Characterization of Myeloid and Plasmacytoid Dendritic Cells in Human Lung¹

The lung is the largest epithelial surface in the body and constitutes a major portal of entry for microbes and other environmental Ags. Dendritic cells (DCs)⁵ are critical APCs that respond immediately with innate immune responses and then initiate and regulate acquired immune responses (reviewed in Ref. 1). Thus, DCs almost certainly play an important role in how the lung responds to infectious challenges. The microenvironment of the lung likely influences the phenotype and function of DCs, including determining their ability to drive tolerance vs productive immunity (2, 3).

The role of DCs in regulating pulmonary immunity has been studied by a number of laboratories, primarily by using experimental animals (reviewed in Refs. 4, 5, 6, 7, 8), and studies on human lung DCs are limited. In man, two major types of DCs have been identified, myeloid DCs (MDCs) and plasmacytoid DCs (PDCs) (9, 10), but the location of both types in the lung, and their phenotypes and functions are incompletely described. In this study we did the following: 1) determined the numbers of MDCs and PDCs in lung tissue excised during the performance of surgical excision for medical reasons; 2) evaluated accessory molecule expression on MDC- and PDC-enriched populations, 3) assessed the difference in allostimulation between MDC-enriched and MDC-depleted monocyte-enriched lung cell populations; 4) determined the cytokine secretion profile of MDCs and PDCs in response to TLR-4 and TLR-7 agonists; and 5) characterized by immunohistology the sites within the airways and parenchyma where MDCs and PDCs might exist. CD11c is a sensitive and specific marker for murine DCs, including both MDCs and PDCs, but it does not identify PDCs in man. Furthermore, CD11c is not specific for MDCs, because it is present on many monocytes. Thus, to identify MDCs, CD11c⁺CD14⁻HLA-DR⁺ lineage-negative mononuclear cells were selected as containing MDCs but not monocytes and then further characterized with CD1a and/or CD1c expression to specifically identify MDCs. PDCs were identified as CD11⁻CD14⁻HLA-DR⁺ lineage-negative mononuclear cells that also expressed CD123. In our hands, unlike the previous experience reported by Demedts et al. (11), we found that blood DC Ag (BDCA)-2 was an insensitive marker for PDCs in the lung, possibly because as previously reported (12) it is down-regulated on PDCs as they mature, at least in culture. Using this labeling strategy, we have enumerated MDCs and PDCs and have begun to functionally characterize and locate them within human lung.

Lung tissue samples were obtained from 22 patients undergoing diagnostic or therapeutic thoracotomies. Signed informed consent was obtained from each subject. Approval of the protocol and consent form was granted by the University of New Mexico’s Human Research Review Committee (HRRC). Tissue not involved with the primary lesion for which the tissue was resected was excluded by the pathology assistant or pathology resident. The distribution of final pathologic diagnoses for the patients is as follows: granulomatous inflammation, 5; emphysema, 1; carcinoid, 2; primary lung cancer, 9; and metastatic cancer, 3. The distribution of the smoking status of the patients is as follows: never smoked cigarettes but an occasional cigar, 2; stopped smoking for greater than 6–12 mo, 4; and current smoker, 14.

Lung tissue was transported in cRPMI (RPMI 1640 supplemented with 2 mM l-glutamine, 1 mM nonessential amino acids, 1 mM sodium pyruvate, 100U/ml penicillin, and 100 μg/ml streptomycin) with 10% heat inactivated FBS (Invitrogen Life Technologies). Small pieces of the tissue were removed for immunohistochemistry studies. The remaining tissue was weighed and minced into small cubes and placed in cRPMI with 10% heat inactivated FBS, collagenase A (0.7 mg/ml; Roche), and type IV bovine pancreatic DNase I (30 μg/ml, Sigma-Aldrich) for enzymatic digestion at 37°C for 90 min. The tissue was then pressed through a metal screen and particulate matter was removed by rapid filtration over a nylon wool plug. An aliquot of cells was removed for total cell enumeration using trypan blue for dead cell exclusion, and cytospins were made for a differential count. Cells were resuspended in high-density Percoll (ρ = 1.075 g/ml; Amersham Biosciences), and an equal amount of low-density Percoll (ρ = 1.030 g/ml) was layered over the cells and centrifuged at 400 × g for 20 min. The mononuclear cell layer at the 1.075/1.030 Percoll interface was then removed and washed with HBSS (Invitrogen Life Technologies). Any contaminating RBCs were lysed with sodium chloride potassium lysis buffer (0.826% NH₄Cl, 0.1% KHCO₃, and 0.0037% EDTA-disodium salt in distilled water (pH 7.3)). The post-Percoll cells were counted on a hemocytometer with trypan blue for dead cell exclusion and cytospins were made for a differential count. Cells were then used for four-color immunophenotyping or frozen at −80°C in 90% FBS and 10% DMSO for later use.

