Molecular Expression of the (cid:97) -Chemokine Rabbit GRO in Escherichia coli and Characterization of Its Production by Lung Cells in Vitro and in Vivo *

GRO proteins are (cid:97) -chemokine cytokines that attract neutrophils and stimulate the growth of a variety of cells. Previously, we observed that rabbit alveolar macrophages transcribe the genes for at least two GRO homologues. In order to study the role of GRO cytokines in lung inflammation, we cloned the predominant rabbit GRO cDNA (RabGRO) from alveolar macrophages, expressed bioactive recombinant protein (rRabGRO) in Escherichia coli , and developed a sensitive and specific enzyme-linked immunosorbent assay for RabGRO pro- tein. We found that rabbit AM express and secrete GRO in vitro in response to both exogenous ( e.g. lipopolysac- charide, heat-killed Staphylococcus aureus , and crystalline silica) and endogenous inflammatory stimuli ( e.g. tumor necrosis factor- (cid:97) ) as determined by both radioimmunoprecipitation and enzyme-linked immunosorbent assay. Biologically significant amounts of GRO are present in vivo in the bronchoalveolar lavage fluid of rabbits with E. coli pneumonia; by in situ hybridization, GRO mRNA is detectable in infiltrating pulmonary leuko- cytes and bronchial epithelial cells. These results indicate that GRO chemokines are likely to be important mediators of the inflammatory response that accompanies acute and humidified Experiments quadruplicate, chemotaxis measured as the total number of leukocytes migrating through the filters in 10 consecutive high power microscopic fields. Results were expressed as percent maximal chemotaxis, in which the average number of leukocytes migrating toward PBS for each chamber was subtracted from each measurement (including ZAS) and then each measurement was divided by the adjusted value for ZAS. The value for ZAS was set as 100% maximal chemotaxis. Results are reported as (cid:54) S.E. Endobronchial Instillation of rRabGRO Protein— To test the bioactivity of rRabGRO in the lung, 1.0 (cid:109) g of LPS-free protein in 1.0 ml sterile, pyrogen-free 0.9% NaCl (approximately 10 (cid:50) 7 M solution) was instilled into the right lung of an anesthetized rabbit through an intra- tracheal catheter directed toward the right mainstem bronchus. For comparison, another rabbit was treated with 0.9% NaCl alone. The catheters were removed immediately following the instillations, and the animals were closely monitored to ensure adequate recovery from the procedure. The rabbits were euthanized 4 h later with pentobarbital, and selective right lung lavage was performed to obtain BALF for total and differential cell counts using methods as described (16).

It is likely that GRO proteins are relevant in lung inflammation, as human alveolar macrophages (AM) express GRO mRNA and protein in response to stimulation with lipopolysaccharide (LPS) (10), toxic shock syndrome toxin-1 (11), and tumor necrosis factor-␣ (TNF-␣) (12). Although these in vitro studies suggest that GRO chemokines are mediators of the lung response to both exogenous and endogenous inflammatory stimuli, demonstration of their biological relevance in vivo is lacking.
Rabbits, like humans and in contrast to rodents (1), produce a spectrum of ␣-chemokines that includes IL-8 (13) and GRO (6,14,15). We reported previously the molecular cloning of two distinct rabbit GRO cDNA homologues from LPS-stimulated AM (RabGRO and rabbit permeability factor-2 or RPF2) (15). Of the two transcripts, RabGRO was the predominant AMderived GRO homologue. We now report the expression in Escherichia coli of bioactive recombinant RabGRO protein (rRabGRO) and the development of a sensitive and specific RabGRO immunoassay. Using this assay, we have found that rabbit AM secrete native GRO protein in vitro in response to stimulation with a spectrum of biologically relevant agonists. We also report that biologically significant quantities of GRO protein are present in the bronchoalveolar lavage fluid (BALF) of rabbits with acute bacterial pneumonia and identify the sites of GRO mRNA expression in the lungs. These data suggest that GRO chemokines make important contributions to acute inflammatory responses in the lungs.

