Dendritic cells derived from bone marrow cells fail to acquire and present major histocompatibility complex antigens from other dendritic cells

Abstract

Dendritic cells stimulate primary T-cell responses and a major activation route is via presentation of antigens pre-processed by other dendritic cells. This presentation of pre-processed antigens most likely proceeds through transfer of functional major histocompatibility complex (MHC) antigens through exosomes, ‘live nibbling’ or apoptotic vesicles. We hypothesized that not all dendritic cell populations may both donate MHC antigen to dendritic cells and present antigens acquired from other dendritic cells. All populations tested, including those derived from bone marrow precursor cells stimulated primary, allogeneic T-cell responses and acted as accessory cells for mitogen stimulation. Populations of responder type, splenic dendritic cells promoted allogeneic responses indirectly but those derived from bone marrow cells blocked rather than promoted T-cell proliferation. To identify mechanisms underlying this difference we studied transfer of I-A antigens between cells. Active, two-way transfer of allogeneic I-A occurred between splenic primary antigen presenting cells including CD8α⁺ lymphoid dendritic cells, CD8α⁻ myeloid dendritic cells and B220⁺ cells; all these cell types donated as well as acquired MHC molecules. By contrast, the bone marrow-derived dendritic cells donated I-A antigens but acquired negligible amounts. Thus, dendritic cells derived directly from bone marrow cells may stimulate primary T-cell responses through transferring functional MHC to other dendritic cells but may not be able to acquire and present antigens from other dendritic cells. The evidence suggests that T-cell activation may be blocked by the presence of dendritic cells that have not matured through lymphoid tissues which are unable to acquire and present antigens pre-processed by other dendritic cells.

Introduction

Primary responses to alloantigens in the mixed leucocyte reaction (MLR) and to antigens such as viral and bacterial antigens are stimulated by antigens expressed on dendritic cells (DC).^1–4 Antigen transfer between DC is a major factor in stimulation of primary responses. A requirement was initially reported for ‘monocytes’ of responder type as well as stimulator type contributing to primary MLR.^5,6 It was later shown that the requirement was for DC of both responder and stimulator type, mimicking putative involvement of both direct and indirect pathways of stimulation of T-cells in allografting.^7,8 Primary T-cell responses to other antigens (e.g. influenza virus and contact sensitizers) were also only achieved in vitro by participation of DC not exposed directly to antigen which may acquire processed antigen.^8,9 These in vitro studies model transfer of antigen between DC for development of primary immune responses in vivo. Thus, collaboration between allogeneic antigen-presenting cells in allografts is evident and may vary depending on site.¹⁰ In addition, the DC of the recipient may be required for production and/or amplification of primary immune responses to antigens delivered by DC or from DC-derived exosomes.^11–15 Endogenous DC are also required for effectiveness of DC in cancer immunotherapy.¹⁶ Predominance of this indirect route for stimulation of primary responses⁹ may be designed to prevent stimulation of responses by DC to antigens in their own immediate environment.¹⁷

Transfer of antigens between DC may be achieved through exosomes. These exosomes shed from DC can be acquired and presented by DC to stimulate T cells.^18–20 Acquisition of functional major histocompatibility complex (MHC) molecules occurs from freshly harvested DC supernatants and this material is removed by ultra-centrifugation supporting the concept that secreted material is transferred in the form of exosomes.⁷ Transfer of apoptotic vesicles from monocytes to prime DC for T-cell stimulation also occurs.²¹ An alternative route of antigen transfer between live cells is by ‘live nibbling’, which involves acquisition of antigen by DC via scavenger receptors.²² The transfer of antigens between DC in production of primary immune responses means that, after initial acquisition by DC of antigens and antigen processing, the stimulation of primary T-cell proliferation occurs indirectly.⁹ Transfer of processed antigen/MHC complex to other DC, possibly via exosomes or ‘live nibbling’, is followed by presentation of antigen to stimulate T cells.

