Human Monocytes Infected with Yersinia pestis Express Cell Surface TLR9 and Differentiate into Dendritic Cells¹

Most human infections caused by the Gram-negative bacillus Yersinia pestis are of the bubonic form, arising after the bite of a flea that previously fed on the blood of an infected rodent (1). However, some cases of bubonic plague progress to a pneumonic form, which has a high transmission and fatality rate. Inhaling an infectious aerosol can also cause direct progression to pneumonic plague (2, 3). Monocytes and macrophages are poised to interact with plague bacilli immediately upon injection of the bacteria by a flea bite (4, 5). At the earliest stage of infection, Yersinia pestis bacilli are readily phagocytosed (6) and can grow in the phagolysozome of infected monocytes (5). Multiple virulence factors targeting host immunity are injected into eukaryotic cells by Y. pestis using a cell contact-dependent secretion apparatus (7) found also in Escherichia coli, Salmonella, Shigella spp., and several other pathogenic bacteria. Following a brief intracellular stage, Y. pestis proliferate extracellularly in lymphatic tissues (8). The contribution of monocytes and monocyte-derived dendritic cells (DCs)³ to controlling disease progression is unclear.

Rapid DC recruitment is a hallmark of the acute inflammatory response to microbial infections (9). Activated DCs contribute substantially to inflammatory responses and subsequent resolution of infections, while infection may cause the DC to exhibit a mature phenotype (10, 11). DC and other cells are activated when TLRs and other innate immune receptors respond to pathogen-associated molecular patterns (PAMPs) recognized on macromolecules of pathogens, for example lipids, proteins, and nucleic acids (12). The TLRs are defined by the presence of a cytosolic Toll/IL-1R domain, involved in signal transduction, and an extracellular ligand-binding domain comprised of leucine-rich repeats. Signals transduced through most TLRs trigger the induction of key inflammatory and immune genes by activation of a common molecular pathway that is anchored by the adaptor protein MyD88, and involves NF-κB, MAPK p38, and JNK (13). In addition, cells discriminate between different types of pathogens by combining TLR signals (14, 15).

TLR9 is an integral membrane protein (16) associated with endosomal maturation (17) and is observed on the cell surface of B lymphocytes. In contrast, TLR9 expression by resting mouse macrophages is intracellular and restricted to endosomes (18). The presence of multiple leucine-rich repeats and a cysteine-rich domain in all TLRs suggests a degree of conservation in protein structure. However, there are no obvious sequence motifs indicative of sequestration in any one cellular organelle. Indeed, transfected cells expressing TLR9 on the cell surface respond to hypomethylated CpG-containing oligonucleotides (CpG NT) equivalently to naturally expressed receptors (16), confirming that the site of cellular localization does not affect biological activity. All TLRs, with the exception of TLR7 and TLR9, were reported to be present in human myeloid DCs, whereas TLR7 and TLR9 were selectively expressed by plasmacytoid DCs (19, 20, 21), representing a possible dichotomy in function for these two DC subsets. Because activation of plasmacytoid DCs by TLR9 agonists results in release of type I IFNs (19), these cells are considered to be the primary in vivo targets for mediating the adjuvant effects of bacterial DNA and synthetic CpG NT in humans. However, the mechanisms controlling cell surface expression of TLR9 in B lymphocytes and the apparent intracellular retention of this receptor in plasmacytoid DCs are currently unknown. Furthermore, insufficient data are available concerning the cellular distribution and function of TLR9 in actively infected cells. Although monocytes are important scavengers and hosts for intracellular infection with Y. pestis, they are also an abundant source of precursors for macrophages and DCs. Our results indicate that human monocytes express TLR9 and respond to infection by elevating cytoplasmic and cell surface levels of this innate immune receptor. In addition, cell surface TLR9 expression may function as a proinflammatory activation marker during the transition from monocyte to DCs.

