Identification of a Novel Eosinophil Chemotactic Cytokine (ECF-L) as a Chitinase Family Protein*

A novel eosinophil chemotactic cytokine (ECF-L) was purified from the culture supernatant of splenocytes of mice by a combination of anion-exchange chromatography, Procion red-agarose affinity chromatography, size exclusion high performance liquid chromatography (HPLC), and reverse phase HPLC. The NH2-terminal amino acid sequence was determined by direct protein sequencing. An ECF-L cDNA clone of 1,506 nucleotides was isolated from a cDNA library, and the nucleotide sequence predicted a mature protein of 397 amino acids. A recombinant ECF-L showed a level of eosinophil chemotactic activity comparable with that of natural ECF-L, and the activity was inhibited by a monoclonal antibody to ECF-L. ECF-L also attracted T lymphocytes and bone marrow polymorphonuclear leukocytes in vitro, whereas it caused selective extravasation of eosinophils in vivo. ECF-L mRNA was highly expressed in spleen, bone marrow, lung, and heart. A comprehensive GenBank data base search revealed that ECF-L is a chitinase family protein. ECF-L retains those amino acids highly conserved among chitinase family proteins, but Asp and Glu residues essential for the proton donation in hydrolysis were replaced by Asn and Gln, respectively. Although ECF-L contains a consensus CXC sequence near the NH2 terminus akin to chemokine family proteins, the rest of ECF-L shows poor homology with chemokines.

animals were pooled and used for the respective chemotaxis experiments.
T Lymphocytes-T lymphocytes for chemotaxis indicator cells were prepared as a T cell line. Female C57BL/6 mice were immunized with 20 g of ovalbumin (Sigma) in complete Freund's adjuvant. 12 days later, the spleen was removed, ground to single cell suspensions, and washed twice with PBS. The splenocytes were cultured with 30 g/ml OVA in RPMI 1640 containing 2% FBS. Blast cells were separated on a Ficoll (Amersham Pharmacia Biotech) density gradient and then cultured without antigen for 7 days in RPMI 1640 containing 2% FBS and 5% rat growth factor (supernatant of rat splenocytes cultured with 1 g/ml concanavalin A in RPMI 1640 containing 2% FBS for 2 days). The cells (5 ϫ 10 5 /ml) were restimulated with 10 g/ml ovalbumin in the presence of 5 ϫ 10 6 /ml irradiated splenocytes. The blast cells were cultured in RPMI 1640 containing 2% FBS and 5% rat growth factor for 7 days. After washing twice with PBS, they were used for chemotaxis experiments.
Chemotaxis Assay-Chemotactic activities in vitro were measured as described previously (10) with slight modifications. Briefly, multiwell microchemotaxis chambers (Neuro Probe, Bethesda, MD) were equipped with Millipore membrane filters (Millipore Co., Bedford, MA) with a pore size of 3 m for eosinophil and neutrophil chemotaxis or 8 m for macrophage and T lymphocyte chemotaxis. The number of indicator cells was adjusted to 1 ϫ 10 6 /ml with RPMI 1640 medium containing 2% heat-inactivated FBS. Incubation was performed at 37°C for 1 h in neutrophil chemotaxis or for 2 h in eosinophil, macrophage, or T lymphocyte chemotaxis. The membranes were stained according to Litt's procedure (51). The indicator cells that migrated into the membrane were counted from a fixed level from the upper surface down to the lower surface. The distance from the upper surface was selected as a value between 60 and 90 m from the upper surface to achieve background counts of less than 20 cells/10 high power field at 400 x. The total number of migrated cells was counted in 10 randomly selected high power fields. The chemotaxis experiments were done in quadruplicate and repeated at least twice. RANTES (Pepro Tech, London, U. K.), GRO (Diaclone Research, Besamcom Cedex, France), or MCP-1 (R & D Systems, Minneapolis, MN) was used as chemotaxis positive control.
In Vivo Study for Eosinophil Accumulation-Samples (25 l) were injected intradermally on the back of ddY mice that had been injected with 100 tetrathyridiae of M. corti 3 weeks earlier. 2 h later, mice were sacrificed, and the injection sites were collected. The tissue samples were fixed in 10% formalin, embedded in paraffin, sectioned at 4 m, and stained with hematoxylin and eosin.
