Human Tumor Necrosis Factor-α Gene 3′ Untranslated Region Confers Inducible Toxin Responsiveness to Homologous Promoter in Monocytic THP-1 Cells*

To better define the role of 3′ untranslated region (3′UTR) on transcriptional regulation of the human tumor necrosis factor (TNF)-α gene, monocytic human THP-1 cells were transfected with two TNF-α promoter constructs spanning base pairs −1897/−1 and −1214/−1, respectively, and linked to the rabbit β-globin gene. Quantitative globin gene expression of chimerae was measured by reverse transcription-polymerase chain reaction. A construct linking the chicken β-actin promoter and a deleted portion of the β-globin gene was cotransfected and used as internal standard. Unexpectedly, when THP-1 cells were stimulated with lipopolysaccharide or toxic shock syndrome toxin-1, gene regulation was hardly detected. In contrast, endogenous TNF-α gene regulation measured by the same reverse transcription-polymerase chain reaction procedure was vigorous. Remarkably, ligation of 3′UTR to chimeric constructs led to a drastic drop in the basal level of chimeric gene expression, resulting in a 15- to 40-fold induction of the reporter gene. Consistently, when the TNF-α promoter was replaced by the cytomegalovirus early immediate promoter, gene expression was also uniformly reduced but was no longer up-regulated upon stimulation with lipopolysaccharide and toxic shock syndrome toxin-1. These data provide the first line of evidence that, in addition to its role in TNF-α transcript stability and translation, human TNF-α 3′UTR also participates in modulating gene expression at the transcriptional level.

TNF-␣, 1 a pleiotropic cytokine produced mainly by macrophages, plays a central role in cell immune responses (1)(2)(3), host defense (4), and inflammation. TNF-␣ gene expression mediated by lipopolysaccharide (LPS) or by MHC class II ligands such as bacterial superantigen toxic shock syndrome toxin-1 (TSST-1) and staphylococcal enterotoxin B (SEB) appears to be regulated at both the transcriptional and posttranscriptional levels (5)(6)(7). Remarkably, TNF-␣ promoter responses were only weakly induced by LPS, and even in nonstimulated cells, a significant constitutive gene expression was detected (8,9). The role of human TNF-␣ 3Ј untranslated re-gion (3ЈUTR) in post-transcriptional control of TNF-␣ mRNA has been well documented (10). A conserved sequence element in the 3ЈUTR of several cytokines in several species, the TTATTTAT element, normally confers translational repression (10,11). Whereas this UA-rich motif also confers instability to many cytokine mRNAs, TNF-␣ transcript stability appears to be unchanged after cell stimulation by LPS (10,12,13). Interestingly, it was first demonstrated in the mouse that TNF-␣ promoter and 3ЈUTR synergized to regulate murine TNF-␣ gene expression (12). Moreover, constitutive expression of mouse TNF-␣ promoter in non-macrophage cell line L929 could be suppressed by ligating mouse 3ЈUTR to chimeric CAT constructs, indicating that mouse 3ЈUTR also played a role in silencing TNF-␣ gene in cells in which it was not expressed (14). In this study, we extend these observations to the human cell line THP-1 and demonstrate, using constructs derived from the human TNF-␣ promoter, that despite a vigorous induction of endogenous TNF-␣, no significant regulation of a reporter gene could be detected upon cell stimulation by LPS or TSST-1. Regulation of two TNF-␣ chimeric constructs could only be obtained after TNF-␣ 3ЈUTR ligation to promoter constructs. These data indicate that the TNF-␣ gene promoter and its 3ЈUTR cooperate in regulating gene expression at the transcriptional level in humans as well.