Mononuclear cells were incubated with FcR blocking reagent (Miltenyi Biotec) to block FcR-mediated binding of staining Ab and cell aggregation by FcγR binding by macrophages of Ab-stained cells. Subsequently, cells were stained with the mAbs anti-CD14-FITC (M5E2, mouse IgG2a), anti-HLA-DR-PerCP (L243, mouse IgG2a), anti-CD11c-allophycocyanin (B-ly6, mouse IgG1), and a PE-conjugated mAb at the surface molecule of interest as follows: anti-CD86 (IT2.2, mouse IgG2b), anti-CD80 (L307.4, mouse IgG1), anti-CD40 (5C3, mouse IgG1), anti-CD1a (HI149, mouse IgG1), anti-CD1c (M241, mouse IgG1 (Ancell); or AD5–8E7, mouse IgG2a (Miltenyi Biotec)), anti-CD83 (HB15e, mouse IgG1), anti-CD123 (AC145, mouse IgG2a (Miltenyi Biotec)), anti-BDCA-2 (AC144, mouse IgG1 (Miltenyi Biotec)), anti-CD54 (HA58, mouse IgG1), anti-CD3 (HIT3a, mouse IgG2a), anti-CD19 (HIB19, mouse IgG1), anti-CD56 (B159, mouse IgG1), anti-CD11b (ICRF44, mouse IgG1), mouse IgG1 (MOPC-21), mouse IgG2a (G155-178), or mouse IgG2b. One set of cells was stained with mouse IgG2a-FITC (G155-178), mouse IgG2a-PerCP (X39), and mouse IgG1-allophycocyanin (MOPC-21) to set the appropriate gates by flow cytometry. All Abs were purchased from BD Biosciences unless otherwise noted. Cells were stained at 4°C followed by two washes with PBS containing 2% FBS and 40 μg/ml EDTA. Cells were fixed with 2% paraformaldehyde. Data acquisition was performed on a BD Biosciences FACSCalibur and analysis was performed with WinList software (Verity Software House).

Mononuclear lung cells were washed with HBSS and resuspended in cRPMI containing 10% FBS. The cells were then layered over 30% Percoll and centrifuged for 20 min at 800 × g to remove dead cells and debris. Cells were first incubated with FcR blocking reagent and subsequently stained with a lineage mixture of FITC-labeled mAbs against CD3, CD14, CD16, CD19, CD20, and CD56 and with anti-CD123-PE, anti-HLA-DR-PerCP, and anti-CD11c-allophycocyanin. Cells were then sorted using a MoFlo cytometer (DakoCytomation) to obtain a CD123⁺ PDC (lineage-negative CD11c⁻HLA-DR⁺) population and/or a CD123⁺ and/or a CD123⁻ MDC (lineage-negative CD11c⁺HLA-DR⁺) population. The forward scatter/side scatter profile was used to gate out debris and large, highly autofluorescent cells. A second gate was set to exclude lineage-positive T cells (CD3), monocytes (CD14), B cells (CD19), and NK cells (CD16 and CD56) and collect HLA-DR⁺ cells. The lineage-negative HLA-DR⁺ population was used to establish additional gates to collect CD11c⁻CD123⁺, CD11c⁺CD123⁻, or CD11c⁺CD123⁺ cells.

Cells (either 1.25 × 10⁴ or 2.5 × 10⁴) were stimulated overnight in cRPMI with 10% FBS and 2-ME in a volume of 200 μl in 96-well round-bottom plates. At the time of plating either LPS (lipopolysaccharides from Escherichia coli 0111:B4; Sigma-Aldrich catalog no. L 4391) was added at a final concentration of 1 μg/ml or influenza A (influenza A/PR/8/34 (H1N1); Advanced Biotechnologies catalog no. 10-210-000) was added at an multiplicity of infection (MOI) of 10. Culture supernatants were collected and multianalyte profiling was performed and acquired on the Luminex LX100 system (software version 1.7) with the XY Platform (Luminex). Calibration microspheres and sheath fluid were also purchased from Luminex. Acquired fluorescence data were analyzed using StatLia from Brendan Scientific (version 3.2). Supernatant concentrations of cytokines (IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12 p40, IFN-α, IFN-γ, and TNF-α) and chemokines (eotaxin, GROα, IP-10, MCP-1, MCP-2, MIP-1α, MIG, and RANTES) were measured by combining single bead kits for IL-1α, IFN-α, and IL-12 p40 with a 10-plex cytokine panel and a 10-plex chemokine panel, all from BioSource International. All analyses were performed according to the manufacturer’s protocol.

Mononuclear lung cells were washed with HBSS and resuspended in cRPMI containing 10% FBS. The cells were then layered over 30% Percoll and centrifuged for 20 min at 800 × g to remove dead cells and debris. Cells were first incubated with FcR blocking reagent and subsequently stained with anti-CD14-FITC, anti-CD3-PE, anti-CD19-PE, anti-CD56-PE, and anti-CD11c-PE-Cy5. Cells were then sorted using a MoFlo cytometer (DakoCytomation) to obtain an enriched MDC (CD11c⁺CD14⁻) population and an enriched monocyte/macrophage (CD11c⁺CD14⁺) population. The forward scatter/side scatter profile was used to gate out debris and large highly autofluorescent cells. A second gate was set to exclude T cells (CD3), B cells (CD19), and NK cells (CD56). A third gate was used to collect the CD11c⁺ population. Finally, two additional gates were set for cell collection of CD14⁻ cells and CD14⁺ cells.

