- Email: [email protected]
Comparison of cytokine ELISpot assay formats for the detection of islet antigen autoreactive T cells
Journal of Autoimmunity 21 (2003) 365–376 www.elsevier.com/locate/issn/08968411
Comparison of cytokine ELISpot assay formats for the detection of islet antigen autoreactive T cells Report of the third immunology of diabetes society T-cell workshop Nanette C. Schloot a, Guido Meierhoff a, Maria Karlsson Faresjo¨ b, Patrick Ott c, Amy Putnam d, Paul Lehmann c, Peter Gottlieb d, Bart O. Roep e, Mark Peakman f*, Timothy Tree f a
German Diabetes Research Institute at the Heinrich Heine University, Du¨sseldorf, Germany b Division of Pediatrics, Clinical Research Center, Health University, Linko¨ping, Sweden c Department of Pathology, Case Western Reserve University, Cleveland, OH, USA d Barbara Davis Center, University of Colorado Health Sciences Center, Denver, CO, USA e Leiden University Medical Centre, Department Immunohaematology and Blood Transfusion, Leiden, The Netherlands f Department of Immunology, Guy’s, King’s and St Thomas’ School of Medicine, King’s College, Denmark Hill Campus, Rayne Institute, 123 Coldharbour Lane, London SE5 9NU, UK Received 2 June 2003; revised 19 June 2003; accepted 21 July 2003
Abstract The identification of sensitive assay formats capable of distinguishing islet autoreactive T cells directly ex vivo in blood is a major goal in type 1 diabetes research. Recently, much interest has been shown in the cytokine enzyme linked immunospot assay (CK ELISpot), an assay potentially capable of fulfilling these difficult criteria. To address the utility of this assay in detecting autoreactive T cells, a ‘wet’ workshop was organized using the same fresh blood sample and coded antigens. Five different laboratories participated, using three distinct CK ELISpot assay formats. Samples from two subjects were pre-tested for responses to sub-optimal concentrations of tetanus toxoid, representing a low frequency recall response, and peptides from diabetes associated autoantigens GAD65, IA-2 and HSP60. All participants measured interferon- production and combinations of interleukins-4, -5, -10 and -13. In the workshop 4 of 5 laboratories detected low frequency recall responses in both subjects and 3 of 5 detected at least one of the autoreactive peptide responses concordant with pre-testing. Significant assay format related differences in sensitivity and signal-to-noise ratio were observed. The results demonstrate the potential for detection of low-level autoreactive T cell responses and identify assay characteristics that will be useful for studies in type 1 diabetes. 2003 Elsevier Ltd. All rights reserved. Keywords: ELISpot; T cell; Diabetes; Workshop
1. Introduction Type 1 diabetes (T1D) is an autoimmune disease characterized by the selective destruction of the insulin producing pancreatic cells [1]. T cells are believed to * Corresponding author. Tel.: +44-2078485956; fax: +44-2078485953. E-mail address: [email protected] (M. Peakman). 0896-8411/03/$ - see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0896-8411(03)00111-2
play a key role in this process, and evidence for this assertion includes a predominance of these cells in the islet infiltrate close to diagnosis [2], the effects of immunosuppressive agents on disease progression [3], ‘adoptive transfer’ case reports on the development of T1D after transplantation of bone marrow [4] or pancreas from diabetic to non-diabetic recipient [5], and more recently a report of the development of T1D in an individual lacking B lymphocytes [6]. Moreover,
366
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Table 1 Methods of PBMC isolation and culture media used by Workshop participants Laboratory
Dilution of blood Separation reagent Spin Wash Assay media Serum
CTL-1
CTL-2
Mabtech
U-Cytech-1
U-Cytech-2
1:2 in PBS Cell Sept 1800 g/30 min 2 in RPMI RPMI 5% HI AB
None CPT tubes 1800 g/20 min 2 in RPMI RPMI 1% HI FCS
1:2 in RPMI Pharmacia hypaque 400 g/30 min 3 in RPMI IMDM 5% HI FCS
1:1.5 in PBS Maxiplus 999 g/20 min 3 in PBS RPMI 10% HI AB pooled
1:2 in RPMI Maxiplus 400 g/40 min 2 in RPMI RPMI 10% HI AB pooled
Abbreviations used: PBS, phosphate buffered saline; HI, heat inactivated; FCS, foetal calf sera; AB, human AB+ve sera.
numerous abnormalities of the immunophenotype and function of T cells in prediabetic subjects and patients with T1D have been reported, including abnormal distributions of cellular subsets characterized by cell surface expression of activation and differentiation marker such as CD25, CD45RA/CD45R0, CD69 and HLA-DR [7–11]. Together with reported imbalances in cytokine production that suggest a dominance of T helper 1 (Th1) over T helper 2 (Th2) type responses in individuals with or at risk from T1D [12–14], these studies support the contention that T cell function is abnormal in T1D, and that this underlies the processes that lead to cell loss. However, evidence in relation to the pathogenetic role of islet autoantigen reactive T cells remains indirect and circumstantial. The study of autoreactive T cells may give important information on disease aetiopathogenesis, identify potential targets for immunotherapy and provide surrogate markers of the islet destructive process, but the characterization of such cells in the peripheral blood of individuals with T1D has proved problematic and yielded less than consistent results [15]. There are many possible explanations for these difficulties, including the inability to assay cells from the inflammatory lesion, the low precursor frequency of autoreactive cells in the periphery and inhibition by populations of regulatory T cells, as well problems associated with assay format and antigen selection. Recently, novel assays capable of identifying disease related autoreactive T cells in individuals with T1D have been reported [16,17]. However, both of these studies relied on at least one cycle of antigen specific expansion prior to detection of T cell reactivity, compromising the ability to draw conclusions on the precursor frequency and original phenotype of these cells. Therefore, the development of sensitive assays able to identify and characterize autoreactive T cells directly ex-vivo remains a major goal in the study of human diabetes. An assay of particular promise in this regard is the enzyme linked immunospot assay for detection of cytokine secreting cells (CK ELISpot) [18]. This assay allows the sensitive detection of antigen specific T cells through their cytokine secretion and may be able to detect responder cell frequencies as low as 1:1,000,000 [19]. Several reports on
the use of the CK ELISpot assays to measure T cell responses in T1D have been published [20–23] and many more are sure to follow. However, as has been the case with proliferation assays for the detection of autoreactive T cells, difficulties may arise in comparing results obtained in different studies, due to many factors including: assay format, nature of cells analysed, choice of antigens, the method of plate reading and the analysis of results. Previously, the use of T cell workshops has been a successful platform for evaluating assay formats and antigen preparations capable of discriminating islet antigen-specific T cell responses [24–26]. In the current study, we report the comparison of CK ELISpot assays in the controlled environment of a ‘wet workshop’, with all participants using blood from the same donors and the same antigen preparations. The specific aims of this workshop were (i) to evaluate the performance characteristics of different CK ELISpot assay formats, (ii) to identify the strengths and weaknesses of differents assays and (iii) to examine different methods of analysing CK ELISpot data. The results of this workshop demonstrate the detection of low-frequency autoreactive T cell responses and show the existence of format-related differences in assay sensitivity, which may guide the selection of CK ELISpot assays for future studies.
2. Materials and methods 2.1. Subjects and participants Fresh heparinized blood samples (300 ml) were drawn from two subjects on the same day at the site of the workshop (Dusseldorf, 24–28th July 2002): one subject has T1D (age 32 years, duration of T1D 20 years, HLA genotype DRB1*0401, DQA1*0301 DQB1*0302) whilst the other has no family history of T1D (age 36 years, HLA genotype DRB1*0401 DRB1*08, DQA1*03 DQB1*0302). Blood samples were immediately divided into aliquots for each workshop participant to prepare fresh peripheral blood mononuclear cells (PBMC) as detailed in Table 1. Where possible, all cytokine assays
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
were performed by all participants on all samples, dependent upon the availability of an appropriate assay format in their laboratory and sufficient blood. There were five participating laboratories. Two performed an assay based on previously published conditions [22,27] (Cellular Technologies Ltd, Cleveland, OH, USA) and are referred to as CTL-1 and CTL-2. Two further participants used the U-CyTech assay format [28] (U-CyTech, Utrecht, NL) and one participant used an assay based on previously published conditions [21] using antibodies available from Mabtech (referred to as the Mabtech assay).
367
numbers per well. To determine a significant response, spot numbers in the presence of stimulus were compared to the relevant control using Mann–Whitney U test (one tailed). The concordance between results obtained by different assays was analysed using contingency tables and Fisher’s exact test (two-tailed). For the comparison of spot numbers counted by different automated readers, linear regression analysis was performed and analysed by Spearman’s rank correlation. All statistical analysis was performed using GraphPad Prism (GraphPad Software Incorporated) and P values <0.05 were considered significant.
