Critical Involvement of the Chemotactic Axis CXCR4/Stromal Cell-Derived Factor-1α in the Inflammatory Component of Allergic Airway Disease

Stromal cell-derived factor-1α/β (SDF-1α/β;² also known as pre-B cell growth stimulating factor), the only chemokine described to bind CXCR4 (fusin, LESTR, npy3r) (1, 2), is expressed constitutively in a wide variety of tissues and is also a powerful chemoattractant for T lymphocytes, among other cell types (3). Mice genetically deficient for SDF-1α/β exhibit a severe defect in cardiac development and die perinatally (4). CXCR4 is expressed on human PBLs, monocytes, neutrophils, and CD34⁺ hemopoietic progenitors and has been described as the major coreceptor for HIV-1/2 on T lymphocytes (1, 2, 3, 5, 6). Similar to mice lacking SDF-1α/β, CXCR4-deficient mice die in utero (7, 8), complicating analysis of the contribution of this chemotactic axis in vivo.

Lung inflammation and other disorders associated with asthma are dependant on the actions of chemokines and their specific receptors (9, 10, 11, 12, 13, 14, 15, 16). During lung allergic inflammation, chemokine-chemokine receptor interactions direct both the recruitment of different infiltrating cell types to the lung and the production of inflammatory mediators that intensify pulmonary damage (10, 12, 15, 16, 17, 18, 19, 20).

Signals delivered through the chemokine receptor CXCR4 upon interaction with its ligand, SDF-1α/β, result in the most efficacious chemoattraction of T lymphocytes detected to date (1, 2, 3, 21, 22). Most recently, other axes, such as macrophage inflammatory protein-3β/CCR7 has been described to participate in naive T cell chemoattraction as well as dendritic cell/recirculating T cell interactions (23, 24).

T lymphocyte-derived products such as IL-4 and IL-5 are critical players in the development and progression of lung eosinophilia and airway hyper-responsiveness (25, 26, 27, 28, 29, 30). In fact, the in vivo depletion of T lymphocytes in mice or the lack of lymphocytes in genetically deficient mice prevents this pathological response (12, 26, 31). In humans, postmortem examination of the airway or bronchial biopsies of asthmatic patients reveal large T lymphocyte infiltrates (32).

The constitutive expression of SDF-1α in a wide variety of tissues, including lung, together with the lack of modulation of its expression during inflammation suggest a putative noninflammatory role of this chemokine, possibly related to the maintenance of normal leukocyte recirculation (3, 33). However, because 1) chemokine expression in the lung promotes the recruitment of inflammatory leukocytes to this organ; 2) lymphocytes are absolutely required for the development of lung allergic reactions; and 3) SDF-1α is a very efficient chemoattractant for lymphocytes, we decided to analyze the relevance of CXCR4-SDF-1α/β interactions during inflammation, specifically on leukocyte recruitment to the lung and the subsequent impact during allergic airway disease (AAD). To assess the role of CXCR4/SDF-1α in inflammation we used a mouse model of lung inflammation based on the repeated exposure of mice to aerosolized OVA (34). In this report we 1) studied the effects of the blockage of CXCR4 on three end points of AAD: leukocyte accumulation in interstitium and airway lumen as well as airway hyper-responsiveness; 2) evaluated AAD progression after CXCR4 overexpression in leukocytes following retroviral gene transduction; and 3) examined the correlation between the accumulation of specific leukocyte types and the expression of CXCR4 in the inflamed lungs. The findings presented here establish an important role for CXCR4 and SDF-1α during the development of AAD and suggest a potential relevance of this receptor and its ligand in other lymphocyte-mediated inflammatory reactions.

Rabbit polyclonal Abs against murine CXCR4 were prepared according to standard methods (35). This Ab was generated against the 16-aa peptide corresponding to aa 181–196 in the mCXCR4 peptide sequence. The sequence of the peptide was QGDISQGDDRYICDRL. This 16-aa peptide shares 85 and 69% homology with its rat and human orthologue genes, respectively, and does not appear in any other known gene described to date. The ability of the neutralizing Abs raised against the second extracellular domain of human CXCR4 to block T cell infection by certain HIV-1 and HIV-2 strains is consistent with the role of this second extracellular loop in the coreceptor activity of CXCR4 (36, 37). Rabbit serum was passaged over a protein A column, and then anti-CXCR4 Abs were purified from the flow-through on an affinity column using the same peptide (Research Genetics, Huntsville, AL).

