Physiological and receptor-selective retinoids modulate interferon gamma signaling by increasing the expression, nuclear localization, and functional activity of interferon regulatory factor-1.

Synergistic actions between all-trans-retinoic acid (atRA) and interferon gamma (IFNgamma) on modulation of cellular functions have been reported both in vitro and in vivo. However, the mechanism of atRA-mediated regulation of IFNgamma signaling is poorly understood. In this study, we have used the human lung epithelial cell line A549 to examine the effect of atRA on IFNgamma-induced expression of IFN regulatory factor-1 (IRF-1), an important transcription factor involved in cell growth and apoptosis, differentiation, and antiviral and antibacterial immune responses. At least 4 h of pretreatment with atRA followed by suboptimal concentrations of IFNgamma induced a faster, higher, and more stable expression of IRF-1 than IFNgamma alone. Actinomycin D completely blocked the induction of IRF-1 by the combination, suggesting regulation at the transcriptional level. Further, we found that activation of signal transducer and activator of transcription-1 was induced more dramatically by atRA and IFNgamma than by IFNgamma alone. Expression of IFNgamma receptor-1 on the cell surface was also increased upon atRA pretreatment. Experiments using receptor-selective retinoids revealed that ligands for retinoic acid receptor-alpha (RARalpha), including atRA, 9-cis-retinoic acid, and Am580, sequentially increased the levels of IFNgamma receptor-1, activated signal transducer and activator of transcription-1, and IRF-1 and that an RARalpha antagonist was able to inhibit the effects of atRA and Am580. In addition, atRA pretreatment affected the transcriptional functions of IFNgamma-induced IRF-1, increasing its nuclear localization and DNA binding activity as well as the transcript levels of IRF-1 target genes. These results suggest that atRA, an RARalpha ligand, regulates IFNgamma-induced IRF-1 by affecting multiple components of the IFNgamma signaling pathway, from the plasma membrane to the nuclear transcription factors.

through retinoic acid receptors (RARs) to transcriptionally activate target genes (4). Interferons (IFNs), on the other hand, signal through interferon signaling pathways and use signal transducer and activator of transcription (STAT) proteins as transcription factors (5). Recently, more attention has been drawn to potential interactions between these two pathways, since synergistic effects of atRA and IFNs have been observed in both animal models (6,7) and cultured cell lines (8 -10). However, the molecular events involved in atRA-dependent regulation of IFN␥ signaling or vice versa are poorly understood.
IFN regulatory factor-1 (IRF-1) was discovered in studies of virusinduced IFN␣/␤ gene regulation and IFN-mediated antiviral responses (11). IFN␥ is one of the strongest inducers of IRF-1. Upon binding of IFN␥ to its receptor, IRF-1 is induced through activation of STAT-1 and binding of activated STAT-1 to the ␥-interferon-activated site within the IRF-1 promoter. IRF-1 protein, a transcription factor itself, has a very short half-life and is degraded via the ubiquitin-proteasome pathway (12). IRF-1 is known to regulate cell growth and apoptosis (13,14). The tumor suppressor function of IRF-1 was first demonstrated in oncogenic transformation of primary IRF-1 Ϫ/Ϫ mouse embryonic fibroblasts (15). Recently, it was also shown that IRF-1 reversed the transformed phenotype of tumor cells not only in vitro but also in vivo (16).
RA synergizes with IFN␥ to increase the level of IRF-1, which subsequently activates promoters of IRF-1 target genes, such as tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) (17). Caspase-1 (previously termed interleukin (IL)-1␤-converting enzyme, or ICE) is another target gene of IRF-1 (18), which plays a critical role in the production of proinflammatory cytokines IL-1␣, IL-1␤, IL-18, and IFN␥ (19). However, the molecular basis of IRF-1 regulation by the combination is not fully understood. atRA by itself activates IRF-1 gene expression in NB4 promyelocytic leukemia cells (20). It induces IRF-1 through a ␥-interferon-activated site motif, but not a putative retinoic acid response element, of the IRF-1 promoter (21), indicating a role of STAT-1 between atRA and IRF-1. However, IRF-1 induction by atRA in squamous carcinoma cells is STAT-1-independent (22). In cervical squamous carcinoma SiHa cells, STAT-1 dependence was observed only when high doses of atRA (1 M and above) were used (23). Collectively, it is still unclear how atRA regulates IRF-1 transcription and whether STAT-1 is involved. RARs, common regulators of atRA functions, may also play important roles in the regulation of IRF-1.