Preparation of Conditioned Medium-The spleen was removed from normal or S. japonicum-infected mice 8 weeks after infection and gently squashed between two frost-ended slides in cold Hanks' balanced salt solution. The cell suspensions were washed with Hanks' balanced salt solution and suspended in serum-free RPMI 1640. Culturing was carried out at 37°C for 24 h in a 5% CO 2 and air environment in the presence of purified egg antigen J1 (10 g/ml). Conditioned medium was obtained by centrifugation at 1,200 ϫ g for 10 min. The supernatant was sterilized by filtration through a 0.45-m membrane filter (Millipore Co.) and stored at Ϫ30°C until used.
Anion-exchange Chromatography-DE52 (Whatman, Springfield, UK) anion-exchange chromatography was carried out at 4°C using a 2.5 ϫ 60-cm column equilibrated with 0.016 M Tris-HCl buffer, pH 7.7. Elution was performed with a linear gradient of NaCl (0 -0.3 M) in the Tris-HCl buffer at a flow rate of 10 ml/h, and 8.5-ml fractions were collected.
Affinity Chromatography on Procion Red-Agarose-Procion red-agarose (Life Technologies, Inc.) was packed in a column (1.2 ϫ 6-cm) and was equilibrated with 0.016 M Tris-HCl buffer, pH 7.7. Samples were dialyzed against the same buffer and applied to the Procion red-agarose column. The column was first washed with 60 ml of starting buffer, and then elution was performed with a linear gradient of NaCl (0 -1.5 M) in the Tris-HCl buffer at a flow rate of 10 ml/h, and 6-ml fractions were collected.
Size Exclusion HPLC-A sample was concentrated to 0.5 ml and applied to a size exclusion HPLC column (G3000SW, 0.78 ϫ 30-cm, Toso, Tokyo, Japan) equilibrated with PBS. Chromatography was performed at a flow rate of 1.0 ml/min, and fractions of 0.7 ml were collected.
Reverse Phase HPLC-A sample was concentrated to 0.5 ml and applied to an Cosmosil 5 C 18 column (4.6 ϫ 25-cm, Nacalai Tesque). Elution was performed with a linear gradient of 0 -70% acetonitrile in 0.1% trifluoroacetic acid at 25°C and a flow rate of 0.7 ml/min, and 0.7-ml fractions were collected.
Amino Acid Sequence Analysis-The amino acid sequence was determined by automated Edman degradation on a gas-phase protein microsequencer (ABI470A, Perkin-Elmer, Foster City, CA).
Antisera-Female Lewis rats were immunized twice at a monthly interval in both hind footpads with a total of 0.2 ml of a 1:1 emulsion of the purified ECF-L (10 g) in PBS and H37Ra adjuvant (Difco Laboratories, Detroit, MI). 2 weeks after the second immunization, they were bled, and the serum was separated. Establishment of a monoclonal antibody (mAb) to ECF-L was carried out according to a method described previously (14) with modifications. In short, female Lewis rats were immunized twice with the purified ECF-L (10 g) emulsified in H37Ra adjuvant. The spleens were removed 2 weeks after the last immunization and were hybridized to Sp2/0. A clone (E3.1) was selected, and the mAb was purified from the culture supernatant of hybridomas by gel chromatography on Sephacryl S300 (Amersham Pharmacia Biotech) for experimental use. The isotype of E3.1 was determined as IgM by a kit (Binding Site, Birmingham, U. K.).
Construction of the cDNA Library-The total RNA of bone marrow cells was isolated by the guanidine isothiocyanate/CsCl 2 method (15). Poly(A) ϩ RNA was selected by two passages over oligo(dT)-cellulose (Amersham Pharmacia Biotech). cDNA synthesis was performed using a cDNA synthesis kit (Amersham Pharmacia Biotech). The cDNAs were cloned into the EcoRI site of gt11 vector (Promega Biotec, Madison, WI). Viral DNA was then packaged in vitro by means of Gigapack (Stratagene, La Jolla, CA).
