Characterization of interferon-alpha 13, a novel constitutive murine interferon-alpha subtype.

Type-I interferons (IFNs), also called IFNs-alpha/beta, are a family of cytokines induced by viral infection and are primarily involved in antiviral defense of the cells. IFNs-alpha/beta were also reported to be produced constitutively at low levels in mouse and human cells. These so-called endogenous or constitutive IFNs are thought to exert important homeostatic functions in the uninfected host. By searching IFN genes that were not repressed by the leader protein of Theiler's virus, we identified three uncharacterized IFN-alpha genes that are constitutively expressed in uninfected mouse cells, in vitro and in vivo. Two of these genes corresponded to pseudogenes and were tentatively called IFN-alpha(psi2) and IFN-alpha(psi3). IFN-alpha(psi2) transcripts are the most abundant IFN-alpha transcripts detected in several mouse organs in the absence of viral infection. The third gene codes for a new IFN-alpha subtype called IFN-alpha13, which exhibits acid-stable antiviral activity against Theiler's virus, Mengo virus, and vesicular stomatitis virus. IFN-alpha13 displays unusual characteristics, suggesting that it might have a particular function. Firstly, it is transcribed constitutively, independent of viral infection and of interferon regulatory factor-7 induction. Secondly, it contains two N-glycosylation sites, in contrast to other murine IFN-alpha subtypes that contain either one or no N-glycosylation site. In addition to the genes described here above, several other IFN-alpha subtype genes, including a new gene (IFN-alpha14), were expressed in tissues of uninfected mice. In contrast to IFN-alpha13, IFN-alpha14 was found to lack N-glycosylation and have its expression induced in response to viral infection.

However, accumulating experimental evidence shows that low levels of IFN-␣ and IFN-␤ are constitutively expressed in mouse and humans (14,15), notably during embryogenesis (16,17). This finding suggests that in addition to their role as inducible inflammatory mediators, interferons exert important homeostatic functions in the uninfected host. A variety of regulatory functions have been proposed for constitutive IFN-␣/␤ including regulation of hematopoietic cell growth (18), MHC-I class-I antigens expression, dendritic cell maturation (19,20), or bone formation (21). Gresser and Belardelli (22) have investigated the antitumoral role of constitutive IFN-␣/␤. More recently, Takaoka and coworkers (23)(24)(25)(26) have elegantly shown that low levels of constitutive IFN-␣/␤ are essential for effective IFN-␥ and IL-6 signaling and for maintaining cells ready to switch on the antiviral response. However, the precise nature of the constitutive IFN-␣ in murine cells has never been determined.
According to current models, induction of interferon transcription occurs as a two-step mechanism. In the mouse, IFN-␣4 and IFN-␤, the so-called "immediate-early interferons," can be produced by naive cells upon viral infection. Transcription of these immediate-early interferons involves phosphorylation, dimerization, and subsequent nuclear translocation of a constitutively expressed transcription factor called interferon regulatory factor-3 (IRF-3) (27). In contrast to IFN-␣4 and IFN-␤, the other IFN subtypes can only be produced in cells that have been "primed" through the binding of IFNs-␣/␤ on their surface receptor. Priming induces the production of IRF-7, a transcription factor which upon viral infection will participate in the transcriptional activation of late IFN genes along with IRF-3 (28).
We have previously shown that in murine fibroblasts infected with Theiler's murine encephalomyelitis virus (TMEV or Theiler's virus), IFN-␣ was transcribed despite the inhibition of the immediate-early IFN production by the leader peptide en-□ S The on-line version of this article (available at http://www.jbc.org) contains Supplemental Tables A-C.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  coded by this virus (29). This result suggested that contrary to the predictions of the model, certain IFN-␣ subtype genes could be transcribed independently of the autocrine or paracrine loop mediated by the immediate-early IFNs.
