A 5′ Extended IFN-Stimulating Response Element Is Crucial for IFN-γ–Induced Tripartite Motif 22 Expression via Interaction with IFN Regulatory Factor-1

Tripartite motif (TRIM) family proteins are involved in diverse cell processes, including apoptosis, differentiation, and transcriptional regulation (1, 2). They are characterized by a combination of RING, B-box, and coiled-coil domains (1). The RING domain of many TRIM proteins has been shown to possess E3 ubiquitin ligase activity (2–5), whereas the B-box and coiled-coil domains may be involved in protein interactions and homo/heterodimerization (1, 2). Recent studies have demonstrated that many members of the TRIM family play important roles in innate antiviral immunity. For example, TRIM5α has been shown to block the infectivity of a range of different retroviruses in a species-specific manner (6, 7); TRIM25 is essential for RNA helicase RIG-I–mediated antiviral activity (5, 8); TRIM28 can inhibit the replication of murine leukemia viruses or related retroelements in embryonic cells by transcriptional silencing (9, 10). It is speculated that the TRIM family may represent a new, widespread class of proteins involved in antiviral innate immunity (11, 12).

In our previous study, we demonstrated that one of the TRIM family members, TRIM22 (also named Staf50), could inhibit hepatitis B virus (HBV) gene expression and replication efficiently by significantly inhibiting the activity of HBV core promoter in a RING domain-dependent manner (13). We also demonstrated that TRIM22 was a RING finger E3 ubiquitin ligase (4), and its E3 ligase activity was responsible for its antiviral activity against encephalomycocarditis virus as reported by another research group (14). Additionally, several studies indicated that TRIM22 possessed antiretroviral activity depending on certain cell types (15–17).

IFNs have an important role in immune system to defend against viral infections. They consist of two main classes of related cytokines: type I IFNs and type II IFNs. Although there are many type I IFNs, including IFN-α, IFN-β, IFN-λ, etc., IFN-γ is the only type II IFN family member (18–20). The biological effects of both IFN-α/β and IFN-γ are mediated by IFN-stimulated genes (ISGs), but the induction of ISGs by IFN-γ is often more complex than that of IFN-α/β, largely due to the fact that many of the IFN-γ–stimulated genes are induced with variable kinetics unlike those stimulated by IFN-α/β, and the expression of some IFN-γ–regulated genes requires de novo protein synthesis (20–23).

To date, nearly 70 TRIM family members have been identified, and two classes of TRIMs can already be distinguished: IFN-inducible TRIMs and constitutive TRIMs (12). The IFN-inducible TRIMs are involved in a broad range of biological processes that are associated with innate immunity, and the constitutive TRIMs, such as TRIM1 and TRIM28, can also block virus replication at early stages. The two classes of TRIMs may contribute to the overall antiviral defense (11, 12). TRIM22 was demonstrated to be one of the most strongly induced TRIMs in HepG2 cells after treatment with IFNs in our previous study (13). Similar to our results, Barr et al. (15) found that TRIM22 was the most upregulated TRIM queried on the microarray prepared from IFN-treated human osteosarcoma cells. Additionally, TRIM22 was also reported to be strongly induced by IFNs in several other cell lines, such as Hela, MCF-7, and Daudi, etc. (17, 24). However, at present, little is known about how IFNs regulate TRIM22 gene expression.

In this paper, we mainly focused our study on the molecular mechanisms of transcriptional regulation of TRIM22 gene by IFN-γ. We identified a novel cis-element termed 5′ extended IFN-stimulating response element (eISRE), which was crucial for the IFN-γ inducibility of TRIM22. In addition, the 5′ eISRE was also implicated in the constitutive transcriptional activity of TRIM22. Furthermore, we demonstrated that IFN regulatory factor-1 (IRF-1) could bind to the 5′ eISRE both in vitro and in vivo and played a key role in the transcriptional regulation of TRIM22.

The human hepatoma HepG2 cells were maintained in DMEM supplemented with 10% FBS, 2 mM l-glutamine, and 100 U/ml penicillin-streptomycin. HepG2 cells were incubated at 37°C with 100% humidity in 5–7% CO₂ and passaged using standard cell-culture techniques. HepG2 cells were stimulated by IFN-γ for various times or with different concentrations and then prepared for RT-PCR and immunoblot analysis.

