Contribution of Naïve and Memory T-Cell Populations to the Human Alloimmune Response


* Corresponding author: Diana Metes,


T-cell alloimmunity plays a dominant role in allograft rejection. The precise contribution of naïve and memory T cells to this response however remains unclear. To address this question, we established an ex vivo flow-cytometric assay that simultaneously measures proliferation, precursor frequency and effector molecule (IFNγ, granzyme B/perforin) production of alloreactive T cells. By applying this assay to peripheral blood mononuclear cells from healthy volunteers, we demonstrate that the CD4+ and CD8+ populations mount similar proliferative responses and contain comparable frequencies of alloreactive precursors. Effector molecule expression, however, was significantly higher among CD8+ T cells. Analysis of sorted naïve and memory T cells showed that alloreactive precursors were equally present in both populations. The CD8+ effector and terminally differentiated effector memory subsets contained the highest proportion of granzyme B/perforin after allostimulation, suggesting that these cells present a significant threat to transplanted organs. Finally, we demonstrate that virus-specific lymphocytes contribute significantly to the alloresponse in certain responder–stimulator HLA combinations, underscoring the importance of T-cell cross-reactivity in alloimmunity. These results provide a quantitative assessment of the roles of naïve and memory T-cell subsets in the normal human alloimmune response and establish a platform for measuring T-cell alloreactivity pre- and posttransplantation.


Humans and experimental animals mount a vigorous immune response to tissues transplanted from genetically disparate individuals. This response, referred to as alloimmunity, is mediated principally by T lymphocytes that recognize nonself-histocompatibility antigens (alloantigens) including the highly polymorphic human leukocyte antigens (HLA) encoded by the major histocompatibility complex (MHC) as well as multiple minor histocompatibility antigens encoded by non-MHC genes (1).

A salient feature of T-cell alloimmunity is its extraordinary potency which is attributed to the presence of a high frequency of host T cells (1–10% of the total T-cell repertoire) that react to donor alloantigens (2–4). While immune responses to nominal antigens are dependent on a very small proportion of T cells that recognize foreign peptides presented in the context of self-MHC molecules, alloresponses are in addition generated by a much larger population of T cells that recognize peptides complexed to allogeneic MHC on donor cells—a phenomenon known as direct allorecognition (5,6). The high precursor frequency of directly alloreactive T cells is likely the result of the inherent bias of the prethymic T-cell repertoire toward MHC recognition (7), and the presence of cross-reactive T cells in the postthymic repertoire (6,8). The latter are self-MHC restricted mature T cells that are specific to microbial-derived peptides but cross-react with allogeneic MHC. Support for the cross-reactivity hypothesis is provided by molecular evidence that cytotoxic T cells specific to common viruses such as the Epstein-Barr Virus (EBV) exhibit strong alloreactivity to particular HLA molecules (9,10). Since humans are exposed to a large array of pathogens beginning very early in life, it is inferred that any given individual harbors a significant number of pathogen-specific T cells that are potentially alloreactive.

The development of ex vivo assays to identify, quantitate and analyze alloreactive T cells has been a longstanding interest of human immunologists. It is hoped that such assays will lead to better understanding of human alloimmunity and enable clinicians to predict allograft outcomes and individualize immunosuppression safely. Earlier attempts at studying the human alloimmune response have relied on assays that gauge one parameter or one cell population at a time; for example, measuring T-cell proliferation in the mixed lymphocyte reaction (MLR), IFNγ production in the enzyme-linked immunosorbent spot (ELISPOT) assay or cytotoxicity in the limiting dilution analysis (11). Although these assays have been instrumental in elucidating some of the fundamental features of human alloimmunity (12–15), several questions remain unanswered. Importantly, it is still unclear what the relative contributions of naïve and memory CD4+ and CD8+ T cells are, and which memory T cell subsets play the most prominent role in the alloimmune response.

In this manuscript, we established a reproducible, ex vivo flow cytometric multiparameter assay that allows simultaneous measurement of proliferation, precursor frequency and effector molecules of CD4+ and CD8+ T-cell subsets sorted from healthy volunteers, investigated the relative contribution of the naïve and memory T-cell pools to the alloimmune response, and directly tested whether virus-specific T cells exhibit alloreactivity.

