Pivotal role of IL-6 in the hyperinflammatory responses to subacute ozone in adiponectin-deficient mice

Abstract

Adiponectin is an adipose-derived hormone with anti-inflammatory activity. Following subacute ozone exposure (0.3 ppm for 24–72 h), neutrophilic inflammation and IL-6 are augmented in adiponectin-deficient (Adipo^−/−) mice. The IL-17/granulocyte colony-stimulating factor (G-CSF) axis is required for this increased neutrophilia. We hypothesized that elevated IL-6 in Adipo^−/− mice contributes to their augmented responses to ozone via effects on IL-17A expression. Therefore, we generated mice deficient in both adiponectin and IL-6 (Adipo^−/−/IL-6^−/−) and exposed them to ozone or air. In ozone-exposed mice, bronchoalveolar lavage (BAL) neutrophils, IL-6, and G-CSF, and pulmonary Il17a mRNA expression were greater in Adipo^−/− vs. wild-type mice, but reduced in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice. IL-17A⁺ F4/80⁺ cells and IL-17A⁺ γδ T cells were also reduced in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice exposed to ozone. Only BAL neutrophils were reduced in IL-6^−/− vs. wild-type mice. In wild-type mice, IL-6 was expressed in Gr-1⁺F4/80⁻CD11c⁻ cells, whereas in Adipo^−/− mice F4/80⁺CD11c⁺ cells also expressed IL-6, suggesting that IL-6 is regulated by adiponectin in these alveolar macrophages. Transcriptomic analysis identified serum amyloid A3 (Saa3), which promotes IL-17A expression, as the gene most differentially augmented by ozone in Adipo^−/− vs. wild-type mice. After ozone, Saa3 mRNA expression was markedly greater in Adipo^−/− vs. wild-type mice but reduced in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice. In conclusion, our data support a pivotal role of IL-6 in the hyperinflammatory condition observed in Adipo^−/− mice after ozone exposure and suggest that this role of IL-6 involves its ability to induce Saa3, IL-17A, and G-CSF.

ozone is an air pollutant produced by reactions between automobile exhaust and sunlight. Once inhaled, ozone oxidizes lipids, proteins, and other species present in the lung lining fluid (18, 34, 53). Products of these reactions damage epithelial cells, cause the generation of cytokines and chemokines, and induce pulmonary inflammation characterized by neutrophil and macrophage influx into the lungs. Ozone also causes pulmonary injury leading to increased permeability of the alveolar/capillary barrier (20, 31).

Adiponectin is an insulin-sensitizing hormone secreted almost exclusively by adipocytes. Serum adiponectin is reduced in obese humans and animals and is elevated upon weight loss (2, 17, 23, 65). Adiponectin has anti-inflammatory properties in vitro and in vivo (51, 60, 63). Indeed, mice lacking adiponectin (Adipo^−/− mice) have augmented neutrophilic influx into the lungs upon subacute exposure to ozone (0.3 ppm for 24–72 h) (28, 29). This augmented neutrophilia is dependent on IL-17A (28): ozone causes greater Il17a mRNA abundance in Adipo^−/− vs. wild-type (WT) mice, and anti-IL-17A mAb treatment prior to ozone exposure reduces bronchoalveolar lavage (BAL) neutrophils in Adipo^−/− mice to levels close to those observed in WT mice. We also demonstrated increased numbers of IL-17A⁺ cells (mainly interstitial macrophages/monocytes and γδ T cells) in the lungs after ozone exposure. These IL-17A⁺ cells were augmented in Adipo^−/− vs. WT mice (28). Anti-IL-17 mAb treatment also resulted in reduced granulocyte colony-stimulating factor (G-CSF) in Adipo^−/− mice exposed to ozone, whereas other IL-17A-dependent neutrophil chemotactic factors were not affected (28).

IL-6 is a pleiotropic proinflammatory cytokine induced by infectious agents (22), in autoimmunity (47), and upon exposure to environmental toxicants, including ozone (3, 27). Indeed, the pulmonary neutrophil recruitment induced by ozone (0.3 ppm for 72 h) is partly IL-6 dependent (27, 50). Compared with WT mice, Adipo^−/− mice not only have augmented neutrophil influx but also have increased BAL IL-6 after ozone exposure (28, 29). IL-6 is important in polarizing T cells toward a Th17 phenotype (62). Hence, the purpose of this study was to examine the hypothesis that IL-6 is required for the increases in IL-17A expression and subsequent neutrophil recruitment observed in Adipo^−/− mice exposed to ozone. To test this hypothesis, we exposed WT, IL-6^−/−, and Adipo^−/− mice as well as mice deficient in both adiponectin and IL-6 (Adipo^−/−/IL-6^−/− mice) to ozone (0.3 ppm) or to room air for 72 h and assessed pulmonary inflammation and IL-17A expression. We also conducted a microarray analysis of lungs of Adipo^−/− and WT mice exposed to ozone or air. We found that serum amyloid A3 (SAA3), an acute-phase protein induced by inflammation and capable of inducing IL-17A expression in the lungs (1), was higher in Adipo^−/− mice vs. WT mice exposed to ozone and that IL-6 deficiency virtually ablated this effect of adiponectin deficiency.

METHODS

Animals.

This study was approved by the Harvard Medical School Standing Committee on Animals (IACUC). Adiponectin-deficient (Adipo^−/−) mice were originally obtained from Dr. Y. Matsuzawa (41). IL-6-deficient (IL-6^−/−) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Adipo^−/− and IL-6^−/− mice were crossbred to generate double-deficient mice (Adipo^−/−/IL-6^−/−). Sex- and age (11–13 wk old)-matched WT C57BL/6J mice were obtained from The Jackson Laboratory.

Protocol.

Mice were exposed to either ozone (0.3 ppm) or ambient air for 24, 48, or 72 h, as previously described (28). Immediately after exposure, mice were euthanized with an intraperitoneal overdose of pentobarbital. A tracheostomy was performed and the trachea was cannulated. BAL was performed and the lungs were harvested and used to extract RNA for RT-PCR. In another cohort, after BAL, blood was flushed from the lungs with ice-cold PBS. The lungs were then used for flow cytometric analysis.

Bronchoalveolar lavage.

