The ability of murine dendritic cell subsets to direct T helper cell differentiation is dependent on microbial signals



Dendritic cells (DC) initiate T cell responses and direct the class of T cell immunity through the production of Th-polarizing cytokines. In the mouse, immunization with CD8α+ DC has led to Th1 priming whereas immunization with CD8α DC has been associated with Th2 induction. Here, we use a direct T cell priming assay in vitro to re-examine the Th-directing potential of total DC or purified CD4+ DC, CD8α+ DC or CD4 CD8α (double-negative; DN) DC subsets from mouse spleen. We show that the default Th effector phenotype induced by priming with DC depends on the protocol used for T cell purification, the T cell:antigen-presenting cell ratio and the antigen dose but is only marginally affected by DC subtype. All DC subsets can direct increased Th1 development in response to microbial stimuli known to elicit IL-12 production. Similarly, all subsets can suppress Th1 development and allow Th2 cellsto expand upon exposure to IL-10-inducing microbial agents. The flexibility of DC in directing Th development in function of microbial signals argues against the notion of pre-determined "DC1" and "DC2" subsets and suggests that multiple DC subtypes can direct an appropriate Th response to different classes of infectious agents.


Double negative


Plasmacytoid DC


Propidium iodide


Solubletachyzoite antigen of Toxoplasma gondii

1 Introduction

Antigen-presenting cells (APC) play an important role in regulating Th effector choice during T cell priming. For example, the amount of antigen presented or the type and quantity of costimulatory molecules expressed by APC can all influence subsequent Th development (reviewed in 1). In addition, APC can produce cytokines, such as IL-12, IL-18 and IFN-γ, that profoundly bias the development of newly-activated T cells 2. Prominent among APC are dendritic cells (DC), which are thought to be central to initiating immune responses and to generating and maintaining T cell tolerance 3. DC encompass several populations found in non-lymphoid and lymphoid tissues. In the mouse, the best studied DC subpopulations are found inthe CD11cbright fraction of spleen and comprise at least three subsets defined on the basis of CD4 and CD8α expression: CD8α+ CD4 DC, CD8α CD4+ DC, and CD8α CD4 (double negative; DN) DC 4.

It has been suggested that DC subtypes may be specialized to prime different Th responses 5. In the human, mature human plasmacytoid DC (PDC) favor Th2 responses whereas monocyte-derived DC preferentially induce Th1 responses 6. In the mouse, Moser and colleagues showed that immunization with antigen-pulsed CD8α+ DC induced a Th1 response while immunization with unfractionated CD8α DC induced a Th2 response 7, 8. In parallel studies, Pulendran et al. 9 noticed that both CD8α+ and CD8α DC populations (purified from Flt-3L-treated mice on the basis of differential CD11b expression) induced Th1 development after immunization but that only the CD11b+ CD8α subset promoted Th2 differentiation. Furthermore, selective expansion of CD8α DC in mice treated with GM-CSF was associated with an increasein Th2-dependent antibody production after immunization 9. However, other have found that CD8α+ and CD8α DC subsets behave similarly and prime mainly non-polarized cells after transfer in vivo10.

Despite their elegance and physiologic relevance, in vivo studies of Th differentiation following DC immunization can be hard to interpret 710. It is difficult to exclude the possibility that differential responses result from differences in DC migration or longevity, which affect the amount of antigen delivered to T cells. In vivo studies are further complicated by the prospect of multiple interactions between DC and different cell types, as well as by the possibility, in some cases, of cross-presentation of the antigen by host DC in recipient mice. To circumvent these problems, we have compared the Th-priming ability of DC subsets in vitro under defined conditions using T cells from DO-11.10 TCR transgenic BALB/c mice as reporters 11. We show that CD8α+, CD4+ and DN DC all induce the differentiation of mixed effector Th populations in vitro but allow skewing towards the Th1 or Th2 phenotypes in response to signals from microorganisms. These results suggest that the Th-determining potential of DC is not primarily determined by ontogeny but by pattern recognition, allowing differential responses to infection to be initiated by all DC subtypes.

