Nucleocapsid protein of SARS-CoV activates the expression of cyclooxygenase-2 by binding directly to regulatory elements for nuclear factor-kappa B and CCAAT/enhancer binding protein.

SARS-associated coronavirus (SARS-CoV) causes inflammation and damage to the lungs resulting in severe acute respiratory syndrome. To evaluate the molecular mechanisms behind this event, we investigated the roles of SARS-CoV proteins in regulation of the proinflammatory factor, cyclooxygenase-2 (COX-2). Individual viral proteins were tested for their abilities to regulate COX-2 gene expression. Results showed that the COX-2 promoter was activated by the nucleocapsid (N) protein in a concentration-dependent manner. Western blot analysis indicated that N protein was sufficient to stimulate the production of COX-2 protein in mammalian cells. COX-2 promoter mutations suggested that activation of COX-2 transcription depended on two regulatory elements, a nuclear factor-kappa B (NF-kappaB) binding site, and a CCAAT/enhancer binding protein (C/EBP) binding site. Electrophoretic mobility shift assay (EMSA) and chromatin immunoprecipitation (ChIP) demonstrated that SARS-CoV N protein bound directly to these regulatory sequences. Protein mutation analysis revealed that a Lys-rich motif of N protein acted as a nuclear localization signal and was essential for the activation of COX-2. In addition, a Leu-rich motif was found to be required for the N protein function. A sequence of 68 residuals was identified as a potential DNA-binding domain essential for activating COX-2 expression. We propose that SARS-CoV N protein causes inflammation of the lungs by activating COX-2 gene expression by binding directly to the promoter resulting in inflammation through multiple COX-2 signaling cascades.

Introduction

In March 2003, a novel severe acute respiratory syndrome-associated coronavirus (SARS-CoV) was identified as the causative agent of SARS (Ksiazek et al., 2003). The virus has been demonstrated to induce fever, edema, and diffuse alveolar damage in severely affected individuals (Poutanen et al., 2003). Similar to other coronaviruses in structure, SARS-CoV is an enveloped virus containing a single-stranded, positive-sense RNA genome, 29.7 kb nucleotides in length, that encodes four viral structural proteins including the spike (S) glycoprotein, the matrix (M) protein, the small envelope (E) protein, and the nucleocapsid (N) protein (Marra et al., 2003). The gene encoding for the 46-kDa SARS-CoV N protein directly precedes the 3′-UTR of the viral genome (Rota et al., 2003). For coronaviruses, the N protein plays an important role during viral packaging, viral core formation, and viral RNA synthesis (Narayanan, Chen, Maeda, & Makino, 2003). The SARS-CoV N protein shares homology with N protein from other members of the coronavirus family. However, it contains a short lysine-rich sequence (362-KTFPPTEPKKDKKKKTDEAQ-381) near the carboxyl terminus, a putative bipartite nuclear localization signal that has not been found in the N proteins of any other known coronavirus (Marra et al., 2003). This suggested that the SARS-CoV nucleocapsid protein might have novel functions in pathogenesis. The SARS-CoV N protein has been reported to form a dimmer by self-association (He et al., 2004), to activate the activator protein 1 (AP1) signal transduction pathway (He et al., 2003), and to induce actin reorganization in COS-1 cells (Sutjit et al., 2004).

Cyclooxygenase-2 (COX-2) is the inducible COX isoform that catalyzes the formation of prostaglandins in response to cytokines, and to oncogenic and mitogenic factors (Wu, Zhu, & Wu, 2003). COX-2 has been known to play an important role in inflammation, tissue damage, and tumorigenesis (Murono et al., 2001). The induction of COX-2 expression depends primarily on transcriptional activation by diverse stimuli. There are many consensus cis-elements in the 5′-flanking region that may contribute to transcriptional regulation of the COX-2 gene. Only a limited number of elements including a cyclic-AMP response element (CRE) at −53/−59, a CCCAAT/enhancer-binding protein (C/EBP) element at −124/−132, and two NF-κB sites at −438/−447 and −213/−222 have demonstrated roles in human COX-2 transactivation (Murono et al., 2001).

A number of investigations demonstrated that viral proteins stimulate COX-2 expression. For example, latent membrane protein 1 of Epstein-Barr virus, X protein of hepatitis B virus, Gp120 and Tat of human immunodeficiency virus, core and NS5A proteins of hepatitis C virus up-regulate COX-2 expression in human cell lines (Bagetta et al., 1998, Lara-Pezzi et al., 2002, Murono et al., 2001, Núñez et al., 2004). However, the roles of SARS-CoV proteins in the regulation of COX-2 protein expression and inflammation are still unclear.

Because COX-2 gene expression is associated with inflammatory processes and can be induced by viral proteins, the aim of this study was to investigate the roles of SARS-CoV encoded proteins in the regulation of COX-2 expression and to determine the molecular mechanisms responsible. Our results showed that SARS-CoV N protein was able to activate the COX-2 gene expression and both NF-κB and C/EBP elements were involved in the activation of COX-2 by SARS-CoV N protein.

