TcR-induced regulated secretion leads to surface expression of CTLA-4 in CD4⁺CD25⁺ T cells

Abstract

In this study we show that CD4⁺ T cells develop a functional regulated secretory compartment after differentiation into effector cells, as shown by their increased expression and T-cell receptor-induced exocytosis of lysosomal and cytotoxic effector proteins. We tested the hypothesis that activation-induced surface cytotoxic T-lymphocyte-associated antigen (CTLA-4) expression in CD4⁺CD25⁺ regulatory T cells occurs via a similar regulated secretory pathway. Fluorescence microscopy showed that internal CTLA-4 in these cells was stored in a vesicular compartment distinct from lysosomal vesicles. Rapid activation-induced CTLA-4 surface expression in mouse CD4⁺CD25⁺ T cells is independent of protein synthesis and Rab-27a. When antigen-dependent T-cell–antigen-presenting cell (APC) conjugates were analysed for surface distribution of CD86 on APC, a higher concentration of CD86 molecules was observed in the synapse of APC conjugated to CD4⁺CD25⁺ cells than APC conjugated to CD4⁺CD25⁻ cells. These results demonstrate that fast delivery of mediators by the regulated secretory pathway in CD4⁺ T cells can be used to perform other functions that are not involved in cytotoxic function but that can influence/regulate other cells.

Introduction

Effector function in T lymphocytes is mediated by secretion of cytokines, chemokines or cytotoxic proteins that regulate the function of other cells. Like other cells, lymphocytes utilize two pathways for secretion: the ‘constitutive’ secretory pathway, in which newly synthesized proteins are released without delay by exocytosis of small vesicles after golgi processing and the ‘regulated’ secretory pathway, in which protein mediators are stored in larger intracellular vesicles/granules until activation signals their exocytosis.¹ In CD8⁺ T lymphocytes and natural killer cells the regulated secretory pathway has been closely associated with cytotoxic function, with the major cytotoxic mediators perforin and granzymes stored in secretory granules that also contain lysosomal enzymes. Target cell recognition triggers the fusion of the granule membrane and the plasma membrane, exposing the bound target cell to locally high concentrations of cytotoxic mediators that enter the target to cause lethal damage. Recently, we found that memory and effector CD8⁺ T lymphocytes store the chemokine RANTES (Regulated on Activation Normal T-cell Expressed and Secreted) in distinct non-lysosomal vesicles that are rapidly exocytosed after T-cell receptor (TCR) ligation.² Unlike the exocytosis of the lysosomal granules associated with cytotoxicity, RANTES exocytosis is independent of Rab27a, showing that there are two molecularly distinct TCR-triggered regulated secretion pathways in lymphocytes. In addition, neutrophil-derived factors were recently shown to induce rapid surface expression of the chemokine receptor CXCR1 in CD8⁺ T cells via rapid exocytosis from a third distinct intracellular compartment.³ It is clear that regulated secretion is not exclusively devoted to cytotoxicity in lymphocytes.

The functional mechanisms used by the minor CD4⁺CD25⁺ regulatory T-cell subset that can suppress immune responses have not been defined,^4–6 we wanted to explore the possibility that a regulated secretory event could contribute to such immunoregulatory function. CD4⁺ lymphocytes generally show low cytotoxic activity, and the presence of internal compartments that undergo activation-induced exocytosis has not been examined. CD4⁺ suppressor cells have distinctive properties as seen by their expression of surface CD25, the FoxP3 transcription factor, glucocorticoid-induced tumor necrosis factor receptor family-related gene, and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4). The last of these, CTLA-4, is a membrane protein that binds the costimulatory ligands CD80 and CD86 with more affinity than the costimulatory receptor CD28,⁷ resulting in the suppression of immune responses.⁸ Its potential functional role in mediating immune suppression by regulatory T cells has been controversial.^9,10 In CD4⁺CD25⁺ regulatory T cells CTLA-4 is constitutively expressed on internal membranes, which must undergo exocytosis to allow functional surface expression of CTLA-4.

In this study, we have examined regulated secretion in CD4⁺ T cells, involving both exocytosis of the lysosomal vesicles and CTLA-4 derived from its internal stores. We found that CD4⁺ T cells underwent an activation-induced exocytosis of their lysosomal granules, while ex vivo isolated bulk CD4⁺ T lymphocytes had minimal, if any, such activity. However, ex vivo CD4⁺CD25⁺ regulatory T (Treg) cells store CTLA-4 internally in a non-lysosomal compartment that undergoes Rab-27a-independent activation-induced exocytosis. Finally, in conjugates with antigen-presenting cells (APC), TCR-triggered surface expression of CTLA-4 by exocytosis of CTLA-4 vesicles in CD4⁺CD25⁺ T cells allowed significant recruitment of CD86 ligands to the synapse, depleting costimulatory molecules from the remaining APC. We propose that regulated secretion of CTLA-4 is an early event in the regulatory mechanism for suppression of CD4⁺CD25⁻ T cells.

