Oncogene (1999) 18, 5455 ± 5463 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00 http://www.stockton-press.co.uk/onc Identi®cation of the secretory leukocyte protease inhibitor (SLPI) as a target of IRF-1 regulation

Hannah Nguyen1,2, Lindsay Teskey1,2, Rongtuan Lin1,3 and John Hiscott*,1,2,3

1Terry Fox Molecular Oncology Group, Lady Davis Institute for Medical Research, Sir Mortimer B Davis Jewish General Hospital, McGill University, Montreal, Quebec H3T 1E2, Canada; 2Department of Microbiology and Immunology, McGill University, Montreal, Quebec H3T 1E2, Canada; 3Department of Medicine, McGill University, Montreal, Quebec H3T 1E2, Canada

Interferon Regulatory Factor (IRF)-1 is a multifunc- distinct roles in biological processes such as pathogen tional transcription factor, involved in cell growth response, cytokine signaling, cell growth regulation and regulation, pathogen response and immune activation. hematopoietic di€erentiation (reviewed in Nguyen et To identify novel targets that may contribute to al., 1997a). IRF-1 mediated activities, RNA ®ngerprinting was IRF-1 and IRF-2 are by far the best characterized performed using NIH3T3 cells that inducibly express a members of the IRF family of transcription factors hybrid form of IRF-1 under tetracycline regulated (reviewed in Nguyen et al., 1997a; Taniguchi et al., control. Secretory leukocyte protease inhibitor 1997). These IRFs bind to similar DNA binding motifs, G T G T (SLPI) ± a regulator of in¯ammation and an inhibitor termed IRF-E (5'-G(A)AAA /C /CGAAA /C /C-3') pre- of the LPS response ± was identi®ed as a gene repressed sent in the promoters of many IFN and/or - after doxycycline induced IRF-1 expression. Preliminary inducible . Structurally, the IRF-1 and IRF-2 analysis of the human SLPI promoter identi®ed an are similar, sharing 76% identity in the ®rst 154 ISRE-like site located within the 7221 to 7200 region N-terminal aa and 8% identity at the C-terminal end to which IRF-1 binds and a second, putative IRF-1 (Harada et al., 1989), but possess very di€erent activities binding site upstream of the TATA box. Furthermore, (Fujita et al., 1988; Miyamoto et al., 1988; Harada et co-transfection studies demonstrated that SLPI expres- al., 1989; Reis et al., 1992). IRF-1 serves as a sion was inhibited by IRF-1 co-expression. The transcriptional activator, whereas IRF-2 acts as an identi®cation of SLPI as a target of IRF-1 regulation antagonistic transcriptional repressor. The functional reveals a unique involvement of IRF-1 in repression of di€erences between IRF-1 and IRF-2 extend further to gene transcription and assigns a novel role for IRF-1 in include an important role in cellular growth regulation; in¯ammation. IRF-1 and IRF-2 tumor suppressor and oncogenic activity were demonstrated by IRF-2 induction of Keywords: interferon regulatory factor (IRF)-1; secre- cellular transformation in NIH3T3 cells and tumor tory leukocyte protease inhibitor (SLPI); transcription formation in nude mice, and reversal of the IRF-2- mediated tumorigenicity by IRF-1 (Harada et al., 1993). Studies performed with knockout mice illustrate that IRF-1 plays key roles in the regulation of various Introduction immune processes. Mice de®cient in the IRF-1 gene, while having normal numbers of immature CD47CD8+ The Interferon Regulatory Factors (IRFs) are a family and CD4+CD87 thymocytes, are 90% de®cient in of transcription factors consisting of nine members mature CD8+ T-cells in the thymus (Matsuyama et al., (IRF-1, IRF-2, IRF-3, ISGF3g/p48, ICSBP, Pip/ 1993), suggesting impairment of CD8+ cell maturation ICSAT/IRF-4, IRF-5, IRF-6 and IRF-7) and virally in the IRF-1 knockout mice. This phenotype may be encoded forms (vIRFs). All members of this family explained by the fact that IRF-1 upregulates expression share signi®cant in the N-terminal 115 of TAP1 and LMP2, genes which are important for amino acids which comprise the DNA binding positive selection of CD8+ cells (Penninger and Mak, domain; for the IRF-3, IRF-4, IRF-5, ISGF3g and 1994). Furthermore, immune cells from IRF-17/7 mice ICSBP proteins, the homology extends into the C- exhibit defective Th1 responses ± impaired macrophage terminal IRF association domain with which these production of IL-12, de®cient CD4+ T-cell response to IRFs interact with other proteins or family members. IL-12, ablated natural killer (NK) cell development, IRFs can be classi®ed as transcriptional activators and exclusive Th2 di€erentiation of macrophages and (IRF-1, IRF-3, IRF-7 and ISGF3g), repressors (IRF-2, CD4+ T-cells in vitro (Taki et al., 1997), implicating ICSBP, and vIRF) or both (IRF-4). Studies character- IRF-1 multiple stages of Th1 di€erentiation. IRF-1 is izing IRF-expressing cell lines and IRF knockout mice also essential for the induction of NK cell-mediated reveal that each member of the IRF family exerts cytotoxicity in vivo, since cytolytic activity of IRF-17/7 NK cells is defective, even following induction by virus infection, double-stranded RNA (dsRNA), IFN-b, IL- 2 and IL-12 (Duncan et al., 1996). Finally, IRF-1 *Correspondence: J Hiscott, Lady Davis Institute for Medical exhibits an important function in the establishment of Research, 3755 Cote Ste. Catherine, Room 526, Montreal, Quebec H3T 1E2, Canada the antiviral state. Inhibition of encephalomyocarditis Received 30 November 1998; revised 14 March 1999; accepted 20 virus replication by type I and II IFN is dramatically April 1999 impaired in IRF-17/7 mice. Furthermore, induction of IRF-1 regulation of SLPI expression HNguyenet al 5456 the inducible iNOS, guanylate-binding (GBP) from 72 h Dox-induced IRF-1/RelA cells (Figure 1, and 2'5'oligoadenylate synthetase (OAS) genes by IFN- lanes 3 and 4). Primer sets 2 and 3 generated many g is severely impaired in IRF-17/7 MEFs (Reis et al., distinct bands by RNA ®ngerprinting but none were 1994; Kimura et al., 1994). di€erentially expressed (Figure 1, lanes 5 ± 12). The In a recent study, IRF-1 expressing cell lines were 400 bp band was isolated, subcloned and sequenced, generated with the tetracycline-inducible expression and the resulting sequence was 100% homologous to a system. Using these cell models, it was demonstrated fragment of the murine secretory leukocyte protease that inducible IRF-1, IRF-1/RelA and IRF-2/RelA inhibitor (SLPI) gene. expression correlated with antiproliferative activity and The 400 bp fragment was used as a probe for upregulation of several growth regulatory genes Northern blot to analyse SLPI expression following (Nguyen et al., 1997b). Since they exhibited similar induction of IRF-1 and IRF/RelA at varying times e€ects, the use of the IRF/RelA fusion genes serve as following Dox induction. Figure 2 illustrates the useful tools to examine IRF-1 mediated activities. In kinetics of expression of SLPI mRNA following this study, the IRF-1 and IRF-1/RelA cell lines and the induction of IRF-1/RelA. Uninduced cells exhibited method of di€erential display/RNA ®ngerprinting were detectable endogenous levels of SLPI. Interestingly, exploited to search for novel gene targets of IRF-1. As following Dox induction of IRF-1/RelA, two distinct a result, secretory leukocyte protease inhibitor (SLPI) patterns of SLPI mRNA expression were observed. was identi®ed as a gene downregulated by IRF-1. Two hours after Dox treatment (Figure 2, lane 2), Belonging to the family of a-1 antitrypsin antiprotei- SLPI levels increased, attaining ®vefold higher levels by nases, SLPI plays a primary role in the regulation of 4 h after Dox induction, when IRF-1/RelA mRNA neutrophil-mediated in¯ammation through proteolysis levels were detected (Figure 2, lane 3). However, 12 ± and subsequent inhibition of the serine leukocyte 48 h following Dox induction, SLPI mRNA levels proteases (Thompson and Ohlsson, 1986). SLPI is decreased to undetectable levels (Figure 2, lanes 4 ± 10). also a regulator of the LPS response; although Similar kinetics of SLPI expression were observed upon activated by LPS in wild-type macrophages and Dox induction of rtTA-IRF-1 cells (data not shown). neutrophils, SLPI antagonizes the LPS response when Several experiments con®rm that the increase in SLPI overexpressed in macrophages (Jin et al., 1997). expression observed early in induction is not due to the Preliminary characterization of the role of IRF-1 in e€ects of doxycycline itself (Nguyen et al., 1997b). SLPI expression discloses a major IRF-1 binding site at From these results, SLPI is a gene which is dually the 7221 to 7200 SLPI promoter region and reveals regulated by IRF-1 in a dose-dependent manner: SLPI complex regulation of SLPI expression involving both expression appears to be activated early after Dox activation and repression by IRF-1. The identi®cation induction of IRF-1/RelA when transgene concentra- of SLPI as a target of IRF-1 regulation attributes a tions are low, and later suppressed when peak, unique repressor function to IRF-1 and assigns a novel relatively high levels of IRF-1/RelA are attained. role for IRF-1 in the in¯ammatory process. dsRNA inducibility of SLPI mRNA in the IRF-1 and IRF-1/RelA inducible cell lines Results To investigate the direct e€ect of dsRNA on SLPI RNA ®ngerprinting of cell lines inducibly expressing expression in the presence or absence of IRF-1 or IRF-1/RelA IRF-1/RelA, control rtTA, rtTA-IRF-1 and rtTA- IRF-1/RelA cells were Dox-induced and/or treated To identify genes regulated by IRF-1, RNA with dsRNA [poly (I):poly (C)] in the presence of ®ngerprinting was performed on the recently char- cycloheximide (CHX) and analysed for SLPI expres- acterized tetracycline-inducible control rtTA, rtTA- sion by Northern blot analysis (Figure 3). Uninduced IRF-1 and rtTA-IRF-1/RelA expressing NIH3T3 cells or Dox-induced control cells exhibited basal levels of (Nguyen et al., 1997b). IRF-1/RelA is a chimeric SLPI (Figure 3, lanes 1 and 3) which were increased protein consisting of the N-terminal DNA-binding approximately 2.5-fold upon treatment with dsRNA/ domain of IRF-1 fused to the C-terminal transactiva- CHX (Figure 3, lanes 2 and 4). Similar increases in tion domain of the RelA subunit of the NF-kB family SLPI mRNA expression were observed in control cells of transcription factors (Lin et al., 1994). Structural in response to dsRNA alone, con®rming that the similarity to IRF-1 led to the prediction and observed increases were not due to the e€ects of CHX subsequent con®rmation that IRF-1/RelA was able (data not shown). In IRF-1 and IRF-1/RelA cells, to mimic IRF-1 as a transactivator and tumor basal SLPI levels in the absence (Figure 3, lanes 5 and suppressor (Nguyen et al., 1997b), making this 9) or presence of Dox (Figure 3, lanes 7 and 11) were transgene a complementary tool to investigate IRF- consistently lower (or absent) than in control cells, 1-mediated activities. rtTA-IRF-1/RelA cells were used presumably due to the e€ects of low basal IRF-1 or preferentially for RNA ®ngerprinting due to their low IRF-1/RelA expression. Following treatment with uninduced transgene levels and high inducibility of dsRNA, SLPI levels were elevated in uninduced expression (Nguyen et al., 1997b). IRF-1 (Figure 3, lane 6) and IRF-1/RelA cells An RNA ®ngerprint resulting from cDNA amplifi- (Figure 3, lane 10) to levels comparable to that of cation using three di€erent primer sets is presented in dsRNA treated control cells. However, in the presence Figure 1. A prominent 400 (bp) band of Dox, the increase in SLPI mRNA following ampli®ed from primer set 1 in undiluted and 100-fold dsRNA treatment of IRF-1 (Figure 3, lane 8) and diluted cDNA from 72 h Dox-induced control rtTA particularly IRF-1/RelA cells (Figure 3, lane 12) was cells (Figure 1, lanes 1 and 2) was absent in cDNA less dramatic, yielding two and threefold lower levels IRF-1 regulation of SLPI expression HNguyenet al 5457