However, DC are heterogeneous with at least three subsets of DC in spleen. DC can be ‘matured’in vitro from precursor cells using granulocyte–macrophage colony-stimulating factor (GM-CSF) and tumour necrosis factor-α (TNF-α) and these cells are frequently used as surrogate ‘mature’ DC. The functional hallmark of DC is their capacity to stimulate primary T-cell proliferation. We hypothesized that DC of different subsets or from different sources might differ in capacities to donate or acquire processed antigens and in direct stimulatory ability for T-cells. We therefore studied different subsets of spleen antigen-presenting cells and DC derived from bone marrow stem cells. We found that DC matured from bone marrow stem cells with GM-CSF, with or without interleukin-4 (IL-4) or TNF-α, stimulated a primary T-cell response efficiently through transferring MHC to other DC, but not by indirect antigen acquisition. The studies raise the possibility that not all cells currently designated DC may directly stimulate T-cell proliferation.

Materials and methods

Animals

CBA and BALB/c mice, between 6 and 10 weeks of age were obtained from Harlan UK (Bicester, UK). Mice of the same sex were used within experiments.

Cell suspensions

Purified T cells. Single cell suspensions were prepared from inguinal, brachial and axillary lymph nodes in medium [RPMI-1640, Dutch modification; Sigma, Poole, UK with penicillin (100 U/ml) streptomycin (100 μg/ml), l-glutamine (2 mm), 5 × 10⁻⁵ m 2-mercaptoethanol and 10% fetal calf serum (FCS)] by pressing nodes through wire mesh. Enriched T cells (>90%) were obtained by passage of cells over nylon wool columns.²³ T-cell preparations were labelled with a mixture of fluoroscein isothiocyanate (FITC)-conjugated antibodies directed against DC (anti IA^k or anti IA^d (Becton Dickinson, Oxford, UK), CD205 (Serotec, Oxford, UK), and 33D1 (Universal Biologicals, Cambridge, UK) and washed. Anti-FITC magnetic beads were added and cells removed over a magnetic column (Miltenyi Biotec, Bisley, UK). Eluted cells were used as DC depleted T cells.

Spleen and lymph node dendritic cells. DC from lymph nodes, cell suspensions, were prepared by pressing cells through a metal strainer, collecting in medium (5–8 ml) at 5 × 10⁶ per ml and layered onto 2 ml gradients of metrizamide (Sigma; 14·5% w/v solution) and centrifuging for 10 min at 600 g. Interface cells were collected, washed once and re-suspended in medium. In specific pathogen-free animals, the suspensions obtained in this manner were between 70 and 90% CD11c⁺ DC.²⁴

Spleen cells were used to prepare subpopulations of DC as previously characterized in our laboratory.²⁴ To release the maximum numbers of cells of different subsets, spleens were digested with 1 mg/ml collagenase D and 20 μg/ml DNAase (Boehringer Mannheim, Lewes, UK) in medium for 2 hr at 37° with gentle agitation and then passed through a strainer, washed twice and re-suspended in medium. Cells were incubated for 24 hr at 37° in 25 cm² Falcon tissue culture flasks (one spleen per flask) and enriched DC separated over metrizamide as for lymph nodes. The cells obtained consisted largely of CD11c⁺ DC and a population of B220⁺ CD19⁺ cells which were a mixture of CD11c⁺ and CD11C⁻ cells; the latter had both B-cell and DC-like properties of stimulating primary T-cell proliferation.²⁴ The B220⁺ cells were separated by positive selection using biotinylated B220 antibody. The metrizamide-separated cells were labelled with antibody for 30 min at 4° and passed over a miniMACS column according to the manufacturer’s protocol (Miltenyi Biotech). The B220⁺ cells obtained were over 95% B220⁺. The B220⁻ cells were then labelled for CD8α and the cells sorted using a FACScalibur flow cytometer (Becton Dickinson, San Jose, CA). The resulting CD8α⁺ cells were >95% pure, and the CD8α⁻ cells were >95% CD11b⁺ myeloid DC.²⁴ These cell populations were used to study antigen transfer between DC populations.

Generation of bone marrow-derived DC (BMDC)

DC were derived from bone marrow precursors as previously described.²⁵ Briefly, bone marrow cells from mouse femurs and tibias were flushed into medium, and resuspended at a concentration of 1 × 10⁶ cells/ml in complete culture medium supplemented with GM-CSF (40 U/ml) and TNF-α (50 U/ml; Peprotech, London, UK). At day 3 of culture, 75% of the non-adherent population was removed and centrifuged at 300 g for 5 min at room temperature; cells were resuspended in medium supplemented with cytokines (GM-CSF and TNF-α) and replaced in the original tissue culture flask. After 8 days in culture, non-adherent cells were centrifuged on metrizamide (14·5% w/v; Sigma). Interface cells were counted in trypan blue and viability was over 90% and from light scatter and phenotype they were >95%, mature, CD11c⁺, CD11b⁺ DC, expressing MHC class II and co-stimulatory molecules.

