Diverse and Potent Chemokine Production by Lung CD11bhigh Dendritic Cells in Homeostasis and in Allergic Lung Inflammation1

Lung CD11chigh dendritic cells (DC) are comprised of two major phenotypically distinct populations, the CD11bhigh DC and the integrin αEβ7+ DC (CD103+ DC). To examine whether they are functionally distinguishable, global microarray studies and real-time PCR analysis were performed. Significant differences between the two major CD11chigh DC types in chemokine mRNA expression were found. CD11bhigh DC is a major secretory cell type and highly expressed at least 16 chemokine mRNA in the homeostatic state, whereas CD103+ DC highly expressed only 6. Intracellular chemokine staining of CD11chigh lung cells including macrophages, and ELISA determination of sort-purified CD11chigh cell culture supernatants, further showed that CD11bhigh DC produced the highest levels of 9 of 14 and 5 of 7 chemokines studied, respectively. Upon LPS stimulation in vitro and in vivo, CD11bhigh DC remained the highest producer of 7 of 10 of the most highly produced chemokines. Induction of airway hyperreactivity and lung inflammation increased lung CD11bhigh DC numbers markedly, and they produced comparable or higher amounts of 11 of 12 major chemokines when compared with macrophages. Although not a major producer, CD103+ DC produced the highest amounts of the Th2-stimulating chemokines CCL17/thymus and activation-related chemokine and CCL22/monocyte-derived chemokine in both homeostasis and inflammation. Significantly, CCL22/monocyte-derived chemokine exhibited regulatory effects on CD4+ T cell proliferation. Further functional analysis showed that both DC types induced comparable Th subset development. These studies showed that lung CD11bhigh DC is one of the most important leukocyte types in chemokine production and it is readily distinguishable from CD103+ DC in this secretory function.

A llergic asthma is caused by dysregulated immune responses to airway allergens in house dust mites, molds, and animal danders, which leads to a disease characterized by airway hyperresponsiveness (AHR), 3 reversible airway obstruction, and lung inflammation (1,2). Dendritic cells (DC) are considered the major APC type in the capture, internalization, and transport of airway Ag to the lung-draining lymph node for Agspecific T cell activation (3)(4)(5)(6)(7) and airway DC depletion abrogates Ag-induced AHR and lung inflammation (8,9). Lung DC have been shown to be occuring in or tightly apposed to the bronchial epithelium and DC extensions form a lattice-like pattern that line the lumenal surface of the airway epithelium (10,11). Isolation of these epithelial airway DC by microdissection showed that they expressed distinct surface markers from nonepithelial lung DC and were more potent in Ag presentation than the latter DC type (12). These early data suggest the occurrence of at least two major lung DC types. In support of this occurrence, Flt3 ligand treatment was found to increase a lung CD11b ϩ but not a CD11b Ϫ DC population preferentially, thus suggesting that CD11b may be considered a distinguishing marker (13). More detailed lung DC subset analysis have confirmed the existence of CD11b high and the CD11b low populations (14). However, this marker was insufficient to differentiate the lung DC populations. CD103, an integrin protein subunit that associates only with integrin ␤ 7 to form the E-cadherin ligand ␣ E ␤ 7 , is a more discriminating marker for the two major lung DC subsets. The basolateral expression of E-cadherin by bronchial epithelial cells immediately explains the epithelial localization of integrin ␣ E ␤ 7 ϩ DC (CD103 ϩ DC) and this localization suggests that the CD103 ϩ DC type corresponds to the epithelial DC reported earlier and is largely responsible for the presentation of airway Ag. Their high expression of tight junction proteins that allows the ready access of DC extensions into the airway lumen further supports their Ag presentation role (14). CD103 ϩ DC have been described in intestinal lamina propria, mesenteric lymph nodes, and skin-draining lymph nodes, but not in spleen or skin (15)(16)(17)(18). In Ag-or pathogenic T cell-induced intestinal inflammation, the generation of gut-tropic T cells expressing CCR9 and integrin ␣ 4 ␤ 7 (lymphocyte Peyer's patch high endothelial venule adhesion molecule 1 (LPAM-1)) requires the participation of CD103 ϩ DC while CD103 Ϫ DC failed to induce these T cells (17, with cytokines or other stimulants and the occurrence of CD11b high and CD103 ϩ DC as distinct functional subset has not yet been established. One possibility is that these DC subsets are derived from I-A Ϫ precursor found in neonatal rats and in adult mice airways (19,20), but this development has not been formally shown in adults.
Allergic diseases are characterized by Th2-mediated inflammatory responses (1,2). In asthma, lung and peripheral blood T cells are dominated by allergen-specific Th2 cells. The basis for this biased-Th subset response has not been established but may be due to lung chemokine production that preferentially attract Th2 cells (21)(22)(23), or the Th2-directing function of lung DC (24,25). In spleen and lymph nodes, CD8 ϩ DC have been found to be Th1biased because of their ability to secrete high levels of IL-12 while CD11b ϩ DC are Th2-directing (26). However, no Th1-directing CD8 ϩ DC have been found in lungs (14). This absence may in itself be the reason for the lung Th2-biased priming. However, because Th subset response studies thus far have used bulk CD11c ϩ lung cells composed of CD11b high DC, CD103 ϩ DC, and macrophages, the DC subset functions in Th2 development have not been examined individually and need to be established.
To elucidate the functional difference between lung CD103 ϩ and CD11b high DC, we have examined lung DC subset functions with the aid of mRNA analysis by microarrays. The results in this report showed that chemokine production function in normal lungs was assumed primarily by CD11b high DC with little contribution from CD103 ϩ DC. Further studies of chemokine production by CD11c ϩ lung cell subsets showed that in homeostasis, CD11b high DC produced more chemokine species and at higher level than CD103 ϩ DC and even than macrophages. In intratracheal LPSinjected mice and in the OVA-induced AHR and lung inflammation model, CD11b high DC remained a major chemokine producer with higher or comparable secretion of most of the chemokines compared to macrophages. The results indicate that lung CD11b high DC is a major chemokine producer and dissociate their functions from CD103 ϩ DC in their role in directing lung leukocyte trafficking during lung inflammation, although both DC subsets directed a Th2-biased response.