The full coding sequence of mCXCR4 was subcloned into the mammalian cell expression vector pcDNA3.1 (Invitrogen, San Diego, CA). Stable cell lines were generated following transfection of the expression vector into HEK-293 cells using Lipofectamine (Life Technologies, Gaithersburg, MD) and selected with G418 (0.8 mg/ml; Life Technologies). Clones expressing high levels of mCXCR4 were identified by binding to ¹²⁵I-labeled SDF1-α (Amersham, Arlington Heights, IL), selected, and expanded for the experiments described here.

The mouse cell lines BAF-3 and L1.2 were grown in RPMI 1640 supplemented with 10% FCS. IL-3 (5 ng/ml) was added to cultured BAF-3 cells (38). Murine CD4⁺ T splenocytes were purified using the R&D Systems separation kit (Minneapolis, MN) and were cultured in RPMI 1640 supplemented with 10% FCS.

Cells (1 × 10⁵) were resuspended in staining buffer (0.1% BSA, 0.02% sodium azide, and PBS) incubated with 10 μg/ml (1/50) of purified anti-mouse CD16/CD32 (FcR; PharMingen, San Diego, CA) for 20 min at 4°C, and then incubated with 10 μg/ml of either anti-mouse CXCR4 Abs or normal rabbit IgG for 30 min at 4°C. Cells were washed and stained with FITC-conjugated goat anti-rabbit IgG (Dako, Glostrup, Denmark) for 20 min at 4°C. The stained cells were washed twice, collected by FACScan flow cytometer, and analyzed by CellQuest software (Becton Dickinson, Mountain View, CA). Murine bone marrow, blood, or spleen cells were stained with FITC/PE-labeled mAbs specific for CD3, CD4, CD8, B220, MAC-1, and GR-1 (PharMingen, San Diego, CA) following the protocol described above.

The in vitro migration of BAF-3 cells and CD4⁺ T splenocytes to different concentrations of SDF-1α (R&D Systems) was evaluated in duplicate using Costar Transwells (Cambridge, MA) as previously described (34). In the blocking experiments cells were preincubated with either 10 μg/ml of anti-mCXCR4 Ab or control Ab at 4°C for 30 min before their addition to the Transwell inserts.

Cells were labeled with the fluorochrome fluo-3/AM (Molecular Probes, Eugene, OR) according to the manufacturer’s recommendations. Briefly, 50 μg of fluo-3 AM was dissolved in 44 μl of DMSO and diluted to 10 μM with modified Gay’s buffer (MGB; 5 mM KCl, 147 mM NaCl, 0.22 mM KH₂PO₄, 1.1 mM Na₂HPO₄, 5.5 mM glucose, 0.3 mM MgSO₄-7H₂O, 1 mM MgCl₂, 10 mM HEPES (pH 7.4), and 0.1% BSA). Cells (10⁷) were resuspended in 1 ml of MGB and incubated with an equal volume of 10 mM fluo-3 mix for 30 min at room temperature. Excess dye was removed by centrifugation, and cells were resuspended at a concentration of 2–10⁶/ml in MGB buffer and 1.5 mM CaCl₂. Calcium influx was measured on the FACScan by analyzing FL1 (linear scale) vs time. In the blocking experiments cells were preincubated with either 10 μg/ml of anti-mCXCR4 Ab or control Ab at 4°C for 30 min before evaluation of Ca²⁺ mobilization.