Pooled human AB sera were obtained from Pel-Freez Biologicals (Brown Deer, WI). Mouse anti-human CD14, CD19, and CD3 mAbs conjugated with magnetic beads were purchased from Miltenyi Biotec (Auburn, CA). The FITC-conjugated mAbs Leu M3 (anti-CD14), Leu HLA-DR (anti-DR), anti-human CD19, anti-CCR7, and Ig isotype control Abs were purchased from BD Biosciences (San Jose, CA). Goat anti-mouse IgG, FITC labeled, was purchased from Zymed Laboratories (San Francisco, CA). PE-conjugated anti-CD83 and FITC-conjugated anti-CD86 were purchased from BD Pharmingen (San Diego, CA). Primary mAbs for TLR9, TLR2, and TLR4 were obtained from Active Motif (Carlsbad, CA) or from e-Biosciences (San Diego, CA). Anti-MyD88 Ab was obtained from Neo Markers (Freemont, CA). IL-15 was purchased from PeproTech (Rocky Hill, NJ). Ficoll-Hypaque was purchased from Pharmacia (Uppsala, Sweden). CpG-NT 10103 is a previously reported (22) type B CpG oligonucleotide specifically optimized for vaccine applications and kindly provided by C. Roy (U.S. Army Medical Research Institute of Infectious Diseases). Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI) was purchased from Vector Laboratories (Burlingame, CA). Monoclonal (YPF19) mouse anti-F₁ Ag Y. pestis was obtained from Research Diagnostics (Flanders, NJ).

PBMC were obtained from consenting healthy donors. Blood components for in vitro research use were collected from healthy, consented volunteers in accordance with the guidelines of an Institutional Review Board-approved research donor protocol.

Mononuclear cells were isolated by standard density gradient centrifugation with Ficoll-Hypaque (Pharmacia). Mononuclear cells were harvested from the interface, washed, and suspended in RPMI 1640 medium. Monocytes (CD14⁺) and B cells (CD19⁺) were isolated, as previously described (23). Briefly, unseparated mononuclear cells were suspended (10⁷ cells/80 μl) in cold PBS supplemented with 2 mM EDTA and 0.5% BSA (fraction V; Sigma-Aldrich, St. Louis, MO). Paramagnetic beads coated with anti-CD3 or anti-CD19 mAbs (Miltenyi Biotec) were mixed with the mononuclear cells (10⁷ cells/20 μl). The Ab-labeled cells were incubated for 15 min (4°C), washed, and passed through a type LS or MS iron-fiber column placed within a strong magnetic field (Miltenyi Biotec). CD19⁺ B cells bound to the column were eluted with buffer. Monocytes were isolated from cells that were not bound to the anti-CD3 column, using anti-CD14 mAb, followed by separation in a magnetic column (as above). CD14⁺ monocytes were incubated 48 h with 100 ng/ml rIL-15 (PeproTech) to prepare DCs, as previously described (23). Cell purities were >98% for isolated mononuclear cell subsets used in all reported experiments. Although representative results are presented, all experiments were performed at least three times.

The attenuated Y. pestis strain CO92, Pgm⁻, Pla⁻ was previously described (24). Bacteria were grown from a single colony in heart infusion broth (Difco Laboratories, Detroit, MI) supplemented with 0.2% xylose and 2.5 mM CaCl₂. The cultures were grown for 12 h at 37°C, and bacteria were pelleted by centrifugation and washed with RPMI 1640 medium before use in experiments.

