Abstract

In this report we identify an accessory cell that interacts with primed and memory T cells at sites where they collaborate with B cells. These cells are distinguished from conventional dendritic cells by their lack of response to Flt3 ligand and their inability to process antigen. Unlike dendritic cells, the CD4+CD3− cells have little CD80 or CD86 expression but do express high levels of the TNF ligands, OX40 ligand and CD30 ligand. We show that Th2-primed cells express the receptors for these TNF ligands and preferentially survive when cocultured with these cells. Furthermore, we show that the preferential survival of OX40+ T cells and support of memory T cell help for B cells are linked to their association with CD4+CD3− cells in vivo.

Introduction

The generation of antibody responses to protein antigens requires the rapid expansion of generally rare antigen-specific CD4 T cells and their recruitment to the two principle sites where they collaborate with B cells. In the T zone they provide help for B cells, enabling the rapid generation of primary nonsomatically mutated plasma cells in extrafollicular foci Jacob et al. 1991, Liu et al. 1991. In B follicles CD4 T cells initiate the development of germinal centers (GCs), which subsequently give rise to the precursors of long-lived plasma cells Ho et al. 1986, Slifka et al. 1998. Furthermore, following reexposure to antigen, primed memory CD4 T cells provide help to both memory and naive B cells for accelerated secondary immune responses.

The cellular and molecular interactions that direct T cells to help B cells during priming and to survive to provide help in secondary responses are still incompletely understood. In vitro differentiated Th2 CD4 cells are more effective inducers of GCs than Th1 cells Randolph et al. 1999, Secord et al. 1996, and this is associated with their preferential migration to the B:T interface, a major site of B:T collaboration (Randolph et al., 1999). The CD28 costimulatory signal to T cells is essential for GC development Ferguson et al. 1996, Lane et al. 1994 and is almost certainly provided by professional antigen-presenting dendritic cells (DCs) during priming. CD28-activated CD4 cells upregulate OX40 (Walker et al., 1999), which is not expressed on naive cells (Mallett et al., 1990), permitting secondary signals from OX40 ligand expressed on CD40-activated DCs Brocker et al. 1999, Ohshima et al. 1997. In vitro, OX40 signals can promote Th2 development Baum et al. 1994, Flynn et al. 1998, Ohshima et al. 1998, and OX40 signals to CD4 T cells in vitro (Flynn et al., 1998) and in vivo (Brocker et al., 1999) upregulate expression of the chemokine receptor CXCR-5, which directs their migration to B follicles (Forster et al., 1996) where the ligand is expressed (Gunn et al., 1998). CXCR-5+ CD4 T cells are associated with help for GC development Breitfeld et al. 2000, Campbell et al. 2001, Kim et al. 2001, so it is plausible that sequential signals from DCs through CD28 and then OX40 are components of the signaling pathway that directs T cells to migrate to and interact with B cells.

Since conventional DCs are restricted to the T zone, it seemed possible that other cell types at the B:T interface and in B follicles might mediate B:T interactions. In this report we have characterized such an accessory cell that is phenotypically and functionally distinct from DCs. We provide evidence that suggests that these cells are the adult equivalent of CD4+CD3− cells found to colonize fetal secondary lymphoid organs early in ontogeny (Mebius et al., 1997). We show that these cells are intimately associated with antigen-specific T cells during primary and memory phases of immune responses to Th2 antigens and find that the survival of OX40+ memory CD4 T cells providing help for secondary B cell responses is linked to OX40 ligand signals provided by these cells.

Results

Identification of a Subset of CD4+CD3− Cells in B Follicles

We first set out to determine whether there were conventional CD11c+ DCs in B cell areas using confocal microscopy of mouse spleen tissue sections (see Figures 1A and 1B). This showed that the majority of CD11c+ DCs were located in and around the T zone or in the red pulp, with CD11c+ cells being rare in B follicles (Figure 1A). However, there were many CD4+ cells in B follicles that were CD11c−, and most of these CD4+ follicular cells were not T cells as they lacked CD3 (Figure 1A). The presence of CD4+CD3− cells in B cell areas did not require GCs, as mice lacking CD40 or CD28 signaling had normal numbers of CD4+CD3− cells (data not shown).

Image Processing of Confocal Images to Identify Positively the Location of CD4+CD3− Cells in Relation to CD11c+ DC Subsets

Because it was hard to identify CD4+CD3−CD11c− cells unequivocally in areas where there were significant numbers of CD4+ T cells, we resorted to image processing of confocal images (see Experimental Procedures). Serial sections were stained to identify CD11c+ DCs and IgM+ B cells. To identify the relative locations of the respective CD4+CD3−CD11c− cells and CD11c+ DC populations, the CD4+CD11c−CD3− digital image was merged with the CD11c+ digital image. The IgM digital matrix from the two images was used to align B follicle areas, and then the individual overlapping matrices were assigned pseudocolors to identify IgM (gray), CD4+CD3−CD11c− (red), CD11c+ (green).