Antigen transfer

Antibodies directed against class II were haplotype specific and in preliminary experiments did not cross-react with other haplotypes. All fluorescence-activated cell sorting staining was carried out using the following directly conjugated antibodies, anti-I-A^k clone 11–52-specific for A_α^k, I-A^d, AMS-32.1 (PharMingen, San Diego, CA) and their isotype controls. The antigen transfer was measured using a cut off on the labelling with isotype control for each MHC type and looking at the proportion of the total cell population moving into the double positive quadrant after cell mixing. Alternatively, a quadrant around cells from individual strains was made from their high levels of individual MHC antigen. The increase in numbers of double-labelled cells in that population on mixing was measured using a subtraction algorithm (Winlist 5; Verity Software House, Topsham, ME), which also gave the intensity positive/control intensity ratio of the staining and the values for Maximum difference so that significant changes could be assessed using Kolmogorov–Smirnov statistics. This technique provides a sensitive way of quantifying small differences in fluorescence between samples.²⁶

Proliferation assays

Cultures (20 μl hanging drops in inverted Terasaki plates) contained 6·25–100 × 10³ T cells or DC depleted T cells and received varying numbers of DC. The cultures were pulsed with [³H]thymidine (Amersham International, Amersham, UK; 2 Ci/mm, 1 μl added per culture to give a final concentration of 1 μg of thymidine/ml) for 2 hr on day 3 of culture and harvested by blotting onto filter paper. This amount of thymidine allows free availability of this alternative pathway precursor throughout the pulse time and low specific activity prevents significant radiation damage to cells taking up the radioactive thymidine so that counts, although lower than those obtained using other techniques, accurately reflect the relative amounts of DNA synthesis.^27,28 The small surface area on the meniscus allows rapid interaction between cells to produce an early MLR. The filter was exposed to a tritium screen for 4 hr and imaged in a phosphor imager (Amersham Biosciences). Quantification of accumulated counts on the storage phosphor screen was done using ImageQuant software (Molecular Dynamics, Amersham, UK).²⁹ Counts measured in this way are well correlated with those obtained by scintillation counting (correlation coefficient >0·98).²⁹ In this system, responses to concanavalin A were between 1000 and 6000 counts. Differences in log counts per minute significantly greater than replication variability were calculated using analysis of variance and Student’s t-test using the log-transformed, normally distributed data from the multiple stimulator and responder cell concentrations. Data from a single, representative cell concentration is presented for ease of visualizing comparisons in some figures. Results were representative of those from at least three experiments.

An additional method used responder cells labelled with the dye 5- (and 6) carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, OR) in 200 μl cultures in flat-bottomed micro plates (Falcon). Cell division was then assessed by flow cytometry from the halving of the fluorescent labelling in dividing daughter cells. Count beads were added to the flow cytometry tubes so that absolute numbers of dividing cells, rather than just the proportion of dividing cells, could be assessed.³⁰

Results

Responses to concanavalin A and to anti-CD3

Responses to concanavalin A, measured from uptake of tritiated thymidine, were abrogated by procedures to remove DC from lymph node cells (Fig. 1a) indicating dependence on DC. Responses were restored by addition of spleen or BMDC (Fig. 1a) or by lymph node (LN) DC (not shown). The effective stimulation by DC was confirmed using CFSE labelling of T-cells and assessment of stimulation using flow cytometry as shown for LNDC and BMDC (Fig. 1b). Similar experiments were performed with anti-CD3 antibody as T-cell stimulant (Fig. 1b). These experiments confirmed the efficiency of strategies to remove DC to abrogate responsiveness to stimulation with mitogen or antibody to surface receptor and showed that BMDC as well as other DC populations acted as efficient accessory cells for these responses.

Figure 1.

Effect of depletion of DC from T-cells responding to concanavalin A (ConA) or anti-CD3 antibody. (a) Incorporation of [³H]thymidine by 50 000 LN T cells (open bars) or DC depleted T cells (closed bars) cultured for 3 days with no additions, 2·5 μg/ml ConA, ConA plus 1000 syngeneic spleen DC (spDC), or ConA plus 1000 syngeneic BMDC (bmDC). *Indicates a P value of <0·001 by one way repeated measures analysis of variance. (b) Flow cytometry profiles of 40 000 CFSE-labelled lymph node cells or DC depleted lymph node T cells either unstimulated or stimulated with 5 mg/ml ConA or 4 μg/ml anti CD3. The effect of adding 8000 syngeneic LN DC or BM DC to the DC depleted population is shown.