Freshly isolated CD14⁺ monocytes were allowed to adhere to sterile glass coverslips in a 6-well cell culture plate (Corning Glass, Corning, NY) and were then incubated (1 h, 37°C) with 0.1 ml of a Y. pestis suspension in RPMI 1640 medium. Cells were gently washed with PBS (pH 7.4, 37°C) to remove nonadherent cells and bacteria, and were fixed with 1% paraformaldehyde (Tousimis Research, Rockville, MD), 0.1% glutaraldehyde (Sigma-Aldrich) in PBS (5 min, 20°C). The fixed cells were washed with PBS containing 0.5% BSA and incubated (30 min, 20°C) in a permeabilization solution (BD Biosciences). Cells were then washed and incubated (15 min, 20°C) with PBS containing 5% BSA to block nonspecific Ab binding. Cells were incubated with labeled primary anti-F1 Ab for 1 h, washed (PBS, 0.5% BSA), followed by FITC-conjugated goat anti-mouse secondary Ab, and counterstained with DAPI to detect bacteria within the monocytes. Cells were mounted by inverting the coverslips on a slide, and labeled cells were imaged using a Bio-Rad (Hemel Hempstead, U.K.) Radiance 2000 MP laser scanning confocal system connected to a Nikon (Melville, NY) TE 300 inverted microscope. FITC fluorescence was visualized using 488-nm wavelength excitation from a krypton/argon laser, and DAPI images were collected using 2-photon excitation from a Ti/sapphire laser tuned to 800 nm. Images were collected using Bio-Rad LaserSharp software. Images taken for the purpose of comparing uninfected with Y. pestis-infected cells were collected using identical settings, and any subsequent processing of the images was likewise performed identically for the two conditions.

To examine cell surface expression of proteins on primary cells or after infection, cells were incubated (20 min, 4°C) with FcR-blocking reagent (Miltenyi Biotec), washed twice with HBSS containing 1% BSA, and then incubated (30 min, 4°C) with FITC-labeled or PE-labeled control or isotype-matched mAbs. For unlabeled mAbs (detecting TLR9, CCR7, F₁), cells were first incubated (30 min, 4°C) with primary Ab, washed briefly by centrifugation, and then incubated with goat anti-mouse IgG (FITC or PE conjugated). Unbound Ab was removed by washing the cells with HBSS (4°C) and centrifugation. After two additional washes, the labeled cells were fixed with 1% paraformaldehyde in PBS, and the cell-associated immunofluorescence was measured by flow cytometry (FACScan; BD Biosciences).

For detecting T cell proliferative responses to Yersinia Ags, CD14⁺ monocytes from an HLA-A2 donor were infected with Y. pestis at the start of culture. Gentamicin (10 μg/ml) was added to cultures after 2 h of infection. Monocytes were washed extensively to remove extracellular bacteria before adding autologous T cells to the culture at 24 h. Primary T cell cultures were then collected after 6 days of incubation. After 6 days of culture with purified recombinant Yersinia F1 Ag presented by fresh DCs, cultures were pulsed with 1 μCi of [³H]thymidine/well for 10 h, and proliferation was measured by harvesting the cells onto microplate unifilters and measuring radioactivity in a liquid scintillation counter (Packard Instrument, Meriden, CT).

Cytolytic activity was measured by a luciferase-based assay of ATP present in metabolically active target cells (Lumitech, Nottingham, U.K.). Briefly, monocytes were obtained from an HLA-A2 donor. To transform monocytes to DCs, monocytes were cultured with IL-15 for 48 h, as previously described (23). Cells were washed extensively to remove the IL-15 and any nonadherent cells. Monocytes were also cultured in the absence of IL-15 and remained as monocytes throughout the culture. Autologous human T cells and Ags (10 μg/ml) were added to the monocyte or DC culture and incubated for 6 days. T cells were collected and used as effector cells in cytotoxicity assays. HLA-A2-matched B lymphoblastoid cell line (BSM) were pulsed with soluble recombinant Ag (10 μg/ml, final Ag concentration), or no Ag for 1 h at 37°C and used as target cells. Target cells (BSM cells, 5 × 10³/well) were incubated (4 h; 100 μl volume) with specific effector T cells (1:1, 1:5, and 1:10 of BSM:T cell). Cells were then lysed with 100 μl of a nucleotide-releasing agent (Lumitech); then 20 μl of an ATP-monitoring agent was added. Light emission was measured in a luminometer (Wallac 1420 Victor; PerkinElmer, Shelton, CT). Cytotoxicity was measured after CTL killing, after subtraction of experimental from maximum lysis of target cells.