2.2. Antigens Phytohaemagglutinin (PHA) (Murex Biotech Ltd, Dartford, UK) at a final concentration of 10 µg/ml and phorbol myristate acetate (PMA)/ionomycin (Sigma Chemical Co, Poole, Dorset, UK) at final concentrations of 10 ng/ml and 1 µM, respectively, were used as polyclonal stimuli. Tetanus toxoid (Pasteur Merieux, Maidenhead, UK) (final concentrations 100 ng/ml and 10 ng/ml) was used as a control antigen. Responses to peptides GAD65 (554–575), IA-2 (853–872) and HSP60 (277–300) were tested at a final concentration of 1 and 5 µM. Peptides were synthesized by solid phase strategies on an automated multiple peptide synthesizer (Abimed AMS 422) and subsequently purified by reverse phase high pressure liquid chromatography. Purity was >95% for all peptides. Prior to distribution to the participants, the peptide preparations were coded for blind testing. Peptides were dissolved in dimethyl sulfoxide (DMSO) and control preparations comprising concentrations of DMSO equivalent to those in the peptide preparations were included as controls (DMSO at 0.01% and 0.05% final concentrations for peptide concentrations of 1 µM and 5 µM respectively). 2.3. CK ELISpot assay and statistical analysis Each participant performed CK ELISpot assays according to their own protocol as detailed in Table 2. Wherever possible, all participants performed assays for each CK (IFN-, IL-4, IL-5, IL-10 and IL-13) but in some cases this was not possible due to insufficient cells or assay availability. Assays were performed in sextupilcate or triplicate wells as indicated. After development, CK ELISpot plates were analysed using the reader of choice for each assay and, in addition, by other readers to allow cross-reader comparison. The following automated plate readers were used: AID Elispot Reader System (AID EliSpot Reader System ELR02, Software Version 2.9.14; Autoimmun Diagnostika GmbH, Strassberg, D); ImmunoSpot Series 2.0 Analyzer (Cellular Technologies Ltd) and Bioreader 3000 (BIOSYS, Karben, D). Results are reported as mean spot
3. Results 3.1. Detection of CK secreting cells in the presence of negative and positive control preparations Results for all CK ELISpot assays, expressed as mean spot number/well, are shown in Tables 3 and 4 for the non-diabetic (control) subject and subject with T1D, respectively. Levels of cytokine secretion in the presence of media alone (with or without DMSO) varied considerably across the assay formats, but were similar within a format independent of operator (i.e. comparing U-CyTech-1 and U-CyTech-2). For IFN- the lowest background values were seen in the U-CyTech assays (combined mean of both participants 0.6 spots/well, range 0.2–1.7), followed by CTL assays (combined mean 3.4 spots/well, range 1.2–6.5) with the Mabtech assay having the highest values (mean 27.7 spots/well, range 11–39). Conversely, for IL-10, responses in the presence of media alone (with or without DMSO) were higher in the U-CyTech assays (combined mean 33 spots/well, range 12–81) than in the CTL assay (mean 2.2 spots/ well, range 2.0–2.5). IL-4 and IL-5 responses in the presence of media alone (with or without DMSO) were low and similar between different assay formats. In the presence of the positive control preparations, PHA and PMA/ionomycin, cytokine secreting cells were consistently detectable across all assay formats (data not shown). 3.2. Detection of CK secreting cells in the presence tetanus toxoid Responses to tetanus toxoid (TT) were measured in order to assess assay performance in the detection of physiological antigen specific T cell recall responses (Tables 3 and 4). The concentrations of TT used were sub-optimal and were selected in order to examine assay sensitivity in the detection of low-level responses. Examples of the detection of IFN- and IL-4 secreting cells in response to TT are shown in Fig. 1. Compared with media alone, significant IFN- responses were
368
Table 2 Culture conditions and CK ELISpot methods used by Workshop participants Laboratory
Culture conditions
CTL-2
Mabtech
U-Cytech-1
U-Cytech-2
No
No
No
Yes
Yes
Pre-incubation plate Pre-incubation cell concentration Pre-incubation time Additional pre-incubation steps
n/a n/a n/a n/a
n/a n/a n/a n/a
n/a n/a n/a n/a
Cell concentration/well on ELISPOT plate Incubation time of cells on ELISPOT plate
1105 cells in 50 µl
3105 cells in 200 µl
IFN- 24 h; IL-4/IL-10 48 h
IFN- 22 h; IL-5 51 h
1105 cells in 200 µl 48 h
48 well 3.7106/well in 500 µl 48 h 500 µl media added after 24 h Non adherant cells from 3105 PBMC 5h
48 well 2106/well in 500 µl 48 h 500 µl media added after 24 h Non adherant cells from 3.3105 PBMC 6h
ELISPOT plate
Whatman small well
Whatman-polyfiltronics
Pall
Nunc Maxisorb
Nunc Maxisorb
ELISPOT plate type Coating antibody
PVDF membrane IFN- Endogen; IL-4/IL-10 Pharmingen O/N@4 (C PBS/1% BSA 1 h@RT
PVDF membrane IFN- Endogen; IL-5 Pharmingen O/N@4 (C PBS/1% BSA 1 h@RT
PVDF membrane Mabtech
Clear U-Cytech
Clear U-Cytech
O/N@4 (C PBS/1%BSA O/N@4 (C
O/N@4 (C PBS/1%BSA O/N@4 (C
IFN- Endogen; IL-5 Pharmingen O/N@4 (C Streptavidin-HRP
U-Cytech
U-Cytech
Incubation time Tertiary reagent
IFN-/IL-4 Endogen; IL-10 Pharmingen O/N@4 (C Streptavidin-HRP
O/N@4 (C IMDM complete 1 h@37 (C Mabtech 2–4 h@37 (C Streptavidin-AP
Incubation time Development reagent
1 h@RT AEC
1 h@RT AEC
1 h@37 (C Goat anti biotin antibody (U-Cytech) 1 h@37(C U-Cytech
1 h@37 (C Goat anti biotin antibody (U-Cytech) 1 h@37(C U-Cytech
Development time Preferred reader
Visually determined CTL
Visually determined CTL
Visually determined BIOSYS
Visually determined BIOSYS
Coating conditions Block reagent Detector antibody
1 h@RT AP Substrate Kit (Bio-Rad) 15 min AID
Abbreviations used: n/a, not applicable; PBS, phosphate buffered saline; BSA, bovine serum albumin; HRP, horseradish peroxidase, AP, alkaline phosphatase; RT, room temperature; O/N overnight; AEC, 3-amino-9-ethylcarbazole.
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
ELISPOT conditions
CTL-1 Pre-incubation step used
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
369
Table 3 Results of CK ELISpot assays performed on control blood sample Culture conditions
Medium DMSO (0.05) DMSO (0.01) TT (100) TT (10) GAD65 (5) GAD65 (1) IA-2 (5) IA-2 (1) HSP60 (5) HSP60 (1)
U-Cytech-1
U-Cytech-2
Mabtech
CTL-1
CTL-2
IFN-
IL-5
IL-10
IL-13
IFN- IL-4
IL-13
IFN- IL-4
IFN- IL-4
IL-10
IFN- IL-5
0.3 1.3 1.7 7.0** 2.8* 0.5 1.8 2.2 1.7 4* 1.3
0.33 0.33 0.67 1.17 1.5 1.0 0.7 1.3 0.2 0.5 1.2
25 12 33 15** 9** 22 35 18 20* 12 17
2.5 1.7 1.3 0.3 2.4* 0.3 1.5 0.3 0.2 0.3 0.3
0.5 0.3 0.3 2.5* 1.7* 3.7* 0.0 0.7 0.3 2.2* 1.7*
0.1 0.3 0.0 0.5 0.3 0.0 0.7 0.0 0.0 0.2 0.0
29 39 39 97** 75* 34 104 40 27* 35 35
4.8 3.6 5.8 4.0 2.8 6.4 6.4 6.2 6.4 5.6 4.0
2.0 2.0 2.5 3.0 4.3 2.0 1.3 3.3 2.3 3.0 2.0
3.8 2.3 4.1 5.5 11.5 2.7 2.5 5.2 3.2 1.8 3.2
1.2 1.7 3.7 4.2** 3.0* 2.3 2.3 2.0 3.7 4.0* 2.3
1.5 2.0 2.0 5.3** 1.5 2.0 0.7 1.3 0.7 3.5 1.0
2.2 2.0 2.8 1.5 2.0 1.7 1.2* 2.5 1.2 1.8 1.7
0.7 0.7 0.5 2.8** 2.0 1.5 0.5 0.7 0.7 0.7 0.5
Abbreviations used: DMSO, dimethyl sulphoxide (%); TT, tetanus toxoid, (ng/ml); GAD65, GAD65 (554–575) (µM); IA-2, IA-2 (853–872) (µM); HSP60, HSP60 (277–300) (µM). Results are expressed as mean spot numbers/well. Levels of significance are indicated: ** P<0.01. * P<0.05. Table 4 Results of CK ELISpot assays performed on T1D blood sample Culture conditions
Medium DMSO (0.05) DMSO (0.01) TT (100) TT (10) GAD65 (5) GAD65 (1) IA-2 (5) IA-2 (1) HSP60 (5) HSP60 (1)
U-Cytech-2
Mabtech
CTL-1
CTL-2
IFN-
IL-4
IL-10
IL-13
IFN-
IL-4
IFN-
IL-4
IL-10
IFN-
IL-5
0.2 0.8 0.5 30** 11** 1.3 2.0 1.5 0.8 1.0 1.5
8.3 6.8 7.3 19** 10 5.5 6.2 7.0 7.0 6.8 6.3
81 31 18 32** 12** 55** 33** 25 22 20 21
0.2 0.3 0.0 3.5** 0.5 0.0 0.0 0.3 0.3 0.0 0.7
11 27 21 43** 27* 24 28 23 19 24 25
0.8 3.3 2.0 3.8** 1.2 1.7 0.7 1.0 1.3 1.5 2.3
1.2 1.4 1.2 5.2** 3.2 0.6 1.6 3.8* 2.4* 1.6 0.8
1.5 4.3 5.0 5.3** 3.8* 6.0 4.3 5.5 5.7 3.0 4.7
2.0 2.0 2.5 3.0 4.3 2.0 1.3 3.3 2.3 3.0 2.0
6.5 3.5 2.7 4.7 3.0 5.0 0.7 6.0 1.2 6.8 2.0
0.5 1.3 1.7 2.8 2.0 1.5 0.5 0.7 0.7 0.7 0.5
Abbreviations used: DMSO, dimethyl sulphoxide (%); TT, tetanus toxoid (ng/ml); GAD65, GAD65 (554–575) (µM); IA-2, IA-2 (853–872) (µM); HSP60, HSP60 (277–300) (µM). Results are expressed as mean spot numbers/well. Levels of significance are indicated: ** P<0.01. * P<0.05.
detected to both concentrations of TT by both U-CyTech-1 and U-CyTech-2 assays and the Mabtech assay for both the T1D and control sample (where tested). The CTL-1 assay detected an IFN- response in the T1D sample only, and only at the higher concentration of TT. No responses to TT were detected by the CTL-2 assay. The frequency of responses detected varied considerably between assay formats, with the Mabtech assay consistently giving a higher number of TT specific spots/well than the U-CyTech or CTL assays. Tetanus toxoid specific IL-4 responses were also detected in both samples by the U-CyTech-2 (analysis not performed by U-CyTech-1) and Mabtech assays and by CTL-1 in the T1D sample (assay not performed by CTL-2). The frequency of the responses was similar in all assay formats. A significant reduction in the number
of IL-10 producing cells was detected at both concentrations of TT using the U-CyTech assay format in the T1D sample (assay only performed by U-CyTech-2) and control sample (assay only performed by U-CyTech-1). This reduction in CK secreting cells was not seen for either sample in the only other assay for IL-10 (CTL-1). Of the assays performed to detect TT-specific IL-5 secretion (CTL-2 and U-CyTech-1) only CTL-2 detected a significant increase in spot number. Only U-CyTech assays analysed IL-13 responses, which were detectable in both subjects. 3.3. Responses to autoantigen peptides Two weeks prior to the workshop, samples from the T1D and control subjects were tested against a panel of
370
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 1. IFN- and IL-4 responses to tetanus toxoid. Responses from T1D are shown in the upper three panels and responses from control individual (Ctrl) are shown in the lower three panels. Filled squares=CTL-1, open squares=CTL-2, open circles=U-CyTech-1, closed circles=U-CyTech-2 and closed triangles=Mabtech. Asterisks indicate the degree of significance (*P<0.05, **P< 0.01) comparing antigen-stimulated response versus spot formation in presence of medium.