The resulting image (Figure 1B) shows clearly the relative location of CD4+CD3− cells. In addition to being found in B cell areas, these cells are found at the B:T interface, in the marginal sinus, and a few cells occur in the central T zone. In contrast, virtually all CD11c+ DCs are found in either the T zone or in the red pulp.

CD4 T Cells Contact Cell Membranes of CD4+CD3− Cells Both at the B:T Interface and in GCs

To determine whether antigen-specific CD4 T cells were in contact with CD4+CD3− cells at the B:T interface, CD45.1-marked transgenic OTII cells were transferred into CD40 ligand-deficient congenic recipients which were immunized a day later with the Th2 immunogen alum-precipitated ovalbumin (Toellner et al., 1998). Figure 1C shows a low-power and the inset shows a high-power image of a CD4+CD3− cell interacting with an antigen-specific transferred CD4 T cell 3 days after immunization. What is clearly shown in this confocal image (Figure 1C) is the dendritic morphology of the CD4+CD3− cell and the intimate contact of membranes between it and the antigen-specific transferred T cell. Similar interactions were also seen in GCs (Figure 1D). As the recipient mice were CD40 ligand-deficient, host cells cannot provide help for GC B cell proliferation, so there are no background GCs. Figure 1D and the inset show low- and high-power images of a nascent GC showing a transferred antigen-specific T cell interacting with a CD4+CD3− cell, illustrating again that these cells interact with primed T cells at sites of T:B collaboration in B follicles.

The associations of antigen-specific T cells with DCs and CD4+CD3− cells in confocal images that scan a 2.6 μM slice through the tissue when viewed with 63× objective were enumerated. As only part of each OTII CD4 T cell membrane is analyzed, this underestimates the true association percentage. Nevertheless, on day 2 after transfer 50/56 (89%) of transferred CD4 T cells are found associated with DCs but only 5/38 (13%) with CD4+CD3− cells. By day 5 at the peak of the OTII CD4 T cell expansion (data not shown), this pattern had changed. While 79/115 (69%) OTII CD4 cells still had contacts with DCs, 74/110 (67%) were now found associated with CD4+CD3− cells, consistent with the idea that primed CD4 T cells migrate to and associate with CD4+CD3− cells after priming on DCs.

Purification and Immunophenotype of CD4+CD3− Cells versus CD11c+ DCs

CD4+CD3− cells were prepared by negatively selecting CD11c+ cells from spleen suspensions from Rag1-deficient (Rag1−) mice. CD4+ cells were then isolated by positive selection from the CD11c− fraction. Purified cells were incubated overnight in culture medium at 37°C and immunostained the following day. Two cell populations are purified by this procedure (see Figure 2A). The first is a population that expresses CD11c+ and little CD4; this can be further subdivided by expression of CD8 (Figures 2Ab and 2Ac). The second population is CD4+CD11clow/− (Figure 2Aa). The immunophenotype of the CD11c+ DCs shown is similar to conventional CD11c+ DCs (data not shown). Unlike the CD11c+ fraction, the CD4+CD3− cells express lower levels of MHC class II, CD40, CD80 and CD86 (Figure 2Aa). This was independently confirmed by staining with CTLA-4Ig (data not shown). In marked contrast, CD4+CD3− cells express high levels of the TNF family members, OX40 ligand and CD30 ligand (Figure 2Aa). This again was independently confirmed using a murine OX40-Ig fusion protein in addition to mAbs to murine OX40 ligand and CD30 ligand (data not shown). Unlike CD11c+ DCs that can express modest levels of OX40 ligand but not CD30 ligand after activation through CD40 Brocker et al. 1999, Ohshima et al. 1997, CD4+CD3− cells do not require additional activation beyond simply being put in culture for 12 hr.

Freshly Isolated CD4+CD3− Cells Express High Levels of mRNA for OX40 Ligand and CD30 Ligand

Freshly isolated CD4+CD3− cells do not express detectable OX40 ligand or CD30 ligand. There were two possible explanations for this: (1) TNF ligands are known to be cleaved readily by metalloproteases (Crowe et al., 1995), and this could have occurred during the purification process in vitro, or (2) the CD4+CD3− cells might only upregulate these molecules after in vitro culture. To differentiate between these two possibilities, we FACS sorted freshly isolated CD4+CD3− cells and CD11c+ DC subsets. mRNA was made from sorted populations which were >95% pure, and RNA message for TNF ligands was quantified using PCR. After correction for β-actin levels, CD4+CD3− cells expressed ∼100-fold greater OX40 ligand mRNA than CD11c+ DCs and 50-fold excess of CD30 ligand mRNA (Figure 2B). This is direct evidence that CD4+CD3− cells synthesize OX40 ligand and CD30 ligand mRNA in vivo.