Isolation and Analysis of a cDNA Clone-Independent recombinant clones (2 ϫ 10 5 ) were screened using 1,000 ϫ diluted anti-ECF-L polyclonal antibody as an antibody probe. Clones reacting with the antibody were isolated from the cDNA library and subcloned into Bluescript SKϪ (Stratagene). Nested deletions were prepared by using a nested deletion kit (Amersham Pharmacia Biotech). The sequences of suitable clones were determined by the dideoxy chain termination method using an automated DNA sequencer (ABI310, Perkin-Elmer).
Construction of the Baculovirus Transfer Vector and Generation of the Recombinant Virus-A BamHI site was created 6 bp upstream of the translation initiation codon and an EcoRI site 6 bp downstream of the stop codon to obtain a full-length cDNA by polymerase chain reaction, and the cDNA was digested with the BamHI and EcoRI. The BamHI-EcoRI fragment was then ligated to pVL1393 (Invitrogen). The recombinant pVL construct was co-transfected into Spodoptera frugiperda Sf21 insect cells with a modified Autographa californica nuclear polyhedrosis virus DNA (BaculoPlus, Pharmingen), and the resultant viral pool was collected 4 days later.
Preparation of Recombinant Proteins-Sf21 insect cells were cultured in Grace medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.) at 27°C. To produce recombinant proteins, Sf21 cells seeded at a density of 10 6 cells/ml were infected with the recombinant virus. The cells were harvested 96 h after infection, washed with a solution of 0.15 M NaCl in 0.02 M phosphate buffer, pH 7.4, homogenized in a Tris-HCl, pH 7.6, and centrifuged at 10,000 ϫ g for 20 min. Recombinant ECF-L was purified from the supernatant using DE52, Procion red-agarose, G3000SW, and 5 C 18 columns following the same steps as for the purification of natural ECF-L from the culture supernatant of splenocytes.
Northern Blot Analysis-Total RNA from each organ was isolated using Isogen (Nippon Gene, Tokyo, Japan) from female C57BL/6 mice infected with 100 tetrathyridiae of M. corti 2 weeks earlier. 10 g of total RNA from each organ was electrophoresed in 1% formaldehydeagarose gel in MOPS and transferred onto a Hybond-N membrane (Amersham Pharmacia Biotech). After prehybridization for 3 h in 50% formamide, 5 ϫ Denhardt's solution, and 100 g/ml denatured salmon sperm DNA at 65°C, blots were hybridized with random primed 32 Plabeled probes for 16 h under identical conditions. Filters were washed three times with 1 ϫ SSC containing 0.1% SDS at 65°C and then exposed to x-ray film (BioMax, Eastman Kodak, Rochester, New York). 32 P-Labeled polymerase chain reaction fragments of ECF-L and mouse glyceraldehyde-3-phosphate-dehydrogenase cDNAs were used for probes.
Analysis of Interaction with Chitins-Chitinase activity was measured by a chromogenic assay using 200 mM p-nitrophenyl ␤-D-N,NЈ,NЉtriacetylchitotriose (Sigma) in 0.05 M phosphate buffer, pH 6.8, as the substrate. Chitinase from Serratia marcescens (Sigma) was used as a positive control. Chitin bead (Life Technologies, Inc.) chromatography (1.0 ϫ 3.3-cm column) was performed with PBS at a flow rate of 3.0 ml/h, and 0.6-ml fractions were collected.

RESULTS
Purification of ECF-L-For purification of ECF-L protein, the medium conditioned by spleen cells obtained from S. japonicum-infected mice was concentrated, dialyzed against Tris-HCl buffer, and applied to a DE52 anion-exchange column. ECF activity was eluted at around the 0.15 M NaCl region of the gradient (Fig. 1A). Active fractions were pooled, concentrated by ultrafiltration, dialyzed against Tris-HCl buffer, and then applied to the Procion red-agarose affinity column. Elution profiles are shown in Fig. 1B. The major ECF activity was eluted at around 0.5 M NaCl. ECF-positive fractions were pooled, concentrated, and chromatographed by G3000SW HPLC (Fig. 1C). ECF-positive fractions were pooled and purified further on a reverse phase HPLC column (Fig. 1D). A single absorbance peak at 230 nm coincided with the peak of ECF activity. The purified ECF-L showed homogeneity on SDS-PAGE analysis (Fig. 1E) with a molecular mass of 43-kDa, and the specific eosinophil chemotactic activity was approximately 100 times that of the crude supernatant (Fig. 1F).