MATERIALS AND METHODS
Expression of rRabGRO Protein-The coding sequence for mature RabGRO protein (219 base pairs, 72 amino acids) (15) was amplified from LPS-stimulated rabbit AM mRNA by reverse transcriptase-polymerase chain reaction techniques using a 5Ј-primer with a BamHI restriction enzyme site (5Ј-CGGGATCCCGCGCTCACCGAGC-3Ј) and a 3Ј-primer with a BglII restriction enzyme site (5Ј-GAAGATCTTCCTC-CTTTCCCAGG-3Ј). The primer sequences written in boldface correspond to base pairs coding for RabGRO. The RabGRO polymerase chain reaction product was confirmed by sequencing and directionally cloned into the expression vector pRSET (Invitrogen Co., San Diego, CA). The pRSET vector encodes a fusion protein that contains an N-terminal fusion peptide sequence with polyhistidine residues for Ni 2ϩ affinity column purification of the fusion protein and an enterokinase cleavage site for removal of the fusion peptide sequence to generate the mature recombinant protein. The transformants were cultured in the presence of isopropyl-␤-D-thiogalactopyranoside and M13/T7 bacteriophage to provide the T7 RNA polymerase required for transcription of the rRa-bGRO fusion protein (rRabGRO-fp) mRNA. After induction for 10 h, the bacterial pellet was resuspended in 20 mM PBS, pH 7.8, with lysozyme (100 g/ml, Sigma) and then incubated on ice for 15 min with 10 mM EDTA and 1% Triton X-100, sonicated, flash-frozen in liquid nitrogen, and thawed at 37°C three times in succession. The supernatant was dialyzed overnight against PBS, pH 7.8 (M r 1000 limit membrane, Spectrum Medical Industries, Inc., Los Angeles, CA). Following dialy-sis, the supernatant was mixed with Ni 2ϩ -activated resin and incubated at room temperature for 1 h. The resin was then washed with PBS at pH 7.8, 6.0, and 5.0 and then packed into a column. The rRabGRO-fp was eluted from the column in 1.0-ml fractions by rinsing with PBS at pH 2.0. Each fraction was analyzed by SDS-PAGE, and the fractions containing rRabGRO-fp were pooled and dialyzed against 100 mM NaCl overnight. The rRabGRO-fp was incubated overnight with enterokinase (Biozyme, South Wales, United Kingdom) at a concentration of 1 unit/g fusion protein in 10 mM Tris-HCl, pH 8.0 and 10 mM CaCl 2 to cleave the fusion peptide sequence. The reaction mixture was then loaded onto a C4 column (Dynamax-300A, Rainin Co., Emeryville, CA) and reverse-phase high performance liquid chromatography (HPLC) was performed using a 5-95% acetonitrile gradient. The resulting HPLC peaks were analyzed by SDS-PAGE, and the N-terminal amino acid sequences were determined by Edman degradation with a 475 Å pulse liquid protein sequencer (Applied Biosystems, Foster City, CA).
Animals-Specific pathogen free female New Zealand White rabbits weighing approximately 3.0 -3.5 kg (Western Oregon Rabbit Co., Philomath, OR) and 6 -8-week-old female BALB/c mice (Simonsen Laboratories, Gilroy, CA) were housed in the Seattle Veterans Affair Medical Center vivarium prior to all experiments. A female goat was purchased locally and housed at the University of Washington vivarium.
Cell Purification-Rabbit AM were recovered by whole lung lavage as described previously (16). Rabbit and human neutrophils were recovered from heparinized venous blood using Ficoll-Hypaque density gradient centrifugation (Mono-Poly resolving medium, ICN Biomedicals, Aurora, OH). The cells were washed twice in PBS and resuspended at the indicated concentrations for chemotaxis or immunoprecipitation studies.