CD4+CD3− Cells Resemble a Population Described to Colonize Neonatal Lymph Nodes but Are Distinct from CD4+CD3−B220+ Plasmacytoid Cells

The population of CD4+CD3− cells that we isolated from Rag1− spleens was phenotypically very similar to a population characterized in neonatal lymph nodes (Mebius et al., 1997) (see Table 1). They express CD45 but not macrophage, B, or T cell markers. They were also distinguished from CD4+CD3− B220+CD11clow murine plasmacytoid cells that secrete large quantities of type I interferons in response to viral stimuli Asselin-Paturel et al. 2001, Nakano et al. 2001. Whereas our cells were strongly positive for OX40 ligand and CD30 ligand, the CD4+CD3− B220+CD11clow fraction was negative for these ligands (Figure 2C).

Table 1. Phenotypic Characterization of CD4+CD3− Cells

Surface AntigenCD4+CD3− CellsMebius et al., 1997
CD4++
MHC class II++
CD25 (IL-2Rα)++
CD132 (IL-2Rγc)++
CD44++
Thy1.2 (CD90.2)++
CD3−−
CD11c−−
CD40−−
CD80−−
CD86−−
B220−−
NK1.1−−
L-selectin (CD62L)+/−−
CD28−n.d.
B220−−
CXCR5−+mRNA
Pan-NK DX5 mAb−n.d.
CD8−n.d.
CD45++
Macrophage F4/80−n.d.
MARCO−n.d.

Unlike CD11c+ DCs, CD4+CD3− Cells Are Not Expanded by Flt3 Ligand but Are Dependent on CD132

The marked phenotypic differences between CD4+CD3− cells and conventional CD11c+ DCs suggested that they might be derived from different cell lineages. To investigate this further, we examined the in vivo response to the fms-like tyrosine kinase receptor 3 ligand (Flt3 ligand), which has been reported to expand murine CD11c+ DCs (Pulendran et al., 1997). All CD11c+ DC subsets were increased ∼10-fold in Flt3 ligand-treated mice compared with PBS-injected controls (Figure 3A). However, Flt3 ligand had little or no effect on the total numbers of CD4+CD3− cells/spleen, showing the CD4+CD3− population to be unresponsive to Flt3 ligand as a growth factor.

Neonatal CD4+CD3− cells (Mebius et al., 1997) express CD132 (the common cytokine receptor γ [γ] chain for the cytokines IL-2, IL-4, IL-7, IL-9, and IL-15), and IL-7 receptor-deficient mice have fewer CD4+CD3− cells (Adachi et al., 1998). To test whether the CD4+CD3− cells that we found in adult mice were dependent on CD132 signals, we compared spleens of Rag1− mice sufficient or deficient in CD132. Although the spleens of CD132− Rag1− mice were smaller than those from CD132+ Rag1− mice, this translated into a 2.4-fold reduction in the numbers of CD8−CD11c+ DCs and a 2.8-fold reduction in CD8+CD11c+ DCs. In contrast, the number of CD4+CD3− cells isolated from each spleen was reduced 6-fold in CD132− Rag1− mice (Figure 3B).

OX40 and CD30 Are Preferentially Expressed on Th2 CD4 T Cells

The expression of CD30 ligand and OX40 ligand on CD4+CD3− cells suggested that they would preferentially interact with OX40+ and CD30+ T cells. We first differentiated OVA-specific transgenic CD4 T cells (OTII) under nonpolarizing Th0 and polarizing Th1 and Th2 conditions (Figure 4A). Th0, Th1, and Th2 primed cells all expressed OX40 although expression was greatest on Th2 cells. Th0 and Th1 cells expressed little CD30, which in contrast was strongly expressed on Th2 cells. Using allotypic markers to distinguish OX40+ and OX40− transgenic CD4 T cells, we found that expression of CD30 on Th2 cells did not depend on OX40. Experiments with OX40+ and OX40− T cells cultured alone gave identical results (data not shown).

In initial cultures, the ratio of OX40+ to OX40− CD4 T cells was ∼1:1 (1:0.9 in the experiment shown). After 6 days in culture in vitro, OX40− CD4 T cells outnumbered their normal counterparts under Th0 (5.5:1) and Th1 conditions (2.5:1), and labeling of cells prior to culture with a fluorescent dye showed this to be attributable to increased proliferation of OX40− T cells (data not shown). In contrast, OX40+ Th2 CD4 cells outnumbered OX40− cells by a small margin (1:0.7).

Deficiency of OX40 did not affect the capacity of exogenously provided IL-4 to induce Th2 (IL-4 and IL-10) or exogenously provided IL-12 to induce Th1 (IFN-γ) cytokines, although OX40+ Th2 IL-4-producing cells outnumbered OX40− cells by 2.6:1, whereas the reverse was true for IFN-γ-expressing cells where OX40− CD4 cells outnumbered OX40+ cells by 2.2:1.