2 Results and discussion

2.1 Establishment of a DC-driven Th differentiation assay in vitro

A series of experiments was carried out to determine the optimal assay conditions for studying DC-driven Th differentiation in vitro. DO-11.10 T cells were purified by FACS-based negative selection of contaminating leukocytes rather than positive selection for CD4+ cells 11 to avoid blocking subsequent CD4-I-Ad interactions. T cells purified in this manner gave rise to a predominant Th1 phenotype upon priming with splenic DC-enriched populations (Fig. 1A), a surprising finding given that DO-1 l.10 T cells on a BALB/c background tend to default towards Th2 12. The frequency of Th1 cells could be increased by the addition of IL-12 and anti-IL-4 ("Th1 cocktail") to the priming culture whereas addition of IL-4 and anti-IL-12 ("Th2 cocktail") promoted the development of IL-4-producing Th2 cells (Fig. 1A).

Despite being bred onto a SCID background, the DO-11.10 mice used in these experiments contained a small (<5%) proportion of CD62L T cells, which could represent differentiated Th cells responding to cross-reactive environmental antigens (not shown). To exclude them, purified DO-11.10 T cells were selected for high expression of CD62L. Remarkably, sorted CD62Lhigh T cells that were primed in "neutral" conditions no longer became polarized towards Th1 effectors but gave rise to a mixture of both Th1 and Th2 cells in variable frequency, generally under 30% of either type (Fig. 1A). This change in differentiation under neutral conditions was not just a consequence of the loss of CD62L T cells as a similar phenomenon was seen when the starting T cell population was stained with anti-CD62L but not selected on the basis of CD62L expression (Fig. 1A). Addition of either Th1 cocktail or Th2 cocktail overrode the default tendency in all cases and polarized the cells in the appropriate direction (Fig. 1A). It is not clear at present whether the effects of anti-CD62L are due to cross-linking or to blocking of interactions between CD62L and a potential ligand. CD62L has not been reported to play a role in Th differentiation although differences in CD62L expression have been observed between Th1 and Th2 cells 13.

Th development in neutral conditions was also very dependent on the amount of stimulation, as reported 1416. CD62L+ T cells developed into a mixture of Th1 and Th2 effectors at low APC:T cell ratios (Fig. 1B) or low doses of peptide (not shown) while Th1 development was promoted at high ratios (Fig. 1B) or high antigen doses (not shown). As before, default development under neutral condition could be overridden by addition of an appropriate Th cocktail to the priming cultures (not shown). Together, these experiments suggested that the procedure used for T cell purification, the T cell:APC ratio and the dose of antigen used in priming can all profoundly affect the Th phenotype that emerges upon priming of DO-11.10 T cells with DC. It is unclear at present which, if any, of the conditions tested here most closely approximates a physiological situation. Cultures using T cells purified on the basis of CD62L-expression and primed at a low T cell:DC ratio (10:1) with low concentrations of peptide (0.05–0.1 μg/ml) showed maximum T cell proliferation with the least amount of cell death, resulting in a high yield of effector cells (data not shown). In addition, these culture conditions led to the least degree of Th polarization in the absence of added cytokines while permitting polarization in either direction in the presence of the appropriate cytokine combination (Fig. 1). Because such culture conditions appeared optimally suited to study the influence of APC properties on Th differentiation, they were chosen for subsequent experiments. Nevertheless, it is worth pointing out that, depending on T cell purification procedures or culture conditions, any one phenotype of "default" differentiation can be observed upon priming with DC. Transposed to an in vivo setting, this degree of variation could account for discrepancies in the literature regarding the role of different DC types in Th development.

Figure 1.

Default Th development of DO-11.10 cells in vitro is affected by the T cell purification procedure and T : DC cell ratios. (A) Pooled lymph node and splenic cells from D011.10/scid mice were stained for MHC class II, CD11c, FcγRII/RIII, Gr-1 and CD11b and were FACS sorted to exclude all positive cells. Where indicated, the cells were also stained with anti-CD62L antibody and sorted but not selected for CD62L+ cells or were stained with anti-CD62L and sorted while selecting for CD62L+ cells. All T cell populations were cultured in parallel cultures with the same population of CD11c-enriched BALB/c splenocytes and OVA peptide for 5 days alone or in the presence of IL-12 (10 ng/ml) and anti-IL-4 (10 μg/ml) or IL-4 (50 ng/ml) and anti-IL-12 (10 μg/ml). Expanded cells were subsequently restimulated with PMA/ionomycin and stained for intracellular IFN-γ and IL-4. Dot plots are gated on CD4+/KJ1–26+ cells. Numbers indicate the % of events within each quadrant. Data are representative of three experiments. (B) DO11.10 T cells sorted for CD62L+ cells as above were cultured with CD11c-enriched splenocytes and peptide for 5 days at the indicated T : DC ratios. Expanded cells were subsequently restimulated and analyzed as in (A). Data are representative of two experiments.