Materials and methods

Plasmid construction and cell culture

The 5′ flanking sequence −891 to +9 containing the core promoter region and a series of truncation mutants of the human COX-2 gene were constructed into a promoterless luciferase expression vector PGL3 (Promega) as described previously (Saunder, Sansores-Garcia, Gilroy, & Wu, 2001). Site-specific mutations of CRE, C/EBP and two NF-κB sites were performed by using QuikChange site-directed mutagenesis kit (Stratagene). The CRE site within −891/+9 fragment was mutated from -59TTCGTCA-53 to TTgagCA, the C/EBP site was mutated from -132TTACGCAAT-124 to gcgatagcT, one of NF-κB sites was changed from -222GGGACTACCC-213 to aattCTACCC (NF-κB-A), and the other was altered from -447GGGGATTCCC-438 to attcATTCCC (NF-κB-B). The mutations and their corresponding primers are listed in Table 2. Single or double mutants were also constructed into luciferase expression vector PGL3.

Table 2

Primers used for site-specific mutations in this study

MutN1u	5′-TGGAGGACGCAATGGGGCAAGGGGTTTACCCAATAATACTGCGT-3′
MutN1d	5′-ACGCAGTATTATTGGGTAAACCCCTTGCCCCATTGCGTCCTCCA-3′
MutN2u	5′-GGCTAGCGGAGGTGGTGAAACTGCCGAGAGCAAAGTTTCTGGTAAAGGCC-3′
MutN2d	5′-GGCCTTTACCAGAAACTTTGCTCTCGGCAGTTTCACCACCTCCGCTAGCC-3′
MutN3u	5′-TAAGAAATCTGCTGCTGAGGCATCTACTGCCACAAAACAGTACAACGTCA-3′
MutN3d	5′-TGACGTTGTACTGTTTTGTGGCAGTAGATGCCTCAGCAGCAGATTTCTTA-3′
MutN4u	5′-CAAAACATTCCCACCAACAGAGCCTACTGATGAAGCTCAGCCTTTGCCGC-3′
MutN4d	5′-GCGGCAAAGGCTGAGCTTCATCAGTAGGCTCTGTTGGTGGGAATGTTTTG-3′
CREu	5′-AGGCGGAAAGAAACAGTCATTTGAGCTCATGGGCTTGGTTTTCAGTC-3′
CREd	5′-GACTGAAAACCAAGCCCATGAGCTCAAATGACTGTTTCTTTCCGCCT-3′
C/EBPu	5′-AACCCTGCCCCCACCGGCGCGATAGCTTTTTTTAAGGGGAGAGGAG-3′
C/EBPd	5′-CTCCTCTCCCCTTAAAAAAAGCTATCGCGCCGGTGGGGGCAGGGTT-3′
NF-κB-Au	5′-GGATCAGACAGGAGAGTGAATTCTACCCCCTCTGCTCCCAAATTGG-3′
NF-κB-Ad	5′-CCAATTTGGGAGCAGAGGGGGTAGAATTCACTCTCCTGTCTGATCC-3′
NF-κB-Bu	5′-GGCGGCGGCGGCGGGAGAATTCATTCCCTGCGCCCCCGGACCTCAG-3′
NF-κB-Bd	5′-CTGAGGTCCGGGGGCGCAGGGAATGAATTCTCCCGCCGCCGCCGCC-3′

Genes of SARS-CoV strain WHU (GenBank Accession No. AY394850) were amplified by RT-PCR with RNA isolated from SARS-CoV infected Vero E6 cells (Zhu et al., 2005). The amplified genes and their corresponding primers are listed in Table 1 . The PCR products were purified using a DNA extraction kit (Fermentas), digested with EcoRI and SalI or BamHI and EcoRI (underlined sequence in primers), and cloned into the vector pCMV-Tag2 (Stratagene). Site-specific mutations of the N gene were performed by using QuikChange site-directed mutagenesis kit (Stratagene). Four site-specific mutations of the N gene were as follows: MutN1 (Δ38PKQRRPQ44); MutN2 (Δ220LALLLLDRLNQL231); MutN3 (Δ257KKPRQKR263); MutN4 (Δ369KKDKKKK376). The deleted regions of MutN3 and MutN4 comprised lysine/arginine-rich motifs. The mutations and their corresponding primers are listed in Table 2 . Forward primers (N1–N5) for sequential truncations of the N gene from the 5′ end are listed in Table 3 , and the reverse primer is Nd (Table 1). Reverse primers (N6–N9) for sequential truncations of the N gene from the 3′ end are listed in Table 3, and the forward primer was Nu (Table 1). Human embryo kidney cell line (HEK293T) was cultured in Dulbecco's modified Eagle's medium, supplemented with 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin. All cells were maintained in a humidified 5% CO2 incubator at 37 °C.

