Regulatory Activities of the 5′- and 3′-Untranslated Regions and Promoter of the Human Aggrecan Gene*

Identification and characterization of the regulatory elements of the human aggrecan gene are necessary first steps in addressing the molecular mechanisms through which the gene is regulated. Using luciferase reporter constructs driven by the human aggrecan promoter or the cytomegalovirus promoter, the 5′- and 3′-untranslated regions of the human aggrecan gene were found to regulate gene expression transcriptionally in a promoter- and/or cell type-specific manner. Independent of cell type, the 5′-untranslated region was inhibitory with respect to the cytomegalovirus promoter, but it was stimulatory to the human aggrecan promoter. The 5′-untranslated region inhibited the cytomegalovirus promoter by approximately 60% in both chondrocytes and NIH 3T3 cells, but it stimulated the activity of the human aggrecan promoter about 8-fold in chondrocytes and 40-fold in NIH 3T3 cells. In contrast, the 3′-untranslated region inhibited the activities of the human aggrecan promoter by 40–70% in both cell types, but it stimulated the cytomegalovirus promoter activities by 50–60% in NIH 3T3 cells and inhibited its activity by 70% in chondrocytes. The differential effects of the untranslated regions on the two types of promoters may be a reflection of differences in regulation of TATA-less promoters, such as the human aggrecan promoter, and TATA-containing promoters, such as the cytomegalovirus promoter.

Expression of aggrecan by chondrocytes is subject to modulation by several factors, including cytokines, growth factors, and mechanical stimuli. Interleukin-1 and tumor necrosis factor-␣ both suppress aggrecan gene expression and synthesis, whereas insulin-like growth factor I and transforming growth factor-␤ stimulate aggrecan gene expression and synthesis. Mechanical loading of cartilage has been shown to modulate (increase and decrease) aggrecan gene expression and synthesis depending on the type and extent of the load imposed (1-7). 1 Changes in aggrecan synthesis have also been identified as important events in cartilage pathology (8 -10). Although aggrecan is expressed predominantly in cartilage, it is also known to be expressed in other tissues such as neural tissues (11)(12)(13)(14) and neonatal lungs (15). In the embryonic central nervous system, aggrecan is involved in regulating neural crest migration (14,16). Exposure of the lungs of neonatal rats to hyperoxia up-regulates aggrecan expression (15). Mutations that preclude the proper synthesis and deposition of aggrecan into the extracellular matrix are often lethal. A point mutation identified in exon 12 of the aggrecan core protein gene of the nanomelic chick introduces a premature translation terminal codon that precludes synthesis of the third globular domain (G3) 2 (17), and a 7-bp deletional mutation in exon 5 of the cartilage matrix-deficiency (cmd) mouse aggrecan gene leads to a premature translation termination codon in exon 6 (18). Despite these observations, the molecular mechanisms governing the regulation of aggrecan gene expression and synthesis remain obscure.
The core protein of aggrecan consists of an amino-terminal globular domain (G1) that binds hyaluronan, an interglobular domain, a second globular domain (G2) whose function is yet unknown, an extended glycosaminoglycan attachment region, and G3, which is located at the carboxyl-terminal end of the molecule (19 -21). Comparison of the aggrecan core protein structure and deduced amino acid sequences from various species (22)(23)(24)(25) with the recently determined exon and intron organizations of the chick (26), human (27), mouse (25), and rat (28) aggrecan have revealed a highly modular structure of the gene. Various domains of the protein are coded by groups of exons or single exons, most of which must be expressed as modules of two or more exons because of restrictive exon splicing requirements (27). Alternative splicing of the epidermal growth factor-like (23,29,30) and complement-regulatory protein-like motifs (23) of the G3 domain in some species have, however, been reported.
The coding region of the aggrecan gene is flanked by exons 1 and 19 (27), from which the 5Ј-and 3Ј-untranslated regions (UTRs) are transcribed (23,27). The UTRs of other genes, for example ferritin, have been shown to function as regulatory elements of the genes (31)(32)(33)(34)(35). We hypothesize that the UTRs of the aggrecan gene are involved similarly in regulating expression of the gene, potentially in concert with the promoter. Thus, in this study we have cloned and investigated the regulatory activities of the promoter and the 5Ј-and 3Ј-UTRs of the human aggrecan gene. * This work was supported in part by Grants AR43597 and AR42003 from the NIAMSD, National Institutes of Health, and Grant 95-024 from the Orthopaedic Research and Education Foundation. 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.
