Two Discrete Promoters Regulate the Alternatively Spliced Human Interferon Regulatory Factor-5 Isoforms

Interferon regulatory factor-5 (IRF-5) is a mediator of virus-induced immune activation and type I interferon (IFN) gene regulation. In human primary plasmacytoid dendritic cells (PDC), IRF-5 is transcribed into four distinct alternatively spliced isoforms (V1, V2, V3, and V4), whereas in human primary peripheral blood mononuclear cells two additional new isoforms (V5 and V6) were identified. The IRF-5 V1, V2, and V3 transcripts have different noncoding first exons and distinct insertion/deletion patterns in exon 6. Here we showed that V1 and V3 have distinct transcription start sites and are regulated by two discrete promoters. The V1 promoter (P-V1) is constitutively active, contains an IRF-E consensusbinding site, and is further stimulated in virus-infected cells by IRF family members. In contrast, endogenous V3 transcripts were up-regulated by type I IFNs, and the V3 promoter (P-V3) contains an IFN-stimulated responsive element-binding site that confers responsiveness to IFN through binding of the ISGF3 complex. In addition to V5 and V6, we have identified three more alternatively spliced IRF-5 isoforms (V7, V8, and V9); V5 and V6 were expressed in peripheral blood mononuclear cells from healthy donors and in immortalized B and T cell malignancies, whereas expression of V7, V8, and V9 transcripts were detected only in human cancers. The results of this study demonstrated the existence of multiple IRF-5 spliced isoforms with distinct cell type-specific expression, cellular localization, differential regulation, and dissimilar functions in virus-mediated type I IFN gene induction.


Interferon regulatory factor-5 (IRF-5) is a mediator of virus-induced immune activation and type I interferon (IFN) gene regulation. In human primary plasmacytoid dendritic cells (PDC), IRF-5 is transcribed into four distinct alternatively spliced isoforms (V1, V2, V3, and V4), whereas in human primary peripheral blood mononuclear cells two additional new isoforms (V5 and V6) were
identified. The IRF-5 V1, V2, and V3 transcripts have different noncoding first exons and distinct insertion/ deletion patterns in exon 6. Here we showed that V1 and V3 have distinct transcription start sites and are regulated by two discrete promoters. The V1 promoter (P-V1) is constitutively active, contains an IRF-E consensusbinding site, and is further stimulated in virus-infected cells by IRF family members. In contrast, endogenous V3 transcripts were up-regulated by type I IFNs, and the V3 promoter (P-V3) contains an IFN-stimulated responsive element-binding site that confers responsiveness to IFN through binding of the ISGF3 complex. In addition to V5 and V6, we have identified three more alternatively spliced IRF-5 isoforms (V7, V8, and V9); V5 and V6 were expressed in peripheral blood mononuclear cells from healthy donors and in immortalized B and T cell malignancies, whereas expression of V7, V8, and V9 transcripts were detected only in human cancers. The results of this study demonstrated the existence of multiple IRF-5 spliced isoforms with distinct cell typespecific expression, cellular localization, differential regulation, and dissimilar functions in virus-mediated type I IFN gene induction.
Virus infection results in the activation of a defined set of cellular genes involved in host antiviral defense. Transcription factors of the interferon (IFN) 1 regulatory factor (IRF) family have been identified as having important roles in innate immunity by participating in both the immediate-early transcriptional response to infection and the secondary response to cytokines (1)(2)(3)(4). While playing an important role in the antiviral immune response, IRFs also participate in cell growth regulation and apoptosis (5,6). IRF-5 is one of the most recently characterized members of the IRF family (2,(7)(8)(9)(10)(11)(12)(13)(14). It encodes a 60 -63-kDa polypeptide that was originally identified as a regulator of type I IFN gene expression (7,9). However, recent studies have indicated that it plays a role in many aspects of host defense (11), including induction of multiple cytokines and chemokines involved in the recruitment of T lymphocytes (8). Subsequently, it has been shown that IRF-5 itself is regulated by type I IFN (8,10), indicating an important regulatory pathway for the controlled induction of multiple immunomodulatory genes (11).
The role of IRF family members in the induction of cell arrest and cell death has been well established (5,10,12,(15)(16)(17). Although IRF-5 participates in the virus-induced signaling cascade, its expression is also modulated by p53 (12), and it has a role in the apoptotic signaling pathway that is p53-independent (10). Thus, IRF-5 modulates the expression of a number of factors involved in cell cycle regulation and apoptosis, independent of viral infection (10). The ability of IRF-5 to induce expression of proteins involved in the regulation of cell growth, apoptosis, and immunomodulation represents an important defense mechanism against extracellular stress, including viral infection. How significant is the overlapping cross-talk between IFN and apoptotic signaling pathways is currently unknown, yet it has been suggested recently that type I IFN can also induce p53 (6). Altogether, the current data suggest an important role for IRF-5 in both IFN-and p53-mediated signaling pathways.
To date, a large number of IRF-5 sequences have been deposited to GenBank TM . Most interesting, the first complete coding sequence of human IRF-5, isolated from lymphocytes and deposited in 1996 to GenBank TM (accession number U51127), was not functionally characterized. Subsequently, we had cloned two new isoforms of IRF-5, here termed variants 3 (V3; GenBank TM accession number AY504946) and 4 (V4; Gen-Bank TM accession number AY504947) from a dendritic cell library (7), and we have characterized the identical polypeptide they encode (2,(7)(8)(9)(10)(11). In the past 2 years, two additional isoforms, termed variant 1 (V1; isoform a; NM_002200; U51127) and variant 2 (V2; isoform b; NM_032643), have been deposited to GenBank TM . These two cDNAs encode proteins distinct from the isoforms 3 and 4 that we have identified, and their functions have yet to be characterized. Sequence analyses of these four isoforms reveal that they differ in their first exon and in the pattern of deletion(s) in exon 6. Alternative splicing of pre-mRNA has been described for other IRF family members, including IRF-1, -3, and -7, in which each isoform may be differentially expressed depending on the cell or tissue type (18 -22). However, differential isoform expression and regulation of these IRFs at the promoter levels have not been addressed yet, even though the promoters of each have been isolated and partially characterized (23)(24)(25).
The aim of the present study was to examine the regulated transcription of the human IRF-5 gene. The transcription analyses demonstrate that two distinct promoter regions differentially regulate exon 1 of V1-associated transcripts and exon 1 of V3 transcripts, where the P-V1 promoter responds to viral infection and the P-V3 promoter responds to IFN stimulation. However, the transcription pattern of the IRF-5 gene is more complex, and we have identified nine distinct alternatively spliced IRF-5 mRNAs (V1-V9). Here we describe the alternative splice patterns of the new IRF-5 isoforms, their cell typespecific expression, localization, inducibility, and function in virus-mediated type I IFN gene induction.