Chemotaxis Assays-Rabbit and human neutrophil chemotaxis to rRabGRO was measured by the modified Boyden method using 48-well microchemotaxis chambers and nitrocellulose membranes with 3.0-m pores for neutrophils as described (17). The samples were diluted with PBS, and 25 l of each sample were added to the bottom well. Controls included 10% zymosan-activated human or rabbit serum (ZAS, positive control) or PBS alone (negative control). Neutrophils were added to the top wells, and the chambers were incubated for 2 h at 37°C in 5% CO 2 and humidified air. Experiments were performed in quadruplicate, and chemotaxis was measured as the total number of leukocytes migrating through the filters in 10 consecutive high power microscopic fields. Results were expressed as percent maximal chemotaxis, in which the average number of leukocytes migrating toward PBS for each chamber was subtracted from each measurement (including ZAS) and then each measurement was divided by the adjusted value for ZAS. The value for ZAS was set as 100% maximal chemotaxis. Results are reported as mean Ϯ S.E.
Endobronchial Instillation of rRabGRO Protein-To test the bioactivity of rRabGRO in the lung, 1.0 g of LPS-free protein in 1.0 ml sterile, pyrogen-free 0.9% NaCl (approximately 10 Ϫ7 M solution) was instilled into the right lung of an anesthetized rabbit through an intratracheal catheter directed toward the right mainstem bronchus. For comparison, another rabbit was treated with 0.9% NaCl alone. The catheters were removed immediately following the instillations, and the animals were closely monitored to ensure adequate recovery from the procedure. The rabbits were euthanized 4 h later with pentobarbital, and selective right lung lavage was performed to obtain BALF for total and differential cell counts using methods as described (16).
Preparation of rRabGRO Antisera-A female goat was immunized with successive doses of rRabGRO-fp. Polyclonal goat IgG was purified from serum using protein G columns (Pierce), and a portion was biotinylated (Pierce). The specificity of the goat antisera was tested by Western blotting using biotinylated goat anti-RabGRO IgG or biotinylated nonimmune IgG and recombinant rabbit IL-8, rabbit monocyte chemotactic protein-1 (MCP-1) (18), recombinant human GRO-␣ (R & D Systems, Inc., Minneapolis, MN), and rRabGRO protein.
Rabbit GRO Immunoassay-A sandwich enzyme-linked immunosorbent assay (ELISA) was constructed using goat polyclonal anti-rRab-GRO IgG as both the capturing and detecting antibody. After coating and overnight incubation at 4°C with nonbiotinylated immune goat IgG (1:350 dilution), microtiter plates were blocked with 10% non-fat milk for 1 h at 37°C. The blocking buffer was removed, samples and standards (serial dilutions of rRabGRO protein) were added, and the plates were incubated for 1 h at 37°C. After washing four times, biotinylated immune goat IgG (1:800 dilution) was added to the plates and they were again incubated for 1 h at 37°C. After washing, peroxidase-labeled streptavidin (Zymed Laboratories, Inc., South San Francisco, CA) was added, followed by incubation for 1 h at 37°C. The chromogen 3,3Ј,5,5Ј-tetramethylbenzidine (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD) was added, and the optical density (A 450 ) in each well was measured using a microtiter plate reader (Dynatech Co., Chantilly, VA). Based on a standard curve generated with rRabGRO protein, the lower limit of detection for the ELISA was 22 pg/ml. There was no cross-reactivity with either recombinant rabbit IL-8 or recombinant rabbit MCP-1 at concentrations varying from 10 pg/ml to 1 mg/ml in spiked rabbit BALF samples. For statistical analysis, undetectable GRO levels were assigned a value of 11 pg/ml (onehalf the limit of detection of the assay).
GRO Protein Measurements in Rabbits with E. coli Pneumonia-Anesthetized rabbits were treated with either a 1.0 ml bolus of 2 ϫ 10 10 colony-stimulating factor E. coli (serotype K1) in sterile, pyrogen-free 0.9% NaCl (n ϭ 8) or NaCl alone (n ϭ 8) through an intratracheal catheter directed toward the right mainstem bronchus. The animals were euthanized 4 or 24 h later, and selective right lung lavage was performed. Bronchoalveolar lavage fluid cell counts and differentials were calculated and GRO protein levels were measured by ELISA.