Th2 but Not Th1 Cells Preferentially Survive When Cocultured with CD4+CD3− Cells

To investigate directly the effects of CD4+CD3− cells on T cells, we primed OX40+ and OX40− OTII cells (initial ratio 1:1) under Th1 and Th2 conditions in the presence and absence of CD4+CD3− cells (Figure 4B). As in previous experiments, under Th1 conditions, OX40− OTII cells outnumbered their OX40+ counterparts by ∼2:1, and this was not changed by adding CD4+CD3− cells (Figure 4B). However, on day 6 T cells cultured with CD4+CD3− cells did show a slight reduction in forward scatter (FSC), and this was independent of OX40 signals. This reduction in FSC was more marked on Th2 cells, but again this was independent of signaling through OX40. However, the survival of OX40+ Th2 cells but not Th1 cells was clearly augmented by coculture with CD4+CD3− cells (Figures 4B and 4C). The reason for the preferential survival of OX40+ Th2 cells with CD4+CD3− cells was attributable to decreased annexin-positive apoptotic cells in this population (26% versus 43%). In contrast, the percentage of annexin-positive cells in OX40+ and OX40− Th1 cells was similar (29% versus 30%) (Figure 4B).

OX40+ CD4 Cells Associate with CD4+CD3− Cells In Vivo

OX40+ CD4 T cells form a minor fraction of the total CD4 cell pool in normal mice (in our hands approximately 5%). To identify the cellular associations of these OX40+ CD4 cells, we developed techniques to stain for OX40 by confocal microscopy, using tissue sections from OX40− mice to confirm specificity of staining (data not shown). Figure 1E shows a low-power and the inset shows a high-power image of an OX40+ CD4 cell interacting with a CD4+CD3− cell. In separate experiments, we found 17% and 29% of OX40+ CD4 cells to have membrane associations with CD4+CD3− cells. This demonstration of the physical association of CD4+CD3− cells and OX40+ CD4 cells is proof of the opportunity for OX40 signaling between them.

To visualize the functional relevance of intact OX40 signals to primed transgenic CD4 T cells, we transferred allotype marked OX40+ and OX40− transgenic OVA-specific T cells into Rag1− recipients and compared their survival after immunization with the Th2 antigen, alum-precipitated OVA (Figure 5A). Following immunization, there was rapid expansion of both OX40+ and OX40− transgenic T cells, and the majority of transferred T cells downregulated expression of CD62L (L-selectin). Initially, survival was not OX40 dependent, but after a week there was preferential survival of OX40+ cells; after 4 weeks they outnumbered their negative counterparts by 3:1 (Figure 5B). This preferential survival was associated with stable expression of OX40 on ∼40% of CD4 T cells (Figure 5A), a much higher proportion than in normal mice, reflecting the fact that virtually all CD4 cells are activated in this model. Analysis of where these OX40+ CD4 cells were located revealed that 53/173 (∼31%) were associated with CD4+CD3− cells (Figure 1F). Of these cells the great majority (50/80 [∼63%]) were associated with CD4+CD3− cells in the outer T zone and B cell areas. Of the 93 OX40+ cells remaining in the inner T zone, only three (∼3%) were associated with CD4+CD3− cells. By comparing the ratio of OX40+ to OX40− T cells in B and T cell areas (Figure 5C), it is also clear that the increased survival of OX40+ cells is associated with their location in B follicles (median OX40+/OX40− ratio 12.08) and the outer T zone (median OX40+/OX40− ratio 4.02). In contrast, their survival in the inner T zone is relatively independent of OX40 signals (median OX40+/OX40− ratio 1.36).

OX40 Signals Are Critical for Sustaining T Cell Memory for Secondary Antibody Responses

The above data showed direct interactions between CD4+CD3− cells and OX40+ CD4 T cells at the B:T interface and in B follicles, and this was correlated with the preferential accumulation of OX40+ cells at these locations. To test whether OX40 signals to T cells were required to maintain T help for B cells, we investigated antibody production to protein antigens in mice sufficient and deficient in OX40 (Figure 6). There was little difference in the primary IgM and IgG antibody responses in the two groups of mice (Figure 6Aa), but after boosting (Figures 6Ab and 6Ac), median IgG- and IgM-specific antibody titers were reduced in OX40-deficient mice.

We reasoned that if OX40− signals from CD4+CD3− cells were important in keeping memory T cells alive, OX40-deficient mice would have an impaired capacity to provide T cell help to a new B cell determinant on the carrier protein. To test this, we analyzed the capacity of mice primed with a carrier protein to provide help to NP-specific B cells following immunization with NP conjugated to the carrier protein. Again, both IgG and IgM anti-NP responses were impaired in OX40-deficient mice compared to controls (Figure 6B).

These data pointed to defective T cell memory for B cell help. To pinpoint further the defect in OX40-deficient mice, we exploited a system in which we can measure the capacity of a cohort of memory T cells to provide help for secondary antibody responses. We use mice deficient in CD28 signaling by virtue of transgenic expression of CTLA-4Ig (CTLA-4Ig+) (Lane et al., 1994). Injection of an agonistic CD28 mAb into CTLA-4Ig+ mice transiently reverses CD28 blockade and generates a cohort of CXCR-5+ CD4 T cells, which induce GCs and the production of high-affinity IgG antibodies (Walker et al., 1999). Clearance of the CD28 mAb (via a mouse anti-hamster CD28 mAb antibody response) restores CD28 blockade and effectively blocks the recruitment of new T cells into antibody responses, as naive T cells are dependent on CD28 signaling (Lane et al., 1994). Comparison of the effects caused by injecting CD28 mAb into OX40-deficient or -sufficient CTLA-4Ig+ mice provides a direct readout of the contribution of OX40 signals to T cell memory for antibody production (Figure 6C). There was little difference in primary (day 7) IgM and IgG antibody responses in the two groups of mice. However, sustained IgG antibody production (day 46) was markedly impaired in CTLA-4Ig+ mice deficient in OX40 signaling.