Materials and methods

T cells

Human peripheral blood mononuclear cells (PBMC) were obtained by Ficoll-paque (Amershan Pharmacia Biotech, Pistacaway, NJ) from apheresis preparations of normal healthy donors provided by the National Institutes of Health blood bank. The T-cell blasts were obtained by PBMC culture in RPMI-1640 with 10% fetal calf serum (FCS), antibiotics and non-essential amino acids, in the presence of 2·5 μg/ml of phytohaemagglutinin (Sigma-Aldrich, St Louis, MO) and 20 U/ml recombinant interleukin-2 for 10 days. CD8⁺ and CD4⁺ blast subpopulations were purified by magnetic beads (Miltenyi Biotec, Auburn, CA). Ex vivo CD4⁺ and CD8⁺ T cells were purified from PBMCs by negative selection (Miltenyi Biotec). CD4⁺CD25⁺ and CD4⁺CD25⁻ T-cell subsets from human and mouse lymph nodes, C3H, ashen (generously provided by Dr John A. Hammer 3rd) and TCR,¹¹ were isolated using a positive selection-based CD4⁺CD25⁺ T-cell isolation kit (Miltenyi Biotec). B cells from B10.A mice were obtained from spleens by negative selection (Miltenyi Biotec). Subpopulation purity, tested by flow cytometry, was > 90%.

Degranulation

Purified CD8⁺ and CD4⁺ T-cells blasts were resuspended in degranulation buffer (Hanks’ balanced salt solution + 1 mg/ml bovine serum albumin) at 1 × 10⁷ cell/ml. Aliquots were treated either with medium or with 7·5 μg/ml cycloheximide (CHX) (Sigma-Aldrich). After 1 hr at 37°, cells were washed once with degranulation buffer, and adjusted to 3 × 10⁶ cells/ml with new drug added. Degranulation was stimulated by suspending cells in 24-well plates whose wells were precoated with 10 μg/ml anti-CD3 (clone UCHT1) and 5 μg/ml anti-CD28 (clone CD8.2; BD Pharmingen, San Jose, CA). The supernatants were harvested at different time-points and stored at −20° until assay. Assays of total cell contents were obtained from extracts in 0·2% Triton X-100. β-Hexosaminidase and granzyme A activity was measured by enzymatic assay as described previously (BLTE: N-alpha-benzyloxycarbonyl-L-lysine thiobenzyl esterase) in refs ² and ¹², respectively.

Flow cytometry

Surface staining was performed using fluorescein isothiocyanate-conjugated CD45RA (CD45RA-FITC), phycoerythrin-conjugated CD25 (CD25-PE), allophycocyanin-conjugated CD4 or CD8 (Caltag, Burlingame, CA), or CD27-PE (BD Pharmingen). Internal perforin, granzyme A, granzyme B and lysosomal markers were quantitatively analysed by flow cytometry of permeabilized cells. First, 1 × 10⁶ cells were fixed and permeabilized using Cytofix/Cytoperm (BD Pharmingen) followed by flow cytometry using monoclonal anti-perforin-FITC, anti-granzyme A-FITC, anti-CD107a-FITC, anti-CD107b-FITC and anti-CD63-FITC monoclonal antibodies (BD Pharmingen). In other experiments, CD107a-biotin, CTLA-4-biotin (BD Pharmingen) and granzyme B-biotin (Caltag) were used, followed by streptavidin-Alexa 594 and analysis with a FACScan. For internal expression of CTLA-4 and CD107a in mouse cells, CD4⁺ T lymphocytes were isolated from lymph node surface labelled with CD4-FITC or CD4-PE and CD25-biotin followed by streptavidin-Cychrome (BD Pharmingen). The cells were fixed and permeabilized as above and stained with anti-CD107a-FITC and anti-CTLA-4-PE before analysis with the FACScan.

