Ultra-sensitive class I tetramer analysis reveals previously undetectable populations of antiviral CD8+ T cells



A major breakthrough in cellular immunology has been the development of HLA class I tetramers to analyze CD8+ T cell responses. However, in many situations, including persistent virusinfection, specific T cell responses are rarely detected using this technology. This raises the question of whether such responses are ‘deleted’ (or ‘exhausted’) or present below the conventional detection limit for class I tetramer staining. In particular, persistent hepatitis C virus (HCV) infection is characterized by very weak or apparently absent specific CD8+ T cell responses, even though they are readily detectable in acute disease. Therefore, we assessed the use of anti-PE-labeled magnetic beads to enrich tetramer-positive HCV-specific T cells and identify previously undetectable populations. Using the enrichment technique, HCV-specific T cells could be detected in the majority of infected individuals, whereas these responses were not detected using conventional tetramer staining (8/15 vs. 1/15; p=0.01). Magnetic enrichment could reliably detect very rare HCV-specific responses at frequencies of >0.0011% of CD8+ T cells (∼1/million PBMC), and phenotypic analysis of these rare populations was possible. Therefore, this direct ex vivo technique revealed the persistence of very low frequencies of virus-specific CD8+ T cells during chronic virus infection and is readily transferable to the study of other viral, self- or tumor-specific T cells.


Hepatitis C virus

1 Introduction

The recent development of HLA class I tetramer technology 1 has enabled the detection of antigen-specific T cells at the single-cell level. Quantification, phenotypic analysis, and selection of these cells using FACS sorting or magnetic separation 24 is now possible and is providing new insights into mechanisms of disease pathogenesis.

Using HLA class I tetramers, we and others have previously shown that hepatitis C virus (HCV)-specific CD8+ T cells are detected during acute HCV infection and that multiple epitopes are targeted 5, 6. However, once persistent infection develops, HCV-specific CD8+ T cell responses become undetectable ex vivo (using tetramers or IFN-γ ELISpot) in the majority of infected individuals 7. In some persistently infected individuals with undetectable ex vivo responses, these responses are clearly detectable following a single round of peptide stimulation in vitro8 or following the treatment of these patients with IFN-α and ribavirin 9, suggesting that memory populations of antigen-specific CD8+ T cells are present in some infected individuals but lie below the threshold of detection using conventional HLA class I tetramer staining techniques.

HLA class I tetramers will detect antigen-specific T cells at a lower limit of 0.02% of CD8+ T cells 5. Below this frequency it is usually not possible to discern adiscrete tetramer+ population, either because the response is truly negative or because even a small amount of background staining makes interpretation problematic. Studies of HCV-specificCD8+ T cell responses using these tetramers frequently encounter responses at around this cut-off 7, 10, and interpretation of these very small populations can be difficult.

The problem of assessing low frequencies of antigen-specific T cells does not only apply to HCV infection. Low frequencies may also be encountered in HIV infection while patients are on therapy, in influenza virus infection, and in response to self- and tumor-reactive antigens.

In this study, we first assess the use of anti-PE-labeled magnetic beads in the enrichment of anti-viral-specific CD8+ T cells bound to PE-labeled HLA class I tetramers when tetramer populations are clearly visible ex vivo. We then address the question of whether rare populations of antigen-specific CD8+ T cells can be detected in situations when tetramer populations are absent or equivocal ex vivo.

2 Results

2.1 Magnetic selection of tetramer+ virus-specific cells markedly enriches the tetramer+ cells in the CD8+ T cell population and reduces background staining

Initially, we assessed the use of anti-PE-labeled magnetic beads to enrich tetramer+ cells from three individuals (C-HCV1, HI1, and C-HCV2), in which responses were detectable ex vivo using conventional tetramer staining (Fig. 1A). For this analysis, we used EBV-, parvovirus- and CMV-specific tetramers. Using conventional tetramer staining, virus-specific T cells could be detected at a frequency between 0.11% and 1.05% of CD8+ T cells. At lower frequencies, visualization of a discrete tetramer+ population (for example EBV tetramer staining on HI1 with a frequency of 0.11%, Fig. 1A) became difficult. Following the enrichment of the tetramer+ populations, distinct tetramer+ populations could be detected at frequencies of between 44.49% and 90% of CD8+ T cells.