Cell Culture and Reagents
Human HeLa, 293T, A549, and Madin-Darby bovine kidney cells were obtained from the American Type Tissue Collection (ATCC). Human fibroblasts (2fTGH) were from G. Stark (Cleveland Clinic), and primary human fibroblasts were from G. Hayward (Johns Hopkins University). These adherent cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Daudi, Namalwa, and BJAB B cell lymphomas, THP-1 acute monocytic leukemia, U937 monocytic lymphoma, and BC-3 and BCBL-1 primary effusion lymphomas (PEL) were from ATCC and cultivated in RPMI with 10% fetal bovine serum. Peripheral blood mononuclear cells (PBMC), plasmacytoid dendritic cells (PDC), monocytes (MO), natural killer (NK), and B and T cells were isolated and purified as described previously (26 -28) from fresh heparinized peripheral blood obtained with informed consent from healthy volunteers. PDC (Lin Ϫ , CD123 ϩ , CD11c Ϫ , and human leukocyte antigen-DR ϩ cells) were isolated by using BDCA-4 microbeads, MO by using CD14 beads, NK by using CD16 beads, T cells by using CD3 beads, and B cells by using CD19 beads (Miltenyi Biotec). Cell purities were all greater than 95%, as determined by flow cytometry. The human studies were approved by the Institutional Review Board of the New Jersey Medical School. Newcastle Disease virus (NDV) was obtained from the ATCC (VR-699); vesicular stomatitis virus was from P. Marcus (University of Connecticut); recombinant IFN-␣2B was purchased from Schering Plough (Kenilworth, NJ), and IFN-␥ was from R&D Systems, Inc. Herpes simplex virus type I (HSV-1) strain 2931 was originally obtained from Carlos Lopez then of the Sloan-Kettering Institute for Cancer Research (New York, NY); CpG (ODN 2336) was from the Coley Pharmaceutical Group (Wellesly, MA); Resiquimod (R848) was from GLSynthesis Inc. (Worcester, MA).

5Ј-Rapid Amplification of cDNA Ends (RACE)
The 5Ј ends of the IRF-5 mRNAs were determined by using RNA from unstimulated or IFN␣2-stimulated (500 units/ml for 16 h) BJAB cells using the 5Ј/3Ј-RACE Kit, 2nd Generation (Roche Applied Science), as described by the manufacturer. For amplification from each first exon (Ex1), the RACEV1R primer was used for Ex1V1, RACEV2R for Ex1V2, and RACEV3R for Ex1V3 (primers 1-3, Table I). RNA was reverse-transcribed using these three primers that are specific for each first exon, with subsequent RNase I degradation of the RNA template.
Following the addition of a homopolymeric A-tail to the antisense strand of the 5Ј end of cDNAs, the tailed cDNA was amplified using a sense oligo(dT)-anchor primer and the same antisense primers specific for each first exon as named above. A second nested PCR was then performed using the oligo(dT)-anchor primer and a nested antisense primer specific for each first exon (primers 4 -6, Table I). PCRs were carried out using Pfu proofreading polymerase for the generation of blunt-ended products, and an initial 2-min denaturation step at 94°C followed by 10 cycles of successive incubations at 94°C (15 s), 70°C (30 s), and 72°C (1 min) and 15 cycles of successive incubations at 94°C (15 s), 70°C (30 s), and 72°C (initially 40 s with an increase of 20 s in each successive cycle), and a final 7-min elongation at 72°C. PCR products were either electrophoresed on a 1.5% agarose gel or cloned in the pCR 4Blunt-TOPO vector using the Zero Blunt TOPO PCR cloning kit for sequencing (Invitrogen) and sequenced.
In order to distinguish between V1-and V4-specific expression and to determine expression of new IRF-5 variants associated with their specific deletion/insertion pattern(s) in exons 5-7, exons 1-7 were amplified with each specific exon 1 sense primer (primers 21-23, Table I and Fig. 1B) and a common antisense primer (primer 26, Table I and Fig. 1B). Optimal PCR conditions were identical to the exon 1-4 PCR amplification, except 37 cycles were used. PCR products were either electrophoresed on a 1.5% agarose gel and/or cloned to the pCR®2.1-TOPO vector using the TOPO TA cloning kit (Invitrogen) and sequenced.

Oligonucleotide Pull-down and Immunoblot Assays
Biotinylated antisense oligonucleotides were annealed with the corresponding sense oligonucleotides (primers 28 -31, Table I). Briefly, 4 g of biotinylated DNA was incubated with 100 g of DynaBeads M-280 streptavidin (Dynal Inc.) for 16 h in 200 l of TEN buffer (20 mM Tris, pH 8.0, 1 mM EDTA, 0.1 M NaCl), and unbound DNA was removed by extensive washing. Cell extracts were prepared as described previously (29) from control untreated BJAB cells or cells stimulated with 500 units/ml IFN␣2 for 16 h. Extracts (300 g of protein) were incubated with bound DNA (7,24) and washed, and the bound proteins were identified by immunoblotting. Rabbit polyclonal antibodies against ISGF3␥p48, STAT1, and STAT2 were purchased from Santa Cruz Biotechnologies. Signals were visualized using the ECL detection reagents (Amersham Biosciences). Dual Luciferase Assay-293T, 2fTGH, or HeLa cells were transfected in 60-mm dishes using SuperFect. 2 g of the reporter plasmid DNA and 0.1 g of pRL-CMV (internal control; Promega) were used for each transfection. In co-transfection experiments a 1:1 ratio of the reporter and expression plasmid (2 g each) were used. The final concentration of transfected DNA was kept constant in all co-transfection assays. Cells were harvested 24 h post-transfection unless additional treatments were performed. For further treatments, cells were split 12-16 h post-transfection into 6-well plates and incubated for another 8 h, after which medium was replaced either with Dulbecco's modified Eagle's medium containing recombinant IFN-␣2 (500 units/ml), IFN-␥ (1000 units/ml), or NDV (50 l/ml). Treatments/infections were incubated an additional 16 h, and cells were then harvested and lysed with 1ϫ Passive Lysis Buffer (Promega). Protein concentrations were determined using the Bio-Rad Protein Assay according to manufacturer's instructions. Luciferase assays were carried out using the Dual Luciferase Assay kit according to the manufacturer's specifications (Promega). Experiments were repeated at least three times in duplicate.

Treatments, Transient Transfections, Infections, and Reporter Assays
SAP Assay-2 ϫ 10 5 2fTGH or 2fTGH/IRF-5 V3 (7) stable expressing cells were transfected with a constant amount of DNA (2 g/6-well plate) by using the Superfect transfection reagent (Qiagen). Equal amounts of the indicated SAP reporter plasmids and IRF5-expressing plasmids were co-transfected with the ␤-galactosidase expression plasmid (50 ng). Transfected cells were split 16 h later, incubated for an additional 6 h, and either left uninfected or infected with NDV for another 16 h. The SAP was determined as described (29,33). Each experiment was repeated at least three times; ␤-galactosidase expression levels were used to normalize the difference in transfection efficiency.
