Adaptive immunity is mediated by a complex network of cellular and molecular interactions that sense and respond to antigenic stimuli derived from dangerous entities. To deconvolute this system, it is important that the necessary tools are available to enable the accurate and reproducible measurement of antigen-specific cell populations directly ex vivo; however, it is equally important that these tools are used with an understanding of their limitations. One of the most significant advances in immunotechnology over the past few years has been the development of soluble recombinant peptide-major histocompatibility complex class I (pMHCI) multimers (1, 2). These reagents bind stably to cognate T cell receptors (TCRs) expressed on the surface of antigen-specific CD8+ T cells, despite the low affinity and rapid kinetics of monomeric TCR/pMHCI binding, and are internalized at physiological binding temperatures (3, 4); thus, in fluorochrome-conjugated form, pMHCI multimers allow the visualization of cognate CD8+ T cells by flow cytometry. In general, pathogen-specific CD8+ T cell populations, which tend to express TCRs that bind cognate pMHCI with high affinities (5) and comprise a substantial proportion of the memory T cell pool, are easily identified with relatively basic flow cytometers and associated software. However, it is sometimes difficult to distinguish true antigen-specific CD8+ T cells from background binding events, especially when the target population is present at low frequency and binds cognate pMHCI with low avidity. Resolution of these issues is critical for the analysis and interpretation of pMHCI multimer-based data. In this brief review, which is necessarily selective in scope, we highlight a few techniques that might help to improve the ex vivo detection of true pMHCI multimer binding events.
The ability to quantify and characterize antigen-specific CD8+ T cells irrespective of functional readouts using fluorochrome-conjugated peptide-major histocompatibility complex class I (pMHCI) tetramers in conjunction with flow cytometry has transformed our understanding of cellular immune responses over the past decade. In the case of prevalent CD8+ T cell populations that engage cognate pMHCI tetramers with high avidities, direct ex vivo identification and subsequent data interpretation is relatively straightforward. However, the accurate identification of low frequency antigen-specific CD8+ T cell populations can be complicated, especially in situations where T cell receptor-mediated tetramer binding occurs at low avidities. Here, we highlight a few simple techniques that can be employed to improve the visual resolution, and hence the accurate quantification, of tetramer binding CD8+ T cell populations by flow cytometry. These methodological modifications enhance signal intensity, especially in the case of specific CD8+ T cell populations that bind cognate antigen with low avidities, minimize background noise, and enable improved discrimination of true pMHCI tetramer binding events from nonspecific uptake. © 2008 International Society for Advancement of Cytometry
WHAT IS REAL?
A priori, the simplest way to confirm both the presence and magnitude of a specific CD8+ T cell population detected by pMHCI multimer staining is to demonstrate and measure the functional consequences of cognate antigen engagement at the single cell level; this can be achieved relatively easily with a variety of readouts using flow cytometric platforms (Figure 1). However, there are a number of issues that complicate the use of functional verification as a “gold standard” test.
Antigen-experienced CD8+ T cells are primed to execute a programmed array of effector functions when triggered by cognate pMHCI molecules; nevertheless, not all cells and not all functions are equivalent. Thus, it is possible that a truly cognate CD8+ T cell might not be functional in a given ex vivo bioassay (Figure 1). Indeed, many studies have now shown that pMHCI multimer binding CD8+ T cells can exhibit various degrees of dysfunction, for example due to the effects of persistent antigen exposure [Ref. (6) and citations therein]. Furthermore, such discrepancies between function and physical presence may vary with circumstance; thus, distinct changes in functional capabilities can accompany both generic T cell differentiation (7, 8) and the unique ontogeny of individual antigen-specific T cells (9), thereby giving rise to a vast heterogeneity of phenotypic characteristics and functional profiles within the memory T cell pool. Similarly, certain functions are triggered at different thresholds and thus can be differentially elicited in CD8+ T cells with different antigen sensitivities (10–12).
A further complicating issue arises from the observation that CD8+ T cells bearing cognate TCRs can exhibit functional activation in response to agonist ligands, yet fail to bind the corresponding pMHCI multimer (13–15). Recent data indicate that these discrepancies can arise as a consequence of the differential TCR/pMHCI affinity thresholds and kinetic rules that govern pMHCI multimer binding and functional activation (15). Thus, pMHCI multimer binding is dependent on both the expression of cognate TCRs that bind monomeric pMHCI above a certain affinity and a corresponding compound cellular avidity for antigen that is determined by multiple additional factors (16). In contrast, pMHCI antigens with monomeric affinities for TCR that lie below the threshold required for binding of the corresponding multimer can act as agonist ligands in functional assays (15).
Thus, even with the assumption that an optimal antigenic stimulus is delivered with current protocols, functional measurements, either individually or in combination, are insufficient in isolation to determine the integrity of pMHCI multimer binding. How else, then, can we provide a “reality check” to discriminate signal from noise in pMHCI multimer-based assays? The purpose of what follows is to provide some potential solutions to these issues.