peptides based on sequences from the diabetesassociated autoantigens IA-2, GAD65, proinsulin and HSP60. In this pre-testing, the control subject showed increased numbers of IFN- secreting cells in response to HSP60 (277–300), and the T1D subject in response to GAD65 (554–575) and IA-2 (853–872). In all cases, these responses represented a low frequency of autoreactive cells close to the limit of detection of the assay used (data not shown). In the workshop, a significant IFN- response to HSP60 (277–300) was seen in the control individual by both the U-CyTech-1 and U-CyTech-2 assays (Fig. 2) but not detected by other assays. In addition, an IL-4 specific response was detected to the same peptide in the U-CyTech assay (assay performed by U-CyTech-2 only)
but not by Mabtech or CTL-1 (assay not performed by CTL-2). As in pre-testing, the responses detected were present at low levels corresponding to <5 spots well. Reproducibility of these control subject responses for the U-CyTech-1 assays performed in pre-testing and during the workshop and the U-CyTech-2 assay performed during the workshop is shown in Fig. 2B. In the T1D sample, a significant IFN- response to IA-2 (853–872) was observed at both concentrations in one assay (CTL-1). Although no assays detected an IFN- response to GAD65 (554–575) in the T1D sample, an IL-10 specific response to this peptide was detected by one assay (U-CyTech-2) at both concentrations. As in the pre-testing, no response to HSP60 (277–300) was observed in the T1D sample.
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
371
Fig. 2. IFN- response to HSP60 (277–300) peptide by control subject and assay reproducibility. (A) Filled squares=CTL-1, open squares=CTL-2, open circles=U-CyTech-1, closed circles=U-CyTech-2 and closed triangles=Mabtech. Asterisks indicate the degree of significance (*P<0.05) comparing antigen-stimulated response versus spot formation in presence of medium control. (B) Results of U-CyTech IFN- ELISPOT analysis of response of control subject to HSP60 (277–300) performed at two different time-points and by two different operators. Open circles=U-CyTech-1, closed circles=U-CyTech-2, for assays performed during workshop; closed diamonds=U-CyTech-1, for assay performed during pre-testing 2 weeks before workshop.
3.4. Combinatorial analysis of cytokine responses A major advantage of CK ELISpot analysis over techniques such as proliferation and limiting dilution analyses is that they provide information on the class of the effector response, which is a reflection of the balance of cytokines secreted in response to a specific antigen (e.g. Th1, Th2 or Th0 profiles). In addition to assessing differences in assay performance, the workshop also aimed to identify novel ways of presenting the data generated by CK ELISpot analysis. We therefore explored the utility of combining responses defined by different cytokines to a given antigen in stacked bars (Fig. 3). This analysis provides an overall representation of the immune response to a specific antigenic stimulus, showing both increases and decreases in cytokine production (e.g. TT response comprises increased secretion of IFN-, IL-4 and IL-13 but a decrease in IL-10; Fig. 3) and should enable better discrimination of an antigen specific response. 3.5. Concordance between assays The degree of concordance of responses (including results obtained with the positive control preparations PHA, and PMA/ionomycin) between all assay formats was assessed using contingency tables (Table 5). Although this comparison is complex, due to the different numbers of assays performed in common by different participants, several conclusions can be drawn. Concordance is stronger between assays of the same type than between assay types. For example, concord-
ance between U-CyTech-1 and -2 assays is more significant than that between either of these and either of the CTL assays. Similarly, the concordance between CTL-1 and -2 assays is more significant than that between either of these and either of the U-CyTech assays. In addition, there is good concordance between the Mabtech assay and assays of other formats. The concordance for IFN- assays is also represented graphically using a Venn diagram approach (Fig. 4). These analyses suggest that there is greater similarity between the Mabtech and U-CyTech assays than between either of these formats and the CTL assays. 3.6. Comparison of CK ELISpot readers A key factor in the analysis of CK ELISpot assays is the objective enumeration of spots, for which several automated readers are available commercially. To compare results obtained with different readers, plates generated during the workshop were analysed by Autoimmun Diagnostika (AID), Cellular Technologies Ltd (CTL) and BIOSYS readers. All three readers were able to read PVDF membrane plates from the Mabtech and CTL assays. Transparent ELISA plates, used in the U-CyTech assay, were only readable by the AID and BIOSYS machines. Results comparing the reading of the CTL-1 and -2 and Mabtech IFN- assays are shown in Fig. 5. Although the numbers of spots read varied, a highly significant correlation between the spot numbers obtained by all readers for each well was observed (P<0.0001).
372
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 3. Cytokine profile of response to tetanus toxoid (TT) (100 ng/ml) by T1D subject. Results are shown as accumulated mean spot numbers/well. Solid bar=IFN-, vertical stripes=IL-4, diagonal stripes=IL-5, un-shaded=IL-10 and hatched=IL-13. Table 5 Concordance between different assay formats
U-CyTech-2 Mabtech CTL-1 CTL-2
U-CyTech-1
U-CyTech-2
Mabtech
CTL-1
P=0.0045 (n=20) n.s. (n=10) P=0.033 (n=20) P=0.031 (n=20)
P<0.0001 (n=40) P=0.021 (n=40) n.s. (n=20)
P=0.0005 (n=40) P=0.026 (n=20)
P=0.014 (n=20)
Values are calculated by comparison of significant results using contingency tables. Number of results used for construction of contingency tables is shown in brackets.
4. Discussion The identification of assays capable of distinguishing islet antigen specific T cell responses is an important research goal, and has been facilitated to date by the performance of a number of workshop studies. The report on the first IDS T Cell Workshop highlighted a number of major limitations with both the antigen preparations and assay methodologies employed [26]. The second Workshop evaluated a large number of antigen preparations and subsequently, a panel of islet antigens and islet antigen derived peptides were made available for assay evaluation [24]. An assay format that has received particular recent interest is the CK ELISpot. In a recent review [29] the relative merits of different techniques for the detection of islet autoreactive T cells was discussed, with particular emphasis on CK ELISpot, and therefore a detailed discussion of the advantages and limitations of the various assay formats will not be entered into here. As a direct means of comparison of different CK ELISpot assay formats a wet workshop was organized,
in which all participants used the same fresh blood samples and identical blinded antigen preparations. A number of factors, including need to pre-test to assess responsiveness, a preference for using fresh rather than cryopreserved samples, size of blood sample required and the timing constraints of the workshop precluded the use of a blood sample from a newly diagnosed T1D patient. As an alternative, samples were obtained from two of the workshop participants, demonstrated in pre-testing to have responses to selected test antigens. It should be noted that in the design of this workshop, the pre-testing of subjects was performed only using one assay format (U-CyTech) and this may have favoured these assays in the ability to reproduce these results. The responses to islet antigen peptides by autoreactive T cells detected in pre-testing demonstrated very low frequencies of responder cells (in the range of 1:100,000) and therefore provided a great challenge to the sensitivity of the assays. Low dose tetanus toxoid specific responses were, therefore, also included to provide a robust means of assessing sensitivity. Assays may vary in
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
373
Fig. 4. Relationship between results of IFN- CK ELISpot from different assay formats. Comparison of IFN- CK ELISpot results from (A) T1D and (B) control individuals. High concentrations (TT 100 ng/ml peptides 5 µM) are shown in bold and low concentrations (TT 10 ng/ml peptides 1 µM) in normal script. Abbreviations used: TT, tetanus toxoid; GAD65, GAD65 (554–575); IA-2, IA-2 (853–872); HSP60, HSP60 (277–300).
the concentration of antigens required to give optimal performance [21]. Whilst there are clearly differences in assay performance, the results presented demonstrate that low level recall antigen- and peptide-specific autoreactive T cell responses are detectable by CK ELISpot and may be reproduced over time within the same assay format. Good levels of concordance were observed both between assays of a similar format and those of different formats. As expected the level of concordance was generally greatest between similar assays. All assays were able to detect significant cytokine secretion in response to the polyclonal stimulators PHA and PMA/ionomycin. These stimuli are sufficiently powerful that in some cases accurate spot numbers were difficult to assign due to the high number of spots and background signal resulting from secreted cytokine in the media. Therefore, for accurate studies of the total numbers of cytokine producing cells in response to these stimuli it may be necessary to titrate the number of cells used per well. At the higher tetanus concentration four of the five assays were able to detect significant IFN- specific T cell responses although the level of significance and total spot numbers varied considerably. At the lower tetanus toxoid concentration only the Mabtech and U-CyTech assays detected IFN- reactivity. Reproducible autoreactive T cell responses (i.e. concordant with pretesting) were detected in the control individual by the U-CyTech assays (IFN-) at both peptide concentrations and also by IL-4 secretion. Reproducible peptide responses in the T1D sample were detected to the IA-2 peptide by CTL-1 (IFN-). Although no IFN- response was observed to the GAD65 peptide, this stimulus induced IL-10 secretion measurable by the U-CyTech-2 assay, indicating the detection of a response. These
results suggest that different assays have different sensitivities, that may vary according to the type or strength of stimulus used. At low levels of recall antigen the Mabtech and U-CyTech assays displayed a higher level of sensitivity than the CTL assays. However, for the detection of extremely low-level peptide specific autoreactivity the U-CyTech and CTL assays were superior. A major factor in the ability of CK ELISpot assays to distinguish low frequency antigen specific T cell responses is the nature of the signal-to-noise ratio. Clearly, variables capable of influencing this ratio arise at all stages of the assay with a consequent influence on spot detection. These include: composition of cell preparation, number and density of cells incubated; the total time and window of cytokine capture and the reagents used for capture and detection. With such a high number of variables it is difficult for a single workshop to identify their relative contributions to differences in assay performance. However, some general conclusions may be drawn. A wide range in the number of spots observed in unstimulated wells (noise) was observed, and was associated with assay type as well as being consistent for both samples. In the case of IFN- assays, the unstimulated spot number correlates well with the time of incubation of cells on the CK ELISpot plate: for example, spot number for Mabtech (48 h incubation)>CTL (24 h)>U-CyTech (5 h). Similarly, the long capture time of the Mabtech assay may have contributed to the large number of spots in response to tetanus toxoid (signal). Cytokine capture time may thus explain the wide differences in spontaneous secretion reported by different groups for T1D patients, at risk and control subjects [21,22]. Different times of incubation may be also be required for optimal detection of responses to whole protein antigens, which require
374
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 5. Comparison of spot numbers obtained by three different readers. (A) plot of values obtained individual readers against mean values for the three readers. Values from the CTL reader are shown as square symbols, AID by triangles and Biosys by circles. (B–D) comparison of values obtained by individual readers. Linear regression analysis was performed and goodness of fit is expressed as r2. For all comparisons P<0.0001 (Spearman’s rank correlation).