Materials and Methods

Human subjects

Twenty-nine healthy volunteers were recruited following informed consent under an IRB-approved protocol at the University of Pittsburgh (IRB#00608014). The mean age ± SD of healthy volunteers was 39.8 ± 10.6 years and included 12 males and 17 females. In addition, nine leukocyte concentrates from healthy volunteers were purchased from the Pittsburgh Blood Bank.

Peripheral blood mononuclear cells (PBMC) isolation and cell sorting

PBMC were isolated by density gradient centrifugation (16). To obtain CD3-depleted PBMC, magnetic cell separation was performed using CD3 microbeads (Miltenyi-Biotech, Auburn, CA). For T-cell memory subset separation, CD3+ T cells were sorted by negative selection using a pan T-cell isolation kit II (Miltenyi-Biotech). These cells were further exposed to either CD45RA or CD45RO and subsequently to CD62L microbeads (Miltenyi-Biotech). Accordingly, naïve (TN-CD45ROloCD62Lhi), effector memory (TEM-CD45ROhiCD62Llo), terminally differentiated effector memory (TEMRA-CD45ROloCD62Llo) and central memory (TCM-CD45ROhiCD62Lhi) cell subsets were obtained. The purity of TN cells was >93% (<7% contamination with TEMRA cells), TEM cells >91% (<9% TCM contamination), TCM cells >89% (<11% TEM contamination) and TEMRA cells >83% (<17% TN cell contamination).

HLA typing

Molecular HLA typing for class I (A and B loci) and class II (DR locus) was done on all participants in the Tissue Typing Laboratory, University of Pittsburgh Medical Center.

Media and reagents

RPMI-1640 (Cellgro, Manassas, VA) was supplemented with 2 mM l-glutamine, 10 mM HEPES, 100 IU/mL penicillin/streptomycin (Gemini Bio-products, West Sacramento, CA) and 5% normal human AB serum (NHS) (Nabi, Boca Raton, FL). Carboxyfluorescein succinimidyl ester (CFSE) was purchased from Invitrogen (Eugene, OR) and PKH-26 kit from Sigma (St. Louis, MO). HLA-B08-restricted EBV peptide EBNA-3A (FLRGRAYGL) and HIV peptide (SLYNTVATL) were synthesized at the Peptide Synthesis Facility (University of Pittsburgh, Pittsburgh, PA). These peptides were employed at 10 μg/mL to stimulate T cells in the functional assays. PE-HLA-B08 tetramer incorporating EBNA-3A peptide was generated at the NIAID MHC-Tetramer Core Facility (Emory University, Atlanta, GA) and used to detect EBV-specific CD8+ T cells by flow cytometry. Negative tetramer was purchased from Beckman Coulter (Fullerton, CA). Tacrolimus was purchased from Sigma.

One-way MLR

CFSE-labeled (2 μM) bulk PBMC or sorted memory T-cell subsets (17) used as responders were incubated in with γ-irradiated allogeneic PKH-26-labeled (according to manufacture's protocol) stimulator PBMC (1:1 ratio) for 5 days. To ascertain allogeneic cross-reactivity to an EBV epitope, CFSE-labeled bulk PBMC from HLA-B08+ EBV+ healthy subjects were incubated with γ-irradiated allogeneic stimulator PBMC from HLA-B44+ or HLA-B44 subjects.

Flow cytometric analysis

Proliferation of allo-activated CD4+ or CD8+ T cells was identified after cell surface staining with mAbs anti-CD3-PeCy7, anti-CD8-PerCP-Cy5.5, anti-CD45RO-APC, anti-CD14-PB (BD Biosciences, San Jose, CA) and with fixable dead cell stain kit-PB (Invitrogen) and by quantifying CFSE dilution. To detect EBV-specific CD8+ T cells, staining with HLA-B08 EBNA-3A tetramer was performed prior to other cell surface staining (18). For intracellular staining, aliquots of surface stained and fixed cells were further permeabilized and incubated with antigranzyme B-Alexa700 (BD) and antiperforin-APC (Biolegend, San Diego, CA). For IFNγ-Alexa700 (BD) intracellular staining, cultured cells were re-stimulated in vitro for 4 h with either 4 μg/mL anti-CD3 (BD) and 2 μg/mL anti-CD28 (R&D, Minneapolis, MN) mAbs or with PKH26-labeled, CD3-depleted, γ-irradiated allogeneic PBMC (ratio 1:1) in the presence of Golgi-Plug (BD). Data acquisition was performed on a LSR II flow cytometer (BD) and data analysis was carried out using FlowJo (Tree Star, Ashland, OR) or ModFit LT® (Verity Software, USA).