Table 1

Primers used for amplification of genes of SARS-CoV strain WHU in this study

X1u	5′-CGTTCCGAATTCGATGGATTTGTTTATGAGATTTTTTAC-3′
X1d	5′-CGTTATGTCGACGTTTCTTCCGAAACGAATGAGTAC-3′
X2u	5′-GGTCCTGAATTCGATGCCAACTACTTTGTTTGCTG-3′
X2d	5′-CGATTGGTCGACGAATACCACGAAAGCAAGAAAAAG-3′
X3u	5′-GGTTCCGAATTCGATGTTTCATCTTGTTGACTTCCAGG-3′
X3d	5′-GAGCACGTCGACTGATGGGCAAGGTTCTTTTAGTAGT-3′
X4u	5′-GTGCGTGAATTCGATGAAAATTATTCTCTTCCTGAC-3′
X4d	5′-CTCGTAGTCGACAGGCTAAAAAGCACAAATAGAAG-3′
X5u	5′-GGGACGGAATTCCATGTGCTTGAAGATCCTTGTAAGGTAC-3′
X5d	5′-GTCGTAGTCGACGTCCACCAAATGTAATGCGGGGGGC-3′
ORF1u	5′-GCGCGCGAATTCGATGAATGAGCTCACTTTAATTGAC-3′
ORF1d	5′-AGTGCCGTCGACCAAAGCCAAGCAGTGCTATAAG-3′
ORF2u	5′-AGTCTGGAATTCGATGAAACTTCTCATTGTTTTGACTT-3′
ORF2d	5′-TGCGTAGTCGACTGACAGTTGATAGTAACATTAGGTG-3′
ORF3u	5′-ATGCTGGAATTCGATGGACCCCAATCAAACC-3′
ORF3d	5′-CCATGTCGACGTAATAGAAGTACCATCTGGGGCTG-3′
ORF4u	5′-ATGCTGAATTCGATGCTGCCACCGTGCTACAAC-3′
ORF4d	5′-TACGTAGTCGACCTCAGCAGCAGATTTCTTAGTGACAG-3′
Eu	5′-AGCTGGATCCATGTACTCATTCGTTTCGGAAGAAAC-3′
Ed	5′-AGCTGAATTCTTAGTTCGTTTAGACCAGAAGATC-3′
Mu	5′-AGCTGGATCCGCTTATCATGGCAGACAACGGTACT-3′
Md	5′-AGCTGAATTCCATCTGTTGTCACTTACTGTACTAGC-3′
Nu	5′-AGCTGGATCCATGTCTGATAATGGACCCCAATCAAAC-3′
Nd	5′-AGCTGAATTCCATCATGAGTGTTTATGCCTGAGT-3′

Table 3

Primers for sequential deletions of N gene

N1	5′-ACAGGATCCCAAGGAGGAACTTAGATT-3′
N2	5′-AGTGGATCCGAGGGAGCCTTGAATACA-3′
N3	5′-ATAGGATCCGGGAAATTCTCCTGCTCG-3′
N4	5′-TTAGGATCCGAACGTCACTCAAGCATTT-3′
N5	5′-ATAGGATCCATATCATGGAGCCATTAAAT-3′
N6	5′-GGTGAATTCTTTAATGGCTCCATGATA-3′
N7	5′-ATAGAATTCGTTTTGTGGCAGTACGTTT-3′
N8	5′-TAAGAATTCAGGAGAATTTCCCCTACTG-3′
N9	5′-TATGAATTCTATTCAAGGCTCCCTCAGT-3′

Transfection and luciferase assays

Cotransfection of luciferase reporter vectors with relevant recombinant plasmids into cells were carried out by mixing 0.2 μg of luciferase reporter vectors and 0.4 μg plasmids with 2 μl Sofast™ transfection reagent (Xiamen Sunma Biotechnology Co., Ltd.). The mixture was then added to each well of 24-well plates with 293T cells growing at 70% confluence. After incubation for 24 h, the cells were serum-starved for 24 h before harvesting for luciferase activity assays.

Western blot analysis

Cell samples were washed with cold PBS and dissolved in lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl2, 1 mM EDTA, 0.1 mM phenylmethulsulfonyl fluoride, 50 mM NaF, 1 mM sodium orthovanadate, and 0.06 mg/ml aprotinin) on ice for 30 min. After centrifugation at 12,000 rpm for 15 min, the supernatants were separated and protein concentration was determined. Proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred to nitrocellulose membranes. The membranes were probed with a rabbit polyclonal anti-COX-2 (Santa Cruz) or rabbit anti-N polyclonal antibody (prepared in this study) or mouse monoclonal anti-flag antibody (Santa Cruz), followed by incubation with horseradish peroxidase conjugated goat anti-rabbit IgG or goat anti-mouse IgG (Santa Cruz). Immunoreactivity was visualized by chemiluminescent detection (Pierce).

Preparation of nuclear extracts

Nuclear extracts were prepared as described previously (Kin & Fischer, 1998) with some modifications. Cells incubated in serum-free media for 24 h were washed with cold PBS twice and scraped into 1 ml of cold PBS, followed by centrifugation at 2000 rpm for 10 min in a microcentrifuge and incubated in buffer A (10 mM HEPES, pH 8.0, 0.5% Nonidet P-40, 1.5 mM MgCl, 10 mM KCl, 0.5 mM DTT, and 200 mM sucrose) for 10 min on ice. Nuclei were collected by centrifugation in a microcentrifuge at 12,000 rpm for 15 s. Pellets were rinsed with buffer A, resuspended in buffer B (20 mM HEPES, pH 7.9, 1.5 mM MgCl, 420 mM NaCl, 0.2 mM EDTA, and 1.0 mM DTT), and incubated on a rocking platform for 30 min at 4 °C. Nuclei were clarified by centrifugation at 12,000 rpm for 15 min, and the supernatants were diluted 1:1 with buffer C (20 mM HEPES, pH 7.9, 100 mM KCl, 0.2 mM EDTA, 20% glycerol, and 1 mM DTT). Protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 50 mg/ml of both aprotinin and leupeptin) and phosphatase inhibitors (10 mM NaF, 10 mM β-glycerophosphate, 0.1 mM sodium orthovanadate, and 1 mM EGTA) were added to each buffer.