Eight- to 10-wk-old C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and kept in a specific pathogen-free mouse facility (Millennium Pharmaceuticals, Cambridge, MA). The mouse model of lung inflammation used here consists of a sensitization phase (OVA, 0.1 mg/mouse i.p. on day 0; Sigma, St. Louis, MO) and an induction of the response phase (2% OVA for 5 min i.n. on day 8 and 1% OVA for 20 min. i.n. on days 15–21; Fig. 3,A). PBS (i.p. and/or i.n.) was administered to mice as a negative control. For the blocking experiments, mice also received 10 μg/mouse of neutralizing polyclonal Abs against either CXCR4 (described here) or SDF-1α (R&D Systems). These Abs were administered i.v. and 30 min before OVA provocation on day 8 and days 15–21 or on days 20–21 (Fig. 3 A). OVA-treated control mice were injected with the same amount of control Ab at the same time points indicated during treatment. Rabbit Ig fraction (Dako, Santa Barbara, CA) was used as a control for the CXCR4 Ab. Three hours after OVA administration on day 21, mice were sacrificed by CO₂ asphyxiation and analyzed for leukocyte accumulation and AHR.

Bronchoalveolar lavage (BAL) and AHR analysis were performed as previously described (16, 34). AHR was expressed as the enhanced pause (Penh), a calculated value, which correlates with measurements of airway resistance, impedance, and intrapleural pressure in the same mouse: Penh = (Te/Tr-1) × (Pef/Pif) (Te, expiration time; Tr, relaxation time; Pef, peak expiratory flow; Pif, peak inspiratory flow × 0.67 = coefficient). The relaxation time is the time it takes for the box pressure to change from a maximum to a user-defined percentage of the maximum. Here, Tr measurement begins at the maximum box pressure and ends at 40%. AHR was measured 3 h after the last Ag challenge by recording respiratory pressure curves by whole body plethysmography (Buxco Technologies, Sharon, CT) in response to inhaled methacholine (Aldrich, Chicago, IL).

Total BAL cell counts were performed, and aliquots (5 × 10⁵ cells/slide) were pelleted onto glass slides by cytocentrifugation. To determine the number of eosinophils and neutrophils, slides were stained with Wright-Giemsa (Fisher Diagnostics, Pittsburgh, PA). T lymphocytes, B lymphocytes, and mononuclear phagocytes were identified by Thy 1.2 (53-2.1; PharMingen, San Diego, CA), IgM (II/41; PharMingen, San Diego, CA), and Moma-2 (BioSource, Camarillo, CA) staining, respectively, as previously described (12). The percentages of eosinophils, lymphocytes, neutrophils, and macrophages were determined by counting their number in eight high power fields (×40 magnification; total area, 0.5 mm²) per area randomly selected and dividing this number by the total number of cells per high power field. To obtain the absolute number of each leukocyte subtype in the lavage, these percentages were multiplied by the total number of cells recovered from the BAL fluid.

Lung sections from the different experimental groups of mice were prepared as previously described (12). Sections (4 μm) were stained with hematoxylin/eosin according to standard protocols. An estimation of the percentage of each leukocyte subtype within the infiltrate in OVA- plus Ab-treated mice or OVA- plus rabbit Ig-treated controls was made by counting 200 cells in one randomly selected peribronchiolar infiltrate and determining the numbers of eosinophils, monocytes, and lymphocytes present.

Total RNA from the lungs of OVA-treated mice or control littermates at different time points was extracted by a single-step method using RNA STAT-60 (Tel-Test, Friendswood, TX). CXCR4 mRNA expression was determined by Multiprobe RNase protection assay as previously described (16). A 403-bp mCXCR4 probe was derived by PCR using the following primers: 5′-GTAATACGACTCACTATAGGGAACGCTGCTGTAGAGGTTGAC-3′ and 5′-GTAACCACCACGGCTGTA-3′. The identity and quantity of each mRNA species in the original RNA sample were determined based on the signal intensities given by the appropriately sized, protected probe fragment bands. Values were created by expressing mCXCR4 up-regulation relative to its expression in normal tissue. The sample loading was normalized by the housekeeping gene, GAPDH, which is included in each template set.

