Abstract
The evaluation of antigen-specific T-cell responses is helpful for both research and clinical settings. Several techniques can enumerate antigen-responsive T cells or measure their products, but they require remarkable amounts of peripheral blood mononuclear cells (PBMCs). Since screening numerous antigens or testing samples from pediatric or lymphopenic patients is hampered in clinical practice, we refined a miniaturized, high-throughput assay for T-cell immunity. Antigens and cells in 10-μl volumes were dispensed into 1,536-well culture plates precoated with anti-gamma interferon (anti-IFN-γ) antibodies. After being cultured, the wells were developed by enzyme-linked immunosorbent assay for bound cytokine. Miniaturization and automation allowed quantitation of antigen-specific responses on 104 PBMCs. This method was applied for epitope mapping of mycobacterial antigens and was used in the clinic to evaluate T-cell immunity to relevant opportunistic pathogens by using small blood samples. A comparison with conventional methods showed similar sensitivity. Therefore, current flow cytometric methods that provide information on frequency and phenotype of specific T cells can be complemented by this assay that provides extensive information on cytokine concentrations and profiles and requires 20- to 50-fold fewer PBMCs than other analytical methods.
Numerous methods are currently available to enumerate antigen-specific T cells and to measure their functions, most of them being based on flow cytometric analysis. Such methods have been compared and critically discussed in recent reviews (4, 17). A common feature of these methods is the requirement, on average, of 2 × 105 to 5 × 105 peripheral blood mononuclear cells (PBMCs) to test a single antigen. This imposes a limitation on the analytical antigenic breadth, which depends on the number of available PBMCs. In the case of lymphopenic subjects, the population that most frequently needs to be monitored for immunocompetence, this is particularly crucial. Also, the small volumes of pediatric blood samples may result in inadequate numbers of PBMCs for extended analysis of specific T-cell immunity. By the same token, screening of large peptide panels for epitope mapping on PBMCs from healthy donors (blood donors or vaccinees) is also limited to a few hundred peptides (16) instead of the thousands of peptides needed to cover several immunodominant proteins (14). Therefore, we felt the need to develop a miniaturized assay that makes use of limited numbers of PBMCs, thereby enhancing our screening power.
A miniaturized assay based on 384-well plates instead of conventional 96-well plates or 5-ml tubes was developed in our laboratory and named cell-enzyme-linked immunosorbent assay (cell-ELISA) (8). This method was based on a previous observation that lymphocyte cultures can be run in 96-well plates in which the wells had been precoated with an anticytokine antibody. The cytokine secreted by antigen-specific T cells was captured by the solid-phase antibody, and the plate was developed as a conventional ELISA plate (11). Miniaturization from 96- to 384-well plates (8) resulted in a reduction of tested lymphocytes from 2 × 105 to 5 × 105 down to 5 × 104 per single antigen, corresponding to a scaling-down factor of 4- to 10-fold.
In this report, we describe further miniaturization in 1,536-well plates by seeding 104 PBMCs per well, meaning a reduction in cell number by a factor of 20- to 50-fold over conventional assays. We also describe here the use of pathogen-specific, established CD4 and CD8 T-cell lines as possible standards for T-cell assays, the comparison of cell-ELISA with other established methods with respect to sensitivity, and the preliminary application of cell-ELISA to clinical samples and to peptide screening for epitope mapping in a miniaturized format.
MATERIALS AND METHODS
The cell-ELISA method, based on an assay originally designed for mouse splenocytes in 96-well plates (12), has been described in detail for 384-well plates (8). The adoption of 1,536-well plates required extensive automation of all procedures for antigen dispensing (proteins or peptides), for liquid and cell handling, and for plate development and scanning. The assay was performed with sterile reagents under a laminar flow cabinet.
Automated instrumentation.
The four-channel MultiProbe (Perkin-Elmer, Shelton, CT) was used to transfer proteins and peptides from vials to 96-well master plates. The 96-channel Hydra II (Matrix) was used for simultaneous dispensing of antigens from 96-well master plates to 384-well and 1,536-well plates. The eight-channel MultiWell (Matrix Technologies, Hudson, NH) was used for dispensing reagents and cells in 384-well plates (catalogue no. 164688; NUNC, Roskilde, Denmark). The eight-channel MultiDrop (Thermo OY, Finland) was used for dispensing reagents and cells in 1,536-well plates (NUNC catalogue no. 1536.13). The ELx800 (Biotek, Winooski, VT) was used for 384-well plate scanning, and the Victor 3V (Perkin-Elmer) was used for 1,536-well plate scanning.
Reagents.
