Analysis of proteomic profiles and functional properties of human peripheral blood myeloid dendritic cells, monocyte-derived dendritic cells and the dendritic cell-like KG-1 cells reveals distinct characteristics

Important proteomic and functional differences between peripheral blood myeloid dendritic cells, monocyte-derived dendritic cells (moDC) and KG-1 cells have been identified.


Background
Dendritic cells (DCs) are highly specialized antigen presenting cells that originate from bone marrow progenitor cells. They represent a major cellular component of the innate immune system and their interaction with cells of the adaptive immune system (for example, T cells) is critical for initiating immune responses and maintaining tolerance [1]. DCs exist in two stages of maturation. Immature cells are found throughout the body where they act as sentinels, continuously taking up antigen and undergoing activation [2]. Activation leads to the secretion of pro-inflammatory cytokines, resulting in up-regulation of co-stimulatory molecules and migration to the lymph nodes. During their maturation, DCs lose their antigen-capturing capacity and become mature immuno-stimulatory cells that have the ability to activate naïve T cells.
There are two main DC types in human peripheral blood, known as myeloid DCs (mDCs) and plasmacytoid DCs (pDCs). mDCs are the major subset, representing around 80% of blood DCs [3]. For ex vivo studies, mDCs can be isolated from peripheral blood using immunomagnetic cell separation [3]. However, the main obstacle here is that DCs represent only around 1% to 3% of peripheral blood mononuclear cells (PBMCs). This has, therefore, prompted researchers to use other model systems for studying mDC biology. For instance, DCs can be generated in vitro from peripheral blood monocytes by culturing them for six days in the presence of interleukin (IL)-4 and granulocyte-macrophage colony stimulating factor (GM-CSF). Under such culture conditions, cells acquire an immature DC morphology and express DC differentiation antigens [4]. These monocyte-derived DCs (moDCs) are routinely used as an mDC model in DC research.
Several human monocytic cell lines are also available, including U937, THP-1, MUTZ-3, HL-60, KG1 and MM6, and some of these have been shown to be able to differentiate into DClike cells [5][6][7][8][9]. KG1 cells, which acquire a DC-like phenotype after stimulation with phorbol 12-myristate 13-acetate (PMA) and ionomycin [6], are probably the most widely used in DC research. PMA-and ionomycin-stimulated KG1 cells show typical DC morphology and become adherent with long neurite processes. They also show up-regulation of major histocompatibility (MHC) class I and II molecules, co-stimulatory molecules and DC-specific markers [6]. Furthermore, they are able to stimulate allogeneic T cell proliferation at levels similar to PBMC-derived DCs [6]. It has also been shown that KG1 cells are able to cross-present exogenous antigen to CD8+ T cells and display similar regulation of MHC class II trafficking to DCs [8]. Therefore, KG1 cells are considered to be a good model system to study human DC biology.
Despite the extensive use of both moDC and KG1 cells as mDC models, their similarity to peripheral blood DCs is yet to be properly defined. This study aims to assess the suitability of both moDC and KG1 cells as model cells for peripheral blood DCs by comparing their proteomes in relation to their surface phenotypes, cytokine profiles and T cell activation ability.

Results and discussion
DCs are sentinels of the immune system and play a pivotal role in bridging innate immunity with the adaptive immune response. Given the scarcity of peripheral blood DCs and the ethical and technical difficulties involved in obtaining tissuederived DCs from human sources, investigators have resorted to using different model systems for studying DC biology. Although moDC and KG-1 cells are routinely used as mDC models [8,10,11], a thorough comparison of these cells has not yet been carried out. A number of phenotypic and functional comparisons have previously been made between mDCs and moDCs [12] and moDCs and KG-1 cells [13], but no studies have compared the proteomes of all three cell types. In this study we compare the proteomes of mDCs, moDCs and KG-1 cells, and then attempt to relate this to the functional properties of these cells. Figure 1 shows the workflow and the way in which each cell population was generated or separated.