Protein expression was determined in noninflamed and inflamed mouse lung tissue samples with either polyclonal rabbit anti-mCXCR4 Ab or polyclonal goat anti-mouse SDF1α (R&D Systems) followed by a modified avidin/biotin staining method as previously described (12). Peripheral blood smears from control mice (BM-mock RV) and CXCR4 overexpressing littermates (CXCR4 RV) were fixed in acetone and immunostained by overnight incubation at 4°C with the anti-mouse CXCR4 Ab diluted in PBS/0.05% Tween. Slides were washed in PBS, then incubated for 30 min at room temperature in PBS/Tween containing a 1/200 dilution of FITC-labeled goat anti-rabbit IgG (Southern Biotechnology Associates, Birmingham, AL) and a 1/30 dilution of PE-labeled mouse anti-CD45.1 (Ly5.2). Slides were washed twice in PBS/Tween and mounted in Fluoromount-G (Southern Biotechnology Associates) for microscopy. Control slides were stained with an isotype-matched negative control Ab instead of primary Ab, or biotinylated anti-rabbit or goat Ig or streptavidin complex was selectively omitted.

The open reading frame of CXCR4 cDNA was cloned into pMSCVneoEB (39). Clones containing sense orientation of inserts were identified and then purified by MaxiPrep (Qiagen, Valencia, CA) and transfected into 293 Ebna (Invitrogen) using Lipofectamine (Life Technologies) reagent along with pN8e gagpol and pN8e EnvE wobble. Viral supernatants were harvested through 0.2-mm pore size filter and then used to infect tunicamycin-pretreated ΩE cells (50 mg/ml) with polybrene (8 mg/ml; Sigma) to generate pools of stable virus-producing cells under G418 selection (1 mg/ml; Life Technologies).

For donor BM cells, BL6.SJL (ly5.1) mice were injected with 5-fluorouracil (15 mg/kg), and after 4 days BM cells were harvested (recovery being around 2 × 10⁶ cells/mouse). BM cells (4 × 10⁶ cells) were cocultured for 72 h in a 100-mm dish with mitomycin C-treated CXCR4 producer cells at 80% confluence in RPMI containing 10% FCS, 1% penicillin-streptomycin, 1% essential amino acids, and the following growth factors and cytokines at 0.2 μg/ml: IL-3, SCF, and IL-6. At the time of transplant, lethally irradiated C57/B6 mice (950 rad) were injected i.v. with 1 × 10⁵ virally infected bone marrow cells. Mice were monitored at 12 days and 5 wk posttransplant in the peripheral blood for evidence of reconstitution and gene expression before analysis at 9 wk. At this time mesenteric lymph nodes and bone marrow were phenotyped to demonstrate that neither viral infection nor expression of transgene affected the normal distribution of leukocyte/lymphocyte subsets. Nine weeks after reconstitution, expression studies and migration studies were performed on retrovirally transduced mice following the protocol explained above.

Because CXCR4-deficient mice die in utero (7, 8), neutralizing Abs against CXCR4 were designed to study the contribution of this chemokine receptor to inflammatory allergic reactions in vivo. Affinity-purified anti-mCXCR4 Abs that reacted positively with the peptide against which they were prepared gave positive immunostaining with a variety of mouse hemopoietic cells. Anti-mCXCR4 Abs strongly immunostained K293 cells transfected with a cDNA encoding the full coding sequence of the mCXCR4 gene, B cell lines such as BAF-3 and L1.2, monocytic cell lines such as RAW287 and J774, T cell lines such as AE-7 and Dorris, as well as CD4⁺ purified splenocytes, blood mononuclear cells, and blood neutrophils (Fig. 1, A and B, and data not shown). All these cell types also show CXCR4 mRNA expression (data not shown). Eosinophils from IL-5 transgenic mice, CHO cells, and mock-transfected K293 cells that did not show CXCR4 mRNA expression by PCR were not stained by the same Ab (Fig. 1, A and B, and data not shown). Similarly, K293 cells transfected with a mouse and/or human cDNA encoding CCR1, CCR2, CCR3, CCR4, CCR6, CCR7, CCR9, D6, and CXCR2 were not stained by the anti-mCXCR4 Abs (data nor shown), indicating that cross-reactivity with other chemokine receptors can be excluded.

