Regulation of IFN-λ1 Promoter Activity (IFN-λ1/IL-29) in Human Airway Epithelial Cells

The type III (λ) IFNs 1, 2, and 3 (IFN-λ1, IFN-λ2, and IFN-λ3), also known as IL-29, IL-28A, and IL-28B, respectively, are the most recently defined IFN ligands in humans (1, 2). In addition to antiviral properties, this class of IFNs has important but poorly characterized functions in innate and adaptive immunity. Type III IFNs are four-helical-bundle cytokines that are distantly related to both the IL-10 family and the type I IFN family (3). In humans, the three IFN-λ proteins are encoded by genes on chromosome 19q13; however, IFN-λ1 is not functionally conserved in mice. Although all three gene products share a high degree of amino acid similarity, the IFN-λ2 and IFN-λ3 proteins are >95% identical and may have arisen from a recent gene-duplication event (1–4).

Studies of IFN-λ gene regulation have focused on comparing it with that of the type I IFN genes (5, 6). Onoguchi et al. (5) demonstrated dependence on NF-κB and IRF sites within the first 300 bases of the IFN-λ1 promoter following infection of transformed cells with Newcastle disease virus. Osterlund et al. (6) extended these findings to suggest that IFN-λ1, like IFN-β, is regulated by IRF-3, whereas IFN-λ2/3 are more dependent upon IRF7, similar to IFN-α. In addition, a recent study using bacterial infection of human monocyte-derived dendritic cells (MDDC) showed that, in this case, IFN-λ1 activation was more dependent on distal NF-κB sites within the promoter (−1137, −1182 bp) (7). These data suggested a regulation that may be both cell type and stimulation specific, perhaps necessary to maintain efficient control over IFN-λ expression and its role in the immune response.

The tightly targeted nature of the IFN-λ response is distinct from the more ubiquitous secretion and response to type I IFNs. Expression of IFN-λ is restricted mainly to plasmacytoid dendritic cells and epithelial cells (8, 9). Furthermore, expression of the IFN-λ receptor (in which both chains are distinct from those used in either the type I or type II IFNRs) is almost completely restricted to immune cells and epithelial cells (9–12). Thus, the restricted expression pattern of the IFN-λ ligands and their receptor confers tight control on IFN-λ responsiveness, in contrast to the situation with the ubiquitous type I IFNs.

Although the type I and type III IFNs use distinct receptor complexes, their downstream-signaling events are very similar. Type III IFNs primarily trigger STAT phosphorylation; pSTAT1 homodimers and pSTAT1/2 heterodimers are both formed, and target-gene promoter γ IFN activation site (GAS) element and IFN-sensitive response element (ISRE) are both activated (1). IFN-stimulated genes, such as myxovirus resistance protein (MX1) and 2′-5′-oligoadenylate synthetase (OAS), are IFN-λ responsive (1, 13–15); consequently, IFN-λs induce biological activities, including antiviral, antiproliferative, and proapoptotic effects that are similar to those of the type I IFNs (16). The antiviral properties of the three IFN-λs vary (17) but are generally held to be weaker than those of the type I ligands (18).

The IFN-λs were recently shown to have immunoregulatory functions (19, 20), particularly an ability to downregulate human Th2 responses. These studies showed that IFN-λ1 acts to limit Th2 polarization and the secretion of the Th2 cytokines IL-4, IL-5, and IL-13. These effects can be direct on CD4⁺ T cells or indirect, via MDDC. Further, IFN-λ1 secretion is itself Th2 cytokine responsive (11, 12, 21–23). Recent studies also showed that IFN-λ1 expression is reduced in alveolar macrophages and bronchial epithelial cells of rhinovirus-infected asthmatic individuals (24, 25). Two key contributing pathologies in asthma are the upregulation of Th2 responses and remodeling of the airway epithelial cells (26). Thus, the observed lack of a key regulator (IFN-λ1) may be critical in asthmatic individuals, because it normally would be produced primarily by plasmacytoid dendritic cells and epithelial cells to both promote viral responsiveness and limit Th2 responses (8–11).