In Situ Hybridization of Lung Tissue from Rabbits with E. coli Pneumonia-In situ hybridization was performed on right lung tissue from rabbits treated with endobronchial E. coli (1 ϫ 10 9 colony-stimulating factor) and sacrificed after 4 h as described above. Lung tissue samples were frozen in liquid Freon ® and placed in O.C.T. Compound (Miles, Inc., Elkhart, IN). The embedded samples were then sectioned (9-m slices) and mounted onto gelatin-coated slides (Fisher). After treatment with a 4% paraformaldehyde solution, the slides were dehydrated with graded ethanol rinses. Antisense and sense (base pairs 111-140 (15)) RabGRO cDNA probes were made by end labeling synthesized oligonucleotides with 35 S-dATP using terminal deoxynucleotidyltransferase (DuPont). The specificity of the probe for rabbit GRO cDNA was confirmed by the absence of hybridization signals with rabbit IL-8, IL-1, and TNF-␣ cDNA on Southern blots (data not shown). Hybridization was performed at 48°C overnight by incubating each slide in a buffer containing 50% formamide and 1 ϫ 10 6 cpm antisense or sense (negative control) RabGRO cDNA. Post-hybridization washings were performed twice in 1 ϫ standard saline citrate solution at 66°C for 30 min and once at room temperature for 1 h. The slides were then dehydrated with successive ethanol rinses and dipped into warmed Kodak NTB2 emulsion solution (Eastman Kodak Co.) for autoradiographic analysis. After exposure for 9 days at 4°C, the slides were developed, fixed, counterstained with hematoxylin-eosin, and visualized by light microscopy.

RESULTS
Expression of rRabGRO-The coding sequence for mature RabGRO was amplified by polymerase chain reaction and cloned into the pRSET expression vector. After confirming the sequence for the RabGRO cDNA, the recombinant protein was expressed in JM109 E. coli and purified from bacterial lysates by Ni 2ϩ affinity chromatography. Analysis by SDS-PAGE confirmed that the purified protein was of the expected size for rRabGRO-fp (approximately 14,000 kDa) (Fig. 1A). The rRab-GRO-fp was then treated with enterokinase, and the products were purified by reverse-phase HPLC (Fig. 1B). The HPLC peak labeled "rRabGRO" was found to contain mature rRab-GRO of the expected sequence as determined by N-terminal amino acid sequencing.
Bioactivity of rRabGRO-The chemotactic activity of rRab-GRO was tested in vitro using rabbit and human neutrophils. For comparison, recombinant human GRO-␣ (R & D Systems) was also tested. Recombinant RabGRO protein stimulated chemotaxis of rabbit neutrophils at concentrations as low as 3 ϫ 10 Ϫ10 M, with maximal activity at approximately 10 Ϫ7 M and a decline in activity at 10 Ϫ6 M (Fig. 2A). In contrast, rRabGRO had little chemotactic activity for human neutrophils over the same concentration range. Similarly, human GRO-␣ showed no significant chemotactic activity for rabbit neutrophils, despite being chemotactically active for human neutrophils (Fig. 2B).
To assess in vivo bioactivity, rRabGRO was instilled endobronchially into a rabbit. This produced a brisk influx of neutrophils into the lung (45% neutrophils in the BALF cell count from the treated lung at 4 h). For comparison, the BALF from the untreated lung in the same rabbit and from the lungs of a rabbit treated with sterile 0.9% NaCl alone contained only 6 and 5% neutrophils, respectively.
Generation and Specificity of Goat Polyclonal Anti-rRabGRO IgG-After isolating total IgG from immune goat serum, the specificity of the anti-rRabGRO IgG was tested by Western analysis using rRabGRO, recombinant rabbit MCP-1, rabbit IL-8 (each produced using similar methods (18)), and recombinant human GRO-␣. The anti-rRabGRO IgG reacted strongly with both the fusion protein and the cleaved mature rRabGRO protein (data not shown). Although there was very weak crossreactivity with human GRO-␣, there was no cross-reactivity with either rabbit IL-8 or rabbit MCP-1. The specificity of the goat anti-rRabGRO antibodies for native rabbit GRO protein was further confirmed by RIP (described below).