Total RNA was extracted from untreated and Y. pestis-infected monocytes using Tri-Reagent (Molecular Research Center, Cincinnati, OH) and reverse transcribed into cDNA with Maloney murine leukemia virus reverse transcriptase (Promega, Madison, WI), according to the manufacturer’s instructions. Amplification of IFN-α-specific and β-actin control sequences was performed, as previously described (25). PCR products were run on a 1.5% agarose gel, stained with ethidium bromide, and photographed.

Cytokines in culture supernatants were measured using BD Cytometric Bead Array kit (BD Pharmingen) using capture beads coated with Abs specific for cytokines and flow cytometry analysis, according to the manufacturer’s method. Briefly, the cytokine standards were reconstituted in the assay dilution buffer for 15 min, and then were diluted by serial dilution from 5 to 5000 pg/ml. The cytokine capture beads were mixed together (10 μl/test) and then transferred, 50 μl of mixed beads to each assay tube. The PE detection reagent was added next to each assay tube (50 μl/test). The standard dilutions and test samples were added to the appropriate sample tubes (50 μl/test). The capture beads, PE-conjugated detection Abs, which form a sandwich complex, and recombinant standards or test samples were incubated together for 3 h at room temperature and protected from the light. After the 3-h incubation, the samples were washed with 1 ml of wash buffer and centrifuged for 10 min at ∼200 × g. The supernatant was decanted, and 300 μl of wash buffer was added to each assay tube for analysis. After sample data acquisition by the BD FACSCalibur flow cytometer, the sample results were plotted in graphical and tabular format by the BD Cytometric Bead Array analysis software.

Resting monocytes were selected for our study from peripheral blood of volunteers, based on expression of the LPS-binding receptor CD14. Fresh cultures of Y. pestis, grown at 37°C under optimal conditions for production of bacterial capsules and most virulence factors, were washed in cell culture medium and mixed with monocytes. Bacteria labeled with an Ab reacting with the capsular protein F1 were clearly visible within the cytoplasm of infected monocytes by 1 h of culture (Fig. 1), and we observed a consistently high level of monocytes infected with bacteria (>87% average). The infected monocytes remained viable after 48 h of infection, and acquired a DC-like morphology, exhibited a loss of CD14, and increased cell surface expression of HLA-DR, CD86, and the DC maturation markers CD83 and CCR7 (Figs. 2,A and 5,A). These phenotypic changes indicated that infection resulted in differentiation of the monocytes into DCs. The Y. pestis-infected monocytes also expressed IFN-α (Fig. 2,B). Significant levels of TNF-α, IL-1β, and IL-6 were produced by monocytes infected for 24 h (Fig. 3), suggesting a possible contribution of proinflammatory cytokines to maturation of the infected monocytes to DCs.

The capsular F1 protein of Y. pestis is a major protective Ag, and immunity to F1 Ag confers protection from plague infection by either the s.c. or respiratory route of exposure (26). We cocultured Y. pestis-infected monocytes or DCs with autologous T cells to evaluate Yersinia-specific T cell responses. T cells primed in vitro were isolated and tested with F1 Ag in proliferation and cytotoxicity assays. Our results showed that Y. pestis-infected DCs induced F1-specific T cell proliferation (Fig. 4). In addition to Th responses, monocytes and DCs infected with Y. pestis stimulated F1 Ag-specific CTL responses (Fig. 4). These results indicated that DCs derived from infected monocytes were functionally equivalent to DCs infected after maturation from cultured monocytes.