Materials and Methods
Materials TLR ligands were from InvivoGen and have been tested to be LPS free by the manufacturer. All mAb were from eBioscience, except the following: anti-integrin ␤ 7 and SiglacF mAb were from BD Pharmingen; anti-mPDCA-1mAb was from Miltenyi Biotec; and rabbit or goat antichemokine Ab were from PeproTech or R&D Systems. The polyclonal anti-chemokine Ab were prepared by immunization with whole chemokine proteins and affinity purified. Alexa Fluor-conjugated secondary Abs were from Molecular Probes. Collagenase D and DNase I were from Roche. Hyaluronidase was from Sigma-Aldrich. TaqDNA polymerase was from New England Biolabs.

DC isolation and analyses by flow cytometer
Lung CD11c ϩ cells from naive BALB/cAnNCr (Charles River Laboratories) were enriched as previously described (14) using anti-CD11cconjugated magnetic microbeads (Miltenyi Biotec) and were comprised of 10 -15% CD103 ϩ DC, 10 -12% CD11b high DC, and 50 -55% alveolar macrophages. PDC were enriched from the flow-through CD11c Ϫ lung cells using anti-mPDCA-1 magnetic microbeads. Magnetic bead-selected DC fractions were further separated into subpopulations by cell sorting on a Vantage SE flow cytometer with DiVa configuration (BD Biosciences). CD103 ϩ DC, CD11b high DC, macrophages, and an additional minor CD11c ϩ cell population were isolated in a four-way sort. CD103 ϩ DC, CD11b high DC, and macrophage populations were gated on CD103 ϩ IA ϩ CD11b low , CD103 Ϫ I-A ϩ CD11b high , and CD103 Ϫ I-A low CD11b low cells, respectively. PDC were sorted by FACS from anti-mPDCA-1-magnetic bead-selected cells for PDCA-1 ϩ B220 ϩ I-A int and low side-scatter cells. For flow cytometry analyses, lung cells were blocked with anti-Fc␥RII/ Fc␥RII mAb 2.4G2, stained with fluorophore-conjugated mAb, and analyzed on a FACSCalibur flow cytometer (BD Biosciences). Dead cells were excluded from analyses by 7-aminoactinomycin D (7-AAD) staining. Flow cytometry results were analyzed by the program FlowJo (Tree Star).

Microarray analysis of CD103 ϩ DC and CD11b high DC mRNA
Sorted DC populations (Ͼ98% purity) were lysed and the total RNA extracted immediately with an RNeasy kit (Qiagen). Good-quality RNA were obtained based on the profile of 28S and 18S RNA fractionated on an Agilent Bioanalyzer and the 5Ј to 3Ј ratios of housekeeping genes in Affymetrix gene chip analysis. Gene expression levels in the two CD11c high DC populations were probed with the Affymetrix Gene Chip mouse genome 430 2.0 array by the University of Virginia Biomolecular Facility and the analysis has been described previously (14). Three independent pairs of CD103 ϩ DC and CD11b high DC from different isolations were analyzed by the same chip batch. The preparation of cRNA probes and the hybridization conditions follow that recommended by the manufacturer. The CEL files provided by the Affymetrix MAS program were further normalized and background subtracted to provide gene expression levels using the program dChip (41).

Real-time PCR analysis of DC mRNA
Total RNA from FACS-sorted DC subsets (Ͼ98% purity) were reversetranscribed by the Advantage RT-for-PCR kit (BD Clontech). Real-time PCR analysis was performed in a Bio-Rad iCycler Thermal Cycler using Sybr green fluorescence as the readout, and data were analyzed by the iCycler program provided (Bio-Rad). PCR conditions were: 94°C, 22 s; 58 -62°C as appropriate, 30 s, and 72°C, 30 s for 39 cycles; 94°C, 22 s; 62°C, 30 s; and 72°C, 5 min for 1 cycle. Melt curves were obtained by increasing the temperature from 65 to 95°C in 0.5°C steps for 10 s. The primer sequences were generated by the program Primer 1 and the primers synthesized by IDT. The cDNA amplified, 5Ј-primer sequence, 3Ј-primer sequence, primer melting temperature and product size are shown in Table  I. Agarose gel (1.8%) electrophoresis was used to confirm the correct molecular size of the PCR products.

