Identification of CRG-2 AN INTERFERON-INDUCIBLE mRNA PREDICTED TO ENCODE A MURINE MONOKINE*

In order to identify novel proteins produced by activated macrophages, a cDNA library was made from cultures of the mouse macrophage-like cell line RAW 264.7 that had been treated with conditioned medium from mitogen-stimulated spleen cells, and the library was screened by differential plaque hybridization. A cDNA clone was isolated that detected a 1.4-kilobase mRNA that accumulated dramatically in response to the spleen cell conditioned medium. The 1.4-kilobase mRNA encodes a predicted protein of 98 amino acids, designated CRG-2, molecular weight (Mr) 10,781, with a 21-residue signal peptide. The amino acid sequence indicates that CRG-2 is a member of the platelet factor 4 family (PF4) of cytokines. The CRG-2 mRNA was induced by alpha-, beta-, and gamma-interferons (IFNs) and by lipopolysaccharide. In response to IFN-gamma, the CRG-2 mRNA level reached a peak between 3 and 6 h. The accumulation of CRG-2 mRNA was not blocked by cycloheximide. Among the known members of the PF4 family, CRG-2 is most closely related to the interferon-inducible human protein IP-10. The 5'-flanking region of the crg-2 gene was isolated, and comparisons between crg-2 and IP-10 genes, mRNAs, and proteins reveal conserved features of possible functional importance. CRG-2 may play a role in host defense, particularly in the response to viral infection.

In order to identify novel proteins produced by activated macrophages, a cDNA library was made from cultures of the mouse macrophage-like cell line RAW 264.7 that had been treated with conditioned medium from mitogen-stimulated spleen cells, and the library was screened by differential plaque hybridization. A cDNA clone was isolated that detected a 1.4-kilobase mRNA that accumulated dramatically in response to the suleen cell conditioned medium. The 1.4-kilobase mRN*A encodes a predicted protein of 98 amino acids, designated CRG-2, molecular weight (Mr) 10,781, with a 21-residue signal peptide. The amino acid sequence indicates that CRG-2 is a member of the platelet factor 4 family (PF4) of cytokines. The CRG-2 mRNA was induced by CY-, /S, and y-interferons (IFNs) and by lipopolysaccharide.
In response to IFN-7, the CRG-2 mRNA level reached a peak between 3 and 6 h. The accumulation of CRG-2 mRNA was not blocked by cycloheximide.
Among the known members of the PF4 family, CRG-2 is most closely related to the interferoninducible human protein IP-10. The 5'-flanking region of the erg-2 gene was isolated, and comparisons between erg-2 and IP-10 genes, mRNAs, and proteins reveal conserved features of possible functional importance. CRG-2 may play a role in host defense, particularly in the response to viral infection.
Activated macrophages have an enhanced ability to perform many functions including phagocytosis, antigen presentation, and pathogen and tumor cell killing. Macrophages can be activated by factors such as interferons (IFN)' (l-3), granu- locyte/macrophage colony-stimulating factor (GM-CSF) (4), tumor necrosis factor (TNF) (5), , , lipopolysaccharide (LPS) (a), muramyldipeptide (9), and target cellbound IgG (10). The expression by macrophages of the activated phenotype is due in large measure to the expression of genes induced by macrophage activating factors. Examples of genes induced by macrophage activators include: those for the major histocompatibility class II antigens (11,12) and for the Fe-y receptor-a induced by ; the IL-6 gene induced by ; the FCC-receptor gene induced by ; the TNF gene induced by , LPS (16), and IFN--y (17); the IL-1 gene induced by LPS (18) and ; and the p-IFN gene induced by viruses (19). The products of many of these inducible genes are cytokines and cell surface receptors involved in cell interactions important for immune and inflammatory responses, and some of the inducible products can themselves serve as autocrine macrophage activators.