Dendritic cell proteomes
Proteomic data are scarce in relation to DC biology, and where available they only focus on moDCs [14][15][16]. Others have focused on gene expression, as well as obtaining some proteomic data, in monocytes and moDCs [17][18][19]. The present study compared the whole cell proteome of immature mDCs, moDCs and KG-1 cells. Clearly, a major challenge in proteomic studies of DCs is obtaining enough protein for performing two-dimensional electrophoresis. This limitation was partly overcome by using a large volume of blood (approximately 120 ml) for cell separation. We also pooled whole cell lysates of DCs from seven individuals to obtain sufficient quantities of protein and to eliminate inter-individual variations. We found that peripheral blood mDCs have sixand five-fold lower protein content per cell than moDCs and KG-1 cells, respectively (data not shown). Unfortunately, the low numbers of mDCs in peripheral blood (approximately 1% of PBMCs), together with their lower protein content, meant that, despite pooling samples, we were able to run only duplicate gels for mDCs. Figures 2 and 3 show three representative two-dimensional gel images of the different cell types. Gel images were analyzed using PDQuest software and all images were normalized before any comparisons between gels were made. The total number of spots in the gels were 661, 619, and 770 for mDCs, moDCs and KG-1 cells, respectively. To analyze the comparability of gels, the densities of spots matched in all three gels were plotted and a correlation coefficient value was calculated. The proteome of mDCs showed different levels of similarity compared with those of moDCs and KG-1 cells (correlation coefficient 0.68 and 0.62, respectively) ( Figure  4). Duplicate gels of mDCs were reproducible (correlation coefficient >0.90), as were triplicate gels of moDC and KG-1 Thirty-five spots were up-regulated more than four-fold in mDCs compared with the DC models, and fifty were downregulated by the same amount (Table 1). A number of differentially expressed proteins, which appeared to be more than two-fold different in intensity (either up-or down-regulated) in the DC models compared to mDCs, were excised from the gels and subjected to trypsin digestion and MALDI-TOF (matrix-assisted laser desorption/ionisation-time of flight) mass spectrometric analysis; Table 2 shows the corresponding protein data. The factor of difference was calculated by dividing the intensity of the protein spot in mDCs by that of the corresponding spot in moDCs or KG-1 cells. Eighteen spots were successfully identified using MALDI-TOF mass spectrometry. These proteins are known to be involved in a wide spectrum of biological processes, including functions related to cell integrity and metabolism ( Table 2).
The majority of the proteins that showed higher levels of expression in mDCs are known to be involved in cell growth and maintenance, including FGG, ubiquinol cytochrome c reductase, glutathione S transferase, nuclease isoform sm3 and annexin A1. Some of these differentially expressed proteins also appear to be involved in DC maturation. Pereira et al. [15] have shown higher expression of FGG in the proteome of immature moDCs compared to mature moDCs. Furthermore, fascin and actin, which showed substantially lower expression (8-and 16-fold, respectively) in mDCs compared with both DC models, are known to play important roles in maintaining cell structure and in the formation of immunological synapses between DCs and T cells [20][21][22]. Al-Alwan et al. [21] have previously shown that increased fascin expression correlates with DC maturation state, and recent work supports this, suggesting that fascin is a mature DC marker [23]. This, together with our data on FGG expression, suggests that, at least in their resting state, mDCs have a less mature phenotype compared to moDCs and KG-1 cells.
Cell culture work-flow Figure 1 Cell culture work-flow. Overview of the methods used for isolation/generation of mDCs, moDCs and DC-like KG-1 cells.

Dendritic cell lysate ELISA
To confirm the proteomics data, we used a capture ELISA to assess the relative expression of five proteins that had an induction factor greater than two, namely actin related protein 2/3 complex 2 (ARPC2), phosphoglucomutase 1 (PGM1), fascin, FGG and carbonic anhydrase 2 (CAH2) ( Table 2). The pattern obtained was in general agreement with the proteomes obtained for each of the three cell types. Thus, as expected, ARPC2, PGM1, fascin and CAH2 were found to be lower in mDCs compared to the two models, whereas FGG was higher ( Figure 6).

