Characterization of the promoter for the human long pentraxin PTX3. Role of NF-kappaB in tumor necrosis factor-alpha and interleukin-1beta regulation.

The “long pentraxins” are an emerging family of genes that have conserved in their carboxy-terminal halves a pentraxin domain homologous to the prototypical acute phase protein pentraxins (C-reactive protein and serum amyloid P component) and acquired novel amino-terminal domains. In this report, a genomic fragment of 1371 nucleotides from the human “long pentraxin” gene PTX3 is characterized as a promoter on tumor necrosis factor-α (TNFα) and interleukin (IL)-1β exposure in transfected 8387 human fibroblasts by chloramphenicol acetyltransferase and RNase protection assays. In the same cells, the PTX3 promoter does not respond to IL-6 stimulation. Furthermore, IL-1β and TNFα responsiveness is not seen in the Hep 3B hepatoma cell line. The minimal promoter contains one NF-κB element which is shown to be necessary for induction and able to bind p50 homodimers and p65 heterodimers but not c-Rel. Mutants in this site lose the ability to bind NF-κB proteins and to respond to TNFα and IL-1β in functional assays. Sp1- and AP-1 binding sites lying in proximity to the NF-κB site do not seem to play a major role for cytokine responsiveness. Finally, cotransfection experiments with expression vectors validate that the natural promoter contains a functional NF-κB site.

the 5Ј half of the protein does not show significant homology with other known proteins. PTX3 is indeed the first isolated member of a new group of proteins, known as "long pentraxins," which have different 5Ј-termini upstream from their pentraxin domains (5)(6)(7)(8)(9).
While the classical pentraxins CRP and SAP are almost exclusively produced by the liver in response to IL-6 in combination with IL-1 and TNF, PTX3 shows a more promiscuous response in that its expression in vitro can be induced in endothelial cells, hepatocytes, fibroblasts, and monocytes. In all cases, the gene is rapidly and directly induced by exposure to IL-1␤, TNF␣, and lipopolysaccharide (LPS), but not by IL-6, the mRNA peaking 4 -6 h after the stimulation (1,4,10). This induction is paralleled by de novo transcription of the gene (10) and is transient in that no more message is detectable after 24 h (1, 10).
The mouse homologue, mPTX3, shows a similar exon/intron organization and 82% identity at the amino acid level with hPTX3 (4). When C57BL mice were injected i.v with LPS to induce an acute phase response, mPTX3 expression was markedly induced in vivo after 4 h in several muscular organs, including the heart and the thigh (4). In situ hybridization studies showed that endothelial cells within the muscular tissues were the major responder cell type. Interestingly, in striking contrast with CRP and SAP, no mRNA for mPTX3 could be detected by Northern analysis in the liver (4).
Similar promiscuous in vivo expression has also been observed for the other "long pentraxins" in organs as diverse as the brain and the testis (6 -9).
To begin to understand the molecular mechanisms underlying the regulated expression of the first cloned long pentraxin PTX3, we cloned and characterized the promoter of hPTX3.
Plasmid Construction-A 1.37-kilobase EcoRI-PvuII genomic fragment from P2 phage (1), which spans nucleotides Ϫ1317 to ϩ54 relative to the transcription start site, was blunted and subcloned into the XhoI-digested and blunted pBL CAT 3 vector (13) (giving Ϫ1317-CAT). Following PstI-XbaI digestion, the latter plasmid was used as a substrate to generate a set of deletion clones with ExoIII (Stratagene, La Jolla, CA) digestion, mung (Stratagene) blunting, and subsequent religation. Deletion clones covering the whole 5Ј-flanking region of the hPTX3 gene were sequenced by the Sanger dideoxy method (14). Three of them, Ϫ387-CAT, Ϫ180-CAT, and Ϫ74-CAT, were subsequently used in functional assays.