We report here the identification and the characterization of such genes. Two of these genes correspond to pseudogenes. The third gene encodes a novel IFN-␣ subtype that we tentatively called IFN-␣13. This gene appears to be constitutively transcribed in vitro as well as in several mouse tissues, suggesting that it might correspond to one of the constitutive IFN-␣ gene. Interestingly, IFN-␣13 displayed an unusual glycosylation pattern, suggesting that it could play a particular role as a homeostatic cytokine produced at very low levels by various cell types. We also investigated the nature of other IFN-␣ subtypes expressed in vivo in the absence of viral infection. Among these subtypes, we identified another undescribed murine IFN-␣ subtype that we called IFN-␣14.

EXPERIMENTAL PROCEDURES
Viruses and Cell Culture-Viruses used in this study were Theiler's virus DA1 and its L cys mutant, TM598, carrying two Cys to Arg mutations, disrupting the zinc finger of the L protein (29,30). The KJ6 and TM659 are the corresponding wild-type and L cys -mutant viruses carrying a capsid adapted to infect L929 cells (29,31). These viruses were produced as described previously by transfection of BHK-21 cells with viral RNAs transcribed in vitro from the corresponding infectious cDNA clones, pTMDA1, pTM598, pKJ6, and pTM659 (29). An attenuated strain of Mengo virus was produced in a similar way from the pMC16 infectious clone (32). VSV (Strain Indiana) was a gift from Eliane Meurs (Pasteur Institute, Paris, France). Viruses were titrated on BHK-21 cells by a standard plaque assay. BHK-21, COS-7, BALB/3T3, and L929 cells were cultured as described previously (29).
RNA Extraction and RT-PCR-For the detection of cytokine mRNA, total RNA was extracted from cells using either the technique of Chomczynski and Sacchi (33) or the Microprep® kit (Stratagene). Because the interferon genes are intronless, RNA samples (5-10 g) were additionally treated once or twice with 20 units of DNase I (fast protein liquid chromatography-pure, Amersham Biosciences) prior to RT-PCR as described previously (34). RT-PCR reactions were performed with and without reverse transcriptase (Superscript-II, Invitrogen) to rule out that PCR products were amplified from genomic DNA contamination. For cloning purposes, enzymes used for amplification were Pfu polymerase (Promega) or Expand high fidelity (Roche Applied Science). For analytical purposes, Taq polymerase (Promega) or Dynazyme (Finnzymes, Thermo-life Science) were used. Primers used for PCR are shown in Supplemental Tables A-C. Primers used for IRF-7 amplification were as described by Marié et al. (28). Primers for total IFN-␣ and IFN-␤ were as described by Chinsangaram et al. (35), and primers for IFN-␣4 were as described by Deonarain et al. (36).
IFN-␣ Subtype Expression Profiling-Total IFN-␣ was amplified by RT-PCR using consensus primers for IFN-␣ (TM235-236). The fragment amplified with Dynazyme (Finnzymes, Thermo-life Science) was gel-purified and subsequently cloned by TOPO-TA cloning (Invitrogen). A series of clones containing an insert of the expected size was then sequenced with the CEQ sequencing kit (Beckman) using a Beckman CEQ2000 8-capillary sequencer. The identity of the cloned IFN fragments was then determined by BLAST homology search.
Determination of New Murine Gene Sequences and Multiple Alignments-Using the BLAST algorithm (37), we identified fragments from the Whole Mouse Genome Sequence (WGS) data base that aligned perfectly with our clones. The WGS fragments were further assembled using the MULTIALIN software (38) into four new IFN-␣ gene sequences tentatively named IFN-␣(2), IFN-␣(3), IFN-␣13, and IFN-␣14.