Cell surface marker expression
We compared the three cell types by studying their surface phenotypes. Immature cells were cultured in the presence of lipopolysaccharide (LPS) for 24 h to produce a mature cell type. The cell markers used for characterization were CD11c, CD40, CD62L, CD80, CD83, CD86, CD206, CD209, HLA DR and Toll-like receptor (TLR)-4, which have all been reported to be found on dendritic cells [4,24].
As with our proteomic data, cell surface marker expression suggested that immature mDCs expressed lower levels of the usual DC maturation markers compared with both moDCs and KG-1 cells. The mDC models, moDcs and KG-1 cells, expressed significantly higher levels of CD11c, CD40, CD80, CD83 and CD209 than mDCs ( Figure 7). However, mDCs showed significant up-regulation of the classic DC maturation markers CD40, CD80, CD83 and CD86 after 24 h stimulation with LPS; levels of these markers were more than ten-fold higher in mature compared to immature cells. The mDC models also showed up-regulation of these markers, but to a lesser extent (more than three-fold). The expression of cell surface markers on mature KG-1 cells was lower than on both mDCs and moDCs, with CD11c, CD40, CD80 and CD86 being expressed at significantly (p < 0.05) lower levels than on mDCs ( Figure 7).
Myeloid DCs showed a more mature phenotype after stimulation with LPS (as shown by higher expression of CD40, CD80, CD83 and CD86) compared with moDCs and KG-1 cells. Interestingly, the mannose receptor (CD206), which has important functions in endocytosis, antigen recognition and  Two-dimensional electrophoresis gels. Three representative twodimensional gel images of (a) mDCs, (b) moDCs and (c) KG1 cells. Whole cell lysate protein (30 μg) was applied to immobilized pH gradient strips (pH 5-8), subjected to isoelectric focusing and separated on 10% to 20% polyacrylamide gel before silver staining. Images were analyzed using PDQuest and normalized by total quantity in valid spots. Highlighted spots were excised and protein identifications attempted using MALDI-TOF mass spectrometry. Boxed areas are shown in detail in Figure 3. Further gel information and protein identifications are shown in Table 2 binding and MHC class II presentation [25][26][27], was only detectable on moDCs and was down-regulated by 4-fold after stimulation with LPS for 24 h; only negligible levels were found on mDCs and KG-1 cells. This would, therefore, suggest that in in vitro assays, moDCs could bind and internalize certain antigens, particularly glycoproteins, more efficiently. These findings are in keeping with those of Hajas et al. [13] showing that moDCs express much higher levels of CD206 than KG-1 cells and they could internalize antigens relatively more efficiently. The expression of DC-SIGN (DC-specific intercellular adhesion molecule-3-grabbing non-integrin or CD209) was low on all three cell types, but significantly higher on immature moDCs and KG-1 cells compared with mDCs.
Our finding of negligible levels of TLR4 on all three cell types is somewhat different from those of others [28,29] who found no expression of TLR4 on mDCs, but did show expression on moDCs. However, there are studies showing TLR4 expression by both mDCs and moDCs, but not on pDCs [30]. This discrepancy in data could have been caused by the use of different monoclonal antibodies and experimental conditions.

Cytokine expression profile
Peripheral blood mDCs were found to express significantly higher levels of key inflammatory (IL-1β, IL-6 and IL-8) and regulatory (IL-10) cytokines, compared to moDCs and KG-1 cells. Levels of IL-1β, IL-6, IL-8 and IL-10 were dose dependent, and following 24 h culture with either 50 or 100 pg/ml LPS were significantly higher in mDCs than in moDC and KG-1 cells (Figure 8). The IL-6, IL-10 and IL-12 data are at variance with a previous study [12], but this may be due to the use of different stimuli (for example, intact Escherichia coli rather than LPS), culture conditions and cytokine detection method by the authors. This pattern of cytokine production clearly makes mature mDCs more efficient in the cross-talk with T cells [31,32] and other cells of the innate immune system (for example, natural killer cells), as well as in exerting Detailed view of two-dimensional gels Figure 3 Detailed view of two-dimensional gels. Detailed areas of the mDCs, moDCs and KG-1 gels. The areas correspond to boxed areas in Figure 2.
inflammatory and/or regulatory effects mediated through cytokine production. This again is in line with our proteomic data suggesting that mDCs have a less mature phenotype, at least in their resting state, compared to the two DC models, moDCs and KG-1 cells [33].