Plasmids and Chimeric Constructs-Gene expression was measured by quantitative reverse transcription-PCR (15) using the rabbit ␤-globin gene as a reporter gene (16). Plasmid pG␤Ac␤G1D, hereafter referred to as ⌬␤ (Fig. 1), contains the chicken ␤-actin promoter driving a truncated rabbit ␤-globin gene in which 40 nucleotides were deleted in the second exon and generates a shorter PCR amplification product (17). A second plasmid, pG␤G(ϩ), containing the undeleted rabbit ␤-globin gene, was used as a vector for TNF-␣ gene promoter constructs. Human TNF-␣ promoter fragments were generated by PCR using the High Fidelity TaqI DNA Polymerase system from Roche Molecular Biochemicals. The DNA template used for PCR contained the entire TNF locus (kindly provided by V. Jongeneel (Ludwig Institute, Lausanne, Switzerland)). Oligonucleotides used as primers were purchased from Microsynth (Balgach, Switzerland) and derived from published sequences of the human TNF locus (18,19). Sequences were as follows: 5Ј-1897 oligo, 5Ј-GCTCGGTACCCTGTCTTCTTTGGAGC-3Ј; 5Ј-1214 oligo, 5Ј-GCTCGGATCCGTCTGGGAGTGAGAAC-3Ј; and 3Ј-1 oligo, 5Ј-GCGCAGATCTGGGTGTGCCAACAACT-3Ј. Promoter fragments in chimerae 1 and 2 starting respectively at positions Ϫ1897 and Ϫ1214 from the transcription initiation site (ϩ1) and terminating at position Ϫ1 were introduced in pIAL1 vector (7) partially digested by BamHI and totally digested by KpnI. Cytomegalovirus enhancer and immedi-* This work was supported by the Swiss National Fund for Scientific Research Grants 32432.91 and 49678.96. 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.
Transfection and Cell Stimulation-THP-1 cells were transiently transfected with a defined ratio of TNF-␣ promoter/rabbit ␤-globin gene chimera and of the reference plasmid ⌬␤ (10 g/ml:0.25 g/ml, respectively, in 1 ml of fetal calf serum-free medium). Cells were grown at a density of 5-8 ϫ 10 6 cells/ml and washed thoroughly with fetal calf serum-free medium. After a 4-h incubation (5-8 ϫ 10 6 cells in 1 ml) in the presence of 300 g/ml DEAE-dextran and the appropriate plasmid DNAs, cells were washed, resuspended in 15 ml of fresh medium, and incubated for 20 -22 h at 37°C in a 5% CO 2 incubator. Transfected cells were then stimulated with either 2 g/ml LPS or 30 g/ml TSST-1 or SEB for 90 min.
Preparation of Total Cellular RNA-Total RNA was extracted as described previously (20). Briefly, after phenol:chloroform extraction, total RNA was concentrated by precipitation and resuspended in DNase I buffer. A RNase-free-DNase I (Roche Molecular Biochemicals) digestion was then performed in 50 mM Tris-HCl, pH 7.5, and 10 mM MgC1 2 to eliminate contaminating plasmid DNA. Resulting deoxynucleotides were removed by precipitation in 2.5 M ammonium acetate and 2.5 volumes of 100% ethanol. The quality of purified RNA was checked on an agarose gel, and RNA concentration was measured by spectrophotometry at 260 nm.