HepG2 cells were harvested, and total RNA was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. One microgram total RNA was reverse transcribed using a commercially available cDNA synthesis kit and oligo(dT) primer (MBI Ferments, St. Leon-Roth, Germany). Subsequently, cDNA was subjected to PCR in a 20 μl reaction mixture with the following primers: TRIM22, forward, 5′-ACCAAACATTCCGCATAAAC-3′ and reverse, 5′-GTCCAGCACATTCACCTCAC-3′; and GAPDH, forward, 5′-ATCCCATCACCATCTTCCAG-3′ and reverse, 5′-GAGTCCTTCCACGATACCAA-3′. The cycling parameters were as follows: denaturation at 94°C for 5 min, amplification at 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s for 32 cycles. PCR products were subjected to electrophoresis by using 2% agarose gels in Tris-borate–EDTA and visualized by ethidium bromide staining. Real-time quantification of cDNA targets was performed using a Light Cycler 480 and SYBR Green system (Roche Diagnostic Systems, Mannheim, Germany) according to the manufacturer’s instructions. TRIM22 expression was calculated following normalization to GAPDH levels by the comparative ΔΔ threshold cycle method. The specificity of the amplification reactions was confirmed by melt curve analysis. Results were representative of three independent experiments.

Protein was denatured in SDS, electrophoresed on SDS-PAGE (8 or 10% gel), and transferred onto a nitrocellulose membrane. Nonspecific binding was blocked with TBST [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.1% Tween 20] containing 5% (w/v) nonfat milk for 2 h at room temperature. After overnight incubation at 4°C with indicated Ab, membranes were washed three times with TBST and incubated at room temperature for 2 h with the corresponding HRP-conjugated secondary Ab diluted 1:1000 in blocking buffers. After washing, signals were detected by an ECL kit (Pierce, Rockford, IL).

A series of deletions of the TRIM22 promoter were generated by PCR amplification using Takara Ex Taq DNA polymerase (Takara, Dalian, China) with the TRIM22 promoter dependent luciferase plasmid [which has been donated by Professor Urban Gullberg (Lund University, Lund, Sweden) (25), designated as pLuc-500 in this study] as a template. All of the PCR fragments were inserted into the SacI and HindIII sites of the pGL3 Basic vector (Promega, Madison, WI), and these reporter plasmids are designated as pLuc-400 (−254 to +146), pLuc-320 (−174 to +146), pLuc-240 (−94 to +146), pLuc-160 (−14 to +146), and pLuc-80 (+67 to +146) according to the length of the remaining 5′-flanking region. Mutations of ISRE1, IFN-γ activation site (GAS), and 5′ eISRE were performed using QuikChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA).

HepG2 cells were transfected using Lipofectamine 2000 as recommended by the manufacturer (Invitrogen). Briefly, HepG2 cells were plated at 1.5 × 10⁵ cells per well in a six-well plate 24 h pretransfection. A total of 1.5 μg plasmid was mixed with 3 μl Lipofectamine 2000 in serum-free medium and added into HepG2 cells. In all transfection assays, pCMV–β-gal was cotransfected to normalize the transfection efficiency. Cells were incubated for 6 h at 37°C in the presence of the transfection complex, washed twice with serum-free medium, and grown in fresh supplemented medium. For IFN-γ–treated samples, IFN-γ was added 24 h posttransfection, and the cultures were incubated for another 24 h. The cells were harvested, and the luciferase activity in the cell lysates was determined with the Luciferase Reporter Assay System (Promega). Reporter activities are presented as means ± SD of at least three independent experiments.

A total of 5 × 10⁶ IFN-γ–treated or untreated HepG2 cells were washed in cold PBS and resuspended in 500 μl buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 0.5 mM PMSF). Postincubation for 15 min on ice, cells were lysed in 0.6% Nonidet P-40 (NP-40) by vigorous vortex for 10 s and centrifuged at 5000 rpm for 5 min. Pellets contained the nuclei and were washed in buffer A two times and resuspended in 50 μl buffer C (20 mM HEPES [pH 7.9], 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF). After shaking for 30 min at 4°C and centrifugation for 10 min at 10 000 rpm, supernatants were used as nuclear extracts.