Peripheral blood was obtained from donors unrelated to the patients undergoing thoracotomies. Signed informed consent was obtained from each donor. Approval of the protocol and consent form was granted by the University of New Mexico’s Human Research Review Committee. To obtain T cells, peripheral blood was mixed with an equal amount of 0.9% NaCl, layered over Lymphoprep (Axis-Shield), and centrifuged at 800 × g for 20 min at room temperature. The mononuclear layer was recovered, and CD4⁺ T cells were magnetically sorted to a purity of 86% or greater using the CD4⁺ T cell isolation kit II (Miltenyi Biotec). CD4⁺ T cells (3 × 10⁵ cells/well) were admixed with sorted lung cells in triplicate in 96-well flat-bottom plates at 37°C with 5% CO2 for 5 days. Eighteen hours before harvesting, 0.5 mCi of [methyl-³H]thymidine was added to each well. Controls included each cell type with medium only and T cells with medium containing PMA and PHA. Proliferation results were reported as the percentage of maximal T cell proliferation, which is defined as ((APC-induced proliferation)/(mitogen-induced proliferation)) × 100.

Fresh, unfixed lung tissue samples were randomly acquired from grossly normal lung tissue from seven different lung resections. The samples were immediately embedded in Tissue-Tek OCT compound (Sakura Finetek), snap frozen with liquid nitrogen, and stored at −80°C until sectioned. A cryostat was used to cut 6- to 7-μm serial sections with an additional slide obtained every 10th section for H&E staining to ensure the persistence of at least one airway, lung alveoli, and an intraparenchymal vessel in each of the serial sections. Sections were mounted on precleaned Superfrost Plus micro slides (VWR Scientific) and frozen until use. Sections were dried at room temperature for 10 min, fixed in cold acetone for 10 min, and washed in TBS (0.05 mol/L Tris-HCl, 0.15 mol/L NaCl, and 0.05% Tween 20 (pH 7.6)). Immunohistochemical staining was visualized using the EnVision+ System-HRP (diaminobenzidine) (DakoCytomation). When staining for CD11c, HLA-DR, CD14, and CD123, sections were treated with peroxidase block for 20 min. This procedure resulted in some low-level background staining, but because the specific staining for these Abs was so high the background was of less significance. When staining for CD1a, CD1c, CD83, and BDCA-2, the sections were first treated with 3% H₂O₂ for 5 min and then with peroxidase block (DakoCytomation) for 30 min to completely quench endogenous peroxidase and successfully reduced all nonspecific staining. We did see BDCA-2 staining on cells with characteristic DC morphology; however, the level of BDCA-2 staining was equal to background. Less stringent blocking protocols did not improve specificity. Purified monoclonal mouse anti-CD11c (KB90, IgG1) (DakoCytomation), anti-HLA-DR (G46-6, IgG2a), anti-CD123 (9F5, IgG1), anti-CD1a (HI149, IgG1), anti-CD83 (HB15e, IgG1) (BD Biosciences), anti-CD14 (B365 (BA8), IgG1) (Biomedia), anti-CD1c (M241, IgG1) (Ancell Immunology), anti-BDCA-2 (AC144, IgG1) (Miltenyi Biotec), IgG1 (X40), and IgG2a (C1.18.4) (BD Biosciences) were diluted to working concentrations in Ab diluent containing Tris-HCl buffer with stabilizing protein and 0.015 mol/L sodium azide (DakoCytomation). The isotype control Abs were diluted to the same concentration as their respective specific Ab. Sections were incubated using the EnVision+ System-HRP, and positive cells were visualized using diaminobenzidine per the manufacturer’s instructions. Sections were counterstained with Mayer’s hematoxylin (Lillie’s modification) (DakoCytomation). H&E staining was performed by TriCore Reference Laboratories using standard techniques.

A semiquantitative scale from 0+ to 3+ was arbitrary chosen: 0+, no positive cells; 1+, 1–4 positive cells; 2+, 5–9 positive cells; and 3+, >10 positive DCs. In each slide, one bronchovascular bundle, one random alveolus at ×20 magnification, and one venule (defined as vascular space lacking a muscular wall and not in a peribronchiolar connective location) were evaluated. Positive cells had to also have histological features of DCs (specifically, to show a nucleus and demonstrate interdigitating cytoplasmic processes) to be considered positive DCs. Raw counts were converted into the semiquantitative scale described above. The mean ± SEM for each marker was then calculated and tabulated. Due to specimen adequacy, not all markers were evaluated for each of the seven cases. Nonetheless, the fewest number of cases evaluated for a given marker was five.

All data are reported as the mean ± SEM. Comparisons between measured phenotype parameters between experiments were done by Student t tests. The stimulation data were analyzed using a one-way ANOVA to establish differences among groups. If there was an overall difference by Fisher’s least significant difference method, then a Student t tests was done to determine where the difference(s) were. Results with p < 0.05 were considered statistically significant.

Cells were isolated from grossly uninvolved lung tissue collected from patients undergoing partial lung resections for therapeutic or diagnostic purposes. The final diagnoses were predominantly primary or metastatic tissue, although ∼25% were from patients with granulomatous inflammation. After enzymatic digestion of the lungs, we obtained 20.0 ± 2.5 × 10⁶ cells per gram of wet lung tissue (Fig. 1). After a Percoll gradient designed to remove the granulocytes and RBCs, the remaining mononuclear cells totaled 12.8 ± 1.7 × 10⁶ cells per gram of wet lung tissue. No significant differences in either the pre-Percoll or post-Percoll cells per gram were noted when comparing lung cell numbers from resections for cancers vs granulomatous diseases (data not shown).