2.2 The ability of spleen DC to drive Th differentiation is modulated by microbial products

We have previously shown that an extract of Toxoplasma gondii (STAg), purified protein derivative (PPD) from Mycobacterium tuberculosis or oligodeoxynucleotides containing CpG motifs (CpG DNA) can elicit production of IL-12 p70 by DC, particularly when combined with feedback signals from newly-activated T cells 17. Conversely, heat-killed yeast or zymosan particles elicit production of IL-10 under the same circumstances 17. To examine the consequence of differential cytokine production by DC on subsequent T cell differentiation, DO-11.10 T cells were primed with spleen DC-enriched populations in the presence or absence of different microbial extracts. Inclusion of PPD, STAg or CpG DNA in the priming culture invariably increased the percentage of Th1 IFN-γ-producing cells and decreased the frequency of IL-4-producing Th2 cells compared to cultures with antigen alone (Fig. 2A). Conversely, zymosan particles or heat-killed yeasts decreased the frequency of IFN-γ-producing T cells and caused a modest increase in the frequency of IL-4 producing cells (Fig. 2A and data not shown). CD11cbright DC purified to >99% purity behaved exactly as the DC-enriched spleen APC used in most experiments, demonstrating that the effects observed were not due to the activity of a contaminating cell (not shown). The Th-polarizing effects of all microbial stimuli were markedly reduced if mature, overnight-cultured DC were used instead of fresh cells, suggesting that the stimuli exerted their effect via the APC and did not act directly on the T cells (not shown). This was further supported by the fact that STAg-, CpG DNA- or PPD-driven Th1 development was largely blocked by addition of a neutralizing anti-IL-12 antibody whereas anti-IL-10 relieved the yeast-driven inhibition of Th1 development (Fig. 3).

Figure 2.

DC conditioning by different microbial products leads to differential Th responses. D011.10 T cells were cultured with (A) CD11c-enriched BALB/c splenocytes or (B) sorted DC subsets and 0.05 μg/ml OVA peptide for 5 days in the presence or absence of STAg (1 μg/ml), PPD (5 μg/ml) CpG DNA (1 μg/ml) or zymosan (10 μg/ml). Expanded cells were subsequently restimulated and analyzed as in Fig. 1. Data in (A) and (B) are representative of four and three experiments, respectively.

Figure 3.

Blocking DC-derived IL-10 or IL-12 abrogates the effect of microbial conditioning on T cell priming. DO-11.10 T cells were primed with peptide and (A) heat-killed Schizosaccharomyces pombe (1×105–2×105 particles/ml) or (B) STAg as in Fig. 3 in the presence or absence of anti-IL-10 (10 μg/ml) or anti-IL-12 (20 μg/ml), as indicated. After 5 days, cells were restimulated and stained for intracellular IFN-γ. Results are expressed as fold change in the percentage of IFN-γ+ cells. Data are representative of two experiments.

2.3 Individual DC subsets prime naive T cells in vitro without inducing overt Th polarization unless conditioned by microbial products

To address whether the differential effect of microbial products on T cell differentiation reflected the activity of one or another DC type, the experiments were repeated using cells separated into CD11cbright subsets. In the absence of added microbial stimuli, all subsets led to a pattern of DO-11.10 differentiation in neutral conditions that was similar to that seen using unfractionated DC and was characterized by the emergence of a large fraction of unpolarized Th cells (Fig. 2A, B). However, some subtle differences were noted. In particular, priming with CD8α+ DC generally increased the frequency of Th1 cells developing under neutral conditions while both subsets of CD8α DC tended to promote more Th2 differentiation than unfractionated DC (Fig. 2B). Although not very pronounced, these biases agree overall with the results by Maldonado-López et al. 7, 8 and Pulendran et al. 9 while the overwhelming preponderance of unpolarized cells is similar to the pattern observed by Schlecht et al. 10 after immunization with CD8α+ or CD8α DC.