Figure 1.

(A) Magnetic selection of HLA class I tetramers when tetramer+ populations are visible ex vivo. Anti-PE-labeled magnetic beads were used to enrich tetramer+ cells from three individuals (C-HCV1, HI1 and C-HCV2) in which responses were detectable ex vivo using conventional tetramer staining. EBV-, parvovirus- and CMV-specific tetramers were used. Using conventional tetramer staining (pre-enrichment column), virus-specific T cells could be detected at frequencies of between 0.11% and 1.05% of CD8+ T cells. Following the enrichment of the tetramer+ populations (post-enrichment column), tetramer+ cells could be detected at frequencies of between 44% and 90% of CD8+ T cells. The percentages of tetramer+/CD8+ cells are given in the upper right quadrant of each FACS plot. (B) Magnetic selection of naive Melan-A-specific CD8+ T cells. Anti-PE-labeled magnetic beads were used to enrich naive Melan-A tetramer+ cells from a healthy individual (HI2) after conventional tetramer staining. Low-frequency Melan-A-specific cells (pre-enrichment frequency: 0.03%) could be enriched to a frequency of 83.78% of CD8+ T cells (post enrichment). The percentages of tetramer+/CD8+ cells pre and post enrichment are given in the upper right quadrant of each FACS plot.

2.2 Magnetic selection of naive populations of Melan-A-specific T cells

Self- or tumor-reactive T cells are reported to be of low avidity 11. To ensure that this technique could be applied to the enrichment of such a T cell population, we used Melan-A-specific T cells. Phenotypic 12, functional 13 and molecular techniques 14 have confirmed unequivocally that in normal individuals, these T cells are truly antigen naive. Following enrichment, Melan-A-specific T cells could be found at a frequency of 83.78% of CD8+ T cells (Fig. 1B).

Tetramer enrichment preferentially selected the tetramer+ cells with the highest level of fluorescence. This reduces the amount of background staining in the post-enrichment samples and allows the identification of discrete tetramer+ populations. This is most clearly observed for the EBV tetramer staining in individuals C-HCV1 and HI1, where the mean fluorescent index increases from 441.99 to 490.66 and 163.75 to 248.45, respectively, in the tetramer+ populations (Fig. 1A). An alternate strategy to detect rare populations of T cells might be to tetramer stain large numbers (e.g. 4–5×106) of cells. However, using this strategy, the relatively high background staining that was obtained in the CD8high population made interpretation of this data difficult (Fig. 2).

Figure 2.

Comparison of enrichment versus conventional staining of high cell numbers. The left panel shows CMV tetramer/CD8 staining of 4.5×106 PBMC (2.2×106 events from a conventional lymphocyte gate are shown) from a normal individual who lacks a CMV response by ELISpot assay or after in vitro peptide restimulation. A background staining with an indistinct tetramer population of 0.012% (i.e. less than 0.02% conventional cut-off) is seen. A further 6×106 PBMC from this patient were used for enrichment. Of this sample, 10% (6×105 cells) was set aside for the pre-enrichment analysis (middle panel; 2.8×105 cells from a live lymphocyte gate are shown). The remaining 90% of the sample was used for enrichment as described in Sect. 4. From this, 3,600 cells were obtained in the post-enrichment sample (right panel; 186 cells are shown from the identical lymphocyte gate used for pre-enrichment analysis).

2.3 Magnetic selection of tetramer+ cells allows antigen-specific T cells to be detected when they are not observed using conventional tetramer staining.