Cloning of Constructs Expressing IFN-␣ Subtypes-The coding sequence of IFN-␣4, IFN-␣5, IFN-␣6T, IFN-␣13, and IFN-␣14 were amplified by PCR from 129/Sv mouse genomic DNA and cloned into the pcDNA3 vector (Invitrogen) downstream from the immediate-early promoter of the human cytomegalovirus (CMV). Cloning was performed using BamHI and XhoI restriction sites introduced in the PCR primers. Constructs were done in such a way that all of the clones contained identical sequences upstream and downstream from the IFN-coding sequences. The sequence from the BamHI restriction site to the second codon of the IFN reading frame was 5Ј-GGATCCACCATGGCT-3Ј. In this sequence, the CCACCATGG motif and in particular the underlined nucleotides (purine at position Ϫ3 and G at position ϩ4) match a bona fide consensus defined by Kozak for efficient translation initiation. For IFN-␣13, a second construct called pcDNA3-IFN-␣13 or was made in which the original translation initiation context of the gene was retained. In this clone, the sequence from the BamHI site to the second codon is 5Ј-GGATCCTTTTTAATGGCT-3Ј. In that sequence, the nucleotide in position Ϫ3 is a T instead of a purine. All of the clones were sequenced to confirm that no unexpected mutation occurred in the cloned fragment during the PCR amplification and cloning steps.
Metabolic Labeling and Glycosylation Analysis-COS-7 cells seeded in 6-well plates were transfected with the different IFN-␣-expressing constructs using FuGENE 6 reagent (Roche Applied Science). 24 h after transfection, cells were washed twice and incubated for 24 h in 1 ml of methionine-and cysteine-deficient Dulbecco's modified Eagle's medium (ICN) containing one-tenth of the normal methionine concentration (3 mg/liter) and 70 or 80 Ci/ml 35 S-labeled methionine-cysteine mixture (Redivue TM Pro-mix TM , Amersham Biosciences). After a 24-h incubation at 37°C, supernatants were collected, centrifuged at 15,000 ϫ g to remove cell debris, and stored at Ϫ70°C.
Antiviral Assay-Interferon antiviral activity was quantified by a cytopathic effect reduction assay adapted from Meager (40), performed in 96-well plates on BALB/3T3 cells. Cells were seeded in 96-well plates at a density of 1.5 10 4 cells/well. After 24 h, cells were incubated for an additional 24 h with 2-fold serial dilutions of the IFN-containing supernatants to be tested. Cells were then infected with Mengo virus at a multiplicity of infection of 0.5 pfu/cell or with VSV at 100 pfu/well. Cytopathic effect was monitored by optical microscopy, and samples were compared at the time cells incubated without IFN were lysed. The well in which cytopathic effect was the closest to 50% was used to compare the relative antiviral activities of the IFN-␣ subtypes. IFNs to be compared were always tested in parallel. When necessary, a commercially available IFN-␤ preparation (PBL Biomedical Laboratories) of known titer was added to the series to calculate IFN activity.
Accession Numbers-Sequences were deposited in GenBank TM or in the third party annotation database subset of GenBank™ under the following accession numbers:

Identification of IFN-␣ Genes Transcribed
Independently of the Immediate-early IFNs-We observed previously that in murine L929 fibroblasts infected with Theiler's virus, total IFN-␣ was transcribed despite the inhibition of immediate-early IFN production by the leader protein (L protein) of this virus (29). This observation suggested that, contrary to the predictions of actual models, some IFN-␣ subtypes could be expressed in the absence of autocrine or paracrine activation by immediateearly IFNs.
To identify the nature of such IFN-␣ subtypes, total RNA was extracted from L929 or BALB/3T3 cells infected for 7 h with Theiler's virus expressing the wild-type L protein. Total IFN-␣ sequences were then amplified by RT-PCR, cloned, and sequenced. To our surprise, a majority of clones did not corre-spond to any of the known IFN-␣ sequences deposited in Gen-Bank TM but their nucleotide sequences shared approximately 90% identity with that of IFN-␣1. We identified fragments from the WGS data base that aligned perfectly with our clones. The WGS fragments were further assembled into three new IFN-␣ gene sequences that subsequently appeared in the supercontig of chromosome 4 (GenBank™ accession NT_039262).
Two of these sequences corresponded to pseudogenes and were tentatively named IFN-␣(2) and IFN-␣(3). The third gene showed a complete open reading frame coding for a new IFN-␣ subtype that we tentatively named IFN-␣13.