The specific staining of BAF-3 cells (like that of K293 CXCR4 transfectants (not shown)) by anti-CXCR4 Abs was completely blocked by preincubating the Ab with the peptide to which the Abs were prepared against (Fig. 1,C) and was almost completely blocked by preincubating BAF-3 cells with SDF-1α (Fig. 1 C).

The ability of the anti-CXCR4 Abs to neutralize the action of SDF-1α on different cell lines expressing CXCR4 was analyzed. BAF-3 cells were found to migrate (25–30%) in response to SDF-1α (Fig. 2,A). This migration was completely abrogated after preincubation of these cells with anti-mCXCR4 Abs (Fig. 2,A). Blockage of SDF-1α-induced BAF-3 cell migration was achieved by the anti-mCXCR4 Abs in a dose-dependent manner (data not shown). Eotaxin-induced migration of eosinophils from IL-5 transgenic mice was not affected by the anti-mCXCR4 Abs, confirming the specificity of these Abs for CXCR4 (Fig. 2 A).

L1.2 cells that were strongly stained by the anti-mCXCR4 Abs responded well to SDF-1α, as shown by Ca²⁺ influx (Fig. 2,B and data not shown). Anti-mCXCR4 Abs (10 μg/ml) also blocked the SDF-1α-induced Ca²⁺ influx of L1.2 cells (Fig. 2 B).

Taken together, these results demonstrate the ability of the anti-mCXCR4 Abs to neutralize functional responses induced by SDF-1α via CXCR4 in a specific manner.

Monocytes and lymphocytes are central to the inflammatory process that results in eosinophilia and AAD (12, 16, 26, 27, 28, 29). These cell types also respond functionally to SDF-1α via CXCR4 in chemotaxis assays in vitro (40). In vivo blockage experiments of this chemokine receptor during AAD were performed using the specific neutralizing anti-CXCR4 Abs that were characterized above (Figs. 1 and 2). The mouse model of OVA-induced inflammatory response studied here induces the accumulation of monocytes and macrophages in both lung interstitium and airways that becomes maximal at early stages (sensitization phase) of the response (3 h after OVA challenge on day 15) and an increasing interstitium and airway accumulation of eosinophils and T lymphocytes that reaches its maximum at later stages (challenge phase) of the response (12, 16) (3 h after OVA challenge on day 21; Fig. 3). Development of airway hyper-responsiveness is another feature of this model, reaching its maximum at later stages of the response (12, 16).

Neutralizing anti-CXCR4 Abs (10 μg/mouse/day) were delivered i.v. daily either during the whole OVA i.n. treatment (days 8 and 15–21, 30 min before OVA challenge) or exclusively at very late stages of the response (days 20–21; Fig. 3 A) to evaluate the involvement of CXCR4 in the sensitization phase of the phenotype as well as during the challenge phase. In either case, analysis was performed 3 h after OVA challenge on day 21. Because the location of infiltrating cells within the lung correlates strongly with the severity of the inflammatory response and AHR (41, 42, 43), leukocyte enumeration was performed in the airway lumen (BAL fluid) as well as in the interstitium (lung sections) after OVA treatment.

Effect of CXCR4 neutralization on the pulmonary recruitment of lymphocytes and monocytes. Mononuclear cell numbers were reduced by 50% in the BAL fluid of OVA-treated mice following CXCR4 neutralization (Fig. 3,B). Because 90% of these mononuclear cells are CD4⁺ T lymphocytes (12), this cell subset is clearly affected by the CXCR4 blockage. A decrease in mononuclear cell infiltration in the lung interstitium of these mice was also detected compared with that in OVA-treated control littermates (Fig. 4, A and B, and Table I). In fact, when sections were examined blindly and assigned a morphological score based upon the extent and size of peribronchiolar infiltrates, CXCR4 blockage was seen to decrease the average score to 2.1 ± 0.6 compared with a mean score of 4.5 ± 0.5 in control mice (Fig. 4 and Table I). Resident macrophage numbers in the lung of OVA-treated mice were not affected by the CXCR4 blockage (Fig. 3,B and 4). No neutrophil infiltration was detected in the lung in these groups of experimental mice (Fig. 4 and data not shown).