Degranulation-induced changes in surface marker expression were monitored by flow cytometry of non-permeabilized cells. Purified CD8⁺ and CD4⁺ T-cell blasts were adjusted to 3 × 10⁶ cell/ml and stimulated by incubation on anti-CD3/anti-CD28-coated wells. After various times, cells were harvested and centrifuged at 4° to collect the supernatants. The cells were then surface stained using the CD107a-FITC and CD107b-FITC monoclonal antibodies from BD Pharmingen, with isotype immunoglobulin G (IgG)-FITC used as a negative staining control, and analysed using a propidium iodide gate for live cells. The results were plotted as median fluorescence values against time.

Activation-induced surface expression of CTLA-4 on mouse T cells was carried out with purified CD4⁺CD25⁺ and CD4⁺CD25⁻ T cells suspended in degranulation buffer at 1 × 10⁷ cell/ml and pretreated either with medium or 10 μg/ml CHX for 1 hr at 37°. After washing and resuspending in degranulation buffer at 2·5 × 10⁶ cells/ml with new drug added, cells were stimulated in 96-well plates with medium or 20 ng/ml phorbol 12-myristate 13-acetate and 1 μm ionomycin for 1 and 3 hr. The incubation was performed in the presence of IgG control or anti-CTLA-4-PE (BD Pharmingen).¹³ The reaction was stopped by washing the cells with cold staining medium and surface staining with anti-CD4-Alexa 647 (Caltag) and anti-CD25-biotin (BD Pharmingen) follow by streptavidin-Alexa 594 (Caltag). CD69 upregulation was followed using anti-CD69-FITC and IgG control-FITC (BD Pharmingen). All samples were acquired using gating on propidium iodide-negative cells.

Microscopy

Purified human CD8⁺ and CD4⁺ T-cell blasts were plated on coverslips as described.² Briefly, cells were fixed in suspension with 2% paraformaldehyde in phosphate-buffered saline, and allowed to adhere to coverslips precoated with poly-l-lysine. The coverslips were fixed with 4% paraformaldehyde, and quenched with 50 mm NH₄Cl. Cells were permeabilized using 1% nonidet P-40 in wash buffer (PBS + 3% FCS + 0·01% saponin) for 10 min. The cells were labelled with the following antibodies: FITC-anti-CD107a, FITC-anti-EEA1 (BD Pharmingen) followed by Alexa 488-rabbit anti-fluorescein (Molecular Probes, Eugene, OR); rabbit anti-cathepsin D (Dako, Carpenteria, CA) followed by Alexa 488-donkey anti-rabbit IgG; and anti-RANTES followed by Alexa 488-donkey anti-goat; double stained with anti-CTLA-4 follow by Alexa 568 goat anti-mouse IgG2a.

Mouse CD4⁺CD25⁺ and CD4⁺CD25⁻ T cells were isolated as described above and treated twice with 0·04% Pronase (Calbiochem, San Diego, CA) in Hanks’ balanced salt solution for 15 min to remove surface antibody. The reaction was quenched with RPMI-1640 + 10% FCS. The cells were washed in PBS and allowed to adhere to poly-l-lysine-coated coverslips, fixed with 2% paraformaldehyde and permeabilized. The cells were stained with anti-CTLA-4 (BD Pharmingen) followed by Alexa 488-goat anti-hamster IgG; anti-CD107a (clone 1D4B hybridoma bank, University of Iowa) followed by Alexa 568-goat anti-rat IgG; anti-EEA1 followed by Alexa 568-goat anti-mouse IgG (BD Transduction Lab, San Jose, CA). Secondary antibodies were obtained from Molecular Probes, Eugene, OR. After staining, coverslips were mounted with Prolong antifade (Molecular Probes) and cells were viewed on a DeltaVision imaging system (Applied Precision, Issaquah, WA) as described.²

Conjugate formation

B cells from B10.A mice were pulsed with 5 μm Piogen cytochrome C (PCC) peptide for 2 hr before mixing them with CD4⁺CD25⁺ and CD4⁺CD25⁻ from AND TCR transgenic mice at a 1 : 1 ratio and centrifuging for 3 min to allow the formation of conjugate.¹¹ The conjugates were incubated for 0·5 or 1 hr at 37° and fixed with 2% paraformaldehyde for 15 min. Conjugates were transferred to poly-l-lysine-coated coverslips and fixed for an additional 30 min, followed by two quenching steps of 5 min each with 50 mm of NH₄Cl. The coverslips were stained with Alexa 488 rat anti-mouse CD86 and hamster anti-mouse CD152 followed by Alexa 568-goat anti-hamster. Images were acquired using a confocal LSM510 with z-steps of 0·3 μm. For colocalization analysis, images were collected with a Delta Vision wide-field microscope with z-steps of 0·069 μm; deconvolved with Delta Vision software and reconstructed three-dimensionally as above.

