Mechanisms of murine RANTES chemokine gene induction by Newcastle disease virus.

We have previously defined the lipopolysaccharide (LPS)-responsive element (LRE) in the promoters of murine RANTES (regulated on activation normal T-cell expressed) (MuRantes) and murine IP-10/crg-2, chemokines which have potent chemotactic properties for inflammatory cells including monocytes and T lymphocytes. In the present work, we studied the transcriptional mechanism of MuRantes gene induction by virus and compared it with that of LPS in an effort to understand the host responses to virus and bacterial toxins at the molecular level. MuRantes mRNA expression is induced by Newcastle disease virus (NDV) and LPS in the RAW 264.7 macrophage cell line and peritoneal macrophages of LPS-responsive C3HeB/FeJ mice. In LPS-hyporesponsive C3H/HeJ mice, only NDV induces this chemokine gene, indicating that the pathways of transcriptional activation by NDV and LPS are not identical. Using a transient transfection assay, the minimal virus-responsive element (VRE) was localized between nt −175 and −116. The VRE contains previously defined LRE motif 1 (TCAYRCTT) and motif 3 ((T/A)GRTTTCA(G/C)TTT), which were shown to also be important for initiation of transcription by virus. NDV-stimulated nuclear extracts were tested for trans-activating factors able to bind the VRE. The chromosomal protein HMG-I(C) was shown to bind the 3′-A·T-rich domains of the VRE, and the presence of HMG-I(C) was demonstrated in the VRE-protein complex formed with nuclear extracts from NDV-stimulated, but not unstimulated cells. These findings demonstrate the role of HMG-I(C) in activation of MuRantes promoter by NDV.

In order to understand the early host responses to infection, much effort has been focused on the transcriptional activation of inducible genes by viruses. DNA viruses such as adenovirus and herpesviruses encode regulatory proteins that affect the activities of cellular transcription factors. These factors play a role in regulating cell growth and neoplastic transformation through activation of inducible genes (1)(2)(3)(4)(5). Retroviruses are known to affect activation of cellular genes in part through regulating enhancer activities (6). The Paramyxoviridae, a family of negative sense RNA viruses, are potent activators of inducible genes, in particular cytokines (7)(8)(9)(10). This family includes measles virus, canine distemper virus, Sendai virus, and Newcastle disease virus (NDV), 1 all of which are neurotropic (11,12).
Regarding transcriptional activation of cellular genes by Paramyxoviruses, a large body of work has centered on elucidating the mechanisms of IFN-␤ gene induction. In addition to NF-B, interferon regulatory factor 1, c-Jun, and ATF-2 (activating transcription factor 2), a DNA-binding protein HMG-I(Y) is required for IFN-␤ gene induction by Sendai virus (10,13). HMG-I(Y) is a member of the high mobility group (HMG) family of chromosomal, non-histone proteins (for review, see Refs. 14 and 15). HMG-I(Y) binds to A⅐T-rich regions of doublestranded DNA and increases the binding affinity of NF-B and ATF-2 for their respective sites. HMG-I(Y) also induces DNA bending, presumably allowing interactions among multiple transcriptional activators bound at distant sites on the promoter (10).
It has been noted that Paramyxoviruses and bacterial LPS stimulate overlapping, but distinct, sets of cytokine genes (8, 11, 16 -21). In order to determine the similarities and differences between antiviral and antibacterial host response mechanisms at a molecular level, we have examined the inducibility of the murine homolog of RANTES (MuRantes), a member of ␤ or C-C chemokine superfamily, in murine macrophages upon stimulation by NDV as compared with LPS. Involvement of RANTES in inflammation through its chemotactic activity for monocytes and eosinophils has been well documented (22)(23)(24). In addition, RANTES is chemotactic for memory T lymphocytes (22). RANTES gene is expressed in a number of cell types including fibroblasts (25), T-cell lines (26), endothelial cells (27), macrophages and macrophage-like cell lines (28), mesangial cells (29), and brain astrocytes and microglia (30).
(Ϫ175/Ϫ125), with an additional A⅐T-rich domain in the 3Ј border (28). This VRE contains 2 DNA motifs, TCAYRCTT and (T/A)GRTTTCA(G/C)TTT, previously shown to be important for the LRE activity of MuRantes and IP-10/crg-2 chemokine genes (28). Examination of nuclear factors binding to the VRE showed the presence of chromosomal protein HMG-I(C) in NDV-stimulated cells. Although purified HMG-I(C) binds to multiple A⅐T-rich sequences at the 3Ј border of MuRantes VRE, not all of them were required for transcription.