Infection of monocytes with Y. pestis resulted in reduced cell surface CD14, whereas levels of CD14 on cells treated only with CpG NT remained similar to untreated control cultures (Fig. 5,A). Because molecular signatures of pathogens are recognized by TLRs, which in turn activate DCs, we next examined changes in expression of TLRs on monocytes from multiple donors. Primary monocytes were infected for 48 h with Y. pestis and examined by flow cytometry. Infection with Y. pestis increased cell surface TLR9, whereas no change was observed for expression of TLR2 or TLR4 compared with levels in untreated cells (Fig. 5,A). Similarly, treatment with CpG NT, containing motifs optimal for human biological activity, increased cell surface expression of TLR9. Adding CpG NT agonists of TLR9 to infected monocytes decreased cell surface TLR9 (Fig. 5,A). A significantly increased colocalization of MyD88 with TLR9 was observed in the cytoplasm of cells harboring intracellular bacteria (Fig. 5,B), whereas most uninfected monocytes retained a low, but detectable, level of TLR9 with little colocalization of MyD88. We also noted that levels of intracellular MyD88 were quite high compared with the noninfected controls (Fig. 5 B), suggesting that neoexpression of the adaptor protein may be essential for the TLR9-mediated response to infection. These results indicated that TLR9 signaling was activated by infection and coincided with DC differentiation.

Because our results showed a significant increase in cell surface expression of TLR9 with Y. pestis infection, this prompted us to examine expression on freshly isolated human blood cells. By flow cytometry, cell surface TLR9 was observed on 5–7% of the CD14⁺ monocytes and B cells (Fig. 6 A), varying from donor to donor (n = 16). Similar results were obtained with two independent mAbs against human TLR9 (data not shown). These results confirmed that freshly isolated monocytes are poised to respond to TLR9 ligands.

As expression of TLR9 is a prerequisite for CpG NT responsiveness in B cells, we next examined TLR9 in primary cultures of B (CD19⁺) lymphocytes, depleted of monocytes. A low level of TLR9 expression was observed by flow cytometry (Fig. 6), which appeared to be greater than that displayed by monocytes (Fig. 5,A). We observed a dramatic increase in CD86 and HLA-DR, and a more modest increase in CD83 expression, in B cells responding to TLR9 agonist treatment (Fig. 6). Like monocytes, the change in surface markers in response to CpG NT was dose dependent (data not shown). Also similar to the results obtained with monocytes (Fig. 3), B cells treated with CpG NT (specific for human receptor) exhibited a significant increase in cell surface TLR9 (Fig. 6,B). Because the CpG NT-induced increase in CD86 expression by B cells was so dramatic, we re-examined expression of this costimulatory molecule in monocytes (Fig. 6 B). An increase in CD86 was observed in CD14⁺ cells treated with CpG NT, although the level was somewhat less than that obtained with B cells. These data also indicate that increased TLR9 cell surface expression was not dependent on phagocytosis. Collectively, our results suggest that agonist-induced changes in expression of costimulatory molecules and TLR9 are similar in monocytes and B lymphocytes.

A thorough understanding of the initial interactions between host and pathogen is essential for devising methods of controlling the high mortality associated with Y. pestis infections. Our results demonstrate that human monocytes infected with Y. pestis differentiate into DCs, suggesting that adaptive immunity may still function during the intracellular stage of plague. Primary helper and CTLs were activated by monocytes matured to DCs by infection and by myeloid-derived DCs that were directly infected with Y. pestis. The capsular F1 protein, which is a major protective Ag and vaccine component (27), was recognized by lymphocytes responding to infected monocytes. Activation of cells expressing TLR9 by hypomethylated CpG NT motifs present in bacterial DNA stimulates a rapid inflammatory response (28). Our results also indicate that Y. pestis infection activates the TLR9 signal transduction pathway, resulting in a significant increase in expression of TLR9, both intracellularly and at the cell surface. Expression of intracellular MyD88 was also activated by infection, suggesting that increased amounts of this adaptor may be necessary to accommodate the elevated levels of TLR9 resulting from the host immune response to infection. Phagocytic internalization of bacteria is also strongly promoted by TLR9 (29). In addition, we conclude that both monocytes and myeloid DCs express functional TLR9, a property previously thought to be restricted to B cells and plasmacytoid DCs.