In this study, we hypothesized that atRA increased IFN␥-induced expression and functions of IRF-1 through, at least in part, its regulation of the key molecules of IFN␥ signaling pathways. Using a human lung epithelial cell line, A549, we have observed that overnight pretreatment with atRA increases the levels of IFN␥ receptor (IFNGR)-1 on the cell surface, thereby enhancing tyrosine phosphorylation (activation) of STAT-1 upon low-dose IFN␥ stimulation. Faster, higher, and more stable levels of IRF-1 are then induced by the combination of atRA and IFN␥ compared with IFN␥ alone. RAR␣ mediates the effect of atRA in increasing cell surface IFNGR-1, activated STAT-1, and IRF-1. atRA pretreatment also potentiates the transcriptional activity of IFN␥-induced IRF-1, increasing its nuclear localization and DNA binding activity. As anticipated, IFN␥-induced caspase-1 and TRAIL are further increased by pretreatment with atRA. Collectively, these results indicate that atRA regulates multiple components of the IFN␥ IRF-1 pathway, from the plasma membrane to the nuclear transcription factors.
Western Blot Analysis-Whole cell lysates (25 g) or nuclear extract (15 g) was denatured and separated by polyacrylamide gel electrophoresis. After separation, proteins were electrophonically transferred to nitrocellulose membranes, which were then sequentially incubated in primary antibody and horseradish peroxidase-conjugated secondary antibody (24). Detection of the horseradish peroxidase conjugate was done using the ECL system (Pierce). For equal loading controls, the membranes were blotted with an anti-␤-actin antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or stained by Ponceau S.
RT-PCR-Total cellular RNA was isolated using a Qiagen RNeasy kit (Qiagen Inc., Valencia, CA) according to the manufacturer's instructions. Total RNA (0.5 g) was subjected to reverse transcription, and one-tenth of the reaction mixture was used for PCR analysis. A pair of primers was designed to detect differential expression of IRF-1 mRNAs: 5Ј-GGC TGG GAC ATC AAC AAG GAT G-3Ј (forward) and 5Ј-GAG CTG CTG AGT CCA TCA GAG AA-3Ј (reverse), amplicon size 330 bp. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was employed as an internal control: 5Ј-TGA AGG TCG GAG TCA ACG GAT TTG GT-3Ј (forward) and 5Ј-CAT GTG GGC CAT GAG GTC CAC CAC-3Ј (reverse), amplicon size 980 bp. Primer sequences for IRF-1 target genes were as follows: TRAIL, 5Ј-TGC GTG CTG ATC GTG ATC TT-3Ј (forward) and 5Ј-CCA ACT AAA AAG GCC CCG AA-3Ј (reverse), amplicon size 800 bp; caspase-1, 5Ј-AGG ACA AAC CGA AGG TGA TC-3Ј (forward) and 5Ј-TGT CCT GGG AAG AGG TAG AA-3Ј (reverse), amplicon size 390 bp. During PCR amplification, 0.5 Ci of [␣-33 P]dATP was added to each reaction as described previously (24). PCR products were separated on a 5% native polyacrylamide gel. The gel was then dried and exposed to Kodak Biomax MS film (Eastman Kodak Co.). Individual bands were cut from the dried gel and counted in 3 ml of ScintiVerse scintillation fluid (Fisher) using a liquid scintillation counter (Beckman Instruments) to quantify relative gene expression levels.
Flow Cytometry Analysis of IFNGR-1-Freshly harvested A549 cells were washed twice in cold wash buffer (0.1% NaN 3 , 1% fetal bovine serum in PBS, pH 7.2). Anti-IFNGR-1 monoclonal antibody (0.1 g/reaction) was added at 25 l/well to a U-bottom 96-well plate. Washed cells (4 ϫ 10 5 in 25 l of low serum medium) were then added to each corresponding well, and the plate was incubated for 1 h at room temperature with mixing several times during the incubation. Cells were pelleted and washed once before the addition of 50 l of wash buffer containing PBXL-3L-conjugated anti-mouse secondary antibody (5 l/reaction; Martek Biosciences, Columbia, MD). The cells were further incubated for 30 min at room temperature in the dark. After they were pelleted again and washed twice in wash buffer, 200 l of wash buffer was added to each well to resuspend and transfer the cells to a flow tube (130 ϫ 10 mm) containing 400 l of fixing solution (1% paraformaldehyde in PBS). Fixed samples were stored at 4°C in the dark before being analyzed using a Coulter Elite ESP flow cytometer (heliumneon laser) within 3 days.
