUA motif (Alberta et al., 1994). It appears that U-richness is the most prominent characteristic of these AU elements. Binding of AUBF, a 44-kDa protein, to an artificial 80-bp reiterating AUUUA RNA transcribed from probe AU80, was stimulated in extracts from cyclic AMP analog-treated adipocytes (Stephens et al., 1992), but enhanced binding to the naturally occurring message was not demonstrated. Our results suggest that AU80 and c-fos ARE RNAs do not specifically interact with SG-URBP even though both can effectively compete with the binding of SG-URBP to SGLT1 3Ј-URE mRNA. Since SG-URBP differs from other known mRNA-binding proteins not only in apparent molecular size, but also in activation by agents that elevate intracellular cyclic AMP, it is probably a new member of the U-rich region mRNA-binding protein inventory.
At present, we do not know if SG-URBP can also interact with a subset of mRNAs other than SGLT1 to coordinately influence their stability in response to cyclic AMP. Such a possibility would extend the Second Messenger Hypothesis for cyclic AMP-mediated regulation to the post-transcriptional level and involve RNA-binding phosphoproteins as components of this intracellular cascade mechanism.  Peng and Lever (1995). b UV-cross-linking assay was carried out in parallel using lysates from late-confluent cells treated with HMBA or IBMX for 4 days, and levels of the 48-kDa complex were estimated by scanning densitometry. Values are expressed as fold increase relative to untreated controls.