Recently, several mouse and human macrophage-derived cytokines have been discovered that are members of a family of inducible factors related to platelet factor 4 (PF4) and platelet basic protein (reviewed in Ref. 20). The macrophagederived family members include the LPS-induced chemotactic factors, macrophage inflammatory protein-2 (MIP-2) (21), and  and the IFN-r-induced IP-10 (23). Other members of the family include the melanoma cell growth factor GRO/melanoma growth stimulatory activity (24,25) and 9E3, a protein associated with transformation by Rous sarcoma virus (26). These proteins all contain 4 similarly spaced cysteines, with the first 2 cysteines in a Cys-X-Cys configuration. The Cys-X-Cys proteins are related to another family of cytokines that includes macrophage inflammatory protein-l (MIP-1) and JE (reviewed in Ref. 20) and whose members also contain 4 similarly positioned cysteines but with the first 2 cysteines immediately adjacent.
In order to identify novel mediators synthesized by lymphokine-activated macrophages, we constructed a cDNA library from cultures of the mouse macrophage-like cell line RAW 264.7 (27) that had been treated with conditioned medium from concanavalin A (ConA)-stimulated spleen cells and screened the library by differential plaque hybridization. One of the cDNA clones that was isolated hybridized to a 1.4kb mRNA species induced following stimulation with the spleen cell supernatants and encoded a predicted protein, designated CRG-2, as one of the collection of proteins encoded by cytokine responsive genes. The predicted CRG-2 protein belongs to the PF4 family of cytokines and shows extensive homology with the IFN-y-induced human protein IP-10. We describe below an analysis of the structure and expression of the CRG-2 mRNA as well as an analysis of the 5'-regulatory region of the erg-2 gene. Comparisons between CRG-2 and IP-10 reveal conserved features that may identify structures of CRG-2 of particular functional importance in the erg-2 and IP-10 genes, mRNAs, and proteins.

RESULTS
Isolation of CRG-2 cDNA Clones-A cDNA library in the XgtlO vector was prepared from cultures of the mouse macrophage-like cell line RAW 264.7 that had been treated for 3 h with 20% conditioned medium from ConA-stimulated mouse spleen cells. Initial screening of the cDNA library by differential plaque hybridization led to the identification of a 0.6-kb CRG-2 cDNA clone that hybridized to an mRNA species induced in RAW cells by the spleen cell conditioned medium but not detectable in control RAW cells. Multiple additional CRG-2 cDNA clones were isolated, both in the course of further differential screening and by screening the RAW cDNA library with CRG-2 cDNA probes, and two near full length cDNA clones of approximately 1.1 kb were isolated and designated 1.1-1 and 1.1-2, which were analyzed in greater detail.

Induction of CRG-2 mRNA-On
Northern blot analysis, the CRG-2 cDNA probe identified a major 1.4-kb mRNA induced in RAW 264.7 cells following exposure for 3 h to conditioned medium from ConA-stimulated spleen cells (Fig.  1). Minor inducible species of lower mobility were also detected, presumably corresponding to precursors of the 1.4-kb mRNA. The inducing activity was not limited to a single preparation of spleen cell conditioned medium. Induction of the CRG-2 mRNA was not inhibited by 10 rg/ml CHX added simultaneously with the conditioned medium. Treatment of RAW 264.7 cells with 10 pg/ml CHX resulted in a 93% 1 I) inhibition of protein synthesis at 5 min and a 96% inhibition at 1 h (data not shown), suggesting that new protein synthesis was not required for induction of the CRG-2 mRNA.
To identify the lymphokines in the spleen cell conditioned medium that may have been responsible for inducing the CRG-2 mRNA, as well as to determine which other macrophage-activating factors were capable of enhancing expression of the erg-2 gene, total RNA was prepared from RAW cells treated with a variety of agents and the RNA analyzed by Northern blot. As shown in Fig. 2, the CRG-2 mRNA was strikingly induced by the three IFNs and, although less so, by LPS. In a separate experiment, poly (1) .poly(C) also led to the conspicuous accumulation of the CRG-2 mRNA. While polymyxin B significantly diminished the induction of CRG-2 mRNA by LPS, it had no effect on the induction by the IFNs, suggesting that trace amounts of contaminating LPS were not contributing to the IFN responses. The blot shown in Fig.  2 was also hybridized with an aldolase A cDNA probe, and the aldolase A mRNA signals serve as controls since they remain unchanged in response to all stimuli except poly (1). poly(C). Poly(I).poly(C) led to a decrease in the level of aldolase A mRNA.