The lungs were lavaged with two 1-ml instillations of ice-cold PBS. The lavage fluid was spun down at 1,500 rpm for 10 min at 4°C and supernatant was stored at −80°C for subsequent analysis. BAL cells were resuspended in 1 ml of ice-cold PBS and counted with a hemacytometer. Cytospin slides were prepared and stained with hematoxylin and eosin, and differential cell counts were performed on at least 300 cells. Adiponectin and proinflammatory cytokines (IL-6, CCL20, and G-CSF) were evaluated in BAL fluid by commercial ELISAs (R&D Systems). For measurements of BAL SAA3, BAL fluid was concentrated 10× by use of an ultrafiltration system with retention of molecules sized above 3 kDa (Amicon Ultra 0.5 ml 3 kDa, Millipore); 25 or 50 μl of concentrated BAL fluid was then used to assay SAA3 by ELISA (Millipore).

Extraction of RNA and gene expression analysis.

Lung RNA was extracted, cDNA was synthesized, quantitative PCR was performed, and data were analyzed as previously reported (28). We determined the expression of the following genes: Il17a, Il23, Saa3, Ccl20, and Timp1. The data were normalized to expression of 18S and expressed relative to ozone-exposed WT mice by the ΔΔCt method (38). Primer sequences for Saa3, Timp1, Il17a, and 18S were previously described (29). Primers for Ccl20 were as follows and were designed to detect both variants of this transcript: forward 5′-aagacagatggccgatgaag-3′ and reverse 5′-aggttcacagcccttttcac-3′ (product size 114 bp). The primers used for Il23 were forward 5′-cccatggagcaacttcacac-3′ and reverse 5′-gctgccactgctgactagaac-3′ (115 bp). In cases where we examined gene expression after 24 or 48 h of ozone exposure, RNA was from lungs of ozone-exposed mice as described previously (28).

Flow cytometry.

Experiments to identify the source of IL-6 were performed on lungs of WT and Adipo^−/− mice exposed to ozone or air. Briefly, after flushing of the pulmonary circulation with ice-cold PBS to remove blood cells, the lungs were excised and digested with collagenase D in complete RPMI 1610 media at 37°C for 1 h in the presence of monensin (Golgi stop, BD). Red blood cells were lysed and cells were suspended in 10 ml of media for total cell counting and then fixed and permeabilized with Perm/Fix solution (BD). Cellular surface Fc receptors were blocked with anti-CD16/32 and cells were stained with the following antibodies for intracellular IL-6 identification: CD45-PE/Cy7 (clone 30-F11, Biolegend), IL-6-APC (clone MP5–20F, Biolegend), F4/80-Alexa Fluor 488 (clone BM8, Biolegend), CD11c-PE (clone N418, Biolegend), and Gr-1-APC/Cy7 (clone RB6–8C5, Biolegend). A flow cytometer (BD FACSCanto II, BD) was used to detect fluorescence labeling of cells, and resulting data were analyzed with FlowJo software (Tree Star, Ashland, OR).

Flow cytometry was also used to compare the number of IL-17A⁺ lung cells in ozone-exposed Adipo^−/− and Adipo^−/−/IL-6^−/− mice, except that cells were stimulated with PMA and ionomycin in the presence of monensin for 5 h to increase intracellular IL-17A in γδ T cells. For intracellular detection of IL-17A we used the following antibodies: CD45-Texas Red, F4/80-FITC, CD11c-Alexa Fluor 780, IL-17A-Alexa Fluor 700, CD3-PE/Cy5, and TCRδ-PE, as previously reported (28).

Microarray.

Sixteen RNA samples were used in this analysis. RNA samples were from lungs from four C57BL/6J (WT) air-exposed mice, four Adipo^−/− air-exposed mice, four C57BL/6J ozone-exposed mice, and four Adipo^−/− ozone-exposed mice. Ozone exposure was at 0.3 ppm for 72 h. In each group, two mice were male and two were female. Gene expression analyses on these 16 RNA samples were performed using the GeneChip Mouse Genome 430A 2.0 Array platform (Affymetrix, Santa Clara, CA). Sample labeling, hybridization, and array scanning were performed by the Harvard Medical School-Partners Healthcare Center for Genetics and Genomics according to standard protocols (www.hpcgg.org/microarrays/resources). For microarray analysis, expression values were extracted from .cel files by using Robust Microarray Analysis (25). Log (base 2) expression values were imported into the program dCHIP for statistical analysis (35, 36). We used a paired analysis to control for sex. We used the DAVID Bioinformatics Resources (12) (http://david.abcc.ncifcrf.gov) to classify the ontology of genes that were significantly different between WT and Adipo^−/− mice, or between air and ozone-exposed mice, and to perform functional annotation clustering. The latter analysis was performed by using high classification stringency with all probe sets on the 430A 2.0 chip as background. The microarray data are available at Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE50183).

Statistical analysis.

Multiple comparisons were performed by factorial ANOVA combined with LSD Fisher as post hoc analysis. P < 0.05 was considered significant. All values are expressed as means ± SE. BAL cells were log transformed before statistical analysis to conform to a normal distribution. Appropriate standard error propagation was calculated when plotting graphics from log-transformed data. The statistics were analyzed by using STATISTICA (Statsoft, Tulsa, OK).

RESULTS

Effect of IL-6 deficiency in adiponectin-deficient mice.

Exposure to ozone resulted in significantly greater increases in BAL neutrophils in Adipo^−/− vs. WT mice (Fig. 1A), as previously described (28, 29), whereas ozone-induced changes in BAL macrophages (Fig. 1B) and BAL protein (Fig. 1C), an index of lung injury (20), were not affected by adiponectin deficiency. Compared with air, ozone caused significant increases in BAL IL-6 in both WT and Adipo^−/− mice, but BAL IL-6 was significantly greater in Adipo^−/− vs. WT mice (Fig. 1D).

Fig. 1.