EXPERIMENTAL PROCEDURES
Reagents-All reagents were obtained from Sigma unless specifically noted otherwise.
Viral Stimulation-Cultures were stimulated with various doses of New Jersey La Sota strain of NDV (ATCC). Virus was harvested from 9-day-old, fertilized chick eggs as described previously (32). Mock NDV was from chick amniotic sac aspirates of sham-infected eggs. The viral titer was determined as described in detail elsewhere (32). The infectivity (m.o.i.) was determined in L-cell cytotoxic assay, and the doses used for RAW 264.7 cells were indicated as m.o.i. equivalent to that of L-cell assay. The viral genome was cross-linked under sterile conditions by irradiating virus stock in a 10-cm Petri dish with 300 ergs/cm 2 of shortwave ultraviolet (UV) light for 8 -10 min. For stimulation, cells were infected with NDV in serum-free RPMI 1640 with gentle rocking every 20 min for 2 h, then an equal volume of RPMI containing 10% FBS was added. Cycloheximide (CHX) and polymyxin B (PB), used at a final concentration of 10 g/ml, were added with the virus. The tyrosine kinase inhibitor herbimycin A (Life Technologies, Inc.) was added to cells 14 h prior to infection. The protein kinase C inhibitor H7 and its isomer HA1004 (Seikagaku America Corp., Rockville, MD) were added to cells 30 min prior to infection.
Northern Analysis-Total RNA, isolated using RNAzol B (Tel-Test, Inc., Friendswood, TX), was fractionated on a 1% agarose/formaldehyde gel (10 g of RNA/lane), followed by transfer onto a nitrocellulose membrane (Millipore) with a Posiblot system (Stratagene). The RNA was cross-linked to the membrane with a Stratalinker (Stratagene) and was subsequently probed with a 0.5-kilobase EcoRI, XhoI fragment of MuRantes cDNA and a 1.4-kilobase EcoRI fragment of aldolase A cDNA (a gift of D. Nathans, Johns Hopkins University). The relative amounts of mRNA were quantified on a Computing Densitometer (Molecular Dynamics, Sunnyvale, CA), the integrated volumes were calculated using Imagequant software (Molecular Dynamics), then the density ratio of MuRantes to aldolase A mRNA was obtained.
Plasmids and Mutants-The pG CAT m expression vectors carrying MuRantes cDNA spanning Ϫ2100/ϩ8, its 5Ј and 3Ј deletions, or mutated nucleotides have been prepared previously (28), as described in detail (28,33). Additional constructs consist of MuRantes promoter containing mutation 11 or 12. As described in Ref. 28, the mutated sequences were generated by polymerase chain reaction-directed mutagenesis in which the 5Ј primer from Ϫ185 was flanked by an SalI site and the 3Ј primer contained the sequence to ϩ8 and was flanked by an ApaI site. These fragments were then cloned into the pG CAT m expression vector.
Transfection and CAT Assay-Transfection using a modified DEAEdextran method (35) was identical as reported previously (28), except that the transfected cells were infected with NDV. In brief, RAW 264.7 cells transfected with various plasmid constructs were incubated in 10% RPMI for 27 h. After the medium was replaced with serum-free RPMI, cells were scraped, divided into three culture dishes, and incubated for additional 1 h. Each plate was incubated with medium alone, 1 m.o.i. of NDV, or 100 ng/ml LPS. The doses of NDV and LPS were chosen to give comparable CAT activity. Dishes were rocked every 20 min for 2 h. After adding 4.5 ml of 10% RPMI, cells were incubated for 10 h at 37°C before lysing for CAT assay. Efficiency of transfection was standardized to constitutive levels of ␤-galactosidase activity which is not affected by stimulation (28). Cell lysates were assessed for CAT activity according to described methods (36), with minor modifications as used previously (28).
DNA Probes-MuRantes probes (Ϫ185/Ϫ116 and Ϫ215/Ϫ116) were generated by polymerase chain reaction using 5Ј primers including a flanking SalI site and 3Ј primers with BamHI site. Oligonucleotides were appropriately digested for 1 h at 37°C and labeled by filling in with [ 32 P]dCTP and [ 32 P]dGTP as described previously (28). The NF-B probe is HIV-long terminal repeat, TCAA GGGACTTTCCGCT-GGGGACTTTCCCTCTCCTT.