Phosphate-buffered saline (PBS) and RPMI 1640 medium were purchased from BioWhittaker, Verviers, Belgium. RPMI 1640 medium was enriched with 10 mM l-glutamine and with 5% fetal calf serum selected for low background in the cell-ELISA assay (complete medium). Antibody pairs were from Mabtech (Stockholm, Sweden). Antibodies for phenotyping and for intracytoplasmic cytokine staining (ICS) were from Becton Dickinson (San José, CA). Pentamers were from Proimmune, Oxford, United Kingdom. Protein antigens, derived from different opportunistic pathogens or whole inactivated pathogen bodies, have been described previously (6, 7, 8, 9). Pathogens used as whole bodies or pathogens from which proteins were derived included the following: Candida albicans, Aspergillus fumigatus, Aspergillus niger, Cryptococcus neoformans, Toxoplasma gondii, Mycobacterium bovis (“Mycobacterium tuberculosis var. bovis” Bergey et al. 1934 pro synon.), Mycobacterium tuberculosis (“Mycobacterium tuberculosis var. hominis” Bergey et al. 1934 pro synon.), cytomegalovirus (CMV), and human immunodeficiency virus type 1 (HIV-1). Peptides were synthesized by INBIOS, Naples, Italy, or by JPT, Berlin, Germany (6, 7, 8, 9). Final antigen concentrations were 10 μg/ml for proteins and 1 μg/ml for peptides.
Blood samples.
Blood was obtained from healthy donors who gave informed consent. Blood from patients was obtained after approval of the Institutional Ethical Committee and after the informed consent was signed.
Plate coating.
Plates (1,536 wells) were coated with 5 μl of the first anti-gamma interferon (anti-IFN-γ) antibody at 1 μg/ml in PBS. After overnight incubation at 4°C, the plates were washed twice with PBS and saturated with 10 μl PBS-1% human serum albumin (20% solution; Kedrion, Lucca, Italy) at room temperature (RT) for 1 h. After saturation, the plates were washed twice with 13 μl Hanks’ balanced salt solution and received 4 μl of complete medium.
Antigen dispensing.
Antigens, dispensed with predefined patterns in 96-well master plates, were transferred into 1,536-well plates using the Hydra II liquid handler. One microliter of 10× antigen solution was dispensed into the wells that already contained 4 μl of complete medium. The plates were sealed with a plastic membrane and frozen at −80°C, ready for use.
Cell-ELISA.
PBMCs obtained from heparinized peripheral blood by the conventional Ficoll gradient were brought to 2 × 106 cells/ml in complete medium and automatically dispensed into thawed plates at 5 μl per well. The plates were incubated for 2 days in a 5% CO2 atmosphere. At the end of the culture, supernatants or cells could be collected for further experiments. After washing twice with PBS-Tween 20 0.05% (Tween 0.1% when PBMCs were derived from HIV-positive donors), the plates were spun upside down on filter paper at 500 × g for 2 min for complete removal of washing buffer. Then, the wells received 5 μl biotinylated antibody at 1 μg/ml for 1 h. After washing twice with PBS-Tween and three times with PBS and drying by centrifugation, the wells received 5 μl alkaline phosphatase-streptavidin at 1 μg/ml. After being incubated at RT, the plates were washed as described in the previous step and the wells received 10 μl p-nitrophenylphosphate. After a 1-h incubation at RT, the plates were scanned. Results are shown as the optical density at 405 nm × 1,000 or as picograms/milliliter of cytokine, referred to as a titration curve.
Methods for specific T cells.
ICS, enzyme-linked immunospot assay (ELISPOT), and pentamer staining were performed according to standard procedures, following the manufacturer's instructions (Becton Dickinson, Mabtech, and ProImmune). Flow cytometric analysis was performed with a CyAn instrument (Dako) using Spectraline and Flow-Check FluoroSpheres (Dako) for instrument setting and compensation and with standard gating practices. Staining with carboxyfluorescein succinimidyl ester (CFSE) was performed as described previously (17). Lymphoproliferation was performed as described previously (7, 9). CD4 and CD8 T-cell lines were produced and maintained as described previously (7, 9).
Cytokines were tested in microsamples from PBMC supernatants or T-cell line supernatants by using the Luminex assay (M. Templin, NMI Natural and Medical Science Institute, Reutlingen, Germany) or by using the Cytomix Assay (11-plex kit, catalogue no. BMS810FF; Bender, Vienna, Austria).
In order to appreciate the quality or power of a high-throughput screening (HTS) assay, the Z factor is currently used (20). In an HTS approach, large numbers of single measurements of unknown samples are compared to well-established positive or negative control samples. A Z factor score estimates the significance of the difference between positive and negative values. To determine Z factor scores (20), we performed assays in 1,536-well plates with a CD4 T-cell line specific for HIV gp120 peptide PAGFAILKCNN generated from donor 2 (10) diluted into autologous, nonresponding, irradiated PBMCs and with a CD8 T-cell line specific for CMV peptide NV9 generated from donor 1 (9) diluted into autologous, nonresponding, irradiated PBMCs. Each well in the plates received 104 PBMCs plus the antigenic peptide. Wells in plate A received five T cells from the CD4 or CD8 T-cell lines, while wells in plate B received no T cells. Thus, responses above PBMCs alone could be assigned to the specific T-cell line diluted 1:2,000 (0.05%) into PBMCs. The plates were incubated, developed, and scanned. The mean value of a plate containing the cell-ELISA reagents without any cells was subtracted from the plate readings. The Z factor was calculated as indicated in reference 20.