To assess the background level of tetramer+ cell staining following magnetic selection and thus define an appropriate lower limit of detection for this assay, we recruited nine individuals as negative controls and assessed them using six different HCV-specific tetramers. The mean frequency of tetramer+/CD8+ T cells in this group, following magnetic enrichment, was 0.0001037% (standard deviation 0.0002667%, range 0–0.0011%) of CD8+ T cells. Based on these findings, we defined the stringent lower limit of detection of tetramer+ cells using magnetic enrichment as >0.0011% of CD8+ T cells.

We then proceeded to evaluate this technique to detect HCV-specific responses that could not be detected using conventional tetramer staining (Fig. 3; first column of FACS plots) in two individuals with resolved HCV infection (R-HCV1 and R-HCV2). Following the selection of tetramer-specific responses, small discrete tetramer+ populations could be clearly observed (Fig. 3; second column of FACS plots). Despite some of the small cell populations shown (upper two rows, middle column), these populations are well above the background level seen in negative controls. The maximal number of PBMC was used per column and, hence, the total number of enriched CD8+/tetramer+ cells shown is limited by the column capacity. However, to ensure that these small tetramer+ populations were not artefacts due to the enrichment process, we simultaneously performed a single round of peptide stimulation (using identical peptides to those used in the tetramer constructs), followed by conventional tetramer staining 10 days later (Fig. 3; third column of FACS plots). Following peptide stimulation, large populations of tetramer+ cells were clearly identified with frequencies as high as 9.4% of CD8+ T cells. In addition, the tetramer frequency observed following enrichment correlated with the tetramer frequency observed following peptide stimulation, in that the larger the tetramer+ population observed following magnetic selection, the larger the tetramer+ population observed following peptide stimulation.

Figure 3.

 Magnetic selection of HCV-specific class I tetramers when responses are invisible ex vivo using conventional tetramer staining. Anti-PE-labeled magnetic beads were used to enrich tetramer+ cells from two individuals with resolved HCV infection (R-HCV1 and R-HCV2), in which responses were below the level of detection when using conventional tetramer staining (first column). Following enrichment (second column), the detection of discrete tetramer+ populations was possible. Peptide stimulation by using the same peptide as that used to construct the respective tetramer, followed by 10-day culture and then conventional tetramer staining (third column) confirmed that the small populations observed post enrichment were not artefacts. The percentages of tetramer+/total CD8+ cells are given in the upper right quadrant of each FACS plot.

2.4 Magnetic selection enables phenotyping of very small populations of HCV-specific T cell responses ex vivo

One key advantage associated with tetramer staining over other methods of detection is the ability to determine the phenotype of antigen-specific CD8+ T cells. This is technically difficult in situations when such cells are directly detectable but low in frequency. We therefore determined whether phenotyping of tetramer-enriched populations was possible.

For this, we tested the enriched HCV-specific T cells for markers associated with T cell differentiation, namely CD27 and CD28. We focused on these markers as we have previously shown, using samples in which tetramer+ clouds were readily visible without enrichment 15, that HCV-specific CTL are generally high in CD27 and CD28. This distinguishes them from antiviral tetramer+ cells in other infections such as CMV 15.

Enriched HCV-specific cells typically displayed an "early memory" phenotype with tetramer+ cells being positive for both CD27 and CD28 (Fig. 4). Naive CD8+ T cell populations such as Melan-A-specific T cells also exhibited a CD27high phenotype (data not shown). This was distinct from the phenotype observed in enriched CMV-specific responses with consistently lower CD27 and CD28 expression on tetramer+ cells, indicating a later stage of maturation 15. No activation of enriched tetramer+ T cells as measured by surface expression of CD25 was observed, suggesting that the enrichment process per se does not lead to substantial activation of these cells over the course of the experiment (data not shown).

Figure 4.

Magnetic selection enables the phenotyping of rare populations of HCV-specific T cells. These FACS plots are gated on CD8+/tetramer+ cells following magnetic enrichment. They show the expression of CD28 (top panels) and CD27 (lower panels) on CMV- and HCV-specific T cells in individuals with resolved and persistent HCV infection. Quadrants are based on distinct CD27+ and CD28+ populations seen when gated on the total lymphocyte population.