Electrophoresis mobility shift assay (EMSA)

Probes were generated by annealing single-strand oligonucleotides (sequences are listed in Table 4 ) containing the cognate promoter regions of the COX-2 gene and labeling the ends with [γ-32P] ATP using T4 polynucleotide kinase (TaKaRa). The COX-2 cis-regulatory motifs, C/EBP1, C/EBP2, NF-κB1, and NF-κB2, were analyzed.

Table 4

Sequences of primers for EMSA and ChIP

C/EBP1	-138ACCGGCTTACGCAATTTTTTTAAG-115
C/EBP2	-115GAGCAGAGGGGGTAGTCCCCACTCT-138
NF-κB1	-228AGAGTGGGGACTACCCCCTCTGCTC-204
NF-κB2	-204GAGCAGAGGGGGTAGTCCCCACTCT-228
CP1	-502ACTTCCTCGACCCTCTAAAGACGT-479
CP2	-2TCGCTAACCGAGAGAACCTTCCTT-25
C/EBP-CP1	-155TAAAAAACCCTGCCCCCACCGG-134
NF-κB-CP1	-243GAGGGATCAGACAGGAGAGT-224
NF-κB-CP2	-136GGTGGGGGCAGGGTTTTTTA-155

EMSAs were performed with 4 μg of nuclear extract in binding buffer (20 mM Hepes, pH 7.9, 0.1 mM EDTA, pH 8.0, 75 mM KCl, 2.5 mM MgCl2, and 1 mM DTT) containing 1 μg of poly(dI-dC). To assure the specific binding of transcription factors to the probe, unlabeled double-stranded oligonucleotide competitors were preincubated at a 50-fold molar excess for 10 min prior to probe addition. For supershift experiments, 2 μg of purified polyclonal antibody was incubated with nuclear extracts on ice for 30 min before adding to the binding buffer. Samples were then electrophoreses on 5% nondenaturing polyacrylamide, 0.25× Tris borate/EDTA gels, and the gels were dried and subjected to autoradiography.

Chromatin immunoprecipitation (ChIP)

The assay was done as previously described with slight modifications (Wu et al., 2003). Monolayer of 293T cells (80% confluent) were incubated for 24 h after transfection, and then were serum-starved for 24 h. Formaldehyde was added to the culture medium to a final concentration of 1%. The cells were then washed twice with PBS, scraped, and lysed in lyses buffer (1% SDS, 10 mM Tris–HCl, pH 8.0, 1 mM PMSF, 50 mg/ml of both aprotinin and leupeptin) for 10 min on ice. The lysates were sonicated on ice and the debris was removed by centrifugation at 12,000 rpm for 15 min at 4 °C. One-fourth of the supernatant was used as DNA input control. The remaining supernatant was diluted 10-fold with dilution buffer (0.01% SDS, 1% Triton X-100, 1 mM EDTA, 10 mM Tris–HCl, pH 8.0, and 150 mM NaCl) and incubated with antibody against N protein overnight at 4 °C. Immunoprecipitated complexes were collected using protein A/G agarose beads. The pellets were washed with dialysis buffer (2 mM EDTA, 50 mM Tris–HCl, pH 8.0). Samples were incubated at 67 °C for 5 h to reverse formaldehyde crosslink. DNA was precipitated with ethanol and extracted three times with phenol/chloroform. Finally, pellets were resuspended in TE buffer and subjected to PCR amplification using COX-2 promoter specific detection primer (Table 4).

The PCR products were resolved by agarose gel electrophoresis.

Results

SARS-CoV N protein activates COX-2 promoter and induces COX-2 protein expression

To investigate the roles of proteins encoded by SARS-CoV in the regulation of COX-2, we constructed a series of plasmids carrying 12 individual SARS-CoV genes or potential open reading frames (Fig. 1A). Each of these plasmids was cotransfected into 293T cells with a reporter plasmid carrying the luciferase gene under the control of the COX-2 promoter. Cells were incubated for 24 h, serum-starved for an additional 24 h, and harvested. Cell lysates were analyzed for luciferase activity. Luciferase activity assays demonstrated that N protein activated the COX-2 promoter (Fig. 1B, lane 10). COX-2 promoter activity was about 12-fold higher in cells transfected with the plasmid expressing the N protein than that of vector control (Fig. 1B, lanes 10 and 13), while the rest of the expression plasmid constructs for viral proteins (X1, X2, E, M, X3, X4, ORF1, ORF2, X5, ORF3, and ORF4) tested in this study had no significant effects on the expression of luciferase (Fig. 1B, lanes 1–9, 11, and 12).