The conclusion that OX40 signals are required to sustain T cell memory for long-term antibody production was independently verified by transferring either OX40+ or OX40− transgenic T cells into Rag1− recipients and testing whether they could provide help for IgG responses following immunization with NP-ovalbumin (Figure 6D). As in the above experiment, mice that received OX40− T cells showed a marked deficiency in their capacity to sustain IgG antibody responses.

Discussion

In this paper, we have characterized an accessory cell that interacts with antigen-specific CD4 T cells after priming on DCs and continues to be associated with memory T cells 4 weeks after priming. These CD4+CD3− accessory cells are distinct from conventional CD11c+ DCs known to present antigen to CD4 T cells in the T zone and are specifically located at sites of T:B collaboration: at the B:T interface and in B follicles. Their localization was not dependent on immunization or on intact signaling through CD28 and OX40. We also found them in Rag1− mice, suggesting that their presence was not dependent on lymphocytes.

Phenotypically, the CD4+CD3− cells that we isolated closely resembled those identified in neonatal mice (Mebius et al., 1997). These cells affect the recruitment and organization of lymphoid cells in lymph nodes and Peyer’s patches by virtue of expression of lymphotoxin α1β2 (Futterer et al., 1998). Unlike conventional DCs (Pulendran et al., 1997), our CD4+CD3− cells were not expanded by Flt3 ligand, nor were we ever able to demonstrate antigen presentation to naive T cells (data not shown). They were also clearly different from the recently described B220+ population of CD4+CD11clowCD3− plasmacytoid cells that secrete large quantities of type I interferons in responses to viral stimuli Asselin-Paturel et al. 2001, Nakano et al. 2001 and which are expanded by Flt3 ligand (Gilliet et al., 2002).

Isolated CD4+CD3− cells expressed little CD80 or CD86, costimulatory molecules expressed by DCs, but they did express high levels of the TNF ligands, OX40 ligand, and CD30 ligand, although not 4-1BB ligand and CD27 ligand, after in vitro culture. Freshly isolated cells expressed high levels of mRNA encoding these proteins, indicating that they had the potential to express these proteins in vivo. It has been reported that primed Th2 cells preferentially migrate to sites where CD4+CD3− cells are located (Randolph et al., 1999) suggesting that they might receive signals from this cell type. In support of this, we found CD30 and OX40 were preferentially expressed on Th2 cells cultured in vitro and found that survival of Th2 but not Th1 cells was substantially augmented by OX40 signals from CD4+CD3− cells.

Our in vitro data was further supported by our observations that OX40-expressing T cells primed by a Th2 immunogen remain strongly associated with CD4+CD3− cells in the outer T zone and B cell areas 4 weeks after priming and that the preferential survival of CD4 cells that could signal through OX40 Chen et al. 1999, Kopf et al. 1999, Murata et al. 2000, Rogers et al. 2001 occurred where CD4+CD3− cells were located, in the outer T zone and B follicles.

We did find significant numbers of OX40-expressing T cells in the inner T zone where OX40 ligand-expressing CD11c+ cells have been found (Malmstrom et al., 2001). However, in our studies we found that the survival of primed CD4 cells in the central T zone was essentially OX40 independent.

The role of CD30 ligand expressed by CD4+CD3− cells is currently unknown, but it is possible that there is a level of redundancy between OX40 and CD30. We found that the reduction in FSC in Th2 cells cultured with CD4+CD3− cells did not depend on OX40 signals. However, as CD30 expression is normal on OX40-deficient Th2 cells, it is possible that signaling through CD30 is responsible for the decrease in FSC. This is currently under investigation.

In summary, we propose an additional role for the CD4+CD3− cells already shown to support the organization of lymphoid tissue. We suggest that primed CD4 T cells migrate to CD4+CD3− cells where they receive OX40 and CD30 signals. This interaction could be a checkpoint to ensure that T cells are fully committed to the provision of B cell help. In addition we suggest that these cells provide OX40 survival signals to memory T cells when antigen is scarce, so that they are subsequently available for secondary responses.