Isolation of ECF-L cDNA-The cDNA library of bone marrow of mice was screened with an anti-ECF-L polyclonal antibody or mAb E3.1b. Of approximately 1,000,000 plaques screened, one plaque was found to be positive. The positive clone was picked and purified through three successive rounds of replating antibody screening. The insert was purified and subcloned into Bluescript SKϪ. As shown in Fig. 2, sequence analysis revealed that the cDNA was 1,506 bp long with a 1,197-bp open reading frame that could encode a protein 397 amino acids long. The putative amino acid sequence included the NH 2terminal 21-amino acid endoplasmic reticulum signal peptide followed immediately by the NH 2 -terminal sequence of natural ECF-L as determined by a protein microsequencer (YQLMXYYTSWAKDRPIEG). The molecular mass of mature ECF-L, calculated from the deduced amino acid sequence, was 42,370 Da. This value is close to the one determined experimentally by SDS-PAGE (Fig. 1E). The calculated isoelectric point (pI 5.3) was somewhat higher than that determined experimentally (pI 3.6) by isoelectric focusing (11). The sequence does not contain any consensus N-linked glycosylation sites (NXS, NXT) (18). In addition, the mRNA contains an approximately 300-bp-long untranslated region with a polyadenylation signal (AAUAAA) toward the 3Ј-end.
Recombinant ECF-L was prepared by a baculovirus expression system and purified by a combination of anion-exchange chromatography on DE52, affinity chromatography on Procion red-agarose, gel chromatography on G3000SW, and reverse phase HPLC. The purified recombinant protein showed a single band on SDS-PAGE analysis with a molecular mass of 43 kDa (data not shown). The ECF activity of the recombinant ECF-L was comparable with that of the natural ECF-L (Fig. 3A) and was completely inhibited by mAb to ECF-L (E3.1b) as well as natural ECF-L (Fig. 3B).
We next examined the cellular specificity of ECF-L as a chemoattractant. Because the optimum chemotactic activity of RANTES, GRO, and MCP-1 was achieved at 10 -8 M in our assay system, the same molar concentration of ECF-L was employed for the comparative chemotaxis assays. As shown in Fig. 4, both natural and recombinant ECF-L exhibited chemotactic activity for T lymphocytes and bone marrow cells as well as for eosinophils. Most of the migrated bone marrow cells in the membrane showed PMN-like features with segmented nuclei. In contrast, ECF-L exhibited a limited chemotactic activity for mature neutrophils and no detectable chemotactic activity for macrophages. This cellular specificity of ECF-L was similar to that of RANTES. On the other hand, GRO showed chemotactic activity for mature neutrophils but not for bone marrow cells. mAb to ECF-L (ECF-3) could inhibit most of the chemotactic activity of ECF-L for eosinophils, T lymphocytes, and bone marrow cells but not the chemotactic activities of RANTES, GRO, or MCP-1.
To examine whether ECF-L could induce eosinophils to extravasate into the inflammatory site, recombinant ECF-L was injected intradermally on the back of mice infected with M. corti. As shown in Fig. 5, large numbers of permeated eosinophils were observed at the site where ECF-L was injected. 90 Ϯ 7% of the migrated leukocytes outside the blood vessel were eosinophilic when 29 Ϯ 4% of all leukocytes in the blood were eosinophils in the treated mice. The number of migrated eosinophils increased dose-dependently with ECF-L. The increase in eosinophil migration was observed at as low as 5 ϫ 10 Ϫ8 M (1 pmol).
