Functional activity of CD40 antibodies correlates to the position of binding relative to CD154

Abstract

In this study we describe the characterization of a panel of 12 anti-mouse CD40 monoclonal antibodies (mAb). Characterization was performed in terms of antibody-binding site relative to the CD154 ligand, and the relationship between position and functional outcome of binding. The antibodies divided into three groups. The first were strong inhibitors of CD154 binding, and induced strong proliferative and activation signals to B cells. Two antibodies gave intermediary inhibition and comparable levels of activation. The remaining antibodies were found to bind outside the CD154 binding site and were poor inducers of B-cell activation. Data presented show a strong correlation between location of mAb binding and the resultant activation signal delivered. This correlation is shown to be independent of the isotype of the antibody involved and of its affinity. Implications of these findings are discussed.

Introduction

The CD40 antigen was first discovered in 1985 by Paulie et al.¹ on bladder carcinoma cells. Since its discovery, this antigen has been extensively studied and is recognized as having a central role in the control of T-dependent B-cell responses. The natural ligand for CD40, known as CD154 (CD40L, gp39) was discovered in 1992 and is primarily expressed on activated CD4 positive T lymphocytes.² Cognate interactions between this receptor–ligand pair are crucial for isotype switching and germinal centre formation, as illustrated by hyper-immunoglobulin M (IgM) syndrome^3,4 (see ^5–9 for reviews).

The first anti-CD40 monoclonal antibodies (mAb; designated S2C6 and G28-5) were described in 1985¹ and 1986¹⁰ and are tools that have been used for many years in the study of human CD40 functions. Since the production of these mAb, many more anti-human CD40 antibodies have been described and characterized. These include mAb89,¹¹ 17 : 40,¹² Ro1, 2, 3, 4 and 5,¹³ 5C3 and EA5.¹⁴ Work with these antibodies has produced interesting data on the possible outcomes of CD40 ligation, particularly regarding the quantitative and qualitative differences in responses following binding to different epitopes.

Bjorck et al.¹² illustrated that the 17 : 40 mAb shared the stimulatory ability of other previously described antibodies, but that this effect was directed through a distinct epitope. Further work by these researchers has also shown qualitative differences in stimulation using S2C6 and mAb89.¹⁵ Similarly, Pound et al. have illustrated distinct outcomes of CD40 stimulation with regards suppression of apoptosis, homotypic adhesion¹⁶ and regulation of CD23 expression and IgE synthesis.¹⁴ In addition, work by Schwabe et al. points to the possibility of three CD40 epitopes.¹³ Antibody against the first is able to act in a stimulatory manner, binding outside the CD154 binding site and enhancing effects mediated by CD154. Antibody against the second, acting as a stimulatory signal but binding within the CD154 binding site, while antibodies that bind the third epitope, outside of the natural ligand binding site, induce no stimulatory signal, unless cross-linked, indicating the importance of receptor aggregation.

The first mAb against mouse CD40 were described by Heath et al.¹⁷ As with studies performed using mAb against human CD40, this work illustrated the importance of the site of antibody binding to the outcome of stimulation. The two mAb described in that paper (1C10 and 4F11) define distinct epitopes and possess properties similar to those described for anti-human CD40 mAb. Binding to the first epitope (as recognized by 1C10), leads to direct proliferation, activation and rescue from apoptosis while binding to the second (as recognized by 4F11), is unable to induce any of these effects alone but is capable of synergizing with 1C10. In contrast to the extensive characterization of anti-human CD40 mAb and their use in elucidating functional outcomes of CD40 ligation, work on anti-mouse CD40 mAb has been limited. This is largely due to the restricted number of appropriate reagents. Given that the mouse represents a well-studied immunological model, in depth investigation into the outcomes of CD40 stimulation is an important and somewhat neglected area of research. In this paper we describe the various effects of a panel of anti-mouse CD40 mAb, with particular emphasis on the differences mediated with respect to their position of binding relative to the natural ligand.