Reporter Gene Assay by Quantitative Reverse Transcription-PCR Assay-A predetermined quantity of RNA was used for each sample of a given experiment. The assay was carried out as described previously (7,21), with minor modifications. Reverse transcription was performed in 30 l of Perkin-Elmer PCR buffer containing 0.3-0.5 g of total RNA supplemented with 0.2 mM deoxynucleotide triphosphate, 1 mM dithiothreitol, 12.5 g/ml oligodeoxythymidylic acid (Amersham Pharmacia Biotech), and 1 unit of avian myeloblastosis virus reverse transcriptase (Amersham Pharmacia Biotech). RNA pAW 109 (0.5 l of Gene Amplimer pAW 109 RNA/sample; 1 ϫ 10 6 copies/l; Perkin-Elmer) was used as an internal standard (22). The PCR product from pAW109 RNA amplified with TNF-␣ primers was 301 base pairs (bp) long and was designed to be shorter than the PCR product from target TNF-␣ mRNA (325 bp). After extension, a reverse transcription reaction aliquot (7 l) was diluted 4-fold in PCR buffer mix (Perkin-Elmer) supplemented with 0.2 mM deoxynucleotide triphosphate, 1-2 Ci of [ 32 P]dCTP (Amersham Pharmacia Biotech; Ͼ 3000Ci/mmol), and a pair of oligonucleotides (2 nM) specific for TNF-␣ (5Ј oligonucleotide, 5Ј-CAGAGGGA-AGAGTTCCCCAG-3Ј; 3Ј oligonucleotide, 5Ј-CCTTGGTCTGGTAGGA-GACG-3Ј) and amplified for 22 cycles using 1.5 units of Taq DNA polymerase (Perkin-Elmer). Another reverse transcription reaction aliquot (7 l) diluted in PCR buffer was amplified as described above for 20 cycles with a pair of oligonucleotides hybridizing to the rabbit ␤-globin gene (5Јoligonucleotide, 5Ј-TCCCCCAAAACAGACAGAATGG-3Ј; 3Јoligonucleotide, 5Ј-ACGTTGCCCAGGAGCCTGAAGT-3Ј). After a 15fold dilution in the buffer mixture, this first PCR product was further amplified for an additional 10 -20 cycles with another 5Ј oligonucleotide (5Ј-GGTGGTGAGGCCCTGGGCAGG-3Ј) and the same 3Ј oligonucleotide. In cotransfections with chimeric constructs and ⌬␤, the chimeric gene expression level was related to the expression level of ⌬␤, which remained unchanged regardless of THP-1 cell stimulation. PCR conditions for each experiment were chosen to provide a linear relationship between RNA levels and observed signals. The number of amplification cycles was at least five cycles below the saturation point for each experiment. Amplified samples were analyzed on a 6% polyacrylamide/ 7.5 M urea sequencing gel, and signals were quantified using an Instant Imager ® (Packard Instruments, Meriden, CT).

TNF-␣ Promoter Is Not Sufficient to Confer Toxin Responsiveness to Rabbit ␤-Globin
Reporter Gene-Stimulation of THP-1 cells transfected with control DNA ⌬␤, with chimera 1 or 2 ( Fig. 1) alone, or with a combination of either ⌬␤ or one of the two chimerae resulted in a strong stimulation of endogenous TNF-␣ transcripts (TSST-1, 25-to 70-fold induction; SEB, 10-to 35-fold induction; LPS, 30-to 97-fold induction) ( Fig. 2A). In contrast, although chimeric constructs 1 and 2 were strongly expressed, their expression was not regulated or was extremely poorly regulated (1.1-to 1.6-fold induction), irrespective of the stimulus and the presence or absence of the internal standard ⌬␤ (Fig. 2B). As compared with endogenous wild type gene, the absence of reporter gene regulation suggested a lack in regulatory element(s) in the chimerae.
huTNF-␣ 3ЈUTR Is Necessary to Reconstitute Toxin Responsiveness-Because promoter regions were unable to regulate human ␤-globin reporter gene expression by themselves, we transfected THP-1 cells with chimeric constructs 1 and 2 linked to human TNF-␣ 3ЈUTR gene sequences in a sense (s) or antisense (a) orientation (Fig. 1, chimerae 1/3Јs, 2/3Јs, 1/3Јa, and 2/3Јa) and examined whether induction could be recovered under these conditions. Endogenous TNF-␣ expression was strongly enhanced upon stimulation with TSST-1 or LPS ( Fig.  3A; Table I), whereas chimerae 1 and 2 were again strongly expressed, but poorly regulated (1-to 2.5-fold induction; Fig. 3, B and C; Table I). However, the addition of TNF-␣ 3ЈUTR drastically affected their level of expression; in contrast to the strong expression of chimerae 1 and 2, the expression of chimerae linked to TNF-␣ 3ЈUTR (sense or antisense) was 10-to 30-fold lower (Fig. 3, B and C; Table I). Furthermore, chimerae 1/3Јs and 2/3Јs from unstimulated THP-1 cells were hardly expressed. This finally resulted in chimerae 1/3Јs and 2/3Јs induction levels comparable to those observed for the endogenous TNF-␣ gene: a 10-to 45-fold induction for chimera 1/3Јs, and up to a 17-to 57-fold induction for chimera 2/3Јs (Table I).
