Bacterial Peptidoglycan Induces CD14-dependent Activation of Transcription Factors CREB/ATF and AP-1*

Peptidoglycan (PGN), the major cell wall component of Gram-positive bacteria, induces secretion of cytokines in macrophages through CD14, the pattern recognition receptor that binds lipopolysaccharide and other microbial products. To begin to elucidate the mechanisms that regulate the transcription of cytokine genes, we wanted to determine which transcription factors are activated by PGN in mouse RAW264.7 and human THP-1 macrophage cells. Our results demonstrated that: (i) PGN induced phosphorylation of the transcription factors ATF-1 and CREB; (ii) ATF-1 and CREB bound DNA as a dimer and induced transcriptional activation of a CRE reporter plasmid, which was inhibited by dominant negative CREB and ATF-1; (iii) PGN induced phosphorylation of c-Jun, protein synthesis of JunB and c-Fos, and transcriptional activation of the AP-1 reporter plasmid, which was inhibited by dominant negative c-Fos; and (iv) PGN-induced activation of CREB/ATF and AP-1 was mediated through CD14. This is the first study to demonstrate activation of CREB/ATF and AP-1 transcription factors by PGN or by any other component of Gram-positive bacteria.

TNF-␣ (6,16), IL-1 (7), and IL-6 (7,16) are the main cytokines released from PGN-activated macrophages. However, the signal transduction pathways that culminate in transcriptional activation of these cytokine genes are so far unknown. The promoters for these cytokine genes contain binding sites for various transcription factors, including NF-B, AP-1, and CREB. We have previously shown activation of NF-B by PGN (8), but it is not known if PGN induces activation of any other transcription factors. It is also not known which transcription factors are required for PGN-induced activation of cytokine genes.
The CREB/ATF family of transcription factors are leucine zipper proteins that bind to the cAMP response element (CRE) with the consensus sequence, 5Ј-TGACGTCA-3Ј (17). CREB, the most extensively studied CRE-binding protein, is phosphorylated at serine 133 by protein kinase A in response to cAMP, and this leads to transcriptional activation (17) of genes whose promoters contain the CRE sequence. There are other signaling pathways that lead to phosphorylation and activation of CREB, such as calmodulin kinase, which phosphorylates CREB in response to increased intracellular Ca 2ϩ (17), or RSK2 which is activated by mitogen-activated protein (MAP) kinases (18). ATF-1, another member of this family of transcription factors, has significant sequence similarity to CREB, including a phosphorylation domain (17). ATF-1 forms heterodimers only with CREB; however, the other ATF proteins can also form heterodimers with specific members of the AP-1 family of transcription factors. In addition, different heterodimers may bind variant CRE sequences, thus increasing the number of potential regulatory mechanisms.
The AP-1 family of transcription factors consists of the Jun and Fos families of proteins that bind the 12-O-tetradecanoylphorbol-13-acetate response element (TPA-RE), 5Ј-TGAC-TCA-3Ј and induce transcription in response to many different stimuli, including phorbol esters (19). These proteins bind DNA as dimers, which are formed through leucine zippers. The Jun proteins can bind DNA as homodimers or as heterodimers with Fos proteins; however, the Fos proteins can only bind DNA as heterodimers (19). In addition to forming heterodimers within the AP-1 family, Jun proteins can heterodimerize with certain members of other transcription factor families, such as ATF and C/EBP␤ (19). Different dimers can bind different sequences, e.g. Jun-Jun and Jun-Fos dimers preferentially bind TPA-RE, while Jun-ATF dimers prefer to bind the CRE sequence (19).
The objective of this study was to: (i) determine if PGN activates the transcription factors CREB/ATF and AP-1, (ii) identify the specific members of these two families of transcrip-* This work was supported by National Institutes of Health Grant AI28797 (to R. D.).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 (20) and purified by vancomycin affinity chromatography, and its purity has been previously described (21). sPGN contained Ͻ24 pg of endotoxin/mg, as determined by the Limulus lysate assay (21). LPS from Salmonella minnesota Re 595 (ReLPS, a minimal naturally occurring endotoxic structure of LPS), obtained by phenol-chloroform-petroleum ether extraction, was purchased from Sigma. All other chemicals were from Sigma, unless otherwise indicated.