Generation of antigen-specific T-cell lines. Antigen-specific CD4 T-cell lines were generated by pulsing PBMCs at 107 cells/ml with the relevant antigens for 4 h at 37°C. After a 1:10 dilution with medium, PBMCs were dispensed in 24-well plates at 2 × 106 cells/ml (2 ml per well), as described previously (7). Interleukin 2 (IL-2) was added 4 days later. The cells were split 1:4 when concentration exceeded 106 cells/ml. Restimulation cycles were performed every 3 weeks on 3 × 105 T cells cultured with 106 autologous 30 Gy of irradiated PBMCs plus antigen. A similar protocol was used to generate a CD4 T-cell line specific for an immunodominant epitope of HIV gp120 from a nonimmune donor (10). CMV-specific CD8 T-cell lines responding to the immunodominant pp65 peptide NV9 were produced as described previously (9). Briefly, PBMCs at 107 cells/ml were pulsed with NV9 peptide for 2 h at 37°C. Pulsed PBMCs (106) and untreated PBMCs (106) were cocultured in 2 ml of medium in a 24-well plate. One week later, the cells were restimulated with 106 peptide-pulsed, 30 Gy-irradiated PBMCs, and IL-2 was added after 1 day. The cells were restimulated weekly with NV9-pulsed, antigen-presenting cells (APCs) and IL-2. The cultures were split 1:4 when the T-cell concentration exceeded 106 cells/ml. Cytolytic function was tested at different effector/target ratios using peptide-pulsed, 51 Cr-labeled phytohemagglutinin blast targets (9).
RESULTS
Use of cell-ELISA to test antigen-specific T-cell lines.
The accuracy of assays that measure T-cell responses is limited by the fact that no cellular standard is generally used. Established T-cell lines that contain 100% antigen-specific T cells represent a valuable, but not easily available, standard. We have produced CD4 and CD8 T-cell lines with different specificities (7, 9).
The consistent production of IFN-γ by T-cell lines at different restimulation cycles was tested in preliminary experiments. CD8 lines specific for pp65 NV9 peptide and CD4 lines specific for tetanus toxoid (TT) or for pp65 pep30 (amino acids 117 to 131) were stimulated in 1,536-well plates. Supernatants from replicate wells were collected and tested for IFN-γ concentration by the Luminex assay (CD8 supernatants) and by the Cytomix assay (CD4 supernatants). Cytokine concentrations fluctuated within a limited range from 1,500 to 7,500 pg/ml in the T-cell lines tested between the second and seventh restimulation cycles.
CD4 or CD8 T cells from established T-cell lines were cultured for 3 weeks after restimulation with antigen and APCs, followed by expansion with IL-2. Once the cells had reverted to a resting state (loss of proliferation and loss of blast-like appearance), they were used for the restimulation experiments shown here. T cells were labeled with the vital dye CFSE and mixed 1:10 with autologous PBMCs in the presence or in the absence of antigen. After a 2-h incubation, brefeldin was added overnight and the cells were finally stained for IFN-γ.
Figure 1 demonstrates that representative T-cell lines contain up to 100% specific T cells. In the case of an HIV-specific CD4 line (Fig. 1A and B), >97% of the input T cells labeled with CFSE and spiked into nonresponding PBMCs produced IFN-γ upon antigenic stimulation. Similarly, in the case of a CMV-specific CD8 T-cell line, both IFN-γ ICS (Fig. 1C and D) and tetramer enumeration (Fig. 1E) concurred to define that the line contained >98% specific cells. While a scattered staining for intracytoplasmic IFN-γ production is generally observed with PBMCs, it can be noted that, as shown in the plots in Fig. 1, T-cell lines were highly homogeneous with respect to IFN-γ staining. Since resting T cells overlapped with the lympho gate of PBMCs, the same gate was used for analysis of the mixtures (CFSE-labeled T cells plus IFN-γ phycoerythrin staining as shown in Fig. 1A to D). No differences were detected by physical analysis between mixtures with and without antigen when the cell mixtures were analyzed after incubation (data not shown).
FIG. 1.
Sensitivity of cell-ELISA. T-cell lines were used as possible standards. A CD4 T-cell line specific for peptide 24 of HIV gp120 (A and B) and a CD8 T-cell line specific for peptide NV9 of CMV pp65 (C, D, and E) were tested to measure the frequency of antigen-responsive cells within the T-cell line. Panels A, B, C, and D show intracytoplasmic staining for IFN-γ on T-cell lines diluted 1:10 in autologous irradiated PBMCs as a source of APCs. The T-cell lines were labeled with CFSE for identification. After antigen stimulation (+ag), only CFSE-positive T cells costained for IFN-γ. T-cell lines are more homogeneous than PBMCs with respect to intracytoplasmic staining. In addition, the CD8 T-cell line was checked by NV9 pentamer staining, showing that also according to this parameter the line contains antigen-specific cells only (E). The panels on the right depict Z factor scores by using liminal dilutions of CD4 and CD8 T-cell lines into autologous, nonresponding PBMCs in 1,536-well plates. We tested 1,536 replicates of zero and five specific T cells diluted into 104 PBMCs. Significantly positive signals were detected with five cells per well with both CD4 and CD8 T-cell lines (Z = 0.63 and Z = 0.51). OD, optical density; FITC, fluorescein isothiocyanate; PE, phycoerythrin.