TNF-␣ 3ЈUTR Does Not Confer Toxin Inducibility to Unrelated Promoters-To determine whether this regulation occurred at the transcriptional and/or post-transcriptional level, we generated chimeric constructs in which the TNF-␣ promoter was replaced by other unrelated promoters. Because construct ⌬␤, driven by the chicken ␤-actin promoter, was not regulated by LPS or TSST-1, ⌬␤ was linked to TNF-␣ 3ЈUTR in the sense and antisense orientations (⌬␤/3Јs and ⌬␤/3Јa) (Fig. 1). RNA levels for ⌬␤/3Јs and for ⌬␤/3Јa were extremely weak, and no signal could be detected (Fig. 4B), although TNF-␣ endogenous induction was strong (Fig. 4A). This strongly suggested that TNF-␣ 3ЈUTR, irrespective of its orientation, destabilized chimeric RNA (12). We then replaced the TNF-␣ promoter in chimerae 1, 1/3Јs, and 1/3Јa with the cytomegalovirus (CMV) immediate early promoter, which is known to be much stronger than the chicken ␤-actin promoter. This new series of chimerae, called CMV, CMV/3Јs, and CMV/3Јa (Fig. 1), was designed to generate RNA transcripts almost identical to those produced by chimerae 1, 1/3Јs, and 1/3Јa, but controlled by a different promoter. Only a few nucleotides differed in the very 5Ј end of the resulting chimeric RNAs. A strong expression of chimera CMV ( Fig. 5B) was observed, which was unaffected by cell stimulation (TSST-1 induction ratio, 1.0 -1.6; LPS induction ratio; 1.7). Signals from cells transfected with chimerae CMV/3Јs and CMV/3Јa were about 10-to 30-fold weaker than signals from cells transfected with chimera CMV (Fig. 5B). However, a strong expression was detected in unstimulated cells even when transfected with chimera CMV/3Јs, in contrast to our observations with chimerae driven by the TNF-␣ promoter (Fig. 3B). The presence of the unrelated CMV promoter in chimerae CMV/3Јs and CMV/3Јa strongly affected reporter gene regulation because the induction level in cells stimulated with TSST-1 was abrogated or only minimally enhanced upon treatment with LPS (induction ratio, 2.4 -3.7; Table I). In any case, this latter increase was significantly lower than the induction level of constructs driven by the TNF-␣ promoter (chimera 2/3Јs, 30-fold induction) ( Fig. 3; Table I). Altogether, these results suggested that the TNF-␣ 3ЈUTR not only played a role at a post-transcriptional level by destabilizing RNA in the absence of induction but also played a role at a transcriptional level by modulating the use of the TNF-␣ promoter. This regulation was not specific for the type of stimulus applied. DISCUSSION We have shown here that the TNF-␣ 3ЈUTR plays a crucial role in human TNF-␣ gene transcriptional regulation, an observation that further extends its role in RNA transcript stability and translation as described previously (23). In our approach, we took advantage of a reporter gene system that allows direct analysis of RNA expression, in contrast to CAT or luciferase gene assays, which are less appropriate to tackle this issue. These latter systems involve protein enzymatic assays that not only reflect transcriptional regulation but, depending on the transfected chimera, also reflect the post-transcriptional and/or translational regulations that can affect the final enzymatic activity. Because TNF-␣ 3ЈUTR contains sequences that can modulate translation (10), we overcame this pitfall by directly measuring RNA transcripts. Thus, it appeared that chimeric constructs driven by large fragments of the human TNF-␣ promoter were strongly expressed in the monocytic cell line THP-1 but only weakly regulated by stimuli such as TSST-1 and LPS. These data are consistent with results from Goldfeld et al. (8), who transfected human TNF-␣ promoter-CAT chimeric constructs in the murine monocytic cell line P3888D1 and found a significant level of expression of the chimeras in uninduced cells. In agreement with our data, only FIG. 1. Schematic representation of the different constructs used in the study. First line, the human TNF-␤ and TNF-␣ locus (not drawn to scale). Ⅺ, 5Ј and 3Ј untranslated regions (UTR); f, exons. The TATA box, initiation codon (ATG), stop codon, and polyadenylation sites are indicated. Chimerae 1 and 2, two constructs containing 1896 and 1213 bp of the TNF-␣ promoter cloned upstream of the genomic sequences coding for the rabbit ␤-globin gene. u, exons. The ATG, stop codon and polyadenylation sites are indicated. 