Phosphorylation of ATF-1 and CREB and Western Blots-RAW264.7 cells were seeded at 0.35-0.4 ϫ 10 6 /ml in 24-well plates (2 ml/well) and cultured for 16 -20 h. THP-1 cells were seeded at 0.15 ϫ 10 6 /ml in 24-well plates (2 ml/well) and allowed to differentiate as above for 72 h. All cells were activated with the stimulants indicated under "Results" and then washed and lysed as before (6). In some experiments, sPGN and ReLPS were incubated with 5 g/ml polymyxin B for 30 min and then added to cells. In other experiments, THP-1 cells were incubated with 10 g/ml anti-CD14 monoclonal antibodies, MY4 (Coulter, Hialeah, FL) or MEM18 (Sanbio-Monosan, Uden, The Netherlands), or the isotype control IgG2b (clone MPC-11; Coulter) for 30 min at 37°C before stimulation. Cell lysates were separated on 12% SDS-PAGE and transferred to Immobilon P (6). Phosphorylation of ATF-1 and CREB was determined by Western blotting with 0.5 g/ml rabbit anti-pCREB antibody (Upstate Biotechnology, Inc., Lake Placid, NY), and detected by the ECL system. This antibody specifically recognizes the phosphorylated forms of both ATF-1 and CREB. Control blots were done with 0.5 g/ml rabbit anti-CREB (Upstate Biotechnology) or rabbit anti-ATF-1 monoclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), which recognize both phosphorylated and nonphosphorylated forms of CREB or ATF-1, respectively.
Activation of AP-1 and Western Blots-Cells were cultured, activated, and lysed as described for ATF-1/CREB. Phosphorylation of c-Jun protein was detected using a monoclonal antibody generated against a c-Jun peptide containing phosphorylated serine 63 (Santa Cruz Biotechnology). Nonphosphorylated c-Jun, Jun B, Jun D, and c-Fos were also detected by Western blots using antibodies from Santa Cruz Biotechnology.
Enzyme Digestions-30-g aliquots of sPGN biosynthetically labeled with [ 14 C]alanine (20) were digested for 72 h at 37°C with 1 mg/ml affinity-purified lysostaphin or 1 mg/ml lysozyme (Grade I from chicken egg, from Sigma), or buffer alone (as a control), and were dialyzed four times (10 -12-kDa cut-off) against Dulbecco's phosphate-buffered saline at 4°C (10). The extent of digestion was determined by measuring the amount of 14 C remaining in the samples after dialysis.
Phosphatase Treatment-Stimulated and control RAW264.7 cells were lysed in three different buffers depending on the phosphatase treatment (22). The first group of cells was lysed in the same buffer as above with 0.8% Nonidet P-40 for control samples. The second group of cells was lysed in 50 mM Tris, pH 7.5, 1 mM MgCl 2 , 0.8% Nonidet P-40, and protease inhibitors, and digested with 250 units/ml alkaline phosphatase for 15 min at 30°C. The third group of cells was lysed in 50 mM Tris, pH 7.5, 1 mM MnCl 2 , 1 mM dithiothreitol, with 0.8% Nonidet P-40, 1.0 nM okadaic acid, and protease inhibitors, and digested with 2.9 units/ml protein phosphatase 2A (Calbiochem, San Diego, CA) for 20 min at 37°C. The final Nonidet P-40 concentration in all samples was 0.6%. The samples were separated on 12% SDS-PAGE and analyzed for phosphorylated and nonphosphorylated c-Jun by Western blots.
Electrophoretic Mobility Shift Assays-Cells were cultured as described above for Western blots, and nuclear extracts were prepared as described previously (8). 5 g of nuclear protein were first incubated with different antibodies as indicated under "Results," for 20 min at 22°C. The nuclear proteins were then incubated with 32 P-labeled oligonucleotide containing consensus CRE (17) or TPA-RE (19) binding sites (AGAGATTGCCTGACGTCAGAGAG and AGCTTGATGACT-CAGCCG, respectively), for 30 min at 22°C. Anti-Jun antibody was from Santa Cruz Biotechnology, and all other antibodies used in these experiments were as described above. In some experiments, the samples were preincubated with an excess of unlabeled specific or nonspecific oligonucleotide for 20 min at 22°C. All samples were then separated on 5% nondenaturing polyacrylamide gels, and the DNA-protein complexes were visualized by autoradiography.