Primary T-cell responses in the MLR

Figure 2a shows a comparison of DC from spleen or bone marrow in stimulating primary allogeneic MLR in lymph node cells. DC from each source stimulated significant MLR. This stimulation by BMDC provided further confirmation that they were ‘functional DC’ although they were less efficient primary stimulators than spleen DC. When DC from spleen (Fig. 2b) or lymph node (not shown) of responder cell type were added to primary MLR there was significant additional proliferation, indicating a contribution to MLR of responder type DC. In contrast to this additional response with spleen or LNDC, responder type BMDC not only failed to enhance responses but also caused significant inhibition of ongoing proliferation (Fig. 2b).

Figure 2.

Effects of stimulator- or responder-type splenic or bone marrow-derived DC in the MLR. (a) Incorporation of [³H]thymidine by 6250–100 000 LN cells cultured for 3 days alone (⋆) or with 500 allogeneic spleen DC (○), or 500 allogeneic BMDC (▴). (b) Incorporation of ³H thymidine by 100 000 LN T cells (open bars) or DC depleted T cells (closed bars) cultured for 3 days with no additions, 500 allogeneic spleen DC, 500 allogeneic spleen DC plus 1000 syngeneic spleen DC (spDC), or 500 allogeneic spleen DC plus 1000 syngeneic BMDC (bmDC). *P value of <0·001 by one way repeated measures analysis of variance.

Removal of DC from responder populations reduced responsiveness suggesting that at least part of responses were stimulated by indirect presentation of antigen (Fig. 2b). Our previous work had shown that DC of responder type, as well as stimulator type, contribute to allogeneic MLR.^7,8 In Figure 2b we show that responder cells depleted of DC gave a reduced response and confirm significant enhancement by adding responder-type spleen DC to the allogeneic MLR in these cells. However, in contrast, BMDC studied in parallel failed to enhance the MLR (Fig. 2b).

Material shed into supernatants of DC cultured for 24 hr stimulated proliferation of allogeneic lymph node T cells in a dose-dependent way but failed to stimulate proliferation of syngeneic T cells (Fig. 3a). Effects of this type are lost if DC are removed from the responder cells.⁷ Antigen-presenting cells were enriched from spleen cells on metrizamide gradients and three populations of cells were then purified to >95% purity as previously described.²⁴ These populations were CD11c positive populations of CD8α⁺DC, CD8α⁻ DC and CD11c low/negative, B220⁺ population of low density cells, which are also potent stimulators of primary MLR; the CD8α⁺ DC are the most potent primary stimulators.²⁴ Supernatants (24 hr) from each of these cell types slightly promoted proliferation in allogeneic T cells although these effects were minimal in comparison with effects of whole cells (Fig. 3b). Any background proliferation and enhancement of this proliferation by supernatants of allogeneic DC were abrogated by removal of DC from responder T-cell populations (Fig. 3b).

Figure 3.

Effects of supernatants from DC in the mixed leukocyte reaction. (a) Incorporation of [³H]thymidine by 25 000–100 000 LN cells cultured for 3 days alone (○) or with 1 μl DC supernatant (△), or 2 μl DC supernatant (▴). Left hand panel shows response to allogeneic supernatant, right hand panel syngeneic supernatant. (b) Incorporation of [³H]thymidine by 50 000 LN cells (open bars) or DC depleted T cells (closed bars) cultured for 3 days with no additions, 1 μl cell-free supernatant from syngeneic spleen DC, CD8α⁺ DC, CD8α⁻ DC, or B220⁺ DC like cells. *P value of 0·002 and ^♦P value of 0·003 by one-way repeated measures analysis of variance.