The plasmid carrying a mutation in the Ϫ96 NF-B site (Ϫ180 m NF-B-CAT) was produced by polymerase chain reaction (PCR) tech-* This work was supported in part by Piano Nazionale Farmaci II, tema 5; F. B. was supported by NATO Grant CRG 941228. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) X97748.
nique. The plasmid Ϫ180-CAT was used as a substrate for amplification with four synthetic oligonucleotides (Duotech srl, Milan, Italy): oligonucleotide I (5Ј-CTATTACGCCAGCTGGCG-3Ј) was designed on a pUC-derived sequence from the pBL CAT 3 vector in sense orientation; oligonucleotide IV (5Ј-CAACGGTGGTATATCCAGTG-3Ј) was designed on the CAT sequence in antisense orientation. Oligonucleotide II (5Ј-TGCTCTAGAATCTGAATTTGGTGGGGGAGG-3Ј) was designed in antisense orientation on hPTX3 promoter from nucleotide Ϫ115 to nucleotide Ϫ97 with an additional 11 nucleotides at the 5Ј end; 6 of them (underlined) give rise to a XbaI restriction site. Oligonucleotide III (5Ј-TGCTCTAGACCGTTACCGCAGTGCCACC-3Ј) was designed in sense orientation on the hPTX3 promoter from nucleotides Ϫ88 to Ϫ69; an additional tail of 9 nucleotides includes (underlined) a XbaI site. Two PCR reactions were performed with oligonucleotides I/II and III/IV, respectively; the fragments obtained were separated on 5% native polyacrylamide gel, eluted, and cut with HindIII/XbaI and XbaI/XhoI, respectively. Finally, the digested fragments were inserted into a HindIII/ XhoI-digested pBL CAT 3 vector. The resulting construct (Ϫ180 m NF-B-CAT) carries a substitution of 8 bases at the NF-B site (AT-TCTAGA instead of GGGAACTC).
To obtain the mutant plasmid Ϫ180 m Sp1-CAT, we used an oligonucleotide carrying two mutations (underlined) in the Ϫ123 Sp1 site (oligonucleotide V 5Ј-CTCTCCCACCCATACCCCTCCCCCACCAAAT-3Ј, spanning from nucleotide Ϫ131 to nucleotide Ϫ101). This mutagenic primer, together with the flanking oligonucleotides I and IV and with the SmaI linearized wild-type Ϫ180-CAT plasmid as a substrate were used to produce a DNA fragment carrying the desired mutations according to the PCR mutagenesis technique as described (15). Similarly, to obtain the Ϫ180 m AP-1-CAT mutant plasmid, we used a mutagenic primer (oligonucleotide VI 5Ј-CCACCAGCATTACGTTTTCATC-CCCATTC-3Ј spanning from nucleotide Ϫ74 to nucleotide Ϫ46) carrying three substitutions (underlined) in the Ϫ65 AP-1 site. The double mutant Ϫ180 m NF-B/Sp1-CAT was obtained using oligonucleotides I, IV, and V and the SmaI linearized Ϫ180 m NF-B-CAT plasmid as a substrate for PCR reaction. The mutated fragments obtained were purified by polyacrylamide gel electrophoresis, eluted, digested with HindIII/XhoI, and cloned into a pBL CAT3 vector. All the PCR reactions above described were carried out using Pfu DNA polymerase (Stratagene).
Transfection and CAT Assays-8387 human fibrosarcoma (11) and Hep 3B (12) human hepatoma cells were grown to confluence, collected after trypsinization, and cultured at a density of 8 ϫ 10 5 (8387 cells) or 1 ϫ 10 6 (Hep 3B) in 100-mm dishes 24 h before the transfection. Cells were transfected by the calcium phosphate precipitation method (16) with 15 g of CAT reporter plasmid together with 2 g of pSV-␤ plasmid carrying the ␤-galactosidase gene under the control of the SV40 promoter (Promega Corp., Madison, WI). Cells were left in contact with DNA for 12 (8387 cells) or 16 (Hep 3B) h. The medium was then replaced with fresh medium. Cells were left to recover for 5 h and then stimulated with human recombinant TNF␣ (500 U/ml, BASF/Knoll, Ludwigshafen am Rhein, Germany) or with human recombinant IL-1␤ (100 ng/ml, Dompé, L'Aquila, Italy) for 24 h. Human recombinant IL-6 (Immunex Corp., Seattle, WA) was used at 50 units of Cess/ml. The cotransfection assays were performed by transfecting 8387 cells with 15 g of CAT containing constructs together with 0.5 g of pRSPA expression vector containing the cDNAs coding for NF-B p50 and p65 driven by Rous sarcoma virus promoter (17) (a kind gift of Dr. Gary J. Nabel), either alone or in combination in a 1:1 ratio. After transfection the cells were stimulated with TNF␣ (500 units/ml) or left untreated.