A total of 5 μg nuclear proteins was preincubated on ice with 2 μg poly(deoxyinosine-deoxycytosine) as an unspecific competitor and 1 μg sonificated sperm DNA in band shift buffer (50 mM Tris, 150 mM KCl, 5% glycerol, 10 mM MgCl₂, and 0.1% NP-40, 5 mM EDTA, 2.5 mM DTT) for 15 min. Biotin-labeled oligonucleotides (+28 to +52, GAGAATTTAACTTTCGTTTCTCACT) were then added in a total volume of 20 μl, incubated on ice for 20 min, and loaded onto 5% native polyacrylamide gels in 0.5× Tris-borate–EDTA buffer. The gels were blotted on nylon membrane, and the blot was cross-linked by UV irradiation. Biotin-labeled probe was detected by a Lightshift Chemiluminescent EMSA kit (Pierce) according to the manufacturer’s recommendations. In competition assays, a 5 or 25 M excess of unlabeled competitor oligonucleotides was added prior to the addition of the probe to the mixture, which was then preincubated on ice for 15 min. For supershift experiments, 2 μg STAT1, STAT2, STAT3, IRF-1, IRF-2, or IRF-9 was added to the preincubation mixture, and the preincubation time was extended to 30 min.

The binding of above-mentioned transcription factors to the 5′ eISRE was further assessed with the ELISA-based transcription factor assay according to the method as described by McKay et al. (26). In brief, 200 ng biotin-labeled oligonucleotides (+28 to +52) was added into each well of a streptavidin-coated microtiter plate (Boehringer, Mannheim, Germany). The plate was incubated at room temperature for 1 h, then immersed in wash buffer (0.05% Tween in PBS) and washed three times for 10 min in a shaking bath at room temperature. Each well of the plate was then incubated at 37°C for 30 min with 30 μg nuclear extract in 60 μl 1× protein binding buffer (25 mM HEPES–KOH [pH 7.5], 50 mM KCl, 4 mM MgCl₂, 20% glycerol, 250 mg/ml BSA, 250 mg/ml poly[deoxyinosine-deoxycytosine]). After washing, 60 μl primary Ab (1:1000) in Ab dilution buffer-2% BSA, 0.05% Tween in PBS was incubated with each well at room temperature for 1 h. The bound Abs were detected with HRP-labeled secondary Ab (1:5000) followed by measurement of enzymatic color reaction in a standard microplate reader.

Chromatim immunoprecipitation (ChIP) experiments were performed according to the manufacturer’s recommendations (Upstate Biotechnology, Lake Placid, NY). In brief, 5 ×10⁶ HepG2 cells were incubated with or without IFN-γ (1000 U/ml) for 6 h. The cells were fixed with 1% formaldehyde for 10 min at room temperature to cross-link transcription factors to DNA. The cross-linking reaction was stopped by adding glycin with a final concentration of 0.125 M. After washing by cold PBS, the cells were resuspended in cell lysis buffer [10 mM Tris-HCl (pH 8), 10 mM NaCl, 0.2% NP-40 and protease inhibitors), and nuclei were centrifuged at 2500 rpm for 5 min. The nuclei were isolated and sonicated on ice to shear the DNA to 200–1000 bp. A small aliquot (20 μl) was saved as input DNA for PCR analysis by reversing histone-DNA cross-links by heating at 65°C for 4 h. Chromatin was immunoprecipitated from 200 μl aliquots at 4°C by mild agitation overnight with 5 μg Abs specific for IRF-1, IRF-9, STAT1, STAT2, STAT3, IRF-2, or IRF-3. Immune complexes were collected by incubation with protein A agarose. To analyze the target region, the immunoprecipitated chromatin DNA samples were amplified by PCR using the primer pair forward, 5′-TTCCCCAGGGTTTATTGTTATG-3′, and reverse, 5′-GCCG GTGTAAACCAGATTCAC-3′, spanning the 5′-flanking region of TRIM22 (−21 to +93) containing 5′ eISRE.