Antibodies-The antibodies to JunD, CREB, c-Fos, p50, and p65 were from Santa Cruz Biotechnology (Santa Cruz, CA). The c-Jun and JunB antibodies were obtained from Oncogene Science (Uniondale, NY). Anti-HMG-I(C) IgG was a gift from J. Maher.

NDV Induced MuRantes mRNA Accumulation in RAW 264.7
Cells-The kinetics of NDV-induced MuRantes gene expression and the dose of NDV required were examined in RAW 264.7 cells by Northern analysis (Fig. 1). MuRantes transcripts were detected at 4 h, reached the maximum between 8 and 24 h (Fig. 1A), then markedly reduced at 48 h (data not shown). At 6 h of stimulation, MuRantes mRNA expression was increased with increasing m.o.i. of NDV (Fig. 1B). MuRantes mRNA was induced by live virus, but not by mock NDV or NDV-depleted supernatants (Fig. 2). Induction also occurred when cells were stimulated by NDV in the presence of cycloheximide (CHX) or polymyxin B (PB), indicating that MuRantes gene induction by NDV occurred in an immediate early manner, as has been shown for LPS stimulation (28), and was not due to contaminating LPS. UV-irradiated virus (UV-NDV) also induced Mu-Rantes. Induction of MuRantes gene by NDV does not require viral replication, since UV and CHX treatments render the NDV unable to replicate, a finding also seen in induction of complement C3 and IP-10/crg-2 genes by UV-NDV (37,38).
Induction of MuRantes mRNA Expression in Peritoneal Macrophages-To examine possible differences between NDV and LPS as stimuli, macrophages were obtained from LPSresponsive strain C3HeB/FeJ mice and C3H/HeJ mice which carry a mutation of the lps gene rendering them hyporesponsive to LPS (39). NDV-induced MuRantes mRNA accumulation in both strains of mice, C3HeB/FeJ and C3H/HeJ, while LPS failed to elicit any induction in C3H/HeJ, as expected (Fig. 3).
In the presence of CHX, NDV superinduced MuRantes in macrophages of both strains, as in primary rat astrocytes (30), but not in RAW 264.7 cell line (28). Since NDV induced Mu-Rantes mRNA in LPS-hyporesponsive macrophages, the pathways of transcriptional activation by LPS and NDV cannot be identical. Studies to evaluate this issue by examining signal pathways using inhibitors are shown in Fig. 4. Tyrosine kinase inhibitor herbimycin A had a potent inhibitory effect on the NDV-mediated mRNA expression, whereas the LPS-induced expression was much less sensitive (Fig. 4A). The potent pro- tein kinase C inhibitor H7 (40) blocked the induction by both stimuli at the same concentration (Fig. 4B). HA1004, an H7 isomer which strongly inhibits cyclic nucleotide-dependent kinases (40), markedly inhibited the LPS-induced induction, but was minimally effective for NDV (Fig. 4C). These data indicate involvement of distinct multiple effector pathways in inducing MuRantes gene. The inhibitors were not cytotoxic, as determined by the level of lactate dehydrogenase in supernatants, which is released upon cell death (data not shown).