Figure 1 RNA ®ngerprint of rtTA and rtTA-IRF-1/RelA cells. Total RNA (2 mg) from 72 h Dox-induced rtTA (lanes 1, 2, 5, 6, 9, and 10) or rtTA-IRF-1/RelA (lanes 3, 4, 7, 8, 11 and 12) cells was used for RNA ®ngerprinting reactions using three di€erent primer sets. To ensure reproducibility, PCR reactions were also performed on 1 : 100 dilutions of the cDNA product from each cell sample. The 400 bp band which corresponds to SLPI is identi®ed

the consistently lower Dox-induced levels of IRF-1 compared to those of IRF-1/RelA (Nguyen et al., 1997b). Thus, dsRNA is a strong activator of SLPI expression, and dsRNA-mediated induction of SLPI expression is suppressed by induced IRF-1 or IRF-1/ RelA expression.

Detection of IRF-1 binding activity at two regions of the SLPI promoter

Figure 2 Expression of SLPI mRNA in rtTA-IRF-1/RelA cells. To characterize the role of IRF-1 in the regulation of Twenty mg of total RNA prepared from rtTA-IRF-1/RelA cells SLPI expression, the available human SLPI promoter 0 ± 96 h following Dox induction was used for Northern blot was analysed for transcription factor consensus sites. analysis with the 400 bp SLPI DNA fragment isolated from the The sequence of the SLPI promoter is shown in Figure RNA ®ngerprint or a 5' 300 bp IRF-1 DNA fragment as a probe. 4. Initially, one ISRE-like and two NF-kB-like sites As a control of RNA loading, 28S RNA obtained from migration of total RNA on the agarose gel used for Northern blot analysis were detected in the 789 to 745 region, and two is presented oligonucleotides were tested for binding by EMSA. BS1 (nt 789 to 761) encompassed the upstream NF- kB-like site and the ISRE-like site, while BS2 (nt 779 to 745) contained the same ISRE-like site and the of SLPI, respectively, compared to Dox-induced downstream NF-kB-like site. EMSA analysis showed control cells. The relatively moderate repressive e€ect that polyhistidine-tagged recombinant IRF-1 or IRF-2 on dsRNA-mediated SLPI expression observed in protein did not bind to either BS1 nor BS2 (data not rtTA-IRF-1 cells (Figure 3, lane 8) correlates with shown). IRF-1 regulation of SLPI expression HNguyenet al 5458 To examine the SLPI promoter sequence for IRF-1 entire BS2 sequence. Strikingly, IRF-1 bound to both binding sites, EMSAs were performed with recombi- fragments of the promoter (Figure 5a, lanes 3 and 5). nant IRF-1 protein using the entire promoter sequence The speci®city of the IRF-1 DNA binding complex was as a probe (Figure 5a). Interestingly, IRF-1 bound to con®rmed by supershift analysis using an IRF-1 the promoter sequence (Figure 5a, lane 1). To isolate speci®c antibody (Figure 5a, lanes 2, 4 and 6). No the region to which IRF-1 may bind, the promoter was IRF-1 DNA binding activity was detected on BS12, a digested with the EcoRI restriction into two probe consisting of BS1 and BS2 together (789 to fragments: SLPI-H/E (the 5' 207 nt region; nt 7276 to 745 nt), and the weak band observed was not shifted 769) and SLPI-E/A (the remaining 3' 69 nt region; nt in the presence of IRF-1 antibody (Figure 5a, lanes 7 769 to +1), which includes 9 nt of the BS1 and the and 8). IRF-2 did not bind to the SLPI promoter despite the fact that it recognizes a similar DNA binding motif as IRF-1 (Tanaka et al., 1993). These results demonstrate the presence of at least two IRF-1 binding sites in the SLPI promoter: one site is located in the upstream 7200 region, and the other site is located within the proximal 745 region, adjacent to the TATA box. To further delineate the IRF-1 site in the SLPI 57 promoter, in vitro footprinting was performed using recombinant IRF-1 protein and a 5'-radiolabeled SLPI promoter sequence as a probe. A representative DNase I footprint is presented in Figure 5b. SLPI promoter DNA in the absence of IRF-1 protein displayed a characteristic DNase I pattern compared to the control DNA ladder (Figure 5b, lanes 1 and 2). In the presence of increasing amounts of IRF-1 protein, a region between nt 7221 and 7200 was protected from DNase I treatment (Figure 5b, lanes 5 and 6). The protected sequence is rich in cytosine (C) and thymidine (T) bases, consistent with the 5'-GAAA-3' stretches found in the IRF-1 consensus sequence. To con®rm binding of IRF-1 to the region between nt 7221 and 7200, an oligonucleotide representing the SLPI promoter sequence between nt 7224 and Figure 3 SLPI mRNA in dsRNA treated IRF-1 and IRF-1/ 7201 (F1A) was synthesized and tested for binding by RelA cells. Zero or 40 h following Dox induction, rtTA, IRF-1 EMSA. As shown in Figure 5c, recombinant IRF-1 and IRF-1/RelA cells were left untreated or were treated for 5 h protein bound to F1A, as addition of increasing with dsRNA (25 mg/ml)/cycloheximide (50 mg/ml). Total RNA amounts of IRF-1 resulted in increased F1A DNA was prepared and 20 mg was used for Northern blot analysis with a 400 bp SLPI DNA fragment as a probe. As a control of RNA binding activity (Figure 5c, lanes 1 to 3). The speci®city loading, 28S RNA obtained from migration of total RNA on the of the IRF-1 DNA binding complex was con®rmed by agarose gel used for Northern blot analysis is presented supershift analysis using an IRF-1 speci®c antibody (Figure 5c, lane 4). As predicted, F1A did not bind recombinant IRF-2 protein (Figure 5c, lanes 5 ± 7). Three thymidine residues at nt 7211 to 7209 are important for IRF-1 binding to F1A, since no IRF binding was detected on DF1A, an F1A oligonucleo- tide in which the three thymidine residues at nt 7211 to 7209 were substituted for adenines (Figure 5c, lanes 8 ± 10). Whole cell extracts from 7 h and 48 h Dox- induced IRF-1/RelA cells also bound to F1A (data not shown). These observations demonstrate the existence of an IRF-1 binding site situated between nt 7221 and 7200 of the SLPI promoter. Since the only other potential ISRE-site in the SLPI-H/E promoter frag- ment (nt 7276 to 769) between 774 and 765 did not bind IRF-1 (Figure 5a, lanes 7 and 8), F1A likely accounts for the DNA binding activity observed in the SLPI-H/E region. IRF-1 protein also bound to a second site on the SLPI promoter, as shown by IRF-1 binding to SLPI-E/ Figure 4 Sequence of the human SLPI promoter. Potential NF- A (Figure 5a, lanes 5 and 6); analysis of the SLPI-E/A kB and ISRE-like and/or IRF-1 binding sites are highlighted, sequence revealed a 5'-GAAA-3' sequence at nt 742 to while the CAAT and TATA boxes are underlined. Sequences of 739. The F1B oligonucleotide which encompasses this the BS1, BS2, F1A and F1B oligonucleotides used in EMSA are delineated with solid lines. The star represents the EcoRI site used 5'-GAAA-3' sequence was found to be bound by IRF-1 to divide the promoter fragment into two parts for EMSA with about tenfold lower anity than F1A (data not analysis in Figure 6a. Exon 1 of the SLPI gene is shown in italics shown), suggesting the existence of a second, putative IRF-1 regulation of SLPI expression HNguyenet al 5459 IRF-1 site in the 751 to 722 region of the SLPI indirectly, by inducing expression of or activating a promoter. SLPI repressor protein. IRF-1 was previously shown to induce repressor IRF-2 expression (Harada et al., 1994; Cha and Deisseroth, 1994). To test the possibility that IRF-2 is not a direct IRF-1-induced repressor of SLPI SLPI expression may be suppressed directly by IRF-1 expression induced IRF-2, the kinetics of IRF-2 expression was IRF-1-mediated downregulation of SLPI may occur by examined in Dox-induced IRF-1/RelA cells (Figure 6). two mechanisms ± directly, by binding to SLPI As shown in Figure 6a, IRF-2 protein levels did not promoter sequences and repressing transcription, or change signi®cantly following Dox induction of IRF-1/