Effect of IL-6 deficiency in adiponectin-deficient (Adipo^−/−) mice. Bronchoalveolar lavage (BAL) was performed immediately after a 72-h exposure to room air or ozone (0.3 ppm). BAL neutrophils (A), macrophages (B), protein (C), IL-6 (D), and granulocyte colony-stimulating factor (G-CSF; G) were assessed. Lungs were harvested and used to isolate RNA for measurement of Il17a (E) and Il23 (F) mRNA abundance by RT-PCR. Means ± SE, n = 4–8 per group. PCR data were normalized for expression of 18s and expressed relative to wild-type (WT) ozone-exposed mice. *P < 0.05 vs. air-exposed mice of the same genotype; #P < 0.05 vs. WT with the same exposure; and +P < 0.05 vs. Adipo^−/− mice with the same exposure. For IL-6 assay N = 13 for WT ozone and 12 for Adipo^−/− ozone.

To determine the impact of the augmented IL-6 production in Adipo^−/− mice, we crossbred Adipo^−/− mice with IL-6^−/− mice and exposed these Adipo^−/−/IL-6^−/− mice to ozone. Compared with ozone-exposed Adipo^−/− mice, BAL neutrophils were reduced to levels not significantly different from WT in Adipo^−/−/IL-6^−/− mice (Fig. 1A). BAL macrophages (Fig. 1B) and protein (Fig. 1C) were increased by ozone but were not different among WT, Adipo^−/−, and Adipo^−/−/IL-6^−/− mice. Ozone-induced increases in Il17a mRNA abundance (Fig. 1E) and BAL G-CSF (Fig. 1G) were significantly greater in Adipo^−/− vs. WT mice, but lower in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice. IL-17A can also be induced by IL-23 (39), and ozone did induce a significant increase in Il23 mRNA abundance, but there was no significant difference in Il23 mRNA among WT, Adipo^−/−, and Adipo^−/−/IL-6^−/− mice (Fig. 1F). Interestingly, we also observed a significant increase in neutrophils and Il17a mRNA in air-exposed Adipo^−/− and Adipo^−/−/IL-6^−/− vs. WT mice (Fig. 1, A and E).

Similar to the Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice (Fig. 1A), and consistent with previous observations (27, 50), ozone exposed IL-6^−/− mice had ∼50% less BAL neutrophils than WT mice (Fig. 2A). However, we did not observe any differences between ozone-exposed IL-6^−/− and WT mice in BAL macrophages, BAL protein, BAL adiponectin, BAL G-CSF, Il17a, or Il23 mRNA expression (Fig. 2, B–G).

Fig. 2.

Effect of IL-6 deficiency in adiponectin-sufficient mice. WT and IL-6^−/− mice were exposed to ozone for 72 h. Shown are BAL neutrophils (A), macrophages (B), protein (C), adiponectin (D), and G-CSF (E). We also measured Il17a mRNA (F) and Il23 (G) by RT-PCR. Results are means ± SE (n = 5 for air-exposed and n = 6–7 for ozone-exposed mice). *P < 0.05.

We compared IL-17A producing IL-17A⁺ CD11c⁻ F4/80⁺ cells (interstitial macrophages/monocytes)(Fig. 3A) and IL-17A⁺TCRδ⁺ cells (γδ T-cells) (Fig. 3B) in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice by flow cytometry. We have previously reported that these cells are the source of the IL-17A⁺ induced after subacute ozone (28). Both cell types were significantly lower in ozone-exposed Adipo^−/−/IL-6^−/− compared with Adipo^−/− mice.

Fig. 3.

IL-6 deficiency decreases IL-17A expression in macrophages and γδ T cells of ozone-exposed Adipo^−/− mice. Cells from lung digests of Adipo^−/− and Adipo^−/−/IL-6^−/− mice exposed to ozone (0.3 ppm) for 72 h were stimulated with PMA and ionomycin for 5 h in presence of monensin. After fixation and permeabilization, flow cytometry was used to quantitate total IL-17A⁺ F4/80⁺ cells (macrophages; A) and IL-17A⁺ CD3⁺ TCDδ⁺ cells (γδ T cells; B). Data are means ± SE of n = 8–9 mice per group. *P < 0.05.

We also used a flow cytometric method to examine the source of IL-6 induced by subacute ozone exposure in both WT (Fig. 4) and Adipo^−/− (Fig. 5) mice. In WT mice, there was little or no IL-6 in CD45⁻ cells (less than 1% of these cells were IL-6⁺) (Fig. 4B), whereas CD45⁺ cells did express IL-6 (Fig. 4C). A majority of these CD45⁺ IL-6-producing cells were positive for Gr-1 (Ly-6_C/Ly-6_G), a cell surface marker expressed on neutrophils (15), myeloid-derived suppressor cells (MDSC) (66), and monocytes (52), but these IL-6⁺ cells were also F4/80⁻ and CD11c⁻ (Fig. 4A). Compared with air, ozone exposure significantly increased the percentage of these IL-6⁺Gr-1⁺F4/80⁻CD45⁺ cells (Fig. 4D).

Fig. 4.

WT mice were exposed to ozone or air for 72 h and IL-6-producing cells were identified by flow cytometry. A: gating strategy used to identify IL-6 expressing cells. Cells from lung digests of WT mice were gated by use of side scatter (SSC-A) vs. forward scatter (FSC-A), followed by SSC-A vs. CD45. After accounting for nonspecific staining with control antibodies, a single population of IL-6-expressing cells was identified (histogram). These cells were Gr-1⁺, but F4/80⁻ and CD11c⁻. B: virtually no CD45⁻ cells (including epithelial and endothelial cells) expressed IL-6. C: ozone significantly increased the number of IL-6⁺ CD45+ cells. D: percentage of total cells that were IL-6⁺Gr-1⁺CD45⁺. Means ± SE of n = 3 mice per group. *P < 0.05.

Fig. 5.

A: identification of the cellular source of IL-6 in ozone-exposed Adipo^−/− mice by flow cytometry. Cells from lung digests of Adipo^−/− mice were gated by forward scatter vs. CD45⁺ (not shown), followed by gating SSC-A vs. IL-6⁺. From the latter, 2 populations of IL-6-producing cells were identified (IL-6^hi and IL-6^low). The IL-6^low cells had Gr-1 expression (blue histogram) similar to the IL-6-producing cells observed in WT mice and like those cells were F4/80⁻ and CD11c⁻. The IL-6^hi cells had somewhat lower Gr-1 expression (red histogram) and were F4/80⁺ and CD11c⁺. B: from both IL-6⁺ cell populations, we determined mean fluorescence intensity (MFI) (*P < 0.05 vs. F4/80⁻CD11c⁻ cells). C: total number of IL-6^hi CD11c⁺ F4/80⁺ cells from right lung homogenates of Adipo^−/− mice exposed to air or O₃. Means ± SE, n = 3 per group.