Transfection of short interference RNAs (siRNAs) targeting IRF-1 or IRF-9 was performed using Lipofectamine 2000 with a final concentration of siRNA of 100 nM according to the manufacturer’s instructions. The siRNAs targeting IRF-1 or IRF-9 and the scrambled siRNA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).

Results were reported as means ± SD. t test was applied to comparisons between groups; a p value <0.05 was considered statistically significant.

We first examined the expression of TRIM22 gene induced by IFN-γ in human hepatoma HepG2 cells. It was found that the expression of TRIM22 could be induced by IFN-γ in a dose-dependent manner at both the mRNA and protein level (Fig. 1A, 1B). To determine whether the induction of TRIM22 gene by IFN-γ was a transcriptional response, we treated HepG2 cells with actinomycin D (Act.D), an inhibitor of gene transcription. It was found that pretreatment with Act.D could abolish the IFN-γ–induced TRIM22 mRNA expression at the indicated time points, implying that the induction was at the transcriptional level (Fig. 1C).

To further characterize the induction of TRIM22 by IFN-γ, cycloheximide (CHX), an established inhibitor of protein synthesis, was added to cells pretreatment with IFN-γ. As shown in Fig. 1D, pretreatment of HepG2 cells with CHX dramatically decreased levels of TRIM22 mRNA at the indicated time points, suggesting that de novo protein synthesis was essential for IFN-γ–induced TRIM22 gene expression.

To identify the cis-acting elements required for the IFN-γ inducibility of TRIM22, a series of DNA sequences derived from the 5′ flanking region of the TRIM22 gene were produced and cloned into the promoterless luciferase reporter plasmid pGL3-Basic (Fig. 2A, left panel). These luciferase reporter plasmids were then transiently transfected into HepG2 cells, treated with IFN-γ or left untreated, and assayed for luciferase activity. Reporter assays revealed a significantly increased transcriptional activity of the construct pLuc-500 (−354 to +146) as compared with the empty vector, and an induction of the transcriptional activity by ∼5-fold after IFN-γ stimulation, indicating the 500-bp 5′ flanking region of TRIM22 gene was sufficient to ensure high levels of expression and IFN-γ inducibility. It was also found that the IFN-γ inducibility was not significantly altered by the 5′ deletions leading to the 400, 320, 240, and 160 bp fragments. In contrast, the IFN-γ inducibility was completely lost in the case of the construct pLuc-80 (+67 to +146), and this construct only retained ∼25% of the transcriptional activity of the construct pLuc-160 (−14 to +146) (Fig. 2A, right panel), indicating the region (–14 to +66) was of critical importance not only for the IFN-γ inducibility, but also for the constitutive expression of TRIM22 gene.

The region (−14 to +66) was then subjected to computer search and manual analysis based on sequence homologies to identify potential transcription factor binding sites. As shown in Fig. 2B, one region was found matching GAS consensus sequence TTCNNNG/TAA, GAS (+24 to +32, TTCTGAGAA), and two regions that match ISRE consensus sequence, AGTTTCN_(1–2)TTTCNY: ISRE1 (+16 to +28, TCTTTCACTTCTG) and ISRE2 (+37 to +49, ACTTTCGTTTCTC). To investigate the role of each putative IFN-γ–responsive element in the transcriptional regulation of TRIM22 gene, mutations were introduced into the above sites with reporter plasmid pLuc-500 or pLuc-160 as the template, respectively (Fig. 2B), and the transcriptional activity of the resulting constructs was examined by transient transfection experiments. Results showed that mutation at ISRE1 had no influence on the IFN-γ inducibility of TRIM22. In contrast, mutation at ISRE2 abolished the IFN-γ inducibility and led to the decrease in the basal transcriptional activity of TRIM22 gene. Unexpectedly, mutation at GAS did not have significant influence on the IFN-γ inducibility (Fig. 2C). These results suggested that IRES2 might be required for the basal and IFN-γ–induced transcriptional activity of TRIM22 gene.