In order to estimate a sensitivity threshold, the CD4 and CD8 T-cell lines were diluted into irradiated, autologous, nonresponding PBMCs (right panels). In both cases, it can be seen that as few as five specific T cells in 104 PBMCs per well, corresponding to a frequency of 0.05%, could be detected by cell-ELISA. Z factor scores for the two experiments were 0.63 and 0.51. Since a Z factor score between 0 and 0.5 indicates a marginal assay and a Z factor score between 0.5 and 1 indicates an excellent assay (20), a 0.05% frequency of specific T cells can be reliably detected by our method. To further demonstrate their usefulness as reference standards, T-cell lines at different restimulation cycles were challenged with antigen and IFN-γ was measured in supernatants by Luminex or Cytomix assays. IFN-γ levels fluctuate within a restricted range between 1,500 and 7,500 pg/ml, irrespective of the restimulation cycle.
Sensitivity of cell-ELISA using PBMCs.
Since T cells from established lines may not mirror T cells from PBMCs, we used donors with steady frequencies of CMV NV9-positive, pentamer-positive CD8 cells (donor 3, 0.5%; donors 1, 4, and 5, 1%) over time. Untreated PBMCs were diluted into autologous, irradiated, nonresponding PBMCs as a source of APCs. Pentamer staining was taken as a reliable measure of NV9-positive CD8 T cells. As shown in Table 1, the different methods were comparable with respect to sensitivity, with around five specific T cells out of 5 × 104 PBMCs (0.01%) detected by the different assays. By considering the frequency of specific T cells (either assumed by pentamer staining, as in the case of PBMCs, or experimentally determined, as in the case of T-cell lines) and the amount of IFN-γ produced in the culture vessel, we can estimate the amount of cytokine produced by a single cell in a time unit. As shown in Table 2, we tested PBMCs from donors 3 and 1 that were undiluted or diluted 1:5 into autologous, irradiated PBMCs. The total amount of IFN-γ produced upon NV9 stimulation was divided by the number of specific T cells present in the culture, showing that a single T cell produced 1.1 to 3.7 pg cytokines in 36 h. Similar results were obtained with a CD4 T-cell line serially diluted into autologous, nonresponding PBMCs, showing that one single T cell produced 1.3 to 2.7 pg IFN-γ in 36 h (Table 3). Assuming a mass of 20 kDa for human IFN-γ (2), this corresponds to approximately 3 × 107 IFN-γ molecules secreted by a single T cell.
TABLE 1.
Comparative sensitivity of different assaysa
| Assay | No. of NV9-specific cells inb:
|
|||
|---|---|---|---|---|
| 3 | 1 | 4 | 5 | |
| ICS | 16-40 | 2-10 | 2-10 | 2-10 |
| ELISPOT | 1-5 | 2-10 | 2-10 | 1-5 |
| Proliferation | 1-5 | 2-10 | 2-10 | 2-10 |
| Cell-ELISA | 1-5 | 1-5 | 1-5 | 1-5 |
PBMCs containing 0.5% (donor 3) and 1% (donors 1, 4, and 5) CMV NV9 pentamer-positive CD8 cells were serially diluted into autologous, irradiated, nonresponding PBMCs and tested by ICS, ELISPOT, proliferation, and cell-ELISA assays.
The highest dilutions resulting in a detectable positive signal were considered, and the corresponding absolute range numbers of NV9-specific cells out of 5 × 104 PBMCs are given. The lowest detection range is between 1 and 5 specific cells out of 5 × 104 PBMCs.
TABLE 2.
IFN-γ produced by NV9-specific single CD8 T-cells from PBMCsa
| Donor | Dilution | No. of NV9-positive CD8 cells | IFN-γ (pg/ml) | IFN-γ (pg/50 μl) | IFN-γ (pg/cell) |
|---|---|---|---|---|---|
| 3 | 1:1 | 250 | 10,400 | 520 | 2.1 |
| 1:5 | 50 | 1,120 | 56 | 1.1 | |
| 1 | 1:1 | 500 | 32,700 | 1,610 | 3.2 |
| 1:5 | 100 | 7,440 | 372 | 3.7 |
PBMCs from donors 3 and 1 were used undiluted or diluted 1:5 into autologous, irradiated, nonresponding PBMCs. Cultures stimulated with NV9 peptide were set up in 384-well plates. The number of NV9-positive CD8 cells was determined based on the pentamer assay (0.5% for donor 3; 1% for donor 1). Concentration of IFN-γ in the supernatants was calculated according to a standard titration curve. The total amount of IFN-γ in 50 μl culture was divided by the number of NV9-specific cells present in 5 × 104 PBMCs in each well to define the amount of IFN-γ as picograms produced by one single T cell. IFN-γ produced by one specific cell ranged from 1.1 to 3.7 pg.