Immunocytochemistry Analysis-A549 cells were plated in 12-well plates at 50% confluence and treated appropriately. At times of harvesting, cells were washed twice in PBS and immediately fixed with 3.7% formaldehyde (w/v) in PBS for 20 min at room temperature. They were then permeabilized with 0.2% Triton X-100 in PBS for 5 min, and the reactions were quenched with freshly prepared 0.1% sodium borohydride in PBS for 5 min. Afterward, the fixed and permeabilized cell monolayer was sequentially incubated in blocking buffer containing 10% fetal bovine serum, 1% bovine serum albumin, 0.02% NaN 3 in PBS, anti-IRF-1 polyclonal antibody in 1% bovine serum albumin in PBS, and fluorescein isothiocyanate-labeled anti-rabbit secondary antibody in 1% bovine serum albumin in PBS in the dark. After washing the monolayer twice in PBS, cells were counterstained with 1.5 g/ml 4Ј,6Ј-diamidino-2-phenylindole for 5 min in the dark. Samples were visualized under an Olympus IX70 inverted system fluorescence microscope equipped with appropriate filters (Hitech Instruments; Edgemont, PA). Images were captured using a SPOT camera (Diagnostic Instruments, Sterling Heights, MI).
Electrophoretic Mobility Shift Assay-Nuclear extract was prepared as described earlier, aliquoted, and stored at Ϫ80°C. For each electrophoretic mobility shift assay reaction, 5 g of nuclear protein was incubated with 30,000 cpm of [␥-32 P]ATP-labeled consensus IRF-1 gel shift oligonucleotide (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for 30 min on ice. For competition or supershift assay, unlabeled IRF-1 consensus or mutant oligonucleotides (in 10-or 50-fold in excess) or anti-IRF-1 monoclonal antibody (1 l) was incubated with nuclear extracts for 10 min on ice prior to the addition of radiolabeled oligonucleotides. Reaction mixtures were then separated on a 5% native polyacrylamide gel. After electrophoresis, the gel was dried and subjected to autoradiography (24).
Statistical Analysis-Statistical analysis was performed by using SuperANOVA software (Abacus Concepts, Berkeley, CA) for Student's t test, one-way analysis of variance (ANOVA), two-way ANOVA, and simple regression. Three-dimensional plot and multiple regression were performed by using SigmaPlot software (SPSS Inc., Chicago, IL). All data, unless specified, are shown as the mean Ϯ S.E., and difference was considered statistically significant when the p value was less than 0.05.

RESULTS
atRA Increases IFN␥-Induced IRF-1 Protein Expression-The dosedependent regulation by IFN␥ on IRF-1 protein in the presence or absence of atRA was investigated by pretreating A549 cells with atRA or vehicle overnight, and then with atRA and different concentrations of IFN␥ for 4 h (Fig. 1A). IRF-1 protein was induced dose-dependently by IFN␥. atRA by itself slightly induced IRF-1 transcripts (Fig. 1B), consistent with the reported findings (20); however, a more marked increase was found in the combination between atRA and IFN␥, where atRA was able to synergize with each concentration of IFN␥ to further increase IRF-1 expression (Fig. 1A). Maximal induction of IRF-1 was observed around 2-4 units/ml IFN␥, with similar to a ϳ5.5-fold increase with IFN␥ alone and ϳ9-fold with IFN␥ plus atRA. Since half-maximal induction was produced by 0.5 units/ml IFN␥, this concentration was chosen as the suboptimal concentration of IFN␥ for use in most of the subsequent experiments. These results demonstrate that pretreatment with atRA can augment the effect of low doses of IFN␥ in IRF-1 induction.
The kinetics of atRA and IFN␥-regulated IRF-1 were investigated by pretreating A549 cells with atRA or vehicle overnight and then with atRA and IFN␥ (0.5 units/ml (suboptimal) or 2 units/ml (maximal)) for 2, 4, and 8 h. Both mRNA (Fig. 1B) and protein ( Fig. 1, B and C) results showed that IRF-1 was induced transiently by low concentrations of IFN␥. The expression peaked at 4 h. Treatments with atRA significantly increased IFN␥-induced production of IRF-1, with an overall increase of 72% in the case of 0.5 units/ml and 35% in the case of 2 units/ml IFN␥ (Fig. 1C). Specifically, the combination of atRA and IFN␥, compared with IFN␥ alone, induced significantly higher levels of IRF-1 protein at 4 h for 0.5 units/ml IFN␥ and 4 and 8 h for 2 units/ml IFN␥. Thus, atRA promotes an increase of IFN␥-induced IRF-1 expression that is faster, higher, and more stable than that induced by IFN␥ alone.
Induction of IRF-1 by atRA and IFN␥ Is Transcriptionally Regulated-Actinomycin D (5 g/ml) completely blocked the protein expression of IFN␥-induced IRF-1 either with or without atRA (Fig. 1D), indicating that production of nascent transcripts is required for IRF-1 protein induction. This was confirmed by correlation analysis (Fig. 1E) between IRF-1 mRNA and protein (R 2 ϭ 0.89). These results suggest that regulation of IRF-1 transcription by atRA and IFN␥ is important for their modulation of IRF-1 protein.