Figure 5 Detection of IRF-1 DNA binding to two regions of the SLPI promoter. (a) IRF-1 binding to whole or half fragments of the SLPI promoter. Twenty-®ve ng of recombinant IRF-1 protein was incubated with 32P-labeled probes corresponding to either the entire SLPI promoter (7276 to +1; lanes 1 and 2), the 5' 207 bp of the SLPI promoter (7276 to 769 or SLPI-H/E; lanes 3 and 4), the 3' 69 bp of the SLPI promoter (768 to +1 or SLPI-E/A; lanes 5 and 6) or BS12, a combination of the BS1 and BS2 sites (789 to 745; lanes 7 and 8). To con®rm speci®city of binding, IRF-1 protein was preincubated with 1 ml of anti-IRF-1 antisera (lanes 2, 4, 6 and 8) prior to addition of radiolabeled probe. The arrows indicate the IRF-1 speci®c complexes in lanes 1, 3, and 5, respectively. (b) Localization of IRF-1 binding between nt 7221 to 7200 by DNase I footprinting. An adenine-guanine (A+G) sequence of (+)-sense 32P-end-labeled SLPI promoter DNA is used as a DNA ladder (lane 1). In the footprint reactions, (+)- sense 32P-end-labeled SLPI promoter DNA was incubated in the presence of DNA binding bu€er and 0 ± 100 ng of recombinant IRF-1 protein (lanes 2 ± 6), followed by the addition of DNase I. The sequence of the SLPI promoter between 7200 and 7221 is shown. A DNase I-protected region representing potential IRF-1 binding is depicted on the right. (c) Binding of IRF-1 and IRF-2 protein to F1A and DF1A. 0.125 ± 1.25 ng of IRF-1 (lanes 1 ± 3) or 1 ± 4 ng of IRF-2 (lanes 5 ± 7) recombinant protein was incubated with a 32P-labeled probe corresponding to F1A (nt 7221 to 7200 on the SLPI promoter). Speci®city of binding was con®rmed by preincubation of IRF-1 protein with 1 ml of anti- IRF-1 antisera (lane 4) prior to addition of radiolabeled probe. Comparable amounts of IRF protein were also incubated with 32P-labeled DF1A probe, in which the three thymidine residues at nt 7211 to 7209 had been changed to adenines IRF-1 regulation of SLPI expression HNguyenet al 5460 RelA. Furthermore, IRF-2 DNA binding to the IRF-2- promoter was analysed in transient transfection studies speci®c Th probe was only detected in whole cell together with the CMV-IRF-1 plasmid, or transfected extracts from uninduced cells, as con®rmed by super- with CMV BL empty vector DNA and subsequently shift analysis using an IRF-2-speci®c antibody (Figure treated with e€ector molecules or infected with Sendai 6b, lanes 1, 8 and 10). IRF-1/RelA DNA binding virus (Figure 7). Co-transfection with increasing activity was not detected in the absence of Dox (Figure amounts of IRF-1 resulted in a dose-dependent 6b, lane 7) but after Dox induction for 48 and 72 h ± decrease in CAT activity (up to 3.8-fold). Tenfold when SLPI expression is completely suppressed (Figure lower amounts of IRF-1 did not activate the SLPI 1b) ± IRF-1/RelA strongly bound to Th (Figure 6b, promoter; it is possible that constitutive, rather than lanes 2 to 6 and lane 9). These results, together with inducible expression of even low amounts of IRF-1 the ®nding that IRF-2 does not bind to the SLPI occuring during a transient transfection assay resulted promoter, indicate that IRF-2 does not play a direct in repression of SLPI expression. Surprisingly, treat- role in IRF-1-mediated downregulation of SLPI ment of cells with the IFNg, LPS, dsRNA or Sendai expression. virus infection did not a€ect hSLPI-CAT activity.