Mice and immunization
The immunization of BALB/cByJ mice (The Jackson Laboratory) for the induction of allergic AHR and lung inflammation was performed as previously described (42) with two notable exceptions: 1) animals were adoptively transferred i.v. with 2.5 ϫ 10 6 splenic CD4 ϩ DO11.10 transgenic T cells 2 days before DC sensitization; and 2) mice were sensitized intratracheally with bone marrow-derived DC (43) instead of splenic DC. For in vivo LPS treatment, mice were injected intratracheally with 10 g of LPS in 50 l of saline, and 20 h later, lungs were harvested for DC isolation. The protocols in this study have been approved by the University of Virginia Institutional Use and Care of Animals Committee.

Measurements of chemokine production by DC
For measuring chemokine production of DC and macrophage populations by intracellular staining, CD11c ϩ lung cells were isolated by magnetic microbeads as described above and plated on Teflon dishes in medium containing 2 M monensin and with or without TLR ligands (5 g/ml LPS and 25 g/ml poly(deoxyinosinic-deoxycytidylic acid) (poly(I:C)). After incubation for 4 h to allow chemokine induction and intracellular accumulation to occur, chemokine staining was performed essentially as described previously (44,45). After induction, cells were stained with anti-I-A-FITC and anti-CD103-PE for CD11c ϩ DC at 4°C. Cells were incubated with 1 g/ml 7-AAD, washed, and fixed in 2% paraformaldehyde containing 80 g/ml actinomycin D. For chemokine staining, cells were permeabilized in 0.2% saponin and blocked with nonfat dry milk, incubated with antichemokine Ab, washed, and further stained with Alexa Fluor 647-conjugated 2°donkey anti-rabbit or anti-goat IgG Ab. Intracellular chemokine levels of DC populations were analyzed on a FACSCalibur with gates set on 7-AAD Ϫ cells that are CD103 ϩ I-A ϩ for CD103 ϩ DC, CD103 Ϫ I-A ϩ cells for CD11b high DC, or I-A int CD103 Ϫ cells for macrophage popula-tions. The specificities of selected Ab for chemokine staining were confirmed by blocking chemokine staining of DC by graded doses of chemokines from 3.2 to 12.5 g/ml. The chemokine staining positivity of DC and macrophage populations in flow cytometry was expressed by Fluorescence Intensity Index which is the product of the percentage of positive cells multiplied by the geometric mean of the positive-staining cells. This measure roughly correlates with the number of synthesized chemokine molecules in the cells.
ELISA was used to measure the secretion of chemokines by sorted DC and macrophage populations. For the assay, FACS-purified DC or macrophages (Ͼ98% purity) were suspended at 2-5 ϫ 10 5 cells/ml in complete medium and cultured with or without 5 g/ml LPS and 25 g/ml poly(I:C) for 8 -10 h. Supernatants were collected and assayed for chemokines by ELISA (R&D Systems). The sensitivities of the assays ranged from 1 to 10 pg/ml.

(25)
Ϫ1.21 Total RNA were prepared from FACS-purified CD11b high DC and CD103 ϩ DC populations and used in microarray or real-time PCR analyses as described in Materials and Methods.
The purities for each of the DC subsets were Ͼ98%. The chemokines were groups into three types: differentially expressed chemokines that were expressed in CD11b high DC at Ͼ3-fold higher levels than in CD103 ϩ DC in microarray analyses; comparably expressed chemokines whose intensity levels in CD11b high DC and CD103 ϩ DC were within a ratio of 3.0; and weakly expressed chemokines whose microarray intensities in both DC types were Ͻ500. Chemokines with microarray intensities of Ͻ100 were not shown. The microarray data were averages from three independent experiments with standard error of the mean in parentheses. Negative values in microarray ratios represented the reciprocal value of the indicated ratio. For chemokines mRNA intensities measured by more than one probe, only one was presented. Real-time PCR for each chemokine mRNA has been performed at least twice, and one representative experiment was presented as mean of triplicate samples (SE). mRNA with the higher value between the two specimens and mRNA ratios were stained with Alexa Fluor 647-conjugated anti-DO11.10 mAb (eBioscience) and directly assayed for proliferation by flow cytometry or fixed and permeabilized for intracellular cytokine staining as described earlier for intracellular chemokine staining. Cell division indices, the product of proliferation indices and percentage of divided cells, were determined by the program FlowJo. Proliferation index is the average number of divisions that dividing cells have undergone.

Statistics
The mean and SD of multiple trials were calculated by the program Excel (Microsoft). Statistical significance using Student's t test was determined by the programs SlideWrite plus (Advanced Graphics Software) and InStat 3 (GraphPad). For multiple comparisons in a data set, significance was further assessed by the false discovery rate procedure (48).