Pretreatment with atRA Is Required to Sensitize A549 Cells-To test whether pretreatment with atRA was required for sensitizing A549 cells to better respond to low concentrations of IFN␥, cells were pretreated with atRA for periods of 0 -16 h, followed by treatments with atRA and IFN␥ (0.5 units/ml) for 4 h (Fig. 2). Shorter than 4 h of atRA pretreatment did not produce a significant effect; however, 4 h of atRA pretreatment was able to significantly increase IFN␥-induced IRF-1. The increase was even more marked with longer pretreatments (8 -16 h). This demonstrates that at least 4 h of pretreatment with atRA is required in atRA-mediated increase of IFN␥-induced IRF-1 protein. We FIGURE 1. Overnight pretreatment with atRA followed by suboptimal concentrations of IFN␥ induces a faster, higher, and more stable expression of IRF-1. A, A549 cells were pretreated with either atRA (0.1 M) or vehicle overnight and then with IFN␥ (0, 0.03125, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, or 100 units/ml) and atRA for 4 h. Cell lysates were prepared and assayed by Western blot. -Fold increases of IRF-1 protein compared with the untreated cells are shown (n ϭ 3; except for 100 units/ml of IFN␥, n ϭ 1). Two-way ANOVA showed significant differences for atRA (p Ͻ 0.002) and IFN␥ dose (p Ͻ 0.0001). B, A549 cells were pretreated with either atRA (0.1 M) or vehicle overnight and then with atRA and 0.5 or 2 units/ml of IFN␥ for 2, 4, and 8 h. Cell lysates were prepared and assayed for mRNA and protein levels of IRF-1 by RT-PCR and Western blot, respectively. GAPDH (for RT-PCR) and ␤-actin (for Western blot) were used as negative controls. The Western blot membranes were also stained by Ponceau S to assure equal loading of proteins. C, -fold increases of IRF-1 protein within each IFN␥ concentration group (n ϭ 6) are shown; the asterisks indicate significant differences between atRA and vehicle groups (Student's t test); AUC, area under the curve. D, cells were pretreated with either atRA or vehicle overnight and then with actinomycin D (Act D; 5 g/ml) for 1 h before the addition of IFN␥ (0.5 units/ml). IRF-1 protein at 4 h (time of maximal expression) is shown. E, a correlation test between -fold increases of mRNA and protein levels of IRF-1 was performed by using simple regression (R 2 ϭ 0.89).
hypothesize that protein(s) related to IFN␥ signaling, such as those important for STAT-1 activation, may be regulated during atRA pretreatment.
Pretreatment with atRA Increases IFN␥-induced STAT-1 Tyrosine Phosphorylation-IFN␥ signals through binding of the cytokine to IFN␥ receptors and generation of a kinase cascade that phosphorylates and activates STAT-1, which in turn transactivates the IRF-1 promoter. STAT-1 phosphorylation at residue Tyr-701 is required for its activation and transcriptional functions. Thus, we measured tyrosine-phosphorylated STAT-1 (pY-STAT-1) to investigate its involvement in atRA-mediated IRF-1 induction (Fig. 3A). pY-STAT-1 protein bands were enumerated by densitometry and plotted in Fig. 3B. As anticipated, neither vehicle nor atRA in the absence of IFN␥ could induce pY-STAT-1. IFN␥ at 0.5 units/ml, on the other hand, induced STAT-1 phosphorylation. pY-STAT-1 was visible after only 20 min of stimulation with IFN␥ and peaked between 1-2 h. Comparisons between any two time points tested showed significant differences except for 20 min and 8 h, indicating rapid and transient increase of pY-STAT-1. More importantly, pretreatment with atRA significantly increased IFN␥-induced pY-STAT-1; however, the effect of a signal dose of atRA at time 0 h was similar to that of IFN␥ alone, confirming the requirement of atRA pretreatment. Protein expression of total STAT-1 was not affected by any of the treatments (Fig. 3A).
When compared with the expression pattern of IRF-1 mRNA or protein, induction of pY-STAT-1 occurred earlier, suggesting that STAT-1 activation was upstream of IRF-1 expression. Overall, these results suggest that atRA pretreatment increases IFN␥-induced IRF-1 through a STAT-1-dependent pathway.