Fluorescence intensity measurements in the synapse were made with metamorph software (Molecular Devices, Sunnyvale, CA). The focal plane with the brightest signal in the synapse region was selected for measurements. Average intensity was measured in three regions of interest: synapse, membrane outside the synapse, and cytoplasm. Background intensity (the cytoplasmic region of interest) was subtracted from the average intensities of the synapse and outer membrane. The ratio of the average intensities with cytoplasmic background subtracted was obtained for (synapse/membrane outside of the synapse) each cell. The data distributions for the experiment and control were plotted as histograms. The mean ratio values for the experiment and the control were statistically evaluated two-sample t-test assuming unequal variances of the conjugates.

Results

Both CD4⁺ and CD8⁺ T cells are capable of activation-induced exocytosis

Human CD4⁺ T-cell blasts were compared to CD8⁺ T-cell blasts with regard to the expression of various proteins of lysosomal cytotoxic granules, as analysed by flow cytometry of permeabilized cells and by biochemical assays (Fig. 1a). The two blast subpopulations had generally similar expression levels of the lysosomal membrane markers CD107a (LAMP1), CD107b (LAMP2); while CD63 (LAMP3) was expressed about 40% less in CD4⁺ than in CD8⁺ T-cell blasts. The lysosomal enzyme β-hexosaminidase was found in lysates of CD4⁺ blasts at approximately 85% of the activity level in CD8⁺ blasts. Comparing cytotoxic mediators by flow cytometry, CD4⁺ blasts had approximately 40% of the levels of granzyme A and granzyme B found in CD8⁺ blasts, a result confirmed by enzymatic assay of granzyme A (BLTE). Perforin expression in CD4⁺ T-cell blasts was also approximately 40% of the level seen in CD8⁺ T-cell blasts. Overall, these results suggest that CD4⁺ T-cell blasts have roughly similar amounts of lysosomal components compared to CD8⁺ T-cell blasts, but the cytotoxic mediators were expressed at somewhat lower levels.

Figure 1.

Lysosomal granule compartment in CD4⁺ T-cell blasts. (a) Expression of intracellular granule markers in CD4⁺ blasts relative to CD8⁺ blasts. Intracytoplasmic staining of the indicated granule markers was analysed in parallel in both blast subsets. Results are expressed as the mean fluorescence intensity of each marker in CD4⁺ relative to CD8⁺ blasts. β-Hexosaminidase and BLTE (granzyme A) were measured by enzymatic assay. (b) CD4⁺ T-cell blasts were cultured on uncoated wells (empty symbols) or stimulated by plate-bound CD3/CD28 monoclonal antibody (mAb; filled symbols). β-Hexosaminidase secretion was measured by enzymatic assay of the supernatants after stimulation and is expressed as a percentage of total enzyme activity in the Triton X-100 lysate. Incubation was carried out with (triangles) or without (diamonds) the protein synthesis inhibitor cycloheximide (CHX). (c) Purified CD4⁺ T-cell blasts were incubated on plate-bound anti-CD3⁺ anti-CD28 for the indicated times, harvested, and live cells were stained with fluorescein isothiocyanate (FITC)-labelled anti-CD107a and analysed by flow cytometry (solid lines). Dashed lines show staining with FITC-isotype control mAb. Histograms are gated on propidium iodide-negative cells. (d) The increase in median channel fluorescence after T-cell receptor stimulation for CD4⁺ (solid triangles) and CD8⁺ (solid squares) T-cell blasts from the experiment in panel (c). Lower panel shows the β-hexosaminidase secretion in this experiment.

To examine whether CD4⁺ T cells have a functional regulated secretion of lysosomal vesicles similar to CD8⁺ T cells, supernatant β-hexosaminidase was measured after stimulation of CD4⁺ T-cell blasts by plate-bound anti-CD3 and anti-CD28 monoclonal antibodies. Secretion via the regulated secretory pathway was distinguished from the constitutive pathway by examining β-hexosaminidase secretion in the presence of the protein synthesis inhibitor CHX, which blocks the constitutive pathway. After activation by the TCR, β-hexosaminidase was detectable in the supernatants of both CD4⁺ and CD8⁺ blasts (data not shown) within 30 min, and increased further with time (Fig. 1b). Activation induced a smaller percentage of the total β-hexosaminidase secretion in CD4⁺ blasts than in CD8⁺ T-cell blasts (12% versus 21%, data not shown). As expected, CHX treatment had a negligible effect on the secretion of this lysosomal marker in either T-cell subset. In accord with Fig. 1a,b, the cytotoxic activity of CD4⁺ blasts in redirected killing assays with Fas negative L1210 target cells was approximately one-sixth that of CD8⁺ T-cell blasts when compared on a lytic units basis (data not shown).

