Dendritic cell-based assays, but not mannosylation of antigen, improves detection of T-cell responses to proinsulin in type 1 diabetes

Abstract

In vitro detection of T-cell responses to autoantigens in type 1 diabetes is recognized as being technically challenging. We aimed to accurately measure cellular responses to proinsulin in patients with diabetes, and speculated that presentation of antigen by dendritic cells (DCs) would enhance the sensitivity of the peripheral blood assay. Antigen was mannosylated to facilitate uptake through DC surface mannose receptors to further improve the assay. Whole proinsulin, as well as mannosylated peptides of proinsulin, were combined with peripheral T cells and autologous immature DCs in a proliferative assay in a panel of newly diagnosed type 1 diabetic patients. The DC-based assay detected responses to proinsulin in five of 15 diabetic patients compared to one of 15 diabetic patients detected using the standard mononuclear cell assay. When the results of all patients were combined, the DC assay, but not the mononuclear cell assay, had a proinsulin response that was significantly higher than background (P < 0·001). The DC assay was, however, associated with high autologous mixed lymphocyte reactions that possibly masked responses in individual patients. Mannosylated antigen was taken up in larger quantities than non-mannosylated antigen, but not presented any more powerfully. Our data suggest that autologous DC-based assays are more powerful than standard peripheral blood mononuclear cell assays. However, they are compromised by high autologous mixed lymphocyte reactions and this requires addressing before they can be used as a routine readout of in vitro peripheral T-cell responses.

Introduction

Type-1 diabetes (T1D) is a T-cell-mediated autoimmune disease characterized by pancreatic β-cell destruction.¹ The ability to measure T-cell responses to pancreatic autoantigens is therefore critical for understanding and intervening in this condition. However, the measurement of T-cell responses to autoantigens in vitro, in T1D as well as in other autoimmune diseases, has proved problematic. This is attributed to a variety of factors, including a low precursor frequency of disease-specific T cells in the peripheral blood,² T cells being present but in an anergic state,³ a low affinity of the T cells for their peptide ligand, and inhibition by regulatory T cells.⁴

Proinsulin is an important autoantigen in T1D.⁵ We have previously measured T-cell responses to proinsulin using the standard peripheral blood mononuclear cell (PBMC) proliferative assay and were unable to detect an increased antibody response in newly diagnosed T1D patients compared with healthy subjects, even though the T1D patients had proinsulin autoantibodies in contrast to healthy subjects who did not.⁶ We assumed that the inability to detect an increased response was the result of a low sensitivity of the standard PBMC assay and sought to design a more sensitive assay.

T cells recognize antigen following processing and presentation by antigen-presenting cells (APCs), notably dendritic cells (DCs). However, DCs account for <1% of PBMCs⁷ and it is not clear whether they are functionally adapted to antigen uptake. In an attempt to amplify both antigen uptake and subsequent T-cell responses, we employed immature blood monocyte-derived DCs – generated by culture ex vivo in the presence of granulocyte–macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4) – and purified peripheral blood T cells. In addition, we further attempted to increase the sensitivity of the DC-based T-cell assay by mannosylation of antigen. The mannose receptor (MR), a membrane-bound calcium-dependent lectin, is expressed predominantly on professional APCs, in particular on DCs,⁸ and rapidly internalizes mannosylated antigens and targets them to the major histocompatibility complex (MHC) class II pathway for presentation to T cells.⁹ This route of targeting the DC has been found to be particularly useful for the gene delivery of antigens.¹⁰

Materials and methods

Subject recruitment

Twenty, type 1 diabetic patients (mean age 25·8 years; range: 18–36 years) were recruited at a mean of 6 weeks (range: 1–28 weeks) after onset of the first symptoms associated with diabetes (polyuria, polydipsia, weight loss, thirst). All patients were on insulin treatment at the time when blood samples were taken. Blood samples were taken from unmatched laboratory volunteers as healthy controls. All subjects gave full informed consent and the study received approval from the local ethical committees.