Structure of the IFN-␣13, IFN-␣(2), and IFN-␣(3) Genes-The IFN-␣(2) region lacks the AUG codon and contains multiple deletions and one insertion, disrupting the potential coding sequence (Fig. 1A). The IFN-␣(3) sequence contains one G to A mutation at nucleotide 50 that interrupts the open reading frame at codon 17. It contains two additional point deletions that create further frameshifts.
The IFN-␣13 gene codes a putative 189 amino acid-long protein containing a predicted signal sequence of 23 amino acids. The amino acid sequence aligns perfectly with that of other IFN-␣ subtypes. It shares 87% identity with IFN-␣1 (Table I).
The putative promoter sequences of the IFN-␣13, IFN-␣(2), and IFN-␣(3) genes contain potential virus-responsive element (VRE) in similar positions relative to those of classical IFN-␣ genes. However, in the three promoters, point mutations occurred in the modules known to regulate virus inducibility (Fig.  1B). No TATA box aligning with that of other IFN-␣ genes was detected in the putative promoters of the two pseudogenes, and no polyadenylation signal was found in their 3Ј-non-coding region. In contrast, the IFN-␣13 sequence contains a TATA box and a polyadenylation signal aligning with those of other IFN genes. In contrast to all of the IFN-␣ genes sequenced so far, the nucleotide sequence preceding the AUG codon of IFN-␣13 does not match the optimal Kozak consensus. In particular, the nucleotide in position Ϫ3 is a T, whereas it would be expected to be a purine. None of the three genes possesses the consensus AU-rich elements (ARE) sequence dictating mRNA degradation that is found in some of the other IFN-␣ sequences (Fig. 1C) (41).
Expression of IFN-␣ Subtypes in L929 Cells-To determine the  conditions in which IFN-␣13 and the two pseudogenes are expressed, we compared the proportions of the different IFN-␣ subtypes transcribed in infected and uninfected L929 cells (Fig. 2).
Interestingly, IFN-␣13 and IFN-a(3) transcripts were found in similar proportions in uninfected cells (Fig. 2C) and in cells infected with the wild-type virus ( Fig. 2A), suggesting that the IFNs produced in the presence of the leader protein corresponded to constitutively expressed IFNs.
We next analyzed the profile of IFN-␣ subtypes expressed in cells primed with IFN-␣/␤ to evaluate the influence of IRF-7 activation on the transcription of these genes (Figs. 2 and 3). RT-PCR analysis confirmed transcriptional activation of IRF-7 in response to priming (Fig. 3). In unprimed cells, IFN-␣4 and  (Fig. 2, D and E). Following IRF-7 induction, a panel of classical IFN-␣ subtypes was expressed. IFN-␣4 was the predominant subtype induced in primed cells following infection with both the wild-type and the L-mutant viruses. The proportion of IFN-␣13 or of IFN pseudogenes transcripts strongly decreased after priming, showing that these genes are unresponsive to IRF-7 contrary to the other IFN-␣ subtype genes.
In uninfected cells, IFN-␣13 and IFN-␣(2) promoters did not show any activity above base line compared with a promoterless control (data not shown), suggesting very low constitutive expression of these promoters.
To evaluate virus inducibility of the promoters, luciferase activity was measured 24 h after infection with either wild-type or mutant viruses (Fig. 4A). IFN-␣4 promoter activity increased at least 10-fold when the cells were infected with the L cys virus but not with the wild-type virus, encoding a functional leader peptide. The mutant virus also induced the transcription from the IFN-␣5 and the IFN-␤ promoters more than the wild-type virus did. However, the activation of these promoters was of lower magnitude (2-4-fold), possibly because these promoters had a higher basal transcription rate than the IFN-␣4 promoter in uninfected cells. Inhibition of transcription by the leader protein was specific for IFN promoters, because it was not observed with a luciferase expressed under the control of the SV40 early promoter.
Results were further confirmed using Mengo virus as viral inducer. In the conditions used, a typical promoter responsive to IRF-7, such as the IFN-␣5 promoter, was highly induced (Fig. 4B). Again, neither IFN-␣13 nor IFN-␣(2) promoter was activated by viral infection.