The recruitment of MyD88 to endosomes containing TLR9 confirmed that these receptors of innate immunity were functional in Y. pestis-infected monocytes. All TLR9-mediated responses are dependent on MyD88 (30), and colocalization of TLR9, MyD88, and CpG NT in late endosomes is necessary for signaling (17, 28). The colocalization of agonist, adaptor, and receptor was reported to be inhibited by wortmannin, suggesting a critical involvement of PI3K (31). Agonists of TLR9 bind to the plasma membrane (32), while other reports suggest that CpG NT-mediated signaling occurs in acidified endocytic vesicles (33, 34). The precise cellular mechanism delivering TLR9 to the cell surface is presently not known. Furthermore, it is possible that TLR9 turnover from cytoplasm to the cell surface is not critical to function and that signal transduction only occurs in endosomes. Macrophage phagosomes are formed by direct association and fusion of endoplasmic reticulum to the plasma membrane during early phagocytosis (35). Both endoplasmic reticulum and endosomes contribute elements to the phagosome membrane (35), bringing proteins from these intracellular compartments to the cell surface during phagocytosis. Therefore, it is possible that phagocytosis may cause TLR9 to cycle between endosomal and plasma membranes. However, phagocytosis of inert beads by monocytes did not result in activation of TLR9 expression or turnover, and a previous study of 293 cells transfected with a hemagglutinin-tagged TLR9 demonstrated receptor in the plasma membrane of these nonphagocytic cells (28). Taken together, these observations suggest a potential role for TLR9 in innate immune recognition of both intracellular and extracellular PAMPs, perhaps serving an additional regulatory function in vivo. Interestingly, TLR2 can bind Ags at the cell surface (36) and thus contribute to CD4⁺ T cell recognition of cognate Ag (36). In a similar manner, the high expression level of TLR9 on infected monocytes compared with the reduced basal level found on resting monocytes may facilitate processing of stimulatory PAMPs or other Ags and serve as a signal of infection to other cells of the immune system.

Although the host activation signals produced during bacterial infection are likely to be complex, previous studies have demonstrated an involvement of TLRs in differentiation of DCs (37, 38, 39). In addition, the recent observation that CpG NT serve as maturation signals for activated myeloid DCs (40) supports our conclusion that functional TLR9 is also expressed by nonplasmacytoid human DCs. Agonists of TLR9 clearly function as differentiation signals for B cells, which also express this receptor on the cell surface. The rapid increase in expression of TLR9 and TLR10 by naive B cells triggered by cross-linking of cell surface Ig allows them to proliferate in response to CpG NT (41). In contrast, memory B cells express high levels of most TLRs, proliferate directly in response to CpG NT, and differentiate into Ig-secreting cells. Proliferation or secretion of IgG, IgM, IL-10, and IL-6 by CpG NT-treated B cells required costimulation by CD40L (42). Our results and previous reports (41) demonstrated increased cell surface expression of costimulatory molecules and HLA-DR on B cells activated with CpG NT. In a manner similar to our results with monocytes, cell surface TLR9 expression was also increased on B cells by treatment with CpG NT. However, despite the many similarities between B cells and monocytes regarding TLR9 signaling, cell activation by CpG NT alone was not sufficient to drive differentiation of monocytes to DCs.

Many obligate and facultative intracellular bacteria are potent inducers of DC maturation in vitro (43). For example, maturation was reported to be induced by in vitro infection of human myeloid DCs with the intracellular organism Listeria monocytogenes (44). DC maturation during infection may be mediated by proinflammatory cytokines, for example TNF-α, IL-1β, and IL-6 measured in our study and in previous reports (45), or IL-15, as previously reported (23). In contrast, virulent Mycobacterium tuberculosis and Mycobacterium leprae-infected DCs inhibited cellular activation induced by CD40L, a potent stimulant of maturation and Ag presentation (46, 47). Significantly, this is the first report of differentiation of human monocytes to DCs by a bacterial infection. It is likely that the differentiation of monocytes to DCs, as well as the induction of TLR9 expression may be observable with other microbial infections.

We thank Larry Ostby for figure preparation.