In contrast to the WT mice, two populations of IL-6⁺ cells were observed in Adipo^−/− mice (Fig. 5A). One population of IL-6⁺ cells (blue arrows and histogram in Fig. 5A) were Gr-1⁺, and F4/80⁻ and CD11c^−, similar to the IL-6⁺ cells observed in WT mice (Fig. 4). The other population (red arrows and histogram in Fig. 5A) had higher IL-6 expression and were Gr-1^intermediate. These cells were F4/80⁺ and CD11c⁺ consistent with alveolar macrophages (Fig. 5A). The intracellular expression of IL-6, measured by mean fluorescence intensity, was significantly higher in F4/80⁺CD11c⁺ (IL-6^hi) than in F4/80⁻CD11c⁻ (IL-6^int) cells (Fig. 5B). Compared with air, ozone exposure increased the number of IL-6^hiCD11c⁺F4/80⁺ cells (Fig. 5C). Similarly to WT mice, negligible numbers of IL-6⁺ CD45⁻ cells were found in Adipo^−/− mice (data not shown). Taken together, our data indicate that adiponectin normally serves to suppress IL-6 expression in alveolar macrophages.

Microarray analysis.

To obtain a comprehensive picture of the gene expression changes induced by subacute ozone exposure, and the impact of adiponectin deficiency on the response to ozone, we performed a microarray analysis on lung mRNA. Comparison of expression values of arrays from WT mice exposed to air vs. ozone (0.3 ppm) for 72 h indicated 340 Affymetrix (Affy) IDs that were significantly altered by ozone (P < 0.05) with at least a twofold change. Of these, 166 were increased by ozone and 174 were reduced. Functional annotation clustering indicated that of the genes either induced (Table 1) or reduced (Table 2) by ozone were mainly involved with the cell cycle and cell replication, the extracellular matrix, transcription, and apoptosis, consistent with previous reports (6) and with the epithelial injury and subsequent proliferation and wound repair that characterizes this type of ozone exposure (32).

Table 1.

Top 15 cluster categories (by enrichment score) of genes significantly increased by O₃ in WT mice

Cluster Category	Enrichment Score	Gene Count	P Value
Cell division	19.57	28	5.6E-22
Condensed chromosome kinetochore	4.2	6	4.0E-5
Platelet-derived growth factor binding	3.1	4	3.5E-5
ATP binding	3.03	23	9.1E-5
Extracellular matrix	2.88	9	1.4E-4
DNA-dependent ATPase MCM	2.49	3	1.3E-3
DNA-dependent DNA replication	2.4	4	2.9E-4
Negative regulation of apoptosis	2.15	9	9.6E-4
Cyclin	2.11	3	2.0E-3
Establishment of chromosome localization	1.97	3	1.9E-3
Keratinocyte differentiation	1.94	4	7.7E-3
Kinesin, motor region, conserved site	1.75	4	2.9E-3
von Willebrand factor, type C	1.62	3	3.3E-2
Chromatin assembly	1.29	10	2.7E-2
Cyclin-dependent protein kinase regulator activity	1.45	3	5.6E-3

Functional annotation clustering was performed by using the program DAVID on high stringency.

WT, wild-type.

Table 2.

Cluster categories (by enrichment score) of genes significantly decreased by O₃ in WT mice

Cluster Category	Enrichment Score	Gene Count	P Value
mRNA processing	2.69	9	3.6E-4
Transcription regulation	2.57	24	6.6E-4
Zinc finger, RanRB2-type	1.93	3	5.8E-3
Cell migration	1.8	7	6.7E-3
RNA recognition motif, RNP-1	1.64	6	1.7E-2
Transcription regulator activity	1.23	3	1.3E-3
DNA-dependent DNA replication	1.23	16	4.5E-2
Apoptosis	1.21	8	4.4E-2
Blood vessel morphogenesis	1.12	5	5.0E-2
Nuclear lumen	1.02	12	3.1E-2
Regulation of apoptosis	1.02	8	9.2E-2
Positive regulation of apoptosis	1.01	5	9.6E-2

Functional annotation clustering was performed by using the program DAVID on high stringency.

In Adipo^−/− mice, 304 Affy IDs were significantly altered by ozone (P < 0.05) with at least a twofold change. Of these, 217 were increased by ozone and 87 were reduced. There was substantial overlap among genes affected by ozone exposure in Adipo^−/− and WT mice. The list of genes that were significantly affected by ozone in either WT or Adipo^−/− mice contained 432 Affy IDs. Figure 6 shows the average fold change value (ozone/air) for each of these Affy IDs in the WT and Adipo^−/− mice. The solid diagonal line is the line of identity. For genes induced by ozone (ozone/air >1), values to the right of this line indicate a greater effect of ozone in the WT than the Adipo^−/− mice. For genes reduced by ozone (ozone/air < 1), values to the left of the line of identify indicate a greater effect of ozone in the WT than the Adipo^−/− mice. For most Affy IDs, the effect of ozone was similar in WT and Adipo^−/− mice: most points fell near the line of identity. DChip identified 55 Affy IDs that were affected by ozone and were significantly different (P < 0.05) with at least a 1.25-fold change between the ozone-exposed WT and Adipo^−/− mice. Of these, 40 were induced by ozone and 15 were reduced by ozone. A heatmap showing relative expression of these 55 Affy IDs in each of the individual mice is shown in Fig. 7. Notably, of the genes induced by ozone and different in ozone exposed WT vs. Adipo^−/− mice, almost all (38 of 40) were affected to a greater extent in Adipo^−/− than WT mice. Of the genes that were reduced by ozone and different in ozone exposed WT vs. Adipo^−/− mice, almost all (14 of 15) were affected to a greater extent in WT than Adipo^−/− mice. Tables 3 and 4 show the top 10 of these, as ranked by the magnitude of the difference in the effect of ozone (ratio of ozone to air) between Adipo^−/− and WT mice for genes that were either induced (Table 3) or reduced (Table 4) by ozone.