1/3Јs, 1/3Јa, 2/3Јs, and 2/3Јa, constructs derived from chimerae 1 and 2 carrying the TNF-␣ 3ЈUTR in the sense (3) or antisense (4) orientation at the 3Ј end of the rabbit ␤-globin gene. ⌬␤, modified rabbit ␤-globin gene carrying a 40-bp deletion in the second exon. o, chicken ␤-actin promoter. ⌬␤/3Јs and ⌬␤/3Јa, the same constructs with the TNF-␣ 3ЈUTR fused at the 3Ј end of the last exon. CMV, a construct in which the TNF-␣ promoter in chimera 1 is substituted with the CMV promoter (p). CMV/3Јs and CMV/3Јa, the same constructs with the TNF-␣ 3ЈUTR hooked to the 3Ј end of the rabbit ␤-globin reporter gene. weak (1.5-to 2-fold) induction by LPS could be detected. Takashiba et al. (9), who also used CAT assays, found similar induction ratios. Very different results were obtained for the mouse gene because a strong induction by LPS was detectable in chimeric constructs containing mouse TNF promoter only (24). Furthermore, kB-type enhancers were involved in the transcription of the murine gene (25) but not the human gene (8). Although the regulation of human and mouse TNF genes differs in many respects, we took advantage of previous observations on the murine gene to design the chimeric constructs studied here. The role of mouse 3ЈUTR on chimeric gene expression has been well documented (12-14, 23, 26). In partic-ular, mouse TNF-␣ 3ЈUTR is able to suppress TNF-␣ promoter constitutive activity in non-macrophage cell lines (14). Furthermore, it interacts with the mouse TNF-␣ 5Ј end to modulate chimeric CAT construct expression (12). We have shown in this study that human TNF-␣ 3ЈUTR suppressed the strong basal expression of TNF-␣ promoter constructs observed in its absence in unstimulated THP-1 cells and partially restored gene regulation in induced cells. This regulation could be due to transcriptional and/or post-transcriptional phenomena. Evidence for transcriptional regulation was provided by the analysis of constructs in which TNF-␣ promoter fragments were replaced with the CMV immediate early promoter. CMV chi-  meric constructs linked to the TNF-␣ 3ЈUTR were expressed at a level comparable to TNF-␣ promoter constructs. However, the expression of CMV chimeric construct CMV/3Јs was poorly regulated or was not regulated, in contrast to chimeric constructs 1/3Јs and 2/3Јs. These results are in agreement with previous studies on the murine TNF-␣ gene (12,14) and indicate that these observations hold true for human TNF-␣ gene regulation. Taken together, our data suggest that 5Ј TNF-␣ promoter and 3ЈUTR regions interact with one another to regulate TNF-␣ gene expression at transcriptional level as well. The interaction between the 3Ј and 5Ј regions appears to be independent of the type of stimulation. DNA fragments used in these experiments are large, and the identification of potential cis-and trans-regulatory elements in both the 5Ј or 3Ј regions will require the study of smaller gene fragments. TNF-␣ transcription may be enhanced by stimulatory mechanisms and/or the release of a pre-existing block. The precise characterization of transcription factors binding to DNA regions essential for regulated expression of the TNF-␣ gene may open new perspectives in the understanding of human diseases related to abnormal TNF-␣ gene expression. Interestingly, mutated genomic sequences in regions flanking the 3Ј TTATTTAT signal element of TNF-␣ gene have been described in murine models (27), but not in young patients with autoimmune diseases (28). Because our results demonstrate the presence of other crucial regulatory domains interacting with promoter regions, the precise identification of discrete regulatory elements may help us to better understand the regulation of the TNF-␣ gene in inflammatory diseases.