Expression of GRO Protein by AM-We performed RIP and ELISA measurements of stimulated rabbit AM supernatants to measure the expression of newly synthesized and secreted GRO protein. The immunoprecipitation experiments were conducted using the supernatants of adherent rabbit AM metabolically labeled with [ 35 S]cysteine and stimulated for 19 h with a variety of endogenous and exogenous proinflammatory agents. Immunoprecipitation of these supernatants with the goat anti-rRabGRO IgG resulted in a major band of the expected size (approximately 7 kDa) for each stimulus (Fig. 3). Further analysis of the gels by phosphorimaging allowed quantitation of GRO protein in the supernatants. SiO 2 , Al 2 O 3 , LPS, human TNF-␣, and HKSA were each strong inducers of GRO secretion, while ConA and heat-inactivated human TNF-␣ were very weak stimuli. GRO protein was not detected when RIP was performed with nonimmune goat serum.
Because adherence alone appeared to be a stimulus for GRO protein production in the RIP studies, we measured GRO levels by ELISA using the supernatants of rabbit AM cultured in polypropylene tubes which were periodically agitated. These results showed significant time-and stimulus-dependent differences in the accumulation of GRO protein in the supernatants of relatively nonadherent AM (Fig. 4). relatively small amounts of GRO protein during the first 6 h, but relatively high concentrations of GRO were present in the supernatants at 20 h. Stimulation with ConA and Al 2 O 3 did not result in significant secretion of GRO when compared with incubation with 0.9% NaCl (negative control). Likewise, human TNF-␣ and rabbit IFN-␥ (both intact and heat-inactivated) failed to stimulate increased amounts of GRO in the supernatants of nonadherent AM (data for heat-inactivated proteins not shown).
GRO Protein Levels in E. coli Pneumonia-Significantly increased amounts of GRO protein were detected in the BALF of rabbits treated with endobronchial E. coli when compared with rabbits treated with NaCl alone at both 4 and 24 h (p Ͻ 0.03 by unpaired t tests for both groups) (Table I). There was also a significant increase in the number and percentage of neutrophils in the E. coli-treated animal BALF at both 4 and 24 h when compared with the NaCl-treated animals (p Ͻ 0.0001 for both groups) and a significant positive relationship between GRO protein levels and the number and percentage of BALF neutrophils (r ϭ 0.82, p Ͻ 0.001 for BALF total neutrophils versus BALF GRO concentration).
Localization of GRO Gene Expression in the Lung in E. coli Pneumonia-To better define which lung cells in the lung express GRO mRNA in acute pulmonary inflammation, we used in situ hybridization techniques to localize GRO mRNA in lung tissue from rabbits with E. coli pneumonia. Hybridization signals with the RabGRO antisense probe were detected in bronchial epithelial cells and in inflammatory cells in the lung parenchyma around the airways (Fig. 5, A and C). The tissue inflammatory cells were strongly positive and appeared to be either macrophages or neutrophils infiltrating the peribronchial lung parenchyma. Hybridization with the negative control (sense) RabGRO probe produced virtually no signals (Fig.  5, B and D). DISCUSSION Alveolar macrophages are important cellular mediators of pulmonary inflammation, in part due to their ability to elaborate a variety of leukocyte chemotactic factors (20). Among these factors, IL-8 is thought by many investigators to be the major neutrophil chemoattractant in the lung (21). However, studies using specific antisera to neutralize IL-8 in the supernatants of activated human AM (22) and BALF from patients with adult respiratory distress syndrome (23) suggest that a significant proportion of AM-derived neutrophil chemotactic activity is due to other non-IL-8 chemoattractants. Possible candidates include the GRO subgroup of ␣-chemokines, which are closely related to IL-8 and are also potent neutrophil chemoattractants and activators (2). In this study, our goals were: 1) to express recombinant rabbit GRO protein and develop a specific and sensitive ELISA; 2) to investigate GRO protein expression by rabbit AM in response to stimulation with both endogenous and exogenous inflammatory agents; and 3) to determine whether GRO protein and mRNA can be detected in the lung in vivo during the acute inflammatory response that accompanies bacterial pneumonia.