IFN Cytopathic Effect Assay-2fTGH cells (0.5 ϫ 10 5 cells/well of a 6-well plate) were transiently transfected with 50 ng of each FLAGtagged IRF-5 full-length variant expression vector. 16 h later, cells were split in duplicate to a 24-well plate, incubated for 8 h, and either left uninfected or infected with NDV for an additional 16 h. The levels of biologically active type I IFN or IFN-␣ were determined in the cell culture supernatants by the viral cytopathic effect assay (34). VSV was used as the challenging virus, and the cytopathic effect was determined in human fibroblasts (type I IFN) or Madin-Darby bovine kidney cells (IFN-␣).

RESULTS
Structural Analysis of Isoform-specific Exons of the Human IRF-5 Gene-Although cDNA sequences for human IRF-5 isoform 1 (V1) and 2 (V2) have been deposited to GenBank TM , the proteins encoded by the cDNAs have not been functionally characterized. We had identified previously two new mRNA isoforms (V3 and V4), and we functionally characterized the identical polypeptide encoded by the two cDNAs (2, 7-11).
Here we examine the transcriptional regulation of these four isoforms (Fig. 1).
Sequence alignments between the four IRF-5 isoforms indicate two major differences: 1) the utilization of distinct alternative first exons in V1 (Ex1V1), V2 (Ex1V2), and V3 (Ex1V3), where V1 and V4 share the same first exon; and (2) the distinct pattern of deletion(s) in exon 6 of V1, V2, and V3, where V3 and V4 share the identical deletion pattern (Fig. 1B). Considering these differences in cDNAs, we first determined the organization of introns and exons in the human IRF-5 gene (Fig. 1A). Searches for homologous sequences using BLAST software mapped the V1, V2, and V3 exon sequences to the recently described human chromosome 7 (35). This mapping is consistent with our previous chromosomal assignment of the IRF-5 gene (7). Based on the cDNAs of IRF-5 deposited to Gen-Bank TM , there are at least three first exons (Ex1V1, Ex1V2, and Ex1V3) and eight common exons thereafter (Fig. 1, A and  B). Sequences at acceptor and donor splice sites usually include GU and AG at the beginning and end of the intron, as well as AG and GU at the end and beginning of the flanking exons. The 3Ј-flanking region of each IRF-5 first exon ends with an AG dinucleotide that is consistent with a splicing donor site. Multiple sequence alignments of the IRF-5 cDNAs demonstrate that the four different IRF-5 transcripts result from alternative usage of the first exon (Fig. 1B). In addition, when compared with the IRF-5 genomic DNA, these IRF-5 isoforms have distinct deletions in exon 6; the IRF-5 V1 isoform has a deletion of 30 nucleotides (nts); the IRF-5 V2 isoform has a deletion of 48 nts, and the IRF-5 V3 and V4 isoforms are missing both the 30and 48-nt regions (Fig. 1B).
Characterization of the 5Ј-UTR of IRF-5 mRNA-To investigate the transcriptional mechanism(s) underlying the regulation of V1, V2, V3, and V4 transcript expression, we identified and characterized the IRF-5 gene promoter sequences. To this effect, we first determined the transcription start site(s) of the IRF-5 gene by 5Ј-RACE. Because BJAB cells contain a functional/inducible IRF-5 promoter, as demonstrated by the stimulation of IRF-5 gene expression after IFN treatment (8, 10), we used RNA extracted from BJAB cells for the 5Ј-RACE experiments. Based on the IRF-5 cDNA sequences deposited to GenBank TM , we performed three 5Ј-RACE experiments, one for each of the known first exons (Fig. 1A). As shown in Fig. 1C, when unstimulated BJAB cells were subjected to 5Ј-RACE using an antisense primer specific for Ex1V1, three distinct products were visualized. Following nested PCR, one distinct band was detected (Fig. 1C, left panel), indicating the presence of a transcription start site for IRF-5 isoforms containing Ex1V1 in unstimulated BJAB cells. When we used the same approach to identify the 5Ј-ends of IRF-5 V2 and V3 tran-  Table I. C, characterization of the 5Ј-UTR of the IRF-5 mRNA. 5Ј-RACE was performed using RNA isolated from either unstimulated or IFN-stimulated BJAB cells. Three 5Ј-RACE experiments were performed, one for each first exon of IRF-5; the two successful experiments are shown. Electrophoretic analysis of Ex1V1 5Ј-RACE products from unstimulated BJAB cells shows a single band, resulting from nested PCR amplification of the three original bands. The identification of a nested PCR product from IRF-5 Ex1V3 following 5Ј-RACE was possible in IFN␣2-stimulated BJAB cells. D, sequencing of 5Ј-RACE products demonstrated that the V1 and V3 transcription start sites match the consensus sequence for transcription initiation. The A residue in each site has been assigned as position ϩ1 of the transcripts. scripts, we were unable to amplify either of the two 5Ј-UTRs in unstimulated BJAB cells. This is likely because of the very low levels of general IRF-5 transcripts and each of these isoforms present in unstimulated BJAB cells (8,10). Only when RNA was isolated from IFN-treated BJAB cells could the IRF-5 V3 5Ј-UTR be amplified (Fig. 1C, right panel). However, amplification of the IRF-5 V2 5Ј-UTR was not detected in either IFN-stimulated or unstimulated BJAB cells.
In summary, electrophoretic migration of 5Ј-RACE products amplified with nested PCR revealed one distinct band for each of the V1 and V3 first exons (Fig. 1C), suggesting that there is a discrete transcription start site for each of these two transcripts. Sequence analysis of the 5Ј-RACE products demonstrated that both transcription start sites matched the consensus sequence for transcription initiation, YYCAYYYYY, and the A residue in each has been assigned as position ϩ1 (Fig. 1D).
Isolation and Cloning of Two Alternative Promoters of IRF-5-As depicted in Fig. 1A, there are four 5Ј-flanking regions of the IRF-5 gene that have the potential to control/ regulate IRF-5 gene transcription; these regions are located as follows: (i) upstream of the Ex1V1 (P-V1); (ii) upstream of the Ex1V2 (P-V2); (iii) upstream of the Ex1V3 (P-V3); and (iv) upstream of exon 2 (P-Ex2), which contains the translation initiation ATG codon of human IRF-5. By using the TRANS-FAC data base (www.gene-regulation.com), we have identified a TATA box upstream of Ex1V1 and Ex1V3 but not upstream of Ex1V2 or exon 2. This analysis, combined with the 5Ј-RACE data, suggests that the IRF-5 gene contains only two promoters, as defined by the presence of the TATA boxes and transcription start sites upstream of Ex1V1 and Ex1V3. However, it is also possible that the Ex1V2 and exon 2 5Ј-flanking regions may be functioning TATA-less promoters. It has been shown previously that mRNA transcription initiating from multiple sites, as 5Ј-RACE demonstrates, is usually driven by TATAless promoters, and thus, we cannot exclude the possibility that additional transcripts initiating from these two 5Ј-flanking regions will be identified (36).