Fig. 6.

Identification of genes induced by ozone and augmented by adiponectin deficiency. Mean gene expression changes in obese vs. lean mice for 432 genes whose expression was significantly affected by subacute ozone exposure in either WT or Adipo^−/− mice. Solid line is the line identity. For genes induced by ozone (ozone/air >1), values to the right of this line indicate a greater effect of ozone in the WT than the Adipo^−/− mice. For genes reduced by ozone (ozone/air < 1), values to the left of the line of identify indicate a greater effect of ozone in the WT than the Adipo^−/− mice. Data point with arrow is Saa3.

Fig. 7.

Heatmap showing relative expression values of genes that were significantly affected by ozone (P < 0.05 with at least a 2-fold change) in either WT or Adipo^−/− mice and in which the expression values from the ozone-exposed WT and Adipo^−/− mice were significantly different (P < 0.05 with at least a 25% change).

Table 3.

Top 10 genes induced (ozone/air >1) by subacute ozone exposure and with significantly different expression in ozone exposed WT and Adipo^−/− mice

Affymetrix ID	Genes Induced by Ozone	Gene Symbol	Log2 Ozone-Air WT	Log2 Ozone-Air Adipo−/−
1450826_a_at	serum amyloid A 3	Saa3	1.56	3.99
1421074_at, 1421075_s_at	cytochrome P-450, family 7, subfamily b, polypeptide 1	Cyp7b1	0.58, 0.72	1.57, 1.2
1460227_at	tissue inhibitor of metalloproteinase 1	Timp1	3.0	3.79
1452242_at	centrosomal protein 55	Cep55	2.15	2.79
1429295_s_at	thyroid hormone receptor interactor 13	Trip13	0.87	1.49
1437716_x_at	kinesin family member 22	Kif22	1.34	1.89
1419513_a_at	ect2 oncogene	Ect2	1.97	2.47
1433893_s_at	sperm associated antigen 5	Spag5	0.65	1.15
1449227_at	cholesterol 25-hydroxylase	Ch25 h	0.63	1.12
1422462_at	ubiquitin-conjugating enzyme E2T	Ube2t	0.7	1.19

Adipo^−/−, adiponectin deficient.

Table 4.

Top 10 genes reduced (ozone/air <1) by subacute ozone exposure and with significantly different expression in ozone exposed WT and Adipo^−/− mice

Affymetrix ID	Genes Reduced by Ozone	Gene Symbol	Log2 Air-Ozone WT	Log2 Air-Ozone Adipo−/−
1449434_at	carbonic anhydrase 3	Car3	0.73	2.61
1451644_a_at 1431008_at	histocompatibility 2, Q region locus 1/6/7/8/9*	H2-Q1,6,7,8,9	1.5, 1.1	0.1, 0.2
1436898_at	splicing factor proline/glutamine rich (polypyrimidine tract binding protein associated)	sfpq	1.85	0.99
1417898_a_at	granzyme A	gzma	1.1	0.24
1425329_a_at	cytochrome b5 reductase 3	Cyb5r3	2.77	1.96
1427202_at	RIKEN cDNA 4833442J19 gene	4833442J19Rik	1.1	0.57
1450459_at	RIKEN cDNA 2010106G01 gene	2010106G01Rik	1.27	0.78
1423155_at	sorcin	sri	1.17	0.7
1422411_s_at	eosinophil-associated, ribonuclease A family, member 1/2/3	Ear1,ear2,ear3	1.31	0.87
1452125_at	thyroid hormone receptor associated protein 3	Thrap3	1.15	0.74

Change in gene expression values are organized by the magnitude of the difference in the ozone induced change in Adipo^−/− versus WT mice.

^*This Affy ID is included in each of these histocompatibility 2, Q region locus genes.

Among the genes induced by ozone, Saa3 (serum amyloid A3), an acute-phase protein, was the gene for which adiponectin deficiency caused the greatest difference in induction by ozone (Table 2 and Fig. 6, arrow). RT-PCR also confirmed a robust time-dependent induction of Saa3 by ozone that was greater in Adipo^−/− vs. WT mice (Fig. 8, A and B). Elevated Saa3 mRNA in Adipo^−/− vs. WT mice was observed not just after 72-h exposure (Fig. 8B), but also after a 24 or 48-h ozone exposure (Fig. 8A) and was also observed in air-exposed mice (Fig. 8, A and B). Because IL-6 can induce Saa3 expression (16), we examined the role of IL-6 in ozone-induced changes in Saa3 mRNA expression in both WT and Adipo^−/− mice. Saa3 expression was significantly lower in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice (Fig. 8B), but there was no difference in Saa3 expression in IL-6^−/− vs. WT mice (data not shown). To confirm our RT-PCR data, we performed an ELISA assay for BAL SAA3 (Fig. 8C). Compared with air, ozone caused a significant increase in BAL SAA3 in Adipo^−/− mice. After ozone exposure, BAL SAA3 was significantly higher in Adipo^−/− vs. WT mice, and significantly lower in Adipo^−/−/IL-6^−/− than Adipo^−/− mice.

Fig. 8.

Adiponectin deficiency increases expression of Saa3 and Ccl20 mRNA. Expression of Saa3 (A) and Ccl20 (D) mRNA in lungs of Adipo^−/− and WT mice exposed to air or ozone (0.3 ppm) for 24 and 48 h. Expression of Saa3 mRNA (B) and Ccl20 (E) mRNA measured by RT-PCR and SAA3 (C) and CCL20 (F) measured by ELISA in WT, Adipo^−/−, and Adipo^−/−/IL-6^−/− mice exposed to air or ozone (0.3 ppm) for 72 h. G: Timp1 mRNA expression from lungs of mice exposed to ozone (0.3 ppm) or air for 72 h. Data are means ± SE; *P < 0.05 vs. air-exposed mice with same genotype; #P < 0.05 vs. WT mice with same exposure; and +P < 0.05 vs. Adipo^−/− mice with same exposure. In the case of 24- and 48-h data, results are normalized to the WT 48-h ozone mice. In the case of 72-h data, results are normalized to the WT 72-h ozone mice; n = 4–8 per group for time course in A and D and n = 3 for air and 6–7 for ozone in the other panels.