Primary T-cell responses to picryl sulphonic acid (PIC)

Figure 4a shows that splenic DC from specific pathogen-free animals stimulated negligible proliferation in syngeneic lymph node cells. However, splenic DC pulsed with PIC stimulated marked primary proliferative responses. By contrast, BMDC exposed to PIC failed to stimulate primary proliferative responses even in the presence of LNDC in responder cells. In primary stimulation using splenic DC, addition to cultures of splenic DC without any direct exposure to antigen enhanced primary proliferative response. Once again addition of BMDC inhibited responses. When DC were removed from responder T cells, primary response was again reduced significantly. When DC from spleen were added in to these depleted cells there was an increase of primary responses. By contrast, BMDC added to cultures failed to support responses and once again caused significant inhibition rather than promotion of proliferation (Fig. 4b).

Figure 4.

Differences between spleen and bone marrow DC in primary responses to antigen. (a) Incorporation of [³H]thymidine by 6250–100 000 LN T cells cultured for 3 days alone (⋆) or with 1000 syngeneic spleen or BMDC (•), or 1000 syngeneic spleen or BM PIC DC (□). (b) Incorporation of [³H]thymidine by 12 500 LN T cells (open bars) or DC-depleted T cells (closed bars) cultured for 3 days with no additions, 1000 syngeneic PIC DC, 1000 PIC DC plus 1000 syngeneic spleen DC, or 1000 PIC DC plus 1000 syngeneic BMDC. *Indicates a P value of <0·001 by one way repeated measures analysis of variance.

Antigen transfer between splenic DC

One mechanism involved in development of primary T-cell stimulation including MLR is transfer of antigens between DC. In the MLR, MHC molecules are transferred from allogeneic DC to DC of responder type, which then stimulate T cells syngeneically.^7,8 This secondarily acquired antigen may be the basis for higher responses produced on addition of responder type DC in the MLR as it is blocked by antibody to I-A specific to responder type as well as to stimulator type. In Figure 5a the transfer of MHC class II molecules between enriched DC populations from mouse spleens is shown. A small level of transfer is shown between allogeneic DC after 2 hr incubation in the cold and transfer increases dramatically when incubation is at 37°; fewer than 2% of cells were labelled with mixtures of immunoglobulin G2b (IgG2b)–FITC or IgG2b–phycoerythrin (PE) antibodies which were appropriate isotype controls for I-A^k-PE or I-A^d–FITC antibodies respectively (not shown). If individual cell populations were labelled with the two antibodies, fixed and then mixed there was little evidence of any double labelling of cells for the two types of MHC class II molecules (Fig. 5b). Aggregation of cells did not account for the double labelling because there was no evidence of increased size of the double-labelled cells and acquired antigen was at a lower level than that present on individual cell populations.

Figure 5.

Transfer of antigen between allogeneic DC. (a) Equal numbers of CBA spleen DC (IA^k) and BALB/c spleen DC (IA^d) were mixed and double labelled with FITC-conjugated anti IA^d and PE-conjugated anti IA^k either immediately (left-hand panel) or after 2 hr incubation at 37° (right-hand panel). Percentages of double positive cells are indicated. (b) Equal numbers of CBA spleen DC and BALB/c spleen DC were mixed after individual cell populations had been double-labelled with anti IA^d-FITC and anti IA^k-PE and fixed (left-hand panel) or cell populations were mixed and incubated at 37° for 2 hr prior to labelling (right-hand panel). Percentages of double positive cells are indicated. (c) Equal numbers of CBA spleen DC and BALB/c spleen DC were mixed and double labelled with anti IA^d-FITC and anti IA^k-PE after 2 hr incubation at 37° in the presence (right-hand panel) or absence (left-hand panel) of 10 μl anti-CD204. Percentages of double positive cells are indicated. d) Equal numbers of CBA spleen DC and BALB/c spleen DC were mixed and double labelled with anti IA^d-FITC and anti IA^k-PE after 2 hr incubation at 37° in the presence (right-hand panel) or absence (left-hand panel) of 10 μl anti MARCO. Percentages of double-positive cells are indicated.

‘Live nibbling’ of antigens by DC has been described and scavenger receptor molecules involved in this antigen transfer between DC identified by blocking with receptor antibodies.²² We therefore checked whether antigen transfer between DC could be blocked by these anti scavenger receptor antibodies. Allogeneic splenic DC were mixed in the presence and absence of scavenger receptor antibodies, anti-CD204 (Fig. 5c), or anti-macrophage receptor with collagenous structure (MARCO) (Fig. 5d). There were no differences in numbers of cells double labelling for MHC class II of the two specificities providing no evidence suggesting that DC utilized either of these two molecules in acquisition of antigen. However, the presence of monensin in the cultures during mixing, significantly down-regulated transfer (data not shown), indicating that active protein release may be involved. This concept was also indicated by low levels of transfer seen between cells held at 4° (Fig. 5a).