Studies of promoter function continue to shed light on the role of immune-system components in health and disease (27–30). To begin to explore the nature of IFN-λ1 production by airway epithelial cells, we characterized the regulation of IFN-λ1 transcription in a model of human airway viral infection.

Four kilobases 5′ of the IFN-λ1 translation start site (TSS) were PCR amplified, sequence-verified, and cloned upstream of the firefly luciferase gene in the pGL4.10 reporter vector (Promega, Madison WI). Truncations (5′) were generated by restriction digestion using conventional methods. Sixteen hours prior to transfection, 2.0 × 10⁴ BEAS-2B cells (American Type Culture Collection catalog no. CRL-9609) were plated in LHC-9 culture medium. Immediately prior to transfection, this was changed to serum-free RPMI 1640 medium. Reporter-construct DNA was mixed with Opti-MEM and Lipofectamine and applied to the cells. Six hours after transfection, the medium was replaced with LHC-9; 18 h thereafter, polyinosinic-polycytidylic acid (poly I:C) was added to a final concentration of 50 μg/ml. Firefly luciferase activity was measured using the Dual-Glo luciferase assay kit (Promega) at the indicated time points. The Renilla luciferase-containing pGL4.74 vector was used for normalization, and the pGL4.14 vector containing an SV40 promoter-driven firefly luciferase gene was included as a positive control. All media and transfection reagents were obtained from Invitrogen (Carlsbad, CA).

BEAS-2B cells grown in LHC-9 medium were stimulated with poly I:C for various lengths of time. The chromatin immunoprecipitation (ChIP) assays were performed using the E-Z ChIP kit (Millipore, Billerica, MA), according to the manufacturer’s instructions with minor modifications. A total of 0.2 × 10⁶ cells per immunoprecipitation was sonicated at 50% power for four 30-s bursts. Binding site occupancy was monitored by SYBR Green-based quantitative PCR, according to conventional methods. Primers were designed to amplify specific regions of interest. All ChIP assay data presented were calculated from triplicate evaluations of two or three replicate experiments. Binding was considered relevant if it achieved ≥2-fold enrichment of specific Ab: IgG.

BEAS-2B cells were cultured in LHC-9 medium. Cells were pretreated for 1 h with Bay11-7082 (Sigma, St. Louis, MO), and poly I:C stimulation was conducted at 50 μg/ml.

SmartPool small interfering RNA (siRNA) to target NF-κB1 (p50), RELA (p65), ZEB1, B lymphocyte-induced maturation protein 1 (BLIMP-1), GAPDH, or control nontargeting (NT) siRNA were purchased from Thermo-Scientific. These were transfected into BEAS-2B or A549 cells using Lipofectamine 2000 (Invitrogen), according to the manufacturer’s instructions. siGLO (Thermo-Scientific) was used to optimize transfection efficiency (87%; data not shown). At 24 h posttransfection, cells were reseeded at 2.0 × 10⁵ cells/ml, and poly I:C stimulation or viral infection was initiated 12 h after reseeding. Supernatants, protein from whole-cell extracts, and total RNA were harvested following 0, 3, 4.5, 8, 24, or 32 h of poly I:C stimulation or human rhinovirus 1B (RV1B) infection. RV1B (American Type Culture Collection) infection was performed at a multiplicity of infection of 1, where the virus was incubated with the cells grown in RPMI 1640 containing 2% FBS. Quantitative RT-PCR (qRT-PCR) was then used to analyze the mRNA levels of genes of interest, using hypoxanthine phosphoribosyltransferase as the normalizer. For all qRT-PCR experiments, the data represent normalized fold changes calculated using the efficiency-calibration method (31). The IFN-λ1 ELISA was performed using the Duo Set ELISA (R&D Systems) or the Ready-Set-Go ELISA (E-Bioscience) kit, according to the manufacturers’ protocols (R&D Systems).

Nuclear lysates were prepared from poly I:C-stimulated BEAS-2B cells over the indicated time course and then incubated with a Cy5-labeled DNA probe for 30 min at room temperature. The reactions were formaldehyde cross-linked and purified using a Qiagen (Valencia, VA) reaction clean-up kit. DNase 1 digestion (New England Biolabs, Ipswich, MA) was performed for 10 min at 37°C. The reactions were then reverse cross-linked, purified, and analyzed using a CEQ genetic analyzer (Beckman Coulter, Fullerton, CA).