Antigens

Tetanus toxoid (TT) was obtained from the Swiss Serum Institute (Berne, Switzerland). A promiscuous T-cell epitope peptide in TT (TTp, amino acids 947–967)¹¹ was synthesized, by standard Fmoc chemistry, by Professor D. C. Wraith (Department of Pathology and Microbiology, University of Bristol, Bristol, UK). Proinsulin was a kind gift from Dr Ron Chance (Eli Lilly Pharmaceuticals, Indianapolis, IN). This source of proinsulin was confirmed to contain <3 IU/ml of endotoxin (NIBSC, Potters Bar, UK). Two overlapping 30-mer peptides spanning sequences 22–51 (proinsulin-p1) and 42–71 (proinsulin-p2) were synthesized by INBIOS (Naples, Italy) and were terminally acetylated. These sequences cover the T-cell epitopes previously identified.^6,12–16

Mannosylated antigens

TTp1, proinsulin-p1 and proinsulin-p2 were mannosylated in a targeted manner at their N-terminus so that the mannose moiety would not interfere with T-cell epitope processing and presentation. Mannosylation was achieved by the addition of a lysine construct containing 0, 1 or 2 mannose units, to the end of the peptide-synthesis reaction. This produced non-mannosylated, mono-mannosylated and bis-mannosylated versions of each of TT, proinsulin-p1 and proinsulin-p2. These were termed TTp, TTp(M), TTp(2M), proinsulin-p1, proinsulin-p1(M), proinsulinp1(2M), proinsulin-p2, proinsulin-p2(M) and proinsulin-p2(2M), respectively. In the construct containing two mannose units, the distance between the units was optimal for binding to the mannose receptor. Purity, as measured by high performance liquid chromatography, was >95% for all mannosylated peptides. The variably mannosylated version of TTp was also synthesized conjugated to fluorescein isothiocyanate (FITC), for use in flow cytometric uptake studies.

PBMC proliferation assay

Fifty millilitres of peripheral blood was obtained from informed, consenting adults, into sterile heparinized tubes and processed within 4 hr. PBMC were isolated by buoyancy centrifugation over Ficoll–Hypaque (Pharmacia Biotech, St Albans, UK) and resuspended in 10% (v/v) heat-inactivated non-autologous human serum (National Blood Service South-west, Bristol, UK) in RPMI-1640 (Gibco, Paisley, UK) supplemented with 2 mm l-glutamine (Sigma, Dorset, UK), 100 U/ml penicillin (Brittania, Redhill, UK) and 0·1 mg/ml streptomycin (Evans, Greenford, UK). Stimulation assays were performed by incubating 2 × 10⁵ PBMC per well with 10 µg/ml of antigen in a final volume of 200 µl. All assays were conducted in triplicate wells of a round-bottomed 96-well plate (NUNC, Rochester, NY). Plates were incubated for 5 days at 37° in a humidified 5% CO₂ incubator. Cells were then pulsed for 6 hr with 10 µl (0·5 µCi) of [³H]thymidine (TRK686; Amersham, Bucks, UK), harvested onto glass-fibre filters, and the amount of incorporated radiolabel was measured in a β-counter (1450 Microbeta-plus; Wallac, Turku, Finland).

Generation of oligoclonal T-cell lines to TT

PBMC were cultured with 10 µg/ml TT for 5 days, followed by further stimulation (5-day cycles) with 2% v/v Lymphocult HP (LC; Biotest, Birmingham, UK), alternating with 2% v/v LC + 10 µg/ml TT, with fresh irradiated (2000 rads) autologous PBMC in a 2 : 1 cell ratio. T-cell lines were then screened for TT responsiveness with 10 µg/ml TT and irradiated (2000 rads) autologous PBMC, at a ratio of 2 : 1, in a 2-day proliferation assay. To generate oligoclonal cell lines, responsive T-cell lines were plated at limiting dilution (one cell per well) in round-bottomed 96-well plates (NUNC) with 2% v/v LC, 10⁵ irradiated fresh non-autologous PBMC and 10 µg/ml phytohaemagglutinin (PHA; Sigma). The wells were fed in 7-day cycles, alternating between 2% v/v LC, or, 2% v/v LC, 10 µg/ml PHA and 10⁵ irradiated (2000 rads) non-autologous PBMC. The contents of wells showing growth were transferred and propagated to sufficient numbers before screening for TT responsiveness, as described above.

