Transforming Growth Factor- (cid:98) 1 Stimulates Multiple Protein Interactions at a Unique cis -Element in the 3 (cid:42) -Untranslated Region of the Hyaluronan Receptor RHAMM mRNA*

The receptor for hyaluronan mediated motility (RHAMM) gene expression is markedly elevated in fibrosarcomas exposed to transforming growth factor- (cid:98) 1 (TGF- (cid:98) 1 ). The half-life of RHAMM mRNA was increased by 3 fold in cells treated with TGF- (cid:98) 1 , indicating that growth factor regulation of RHAMM gene expression at least in part involves a posttranscriptional mechanism. Our studies demonstrated that a unique 30-nucleotide (nt) region that has three copies of the sequence, GCU-UGC, was the TGF- (cid:98) 1 -responsive region in the 3 (cid:42) -un-translated region (3 (cid:42) -UTR) that mediated message stability. This region interacted specifically with cyto- plasmic trans -factors to form multiple protein complexes of approximately 175, 97, 63, 26, and 17 kDa post- TGF- (cid:98) 1 treatment, suggesting a role for these complexes in the mechanism of action of TGF- (cid:98) 1 -induced message stabilization. Insertion of the 3 (cid:42) -UTR into the chloramphenicol acetyltransferase gene conferred TGF- (cid:98) 1 in- duced stability of chloramphenicol acetyltransferase-hybrid RNA in stably transfected cells, while the same insert carrying a deletion containing the 30-nt region had no significant effect on mRNA stability. These re- sults provide a model of RHAMM 0.5 m M EDTA) and lysed by repetitive cycles of freeze-thaw. Nuclei and cytoplasmic extracts were obtained (5, 8), and protein concentrations were determined according to previously described methods (16) and imme-diately frozen on dry ice and stored at (cid:50) 70 °C. In Vitro Transcription— A 1.9-kilobase pair RHAMM cDNA clone (1),

RHAMM 1 is a cell membrane receptor for hyaluronan (HA)mediated motility, which has no primary sequence similarity to other HA-binding proteins, or to CD44, another commonly expressed HA receptor (1,2). RHAMM expression is elevated in ras-transformed cells, and in these cells it mediates increased cell locomotion through a pathway involving tyrosine phosphorylation within focal adhesions of lamellipodia (3). RHAMM is also regulated by growth factors such as TGF-␤ 1 , and its expression in ras-transformed cells is necessary for TGF-␤ 1 stimulation of cell motility (4). A better understanding of the process of cell locomotion and RHAMM-regulated transformation requires more information about the mechanisms that control RHAMM gene expression.
The regulation of mRNA stability has emerged as an important control mechanism of gene expression. The altered stability of mRNAs is regulated by interactions among trans-acting factors and cis-elements often located within the 3Ј-UTRs of mRNAs (5)(6)(7)(8)(9). Growth factors such as TGF-␤ 1 exert their effects by binding to specific cell surface receptors (10,11), and altering the expression of a discrete set of genes important for the regulation of a variety of characteristics, including cell proliferation and locomotion (4,12). These TGF-␤ 1 effects on gene expression may be mediated, at least in part, posttranscriptionally through message stability alterations (13,14). For example, we have demonstrated that the TGF-␤ 1 -increased elevation in ribonucleotide reductase R2 mRNA levels, at least partly, is due to the stabilization of R2 mRNA (15) and is mediated by a 9-nt cis-element within the 3Ј-UTR (16). In this study we demonstrate that RHAMM mRNA forms specific complexes with multiple cytosolic proteins. The data suggest that the RHAMM mRNA-protein interactions play a critical role in a mechanism that leads to the TGF-␤ 1 -induced stabilization of RHAMM mRNA.

EXPERIMENTAL PROCEDURES
Cell Culture-C1, C2, and C3 fibrosarcomas derived from 10T 1 ⁄2 cells following T24 Ha-ras transfections (17), and cells transfected with chloramphenicol acetyltransferase (CAT) hybrid RHAMM plasmids were routinely cultured in a ␣-minimal essential medium (Flow Laboratories) supplemented with antibiotics and 10% (v/v) fetal bovine serum (18). For investigating TGF-␤ 1 effects on RNA levels and RNAprotein binding activity, cells were grown overnight in a serum-free medium containing 0.4 mg of transferrin and 0.2 mg of insulin (Sigma-Aldrich Canada) in 100 ml of ␣-MEM. TGF-␤ 1 (R & D Systems Inc.) was dissolved in 1.0 mg/ml bovine serum albumin, and 4 mM HCl and was added at predetermined times. The control cells received 1.0 mg/ml bovine serum albumin, 4 mM HCl, 0.4 mg of transferrin and 0.2 mg of insulin in 100 ml of the serum-free medium. The cells were harvested from the tissue culture plates with 0.3% buffered trypsin solution after centrifugation, washed once in phosphate-buffered saline (pH 7.2), and transferred to Eppendorf tubes.