Materials and methods

Production and purification of anti-CD40 mAb

Rat anti-mouse CD40 hybridomas were produced as described previously.¹⁷ Anti-CD40 mAb producing hybridoma cell lines were expanded in Integra CL-6 growth chambers (Integra Biosciences Ltd, Letchworth, Hertfordshire, UK). This apparatus consists of two compartments separated by a permeable membrane that allows transfer of nutrients to the cell compartment but retains secreted antibody within the growth chamber, thus allowing production of higher concentration supernatants. Once sufficient antibody-containing supernatant had been collected mAb were concentrated and purified by standard techniques.

Briefly, following 50% ammonium sulphate precipitation, antibody was purified on protein G columns (Pharmacia Biotech, Uppsala, Sweden) and antibody containing fractions pooled and dialysed against phosphate-buffered saline (PBS). Purified antibody was concentrated to approximately 2 mg/ml using 30 000 MW cut-off centrifugal filters (Millipore, Bedford, MA), filter sterilized and stored at −20° until used.

Isotyping mAb

The isotype of each mAb was determined using a commercially available enzyme-linked immunosorbent assay (ELISA)-based kit (Zymed, San Francisco, CA).

Titration of mAb

A titration of each mAb was performed in order to determine functional affinity of mAb and to ensure that the concentration (10 µg/ml) used for subsequent experiments was saturating. CD40 transfected fibroblasts and untransfected (L929) control cells were incubated with mAb samples at a range of dilutions (10 µg/ml to 0·1 ng/ml). Following three washes with fluorescence-activated cell sorting (FACS) buffer (0·01% sodium azide, 3% bovine serum albumin (BSA) in PBS), cells were incubated with goat anti-rat immunoglobulin fluorescein isothiocyanate (FITC)-conjugated antibody (Pharmingen, San Diego, CA) and mean fluorescent intensity (MFI) determined by flow cytometry. MFI was plotted against antibody concentration to produce a titration curve for each antibody.

Determination of relative affinity of mAb

Relative affinity of mAb was determined using a flow cytometric based technique as reported by Benedict et al.¹⁸ This technique has the advantages of being rapid and simple to perform, without the requirement for large quantities of purified antigen. Affinity was calculated from the linear region of titration curves (as described above) using the following equation: $K_D=[1/(f–f_{\mathrm{back}})–1/f_{\max }\times [\mathrm{Ab}]\times f_{\max }$ where f is fluorescence; f_back is background fluorescence; f_max is maximum fluorescence (determined from mAb titration curve); [Ab] is antibody concentration (ng/ml) for value f.

Inhibition of CD154 binding

The binding site of each mAb, relative to the natural ligand (CD154), was determined by a competitive binding assay and subsequent FACS analysis. CD40 expressing L cells were incubated with each of the mAb at 10 µg/ml for 20 min on ice. An equal volume of CD154–CD8-containing supernatant¹⁹ (1/2 final dilution, shown to be a saturating concentration. Data not shown) was then added to each reaction vessel and left for a further 20 min. Cells were then washed and incubated with anti-CD8 FITC and analysed using a FACScalibur flow cytometer and Cell Quest software (Becton-Dickinson, Mountain View, CA). Each sample was tested three times and mean values for each mAb determined. Percent inhibition was calculated as below:

Activation of B cells

Splenocytes (2 × 10⁵ per well) were incubated with each of the mAb at 10 µg/ml for 24 hr at 37° in 96-well cell culture plates (Nalge NUNC International, Roskilde, Denmark). Samples were then double-stained using phycoerythrin (PE)-conjugated anti-B220 and FITC-conjugated antibodies for the following B-cell activation markers; major histocompatibility complex (MHC) II, intracellular adhesion molecule-1 (ICAM-1) and CD23 (Pharmingen, San Diego, CA). Samples were then analysed by flow cytometry and the percent increases in MFI of the activation markers calculated for the B220 positive population. This experiment was repeated three times and mean values for each sample calculated.