2.5 Conventional tetramer staining significantly underestimates the frequency of HCV-specific responses in individuals with HCV infection

We used the technique of magnetic enrichment to assess the frequency of HCV-specific responses in 15 individuals, 12 with persistent HCV infection and 3 with resolved HCV infection, and compared this to the frequency of HCV-specific responses detected using conventional tetramer staining (Fig. 5). Using conventional tetramer staining with six HCV-specific tetramers, responses were detected in only 1/39 assays performed. In contrast, using magnetic tetramer enrichment, we could detect HCV-specific responses in 9/39 HCV assays (p=0.0032; Fisher's exact test). Overall, we identified responses in 8/15 individuals (3/3 individuals with resolved infection and 5/12 individuals with chronic infection) using magnetic selection, compared to a single individual using conventional tetramer staining (p=0.0142; Fisher's exact test). The frequency of the HCV-specific tetramer+/CD8+ responses detected using magnetic enrichment was low (mean 0.0137%; range 0.0018–0.0660% of CD8+ T cells). We were able to confirm that the very lowest frequency of tetramer+/CD8+ cells (0.0018%) post enrichment, that was above our previously defined cut-off, was truly positive by conventional tetramer staining following peptide stimulation. There was one assay in which a small (0.0004% of CD8+ T cells) but distinct tetramer+ population was observed that fell below our previously defined cut-off (0.0011%). Here, we were able to confirm that this was in fact a truly positive tetramer+ population by peptide stimulation followed by conventional tetramer staining. This suggests that the ex vivo HCV-specific response may be extremely small, even falling below our stringent cut-off.

Figure 5.

The frequency of HCV-specific responses after using magnetic selection compared to conventional tetramer staining. The magnetic selection of six different HCV-specific HLA class I tetramers was used to assess the frequencies of HCV-specific responses in 15 individuals, 12 with persistent HCV infection and 3 with resolved HCV infection. The resulting frequencies are compared to the frequencies of HCV-specific responses detected using conventional tetramer staining. The limit of detection of these assays is shown by the dotted line. The bold line in the post-enrichment group indicates the mean positive response. Significant differences (Fisher's exact test; p<0.05) between pre and post enrichment are indicated with their p value.

3 Discussion

Using conventional tetramer staining, we have previously shown that HCV-specific CD8+ T cell responses are at or below the level of detection in the large majority of individuals with persistent or resolved HCV infection 5. Nonetheless, it is clear that in some cases, these responses do exist as detection becomes possible following a single round of peptide stimulation in vitro8, suggesting that memory populations of antigen-specific CD8+ T cells are present in some infected individuals but lie below the threshold of detection by use of conventional HLA class I tetramers. Furthermore, current analysis using conventional tetramer staining has been based on unusually large tetramer+ populations that are detected in a very small minority of HCV-infected individuals and that may distort the true phenotype and function of HCV-specific cells typically arising during HCV infection. The technique of magnetic tetramer enrichment has been demonstrated before 3 in situations when tetramer+ populations are large and clearly defined. We show here that this technique can be readily applied to detect very rare populations ex vivo, when tetramer+ populations are not detectable using conventional tetramer staining. Phenotypic and functional analysis of these rare populations is now possible.

Magnetic selection reduces the "background" tetramer staining by preferentially selecting tetramer+ cells with a high tetramer+ fluorescent index, allowing discrete tetramer+ populations to be quantified and characterized. Furthermore, magnetic selection helps to distinguish between background staining and real tetramer+ T cell populations. This is perhaps not surprising, as cells with more tetramer binding to their cell surface will preferentially bind to the anti-PE-labeled magnetic beads. This may be of biological significance in that high- and low-avidity T cells may possess different effector capacity in vivo11. Additionally, nonspecific background staining of CD8 cells with tetramer can be reduced by the more extensive washing procedure used in the enrichment method.