Preparation of DCs

Immature DCs were generated from peripheral blood monocytes, according to the technique of Sallusto et al.⁹ Briefly, fresh PBMC were incubated for 75 min at 37°, in humidified 5% CO₂, in wells of a six-well plate (COSTAR, Bucks, UK), after which non-adherent cells were removed by gently swirling with warm RPMI-1640. DCs were generated from the adherent cells (enriched from ≈12% to 50% in CD14 monocytes) cultured with 10% fetal calf serum (FCS; Sigma), in the presence of 1000 U/ml interleukin-4 (Genzyme, West Malling, UK) and 800 U/ml recombinant human GM-CSF (Leucomax, Sandoz, Frimley, UK) for 6 days. Half of the culture medium was replaced with fresh cytokines and medium after 3 days.

DC-potentiated T-cell proliferation assays

T cells were purified from PBMC by immunomagnetic negative selection following incubation with bead-conjugated antibodies to CD11b, CD16, CD19, CD36 and CD56 (Miltenyi Biotec, Bisley, UK). Immature DCs were washed twice, irradiated (2000 rads) and combined with these freshly purified autologous T cells in triplicate round-bottomed wells of a 96-well plate (NUNC). Each well contained 8000 DCs, 4 × 10⁵ T cells and 10 µg/ml antigen, in 200 µl of 10% human serum. Hence, although the DCs were generated in FCS, they were used in the T-cell proliferation assay with human serum. Plates were incubated for 5 days and proliferation was measured by thymidine incorporation, as described above. The numbers of DCs and T cells had been optimized to provide the highest antigen-specific proliferation with the lowest background counts (data not shown).

Flow cytometry

Uptake of mannosylated antigen by DCs was measured following incubation of 10⁵ freshly prepared DCs with a range of concentrations of mannosylated or non-mannosylated FITC-conjugated TTp, at 37° in the dark. Uptake was quickly terminated by the addition of a large volume of cold buffer [phosphate-buffered saline containing 5% (v/v) heat-inactivated FCS], and the DCs were pelleted by centrifugation and analysed by single-colour flow cytometry (FACSCalibur; Becton Dickinson, Cowley, UK). To study the effect of blockade of the MR, DCs were preincubated for 10 min with 10 µg/ml mannan (Sigma) prior to the addition of FITC-conjugated TTp peptides. The data were analysed using CellQuest software (Becton Dickinson).

Statistical analysis

Results of proliferation assays were expressed in one of two ways. First, proliferation was considered ‘positive’ when the mean counts per minute (c.p.m.) in the presence of antigen was greater than the mean value plus two standard deviations in the absence of antigen, and, in addition, was >1000 c.p.m. Second, responses of subjects were combined and proliferation in the presence and absence of antigen was compared by two-way analysis of variance (anova).

Results

DC-potentiated T-cell responses to TT and proinsulin in diabetic patients

The DC-potentiated T-cell assay and the standard PBMC assay were used to measure responses to proinsulin (n = 15) and TT (n = 10) in newly diagnosed diabetic subjects (Table 1). The DC assay was associated with very high autologous mixed lymphocyte reactions (AMLRs), as shown by spontaneous T-cell proliferation in the presence of autologous DCs and the absence of antigen. Nevertheless, the DC-based assay detected a greater number of responses to both proinsulin and TT; in five of 15 (33%) and 10/10 (100%), respectively, compared with one of 15 (7%) and eight of 10 (80%) in the PBMC assay. When responses across all 15 patients were combined and compared by two-way anova, the DC assay detected a highly significant difference between control and proinsulin (P < 0·001), whereas the PBMC assay did not detect a difference (P = 0·61). On an individual patient basis, however, it is possible that the AMLR masked proinsulin-specific proliferation. Interestingly, the high background was not seen when a TT-specific oligoclonal line, rather than PBMC-derived polyclonal T cells, was used in the DC assay, resulting in a better definition of the antigen-specific response (Fig. 1). This is consistent with previous studies using T-cell clones rather than PBMC.¹⁷ It is also consistent with AMLR proliferation being directed towards proteins (for example in the culture media) being processed and presented by DCs, or towards self-antigens presented by the DC.

Table 1.