Since granule exocytosis involves a fusion of granule and plasma membranes, internal membrane proteins become expressed on the cell surface, and the surface exposure of the lysosomal membrane protein CD107a has been used to quantify the exocytosis of lysosomal granules.¹³ TCR stimulation increases rapid surface expression of CD107a in CD4⁺ T-cell blasts (Fig. 1c), with kinetics comparable to β-hexosaminidase secretion (Fig. 1d). The homogeneous histograms make it clear that granule exocytosis was not the result of a particular subset of CD4⁺ T-cell blasts. The TCR-induced increase in surface CD107b expression was also detectable, but was lower than that of CD107a, while no increase in surface CD63 was detected (data not shown).

While the above results show that the granule exocytosis pathway operates in CD4⁺ T-cell blasts activated in vitro, it was important to examine T cells directly isolated from human blood. Flow cytometry was used to identify naïve, memory and effector subsets based on the surface expression of CD45RA and CD27, followed by permeabilization and analysis of granule markers (Fig. 2). Expression of the lysosomal membrane protein CD107a showed a single positive peak within each subset, with higher expression levels in the CD45RA⁺ CD27⁻‘effector’ subset of CD8⁺ T cells and also in the few (< 1% of total) CD4⁺ cells with this phenotype. Granzyme B expression was negative in naïve cells and memory CD4⁺ cells, but heterogeneous in the memory CD8⁺ T cells. The CD8⁺ effector cells and some of the parallel CD4⁺ subset had markedly higher granzyme B expression. Perforin expression was not detectable in any CD4⁺ T-cell subset analysed directly ex vivo (data not shown) so unlike blasts, most blood donors have relatively few circulating CD4⁺ T cells that appear to contain internal compartments with cytotoxic potential.

Figure 2.

Lysosomal granule proteins expressed by human CD4⁺ and CD8⁺ T-cell subsets. Four-colour flow cytometry analysis of purified CD4⁺ and CD8⁺ T cells. T-cell subsets, naïve, memory and effector, were defined by surface expression of CD45RA and CD27 and analysed for cytoplasmic content of lysosomal protein CD107a and granzyme B. T-cell subsets were gated on CD4⁺ or CD8⁺ T cells. Dotted blue histogram represents isotype control and red solid histogram represents specific monoclonal antibody CD107 and granzyme B.

Expression of CTLA-4 in human CD4⁺CD25⁺ blood lymphocytes

The CD4⁺CD25⁺‘Treg’ cells express intracellular CTLA-4/CD152,¹⁴ which in activated T cells has been associated with lysosomal granule markers;^15,16 it therefore seemed likely that the same regulated secretory pathway was responsible for CTLA-4 surface expression. The Treg subset was examined by flow cytometry for the intracellular expression of the granule proteins CD107a and granzyme B as well as CTLA-4 (Fig. 3a). Expression of CTLA-4 was clearly positive in most of the cells in the small CD4⁺CD25^high subpopulation, and negative in the much more numerous CD4⁺CD25⁻ subpopulation. It was not detectably expressed on the surface of either of these subpopulations (data not shown). CD107a expression was equivalent in these two subpopulations, while granzyme B was not detectably expressed in either. Furthermore, we examined the nature of the CTLA-4 storage compartment by deconvolution microscopy. CTLA-4 was detectable in multiple small vesicles in the CD4⁺CD25⁺ cells, while as expected the CD4⁺CD25⁻ cells were not stained (Fig. 3b). Costaining of the CD4⁺CD25⁺ subset did not show significant CTLA-4 co-localization with the lysosomal markers CD107a and cathepsin D, while costaining with the early endosomal marker (EEA1) revealed minimal colocalization. Based on our previous description of intracellular RANTES storage vesicles in CD8⁺ T-cell memory and effector subsets, we tested for the possibility that CTLA-4 is stored in RANTES storage vesicles. Although only a minority of CD4⁺CD25⁺ T cells expressed RANTES storage vesicles, those that did showed minimal colocalization with CTLA-4 (Fig. 3b).

Figure 3.

Cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) expression in human CD4⁺CD25⁺ T cells. (a) Purified CD4⁺ T cells were surface stained for CD4 and CD25 and analysed for cytoplasmic expression of CD107a and CTLA-4 and granzyme B. Dotted blue histogram represents isotype control and red solid histogram respresents the specific staining for CD107, CTLA-4 and granzyme B. (b) Deconvolution microscopy and three-dimensional reconstruction of 100 planes (0·069 μm) were performed with purified CD4⁺CD25⁻ and CD4⁺CD25⁺ human T cells stained for: the lysosomal proteins cathepsin D, CD107a, the early endosome marker EEA1 and RANTES (green) combine with CTLA-4 staining (red).

Activation of mouse lymph node CD4⁺CD25⁺ T cells triggers Rab27a-independent exocytosis of internal CTLA-4

Previous studies in humans and mice have shown that the exocytosis of lysosomal/cytotoxic granules in activated CD8⁺ and CD4⁺ T cells requires Rab27a.¹² The availability of Rab27a-defective ashen mice allowed a test of whether activation-induced CTLA-4 surface expression in Treg cells is also Rab27a-dependent. As previously reported,¹⁷ mouse lymph node CD4⁺ T cells have a more distinct population of CD25⁺ cells, and internal CTLA-4 was expressed only in the positive subset (Fig. 4a). Internal CD107a expression was homogeneous and very similar in the CD25⁺ and CD25⁻ subsets. Flow cytometry of permeabilized lymph node T cells showed normal positive CD107a and CTLA-4 expression in the CD4⁺CD25⁺ subpopulation from both ashen and control C3H mice. When purified CD4⁺CD25⁺ T cells were examined by fluorescence microscopy using image deconvolution, CTLA-4 was seen in similar cytoplasmic vesicles in CD4⁺CD25⁺ cells from both mouse strains (Fig. 4b). Costaining with lysosomal markers CD107a (red) showed very little colocalization of the two vesicular compartments in either case, confirming the results with human cells. When costained with the EEA1 (red), partial colocalization with CTLA-4 was seen, although the great majority of both stains did not colocalize (Fig. 4b).

Figure 4.

Cytoplasmic expression of cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) by CD4⁺CD25⁺ T cells from ashen and C3H mice. (a) Intracellular expression of CTLA-4 and CD107a (red histrogram) and isotype control (blue histogram) was analysed by flow cytometry of permeabilized CD4⁺CD25⁻ and CD4⁺CD25⁺ T-cell subsets isolated from lymph nodes of ashen and C3H mice. (b) Expression of CTLA-4 by CD4⁺CD25⁺ T cells from ashen and C3H mice. Deconvolution and three-dimensional reconstruction of fluorescence images were performed with purified CD4⁺CD25⁻ (top) and CD4⁺CD25⁺ (bottom) T-cell subsets and stained for CD107a or EEA1 (red), costained for CTLA-4 (green).

Activation-induced surface expression of CTLA-4 was measured on purified CD4⁺ T-cell lymph node cells (Fig. 5). Before activation, CTLA-4 was not detectably expressed on the surface, and as expected, activation of CD4⁺CD25⁻ T cells gave no detectable surface CTLA-4 expression (data not shown). In contrast, surface CTLA-4 was detected at 1 and 3 hr after activation in CD4⁺CD25⁺ cells, and was roughly equivalent in C3H and ashen cells. In the presence of CHX, surface CTLA-4 expression was slightly inhibited after 1 hr and more markedly after 3 hr, but the protein synthesis-independent surface expression was similar in C3H and ashen cells. Activation-induced CD69 expression served as a control for CHX inhibition of newly synthesized proteins (Fig. 5). These results confirm that CTLA-4 is stored in a compartment different from the lysosome-like granules.

Figure 5.

Regulated secretion of cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) by C3H and ashen mice. Surface expression of CTLA-4 was assessed after T-cell receptor stimulation in the absence or presence of protein synthesis inhibitor cycloheximide (CHX). The constitutive secretion pathway was analysed by surface staining of CD69 and its inhibition by CHX. Histograms represent cells cultured in medium (blue) or activated with phorbol myristate acetate/ionomycin (red), and the isotype monoclonal antibody control (black).