Analysis of hSLPI-CAT promoter activity Discussion To examine the potential role of IRFs and NF-kBon transcriptional regulation of SLPI expression, a CAT The combination of tetracycline inducible transgene reporter construct driven by the human SLPI (hSLPI) expression and di€erential display/RNA ®ngerprinting were used to identify the secretory leukocyte protease inhibitor (SLPI) as a novel IRF-1-regulated gene. Characterization of the role of IRF-1 in the regulation of SLPI expression demonstrated that: (1) SLPI mRNA expression was activated early (2 ± 4 h) and then later repressed (12 ± 48 h) following Dox induction of IRF-1/RelA, mediated via two putative IRF-1 sites

Figure 6 Expression and DNA binding activity of IRF-2 following Dox induction of IRF-1/RelA. (a) Expression of endogenous IRF-2 protein in rtTA-IRF-1/RelA cells. Twenty mg Figure 7 Repression of SLPI promoter activity in response to of whole cell extract prepared from 0 ± 72 h Dox induced rtTA- IRF-1 co-expression. NIH3T3 cells were either cotransfected with IRF-1/RelA cells was subjected to SDS ± PAGE and transferred 5 mg of hSLPI-CAT DNA and 15 mg of CMV BL empty vector to nitrocellulose membrane. IRF-2 levels were detected using an DNA (considered as 100% relative CAT activity) or cotransfected IRF-2 antibody. (b) DNA binding activity of endogenous IRF-2 with 2, 5 or 10 mg of CMV-IRF-1 expression plasmid. In some in rtTA-IRF-1/RelA cells. The extracts described in (a) were experiments, cells were treated with either 100 mg/ml IFNg and/or incubated with a 32P-labeled probe corresponding to Th 10 ng/ml LPS, 25 mg/ml dsRNA or infected with 60 hemaggluti- [(AAGTGA)4]. Speci®city of binding was con®rmed by preincu- nating units/ml of Sendai virus 24 h post transfection for a period bation of extract with 1 ml of anti-IRF-2 (lanes 8 and 10) or anti- of 16 ± 24 h. All transfections were performed 3 ± 5 times. CAT RelA antibody (lanes 7 and 9) prior to addition of radiolabeled activity was assayed using 5 mg of total protein extract for 1 h at probe 378C IRF-1 regulation of SLPI expression HNguyenet al 5461 in the SLPI promoter; (2) dsRNA induced expression 1 response elements in the iNOS promoter (Kamijo et of SLPI, and expression of IRF-1 or IRF-1/RelA al., 1994; Martin et al., 1994). These above observa- repressed dsRNA-mediated induction of SLPI; (3) an tions are consistent with the role of IRF-1 described in IRF-1 binding site was identi®ed in the 7221 to 7200 the present study, as a repressor of SLPI expression region of the SLPI promoter; and (4) SLPI transcrip- which in turn interferes with NO production. tion was negatively regulated by IRF-1 in co- The RNA ®ngerprint leading to the identi®cation of expression assays. SLPI, the patterns of SLPI expression late after SLPI plays a primary role in the regulation of induction of IRF-1/RelA, and the co-transfection neutrophil-mediated in¯ammation through proteolysis studies all reveal a unique involvement of IRF-1 in and subsequent inhibition of the leukocyte serine repression of SLPI gene transcription. A similar proteases, including the neutrophil proteases (cathe- phenomenon was observed previously with IRF-2. psin G and elastase) and the pancreatic proteases Although viewed as a transcriptional repressor, two ( and chymotrypsin) (Thompson and Ohlsson, reports demonstrated that IRF-2 also acts as an 1986). Because of its potent antiprotease activity, SLPI activator. Human histone H4 gene FO108 was found serves as a potential therapeutic agent for the to be directly activated by IRF-2 through binding to a treatment of proteolytic tissue damage seen in cell-cycle element (CCE) present in the H4 promoter degenerative and in¯ammatory diseases such as cystic (Vaughan et al., 1995). The Qp promoter region of the ®brosis, allergic rhinitis and (Lee et al., 1993); Epstein-Barr Virus (EBV)-encoded EBNA-1 gene was reviewed in Vogelmeier et al. (1996). This nonglycosy- also activated by IRF-2 (Nonkwello et al., 1997). From lated 11.7 kDa enzyme is produced by epithelial cells our studies, IRF-1 repression of SLPI occurs when and resides in parotid secretions, bronchial, nasal and IRF-1 levels are high. Dox-induced levels of IRF-1 and cervical mucus, and seminal ¯uid (Thompson and IRF-1/RelA in the cell inhibited dsRNA-mediated Ohlsson, 1986). The relatively high basal levels of SLPI induction of SLPI; interestingly, LPS-mediated SLPI mRNA in ®broblast NIH3T3 cells observed in our induction was also suppressed by IRF-1 fusion protein experiments is consistent with high production of SLPI expression (data not shown). It is possible that cells by epithelial cells. Phorbol 12-myristate 13-acetate which are `primed' by IRF-1 expression are refractory (PMA) and tumor necrosis factor-alpha (TNFa) to inducers of SLPI expression as a mechanism of treatment increases SLPI levels in bronchial epithelial maintaining the pathogen or antiviral response. cells (Maruyama et al., 1994; Sallenave et al., 1994). Consistent with the direct role of IRF-1 in SLPI Activation of SLPI expression by these in¯ammatory regulation, two regions of the SLPI promoter (7221 to stimuli is concordant with its role in in¯ammation. 7200 and 751 to 722) were identi®ed that bind IRF- SLPI mRNA levels were also upregulated in response 1; in vitro footprinting and EMSA analysis determined to dsRNA and Sendai virus infection (data not shown), the 7200 site as the major site for IRF-1-DNA suggesting a potential involvement of SLPI in antiviral complex formation. IRF-1 had about tenfold higher defense as well as pathogen response. It would be of anity for the 7221 to 7200 region than the interest to determine whether this induction is mediated downstream 751 to 722 element, concordant with through dsRNA mediated activation of the dsRNA the fact that the single 5'-GAAA-3' stretch in the activated protein kinase PKR (Clemens and Elia, downstream sequence represents a half site for IRF-1 1997). binding. It is possible that IRF-1 may bind the 751 to Recently, a novel function was assigned to SLPI 722 region but only in association with an induced with its characterization as a macrophage derived `partner' protein. DNA binding of IRF-1 and IRF-2 inhibitor of macrophage response to LPS (Jin et al., was previously shown to be enhanced by interaction 1997). SLPI in this context was identi®ed by di€erential with the TFIIB component of the basal transcription display as a gene that was overexpressed in primary machinery (Wang et al., 1996). The proximity of the macrophages from LPS hyporesponsive (Lpsd) C3H/ 751 to 722 IRF-1 site to the TATA box suggest that HeJ mice. Although activated by LPS, SLPI antag- TFIIB may be a putative candidate as the IRF-1 onized the LPS response when overexpressed in `partner' protein. macrophages (Jin et al., 1997). One of the components Based on available evidence and data presented in of the LPS pathway inhibited by SLPI was the this paper, the following model of SLPI regulation is production of nitric oxide (NO), a major constituent proposed (Figure 8). Response to LPS involves the of macrophage cytotoxicity for bacteria and other induction of expression of TNFa and nitric oxide as foreign particles (reviewed in Nathan and Hibbs, 1991) well as the activation of NF-kB transcription factors. and an agent of antiviral defense (Croen, 1993; LPS also activates expression of SLPI (Jin et al., 1997). Karupiah et al., 1993). Interestingly, studies in NF-kB may mediate LPS-induced activation of SLPI macrophages from IRF knockout mice revealed an expression (+) or regulate the LPS response by essential role for IRF-1 in the transcriptional induction suppressing SLPI expression (7). IFNg and dsRNA/ of the nitric oxide synthase gene (iNOS) in macro- Sendai virus infection also activate SLPI expression. phages (Kamijo et al., 1994; Martin et al., 1994). Since IRF-1 protein expression correlates with SLPI Macrophages from IRF-1 de®cient mice produced little mRNA expression in IFNg and dsRNA/Sendai virus or no nitric oxide and synthesized low levels of iNOS infected cells (data not shown), IRF-1 serves as a mRNA in response to IFNg stimulation. Furthermore, potential mediator of SLPI activation in response to IRF-1 knockout mice were more severely a€ected by these stimuli. On the other hand, high levels of IRF-1 Mycobacterium bovis infection than their wild-type may trigger a feedback mechanism to regulate SLPI counterparts (Kamijo et al., 1994). The role of IRF-1 transcription, resulting in suppression of induction by in the IFN-g induced activation of iNOS expression is dsRNA and perhaps other SLPI inducers. IRF-1 may further supported by the presence of two adjacent IRF- repress directly by binding to the IRF-1 binding site(s) IRF-1 regulation of SLPI expression HNguyenet al 5462 using the ®rst 300 bp of the IRF-1 cDNA or a 400 bp fragment of SLPI cDNA (nucleotides 786 ± 1089), respec- tively, labeled with a-32P-dCTP by nick translation using the Oligolabelling Kit (Pharmacia Biotech).