Comparison of peripheral blood mononuclear cell (PBMC)- and dendritic cell (DC)-based T-cell responses to tetanus toxoid (TT) and proinsulin in newly diagnosed diabetic patients

	PBMC			DC + T cell
Patient	Control	TT	Proinsulin	Control	TT	Proinsulin
1	354 ± 46	6721 ± 347*	381 ± 79	2097 ± 345	13 789 ± 2715*	3445 ± 853*
2	1741 ± 662	3660 ± 1489	627 ± 21	6654 ± 1518	25 835 ± 541*	5828 ± 2131
3	631 ± 38	11 847 ± 1552*	724 ± 63	6477 ± 1403	25 997 ± 630*	6047 ± 2306
4	2476 ± 510	35 555 ± 3486*	4628 ± 933	4601 ± 771	36 808 ± 3641*	6459 ± 1423
5	2598 ± 581	9137 ± 1924*	2491 ± 421	4541 ± 1753	25 333 ± 1434*	6571 ± 1518
6	353 ± 27	25 657 ± 1664*	291 ± 20	2250 ± 258	22 932 ± 6431*	3723 ± 124*
7	3288 ± 384	12 104 ± 1907*	2678 ± 638	13 776 ± 1631	26 156 ± 1921*	16 817 ± 3387
8	165 ± 46	19 692 ± 851*	132 ± 33	14 744 ± 2457	33 938 ± 1610*	17 376 ± 1574
9	392 ± 66	625 ± 87	603 ± 30	6285 ± 1196	21 964 ± 1264*	7165 ± 1689
10	264 ± 29	24 109 ± 8626*	351 ± 64	7025 ± 1344	19 478 ± 658*	5622 ± 1105
11	1993 ± 95	NT	2580 ± 530*	3956 ± 1660	NT	9730 ± 3624*
12	5197 ± 290	NT	4436 ± 1078	9538 ± 801	NT	21 411 ± 1444*
13	658 ± 73	NT	427 ± 108	17 906 ± 2627	NT	14 976 ± 3128
14	2291 ± 329	NT	2508 ± 781	7225 ± 986	NT	7875 ± 570
15	2083 ± 526	NT	3031 ± 455	3635 ± 1946	NT	19 766 ± 2313*

Data are expressed as counts per minute ± standard error of the mean (c.p.m. ± SEM).

Control values indicate proliferation in the absence of antigen.

^*Antigen-specific responses that are significantly different from control responses.

NT, not tested.

Figure 1.

Proliferation of a tetanus toxoid (TT)-specific T-cell line and peripheral blood mononuclear cells (PBMC) to TT (10 µg/ml) presented by irradiated autologous dendritic cells (DCs). Proliferation in the presence (TT) and absence (control, C) of antigen is shown (results expressed as mean ± standard error of the mean), and their calculated ratio indicated as a stimulation index (SI). c.p.m., counts per minute.

Uptake of mannosylated peptides by DCs

In attempting to increase the sensitivity of detection of antigen-specific T cells, we took advantage of the ability of the MR to take up and target antigen for processing and presentation. Flow cytometry of cells incubated with FITC-labelled mannosylated TT peptide was conducted to investigate whether mannosylation could increase uptake of peptide. DCs were ‘flash’ pulsed for 1 second with a fixed concentration of labelled peptide and subsequent uptake was measured. Mannosylation increased uptake by 2·9-fold compared with non-mannosylated peptide, as assessed by measuring the mean fluorescence intensity of peptide-pulsed cells (Fig. 2a). Bis-mannosylated and mono-mannosylated peptides were taken up to an equivalent extent.

Figure 2.

Uptake of mannosylated peptide by dendritic cells (DCs). (a) DCs were pulsed for 1 second with 10 µg/ml of non-mannosylated (dotted line), mono-mannosylated (dashed line) or bis-mannosylated (dot-dashed line) peptide TTp, quickly diluted with cold buffer, washed and analysed by flow cytometry. Background fluorescence is shown as a solid line.(b), (c) and (d) DCs were pulsed for 10 min with 1 µg/ml (dotted line), 10 µg/ml (dashed line) or 100 µg/ml (dot-dashed line) of fluorescein-conjugated non-mannosylated tetanus toxoid (TT) peptide (b), mono-mannosylated TT peptide (c), or bis-mannosylated TT peptide (d), washed and analysed by flow cytometry. (e) Uptake of non-mannosyated (TTp), mono-mannosylated [TTp(M)] and bis-mannosylated [TTp(2M)] TT peptides by DCs following a 10-min incubation at 37° in the presence (white boxes) or absence (black boxes) of 10 µg/ml mannan.