intracellular processing, compared to responses to peptides which do not. This may have contributed to the greater sensitivity of Mabtech and U-CyTech assays in detecting tetanus responses. Clearly a good balance of signal-to-noise is required for optimal assay sensitivity and this may vary depending upon the type of response measured. Whereas a large signal may make detection of tetanus responses more impressive in terms of the total number of spots detected, the detection of autoreactive responses, present at very low frequencies, may be facilitated in assays with a low spontaneous response. In the detection of very low frequency responses, the signal may be lost in the ‘noise’ of assays with a high spontaneous response. In addition, the total number of lymphocytes in an assay will clearly have an effect on the power of the analysis to detect a significant response, with large cell numbers required to detect very low frequency responses. As statistical analysis of multiple wells was used to determine positivity in the present
study, it is clear that inter-well variation will be an important factor in assay sensitivity. In addition to time of cytokine capture, a major difference in assay format is the inclusion of a preincubation step in the absence of CK capture antibodies in the U-CyTech assays. This step is included to allow incubation of cells at a high density for a prolonged period without the generation of a high background and is one of several modifications introduced with the specific aim of detecting low frequency cells producing multiple cytokines [28,30] (P van der Meide and B.O. Roep, personal communication). This step may also allow cross-regulatory effects of different cytokines to occur, as they are not captured and takes advantage of the fact that tissue culture plates are coated to optimize the quality of cellular immunological responses. Moreover, the use of an initial bulk-culture step may result in reduced inter-well variability in the final CK ELISpot plate. Differences between assays may also be
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
attributable to plate selection as previously demonstrated [31]. In this workshop PVDF and ELISA plates were used, although the study design does not allow direct comparison of the effects of these two plate types. Ultimately, each of the assay types evaluated in this WS can be further improved to allow detection of low-frequency autoreactive T-cells producing various cytokine profiles. A consensus on the definition of a positive CK ELISpot response remains to be established. Methods used include comparison of absolute spot numbers, with or without subtraction of background, the calculation of stimulation indices or the comparison of different cytokines leading to the calculation of Th1/Th2 ratios. In the present study, the subtraction of background counts was not favoured as it masks the extent of the background response (for example, results for medium versus stimulus of 0 and 5 spots, respectively, clearly represents a different result from 100 and 105 spots) and also precludes statistical analysis for positivity. Likewise, although the generation of stimulation indices does take into account both the signal and noise in an assay, it is less amenable to statistical analysis. In this workshop we chose the comparison of total spot numbers from multiple wells by the Mann–Whitney U test. We suggest that where possible, CK ELISpot studies should report both the magnitude, as mean or median, and statistical significance of responses to allow a full interpretation of the results. However, care will be required in the choice of test for statistical comparison and in the interpretation of the results, particularly when multiple wells are used and inter-well variation is small. In summary, this CK ELISpot workshop has demonstrated the detection of low-level autoreactive T cell responses. Differences between CK ELISpot formats in terms of sensitivity will require further investigation and this will lead to the refinement of these assays as tools for unravelling the complex mechanisms of cell destruction that lead to T1D.
[4]
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
Acknowledgements [17]
We are thankful to the financial support for this workshop by U-Cytech, Utrecht, The Netherlands; CTL, Europe; Biosys, Germany; AID, Germany and Mabtech Sweden.
[18]
[19]
References [1] Castano L, Eisenbarth GS. Type-I diabetes: a chronic autoimmune disease of human, mouse, and rat. Annu Rev Immunol 1990;8:647–79. [2] Gepts W. Pathologic anatomy of the pancreas in juvenile diabetes mellitus. Diabetes 1965;14:619–33. [3] Bougneres PF, Carel JC, Castano L, Boitard C, Gardin JP, Landais P et al. Factors associated with early remission of type I
[20]
[21]
375
diabetes in children treated with cyclosporine. N Engl J Med 1988;318:663–70. Lampeter EF, Homberg M, Quabeck K, Schaefer UW, Wernet P, Bertrams J et al. Transfer of insulin-dependent diabetes between HLA-identical siblings by bone marrow transplantation. Lancet 1993;341:1243–4. Sibley RK, Sutherland DE, Goetz F, Michael AF. Recurrent diabetes mellitus in the pancreas iso- and allograft. A light and electron microscopic and immunohistochemical analysis of four cases. Lab Invest 1985;53:132–44. Martin S, Wolf-Eichbaum D, Duinkerken G, Scherbaum WA, Kolb H, Noordzij JG et al. Development of type 1 diabetes despite severe hereditary B-lymphocyte deficiency. N Engl J Med 2001;345:1036–40. Peakman M, Leslie RD, Vergani D. Immunological studies on type 1 diabetes in identical twins. Arch Dis Child 1993;69:97–9. Smerdon RA, Peakman M, Hussain MJ, Alviggi L, Watkins PJ, Leslie RD et al. Increase in simultaneous coexpression of naive and memory lymphocyte markers at diagnosis of IDDM. Diabetes 1993;42:127–33. Buschard K, Damsbo P, Ropke C. Activated CD4+ and CD8+ T-lymphocytes in newly diagnosed type 1 diabetes: a prospective study. Diabet Med 1990;7:132–6. Petersen LD, Duinkerken G, Bruining GJ, van Lier RA, de Vries RR, Roep BO. Increased numbers of in vivo activated T cells in patients with recent onset insulin-dependent diabetes mellitus. J Autoimmun 1996;9:731–7. Gessl A, Waldhausl W. Increased CD69 and human leukocyte antigen-DR expression on T lymphocytes in insulin-dependent diabetes mellitus of long standing. J Clin Endocrinol Metab 1998; 83:2204–9. Szelachowska M, Kretowski A, Kinalska I. Decreased in vitro IL-4 [corrected] and IL-10 production by peripheral blood in first degree relatives at high risk of diabetes type-I. Horm Metab Res 1998;30:526–30. Halminen M, Juhela S, Vaarala O, Simell O, Ilonen J. Induction of interferon-gamma and IL-4 production by mitogen and specific antigens in peripheral blood lymphocytes of Type 1 diabetes patients. Autoimmunity 2001;34:1–8. Rapoport MJ, Mor A, Vardi P, Ramot Y, Winker R, Hindi A et al. Decreased secretion of Th2 cytokines precedes Up-regulated and delayed secretion of Th1 cytokines in activated peripheral blood mononuclear cells from patients with insulin-dependent diabetes mellitus. J Autoimmun 1998;11:635–42. Roep BO. Autoreactive T cells in endocrine/organ-specific autoimmunity: why has progress been so slow? Springer Semin Immunopathol 2002;24:261–71. Viglietta V, Kent SC, Orban T, Hafler DA. GAD65-reactive T cells are activated in patients with autoimmune type 1a diabetes. J Clin Invest 2002;109:895–903. Reijonen H, Novak EJ, Kochik S, Heninger A, Liu AW, Kwok WW et al. Detection of GAD65-specific T-cells by major histocompatibility complex class II tetramers in type 1 diabetic patients and at-risk subjects. Diabetes 2002;51:1375–82. Klinman DM. ELISPOT assay to detect cytokine-secreting murine and human cells. In: Coico R, editor. Current protocols in immunology, New York: John Wiley & Sons; 1994, p. 6, 19, 11. Karulin AY, Hesse MD, Tary-Lehmann M, Lehmann PV. Singlecytokine-producing CD4 memory cells predominate in type 1 and type 2 immunity. J Immunol 2000;164:1862–72. Kotani R, Nagata M, Moriyama H, Nakayama M, Yamada K, Chowdhury SA et al. Detection of GAD65-reactive T-cells in type 1 diabetes by immunoglobulin-free ELISPOT assays. Diabetes Care 2002;25:1390–7. Karlsson MG, Lawesson SS, Ludvigsson J. Th1-like dominance in high-risk first-degree relatives of type I diabetic patients. Diabetologia 2000;43:742–9.
376
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
[22] Alleva DG, Crowe PD, Jin L, Kwok WW, Ling N, Gottschalk M et al. A disease-associated cellular immune response in type 1 diabetics to an immunodominant epitope of insulin. J Clin Invest 2001;107:173–80. [23] Raz I, Elias D, Avron A, Tamir M, Metzger M, Cohen IR. Beta-cell function in new-onset type 1 diabetes and immunomodulation with a heat-shock protein peptide (DiaPep277): a randomised, double-blind, phase II trial. Lancet 2001; 358:1749–53. [24] Peakman M, Tree TI, Endl J, van Endert P, Atkinson MA, Roep BO. Characterization of preparations of GAD65, proinsulin, and the islet tyrosine phosphatase IA-2 for use in detection of autoreactive T-cells in type 1 diabetes: report of phase II of the Second International Immunology of Diabetes Society Workshop for Standardization of T-cell assays in type 1 diabetes. Diabetes 2001;50:1749–54. [25] Roep B, Atkinson M, Honeyman M, Peakman M. T Cell Workshop update 2002. [26] Roep BO, Atkinson MA, van Endert PM, Gottlieb PA, Wilson SB, Sachs JA. Autoreactive T cell responses in insulin-dependent (Type 1) diabetes mellitus. Report of the first international
[27]
[28]
[29]
[30]
[31]
workshop for standardization of T cell assays. J Autoimmun 1999;13:267–82. Hesse MD, Karulin AY, Boehm BO, Lehmann PV, Tary-Lehmann M. A T cell clone’s avidity is a function of its activation state. J Immunol 2001;167:1353–61. van Halteren AG, van Etten E, de Jong EC, Bouillon R, Roep BO, Mathieu C. Redirection of human autoreactive T-cells upon interaction with dendritic cells modulated by TX527, an analog of 1,25 dihydroxyvitamin D(3). Diabetes 2002;51:2119–25. Meierhoff G, Ott PA, Lehmann PV, Schloot NC. Cytokine detection by ELISPOT: relevance for immunological studies in type 1 diabetes. Diabetes Metab Res Rev 2002;18:367–80. Schmittel A, Keilholz U, Bauer S, Kuhne U, Stevanovic S, Thiel E et al. Application of the IFN-gamma ELISPOT assay to quantify T cell responses against proteins. J Immunol Methods 2001;247:17–24. Ronnelid J, Klareskog L. A comparison between ELISPOT methods for the detection of cytokine producing cells: greater sensitivity and specificity using ELISA plates as compared to nitrocellulose membranes. J Immunol Methods 1997; 200:17–26.