The predominant cells in the total lung cell population before the Percoll gradient separation were moderately large to large mononuclear cells containing monocytes, macrophages and DCs representing well over half of the total cells (Fig. 1). Small mononuclear cells (lymphocytes and NK cells) outnumbered polymorphonuclear cells, eosinophils represented a very small fraction of cells, and basophils were infrequently encountered. We have previously shown that lung cells isolated from nonperfused tissue are minimally contaminated with blood leukocytes at an average of 4% with a range of 1–8% (13). Thus, it is possible that a small number of peripheral blood cells contaminated our lung cell population, because we could not perfuse the tissue. However, the tissue was rinsed before use to reduce external blood contamination. The Percoll gradient step removed most of the polymorphonuclear cells and eosinophils, leaving predominately mononuclear cells for further study.

We next categorized the cell types in the mononuclear population based first on autofluorescence and second by using lineage-specific, accessory molecule and DC maturation markers. Mononuclear cells from each lung tissue were analyzed using four-color flow cytometry after staining with FITC-labeled anti-CD14, PerCP-labeled anti-HLA-DR, and allophycocyanin-labeled anti-CD11c to distinguish DCs from monocytes and with a number of different PE-labeled mAbs to further identify MDCs and PDCs, T cells, B cells, and NK cells. We assumed that CD14 was expressed on monocytes and most macrophages but not on MDCs or PDCs. CD11c is expressed on monocytes and MDCs but not on PDCs. HLA-DR is constitutively expressed on monocytes, macrophages, MDCs, PDCs, and B cells as well as on activated T cells and NK cells. Cells that were HLA-DR⁻ were considered neither DCs nor monocytes.

Forward and side scatter analysis of unstained mononuclear cells consistently revealed three distinct populations as illustrated in two representative lung samples in Fig. 2, A and D. The first population (R1) comprised 42% of total mononuclear cells that were of small to moderately large size and of low to moderate complexity. The second population (R2) comprised 49% of total mononuclear cells that were of moderately large to large size and of high complexity. The unlabeled third population represented 9% of the mononuclear cells that were of small size and low complexity and were RBCs and dead or dying cells. The cells in the R1 gate were of relatively low autofluorescence in the FITC and allophycocyanin channels (Fig. 2,B) and of low to moderate autofluorescence in the PerCP and PE channels (Fig. 2,C), whereas cells in R2 were virtually autofluorescent in all four channels (Fig. 2, C and F). Thus, we define the R1 cells as low autofluorescent (LAF) cells. Others have shown that the high autofluorescent (HAF) cells isolated from lung tissue correspond to large, inclusion-rich macrophages, likely alveolar macrophages, and those cells with functional characteristics of lung DCs are enriched primarily in the low autofluorescent cells (11, 14). The moderately large to large size HAF cells could not be reliably analyzed for surface markers. Thus, in all subsequent analyses the moderately large to large size, highly complex HAF cells were excluded from analysis by positive gating of the R1 population on the forward and side scatter plots.

The LAF cells were grouped based on their expression of CD11c and CD14, assuming that the CD11c⁺CD14⁺ and CD11c⁻CD14⁺ cell populations contained monocytes and macrophages; the CD11c⁺CD14⁻ population was enriched in MDCs and the CD11c⁻CD14⁻ population was enriched in PDCs. In each of the four groups the cells were further analyzed for the expression of CD3, CD19, and CD56 to distinguish T, B, and NK cells, respectively, and to classify each cell type to a LAF subset (Table I). As expected, T cells were found predominantly in the CD11c⁻CD14⁻ group that comprised 40% of the LAF population. Cells expressing CD56 were also found largely in the CD11c⁻CD14⁻ group and in lower percentages in each of the three other groups. B cells were an infrequent population but were identified in the CD11c⁺CD14⁺ and CD11c⁻CD14⁻ groups. Back gating showed that T, B, and NK cells prevailed in the small cell population of LAF and that monocytes, macrophages, and DCs prevailed in the moderately large cell population. If the HLA-DR⁺ cells in the two CD14⁻ groups were considered enriched in MDCs and PDCs and the HLA-DR⁺ NK cells were excluded, then ∼7% of the CD11c⁺CD14⁻ LAF cells might be MDCs and 7% of the CD11⁻CD14⁻ cells might be PDCs.

In each of the four groups the cells were further distinguished by HLA-DR expression and by markers considered highly characteristic for MDCs (i.e., CD1a and CD1c), PDCs (i.e., BDCA-2 and CD123), and mature DCs (i.e., CD83). Cells expressing CD1a or CD1c were most prevalent in the CD11c⁺CD14⁻HLA-DR⁺ MDC-enriched population (See Figs. 3 and 4). However, CD1a⁺ and CD1c⁺ cells were also present in the CD11c⁺CD14⁺HLA-DR⁺ monocyte-enriched subpopulation, although the mean fluorescence intensity (MFI) of CD1a and CD1c was somewhat less in this population than in the CD11c⁺CD14⁻ population.