Cells were harvested and lysed by three cycles of freezing and thawing in 250 mM Tris buffer, pH 7.6. The transfection efficiencies were measured by ␤-galactosidase activity determination in the same amount of cell lysate (usually 40 g of proteins), measured by an enzymatic assay with chlorophenol red-␤-D-galactopyranoside (Boehringer Mannheim GmbH, Mannheim, Germany) as a substrate. Protein concentrations were determined with the Bio-Rad Protein Assay (Bio-Rad, Richmond, CA).
CAT reactions were carried out at 37°C for 3 h in an 80-l reaction mixture containing 40 g of cellular extracts, 0.5 Ci of [ 14 C]chloramphenicol (Amersham International, Little Chalfont, UK), and 5 mM acetyl-CoA (Boehringer Mannheim). The products were separated on a thin layer chromatography sheet, and the percentage of conversion to the acetylated form of chloramphenicol was quantified by scintillation counting. CAT activity values were normalized to the ␤-galactosidase activity.
RNase Protection Assay-A 382-base pair (bp) HindIII-EcoRI fragment from the construct Ϫ74-CAT, comprising the nucleotides spanning from Ϫ74 to ϩ54 (PvuII site) of the hPTX3 promoter, and a portion of the CAT gene (up to the EcoRI site) were subcloned into the pGEM-4 plasmid (Promega). The plasmid was linearized with HindIII, and a 32 P-labeled RNA probe was generated with SP6 RNA polymerase and [ 32 P]UTP according to the manufacturer's instructions.
Six plates per plasmid were transfected with the same precipitation mixture, and then three were treated with TNF␣ (500 units/ml) for 4 h and three were left untreated. At the end of the incubation time, the cells were extracted in guanidinium isothiocyanate, and the RNA was purified as described previously (18).
After inactivation of the enzymes, the hybridization products were extracted in phenol-chloroform, precipitated, and loaded onto an urea/ polyacrylamide 6% gel.
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared from 8387 cells that were stimulated with TNF␣ (500 units/ml) for 3 h or left untreated as described (19). The oligonucleotides utilized (Duotech) correspond to the NF-B site at position Ϫ96 and its flanking sequences both in a wild type (5Ј-AATTCAGGGGAACTCCCGTTACC-3Ј) and a mutated form (5Ј-AATTCAGATTCTAGACCGTTACC-3Ј). 10 g of nuclear proteins were incubated with 50 pg of 32 P-labeled oligonucleotides and 1 g of poly(dI⅐dC) (Pharmacia Biotech, Uppsala, Sweden) in 15 l of binding reaction buffer (40 mM Tris (pH 7.5), 120 mM KCl, 8% Ficoll, 4 mM EDTA, 1 mM dithiothreitol, and 10% glycerol) for 20 min at room temperature. A 1000-fold molar excess of cold oligonucleotide was used for competition assays. Competition assays were performed using oligonucleotides carrying wild-type or mutant NF-B sites and with an oligonucleotide carrying a functional NF-B site from human immunodeficiency virus type I (HIV-I) long terminal repeat (LTR) (5Ј-GATCCAGAGGGGACTTTCCGAGAGGC-3Ј) (20) which was also used for a standard binding reaction as a positive control. The resulting complexes were separated from the free probe by electrophoresis in a 5% native polyacrylamide gel in 0.5% Tris-buffered EDTA. In supershift analysis we used serum 1141 raised against an aminoterminal peptide of human p50 (21), serum 1226 raised against a carboxy-terminal peptide of human p65 (21), and serum 1136 raised against a carboxy-terminal peptide of human c-Rel (21); after 20 min of preincubation on ice, a standard binding assay was performed.