Preparation of the Cytomegalovirus Promoter-driven 5Ј-and 3Ј-UTR Constructs-For investigation of the activities of the human aggrecan 5Ј-and 3Ј-UTRs independent of the human aggrecan promoter, luciferase constructs containing the 5Ј-and 3Ј-UTRs of the human aggrecan mRNA (Fig. 1A) were prepared in the cytomegalovirus (CMV) promoter-driven pcDNA3 eukaryotic expression vector. The luciferase gene cassette of the pGEM-luc plasmid was first subcloned into the BamHI and XhoI sites of pcDNA3 to generate the control luciferase reporter plasmid pLUCneo. The previously cloned 5Ј-UTR of the human aggrecan mRNA (27) was amplified using the primers 5Ј-ATC ACT AAG CTT GGC CCG ACC ACC TAC CTC-3Ј and 5Ј-TCA CGA GGA TCC AGA GTA AAG TGG TCA TAG TTC AC-3Ј. The underlined sequences indicate restriction sites for HindIII and BamHI, respectively. To preserve the Kozak sequence (36) of the human aggrecan mRNA, TRP-2 was designed from nucleotides ϩ78 to ϩ54 of the published human aggrecan cDNA sequence (23). The amplified product was digested with HindIII and BamHI and cloned in the HindIII/BamHI site immediately upstream of the luciferase reporter gene in pLUCneo to generate the human aggrecan 5Ј-UTR construct pCMV/5UTR. Exon 19 (containing the 3Ј-UTR and two polyadenylation signals) of the human aggrecan gene was amplified from the genomic cosmid cosHA-G3 (27), using the primers 5Ј-TTT CGC TCG AGC CAC CAC CTA CAA ACG CAG A-3Ј and 5Ј-TGC ACT GGG CCC TGG AAA GGC ACG ATG G-3Ј, and cloned into the XhoI/ApaI site immediately downstream of the reporter gene in pLUCneo to generate the 3Ј-UTR construct pCMV/3UTR. The construct pCMV/5_3UTR was generated by cloning, respectively, the 5Ј-UTR and exon 19 in the HindIII/BamHI and XhoI/ApaI sites of pLUCneo. Constructs pCMV/5UTR-US and pCMV/3UTR-US, containing the UTRs immediately upstream of the CMV enhancer/promoter, were prepared by individually cloning the 5Ј-and 3Ј-UTRs upstream of the CMV enhancer/promoter in the BglII/MluI site of pLUCneo. The suffix -US denotes the upstream position of the UTRs relative to the CMV promoter. All inserts were cloned, by design, in the sense orientation. The ␤-galactosidase (␤-GAL) gene cassette of the pSV-␤-galactosidase expression vector was subcloned into the HindIII/XhoI site of pcDNA3 for use as transfection efficiency control.
Preparation of the Human Aggrecan Promoter Constructs-A 1.05-kb fragment (containing 0.7 kb of the proximal promoter and 0.35 kb of exon 1) of the upstream region of the human aggrecan gene was amplified by PCR using the human genomic PromoterFinder kit and the Elongase Enzyme Mix according to the suppliers' instructions. The PCR product was cloned in the pCR II TA cloning vector and sequenced in both directions. The promoter/exon 1 fragment was then digested with MluI (located in the PromoterFinder adaptor) and BsrBI (located at ϩ25 of the human aggrecan exon 1 or 5Ј-UTR). The digested human aggrecan promoter (Ϫ701 to ϩ25) was ligated to the 5Ј-UTR at the BsrBI site and subcloned, in place of the CMV promoter, in pCMV/ 5UTR and pCMV/5_3UTR to generate the chimeric luciferase constructs pAGC1(Ϫ701)/5UTR and pAGC1(Ϫ701)/5_3UTR (Fig. 1B). Similarly, constructs pAGC1(Ϫ701) and pAGC1(Ϫ701)/3UTR were prepared by cloning the aggrecan promoter fragment into the MluI/ BamHI site immediately upstream of the luciferase gene in pLUCneo and pCMV/3UTR. The aggrecan gene symbol AGC1 here denotes the human aggrecan promoter.