In order to analyze the time course of accumulation of the CRG-2 mRNA, RAW 264.7 cells were harvested following treatment with IFN--/ for O-24 h. As shown in Fig. 3, the CRG-2 mRNA was induced rapidly and dramatically, reaching a peak between 3 and 6 h. Treatment of RAW cells with recombinant murine IL-la, recombinant murine IL-3, recombinant murine IL-4, recombinant murine GM-CSF, recombinant human CSF-1, PMA, and the calcium ionophore A23187 all failed to induce the CRG-2 mRNA, nor was the CRG-2 mRNA induced by serum, i.e. mitogen, stimulation of serumstarved Balb/c 3T3 fibroblasts (data not shown).
To demonstrate the induction of CRG-2 mRNA during the activation of normal macrophages, erg-2 gene expression was analyzed in the adherent population from starch-elicited peritoneal exudate cells obtained from both C3HeB/FeJ and BALB/cJ mice. As shown in Fig. 4, when these cells (>80% macrophages as determined by morphology) were exposed to conditioned medium from ConA-stimulated spleen cells, CRG-2 mRNA was induced. In the case of the C3HeB/FeJ, but not the BALB/cJ mice, conditioned medium from unstimulated spleen cells to which ConA had been added also induced the CRG-2 mRNA. In addition, the C3HeB/FeJ cells were treated with IFN-7, and this too led to the expression of the erg-2 gene. The electrophoretic mobilities of the CRG-2 mRNA species from peritoneal macrophages as compared to RAW cells were indistinguishable. Nucleotide and Predicted Amino Acid Sequence of CRG-P-As noted above, two CRG-2 cDNA clones of approximately 1.1 kb were isolated. Clone 1.1-1 extended 5 nucleotides 5' beyond the 5' end of clone 1.1-2. As shown in Fig. 5, these cDNAs consisted of 3 EcoRI fragments which, in the case of cDNA clone 1.1-1, were 562, 320, and 215 bp. A partial EcoRI digest of the phage containing cDNA clone 1.1-1 yielded a full length 1.1-1 cDNA clone, which was inserted in pBlueScript, and both strands were sequenced in their entirety using deletions made on the basis of available restriction sites (Fig.  5) and deletions generated using exonuclease III as noted under "Materials and Methods." In addition, the 557-bp EcoRI fragment of cDNA clone 1.1-2, containing the long open reading frame corresponding to the presumed coding sequence for the CRG-2 protein, was also sequenced and no discrepancies were found between the 1.1-1 and the 1.1-2 sequences.
In order to determine the start site for transcription of the erg-2 gene and determine the 5'-most sequence of the CRG-2 mRNA, erg-2 genomic clones were isolated from a BALB/c mouse library. A 1.2-kb internal Hind111 genomic fragment from a genomic clone was found to hybridize on Southern analysis to the 5' EcoRI-BamHI fragment from cDNA clone 1.1-1 (data not shown) and sequence analysis showed that this genomic fragment extended 5' from the end of the cDNA sequence (Figs. 6 and 10 in the long open reading frame, the predicted CRG-2 protein contains 98 amino acids, molecular weight (M,) 10,781. The Nterminal sequence of the CRG-2 protein includes a stretch of hydrophobic residues (amino acids 5 to 14) that is characteristic of the signal peptide of a secreted or transmembrane protein (54). CRG-2 is unlikely to be a transmembrane protein because it lacks a second long hydrophobic region. The (-3,-l) rule of von Heijne (55) predicts signal peptide cleavage after Gly-21. Outside of the signal peptide there are no predicted N-linked glycosylation sites.