We next examined whether the Th-directing potential of DC subsets was affected by microbial stimulation. When zymosan was added to the priming cultures, there was a dramatic suppression of Th1 development induced by all three DC subsets (Fig. 2B). In cultures primed by CD8α but not CD8α+ DC, this was also accompanied by an increase in the percentage of Th2 cells (Fig. 2B). In contrast, addition of PPD or STAg to the priming cultures led to a substantial increase in Th1 differentiation and suppression of Th2 development when CD8α+ DC or, to a lesser degree, DN DC were used as APC (Fig. 2B). CD4+ DC were weaker than CD8α+ or DN DC at promoting Th1 development (Fig. 2B). This may be due to the relative inability of CD4+ DC to produce IL-12 p70 17, 18. Nevertheless, CD4+ DC exposed to PPD clearly promoted a higher degree of Th1 development than CD4+ DC cultured with T cells alone (Fig. 2B). In contrast, STAg did not significantly increase the ability of CD4+ DC to promote Th1 development (Fig. 2B), likely due to the exquisite specificity of this stimulus for the CD8α+ DC subset 1921. In all cases, yields from the cultures were very similar and showed that T cells had expanded approximately 20-fold during priming (Table 1). We conclude that although certain DC subsets may possess an intrinsic weak ability to bias Th development, in all cases this can be either strengthened or overridden by signals from microbes.

There are several reports arguing for flexibility in the Th-directing ability of DC. Pre-exposure to microbial signals, cytokines or prostaglandins can all affect the Th-directing potential of human monocyte-derived DC 22, 23 and mouse bone marrow-derived DC have been shown to induce different classes of effector T cells in response to LPS or filarial worm products 24 and, similarly, to discriminate between bacteria and Schistosoma egg antigens 25. Fewer experiments have been done to examine the flexibility of pre-differentiated DC subsets ex vivo as opposed to DC grown in vitro from progenitors. Nevertheless, human plasmacytoid "DC2" can induce Th1 development in response to viral infection 26, 27 and the Th-skewing ability of murine CD8α+ and CD8α DC can be modulated by pre-treatment with IFN-γ or IL-10 8. The results reported here suggest that all murine CD11cbright DC subsets, at least in spleen, are able to couple Th-directing ability to microbial recognition. This is in line with results showing that the cytokine response of CD11chigh murine DC subsets can be dictated by microbial signals 17, 28 and suggests that DC subsets are highly plastic in their responses to infection.

Table 1. T cell expansion after priming with DC subsets in the presence or absence of microbial modulatorsa)
  1. a) The total number (×10–5) of live DO11.10 T cells at the end of the priming culture was calculated as detailed in Sect. 3. Numbers represent the average ± SD of triplicate wells from the experiment shown in Fig. 2B and are representative of three experiments.

Peptide alone7.3 ± 1.98.5 ± 0.4 7.1 ± 0.4
Peptide + PPD6.4 ± 1.48.4 ± 0.1 7.8 ± 0.3
Peptide + STAg7.6 ± 0.67.5 ± 0.5 8.6 ± 0.6
Peptide + zymosan8.7 ± 1   11 ± 1.110.2 ± 1.5

3 Materials and methods

3.1 Animals

Male and female BALB/c mice, 8–12 weeks old, were obtained from Harlan UK (Bicester, Oxon) or from the breeding unit of Cancer Research UK (Clare Hall Laboratories, South Mimms, GB). DO11.10 mice on a BALB/c-scid background were bred at Cancer Research UK.

3.2 Reagents

STAg prepared from tachyzoites of the RH 88 strain of T. gondii 29 was a gift from Dr. Alan Sher, NIH, Bethesda, MD. Zymosan (Sigma, Poole, UK) was boiled for 30 min and washed twice in PBS. Laboratory cultures of S. pombe (strain 513) were autoclaved and washed twice in PBS. The CpG-containing phosphorothioate-linked oligonucleotide TCC ATG ACG TTC CTG ATG CT 30 was made by the Cancer Research UK oligonucleotide synthesis service. PPD was from Staatens Seruminstitut (Copenhagen, Denmark). The ovalbumin peptide 323–339 (OVA peptide; ISQAVHAAHAEINEAGR) was made by Cancer Research UK peptide synthesis service. Endotoxin levels in all reagents were negligible as determined by the Limulus assay (Biowhittaker, Walkersville, MD). mAb used were: HL3, a hamster IgG mAb against CD11c; RM4–5, and 53–6.7, rat IgG2a mAb against CD4, and CD8α, respectively. 11B11 and XMG1.2, rat IgG1 mAb against IL-4 and IFN-γ, respectively. All mAb were from PharMingen, San Diego, CA, unless otherwise indicated. KJ1–26 (anti-DO-11.10 clonotype) was a kind gift from Dr. Paul Garside, Glasgow, GB.