RESULTS
Cloning and Sequencing of the 5Ј-Flanking Region of hPTX3-To analyze the promoter region of the hPTX3 gene, genomic DNA sequences upstream the cDNA-encoding sequence were studied. An EcoRI-PvuII fragment, which spans nucleotides Ϫ1317 to ϩ54 (and does not include the ATG) (Fig.  1A), was subcloned for further investigation in the pBL CAT 3 expression vector (13) and identified as Ϫ1317-CAT (Fig. 1B). A series of deletion mutants obtained by ExoIII/mung directional deletions was selected as shown in Fig. 1B.
The sequence of the 5Ј-flanking region of the hPTX3 gene ( Fig. 1A) revealed features of an eukaryotic promoter, such as the presence of a number of potential binding sites for transcription factors. We have identified one NF-IL 6, two NF-B, one AP-1, two Pu.1, three PEA 3, one Ets-1, and two Sp1 consensus sequences (Fig. 1A). No obvious TATA or CAAT consensus box was found. The previously identified transcription start site (1), however, corresponds to a pyrimidine-rich 7-nucleotide consensus (22) sequence which has been reported to act in TATA-less promoters and is underlined in Fig. 1A.
While this article was in preparation, the sequence of the mPTX3 became available (23), and the alignment shows an overall 50.1% conservation, with the last 380 nucleotides showing a 66% conservation (Fig. 1A). Furthermore, two of the reported potential binding sites, the NF-B and the AP-1 sites at positions Ϫ96 and Ϫ65, respectively, are maintained at approximately the same positions in both sequences (Fig. 1A).
Functional Analysis of the Promoter-The deletion mutants schematically shown in Fig. 1B were used for expression studies in 8387 human fibrosarcoma cells and in Hep 3B human hepatoma cells. After calcium phosphate-mediated transfection of the cells, cultures were left untreated or were treated with TNF␣ or IL-1␤ for 24 h. Data from at least four separate experiments are shown in Fig. 2. The Ϫ1317-CAT construct shows a 5.3-fold basal activity with respect to the empty vector, and a comparable level is observed also with the Ϫ387 construct, implying that the 1000 intervening nucleotides do not contribute significantly to this basal activity. In contrast, the Ϫ180 construct has a 2-fold higher basal level, while further deletion up to Ϫ74 abolishes almost completely the activity.
TNF␣ exposure (gray bars) results in a 2.5-fold induction with the Ϫ1317, Ϫ387, and Ϫ180 CAT constructs, but it is completely inactive on the Ϫ74 construct. IL-1␤ exposure (hatched bars) induced a quite similar effect in the conditions tested. On the contrary, IL-6 assayed on 8387 cells transfected with the Ϫ1317 construct was completely inactive and did not modify the responsiveness to TNF␣ (data not shown). In the same experimental setting, an artificial CAT reporter construct containing four tandem NF-B sites from IL-6 promoter (24) cloned into a pBL CAT 2 vector (13) gave a mean fold induction of 4.5 Ϯ 0.35 times on TNF␣ induction compared with the untreated cells (data not shown).
To validate these results with a different approach, we analyzed the 8387-transfected cells by RNase protection. As shown in Fig. 3, lane 1, the undigested riboprobe corresponds to the predicted size of 383 nucleotides. In the transfected 8387 cells (lanes 4 -13), the protected band is 309 bp in all cases, thus showing the use of the same transcriptional start site in the artificial constructs as in the wild-type gene (1).
While no precise measurement can be made for the baseline values of the different constructs among them due to possible variation in transfection efficiency, for each experimental group the untreated and treated cells can be compared. TNF␣ indeed increases the transcription of the three responsive constructs by 3-6-fold (as determined by densitometric scanning) (lanes 4 -9), whereas construct Ϫ74 shows no activity in both unstimulated and stimulated cells (lanes 12 and 13). These data have been reproduced in three separate experiments.