Production of ECF-L in a Parasitic Infection-Kinetics of the ECF-L production by splenocytes were examined in M. cortiinfected mice. As shown in Fig. 6A, production of eosinophil chemotactic activity by splenocytes was detectable as early as 2 weeks postinfection by Western blot analysis in an antigenspecific manner and increased with time. The amount of ECF-L detected was correlated with the eosinophil chemotactic activity of the culture supernatant of splenocytes (Fig. 6B). When the ECF-L mRNA expression of each organ was examined at 2 weeks after M. corti infection, high level mRNA expression was detected in spleen, bone marrow, lung, and heart (Fig. 6C). Lower levels of mRNA were also detected in liver, thymus, and small intestine. A message for ECF-L was generally undetectable in normal mice by Northern blot analysis, whereas a low level of mRNA was detected by reverse transcription-polymerase chain reaction in bone marrow, spleen, and thymus in normal mice (data not shown). Nucleotide sequences of the reverse transcription-polymerase chain reaction products 195) from the liver, spleen, and bone marrow mRNA of ddY mouse were identical to the sequence of the isolated cDNA clone from the library (Fig. 2).
Sequence Homology with Other Proteins-A comprehensive search of GenBank or EMBL nucleic acid data bases revealed that ECF-L possesses significant homology with prokaryotic chitinases, class III plant chitinases, fungus chitinase, insect and nematode chitinase, and chitinase family proteins distributed in vertebrate animals. An optimal alignment of 20 repre-FIG. 2. Nucleotide sequence and deduced amino acid sequence of ECF-L. The hydrophobic leader sequence is underlined with dots. The matched residues of the deduced amino acid sequence and the NH 2 -terminal amino acid sequence of natural ECF-L are underlined. The polyadenylation signal is indicated in bold lettering. The reverse type denotes the highly conserved residues (conserved in more than 80% of 20 representative chitinases or chitinase family proteins as shown in Fig. 5). The reverse type with asterisks indicates the completely conserved residues in the 20 representative chitinases and chitinase family proteins. Boxed letters denote the specific amino acid substitution in ECF-L of highly conserved residues (conserved in more than 80% of the 20 chitinases or chitinase family proteins). Double underlining denotes a highly conserved region that includes amino acids essential for chitinase activity. Numbers indicate the nucleotide position. The sequence data have been deposited in the DDBJ data base. sentative chitinases or chitinase family proteins was performed using the multiple sequence alignment program, Clustal X. 49 residues are highly conserved in more than 80% of the 20 representative chitinase family proteins, and 13 residues are completely conserved in all chitinase family proteins including ECF-L (Fig. 2). ECF-L retains 44 residues (90%) out of the 49 highly conserved. Fig. 7 shows an alignment of a conservative region where the catalytic center of the chitinases is located (19). Aspartic acid, essential for catalytic activity (19), is conserved in all of the chitinases and chitinase family proteins other than ECF-L. Although the other essential acid for catalytic activity (19), glutamic acid, is conserved in all of the chitinases, it is generally replaced by leucine or isoleucine in chitinase family proteins that have no chitinolytic activity (20 -25). In ECF-L, both the glutamic acid and the aspartic acid are replaced, by glutamine and asparagine, respectively. ECF-L possesses the CXC consensus sequence near the NH 2 terminus which is typical of CXC chemokines. However, for ECF-L with CC chemokines or CXC chemokines (Fig. 8) poor alignments including for second and third cysteine residues were obtained.
Analysis of Interaction with Chitins-We could not detect any chitinase activity in purified ECF-L at any concentration up to 0.5 mg/ml. On the other hand, natural ECF-L eluted far behind (162%) the bed volume of the chitin column, whereas bovine serum albumin or other unrelated proteins eluted at 47% of the bed volume. DISCUSSION In this paper, we described the purification, cDNA cloning, and molecular characterization of a novel eosinophil chemotactic cytokine, ECF-L, which is produced in parasitic infections upon stimulation of specific antigens. Comparisons of the inferred protein sequence against all of the sequences in the GenBank or EMBL data base revealed a high degree of similarity to chitinase belonging to family 18 of glycosyl hydrolases (26) and vertebrate chitinase family proteins without chitinase activity (20 -25). Sequence alignments revealed that ECF-L is an independent molecule different from other known ECF cytokines such as interleukin-5 (27), RANTES (28), eotaxin (29), or ecalectin (30).