Lung CD11b high DC and CD103 ϩ DC differentially express chemokine mRNA in naive mice
Global mRNA expression analysis of the two major lung CD11c high DC subsets in naive mice by Affymetrix microarray analysis showed that CD11b high DC expressed many chemokine mRNA species at much higher levels than CD103 ϩ DC (Table II), suggesting that CD11b high DC was the major chemokine-producing lung DC. Five type I (C-X-C) and 8 type II (C-C) chemokines were expressed at 3-to 60-fold higher levels by CD11b high DC. These chemokines mediate the migration of all leukocyte types . Several chemokine mRNA were found in CD11b high DC and CD103 ϩ DC at comparable levels. Interestingly, many of these chemokines bind CCR3 or CCR4 (C-10, TARC, MDC) and are responsible for stimulating Th2 cells. A few chemokine mRNA species were expressed at low levels by both CD11c high DC types. Thus, homeostatic CD11b high DC expressed at least 16 chemokine mRNA at very high levels (intensity Ͼ 1000; maximal intensity in array set Х8000), whereas CD103 ϩ DC highly expressed only 6, 4 of which at comparable levels with CD11b high DC, and 3 of which bind to Th2 cells.
The microarray results were confirmed by real-time PCR analysis (Table II). The ratios of chemokine mRNA levels in CD11b high DC vs that in CD103 ϩ DC obtained by real-time PCR were comparable to the microarray results. The PCR products were confirmed to be of expected size by agarose gel electrophoresis (data not shown). Thus, two complementary techniques showed that CD11b high lung DC expressed more species of chemokine mRNA and at higher levels than CD103 ϩ DC with only a few exceptions.

Chemokine production by lung CD11c ϩ DC and macrophages
To study lung DC chemokine production, two independent approaches were taken. The first is to quantitate chemokine ac-cumulation in DC populations by intracellular staining (45). The second is to measure chemokine secretion of sort-purified DC populations by ELISA. In the intracellular chemokine assays, a panel of affinity-purified anti-chemokine protein Ab was used to stain cells with and without incubation in monensincontaining medium. Without further culturing, DC and macrophages already accumulated detectable amounts of chemokines, the most prominent of which were MIP-2, IP-10, MIP-1␣, MIP-1␤, RANTES, CXCL16, C10, TARC, and MDC for CD11b high DC, MIP-1␣, CXCL16, TARC, and MDC for CD103 ϩ DC, and IP-10, MIP-1␥, and C10 for macrophages (Fig. 1, B and C). Time course studies showed that peak chemokine accumulation was reached at 4 h. The same percentages of CD11b high DC, CD103 ϩ DC, and macrophage populations were found with or without culturing, indicating that the three lung CD11c ϩ subsets remained distinct populations in short term cultures (Fig.  1A). However, there were some up-regulation of I-A and downmodulation of CD103 within the CD103 ϩ DC population (cf Fig. 1, A-b and A-d). The 4-h increase in chemokine staining used as a measure of production rate was the highest in CD11b high DC in 8 of 10 differentially expressed chemokines (Table III), and the fluorescence intensity indices were 4-to 20-fold higher than that of CD103 ϩ DC. Interestingly, CD103 ϩ DC produced slightly higher or comparable amounts of Th2stimulating chemokines TARC and MDC, in addition to CXCL16. Chemokine production by resting macrophages was generally low. However, they produced the highest amounts of MIP-1␥ and C10. These results showed that chemokine production by DC subsets and macrophages is in general agreement with their mRNA accumulation and CD11b high DC produced the highest amount and most number of species of chemokine in homeostasis.
Chemokines in 8-to 10-h culture supernatants of FACS-purified CD11b high DC, CD103 ϩ DC, and macrophages were also measured by ELISA, and the results of the most highly produced chemokines are shown in Fig. 2. The three lung CD11c ϩ populations  were Ͼ98% purity and showed the expected staining patterns after purification ( Fig. 2A). In control medium, CD11b high DC produced the highest amounts of KC, MIP-2, MIP-1␣, MIP-1␤, and RANTES, whereas CD103 ϩ DC produced much small amounts of the same chemokines but a similar amount of MDC (Fig. 2B). Macrophages, in contrast, produced substantial amounts of the type I chemokines KC and MIP-2 and the highest amount of C10. These ELISA results are similar to those obtained by intracellular staining and are consistent with the mRNA accumulation results. The results again support the thesis that CD11b high DC is the major chemokine producing lung DC population and is a major myeloid cell type in homeostatic chemokine production.

TLR ligands stimulate chemokine production by CD103 ϩ DC in vitro
Because TLR ligands are major DC stimuli and may provide the initial stimuli for airway antigenic responses, the response of DC subsets to these ligands were tested in vitro. LPS and poly(I:C) were chosen as the stimuli because the receptors for these ligands have been shown to be on lung DC and they induced DC production of IL-12 efficiently (14). Stimulation with these ligand induced little increase in chemokine production by CD11b high DC, perhaps because their chemokine production is already near the maximal rate (Fig. 3A). However, LPS and poly(I:C) induced large increases in chemokine production by CD103 ϩ DC. The most notable chemokine increases were MIP-1␣, MIP-1␤, and RANTES (Fig. 3B). The most prominent increases in macrophage chemokine production were KC, MIP-2, and MIP-1␣.  The increases in intracellular chemokine staining induced by LPS plus poly(IC) were similar to that obtained by ELISA measurements of stimulated purified DC and macrophage populations (Fig. 2B). KC, MIP-2, and MIP-1␣ production by CD11b high DC showed little increases, perhaps because of the high baseline production in unstimulated cultures or the reutilization by the DC (Figs. 2B and 3). However, MIP-1␣, MIP-1␤, and RANTES production by CD103 ϩ DC increased several fold (Figs. 2B and 3B). Notable increases of KC, MIP-2, and MIP-1 production up to 2-fold were also detected in stimulated macrophage cultures (Fig.  2B). Thus, CD103 ϩ DC have the potential to produce higher amounts of type I and type II chemokines besides MDC and CXCL16. However, in these stimulated cultures, CD11b high DC were still the predominant chemokine producer compared to CD103 ϩ DC.