REAGENT PREPARATION, ASSAY CONDITIONS, AND INTERFACING WITH THE FLOW CYTOMETRIC PLATFORM
Many factors can influence the reliability and reproducibility of pMHCI multimer-based assays. A complete discussion of these factors is beyond the scope of this article; however, we have recently reviewed many of these issues separately (16) and optimization of some related parameters is also considered in a companion article by Keeney and colleagues in this issue of Cytometry. For the purposes of this review, we will assume that: (i) all recombinant pMHCI reagents are produced to a high degree of purity and fully tetramerized, albeit with an integral degree of heterogeneity (17), by conjugation to high quality, bright fluorochrome-labeled streptavidin preparations; (ii) all stains are performed at physiological binding temperature (37°C) for a maximum of 20 min as described previously (3) unless stated otherwise; (iii) all reagents are optimally titrated to ensure maximal discrimination between signal and noise; (iv) all flow cytometry-related variables are optimally configured; and, (v) sufficient numbers of events are collected to allow accurate data interpretation (18). It should be noted that the techniques discussed below are applicable to pMHCI multimers of different valencies; the commonly used tetrameric scaffold is considered here for simplicity.
As the frequency of antigen-specific CD8+ T cells decreases, the true pMHCI tetramer signal becomes increasingly obscured within the inherent background noise that assumes greater relative proportionality within the system; this can lead to erroneous measurements of quantity and quality. Thus, it is essential to minimize noise through the elimination of aberrant binding events from the analysis. This can be achieved through the incorporation of a “dump” channel, in which specific stains are used to identify confounding events that are subsequently gated out en masse (Figure 2). A viability marker should always be included in the dump channel because dead/dying cells with impaired membrane integrity are a major source of nonspecific binding events; the fixable amine reactive dyes are particularly useful for this purpose as they are compatible with fixation/permeabilization procedures (19). Similarly, it is generally useful to exclude monocytes and B cells with αCD14 and αCD19 mAbs, respectively (20, 21). Multiple additional markers can also be included within a single dump channel as circumstances dictate; for example, the addition of αCD33 and αCD34 might eliminate further nonspecific binding events in bone marrow mononuclear cell samples. Overall, the use of a dump channel increases measurement sensitivity by reducing compensation-induced spreading error and eliminating irrelevant cells that artifactually masquerade as true signal with traditional gating strategies (Figure 2).
Even once aberrant binding events have been eliminated, a degree of background staining frequently persists within the CD8+ T cell population that can be difficult to resolve. A visual clue to the extent of the problem can be attained from simple inspection of the corresponding CD3+CD8− cell population; thus, if a similar degree of pMHCI tetramer staining is observed, then it is less likely that the signal emanating from the CD3+CD8+ cell population is real. Indeed, this information should always be disclosed in the presentation of pMHCI tetramer-based data. However, such observations do not necessarily exclude the presence of cognate CD8+ T cells; rather, they provide an indication of the general level of noise from which any true binding events must be extracted. Assuming that such nonspecific staining is, in part, random, one empirical approach to this task is to use the same pMHCI tetrameric reagent labeled in different colors (Figure 3). The principle here is that cognate CD8+ T cells will bind specifically regardless of the label, whereas the random element of the staining pattern will not associate equally with each version of the reagent. Thus, although there is inevitably a concomitant reduction in the intensity of specific labeling with each individual pMHCI tetramer, a degree of confidence is added by the observation of dual fluorochrome uptake. This strategy can be especially useful in the case of infrequent CD8+ T cells that are not easily amenable to functional verification and exhibit poor separation as a consequence of low antigen binding avidities (22, 23).
In addition, however, there is often an intrinsic element to background staining that can vary idiosyncratically with different pMHCI tetramers in different individuals. To some extent at least, this is a CD8 coreceptor-mediated effect (24). Consequently, tetramers produced from pMHCI mutants with impaired CD8 binding properties exhibit diminished background staining and can thereby lead to enhanced signal resolution (25–27). In general, this has proven to be an effective strategy for pathogen-specific CD8+ T cells that typically express high affinity TCRs and bind antigen with high avidities (5, 28). However, CD8 stabilizes and enhances TCR/pMHCI interactions at the cell surface through effects on both the off-rate and the on-rate (15, 24, 29), despite the fact that these trimolecular interactions are structurally distinct and noncooperative in three dimensions (30–32). Therefore, modification of CD8 coreceptor binding also alters the avidity range of antigen-specific CD8+ T cells that can bind pMHCI tetramers; given that avidity for antigen is at least partially dependent on the corresponding TCR/pMHCI monomeric affinity, intact or even enhanced CD8 binding properties (see below) are likely to be required for optimal pMHCI tetramer-based detection of tumor-specific and autoreactive CD8+ T cell populations (5, 15). How, then, can we distinguish background staining from signal without compromising the detection of low avidity CD8+ T cells? Perhaps the most useful technique here is to include phenotypic markers in the flow cytometric panel. The “shoulder” that protrudes from the noncognate CD8+ T cell population and represents background staining comprises a mixture of both memory and naïve cells; i.e. the phenotypic distribution follows that of the CD8+ T cell population as a whole. In contrast, true cognate CD8+ T cells that bind pMHCI tetramer generally exhibit an almost exclusive memory phenotype (Figure 4). It should be noted that there are some exceptions to this general dichotomy; for example, CD8+ T cell populations specific for Melan-A/MART-1 can contain substantial numbers of naïve cells (33). Nevertheless, “memory gating” can help to distinguish true binding events from background staining in most experimental settings.