Chitinases catalyze the hydrolysis of ␤-1,4-N-acetylglucoside linkages in chitin and chitodextrins and are widely distributed among a variety of species such as bacteria, fungi, nematodes, plants, insects, and vertebrates. In plants, chitinases/lysozymes themselves likely form part of innate immune system important for host defense against invading pathogenic bacteria and fungi (31). Insect or nematode chitinases are likely essential to certain life cycle events such as molting of the larval exoskeleton or casting of the egg shell (32). Similarly, protozoans increase chitin synthesis (33) and chitinase activity (34) during encystation. On the other hand, proteins of the so-called chitinase family are distributed in mammals and have no detectable chitinolytic activity (20 -25). The actual physiological roles of the mammalian chitinase family proteins remain to be clarified. The present study has revealed a signif-   (38). Production of ECF-L at local sites would contribute to the triggering of the influx of eosinophils that act as effectors of parasite killing. Thus, ECF-L may have evolved from a chitinase as an immune molecule from invading parasites by means of accumulating eosinophils followed by secreting toxic substances rather than directly digesting parasites.
Chitinase-related proteins of vertebrates share relatively weak amino acid sequence homology with chitinases as a whole. To elucidate the possible function of the conserved residues of ECF-L, the alignment of ECF-L and 19 sequences of representative chitinase-related proteins was performed (Figs. 2 and 5). 13% of the amino acids (49 residues) were highly conserved in more than 80% of the representative chitinase or chitinase family proteins. The ratio of aromatic amino acids (Phe, Tyr, Trp) in the highly conserved residues (34.7%) is relatively high compared with those among all of the residues (14.8%) of ECF-L. Glycine was also prominent in the conserved residues (20.4%) compared with that in total ECF-L (8.5%).
The tertiary structures of some chitinase proteins have been resolved: these proteins all share the feature of a (␤␣) 8 barrel topology (39,40). Based on the amino acid sequence homology, the locations of these conserved amino acids of ECF-L are assumed by analogy. Some highly conserved aromatic amino acids of ECF-L (Tyr-6, Trp-10, Phe-37, Tyr-246, Trp-339) likely form hydrophobic clusters in the active cleft, and two (Phe-37, Trp-339) of these residues form a cis-peptide bond that is assumed to play a key role in substrate recognition. In terms of position, the glycine residues (Gly-76 and Gly-77) are highly conserved; they likely locate at the end of the third ␤-strand where a conserved phenylalanine (Phe-37) is closely located. All but one of the highly conserved glycine residues are also located in the ␤-strand or ␤␣-loops, indicating that replacement by other residues may not be permitted to avoid a steric hindrance. ECF-L shares the highly conserved aromatic amino acids and glycines that form a substrate recognition site, and this fact suggests that ECF-L retains a similar ability for the recognition or conformational change of chitin-related carbohydrates. In fact, elution of ECF-L from the chitin beads column clearly took longer than estimated probably because of the interaction with chitins. As for the carbohydrate recognition of ECF, human ecalectin that binds to carbohydrates (␤-galactosides) possesses ECF activity (30). The ability to recognize carbohydrate may be another aspect of ECF-L in the protective immunity against parasitic infections because chitin-related carbohydrates are an important component of the exoskeleton of larvae and egg shells in nematodes (32).
As for the catalytic activity of chitinases, a site-directed mutagenesis study of chitinase A1 from Bacillus circulans showed that artificial mutation of either glutamic acid at position 204 (Glu 3 Gln or aspartic acid at position 200 (Asp 3 to Asn) resulted in a remarkable loss of catalytic activity (19), indicating that the carbonic acids of both residues are essential for proton donation in hydrolysis. In ECF-L, both the essential glutamic acid (corresponding location in ECF-L: Gln-119) and aspartic acid (corresponding location in ECF-L: Asn-115) are modulated in the natural protein, explaining the undetectable level of chitinolytic activity. In other vertebrate chitinase family proteins that have no catalytic activity, glutamic acid is replaced by a similar sized hydrophobic side chain such as leucine in synovial protein (21) and oviductal glycoprotein (22), or with isoleucine, such as in major secreted protein of human articular chondrocytes (20) and heparin-binding glycoprotein (25). On the other hand, the essential aspartic acid is generally conserved. Thus ECF-L possesses a unique blocking form of two essential carbonic acids in contrast to other vertebrate chitinase family proteins, and this fact may suggest a unique function of ECF-L among chitinase family proteins without chitinase activity.