Electrophoretic mobility shift assays Whole cell extracts for analysis of DNA binding were prepared as described in (Cohen et al., 1991). Twenty micrograms (20 mg) of whole cell extract or one to ten nanograms (1 ± 10 ng) of recombinant protein were preincu- bated for 10 min at room temperature in the presence of binding bu€er (10 mM: Tris-HCl (pH 7.5), 50 mM NaCl, Figure 8 Model for the regulation of SLPI expression. `7', 1mM EDTA, 1 mM DTT, 5% glycerol, 0.5% Nonidet P-40, repression; `+', activation. Response to LPS involves the and 62.5 mg/ml of the non-speci®c DNA competitor poly (dI- induction of expression of TNFa and nitric oxide as well as the dC) (Pharmacia Biotech) in a total volume of 20 ml. Reaction activation of NF-kB transcription factors. LPS also activates mixtures involving the recombinant proteins also contained expression of SLPI. NF-kB may mediate LPS-induced activation of SLPI expression or regulate the LPS response by suppressing 10 mg/ml bovine serum albumin. Following 30 min of SLPI expression. IFNg and dsRNA also activate SLPI expression. incubation with 75 000 c.p.m. of labeled probe at room IRF-1 serves as a potential mediator of SLPI activation in temperature, the mixtures were loaded on a 5% polyacryla- response to these stimuli, since IRF-1 protein expression mide gel (60:1 cross-link) prepared in 0.56TBE. After correlates with SLPI mRNA expression in IFNg and dsRNA/ running at 200 V for 2.5 h, the gel was dried and exposed Sendai virus infected cells. High levels of IRF-1 may trigger a to Kodak ®lm at 7708C overnight. To test the speci®city of feedback mechanism to regulate SLPI transcription, resulting in DNA binding, recombinant protein or whole cell extract suppression of induction by dsRNA and perhaps other SLPI from cells treated with 100 units/ml IFNg (Sigma) for 0 ± 18 h inducers. IRF-1 may repress directly by binding to the IRF-1 was pre-incubated with antiserum or 500-fold excess of binding site(s) in SLPI promoter and blocking transcription or unradiolabeled competitor DNA for 10 min at room indirectly through activation of an unde®ned repressor `X' which would bind directly to the SLPI promoter and repress temperature in binding bu€er, prior to the addition of transcription probe. The sequence of the probes used are as follows: Th (synthetic tetrahexamer of minimal IRF binding site): 5'-