Comparison of cytokine ELISpot assay formats for the detection of islet antigen autoreactive T cells Report of the third immunology of diabetes society T-cell workshop Nanette C. Schloot a, Guido Meierhoff a, Maria Karlsson Faresjo¨ b, Patrick Ott c, Amy Putnam d, Paul Lehmann c, Peter Gottlieb d, Bart O. Roep e, Mark Peakman f*, Timothy Tree f a
German Diabetes Research Institute at the Heinrich Heine University, Du¨sseldorf, Germany b Division of Pediatrics, Clinical Research Center, Health University, Linko¨ping, Sweden c Department of Pathology, Case Western Reserve University, Cleveland, OH, USA d Barbara Davis Center, University of Colorado Health Sciences Center, Denver, CO, USA e Leiden University Medical Centre, Department Immunohaematology and Blood Transfusion, Leiden, The Netherlands f Department of Immunology, Guy’s, King’s and St Thomas’ School of Medicine, King’s College, Denmark Hill Campus, Rayne Institute, 123 Coldharbour Lane, London SE5 9NU, UK Received 2 June 2003; revised 19 June 2003; accepted 21 July 2003
Abstract The identification of sensitive assay formats capable of distinguishing islet autoreactive T cells directly ex vivo in blood is a major goal in type 1 diabetes research. Recently, much interest has been shown in the cytokine enzyme linked immunospot assay (CK ELISpot), an assay potentially capable of fulfilling these difficult criteria. To address the utility of this assay in detecting autoreactive T cells, a ‘wet’ workshop was organized using the same fresh blood sample and coded antigens. Five different laboratories participated, using three distinct CK ELISpot assay formats. Samples from two subjects were pre-tested for responses to sub-optimal concentrations of tetanus toxoid, representing a low frequency recall response, and peptides from diabetes associated autoantigens GAD65, IA-2 and HSP60. All participants measured interferon- production and combinations of interleukins-4, -5, -10 and -13. In the workshop 4 of 5 laboratories detected low frequency recall responses in both subjects and 3 of 5 detected at least one of the autoreactive peptide responses concordant with pre-testing. Significant assay format related differences in sensitivity and signal-to-noise ratio were observed. The results demonstrate the potential for detection of low-level autoreactive T cell responses and identify assay characteristics that will be useful for studies in type 1 diabetes. 2003 Elsevier Ltd. All rights reserved. Keywords: ELISpot; T cell; Diabetes; Workshop
1. Introduction Type 1 diabetes (T1D) is an autoimmune disease characterized by the selective destruction of the insulin producing pancreatic cells [1]. T cells are believed to * Corresponding author. Tel.: +44-2078485956; fax: +44-2078485953. E-mail address: [email protected] (M. Peakman). 0896-8411/03/$ - see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0896-8411(03)00111-2
play a key role in this process, and evidence for this assertion includes a predominance of these cells in the islet infiltrate close to diagnosis [2], the effects of immunosuppressive agents on disease progression [3], ‘adoptive transfer’ case reports on the development of T1D after transplantation of bone marrow [4] or pancreas from diabetic to non-diabetic recipient [5], and more recently a report of the development of T1D in an individual lacking B lymphocytes [6]. Moreover,
366
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Table 1 Methods of PBMC isolation and culture media used by Workshop participants Laboratory
Dilution of blood Separation reagent Spin Wash Assay media Serum
CTL-1
CTL-2
Mabtech
U-Cytech-1
U-Cytech-2
1:2 in PBS Cell Sept 1800 g/30 min 2 in RPMI RPMI 5% HI AB
None CPT tubes 1800 g/20 min 2 in RPMI RPMI 1% HI FCS
1:2 in RPMI Pharmacia hypaque 400 g/30 min 3 in RPMI IMDM 5% HI FCS
1:1.5 in PBS Maxiplus 999 g/20 min 3 in PBS RPMI 10% HI AB pooled
1:2 in RPMI Maxiplus 400 g/40 min 2 in RPMI RPMI 10% HI AB pooled
Abbreviations used: PBS, phosphate buffered saline; HI, heat inactivated; FCS, foetal calf sera; AB, human AB+ve sera.
numerous abnormalities of the immunophenotype and function of T cells in prediabetic subjects and patients with T1D have been reported, including abnormal distributions of cellular subsets characterized by cell surface expression of activation and differentiation marker such as CD25, CD45RA/CD45R0, CD69 and HLA-DR [7–11]. Together with reported imbalances in cytokine production that suggest a dominance of T helper 1 (Th1) over T helper 2 (Th2) type responses in individuals with or at risk from T1D [12–14], these studies support the contention that T cell function is abnormal in T1D, and that this underlies the processes that lead to cell loss. However, evidence in relation to the pathogenetic role of islet autoantigen reactive T cells remains indirect and circumstantial. The study of autoreactive T cells may give important information on disease aetiopathogenesis, identify potential targets for immunotherapy and provide surrogate markers of the islet destructive process, but the characterization of such cells in the peripheral blood of individuals with T1D has proved problematic and yielded less than consistent results [15]. There are many possible explanations for these difficulties, including the inability to assay cells from the inflammatory lesion, the low precursor frequency of autoreactive cells in the periphery and inhibition by populations of regulatory T cells, as well problems associated with assay format and antigen selection. Recently, novel assays capable of identifying disease related autoreactive T cells in individuals with T1D have been reported [16,17]. However, both of these studies relied on at least one cycle of antigen specific expansion prior to detection of T cell reactivity, compromising the ability to draw conclusions on the precursor frequency and original phenotype of these cells. Therefore, the development of sensitive assays able to identify and characterize autoreactive T cells directly ex-vivo remains a major goal in the study of human diabetes. An assay of particular promise in this regard is the enzyme linked immunospot assay for detection of cytokine secreting cells (CK ELISpot) [18]. This assay allows the sensitive detection of antigen specific T cells through their cytokine secretion and may be able to detect responder cell frequencies as low as 1:1,000,000 [19]. Several reports on
the use of the CK ELISpot assays to measure T cell responses in T1D have been published [20–23] and many more are sure to follow. However, as has been the case with proliferation assays for the detection of autoreactive T cells, difficulties may arise in comparing results obtained in different studies, due to many factors including: assay format, nature of cells analysed, choice of antigens, the method of plate reading and the analysis of results. Previously, the use of T cell workshops has been a successful platform for evaluating assay formats and antigen preparations capable of discriminating islet antigen-specific T cell responses [24–26]. In the current study, we report the comparison of CK ELISpot assays in the controlled environment of a ‘wet workshop’, with all participants using blood from the same donors and the same antigen preparations. The specific aims of this workshop were (i) to evaluate the performance characteristics of different CK ELISpot assay formats, (ii) to identify the strengths and weaknesses of differents assays and (iii) to examine different methods of analysing CK ELISpot data. The results of this workshop demonstrate the detection of low-frequency autoreactive T cell responses and show the existence of format-related differences in assay sensitivity, which may guide the selection of CK ELISpot assays for future studies.
2. Materials and methods 2.1. Subjects and participants Fresh heparinized blood samples (300 ml) were drawn from two subjects on the same day at the site of the workshop (Dusseldorf, 24–28th July 2002): one subject has T1D (age 32 years, duration of T1D 20 years, HLA genotype DRB1*0401, DQA1*0301 DQB1*0302) whilst the other has no family history of T1D (age 36 years, HLA genotype DRB1*0401 DRB1*08, DQA1*03 DQB1*0302). Blood samples were immediately divided into aliquots for each workshop participant to prepare fresh peripheral blood mononuclear cells (PBMC) as detailed in Table 1. Where possible, all cytokine assays
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
were performed by all participants on all samples, dependent upon the availability of an appropriate assay format in their laboratory and sufficient blood. There were five participating laboratories. Two performed an assay based on previously published conditions [22,27] (Cellular Technologies Ltd, Cleveland, OH, USA) and are referred to as CTL-1 and CTL-2. Two further participants used the U-CyTech assay format [28] (U-CyTech, Utrecht, NL) and one participant used an assay based on previously published conditions [21] using antibodies available from Mabtech (referred to as the Mabtech assay).
367
numbers per well. To determine a significant response, spot numbers in the presence of stimulus were compared to the relevant control using Mann–Whitney U test (one tailed). The concordance between results obtained by different assays was analysed using contingency tables and Fisher’s exact test (two-tailed). For the comparison of spot numbers counted by different automated readers, linear regression analysis was performed and analysed by Spearman’s rank correlation. All statistical analysis was performed using GraphPad Prism (GraphPad Software Incorporated) and P values <0.05 were considered significant.