DCs were pulsed for an extended period of 10 min with log-fold increasing concentrations of TT peptides. Mannosylation increased the amount of internalized peptide across the concentration range 1–100 µg/ml (Fig. 2b, 2c, 2d). However, mannosylation did not increase uptake when the peptide concentration was ≤ 0·1 µg/ml (data not shown). Uptake of mono-mannosylated peptide was similar to bis-mannosylated peptide across the concentration range studied (Fig. 2c, 2d). Blockade of the MR with mannan significantly reduced the uptake of mannosylated peptides without affecting the uptake of non-mannosylated peptide (Fig. 2e).

Responses of healthy subjects to mannosylated TT

Having confirmed an increased antigen uptake following mannosylation, the PBMC from five healthy subjects, previously immunized to TT, were stimulated with the promiscuous epitope of TT (TTp) in both mannosylated and non-mannosylated form. All subjects tested responded (Fig. 3a) in a dose-dependent manner. However, mannosylation of TTp did not improve the sensitivity of the T-cell proliferation assay (Fig. 3a).

Figure 3.

Tetanus toxoid (TT) peptide-specific responses in healthy subjects. Peripheral blood mononuclear cell (PBMC)-based (a) and dendritic cell (DC)-based (b) assays were used to measure proliferative T-cell responses to increasing concentrations of TT peptide. Responses to non-mannosylated (continuous line) and mono-mannosylated (dotted line) TT peptide were measured, and results from a representative subject are shown. In the PBMC assay, responses to both non-mannosylated and mono-mannosylated TT become significantly different from background at 1 µg/ml. In the DC-based assay, the response becomes significant at 10 µg/ml for the non-mannosylated TT, but not at all for the mono-mannosylated TT. c.p.m., counts per minute.

Because the number of MR-expressing APCs is low in PBMCs, the effect of peptide mannosylation was assessed in the DC-based assay. Over the range of antigen concentrations used, mannosylation of TTp did not enhance T-cell proliferation (Fig. 3b).

Responses of diabetic patients to mannosylated proinsulin

Although mannosylation did not improve the T-cell response to TTp, it was possible that it might enhance the responses to a weaker (auto-)antigen, such as proinsulin. To minimize masking potential T-cell epitopes with bulky mannose groups, it was important to target mannosylation to one end of proinsulin. At 87 amino acids, proinsulin is too long to be efficiently synthesized as a single peptide. As mannose constructs can only be added in a targeted manner to proteins synthesized through standard Fmoc chemistry, we synthesized two overlapping 30-mer peptides (residues 22–51 and 42–71), spanning most of proinsulin, for N-terminal mannosylation. Previous studies have reported the detection of responses to epitopes within the amino acid region encompassed by these two peptides.^6,12–16 These two 30-mers were mono- or bis-mannosylated at the N-terminus to form proinsulin-p1(M) or proinsulin-p1(2M) and proinsulin-p2(M) or proinsulin-p2(2M), respectively. The response to these proinsulin peptides compared to whole proinsulin was measured in the PBMC (Table 2, n = 10) and DC-based T-cell (Table 3, n = 5) assays in newly diagnosed diabetic patients (Table 1).

Table 2.

Peripheral blood mononuclear cell (PBMC) responses to proinsulin and mannosylated proinsulin peptides in newly diagnosed diabetic patients

			Proinsulin peptide 1			Proinsulin peptide 2
Patient	Control	Proinsulin	Proinsulin-p1	Proinsulin-p1(M)	Proinsulin-p1(2M)	Proinsulin-p2	Proinsulin-p2(M)	Proinsulin-p2(2M)
11	1993 ± 95	2580 ± 530*	1075 ± 546	700 ± 189	3296 ± 826*	3093 ± 529*	157 ± 42	2142 ± 232
12	5197 ± 290	4436 ± 1078	1566 ± 602	1358 ± 201	3193 ± 473	2354 ± 262	4205 ± 1059	2842 ± 496
13	658 ± 73	427 ± 108	329 ± 72	580 ± 158	161 ± 26	363 ± 83	368 ± 122	432 ± 53
14	2291 ± 329	2508 ± 781	1434 ± 380	1224 ± 468	1372 ± 102	3300 ± 477	2622 ± 952	5790 ± 1474*
15	2083 ± 526	3031 ± 455	1418 ± 459	1017 ± 179	2438 ± 364	4851 ± 2301*	3064 ± 693	2733 ± 533
16	2066 ± 614	NT	1709 ± 83	1763 ± 606	1833 ± 370	707 ± 349	904 ± 42	2887 ± 78
17	3202 ± 1471	NT	2198 ± 492	1725 ± 211	2271 ± 543	2549 ± 144	2644 ± 292	2683 ± 432
18	4436 ± 1281	NT	5118 ± 943	4340 ± 637	3676 ± 63	3743 ± 540	4910 ± 1085	4969 ± 1359
19	5090 ± 814	NT	3504 ± 147	3748 ± 469	4951 ± 549	4080 ± 529	3565 ± 584	3759 ± 651
20	4759 ± 545	NT	4878 ± 433	2247 ± 147	2351 ± 334	2245 ± 351	3646 ± 350	3126 ± 530