Stimulation of MuRantes Promoter Activity by NDV-After establishing MuRantes gene induction by NDV, the nucleotide sequences of the 5Ј-regulatory region of MuRantes gene required for the NDV-VRE were analyzed by CAT assay in RAW 264.7 cells. Cells transfected with pG CAT m carrying cDNA spanning nt Ϫ2100 to ϩ8 showed significant CAT activity upon NDV stimulation compared to unstimulated or mock NDVstimulated cells (Fig. 5). The presence of FBS enhanced the efficiency of induction by LPS and NDV. The results of deletion analysis to determine the 5Ј and 3Ј boundaries required for viral inducibility, are presented in Fig. 6, A and B, as a histogram. Deletion from nt Ϫ2100 to Ϫ215 relative to the transcriptional start site had virtually no effect on virus-induced CAT activity (Fig. 6A). Full activity was maintained until a deletion of 3 nt from Ϫ175 to Ϫ172, which reduced the relative CAT activity to 37.3% Ϯ 8.2. A further deletion from nt Ϫ167 to Ϫ155 completely abolished the activity. The 3Ј boundary was delineated using 3Ј deletion constructs containing heterologous hamster sarcoma virus-thymidine kinase minimal promoter (34). When the promoter activity of nt Ϫ185 to Ϫ116 was considered as 100%, deletion to nt Ϫ125 reduced the activity to 52.3% Ϯ 17.5 and deletion to nt Ϫ138 completely abolished the activity (Fig. 6B). The activity of a longer nt Ϫ185 to Ϫ60 segment was consistently lower (52.1% Ϯ 10.3), suggesting a repressor site within nt Ϫ116 and Ϫ60 (see Fig. 7B). Therefore, MuRantes VRE lies between nt Ϫ175 and Ϫ116 and contains a 6/7 AP-1 site, a 8/10 NF-B site, and a 9/12 ISRE site (nt Ϫ114 to Ϫ156) (Fig. 7). MuRantes induction is unlikely to occur through the ISRE core since this promoter is not inducible by IFNs (28). Cells were stimulated by LPS and NDV side by side, whereby the previously described 5Ј and 3Ј boundaries of LRE (Ϫ175/Ϫ125) (28) have been verified in this study (data not shown). The pattern of relative CAT activity induced in these constructs by LPS and NDV show marked similarities. Therefore, the 5Ј boundary of the VRE (Ϫ175) was closely aligned to that of the LRE and the 3Ј boundary of VRE (Ϫ116) was found to extend further than the LRE.
Mutational Analysis of the MuRantes VRE-Both the VRE and LRE contain 2 conserved sequences shared with muIP-10/ crg-2 LRE (28): motif 1, TCAYRCTT, and motif 3, (T/A)GRTT-TCA(G/C)TTT (Fig. 7). Since the VRE also contains an AP-1 site (6/7) and an NF-B-like site (8/10), possible utilization of these sites for transcription was examined by mutational analysis. Results from 4 separate experiments revealed two regions critical for NDV inducibility (Fig. 8). A half AP-1 site (TCA) in motif 1 is flanked on the 3Ј side by 5 nt, and on the 5Ј side by the other half of the AP-1 site (Fig. 8A). The first required domain was identified by mutations 1 through 4 (M1-M4). Mutation 1 (M1) in the 5Ј-half AP-1 site resulted in a moderate

FIG. 5. Induction of CAT activity by NDV in RAW 264.7 cells transiently transfected with MuRantes promoter-CAT constructs.
A, the 5Ј-flanking region of the MuRantes gene spanning Ϫ2100 to ϩ8 which includes the MuRantes minimal promoter (28) was ligated upstream of the CAT coding region in the pG CAT m plasmid. B, this construct was transiently transfected into RAW 264.7 cells as described under "Experimental Procedures." Cells were then stimulated with 30 m.o.i. NDV, mock NDV, or 300 ng/ml LPS in serum-free medium for 2 h. One set of cultures had RPMI/10% FBS added after 2 h, for the remaining 10 h (ϩFBS). The CAT activity was determined as described under "Experimental Procedures." C, relative CAT activity of each assay was determined by correcting the value for cell number by standardizing for ␤-galactosidase activity and subtraction of the values of the unstimulated cells. The CAT activity in cells stimulated with NDV ϩ FBS was considered as 100%.

FIG. 6. CAT activity of 5-and 3-deletion constructs of MuRantes promoter stimulated with NDV. A, RAW 264.7 cells transfected
with MuRantes promoter-CAT constructs carrying 5Ј-nested deletions were stimulated with 1 m.o.i. NDV for 12 h. CAT assay was performed with cell lysates as described. Relative CAT activities Ϯ S.E. in NDVstimulated cells from 3 separate experiments are shown. The CAT activity of Ϫ215/ϩ8 construct was considered as 100%. B, the 3Ј-nested deletion constructs of MuRantes promoter were prepared by using a 3Ј polymerase chain reaction primer with the appropriate sequence flanked by a site which was generated by polymerase chain reaction and then inserted into the pBL CAT 2 vector which contains a heterologous thymidine kinase minimal promoter upstream of the CAT coding region. RAW 264.7 cells transfected with MuRantes promoter-CAT constructs were stimulated with 1 m.o.i. NDV, as in A. Relative CAT activities Ϯ S.E. in response to NDV from 4 separate experiments are shown. The CAT activity of Ϫ185/Ϫ116 construct was taken as 100%.