Transfection and Chloramphenicol Acetyltransferase (CAT) and Luciferase Assays-RAW264.7 cells were cultured at 0.35-0.4 ϫ 10 6 /ml in six-well plates (2 ml/well) for 16 -20 h and transfected with 200 g/ml DEAE-dextran and DNA (the amount of DNA used was optimized for different plasmids and is indicated in the figure legends). The following plasmids were used for transfections: (27), and A-ATF1, which was constructed as described previously (24). Cells were allowed to recover for 24 -48 h and then were left unstimulated or were stimulated as described in the figure legends. Lysates were prepared and were assayed for luciferase activity using the Luciferase Reporter kit (Promega, Madison, WI) or for CAT activity as described (22).

RESULTS
sPGN Induces Phosphorylation of ATF-1 and CREB-Activation of the transcription factor CREB is regulated by phosphorylation at serine 133. ATF-1, also a CRE-binding protein, has extensive homology to CREB, including a conserved phosphorylation site. We tested if sPGN induces phosphorylation of CREB and ATF-1 in RAW264.7 cells, using an antibody that specifically recognizes both phosphorylated CREB and phosphorylated ATF-1. sPGN induced rapid and transient dose-dependent phosphorylation of both ATF-1 and CREB (Fig. 1, A and D) with similar kinetics. The control stimulant, ReLPS, also induced phosphorylation of ATF-1 and CREB with kinetics similar to those seen with sPGN. The phosphorylation of ATF-1 was stronger than phosphorylation of CREB for both stimulants. Identical samples analyzed with antibodies that recognize both phosphorylated and nonphosphorylated CREB ( sPGN-induced phosphorylation of ATF-1 and CREB was not inhibited by polymyxin B (an antibiotic that binds LPS and inhibits its biologic effects), unlike the induction by ReLPS, which was almost completely inhibited (Fig. 1D). This confirms that the sPGN-induced effect was not due to endotoxin contamination. In control experiments, the amounts of total (phosphorylated and nonphosphorylated) CREB ( Fig. 1E) and ATF-1 (not shown) proteins were the same in all samples.
Composition of CREB/ATF Complexes That Bind CRE Consensus Sequence-To determine if CREB and ATF-1 bind to the CRE sequence and to determine the composition of the CREbinding complexes, supershift assays were performed with an oligonucleotide containing the CRE consensus sequence and antibodies to specific proteins in the CREB/ATF and AP-1 families of transcription factors. The AP-1 family was included because Jun proteins also bind CRE sequence.
Nuclear extracts from sPGN-or ReLPS-stimulated RAW264.7 cells contained proteins that bind to the CRE consensus site (the protein-DNA complex ran higher than the free oligonucleotide) (Fig. 2A). There was no difference in the binding of proteins to the CRE site between stimulated and control cells ( Fig. 2A, compare lanes 1 for Nil, sPGN, and ReLPS samples). This is characteristic of the protein complex that binds CRE sequence, where activation does not result in a change of the binding of the proteins, but induces phosphorylation of the already bound proteins (17).
The protein-DNA complexes with nuclear extracts from sPGN-or ReLPS-stimulated cells were supershifted by antibodies against CREB-1, pCREB, and ATF-1, but not against ATF-2, Jun, and c-Fos ( Fig. 2A). This indicates that the protein dimer that binds CRE consensus sequence in sPGN-or ReLPSstimulated RAW264.7 cells consists of ATF-1 and CREB. There were no qualitative differences in the specific transcription factors that bind the CRE site between stimulated and nonstimulated cells; however, there was a difference in the amount of CREB and pCREB in the bound complex between stimulated and nonstimulated cells ( Fig. 2A). The amount of pCREB increased upon stimulation by sPGN or ReLPS, as the antibodies against pCREB caused a complete shift of the protein-oligonucleotide complex from the stimulated cells (compare the pCREB lanes between stimulated and control in Fig. 2A). These results confirm that upon stimulation DNA-bound CREB undergoes phosphorylation, which has been shown to activate CREB.