Expression of IFNGR-1 on the Cell Surface Is Increased upon atRA Pretreatment-IFNGR-1 is the ligand-binding chain of IFN␥ receptor complex. It is also important for Janus kinase-1 binding and STAT-1 recruitment (25). Thus, if total or surface expression of IFNGR-1 is regulated by overnight pretreatment with atRA, an increase of IFNGR-1 expression might explain the observed effects of atRA on increasing IFN␥-induced STAT-1 activation and IRF-1 mRNA and protein expression. To test this hypothesis, A549 cells were treated with vehicle or atRA overnight and harvested at time 0 h for Western blot and flow cytometry analysis (Fig. 4). Although the total protein level of IFNGR-1, assessed by Western blot (Fig. 4A), was not changed by atRA, multiple experiments showed consistently that atRA significantly increased the average intensity of IFNGR-1 on the cell surface (Fig. 4, B and C). The increase was modest, but considering that IFNGR-1 is the start point of IFN␥ signaling, a small increase in initial signaling could be amplified by the downstream events. The increase of IFNGR-1 surface expression by atRA is parallel with our earlier observations that both STAT-1 activation and IRF-1 expression induced by IFN␥ were further increased by pretreatment with atRA.
Ligands for RAR␣ Sequentially Increased the Levels of IFNGR-1, Activated STAT-1, and IRF-1-Retinoids regulate transcription of genes through binding to members of the retinoid receptor family. Western   blot analysis showed that at least two retinoid receptors, RAR␣ and RXR␣, were constitutively expressed in A549 cells (Fig. 5A). To sort out which retinoid receptors were involved in the effect of atRA on IFN␥ signaling and IRF-1 expression, we used several receptor-selective retinoids to treat A549 cells overnight and at time 0 h. Cells were harvested after overnight pretreatment (for IFNGR-1 analysis) or at either 1 h 40 min (approximate peaking time of pY-STAT-1), or 4 h (approximate peaking time of IRF-1 protein) after IFN␥ stimulation. The results showed that 9cRA, an RAR␣ agonist (Am580), and the combination of Am580 and a pan-RXR agonist significantly increased cell surface IFNGR-1 and IFN␥-induced pY-STAT-1 and IRF-1 to an extent similar to the effect of atRA (Fig. 5, B and C). RAR␣ antagonist blocked the effect of atRA on pY-STAT-1 and IRF-1 and that of Am580 on all three levels, suggesting an important role of RAR␣ in regulating IRF-1 induction. In fact, we found that atRA increased nuclear levels of RAR␣ (2.1-fold) and RXR␣ (2.4-fold) after overnight pretreatment with atRA, indicating that these retinoid receptors translocate into the nucleus and act to mediate the effect of atRA on IFN␥ signaling. RTMBE, a retinoid analog without known binding activity to retinoid receptors, was used as a negative control. All-transretinol, a precursor of atRA, did not increase either pY-STAT-1 or IRF-1, suggesting limited conversion of all-trans-retinol to atRA during overnight pretreatment. Protein expression of total STAT-1 was not affected by any of the treatments (data not shown).
To determine the correlations among the levels of IFNGR-1, pY-STAT-1, and IRF-1, we used a multiple regression model to analyze these three variables. Retinoid-treated levels of cell surface IFNGR-1 before IFN␥ stimulation and those of pY-STAT-1 and IRF-1 at 1 h 40 min and 4 h after IFN␥ stimulation, respectively, were normalized to the vehicle-treated levels. A three-dimensional plot (Fig. 5D) showed that positive regulators of all three variables (i.e. atRA, Am580, 9cRA, and the combination of Am580 and RXR-agonist) localized in the upper inner portion of the plot. In contrast, atRA or Am580 combined with RAR␣ antagonist was within the lower outer portion, with the vehicle and RTMBE. Multiple regression analysis showed significant correlations among IFNGR-1, pY-STAT-1, and IRF-1 (R 2 ϭ 0.85). This suggests that atRA (together with Am580 and 9cRA) sequentially regulates three components of the IFN␥ signaling pathway: cell surface IFNGR-1 and,

after IFN␥ is given, STAT-1 activation (maximum at 1 h 40 min) and IRF-1 expression (maximum at 4 h).
Pretreatment with atRA Increases Nuclear Localization and DNA Binding Activity of IFN␥-induced Nuclear IRF-1-IRF-1 stimulated by 9cRA and IFN␥ as part of a coactivation complex activates the promoter of TRAIL (17). However, little is known about the trafficking of IRF-1 between cytoplasm and nucleus. To investigate the effect of IFN␥ and atRA on nuclear localization of IRF-1, we performed Western blot of nuclear extracts (Fig. 6, A and B) and immunostaining of formaldehydefixed cells (Fig. 6C). IFN␥ by itself increased the nuclear level of IRF-1 at 4 h, which declined at 8 h, corresponding to the kinetics of IRF-1 protein expression (Fig. 1C). However, the combination of atRA and IFN␥ served to concentrate IRF-1 in the nucleus at both 4 and 8 h. This suggests that atRA not only plays a role in increasing IRF-1 expression but also functions to maintain IRF-1 levels in the nucleus for a longer period of time.