loss of activity (37.3% Ϯ 7.3 in relation to nt Ϫ185 to ϩ8 wild type construct). M3 which involves the 3Ј-half of the AP-1 site and the 5Ј-half of motif 1 caused dramatic loss of activity. M2, which altered both 5Ј-and 3Ј-halves of the AP-1 site as well as motif 1, had an even more profound loss of activity. M4, which changed 3 bases flanking the half AP-1 site in motif 1, reduced the activity to 48.4% Ϯ 4.1. M5 and M6 carrying mutated bases of motif 2 between motif 1 and motif 3, did not influence the CAT activity. A second critical region was identified through the use of M8 through M12 constructs. M8 decreased the VRE activity close to 60%. M9 and M10, which affect A⅐T-rich domains within motif 3, totally abolished the VRE activity. It was significant that M10 which lies within motif 3, but outside of the NF-B-like site, as well as M11 abolished the promoter activity (0.3% Ϯ 0.3). M12 showed significant VRE activity upon NDV stimulation (57%). In addition, M7, which alters the highly conserved nucleotides of the potential NF-B site, had no effect on virus inducibility. These functional data indicate that in addition to motif-1, three A⅐T-rich domains, two of them located within motif 3, are required for the VRE activity. The data also imply that the NF-B like site within the VRE is unlikely to participate in NDV induction of MuRantes, as was previously found for LPS (28), even though both LPS and NDV can induce NF-B DNA binding activity (data not shown), as reported by others (6,41,42). The CAT activity induced by LPS determined along with the virus using M1-M10 constructs, which was closely similar to the previously published values (28), also showed a similarity to the activity induced by NDV, except that mutation M9 produced partial reduction in LPStreated cells, but a near-complete loss in NDV-infected cells. NDV and LPS also produced similar results for M11 and M12.
Examination of the DNA-binding Factors Involved in Mu-Rantes Induction-The properties of the DNA-binding factors were examined by EMSA using poly(dG:dC)(dG:dC) and poly-(dI:dC)(dI:dC) as nonspecific competitors (Fig. 9A). With poly-(dI:dC)(dI:dC), formation of VRE-protein complexes in unstimulated or stimulated nuclear extracts was markedly reduced when compared to those formed with poly(dG:dC)(dG: dC). Using poly(dG:dC)(dG:dC), a single slower mobility band in unstimulated extracts and two closely spaced faster mobility bands in NDV-stimulated extracts were observed, all of which were competed away with 20-fold excess of cold wt VRE (nt Ϫ185 to Ϫ116) (Fig. 9, A and B). As shown in Fig. 9, A and B, unstimulated nuclear extracts formed a single band of slower mobility that can be competed away with wt VRE. The nature of this band is currently unknown. As shown in Fig. 9B, VRE complexes formed with NDV-stimulated extracts that can be competed away with wt VRE (lanes 3 and 4), were not competed with a wt-or mutation 6-containing 30-mer (Ϫ167/Ϫ137) (lanes 5 and 6). However, full-length VRE (Ϫ185/Ϫ116) with M7, M10, and M12 were effective as competitive inhibitors (lanes 7,  8, and 10). M11 VRE partially inhibited the complex formation (lane 9). A significant finding is that M7 (mutation of the NF-B-like site) inhibited the complex formation with similar efficiency as wt VRE. The inability of truncated wt 30-mer sequences to inhibit the VRE-complex formation was correlated with the failure of a wt 30-mer (Ϫ167/Ϫ137), used as a probe, to form DNA-protein complexes (Fig. 9C). These results collectively indicate that both motif 1 and motif 3 plus A⅐T-rich sequences may be required to form VRE-protein complexes in NDV-stimulated nuclear extracts. The finding that complex formation was inhibited by M10, M11, and M12 VRE, each containing mutations affecting a single A⅐T-rich domain, suggested a cooperative binding to the A⅐T-rich domains. We have not tested VRE in which all of A⅐T-rich domains are mutated.