Taken together, our data indicate that inhibition of IFN production by the Leader peptide of Theiler's virus occurs at least in part at the transcriptional level. The putative promoters of the IFN-␣13 and IFN-␣(2) genes are weak and are neither induced nor repressed by viral infection. This is in agreement with the fact that point mutations affect the virusresponsive elements found in these promoters (Fig. 1B).
In Vivo Expression of Constitutive IFN-␣ Subtypes-In view of the constitutive expression of IFN-␣13 in vitro, we wondered whether this IFN subtype might correspond to a so-called endogenous or constitutive IFN expressed in vivo in uninfected mouse tissues. Therefore, we extracted the RNA from organs of two uninfected 129/Sv mice kept under specific pathogen-free conditions and we analyzed the expression profile of IFN-␣ subtypes expressed in the absence of viral infection by the RT-PCR-cloningsequencing procedure used above. We also analyzed the transcription profile of IFN-␣ subtypes expressed in the organs of two IFNAR-1Ϫ/Ϫ mice (42) lacking the IFNAR-1 subunit of the in-terferon receptor. The absence of IFN receptor avoids potential activation of IFN genes by the positive feedback loop mediated by IRF-7 after the binding of IFNs to their receptor (Table II).
Interestingly, in both wild-type and IFNAR-1Ϫ/Ϫ mice, the majority of the constitutive IFN-␣ transcripts corresponded to IFN-␣(2), one of the new pseudogene identified in vitro. IFN-␣13 was also detected, albeit in low proportions, in the thymus, spleen, kidney, and spinal cord, confirming constitutive low expression of this subtype independently of the autocrine loop mediated by the IFN-␣/␤ receptor. Other IFN-␣ mRNA species were also detected in some organs (Table II). Their proportion was notably increased in some spleen, thymus, and spinal cord samples. Finally, a previously undescribed IFN-␣ subtype, tentatively named IFN-␣14, was detected in the thymus of wild-type and IFNAR-1Ϫ/Ϫ mice and in the spleen of IFNAR-1Ϫ/Ϫ mice.
Expression of IFN-␣ in Freshly Isolated Splenic Dendritic Cells- Montoya et al. (20) recently reported the role of constitutive IFN-␣/␤ in the maturation of dendritic cells. As constitutive IFN-␣13 expression was detected in the spleen and thymus of 129/Sv IFNAR-1Ϫ/Ϫ mice, we were interested in determining whether IFN-␣13 was specifically expressed in dendritic cells. RNA was therefore extracted from splenic dendritic cells, which were freshly isolated from C57BL/6 mice treated with FLT3-ligand (kindly provided by Laurent Fraga and Muriel Moser, Free University of Brussels, Gosselier, Belgium). RT-PCR reactions performed with specific primers TM333 and TM334 detected expression of IFN-␣13 in this cell population (data not shown). The identity of the PCR product was confirmed by direct sequencing. However, IFN-␣13 expression was very low and was not detected in these cells by the RT-PCR-cloning-sequencing procedure (0 of 21 clones analyzed). The IFN-␣ transcripts detected in these cells by this way were essentially from the pseudogene IFN-␣(2) (19 of 21 clones). One clone corresponded to IFN-␣4, and one corresponded to IFN-␣5.
Characterization of IFN-␣13-To check whether IFN-␣13 had anti-viral activity, the coding sequence of this gene was cloned from 129/Sv mouse DNA in the pcDNA3 vector under the control of the CMV promoter. The constructed plasmid was named pcDNA3-IFN-␣13. Equivalent plasmids expressing IFN-␣4, IFN-␣5, and IFN-␣6T were constructed as controls.
The IFN-expressing constructs were transiently transfected into COS-7 cells, and culture supernatants were collected 48 h post-transfection. Anti-viral activities were determined by a cytopathic effect reduction assay. Control supernatants issued from cells transfected with the pcDNA3 vector alone did not exhibit any detectable anti-viral activity. In contrast, the supernatants issued from cells transfected with the IFN-␣expressing constructs, including IFN-␣13, showed protective anti-viral activity (up to 50,000 IU/ml) against Theiler's virus, Mengo virus, and VSV.