To study the cellular specificity of the TNF␣/IL-1␤ responsiveness, we transfected the same constructs in the human hepatoma cell line Hep 3B; although PTX3 mRNA is inducible in these cells by TNF␣ and IL-1␤ exposure (1), we could not observe, either in unstimulated or in TNF␣-stimulated cells, a significant CAT activity with respect to the empty vector-transfected cells (data not shown) despite a good transfection efficiency. On the other hand, the artificial CAT reporter construct containing four tandem NF-B sites (see above) showed, under the same experimental conditions, a full responsiveness to TNF␣ (Ͼ15-fold; data not shown). All these data together imply that further genomic elements or posttranslational modifications may crucially contribute to the hepatic transcription of PTX3 detectable in vitro.
NF-B p50 and p65 Can Bind to the hPTX3 Promoter-The sharp difference in the activity between Ϫ180 and Ϫ74, the presence of an NF-B element at position Ϫ96 (Fig. 1A), and the known effect of NF-B in TNF␣-and IL-1␤-mediated responses prompted us to analyze the involvement of NF-B in PTX3 regulation by electrophoretic mobility shift assay. An oligonucleotide corresponding to the sequence from position Ϫ103 to Ϫ81 of hPTX3 was utilized. A low level of binding activity was detectable in untreated 8387 cells as two separate bands (Fig. 4A, lane 1), and they were clearly increased after a 3-h TNF␣ stimulation (lane 2). The specificity of the binding activity is documented by the complete competition with the cold specific oligonucleotides (lane 3) and with an oligonucleotide containing a canonical NF-B site from HIV-1 LTR (lane 5) (20).
We also generated a mutant hPTX3 NF-B oligonucleotide which did not contain a NF-B-binding site. This mutated oligonucleotide was unable to compete for the binding of the NF-B proteins to the wild-type sequence (lane 4) and, on the other hand, when used as a probe, did not show binding activity in either untreated (lane 6) and TNF␣-treated (lane 7) 8387 cells.
Supershifting with antibodies clearly indicated that the two bands correspond to the p50/p65 heterodimer and to the p50/ p50 homodimer, respectively (Fig. 4B, lanes 3 and 4). Furthermore, c-Rel is not present in this complex (lane 5) as demonstrated by the lack of supershifting, similar to what is observed with an irrelevant antibody (lane 6).
We further compared under the same experimental conditions the binding activity of 8387 nuclear extracts on an oligonucleotide containing the PTX3 NF-B site and on an oligonucleotide containing a canonical NF-B-binding site from HIV-1 LTR, which has been previously reported to give rise to only one retarded complex (20). As shown in Fig. 4C, while the binding on the PTX3 oligonucleotide gave rise to two retarded complexes, corresponding to p50/p50 and p50/p65 homo-and heterodimers, only the upper band was present in the binding to the HIV-1 LTR NF-B site.
The NF-B Site Is Functionally Relevant in the hPTX3 Promoter-To more directly assess the functional relevance of the NF-B site, we mutagenized this site in the Ϫ180 construct, (Ϫ180 m NF-B-CAT). The mutated sequence is identical to the degenerated oligonucleotide that we had used in the gel retardation experiments (Fig. 4A, lanes 6 and 7). When the Ϫ180 m NF-B-CAT construct was analyzed by CAT analysis and RNase protection, it was evident that despite a detectable level of basal activity, cytokines exposure did not lead to any significant induction of its transcription (Fig. 2, Ϫ180 m NF-B-CAT, and Fig. 3, lanes 10 and 11), thus implying an NF-B mediated induction of the transcriptional activity of the hPTX3 promoter by TNF␣ and IL-1␤.