Preparation of Protein Extracts from the Cytosol and the Nucleus-Cells transferred to Eppendorf tubes were briefly centrifuged for 1 min and resuspended in hypotonic buffer (25 mM Tris-HCl (pH 7.9), 0.5 mM EDTA) and lysed by repetitive cycles of freeze-thaw. Nuclei and cytoplasmic extracts were obtained (5,8), and protein concentrations were determined according to previously described methods (16) and immediately frozen on dry ice and stored at Ϫ70°C.
In Vitro Transcription-A 1.9-kilobase pair RHAMM cDNA clone (1), * This study was supported by grants from the National Cancer Institute of Canada (to J. A. W. and E. A. T.). 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 1 The abbreviations used are: RHAMM, receptor for hyaluronan mediated motility; HA, hyaluronan; UTR, untranslated region; TGF-␤ 1 , transforming growth factor-␤ 1 ; CAT, chloramphenicol acetyltransferase; nt, nucleotide; PCR, polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis.
containing part of the coding region and the entire 3Ј-UTR, was used as a template to generate PCR-amplified products. These corresponded to different segments of the 3Ј-UTR as shown in Fig. 2A, which spanned the following regions from nucleotides, 2082-3094 (full-length 3Ј-UTR), 2082-2288, 2082-2406, 2082-2490, 2394 -2435, and 2444 -3094 (2). To facilitate cloning, the oligonucleotide primers (20 -40 bases) complementing these regions were flanked at their 5Ј and 3Ј ends, by BamHI and EcoRI restriction sites, respectively. PCR parameters were: denaturation at 94°C for 4 min, denaturation at 94°C for 45 s, annealing at 55°C for 45 s, and extention at 72°C for 2 min. Thirty cycles were used with a final extension time of 7 min. The reaction mixture contained 50 ng of DNA template, 40 pmol of DNA primers, and 5 units of Taq DNA polymerase in a total volume of 100 l. The samples were gel-purified and subsequently extracted once with phenol/chloroform/isoamyl alcohol (25:4:1) and once with chloroform, precipitated in ethanol, resuspended in a suitable amount of Tris-EDTA (pH 7.5), and digested with the restriction endonucleases BamHI and EcoRI. Digests were heatinactivated at 65°C for 15 min and ligated into the pSPT18 in vitro transcription plasmid (Boehringer Mannheim) to generate a plasmid construct.
In brief, run-off RNA transcripts were produced by T7 polymerase activity (Boehringer Mannheim) from 1 g of digested pSPT18 RHAMM cDNA plasmid constructs. These constructs were linearized with the HindIII restriction enzyme, at a unique site in the polylinker of the pSPT18 plasmid. These digested plasmids were then used for in vitro transcription. RNA transcripts were produced as described previously (5,6) and extracted to a specific activity of approximately 3 ϫ 10 8 cpm/g RNA (16).
3Ј End Labeling of Synthetic RNA Oligonucleotides-The RNA oligonucleotides purchased from Dalton Chemical Laboratories Inc., Toronto, Canada, such as the 30-nt cis-element sequence and the 21-nt GCUUGCU repeat, were synthesized on the Applied Biosystems model 392 DNA/RNA synthesizer using phosphoramidite RNA monomers (user Bulletin No. 47, Applied Biosystems). The ribonucleotides were labeled at their 3Ј ends by a previously described method (19).
RNA Mobility Shift Assay-Binding reactions were performed as described previously (5, 6), with 10 -40 g of cytosolic or nuclear protein extract and 50 ϫ 10 3 cpm of 32 P-labeled RNA transcript or oligonucleotides. However, 50 units of RNase T1 (Boehringer Mannheim) were added when the RHAMM 3Ј-UTR was used (but not for binding reactions with the 3Ј end-labeled oligoribonucleotides).
UV Cross-linking Analyses-RNA-protein binding reactions were carried out as described above. Following the addition of heparin, samples were added to microtiter wells placed on ice and UV-cross-linked for 10 min in a Stratalinker Chamber (Stratagene), at a 250-mJ energy level. The samples were separated by electrophoresis on a 10% SDSpolyacrylamide gel (20 -22).