Experimental Procedures

Mice

Normal, CD40 ligand-deficient and Rag1-deficient (Rag1−) C57Bl/6 mice were bred and maintained in our animal facility. OX40-deficient mice (Kopf et al., 1999) were crossed with the ovalbumin (OVA)-specific TCR transgenic (OT-II) mice (Barnden et al., 1998) and were homozygous for the expression of CD45.2. Homozygous OT-II mice were also crossed with CD45.1 congenic C57Bl/6 mice to obtain allotype-marked OVA-specific transgenic T cells that could be distinguished in cultures from OX40-deficient CD45.2 OTII cells. OX40-deficient mice (Kopf et al., 1999) were crossed with the CTLA-4Ig transgenic (CTLA-4Ig+) mice. OX40-deficient and -sufficient CTLA-4Ig+ mice were obtained from the F1 intercross. All mice were used between 6 and 16 weeks of age and were sex and age matched as far as possible in individual experiments. All mice were housed and maintained in the University of Birmingham, Biomedical Services Unit. The spleens from the CD132− Rag1− and CD132+ Rag1− mice were provided by Dr. B. Stockinger (Mill Hill).

Cell Transfer and Immunization

107 splenocytes (∼1–2 × 106 CD4 T cells) from CD45.1 OTII mice were injected i.p. into congenic CD40 ligand-deficient CD45.2 recipients. Twenty-four hours after cell transfer they were immunized i.p. with either 200 μg ovalbumin (OVA) (Sigma, Dorset, United Kingdom) or NP-OVA precipitated in a commercial alum-containing gel, Alu-Gel-S (SERVA, Heidelberg, Germany). Similar transfers were performed using Rag1− mice as recipients and/or by injecting mixtures of OX40+ and OX40− splenocytes.

For other immunizations, mice were immunized with either 100 μg chicken γ-globulin (CγG) (Jackson Immunoresearch via Stratech, Luton, Bedfordshire, United Kingdom) or 100 μg NP-CγG i.p. For priming, proteins were precipitated with Alu-Gel-S as above. Boosts were with soluble protein dissolved in PBS (pH 7.4).

CTLA-4Ig+ mice deficient or sufficient for OX40 were immunized as above except that 100 μg of anti-CD28 mAb was included with the priming protein (clone 37.51, a kind gift of Dr. Jim Allison, Berkeley, CA). Boosts were with antigen alone, and no additional anti-CD28 mAb was given.

ELISA

Plates were coated with 5 μg/ml NP-bovine serum albumin or 10 μg/ml CγG diluted in 0.05 M carbonate buffer (pH 9.6). Serum samples were titrated on ELISA plates blocked with 1% bovine serum albumin in PBS (pH 7.4). Bound IgG and IgM were detected using alkaline phosphatase-conjugated goat anti-mouse IgG and goat anti-mouse IgM (Southern Biotechnology Associates, Birmingham, AL), developed with p-nitrophenylphosphate substrate (KPL, Gaithersburg, MA). The absorbance was read at 405 nm.

Antibodies for Staining

Anti-CD4 PE, anti-CD45.1 FITC, and anti-B220 FITC mAbs and biotinylated mAbs against CD11c, mouse I-Ab, CD80, CD86, CD132, OX40, OX40 ligand, CD30, CD30 ligand, CD62L, IFN-γ, IL-4, and IL-10 were purchased from Pharmingen (Oxford). Anti-CD8α FITC mAb and anti-CD8α PE mAb were obtained from CALTAG laboratories (Burlingame, CA). Biotinylated anti-CD40 mAb was a kind gift of Dr. Ton Rolink (Basel Institute for Immunology), and the fusion proteins mCTLA4Ig and mOX40Ig were made as described elsewhere Brocker et al. 1999, Lane et al. 1994. The macrophage markers F4/80 and MARCO (macrophage receptor with collagenous structure) were obtained from Serotec (Oxford).

As the second-step staining reagents, streptavidin FITC and streptavidin PE were purchased from DAKO (High Wycombe, United Kingdom), and streptavidin cychrome was purchased from Pharmingen. Goat anti-human IgG FITC was obtained from Southern Biotechnology Associates for the second-step antibody in conjunction with the fusion proteins.

Immunohistology for Confocal Microscopy

Spleens from normal and injected mice were snap frozen in liquid nitrogen and stored in sealed polythene bags at −70°C until use. Five micrometer cryostat sections of the spleens were mounted on four-spot glass slides for immunohistology and confocal microscopy.

Sections to be examined by confocal microscopy were stained with fluorochrome-conjugated antibodies diluted in PBS containing 1% bovine serum albumin. Slides were preblocked with 10% goat serum, and anti-rat and anti-hamster second-step reagents were preabsorbed with normal mouse serum. Armenian hamster anti-CD3, hamster anti-CD11c, rat anti-CD4, and rat anti-CD8 were purchased from Pharmingen. Goat anti-rat TRITC was purchased from Southern Biotechnology Associates, and goat anti-Armenian hamster Cy5 and goat anti-mouse IgM AMCA were purchased from Jackson Laboratories Inc. (via Stratech).

To stain OX40+ CD4 T cells, we used biotinylated anti-OX40 mAb (Pharmingen), followed by goat anti-biotin FITC (Vector Labs, CA). This signal was further amplified using rabbit anti-FITC and goat anti-rabbit FITC third and fourth stages. Stained sections were mounted with DABCO in glycerol at pH 7 and examined by confocal microscopy.