Instillation of serum amyloid A into the lungs results in a neutrophilic inflammation that is dependent on recruitment of IL-17A⁺ γδ T cells (1), similar to the effects of ozone (28, 43). Furthermore, SAA3 can induce expression of Ccl20 (MIP-3α) (45), a chemoattractant for IL-17A⁺ γδ T cells (37, 40). Our data indicate that ozone also induced Ccl20 mRNA expression (Fig. 8, D and E). In WT mice, Ccl20 mRNA abundance was greater in mice exposed to 0.3 ppm ozone for 24, 48, or 72 h than in mice exposed to room air (Fig. 8, D and E). Even in air-exposed mice, pulmonary Ccl20 mRNA expression was greater in Adipo^−/− vs. WT mice (Fig. 8, D and E). Indeed, the effect of adiponectin deficiency alone was at least as great as the effect of ozone. Ccl20 mRNA abundance was also higher in Adipo^−/− vs. WT mice exposed to ozone for 24 or 48 (Fig. 8D), but not 72 h (Fig. 8E). There was no difference in Ccl20 mRNA in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice (Fig. 8D) or in IL-6^−/− vs. WT mice deficiency (data not shown). Qualitatively similar results were obtained for BAL CCL20 (Fig. 8F).

Microarray analysis also indicated greater induction of Timp1, a matrix metalloproteinase inhibitor, in Adipo^−/− vs. WT mice (Table 2). RT-PCR confirmed these observations (Fig. 8G) and also indicated that IL-6 deficiency reduced Timp1 expression in Adipo^−/− mice, consistent with the ability of IL-6 to induce Timp1 expression (46).

We also examined differences in gene expression in lungs of Adipo^−/− vs. WT mice exposed only to air. We used less stringent criteria for this analysis to obtain a sufficient number of Affy IDs for functional annotation clustering. There were 180 Affy IDs with significantly different expression (P < 0.05, fold change >1.2) in Adipo^−/− vs. WT air-exposed mice. Of these, 122 were significantly reduced in Adipo^−/− vs. WT mice and the remainder was significantly elevated in the Adipo^−/− vs. WT mice. Functional annotation clustering indicated that in air-exposed mice, the genes affected by adiponectin deficiency were those involved with the extracellular matrix, cell migration, cell adhesion, and respiratory system development (Table 5). These changes in gene expression may be the mechanistic underpinnings of the emphysematous changes that characterize the lungs of Adipo^−/− mice as they age (54). A heatmap showing relative expression in each of the individual air-exposed mice of those Affy IDs for which there was at least a 25% change is shown in Fig. 9.

Table 5.

Top 15 cluster categories (by enrichment score) of genes significantly different in Adipo^−/− versus WT mice exposed to air

Cluster Category	Enrichment Score	Gene Count	P Value
Extracellular matrix	3.25	10	1.4E-4
Cell migration	2.96	10	3.9E-4
Polysaccharide binding	1.98	6	7.2E-3
Extracellular matrix structural component	1.97	5	1.7E-4
Positive regulation of cell-substrate adhesion	1.95	4	1.9E-3
Regulation of body fluid levels	1.61	5	9.4E-3
Monovalent inorganic cation homeostasis	1.6	5	4.1E-4
Hemopoiesis	1.37	7	2.9E-2
Neural crest cell differentiation	1.25	3	3.7E-2
Respiratory system development	1.23	5	2.8E-2

Functional annotation clustering was performed by using the program DAVID on high stringency.

Fig. 9.

Heatmap showing relative expression values of genes that were significantly different in WT vs. Adipo^−/− mice exposed to air (P < 0.05 with at least a 25% change). m, Male; f, female.

DISCUSSION

Our data indicate that IL-6 plays a pivotal role in the augmented ozone-induced inflammation observed in Adipo^−/− vs. WT mice. IL-6 deficiency either partially or completely reversed elevations in ozone-induced recruitment of neutrophils (Fig. 1A), Il17a mRNA expression (Fig. 1E), and BAL G-CSF (Fig. 1G) observed in Adipo^−/− mice. In addition, IL-17A⁺ CD11c⁻ interstitial macrophages/monocytes and IL-17A⁺ γδ T-cells were lower in Adipo^−/−/IL-6^−/− compared with Adipo^−/− mice (Fig. 3), whereas total γδ T cells were unchanged. Our microarray study identified the acute-phase protein, Saa3, which can induce increase IL-17A in the lung, among the genes most affected by adiponectin deficiency in ozone-exposed mice (Table 3, Fig. 6), and IL-6 deficiency reduced the augmented Saa3 mRNA abundance observed in Adipo^−/− mice (Fig. 8B).

In WT mice, BAL IL-6 increased after ozone exposure (Fig. 1D), consistent with data from ourselves and others using a variety of different ozone exposure protocols (10, 14, 27, 30, 33, 50, 57, 64). We now identify Gr-1^hiF4/80⁻CD11c⁻cells as the source of this IL-6 in WT mice (Fig. 4). These cells express surface markers consistent with neutrophils. Murine neutrophils have also been shown to express IL-6 under other conditions (42). However, monocytes also express Gr-1 on their surface (19). A similar but not identical population of Gr-1⁺ macrophages was shown to be induced by acute ozone and have higher expression of CX3CR1 and MARCO compared with alveolar macrophages(55). In recent years, a new granulocytic population have been described as CD11b⁺Gr-1⁺F4/80⁻Ly-6c^intLy-6G^hi with anti-inflammatory role in inflammation induced by tumor, also known as granulocytic myeloid-derived suppressor cells (g-MDSC or MDSC) (8, 48, 66). Adipo^−/− mice bearing EL4 tumors have a significantly decreased recruitment of granulocytic MDSC (21). Importantly, BAL IL-6 was higher in Adipo^−/− vs. WT mice (Fig. 1D), consistent with previous observations (28, 29). In Adipo^−/− mice, IL-6 expression was observed not only in Gr-1^hiF4/80⁻CD11c⁻ cells but also in CD11c⁺ F4/80⁺ alveolar macrophages. Indeed, the mean fluorescence intensity for IL-6 in these CD11c⁺ macrophages was higher than in Gr-1⁺F4/80⁻CD11c⁻ cells (Fig. 5B). The ability of alveolar macrophages to express IL-6 after ozone exposure has also been observed in macrophages exposed to ozone in cell culture (3), perhaps because murine adiponectin was not present in the culture media. Cultured airway epithelial cells also have the capacity to produce IL-6 in response to ozone (14), but our data indicate that CD45⁻ cells, which include epithelial cells, were not major sources of IL-6, at least under these exposure conditions (Fig. 4B). Taken together, the results suggest that adiponectin is exerting its anti-inflammatory role in this model via effects on alveolar macrophages. Indeed, macrophages express adiponectin receptors, and in vitro studies indicate that adiponectin reduces LPS-induced IL-6 expression in porcine macrophages (61).