Fig. 1

SARS-CoV N protein activates the expression of COX-2 in 293T cells. (A) Diagram of individual open reading frames of SARS-CoV with locations on the viral genome. (B) Analysis COX-2 promoter activation by individual SARS-CoV proteins. 293T cells were cotransfected with plasmid expressing different SARS-CoV proteins, respectively, and the reporter plasmid in which the luciferase gene is under the control of COX-2 promoter. Relative luciferase activity was determined by standard procedures. The expressed genes are indicated in panel A: 13 is the negative control pCMV-Tag2 and 14 is the positive control pCMV-Tag2-HBx (X protein of hepatitis B virus). (C) Analysis of dose-dependent of the SARS-CoV N protein in the activation of COX-2 promoter. 293T cells were cotransfected with different amount of plasmids expressing N protein along with the reporter plasmid and relative luciferase activity was determined. Values correspond to an average of at least three independent experiments done in duplicate. Error bars show 1 S.D. (D) Western blot analysis of the expression of COX-2 protein activated by N protein. Cells were transfected with empty vector pCMV-Tag2 as a control (lane 1) or with plasmid pCMV-Tag-N, expressing SARS-CoV N protein (lane 2). Cell extracts were prepared and the expressed proteins were determined using rabbit anti-COX-2 antibody. (E) Western blot analysis of same blot probed with antibodies against N protein.

To determine the activation of COX-2 by N protein was dependent on the amount of N protein, different concentrations of plasmid expressing the N protein along with plasmid carrying the reporter gene were cotransfected into 293T cells. Luciferase activity assays showed that COX-2 promoter activity increased as the concentration of plasmid DNA increased until the concentration reached to 0.4 μg (Fig. 1C), indicating activation of COX-2 promoter by N protein was concentration-dependent.

To determine the role of N protein in the regulation of COX-2 protein production, plasmid (pCMV-Tag2-N) carrying the N gene or control plasmid (pCMV-Tag2) was transfected into 293T cells, respectively. Transfected cells were treated and harvested as described above. Western blot analysis of cell lyses using COX-2 antibody showed that the level of COX-2 protein production increased in the presence of N protein (Fig. 1D, lane 2) relative to control transfection (Fig. 1D, lane 1). To confirm the expression of N protein in transfected cells, Western blot analysis was also carried out using antibody against the N protein (anti-N). N protein was detected in cells transfected with the plasmid pCMV-Tag2-N (Fig. 1E, lane 2), but not present in cells transfected with control plasmid pCMV-Tag2 (Fig. 1E, lane 1). Results from Western blot analysis and luciferase activity assays showed that the N protein of SARS-CoV is sufficient for the activation of COX-2 promoter and for the production of COX-2 protein.

NF-κB and C/EBP binding elements are required for the expression of COX-2 activated by SARS-CoV N protein

To define the COX-2 cis-regulatory elements that were responsive to SARS-CoV N protein, truncation mutants and site-specific mutations of the promoter were generated (Fig. 2 ). Reporter plasmids were then constructed in which the luciferase gene was under the control of each individual mutant COX-2 promoter. To test the functions of mutant promoters, 293T cells were cotransfected with a plasmid carrying the SARS-CoV N gene (pCMV-Tap2-N) and plasmids containing the luciferase reporter gene driven by mutated COX-2 promoters. The mutated COX-2 promoter was determined by luciferase activity assays. Results indicated that mutations in promoter elements C/EBP, NF-κB-A, and a C/EBP-CRE double mutations significantly decreased expression from COX-2 promoter in response to N protein, respectively, while other mutations had little effects on the activation of COX-2 promoter regulated by the N protein, compared to the full-length wild type promoter (Fig. 2). These results indicated that C/EBP recognition site and one of the two NF-κB-A binding sites were required for the activation of COX-2 promoter by SARS-CoV N protein suggesting N protein regulates COX-2 gene expression in a NF-κB and C/EBP recognition element dependent manner.

Fig. 2

Functional analysis of cis-regulatory elements of the COX-2 promoter. Diagram of individual cis-regulatory elements of the COX-2 promoter and its truncated or site-specific mutants are shown in the left panel and results from luciferase activity assay are shown in the right panel. Plasmid carrying the SARS-CoV N gene and plasmids containing the luciferase reporter gene driven by individual COX-2 promoter mutants were cotransfected into 293T cells. Promoter activities were determined by measuring the relative luciferase activity in transfected-cell lysates. pCMV-Tag2 was used as a vector only control. Luciferase activities correspond to an average of at least three independent experiments done in duplicate. The black symbols indicate mutations. Error bars show 1 S.D.