In vivo LPS stimulation increased chemokine production by CD11b high DC
Lung DC have been found to turnover rapidly upon intratracheal stimulation by heat-killed bacteria (51), an effect related to chemokine production by TLR-stimulated lung cells. An in vivo study was performed to assess the chemokine produced and the DC type involved in the production during intratracheal LPS stimulation. Intracellular staining of isolated CD11c ϩ lung populations showed that large increases of the type I chemokines KC and MIP-2 were found in CD11b high DC and macrophages and in lower amounts in CD103 ϩ DC ( Fig. 4 and Table IV). This increase explains the CD11c ϩ cells were prepared from the pooled lung cell suspensions as described in Fig. 1, incubated in medium for 0 or 4 h, and stained for intracellular chemokine accumulation. The figure shows accumulation of the indicated chemokines (ag) by CD11b high DC (left column), CD103 ϩ DC (middle column), and macrophages (right column) in unincubated (dotted line) or in vitro incubated (solid line) CD11c ϩ populations from LPSinjected mice and gates were set as in Fig. 3. The results from salineinjected mice were similar to those in Fig. 1 and not shown. This experiment was repeated twice.

FIGURE 5.
Chemokine production by FACS-purified lung DC and macrophage populations stimulated in vivo by LPS or poly(I:C) stimulation. Groups of 10 mice were injected with saline, 10 g of LPS, or 25 g of poly(I:C) as described in Fig. 4 and 20 h later harvested for CD11c ϩ lung cell isolation as described in Fig. 1A. The unsorted CD11c ϩ lung cells from LPS-injected mice (a) were sorted into CD11b high DC (b), CD103 ϩ DC (c), and macrophages (d) as described in Fig. 2. Sorted cells were 97-99% pure. The contaminating events likely represent cells that died during sorting (b, c). Sorting results of CD11c ϩ lung cells from poly(I:C)injected mice were similar and not shown. Chemokine production by sorted lung CD11c ϩ populations from LPS-injected (B) or poly(I:C)-injected (c) mice were assessed as described in Fig. 2. One of two experiments is shown. Bar represents SD of duplicate samples. The p value designations were the same as in Fig. 2. rapid influx of neutrophils (PMN) into lungs upon bacteria infection because KC and MIP-2 preferentially chemoattract PMN. LPS also markedly increased IP-10, CXCL16, RANTES, C10, and MIP-1␥ production by CD11b high DC but much more moderately in CD103 ϩ DC (Table IV). With LPS stimulation, macrophages produced large amounts of MIP-1␣ and MIP-2 and are the highest producers of C10 and MIP-1␥ (Table IV). Thus, in in vivo stimulation in lungs, both CD11b high DC and macrophages are induced to produce large amounts of chemokines. To further confirm these results, lung DC and macrophages were isolated from mice injected intratracheally with LPS by cell sorting (Fig. 5A) and cultured in vitro. The amounts of several chemokines in the culture supernatants of these purified DC subsets and macrophages were measured (Fig. 5B) and the results again showed that CD11b high DC are the major chemokine-producing cell types. Lung DC and macrophages were also sorted from poly(I:C)-injected mice to 97-98% purity. These poly(I:C)-stimulated DC and macrophages produced similar amounts of chemokines, except that more RANTES was produced by CD103 ϩ DC than those stimulated by LPS in vivo.