Given the importance of IL-6 for neutrophil recruitment following subacute ozone exposure in WT mice (27, 50) and the elevated levels of IL-6 in Adipo^−/− mice (Fig. 1D), we sought to determine whether the augmented ozone-induced neutrophil recruitment observed in Adipo^−/− mice was the result of their enhanced capacity to produce IL-6. Indeed, IL-6 deficiency caused a marked reduction in BAL neutrophils in adiponectin-deficient mice. In fact, BAL neutrophils were not significantly different in ozone exposed WT vs. Adipo^−/−/IL-6^−/− mice (Fig. 1A). Because we have previously reported that the increased neutrophil recruitment observed in Adipo^−/− vs. WT mice was partially dependent on IL-17A, we also examined the role of IL-6 in the augmented IL-17A expression observed in Adipo^−/− vs. WT mice. Il17a mRNA abundance was markedly increased in Adipo^−/− vs. WT mice, but there was no significant effect of ozone exposure on Il17a mRNA abundance in Adipo^−/−/IL-6^−/− mice (Fig. 1E). Consistent with these observations, there were fewer IL-17A⁺ macrophages and fewer IL-17A⁺ γδ T cells in lungs of Adipo^−/−/IL-6^−/− compared with Adipo^−/− mice (Fig. 2). We have previously reported that elevated ozone-induced expression of Il17a in lungs of Adipo^−/− vs. WT mice is the result of IL-17A expression in these two cell types, not in Th17 cells (28). Notably, the F4/80⁺ cells that express IL-17A in response to ozone are CD11c⁻ (28) whereas the F4/80⁺ cells that express IL-6 in Adipo^−/− mice are CD11c⁺ (Fig. 3). F4/80⁺CD11c⁻ cells in the lung have been described as interstitial/inflammatory macrophages whereas F4/80⁺CD11c⁺ cells are alveolar macrophages (7). However, inflammatory monocytes have been described in the past to have similar surface markers interstitial/inflammatory macrophages (19). Given that IL-17A is required for the induction of G-CSF in ozone-exposed Adipo^−/− mice (28), the observation of reduced G-CSF in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice (Fig. 1G) is consistent with the observation that IL-17A is also reduced in these mice (Fig. 1E). Taken together, the data indicate that augmented ozone-induced neutrophil recruitment in Adipo^−/− mice is related to the ability of IL-6 to promote IL-17A expression and consequent G-CSF release.

We observed increased numbers of neutrophils, and increased Il17a, Saa3, and Ccl20 mRNA expression even in air-exposed Adipo^−/− and Adipo^−/−/IL-6^−/− mice (Figs. 1 and 8). However, as opposed to the case of ozone exposure, we observed no difference in BAL neutrophils, or in Il17a, Saa3, and Ccl20 mRNA expression in air-exposed Adipo^−/− vs. Adipo^−/−/IL-6^−/− mice (Figs. 1 and 8), indicating that factors other than IL-6 contribute to this baseline inflammation. Others have reported that, even in the absence of ozone exposure, alveolar macrophages from Adipo^−/− mice have increased expression of TNF-α (54). Taken together, the data suggest that adiponectin plays a role in suppressing inflammation even in the absence of ozone, perhaps by suppressing activation of alveolar macrophages.

We do not know the precise role of IL-6 in the augmented IL-17A expression observed in Adipo^−/− vs. WT mice exposed to ozone. IL-6 is required for biasing CD4⁺ T cells toward a Th17 phenotype (62), but the role of IL-6 in IL-17A expression in γδ T cells, especially murine γδ T cells, is less clear. In γδ T cells that express RORγt, a transcription factor needed to drive IL-17A production, IL-6 deficiency has no effect on the ability of γδ T cells to produce IL-17A (39). However, in naive animals, not all γδ T cells express RORγt (26), and IL-6 may be required to drive RORγt and thus IL-17A expression. The cellular locus of IL-6 expression may also be important. IL-6 was required for Il17a mRNA expression in Adipo^−/− but not WT mice (Figs. 1E and 2F) and macrophages expressed IL-6 in Adipo^−/− but not WT mice (Figs. 4 and 5). Macrophages are found in close apposition to γδ T cells in the lungs (56). Perhaps this proximity permits IL-6 to induce IL-17A expression in γδ T cells. It is also possible that IL-6 deficiency has an indirect effect on IL-17A⁺ γδ T cells. For example, we have shown recruitment/proliferation of γδ T cells to the lungs after subacute ozone (28, 43), and IL-6 may contribute to the generation of chemotactic factors that promote this recruitment (see below). Alternatively, the role of IL-6 in IL-17A expression may be mediated via effects on Saa3 expression.

Microarray analysis identified Saa3, an acute-phase protein, among the genes most differentially expressed in lungs of Adipo^−/− vs. WT mice exposed to ozone (Table 1, Fig. 4). Pulmonary Saa3 expression is also increased in response to other stimuli that induce oxidative stress, including tobacco smoke, bleomycin, and allergen (4, 9, 13). SAA3 is produced in adipocytes, airway epithelium, and macrophages (16, 44, 59), and its expression is induced by IL-6 (16), particularly in combination with IL-1 (24), which is also induced by subacute ozone exposure (27, 50). The ability of IL-6 to promote SAA3 expression (16) is consistent with the reduction in ozone-induced Saa3 mRNA and protein observed in Adipo^−/−/IL-6^−/− vs. Adipo^−/− mice (Fig. 8, B and C). It is possible that IL-6 drives augmented IL-17A expression in ozone-exposed Adipo^−/− vs. WT mice (Figs. 1 and 3) via its ability to promote Saa3 expression (Fig. 8): chronic intratracheal instillation of serum amyloid A causes neutrophil influx into the lungs that is associated with recruitment/proliferation of IL-17A⁺ γδ T cells to the lung and is abolished by anti-IL-17 antibodies (1). Serum amyloid A also induces IL-17A expression in CD4⁺ T cells (4). The mechanistic basis for effects of serum amyloid A on IL-17A expression in the lungs has not been established, but TLR4 may be involved. SAA3 has the ability to cause TLR4 activation (11) and γδ T-cells from Tlr4^−/− mice have impaired expression and secretion of IL-17A (49).