The TESS (http://www.cbil.upenn.edu/cgi-bin/tess/tess) and Genomatix (http://www.genomatix.de) Web sites were used.

To identify key responsive IFN-λ1 promoter regions, firefly luciferase reporter constructs were generated containing up to 4 kb from the IFN-λ1 TSS (i.e., to −4 kb 5′ of the TSS; Fig. 1A). Time course experiments using BEAS-2B airway epithelial cells showed that the peak of endogenous IFN-λ1 expression in response to the TLR3 agonist, poly I:C, occurred at 3 h (Fig. 1B); luciferase activity of the full-length reporter construct also occurred at 3 h (data not shown); this time point was selected for further experiments. To identify regulatory regions involved in IFN-λ1 expression in response to the poly I:C model of viral infection, truncations of the full-length construct were made as depicted (Fig. 1A). BEAS-2B cells were transfected with these constructs and then stimulated for 3 h with poly I:C (Fig. 1C). Typically, a 15–20-fold activation was observed with the full-length construct, and this was also true of truncations at −3.3, −2.0, −1.7, and −1.5 kb from the TSS. All of the promoter constructs from 4.0–1.2 kb showed the same timing of peak poly I:C responsiveness. Interestingly, the degree of luciferase activity was markedly increased in the −1.2-kb IFN-λ1 promoter construct, indicating the presence of a powerful repressive region between −1.5 and −1.2 kb. Responsiveness to poly I:C was completely lost in the −0.6-kb construct, suggesting that critical activation elements lay between 0.6 and 1.2 kb upstream of the TSS. These data also suggested that previously reported proximal regulatory elements within the first 0.6 kb 5′ of the TSS (5, 6) are insufficient for activation of this gene in human airway epithelial cells.

Three NF-κB binding sites were predicted within the potential activating region (i.e., between −1.2 and −0.6 kb) of the IFN-λ1 promoter, at −1183, −1129, and −1016 bp. These more distal NF-κB binding sites [and not the previously suggested proximal ones (5, 6)] were shown to be relevant for LPS induction of IFN-λ1 expression (7).

To clarify the dependence of poly I:C induction of IFN-λ1 on NF-κB, we first made use of BAY11-7082, an inhibitor of IκBα degradation, which consequently prevents translocation of NF-κB to the nucleus. Induction of endogenous IFN-λ1 expression was reduced in the presence of this inhibitor (Fig. 2A). Similarly, induction of the −1.2-kb luciferase reporter was also reduced in the presence of this compound (data not shown). We then confirmed the involvement of NF-κB using ChIP assays for the p50, p52, p65, and c-REL NF-κB family members. Potential binding of these transcription factors to the endogenous IFN-λ1 promoter of BEAS-2B cells is shown in Fig. 2 at various time points (between 0 and 270 min of poly I:C activation) at the distal and proximal regions, using primer pairs F6/R6 and F3/R3 (Fig. 1A), respectively.

Binding of p50 was observed at 0 and 270 min of poly I:C stimulation only, to both the proximal and distal predicted NF-κB sites (Fig. 2B). In contrast, c-REL and p65 bound the promoter most strongly at the 90-min time point (Fig. 2C, 2D, respectively), with c-REL binding both proximal and distal regions but p65 binding distal regions only. Binding of p52 was observed at 90 min to the proximal sites only (Fig. 2E).

NF-κB family members are known to activate gene expression as heterodimers, whereas p50 homodimers in the nucleus are typically repressive (32, 33). By comparing Fig. 1B with Fig. 2B–D, we can see that p50 binding was maximal at times when IFN-λ1 mRNA expression was not yet evident (0 min) or postpeak and being reduced (270 min). In contrast, binding of p52, p65, and c-REL was maximal at 90 min, when transcription was accelerating. These data are consistent with the suggestion that, upon poly I:C activation, p50 homodimers are replaced by p65/c-REL heterodimers at the distal promoter and c-REL/p52 heterodimers at the proximal promoter, which, in turn, are replaced by p50 homodimers as transcription declines and the promoter resets.