Proliferation of B cells

Spleen cell suspensions were prepared as described above and the media supplemented with murine IL-4 (DNAX Research Institute, Palo Alto, CA) at 200 U/ml. Samples (in triplicate) were incubated for 48 hr at 37°. During the last 6 hr of incubation cells were pulsed with tritiated thymidine (New England Biolabs, Hitchin, Hertfordshire, UK) at a concentration of 1 µCi/well. Cells were then harvested for scintillation counts using a Beckman Coulter Top Count (Beckman Coulter (UK) Ltd, High Wycombe, Buckinghamshire, UK). Proliferation rate, expressed as incorporation of radioactive thymidine, was measured and the mean value for each mAb calculated.

Results

Isotyping of anti-CD40 mAb

The isotyping assays showed that the rat mAb in this panel were of three isotypes; five of the mAb were IgG1, four were IgG2a and three were IgM antibodies (see Table 1).

Table 1.

Summary of data obtained from isotyping, affinity determination and inhibition of CD154 binding assays for the panel of anti-CD40 mAb

Determination of antibody affinity

Flow cytometric-based affinity studies of this panel of mAb revealed a range of affinities. Most of the mAb analysed had comparable values, with K_D values between 1·7 × 10⁻¹¹ and 8·0 × 10⁻¹³, with little difference in affinity observed due to the isotype of the antibody. Data from affinity determination experiments are summarised in Table 1. The IgM antibodies were omitted from affinity measurements as the binding valency would affect avidity.

Inhibition of CD154 binding

The ability of each of the mAb to inhibit binding of recombinant CD154-CD8 to CD40-transfected fibroblasts was used to indicate their position of binding relative to the natural ligand. The panel displayed a range of inhibitory activities. Antibodies 1C10, 1D7 and 10C8 gave very high inhibition of CD154 binding (the latter consistently completely inhibiting all binding) indicating that these mAb probably bind epitopes within the CD154 binding site or at a site overlapping it. Antibodies 1G7 and 4A7 were found to give partial inhibition of ligand binding, a phenomenon also demonstrated by others with anti-human CD40 mAb.¹⁶ The remaining antibodies gave low inhibition of CD154 binding, indicating that the epitopes recognized by these antibodies lie largely outside that of the natural ligand. Figure 1 shows example FACS histogram plots for the inhibition of CD154 binding for two of the mAb in the panel. Table 1 summarizes the data obtained from the inhibition assays. Means were based on three experiments and standard deviation is shown in parentheses.

Figure 1.

FACS histogram plot illustrating inhibition of CD154-CD8 binding to CD40 transfected fibroblast cells. The filled histogram represents the positive control (i.e. cells incubated with CD154-CD8 in the absence of blocking reagents) the thin, solid line represents the negative control (i.e. cells incubated with secondary reagents only), the thick, solid line represents complete inhibition by the 10C8 mAb and the broken line indicates partial inhibition by the mAb 1G7.

In vitro effects of anti-CD40 mAb

Spleen cells from BALB/c mice were incubated with each of the mAb in order to determine the effects on B-cell activation and proliferation. As with inhibition data the mAb were found to possess a range of B-cell activating abilities. By comparing results obtained from inhibition assays, proliferation and activation assays it was possible to determine how the position of binding of each antibody affected its in vitro activity. Figure 2 shows combined data from these assays. For induction of CD23 expression (Fig. 2a), antibodies 10C8 and 1C10 proved to be the most potent and these mAb also bind epitopes within the CD154 binding site. That those mAb that bind outside the CD154 binding site are the weakest inducers of CD23 expression, suggests a strong correlation between site of binding and functional outcome. Non-parametric correlation analysis using Spearman's correlation coefficient (r_s) gives a value of 0·810 for CD23 induction (critical value for n = 12 is 0·576). This indicates a correlation between inhibition of CD154 binding and induction of CD23 expression (P < 0·05). Similarly, increase in ICAM-1 expression on B cells shows a correlation between these variables (r_s = 0·613, see Fig. 2b). Interestingly, induction of MHC class II expression (Fig. 2c) gives a less linear relationship (r_s = 0·467, P > 0·05) largely due to the mAb 8F4. Despite this antibody binding an epitope outside the CD154 binding site, it still induces a large increase in MHC II expression. Figure 2(d) shows data obtained from the proliferation assay, and again a range of proliferative signals are demonstrated by these antibodies. As with activation data, a strong correlation between position of mAb binding relative to CD154 and induction of proliferation is apparent (r_s = 0·598). Comparisons of antibody affinity for CD40 expressed on the surface of transfected fibroblasts and the ability of these antibodies to inhibit CD154 binding to this same antigen reveal that inhibitory activity is not dependent on affinity nor is the antibody's ability to induce expression of CD23, ICAM-1 and MHC II on B-cells. Thus, it appears that the major factor influencing signalling through CD40 by these mAb is the relative position of binding.