Transient transfection experiments suggested but did not prove that the IRES2 contributed to the IFN-γ–induced transcriptional activity of TRIM22 gene. To corroborate this suggestion, a synthetic double-stranded oligonucleotide spanning the region (+35 to +52) and containing the ISRE2 motif (ACTTTCGTTTCTC) was inserted into the pGL3-Promoter reporter plasmid (Fig. 3A, left panel). This plasmid contains a luciferase reporter gene under the control of an SV40 promoter without enhancer sequences, and putative transcriptionally active sequences can be cloned upstream of the SV40 promoter. The reporter plasmid was then assayed for the ability to respond to IFN-γ as described above. The result showed that the region (+35 to +52), which harbors ISRE2, failed to confer the IFN-γ inducibility (Fig. 3A, right panel), therefore we added nucleotides upstream of the region (+35 to +52) according the sequence in the 5′ flanking region of TRIM22 gene. It was found that addition of nucleotides to the position +33 could increase the IFN-γ inducibility markedly, and addition of nucleotides to the position +31 could further increase the effect, but further addition of nucleotides upstream of the region (+31 to +52) to the position +26 failed to further increase the effect (Fig. 3A). We also tested the contribution of the nucleotides at the 3′ end of the region (+26 to +52) to the IFN-γ inducibility by deletion analysis and found that deletion of three nucleotides at the 3′ end to the position +49 had no influence on the IFN-γ inducibility, but further deletion of nucleotides at the 3′ end nearly abolished the IFN-γ inducibility (Fig. 3A). We also tested the IFN-γ inducibility of the ISRE of human 6-16 gene upon the SV40 promoter and found that the ISRE of human 6-16 gene also conferred IFN-γ inducibility upon a heterologous promoter (Fig. 3A), consistent with the results reported by Reid et al. (27).

The thymine triplets of ISRE have been reported to play important roles in the IFN-inducibility (28, 29), and the above-mentioned result also demonstrated that mutation of a thymine triplet of IRES2 (GAGAATTTAACTTTCGTTTCTCATC, ISRE2 was shown in bold) disrupted the IFN-γ responsiveness of TRIM22 (Fig. 2B, 2C). As there is an additional thymine triplet (+33 to +35) upstream of ISRE2 (GAGAATTTAACTTTCGTTTCTCATC), we suspected that this thymine triplet might also contribute to the IFN-γ responsiveness of TRIM22 gene. Using site-directed mutagenesis analysis, we found the thymine triplet (+33 to +35) was also crucial for the IFN-γ induciblity of the TRIM22 gene (Fig. 3B). We also investigated the contribution of the thymine triplet (+44 to +46, GAGAATTTAACTTTCGTTTCTCATC) and the nucleotides downstream of IRES2 (+50 to +51, GAGAATTTAACTTTCGTTTCTCATC) to the IFN-γ inducibilty of TRIM22. As shown in Fig. 3B, the thymine triplet (+44 to +46) was crucial for the IFN-γ inducibility of the TRIM22 gene, whereas the nucleotides downstream of ISRE2 seemed to be dispensable.

Taken together, these results indicated that at least six additional nucleotides (AATTTA) upstream of ISRE2 were required, whereas the nucleotides downstream of ISRE2 seemed to be dispensable, and all three thymine triplets contributed to the IFN-γ inducibility of TRIM22 gene. As this novel IFN-γ response element consisted of ISRE2 plus six additional nucleotides upstream of it, we therefore named it 5′ eISRE.

To analyze the binding of nuclear proteins to the 5′ eISRE in response to IFN-γ treatment, we prepared nuclear extracts from HepG2 cells at various times after the addition of 1000 U/ml IFN-γ and performed EMSA with the biotin-labeled fragment from bp +28 to +52 which contains the 5′ eISRE as a probe (Fig. 4A). As shown in Fig. 4B, two DNA–protein complexes, C1 and C2, were formed within 30 min of IFN-γ stimulation with its peak value at 6 h poststimulation, and the DNA-protein complexes could be continuously detected over the 24 h after the IFN-γ addition. The binding of nuclear proteins to the probe was specific because the binding was abolished by an excess of wild-type competitor, but not by an excess of mutant competitor (Fig. 4C). We also performed competition assays with an excess of ISRE2 or GAS competitors, and it was found that both ISRE2 and GAS competitors failed to block the interaction between the IFN-γ–induced proteins and the biotin-labeled 5′ eISRE (data not shown). Furthermore, both biotin-labeled ISRE2 and GAS (Fig. 4A) failed to bind to the IFN-γ–induced nuclear proteins efficiently as shown in Fig. 4D.