To further investigate p50 as a repressor of IFN-λ1, we performed siRNA to specifically target either p50 or p65. BEAS-2B cells with targeted knockdown of p50 showed enhanced IFN-λ1 mRNA levels in response to poly I:C at 3, 6, 24, and 32 h (Fig. 3A). The increase in IFN-λ1 message correlated with enhanced protein secretion throughout the time course (Fig. 3B). Conversely, p65 knockdown led to a decrease in IFN-λ1 mRNA levels in response to poly I:C stimulation at the same time points (Fig. 3A), as well as a complete inability to secrete IFN-λ1 protein, as detected by ELISA (Fig. 3B).

The luciferase reporter construct data (Fig. 1C) had demonstrated the presence of a powerful transcription-repressing region between −1.5 and −1.2 kb in the IFN-λ1 promoter. TESS and Genomatix analysis identified E-box–like sites (−1431 and −1321 bp) and a Z-box–like site (−1471 bp) as candidate binding sites for the zinc-finger transcription factor ZEB1/AREB6/TCF8, known to have repressor function in epithelial cells (34).

We created a series of microdeletion reporter constructs to test each of these potential sites independently (Fig. 4A). Transfection and poly I:C activation clearly showed that deletion of the most proximal (Z-box) site at −1471 bp completely recapitulated the expression phenotype of the highly active −1.2-kb construct (Fig. 4B), demonstrating that this element was essential to the repressive nature of the −1.5 to −1.2-kb region. In addition, ChIP analysis of this region within the BEAS-2B endogenous IFN-λ1 promoter (primer set F7/R7, Fig. 1A) demonstrated binding of ZEB1 at the 0-, 225-, and 270-min time points (Fig. 4C). This promoter occupancy by ZEB1 corresponded to times when transcription was not yet initiated (0 min; Fig. 1B) or had peaked and was undergoing repression (225/270 min). Taken together, our data from luciferase and ChIP experiments shown in Fig. 4B and 4C demonstrated that the repressive element observed between −1.5 and −1.2 kb is most probably mediated by ZEB1 binding to the Z-box site at −1471 bp.

One distal (−3945 bp) and one proximal (−120 bp) element were predicted to contain overlapping positive regulatory domain 1/ISRE sites. Such sites would confer the ability to bind BLIMP-1 and IRF, competitively (35). BLIMP-1 is a transcriptional repressor, whereas IRF family members are typically transcriptional activators. To determine whether this may be the case, we examined the distal region for evidence of transcription factor binding by DNase 1 footprinting.

Nuclear extracts from BEAS-2B cells were prepared over a time course of poly I:C stimulation. A Cy5-labeled probe (−4005 to −3633 bp from the TSS) was used to monitor protection from DNase 1 digestion. Protection was evident throughout the time course (Fig. 5), but the precise sequence involved varied with the time of stimulation, suggesting a dynamic pattern of transcription factor binding that would be consistent with occupation of this overlapping PRDI/ISRE site by different proteins.

Therefore, we examined both the distal and proximal sites for binding of BLIMP-1 and IRF1 by ChIP (using primer sets F4/R4 and Fd/Rd, respectively, Fig. 1A) and compared the binding to the levels of IFN-λ1 mRNA as before. Binding of IRF1 became apparent at 90 min of poly I:C stimulation at the distal (−3945 bp) site; this appeared to increase at 135 min, before declining to background level by 225 min. Interestingly, IRF-1 also bound at the proximal (−120 bp) site at 135 min (Fig. 6A). In stark contrast, BLIMP-1 bound only at 270 min, coincident with resetting of transcription levels to baseline (Fig. 6B). Although BLIMP-1 bound markedly at the distal site, >2-fold enrichment was also observed at the proximal site, reflecting the relative binding of IRF-1 at these two sites. A third predicted site (−250 bp, primers Fb/Rb) failed to show binding of either transcription factor. These data suggested that IRF-1 plays a key role in progressing IFN-λ1 transcription, whereas BLIMP-1 uses the same two sites to conclude the expression of IFN-λ1.