Lung DC and macrophage chemokine production in allergic inflammation
While LPS stimulation results in a biased Th1 response, allergic airway inflammation is Th2 biased. The model of airway inflam-mation induction by DC priming followed by OVA challenges was used to study chemokine production by CD11c ϩ cells. Mice immunized with OVA showed significant AHR, as shown by both FIGURE 6. Chemokine production by lung CD11c ϩ DC and macrophage populations from OVA-immunized mice. Six mice were primed with OVA-pulsed bone marrow-derived DC and challenged with OVA as described in Materials and Methods. A, Mice were sacrificed for CD11c ϩ cell preparation and intracellular chemokine staining was performed as described in Fig. 1. The figure shows accumulation of the indicated chemokines (rows a-l) in cell populations (labeled on top of columns) with (solid line) and without (dotted line) in vitro incubation in the presence of monensin, and the percentages shown were from cells with the incubation. Separate experiments with CD45.1 DC priming showed that Ͻ0.5% lung CD11c ϩ cells were exogenous DC during the first OVA challenge and even less at harvest 7 days later.  airway pressure time index (36,52) and Penh measurements, and marked lung inflammation (14,42). Intracellular chemokine staining of CD11c ϩ lung cells showed that increased chemokine accumulation was readily detectable after 4 h in vitro culture in all three cell types examined (Fig. 6). In contrast to short-term intratracheal stimulation with LPS or poly(I:C), which mostly stimulated chemokine production by CD11b high DC but the production of only limited number of chemokine species by CD103 ϩ DC and macrophages (Fig. 4), this chronic antigenic stimulation and lung inflammation caused a more generalized increase in chemokine pro-duction by all three CD11c ϩ cell populations (Fig. 6). chemokine fluorescence intensity index after culture were markedly increased for 11 of 12 chemokines for CD11b high DC (Table V). Importantly, production of 8 of 12 chemokines were increased from moderate to high levels in CD103 ϩ DC, and 10 of 12 species were increased in macrophages. CD11b high DC remained an important chemokine producer and accumulated severalfold higher chemokines than CD103 ϩ DC. Furthermore, because CD11b high DC numbers is 6-to 10-fold higher in inflamed lungs than CD103 ϩ DC (14), the contribution of the former in chemokine production is much more important. However, CD103 ϩ DC produced the highest levels of the CCR4-binding and therefore Th2-stimulating chemokines TARC and MDC. Macrophages produced the highest levels of the PMN-attracting chemokines KC, MIP-2, and MIP-1␥ and the macrophages-stimulating chemokine C10, similar to LPS stimulation (Table IV).
Chemokine production measurements of sort-purified DC and macrophages yielded results similar to those obtained by intracellular chemokine staining (Fig. 7). Four of the eight chemokines measured were produced at higher levels, all by CD11b high DC and macrophages at comparable levels, except C10, which is produced at 3-to 4-fold higher amounts by macrophages (Fig. 7B). The other four chemokines were produced at lower levels. For these latter chemokines, CD103 ϩ DC produced as much or at somewhat lower levels as CD11b high DC and macrophages. Interestingly, CD103 ϩ FIGURE 7. Chemokine production by sort-purified lung DC and macrophage populations from OVA-immunized mice. Mice were immunized as described in Materials and Methods and sacrificed for CD11c ϩ lung cell preparation as described in Fig. 2. The panels in A show presort (a) and purified postsort cells (bd). a, Eosinophils with high side-scatter were excluded from the DC and macrophage gate in the forward-vs side-scatter gate and any residual eosinophils were CD11c low and excluded in the CD11c ϩ gate. The chemokine production of purified DC and macrophage populations are shown in B and C with the chemokines produced in larger amounts shown in B. The experiments have been performed twice and bars indicate SD. The designations of p values were identical to that in Fig. 2. DC produced the largest amounts of the Th2 cell-attracting chemokine MDC. These results showed that in inflamed lungs, CD11b high DC is a much more important chemokine producer than CD103 ϩ DC. However, the latter produced the most MDC.

MDC effects on in vitro T cell proliferation
The uniquely high level of MDC production by the epithelial CD103 ϩ lung DC suggests that MDC may serve important functions in Ag presentation by this DC type. This possibility was explored in an in vivo T cell proliferation assay (Fig. 8). Under the assay conditions, proliferation was observed on day 4 and maximal between days 5 and 6. CD4 ϩ CD25 Ϫ DO11.10 ϩ transgenic T cells alone with no APC stimulation underwent little proliferation (data not shown), and in the absence of peptides, CD11b high DC or CD103 ϩ DC could stimulate 25-57% of T cells to undergo only one division cycle (Fig. 8, A and E). With OVA 323-339 peptides, CD11b high DC stimulated more T cell proliferation, as observed earlier in [ 3 H]thymidine incorporation (cf Fig. 8, B and F; Ref. 14). The presence of anti-MDC Ab had a small effect on T cell proliferation stimulated by CD11b high DC (Fig. 8C). However, MDC reduced the percentage of dividing cells by 4-to 5-fold, but no change in the profile of cell division, resulting in the reduction of division index by the same magnitude. In T cell proliferation stimulated by CD103 ϩ DC, on the other hand, anti-MDC Ab increased the percentage of dividing cells by 5-to 6-fold while maintaining a similar division profile. MDC in these cultures have a slight inhibitory effect on the percentage of dividing T cells. These results suggest that MDC may serve a regulatory role in T cell proliferation and dampen lung mucosal antigenic responses.

Th subset responses to lung DC stimulation in vitro
Lung CD11c ϩ cells have been shown to express IL-6 and IL-10 mRNA after intranasal parasite inoculation, suggesting that they mediate biased Th2 responses (25). Whether lung DC subsets mediate divergent Th subset responses were examined. In in vitro cultures, T cells secreting both Th1-and Th2-type cytokines were found. Although CD11b high DC are similar to the splenic CD11b ϩ DC and are expected to stimulate a biased Th2 response (26,53), the results showed that they stimulated a higher percentage of IFN-␥-producing cells than CD103 ϩ DC (cf Fig. 9, B-a and C-a). The blockage of MDC functions, which increased T cell proliferation (Fig. 8), reduced the percentage of IFN-␥-producing T cells and increased the percentage of IL-4-and IL-10-producing cells in CD11b high DC-stimulated cultures, thus rendering the cultures more Th2-biased (Fig. 9, B-d and B-f). In CD103 ϩ DC stimulated cultures, anti-MDC Ab also promoted Th2 cell occurrence by increasing IL-10-producing cell percentage by 3-fold (18 -53% dividing cells; compare Fig. 9, C-c and C-f) while elevating IFN-␥producing cells only moderately (13-19% dividing cells; Fig. 9, C-a and C-d). The results showed that there is little distinction between lung CD11b high DC and CD103 ϩ DC in their Th subsetdirecting effects in vitro. However, the blocking of MDC promotes Th2 cell development.