To validate the hypothesis that NF-B is the key responsive element, we also mutagenized the Sp 1 and AP-1 sites which are present in the minimal promoter (as shown in Fig. 1b). The Ϫ180 m Sp 1-CAT was indeed still responsive (Fig. 2), although at a lower level compared with the wild type Ϫ180-CAT construct (1.7 mean fold induction over four separate experiments upon TNF␣ induction). As expected, also the double mutant Ϫ180 m NF-B/Sp 1-CAT is not responsive to TNF␣ and IL-1␤ stimulation, although it retains a basal activity. On the other hand, the AP-1 mutant Ϫ180 m AP-1-CAT shows full cytokine responsiveness (3.8 and 2.7 mean fold induction with TNF␣ and IL-1␤, respectively), but a much lower basal level of CAT expression (Fig. 2).
To quantify the observed stimulations with respect to canonical NF-B sites, we made use in the same experimental setting, of an artificial CAT reporter construct containing four tandem NF-B sites derived from IL-6 promoter (24) and cloned in a pBL CAT 2 vector (13). This reporter gave a mean fold induction of 4.5 Ϯ 0.35 upon TNF␣ stimulation relatively to untreated cells (data not shown), therefore of comparable entity to those observed with the PTX3 constructs.
Cotransfection with Expression Vectors-To further substantiate that the NF-B site is indeed the main functional responsive element, we cotransfected 8387 cells with the Ϫ180-CAT and with the Ϫ180 m NF-B-CAT constructs as reporters together with p50 and p65 NF-B expression vectors (17) either alone or in combination. As shown in Fig. 5, top, cotransfection of p50 alone did not modify the basal CAT activity by the construct in the wild type configuration, as compared with cells transfected with the empty vector pRSPA, while p65 and the combination of the two increased the basal activity by a factor of 3.6 (p50/p65) to 3.9 (p65 alone). The addition of TNF␣ was effective in all the experimental conditions tested (3, 2.6, 1.7, and 1.7 fold induction over the unstimulated cells in pRSPA, p50, p65, and p50/p65 transfected cells respectively). On the other hand, overexpression of p65 alone, or in combination with p50, as well as addition of TNF␣ had no effect on the Ϫ180 m NF-B-CAT construct carrying the mutation in the Ϫ96 NF-B site (Fig. 5, bottom). These data are consistent with the hypothesis that the hPTX3 natural promoter contains a functional NF-B site.
Under the same experimental conditions, the control reporter plasmid containing four NF-B binding sites from the IL-6 promoter was induced by p50/p65 overexpression by 5.9fold, while no induction was detectable against a pSV2 CAT reporter plasmid utilized as a negative control (not containing FIG. 3. RNase protection analysis of the transcriptional activity of hPTX3 promoter. Cells were transfected as in Fig. 2. The CAT mRNA levels in the extracts were quantified by RNase protection assay. That each sample pair contained similar levels of RNA was checked by hybridization with an antisense ␤-actin (␤-ACT) probe. a NF-B element) when cotransfected with p50 and p65 in combination (data not shown).

DISCUSSION
In this report, we characterize the promoter of the human PTX3 gene (hPTX3). A genomic fragment of 1317 bp, located 5Ј to the transcriptional start site, responds to TNF␣ and IL-1␤ stimulation in transiently transfected human 8387 fibroblasts but not in human hepatoma Hep 3B cells, as measured by transfection and CAT assays (more than 2-fold induction) and by RNase protection analysis (3-6-fold induction). Deletion mutants show that the 180 bp more proximal to the start site are sufficient for TNF␣and IL-1␤-inducible transcriptional activity. On the contrary, the last 74 bp are unresponsive. In the intervening 106 bp we show that a classical NF-B binding site is present and furthermore that p50/p50 homodimers and p50/p65 heterodimers can bind to this element after incubation with nuclear extracts from 8387 fibroblasts. TNF␣ exposure increases this NF-B activity, while the minimal construct carrying an inactivating mutation of this site loses the TNF␣ inducibility in the same cells. Finally, we confirmed the hypothesis that NF-B proteins are functionally active on the hPTX3 promoter by cotransfection with p50 and p65 NF-B expression vectors. On the contrary, Sp1 and AP-1 do not seem to play a major role for the cytokine inducibility of the gene.