ZEB1 and BLIMP-1 appear to be important repressors of IFN-λ1 expression, acting at complementary times during the response to poly I:C stimulation. Transfection of specific SmartPool siRNA into BEAS-2B cells reduced ZEB1 mRNA levels by 40% and BLIMP-1 by 30%, at 48 h after transfection. This was reflected in a 71% reduction in ZEB1 protein level and a 53% reduction in BLIMP-1 protein, by Western blotting at 72 h posttransfection (Fig. 7A). Importantly, BLIMP-1 knockdown did not lead to alterations in ZEB1 mRNA levels and vice versa (Fig. 7A).

Next, we sought to determine the extent to which siRNA knockdown of ZEB1 and BLIMP-1 would alter IFN-λ1 expression kinetics. As shown in Fig. 7B, poly I:C stimulation of ZEB1 or BLIMP-1 siRNA-targeted BEAS-2B cells did not result in an overall higher level of IFN-λ1 mRNA at the previously determined peak of transcription, 3 h. However, withdrawal of either repressive transcription factor allowed active transcription to be extended over a longer time period; at 4.5 h (by which time transcription was expected to have ceased), ZEB1 knockdown cells had a 3.7-fold elevated (p = 0.001) transcript level, and BLIMP-1 knockdown cells showed a 3-fold increase (p = 0.0004). Measurement of IFN-λ1 protein levels by ELISA (Fig. 7C) showed that secretion by ZEB1 or BLIMP-1 knockdown cells was almost double that of nontargeted cells at 8 h; this difference was maintained out to ≥32 h.

We then sought to confirm the regulation of IFN-λ1 by BLIMP-1 and ZEB1 in an additional airway epithelial cell line. For these experiments, A459 airway epithelial cells were transfected with ZEB1 or BLIMP-1 siRNA and then stimulated with poly I:C over time. Targeted knockdown of ZEB1 or BLIMP-1 in these cells led to an earlier activation of the IFN-λ1 gene by poly I:C (Fig. 7D). These data again demonstrated that ZEB1 and BLIMP-1 are key repressors of IFN-λ1 transcription in human airway epithelial cells.

Previous reports on IFN-λ regulation suggested that their regulation is very similar to that observed for the type I IFNs (IFN-α and IFN-β) (5, 6). Indeed, it is known that BLIMP-1 is a repressor of IFN-β (36, 37). We first confirmed the regulatory effect of BLIMP-1 in poly I:C-induced IFN-β mRNA expression. siRNA-targeted BLIMP-1 BEAS-2B airway epithelial cells showed a 2-fold increase in IFN-β mRNA levels, significant at both the 3-h (p = 0.021) and 4.5-h (p = 0.031) time points. However, of great interest, ZEB1 knockdown did not alter IFN-β transcription at all, strongly suggesting that ZEB1 regulation is unique to the type III IFNs (Fig. 8A).

To address this latter question, we examined the mRNA levels of IFN-λ2/3 following ZEB1 knockdown. In concordance with our IFN-1λ data, mRNA levels were higher when ZEB1 levels were reduced; elevation was observed at both 3- and 4.5-h time points (p = 0.02 and p = 0.002, respectively). BLIMP-1 knockdown also resulted in elevated IFN-λ2/3 levels at both time points (p = 0.01 and p = 0.051, respectively, Fig. 8B). Thus, ZEB1 and BLIMP-1 appear to be important in the regulation of all type III IFNs, but ZEB1 may not regulate the type I IFNs.

Viral infections are known to exacerbate asthma in certain individuals (38, 39). Interestingly, viral infection of bronchial epithelial cells may preferentially induce type III IFNs over type I IFNs (8). However, asthmatic individuals make less type III IFN than do otherwise healthy people, following viral infection or challenge (24). Thus, it was of interest to determine whether, given the elevation of IFN-λ1 mRNA and protein levels in targeted knockdown BEAS-2B cells, there was any evidence that antiviral cellular defense mechanisms would also be elevated.