Presence of HMG-I(C) in NDV-stimulated Nuclear Extract and Binding of HMG-I(C) to MuRantes VRE-Since formation of the VRE-protein complex was inhibited by poly(dI:dC)(dI:dC) which appears similar to A⅐T-rich DNA with respect to hydrogen bond formation in the minor groove, and since MuRantes VRE also contains A⅐T-rich regions in the 3Ј border, a possible role for the HMG-I family of proteins was explored. The presence of HMG-I(C) in VRE-protein complexes was detected by anti-HMG-I(C) antibody, which supershifted the complexes formed with NDV-stimulated nuclear extracts (Fig. 10A, lanes  9 -12), but not with unstimulated extracts (lanes 5-8). Purified HMG-I(C), 10 nM, formed two bands with wt VRE probe, and the complexes were also supershifted with antibody ( lanes  1-4). In EMSA, purified HMG-I(C) formed a single mobility complex with a wt 30-mer (Ϫ167/Ϫ137) (Fig. 10B), and the HMG-I(C) binding to this 30-mer probe was markedly diminished when any one of the 2 A⅐T-rich domains was mutated (M10, M11). The binding of HMG-I(C) to a 30-mer (Ϫ161/Ϫ131) which contains M12 was also significantly decreased. Although each of these A⅐T rich domains is the target for HMG-I(C) binding, the role of M12 site in transcriptional activity of the VRE may not be as important as the A⅐T-rich sites effected by M9, M10, and M11, based on the results of the CAT assay. The supershift with anti-HMG-I(C) carried out with LPS-stimulated nuclear extracts was far below the level achieved with virus-stimulated nuclear extracts (data not shown).

Studies of AP-1-binding Factors and NF-B in VRE-Protein
Complexes-Proteins capable of binding to NF-B and AP-1 sites were examined using antibodies (Fig. 11). Only anti-c-Jun supershifted the VRE-protein complex (lane 11). Although the effect was detected only in overexposed autoradiograms, this finding was consistently reproduced in 4 separate experiments. Antibodies to c-Fos, JunB, JunD, CREB, p50, and p65 were unable to supershift the VRE-protein complexes. NF-B binding to the HIV-long terminal repeat NF-B probe was induced in NDV-or LPS-stimulated nuclear extracts, and the binding was not competed away with wt VRE or M7 VRE (data not shown).

DISCUSSION
In this study, we have shown that the murine RANTES (MuRantes) chemokine gene is induced by NDV in an immediate early manner in RAW 264.7 cells. MuRantes mRNA induction does not require the synthesis of viral proteins or doublestranded viral RNA since CHX ϩ NDV and UV-NDV showed induction efficiency similar to live virus. These results were reproduced in primary peritoneal macrophages elicited with starch from C3HeB/FeJ mice. However, MuRantes gene was induced by NDV, but not by LPS in macrophages from C3H/ HeJ mice, an LPS-hyporesponsive strain. In addition, MuRantes mRNA expression induced by NDV and LPS showed stimulus-specific susceptibility to signal pathway inhibitors. Herbimycin, a tyrosine kinase inhibitor, preferentially inhibited NDV-stimulated mRNA accumulation, whereas HA1004, a cyclic nucleotide-dependent kinase inhibitor, was more effective in inhibiting the LPS effect. These results indicate that induction of MuRantes gene by NDV and LPS is through distinctly different pathways. In this context, it is interesting that the VRE and LRE of the MuRantes promoter shared extensive sequence homology: the minimal VRE lies between nt Ϫ175 and Ϫ116 and the LRE between nt Ϫ175 and Ϫ125. The VRE also shared with LRE certain DNA motifs, designated motif 1 and motif 3, which were required for MuRantes transcription, as shown by mutational analysis. Motif 3 contains two A⅐T-rich domains, each of which when mutated as M9 or M10 abolished the VRE activity. In contrast, mutation of the first A⅐T-rich domain (M9) reduced the LRE activity only partially (28). The VRE carries additional A⅐T-rich sequence repeats in the 3Ј border, to which HMG-I(C) was shown to bind. It is likely that virus and LPS may use overlapping, but not identical, sets of transcription factors that interact with multiple sites within the VRE and LRE. However, it remains to be determined whether VRE and LRE associate with distinct proteins regulated through different signaling pathways.