In human blood, BDCA-2 is highly specific for PDCs, whereas CD123 is not only present on PDCs but on other blood cells including monocytes (12, 15, 16). However, BDCA-2 is down-regulated as PDCs mature; thus, this marker would likely underestimate significantly the total number of PDCs present in the lung (12). Cells expressing CD123 were present in both the CD11c⁺CD14⁻HLA-DR⁺ and CD11c⁺CD14⁺HLA-DR⁺ subpopulations with a smaller percentage present in the CD11c⁻CD14⁻HLA-DR⁺ subset (see Figs. 3 and 4). In the studies performed in which CD123 was assessed, one of the 13 lung samples demonstrated no CD123⁺ cells in the CD11c⁻CD14⁻HLA-DR⁺ populations. BDCA-2 expression was not convincingly evident in the CD11c⁺CD14⁻HLA-DR⁺ subpopulation but was clearly present in the CD11c⁻CD14⁻HLA-DR⁺ subpopulation as expected, but in only nine of the 13 samples, with BDCA-2 absent in the case where CD123⁺ cells were also absent. In general, the percentage of BDCA-2⁺ cells in the PDC-enriched lung population tended to be less than the percentage of CD123⁺ cells. Unexpectedly, BDCA-2 expression was also seen in the CD11c⁺CD14⁺HLA-DR⁺ subpopulation, raising concerns about the specificity of this marker in lung cell populations.

Mature DCs, as defined by expression of CD83, were present within the MDC, PDC, and monocyte-enriched lung cell subpopulations, although the MFI of CD83 staining was quite low in all subsets. Cells expressing CD83 were present in both the HLA-DR⁺ and HLA-DR⁻ subgroups of the PDC-enriched CD11c⁻CD14⁻ cells (Fig. 4 and data not shown for HLA-DR), which raises the question as to the specificity of CD83 for mature DCs in the lung.

As discussed earlier, CD123 was present on a small percentage of CD11c⁻CD14⁻HLA-DR⁺ lung cells. To demonstrate that CD123⁺ cells within this group were PDCs with the functional characteristic of secreting IFN-α in response to a TLR-7 ligand and to compare this group to CD11c⁺CD14⁻HLA-DR⁺ cells enriched in MDCs, we analyzed the capacity of flow-sorted low autofluorescent lineage negative CD11c⁻HLA-DR⁺CD123⁺ (denoted PDC in Fig. 5) and CD11c⁺HLA-DR⁺ (denoted MDC in Fig. 5) lung cells to respond to influenza A or, as a control, the TLR-4 ligand LPS. In two replicate experiments (one of which is shown in Fig. 5), PDC-enriched lung cells secreted IFN-α in response to influenza stimulation and responded poorly to LPS for all cytokines and chemokines measured, thus supporting the presence of PDCs within this population. In addition, influenza induced PDCs to increase secretion of TNF-α in both experiments and MIP-1α in one experiment. In both experiments, stimulation of MDC-enriched lung cells with LPS induced an increase in IL-12p40, TNF-α, MIP-1α, IL-6, and IL-10 and did not respond to influenza A for all cytokines and chemokines measured. Both MDCs and PDCs spontaneously secreted IL-8.

To determine whether the expression of CD123 on MDCs (low autofluorescent lineage negative CD11c⁺HLA-DR⁺ lung cells) delineates a subset with functional distinctiveness, we examined the cytokine and chemokine secretion profile of CD123⁺ or CD123⁻ MDCs in response to influenza or LPS stimulation (Fig. 6). In two of three experiments, CD123⁺ and CD123⁻ lung cells respond to LPS in with an increase secretion of IL-12p40, TNF-α, MIP-1α, IL-6 and IL-10, with the CD123⁺ population trending toward higher levels of each analyte except for IL-10. A third experiment showed limited responsiveness to these analytes with only an increase in IL-10 for CD123⁺ MDCs and IL-12p40 for both populations, with CD123⁺ MDCs showing the higher level. All three experiments showed that neither the CD123⁺ nor the CD123⁻ lung cells responded to influenza. Both CD123⁺ and CD123⁻ MDCs spontaneously secreted IL-8.

HLA-DR⁺ cells in the CD11c⁻CD14⁻, CD11c⁺CD14⁻ and CD11c⁺CD14⁺ groups were further studied for expression of the costimulatory markers, CD80, CD86, or CD40 (Fig. 7). A small percentage of cells expressing low levels of CD80 were seen in both the CD11c⁺CD14⁻HLA-DR⁺ and CD11c⁺CD14⁺HLA-DR⁺ subgroups, with little to no expression in the CD11c⁻CD14⁻HLA-DR⁺ and CD11c⁻CD14⁺HLA-DR⁺ subgroups. CD86 showed the highest expression and was prevalent in both the CD11c⁺CD14⁻HLA-DR⁺ and CD11c⁺CD14⁺HLA-DR⁺ subgroups, with the monocyte-enriched subgroup showing a slightly higher percentage of cell expression and no significant difference in MFI. Little to no expression of CD86 was seen on CD11c⁻CD14⁺HLA-DR⁺ and CD11c⁻CD14⁻HLA-DR⁺ subgroups. A very small percentage of LAF cells expressed CD40, with the majority of those cells found in the CD11c⁺CD14⁺HLA-DR⁺ monocyte-enriched subgroup and a lesser percentage being CD11c⁺CD14⁻HLA-DR⁺. These findings show that MDCs and monocytes express costimulatory molecules at levels significantly higher than PDCs. Indeed, lung PDCs have little to no expression of these three important costimulatory molecules.