In order to test whether all of these 5Ј-flanking regions ( Fig.  1A, P-V1, P-V2, P-V3, and P-Ex2) function as constitutive or inducible promoters, DNA fragments corresponding to P-V1, P-V2, P-V3, and P-Ex2 (Fig. 1A) were amplified from BJAB genomic DNA and cloned into the promoter-less luciferase reporter pGL3-Basic. The putative promoter constructs were named according to the exon they flank. These four putative promoter constructs were transiently transfected to HeLa, 293T, or 2fTGH cells and assayed for luciferase activity. Of the four constructs, only the P-V1 and P-V3 promoters displayed basal activity, suggesting that only these two 5Ј regions of the IRF-5 genome can function as constitutive promoters in the cell types examined (data not shown). Because the P-V1 promoter contains an IRF-E consensus site and the P-V3 contains an ISRE site, we have focused on the role of these two sites in the regulation of IRF-5 gene expression in virus-infected cells (8,10).
Virus-mediated Activation of the IRF-5 Isoform 1 Promoter (P-V1); Regulation by IRFs-We first investigated the function of the IRF-E site, a known recognition site for members of the IRF family. Reporter constructs containing the entire wild type 0.64-kb P-V1 region or this region in which the IRF-E site was mutated were both inserted upstream of the luciferase gene in the pGL3-Basic vector (Fig. 2, A and B). The activities of these constructs were analyzed in transient transfection assays in 293T, HeLa, and 2fTGH cell lines (Fig. 2, C and D, depict data from HeLa cells, however all cell lines gave similar results). Transfection of the full-length wild type P-V1 promoter resulted in significant luciferase reporter activity that was ϳ100 times higher than the activity of the promoterless pGL3-Basic control vector (Fig. 2C), indicating that in the cell types examined, the P-V1 promoter is constitutively active. Infection with NDV enhanced promoter activity by 3-fold. When the P-V1 promoter construct was co-transfected to HeLa cells with IRF-1, -3, -5 (V3/4), or -7 expression plasmids, only IRF-1 and IRF-5 significantly enhanced promoter activity in NDV-infected cells (Fig. 2C). Co-transfection of IRF-7 had no effect on promoter activity, and IRF-3 only slightly enhanced P-V1 reporter activity in NDV-infected cells. No activation was detected when the IRFs were co-transfected with the full-length P-V1 promoter carrying a mutated IRF-E site (Fig. 2D, P-V1-IRFE-M), indicating that the IRF-E site is essential both for the constitutive and virus-induced activation. This result suggests that in vitro the IRF-E site may be required for the constitutive activation of the IRF-5 P-V1 promoter and constitutive expression of the IRF-5 V1 and V4 isoforms.
Because the P-V1 IRF-E site shows significant homology to the consensus ISRE site, a region identified for binding of the ISGF3 complex that assembles in IFN-␣-stimulated cells, HeLa cells transfected with the wild type P-V1 promoter were also treated with IFN-␣2. However, IFN-␣ did not stimulate the activity of the P-V1 promoter (data not shown), indicating that the IRF-E site in this promoter is not responsive to IFN treatment.
Interferon-mediated Activation of the IRF-5 Isoform 3 Promoter (P-V3); Regulation by IFN-We have observed previously that transcription of the IRF-5 gene is stimulated by human IFN-␣2 (8, 10) (Fig. 1C). Because 5Ј-RACE from the IRF-5 Ex1V3 was only successful in BJAB cells stimulated with IFN-␣2 and the P-V3 promoter contains an ISRE site (Fig. 3A), suggesting its role in the IFN␣2-mediated induction of IRF-5 V3 transcription (Fig. 1C), we examined whether the ISRE site was active. Reporter plasmids containing the entire 1.2-kb P-V3 region or its analogue with a mutated ISRE site (P-V3-ISRE-M) were analyzed for reporter activity by transient transfection to 293T, HeLa, or 2fTGH cells (Fig. 3, B and C). Transfection of the full-length wild type putative P-V3 promoter resulted in significant luciferase reporter activity that was ϳ100 times higher than when the promoterless vector pGL3-Basic control was transfected alone (Fig. 3C), indicating that the P-V3 promoter is constitutively active in these cells. Treatment of the P-V3-transfected cells with IFN-␣2 for 16 h resulted in about a 3-fold induction above the basal level activity (Fig. 3C), whereas IFN-␥ treatment led to only a 2-fold increase. Mutation of the ISRE element (Fig. 3, A-C, P-V3-ISRE-M) abolished the response to IFN-␣2 without affecting the base-line activity of the P-V3 reporter (Fig. 3C). These results indicate that the 5Ј-flanking region of IRF-5 V3 can function as a promoter and that the ISRE site in this promoter confers the IFN-␣2-mediated stimulation of the IRF-5 gene.
The induction of IFN-stimulated gene (ISG) promoters by IFN-␣ is generally mediated by binding of the ISGF3 complex, which consists of p48 (IRF-9), STAT1, and STAT2 proteins, to an ISRE consensus site (37). We have therefore used the oligonucleotide pull-down assay to examine whether the ISGF3 complex binds to the ISRE present in the IRF-5 P-V3 promoter. Because this ISRE differs slightly from the consensus ISRE sequence (Fig. 3B), we analyzed and compared the binding of the ISGF3 complex from whole cell extracts of unstimulated and IFN-␣2-stimulated BJAB cells to the biotin-labeled P-V3 ISRE and the consensus ISRE oligonucleotides immobilized on magnetic beads. The bound proteins separated on SDS gels were identified by immunoblotting. Results show that both the P-V3 ISRE and the consensus ISRE bound p48, STAT1, and STAT2 from the extracts of IFN-treated cells, whereas only low levels of STAT2 bound in uninfected cells (Fig. 3D). These data demonstrate that the P-V3 ISRE is functional and that the ISGF3 complex contributes to the IFN␣2-mediated activation of the IRF-5 P-V3 promoter.

IRF-5 Isoforms Are Differentially Expressed and Regulated in Purified Primary Cell Subpopulations from Healthy Donors-By
Northern blot analysis and RT-PCR, we have shown previously that the constitutive expression of IRF-5 transcripts occurs primarily in normal lymphoid tissue, peripheral blood lymphocytes, and dendritic cells; yet these transcripts were not detected in a variety of immortalized B and T cell leukemias (7) or primary hematological malignancies (10). At the time of these studies, however, the presence or functional role of multiple IRF-5 isoforms was not known. To date, only the functional roles for IRF-5 V3/V4 have been elucidated (2, 7-11).
Recently, it was shown that IRF-5 transcripts were constitutively expressed at high levels in PDC (26,38), and the high levels of IRF-5 expression may contribute to the high levels of IFN-␣ induced in these cells after stimulation with virus. IRF-5 V3 and V4 cDNAs were initially cloned from a dendritic cell library (7). Here we extend our previous findings to characterize further the constitutive and inducible expression of IRF-5 isoforms in purified PBMC, PDC, MO, NK, and T and B cells from healthy donors and in human immortalized cancer cell lines.