SARS-CoV N protein binds directly to C/EBP and NF-κB regulatory elements on the COX-2 promoter

Localization to the nucleus is a common feature of coronavirus nucleoproteins (Wurm, Chen, Hodgson, Britton, Brooks, & Hiscox, 2001). A short lysine-rich region near the carboxyl terminus of SARS-CoV N protein has been identified as a putative bipartite nuclear localization signal (Marra et al., 2003). Since C/EBP and NF-κB regulatory elements are required for the expression of COX-2 gene activated by N protein. It is reasonable to assume that SARS-CoV N protein function may through binding to C/EBP and NF-κB regulatory elements. To confirm this speculation, we conducted electrophoresis mobility shift assay to define protein–DNA binding between N protein and COX-2 promoter. 293T cells were transfected with a control plasmid (Fig. 3A and B, lanes 1–3) or a plasmid containing the N gene (Fig. 3A and B, lanes 4–6). Nuclear extracts were prepared from transfected cells and EMSA was performed with 4 μg of nuclear extract in binding buffer. To assure the specific binding of transcription factors to the probe, unlabeled double-stranded oligonucleotide competitors were added prior to the addition of labeled probe (Fig. 3A and B, lanes 1 and 4). To determine whether N protein was specific bound to the promoter, rabbit anti-N polyclonal antibody was incubated with nuclear extracts before adding the binding buffer (Fig. 3A and B, lanes 3 and 6). DNA probes used in this study contained either the C/EBP element (Fig. 3A) or the NF-κB-A elements (Fig. 3B) from COX-2 promoter. Samples were then electrophoresed on nondenaturing polyacrylamide gels and subjected to autoradiography. Results from EMSA using C/EBP probe showed that a specific protein–DNA complex was supershifted in cells transfected with plasmid expressing N protein (Fig. 3A, lane 6). Similar results were also observed when NF-κB element probe was used (Fig. 3B, lane 6).

Fig. 3

Determination of interaction between SARS-CoV N protein and COX-2 promoter by electrophoretic mobility shift assay (EMSA). EMSA was performed with nuclear extracts of 293T cells transfected with (lanes 4–6) or without (lanes 1–3) the N gene. Probes were generated by annealing single-stranded and end-labeled oligonucleotides containing the cognate COX-2 promoter regions. C/EBP at nucleotides −132/−125 (A) or NF-κB at nucleotides −228/−204 (B) probes were added to all reactions (lanes 1–6). Unlabeled double-stranded oligonucleotide competitors were added during preincubation prior to probe addition (lanes 1 and 5). For supershift experiments, polyclonal antibody was incubated with nuclear extracts before adding to the reaction (lane 6). Samples were electrophoresed on 5% nondenaturing polyacrylamide gel and visualized by autoradiography. Arrows indicate the super shifted protein–DNA complexes.

To confirm N protein-promoter DNA binding, chromatin immunoprecipitation assays were performed. Chromatin fragments were prepared from 293T cells transfected with plasmid expressing the N protein and immunoprecipitated with specific rabbit anti-N polyclonal antibody. The COX-2 promoter region (−502 to −2) containing NF-κB and C/EBP binding sequences in the chromatin precipitates was amplified by PCR using specific primers (CP1 and CP2) (Fig. 4A, lane 2). The COX-2 promoter region (−243 to −136) containing only NF-κB-A binding sequences, excluding C/EBP site was amplified by PCR using NF-κB-specific primers (NF-κB-CP1 and NF-κB-CP2) (Fig. 4B, lane 2). The COX-2 promoter region (−155 to −2) containing C/EBP binding sequences, excluding NF-κB site was also amplified from anti-N protein precipitation by PCR using C/EBP-specific primers (C/EBP-CP1 and CP2) (Fig. 4C, lane 2). All amplified products from immunoprecipitated DNA were specific for cells transfected with plasmid expressing N protein and were the expected sizes, comparing PCR products from immunoprecipitation with PCR products amplified directly from input DNA (Fig. 4A, lanes 2 and 4; Fig. 4B, lanes 2 and 3; Fig. 4C, lanes 2 and 3). These results indicated that the N protein bound to the C/EBP and NF-κB recognition elements in the COX-2 promoter.

Fig. 4

Determination of interaction between SARS-CoV N protein and COX-2 promoter by chromatin immunoprecipitation (ChIP) assays. 293T cells transfected with empty vector pCMV-Tag2 (lane 1 in A; lane 4 in B and C) or with pCMV-Tag2-N expressing the N protein (lanes 2–4 in A; lanes 1–3 in B; lanes 1–3 in C) were lysed and subjected to ChIP assays. Immunoprecipitated complexes were collected, subjected to PCR amplification, and separated by agarose gel electrophoresis. (A) The COX-2 promoter region (−502 to −2) was amplified by PCR using specific primers (CP1 and CP2). (B) The COX-2 promoter region (−243 to −136) amplified by PCR using NF-κB-specific primers (NF-κB-CP1 and NF-κB-CP2). (C) The COX-2 promoter region (−155 to −2) amplified by PCR using C/EBP-specific primers (C/EBP-CP1 and CP2).