The ISRE motif was originally described in the promoters of several IFN-inducible genes and is the recognition site for multiprotein complex IFN-stimulated gene factor 3, composed of IRF-9 and the phosphorylated STAT1 and STAT2 (30, 31). The ISRE motif also binds to other IRF family members, such as IRF-1 and IRF-2 (32–34). To further characterize the protein complexes binding to the 5′ eISRE, we performed supershift assay with Abs against STAT1, STAT2, STAT3, IRF1, IRF2, IRF3, and IRF9. As shown in Fig. 5A, the major DNA–protein complex, C1, was shifted by the anti–IRF-1 Ab, and the minor complex, C2, was shifted by anti-IRF9 Ab, whereas other Abs failed to shift the protein complexes.

To further confirm the results, a more sensitive ELISA-based assay was used to detect the transcription factors binding to the biotin-labeled probe. The results showed that IRF1 could bind to the immobilized probe very efficiently. Consistent with the EMSA results, the binding occurred within 30 min with its peak value at 6 h after the IFN-γ stimulation, and the binding could be continuously detected over the 24 h after the IFN-γ addition. IRF9 could also bind the probe, although to a lesser extent, whereas other transcription factors did not show significant binding (Fig. 5B).

To investigate the expression of these transcription factors in HepG2 cells after IFN-γ stimulation, Western blot was performed to detect their expression in nuclear extracts from HepG2 cells at various times after the addition of 1000 U/ml IFN-γ. Well induction of these transcription factors by IFN-γ was demonstrated (Fig. 5C), thus excluding the possibility that the inability of other transcription factors, such as STAT1, to bind to the 5′ eISRE was due to their weak expression in IFN-γ–induced HepG2 cells or the poor quality of the Abs.

To further determine which transcription factor might bind to the 5′ eISRE in living cells, ChIP assays were used to precipitate transcription factor–DNA complexes with various Abs as mentioned above. The immunoprecipitated DNA was then subjected to PCR analysis using primers flanking the 5′ eISRE sequence. As shown in Fig. 5D, IRF-1 (and to a lesser extent IRF-9) could bind to the 5′-flanking region, which contains 5′ eISRE of the TRIM22 gene, whereas other transcription factors, such as STAT1, could not.

To investigate the role of IRF-1 and IRF-9 in the expression of TRIM22, we constructed the expression plasmid for IRF-1 or IRF-9, respectively, by cloning the appropriate PCR fragments in the pCDNA3.1 vector between NheI–HindIII. The expression of rIRF-1 or IRF-9 was confirmed by immunoblotting (Fig. 6A). We then tested the effect of IRF-1 or IRF-9 on the transcriptional activity of TRIM22 promoter by cotransfecting the luciferase reporter plasmid pLuc-160 into the HepG2 cells together with increasing amounts of pcDNA-IRF-1 or pcDNA-IRF-9. It was found that overexpression of IRF-1 increased the transcriptional activity of pLuc-160 in a dose-dependent manner, whereas overexpression of IRF-9 did not have significant influence on the activity of pLuc-160 (Fig. 6B). Cotransfection with the reporter plasmid pLuc-160 ISRE2mut and IRF-1 or IRF-9 demonstrated the failure of both IRF-1 and IRF-9 to stimulate the transcriptional activity of pLuc-160 ISRE2mut in HepG2 cells (Fig. 6B). Further study showed that overexpression of IRF-1 significantly upregulated the mRNA level of TRIM22 in HepG2 cells, whereas IRF-9 could not (Fig. 6C).

To further investigate the role of IRF-1 and IRF-9 in the IFN-γ inducibility of the TRIM22 gene, we silenced their expression via RNA interference. Functionality of the siRNAs directed against IRF-1 and IRF-9 in HepG2 cells was confirmed by Western blot analysis: protein levels of IRF-1 and IRF-9 following IFN-γ treatment were strongly reduced in the presence of specific siRNA compared with the unspecific control siRNA (Fig. 6D). We then cotransfected the luciferase reporter plasmid (pLuc-160) into HepG2 cells together with 100 ng IRF-1, IRF-9 specific, or control siRNA, respectively. Twenty-four hours posttransfection, the transfected HepG2 cells treated with IFN-γ for additional 24 h. Results showed that the silence of IRF-1 abolished the IFN-γ inducibility and even could lead to the decrease in the constitutive transcriptional activity of the TRIM22 gene. However, silencing the expression of IRF-9 did not have significant influence on the IFN-γ inducibility of the TRIM22 gene (Fig. 6E). Furthermore, IRF-1 siRNA, but not IRF-9 siRNA, completely inhibited the induction of TRIM22 mRNA by IFN-γ in HepG2 cells (Fig. 6F).