Discussion
Although the two major lung CD11c ϩ DC subsets are distinguishable by a number of surface markers and by their localization with respect to the bronchial epithelium, no functional distinction has been described previously (14). This study shows that the CD11b high DC subset is a major producer of a wide variety of chemokines and protein secretion is one of its main functions. CD103 ϩ DC, in contrast, is much more limited in their secretory capacities and produce limited species of chemokines. Microarray data also showed that although both DC types expressed high levels of TNF-␣, IL-1␤, IL-12p40, and IL-18 mRNA, CD11b high DC expressed 3-to 10-fold higher IL-1␣, IL-6, and IL-10 mRNA than CD103 ϩ DC (S.-s. J. Sung, unpublished results), further supporting the more versatile role of CD11b high DC as a secretory cell type. Assessment of the Ag presentation differences between CD11b high DC and CD103 ϩ DC by in vitro T cell proliferation assays has yielded little substantial differences between their Ag presentation potencies and their Th subset-directing capabilities, although CD11b high DC were somewhat more efficient in inducing CD4 ϩ T cell division and Th1 cell development. Additional data from microarray analysis followed by cellular staining showed that CD11b high and CD103 ϩ DC differ in their expression of several chemokine receptors, TLR, and cytokine receptors, thereby further FIGURE 9. Cytokine production by CD4 ϩ T cells stimulated by CD11b high DC or CD103 ϩ DC. CD4 ϩ CD25 Ϫ DO11.10 ϩ TCR transgenic T cells were stimulated by OVA323-339 peptide presented by CD11b high DC (B) or CD103 ϩ DC (C) with (d-f) or without (ac) anti-MDC Ab for 5 day as described in Materials and Methods, stimulated for an additional 5 h for cytokine induction, stained with anti-DO11.10 mAb, fixed by 2% p-formaldehyde, and stained again for the intracellular cytokines (indicated at the top of the panels. The gating of the DO11.10 ϩ live T cell population (7-AAD staining not shown) is shown in A with the lymphoid area gated as shown in a, and the DO11.10 ϩ cells gated as shown in b. Control Ab for cytokine staining showed little background (A-C). One of two similar experiments is shown. differentiating their surface phenotypes and suggesting the functional difference of these two lung DC subsets. Studies of intestinal CD103 ϩ DC have shown that this DC subset is far superior to the CD103 Ϫ DC subset in inducing CCR9 and LPAM-1 expression on gut-tropic T cells (17,18). This current study provides further support that CD11b high DC and CD103 ϩ DC are functionally distinct.
Chemokines are essential mediators for allergic asthma (1,21,54). The progression of lung chemokine production during the development of allergic airway inflammation is well documented (22,55,56), and the requirement of chemokines MDC, TARC, CXCR2, and CCR6 for disease manifestation has been described previously (29,(31)(32)(33). Other chemokine and chemokine receptors such as CCR1 and C10 are involved in inflammation-associated airway remodeling (30,57,58). Previous studies on the cellular sources of chemokines produced during allergic airway inflammation induction have indicated that macrophages are the primary sources of a number of chemokines (22) and lung DC have not been considered a significant contributor. Results on DC production from other studies are fragmentary, with the identification of a limited number of chemokines produced by DC mixtures and from different tissue sources following various stimuli but did not directly support the important role of lung DC as a chemokine producer in allergic airway diseases (37, 38, 40, 59 -62). The results in this study establish that CD11b high lung DC is a major chemokine-producing cell in homeostasis and upon stimulation by airway Ag and TLR ligands. Their role in the homeostatic state is of particular importance because there are few other myeloid cells in normal lungs besides macrophages and DC and resting macrophages produced less chemokine than CD11b high DC except C10 (Table III and Fig. 2). The results also indicate that lung CD103 ϩ DC and PDC are minor chemokine producers, although they may produce certain chemokines (e.g., MDC by CD103 ϩ DC) for specialized functions such as the regulation of T cell responses. Chemokines detected in lung during the induction of allergic airway inflammation can be attributed to a large extent by the chemokine production by CD11b high DC and macrophages (22). It is important to note that there is a continual increase in CD11b high DC in lungs during this increasing inflammation (compare Figs. 2A and 7A). In allergic airway inflammation, when CD11b high DC is as numerous as macrophages in lungs, the two cell types remain the predominant chemokine-producing cells (Table V and Fig. 7). Notable deficiency of chemokine synthesis by CD11b high DC are CCL11/eotaxin and CCL20/MIP-3␣ which are produced by airway epithelial cells (63,64), and CCL21/SLC which is produced by lymphatic endothelial cells (65). These latter three chemokines are important in Ag-induced lung inflammation in inducing eosinophilia and in mediating macrophage and DC influx into lungs by their CCR3, CCR6, and CCR7 binding (31,36,66,67).