Effects of different subsets of DC

We assessed whether DC of different subsets or from different sources differed in their capacity to donate or to acquire allogeneic MHC antigens. In Figure 6 an example of transfer of antigens between LNDC and BMDC is shown. Allogeneic LNDC mixed in culture showed transfer of MHC molecules bidirectionally. After 2 hr at 37° there were increased numbers of double-labelled cells – 31·6% in the example shown in Fig. 6a. In contrast to antigen transfer seen with LNDC, when BMDC were mixed there was no evidence of significant transfer of MHC above that of isotype control labelling (Fig. 6b). When BMDC were mixed with allogeneic LNDC there was little evidence of transfer from BALB/c LNDC to CBA BMDC but BMDC donated MHC molecules to LNDC (Fig. 6c). The transferred antigen could again be distinguished from direct labelling because it was at a lower level than that in individually labelled controls. This effect was also seen in the reverse cell combination when BALB/c BMDC were mixed with CBA LNDC (not shown). DC grown from bone marrow cells in the presence or absence of IL-4 or TNF-α all failed to acquire alloantigen. Thus, BMDC donated to but did not acquire MHC molecules from allogeneic DC. The lack of acquisition of antigen from the allogeneic DC by BMDC is consistent with lack of ability of BMDC syngeneic with responder cells to promote the MLR.

Figure 6.

Capacity of spleen and bone marrow-derived DC to donate and to acquire MHC class II antigen. (a) Equal numbers of CBA LNDC (IA^k) and BALB/c LNDC (IA^d) were mixed and double-labelled with FITC-conjugated anti IA^d and PE-conjugated anti IA^k either immediately (left-hand panel) or after 2 hr incubation at 37° (right-hand panel). Percentages of double-positive cells are indicated. (b) Equal numbers of CBA BMDC (IA^k) and BALB/c BMDC (IA^d) were mixed and double-labelled with FITC-conjugated anti IA^d and PE-conjugated anti IA^k either immediately (left-hand panel) or after 2 hr incubation at 37° (right-hand panel). Percentages of double positive cells are indicated. (c) Equal numbers of CBA BMDC (IA^k) and BALB/c LNDC (IA^d) were mixed and double labelled with FITC-conjugated anti IA^d and PE-conjugated anti IA^k either immediately (left-hand panel) or after 2 hr incubation at 37° (right-hand panel). Percentages of double-positive cells are indicated.

Capacity of DC subsets to transfer antigen

Enriched spleen DC from the whole metrizamide separated population from the two strains of mice were mixed and transfer of MHC molecules to specific subsets of DC assessed by gating on the cells in the flow cytometer. The cells mixed for 2 hr at 4° or at 37° were labelled with I-Ak-PE (PE-labelled), I-Ad (FITC-labelled), CD8α⁻ (phycoerythrin-Texas®x, ECD) and CD19⁻ (PE-Cy5). In Fig. 7, the transfer of antigen to CD8α⁺ lymphoid DC, to CD8α⁻, CD19⁻ myeloid DC and to the CD19⁺ B220⁺ low-density cells is shown. There was transfer of antigen to all the cell types, which was again apparent at low levels in cells incubated at 4° but was much greater at 37°. The most marked transfer was to CD8α⁺ cells.

Figure 7.

Capacity of DC from different subpopulations to acquire allogeneic MHC class II. Equal numbers of CBA DC (IA^k) and BALB/c DC (IA^d) were mixed and labelled with anti IA^d-FITC, anti IA^k-PE, CD8α-ECD, and CD19-PECy5 after incubation for 2 hr at 4° (left-hand panel) 37° (right-hand panel). Dual labelling for class II was then measured in (a) the CD8α+ve gate, (b) the CD8α−ve gate, and (c) the CD19+ve gate. Percentages of double positive cells are indicated.