Functional analysis
Having demonstrated that unstimulated moDCs have a more mature phenotype than freshly isolated mDCs, as shown by proteomics (for example, lower FGG and higher fascin) and surface marker expression (higher CD83), we then proceeded to assess the endocytic and T cell stimulatory abilities of the DCs using dextran uptake and autologous mixed leukocyte reaction, respectively. moDCs were found to be better in endocytosis ( Figure 9) and T cell activation ( Figure 10) compared to mDCs, and this is in keeping with their more advanced maturation status. Others have shown [12] that, upon stimulation, both mDCs and moDCs are equally efficient in autologous T cell activation, which is in agreement with our finding that mDCs acquire a fully mature phenotype after LPS stimulation (Figure 7).

Conclusion
Despite the general similarities between mDCs and the two DC model systems, moDCs and KG-1 cells, our findings identified important differences between the proteomes of these cells, and the findings were confirmed by ELISA detection of a selection of proteins. These differences were particularly noticeable with proteins involved in cell growth and maintenance, as well as those involved in cell-cell interaction, cell integrity and maturation. The scarcity of peripheral DCs meant that we were not able to focus on less abundant proteins in the current study, which could identify differentially expressed proteins involved in other cell functions. The functional relevance of differentially expressed proteins was confirmed by analysis of surface marker expression, cytokine profile, endocytic and T cell activation abilities of the cells, again suggesting differences in the maturation status between mDCs and the DC models. These observations have important functional implications, particularly in relation to DC-T cell interactions, the so-called immunological synapse, and, therefore, need to be considered when interpreting data obtained from model DC systems. This study clearly shows the value of the proteomic approach as a tool for studying the biology of immune cells.

Cell cultures and stimulation
Heparinized whole blood from healthy volunteers (obtained with prior consent and Ethical Committee approval) was used for separation of PBMCs on a Histopaque density gradient (HISTOPAQUE-1077, Sigma, Poole, UK). CD1c + peripheral DCs were isolated using the CD1c dendritic cell isolation kit from Miltenyi Biotech (Bisley, UK). Briefly, this involved depleting PBMCs of CD19 + B cells followed by positive selection of CD1c + cells. CD14 + monocytes were isolated by positive selection from the CD1ccell fraction, and immature CD1a + CD83 -moDCs were generated as previously described [24]. Briefly, this involved culturing CD14 + monocytes in the presence of IL-4 (250 IU/ml; R&D systems, Oxford, UK) and GM-CSF (50 ng/ml; R&D systems) for six days. Cells were cultured at 1 × 10 6 /ml in RPMI 1640 medium (Sigma) supplemented with 2 mM L-glutamine, 100 U/ml penicillin, 100 U/ ml streptomycin (Gibco Life Technologies, Paisley, UK) and 10% (v/v) fetal calf serum (FCS; Harlan Sera-Lab,  Loughborough, UK) at 37°C in 5% CO 2 . On day 3, cultured cells were fed with fresh medium containing relevant cytokines.

Comparison of matched spots in all three cell types
The human monocytic cell line KG1 was purchased from ECACC (Salisbury, UK). Cells were maintained at 0.35 × 10 6 / ml in Iscoves modified Dulbecco's medium (Sigma) supplemented with 2 mM L-glutamine, 100 U/ml penicillin, 100 U/ ml streptomycin (Gibco Life Technologies) and 10% (v/v) FCS (Harlan Sera-Lab) at 37°C in 5% CO 2 . Cells were stimulated with 10 ng/ml PMA and 100 ng/ml ionomycin (both from Sigma) for 24 h, as described previously [6].
Immature peripheral mDCs, immature moDCs and DC-like KG1 cells were cultured in 48-well culture plates at 0.25 × 10 6 cells/ml. Four conditions were set up in parallel, stimulating cells with 0, 10, 50 or 100 pg/ml LPS (Sigma). After 24 h, 250 μl of supernatant was collected and frozen at -80°C, and cells were harvested for cell surface marker staining.