It has been proposed that blocking parasite-derived chitinases could block the transmission of malaria (41) by inhibiting the chitinase activity needed for the penetration of the peritrophic membrane in anopheline mosquitoes. Similarly, microfilarial stage-specific chitinase has been shown to be a candidate antigen for a transmission-blocking vaccine against filariasis (42). The sequence similarity between the chitinase family proteins of the host molecules and parasite-derived chitinases suggests some restriction in the use of chitinases as vaccine molecules.
ECF-L possesses a CXC sequence near the NH 2 terminus of the mature molecule. The CXC or CC sequence is a typical motif shared in many chemokine family proteins but not in other chitinase family proteins. Amino acid sequence alignments of ECF-L with CXC or CC chemokines revealed no sequence similarity even around the CXC sequence (Fig. 7) where the critical receptor binding region exists (43,44). The other two cysteine residues conserved in all of the chemokines are not found in ECF-L. Moreover, the tertiary structure of interleukin-8, a CXC chemokine, exhibits topography similar to that of the ␣1/␣2 domains of the human class I histocompatibility antigen HLA-A2 (45) and to share no similarity with chitinases (39,40). This would consistently indicate that ECF-L is an independent molecule, not one of these chemokines.
We found that ECF-L attracts not only eosinophils but also T lymphocytes and bone marrow PMNs. ECF-L possesses a specificity similar to RANTES as a chemoattractant in that it attracts CD4 ϩ memory T lymphocytes (46), eosinophils, and bone marrow cells (Fig. 4). This indicates that the receptor(s) for ECF-L are related to that for RANTES (47,48). The specific inhibitory effect of anti-ECF-L mAb is likely understood in terms of the steric hindrance that resulted in the obstruction of chemotactic epitope(s) on ECF-L.
Recent evidence suggests that eosinophils bind to vascular endothelial cells, and their transendothelial migrations are mediated by interactions of various adhesion molecules and their ligand molecules (49). ECFs produced in the inflammatory site would play a critical role in the exit of eosinophils from vasculature and their attraction to the inflammatory site. In the present study, we showed that ECF-L could induce the specific extravasation of eosinophils into the injected site at as low as 1 pmol. This result suggests that eosinophil-specific transendothelial migration could be caused by ECF-L when an appropriate amount of ECF-L is produced outside the blood vessel. The discrepancy in the cellular specificity of ECF-L in in vivo (Fig. 5) and in vitro (Fig. 4) studies would be the result of complicated mechanisms of transendothelial processes in vivo including the effects of adhesion molecules (49).
With regard to whether enough ECF-L is generated locally to cause tissue eosinophilia around the larvae, little evidence of this is presently available. Expression of ECF-L mRNA was not limited to the liver where the M. corti larvae locate. So far as we could ascertain, spleen, bone marrow, lung, and heart are the dominant organs of production of mRNA encoding ECF-L. Western blot and flow cytometric analysis also indicate that ECF-L molecules locate on some populations of bone barrow cells (data not shown). Furthermore, ECF-L exhibits chemotactic activity for bone marrow PMN (Fig. 4E). These results together with our previous finding that ECF-L could change the chemotactic reactivity of immature eosinophils located in bone marrow (10) suggests a major role for ECF-L in the priming of naive eosinophils or neutrophils. Alternatively, antigen-specific production of ECF-L by splenocytes and chemotactic reactivity of ECF-L for bone marrow cells would support the idea that the spleen is the site of maturation of immature eosinophils (9) and neutrophils, and thus ECF-L contributes to the recruitment of immature PMNs to this site.
Several eosinophil chemotactic cytokines have been shown to have multiple functions (27,50). In addition to their role in the recruitment of eosinophils from the site where PMNs are differentiated to the inflammatory site and to their priming effects, ECF-L may participate in the removal of excess eosinophils from the circulation to spleen or bone marrow where they can be treated and/or reused. The possible novel functions of ECF-L remain to be clarified.