Northern Blot Analysis and Half-life Measurements-Total cellular RNA was extracted from TGF-␤ 1 -treated (10 ng/ml) and -untreated fibrosarcoma transfectants by a rapid method of RNA preparation (23). For half-life measurements, the cells were stimulated with the growth factor (TGF-␤ 1 ) for 2 h followed by addition of 10 g/ml actinomycin D. RNA was isolated at different times and analyzed for CAT/RHAMM 3Ј-UTR hybrid mRNA. The blots were prehybridized and probed with 32 P-labeled 1.9-kilobase pair EcoRI fragment from the RHAMM cDNA (1) and 32 P-labeled HindIII-BamHI fragment of CAT cDNA derived from the plasmid pSVLacOCAT (16,24), and mRNA half-life was esti-mated as described previously (5,6). Determination of sample loading was performed by probing with glyceraldehyde-3-phosphate dehydrogenase cDNA (6,25).
Plasmid DNA Constructs and Transfection-The plasmid pCAT was prepared by cloning a 715-bp HindIII-BamHI CAT encoding gene fragment derived from the plasmid pSVLacOCAT (24), blunt-ended, and inserted into the BglII site in the polylinker of the pECE expression vector (26). For the preparation of the pCATRH1 hybrid construct, a 1.01-kilobase pair full-length RHAMM 3Ј-UTR PCR-amplified product with SalI and XbaI overlap extensions at the 5Ј and 3Ј ends, respectively, was ligated into the corresponding sites in the pCAT construct. Based on the pCATRH1 construct, pCATRH2 was constructed by digestion at the unique restriction enzyme sites, NcoI (nt position 2377) and BclI (nt position 2671), to delete a 295-bp fragment that contained the putative cis-element. pCATRH3 was generated by digesting the RHAMM 3Ј-UTR at nucleotide position 2684 with the restriction enzyme, AhaIII to remove a 450-bp fragment downstream of the ciselement region. C3 fibrosarcoma cells were stably co-transfected with these plasmid DNA constructs and the selectable marker plasmid py3, the hygromycin gene, at a 30:1 ratio by the calcium phosphate method as described previously (27). Stably transfected cells were selected with 200 g/ml hygromycin, starting at 48 h after the addition of the plasmid calcium phosphate precipitate to the cells. Northern blot and CAT assays were performed to screen permanently transfected cell lines (28).

Effect of TGF-␤ 1 on RHAMM Gene Expression-Northern
blot analysis using the 1.9-kilobase pair EcoRI fragment of RHAMM as a cDNA probe (1) showed that RHAMM message levels in C3 fibrosarcoma cells were significantly elevated during a 3-20-h treatment time period with 10 ng/ml TGF-␤ 1 (Fig.  1A). To determine whether or not the TGF-␤ 1 effect on RHAMM mRNA steady-state level results from, at least in part, changes in RHAMM message stability, Northern blot analysis of RHAMM mRNA level was performed using the total RNA obtained from TGF-␤ 1 -treated and untreated cells, in which transcription was blocked by actinomycin D (15,16). Fig.  1B shows that there was a 3-fold increase in RHAMM message half-life in TGF-␤ 1 -treated cells in comparison with untreated cells.
Identification of Proteins Forming Complexes with a Unique cis-Element in the 3Ј-UTR of RHAMM mRNA-The knowledge that the 3Ј-UTR of other mRNAs have previously been implicated in the regulation of growth factor alterations of message stability (5,6,(13)(14)(15)(16) led us to test the hypothesis that TGF-␤ 1 -responsive trans-acting factors exist that can bind to a ciselement(s) of RHAMM mRNA 3Ј-UTR. RNA mobility shift assays were performed using radiolabeled in vitro transcribed RHAMM mRNA transcripts corresponding to the full-length 3Ј-UTR. As demonstrated in Fig. 2A, the 3Ј-UTR showed multiple RNA-protein interactions post-TGF-␤ 1 treatment. To   FIG. 1. Effect of TGF-␤ 1 on RHAMM gene expression. A, RHAMM mRNA levels in C3 fibrosarcomas incubated in the absence or presence of 10 ng/ml TGF-␤ 1 for 3, 6, 10, and 20 h. Total cellular RNA (60 g) blots were also probed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (cDNA) as a control for loading. Similar results were obtained in two independent experiments. B, C3 fibrosarcoma cells were incubated with actinomycin D (10 g/ml) to block transcription, in the absence (Ϫ) or presence (ϩ) of 10 ng/ml TGF-␤ 1 for the indicated time periods (hours). Autoradiographs of the Northern blots were quantitated and the results of the mRNA half-lives (t 0.5 ) were determined as described previously (5,6).