DNA binding activity of IRF-1 was also tested as an indication of its transcriptional functions. As shown in Fig. 6D, IRF-1 binding to the consensus element (IRF-E) was induced after IFN␥ stimulation, and the induction was further increased by atRA pretreatment. The IRF-1⅐IRF-E complex migrated so close to nonspecific bands that the complex was indistinct; however, supershift experiments clearly identified the complex. atRA pretreatment markedly increased the supershifted band induced by IFN␥. Competition and specificity controls indicated that the complex was specific. Moreover, kinetics experiments showed that atRA markedly increased IFN␥-induced IRF-1 binding to IRF-E at 8 h (Fig. 6E). This is consistent with the earlier demonstration that with atRA pretreatment, IFN␥-induced nuclear IRF-1 stayed at a higher level for a longer period of time (Fig. 6B).
The effect of atRA on IRF-1 binding was abolished by co-pretreatment with RAR␣ antagonist (data not shown), indicating that RAR␣ was involved in mediating the DNA binding activity of IRF-1. Determination of whether this is directly related to the effect of RAR␣ on IRF-1 protein levels requires further investigation.
atRA and IFN␥ Synergistically Increase Transcription of IRF-1 Target Genes-Transcript levels of two IRF-1 target genes, caspase-1 (18) and TRAIL (17), were examined by RT-PCR as indicators of biological  1 M). B, cells were pretreated with different retinoids (0.1 M; except for RAR␣ antagonist, which was 50ϫ in excess) overnight and with 0.5 units/ml of IFN␥ for 1 h 40 min or 4 h. Cell lysates were prepared and assayed for IRF-1 protein and pY-STAT-1 by Western blot. RTMBE was used as a negative control. RAR␣ antagonist is shown as v. C, cells were treated overnight with different retinoids at the same concentrations described above and then stained for the determination of IFNGR-1 cell surface expression by flow cytometry. Percentage increases of the average intensity of PBXL-3L dye compared with the vehicle are shown (n ϭ 6). Comparisons by one-way ANOVA showed that atRA, Am580, Am580/RXR, atRA/RAR␣ antagonist, and 9cRA are significantly different from either no retinoid or RTMBE (*, p Ͻ 0.05). D, -fold increases of the levels of cell surface IFNGR-1, pY-STAT-1, and IRF-1 are plotted on a three-dimensional scale. Multiple regression was performed that suggests significant correlations among IFNGR-1, pY-STAT-1, and IRF-1 (R 2 ϭ 0.85).
effects of atRA and IFN␥-induced IRF-1. Both genes were induced by IFN␥ and the combination of IFN␥ and atRA (Fig. 7A). Specifically, atRA pretreatment synergized with low dose IFN␥ to induce caspase-1 mRNA to more than 20-fold compared with the untreated cells at 24 h, whereas IFN␥ alone induced only 12-fold (Fig. 7B). TRAIL mRNA was induced about 3-fold by the combination. These results support the concept that pretreatment with atRA increases the transcriptional functions of IFN␥-induced IRF-1.

DISCUSSION
Vitamin A is important in the maintenance of normal immune functions. This is supported by studies of vitamin A deficiency, where disease risk is increased (26). However, therapeutic effects of vitamin A supplementation vary, depending on the types of illness and the populations studied (27). In addition, vitamin A has failed to affect the prevalence of illness, even in the presence of a large impact on mortality. In contrast, studies on vitamin A and morbidity have consistently showed an important effect of this nutrient on the severity of illness. In one study monitoring 1455 young children weekly for a year (28), vitamin A-supplemented children had less frequent vomiting and anorexia, although the prevalence of diarrhea or acute respiratory infections was not changed. The supplementation also led to fewer clinic attendances, hospital admissions, and death. Thus, vitamin A appears to reduce the severity of infection, possibly by enhancing functions of innate and cell-mediated immune responses and helping to repair virus-damaged epithelia (1).
Our present study was designed to study how atRA, an active metabolite of vitamin A, increases IFN␥-induced response as part of innate immunity. Type I and II IFNs were discovered as proteins that inhibit virus replication (29). They are induced in response to virus infection, secreted, and function to activate a global antiviral state. Type II IFN (IFN␥) is generally expressed at a low level in the early stages of an infection (30); thus, enhancing the functions of albeit small amounts of IFN␥ during early infection may potentially improve the overall IFN response and encourage faster recovery.