(AAGTGA)4-3'; BS1 (nt 789 to 761 of the SLPI promoter): 5'-AGCTGGGAGAGGCCCGAAAGAATTCTGGT-3'; BS2 in the SLPI promoter and blocking transcription, in (nt 779 to 745 of the SLPI promoter): 5'-GCCCGAAA- which case IRF-1 would be assigned a novel repressor GAATTCTGGTGGGGCCCACCCA CTG-3'; BS12 (nt function distinct from its well characterized role as a 789 to 745 of the SLPI promoter (BS1 and BS2 transcriptional activator (reviewed in Nguyen et al., combined)): 5'-AGCTGGGAGAGGCCCGAAAGAATTC- 1997a). In contrast, IRF-1 may also function by an TGGTGGGGCCACACCCACTG-3'; SLPI-H/E: nt 7276 indirect mechanism through activation of an unde®ned to 769 of the SLPI promoter; SLPI-E/A: nt 768 to +1 of the SLPI promoter; F1A (nt 7224 to 7201 of the repressor (X) which would bind directly to the SLPI SLPI promoter): 5'-CCAATCTCTTCCCTTTCTTTTCTC-3'; promoter and repress transcription. Another possibility DF1A (point-mutated SLPI-F1A): 5'-CCAATCTCTTCCC- is that IRF-1 may indirectly alter the stability of SLPI AAACTTTTCTC-3'; and F1B (nt 751 to 722 of the SLPI mRNA. Our identi®cation of SLPI as a novel IRF-1 promoter). regulated gene and characterization of SLPI illustrate an interesting but complex mechanism of control, with implications for inflamma- Western blot analysis tory response to viral pathogens. Twenty mg of whole cell extract was subject to SDS- polyacrylamide gel electrophoresis (SDS ± PAGE) in a 10% polyacrylamide gel. After electrophoresis, the proteins were transferred to Hybond transfer membrane (Amersham) in a Materials and methods bu€er containing 30 mM Tris, 200 mM glycine and 20% methanol for 1 h. The membrane was blocked by incubation RNA ®ngerprinting in phosphate-bu€ered saline (PBS) containing 5% dried milk The RNA ®ngerprinting reactions were performed using the for 1 h and then probed with IRF-2 (Santa Cruz DeltaTM RNA Fingerprinting Kit (CLONTECH Labora- Biotechnology, Inc.) or C-terminal NF-kB RelA (Pepin et tories, Inc.). The reactions were resolved on a 5% Long al., 1994) antibody in 5% milk/PBS, at a dilution of 1 : 1000. RangerTM gel (50% Concentrate; JT Baker, Inc.) in 0.66 These incubations were done at 48C overnight or at room TBE. After running at 65 W for 1.5 h, the gel was dried and temperature for 1 ± 3 h. After four 10 min washes with PBS, exposed to Kodak ®lm at 7708C overnight. Bands of interest membranes were reacted with a peroxidase-conjugated were removed and reampli®ed, and then subcloned into a secondary goat anti-rabbit or anti-mouse antibody (Amer- Bluescript KS vector containing overhang thymidine (T) sham) at a dilution of 1 : 2500. The reaction was then bases inserted at the EcoRV site in the multiple cloning visualized with the enhanced chemiluminescence detection region. system (ECL) as recommended by the manufacturer (Amersham Corp.).

RNA extraction and Northern blot analysis Transient transfection and reporter gene assay Total cellular RNA was isolated by the guanidium thiocyanate method as described in Chomczynski and Sacchi Transient transfections were carried out in murine NIH3T3 (1987). Ten to twenty mg of total RNA was denatured, cells grown in Dulbecco's Modi®ed Eagles Medium (DMEM) electrophoresed in a denaturing formaldehyde/1.2% agarose supplemented with 10% fetal bovine serum, glutamine and gel, and transferred to nylon membrane. IRF-1/RelA and antibiotics. Subcon¯uent cells were transfected with CsCl SLPI mRNA were visualized by Northern blot hybridization puri®ed chloramphenicol acetyltransferase (CAT) reporter IRF-1 regulation of SLPI expression HNguyenet al 5463 driven by the human SLPI promoter (hSLPI-CAT) and Lin et al., (1992). Essentially, end-labeled DNA was CMV-IRF-1 expression plasmid by lipofection (Lipofecta- incubated in the presence of DNA binding bu€er containing mine, Life Technologies, Inc.) according to manufacturer's 10 mg/ml BSA and 0 to 150 ng of polyhistidine-tagged instructions. In some experiments, cells were treated with recombinant IRF-1 protein. After 10 min at room tempera- either 100 units/ml IFNg (Sigma), 10 mg/ml LPS (Sigma), ture, DNaseI was added and the reaction was incubated for both IFNg and LPS, or 25 mg/ml poly (I):poly (C), or 2 min at 308C. The reaction was then terminated by the infected with Sendai virus (60 hemagglutinating units/ml for addition of stop solution (0.6 M sodium acetate, pH 5.2, 90 min) 24 h post transfection for a period of 16 ± 24 h. All 50 mM EDTA and 1.5 mg/ml salmon sperm DNA). Protein transfections were performed 3 ± 5 times. CAT activity was was removed by phenol-chloroform extraction and the assayed using 5 mg of total protein extract for 1 h at 378C. resulting DNA was precipitated and resuspended in 6 mlof formamide loading bu€er. The DNA ladder and footprinting reactions were then electrophoresed on a 5% polyacrylamide DNAseI footprinting (19 : 1 crosslink)-8 M urea sequencing gel. After running at For DNAseI footprinting of the sense (+) strand, a HindIII/ 65 W for 1.5 h, the gel was dried and exposed to Kodak ®lm AccI fragment encompassing the complete known promoter at 7708C overnight. region (nucleotides 7276 to +1 of the SLPI gene) was end- labeled at the HindIII site. An adenine-guanine (A+G) DNA ladder was formed with 10 000 ± 20 000 c.p.m. of probe using Acknowledgements the chemical (Maxam-Gilbert) sequencing method. The This work was supported by grants from the Medical reaction was then completed by the addition of hydrazine Research Council of Canada and the Canadian Foundation stop solution (0.3 M sodium acetate, pH 7.0; 0.1 mM EDTA, for AIDS Research (CANFAR). H Nguyen is the recipient pH 8.0; and 100 mg/ml yeast tRNA), ethanol precipitated and of an FCAR studentship, R Lin is the recipient of a Fraser, then resuspended in 6 ml of formamide loading bu€er (90% Monat, MacPherson Fellowship from McGill University formamide, 10% bromophenol blue and 10% xylene cyanol). and J Hiscott is the recipient of a Senior Scientist Award The footprinting reactions were performed as described in from the Medical Research Council of Canada.