Primer sets that specifically recognize exon 1 of each isoform (Fig. 1B, Ex1V1, Ex1V2, and Ex1V3) and a common region in exon 4 of IRF-5 were optimized using cloned IRF-5 V1/4, V2, and V3 cDNAs. As shown in Fig. 4A, PCR amplification using the exon 1-specific sense primers was isoform-specific. Because we have detected previously high levels of general IRF-5 transcripts in PBMC and PDC (26), we first examined the levels of constitutive and inducible exon 1-specific IRF-5 isoform expression in similar purified cell populations. The isolated cell populations were greater than 95% pure as determined by flow cytometry. Cells (PBMC, PDC, MO, NK, and T and B cells) were either left unstimulated or stimulated with IFN-␣, HSV-1, or CpG as described under "Materials and Methods." The effect of CpG DNA on IRF-5 gene transcription was examined because the stimulation of PDC with oligodeoxyribonucleotides containing unmethylated CpG motifs has been shown to produce high levels of IFN␣ (38 -40). Results in Fig. 4B show   FIG. 4.

RT-PCR analysis of IRF-5 isoform expression in primary purified subpopulations of peripheral blood mononuclear cells or human immortalized hematological malignancies.
A, optimization of IRF-5 Ex1-specific PCR amplification using V1/4, V2, and V3 control cDNAs, as indicated at the top; primer sets are shown at the bottom, and each was amplified from the specific exon 1 to the conserved exon 4. Lanes 0 depict the water controls for each indicated primer set. B, the expression of endogenous IRF-5 transcripts was determined in purified PBMC, PDC, MO, NK, and T and B cells from a single donor. RNA was isolated from each of the purified cell populations after stimulation with HSV-1, IFN-␣, CpG, or from unstimulated cells. RT-PCR was performed with exon 1 isoform-specific primers that amplify through exon 4. IFNA and IRF-7 transcripts were examined in PBMC and PDC following stimulation with CpG (right panel). The control lane (cont) indicates amplification for contaminating DNA in each of the examined primer sets. C, IRF-5 Ex1-specific isoform expression was examined in unstimulated Daudi, Daudi infected with NDV, or stimulated with IFN-␣, as indicated. Unstimulated BC-3 and BCBL-1 PEL cells, unstimulated THP-1, and THP-1 infected with NDV or VSV, and THP-1 stimulated with R848 were examined, along with unstimulated U937 and Namalwa, as indicated. The levels of ␤-actin were determined as a control for RNA integrity. that the Ex1V1 transcripts were detected in all types of unstimulated cells except T cells; the highest levels were detected in unstimulated MO and B cells (top panel). Although treatment with IFN-␣ or CpG had either no effect (PDC and B cells) or decreased (PBMC, MO, NK, and T cells) IRF-5 Ex1V1 transcript levels, stimulation with HSV-1 enhanced the levels of Ex1V1-associated transcripts in PBMC, NK, and T cells. Unstimulated and stimulated (with HSV or IFN-␣) B cells expressed the highest levels of IRF-5 Ex1V1-associated transcripts that were unaffected by these inducers; expression was nearly identical in all three samples even at lower cycle numbers where amplification was not saturated but was in the linear range (data not shown). In comparison, transcripts associated with Ex1V2 were either very low (MO, NK, and B cells) or nearly undetectable (PBMC, PDC, and T cells) at 37 amplification cycles (Fig. 4B, middle panel). However, transcripts were detected in all cell types examined by increasing the amount of amplified cDNA and by using a higher number of amplification cycles compared with that required for the amplification of Ex1V1 or Ex1V3 transcripts. These results suggest that IRF-5 V2 transcripts, although present, were expressed at significantly lower levels than V1 or V3 transcripts (Fig. 4B). Furthermore, even at the higher number of amplification cycles, levels of the Ex1V2 transcripts were unaffected by stimulation with HSV-1, IFN-␣, or CpG (data not shown). Although Ex1V3 transcripts were detected in unstimulated PDC, MO, and B cells, stimulation with IFN-␣ significantly up-regulated the levels of Ex1V3 transcripts in all cell types examined (Fig. 4B, lower panel). The origin of the multiple amplification bands for Ex1V3 PCR in the IFN-␣-stimulated PBMC, PDC, and MO, and all B cell samples except for the CpG-stimulated B cells is presently unclear. However, these data may suggest the presence of multiple Ex1V3-associated IRF-5 isoforms. Although stimulation with HSV-1 did not affect Ex1V3 transcripts in most of the cells, transcript levels were dramatically decreased in MO and increased in NK, T, and B cells. CpG stimulation had either no effect or slightly decreased Ex1V3 transcript levels in all cell types except NK cells, where levels were up-regulated compared with unstimulated cells. However, stimulation with CpG did induce high levels of IFNA transcripts in PBMC and PDC but not in B cells and IRF-7 transcripts, and although dramatically up-regulated in PBMC, induction was less dramatic in PDC, and levels were down-regulated by CpG in B cells (Fig. 4B,  right panel).
We next examined IRF-5 isoform expression in human immortalized cancer cell lines, Daudi, BC-3 and BCBL-1, THP-1, U937, and Namalwa. For these studies, Daudi cells were either left unstimulated, infected with NDV, or stimulated with IFN-␣2. As shown in Fig. 4C, Ex1V1-associated transcript levels were up-regulated after infection with NDV, whereas V2 and V3 transcript levels were nearly undetectable in either the uninfected or infected samples. In comparison, IRF-5 V3 transcript levels were specifically induced after treatment with IFN-␣2. Constitutive isoform expression was then examined in the two PEL cell lines: where Ex1V1-associated transcripts were expressed at high levels in BCBL-1 and not BC-3 cells, and both cell lines expressed low levels of constitutive IRF-5 V3. On the other hand, THP-1 cells expressed high constitutive levels of both Ex1V1 and Ex1V3 transcripts and low levels of Ex1V2 transcripts. Upon further stimulation with either NDV or VSV, only Ex1V1-associated transcript levels were up-regulated (Fig. 4C). With the recent finding that IRF-5 is critical for TLR7/8-mediated induction of type I IFNs in THP-1 cells, we examined the inducible expression of IRF-5 isoforms by R848 (41). We found that although THP-1 cells indeed expressed the Ex1V1-, V2-, and V3-associated transcripts constitutively, albeit at distinct levels, only the IRF-5 V3 transcripts were up-regulated by R848. Furthermore, the examination of constitutive IRF-5 expression in U937 and Namalwa revealed the overwhelming presence of Ex1V1-associated transcripts, and low levels of V3 transcripts were only detected in U937 cells. Levels of human ␤-actin transcripts are shown as a control for RNA levels.
Taken together, results from the 5Ј-RACE experiments, reporter assays and RT-PCR indicate that only the P-V3 promoter and its Ex1V3-associated transcripts are stimulated by IFN treatment (Fig. 3C and Fig. 4, B and C). Furthermore, these data support the finding that IRF-5 Ex1V1-associated transcript expression is regulated by the virus-mediated activation of P-V1 (Fig. 4, B and C). The differences in the levels of constitutive and inducible IRF-5 isoform expression in purified immune cell subpopulations from healthy donors and in tumor cell lines of different origin suggest that expression of the IRF-5 isoforms is cell type-specific.