CTLA-4 surface expression by CD4⁺CD25⁺ T cells recruits B7-2 to the synapse of conjugated APC

To analyse one potential functional consequence of TCR-induced CTLA-4 surface expression in Treg cells, we tested the possibility that surface CTLA-4 expression on Treg cells sequesters significant numbers of costimulator molecules on the surface of conjugated APC, thus compromising APC helper function. This was modelled by examining conjugates of B cells and PCC (I-E^k restricted) TCR-specific CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells. To generate PCC-specific CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells, AND-TCR mice (H-2^b) were bred with PCC transgenic mice (H-2^a) to generate H-2^bxa mice in which PCC-specific CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells can both differentiate.¹¹ Most of these conjugates are antigen dependent, as approximately 40% of the CTLA-4⁺ T cells formed conjugates in the presence of the PCC peptide compared to 10% in the absence of antigen (Fig. 6a). The distribution of costimulator B7-2 (CD86) on the surface of conjugated B cells was monitored by confocal microscopy. CD4⁺CD25⁺ T cells were detected by cytoplasmic staining of CTLA-4 (red) and CD86 (green). Confocal microscopy shows that CD86 molecules were strongly enriched in the synapse in conjugates formed by CD4⁺CD25⁺ T cells, and the CTLA-4-containing vesicles polarized to the synapse, while in similar conjugates with CD4⁺CD25⁻ T cells, CD86 staining was rather uniformly distributed around the B-cell surface (Fig. 6b).

Figure 6.

CD4⁺CD25⁺ T-cell–antigen-presenting cell conjugates recruit B7-2 molecules to the synapse. (a) Purifed CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells from AND mice were mixed at a 1 : 1 ratio with B cells (B10.A mice) loaded with Piogen cytochrome C (PCC) (2 μm) and incubated for 30 min at 37°. Then the cells were fixed, attached to coverslips coated with poly-l-lysine, and stained for B7-2 (green) and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4; red) before examination under the confocal microscope. (a) The antigen dependence of conjugate formation. (b) The six panels show a confocal z-series using 0·3-μm sections of T–B conjugates with a CD4⁺CD25⁻ cell (upper panel) and a CD4⁺CD25⁺ T cell (lower panel). Control monoclonal antibody staining is shown for each type of conjugate.

To quantify B7-2 recruitment into the antigen-dependent synapse after TCR engagement we measured the mean B7-2 fluorescence intensity on the surface of antigen-loaded B cells conjugated to CD4⁺CD25⁺ compared with CD4⁺CD25⁻. Measurements of average B7-2 intensity were normalized by calculating the ratio of average intensity in the synapse region (R1) to that on the remaining B-cell surface (R2), with background (R3) subtracted from each, (Fig. 7). The histogram shows the results of three independent experiments showing a statistically significant difference between B cells conjugated to the two different CD4⁺ T-cell subsets, with B cells conjugated to Treg cells having a higher synapse : membrane ratio (Fig. 7b). This indicates that the latter cells have a higher concentration of B7-2 in synapse, presumably because of B7-2 recruitment by CTLA-4. Moreover, analysis of integrated B7-2 fluorescence intensity in synapse versus total membrane showed that 53% (± 3% SEM, n = 38) of the total B7-2 intensity was in the APC–Treg cell synapse, so that about half of the costimulatory potential of these APC would be lost to a T-cell recognizing antigen on such APC.

Figure 7.

Quantification of B7-2 in the synapse of CD4⁺CD25⁺ and CD4⁺CD25⁻ T cells. (a) CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells from AND mice were mixed at a 1 : 1 ratio with B cells from B10.A mice loaded with Piogen cytochrome C (PCC) peptide. (b) Conjugates formed by a 1 : 1 ratio of CD4⁺CD25⁻ or CD4⁺CD25⁺ and B-cell or lipopolysaccharide B-cell blasts pulsed with PCC peptide were acquired in z-section steps of 0·067 μm and deconvolved. The regions of interest were selected in the brightest plane and fluorescence intensity was in R1 (synapse), R2 (membrane outside the synapse) and corrected for R3 (background inside the cell). Corrected fluorescence intensities were used to calculate the ratio of synapse : (surface outside the synapse), and the histograms were plotted as frequency (y) of intensity ratio (x). The data show 32 CD4⁺CD25⁻ synapses and 36 CD4⁺CD25⁺ synapses, in three independent experiments. The shift of the frequency ratio in the histogram showed a statistical significance with a P-value of 5 x 10⁻⁹.