Two regions of SARS-CoV N protein play important roles in the activation of COX-2 gene

For coronaviruses, the N protein (NP) plays an important role during viral packaging, viral core formation, and viral RNA synthesis (Narayanan et al., 2003). To determine the roles of different regions of N protein in the activation of COX-2 gene expression, we carried out a functional analysis of the protein by deletion mutagenesis. Sequential N-terminal deletion mutants of the SARS-CoV N protein were generated by deleting nucleotides from the 5′ end of the N protein gene and inserting the truncated gene into vector pCMV-Tag2 (Fig. 5A). The function of each mutant N gene was evaluated by luciferase assay following transfection of 293T cells with a reporter plasmid carrying the luciferase gene driven by the COX-2 promoter. Results from luciferase activity assays (Fig. 5B) showed that deletion of amino acids from 1 to 61 (NΔ1), 1 to 136 (NΔ2), or 1 to 204 (NΔ3) from the N terminal of the protein, respectively, had no effects on its function in terms of COX-2 gene activation, whereas deletions of amino acids from 1 to 269 (NΔ4) significantly decreased the level of reporter gene expression. Further deletion from 1 to 333 (NΔ5) entirely abolished luciferase activity. These results suggested that sequences located from amino acids from 204 to 269 are essential for the activation of COX-2 gene by N protein.

Fig. 5

Functional analysis of SARS-CoV N protein by sequential deletions. (A) Diagram of N-terminal truncated mutants of the SARS-CoV N protein and determination of functions of N-terminal truncated N proteins by measuring their ability on the activation of the COX-2 promoter. (B) Diagram of C-terminal truncated mutants of the SARS-CoV N protein and determination of functions of C-terminal truncated N proteins by measuring their ability on the activation of the COX-2 promoter. 293T cells were cotransfected with plasmids containing individual mutant N gene expressing different truncated protein and the reporter gene. The effects of each mutant protein on the activation of COX-2 promoter were measured by the luciferase activity assay. pCMV-Tag2 was used as a vector control. Values correspond to an average of at least three independent experiments done in duplicate. Error bars show 1 S.D.

Sequential deletion mutants of SARS-CoV N protein from C terminal were constructed in the same manner as N-terminal deletions, but by deleting nucleotides from the 3′ ends of the gene (Fig. 5B). Plasmid expressing these sequential deleted N genes and the reporter plasmid were cotransfected into 293T cells. Results from luciferase activity assays (Fig. 5B) revealed that reporter gene activity decreased two-fold in the presence of the N protein (CΔ1) in which amino acids from 339 to 422 were deleted. Further deletions (CΔ2, CΔ3, CΔ4, CΔ5, and CΔ6) eliminated the N protein function in the activation of COX-2 gene. These results implicated sequences between amino acids 339 and 422 of the SARS-CoV N protein to fully transactivate transcription.

A putative nuclear localization signal domain of N protein is essential for the activation of COX-2 promoter

Sequence analysis revealed that the N protein of SARS-CoV contains two potential nuclear localization signals, individual short lysine-rich sequences, located from residuals 257 to 263 (KKPRQKR) and from amino acids 362 to 381 (KTFPPTEPKKDKKKKTDEAQ), respectively (Fig. 6A). The second one is a putative bipartite nuclear localization signal that has not been found in N protein from any other known coronavirus.

Fig. 6

Determination of the function of putative nuclear localization signals of the N protein. (A) Diagram and location of the potential nuclear localization signals of the N protein and their deletion mutants. MutN1 with 38-PKQRRPQ-44 deleted; MutN2 with 220-LALLLLDRLNRL-231 deleted; MutN3 with 257-KKPRQKR-263 deleted; MutN4 with 369-KKDKKKK-376 deleted. (B) Functional analysis of deletion mutant N proteins. 293T cells were cotransfected with plasmids carrying genes expressing the mutant N proteins and the reporter vector. The effects of the deletions on the N protein were determined by measuring luciferase activity. pCMV-Tag2 was used as a vector control. Values correspond to an average of at least three independent experiments done in duplicate. Error bars show 1 S.D. (C) Western bolt analysis of N proteins expressed in transfected cells using N protein antibody.

To analyze the function of these sequences in the activation of COX-2 gene, we constructed plasmids expressing four mutant N proteins (MutN1, MutN2, MutN3, and MutN4), in which one of the potential nuclear localization signals deleted by site-directed mutagenesis (Fig. 6A). The plasmids expressing mutant N proteins were cotransfected, respectively, with the reporter gene into 239T cells. Results from luciferase activity analysis showed that MutN1 and MutN4 stimulated expression of from the COX-2 gene promoter to similar levels as wild type N protein, whereas MutN2 and MutN3 abolished N protein activation of the COX-2 gene promoter (Fig. 6B and C). These results demonstrated that amino acids 257–263 containing a short lysine-rich sequence (KKPRQKR), deleted in MutN2, and amino acids 220–231 carrying a short lucine-rich sequence (LALLLLDRLNQL), deleted in MutN3, were essential for N protein activation of the COX-2 promoter.

Discussion

Virus infection stimulates the expression of a number of proinflammatory gene products, including COX-2, inducible nitric oxide synthase (iNOS), and proinflammatory cytokines. COX-2 converts arachidonic acids to prostaglandins, as participate in the modulation of inflammation and tissue damage in response to infection (Murono et al., 2001). Several viruses have been reported to stimulate expression of COX-2, including Epstein-Barr virus, HBV, HIV, HCV, pseudorabies virus, and rotavirus (Bagetta et al., 1998, Lara-Pezzi et al., 2002, Murono et al., 2001, Núñez et al., 2004; Ray & Enquist, 2004; Steer & Corbett, 2003).