Cell Culture and Transient Transfection-Primary bovine articular chondrocytes (1 ϫ 10 6 cells/60-mm dish) or NIH 3T3 cells (2 ϫ 10 5 cells/60-mm dish) were maintained in monolayer cultures at 37°C for 24 h in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, amino acids (0.5 ϫ minimal essential amino acids, 1 ϫ nonessential amino acids), buffering agents (10 mM Hepes, 10 mM sodium bicarbonate, 10 mM TES, 10 mM BES), and antibiotics (100 units/ml penicillin, 100 g/ml streptomycin). The cells were transiently transfected, using the ProFection calcium phosphate mammalian transfection kit, with 5 g of each luciferase construct and 2-5 g of ␤-GAL cotransfection plasmid/60-mm dish for 48 h. Within experiments, all dishes were transfected with the same amount of ␤-GAL cotransfection plasmid. At the time of harvesting, each dish was rinsed twice with 5 ml of Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline on ice, and the cells were lysed in 0.3 ml of 1 ϫ Promega reporter lysis buffer. Luciferase activity in each lysate was assayed using a LKB-Wallac liquid scintillation counter and the Promega luciferase assay system. ␤-GAL activity was assayed using the Galacto-Light ␤-GAL assay system. Luciferase activities were normalized to ␤-GAL activities, and the data were subjected to analysis of variance as well as the Student-Newman-Keuls multiple range comparison test.
FIG. 1. Schematic diagram of reporter constructs. Several constructs were prepared to investigate the roles of the human aggrecan promoter and the 5Ј-and 3Ј-UTRs in regulating expression of the gene. Panel A, for investigating the activities of the 5Ј-and 3Ј-UTRs independent of the human aggrecan promoter, chimeric constructs driven by the CMV promoter were prepared. CMV Prom denotes the CMV promoter and enhancer region. Panel B, to determine the effects of the human aggrecan 5Ј-and 3Ј-UTRs on activities of the human aggrecan promoter, the CMV promoter and enhancer were replaced with the human aggrecan promoter (Ϫ701 to ϩ25), using restriction sites in the vector and the human aggrecan 5Ј-UTR (exon 1), to generate the human aggrecan promoter and UTR constructs. The small hooked arrows indicate the start site and direction of transcription.

Measurement of Luciferase mRNA
Half-life-To assess the effects of the UTRs on stability of the luciferase mRNA, primary bovine articular chondrocytes were transfected with 10 g each of the control luciferase plasmid pLUCneo and the pCMV/5UTR and pCMV/3UTR constructs. 45 h later (t ϭ 0), the cells were treated with 65 M DRB, an RNA synthesis inhibitor specific for RNA polymerase II (37,38). Total RNA was subsequently isolated from the cells at 30-min intervals, starting at t ϭ 0, and the luciferase mRNA levels were quantified by PCR (39).