In order to generate the species-specific reagents necessary to measure rabbit GRO protein, we first produced rRabGRO as a fusion protein in E. coli. We used a prokaryotic expression system for producing the recombinant protein because GRO proteins have been shown not to undergo post-translational sulfation, glycosylation, or phosphorylation (24). RabGRO amino acid sequence alignment with all reported full-length GRO homologues demonstrated identities ranging from 41% for 9E3 (chicken GRO homologue) to 78% for human GRO-␤ (Table II). Given the high degree of identity with human GRO proteins, it is perhaps surprising that rRabGRO had little chemotactic activity for human neutrophils. However, this confirms previous observations by other investigators that 125 Ilabeled human GRO-␣ does not bind to rabbit neutrophils (25) and further underscores the necessity of using species-specific reagents when studying ␣-chemokines (26).
The rRabGRO-fp proved to be a good immunogen for raising specific goat anti-rabbit GRO polyclonal antibodies, as Western   FIG. 3. Radioimmunoprecipitation of GRO protein from the supernatants of adherent rabbit AM. Adherent rabbit AM stimulated for 19 h with crystalline SiO 2 (100 g/ml), Al 2 O 3 (100 g/ml), LPS (10 g/ml), human TNF-␣ (500 U/ml), heat-inactivated (hi-) human TNF-␣ (500 units/ml), ConA (5 g/ml), HKSA (1 ϫ 10 8 bacteria/ml), or NaCl (0.9%). The autoradiogram of radiolabeled proteins in the supernatants, resolved by SDS-PAGE, is shown on top, and the corresponding analysis by phosphorimaging is shown below. The "ϩ" lanes denote incubation with goat anti-rRabGRO IgG and the "Ϫ" lanes refer to incubation with nonimmune goat IgG. The vertical axis is the relative intensity of the signal from the major band (7 kDa, the expected size for RabGRO protein) in each lane compared with the background radioactivity of the gel.
FIG. 4. Rabbit GRO protein levels in stimulated nonadherent AM supernatants. Measurement of GRO protein (ng/ml) by immunoassay in the supernatants of rabbit AM incubated with crystalline SiO 2 (100 g/ml), Al 2 O 3 (100 g/ml), ConA (5 g/ml), HKSA (1 ϫ 10 8 bacteria/ml), human TNF-␣ (500 units/ml), LPS (10 g/ml), rabbit IFN-␥ (500 units/ml), or 0.9% NaCl for the times indicated. analysis confirmed that the goat antibodies did not cross-react with either rabbit IL-8 or MCP-1. The antibodies did show a small amount of cross-reactivity for recombinant human GRO-␣, which is not surprising given the high degree of homology between these proteins (Table II). As RabGRO and RPF2 are nearly identical at the mature protein level (93% identity), the anti-rRabGRO antibodies are likely to cross-react with RPF2 as well.
We have also shown that rabbit AM express and secrete GRO protein in response to stimulation by LPS, TNF-␣, SiO 2 , Al 2 O 3 , and HKSA. Adherence and/or culture conditions alone also appear to stimulate GRO expression. Under nonadherent conditions, LPS, HKSA, and SiO 2 were the most potent inducers of GRO protein secretion, whereas human TNF-␣ was a weak stimulus. By contrast, human TNF-␣ was a potent stimulus for GRO protein production in adherent AM. Cell adherence has been shown to induce the expression of both TNF-␣ and IL-1␤ in AM (27,28), and it is possible that increased expression of these cytokines in adherent AM may in turn mediate increased GRO protein expression by these cells. The increase in GRO secretion by rabbit AM in response to the tested stimuli is in agreement with studies performed with AM from other species. In human AM, GRO expression has been described in response to LPS (10,12), TNF-␣ (12), and S. aureus (11). In rat AM, the GRO homologue KC (CINC-1) is induced by LPS, and MIP-2 expression increases following stimulation with TNF-␣, LPS, and SiO 2 (29). Incubation with rabbit IFN-␥ did not cause rabbit GRO expression under nonadherent conditions. It is possible that IFN-␥ may actually inhibit GRO protein secretion by AM, as the measured concentrations were less than those observed with 0.9% NaCl alone. Although this finding needs to be confirmed in further studies, it is consistent with a prior report that IFN-␥ inhibits KC gene expression in mouse peritoneal macrophages (30).