Effect of CXCR4 neutralization on lung eosinophilia. Eosinophil accumulation in the lumen of the airways and in the lung interstitium was reduced by 50 and 60%, respectively, after CXCR4 blockage (Figs. 3,B and 4 and Table I). Because eosinophils do not express CXCR4 or respond to SDF-1α, the observed decrease in eosinophilia is likely to be a secondary consequence of the CXCR4 neutralization-induced reduction of mononuclear cell (monocyte and/or lymphocyte) accumulation in the lung.

In fact, monocytes and lymphocytes have been described to promote eosinophil infiltration in this organ (12, 13, 14, 15, 16, 26, 27, 28, 29). This interpretation was supported by experiments in which neutralization of CXCR4 was performed exclusively in the last 2 days of the treatment (days 20 and 21). Under these conditions, only a small reduction (<10%, which did not reach statistical significance) in OVA-induced eosinophil and mononuclear cell infiltration (in both the airway lumen and lung interstitium) was detected after CXCR4 neutralization (Fig. 3 B and data not shown).

The effect of neutralization of the CXCR4 ligand, SDF-1α, was also assessed in these studies. A similar reduction in the total number of eosinophils and mononuclear cells was detected in both the BAL fluid and lung interstitium of OVA-treated mice following SDF-1α neutralization compared with that induced by OVA following CXCR4 neutralization (Fig. 4 and Table II). As observed in the CXCR4 neutralization experiments, this reduction was only achieved when anti-SDF-1α Ab was administered from early stages of the inflammatory response (data not shown).

Effect of CXCR4 neutralization in the induction of AHR. To determine whether the reduction in lung eosinophilia observed following CXCR4 neutralization also influences airway function, AHR was evaluated in OVA-treated mice after neutralization of this receptor. Mice received 10 μg/mouse of neutralizing polyclonal Abs against either CXCR4 or SDF-1α. These Abs were administered i.v. and 30 min before OVA provocation on day 8 and days 15–21. Rabbit Ig fraction was used as control for the CXCR4 Ab. AHR was evaluated 3 h after the last OVA administration on day 21. Fig. 5 shows that after methacholine provocation on day 21, the experimental group of mice that was subjected to CXCR4 neutralization showed a reduction of AHR of 30–35% compared with that detected in OVA control-treated littermates at the same time point. Similarly, SDF-1α neutralization reduced OVA-induced AHR to a similar extent (Table II). This reduction was observed at the point of maximal lung leukocyte accumulation (day 21, 3 h).

Effect of CXCR4 overexpression in leukocytes during lung allergic inflammation. We next tested whether overexpression of CXCR4 in leukocytes results in an increased inflammatory response during AAD. Mouse C57BL/6.SJL bone marrow progenitor cells were infected with an MSCV retrovirus containing the CXCR4 cDNA. These retrovirally transduced BM progenitors were then used to generate lethal irradiation chimeras using a standard protocol based on the use of these congenic mice with a different allotype in the Ly5 locus (see Materials and Methods for more details).

Nine weeks after transplantation, CXCR4 protein expression was demonstrated in 100% of CXCR4 retrovirally transduced donor derived (Ly5.2⁺) peripheral blood cells (Fig. 6). Conversely, low endogenous CXCR4 expression was detected when an empty virus was used as a control (Fig. 6). No differences in total cell numbers and phenotype were detected when bone marrow, blood, thymus, peripheral lymphoid organs, and BAL fluid from the CXCR4-transduced mice were compared with their equivalent organs in the control littermates (data not shown). This suggests that the overexpression of CXCR4 does not alter leukocyte differentiation and normal distribution of these cells in tissues.