References

Cha Y and Deisseroth AB. (1994). J. Biol. Chem., 269, 5279 ± Matsuyama T, Kimura T, Kitagawa M, Watanabe N, 5287. Kundig T, Amakawa R, Kishihara K, Wakeham A, Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, Potter J, Furlonger C, Narendran A, Suzuki H, Ohashi 150 ± 159. P, Paige C, Taniguchi T and Mak T. (1993). Cell, 75, 83 ± Clemens MJ and Elia A. (1997). J. Interferon Cytokine Res., 97. 17, 503 ± 524. Miyamoto M, Fujita T, Kimura Y, Maruyama M, Harada CohenL,LacosteJ,ParniakM,DaigneaultL,SkupDand H, Sudo Y, Miyata T and Taniguchi T. (1988). Cell, 54, Hiscott J. (1991). Cell Growth Di€er., 2, 323 ± 333. 903 ± 913. Croen KD. (1993). J. Clin. Invest., 91, 2446 ± 2452. Nathan CF and Hibbs Jr JB. (1991). Curr. Opin. Immunol., 3, Duncan GS, Mittrucker HW, Kagi D, Matsuyama T and 65 ± 70. Mak TW. (1996). J. Exp. Med., 184, 2043 ± 2048. Nguyen H, Hiscott J and Pitha PM. (1997a). Cyt. Growth Fujita T, Sakakibara J, Sudo Y, Miyamoto M, Kimura Y Fact. Rev., 8, 293 ± 312. and Taniguchi T. (1988). EMBO J., 7, 3397 ± 3405. Nguyen H, Lin R and Hiscott J. (1997b). Oncogene, 15, Harada H, Fujita T, Miyamoto M, Kimura Y, Maruyama 1425 ± 1435. M, Furia A, Miyata T and Taniguchi T. (1989). Cell, 58, Nonkwello C, Ruf IK and Sample J. (1997). J. Virol., 71, 729 ± 739. 6887 ± 6897. Harada H, Kitagawa M, Tanaka N, Yamamoto H, Harada Penninger JM and Mak TW. (1994). Immunol. Rev., 142, K, Ishihara M and Taniguchi T. (1993). Science, 259, 231 ± 272. 971 ± 974. Pepin N, Roulston A, Lacoste J, Lin R and Hiscott J. (1994). HaradaH,TakahashiE-I,ItohS,HaradaK,HoriT-Aand Virol., 204, 706 ± 716. Taniguchi T. (1994). Mol. Cell Biol., 14, 1500 ± 1509. ReisLFL,HaradaH,WolchokJD,TaniguchiTandVilcek Jin F-Y, Nathan C, Radzioch D and Ding A. (1997). Cell, 88, J. (1992). EMBO J., 11, 185 ± 193. 417 ± 426. Reis LFL, Ru€ner H, Stark G, Aguet M and Weissmann C. Kamijo R, Harada H, Matsuyama T, Bosland M, Gerecitano (1994). EMBO J., 13, 4798 ± 4806. J, Shapiro D, Le J, Koh SI, Kimura T, Green SJ, Mak TW, Sallenave JM, Shulmann J, Crossley J, Jordana M and Taniguchi T and Vilcek J. (1994). Science, 263, 1612 ± Gauldie J. (1994). Am.J.Resp.CellMol.Biol.,11, 733 ± 1615. 741. Karupiah G, Xie QW, Buller RM, Nathan C, Duarte C and Taki S, Sato T, Ogasawara K, Fukuda T, Sato M, Hida S, MacMicking JD. (1993). Science, 261, 445 ± 448. Suzuki G, Mitsuyama M, Furuta T, Kojima S, Taniguchi Kimura T, Nakayama K, Penninger J, Kitagawa M, Harada T and Asano Y. (1997). Immunity, 6, 1 ± 20. H, Matsuyama T, Tanaka N, Kamijo R, Vilcek J, Mak Tanaka N, Kawakami T and Taniguchi T. (1993). Mol. Cell TW and Taniguchi T. (1994). Science, 264, 1921 ± 1924. Biol., 13, 4531 ± 4538. Lee CH, Igarashi Y, Hohman RJ, Kaulbach H, White MV Taniguchi T, Lamphier MS and Tanaka N. (1997). Biochim. and Kaliner MA. (1993). Am. Rev. Resp. Dis., 147, 710 ± Biophys. Acta Rev. Cancer, 1333, M9 ± M17. 716. Thompson RC and Ohlsson K. (1986). Proc. Natl. Acad. Sci. Lin R, Ernsting B, Hirsh®eld IN, Matthews RG, Neidhardt USA, 83, 6692 ± 6696. FC, Clark RL and Newman EB. (1992). J. Bacteriol., 174, Vaughan PS, Aziz F, van Wijnen AJ, Wu S, Harada H, 2779 ± 2784. Taniguchi T, Soprano KJ, Stein JL and Stein GS. (1995). Lin R, Mustafa A, Nguyen H and Hiscott J. (1994). J. Biol. Nature, 377, 362 ± 365. Chem., 269, 17542 ± 17549. Vogelmeier C, Gillisen A and Buhl R. (1996). Chest, 110, Martin E, Nathan C and Xie Q-W. (1994). J. Exp. Med., 180, 261S ± 266S. 977 ± 984. Wang I-M, Blanco JCG, Tsai SY, Tsai M-J and Ozato K. Maruyama M, Hay JG, Yoshimura K, Chu C-S and Crystal (1996). Mol. Cell Biol., 16, 6313 ± 6324. RG. (1994). J. Clin. Invest., 94, 368 ± 375.