We then tested pH 2 stability of IFN-␣13. Therefore, the supernatants containing IFN-␣13 were treated at pH 2 for 24 h as described previously (29) and used in the anti-viral assay in parallel with untreated supernatants. IFN-␣13 turned out to be stable at pH 2 because no loss in anti-viral activity was detected in treated supernatants as compared with untreated samples (data not shown).
As shown in Fig. 1, the sequence of the IFN-␣13 gene is unusual in that the sequence context of the AUG codon used for translation initiation does not match the consensus defined by Kozak. To test the influence of this region on the expression of IFN-␣13, we compared IFN-␣13 production by cells transfected with pcDNA3-IFN-␣13 or with pcDNA3-IFN-␣13 OR , an equivalent plasmid that retained the original suboptimal context of the AUG codon. Unexpectedly, supernatants issued from cells transfected with the IFN-␣13 OR construct did not clearly show less antiviral activity than the supernatants issued from cells transfected with the IFN-␣13 construct carrying an optimal initiation context (data not shown). The amount of 35 S-labeled IFN present in supernatants of transfected cells was also quantified by PhosphorImager analysis after SDS-PAGE (data not shown and Fig. 5). The amount of IFN-␣13 secreted by cells transfected with the IFN-␣13 OR construct was again Ͻ2-fold lower than the amount produced by cells expressing the construct with an optimal initiation context.
Characterization of IFN-␣14 -In the experiment described above (Table II), a yet undescribed IFN-␣ subtype tentatively named IFN-␣14 was found to be expressed in the thymus and in the spleen of uninfected mice. The sequence of the gene was assembled from raw data from the mouse genome-sequencing project as described above for IFN-␣13. The IFN-␣14 coding sequence was then cloned from 129/Sv mouse genomic DNA in the pcDNA3 vector and sequenced. It encodes a predicted protein of 189 amino acids with a putative signal sequence of 23 amino acids (Fig. 5D). Like IFN-␣13, IFN-␣14 turned out to have pH 2-stable anti-viral activity (data not shown).
However, unlike IFN-␣13, the transcription of IFN-␣14 was found to be up-regulated in L929 cells in response to Newcastle disease virus (NDV) infection (data not shown). This observation is in agreement with the fact that the putative promoter of the IFN-␣14 gene contains an IRF-7 binding sequence identical to that found in the inducible IFN-␣4 and IFN-␣5 promoters (Fig. 1B). As in the case of IFN-␣4 and IFN-␣5 genes, the 3Ј-non-coding region of the IFN-␣14 gene contains a typical ARE (Fig. 1C).
Presence of Two N-glycosylation Sites in IFN-␣13-To analyze the glycosylation status of the cloned IFN subtypes, COS-7 cells were transfected with the constructed plasmids expressing the different IFN subtypes. 35 S-Labeled proteins secreted by transfected COS-7 cells were separated by SDS-PAGE and analyzed (Fig. 5, A-C). The bands corresponding to the various IFN-␣ subtypes were identified by comparison with the pcDNA3 control lane. The data shown in Fig. 5 nicely fit Nglycosylation predictions made on the basis of the amino acid sequences (Fig. 5D). Indeed, IFN-␣6T (43) and IFN-␣14 lack any predicted site for N-glycosylation and migrate at ϳ18 kDa as expected for non-glycosylated IFN molecules. IFN-␣4 and IFN-␣5 migrate at ϳ24 kDa. Accordingly, the sequences of these IFN subtypes contain a single potential N-glycosylation site at Asn-78 of the sequence. Interestingly, IFN-␣13 showed a unique profile on SDS-PAGE by forming multiple bands at approximately 26 -30 kDa. This finding correlates with the presence of two putative N-glycosylation sites in this IFN, one site at position 78 aligning with those of IFN-␣4 and IFN-␣5 and one site found at position 71 in the IFN-␣13 sequence, which is lacking in other IFN-␣ subtypes (Fig. 5D). Because migration of IFN-␣13 on SDS-PAGE was slower than that of IFN-␣4 or IFN-␣5, it is likely that both Asn-71 and Asn-78 residues of IFN-␣13 carry N-linked sugars. Following N-glycosidase treatment, IFN-␣4, IFN-␣5, and IFN-␣13 migrated at ϳ18 kDa, showing that they were indeed N-glycosylated. In contrast, we found no evidence for O-glycosylation of these IFNs (Fig. 5C).