Comparison of the Nucleotide and Predicted Protein Sequence of CRG-2 with Other Sequences-The nucleotide sequence of the CRG-2 cDNA was compared to the sequences in the GenBank (United States Dept. of Health and Human Services) and EMBL Nucleotide Sequence Library data banks using the FASTA program from Genetics Computer Group, University of Wisconsin Biotechnology Center, based on the Lipman and Pearson (56) search for similarity.
Greatest homology was found with the human IP-10 cDNA sequence, with which there was a 68.3% identity in a 546-nucleotide overlap. When the CRG-2-predicted protein sequence was used to search the National Biomedical Research Foundation Protein Library, again the most closely related sequence was that of the human IP-10 protein. Comparisons between CRG-2 and other members of the PF4 family were performed according to Oquendo et al. (57) and are shown in Fig. 8. The CRG-2 predicted protein shares with all these sequences the 4 invariant cysteines, with the first 2 cysteines in the Cys-X-Cys configuration that is characteristic of the PF4 family. Because of the high degree of homology between CRG-2 and IP-10, mRNA and genomic sequences of the two were compared in order to identify conserved elements of possible functional importance.
A dot plot comparison (61) of the CRG-2 and IP-10 cDNA sequences is shown in Fig. 9A. Apart from the high degree of homology in the coding regions of the two cDNAs (residues 66-359 for CRG-2 and residues 68-361 for IP-10) there is significant homology in the 3'-untranslated regions corresponding to nucleotides 401 to 523 of CRG-2 and 406 to 540 of IP-10 immediately 3' of the coding sequence and in the more 3' sequences corresponding to nucleotides 842 to 967 of CRG-2 and 879 to 1012 of IP-10 (Fig. 9B). This latter region is particularly A:T-rich (69% over 126 positions) and contains the sequence TATTTAAT similar to the sequence motifs (62) that have been implicated in the control of the stability of mRNAs of oncogenes and lymphokines (63). As described above and under "Materials and Methods," a 1.2-kb Hind111 erg-2 genomic fragment was isolated, that included the transcription start site and 5' flanking sequence. An oligonucleotide complementary to nucleotides 34 to 51 of the CRG-2 mRNA was used to obtain sequence of the noncoding strand, and an additional oligonucleotide corresponding to positions -296 to -279 of the plus strand was used to obtain the plus strand sequence. This erg-2 genomic sequence along with a comparison of the corresponding sequence from the IP-10 gene is shown in Fig. 10. The erg-2 sequence shows a probable TATA box at -24 and a CAAT box at -60, similar to the positions of these elements in the IP-10 gene.
In addition, previously identified enhancer elements are found in similar positions in the two genes. One of them is the interferon-stimulated response element (ISRE; 5' GGAAAGTGAAACCT 3') found in a series of interferoninducible genes (64). A second enhancer element closer to the CAAT box is 5' GGGAAATTCC 3', which matches the positive regulatory domain II (PRDII) element containing the virus-inducible NF-KB binding site in the @-interferon gene (65), also shown to mediate gene activation by poly (1) .poly(C) (66). A second NF-KB site is located at -113 to -104 of the erg-2 promoter. An AP-1 site (67) seen at -84 to -78 in crg-2 is not conserved in IP-10.