identify the sequence within the 3Ј-UTR that forms the RNAprotein complexes, we generated a series of in vitro transcribed fragments corresponding to different segments of the 3Ј-UTR as shown in Fig. 2A (lanes 4 -8), to be used in RNA binding assays as possible RNase T1 protected fragments. The data summarized in Fig. 2, A and B, indicate that the nucleotide sequence fragment with multiple protein complex interactions is located between nucleotide positions 2394 and 2435 (42 nt). Interestingly, this 42-nt binding fragment consists of three copies of the 7-nt sequence, GCUUGCU (Fig. 2, B and C). To more precisely locate the cis-element within the 42-nt RHAMM 3Ј-UTR fragment that forms the TGF-␤ 1 -inducible multiple RNA-protein complexes, we generated a series of synthetic oligoribonucleotides corresponding to different regions of the 42 nt. The data summarized in Fig. 2C indicate that the ciselement binding site, which forms TGF-␤ 1 up-regulated multiple protein complexes as observed with the full-length RHAMM 3Ј-UTR, is within the 30-nt region, nt 2400 -2429.
Effect of Mutations on the cis-Element-Protein Binding Activity-To assess the contributions of the individual nucleotides to the cis-element binding activity, in an attempt to determine a consensus sequence, we focused our attention on the three copies of the sequence GCUUGCU by producing a substitution mutation within this sequence and monitoring binding activity by standard gel shift assays. The results summarized in Fig. 3 showed that UU to AA changes within this sequence, GCU-UGCU, showed no detectable binding (lane 2). Similar results were obtained with UU to GG or CC changes (lanes 3 and 5). Changes of the core U residues within the last two copies of the 7-nt sequence to CC (GCCCGCU) resulted in no detectable binding (lane 6). Changing the G residues within the GCU-UGCU sequence to A residues also abolished binding. Interestingly, substituting the terminal U residue of the sequence GCUUGCU to A did not affect binding (lane 4), indicating that this U residue is not part of the binding motif. This observation further refines the binding requirement to a sequence consisting of GCUUGC.
Characterization of the cis-Element Binding Activity-Standard RNA gel shift assays were used to investigate the properties of the TGF-␤ 1 -mediated RHAMM mRNA-protein binding activity. The results summarized in Fig. 4A showed that preincubation of cytoplasmic extracts from C3 TGF-␤ 1 -treated cells with proteinase K (40 units/ml) or 0.1% SDS abolished the  3). When RNA binding reactions were performed with nuclear extracts instead of cytosolic extracts, the TGF-␤ 1 up-regulation of RNA-protein complexes was not observed (lanes 4 and 5). These results demonstrated the polypeptide component of these complexes and suggested that the proteins involved in complex formation are confined to the cytoplasm. The results in Fig. 4B showed that the related sequences, the RHAMM 3Ј-UTR self-competitor, the 30-nt cis-element fragment, and three copies of the sequence GCUUGCU (21 nt) within the 30-nt fragment (1) effectively prevented formation of the RNA-protein complexes (lane 2, 3, and 5), but the unrelated 3Ј-UTR mRNAs, granulocyte-macrophage colony-stimulating factor (3Ј-UTR) (7) (lane 4), and the ribonucleotide reductase R1 and R2 protein (R1BP, R2BP) binding cis-elements (5, 8) did not prevent binding (lanes 6 and 7). The TGF-␤ 1 up-regulation of multiple RNA-protein binding complexes was also observed for similar ras-transformed fibrosarcomas, C1 and C2 (Fig. 4C,  compare lanes 2 and 4 and lanes 3 and 5). As shown in Fig. 5A, the TGF-␤ 1 -inducible RNA-protein complex binding activities increased with time. The corresponding UV cross-linking results (Fig. 5, B and C) showed that the RHAMM TGF-␤ 1responsive cis-element and the 3Ј-UTR formed five major RNAprotein complexes of 175, 97, 63, 26, and 17 kDa. Most importantly, these complexes were not observed in untreated cells, but only upon TGF-␤ 1 stimulation. This binding activity increased with time, consistent with the gel shift data (Fig. 5A).