2.2. Antigens Phytohaemagglutinin (PHA) (Murex Biotech Ltd, Dartford, UK) at a final concentration of 10 µg/ml and phorbol myristate acetate (PMA)/ionomycin (Sigma Chemical Co, Poole, Dorset, UK) at final concentrations of 10 ng/ml and 1 µM, respectively, were used as polyclonal stimuli. Tetanus toxoid (Pasteur Merieux, Maidenhead, UK) (final concentrations 100 ng/ml and 10 ng/ml) was used as a control antigen. Responses to peptides GAD65 (554–575), IA-2 (853–872) and HSP60 (277–300) were tested at a final concentration of 1 and 5 µM. Peptides were synthesized by solid phase strategies on an automated multiple peptide synthesizer (Abimed AMS 422) and subsequently purified by reverse phase high pressure liquid chromatography. Purity was >95% for all peptides. Prior to distribution to the participants, the peptide preparations were coded for blind testing. Peptides were dissolved in dimethyl sulfoxide (DMSO) and control preparations comprising concentrations of DMSO equivalent to those in the peptide preparations were included as controls (DMSO at 0.01% and 0.05% final concentrations for peptide concentrations of 1 µM and 5 µM respectively). 2.3. CK ELISpot assay and statistical analysis Each participant performed CK ELISpot assays according to their own protocol as detailed in Table 2. Wherever possible, all participants performed assays for each CK (IFN-, IL-4, IL-5, IL-10 and IL-13) but in some cases this was not possible due to insufficient cells or assay availability. Assays were performed in sextupilcate or triplicate wells as indicated. After development, CK ELISpot plates were analysed using the reader of choice for each assay and, in addition, by other readers to allow cross-reader comparison. The following automated plate readers were used: AID Elispot Reader System (AID EliSpot Reader System ELR02, Software Version 2.9.14; Autoimmun Diagnostika GmbH, Strassberg, D); ImmunoSpot Series 2.0 Analyzer (Cellular Technologies Ltd) and Bioreader 3000 (BIOSYS, Karben, D). Results are reported as mean spot
3. Results 3.1. Detection of CK secreting cells in the presence of negative and positive control preparations Results for all CK ELISpot assays, expressed as mean spot number/well, are shown in Tables 3 and 4 for the non-diabetic (control) subject and subject with T1D, respectively. Levels of cytokine secretion in the presence of media alone (with or without DMSO) varied considerably across the assay formats, but were similar within a format independent of operator (i.e. comparing U-CyTech-1 and U-CyTech-2). For IFN- the lowest background values were seen in the U-CyTech assays (combined mean of both participants 0.6 spots/well, range 0.2–1.7), followed by CTL assays (combined mean 3.4 spots/well, range 1.2–6.5) with the Mabtech assay having the highest values (mean 27.7 spots/well, range 11–39). Conversely, for IL-10, responses in the presence of media alone (with or without DMSO) were higher in the U-CyTech assays (combined mean 33 spots/well, range 12–81) than in the CTL assay (mean 2.2 spots/ well, range 2.0–2.5). IL-4 and IL-5 responses in the presence of media alone (with or without DMSO) were low and similar between different assay formats. In the presence of the positive control preparations, PHA and PMA/ionomycin, cytokine secreting cells were consistently detectable across all assay formats (data not shown). 3.2. Detection of CK secreting cells in the presence tetanus toxoid Responses to tetanus toxoid (TT) were measured in order to assess assay performance in the detection of physiological antigen specific T cell recall responses (Tables 3 and 4). The concentrations of TT used were sub-optimal and were selected in order to examine assay sensitivity in the detection of low-level responses. Examples of the detection of IFN- and IL-4 secreting cells in response to TT are shown in Fig. 1. Compared with media alone, significant IFN- responses were
368
Table 2 Culture conditions and CK ELISpot methods used by Workshop participants Laboratory
Culture conditions
CTL-2
Mabtech
U-Cytech-1
U-Cytech-2
No
No
No
Yes
Yes
Pre-incubation plate Pre-incubation cell concentration Pre-incubation time Additional pre-incubation steps
n/a n/a n/a n/a
n/a n/a n/a n/a
n/a n/a n/a n/a
Cell concentration/well on ELISPOT plate Incubation time of cells on ELISPOT plate
1105 cells in 50 µl
3105 cells in 200 µl
IFN- 24 h; IL-4/IL-10 48 h
IFN- 22 h; IL-5 51 h
1105 cells in 200 µl 48 h
48 well 3.7106/well in 500 µl 48 h 500 µl media added after 24 h Non adherant cells from 3105 PBMC 5h
48 well 2106/well in 500 µl 48 h 500 µl media added after 24 h Non adherant cells from 3.3105 PBMC 6h
ELISPOT plate
Whatman small well
Whatman-polyfiltronics
Pall
Nunc Maxisorb
Nunc Maxisorb
ELISPOT plate type Coating antibody
PVDF membrane IFN- Endogen; IL-4/IL-10 Pharmingen O/N@4 (C PBS/1% BSA 1 h@RT
PVDF membrane IFN- Endogen; IL-5 Pharmingen O/N@4 (C PBS/1% BSA 1 h@RT
PVDF membrane Mabtech
Clear U-Cytech
Clear U-Cytech
O/N@4 (C PBS/1%BSA O/N@4 (C
O/N@4 (C PBS/1%BSA O/N@4 (C
IFN- Endogen; IL-5 Pharmingen O/N@4 (C Streptavidin-HRP
U-Cytech
U-Cytech
Incubation time Tertiary reagent
IFN-/IL-4 Endogen; IL-10 Pharmingen O/N@4 (C Streptavidin-HRP
O/N@4 (C IMDM complete 1 h@37 (C Mabtech 2–4 h@37 (C Streptavidin-AP
Incubation time Development reagent
1 h@RT AEC
1 h@RT AEC
1 h@37 (C Goat anti biotin antibody (U-Cytech) 1 h@37(C U-Cytech
1 h@37 (C Goat anti biotin antibody (U-Cytech) 1 h@37(C U-Cytech
Development time Preferred reader
Visually determined CTL
Visually determined CTL
Visually determined BIOSYS
Visually determined BIOSYS
Coating conditions Block reagent Detector antibody
1 h@RT AP Substrate Kit (Bio-Rad) 15 min AID
Abbreviations used: n/a, not applicable; PBS, phosphate buffered saline; BSA, bovine serum albumin; HRP, horseradish peroxidase, AP, alkaline phosphatase; RT, room temperature; O/N overnight; AEC, 3-amino-9-ethylcarbazole.
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
ELISPOT conditions
CTL-1 Pre-incubation step used
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
369
Table 3 Results of CK ELISpot assays performed on control blood sample Culture conditions
Medium DMSO (0.05) DMSO (0.01) TT (100) TT (10) GAD65 (5) GAD65 (1) IA-2 (5) IA-2 (1) HSP60 (5) HSP60 (1)
U-Cytech-1
U-Cytech-2
Mabtech
CTL-1
CTL-2
IFN-
IL-5
IL-10
IL-13
IFN- IL-4
IL-13
IFN- IL-4
IFN- IL-4
IL-10
IFN- IL-5
0.3 1.3 1.7 7.0** 2.8* 0.5 1.8 2.2 1.7 4* 1.3
0.33 0.33 0.67 1.17 1.5 1.0 0.7 1.3 0.2 0.5 1.2
25 12 33 15** 9** 22 35 18 20* 12 17
2.5 1.7 1.3 0.3 2.4* 0.3 1.5 0.3 0.2 0.3 0.3
0.5 0.3 0.3 2.5* 1.7* 3.7* 0.0 0.7 0.3 2.2* 1.7*
0.1 0.3 0.0 0.5 0.3 0.0 0.7 0.0 0.0 0.2 0.0
29 39 39 97** 75* 34 104 40 27* 35 35
4.8 3.6 5.8 4.0 2.8 6.4 6.4 6.2 6.4 5.6 4.0
2.0 2.0 2.5 3.0 4.3 2.0 1.3 3.3 2.3 3.0 2.0
3.8 2.3 4.1 5.5 11.5 2.7 2.5 5.2 3.2 1.8 3.2
1.2 1.7 3.7 4.2** 3.0* 2.3 2.3 2.0 3.7 4.0* 2.3
1.5 2.0 2.0 5.3** 1.5 2.0 0.7 1.3 0.7 3.5 1.0
2.2 2.0 2.8 1.5 2.0 1.7 1.2* 2.5 1.2 1.8 1.7
0.7 0.7 0.5 2.8** 2.0 1.5 0.5 0.7 0.7 0.7 0.5
Abbreviations used: DMSO, dimethyl sulphoxide (%); TT, tetanus toxoid, (ng/ml); GAD65, GAD65 (554–575) (µM); IA-2, IA-2 (853–872) (µM); HSP60, HSP60 (277–300) (µM). Results are expressed as mean spot numbers/well. Levels of significance are indicated: ** P<0.01. * P<0.05. Table 4 Results of CK ELISpot assays performed on T1D blood sample Culture conditions
Medium DMSO (0.05) DMSO (0.01) TT (100) TT (10) GAD65 (5) GAD65 (1) IA-2 (5) IA-2 (1) HSP60 (5) HSP60 (1)
U-Cytech-2
Mabtech
CTL-1
CTL-2
IFN-
IL-4
IL-10
IL-13
IFN-
IL-4
IFN-
IL-4
IL-10
IFN-
IL-5
0.2 0.8 0.5 30** 11** 1.3 2.0 1.5 0.8 1.0 1.5
8.3 6.8 7.3 19** 10 5.5 6.2 7.0 7.0 6.8 6.3
81 31 18 32** 12** 55** 33** 25 22 20 21
0.2 0.3 0.0 3.5** 0.5 0.0 0.0 0.3 0.3 0.0 0.7
11 27 21 43** 27* 24 28 23 19 24 25
0.8 3.3 2.0 3.8** 1.2 1.7 0.7 1.0 1.3 1.5 2.3
1.2 1.4 1.2 5.2** 3.2 0.6 1.6 3.8* 2.4* 1.6 0.8
1.5 4.3 5.0 5.3** 3.8* 6.0 4.3 5.5 5.7 3.0 4.7
2.0 2.0 2.5 3.0 4.3 2.0 1.3 3.3 2.3 3.0 2.0
6.5 3.5 2.7 4.7 3.0 5.0 0.7 6.0 1.2 6.8 2.0
0.5 1.3 1.7 2.8 2.0 1.5 0.5 0.7 0.7 0.7 0.5
Abbreviations used: DMSO, dimethyl sulphoxide (%); TT, tetanus toxoid (ng/ml); GAD65, GAD65 (554–575) (µM); IA-2, IA-2 (853–872) (µM); HSP60, HSP60 (277–300) (µM). Results are expressed as mean spot numbers/well. Levels of significance are indicated: ** P<0.01. * P<0.05.
detected to both concentrations of TT by both U-CyTech-1 and U-CyTech-2 assays and the Mabtech assay for both the T1D and control sample (where tested). The CTL-1 assay detected an IFN- response in the T1D sample only, and only at the higher concentration of TT. No responses to TT were detected by the CTL-2 assay. The frequency of responses detected varied considerably between assay formats, with the Mabtech assay consistently giving a higher number of TT specific spots/well than the U-CyTech or CTL assays. Tetanus toxoid specific IL-4 responses were also detected in both samples by the U-CyTech-2 (analysis not performed by U-CyTech-1) and Mabtech assays and by CTL-1 in the T1D sample (assay not performed by CTL-2). The frequency of the responses was similar in all assay formats. A significant reduction in the number
of IL-10 producing cells was detected at both concentrations of TT using the U-CyTech assay format in the T1D sample (assay only performed by U-CyTech-2) and control sample (assay only performed by U-CyTech-1). This reduction in CK secreting cells was not seen for either sample in the only other assay for IL-10 (CTL-1). Of the assays performed to detect TT-specific IL-5 secretion (CTL-2 and U-CyTech-1) only CTL-2 detected a significant increase in spot number. Only U-CyTech assays analysed IL-13 responses, which were detectable in both subjects. 3.3. Responses to autoantigen peptides Two weeks prior to the workshop, samples from the T1D and control subjects were tested against a panel of
370
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 1. IFN- and IL-4 responses to tetanus toxoid. Responses from T1D are shown in the upper three panels and responses from control individual (Ctrl) are shown in the lower three panels. Filled squares=CTL-1, open squares=CTL-2, open circles=U-CyTech-1, closed circles=U-CyTech-2 and closed triangles=Mabtech. Asterisks indicate the degree of significance (*P<0.05, **P< 0.01) comparing antigen-stimulated response versus spot formation in presence of medium.
peptides based on sequences from the diabetesassociated autoantigens IA-2, GAD65, proinsulin and HSP60. In this pre-testing, the control subject showed increased numbers of IFN- secreting cells in response to HSP60 (277–300), and the T1D subject in response to GAD65 (554–575) and IA-2 (853–872). In all cases, these responses represented a low frequency of autoreactive cells close to the limit of detection of the assay used (data not shown). In the workshop, a significant IFN- response to HSP60 (277–300) was seen in the control individual by both the U-CyTech-1 and U-CyTech-2 assays (Fig. 2) but not detected by other assays. In addition, an IL-4 specific response was detected to the same peptide in the U-CyTech assay (assay performed by U-CyTech-2 only)
but not by Mabtech or CTL-1 (assay not performed by CTL-2). As in pre-testing, the responses detected were present at low levels corresponding to <5 spots well. Reproducibility of these control subject responses for the U-CyTech-1 assays performed in pre-testing and during the workshop and the U-CyTech-2 assay performed during the workshop is shown in Fig. 2B. In the T1D sample, a significant IFN- response to IA-2 (853–872) was observed at both concentrations in one assay (CTL-1). Although no assays detected an IFN- response to GAD65 (554–575) in the T1D sample, an IL-10 specific response to this peptide was detected by one assay (U-CyTech-2) at both concentrations. As in the pre-testing, no response to HSP60 (277–300) was observed in the T1D sample.