Figure 2.

Scatter plots showing relationship between the ability of each mAb to induce expression of various B cell activation markers and its position of binding relative to the natural ligand, as indicated by percent inhibition of CD154 binding. The superimposed lines indicate curve fit as determined by Spearman's correlation coefficient. The critical r_s value (P < 0·05) for a sample size of 12 is 0·576. (a) The activation/position relationship for CD23 expression (r_s = 0·810). (b) Data for induction of ICAM-1 expression (r_s = 0·613). (c) Data for MHC II expression (r_s = 0·467, indicating a lack of correlation between these two variables). (d) Data from the B-cell proliferation assays, as determined by tritiated thymidine uptake assay (r_s = 0·598).

Discussion

The data presented here illustrate the importance of the position of mAb binding to CD40 relative to the natural ligand, CD154. It appears that those mAb binding epitopes at or within the binding site of CD154 most resemble the natural ligand in its ability to activate B cells and induce proliferation. Those antibodies binding outside this site do not share this ability, but have been shown to capable of synergizing with CD154, CD154-like mAb or anti-IgM.¹⁷ We have also shown that the ability of these mAb to activate B cells is not dependent on isotype or affinity of the antibody.

The identification of the potent B-cell activator and complete CD154-binding inhibitor 10C8 represents a potentially powerful tool. Work in our laboratory using these mAb as potential adjuvants^20,21 has focused on the 1C10 antibody, which was previously thought to be the most potent anti-mouse CD40 mAb. Use of 10C8 in this work has shown it to be more effective at enhancing antibody responses against T-independent antigens (unpublished observations).

Interestingly, none of the mAb in this study were capable of blocking CD154 without delivering an activation–proliferation signal. Such a mAb could have great value for the study of CD40/154 interactions and would be particularly useful in systems where suppression of CD40 activation is desirable (e.g. autoimmune diseases and repression of transplant rejection^22,23). In an attempt to generate such an activity, 10C8 Fab fragments have been produced in this laboratory and characterization of these fragments is currently in progress. The work presented here illustrates the pleiotropic nature of the function of CD40 stimulation via antibody binding, and we show that the range of outcomes is strongly correlated to the position of antibody binding relative to the natural counterstructure, CD154.

It has been assumed that signalling through CD40, like other tumour necrosis factor receptor (TNFR) family members, is mediated through simple cross-linking of receptors.^24,25 However, our data indicate that mAb binding to certain sites on CD40 are more effective at inducing signalling than mAb binding other sites. This could be due to conformational changes in CD40 induced by binding at these sites. This possibility is supported by limited data indicating that monomeric Fab or Fv anti-CD40 mAb^26,27 or CD154¹⁶ can induce activation. In addition, differences in the effects of CD40 binding by mAb or CD154 also support this possibility.²⁸ A recent paper by Baccam et al. has shown that interleukin-6 production on CD40 binding could be induced by CD154 but not anti-CD40 mAb, while nuclear factor κB activation could be induced by both.²⁹ Different intracellular regions were required for these effects. It is possible therefore that there are cross-linking dependent and independent effects of CD40 ligation.