Effects of the Human Aggrecan 5Ј-and 3Ј-UTRs on Activities
of the CMV Promoter-The human aggrecan mRNA contains a G/C-rich 382-bp leader sequence and a relatively short 3Ј-UTR (27). Computer analysis predicts a complex secondary structure for the 5Ј-UTR which could influence the rate of translation of the aggrecan gene (40). However, no information is available on how the aggrecan UTRs influence expression of the aggrecan gene. Thus, to determine whether the UTRs of the aggrecan gene have regulatory roles, the activities of the human aggrecan 5Ј-and 3Ј-UTRs were first investigated, independent of the aggrecan promoter, using chimeric luciferase constructs driven by the CMV promoter/enhancer. In bovine articular chondrocytes, the activity of the control pLUCneo luciferase plasmid was 1.84 ϫ 10 4 Ϯ 0.86 ϫ 10 4 units of LUC/ unit ␤-GAL. Relative to the activity of pLUCneo, the luciferase activity of chondrocytes transfected with pCMV/5UTR was in-hibited by 61%, whereas that of the pCMV/3UTR-transfected chondrocytes was inhibited by 71% (p Ͻ 0.05, n ϭ 18) ( Fig. 2A). When the luciferase reporter gene was flanked by both UTRs, the enzyme activity was reduced by 92%. The results demonstrated that the human aggrecan UTRs contain regulatory activities. Transfection of chondrocytes with pCMV/5UTR-US and pCMV/3UTR-US, which contained the 5Ј-UTR and 3Ј-UTR immediately upstream of the CMV promoter/enhancer, respectively, reduced the expression of luciferase activities by 59 and 61% ( Fig. 2A), demonstrating that transcription of the UTRs, and therefore their physical presence in the luciferase transcript, is not required for their regulatory activities. Assessment of the effects of the UTRs on luciferase mRNA stability indicated that the presence of the UTRs in the luciferase transcripts did not alter mRNA stability significantly, as assessed by measurement of mRNA half-lives (Fig. 3). The half-lives of the luciferase transcripts ranged from 39 to 47 min, with or without the UTRs.
The activities of the CMV promoter-driven constructs were also investigated in NIH 3T3 cells. In these cells, the activity of pLUCneo was 5.34 ϫ 10 4 Ϯ 1.04 ϫ 10 4 units of LUC/unit ␤-GAL. The luciferase activities expressed by pCMV/5UTR and pCMV/5UTR-US were, respectively, 40 and 39% lower than the activity of pLUCneo (p Ͻ 0.05, n ϭ 15) (Fig. 2B). In contrast, pCMV/3UTR and pCMV/3UTR-US luciferase activities were higher than control activities by 64 and 50% (p Ͻ 0.05), respectively. The activity of the pCMV/5_3UTR construct was 15% lower than control activities, but the difference was not significant.
Sequence of the Human Aggrecan Promoter-The proximal promoter (701 bp) and exon 1 (375 bp) of the human aggrecan gene have been cloned and sequenced in the forward and reverse directions. The current sequence of exon 1 differs from that obtained previously by cDNA cloning (27) at two nucleotide positions. At position ϩ1, the current sequence reads as a C (Fig. 4), whereas it was determined as a G previously. Also, at position ϩ7 the genomic sequence is a C compared with an A at the same position in the cDNA sequence. To verify the accuracy of these sequences, two separate clones containing the promoter and exon 1 were sequenced. The genomic sequence at position ϩ7 is consistent with those reported for the rat and mouse aggrecan gene (25,28).
The sequence was analyzed for promoter motifs and tran- scriptional factor binding sites using the TSSG 3 and MatInspector (41) programs at the Baylor College of Medicine Search Launcher site. The promoter region is highly G/C-rich and lacks canonical TATA and CAAT motifs (Fig. 4). However, a search of the promoter data base predicted a TATA box at Ϫ31, which coincides with the beginning of the TATGTATG sequence (reverse font). This sequence is conserved in the rat and mouse aggrecan promoters (25, 28) but not in the chick aggrecan promoter (42). The transcriptional start site of the human aggrecan gene was predicted by TSSG to be at ϩ6, 5 bp downstream of the transcriptional start site (Fig. 4) determined previously by PCR-based cDNA cloning (27). Several putative binding sites for the ubiquitous transcription factors SP-1, AP-2, and AP-4 were found in the promoter and exon 1 (Fig. 4). A glucocorticoid/progesterone response element-like sequence is found at position Ϫ663. Two sis/PDGF (platelet-derived growth factor)-inducible factor (SIF) binding sites are located at Ϫ583 and Ϫ39. AP-1/CREB (cAMP response element-binding protein) binding sites are found in both the promoter and exon 1 at Ϫ599 and ϩ107, respectively. Also located in the promoter and exon 1 are four shear stress response elements (SSREs) (43). The SSRE in the promoter is located at Ϫ655; those in exon 1 are located at ϩ73, ϩ124, and ϩ284. Only the SSRE at ϩ284 is in the sense orientation. STAT (signal transducer and activator of transcription) and NF-B as well as Gfi-1 (growth factor independence) and ␦EF1 (␦-crystallin/E2-box factor 1) recognition sequences are also found in the promoter and exon 1. The STAT and NF-B binding sites may confer cytokine respon-siveness to the gene, whereas Gfi-1 and ␦EF1 are likely to function as negative regulators of transcription.