IFN␥ signaling pathway is initiated upon binding of IFN␥ to IFNGR-1. The extracellular portion of IFNGR-1 chain contains multiple disulfide bonds that are required for maximal binding of the ligand. The intracellular portion of IFNGR-1 contains a tyrosine residue that becomes phosphorylated upon ligand binding and acts in STAT-1 recruitment (25). STAT-1 is phosphorylated at Tyr-701, which is required for the activation and dimerization of STAT-1 through Src homology 2 domains (31). The activated STAT-1 homodimer binds and activates transcription of target genes through the conserved response element ␥-interferon-acti-  vated site (32). STAT-1 is also phosphorylated on Ser-727 in order to exert maximal transcriptional activity (33).
In the present study, we found that atRA of a subpharmacologic concentration (0.1 M) increased the effect of low-dose IFN␥ (0.5 units/ml) on IRF-1 induction by about 2-fold, potentially enhancing the antiviral functions of IFN␥. IRF-1, a target of IFN␥ signaling, was originally identified as a mouse nuclear factor that specifically bound to the upstream regulatory region of the IFN␤ gene. In 1988, Taniguchi and co-workers (11) cloned and characterized the cDNA encoding IRF-1, and a role of IRF-1 in viral infections was then suggested, since it possessed a virusinducible promoter (34).
In previous studies, Matikainen et al. (20) showed that atRA (1 M) rapidly up-regulated IRF-1 in promyelocytic leukemia NB4 cells, inducing the mRNA level to nearly 8-fold within 3 h of treatment. There was also a 3-fold increase of IRF-1 protein levels by atRA (1 or 100 M) in cervical squamous carcinoma SiHa cells (23). In our studies of lung epithelial A549 cells, we observed minimal effect of atRA (0.1 M) by itself after 16-h (overnight) treatment or longer, probably due to the transient action of lower dose atRA on IRF-1 induction (23). However, atRA at this concentration significantly potentiated the effect of low dose IFN␥ on IRF-1 induction (Fig. 1).
atRA pretreatment of a certain length of time (4 h or longer) was required for sensitizing these cells to better respond to IFN␥ (Fig. 2). We also used a protein synthesis inhibitor, cycloheximide (10 g/ml), together with atRA pretreatment, to determine whether the increase of IRF-1 transcription was blocked by inhibiting protein synthesis (data not shown). The result showed that both IFN␥-induced and IFN␥ plus RA-induced transcription was blocked by cycloheximide, indicating the need of certain protein(s) in both processes. Such proteins may regulate the migration of IFNGR-1 from the cytoplasm to the cell surface. However, whether or not these proteins are targets of atRA pretreatment requires further investigation. Nevertheless, our study has demonstrated that treatment with atRA increases the levels of cell surface IFNGR-1 (Fig. 4). This is supported by reports that up-regulation of type I IFN receptors by retinoids is associated with increased signaling of IFN␣/␤ (35,36).
STAT-1 is tyrosine-phosphorylated and activated upon IFN␥ stimulation (37). atRA was observed in our model to enhance tyrosine phosphorylation (activation) of STAT-1 induced by low dose IFN␥, without increasing STAT-1 protein levels (Fig. 3). This effect of atRA required pretreatment, suggesting that the increase of STAT-1 activation is due to the increase of cell surface IFNGR-1 during atRA pretreatment. In comparison, the rapid effect of atRA on IRF-1 induction may not involve STAT-1 functions (20,22), despite a report that atRA induces IRF-1 through a ␥-interferon-activated site motif of the IRF-1 promoter (21). However, the present study indicates that STAT-1 activation is required in the regulation of IFN␥-induced IRF-1 by atRA.
The use of receptor-selective retinoids in the present study revealed strong correlations among the levels of IFNGR-1, activated STAT-1, and IRF-1 (Fig. 5), indicating that atRA and RAR␣ together influenced several components of the IFN␥ signaling pathway. Our results identified RAR␣ ligands as important regulators of IFN␥-induced IRF-1. atRA has been shown to enhance transactivation of STAT-1 in the presence of RAR␣ in acute promyelocytic leukemia cells (38). However, it was unknown if RAR␣ mediated other components of IFN␥ signaling. In the present study, atRA also increased nuclear levels of RAR␣, which, together with RXR␣, may mediate the regulatory effect of atRA on multiple components of the IFN␥ signaling pathway. How atRA and RAR␣ function in this context requires further investigation. We hypothesize that during the pretreatment period of 4 h or longer, atRA and RAR␣ regulate the synthesis or posttranslational modification of proteins involved in surface expression of IFNGR-1, tyrosine phosphorylation of STAT-1, and/or transcription of IRF-1.