The specificity of binding to the CRE sequence was confirmed using excess unlabeled specific and nonspecific oligonucleotides (Fig. 2B). In nuclear extracts from both sPGN-or ReLPSstimulated cells, the binding of proteins to the oligonucleotide carrying CRE sequence was inhibited by an excess of unlabeled specific oligonucleotide (CRE), but not by a nonspecific (NS) oligonucleotide with no CRE sequence (Fig. 2B). These data indicate that the proteins that bind the CRE oligonucleotide specifically recognize the CRE sequence. RAW264.7 cells with the plasmid ⌬(Ϫ71)Som-CAT or empty vector (CMV) and tested for chloramphenicol acetyltransferase (CAT) activity in the transfected cells after stimulation with sPGN or ReLPS. The plasmid ⌬(Ϫ71)Som-CAT contains Ϫ71 to ϩ53 bp of somatostatin promoter fused to the CAT gene, and this plasmid has been shown previously to be regulated by CREB (23). Both sPGN and ReLPS induced 14.1 Ϯ 2.5-fold and 13.4 Ϯ 2.1-fold (means Ϯ S.E., n ϭ 6) increase, respectively, in CAT activity in transfected macrophage cells (Fig. 3A). RAW264.7 cells transfected with the empty vector showed no induced CAT activity upon stimulation (Fig. 3A).
To determine if the induced CAT activity in sPGN-or ReLPS-stimulated cells was due to ATF-1 and/or CREB, we co-transfected RAW264.7 cells with ⌬(Ϫ71)Som-CAT and A-ATF1 (a dominant negative inhibitor of ATF-1), A-CREB (a dominant negative inhibitor of CREB), or the empty CMV. These dominant negative proteins were constructed by fusing an acidic extension at the N terminus of the leucine zipper domain. This acidic region of the recombinant protein binds the basic region of the wild type protein, and the basic region is thus no longer available for binding to DNA. Both dominant negative ATF-1 and dominant negative CREB, individually or in combination, inhibited the sPGN-or ReLPS-induced activation of CAT reporter gene (Fig. 3B). This inhibition was specific for A-ATF1 and A-CREB, as the empty vector did not inhibit sPGN-or ReLPS-induced CAT activity. These results provide evidence that sPGN and ReLPS induce functional activation of the transcription factors ATF-1 and CREB.  (Fig. 4, A and C). ReLPS induced phosphorylation of c-Jun similar to that seen with sPGN (Fig. 4, A and C). Stripping and reprobing the blots with anti-c-Jun antibody, which recognizes both phosphorylated and nonphosphorylated protein, revealed that sPGN and ReLPS stimulation also causes a modest increase (1.5-2 times) in the total (phosphorylated and nonphosphorylated) amount of c-Jun protein (Fig. 4, B and D). This is not unexpected as activated c-Jun protein induces transcription of the c-Jun gene (19).
sPGN-induced phosphorylation of c-Jun was not inhibited by polymyxin B, in contrast to the phosphorylation induced by ReLPS, which was completely inhibited by polymyxin B (Fig.  4C). These data confirm that sPGN-induced activation of RAW264.7 cells and phosphorylation of c-Jun is due to PGN and not due to an endotoxin contaminant in our sPGN preparation. In control experiments, the amount of c-Jun protein did not change with polymyxin B treatment, for either stimulant (Fig. 4D).
To confirm that the anti-phosphorylated c-Jun antibody recognizes phosphorylated c-Jun, samples were treated with different phosphatases and then analyzed by Western blot. Cell lysates from both sPGN-and ReLPS-stimulated cells, treated with alkaline phosphatase, a nonspecific phosphatase, did not show any binding to the anti-phosphorylated c-Jun antibody (Fig. 4E). Furthermore, protein phosphatase 2A, a serine phosphatase, strongly diminished binding to the anti-phosphorylated c-Jun antibody (Fig. 4E). However, the binding to the anti-c-Jun antibody was not eliminated or reduced by either

FIG. 2. CREB and ATF-1 bind to CRE consensus sequence in sPGN-or ReLPS-stimulated cells.