SARS-associated coronavirus causes inflammation and tissue damage to the lungs resulting in severe acute respiratory syndrome. However, the molecular mechanisms involved in the viral infection and tissue inflammation are still largely unknown. To determine the correlations between viral infection and lung inflammation, we isolated a SARS-CoV from a SARS patient previously (Zhu et al., 2005). In this study we demonstrated that N protein activates the COX-2 promoter and induces COX-2 protein production in mammalian cells in dosage-dependent manner indicating that N protein was required for the induction of COX-2 gene expression.

It has been suggested recently that N protein is a two-domain protein, with the N-terminal amino acids from 50 to 150 as the RNA-binding domain (Tang et al., 2005). This study revealed that N protein specifically recognized the NF-κB and C/EBP regulatory elements on the COX-2 promoter. In addition, the region from amino acids 204 to 269 at N terminal of the protein appeared to be essential for the activation of COX-2 expression. Based on our results and previously reported findings, it is reasonable for us to suggest that N protein activates COX-2 gene expression by binding either directly or indirectly to NF-κB and C/EBP regulatory elements on COX-2 promoter. Results from electrophoresis mobility shift assay and chromatin immunoprecipitation assays demonstrated that SARS-CoV N protein activated COX-2 gene expression by binding directly to C/EBP and NF-κB regulatory elements on the promoter. It is most likely that amino acids from 136 to 204 of N protein were involved in the protein–DNA or protein–protein binding. This was also in agreement with the finding that through multi-alignment of total nineteen sequences of the coronavirus N proteins including that of SARS-CoV, one conserved structural region at amino acids 81–140 was found to perform critical functions (Wang et al., 2003).

Localization to the nucleolus is a common feature of coronavirus N proteins. This feature helps with disrupt host cell division to promote virus assembly and sequester ribosomes for translation of viral proteins (Wurm et al., 2001). SARS-CoV N protein localized to the cytoplasm and nucleus of insect cells (Ren et al., 2004) and mammalian cells (date not shown). Deletion of the Lys-rich region (257-KKPRQKR-263) in N protein resulted in the loss of function in the activation of COX-2. Because this domain is putative nuclear localization signal, the failure of this mutation to activate COX-2 expression may be due to its inability to target to the nucleus.

We generated two mutations of the N protein (CΔN1 and MutN1), in which a short lysine-rich sequence (362-KTFPPTEPKKDKKKKTDEAQ-381) near the carboxyl terminus and the sequence around it were deleted, respectively. To our surprise, deletions of this domain in N protein had minimal effects on the activation of COX-2 expression relative to full-length N protein. There are at least two explanations for these results. One is that there are two nuclear localization signal sequences in the N protein and the first one 257-KKPRQKR-263 at the N terminal is essential and efficient to transmit the protein into nucleus as demonstrated in this study. The second explanation is that the short lysine-rich sequence (362-KTFPPTEPKKDKKKKTDEAQ-381) and adjacent sequences at the C terminal of the N protein are in fact involved in different functions, such as protein dimerization, rather than nuclear localization.

The N protein has been reported to form a dimmer by self-association (He et al., 2004), activate the AP1 (activator protein 1) signal transduction pathway (He et al., 2003), and induce actin reorganization in COS-1 cells (Sutjit, Liu, Jameel, Chow, & Lal, 2004). It was suggested recently that the C-terminal 169–422 of N protein contains the dimerization domain (Tang et al., 2005). Protein mutation analysis in this study supported that sequences at the C terminal of N protein are most likely involved in protein dimerization, since deletion of this sequence disrupted protein–protein association and resulted in the partially lose of its functions in activation of COX-2 and perhaps other biological activities.

It is well known that C/EBP family has a common structure, an N-terminal transactivation domain, a basic DNA-binding domain, and a C-terminal domain containing a leucine zipper, which allows the homo- or heterodimerization of these factors (Williams, Cantwell, & Johnson, 1991). Our results suggested that the N protein is a three-domain protein, with the N-terminal amino acids 136–204 as the DNA-binding domain, amino acids 257–263 (KKPRQKR) at the middle as the nuclear localization signal domain, and the C-terminal amino acids 169–422 as the dimerization domain.

Studies have shown that CCAAT/enhancer binding protein, cyclic-AMP response element binding protein (CREB), as well as NF-κB was commonly or individually involved in the regulation of COX-2 gene (Rossen, Bouma, Raatgeep, Büller, & Einerhand, 2004; Thomas et al., 2000; Wardlaw, Zhang, & Belinsky, 2002; Williams et al., 1991; Zhu, Saunders, Yeh, Deng, & Wu, 2002). COX-2 gene is regulated through interactions between NF-κB and C/EBP factors (Thomas et al., 2000). This study demonstrated that both NF-κB site and C/EBP site were involved in the activation of COX-2 by the N protein of SARS-CoV. The ability of virus infection to activate multiple signaling cascades (such as PKR, MAPK, iPLA2, and NF-κB) and how these pathways are integrated in the regulation of individual target gene expression have been discussed previously (Steer & Corbett, 2003). However, the details of molecular mechanisms involved in the activation of COX-2 regulated by the N protein need further investigation.