DISCUSSION
We have identified a novel gene, erg-2, that was induced in mouse macrophages following exposure to conditioned medium from mitogen-stimulated spleen cells. The CRG-2 mRNA accumulated rapidly and dramatically in response to 1) -51 1) -3 y-, (Y-, and fi-IFNs, responded less well to LPS, and minimally or not at all to a number of additional cytokines and pharmacologic agents. In response to IFN-7, the CRG-2 mRNA level reached a peak between 3 and 6 h. Induction of CRG-2 mRNA by the spleen cell supernatants and IFN-7 did not require new protein synthesis and therefore CRG-2 is part of the immediate response to IFN-7. The CRG-2 cDNA encodes a predicted protein of 98 amino acids with a 21-amino acid signal peptide that is a member of the PF4 family of cytokines. The sequence identities between the predicted CRG-2 and IP-10 proteins suggest that the two proteins may be mousehuman homologues although, using CRG-2 cDNA probes, no cross-hybridizing species could be detected by Southern analysis of human genomic DNA (data not shown). Sequence comparisons between the CRG-2 and IP-10 cDNAs and the presumed regulatory regions of their genes have identified conserved elements of possible functional importance. The presence of conserved sequences in the 3'- untranslated regions is similar to what has been observed in comparisons of the human-mouse homologues of a number of growth factors, oncogenes, and cytokines such as TNF (62), c-myc (68), GM-CSF (62,63), andgro (24). The region 842 to 967 of the CRG-2 mRNA shows 77% identity with the corresponding region of IP-10 (gaps counted as mismatches), is A:U-rich (69% versus 53% for the mRNA as a whole and 56% over the entire 3'-untranslated region), and contains sequence motifs, i.e. UAUUUA and A&, that have been implicated in the control of mRNA stability (62,63). These observations suggest that the conserved 3' sequences may be important in regulating the stability of the CRG-2 and IP-10 mRNAs. Analysis of the 5'-flanking regions of the erg-2 and IP-10 genes again reveals regions of conserved sequence, some of which contain known regulatory elements. In addition to the TATA and CCAAT boxes (69), both erg-2 and IP-10 contain the conserved ISRE (64) and binding sites for the transcription factor NF-KB (71). The presence of an ISRE in the crg-2 and ZP-10 promoter is consistent with the induction of crg-2 by type I and type II interferons as well as by poly (1). poly(C) (72)(73)(74). Induction of IP-10 has been reported to be a relatively specific response to IFN-7, although induction at 18 h by ~100 antiviral units of IFN-P was in fact demonstrated (23). erg-2 did appear more sensitive to IFN--r than to the other IFNs, compared on the basis of antiviral units, when analyzed at 6 h (Fig. 2), although the differences at 3 h were less marked. erg-2 and IP- 10 have 2 conserved NF-KB sites, identical with sequences that have been shown capable of binding NF-KB when used as isolated elements. One sequence, erg-2 -169 to -160, is identical with that found in the regulatory region of the human IFN-P gene (66,71,72) while the other, located at -113 to -104 of the erg-2 promoter, matches a site in the murine MHC class II End gene (75). The NF-KB site in the IFN-fi gene has been shown to confer inducibility by virus  (1) .poly(C) on a heterologous promoter (66,71,72), and the NF-KB sites may play a role in the induction of erg-2 by poly (1) .poly(C). Likewise, NF-KB is a mediator of LPSinduced activation of transcription in B cells (76) and may be responsible for the induction of erg-2 gene expression by LPS. Despite the presence of the NF-KB sites in erg-2 and IP-10, IP-10 was found not to be induced by PMA (77). In our experiments, we have not found erg-2 to be induced by PMA or by IL-l or TNF, all agents that have been reported to activate NF-KB (76, 78). There have been similar findings in the studies of the IFN-p gene, which is inducible by virus through NF-KB, but is not induced by PMA (71).
There are additional conserved sequences in the 5'-flanking regions of erg-2 and IP-10, such as an A:T-rich region between positions -147 and -139 of the erg-2 promoter and the region between the CCAAT and the TATA boxes. At -84 to -78 of the erg-2 promoter, there is a putative AP-1 site (67) that is not conserved in IP-10. Since the AP-1 site, like the NF-KB site, confers inducibility by phorbol esters (67), it is again of interest that in our initial experiments erg-2 was not induced by PMA, nor was it induced by serum in serum-starved Balb/ c 3T3 fibroblasts.
The biological activities of the PF4 family of secreted proteins are being actively investigated and defined (reviewed in Ref. 20). While activities of IP-10 have not been reported, IP-10 has been shown to be secreted, in response to IFN-7, ' MIG is a new member of the PF4 family induced in macrophages . *-__ ,--.