These data show for the first time that a classical NF-B complex can functionally interact with the "long pentraxin" hPTX3 promoter in human fibroblasts after exposure to TNF␣. The different methods utilized indicate a 2-5-fold transcriptional induction of the gene, which is in agreement with the observed increase in nuclear runoff experiments on isolated monocytes (10).
PTX3 belongs structurally to the family of the classical acute phase protein pentraxins, which include CRP and SAP, in several animal species. Both genes are characteristically induced by IL-6 in combination with IL-1 and TNF mainly, if not exclusively, in hepatocytes (25)(26)(27)(28)(29)(30)(31). PTX3 was the first cloned member of the newly emerging group of "long pentraxins," (5-9) because they show a long amino-terminal domain fused to the carboxy-terminal pentraxin domain (corresponding to most of the classical pentraxin sequence). The significance of this genetic acquisition is far from being understood, but all the long pentraxins do not show a liver-restricted expression pattern and seem to be expressed in a much wider spectrum of organs, such as the brain and the testis (6 -9). Furthermore, hPTX3 was shown to be transcribed after exposure to IL-1␤, TNF␣, and the bacterial product LPS (i.e. all prototypical proinflammatory signals) but not by IL-6, in several different cell types, including endothelial cells, fibroblasts, hepatocytes, and monocytes (1,10). Furthermore, PTX3 expression induced by IL-1␤ is not modified in endothelial cells and hepatocytes by concomitant exposure to IL-6 (data not shown). The same gene was cloned in TNF␣-stimulated fibroblasts, named TSG-14 (2), and demonstrated to be directly induced by TNF␣ (32,33).
Inducibility by IL-1␤ and TNF␣, but not by IL-6, may correlate well with the demonstrated role of NF-B (for review, see Refs. 34 -36) and, furthermore, with the presence of only one NF-IL 6 binding site (37,38) and with the absence of APRF elements (39 -41) in the human promoter. Both elements, in fact, have been demonstrated to be necessary in multiple copies for IL-6 inducibility (42).
The murine gene (82% identical at the amino acid level) shows a similar exon/intron organization and is localized on a syntenic chromosomal region (4). It is induced in vitro only in peritoneal macrophages, in some fibroblasts, and in very few endothelial cell lines, but not in hepatocytes; on the other hand, it was induced in vivo in several muscular tissues after LPS i.v. injection (an acute phase experimental model), but not in the liver (4,32). In addition, in situ hybridization studies have indicated that in the heart and in the thigh, the endothelial cells were the most abundant producer cell type (4).
The alignment between the human and the murine promoters shows a high overall degree of conservation, including few hypothetical binding sites for transcription factors, in particular the NF-B site which is here demonstrated as functionally important for the hPTX3 gene. What are the structural reasons for the differences in the expression of PTX3 between humans and mice is still unclear.
The reported lack of consensus sites for hepatic transcription factors in the murine promoter (and in the human promoter as well) may account in part for the absence of induction in the liver (another obvious difference with the classical CRP and SAP genes (3), but the positive elements required for its inducibility in the endothelial cells of the muscular district have yet to be elucidated. On the other hand, recent work with transgenic animals has shown, in the case of CRP, that the precise characterization of the functional elements required for the in vivo "acute phase" inducibility may require a complex interaction between 5Ј and 3Ј elements (43), which was unexpected on the basis of previous in vitro studies (27)(28)(29)31).
We have described the functional role of the NF-B site in the promoter of the hPTX3 gene for TNF␣ inducibility in fibroblasts. We can only speculate at the moment, on the basis of the large amount of published data, that this same site may be relevant also for LPS inducibility of the gene in fibroblasts as well as in other cell types. NF-B may interact with other factors as suggested by others (44 -48), particularly in view of the presence of AP-1 and Sp1 sites in close proximity to the NF-B site, and of the fact that they both have been reported to interact functionally with NF-B complexes (44,48). Indeed, the Sp1 mutant shows a reduced TNF␣ and IL-1␤ inducibility, while the AP-1 mutant is fully responsive to cytokines, although its basal level of expression is significantly reduced. Further work will be required to directly address the possible interplay of different transcription complexes on the hPTX3 promoter.