Isolation and Cloning of Five New Alternatively Spliced IRF-5 Isoforms-Whereas our previous screening method for IRF-5 isoform expression yielded valuable information with regard to expression and induction of exon 1-associated transcripts, we were unable to distinguish between variants 1 and 4 because they both utilize the same exon 1 (Ex1V1). Moreover, the previous screening method did not allow for the identification of potential new alternatively spliced isoforms. In order to distinguish between V1 and V4 transcripts, we optimized primer sets that amplified from exon 1 to exon 7 that includes the region where the different deletion patterns occur (Fig. 1B). PCR fragments amplified with each of the three primer sets (primers 21-23 and 26, Table I) from THP-1, PDC, PBMC, Namalwa, U937, and/or BCBL1 were purified, cloned, and sequenced. At least 75 positive clones from each ligation reaction were isolated and sequenced. From the sequences of these clones, we identified nine new IRF-5 isoforms, most of which were alternatively spliced from Ex1V1 (Table II). Five of these new IRF-5 isoforms were re-cloned in order to obtain fulllength IRF-5 cDNAs (Fig. 5A). Expression of the IRF-5 isoforms 5 and 6 was detected in PBMC and THP-1 cells, whereas expression of IRF-5 isoforms 7-9 was identified only in THP-1 cells. In addition, we were able to clone the full-length IRF-5 isoform 1 from THP-1 cells. All of the new IRF-5 isoforms share Ex1V1 except for V9 that contains Ex1V2. IRF-5 V5 contains the full genomic sequence of IRF-5 that does not show any deletions in exon 6 (Fig. 5A). IRF-5 V6 is a spliced isoform from Ex1V1 with a coding region identical to V2. IRF-5 V7 is lacking the majority of the DNA binding domain and thus may function as an endogenous dominant negative (DN) mutant of fulllength IRF-5, it has the same deletion pattern as V2. IRF-5 V8 has a large single deletion that spans most of exons 6 and 7. For IRF-5 V9, although we obtained the full-length mRNA, alternative splicing led to the introduction of an early termination codon, and the protein is therefore missing its carboxyl terminus. By using both RT-PCR and sequencing methods, the levels of the IRF-5 isoforms appear to be distinctly expressed. Purified PDC expressed only IRF-5 V1, V2, V3, and V4 transcripts, with V3 and V4 being the most predominant transcripts expressed in unstimulated PDC. In contrast, the most abundant transcripts constitutively expressed in the other cell types/lines examined were IRF-5 V5, V6, and V1, respectively (data not shown). Expression of the IRF-5 isoforms cloned to a FLAG-tagged or gfp-tagged vector was analyzed in transiently transfected 2fTGH cells by immunoblot ( Fig. 5B and data not shown). It can be seen that whereas the IRF-5 V1-V6 isoform cDNAs encode proteins of similar size, ϳ61 kDa, the IRF-5 V7 and V8 encode proteins of ϳ47 kDa, and V9 encodes a protein of only ϳ25 kDa, which correspond to their predicted sizes ( Fig. 1B  and Fig. 5, A and B). Since it has recently been demonstrated that gfp-IRF-5V1 and V2 are expressed in the cytoplasm of uninfected cells (13), and we had shown previously that the identical polypeptide encoded by IRF-5 V3 and V4 cDNAs was primarily localized to the cytoplasm, with low levels detected in the nucleus of uninfected cells (8), we next examined by fluorescent microscopy the subcellular localization of gfp-IRF-5 fusion proteins (V5 and V7-V9) transfected to 2fTGH cells. Fig.  5C shows that similar to IRF-5 isoforms V1-V4, IRF-5 isoforms V5-V7 were localized to the cytoplasm (and data not shown), whereas V8 and V9 were localized to the nucleus. Most interestingly, V8 contains both nuclear localization signals (8) and the recently described nuclear export signal (13), whereas V9 contains only the amino-terminal nuclear localization signal but is lacking the nuclear export signal.
Distinct Roles for IRF-5 Isoforms in Type I IFN Gene Induction-We have shown that cDNA encoding the IRF-5 V3 or V4 polypeptides can regulate the virus-mediated expression of distinct IFN-␣ subtypes, which are encoded by 13 human IFNA genes (2, 7). In a transient co-transfection assay, IRF-5 V3/4 stimulated IFNA1, A2, and A14 SAP reporter activity in both uninfected and NDV-infected cells (7). We have also shown recently that the polypeptides encoded by V3 and V4 cDNAs are important for the virus-mediated induction of IFNB and other immune response genes (11). Given that the predominant sources of type I IFN in human blood are the PDC, and PDC constitutively express IRF-5 isoforms 1-4, we evaluated first their contribution to virus-mediated type I IFN gene induction. The IFNA subtype promoter reporters (IFNA1, A2, A4, and A14) and the IFNB promoter reporter plasmids were transiently co-transfected to 2fTGH cells (which do not express endogenous IRF-5 or IRF-7) with IRF-5 V1, V2, or V3/V4 expression plasmids (Fig. 5B). Transfected cells were either left uninfected or infected with NDV. As shown in Fig. 6A, whereas ectopic expression of V1 stimulated low levels of each of the examined reporter activities (IFNA2 Ͼ IFNB Ͼ A4 Ͼ A14 Ͼ A1) in NDV-infected cells, transactivation was significantly lower than that observed for V3/4 proteins. Ectopic expression of V2 stimulated IFNB Ͼ A14 Ͼ A4 reporter activities but not IFNA1 or A2 after infection with NDV. The polypeptide encoded by V3/4 cDNAs stimulated the IFNB promoter reporter most effectively, whereas the IFNA promoters were stimulated less effectively.
By having shown that IRF-5 V1, V2, V3, and V4 enabled transactivation of type I IFN promoter reporters after infection with NDV, we next measured the ability of the other IRF-5 isoforms to induce IFNA1 or IFNB reporter gene activity. Each variant was co-transfected with the indicated IFN SAP reporter and infected with NDV. Of note, whereas V2 and V6 cDNA encode for the identical polypeptide ( Fig. 1B and Fig. 5, A and C), we have cloned V2 with the first exon (Ex1V2) deleted, whereas the first exon (Ex1V1) of V6 is present. The noncoding first exon in V6 may affect its translational efficiency (Fig. 5B). The activation of the IFNA1 reporter by the V3/4 polypeptide was used as a positive control; this reporter was also activated by V6 and V9 polypeptides and to a lesser degree by V5, V7, and V8 (Fig. 6B). In contrast, IFNB reporter gene activity was only stimulated by V3/4, V6, and to a lower extent by V7, after NDV infection (Fig. 6C). To evaluate further the role of IRF-5 isoforms in mediating type I IFN gene responses, we measured the synthesis of endogenous biologically active type I IFN in NDV-infected 2fTGH cells transiently expressing each isoform. Infection of 2fTGH cells with NDV resulted in only low levels of endogenous type I IFN (Fig. 7A). In contrast, ectopic expression of FLAG-tagged IRF-5 isoforms (V2, V3, V4, and V6) conferred on these cells the ability to induce type I IFN upon NDV infection, albeit low levels by V2 (Fig. 7A). Most importantly, only IRF-5 V3 and V4 induced measurable levels of synthesized IFN-␣ assayed in Madin-Darby bovine kidney cells, indicating that the low levels of type I IFN induced by V2 and V6 are most likely IFN-␤. We also measured the endogenous IFNA and IFNB transcript levels in these cells by semiquantitative RT-PCR. Fig. 7B shows the relative intensity of IFNA and IFNB transcripts induced by each isoform after normalization with ␤-actin levels. Again, measurable levels of IFNA transcripts could only be detected in NDV-infected 2fTGH cells expressing the IRF-5 V3/V4 polypeptide. Altogether, these data indicate that whereas IRF-5 V6 is an efficient inducer of IFNB in infected cells, IRF-5 V3 and V4 are the primary inducers of type I IFN (Figs. 5-7).