Magnetic selection increases the likelihood of detecting HCV-specific cells; i.e., this assay is more sensitive than conventional tetramer staining, with a lower limit of detection of0.0011% of CD8+ T cells as defined in this study, although potentially even lower than this in other settings. As only a proportion of antigen-specific cells produce cytokines in response to peptide stimulus (about 1 in 10 in the case of HCV infection) 16, 17, this assay will also be more sensitive than functional assays such as ELISpot and intracellular cytokine staining.

In this study, we show that HCV-specific responses occur in the majority of patients with persistent or resolved HCV infection, but at very low frequencies (mean 0.0137%; range 0.0018–0.0660% of CD8+ T cells). These responses were not detected by conventional tetramer staining in all but one individual. Studies that, to date, have relied on conventional tetramer staining are likely to have significantly underestimated the frequency of HCV-specific responses, especially in chronic disease. This is of some importance in understanding pathogenesis. The responses are clearly not deleted or "exhausted", but are detectable and appear to retain characteristics of "early" or immature forms. This is consistent with the idea that responses persist in vivo, but that more mature cells are recruited to the liver, where they die 18. Alternatively, these very weak and immature ("stunted") responses may also potentially result from dysfunction of antigen-presenting cells that are unable to prime or to sustain the expansion of HCV-specific CTL, through lack of T cell help or through T cell regulation.

Importantly, this technique will be readily transferable to the study of other rare populations of antigen-specific T cells, e.g. in response to self-reactive and tumor antigens, and other infections such as hepatitis B virus, HIV and influenza virus infection. These rare populations could be of some importance, as low-frequency responses may nevertheless maintain proliferative capacity (as shown here in the restimulation assays) and therefore have protective functions in vivo. Similarly, for auto-reactive and tumor-specific T cells (e.g. Melan-A-specificcells as shown here), both tissue compartmentalization and self tolerance effects will serve to reduce the frequencies of these populations in the blood, so that ultra-sensitive techniques such as that shown here might be necessary. Finally, the technique can be readily applied to the study of other lymphocyte subsets present at low frequencies as e.g., antigen-specific CD4+T helper cells by using MHC class II tetramers 19 and NKT cells by using CD1d tetramers, as recently published 20.

4 Materials and methods

4.1 Subjects

Informed consent in writing was obtained from each patient, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected in the a priori approval from the ethics committee at the John Radcliffe Hospital, Oxford, GB. Blood from three healthy HLA-A2+ individuals (HI 1–3) was used for the analysis of EBV, CMV and parvovirus responses. For the analysis of HCV-specific T cell responses, blood was obtained from nine HLA-A2 and three HLA-A1 patients with known chronic HCV infection (C-HCV 1–12), defined as the detection of HCV RNA by PCR (detection limit of 300 HCV RNA copies/ml of plasma; v2.0 Amplicor assay, Roche Diagnostics Ltd.). Blood was also obtained from three patients with resolved HCV infection (R-HCV 1–3),defined as the absence of detectable HCV RNA by PCR in the presence of HCV antibodies (third generation EIA) on at least two consecutive occasions 6 months apart. One of these patients had a spontaneously resolved HCV infection and the other two patients had a sustained virological response following treatment with an IFN-α + ribavirin combination therapy. Finally, to assess the background level of tetramer+ cell staining following magnetic selection and thus define an appropriate lower limit of detection for this assay, we recruited nine individuals as negative controls, who were assessed by use of six different HCV-specific tetramers; seven of these individuals, who were anti-HCV seronegative, were assessed with HLA-matched and -mismatched tetramers. Two HCV RNA+ individuals were assessed using HLA-mismatched tetramers only. In total, 66 assays were performed, including 39 on the HCV-exposed individuals and 27 on the negative control group.

4.2 Separation of lymphocytes and HLA tissue typing

PBMC were obtained from whole blood by density gradient centrifugation over Lymphoprep (Nycomed, Sweden), used immediately or frozen for future analysis. All subjects were HLA typed by PCR-SSCP 21 or by use of an HLA-A2-FITC antibody (One Lambda, Canoga Park, CA) and subsequent FACS analysis.