To assess the effect of CXCR4 overexpression during AAD, CXCR4 retrovirally transduced mice were subjected to the OVA treatment described above. As in the CXCR4 blockage experiments, lung leukocyte infiltration was evaluated 3 h after OVA challenge on day 21. OVA-induced eosinophil accumulation in the airway lumen (eosinophil BAL counts) was increased in CXCR4-transduced mice (1300 × 10³and 172 × 10³) compared with that observed in the control mice (900 × 10³and 129 × 10³; Fig. 7). Mononuclear cell numbers (lymphocytes and monocytes) were also increased in the BAL fluid of OVA-treated CXCR4-transduced mice (337 × 10³plus 26.7 × 10³) compared with that observed in the control mice (241.6 × 10³ plus 29.1 × 10³; Fig. 7). Increased eosinophilia in the lung interstitium of these mice was also detected compared with that observed in OVA-treated control littermates (Fig. 8). The morphological scores of peribronchiolar, perivascular, and interstitial infiltrates (extent and size) showed that CXCR4 overexpression increases the average score to 4.5 ± 0.3 compared with a mean score of 3.5 ± 0.2 in OVA-treated control mice (Fig. 8). Macrophage numbers were similar in the lungs of both groups of mice (Figs. 7 and 8). It is important to note that no neutrophil infiltration was detected in the lung of CXCR4-transduced mice (Figs. 7 and 8 and data not shown). These data indicate that overexpression of CXCR4 on leukocytes results in an increased response to inflammatory stimuli.

The results presented above show that modulation of the levels of CXCR4 on leukocytes (decreased levels by Ab neutralization or increased levels by gene transduction) correlates well with the intensity of the inflammatory response in the lung following OVA challenge in this particular mouse model. Therefore, we tested whether the detection of CXCR4-positive cells parallels the course of the inflammatory response during AAD.

This is specially relevant, because its ligand, SDF-1α, is expressed in the lung even under noninflammatory conditions, and no significant change in its already high expression levels is observed during the course of this AAD model (see Fig. 10 and data notshown). The main source of SDF-1α protein before or after challenge is alveolar and bronchial epithelium (see Fig. 10).

CXCR4 mRNA expression was studied by RNase protection assay (Fig. 9). Although there was little CXCR4 mRNA expression on day 0, by day 15 there was a 2-fold increase in expression, which peaked on day 21. Day 15 represents the time of peak infiltration of monocytes in this model (12). Cellular localization of CXCR4 protein within inflamed lungs was then determined by immunohistochemistry using anti-CXCR4 Abs described above. Concomitant with the RNA expression pattern, there was minimal expression of CXCR4 in normal lungs (Fig. 9,Bi), but a significant increase in expression was evident on day 21 (Fig. 9 Bii). The main cell type expressing CXCR4 was mononuclear cells infiltrating the tissue and mostly T lymphocytes, which make up the majority of infiltrating mononuclear cells at this time point (data not shown) (12).

We propose that this increase in CXCR4 mRNA and protein is due mainly to the accumulation of CXCR4⁺ cells and to some extent to the increase in CXCR4 levels in these CXCR4⁺ cells. In fact, a higher density of CXCR4 molecules per cell was observed in lung mononuclear cells compared with that in peripheral blood or spleen mononuclear cells during AAD (data not shown).