As discussed above, a small and similar number of lung cells in both the MDC and monocyte-enriched lung cell subpopulations expressed costimulatory molecules important in stimulating naive T cell proliferation. Both groups also expressed HLA-DR at similar levels. We have previously reported that human alveolar macrophages, rich in HLA-DR, and human blood monocytes are both weak APCs in a MLR as compared with lung cell populations enriched in DCs (17). To demonstrate that the CD11c⁺CD14⁻ subpopulation was enriched in MDC with an enhanced capacity to stimulate naive T cells, we compared their capacity to stimulate a MLR with that of CD11c⁺CD14⁺ cells. Lung cells were sorted for the two groups using the following criteria: 1) large autofluorescent cells were excluded; 2) CD3⁺, CD19⁺, and CD56⁺ cells were excluded; and 3) all remaining CD11c⁺ were sorted as either CD14⁺ or CD14⁻. Each of these two groups were then individually admixed in graded doses with allogeneic CD4⁺ peripheral blood T cells and cultured as described. Both lung cell populations stimulated the T cells to proliferate, but the MDC-enriched subset was significantly more efficient (Fig. 8).

In the human lung, HLA-DR⁺ dendritic-shaped cells have been noted to reside within the bronchial epithelium, alveolar septa, visceral pleura, and in vascular walls (18, 19, 20). We were particularly interested in whether both MDCs and PDCs were present at the interface of the airway epithelium with the external environment. To delineate the location of MDCs and PDCs within the lung, we used single color immunohistochemistry and labeled serial sections with HLA-DR, CD11c, CD14, CD123, BDCA-2, CD1a, CD1c, and CD83 mAbs. Only cells with the histological feature of dendritic cells (e.g., having a nucleus and interdigitating cytoplasmic processes) were evaluated (Table II). The data presented in Table II represent the location of these cells relative to airways, septa, and perivenular connective tissue and do not reflect a density (i.e., number of cells per area of airway or vessel). BDCA-2 positive staining could not be distinguished from background, although this stain was strongly positive using human tonsil, and thus no reliable location data was determined for this PDC marker. Cells expressing HLA-DR, CD11c, or CD14 were observed at both the airway (Fig. 9, A, B, and C, respectively) and in parenchymal locations, with CD11c cells (1.8 ± 0.3, mean ± SEM) outnumbering CD14 cells (0.7 ± 0.2) within the epithelium of the airway (p = 0.0107) (Table II). Cells expressing CD123 were present in the airway (Fig. 9, D and G) and parenchyma (Fig. 9 J). One problem with the use of CD123 as a single immunohistochemistry marker for PDCs is that CD123 is also expressed by the monocyte- and MDC-enriched populations.

Based on our flow analysis, ∼54% of CD123⁺ LAF cells expressed CD14, but how many CD123⁺CD14⁺ cells have a DC morphology in tissue cannot be determined. Thus, according to flow cytometry and pattern of cytokine and chemokine secretion, PDCs are present in the human lung but their exact location cannot be determined by single color immunohistochemistry.

CD1a⁺ and CD1c⁺ cells were seen in the airway (Fig. 9,D) and the alveoli and very infrequently in perivascular tissue in the parenchyma. The location of CD1a⁺ cells paralleled CD1c⁺ cells, with CD1a⁺ cells being significantly less numerous than CD1c⁺ cells in the intraepithelium, subepithelium, and alveoli (Table II). One limitation of our single marker immunohistochemical staining is that we could not determine the coexpression of CD1a and CD1c. However, overall, the expression of CD1c was 3.8 times greater than CD1a, indicating that the majority of CD1c MDC did not coexpress CD1a. CD83⁺ DCs were infrequently observed, suggesting that the majority of DCs in the lung are immature.

Our examination of human lung tissue obtained from surgical resections resulted in five major observations. First, based on the identification of MDCs as lineage-negative CD11c⁺HLA-DR⁺ and either CD1a⁺ or CD1c⁺ cells and PDCs as lineage negative CD11c⁻HLA-DR⁺ and CD123⁺ cells, both DC subsets were confirmed to be present in the human lung. Notably, MDCs were about twice as numerous as PDCs. Second, when stimulated through TLR-7, only PDCs secreted IFN-α while MDCs (both CD123⁺ and CD123⁻), but not PDCs, responded to TLR-4 stimulation with the secretion of IL-6, IL-10, IL-12 p40, and MIP-1α, confirming that the two subsets respond to distinct TLR receptors as has been described when they were isolated from blood and other tissue sites (21, 22). Third, MDC-enriched lung cell populations, as well as monocyte-enriched lung populations, expressed higher levels of the costimulatory molecules CD80, CD86, and CD40 than PDC-enriched lung cell populations. Fourth, despite equivalent costimulatory molecules and HLA-DR expression, lung cells enriched in MDCs was more efficient than the monocyte-enriched lung cells in stimulating naive T cell proliferation. Finally, we identified dendritic-shaped cells compatible with both MDCs and PDCs in the airway epithelium. MDCs expressing CD1c were more numerous than MDCs expressing CD1a, and DCs with these markers were present within the airway epithelium, validating the findings of others that MDCs are available at the lung-air interface (23). CD123⁺ dendritic-shaped cells were present in the epithelium and subepithelial regions of the airways, but because this marker is so widely distributed on both MDCs and monocytes, additional studies are needed to clarify the intrapulmonary location of these cells.