To show whether each antigen-presenting cell subset was also capable of donating MHC molecules three cell populations with primary T-cell stimulating function were separated from spleen cells to >95% purity as previously described.²⁴ Purified populations were mixed, like with like. After being subjected to the purification techniques the cells showed only low levels of antigen transfer with intensity ratios for acquired MHC between 1·3 and 2. However, high proportions of cells acquiring antigens could be measured for each mixture suggesting that the two populations of DC (CD8α⁺ lymphoid DC and CD8α⁻, CD19⁻ myeloid DC) and the B220⁺ population of cells could all donate as well as acquire MHC molecules with 15–58% of the cells becoming double labelling with both MHC class II molecules in different mixtures (not shown). The transfer of antigen was greatest when CD8α⁺ cells donated antigen and stimulation of MLR was also greatest when CD8α⁺ cells were used to stimulate the MLR as we have previously reported.²⁴

Discussion

Functional studies of mouse DC grown from bone marrow cells under the influence of GM-CSF and TNF-α indicated that they could stimulate an MLR but, unlike DC populations from spleen, BMDC syngeneic to responder cells blocked rather than promoted MLR. The mechanism underlying this difference appeared to be the inability of BMDC to acquire and present MHC antigens processed by other DC. Even when BMDC were stimulated to mature with TNF-α they still failed to acquire MHC molecules. Dilution of effective interaction of antigen-bearing DC with T cells by engagement with DC not bearing antigen could perhaps explain this loss of responsiveness although an active regulatory interaction of BMDC is possible. It was hypothesized originally that different subpopulations of DC might have different properties in donating and acquiring antigens. However, all subpopulations of DC tested from mouse spleen functioned both to donate and to acquire MHC class II molecules. The CD8α⁺ cells were the most efficient at both donating processed antigens and acquiring them in agreement with their greater efficiency both to stimulate primary MLR and to act as cross-presenters of antigen in other systems.²⁴ There is heterogeneity in the cells because only a proportion of DC from each source acquired antigen. The failure of BMDC of responder type to contribute to MLR is in agreement with the inability of these cells to acquire MHC. It may be that this type of DC is also represented amongst the CD11c positive DC of the spleen or lymph nodes. We do not currently have other markers that distinguish this ‘immature’ BMDC population specifically. Transfer of whole functional MHC molecules has also been reported in transfers from DC to T-cells and between gut epithelial cells and DC.^31,32

A possible explanation for the different properties of BMDC is that the BMDC derived from precursor cells ‘in vitro’ are not ‘real DC’ and that the cells derived in this way bear a poor relationship with cells present ‘in vivo’. However, these cells were fully ‘viable’ using dye exclusion and flow cytometry criteria, had a classic DC phenotype and showed maturation processes when stimulated TNF-α, could restore mitogen responses and stimulate an MLR. It thus seems likely that they represent cells that are ‘matured’ in an inappropriate environment. They should thus be used conservatively as ‘standard’ DC populations but such cells may have other interesting attributes. For example, they appear to have properties similar to those abnormal DC present in cancer patients.³³

The inability of BMDC to acquire processed antigen seems to be the mechanism at the core of the deficiency in BMDC antigen-presenting capacity. We have investigated two possible routes implicated in antigen transfer between DC for their involvement in antigen transfer between DC. Previous studies showed MHC molecules in supernatants of DC acquired by allogeneic DC which then stimulated syngeneic T-cells – a phenomenon blocked by antibody to both stimulator and responder type MHC.^7,8 Similar stimulatory materials were released from subpopulations of spleen DC suggesting that the material might be in the form of exosomes. However, this mechanism was always weak compared with effects seen on mixing live cells. The material in supernatants was also labile and activity was lost in a few hours, on storage or during handling of cells to isolate exosomes. These findings are consistent with the requirement for live active cells for stimulation of MLR. The loss of antigen transfer in the presence of monensin also supports the concept that the transfer involves active protein release from live cells. However, some receptor binding of a released substance may also occur because low levels of antigen transfer were seen between cells on ice. Antibodies to scavenger receptors, which others had shown could block ‘live nibbling’, had no significant effects our system and provided no evidence that this mechanism was involved in the antigen transfer.

Mature LNDC acquire and present contact sensitizers to produce a primary T-cell response but blood and skin DC fail to stimulate primary T-cell proliferation to contact sensitizer.³⁴ A likely mechanism for this difference is that sensitizers bind directly to antigen in the groove of MHC molecules on mature, antigen-bearing DC of the type found in spleen or lymph nodes but not present in blood or skin; these mature DC may stimulate T-cells via this route. The lack of primary T-cell stimulation by BMDC is in agreement with the possibility that BMDC are not truly mature antigen-presenting cells. The BMDC in this system, as in the MLR, also showed evidence of a blocking or ‘tolerogenic’ effect on stimulation of ongoing T-cell responses to contact sensitizer. Deficiencies in BMDC in comparison with splenic DC were not, therefore, confined to an inability to acquire antigen processed by DC because the BMDC also failed to stimulate a primary response to contact sensitizer even in the presence of mature LNDC in responder cells. There may be multiple differences between BMDC ‘matured’in vitro and mature DC in lymph nodes.