The data also show that GRO protein is present in the lungs of rabbits with Gram-negative bacterial pneumonia. All of the rabbits with E. coli pneumonia had elevated concentrations of GRO in BALF, and the amounts correspond to biologically significant levels based on the chemotactic activity of the recombinant GRO protein that we measured in vitro. Moreover, the amount of GRO protein present in the rabbit BALF is roughly equivalent to the levels of IL-8 present in the BALF of patients with the adult respiratory distress syndrome (23). While the in vitro data suggest that AM are probably a significant source of GRO protein in the lung, many other lung cell types can produce GRO chemokines, and each may be an important contributor to the expression of GRO in the lungs, depending on the in vivo circumstances. In rabbits with Gramnegative bacterial pneumonia, the in situ hybridization studies indicate that airway epithelial cells and tissue inflammatory cells are significant sources of GRO gene expression within 4 h of bacterial entry into the lungs. These results agree with studies by Becker and associates who showed increased GRO mRNA expression in human airway epithelial cells in response to LPS-stimulation (12) and with studies by Xing et al. showing localization of GRO mRNA expression in parenchymal inflammatory cells in LPS-treated rat lung tissue (31). In contrast, Rogivue and colleagues (32) have reported in bovine pneumonic pasteurellosis that GRO protein is detectable in type II alveolar epithelial cells and mesothelial cells, but not in bronchial epithelial cells or pleural fibroblasts. Lung fibroblast cell lines (33) and endothelial cells (6) also have been shown to express GRO homologues in vitro, but we did not find evidence of GRO production by these cell types in vivo in bacterial pneumonia.
Using a cDNA library prepared from LPS-stimulated rabbit AM and Northern analysis, we have shown previously that LPS-stimulated rabbit AM express the genes for two GRO FIG. 5. Localization of GRO mRNA in lung tissue in E. coli pneumonia by in situ hybridization. Lung tissue specimens (9-m sections) from a rabbit treated with the endobronchial instillation of 10 9 colony-stimulating factor/ml E. coli at 4 h. A and C, hybridization with the RabGRO antisense probe; B and D, hybridization with RabGRO sense probe (negative control). The large arrowheads indicate bronchial epithelial cells and small arrowheads refer to parenchymal inflammatory leukocytes. Specimens were counterstained with hematoxylin and eosin (magnification: ϫ 450).
In conclusion, we have found that rabbit AM produce two GRO homologues that are distinct from a GRO chemokine produced by rabbit endothelial cells. We have expressed the predominant GRO homologue produced by rabbit AM, developed a specific immunoassay, and used this to show that rabbit AM produce GRO protein in response to both exogenous (LPS, HKSA, and SiO 2 ) and endogenous (TNF-␣) inflammatory stimuli. Finally, we have shown that GRO mRNA and biologically significant levels of GRO protein are present in the lungs of rabbits with E. coli pneumonia. The relative contribution of GRO proteins (particularly in comparison to other ␣-chemokines) to the inflammation accompanying lung injury needs to be clarified. In rabbit models of lung reperfusion injury (38), acid aspiration (39), and bacterial pneumonia (40), the closely related ␣-chemokine IL-8 has been shown to have a prominent role in mediating pulmonary neutrophil infiltration and tissue injury. However, rats pretreated with neutralizing antibodies to GRO homologues have significantly decreased BALF neutrophil accumulation in response to the intratracheal instillation of LPS (41). Apart from their potential role in recruiting neutrophils to the lungs, GRO and other ␣-chemokines have bioactivities that may be important in other aspects of pulmonary inflammation, such as mediating chemotaxis and/or activation of leukocytes other than neutrophils (6,(42)(43)(44), collagen turnover (4), and vascular proliferation (8). The exact roles of GRO chemokines and related proteins in acute and chronic inflammatory processes in the lung warrant further study.