Image Acquisition and Analysis of Confocal Images

Confocal images were acquired using a Zeiss LSM510 laser scanning confocal head with a Zeiss Axiovert 100M microscope. Digital images were recorded in four separately scanned channels with no overlap in detection of emissions from the respective fluorochromes. FITC-conjugated antibodies were excited with the 488 nm line of the argon laser, and emissions were collected in a channel defined by a 505–550 nm bandpass filter. TRITC antibodies were excited with a 543 nm helium laser, and emissions were collected in a channel defined by a 560–615 nm bandpass filter. Cy5-conjugated antibodies were excited with the 633 nM helium laser, and emissions were collected using a 650 nm longpass filter. Finally, AMCA-conjugated antibodies were excited with an Enterprise ultraviolet laser with 351 and 364 nm excitation lines, and emissions were collected in a channel defined by a 385–470 nm bandpass filter.

To identify positively cells that expressed CD4 but lacked CD3 or CD11c (CD4+CD3− CD11c− cells), sections were stained with goat anti-mouse IgM AMCA, rat anti-mouse CD4 identified with goat anti-rat TRITC, and the hamster antibodies to CD11c and CD3 identified with goat anti-hamster Cy5. Staining for CD3 and CD11c mAbs was optimized individually.

Images were processed using the Zeiss LSM510 software, which allows pixel by pixel manipulation of channels within an image. The Zeiss confocal software stores a standard 8 bit (pixel intensities range from 0–255) array of 1024 × 1024 pixels for each channel recorded. To identify positively cells that expressed CD4 but not CD3 or CD11c, the matrix containing values for CD3 and CD11c was divided into the matrix containing values for CD4. In the resulting matrix, pixels carrying only CD4 remain positive, and pixels coexpressing significant levels of CD3 or CD11c are reduced to background levels.

For positive identification of CD11c+ cells, sections were stained with goat anti-mouse IgM AMCA, and the hamster antibodies to CD11c identified with goat anti-hamster Cy5.

In experiments involving the transfer of mixed OX40+ and OX40− lymphocytes, OX40+ cells in spleen sections were positively identified by staining for the congenic CD45.1 marker, CD4 and CD3. OX40− T cells were those CD4+CD3+ cells not staining for CD45.1. B cell areas were defined by staining with IgM (B). The outer T zone (B:T) was defined as an approximately two cell-wide area at the T zone border. The inner T zone was the remaining T cell area (T). More than 200 OX40+ and OX40− T cells were counted from at least three regions of randomly selected white pulp from each of three mice.

Preparation of CD4+CD3− Cells and DCs Enriched Populations for FACS Staining and FACS Cell Sorting

Cell suspensions were made from the spleens of Rag1− and common cytokine receptor γ chain deficient Rag1− mice as follows. Spleens were cut into small (8 mm3) fragments and then cultured in serum-free RPMI with collagenase D (1 mg/ml) (Boehringer Mannheim) for 45 min at 37°C. Digested fragments were crushed between gauze. After depletion of red blood cells with Gey’s solution, the cell suspensions were pooled and resuspended in MACS buffer containing 10% mouse serum as per the manufacturer’s instructions. CD11c+ cells were enriched by incubating the cell suspension with MACS anti-mouse CD11c microbeads (Miltenyi Biotec Ltd., Surrey, United Kingdom) at 100 μl/108 cells for 15 min at 4°C. CD11c+ bead-attached single cells were positively selected using a midiMACS separation unit according to the manufacturer’s protocol. CD4+ cells from CD11c− cells were enriched as above except that anti-CD4 coated microbeads were used (Miltenyi Biotec Ltd.). The resulting populations were stained after overnight culture at 37°C with appropriate antibodies and analyzed by flow cytometry using a FACScan (Becton Dickinson, Mountain View, CA).

Purified CD4+CD3− cells and DC subpopulations were prepared by cell sorting using a FACSVantage (Becton Dickinson): (1) CD4+CD11c−, (2) CD8−CD11c+, and (3) CD8+CD11c+ populations. Purified populations were >95% pure.

Semiquantitative RT-PCR

Total RNA was prepared from FACS-sorted cell subsets using RNAzol B according to the manufacturer’s instructions (Biogenesis, Poole, United Kingdom). cDNA enriched for mRNA precursors was prepared using avian reverse transcriptase (Life Technologies, Paisley, United Kingdom) and oligo dT primers (Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom). cDNA was diluted to a final volume of 100 μl. Gene expression was quantified using the ABI Prism 7700 sequence detection system (PE Biosystems, Warrington, United Kingdom). By amplifying TNF ligands and β-actin in the same PCR tube, variations in cDNA preparations were corrected for, so that levels of TNF ligands between cell subsets could be compared directly. Signals from the specific probes conjugated to the dye VIC (for β-actin) or FAM (6-carboxyfluorescein) (for TNF ligands) were measured at the end of each of the 44 PCR cycles to generate sigmoid plots for TNF ligands and β-actin for each sample. A threshold was set on the exponential phase of the sigmoid plots, and the cycle number (CT) at which each sample crossed this threshold was recorded. The relative concentration of TNF ligands in the samples was compared by correcting for the β-actin signals.