Taken together, these results indicated the critical role of IRF-1 in the transcriptional activation of TRIM22 gene, which may explain the dependence on new protein synthesis for IFN-γ–induced TRIM22 expression as shown in Fig. 1D.

HBV replication is blocked noncytolytically by both IFN-α/β and IFN-γ (35–37). Although IFN-α/β is not strongly induced during acute HBV infection of chimpanzees, IFN-γ from intrahepatic immune cells plays a central role in controlling virus replication (38, 39). TRIM22, for which expression is correlated with HBV clearance in acutely infected chimpanzees (38, 39), was found to play an important role in antiviral immunity against HBV in our previous study (13). Interestingly, we also found that TRIM22 was strongly induced by IFN-γ in human hepatoma HepG2 cells, indicating its role as a downstream effector in IFN-γ–mediated anti-HBV immune responses. However, how IFN-γ induces the expression of TRIM22 remains obscure. In this study, we investigate the molecular mechanism of TRIM22 induction by IFN-γ.

With deletion analysis, we narrowed down the IFN-γ responsive region of the TRIM22 gene to an 80-bp region (−14 to +66). Interestingly, this region also contributed to the basal transcriptional activity of TRIM22 gene. Inspection of the region (−14 to +66) by computer search and manual sequence analysis revealed three putative regulatory motifs, including one GAS and two ISRE elements. It is known that the majority of responsive genes are induced by IFN-γ through the interaction of STAT1α homodimers with GAS element, and, in some cases, ISRE can also contribute to the IFN-γ inducibility (18, 21). In this study, however, we found the GAS element seemed to be dispensable for the IFN-γ induction of TRIM22 gene, as mutation of the GAS element could not significantly decrease the IFN-γ–induced transcriptional activity, and this cis-element was unable to bind to the IFN-γ–induced nuclear proteins as demonstrated by EMSA experiments.

A more interesting finding of this study is that a novel cis-element 5′ eISRE, which consists of ISRE2 (AGTTTCGTTTCTC) and six nucleotides (AATTTA) upstream of it, was crucial for the IFN-γ inducibility. First, when cloned into the reporter plasmid pGL3-Promoter immediately upstream of SV40 promoter, 5′ eISRE, not ISRE2, was found to be sufficient to confer IFN-γ inducibility on a heterologous promoter. Second, site-directed mutagenesis demonstrated that besides two thymine triplets in ISRE2 (AATTTAAGTTTCGTTTCTC), the thymine triplet upstream of ISRE2 (AATTTAAGTTTCGTTTCTC) was also required for IFN-γ inducibility of TRIM22. Third, EMSA experiments showed that it was 5′ eISRE, not ISRE2, that could bind to the IFN-γ–induced nuclear proteins efficiently. Similar to our present findings, the ISRE of several ISGs, such as guanylate-binding protein (GBP), IL-12, β₂-microglobulin, and caspase-8 (34, 40–42), also has the features of an eISRE, although the contribution of the sequence adjacent to the conventional ISRE for the IFN inducibility of those ISGs has not been determined. Additionally, there also exist differences in DNA sequence between the 5′ eISRE of TRIM22 and the eISRE of other ISGs. For example, the eISRE of GBP comprises a conventional ISRE plus another TTTC repeat and that of caspase-8 consists of a conventional ISRE and an additional TTTTTG motif. Interestingly, these eISREs seem to be bound preferentially by IRF members: the eISRE of all above-mentioned ISGs can be bound by IRF-1 and that of IL-12 or β₂-microglobulin can also be bound by IRF-8 in immune cells.