Effects of the Human Aggrecan 5Ј-and 3Ј-UTRs on Activities of the Human Aggrecan Promoter-To investigate the roles of the human aggrecan promoter and the UTRs in expression of the aggrecan gene, chimeric luciferase constructs containing the human aggrecan promoter 5Ј-UTR (exon 1) and/or 3Ј-UTR (exon 19) (Fig. 1B) were prepared and used in transient transfection assays. The level of expression induced by the promoter was 89-fold higher than that of the promoter-less luciferase vector (data not shown), indicating that this was a functional promoter. In the following experiments the activity of pAGC1(Ϫ701), containing the proximal promoter region (Ϫ701 to ϩ25) of the human aggrecan gene, served as base-line activity. The presence of the 5Ј-UTR immediately downstream of the aggrecan promoter, construct pAGC1(Ϫ701)/5UTR, significantly stimulated the expression of luciferase activity 7.7-fold relative to the activity of pAGC1(Ϫ701) (Fig. 5A). The expression of luciferase activity was suppressed by 68% (p Ͻ 0.001, n ϭ 6) when cells were transfected with pAGC1(Ϫ701)/3UTR, the construct containing the human aggrecan exon 19 (3Ј-UTR) immediately downstream of the luciferase reporter gene. The activity of pAGC1(Ϫ701)/5_3UTR, with the UTRs flanking the luciferase gene, was 45% lower than that of pAGC1(Ϫ701) and 93% lower than the activity of pAGC1(Ϫ701)/5UTR (Fig. 5A). In NIH 3T3 fibroblasts transfected with the aggrecan promoter/UTR constructs, a pattern of luciferase expression similar to that in chondrocytes was observed (Fig. 5B). Relative to the activity of the pAGC1(Ϫ701) construct, expression of luciferase activity in fibroblasts transfected with pAGC1(Ϫ701)/5UTR 3 V. V. Solovyev and A. A. Salamov, manuscript in preparation.

FIG. 4. Sequence of the human aggrecan promoter.
The proximal 701 bp of the human aggrecan promoter along with exon 1 were cloned using the PromoterFinder kit and sequenced in both directions. Several putative binding sites for ubiquitous transcription factors such as SP-1, AP-2, and AP-4 are located in both the promoter and exon 1. Recognition sequences for growth factor-and cytokine-inducible factors are also present. The promoter lacks a TATA box, but a TATGTATG sequence at Ϫ31 (reverse font) might function as a TATA element. was 39.6-fold higher (p Ͻ 0.001, n ϭ 12). However, there was no significant difference between the activities of pAGC1(Ϫ701)/ 3UTR and pAGC1(Ϫ701)/5_3UTR compared with that of pAGC1(Ϫ701).