The lung is recognized as an immunologically competent organ generally responding to innocuous aerogenous Ags with tolerance and responding to infectious insults with the appropriate cellular or humoral immune response (1, 24, 25). Failures in immune regulation occur, however, with the development of hypersensitivity disease and the loss of tolerance. Lung DCs have been shown to regulate these responses, but relatively little has been done to determine the roles of DC subsets in human lung immune regulation (11, 24, 25). Important initial studies require identifying the numbers and sites for the two major DC subsets in humans, MDCs and PDCs, each of which exhibit distinct phenotypic and functional properties (9, 10). MDCs from the blood preferentially express TLR-2, TLR-3, TLR-4, or TLR-5; and during bacterial and some viral infections produce IL-12, which promotes Th1 and CTL adaptive immune responses (21). PDCs from the blood express TLR-7 and TLR-9 and, upon viral infection, rapidly produce large amounts of type 1 IFNs, which have a direct inhibitory effect on viral replication and activate NK cells, B cells, T cells, and MDCs, leading to the amplification of innate immune responses and the induction of adaptive immune responses (26, 27). Both MDCs and PDCs are also able to tolerize or prime a Th2 response, depending upon the context in which they see Ag (28, 29, 30). To our knowledge, only a single other study has sought to identify PDCs in human lung together with MDCs (11). This well-done study used mAbs against BDCA-1 to -4 Ags as the primary measure for DC subsets. We, in contrast, used a standard strategy for separating blood MDCs and PDCs from other blood leukocytes while focusing, as did Demedts et al. (11), on the LAF of mononuclear cells. Thus, CD11c, CD14, and HLA-DR were used to identify four major leukocytic cell subgroups, and then the percentage of T cells, B cells, NK cells, and cells with known MDC and PDC markers were determined. BDCA-2 was used as a unique PDC marker but, as discussed, we believe that this marker underestimates the total number of PDCs. We did not use BDCA-3 and BDCA-4, because they can appear on both PDC and MDCs. BDCA-1 recognizes CD1c, which we have also used here. We found that the MDC subset was present in all lungs studied and PDC in 12 of 13 of those lungs studied in which both the CD123 and BDCA-2 markers were assessed. This finding corresponds to that of Demetdts et al. (11). We, like they (11), found that MDC were more frequent than PDCs.

Cells with specific markers for DCs, i.e., CD1a and CD1c for MDCs and BDCA-2 for PDCs, were identified in small subpopulations of human lung mononuclear cells. MDCs predominated in the CD11c⁺CD14⁻HLA-DR⁺ mononuclear subset and PDCs in the CD11c⁻CD14⁻HLA-DR⁺ mononuclear subset. A very low number of possible MDC precursors were seen in the CD11c⁺ CD14⁺HLA-DR⁺ mononuclear subset based on the presence of CD1a and CD1c. Whether these cells are MDC precursors or MDCs that have retained the monocyte/macrophage marker, CD14, which may impart functional uniqueness, needs to be clarified. A recent report by Demedts et al. (11) noted that MDCs have a heterogeneous expression of CD14 and that PDCs do not express CD11c and CD14. The overall number of MDCs noted in the lung was significantly higher than the number of PDCs. This may be an accurate assessment or due to phenotypic changes that PDCs undergo after maturation. Within the PDC-enriched cell population, the expression of CD83 exceeded BDCA-2, indicating that mature non-BDCA-2-expressing PDCs may actually be present. Currently, no reagents are available that uniquely allow the identification of mature PDCs. Our flow data suggest that CD123, unlike BDCA-2, is not lost upon maturation. Thus, the use of CD123 (together with CD83 as a marker for mature PDCs) within the CD11c⁻CD14⁻HLA-DR⁺ population may be the most inclusive definition of lung PDCs. Overall, the presence of both MDCs and PDCs infers that the lung is equipped with APCs that can recognize and respond to a spectrum of infectious threats. The cytokine secretion data derived from sorting cells in the lungs based on lineage negativity, CD11c⁺HLA-DR⁺ cells for MDCs, and lineage-negative CD11c⁻HLA-DR⁺CD123⁺ cells for PDCs indicate that lung MDCs and PDCs are differentially responsive to TLR-4 and TLR-7 ligation, respectively, and that our phenotypic distinctions are valid.

The activation of naive T cells by DCs depends not only on Ag/TCR interactions but also on the relative density and function of accessory molecules (31). We asked whether MDCs and PDCs expressed similar levels of accessory molecules that are known to interact with counter-receptors on T cells to enhance costimulation. A low expression of CD40 by MDCs and a low to no expression by PDCs suggest that both lineages, especially PDCs, are limited in interacting with CD40L expressed on CD4⁺ T cells. MDCs and CD11c⁺CD14⁺HLA-DR⁺ cells (cell population containing monocytes and macrophages) expressed higher levels of CD80 and CD86, suggesting that these APCs are poised to initiate T cell stimulation and adaptive immune responses. The low level of CD80 and CD86 expression on PDCs could hinder the ability of PDCs to initiate an immune response unless stimulated by interaction with TLRs. A mouse study by de Heer et al. (32) supports a role of lung PDCs in tolerance to inert inhaled Ags. Future studies will determine whether human lung PDCs are actually tolerogenic in the absence of TLR-7 and/or TLR-9 engagement.