The specific primers and probe sequences were synthesized by PE Biosystems.

β-actin

Forward primer, 5′(CGTGAAAAGATGACCCAGATCA); reverse primer, 5′(TGGTACGACCAGAGGCATACAG); VIC probe, 5′(TCAACACCCCAGCCATGTACGTAGCC).

OX40 Ligand

Forward primer, 5′(ACGCTAAGGCTGGTGGTCTCT); reverse primer, 5′(TGCTCCTCTGAGTCTTTGGATTG); FAM probe, 5′(CTGTGCTTCATCTATGTCTGCCTGCAACTCT).

CD30 Ligand

Forward primer, 5′(AGGATCTCTTCTGTACCCTGAAAAGTA); reverse primer, 5′(GTTTGGTATTGTTGAGATGCTTTGA); FAM probe, 5′(CAAGAAGTCATGGGCCTACCTCCAAGTG).

4-1BB Ligand

Forward primer, 5′(GGCCGCCTGCACTGAA); reverse primer, 5′(CAGAAGCAGCAAAACCAAAGC); FAM probe, 5′(TCTGTTCCCGCCACCCAAAGCTCT).

CD27 Ligand

Forward primer, 5′(CACGGAGATGTCCTCTGTACCA); reverse primer, 5′(TGAACTCCAAAGAAGGTCTCATCA); FAM probe, 5′(CCTCTGCTGCCGTCCCGCA).

Flt3 Ligand Treatment

Rag1− mice were injected i.p. with recombinant human Flt3 ligand (a kind gift from the Immunex Corporation, Seattle, WA). Flt3 ligand or PBS was injected on days 0, 2, 4, and 7. Mice were sacrificed on day 8, and suspensions of spleen cells were prepared as described above. A total of 106 cells were stained with anti-CD4 PE, anti-CD8α FITC, and biotinylated anti-CD11c and streptavidin cychrome. The numbers of each cell subset isolated/spleen were calculated by measuring by FACS staining the proportion of CD4+CD11c−, CD8+CD11c+, and CD8−CD11c+ cells, and multiplying this percentage by the total numbers of viable cells obtained from each spleen.

T Cell Stimulation and Intracellular Cytokine Staining

For Th0, Th1, and Th2 polarized cell cultures, splenocytes were collected from OX40-sufficient (CD45.1) and -deficient (CD45.2) OT-II mice. OX40+ and OX40− cells were mixed 1:1 (total 2 × 106 cells/ml/well), and the final ratio of CD4 T cells was determined by flow cytometry. Mixtures were incubated in RPMI 1640 medium containing L-glutamine (GIBCO BRL), and 10% FCS (GIBCO BRL) with 1 μM/ml ovalbumin 323-339 peptide at 37°C (Th0 conditions). For Th1 conditions, 10 ng/ml IL-12 (Peprotech EC, Ltd., London) and 10 μg/ml anti-IL-4 mAb were added to cultures, and for Th2 conditions, 10 ng/ml IL-4 (Peprotech EC, Ltd.) and 10 μg/ml anti-IL-12 mAb were added. All cells were cultured for 6 days and then stained to detect the surface expression of OX40 and CD30. For intracellular cytokine staining, cultures were restimulated for 4 hr at 37°C in the presence of GolgiStop (following the manufacturer’s protocol, Cytofix/Cytoperm Plus cytostain kit, Pharmingen) on anti-CD3− coated plates (coated overnight with 5 μg/ml anti-CD3 mAb at 4°C).

After restimulation, the cells were surface stained for CD4 and CD45.1, and then fixed and permeabilized according to the manufacturer’s protocol for the cytokine kit. The cells were then stained optimally with mAbs (listed above) to detect intracellular IL-4, IL-10, and IFN-γ.

Annexin V Staining

Surface molecules were optimally stained with CD45.1 FITC and CD4 CyChrome. The cells were washed twice with cold PBS, resuspended in binding buffer (Pharmingen) at a concentration of 106 cells/ml, and stained with an optimal amount of Annexin V PE (Pharmingen) for 15 min at RT in the dark. Annexin V was detected by flow cytometry within 30 min of staining.

Acknowledgements

This work is funded by the Wellcome Trust. H.E.W. is supported by an endowment fund fellowship. We would like to thank Manfred Kopf for sending us the OX40-deficient mice and also Bill Heath and Frank Carbone for allowing us to obtain the OTII transgenic mice. Recombinant human Flt3 ligand was a kind gift of the Immunex Corporation. Jim di Santo kindly gave permission to use the CD132− mice, and Gitta Stockinger sent us spleens from Rag1− and Rag1− CD132− mice for our analysis. These experiments could not have been done without the expert help of the staff of the Birmingham University animal facility (BMSU), particularly John Atkinson.

References

References

https://doi.org/10.1016/S1074-7613(03)00110-9