Control values indicate proliferation in the absence of antigen.

Data are expressed as counts per minute ± standard error of the mean (c.p.m. ± SEM).

^*Antigen-specific responses that are significantly different from control responses.

NT, not tested. Note that data from patients 11–15 are also shown in Table 1.

Table 3.

Dendritic cell (DC)-based T-cell responses to proinsulin and mannosylated proinsulin peptides in newly diagnosed diabetic patients

			Proinsulin peptide 1			Proinsulin peptide 2
Patient	Control	Proinsulin	Proinsulin-p1	Proinsulin-p1(M)	Proinsulin-p1(2M)	Proinsulin-p2	Proinsulin-p2(M)	Proinsulin-p2(2M)
11	3956 ± 1660	9730 ± 3624*	3732 ± 2452	2754 ± 801	793 ± 644	4885 ± 2091	8866 ± 3018	2066 ± 839
12	9538 ± 801	21 411 ± 1444*	11 269 ± 890*	13 826 ± 793*	9168 ± 1325	10 604 ± 1218	9381 ± 1738	12 085 ± 2208*
13	17 906 ± 2627	14 976 ± 3128	9651 ± 3609	16 874 ± 3496	7150 ± 770	14 383 ± 3830	4719 ± 344	10 046 ± 2825
14	7225 ± 986	7875 ± 570	5899 ± 550	6293 ± 503	5769 ± 401	6741 ± 684	5229 ± 460	8048 ± 1985
15	3635 ± 1946	19 766 ± 2313*	3942 ± 1186	5281 ± 3023	7416 ± 2645	8795 ± 2645	10 107 ± 3520	4484 ± 1585

Control values indicate proliferation in the absence of antigen.

Data are expressed as counts per minute ± standard error of the mean (c.p.m. ± SEM).

^*Antigen-specific responses that are significantly different from control responses.

Note that data from patients 11–15 are also shown in Tables 1 and 2.

In subjects 11, 12 and 15 who responded to whole proinsulin in the DC assay, significant responses (in both the PBMC and DC assays) were seen to the proinsulin-p2 peptides in all three subjects and to proinsulin-p1 peptide in subjects 11 and 12. A response to proinsulin-p2 was also seen in subject 14 in the PBMC assay. The mannosylation of proinsulin peptides did not, however, clearly amplify the response (Tables 2 and 3). High background counts and standard errors (Table 3) may have hindered the ability to detect any advantage of mannosylation.

Discussion

Several reports have suggested that DC-based T-cell assays may be superior to standard PBMC proliferative assays. However, these studies used T-cell clones,¹⁷ T-cell lines⁹ or hybridomas, or in vivo assays.¹⁸ We wanted to use a DC-based assay to detect polyclonal peripheral blood T-cell responses to an autoantigen. Although cytotoxic T-cell responses to tumour antigens have been elicited by DC stimulation,¹⁹ there have been no reports of the use of such a technique to study T-cell responses in autoimmune disease.

We optimized a DC-based assay for use with autologous T cells to measure cellular immunity to proinsulin and compared this to the standard peripheral blood T-cell assay in newly diagnosed T1D patients. Analysis of both individual and pooled results suggest that the DC-based assay is more sensitive; 33% of subjects had cellular responses to proinsulin with the DC-based assay compared to 7% with the PBMC assay. It should be noted that twice as many T cells were present in the DC-based assay (4 × 10⁵ versus 2 × 10⁵). However, our aim was to compare an optimized DC-based assay with the standard PBMC assay.²⁰ Despite optimization, high levels of spontaneous T-cell proliferation in the presence of DCs markedly compromised the effectiveness of the DC assay. DCs were irradiated prior to use and contributed minimally to the total proliferation (<100 c.p.m., data not shown). The DC-based assay still appears to lack the sensitivity that would enable it to be used routinely in individual patients for the detection of cellular immunity to proinsulin.