TABLE 3.
IFN-γ produced by single cells from an HIV gp120-specific CD4 T-cell linea
| No. of HIV-specific T cells/well | IFN-γ (pg/ml) | IFN-γ (pg/50 μl) | IFN-γ (pg/cell) |
|---|---|---|---|
| 100 | 5,400 | 270 | 2.7 |
| 10 | 256 | 13 | 1.3 |
| 1 | 30 | 1.5 | 1.5 |
| 0 | <20 | ND | ND |
Different numbers of cells (100, 10, 1, and 0) from the HIV-specific T-cell line were stimulated with antigen in the presence of autologous, irradiated, nonresponding PBMCs. The IFN-γ concentration and IFN-γ amount per well were calculated to determine cytokine production per cell. IFN-γ produced by one specific CD4 T-cell ranged from 1.3 to 2.7 pg. ND, not determined.
Epitope mapping by cell-ELISA.
The 1,536-well plates were used for a preliminary epitope mapping of different M. tuberculosis strain H37Rv proteins that have been previously reported as being immunodominant in infected or immune subjects. For initial experiments, peptide panels from small proteins (Ag85B of M. bovis, rdESAT6, rdESAT6-like protein 9, culture filtrate protein 10 [CFP-10], and RV 2628 from M. tuberculosis) were used, represented by a total number of 105 overlapping peptides. Sequence from these proteins have the following accession numbers: Ag85B, P12942; rdESAT6, CAE55648; rdESAT6-like protein 9, CAA16104; CFP-10, CAA17966; and RV 2628, NP_217144. Some of these proteins have already been mapped (1, 15, 18). RV 2628 was chosen as one of the proteins encoded by dormant operons that are particularly immunogenic in infected subjects (5).
Figure 2 shows that numerous stimulatory peptides can be defined. These experiments on three donors were not run to provide a thorough epitope mapping for these antigens, but rather as a proof of principle that cell-ELISA is a valuable tool to approach epitope mapping in a miniaturized format (1,536-well plates) by using a small number of PBMCs. In fact, 105 peptides could be tested in duplicate with 2 million PBMCs only.
FIG. 2.
Epitope mapping by cell-ELISA. PBMCs from three PPD-responsive, healthy subjects were dispensed into 1,536-well plates into which overlapping peptides of mycobacterial Ag85 protein (31 peptides), rdESAT6 protein (17 peptides), rdESAT6-like protein 9 (17 peptides), CFP-10 protein (18 peptides) and RV 2628 protein (22 peptides) had been dispensed. Plates were processed as described. Results are shown as the optical density (OD) at 405 nm at 1 h. Several peptides can be identified in all of the proteins as stimulatory for T-cell responses in the tested subjects (donors 6, 7, and 8). The dotted line indicates an arbitrary threshold set at four times the background level without antigen.
Suitability of cell-ELISA to analyze clinical samples.
Blood samples of 5 ml were also sufficient to test a broad panel of recall antigens to appreciate the immunocompetence of selected patients. Results are shown for three tuberculosis patients with active infection and for three HIV patients with active tuberculosis infection. We obtained extended profiles of specific responses (13 antigens in duplicate) with 1.3 × 106 PBMCs only (Fig. 3). It can be noted that the lack of response to the mycobacterial antigens purified protein derivative (PPD), rdESAT6, and Ag85 in HIV-positive patient donor 9 does not imply a generalized immunodeficiency. In fact, responses to CMV antigens (CMV lysate, CD4, and CD8 peptide pools), to C. albicans, and to Aspergillus fumigatus were present.
FIG. 3.
Testing a broad panel of recall antigens with limited numbers of PBMCs. Thirteen antigenic preparations plus PHA were dispensed in duplicate into 384-well plates and tested with 5 × 104 PBMCs/well from patients with active tuberculosis with (HIV+) or without (HIV−) HIV infection. Antigens included CMV lysate, CD4 and CD8 peptide pools from CMV pp65, TT, mycobacterial proteins (PPD, rdESAT6, Ag85), fungi or protozoa (Candida albicans [Ca], Aspergillus fumigatus [Asp fum], Aspergillus niger [Asp niger], Cryptococcus neoformans [Cry], Pneumocystis carinii [Pc], and Toxoplasma gondii [Toxo]). Results are shown as IFN-γ pg/ml ± the standard deviation.
Cytokine profiling on cell-ELISA supernatants.