The intensity of labeling with pMHCI tetramers reflects, at least in part, cellular avidity for antigen (34, 35). This parameter is determined both by TCR/pMHCI affinity (15) and by factors beyond the direct interface with antigen, such as membrane flexibility, molecular density, and cell surface topography [reviewed in Ref. (16)]. Although the clonotypic structure of a given antigen-specific CD8+ T cell population is fixed at the time of analysis, pMHCI tetramer binding to that population can be altered via manipulations that affect either cellular behavior or the structure of the reagent; these technical modifications, which primarily impact the detection of low avidity CD8+ T cells, are not necessarily mutually exclusive and can exhibit useful compound effects in certain situations.
The extent of ligand binding is dependent on the concentration of both partners. For this reason, the maximum concentration of pMHCI tetramer that allows clear visualization with minimal background issues should be selected in reagent titrations for optimal experimental usage; this will facilitate the capture of cognate pMHCI tetramers by lower avidity CD8+ T cells (see below). It is also possible to alter the surface densities of TCR and CD8 by pre-treating cells with the protein tyrosine kinase inhibitor dasatinib (36). This manipulation can substantially enhance the intensity of pMHCI tetramer staining and preferentially benefits the detection of low avidity CD8+ T cells (reference 37 and Figure 5). These phenomena are not unexpected given the effects of CD8 on membrane-constrained TCR/pMHCI interactions and the impact of TCR density on the rate at which the second and third pMHCI components of the tetramer bind to form a stable complex with the cell [reviewed in Ref. (16)]. However, it is surprising that dasatinib can exert these beneficial effects on tetramer binding with incubation times shorter than those required to observe any upregulation of TCR or CD8 on the cell surface; preliminary experiments suggest that the underlying mechanism relates to the prevention of “empty” TCR internalization during pMHCI binding, but other explanations remain possible (37). Regardless, substantial signal enhancement with minimal background effects can be observed with this manipulation, especially in the context of low avidity CD8+ T cell populations. Furthermore, it appears from preliminary observations that dasatinib can reduce the degree of cell death induced by pMHCI tetramers during the course of the staining procedure (Ref. 37 and Figure 5). Such effects may be conducive to the viable sorting of CD8+ T cells labeled with cognate pMHCI tetramers, especially as the effects of dasatinib are reversible (36).
As discussed above, the CD8 coreceptor profoundly affects TCR/pMHCI binding at the cell surface by slowing the dissociation rate and enhancing the association rate (15, 24, 29). These effects influence the range of pMHCI ligands that can be bound in tetrameric form by a given CD8+ T cell. Thus, in the absence of CD8 binding, the monomeric TCR/pMHCI affinity threshold required to observe tetramer binding is substantially increased (15), thereby enabling the selective detection of high avidity antigen-specific CD8+ T cell populations with tetrameric forms of α3 domain-mutated (D227K/T228A) pMHCI molecules that retain TCR binding integrity but fail to engage CD8 (28, 38, 39). Recently, we have constructed pMHCI molecules that bind CD8 with enhanced affinity (24, 40, 41). The Q115E α2 domain mutation increases CD8 binding affinity from KD ≈ 120 μM to KD ≈ 85 μM (24) and extends the range of pMHCI ligands that can be bound in tetrameric form by a given CD8+ T cell; these effects enable the detection of low avidity antigen-specific CD8+ T cells with “coreceptor-enhanced” pMHCI tetramers (42). As discussed above, enhanced CD8 binding can be associated with concomitant increases in background staining; these can be minimized with appropriate reagent titration and resolved by the inclusion of phenotypic markers as shown in Figure 4 (42). However, while background staining is to some extent a CD8-mediated phenomenon, there are clearly additional factors that affect its penetrance in practice; thus, these effects appear to vary between individuals according to the nature of the antigen (42) and are not always a problematic accompaniment to the use of coreceptor-enhanced tetramers (Figure 6). In general, then, coreceptor-enhanced pMHCI tetramers can be usefully employed for signal amplification purposes and seem destined to find widespread applications in fields such as tumor immunology and autoimmunity where low avidity CD8+ T cell populations might prevail.
As immunotechnology continues to advance with the development of multiplex applications for combined ex vivo measurements (21), it is increasingly important that signal and noise are accurately discriminated with basic reagents such as pMHCI tetramers in order to obtain reliable data. In this brief review, we have highlighted a few techniques that might contribute to the accurate detection of true antigen-specific CD8+ T cell populations with pMHCI tetramers directly ex vivo. These modifications to the standard assay approach are neither mutually exclusive nor definitive, but provide some options that can be tested in different scenarios to improve resolution.
DAP is a Medical Research Council (UK) Senior Clinical Fellow. LW is a Wellcome Trust Clinical Intermediate Fellow.