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
371
Fig. 2. IFN- response to HSP60 (277–300) peptide by control subject and assay reproducibility. (A) Filled squares=CTL-1, open squares=CTL-2, open circles=U-CyTech-1, closed circles=U-CyTech-2 and closed triangles=Mabtech. Asterisks indicate the degree of significance (*P<0.05) comparing antigen-stimulated response versus spot formation in presence of medium control. (B) Results of U-CyTech IFN- ELISPOT analysis of response of control subject to HSP60 (277–300) performed at two different time-points and by two different operators. Open circles=U-CyTech-1, closed circles=U-CyTech-2, for assays performed during workshop; closed diamonds=U-CyTech-1, for assay performed during pre-testing 2 weeks before workshop.
3.4. Combinatorial analysis of cytokine responses A major advantage of CK ELISpot analysis over techniques such as proliferation and limiting dilution analyses is that they provide information on the class of the effector response, which is a reflection of the balance of cytokines secreted in response to a specific antigen (e.g. Th1, Th2 or Th0 profiles). In addition to assessing differences in assay performance, the workshop also aimed to identify novel ways of presenting the data generated by CK ELISpot analysis. We therefore explored the utility of combining responses defined by different cytokines to a given antigen in stacked bars (Fig. 3). This analysis provides an overall representation of the immune response to a specific antigenic stimulus, showing both increases and decreases in cytokine production (e.g. TT response comprises increased secretion of IFN-, IL-4 and IL-13 but a decrease in IL-10; Fig. 3) and should enable better discrimination of an antigen specific response. 3.5. Concordance between assays The degree of concordance of responses (including results obtained with the positive control preparations PHA, and PMA/ionomycin) between all assay formats was assessed using contingency tables (Table 5). Although this comparison is complex, due to the different numbers of assays performed in common by different participants, several conclusions can be drawn. Concordance is stronger between assays of the same type than between assay types. For example, concord-
ance between U-CyTech-1 and -2 assays is more significant than that between either of these and either of the CTL assays. Similarly, the concordance between CTL-1 and -2 assays is more significant than that between either of these and either of the U-CyTech assays. In addition, there is good concordance between the Mabtech assay and assays of other formats. The concordance for IFN- assays is also represented graphically using a Venn diagram approach (Fig. 4). These analyses suggest that there is greater similarity between the Mabtech and U-CyTech assays than between either of these formats and the CTL assays. 3.6. Comparison of CK ELISpot readers A key factor in the analysis of CK ELISpot assays is the objective enumeration of spots, for which several automated readers are available commercially. To compare results obtained with different readers, plates generated during the workshop were analysed by Autoimmun Diagnostika (AID), Cellular Technologies Ltd (CTL) and BIOSYS readers. All three readers were able to read PVDF membrane plates from the Mabtech and CTL assays. Transparent ELISA plates, used in the U-CyTech assay, were only readable by the AID and BIOSYS machines. Results comparing the reading of the CTL-1 and -2 and Mabtech IFN- assays are shown in Fig. 5. Although the numbers of spots read varied, a highly significant correlation between the spot numbers obtained by all readers for each well was observed (P<0.0001).
372
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 3. Cytokine profile of response to tetanus toxoid (TT) (100 ng/ml) by T1D subject. Results are shown as accumulated mean spot numbers/well. Solid bar=IFN-, vertical stripes=IL-4, diagonal stripes=IL-5, un-shaded=IL-10 and hatched=IL-13. Table 5 Concordance between different assay formats
U-CyTech-2 Mabtech CTL-1 CTL-2
U-CyTech-1
U-CyTech-2
Mabtech
CTL-1
P=0.0045 (n=20) n.s. (n=10) P=0.033 (n=20) P=0.031 (n=20)
P<0.0001 (n=40) P=0.021 (n=40) n.s. (n=20)
P=0.0005 (n=40) P=0.026 (n=20)
P=0.014 (n=20)
Values are calculated by comparison of significant results using contingency tables. Number of results used for construction of contingency tables is shown in brackets.
4. Discussion The identification of assays capable of distinguishing islet antigen specific T cell responses is an important research goal, and has been facilitated to date by the performance of a number of workshop studies. The report on the first IDS T Cell Workshop highlighted a number of major limitations with both the antigen preparations and assay methodologies employed [26]. The second Workshop evaluated a large number of antigen preparations and subsequently, a panel of islet antigens and islet antigen derived peptides were made available for assay evaluation [24]. An assay format that has received particular recent interest is the CK ELISpot. In a recent review [29] the relative merits of different techniques for the detection of islet autoreactive T cells was discussed, with particular emphasis on CK ELISpot, and therefore a detailed discussion of the advantages and limitations of the various assay formats will not be entered into here. As a direct means of comparison of different CK ELISpot assay formats a wet workshop was organized,
in which all participants used the same fresh blood samples and identical blinded antigen preparations. A number of factors, including need to pre-test to assess responsiveness, a preference for using fresh rather than cryopreserved samples, size of blood sample required and the timing constraints of the workshop precluded the use of a blood sample from a newly diagnosed T1D patient. As an alternative, samples were obtained from two of the workshop participants, demonstrated in pre-testing to have responses to selected test antigens. It should be noted that in the design of this workshop, the pre-testing of subjects was performed only using one assay format (U-CyTech) and this may have favoured these assays in the ability to reproduce these results. The responses to islet antigen peptides by autoreactive T cells detected in pre-testing demonstrated very low frequencies of responder cells (in the range of 1:100,000) and therefore provided a great challenge to the sensitivity of the assays. Low dose tetanus toxoid specific responses were, therefore, also included to provide a robust means of assessing sensitivity. Assays may vary in
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
373
Fig. 4. Relationship between results of IFN- CK ELISpot from different assay formats. Comparison of IFN- CK ELISpot results from (A) T1D and (B) control individuals. High concentrations (TT 100 ng/ml peptides 5 µM) are shown in bold and low concentrations (TT 10 ng/ml peptides 1 µM) in normal script. Abbreviations used: TT, tetanus toxoid; GAD65, GAD65 (554–575); IA-2, IA-2 (853–872); HSP60, HSP60 (277–300).
the concentration of antigens required to give optimal performance [21]. Whilst there are clearly differences in assay performance, the results presented demonstrate that low level recall antigen- and peptide-specific autoreactive T cell responses are detectable by CK ELISpot and may be reproduced over time within the same assay format. Good levels of concordance were observed both between assays of a similar format and those of different formats. As expected the level of concordance was generally greatest between similar assays. All assays were able to detect significant cytokine secretion in response to the polyclonal stimulators PHA and PMA/ionomycin. These stimuli are sufficiently powerful that in some cases accurate spot numbers were difficult to assign due to the high number of spots and background signal resulting from secreted cytokine in the media. Therefore, for accurate studies of the total numbers of cytokine producing cells in response to these stimuli it may be necessary to titrate the number of cells used per well. At the higher tetanus concentration four of the five assays were able to detect significant IFN- specific T cell responses although the level of significance and total spot numbers varied considerably. At the lower tetanus toxoid concentration only the Mabtech and U-CyTech assays detected IFN- reactivity. Reproducible autoreactive T cell responses (i.e. concordant with pretesting) were detected in the control individual by the U-CyTech assays (IFN-) at both peptide concentrations and also by IL-4 secretion. Reproducible peptide responses in the T1D sample were detected to the IA-2 peptide by CTL-1 (IFN-). Although no IFN- response was observed to the GAD65 peptide, this stimulus induced IL-10 secretion measurable by the U-CyTech-2 assay, indicating the detection of a response. These
results suggest that different assays have different sensitivities, that may vary according to the type or strength of stimulus used. At low levels of recall antigen the Mabtech and U-CyTech assays displayed a higher level of sensitivity than the CTL assays. However, for the detection of extremely low-level peptide specific autoreactivity the U-CyTech and CTL assays were superior. A major factor in the ability of CK ELISpot assays to distinguish low frequency antigen specific T cell responses is the nature of the signal-to-noise ratio. Clearly, variables capable of influencing this ratio arise at all stages of the assay with a consequent influence on spot detection. These include: composition of cell preparation, number and density of cells incubated; the total time and window of cytokine capture and the reagents used for capture and detection. With such a high number of variables it is difficult for a single workshop to identify their relative contributions to differences in assay performance. However, some general conclusions may be drawn. A wide range in the number of spots observed in unstimulated wells (noise) was observed, and was associated with assay type as well as being consistent for both samples. In the case of IFN- assays, the unstimulated spot number correlates well with the time of incubation of cells on the CK ELISpot plate: for example, spot number for Mabtech (48 h incubation)>CTL (24 h)>U-CyTech (5 h). Similarly, the long capture time of the Mabtech assay may have contributed to the large number of spots in response to tetanus toxoid (signal). Cytokine capture time may thus explain the wide differences in spontaneous secretion reported by different groups for T1D patients, at risk and control subjects [21,22]. Different times of incubation may be also be required for optimal detection of responses to whole protein antigens, which require
374
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
Fig. 5. Comparison of spot numbers obtained by three different readers. (A) plot of values obtained individual readers against mean values for the three readers. Values from the CTL reader are shown as square symbols, AID by triangles and Biosys by circles. (B–D) comparison of values obtained by individual readers. Linear regression analysis was performed and goodness of fit is expressed as r2. For all comparisons P<0.0001 (Spearman’s rank correlation).