To determine regions of the promoter which are essential for promoter activity, unidirectional deletion of the aggrecan promoter in pAGC1(Ϫ701)/5UTR was performed using endogenous restriction enzyme sites. Deletion of segment Ϫ701 to Ϫ452 moderately stimulated luciferase activity by 47% (p Ͻ 0.001) compared with the activity of pAGC1(Ϫ701)/5UTR (Fig.  6A). Further deletion of the promoter down to Ϫ179 resulted in luciferase activity that was approximately the same as that of pAGC1(Ϫ701)/5UTR. Removal of all but 52 bp of the promoter reduced luciferase expression by only 20% (p Ͻ 0.001). These results indicated that most of the human aggrecan proximal 701 bp promoter activity resides downstream of nucleotide Ϫ53. Deletion of the 5Ј-UTR (exon 1) from the construct containing the 52-bp promoter fragment completely inhibited expression of luciferase activity (Fig. 6B), demonstrating the requirement of sequences within exon 1 (i.e. downstream of ϩ25) for promoter activity. DISCUSSION The structural gene for aggrecan is large and complex, giving rise to an mRNA approximately 8 kb in size for the human gene. With elucidation of the gene structure for aggrecan in several different species (25)(26)(27)(28), isolation and characterization of regulatory regions of the gene can now be performed. We have cloned the human aggrecan promoter and performed functional studies to determine regulatory activities of the promoter and the 5Ј-and 3Ј-UTRs of the gene. The human aggrecan promoter contains several putative SP-1/AP-2 binding sites (-CCCGCC-), including a cluster of SP-1/AP-2 binding sites around the transcriptional start site (Fig. 4). SP-1 plays a key role in regulating transcription initiation of TATA-less promoters (44,45). The transcriptional start site of the human aggrecan promoter is spanned by three overlapping SP-1/AP-2 binding sites. In the rat gene, there is only one copy of the SP-1/ AP-2 site (28) in a position equivalent to that of the overlapping SP-1/AP-2 sites. Nonetheless, in both genes, transcription is initiated 23-24 bp downstream of a conserved TATG direct repeat (Fig. 4, reverse font), suggesting that this sequence is important in transcriptional initiation of the aggrecan gene. Search of the eukaryotic promoter data base using the human aggrecan promoter sequence predicted a TATA box at Ϫ31, the position of the TATG repeat. Thus, this sequence may be the equivalent of a TATA box in TATA-less promoters such as the aggrecan promoter. Transcription initiation of the bovine aggrecan gene, however, occurs at four different sites upstream of this TATG repeat (25), suggesting other mechanisms of transcription initiation. Panel A, the 701-bp proximal human aggrecan promoter in the construct pAGC1(Ϫ701)/5UTR was unidirectionally deleted (5Ј to 3Ј) using endogenous restriction enzymes and blunt end ligated to generate constructs with promoter fragments ending at the indicated positions. Monolayer bovine articular chondrocytes were then transiently transfected with the deletion constructs and assayed for luciferase activities. Panel B, the 5Ј-UTR (exon 1) was deleted from the construct containing 52 bp of the proximal promoter (panel A), and its activity was compared with the full-length (701-bp) promoter with or without the 5Ј-UTR. Note the marked suppression of luciferase activity when the 5Ј-UTR is deleted from the full-length or 52-bp promoter constructs.
Expression of aggrecan can be modulated by mechanical forces, cytokines, and other serum factors. The two NF-B sites in the promoter may confer responsiveness to cytokines. This is suggested by the overlapping NF-B and STAT binding sites at Ϫ375 to Ϫ362 (Fig. 4) and the fact that NF-B is often activated by cytokines such as tumor necrosis factor-␣ (46). The four SSREs (one in the promoter and three in exon 1) are potential sites for regulating aggrecan gene expression by shear stresses. The suppressive effects of interleukin-1 on aggrecan gene expression can be blocked or reversed by treatment with plateletderived growth factor (47). This stimulatory effect of plateletderived growth factor on aggrecan expression may be mediated through the SIF response elements in the promoter (Fig. 4).
The 5Ј-and 3Ј-UTRs of the human aggrecan gene strongly modulated the activities of the human aggrecan promoter (Fig.  5) as well as those of the CMV promoter (Fig. 2). In chondrocytes and NIH 3T3 cells, the 5Ј-UTR stimulated the activities of the human aggrecan promoter greater than 7-fold (Fig. 5). In contrast, the 3Ј-UTR inhibited the activities of the human aggrecan promoter in both chondrocytes and NIH 3T3 cells. The stimulatory activity of the 5Ј-UTR was inhibited in the presence of the 3Ј-UTR. The mechanism by which the 3Ј-UTR suppresses aggrecan promoter activity is unclear, but it might be mediated through negative transcriptional elements including the Gfi-1 response elements found in both the promoter (Fig. 4) and the 3Ј-UTR (not shown). With respect to the CMV promoter, the 5Ј-UTR suppressed the expression of luciferase activity in chondrocytes or NIH 3T3 cells, but the 3Ј-UTR modulated the activity of CMV promoter in a cell type-specific manner. The 3Ј-UTR was inhibitory in chondrocytes ( Fig. 2A) and stimulatory in NIH 3T3 cells (Fig. 2B). This cell typespecific effect of the 3Ј-UTR may be caused by the presence or absence of specific transcription factors in the transfected cells which bind the 3Ј-UTR and regulate transcription.