Gel imaging and analysis
Gels were scanned on a GS-800 calibrated imaging densitometer (Bio-Rad). Gel images were analyzed using PDQuest gel analysis software version 7.1 (Bio-Rad). Spots were automatically detected, and then visually checked for undetected or incorrectly detected spots. All images were normalized according to total quantity in valid spots in each gel before any comparisons were made.

Mass spectrometry In gel digestion
Gel pieces were excised and placed in a 96-well plate, then loaded onto a MassPrep robotic liquid handling system (Waters Corporation, Elstree, UK). This was used to destain gel pieces, reduce and alkylate cysteine residues using DTT and iodoacetamide, carry out an in-gel tryptic digest and extract the resulting peptide mixture into a 96-well PCR plate. The extracted peptide mixture was manually desalted using C18 loaded zip-tips (Millipore, Watford, UK). We routinely spotted 2 μl onto sample wells of a stainless steel MALDI target plate previously spotted with 1 μl matrix solution, comprising 1 mg/ml α-cyano-4-hydroxycinnaminic acid (Sigma) in 50% acetonitrile, 50% ethanol and an internal standard, adeno corticotophic hormone (Sigma), at a final concentration of approximately 100 fmol/μl in 0.1% formic acid (Romil, MASCOT scores >64 were taken to be significant (p < 0.05). *Induction factor corresponds to the factor of difference between spot volume in mDCs compared with the respective mDC model. † Biological process: 1, cell growth/maintenance; 2, metabolism; 3, cell communication; 4, morphogenesis; 5, response to external stimulus; 6, cell motility; 7, response to stress; 8, circulation; 9, regulation of cellular processes; 10, cell differentiation; 11, death; 12, cell death; 13, coagulation; 14, homeostasis. Cambridge, UK). Samples were left to air dry and the plate placed in the MALDI mass spectrometer.

MALDI-TOF mass spectrometry analysis
Samples were analyzed using a MALDI TOF mass spectrometer (Waters Corporation) operating at a resolution of greater than 10,000 full width at half maximum in reflectron mode. Spectra were acquired at 5 Hz using a nitrogen laser (337 nm wavelength). Typically, ten data collection events were combined to generate each spectrum. Data acquisition was achieved by randomly sampling from the target well.

Mass spectrometry data analysis
Peak lists were entered into MASCOT PMF [34] and Expasy [35] database search engines. Search parameters included a peptide mass accuracy tolerance of 0.2 Da and allowed for modifications such as alkylation of cysteine during the tryptic digest procedure and the possible formation of methionine sulfoxide.  analyzed on an EPICS Altra flow cytometer (Beckman Coulter, Luton, UK) within six days of staining. Data were analyzed using WinMDI version 2.8. [36]. Isotype-matched irrelevant antibodies were used to verify the staining specificity.

Cytokine expression
Culture supernatants from 4 independent experiments were collected after 24 h stimulation with 0, 10, 50 or 100 pg/ml LPS. Supernatants for each condition were pooled and a cytokine bead array (CBA; Inflammation kit, Becton Dickinson) was performed in triplicate.

Endocytosis assay
For the analysis of the endocytic activity of the three cell types, 1 × 10 5 cells were incubated with FITC-dextran (40,000 MW; Sigma) for 1 h at 37°C. As a control, 1 × 10 5 cells were cooled to 4°C prior to incubation with dextran at 4°C for 1 h. Cells were washed three times and immediately analyzed on a FACS EPICS Altra cytometer.  Analysis of endocytic activity using a FITC-dextran uptake assay. Freshly isolated mDCs, moDCs and KG-1 cells were pulsed with dextran for 1 h at 37°C and the uptake was measured by flow cytometry. Data represent average values of three experiments for each cell type.

Phenotypic comparison of cells
T cell activation assay