intracellular processing, compared to responses to peptides which do not. This may have contributed to the greater sensitivity of Mabtech and U-CyTech assays in detecting tetanus responses. Clearly a good balance of signal-to-noise is required for optimal assay sensitivity and this may vary depending upon the type of response measured. Whereas a large signal may make detection of tetanus responses more impressive in terms of the total number of spots detected, the detection of autoreactive responses, present at very low frequencies, may be facilitated in assays with a low spontaneous response. In the detection of very low frequency responses, the signal may be lost in the ‘noise’ of assays with a high spontaneous response. In addition, the total number of lymphocytes in an assay will clearly have an effect on the power of the analysis to detect a significant response, with large cell numbers required to detect very low frequency responses. As statistical analysis of multiple wells was used to determine positivity in the present
study, it is clear that inter-well variation will be an important factor in assay sensitivity. In addition to time of cytokine capture, a major difference in assay format is the inclusion of a preincubation step in the absence of CK capture antibodies in the U-CyTech assays. This step is included to allow incubation of cells at a high density for a prolonged period without the generation of a high background and is one of several modifications introduced with the specific aim of detecting low frequency cells producing multiple cytokines [28,30] (P van der Meide and B.O. Roep, personal communication). This step may also allow cross-regulatory effects of different cytokines to occur, as they are not captured and takes advantage of the fact that tissue culture plates are coated to optimize the quality of cellular immunological responses. Moreover, the use of an initial bulk-culture step may result in reduced inter-well variability in the final CK ELISpot plate. Differences between assays may also be
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
attributable to plate selection as previously demonstrated [31]. In this workshop PVDF and ELISA plates were used, although the study design does not allow direct comparison of the effects of these two plate types. Ultimately, each of the assay types evaluated in this WS can be further improved to allow detection of low-frequency autoreactive T-cells producing various cytokine profiles. A consensus on the definition of a positive CK ELISpot response remains to be established. Methods used include comparison of absolute spot numbers, with or without subtraction of background, the calculation of stimulation indices or the comparison of different cytokines leading to the calculation of Th1/Th2 ratios. In the present study, the subtraction of background counts was not favoured as it masks the extent of the background response (for example, results for medium versus stimulus of 0 and 5 spots, respectively, clearly represents a different result from 100 and 105 spots) and also precludes statistical analysis for positivity. Likewise, although the generation of stimulation indices does take into account both the signal and noise in an assay, it is less amenable to statistical analysis. In this workshop we chose the comparison of total spot numbers from multiple wells by the Mann–Whitney U test. We suggest that where possible, CK ELISpot studies should report both the magnitude, as mean or median, and statistical significance of responses to allow a full interpretation of the results. However, care will be required in the choice of test for statistical comparison and in the interpretation of the results, particularly when multiple wells are used and inter-well variation is small. In summary, this CK ELISpot workshop has demonstrated the detection of low-level autoreactive T cell responses. Differences between CK ELISpot formats in terms of sensitivity will require further investigation and this will lead to the refinement of these assays as tools for unravelling the complex mechanisms of cell destruction that lead to T1D.
[4]
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
Acknowledgements [17]
We are thankful to the financial support for this workshop by U-Cytech, Utrecht, The Netherlands; CTL, Europe; Biosys, Germany; AID, Germany and Mabtech Sweden.
[18]
[19]
References [1] Castano L, Eisenbarth GS. Type-I diabetes: a chronic autoimmune disease of human, mouse, and rat. Annu Rev Immunol 1990;8:647–79. [2] Gepts W. Pathologic anatomy of the pancreas in juvenile diabetes mellitus. Diabetes 1965;14:619–33. [3] Bougneres PF, Carel JC, Castano L, Boitard C, Gardin JP, Landais P et al. Factors associated with early remission of type I
[20]
[21]
375
diabetes in children treated with cyclosporine. N Engl J Med 1988;318:663–70. Lampeter EF, Homberg M, Quabeck K, Schaefer UW, Wernet P, Bertrams J et al. Transfer of insulin-dependent diabetes between HLA-identical siblings by bone marrow transplantation. Lancet 1993;341:1243–4. Sibley RK, Sutherland DE, Goetz F, Michael AF. Recurrent diabetes mellitus in the pancreas iso- and allograft. A light and electron microscopic and immunohistochemical analysis of four cases. Lab Invest 1985;53:132–44. Martin S, Wolf-Eichbaum D, Duinkerken G, Scherbaum WA, Kolb H, Noordzij JG et al. Development of type 1 diabetes despite severe hereditary B-lymphocyte deficiency. N Engl J Med 2001;345:1036–40. Peakman M, Leslie RD, Vergani D. Immunological studies on type 1 diabetes in identical twins. Arch Dis Child 1993;69:97–9. Smerdon RA, Peakman M, Hussain MJ, Alviggi L, Watkins PJ, Leslie RD et al. Increase in simultaneous coexpression of naive and memory lymphocyte markers at diagnosis of IDDM. Diabetes 1993;42:127–33. Buschard K, Damsbo P, Ropke C. Activated CD4+ and CD8+ T-lymphocytes in newly diagnosed type 1 diabetes: a prospective study. Diabet Med 1990;7:132–6. Petersen LD, Duinkerken G, Bruining GJ, van Lier RA, de Vries RR, Roep BO. Increased numbers of in vivo activated T cells in patients with recent onset insulin-dependent diabetes mellitus. J Autoimmun 1996;9:731–7. Gessl A, Waldhausl W. Increased CD69 and human leukocyte antigen-DR expression on T lymphocytes in insulin-dependent diabetes mellitus of long standing. J Clin Endocrinol Metab 1998; 83:2204–9. Szelachowska M, Kretowski A, Kinalska I. Decreased in vitro IL-4 [corrected] and IL-10 production by peripheral blood in first degree relatives at high risk of diabetes type-I. Horm Metab Res 1998;30:526–30. Halminen M, Juhela S, Vaarala O, Simell O, Ilonen J. Induction of interferon-gamma and IL-4 production by mitogen and specific antigens in peripheral blood lymphocytes of Type 1 diabetes patients. Autoimmunity 2001;34:1–8. Rapoport MJ, Mor A, Vardi P, Ramot Y, Winker R, Hindi A et al. Decreased secretion of Th2 cytokines precedes Up-regulated and delayed secretion of Th1 cytokines in activated peripheral blood mononuclear cells from patients with insulin-dependent diabetes mellitus. J Autoimmun 1998;11:635–42. Roep BO. Autoreactive T cells in endocrine/organ-specific autoimmunity: why has progress been so slow? Springer Semin Immunopathol 2002;24:261–71. Viglietta V, Kent SC, Orban T, Hafler DA. GAD65-reactive T cells are activated in patients with autoimmune type 1a diabetes. J Clin Invest 2002;109:895–903. Reijonen H, Novak EJ, Kochik S, Heninger A, Liu AW, Kwok WW et al. Detection of GAD65-specific T-cells by major histocompatibility complex class II tetramers in type 1 diabetic patients and at-risk subjects. Diabetes 2002;51:1375–82. Klinman DM. ELISPOT assay to detect cytokine-secreting murine and human cells. In: Coico R, editor. Current protocols in immunology, New York: John Wiley & Sons; 1994, p. 6, 19, 11. Karulin AY, Hesse MD, Tary-Lehmann M, Lehmann PV. Singlecytokine-producing CD4 memory cells predominate in type 1 and type 2 immunity. J Immunol 2000;164:1862–72. Kotani R, Nagata M, Moriyama H, Nakayama M, Yamada K, Chowdhury SA et al. Detection of GAD65-reactive T-cells in type 1 diabetes by immunoglobulin-free ELISPOT assays. Diabetes Care 2002;25:1390–7. Karlsson MG, Lawesson SS, Ludvigsson J. Th1-like dominance in high-risk first-degree relatives of type I diabetic patients. Diabetologia 2000;43:742–9.
376
N.C. Schloot et al. / Journal of Autoimmunity 21 (2003) 365–376
[22] Alleva DG, Crowe PD, Jin L, Kwok WW, Ling N, Gottschalk M et al. A disease-associated cellular immune response in type 1 diabetics to an immunodominant epitope of insulin. J Clin Invest 2001;107:173–80. [23] Raz I, Elias D, Avron A, Tamir M, Metzger M, Cohen IR. Beta-cell function in new-onset type 1 diabetes and immunomodulation with a heat-shock protein peptide (DiaPep277): a randomised, double-blind, phase II trial. Lancet 2001; 358:1749–53. [24] Peakman M, Tree TI, Endl J, van Endert P, Atkinson MA, Roep BO. Characterization of preparations of GAD65, proinsulin, and the islet tyrosine phosphatase IA-2 for use in detection of autoreactive T-cells in type 1 diabetes: report of phase II of the Second International Immunology of Diabetes Society Workshop for Standardization of T-cell assays in type 1 diabetes. Diabetes 2001;50:1749–54. [25] Roep B, Atkinson M, Honeyman M, Peakman M. T Cell Workshop update 2002. [26] Roep BO, Atkinson MA, van Endert PM, Gottlieb PA, Wilson SB, Sachs JA. Autoreactive T cell responses in insulin-dependent (Type 1) diabetes mellitus. Report of the first international
[27]
[28]
[29]
[30]
[31]
workshop for standardization of T cell assays. J Autoimmun 1999;13:267–82. Hesse MD, Karulin AY, Boehm BO, Lehmann PV, Tary-Lehmann M. A T cell clone’s avidity is a function of its activation state. J Immunol 2001;167:1353–61. van Halteren AG, van Etten E, de Jong EC, Bouillon R, Roep BO, Mathieu C. Redirection of human autoreactive T-cells upon interaction with dendritic cells modulated by TX527, an analog of 1,25 dihydroxyvitamin D(3). Diabetes 2002;51:2119–25. Meierhoff G, Ott PA, Lehmann PV, Schloot NC. Cytokine detection by ELISPOT: relevance for immunological studies in type 1 diabetes. Diabetes Metab Res Rev 2002;18:367–80. Schmittel A, Keilholz U, Bauer S, Kuhne U, Stevanovic S, Thiel E et al. Application of the IFN-gamma ELISPOT assay to quantify T cell responses against proteins. J Immunol Methods 2001;247:17–24. Ronnelid J, Klareskog L. A comparison between ELISPOT methods for the detection of cytokine producing cells: greater sensitivity and specificity using ELISA plates as compared to nitrocellulose membranes. J Immunol Methods 1997; 200:17–26.