The effects of the 5Ј-and 3Ј-UTRs were mediated at the transcriptional level since 1) placement of the UTRs upstream of the CMV promoter did not alter their effects on expression of luciferase activities significantly (Fig. 2), and 2) their presence or absence from the luciferase transcript did not affect luciferase mRNA stability. The involvement of the 3Ј-UTR in modulating gene expression at the level of transcription is an unusual finding that has been demonstrated only for a few other genes. Typically, the 3Ј-UTR has been reported to control gene expression by modulating stability and translation of mRNAs (31)(32)(33). However, a role for the 3Ј-UTR in suppressing both basal and growth hormone-induced transcription has been reported in the rat serine protease inhibitor gene (35). Other examples of the transcriptional involvement of the 3Ј-UTR in regulating gene expression include the interaction between the tumor necrosis factor promoter and 3Ј-UTR in mediating the response of the gene to bacterial endotoxin (48) and the identification of an estrogen response element in the c-fos 3Ј-UTR which is capable of conferring estrogen responsiveness to a heterologous promoter such as the thymidine kinase promoter (34).
In the absence of inhibitory modulators, high level expression of the human aggrecan gene is critically dependent on sequences present in exon 1 (5Ј-UTR) (Fig. 6). This is indicated by the fact that sequential deletions of the promoter from Ϫ701 to Ϫ52 have little effect on promoter activity when the 5Ј-UTR is present. However, when the 5Ј-UTR is removed, leaving a 5Ј-flanking sequence of Ϫ701 to ϩ25 or Ϫ52 to ϩ25 (Fig. 6B), luciferase activity is reduced 85-100%. It is possible that sequences within this region bind to general transcription factors required during the assembly of the transcription initiation complex. This region contains several putative SP-1 binding sites, characteristic of TATA-and CAAT-less promoters, and may be involved in recruiting the TFIID protein to the transcription initiation complex.
Aggrecan is a member of a family of hyaluronan-binding proteoglycans that have been designated as hyalectans (49) or lecticans (50). Other members of this group include versican, neurocan, and brevican. The hyalectans share striking structural similarities, both at the protein and genomic levels (49). They differ primarily in the glycosaminoglycan attachment regions and also in the presence of the G2 domain in aggrecan. Promoter sequences for aggrecan (Fig. 3) (25,28,42), versican (51), neurocan (52), and brevican (53) have now been described. The versican (51) and neurocan (52) promoters both contain TATA boxes, whereas the aggrecan (Fig. 3) and brevican (53) promoters do not. Like the aggrecan promoter, the versican promoter contains multiple AP-2 sites (51). However, the 150-bp promoter region immediately upstream of the transcriptional start site of the versican gene lacks AP-2 or SP-1 sites, in contrast to that of the aggrecan promoter region (Fig. 3). This, along with the presence of a TATA box in the versican gene, suggests different mechanisms for transcriptional initiation of the aggrecan and versican genes.
Perlecan, a modular heparan-sulfate proteoglycan that is expressed predominantly in vascular tissues (49,54), is also expressed in developing and adult cartilage (55,56). The perlecan promoter, like the aggrecan promoter (Fig. 3), lacks canonical TATA and CAAT boxes and contains multiple SP-1/ AP-2 binding sites in the promoter region and exon 1 (57). Most of the perlecan SP-1/AP-2 binding sites are in the reverse orientation compared with those of the aggrecan promoter. Notwithstanding, the aggrecan (Fig. 6A) and perlecan (57) proximal promoters exhibit very similar activities. Unidirectional deletion of the promoters to Ϫ52 (for aggrecan) and Ϫ132 (for perlecan) retained 80% and 40 -50% of the wild-type promoter activities, respectively. This is suggestive of the sufficiency of the clusters of SP-1 and AP-2 sites near the transcriptional start sites, in conjunction with sequences in exon 1, in mediating basal activities of the aggrecan and perlecan promoters. The similarities in the organizations and activities of the aggrecan and perlecan promoters may provide clues to the mechanisms governing expression of the aggrecan and perlecan genes, especially in relation to their expression in cartilage (55,56).