DISCUSSION
The Leader Protein of Theiler's Virus Inhibits IFN-␣/␤ Transcriptional Activation-We showed previously that the leader protein of Theiler's virus inhibited immediate-early IFN production in infected L929 cells (29). This work confirms these observations and shows that production of other classical IFN-␣ subtypes is inhibited by the leader protein as well. Indeed, in cells infected with the L-mutant virus, IFN-␣4, IFN-␣5, or IFN-␣6 transcripts represented 7.2, 2.9, and 2.9% IFN-␣ transcripts, respectively, whereas transcripts of these IFN subtypes were not detected in cells infected by the wild-type virus (0 of 66 sequences). Inhibition of late IFNs production might be due to the absence of the immediate-early IFNs as predicted by current IFN induction models. The leader protein does not, however, block the low constitutive transcription of IFN-␣13 or that of the pseudogenes.
Using a gene reporter assay, we showed that the leader protein inhibited IFN production at the transcriptional level. However, this does not rule out the possibility for additional post-transcriptional effects.
Unique Characteristics of IFN-␣13-Like other characterized IFN subtypes, IFN-␣13 was found to be stable at pH 2 and to display antiviral activity protective against Theiler's virus, Mengo virus, and VSV. However IFN-␣13 possesses some unique characteristics among the IFN-␣ gene family.
Firstly, unlike other IFN genes, IFN-␣13 transcription was not up-regulated by IRF-7. Accordingly, in the luciferase reporter assay, the putative promoter of IFN-␣13 was not induced by viral infection. These observations are in agreement with the fact that the putative promoter of IFN-␣13 contains a point mutation in the predicted IRF-7 binding sequence (Fig. 1B).
However, we cannot rule out the possibility that other factors  could regulate IFN-␣13 expression. For instance, in L929 cells but not in BALB/3T3 cells, transcription of IFN-␣13 reproducibly appeared to be up-regulated in response to TMEV infection ( Fig. 3) (29). In addition, using RT-PCR, we observed that at least part of the IFN-␣13 transcripts was driven by a promoter located at least 480 nucleotides upstream of the IFN-␣13 coding sequence. Secondly, the nucleotides preceding the translation initiation site do not match the Kozak consensus (44,45), suggesting that IFN-␣13 could have evolved to maintain low levels of expression. However, we failed to detect a clear effect of the initiation context in transfection experiments, possibly because experiments were done in COS cells, which overproduce IFN mRNA and might render the availability of ribosome-limiting.
Thirdly, IFN-␣13 showed a unique N-glycosylation profile that correlates with the presence of two putative N-glycosylation sites: one site conserved in IFN-␣4 and IFN-␣5 at position 78 and one site at position 71, which is unique to IFN-␣13 among murine IFN-␣ subtypes. The prediction of the IFN-␣13 structure, based on the solved human IFN-␣2b structure, indicates that the two sugar moieties of the IFN-␣13 molecule are clustered on a side of the molecule opposite to the putative IFN-receptor binding site (Fig. 5E). Thus, glycosylation is unlikely to influence binding activity. It might enhance IFN-␣13 stability in vivo. The constitutive transcription and the un-usual post-translational modification of IFN-␣13 suggest that the role of this IFN could be partly distinct from that of the virus-induced IFN-␣ subtypes.