We have noted that poly(dI:dC)(dI:dC) decreased VRE-protein complex formation, while the LRE-protein complexes were less affected (data not shown). This inhibitory effect of poly(dI: dC)(dI:dC) may be due to its structural similarity to the minor groove of double-stranded, A⅐T-rich DNA sequences, thereby allowing it to compete with minor groove binding proteins like HMG-I (43)(44)(45). Earlier studies have demonstrated that HMG-I(Y) participates in transcription of IFN-␤ gene induced by Sendai virus by enhancing the affinity of NF-B and ATF-2 to their respective binding sites (10,13). We have demonstrated that HMG-I(C), a member of the HMG-I family of proteins (46,47), was present in the VRE-protein complex formed with NDV-stimulated nuclear extracts, but not the VRE complex of unstimulated extracts. These data indicated that NDV-induced post-transcriptional activation of HMG-I(C) allows the assembly of the VRE-protein complex needed for transcription rather than induction of HMG-I(C) synthesis, since MuRantes gene induction by NDV was not affected by CHX. The presence of multiple HMG-I(C) binding sites and the failure of VRE containing M9, M10, or M11 to induce transcription suggest possible cooperative activity among HMG-I(C) proteins and between HMG-I(C) and other DNA-binding factors. LPS-induced CAT activity was only partially affected by M9 and the potency of anti-HMG-I(C) to supershift the LRE-protein complexes was reduced significantly compared with that of virus. Thus, the possible effects of HMG-I(C) on MuRantes transcription in response to virus and LPS may be quantitative.
It was unexpected that the 8/10 NF-B binding site failed to play a major role in virus-mediated activation of this protein. It is possible that the binding site sequences are not favorable for NF-B binding. Activated NF-B may still play a role through binding to transcription factors such as c-Jun (48). As shown here, it is likely that c-Jun in the VRE-protein complexes may interact with motif 1 and the 5Ј-flanking half AP-1 site. The interaction among DNA-binding proteins may be regulated by HMG-I(C) in the VRE and the binding specificities, and possible partners that may dimerize with c-Jun in the VRE and LRE may not be identical. Various trans-activating proteins, FIG. 10. The VRE-protein complexes formed with NDV-stimulated nuclear extracts contain HMG-I(C). A, labeled wt VRE Ϫ185/ Ϫ116 and unstimulated (lanes 5-8) or NDV-stimulated nuclear extracts (lanes 9 -12) were incubated with anti-HMG-I(C) IgG at 1/2.5 and 1/5 dilutions as described under "Experimental Procedures." Supershifted complexes were detected by EMSA. The binding of purified HMG-I(C), 10 nM, to wt VRE probe (Ϫ185/Ϫ116) was examined by EMSA using anti-HMG-I(C) (lanes 1-4). B, purified HMG-I(C) binding to the A⅐T-rich sequences of the VRE was evaluated using wt 30-mer (Ϫ167/Ϫ137), a 30-mer (Ϫ167/Ϫ137) with mutated A⅐T-rich sequences, M10, or M11, and a 30-mer (Ϫ161/Ϫ131) with M12.
FIG. 11. Identification of proteins in the VRE-protein complexes using antibodies to AP-1 proteins, NF-B, and HMG-I(C). Unstimulated (lanes 1-9) and NDV-stimulated (lanes 10 -18) nuclear extracts (5 g) were allowed to bind to labeled VRE Ϫ185/Ϫ116. Samples were then incubated with antibodies for 1 h at 4°C, and evidence of supershifting bands was assessed by EMSA. This blot was overexposed to detect any bands supershifted by antibodies. Clearly supershifted complexes are observed with NDV-stimulated samples treated with anti-c-Jun (lane 11) and anti-HMG-I(C) (lane 18). Antibodies including anti-JunB, anti-JunD, anti-CREB, anti-c-Fos, anti-p50, and anti-p65, supplied as IgG were used at 2 g in the binding assay. through homo-or heterodimer formation, are known to generate response diversity by affecting binding affinity and/or selecting different flanking sequences for binding (for review, see Ref. 49). The A⅐T-rich sequences in the VRE that can bind multiple HMG-I(C) may also play a critical regulatory role in generating additional diversity by enhancing the binding of specific transcription factors and by allowing protein-protein interaction through DNA bending.
It would be surprising if the molecular mechanisms used to induce common cytokines by bacterial and viral pathogens evolved as totally separate pathways. Therefore, the extensively overlapping sequence motifs between the LRE and VRE in MuRantes induction are not unexpected. These findings also indicate that host defense mechanisms to viral and bacterial infection may be achieved by utilizing similar or closely overlapping molecular events.