In this report a mouse model of lung allergic inflammation has been used to evaluate in vivo the contribution of the CXCR4-SDF-1α axis during AAD. This model, which has been used in both the mouse (12, 16, 34) and the guinea pig (10), is based on the observation that the administration of OVA induces massive eosinophil infiltration and AHR (10, 12, 28, 29, 31, 34, 44). These are the two principal parameters of lung disease analyzed. Even though most of the models of OVA-induced AAD combine the OVA Ag in alum to induce a stronger response, preimmunization with higher doses of OVA alone (0.1 mg) and subsequent aerosolized challenge induce similar monocyte, CD4⁺ T cell, and eosinophil lung infiltration, Th2 cytokine production, and AHR (10, 12, 28, 29, 31, 34, 44). However, it is also known that the mechanisms involved in the induction of the disease are influenced by the route and dose of administration of the Ag. In fact, depletion of T cells abrogates eosinophil infiltration by modulating eotaxin expression in some AAD models (44) or without affecting chemokine expression in some others (12). The mouse strain used is also critical. Different components, such as IL-4 or IL-5, are central or secondary players in the induction of the inflammatory response, which is influenced by the mouse genetic background (28, 29). As indicated, mouse C57BL/6.SJL bone marrow progenitor cells were infected with an MSCV retrovirus containing the CXCR4 cDNA. Lethal irradiation chimeras were generated using a standard protocol based on the use of these congenic mice with a different allotype in the Ly5 locus. For comparison of the OVA response in the bone marrow reconstitution experiments with CXCR4 or SDF-1α neutralization experiments, the C57BL/6J strain was chosen. C57BL/6J mice show a relatively modest OVA-induced AHR response compared with that observed in BALB/c mice. However, 1) the OVA protocol used here has been previously assessed in both BALB/c and C57BL/6J strains of mice with very similar results (12, 34); 2) significant OVA-induced lung allergic inflammation reactions have been induced in genetically mice made deficient for cell types, cytokines, or adhesion receptors in a C57BL/6J background (12); and 3) BALB/c and C57BL/6J mice subjected to CXCR4 or SDF-1a neutralization during OVA treatment showed similar phenotype (data not shown).

Taken together, our results indicate that despite the high levels of SDF-1α in the lung before or after the induction of an inflammatory response, the CXCR4/SDF-1α axis plays a critical role in the recruitment of inflammatory leukocytes into the lung and the subsequent induction of pathophysiological manifestations characteristic of asthmatic reactions. It is conceivable that the expression of a particular chemokine is necessary, but not sufficient, for the induction of cell recruitment. Eotaxin is highly expressed in the mouse thymus under noninflammatory conditions, and yet no eosinophilia has been detected in this particular organ (34). Thus, SDF-1α is constitutively expressed in the lung, but presumably other events must be required, such as up-regulation of its receptor CXCR4, adhesion receptor overexpression, and production of other inflammatory mediators, to achieve the inflammatory response.

Monocytes and eosinophils have been described to promote lung eosinophilia in this particular mouse model (16). Because CXCR4 is not expressed on eosinophils, and SDF-1α does not induce eosinophil migration in vitro, the CXCR-4-SDF-1α interaction could be indirectly regulating the accumulation of eosinophil by acting through its target cell types, monocytes and T lymphocytes. In this model of AAD, monocyte accumulation occurs at early stages of the inflammatory response and precedes the development of eosinophilia (12). T lymphocyte recruitment in the lung parallels eosinophil accumulation (12). Accordingly, neutralization of CXCR4-mediated interactions should be more critical at early stages of the inflammatory response. No effect or few effects are detected when CXCR4 signals were neutralized late in the induction of the disease (days 20 and 21; Fig. 3). One could argue that the longer administration of the anti-CXCR4 Abs could induce a systemic depletion of CXCR4-expressing leukocytes. Spleen and bone marrow cellularity have been evaluated during anti-CXCR4 blockage. The number and phenotype of cells from both organs were not affected by the Ab compared with those in control littermates (OVA or OVA plus rabbit Ig; data not shown).

In conclusion, we provide three lines of evidence to support the crucial role of CXCR4-SDF1α interactions during AAD: 1) there is a correlation between the influx of CXCR4⁺ cells (supported by an increase in CXCR4 mRNA in the lungs) and the progression of the inflammatory phenotype; 2) the neutralization of CXCR4-mediated signals with either anti-CXCR4 or anti-SDF-1α Abs leads to a reduced inflammatory and AHR response to OVA; and 3) the overexpression of CXCR4 by retroviral delivery of a CXCR4 cDNA in leukocytes augments lung eosinophilia. On the basis of these findings we now propose that, although not considered a typical“ inflammatory chemokine, the CXCR4/SDF-1α axis plays a very significant role in the inflammatory component of AAD.

We thank Dr. Luke O’Neil for the critical reading of this manuscript. We also thank Steve Lin, Gui-Quan Jia, and Nita Bikkal for excellent skilled technical assistance.