We examined the expression of mRNA for two IFN-λ1–induced genes (1), MX1 and OAS, in the targeted knockdown BEAS-2B cells. In nontargeted cells, MX1 expression peaked at 8 h, with an ∼100-fold increase over baseline; it then declined to ∼40-fold elevation by 32 h. Targeted knockdown of ZEB1 resulted in the peak levels of MX1 transcription being sustained out to 32 h, whereas no effect was observed on MX1 expression in the BLIMP-1 knockdown cells (Fig. 9A). Expression of OAS in nontargeted cells followed a different course from MX1; under poly I:C stimulation, OAS expression increased steadily over the 32-h time period, to an ∼40-fold increase. Knockdown of ZEB1 caused this to be elevated to ∼100-fold at 24 h, with only a slight diminution (80-fold elevation) at 32 h. BLIMP-1 knockdown cells were also affected, with an earlier increase at 4.5 and 8 h (from 20- to 70-fold, approximately), which had normalized by 32 h (Fig. 9B). We speculated that these increases in levels of cellular antiviral transcripts may result from the elevated IFN-λ1 secretion observed in the targeted knockdown cells.

We finally sought to confirm that ZEB1 and BLIMP-1 are negative regulators of IFN-λ1 in response to infection with live virus. For these experiments, we transfected BEAS-2B cells with siRNA for ZEB1 or BLIMP-1 and then infected the cells with RV1B to determine the effect of siRNA knockdown on IFN-λ1 mRNA levels. In the nontargeted cells, up to a 19-fold increase in IFN-λ1 mRNA was observed by 32 h postinfection. Cells that were transfected with either ZEB1 or BLIMP-1 siRNA showed a much more robust induction of IFN-λ1 in response to RV1B. Targeted knockdown of ZEB1 resulted in a 150-fold increase in IFN-λ1 mRNA by 32 h post-RV1B infection, and knockdown of BLIMP-1 resulted in a 350-fold increase of IFN-λ1 mRNA (Fig. 10).

The type III IFN family comprises IFN-λ1, IFN-λ2, and IFN-λ3. These ligands were reported to exhibit diminished antiviral activity relative to the type I ligands. However, our laboratory recently characterized interesting immunomodulatory aspects to type III IFN function that have clear potential relevance to asthma (9, 11, 12, 19, 20, 22). Because epithelial and immune cells both produce and respond to IFN-λ (9–11), it is likely to play an important role in the immune/epithelial interface. Aberrant interactions in this location are particularly important in the asthmatic airway (26), a condition in which IFN-λ1 may be deficient (24, 25). The IFN-λ1 that is produced by airway epithelial cells is critical for mounting efficient antiviral responses and, in parallel, moderating Th2 development; therefore, we characterized the regulation of this gene’s expression in this context. In this article, we have provided a description of key transcriptional activators and repressors of the natively configured IFN-λ1 promoter, using BEAS-2B cells stimulated with poly I:C (a TLR3 agonist) as a model system for viral infection or with live RV1B.

An important result of our study is the uncovering of the role of the transcriptional repressor ZEB1. This is a key feature of the regulation of IFN-λ1 transcription and one that appears to set it apart from type I IFNs. This zinc-finger homeodomain transcription factor is known to play a key role in epithelial cells [e.g., regulating epithelial to mesenchymal transition (40)]. The greatly enhanced reporter gene activity in the construct lacking potential E-box and Z-box domains (contained within the −1.5 to −1.2-kb region, Fig. 1) initially highlighted the critical repressive nature of this region. Analysis of ZEB1 binding by ChIP showed a loss of enrichment/binding at 90 and 125 min (Fig. 4C), when IFN-λ1 mRNA levels showed the greatest increase (Fig. 1B); the Z-box element at −1471 bp was shown to be critical for this repressive activity. The strong binding of ZEB1 in the absence of stimulation, as well as its rapid return, suggested that this factor is key to holding the promoter in the resting state and that defective ZEB1 relief may compromise IFN-λ1 expression. In contrast, targeted ZEB1 knockdown increased the duration of transcription and allowed elevated levels of IFN-λ1 protein to be produced (Fig. 7), a factor that may have contributed to the elevated antiviral responses seen under these conditions (Fig. 9). Furthermore, we were intrigued to observe that ZEB1 modulated transcription of all three IFN-λ genes yet did not alter transcription of IFN-β (Fig. 8). These new aspects of ZEB1 activity constitute an important and novel contrast between type III and type I IFN regulation.