Culture supernatants from cell-ELISA can be transferred to a replica plate. Supernatants from positive wells, according to cell-ELISA, can be screened later for extended cytokine profiles by using available micromethods. Figure 4 shows examples of cytokine profiles determined by Cytomix or by Luminex assays on 10 μl of supernatants from three CD4 T-cell lines specific for CMV, for an immunodominant CMV pp65 peptide (pep30; amino acids 117 to 131), and for TT and from two CD8 T-cell lines specific for the NV9 peptide of CMV pp65. The results indicate that most cytokines were produced only upon antigen stimulation and therefore were informative to characterize positive responses. In contrast, other cytokines (e.g., IL-8) with high background production in the absence of antigen (Fig. 4, black bars) were poorly informative. These cytokine profiles do not mirror profiles observed with T-cell lines specific for different antigens and with antigen-stimulated PBMCs (data not shown), suggesting that these profiles are markers of individual T-cell responses and cannot be easily predicted.
FIG. 4.
Cytokine profiles on cell-ELISA supernatants from antigen-specific CD4 and CD8 T-cell lines. CD4 T-cell lines were generated from donor 1 (CMV lysate and pp65 pep30) and donor 2 (TT). CD8 T-cell lines specific for the NV9 peptide of CMV pp65 were generated from donors 3 and 1. T-cell lines were tested at the fourth restimulation cycle. Cultures were performed in 1,536-well plates. Supernatants were collected after 48 h and tested by the Cytomix assay for CD4 lines and by the Luminex assay for CD8 lines. The concentrations of cytokines induced by antigenic stimulation are shown as picograms/milliliter by gray bars, while background concentrations in the absence of antigen are shown by black bars.
Comparative IFN-γ concentration by cell-ELISA and conventional ELISA.
The possibility of using supernatants from cell-ELISA cultures to test IFN-γ by conventional ELISA or by other high-throughput methods is confirmed by data shown in Fig. 5. PBMCs from donor 1 were stimulated with TT, PPD, and CMV and with pp65 immunodominant peptides 45, 62, and 128 (6). Cell-ELISA wells were developed in parallel with a conventional ELISA performed on supernatants from the same cultures. IFN-γ concentrations detected by ELISA (Fig. 5, gray bars) in the supernatants were comparable to or slightly lower than concentrations detected by cell-ELISA (Fig. 5, black bars). This demonstrates that only a small fraction of the cytokine can be sequestered by the solid-phase antibody present in the cell-ELISA plates, thereby remaining available for subsequent cytokine assays based on ELISA-like methods.
FIG. 5.
IFN-γ detected by cell-ELISA and by conventional ELISA in cell-ELISA supernatants (sups). PBMCs from donor 1 were stimulated with TT, PPD, CMV, and with pp65 immunodominant peptides 45, 62, and 128. Cell-ELISA wells were developed in parallel with conventional ELISA performed on supernatants from the same cultures. IFN-γ concentrations detected by ELISA in the supernatants (gray bars) were comparable to or slightly lower than concentrations detected by cell-ELISA (black bars).
DISCUSSION
Improved methods to test T-cell immunity are highly desirable for both basic and clinical immunology. A drawback of current assays is that the number of PBMCs required to test each antigen limits the antigenic analytical breadth. This must be kept in mind, considering that patients who most often need to be examined for cellular immunocompetence are severely lymphopenic. This is also a problem for pediatric patients, from whom small blood samples are available. In addition, when proteins need to be dissected with panels of overlapping peptides to identify immunodominant sequences (19), large numbers of lymphocytes from the responding subjects are required (12, 16). Therefore, assay miniaturization can help enhance the capacity to define the specific T-cell response with the available PBMCs and to extend the analytical power of epitope mapping.
Here we described the miniaturization of the immunoenzymatic assay we have previously reported (8), which allows testing of a single antigen with as few as 104 PBMCs cultured in a volume of 10 μl complete medium in 1,536-well plates. The wells, precoated with a monoclonal antibody specific for a given cytokine (e.g., IFN-γ), received the antigens to be tested and were frozen for months without a loss of function. PBMCs were seeded in the thawed plates and cultured for 2 days. At the end of the culture period, the supernatants could be saved for later cytokine profiling and the wells were developed by conventional ELISA to reveal the captured cytokine produced by antigen-activated T cells.
All of the procedures based on the use of 1,536-well plates were automated by using liquid handlers and dispensers, in addition to appropriate scanners. In fact, while automated handling of reagents and cells in 384-well plates is advisable (8), it becomes mandatory in 1,536-well plates. In order to use a cell density similar to that of conventional 96-well plates, we dispensed 104 PBMCs in 10 μl per well. This value may represent the lowest threshold for a miniaturized T-cell assay. In fact, the frequency of antigen-specific T cells is the limiting factor (4, 11). Assuming a reasonable frequency of 1% to 0.1% of T cells specific for a given recall antigen, we need at least 103 PBMCs to have one specific T cell per well giving a positive signal. Since our method has a sensitivity of around five cells per well, we need at least 5 × 103 PBMCs per well. We may need even more cells if we consider the expected frequencies of CD4 T cells, which are generally lower than the frequencies of CD8 T cells (16).
The frequency of specific T cells in PBMCs cannot be defined with accuracy because cellular standards are not easily available. Therefore, we used established CD4 and CD8 T-cell lines as possible standards, since they contain 100% specific T cells and can be spiked into nonresponding PBMCs to test assay accuracy. In addition, established T-cell lines can also be used to compare sensitivity thresholds of different assays, as previously reported (8). These arguments have relevant implications when it comes to the definition of accuracy and precision of T-cell assays for clinical purposes.