Discussion

In considering a possible role for the regulated secretory pathway in delivering non-cytotoxic effector molecules by CD4⁺ T cells, we first sought evidence for the existence of internal storage of candidate molecules in these cells. While previous studies had suggested that CD4⁺ T cells can utilize the granule exocytosis cytotoxicity pathway,^18–20 and that cloned CD4⁺ helper cells release granzyme A after TCR engagement,²¹ it was not clear whether this was attributable to an atypical subpopulation of CD4⁺ T cells. The flow cytometry data show a roughly equivalent lysosome-like compartment in CD4⁺ and CD8⁺ T-cell blasts, and a functional TCR-regulated secretion of stored lysosomal/granule components in the presence of protein synthesis inhibitors. Such secretion was observed for the lysosomal enzyme β-hexosaminidase, and the granule-associated protease granzyme A (Fig. 1a,b).¹² Homogeneous histograms of surface expression of the lysosomal membrane markers CD107a (Fig. 1c) and CD107b (data not shown), make it clear that the great majority of CD4⁺ T-cell blasts underwent TCR-induced granule exocytosis, although it was less prominent than that of CD8⁺ T cells. In microscopy studies with anti-CD3/CD28-coated beads, the granzyme A/lysosomal vesicles in CD4⁺ T-cell blasts showed a polarization to the region of TCR engagement similar to that seen in CD8⁺ T cells (data not shown).

When permeabilized subsets of CD4⁺ and CD8⁺ T cells isolated directly from blood were analysed by flow cytometry, the expression of the lysosome markers CD107a and granzyme B correlated with the differentiation phenotype in both CD4⁺ and CD8⁺ T-cell subsets of the T cells (Fig. 2). These results parallel microscopy studies of ex vivo subsets of CD8⁺ T cells, showing that staining of the lysosome granule proteins correlates with the differentiation markers of the T cells.²

Because internal CTLA-4 in T-cell blasts has been reported to be partially colocalized with perforin,^16,22,23 and its cytoplasmic domain contains an endosomal targeting motif, it seemed plausible that CTLA-4 is stored in the secretory lysosomal compartment. However, the nature of the internal compartment in which CTLA-4 is stored remains undefined for any T-cell subset. Given current interest in the Treg subset of CD4⁺ T cells, we considered whether CTLA-4 exocytosis could contribute to their suppressor function. Examining CD4⁺ CD45⁺ Treg cells in the human and mouse (Figs 3 and 4), we found that CTLA-4 did not colocalize with lysosomal markers or internal RANTES, and colocalized only minimally with endocytosis markers. The finding that protein synthesis-independent TCR-triggered surface CTLA-4 expression is normal in Rab27a-defective ashen mice (Fig. 5) further strengthens the case that internal CTLA-4 is contained in a non-lysosomal compartment as previously described.^22,24

The functional mechanisms used by regulatory cells to suppress immune responses are unclear, and it seems likely that both suppressive cytokines and mechanisms involving direct cell–cell contact may contribute. The many proposed Treg cell subpopulations⁴ may use different sets of effector mechanisms. We hypothesized that one mechanism might involve antigen-triggered granule exocytosis of internal membrane CTLA-4, resulting in its surface expression on Treg cells. This surface CTLA-4 would then recruit and bind CD80/CD86 molecules on the surface of APC, neutralizing their costimulatory function. We tested these hypotheses by examining conjugates of CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells and APC loaded with PCC and examined the mean fluorescence intensity in the synapse by microscopy. In comparing CD4⁺CD25⁺ Treg cells with conventional CD4⁺CD25⁻ cells, we observed a clear difference in the intensity of CD86 molecules in the T-cell–APC junction in the conjugates, with much more CD86 in the junction compared to the remaining APC surface in the Treg conjugates. Regulated secretion of CTLA-4 on Treg cells can therefore compromise APC function by depriving T cells recognizing antigen on coconjugated APCs of CD28-mediated costimulatory signals. In our experiments, we tested the recruitment of CD86 expressed by resting B cells; because CTLA-4 possesses higher affinity for both CD80 and CD86 than for CD28, we would predict that CTLA-4 would compete efficiently for both ligands at the T-cell–APC synapse.²⁵ Our results support previous proposals for this mechanism of Treg function,^26–28 similar to proposals of Treg-mediated inhibition of APC–T-cell interaction in in vivo studies of autoimmune models.^29,30

Regulated secretion of mediators involves the rapid and localized delivery of proteins independent of protein synthesis. In this paper we demonstrated that CD4⁺ T cells are also able to secrete mediators in lysosomal storage vesicles by the regulated secretory pathway, and this could be important in mediating various effector functions. This paper further demonstrates that CD4⁺CD25⁺ Treg cells have another non-lysosomal CTLA-4-containing internal compartment which utilizes a Rab-27a-independent regulated secretory pathway. By mediating the surface exposure of CTLA-4 after antigen recognition and complexing B-7 molecules on the APC surface, this compartment may contribute to the suppressive activity of Treg cells.