4.3 Class I peptide tetramers

Class I peptide tetramers were prepared and validated as previously described 1. Peptides were obtained from Research Genetics/Invitrogen (USA) and were as follows: HLA-A2-restricted HCV NS3 peptide 1073–1081 (CINGVCWTV), HLA-A2-restricted HCV NS3 peptide 1406–1415 (KLVALGINAV), HLA-A2-restricted HCV NS5 peptide 2592–2602 (ALYDVVTKL), HLA-A2-restricted HCV E2 peptide 614–622 (RLWHYPCTV), HLA-A2-restricted HCV NS4 peptide 1987–1995 (VLDSFKTWL), HLA-A1-restricted HCV NS3 peptide 1435–1443 (ATDALMTGY), EBV HLA-A2-restricted BMLF-1 peptide (GLCTLVAML), HLA-A2-restricted CMV pp65 peptide 495–503 (NLVPMVATV), HLA-B35-restricted parvovirus peptide (QPTRVDQKM), and HLA-A2-restricted Melan-A peptide 27–35 (ELAGIGILTV).

4.4 Tetramer and phenotypic staining

PBMC (4×106–6×106) were washed in RPMI and pelleted following separation. Tetramer (1 μl) was added, followed by a 30-min incubation at room temperature. Cells were washed in PBS, pelleted and stained with CD14-PerCP, CD19-PerCP, and ViaProbe (in order to gate out monocytes, B cells and dead cells, all of which may bind nonspecifically to the anti-PE-labeled magnetic beads), CD8-APC, and CD28- or CD27-FITC (all BD PharMingen, San Diego, CA) and left for a further 20 min at room temperature.

4.5 Magnetic enrichment of tetramer+ cells

Cells were washed in PBS and resuspended in 80 μl of cold MACS buffer (PBS, 0.5% FBS, 2 mmol EDTA). Super-paramagnetic microbeads (20 μl) conjugated with monoclonal mouse anti-PE antibodies (Miltenyi Biotec, Germany) were added to the cell pellet and incubated at 4°C for 15 min. Cells were washed in cold MACS buffer. Of the cell sample, 10% was removed for pre-enrichment FACS analysis. The remainder of the cells was positively selected by passing the cells twice through MS+ columns or an autoMACS (Miltenyi Biotec). To calculate the percentage of tetramer+/CD8+ cells, when tetramer+ cells were not detectable in the pre-enrichment analysis (i.e. in the negative control group and the HCV-exposed individuals), we used the following formula: (absolute number of tetramer+ cells in the post-enrichment sample)/(total number of CD8+ T cells in the pre-enrichment sample)×9. The factor of 9 is due to the acquisition of 10% of the stained sample as the pre-enrichment sample with the remainder (90%) of the cells being the source of the post-enrichment sample. This formula has been previously adopted for the calculation of the percentage of cytokine-secreting cells following magnetic enrichment (Miltenyi Biotec).

4.6 Peptide restimulation

PBMC were restimulated in 48-well plates (5×105 cells per well) with 10 μg/ml peptide in 1 ml RPMI 1640, 10% FCS, 100 μg/ml streptomycin, 100 U/ml penicillin and 2 mM glutamine (RPMI-10) and incubated at 37°C in a humidified CO2 (5%) incubator. On days 3–4, 500 μl RPMI-10 containing 200 U/ml rIL-2 (Chiron) was added, and further media changes with 100 U/ml rIL-2 were performed every 3–4 days. After 14 days, the percentage of tetramer-specific CD8+ T cells ws determined.

4.7 Flow cytometry

Flow cytometric analysis was performed on a FACSCalibur, and analysis was performed using CellQuest software (BD Biosciences).

4.8 Statistical analysis

Statistical analysis (GraphPad Prism 3, CA) utilized Fisher's exact test to compare responses determined by using magnetic enrichment with responses determined by conventional tetramer staining.


This work was supported by the Medical Research Council, GB (E.B. and V.O.K.), Wellcome Trust, (S.M.W. and P.K.), and the European Union (grant code QLK2-CT-1999–00356) (M.L.).


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5