A, RAW264.7 cells were stimulated with 10 g/ml sPGN or 10 ng/ml ReLPS for 30 min and nuclear extracts were preincubated with the indicated Abs for 20 min, followed by a 30-min incubation with a radioactively labeled oligonucleotide containing a CRE consensus sequence. The DNA-protein complexes were separated on nondenaturing polyacrylamide gel. The arrows indicate the shift in the DNA-protein complexes upon Ab binding. B, specificity of binding to CRE sites was demonstrated using an excess of unlabeled specific (CRE) and nonspecific (NS) oligonucleotides. The results are from one of two similar experiments. ATF1). A, RAW264.7 cells were transfected with 1.0 g/ml CREBregulated CAT plasmid, ⌬(Ϫ71)Som-CAT, or empty vector, pUC18; 24 h after transfection, cells were stimulated with 10 g/ml sPGN or 10 ng/ml ReLPS for 36 h or left unstimulated. Cell lysates were prepared, and aliquots were assayed for CAT activity. The results are means of duplicate samples from one out of six similar experiments. B, cells were co-transfected with 1.0 g/ml ⌬(Ϫ71)Som-CAT and 2.0 g/ml A-ATF1, or A-CREB, or empty vector (CMV), or 2.0 g/ml each of A-ATF1 and A-CREB or 4.0 g/ml CMV and were stimulated and assayed as in A. The percent of empty vector ϭ (mean cpm stimulated with dominant negative Ϫ mean cpm unstimulated with dominant negative) ϫ 100/ (mean cpm stimulated with empty vector Ϫ mean cpm unstimulated with empty vector). The results are means Ϯ S.E. from three experiments.

FIG. 3. sPGN and ReLPS induce transactivation of a CREBregulated gene, and this induction is inhibited by dominant negative CREB (A-CREB) and dominant negative ATF-1 (A-
phosphatase treatment (Fig. 4F). These results confirmed that the anti-phosphorylated c-Jun antibody was indeed specific for the phosphorylated c-Jun.
JunB, JunD, and c-Fos are other members of the AP-1 family of transcription factors. sPGN consistently induced a dose-dependent increase of JunB and c-Fos, but not of JunD protein synthesis in RAW264.7 cells (Fig. 5). ReLPS also induced increases in JunB and c-Fos proteins, but not of JunD protein (Fig. 5). The up-regulation of JunB by sPGN was not inhibited by polymyxin B, in contrast to ReLPS-induced increase in JunB protein, which was completely inhibited by polymyxin B (Fig.  5B). c-Jun, JunB, and c-Fos Bind to the TPA-RE Consensus Sequence in sPGN-stimulated Cells-To identify the specific members of the AP-1 family of transcription factors that bind the TPA-RE consensus sequence, gel shift assays were performed. The binding of proteins to an oligonucleotide with the TPA-RE sequence showed no differences between sPGN-or ReLPS-stimulated and unstimulated RAW264.7 cells (Fig. 6; gel shift assays with unstimulated lysates are not shown). Using a supershift assay and a series of antibodies that recognize different members of the AP-1 and CREB family of transcription factors, we determined that in both sPGN-and ReLPS-stimulated cells (Fig. 6A) and in unstimulated cells (data not shown) c-Jun, JunB, and c-Fos bind to the TPA-RE sequence. The specificity of the oligonucleotide that was used in the supershift assays was confirmed by inhibition of binding of nuclear proteins by an excess of unlabeled specific oligonucleotide, but not by a nonspecific oligonucleotide with no TPA-RE sequence (Fig. 6B).
sPGN Induces Transactivation of an AP-1-regulated Gene That Is Inhibited by Dominant Negative c-Fos-To determine if sPGN induced functional activation of AP-1, we transfected RAW264.7 cells with the plasmid pAP1-Luc, which has seven AP-1 binding sites upstream of the luciferase gene, or Ϫ60Col-Luc, which has no AP-1 binding sites. Cells transfected with pAP1-Luc showed an average 9.0 Ϯ 0.6-fold and 9.4 Ϯ 1.4-fold (means Ϯ S.E., n ϭ 3) increase in inducible luciferase activity when stimulated by sPGN or ReLPS, respectively (Fig. 7A). Cells transfected with the control plasmid Ϫ60Col-Luc showed no luciferase activity in the absence or presence of the stimulants (Fig. 7A). We also confirmed activation of AP-1 by sPGN or ReLPS using two additional plasmids: Ϫ73Col-Luc, which has Ϫ73 to ϩ63 bp of the collagenase promoter fused to the luciferase gene, and 2xAP1-Luc, which has two AP-1 binding sites upstream of the luciferase gene (data not shown). These results demonstrate that sPGN and ReLPS induced functional activation of the transcription factor AP-1.

FIG. 7. sPGN and ReLPS induce transactivation of an AP-1regulated gene, and this induction is inhibited by dominant negative c-Fos.