It has been shown previously the presence of alternatively spliced isoforms of IRF-1, IRF-3, and IRF-7 that occur at the level of RNA splicing (18 -19, 21, 42). In the case of IRF-3, an endogenous isoform lacking part of its DNA binding domain was isolated, termed IRF-3a (21)(22), and shown to act as a DN of IRF-3-mediated IFNB gene induction. As such, we examined whether the IRF-5 V7 would act in a similar manner when co-expressed with the IRF-5 V3 or V4 and the IFNB promoter reporter. Data shown in Fig. 7C revealed the IRF-5 V7, which is lacking its DNA binding domain, potently inhibited the virus-induced activation of the IFNB promoter reporter by IRF-5 V3 in a dose-dependent manner. A noted difference, in the absence of endogenous IRF-5 (2fTGH cells; Fig. 6, B and C), the V7 has low transactivation ability indicating that it may form functional heterodimers with IRF-3. The mechanism of this stimulation has yet to be determined. DISCUSSION The family of IRFs was originally identified as regulators and/or mediators of type I and II IFN signaling. IRF-3, IRF-5, and IRF-7 are central mediators controlling the expression of human type I IFNs (2). Whereas IRF-3 is constitutively expressed in all cell types, expression of IRF-5 and IRF-7 has been primarily detected in lymphoid cells and can be further stimulated by type I IFNs (2). It was shown that expression of IRF-3 is sufficient to support induction of IFNB, whereas IRF-5 or IRF-7 were sufficient for stimulation of IFNA gene expression in virus-infected cells (7,29). Consistent with these observations, PDCs express high constitutive levels of IRF-3, IRF-5, and IRF-7 and are thus uniquely preprogrammed to respond rapidly to a range of viral pathogens with high levels of IFN ␣/␤ production (26). IRF-5, like IRF-3 and IRF-7, requires phosphorylation of carboxyl-terminal serines for the nuclear localization and transactivation (8). Most interesting is the fact that unlike IRF-3 or IRF-7, IRF-5 is activated in a virus-specific manner (7), and not by double-stranded RNA, suggesting that IRF-5 may be activated by distinct signaling pathways leading to virus-mediated type I IFN gene expression (8). With the recent finding that IRF-5 exists as multiple isoforms that are expressed in a cell type-specific manner, and the fact that at least two alternative promoters were found to differentially regulate expression of these isoforms, it seems likely that they would have distinct functions. Data from Lin et al. (13), examining for the first time the dissimilar polypeptides encoded by V1 and V2 cDNAs, support this hypothesis. Since it had been shown recently that the IKK-related kinases TBK1 and IKK⑀ phosphorylate and activate IRF-3 and IRF-7 leading to the production of type I IFNs (43)(44), they examined the in vitro phosphorylation and activation of IRF-5 V1 and V2 by TBK1 and IKK⑀ and showed that although these two isoforms were indeed phosphorylated, phosphorylation did not lead to nuclear localization or activation (13). Whereas Lin et al. (13) did not examine localization/activation of these two isoforms by NDV, or their ability to induce type I IFNs after NDV infection, they concluded that IRF-5 might simply not be an activator of IFN gene expression. Important to note, Lin et al. (13) only examined the ability of Sendai virus to induce IRF-5-mediated transactivation of IFN promoter reporters even though it has been shown previously that Sendai virus is not an activator of IRF-5 (7,41).
Additional data from our lab and others indicate that the IRF-5 V3/4 and V5 polypeptides are targets of TBK1 and/or IKK⑀ by in vitro phosphorylation (14,41). In the case of V5, it was shown that phosphorylation by TBK1 or IKK⑀ led to IRF-5 nuclear translocation (14). Moreover, a critical role for the IRF-5 V3/4 polypeptide in TLR7/8-dependent type I IFN response was shown (41). In these studies, R848 stimulated the transactivation of both IRF-5 and IRF-7 but not IRF-3 and induced nuclear translocation of IRF-5 and IRF-7. The TLR7and TLR8-induced activation of IRF-5 V3/4 was inhibited in a dose-dependent manner by the TBK1K38A and IKK⑀K38A kinase-inactive mutants, suggesting that phosphorylation of IRF-5 V3/4 by TBK1 and IKK⑀ is functional. In support of these findings, a recent paper from Takaoka et al. (45) has shown an important role for IRF-5 in TLR signaling downstream of the TLR-MyD88 signaling pathway using spleen-derived dendritic cells and macrophages from wild type and IRF-5 Ϫ/Ϫ mice. Of particular interest, whereas the induction of interleukin-6, interleukin-12, and tumor necrosis factor-␣ in these cells was shown to be dependent on IRF-5 expression after stimulation with CpG ODN (TLR9), poly(I⅐C) (TLR3), lipopolysaccharide (TLR4), flagellin (TLR5), or poly(U) (TLR7/8), only CpG-induced IFN-␣ was shown to be independent of IRF-5 (45). Whether the induction of murine IFN-␣ by the other TLR ligands or virus is dependent on IRF-5 expression was not examined and thus remains a very interesting and highly relevant question.
The presence of multiple IRF-5 isoforms suggests that their functions may not be redundant; instead, these isoforms may have distinct expression, regulation, and/or function, as has been shown for the IRF-3 isoforms (21,22). Alternative splicing is a mechanism that allows for individual genes to express multiple mRNAs that encode proteins with diverse and even antagonistic function. This process generates segments of mRNA variability that can insert or remove amino acids, shift reading frame, or introduce a termination codon. It also affects gene expression by removing or inserting elements controlling translation, mRNA stability, and/or localization. To this effect, we demonstrate that the IRF-5 alternatively spliced isoforms are expressed in a cell type-specific manner and are differentially regulated by at least two alternative promoters. In addition, the subcellular localization of IRF-5 isoforms was dependent on alternative splicing events because the isoforms gave distinct localization patterns, which likely lead to alternate functions (Figs. 5D, 6, and 7). For example, analysis of the IRF-5 V7 (IRF-5 DN) function indicated that in the absence of endogenous or ectopically expressed IRF-5, i.e. 2fTGH cells (Fig. 6), V7 has a low level of transactivation ability. However, in cells that express IRF-5 V3, the IRF-5 V7 acts as a DN mutant and inhibits IRF-5 V3-mediated IFN induction. Most interestingly, because the IRF-5 V7 mRNA contains the noncoding exon 1 of V1 (Ex1V1), which is associated with P-V1, expression of this alternatively spliced isoform in virus-infected cells may provide an alternate mechanism for the virus to shut down type I IFN gene expression mediated by IRF-5 V3/4. We are currently examining this possibility. Altogether, these results imply that transcription of the IRF-5 isoforms is under multiple levels of control and that the alternatively spliced mRNAs encode for IRF-5 isoforms with potentially unique biological functions. A preliminary examination of the expression profile of mouse IRF-5 isoforms by gel electrophoresis of amplified exons 2-7 in RNA from peripheral blood lymphocytes of wild type C57Bl/6 mice or from the A20 murine B cell lymphoma cell line revealed a similar phenomenon of multiple banding patterns, indicating that murine IRF-5 may also exist as multiple isoforms (data not shown). A more detailed analysis is currently underway.