atRA as a panagonist for all RARs may have increased IFNGR-1 surface expression via receptors other than RAR␣, since the antagonist specific for RAR␣ did not block the effect of atRA on IFNGR-1. This is supported by the observation that RAR␤ agonist also increased cell surface IFNGR-1 (data not shown). Hence, additional mechanisms of atRA action might affect IFNGR-1 but not pY-STAT-1 or IRF-1. Nevertheless, our current findings suggest an essential role of RAR␣ in mediating the effect of atRA on IFN␥ signaling. atRA pretreatment also potentiated the transcriptional activity of In A549 cells, atRA pretreatment of 4 h or longer facilitates the shuttling of RAR␣ and its dimerization partner, RXR␣, from the cytoplasm to the nucleus in a ligand-dependent manner (1). This enhances the transactivation activity of RAR␣/ RXR␣ heterodimer, leading to yet to be characterized changes that prepare the cells for increased response to low dose IFN␥. Via the actions of RAR␣, atRA increases the levels of IFNGR-1 on the cell surface (2), enhancing tyrosine phosphorylation (activation) of STAT-1 upon IFN␥ stimulation (3). IRF-1, a target gene of STAT-1, is then induced (4) and translated (5); its nuclear localization and DNA binding activity (6), and target gene (caspase-1 and TRAIL) transcription (7), are also increased.
IFN␥-induced IRF-1, increasing its nuclear localization and DNA binding activity (Fig. 6). Notably, pretreatment with atRA maintained higher IRF-1 levels in the nucleus for a longer period of time, with a nearly 4-fold difference between IFN␥ alone and atRA plus IFN␥ at 8 h. This is possibly due to an unknown role of atRA in facilitating nuclear translocation of IRF-1 and/or preventing shuttling of the protein out of the nucleus. Reports are scarce on this subject; however, a retinoid-mediated increase of nuclear localization of calcyclin-binding protein was observed during neuronal differentiation (39). Another report has shown that nucleocytoplasmic translocation of RAR␤ is facilitated by RXR in a ligand-dependent manner (40). Our results are consistent with these findings, where atRA facilitated or maintained the nuclear localization of RAR␣, RXR␣, and IRF-1 and may thereby prolong the transactivation of IRF-1. IFN␥-induced caspase-1, an enzyme important for the production of IFN␥ during an innate immune response, was further increased by pretreatment with atRA (Fig. 7). Caspase-1 processes the precursors of IL-1␤ (41) and IL-18 (19) to their biologically active forms. IL-1␤ and IL-18, alone or in synergy with IL-12, increases lipopolysaccharide-induced IFN␥ production by natural killer cells, evidenced by marked reduction of lipopolysaccharide-induced serum IFN␥ titers in caspase-1 Ϫ/Ϫ mice (19) and IL-12-deficient mice (42). It is very likely that IFN␥ (low dose)-induced caspase-1, which can be enhanced by atRA pretreatment, feeds forward IFN␥ signaling during early stages of an infection.
TRAIL, another target gene of IRF-1, induces apoptosis selectively in cancer cells, sparing normal cells that are generally TRAIL-insensitive (43). IFN␥-9cRA cotreatment synergistically induces TRAIL mRNA levels due to sustained occupancy of IRF-1 on the TRAIL promoter (17), supporting the use of combination therapies between IFNs and retinoids in cancer treatments. The induction of TRAIL could also be important in the cellular response to infection but has received little attention in this context. Our data indicate that the retinoid status of the cell can affect the induction of TRAIL by low doses of IFN␥ (Fig. 7), suggesting another mechanism by which retinoids may improve the response of the host to infectious disease.
In summary, the present study has detailed the regulating roles of atRA in IFN␥ signaling and IRF-1 induction. Using RAR␣ as a mediator, atRA pretreatment sensitizes A549 cells to better respond to low dose IFN␥, increasing cell surface IFNGR-1, STAT-1 activation, levels of IRF-1 mRNA and protein, and transcriptional functions of IRF-1 (Fig.  8). Could these results in a model system of the cellular response to low dose IFN␥ be relevant to human studies of vitamin A status and infectious disease? In epidemiological studies conducted in populations where vitamin A deficiency remains a public health problem (see World Health Organization review (27)), supplementation with vitamin A has been shown to reduce the severity of infectious diseases (28) and to significantly reduce all-cause mortality rates in children and pregnant women (44,45). It is possible that vitamin A, through conversion to physiological levels of atRA, results in the generation of an "active atRA state," which serves to improve the integrity of epithelial tissue barriers as a first line of host defense and to enhance the initial response to viral and bacterial infections through potentiation of the response to low levels of IFN␥, such as are likely to be present in the early stages of an infection. A heightened host response to IFN␥, produced when the "active atRA state" is adequate, may provide an important advantage in quickly responding to viral and bacterial agents and could be part of the mechanism whereby vitamin A has been shown to be effective in reducing morbidity and mortality in human populations.