A, RAW264.7 cells were transfected with 0.5 g/ml pAP1-Luc plasmid or the empty vector Ϫ60Col-Luc; 48 h after transfection, cells were stimulated with 10 g/ml sPGN or 10 ng/ml ReLPS for 5 h or left unstimulated. Cell lysates were prepared, and aliquots were assayed for luciferase activity. The results are means of duplicate samples from one out of three similar experiments. B, RAW264.7 cells were co-transfected with 0.5 g/ml pAP1-Luc plasmid and 1.0 g/ml dominant negative A-Fos, or empty vector (CMV). Cells were stimulated and assayed as in A. The percentage of control vector was calculated as in Fig. 3. The results are means Ϯ S.E. from three experiments. To confirm that the sPGN-or ReLPS-induced luciferase activity was due to AP-1, we co-transfected RAW264.7 cells with pAP1-Luc and A-Fos, a dominant negative mutant of c-Fos, or empty vector (CMV). A-Fos heterodimerizes with Jun proteins in an AP-1 complex and inactivates their ability to bind DNA. A-Fos, but not the empty vector, inhibited both sPGN-and ReLPS-induced increases in luciferase activity (Fig. 7B). These results confirm that sPGN and ReLPS induce a functionally active AP-1, and that c-Fos and Jun form the active AP-1 complex.
sPGN-induced Phosphorylation of CREB and c-Jun in Human Monocytes Is CD14-dependent-Since we have previously shown that sPGN-induced activation of NF-B is mediated through the membrane receptor CD14 (8), we now determined if sPGN-induced phosphorylation of CREB and c-Jun also requires CD14. We first tested if sPGN induced phosphorylation of CREB and c-Jun in the human monocyte cell line THP-1. THP-1 cells activated with sPGN and ReLPS showed phosphorylation of CREB (Fig. 8A). THP-1 cells differed from the mouse RAW264.7 cells, in that they did not have detectable levels of ATF-1, which in RAW264.7 cells was present in high amounts (Fig. 1C) and was strongly phosphorylated upon stimulation of the cells (Fig. 1A). Total amount of CREB did not change upon cell activation by sPGN or ReLPS (Fig. 8B).
sPGN or ReLPS also induced phosphorylation of c-Jun in THP-1 cells (Fig. 8C) and an increase in the total amount of c-Jun protein (Fig. 8D). The increase in c-Jun protein may be due to the activated c-Jun itself, which is known to induce transcription of the c-Jun gene (19). Anti-CD14 monoclonal antibodies, MY4 and MEM18, inhibited both sPGN-and ReLPS-induced phosphorylation of CREB (Fig. 8E) and c-Jun (Fig. 8F). These data confirm that sPGN and ReLPS activate cells through CD14 and that CD14 is required for both sPGN-and ReLPS-induced phosphorylation of CREB and c-Jun.
Lysostaphin and Lysozyme Reduce sPGN-induced Phosphorylation of ATF-1, CREB, and c-Jun-To confirm the identity of sPGN as the activating molecule that induces phosphorylation of ATF-1, CREB, and c-Jun, sPGN was digested with lysostaphin or lysozyme, enzymes that specifically degrade PGN. Digestion with both enzymes reduced sPGN-induced phosphorylation of ATF-1 and CREB (Fig. 9A) and c-Jun (Fig. 9C), and this reduction was proportional to the extent of digestion of sPGN (Fig. 9E). As expected, the total amount of CREB (Fig.  9B) and c-Jun (Fig. 9D) proteins remained the same in all treated and untreated groups. DISCUSSION Our results demonstrate that: (i) PGN induces phosphorylation and functional activation of the transcription factors ATF-1 and CREB and these proteins bind DNA as a dimer; (ii) PGN induces phosphorylation of c-Jun, protein synthesis of JunB and c-Fos, and functional activation of the transcription factor AP-1; and (iii) PGN-induced activation of CREB/ATF and AP-1 is mediated through CD14. This is the first study to demonstrate activation of these transcription factors by PGN or by any other component of Gram-positive bacteria.
Our data also confirm that: (i) CREB and ATF-1 from LPSactivated nuclear extracts bind the CRE sequence (28); (ii) LPS induces phosphorylation of CREB (28); (iii) LPS induces increase in JunB protein, but not JunD (29); and (iv) JunB and c-Jun bind the TPA-RE sequence (29). In addition, our data demonstrate: (i) that LPS induces phosphorylation of ATF-1 and c-Jun, and (ii) that LPS-induced activation of CREB/ATF and AP-1 is CD14-dependent, which are new findings for LPS.