We also considered the possibility that the role of SAA3 may involve its ability to induce chemotactic factors that promote recruitment of IL-17A⁺ γδ T cells to the lungs. For example, SAA3 induces expression of CCL20 (MIP-3α) (45). CCL20 is a ligand for CCR6, which is expressed on IL-17A⁺ γδ T cells (39). Ligation of CCR6 causes chemotaxis of IL-17A⁺ γδ T cells (37, 40). Indeed, similar to Saa3, Ccl20 mRNA abundance was increased by subacute ozone exposure in WT mice (Fig. 5, C and D). Increased BAL CCL20 is also observed after acute ozone exposure (2 ppm for 3 h) (58). Interestingly, Ccl20 mRNA expression was higher in Adipo^−/− vs. WT mice even after air exposure (Fig. 8, D and E). Consistent with these observations, we have previously reported higher numbers of IL-17A⁺ γδ T cells in the lungs of Adipo^−/− vs. WT mice even after air exposure (28). Ccl20 mRNA expression was higher in Adipo^−/− vs. WT mice after shorter duration ozone exposure (24 and 48 h) (Fig. 8D). However, whereas expression of Saa/SAA3 was also higher in lungs of ozone-exposed Adipo^−/− vs. WT mice after 72 h of ozone exposure and was markedly reduced by IL-6 deficiency (Fig. 8, B and C), that was not the case for Ccl20/CCL20 (Fig. 8, E and F). Thus it may be that CCL20 expression contributes to the recruitment of γδ T cells, but that other factors contribute to the ability of IL-6 and SAA3 to produce IL-17A leading to neutrophil recruitment.

Neutrophil recruitment induced by subacute ozone exposure was also reduced in IL-6^−/− vs. WT mice (Fig. 2A), consistent with previous observations (27, 50). However, BAL G-CSF and Il17a mRNA expression were not different in WT vs. IL-6^−/− mice exposed to ozone (Fig. 2, E and F), indicating that other factors contribute to the ability of IL-6 to recruit neutrophils in WT mice. Nevertheless, IL-17A is induced by ozone even in WT mice (Fig. 1E and Refs. 23 and 24). IL-23 also induces IL-17A in lymphocytes (39), and Il23 mRNA was increased in ozone vs. air exposed WT mice (Fig. 1F), suggesting that IL-23 may be involved in the induction of IL-17A in WT mice. Interestingly, Il23 expression was not affected by adiponectin deficiency (Fig. 1F).

One technical issue requires further consideration and discussion. The studies reported here used a 72-h ozone exposure period to examine the impact of IL-6 on neutrophil recruitment. Differences in the kinetics of O₃-induced pulmonary inflammation have been reported depending on the precise outcome indicator examined (5, 6). Hence, it is conceivable that the IL-6-dependent SAA3/IL-17A/G-CSF-mediated neutrophil recruitment observed in Adipo^−/− mice exposed to ozone, as outlined in Fig. 10, is unique to this particular duration of exposure. However, we consider this to be an unlikely scenario for the following reasons. We have observed increased BAL IL-6 even after only 24 h of ozone exposure (43) and we have also observed elevated neutrophil recruitment and pulmonary Il17a mRNA expression in Adipo^−/− vs. WT mice after only 24 or 48 h of ozone exposure. Furthermore, Saa3 mRNA expression was greater in Adipo^−/− vs. WT mice at all-time points examined (Fig. 8, A and B), suggesting that the outcomes reported here are not specific to a unique temporal setting.

Fig. 10.

Schematic illustration of proposed mechanism whereby ozone induces pulmonary neutrophilic inflammation and the role of adiponectin in modulating the response to ozone. Adiponectin acts mainly to inhibit IL-6 secretion by alveolar macrophages, ultimately decreasing recruitment of γδ T-cells capable of IL-17A expression.

In conclusion, our data support a pivotal role for IL-6 in orchestrating the hyperinflammatory condition observed in Adipo^−/− mice after ozone exposure (Fig. 10). Adiponectin likely acts to inhibit IL-6 release from alveolar macrophages. In the absence of adiponectin, excess IL-6 results in increased induction of SAA3, IL-17A, and G-CSF, resulting in increased neutrophil recruitment. In contrast, IL-6-dependent neutrophil recruitment in ozone-exposed WT mice likely involves other IL-6-dependent events.

GRANTS

This study was supported by the U.S. National Institutes of Health Grants HL-084044, ES-013307, and ES-000002. D. T. Umetsu and H. Y. Kim are supported by RO1 AI026322. C. Hug is supported by DP2 OD008618.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s). N. Verbout is currently an employee of Aronora Inc., but her contribution was before the employment in this company. Aronora does not have any involvement with any part of this study.

AUTHOR CONTRIBUTIONS

D.I.K., D.T.U., and S.A.S. conception and design of research; D.I.K., H.Y.K., J.A.M., N.G.V., A.P.W., F.M.C.N., F.L.N., and L.A.B. performed experiments; D.I.K., N.G.V., F.M.C.N., and S.A.S. analyzed data; D.I.K., D.T.U., and S.A.S. interpreted results of experiments; D.I.K. and S.A.S. prepared figures; D.I.K. and S.A.S. drafted manuscript; D.I.K., H.Y.K., J.A.M., N.G.V., A.S.W., A.P.W., C.H., D.T.U., and S.A.S. edited and revised manuscript; D.I.K., H.Y.K., J.A.M., N.G.V., A.S.W., A.P.W., F.M.C.N., F.L.N., L.A.B., C.H., D.T.U., and S.A.S. approved final version of manuscript.