One major difference between the cells derived in vitro from bone marrow stem cells and those in central lymphoid tissues is that the latter may have been in residence in peripheral tissues. One mechanism involved in stimulation of DC to mature and migrate into lymph nodes is production of TNF-α. This cytokine causes maturation of DC and up-regulation of co-stimulatory molecules. However, BMDC grown in this cytokine, whilst developing a mature phenotype, did not acquire antigens secondarily for stimulation of primary MLR. Thus, DC such as BMDC may be unique in being most efficient at stimulating primary T-cell stimulation by actively transferring antigens to other DC that are capable of direct T-cell stimulation. In most experiments, after depletion of responder DC, there was still some – albeit low – level of residual allogeneic response suggesting that they do have some capacity to cause direct stimulation of allogeneic T cells. However, an occasional complete loss of response on removal of responder-type DC supported the view that DC have little or no capacity directly to stimulate allogeneic T cells; residual response may be caused by potency of small numbers of residual DC in responder populations. The studies thus raise the possibility that BMDC have no capacity directly to stimulate primary responses in T cells. This idea is in agreement with lack of stimulation of primary responses to contact sensitizers discussed above. However, presence of DC in responders is insufficient to promote responses to mitogen and anti-CD3 so this interpretation is not yet clear.

These findings have implications for understanding the role of stem cell-derived DC or DC from other sources in initiating or blocking primary immune responses or for predicting the efficiency of their use for immunotherapy. First, we can speculate that preferential use of this indirect route for stimulation of primary responses, by transfer of antigens to DC not themselves directly exposed to antigen, could prevent stimulation of responses by DC to antigens in their own environment. Such a system would help preserve local integrity of tissues by blocking local immune/autoimmune responses to antigens present on DC throughout a tissue and draining to a local lymph node. However, antigens acquired asymmetrically on some DC or reaching other tissues would be transferred between DC across an antigen gradient and initiate immune activity.¹⁷ Specificity of homing of stimulated lymphocytes to tissues where DC originate would contribute to maintaining appropriate tolerance or immunity to local antigens.³⁰ Second, this study indicates that the use of bone marrow-derived DC may be inhibitory to ongoing lymphocyte stimulation; inhibitory or ‘tolerogenic’ effects are reported for immature DC and exosomes can also have inhibitory as well as stimulatory properties.³⁵ It is not known whether DC matured in vitro from blood monocytes possess properties similar to those of BMDC or to those of lymph node or splenic cells. Finally, there is evidence of changed function of DC in many diseases. The functional attributes of both directly antigen-exposed DC and those that acquire processed antigen secondarily and present it in primary responses now need to be assessed. It should be possible to identify whether changes in acquiring and processing antigen by DC, its transfer or release from these DC or acquisition and presentation of antigens to T-cells by other DC are altered in different disease states. Evidence suggests that DC in human immunodeficiency virus infection may fail to transfer processed antigen to other DC but can acquire and present such antigens; this conclusion results from observations using cells from asymptomatic patients that there is reduction in capacity of DC to stimulate an MLR but that blood T-cells show a full capacity to respond to normal allogeneic DC.³⁶ By contrast, defects in DC acquiring and presenting antigens to T cells in tumour-bearing individuals may occur,³³ which may be a contraindication for the use of DC for therapy.¹⁷

Abbreviations

BMDC

bone marrow dendritic cells

CFSE

carboxyfluorescein diacetate succinimidyl ester

DC

dendritic cells

ECD

phycoerythrin-Texas®x

FCS

fetal calf serum

FITC

fluoroscein isothiocyanate

GM-CSF

granulocyte–macrophage colony-stimulating factor

IL

interleukin

LNDC

lymph node dendritic cells

MARCO

macrophage receptor with collagenous structure

MLR

mixed leucocyte reaction

PIC

picryl sulphonic acid

TNF

tumour necrosis factor