Our data also demonstrate that LPS induces transcriptional activation: (i) of a CRE reporter plasmid and that this activation is mediated by a complex of ATF-1 and CREB, and (ii) of an AP-1 reporter plasmid and that this activation is mediated by a complex containing c-Fos proteins. These data are in agreement with other investigators' results showing that the CRE sites in the promoters of TNF-␣ (30), MIP-1␤ (31), and Pselectin (32) genes are necessary for LPS-induced transcriptional activity of these genes. Also in agreement with our findings are the results showing that site-specific mutations in the CRE site in the IL-1␤ promoter result in a substantial loss in transcriptional induction following combined stimulation with LPS, phorbol myristate acetate, and dibutyryl cAMP (33), and also that the AP-1 site is required for LPS-induced transcrip-tional activation of tissue factor (34) and heme oxygenase (35) genes.
However, the actual protein complex that binds the CRE or AP-1 site and regulates transcription may be different for different genes, e.g. the CRE site in the IL-1␤ promoter binds CREB and ATF-1 (33), while the c-Jun protein binds the CRE site in the human TNF-␣ promoter, and the amount of c-Jun bound to this site increases when cells are stimulated with LPS (30). In addition, different proteins may have different effects on transcription, e.g. c-Jun is an effective transcriptional activator, while JunB is not, and thus JunB may have an inhibitory function (26). These data emphasize the importance of the CRE and AP-1 sites and the CREB/ATF and AP-1 families of transcription factors in LPS-induced transcriptional activation of pro-inflammatory genes. These transcription factors and their binding sites may also play a significant role in PGNinduced inflammatory response.
Although both PGN (10) and LPS (36) bind to CD14 and activate cells through CD14 (8), there are several differences in the function of CD14 as the PGN and LPS receptor. In particular, both the binding sites for PGN and LPS on CD14 and the sites needed for cell activation are partially similar but partially different (8,10), and only LPS-induced, but not PGNinduced cell activation and binding affinity for CD14 are enhanced by the LPS-binding protein (7,10). Moreover, PGN and LPS induce differential activation of MAP kinases, with LPS strongly inducing all three families of kinases (ERK, JNK, and p38), but with PGN only inducing ERK and JNK, but not p38 (37). Furthermore, soluble CD14⅐LPS complexes activate CD14-negative cells, whereas soluble CD14⅐PGN complexes do not (16).
Despite these differences, in this study we did not detect any differences between PGN and LPS in the activation of CREB/ ATF and AP-1 transcription factors. Therefore, our current and previous (8) results indicate that transcription factors NF-B, CREB/ATF-1, and AP-1 are either induced by CD14-dependent signal transduction pathways that are common for PGN and LPS, or that different pathways activated by PGN and LPS converge to activate these three families of transcription factors. Such a convergence of initially different signal transduction pathways to activate the same transcription factors has been demonstrated in the activation of cells through cytokine receptors, e.g. IL-1 and TNF-␣ (38).
The signal transduction pathway(s) through which PGN and LPS activate CREB/ATF and AP-1 are still not clear. Activation of AP-1 is consistent with the strong activation of JNK and ERK1 and ERK2 by both PGN and LPS (37), since JNK can activate c-Jun and both JNK and ERK can induce c-Fos through activation of ternary complex factor/Elk-1 (39). The possible mechanism of activation of ATF-1 and CREB are less clear, since in other systems the main mechanism of activation of ATF-1 and CREB is through protein kinase A, but we could not show any activation of protein kinase A by PGN or LPS (37). Other possible mechanisms of activation could involve calmodulin kinase or the MAP kinases ERK1 and ERK2 (17,18).
The functional significance of the activation of NF-B, CREB/ATF, and AP-1 for the induction of cytokine genes by PGN is still not clear. As seen for LPS, these transcription factors are required for the induction of specific genes coding for pro-inflammatory molecules, and our preliminary data indicate that NF-B, CREB, and ATF-1, but not AP-1, are required for PGN-induced transcriptional activation of TNF-␣. 2 These transcription factors are also likely to play a role in the 2 D. Gupta and Q. Wang, unpublished observation. induction of several other pro-inflammatory molecules, such as cytokines, chemokines, and adhesion molecules in PGN-activated cells.
In summary, we demonstrate that activation of macrophages by PGN leads to the functional activation of the transcription factors CREB/ATF and AP-1 and that this activation is CD14-dependent.