Although poor chemokine producers, CD103 ϩ DC produced the highest amounts of MDC in resting and stimulated conditions. MDC recruits Th2 cells and regulatory T cells by binding to their surface CCR4 (21,68). Considering the localization of CD103 ϩ DC on the lung vascular endothelium by the interaction of their surface integrin ␣ E ␤ 7 with vascular lymphocyte-endothelial-epithelial-cell adhesion molecule (14,69), CD103 ϩ DC secretion of MDC may serve in mediating the extravasation of Th2 cells in blood vessels. The prevention of eosinophil extravasation into lung parenchyma and airway hyperreactivity induction by anti-MDC Ab (33) supports this MDC function. In addition to extravasation induction, MDC may have a role in CD103 ϩ DC regulation. Lung CD4 ϩ intraepithelial T cells express integrin ␣ E ␤ 7 and are adjacent to CD103 ϩ DC at the airway epithelia. Many of these CD103 ϩ CD4 ϩ epithelial T cells may have regulatory functions and express CCR4 (18,68). The secretion of MDC by CD103 ϩ DC to chemoattract and stimulate the adjacent CD103 ϩ CD4 ϩ regulatory T cells may be a significant mechanism for the interaction of CD103 ϩ DC and regulatory T cells in lungs. The current study also defined a regulatory role for the high MDC production by lung epithelial CD103 Ϫ DC. The finding that MDC inhibited in vitro T cell proliferation, further supported by the increase in CD4 ϩ T cell proliferation in the presence of anti-MDC Ab (Fig. 8), suggested that MDC may play a significant role in maintaining the quiescent state in normal lungs by moderating airway T cell responses to airway Ag and TLR ligands. This chemokine may also play a role in promoting the lung Th2-biased antigenic responses (Fig. 9).
In other tissues, Langerhans cells but not splenic DC produce large amounts of MDC (61), suggesting that lung CD103 ϩ DC may be similar to Langerhans cells not only in Langerin expression (14), but also in their chemokine production. CD103 ϩ DC were found to produce large amounts of different chemokines when stimulated in vitro by LPS and poly(I:C) (Fig. 3). Thus, CD103 ϩ DC possess the potential to produce many chemokines at high levels when optimally stimulated. However, they produced little chemokines, except RANTES, when stimulated in vivo by TLR ligands. DC migration studies after intratracheal TLR ligand injection indicated that a major function of TLR ligands may be to recruit DC into lungs, and the major infiltrating DC population was the CD11b high DC (Fig. 5A). CD103 ϩ DC numbers in lungs were markedly reduced by day 2 either due to their exit to the draining lymph nodes, or the fact that activated DC can no longer be identified because of the down-regulation of CD103 ϩ on their surface (S.-s. J. Sung, unpublished results). A possibility for the low chemokine production by CD103 ϩ DC in TLR ligand-treated mice is that CD103 ϩ DC are highly mobile following TLR ligand stimulation and the CD103 ϩ DC studied may be newly recruited cells that have seen little TLR ligand.
The limited capability of CD103 ϩ DC to synthesize chemokines in homeostasis may have a functional requirement. Considering that this DC type is juxtaposed against lung blood vessel endothelial cells, it is undesirable for them to be exquisitely sensitive to stimuli and produce chemokines continually in response to small perturbations to promote excessive migration of blood leukocytes into the lung parenchyma. CD11b high DC, in contrast, are situated farther away in the subepithelia and from the blood vessels and can amplify stimulatory signals from cytokines or from direct T cell contacts to mount a more local immune response without affecting leukocyte extravasation. During inflammation, chemokine production by CD11b high DC is increased not only on a per cell basis but also by the marked increase in their numbers in lungs. Furthermore, in the intestines, CD11b high DC acquire tight junction proteins upon bacterial infection and are essential for the antibacterial functions (70). Thus TLR ligand-stimulated CD11b high DC are more versatile in their migration capabilities and may function at the vascular interface during inflammation to induce rapid leukocyte influx. This diversification of chemokine production by lung DC subsets may insulate blood cells from rapid leukocyte influx during homeostasis and yet maintain a flexible and effective machinery for coping with local responses. During inflammation, rapid mobilization of cellular responses can be readily achieved by the activation of CD11b high DC and macrophages to produce chemokines, cytokines, and other mediators.
Lung macrophages produce the highest amounts of C10 and MIP-1␥ (Tables III and V), both of which bind to CCR1, which is essential for allergic pulmonary inflammation and remodeling (57,58). Thus, macrophages may also play a key role in asthma pathogenesis. It is also interesting to note that, upon LPS stimulation in vivo, both CD11b high DC and macrophages produce high levels of type I chemokines KC and MIP-2 which attract PMN influx. Thus, the observation that bacterial infection is accompanied by rapid influx of PMN can be explained by this chemokine production.
This study clarified the role of CD11c ϩ lung cell subpopulations in chemokine production and identified CD11b high DC as a major chemokine producer in homeostasis and in induced lung inflammation. The results also indicated that CD11b high DC and CD103 ϩ DC have differential roles in effector molecule secretion and chemokine production. This difference further supported CD11b high DC and CD103 ϩ DC as functionally distinct lung DC subsets. The finding of CD11b high DC as a major chemokine producer has also been identified this DC subset as an important target for the intervention of mediator of mediator secretion in asthma therapy.