Another point that is worth considering is whether the frequency of cytokine-producing specific T cells is more or less relevant than the overall amount of effector cytokines produced at the inflammation site. While functional in vitro assays that enumerate specific T cells, including ICS, ELISPOT, and cytokine capture assays, cannot appreciate cytokine concentration (4, 11), cell-ELISA, like conventional ELISA, can provide quantitative data on numerous cytokines. Both pieces of information are likely helpful to better define correlations between in vitro T-cell responses and clinical parameters, but more studies are needed to reach a consensus on this issue.
The high-throughput approach can also be applied to epitope mapping. Automation allows simultaneous manipulation of numerous 1,536-well plates, thereby permitting us to test, in principle, thousands of different peptides. To underline the analytical power of the method, we should keep in mind that a 1,536-well plate corresponds to sixteen 96-well plates. At the same time, reducing the number of tested cells per antigen from 2 × 105 to 5 × 105 PBMCs with conventional methods down to 104 PBMCs in the 1,536-well plate results in 50- to 20-fold-higher screening power with the same number of available PBMCs. The data presented here on epitope mapping with peptide panels and on analyses of patients’ responses with panels of recall antigens are examples of how cell-ELISA can be conveniently applied. In fact, thanks to the small number of PBMCs required to test numerous antigens, this assay is currently applied to extended clinical studies of cohorts of lymphopenic and pediatric patients.
In order to test broad antigenic panels with limited amounts of PBMCs, peptide pools arranged as matrixes can be used (3, 13). Nevertheless, this approach requires sequential steps to define the individual immunodominant peptide(s). Alternatively, cell-ELISA can be used as a high-throughput approach for epitope mapping by testing all of the individual peptides in one shot. As a proof of principle, we have shown here that this is feasible by using small mycobacterial proteins. An additional advantage of cell-ELISA is represented by the fact that saved supernatants can be tested for extended cytokine profiles using high-throughput methods based on flow cytometry or on proteomic microarrays that are required when small volumes are available for testing.
In conclusion, extension of the cell-ELISA assay from the 384- to the 1,536-well format provides a useful tool for analysis of T-cell immunity and for epitope mapping, thanks to the small number of required PBMCs. In addition, the sensitivity of the assays to low frequencies of specific cells and the possibility of defining cytokine profiles, coupled with automation for HTS, suggest that this assay may have numerous applications as a complement to other current methods for the measurement of antigen-specific T-cell responses.
Acknowledgments
This work was supported by the Comitato Interministeriale Programmazione Economica (CIPE), Rome, Italy (2007), Regione Liguria, Genoa, Italy (2007); Ministry of Health, Rome, Italy (Regional Finalized Project 2008; Establishment of a GMP-Validated Biobank); Ministry of Higher Education, Rome, Italy (FIRB RBNEOI-RB9B.003 and FIRB RBIP064CRT-006); Italian Health Institute, Rome, Italy (AIDS 40G.33, 45G.17, and 50G.25); and EU, Brussels, Belgium (grant no. LSHB-CT-2005-018680).
Footnotes
Published ahead of print on 22 October 2008.
REFERENCES
- 1.Chapman, A. L., M. Munkanta, K. A. Wilkinson, A. A. Pathan, K. Ewer, H. Ayles, W. H. Reece, A. Mwinga, P. Godfray-Faussett, and A. Lalvani. 2002. Rapid detection of active and latent tuberculosis infection in HIV-positive individuals by enumeration of M. tuberculosis-specific T-cells. AIDS 16:2285-2293. [DOI] [PubMed] [Google Scholar]
- 2.Kelker, H. C., J. Le, B. Y. Rubin, Y. K. Yip, C. Nagler, and J. Vilcek. 1984. Three molecular weight forms of natural human interferon gamma revealed by immunoprecipitation with monoclonal antibodies. J. Biol. Chem. 259:4301-4304. [PubMed] [Google Scholar]
- 3.Kern, F., T. Bunde, N. Faulhaber, F. Kiecker, E. Khatamzas, I. M. Rudawski, A. Pruss, J. W. Gratama, and R. Volkmer-Engert. 2002. Cytomegalovirus (CMV) phosphoprotein 65 makes a large contribution to shaping the T cell repertoire in CMV-exposed individuals. J. Infect. Dis. 185:1709-1716. [DOI] [PubMed] [Google Scholar]