We have previously shown that the expression of IRF-5 is stimulated by IFN (8,10), and results from the present study demonstrate that this induction is isoform-specific (Figs. 1C, 3C, and 4). As summarized in Fig. 8, only the Ex1V3-specific transcripts were up-regulated by IFN-␣ stimulation or treatment with the TLR7/8 ligand R848. Although stimulation with either of these two inducers had no significant effect on Ex1-associated transcripts, stimulation with HSV-1 primarily up-regulated the Ex1V1-associated transcripts. In addition, NDV-infected Daudi or THP-1 cells specifically up-regulated Ex1V1 transcripts and not Ex1V2 or Ex1V3 (Fig. 4C). Infection of IRF-5 expressing cells with virus or treatment with R848 leads to the phosphorylation and translocation of IRF-5 to the nucleus (7)(8)41). All of these data correlate well with that obtained by promoter reporter assays (Fig. 8). However, in contrast to the data reported by Takaoka et al. (45), we were unable to detect up-regulation of the human IRF-5 isoforms by CpG stimulation. This may reflect differences in the amount and time of CpG ODN stimulation where Takaoka et al. (45) examined IRF-5 transcript levels after only 2 h of treatment at 1 M, and here we examined levels after 6 h at 40 g/ml. Although they observed an up-regulation of murine IRF-5 transcripts by CpG, they found that IRF-5 was not a mediator of CpG-induced IFN-␣ (45). TLR7 and TLR9 signaling are both MyD88-dependent, and are potent inducers of IFN, yet the ability of either R848 or CpG to up-regulate IRF-5 transcripts levels may not be a prerequisite for its involvement in either signaling pathway since unstimulated PBMC and PDC already express IRF-5. Both R848 and CpG are capable of inducing IRF-5 nuclear translocation (41,45). A distinct difference between these two signaling pathways could be in the mechanism of IRF-5 induction, activation, and/or the diversification of transcription factors activated downstream of MyD88/TRAF6/IRF-5-dependent signaling.
A further characterization of IRF-5 isoform expression in purified PBMC and PDC from healthy donors and immortalized tumor cell lines (THP-1, Namalwa, U937, and BCBL-1) led to the identification and cloning of five new IRF-5 isoforms, V5-V9 (Fig. 5A). These data revealed that human PDCs express only IRF-5 isoforms 1-4, whereas PBMC express, in addition, two new isoforms, V5 and V6. Moreover, IRF-5 V7, V8, and V9 were found to be expressed only in hematological malignancies. In comparing the function of these nine IRF-5 isoforms with regard to virus-mediated type I IFN gene expression, we found that the PDC-expressing isoforms did transactivate IFNA and IFNB gene promoters to varying degrees, but IRF-5 V3/4 was the most potent transcriptional activator of virus-induced IFNA and IFNB promoters (Fig. 6A). The IRF-5 V1, V6, and V9 also transactivated the IFNA1 reporter, whereas V2, V6, and V7 transactivated the IFNB reporter. Moreover, the synthesis of endogenous biologically active IFN-␣ in NDV-infected 2fTGH cells transiently transfected with each individual IRF-5 cDNA was only observed in cells transfected with either IRF-5 V3 or V4 cDNA, whereas IRF-5 V2 and V6 induced lower but detectable levels of IFN-␤. Finally, we and others have shown recently that IRF-5 is a component of the p53 pathway where IRF-5 transcripts were up-regulated by p53 overexpression or activation of p53 by DNA-damaging agents (10,12), yet IRF-5 expression can also be up-regulated by DNA-damaging agents in the absence of p53. 2  Constitutive IRF-5 isoform expression associated with Ex1V1 is maintained through the P-V1 IRF-E site; mutation of this site ablates the basal IRF-5 V1 promoter (P-V1) activity. Following viral infection, IRF-5 and other family members are phosphorylated and translocated to the nucleus, where they bind to the IRF-E site in P-V1, thereby increasing Ex1V1-associated transcript expression (V1 and V4 -V8). Consecutively, the phosphorylated IRF family members are also able to bind the IRF-E in IFNA gene promoters, increasing the amount of secreted IFN-␣. The endogenously produced IFN-␣ enters a positive feedback loop, mediating signaling through the ISGF3 complex by binding to the ISRE site in the IRF-5 V3 promoter (P-V3) and other ISREcontaining promoters, thereby up-regulating expression of the IRF-5 V3 transcripts. In addition, the TLR7/8 ligand R848 induces IRF-5 phosphorylation (41) and up-regulates V3 transcripts. Altogether, the findings presented here indicate that IRF-5 transcription is regulated as follows: (i) by virus; (ii) by members of the IRF family of transcription factors, IRF-1 and IRF-5 V3; and (iii) by IFNs via the ISGF3 complex.
transcripts. 3 The factors that are involved in the stimulation of IRF-5 expression in response to stress have yet to be identified. Although the molecular mechanism of IRF-5 induction or activation in uninfected cells is not known, we have identified a number of IRF-5-specific target genes in uninfected cells, such as the cell-cycle regulator p21, pro-apoptotic genes Bax, Bak, and caspase 8 (10). These genes are regulated specifically by the V3-and V4-encoded polypeptides. Currently, it is not clear whether these genes would also be regulated by the polypeptides encoded by V1 or V2 cDNAs. The absence of IRF-5 general transcripts in a number of primary leukemias and lymphomas (10) suggests that IRF-5 may act as tumor suppressor gene in which its function is deleted in certain cancers. Because expression of the IRF-5 V7, V8, and V9 mRNA transcripts has only been observed to date in cancer cell lines, another potential role for the IRF-5 DN (V7) may be to inhibit the cell cycle regulatory, growth inhibitory, and/or pro-apoptotic effects of IRF-5 V3/4. Whether V8 and V9 also have distinct functions specific to cancer cell growth is currently unknown. Altogether, these data suggest that further analysis of the molecular mechanism(s) regulating expression, induction, and biologic function of the IRF-5 isoforms is likely to reveal potential new biomarkers for cancer detection and new immunomodulatory strategies for inflammation-associated disorders.