Our present data show that a mononuclear cell population enriched in MDCs was more effective at stimulating naive allogeneic T cell proliferation than a mononuclear cell population enriched in monocytes and macrophages. The enhanced accessory property of the MDC-enriched population cannot be explained based upon HLA-DR, CD40, CD80, CD86, or CD83 differences. Specifically, both experimental groups showed similar expression of these molecules, both in the percentage of cells within the enriched populations expressing the markers and their mean fluorescence intensity. Currently, we can only speculate on the mechanism(s) underlying the immunostimulatory difference between the two groups. Indeed, Demedts, et al. (11) suggest that CD14 is present on some lung MDCs. One hypothesis is that, in addition to CD1a⁺ and CD1c⁺ cells, other cells in the MDC-enriched population may contribute to the observed immunostimulatory function through direct or indirect augmentation. A second hypothesis is that cells in the monocyte/ macrophage-enriched, but not the MDC-enriched, population express inhibitory surface molecules that engage receptors on T cells to inhibit T cell activation. A third hypothesis is that a disparate production of stimulating or inhibiting cytokines or other inhibitory molecules by these two groups underlies their ability to stimulate T cell proliferation. Still, these findings are consistent with numerous studies in the past showing that peripheral blood monocytes are much less stimulatory of naive T cells than DCs. In view of the capacity of monocytes to become fully competent DCs after culture in GM-CSF and IL-4 followed by activation (33), in an inflamed lung typical monocytes could become effective Ag-presenting MDCs.

We addressed the location of CD1a⁺ or CD1c⁺ DCs in the human lung. CD1a⁺ DCs were reported to localize within the epithelial layer of the airways, whereas CD1c⁺ DCs were almost exclusively in the submucosa. Cells expressing both CD1a and CD1c were noted at each location (34). We found CD1a⁺ DCs in all layers of the airways and, although sparse, in the parenchyma. CD1c⁺ staining demonstrated a similar distribution pattern, but by flow cytometry the overall number of DCs expressing CD1c was 3.8 times higher than those expressing CD1a. In the current study we did not distinguish whether CD1a was coexpressed with CD1c, but others have shown that these markers are coexpressed on some, but not all, MDCs. CD1 molecules are homologous to MHC class I molecules, i.e., glycoproteins composed of a H chain noncovalently linked to β-2 microglobulin (35). Unlike MHC class I, however, CD1 molecules bind lipids. CD1 molecules have cytoplasmic tails that target them to distinct endocytic compartments where they can acquire lipid Ag; therefore, microbial Ags presented by CD1a and CD1c to T cell receptors can be acquired in different endocytic compartments and might play a role in initiating different types of immune responses (36, 37). Recent studies suggest that the Ag-binding grooves of these two CD1 isoforms might inherently demonstrate ligand preferences that can influence the role of the DC expressing one or both of the CD1 molecules in immune responses (38). In any case, the presence of both CD1a⁺ and CD1c⁺ MDCs within the airway wall corresponds with their function as sentinel cells poised to respond to and translate microbial threats to cells of the innate and adaptive immune response.

The possible presence of PDCs as identified by a dendritic-shaped CD123⁺ cell at the interface of the airway and lung interstitium is important in view of their function as an early inhibitor of viral replication through their ability to secrete copious amounts of IFN-α or -β in response to viral challenge (26, 27). Cells consistent with PDCs were also seen infrequently in the parenchyma. Unfortunately, limitations exist in using single stain immunohistochemistry and CD123, because we found CD123 expressed on cells consistent with MDCs as well as on cells in the monocyte-enriched lung cell population. Indeed, CD123 has been reported to be expressed by some monocytes and macrophages (15). Monocytes and macrophages should not have been included in our analysis because we evaluated only lung cells with histological feature of DCs, but it is not known how often monocytes and tissue macrophages can assume a dendritic shape in tissues. Further studies using multicolor confocal techniques and additional macrophage-specific (CD14 and CD163) and DC-specific (CD1a, CD1c, CD83, BDCA-2, fascin, and DC-LAMP) markers in combination with CD123 should help distinguish MDCs and PDCs from one another, identify more precisely their anatomic locations within the lung, and provide information on their state of maturation.

In summary, both MDCs and PDCs were identified in human lung. Although we have drawn our conclusions from grossly normal lung tissue from diseased individuals, initial studies with normal lung tissue from an unused lung removed for lung transplantation show similar results. These data provide the groundwork to address the issue of how each DC subset contributes to immune responses in the lungs of normal and infected individuals and those undergoing therapeutic interventions. The goal would be to eventually exploit lung DCs in the development of protocols for DC-based vaccination studies for the prevention and treatment of pulmonary infections, hypersensitivity disease, and malignancies.

We acknowledge the helpful suggestions of trainees and peers, especially Drs. Rick Lyons and Julie Wilder, and the technical support of Charlene Hensler and Connie Peceny as well as Mark Carter, University of New Mexico Flow Cytometry Core.

The authors have no financial conflict of interest.