- 4.Kern, F., G. Li Pira, J. W. Gratama, F. Manca, and M. Roederer. 2005. Measuring ag-specific immune responses: understanding immunopathogenesis and improving diagnostics in infectious disease, autoimmunity and cancer. Trends Immunol. 26:477-484. [DOI] [PubMed] [Google Scholar]
- 5.Leyten, E. M., M. Y. Lin, K. L. Franken, A. H. Friggen, C. Prins, K. E. van Meijgaarden, M. I. Voskuil, and M. R. Klein. 2006. Human T-cell responses to 25 novel antigens encoded by genes of the dormancy regulon of Mycobacterium tuberculosis. Microbes Infect. 8:2052-2060. [DOI] [PubMed] [Google Scholar]
- 6.Li Pira, G., L. Bottone, F. Ivaldi, R. Pelizzoli, F. Del Galdo, L. Lozzi, L. Bracci, A. Loregian, G. Palu, R. De Palma, H. Einsele, and F. Manca. 2004. Identification of new Th peptides from the cytomegalovirus protein pp65 to design a peptide library for generation of CD4 T cell lines for cellular immunoreconstitution. Int. Immunol. 16:635-642. [DOI] [PubMed] [Google Scholar]
- 7.Li Pira, G., L. Bottone, F. Ivaldi, A. Tagliamacco, S. Fiordoro, A. Ricciardi, G. Barbano, and F. Manca. 2005. Generation of cytomegalovirus (CMV)-specific CD4 T cell lines devoid of alloreactivity, by use of a mixture of CMV-phosphoprotein 65 peptides for reconstitution of the T helper repertoire. J. Infect. Dis. 191:215-226. [DOI] [PubMed] [Google Scholar]
- 8.Li Pira, G., F. Ivaldi, L. Bottone, and F. Manca. 2007. High throughput functional microdissection of pathogen-specific T-cell immunity using antigen and lymphocyte arrays. J. Immunol. Methods 326:22-32. [DOI] [PubMed] [Google Scholar]
- 9.Li Pira, G., F. Ivaldi, L. Bottone, G. Koopman, and F. Manca. 2007. Helper function of cytolytic lymphocytes: switching roles in the immune response. Eur. J. Immunol. 37:66-77. [DOI] [PubMed] [Google Scholar]
- 10.Li Pira, G., L. Bottone, F. Ivaldi, F. Del Galdo, F. Papa, R. Accolla, G. Koopman, G. Abbate, G. De Berardinis, L. D'Apice, R. De Palma, and F. Manca. 2005. Human naive CD4 T-cell clones specific for HIV envelope persist for years in vivo in the absence of antigenic challenge. J. Acquir. Immune Defic. Syndr. 40:132-139. [DOI] [PubMed] [Google Scholar]
- 11.Li Pira, G., F. Kern, J. Gratama, M. Roederer, and F. Manca. 2007. Measurement of antigen specific immune responses: 2006 update. Cytometry B 72:77-85. [DOI] [PubMed] [Google Scholar]
- 12.McKinney, D. M., R. Skvoretz, M. Qin, G. Ishioka, and A. Sette. 2000. Characterization of an in situ IFN-γ ELISA assay which is able to detect specific peptide responses from freshly isolated splenocytes induced by DNA minigene immunization. J. Immunol. Methods 237:105-117. [DOI] [PubMed] [Google Scholar]
- 13.Precopio, M. L., T. R. Butterfield, J. P. Casazza, R. A. Koup, and M. Roederer. 2008. Optimizing peptide matrices for identifying T-cell antigens. Cytometry A 73:1071-1078. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sette, A., and B. Peters. 2007. Immune epitope mapping in the post-genomic era: lessons for vaccine development. Curr. Opin. Immunol. 19:106-110. [DOI] [PubMed] [Google Scholar]
- 15.Silver, R. F., R. S. Wallis, and J. J. Ellner. 1995. Mapping of T-cell epitopes of the 30-kD alpha antigen of Mycobacterium bovis strain bacillus Calmette-Guerin in purified protein derivative (PPD)-positive individuals. J. Immunol. 154:4665-4674. [PubMed] [Google Scholar]
- 16.Sylwester, A. W., B. L. Mitchell, J. B. Edgar, C. Taormina, C. Pelte, F. Ruchti, P. R. Sleath, K. H. Grabstein, N. A. Hosken, F. Kern, J. A. Nelson, and L. J. Picker. 2005. Broadly targeted human cytomegalovirus-specific CD4+ and CD8+ T cells dominate the memory compartments of exposed subjects. J. Exp. Med. 202:673-685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Thiel, A., A. Scheffold, and A. Radbruch. 2004. Antigen-specific cytometry: new tools arrived! Clin. Immunol. 111:155-161. [DOI] [PubMed] [Google Scholar]
- 18.Valle, M. T., A. M. Megiovanni, A. Merlo, G. Li Pira, L. Bottone, G. Angelini, L. Bracci, L. Lozzi, K. Huygen, and F. Manca. 2001. Epitope focus, clonal composition and Th1 phenotype of the human CD4 response to the secretory mycobacterial antigen Ag85. Clin. Exp. Immunol. 123:226-232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vita, R., B. Peters, and A. Sette. 2008. The quality or power of a high-throughput (HTS) assay. Cytometry A 73:1066-1070.18688821 [Google Scholar]
- 20.Zhang, J. H., T. D. Y. Chung, and K. R. Oldenburg. 1999. Simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen. 4:67-73. [DOI] [PubMed] [Google Scholar]