With supershift analysis, an ELISA-based transcriptional factor assay, and ChIP analysis, we found the 5′ eISRE of TRIM22 could be efficiently bound by IRF-1, but not by STAT1, both in vitro and in vivo. It is known that IRF-1 is involved in diverse biological processes, such as antiviral immunity, antiproliferative response, apoptosis, and inflammation, etc., and is critical for the IFN-γ–induced expression of several ISGs, such as iNOS, gp91 phox, and GBP, etc. (43–45). Our study also founded that IRF-1 was crucial for both the basal and IFN-γ–inducible expression of TRIM22, as overexpression of IRF-1 cDNA in HepG2 cells resulted in high-level expression of TRIM22, whereas knocking down IRF-1 expression abolished the IFN-γ inducibility and significantly downregulated the basal expression of TRIM22. The involvement of IRF-1 may also explain the dependence on new protein synthesis for IFN-γ–inducible expression of TRIM22.

As TRIM22 expression can be strongly induced by IFN-α, we have also investigated whether the 5′ eISRE or IRF-1 contributed to the IFN-α–induced TRIM22 expression. We found that, similar to its role in IFN-γ–induced TRIM22 expression, the 5′ eISRE also played an important role in the IFN-α inducibility of the TRIM22 gene, as mutation of any of the thymine triplets of the 5′ eISRE abolished this inducibility. Upon IFN-α stimulation, it was IRF-1, not STAT1, that bound to the 5′ eISRE, and IRF-1 siRNA could abolish the IFN-α–induced TRIM22 expression (data not shown). Interestingly, several other ISGs, such as 9-27 gene, human ISG20, and human STAT1 (27, 32, 46), are also induced by both IFN-α and IFN-γ via a similar molecular mechanism, which may be due to the cross talk between the signaling pathways activated by both types of IFN.

TRIM22 gene is located in chromosome 11p15 in a cluster with other TRIM genes including TRIM3, TRIM5, TRIM6, TRIM21, and TRIM34. It is of interest that all of these TRIMs are IFN-inducible genes, suggesting they may have coevolved to coordinate important antiviral functions (11, 47). TRIM5 was particularly related with TRIM22, as TRIM5 is adjacent to TRIM22 in a head-to-head direction in chromosomes with only a 4.8-kb distance, and they have evolved under positive selection in a mutually exclusive fashion (48). Asaoka et al. (49) have investigated the molecular mechanisms of TRIM5α induction by IFNs. They identified a functional ISRE in the TRIM5α promoter that could be bound by STAT1 upon IFN-β stimulation. However, the ISRE of TRIM5α does not have the characteristics of an eISRE, and the interaction between this ISRE and other transcriptional factors, such as IRF-1, has not been investigated. As the majority of TRIM family members are IFN-inducible genes, and the molecular mechanisms of the induction of most TRIM family members by IFNs remain largely unknown (11, 12, 47), it is of interest to investigate whether the ISRE of other TRIM family members possesses the features of eISRE and whether IRF-1 plays roles in IFN-induced expression of other TRIM family molecules.

Because TRIM22 has been demonstrated with strong inhibitory effect on HBV replication, the 5′ eISRE might possibly be involved in the anti-HBV activity of the IFN-γ–induced TRIM22. It is known that the abilities of spontaneous clearance of viral infection after HBV exposure differ significantly among various subjects (37). Our preliminary data showed that TRIM22 expression varied among different individuals, and more strikingly, the IFN-induced TRIM22 expression was individual specific (data not shown). Such variations in TRIM22 expression and IFN inducibility in liver cells may contribute to differences in response to treatment of HBV infections and different courses of viral pathogenesis in infected individuals. We think the 5′ eISRE may contribute in several ways to such variations. First, there may exist nucleotide mutations in the 5′ eISRE of TRIM22 in IFN low-response individuals. Second, the transcription factors, such as IRF-1, are crucial for the IFN-induced TRIM22 expression through binding with the 5′ eISRE. Varied expression levels of 5′ eISRE-binding transcription factors among individuals may also influence the expression of antiviral protein TRIM22. So it would be of interest to design chemical compounds or small peptides to upregulate TRIM22 expression via mimicking or stimulating the expression of those key transcription factors, with the aim to enhance the clearance of HBV infection.

Disclosures The authors have no financial conflicts of interest.