Maturation of the DCs by exposure to tumour necrosis factor or lipopolysaccharide to enhance antigen presentation further was found to result in even higher spontaneous proliferation (data not shown), making it even less sensitive for the detection for peripheral T-cell responses.

The spontaneous proliferation of polyclonal T cells in the presence of autologous DCs is believed to occur as the result of activation of T cells that are cross-reactive with self-antigen or serum-derived antigens.²¹ This seems to be more of a problem with purified DCs rather than DCs in PBMC, probably because of a higher DC : T-cell ratio.²² When a T-cell line specific for a non-self antigen is used, spontaneous proliferation, as expected, is minimal (Fig. 1).²³ Removal of FCS from the medium, as far as possible, has been shown to reduce the background response, even when mature DCs are used, suggesting that at least part of the polyclonal response is to foreign serum-derived proteins.^21,24 Because of the long half-life of the cell-surface MHC–peptide complex, it is possible that these foreign serum-derived proteins were being presented to the polyclonal T-cell population, even though the DCs are extensively washed before being combined with the T cells. This may explain why the background is not present with an oligoclonal T-cell line and why maturation of the DCs induces an even higher background count. We attempted to generate DCs in human AB serum instead of FCS to eliminate contamination with foreign serum-derived proteins, but were unable to do so. This was despite using different sources, preparations and concentrations of serum.

Attempts were therefore made to further improve the assay sensitivity by antigen mannosylation. Mannosylated antigen is taken up more efficiently by DC through mannose receptors that deliver antigen to the MHC class II processing pathway.⁸ If antigen concentration, processing or loading onto MHC molecules is rate-limiting in in vitro antigen-presentation assays, then this approach, notwithstanding background proliferation, might be expected to increase ‘signal-to-noise’, by amplifying the antigen-specific response. Previous studies with T-cell clones have clearly shown that mannosylation can increase sensitivity to antigen,²⁵ but, to our knowledge, studies using polyclonal T cells have not yet been conducted. N-terminal targeted, rather than random, mannosylation was used to ensure that the bulky carbohydrate group did not obscure T-cell epitopes or antigen-processing sites, and a bis-mannosylated tag with spacing designed to bind two adjacent domains of the MR simultaneously was developed. Despite increasing antigen uptake, we failed to improve the response of PBMCs to antigen. This was true both of standard PBMC- and DC-based T-cell assays, and with TT, a strong recall antigen, as well as with proinsulin, an autoantigen. Mannosylation may have failed to amplify T-cell responses, for several reasons. First, antigen delivery to MHC class II molecules may not be rate-limiting in detecting antigen-specific T-cell responses in polyclonal T-cell populations in vitro, and other factors, such as regulatory cells⁴ and a low T-cell precursor frequency,² have a greater part to play. Second, the peptides used, even though quite long (up to 30 residues) may have been presented in the context of surface MHC II molecules without needing to be internalized and processed. Third, the high spontaneous proliferation may also have masked any advantage of peptide mannosylation in DC assays. Fourth, DCs secrete a soluble form of MR that may facilitate transport of mannosylated antigens to areas of immune surveillance in vivo,^26,27 but which may out-compete DC surface MRs in the in vitro assay. Fifth, individuals with T1D may have abnormalities of mannose binding that reduce their ability to internalize antigens via the mannose receptor, as reported for the mannose-binding lectin.²⁸

In summary, we have shown that the use of DCs in in vitro cultures results in higher spontaneous, as well as antigen-stimulated, T-cell stimulation. This results in an improvement in detection of autoantigen-specific responses, but the high spontaneous proliferation limits the technique. Technical improvements, such as the use of serum-free medium,²⁹ may be required to fully exploit the potential of DC-based assays for the detection of T-cell responses to self-antigens. Targeted mannosylation of antigen enhanced the uptake by DCs, but failed to enhance antigen-specific proliferation in polyclonal T-cell populations.