IFN-␣14 -In the course of this study, we identified an additional subtype of murine IFN-␣ that we tentatively called IFN-␣14. This IFN shares between 78 and 89% identity with the other known IFN-␣ subtypes at the protein sequence levels (Table I). IFN-␣14 was found to display pH 2 stability and antiviral activity. Although it was detected in tissues of uninfected mice, expression of this IFN gene was found to be up-regulated in response to viral infection. Unlike IFN-␣13, IFN-␣14 was found to lack N-glycosylation. The finding of this new IFN subtype further raises the question of the reason for the multiplicity of IFN-␣ genes present in the mouse and human genomes.
Identification of IFN-␣ Genes Transcribed in Uninfected Mice-Constitutive transcription of IFN-␣/␤ has previously been described both in organs and specific cell types. By S1 nuclease mapping, Tovey et al. (14) have identified IFN-␣1 and IFN-␣2 transcripts in human spleen, kidney, and liver. Hida et al. (46) have shown by RT-PCR the presence of IFN-␣/␤ transcripts in several mouse organs (46). At the cellular level, mouse macrophages, dendritic cells, and murine embryonic fibroblasts (MEFs) were found to express constitutive IFN-␣/␤ (20, 23, 47-49). However, the identity of the IFN-␣ subtypes expressed in these cells has not been determined. We found that the IFN-␣ gene most prominently expressed in uninfected mice is the IFN-a(2) pseudogene. Interestingly, a human IFN-␣ pseudogene, LeIFN E, was also identified in a myeloblastoïd cell line cDNA library, indicating that it also retained some transcriptional activity (50). Despite their inability to encode a functional protein, several pseudogenes have nevertheless been reported to be transcribed (51). For example, the PTEN pseudogene transcript was reported to be more abundant in the liver than its functional paralogous transcript (52).
Although IFN-␣13 represented the majority of the transcripts detected in uninfected L929 cells, it only represented a small proportion of the transcripts found in dendritic cells or in tissues of uninfected mice. In these samples, IFN-␣(2) was the most abundantly transcribed subtype. This finding suggests that the relative amounts of the constitutive IFN transcripts could vary according to the cell type.
Beside IFN-␣13 and IFN-␣(2), several IFN-␣ subtypes, including IFN-␣14, were found to be expressed at low levels in tissues of uninfected mice. In contrast to IFN-␣13 and pseudogenes, these classical IFN genes are known to have inducible transcription, notably in response to viral infection.
We noticed that in a few samples, "classical" IFN-␣ subtypes constituted up to 50 or 80% of the IFN transcripts detected. Our interpretation is that an unidentified stress stimulus slightly activated the transcription of inducible IFN genes in these organs. This does not rule out the fact that such classical IFN genes could also be expressed at low levels and play a role in uninfected tissues. The contrast between IFN-␣13 and these genes is that IFN-␣13 appears to have evolved to have its expression maintained at low levels. Viral unresponsiveness might also guarantee low but stable IFN production in conditions where transcription of classical IFNs is blocked by interfering viral proteins like the leader protein of Theiler's virus.
IFNAR-1-independent Expression of Interferons-We observed that a varied panel of IFN-␣ genes were transcribed in IFNAR-1Ϫ/Ϫ mice. Transcription of these IFN genes thus occurred in the absence of the autocrine feedback loop mediated by the immediate-early IFNs. This observation is in concordance with the data obtained by Hata et al. (23) showing IFN-␣ transcription in uninfected IFNAR-1Ϫ/Ϫ MEFs at levels comparable to that seen in uninfected wild-type cells. Significantly, MEFs deficient for both IRF-3 and IRF-9 still transcribe constitutive IFN-␣, whereas induction of IFN-␣/␤ following NDV infection is abolished in these cells. Taken together, these results show that constitutive IFN transcription is independent of the factors that regulate virus inducibility of the IFN-␣ promoters.
In uninfected IFNAR-1Ϫ/Ϫ mice, IFN-␣ mRNA are transcribed in higher proportion in lymphoid organs (spleen and thymus) but surprisingly also in spinal cord. This could be related to the constitutive expression of IRF-7 and IRF-5 in specific cell types. For instance, several reports have shown constitutive expression of IRF-7 in lymphoid organs (53), and expression of IRF-5 in dendritic cells (54).