3.3 Cells

Spleen-cell suspensions were prepared by LiberaseTM CI (Roche Diagnostics Ltd., Lewes, GB) and DNaseI digestion 20. DC-enriched splenocytes (70–95% CD11cbright cells) were prepared by positive selection with anti-CD11c MACS beads (Miltenyi Biotec Ltd., Bisley, GB) 17. To purify DC or DC subsets, CD11c-enriched cells were then stained with PE-anti-CD11c, FITC-anti-CD4, and APC-anti-CD8α and sorted using a MoFlo cytometer (Cytomation, Fort Collins, CO) as described 17. Sorted cells were >99% CD11cbright and each subset was >95% pure.

3.4 T cell priming and restimulation

T cells from pooled lymph nodes and spleen of DO11.10-transgenic mice on a SCID background were purified in different ways as described in Sect. 2. Cells used for most DC priming experiments were purified by a combination of negative selection of contaminating cells and enrichment for CD62L+ cells so that greater than 96% of the resulting population stained with KJ-126 and anti-CD4 and had a naive phenotype (CD62L+, CD44low). T cells (4×104) were cultured with DC (T:DC ratio of 10:1) together with OVA peptide (0.05–0.1 μg/ml) in a 96-well round-bottom plate; triplicate wells containing 200 μl medium were set up for each culture, in the presence or absence of various potential modulators. Cells were split 1:2 on day 3 by transferring 100 μl from each original culture into a fresh well and adding 100 μl of fresh medium to all wells. After an additional 2 days, cells were recovered, counted, and re-stimulated. For counting, 50 μl of cells from each of the original triplicate wells was stained with anti-CD4, KJ1–26 and propidium iodide (PI) and analyzed by flow cytometry after adding a known number of fluorescent reference microbeads immediately before collection (Calibrite®, Becton Dickinson, San Jose, CA). The number of live D011.10 T cells in each sample was calculated by comparing the number of CD4+, KJ1–26+, PI events to the number of fluorescent beads events using the formula (CD4+, KJ1–26+, PI events)×(number of beads added to tube)/(bead events). After averaging across the three samples, this number was multiplied by 8 to account for total expansion in the cultures. The remaining cells from the primary cultures were pooled and restimulated in 48-well plates with 10 ng/ml of PMA and 1 μg/ml of ionomycin for 6 h at 37°C with 5 μg/ml of brefeldin A (Sigma) added for the last 4 h of culture. Cells were washed in PBS, fixed with 4% paraformaldehyde for 15 min at room temperature, resuspended in PBS/EDTA containing 1% FCS and 0.02% sodium azide and left overnight at 4°C before staining.

3.5 Cytokine staining

Fixed T cells were resuspended in PBS/EDTA containing 1% FCS, 0.02% sodium azide and 0.1% saponin (Sigma). Cells were stained with biotinylated-KJ1–26 and washed twice before the addition of FITC-conjugated anti-CD4, PE-conjugated anti-IL-4, TriColor-conjugated streptavidin (Caltag, Burlingame, CA) and APC-conjugated anti-IFN-γ; PE- and APC-labeled rat IgG1 isotype-matched controls were used in each case to determine the extent of background staining. After 30 min at 4°C, cells were washed twice in PBS/EDTA containing 1% FCS and 0.02% sodium azide without saponin.

Cell acquisition was performed on a FACScalibur® flow cytometer (Becton Dickinson, Mountain View, CA) and data were analyzed using FlowJo software (Tree Star Inc., San Carlos, CA). Quadrant gates set on the basis of the isotype-matched control samples were used to determine the percentage of cells positive for either cytokine.


This study was supported by Cancer Research UK. We thank Derek Davies, Gary Warnes, Cathy Simpson and Ayad Eddaoudi for their wonderful assistance in cell sorting. We aregrateful to Anne O'Garra, Andre Boonstra and members of the Immunobiology Laboratory, Cancer Research UK, for discussions.


  1. 1


  2. 2


  3. 3