3Ј-UTR Confers TGF-␤ 1 -mediated Stabilization of CAT Hybrid mRNA-To investigate the possibility of the role of the TGF-␤ 1 -responsive cis-element in the stabilization of RHAMM mRNA in vivo, we have assessed the contribution of the 3Ј-UTR of RHAMM mRNA to the stability of a heterologous CAT mRNA. Using the C3 cell line, various 3Ј-UTR-CAT plasmid constructs shown in Fig. 6 were stably transfected. The halflives of the CAT hybrid transcripts were determined in the absence and presence of TGF-␤ 1 treatment as described previously (5,6). The results showed that in the control experiments there was no significant difference between the half-life of the CAT transcript in the absence or presence of TGF-␤ 1 treatment (Fig. 6A). Northern blot analysis showed a markedly significant difference between the half-life of CAT RH1 in the absence and presence of TGF-␤ 1 stimulation. These results indicated that treatment of cells with 10 ng/ml TGF-␤ 1 increased the half-life of CAT mRNA with the RHAMM 3Ј-UTR by a factor of about 5 relative to message from untreated cells (Fig. 6B). Interestingly, no changes in half-lives were observed after TGF-␤ 1 stimulation with CAT RH2 mRNA, which has a 3Ј-UTR deletion of a 295-bp fragment containing the 30-nt TGF-␤ 1 -responsive cis-element binding site (Fig. 6C). Furthermore, deletion of a 450-bp fragment (nt 2684 -3133) downstream of this ciselement, pCATRH3, did not abolish the TGF-␤ 1 -induced stabilization of CAT hybrid mRNA (Fig. 6D). DISCUSSION In the present study, we have demonstrated for the first time that alterations in message stability through specific cis-trans interactions play an important role in the TGF-␤ 1 -induced elevation of RHAMM mRNA. We have identified a novel 30-nt cis element, 5Ј-GCUUGCUUCCCCUGGCUUGCUCAGCUU-GCU-3Ј (nt 2400 -2429), in the 3Ј-UTR of RHAMM mRNA, which interacts with multiple cytosolic protease sensitive factors in fibrosarcomas exposed to TGF-␤ 1 , to form five major RNA-protein complexes of approximately 175, 97, 63, 26, and 17 kDa. Similar to several previously published mRNA ciselements (8, 15, 16, 29 -31), there is only one copy of the 30-nt cis-element sequence in the RHAMM message. Interestingly, within this 30-nt sequence there are three copies of the 6-nt sequence 5Ј-GCUUGC-3Ј separated by 8 and 3 nts, respectively. This resembles the AU-rich pentamer repeats within the 3Ј-UTR of lymphokine and cytokine mRNAs (32,33). The RHAMM cis-element showed sequence specificity, since the cis-trans interactions mediated by TGF-␤ 1 were diminished in the presence of related but not unrelated mRNA sequences. Mutagenesis data indicated that there are relatively stringent primary sequence requirements for the optimum interaction of  the cis-element with the trans-acting proteins and suggest that the sequence 5Ј-GCUUGC-3Ј is important for binding activity. It is not known if the nucleotides in the 30-nt cis-element observed not to be directly involved in protein binding in these studies play some other roles in message regulation. Further studies will more precisely define their potential roles and also determine the relative importance of the three 5Ј-GCUUGC-3Ј motifs within the 30-nt cis-element sequence, with regard to protein binding and eventually in models of message stability regulation. Knowledge of a cis-element binding motif for TGF-␤ 1 -responsive mRNA-binding proteins may alert us to regulatory sequences present in candidate 3Ј-UTRs that are posttranscriptionally regulated by TGF-␤ 1 (34,35). The appearance of the RNA-protein complexes following TGF-␤ 1 treatment of C3 cells was time-dependent and was also observed with cytosolic extracts from the TGF-␤ 1 -treated ras-transformed fibrosarcomas, C1 and C2.
Insertion of the 3Ј-UTR of RHAMM next to a heterologous CAT sequence resulted in a marked increase in the stability of CAT hybrid mRNA following TGF-␤ 1 treatment, supporting the concept that the RHAMM 3Ј-UTR is directly involved in the process that mediates TGF-␤ 1 -induced stabilization of RHAMM mRNA. This view is consistent with the finding that an internal deletion within the RHAMM 3Ј-UTR, containing the 30 nt cis-element, abolished the TGF-␤ 1 -induced increase in CAT hybrid mRNA half-life, and suggests that the RHAMM cis-element is necessary to mediate the TGF-␤ 1 effects on message stability. Further evidence of the importance of the 30 nt cis-element was the observation that another deletion within the 3Ј-UTR downstream of the cis-element sequence did not prevent the TGF-␤ 1 induced stabilization of CAT hybrid mRNA.
The results of this investigation are consistent with a model in which the 30-nt cis-element functions as a message destabilizer, and the TGF-␤ 1 -responsive proteins act as stabilizing factors by binding to the cis-element sequence, leading to a reduced rate of RHAMM mRNA degradation. It will be important in future studies to further refine the minimal cis-element and to determine how the binding of the TGF-␤ 1 -responsive proteins alter message-targeted nuclease activity in ras-transformed cells expressing RHAMM mRNA.