NF-κB p50 homodimer also appears to be critical to maintaining the resting state of the promoter in airway epithelium, because it was enriched/bound only when transcription was not occurring and no IFN-λ1 mRNA was detectable (Fig. 2B); two sites, defined by our primer sets 3 and 6 (Fig. 1A), appear equally important. Interestingly, p50 is replaced in these regions by different c-REL–containing heterodimers when transcription is active (Fig. 2C, 2D). Our data suggested that the role of the p65/c-REL heterodimer at the more distal site (primer set 6) may be more important for gene activation than is that of the p52/c-REL heterodimer (primer set 3, more proximal to the TSS). Specific dependence upon c-REL has been noted for the transcription of several cytokines, including IL-2, IL-12p40, IL-23p19, and possibly IL-27p28 (41–45). It is well established that NF-κB p50 homodimers are repressive (33, 46, 47), and this may also be the case in this study. Thus, our data present a picture where NF-κB p50 is present on the resting promoter and is replaced upon stimulation by the p65/c-REL and p52/c-REL heterodimers, which support transcription until they themselves are replaced by p50 homodimer, contributing to the reset of the promoter back to resting.

Our findings emphasize the role of moderately distal (primer set 6) NF-κB sites. These data concur with a recent study examining activation by LPS in which p65 was involved in activating IFN-λ1 transcription from these distal NF-κB sites (7) and are in broad consensus with two previous studies, which focused on more proximal sites within the first 300 bp upstream of the TSS (5, 6). Our −0.6-kb reporter construct was not responsive to poly I:C (Fig. 1C), but the endogenous promoter did show binding of both IRF1 and NF-κB proteins (by ChIP) in this proximal region, suggesting that this −0.6-kb region contributes to poly I:C-induced IFN-λ1 transcription but is not sufficient to support this process by itself. Although a number of factors may account for these observed differences, one important element may be the use of human MDDC by previous investigators compared with our use of airway epithelial cells. Minor differences in reporter plasmid construction may also have relevance. It may be of note that our data (not shown) also failed to demonstrate a role for IRF-7, considered important in earlier work with transcription factor ectopic overexpression (6).

However, IRF1 did show enrichment/binding at two sites, at times consistent with an important role in activation and maintenance of transcription (Fig. 6A). As predicted from our DNase footprinting analysis (Fig. 5) and previous reports (35), these same sites were also occupied by BLIMP-1, at nonoverlapping times. (Fig. 6B). BLIMP-1 was initially characterized as a regulator of the antiviral response through repression of IFN-β (36, 37), and (unlike ZEB1) it was also repressive of this gene in our system (Fig. 8A). BLIMP-1 binds both distal and proximal PRDI elements (Fig. 6B; Fig. 1A, primer sets 4 and d), and the data that we present in this article are consistent with a role for BLIMP-1 in the postactivation rerepression of IFN-λ1. In T cells, BLIMP-1 can repress expression of critical Th1 proteins, including IFN-γ (48). In this article, we show that it also represses IFN-λ1 in epithelial cells, thereby controlling the expression of this key Th2 regulator.

In conclusion, to our knowledge, this is the first study to examine the regulation of the human IFN-λ1 promoter in its native state and in a model of viral infection of human airway epithelium. Our study revealed the key role of NF-κB p50 homodimer in the resting state and the hitherto unsuspected role of ZEB1 in this context. IRF-1, p52/c-REL, and p65/c-REL transcription factors are closely associated with active transcription, whereas ZEB1, p50 homodimer, and BLIMP-1 all contribute to repression and resetting.

We thank Drs. Raymond Yu and Joseph Nickels for helpful discussions.

The authors are full-time employees of HUMIGEN.