Cytotoxicity assay

Allostimulated T cells were tested in a standard 51Cr release assay. As targets, autologous (HLA-B08+HLA-B44) and allogeneic (HLA-B08HLA-B44+ or HLA-B08HLAB44) PHA blasts were generated as previously described (9). 51Cr-labeled targets were incubated with 10 μg/mL HLA-B08 EBV EBNA3A peptide (or negative control HIV peptide) (9). Effector:target ratios (100:1, 50:1, 25:1, 12.5:1) were plated in triplicates in 96-well plates (Costar Corning Inc, Corning, NY), and 51Cr release was measured in a gamma counter (LKB, Helsinki, Finland). The percentage of specific lysis was calculated using a previously published formula (19).

Statistical analysis

Medians and means ± standard deviations (SD) were calculated and compared using the two-tail Student t-test. Results were considered statistically significant if the p-value was ≤ 0.05.


Assessment of alloantigen-induced T-cell proliferation by a flow cytometry-based MLR

Alloantigen-induced T-cell proliferation has traditionally been measured by thymidine incorporation in the MLR. A shortcoming of this method is that it does not identify which lymphocyte population or T-cell subset (CD4+ or CD8+) is proliferating or quantify the precursor frequencies of alloreactive CD4+ and CD8+ T cells unless tedious limiting dilution assays are set up in parallel. We therefore implemented a flow cytometry-based MLR method to analyze the alloreactivity of human T cells. The results of one representative experiment are shown in Figure 1 to illustrate the analysis method. Gating strategy is shown in Figure 1A, while the CFSE dilution of the CD4+ and CD8+ T-cell subsets was analyzed separately to measure their proliferation. Results are depicted either as dot plots (Figure 1B) or histograms (Figure 1C). As shown in Figure 1C, both the parent nonproliferating CD4+ and CD8+ T cells as well as 11 generations of dividing cells could be identified using ModFit LT® software. The histogram plots consistently showed that the bulk of proliferated CD4+ and CD8+ T cells clustered in peaks representing, on the average, cells that had divided between 4 and 10 times.

Figure 1.

Identification and quantification of T-cell proliferation by flow cytometry-based one-way MLR. (A) Responder T cells were identified at the end of the 5-day MLR by gating on CD3+ cells followed by exclusion of dead cells and PKH26+ stimulator cells. (B) Proliferation of the CD4+ and CD8+ populations was then quantified by CFSE dilution (% CFSElow cells) (C) and by histogram plots of generations of divided cells as determined by ModFit-LT® software. (D) The specificity of T-cell proliferation in this assay was determined by measuring CFSE dilution after co-culture with either autologous or allogeneic PBMC in the presence or absence of tacrolimus (TAC) (100 ng/mL). Results shown are one example of 41 independent responder–stimulator pairs. Suppression of T-cell proliferation by tacrolimus was demonstrated in 10 independent pairs. The absence of significant proliferative response to autologous cells was confirmed in 24 independent pairs.

To test the specificity of the assay, CFSE-labeled responder PBMC were co-cultured with either autologous PBMC or with allogeneic PBMC in the presence of the T-cell immunosuppressive drug tacrolimus (TAC). The results depicted in Figure 1D show minimal proliferation in response to autologous cells and significant inhibition of both CD4+ and CD8+ T-cell proliferation in response to allostimulation if TAC (100 ng/mL) was added to the MLR. These data indicate that proliferation measured by flow cytometry-based MLR is both alloantigen and T-cell dependent.

CD4+ and CD8+ T cells proliferate at comparable levels in response to allostimulation and contain similar alloreactive precursor frequencies

To characterize and quantify the normal human T-cell alloimmune response, we obtained PBMC from 29 HLA-typed, healthy, nontransplanted volunteers and performed flow cytometry-based MLR on 41 different responder–stimulator pairs. In each case, the proportion (%) of either CD4+ or CD8+ T cells that had undergone at least one division (second-generation cells or greater) was measured. The precision of this measurement was determined by performing five replicate assays on four responder–stimulator pairs (data not shown) that yielded a low mean coefficient of variability of 14% (range = 6–23%), indicating that the assay is precise. As summarized in Figure 2A, overall CD4+ and CD8+ T-cell subsets proliferated equally in response to allostimulation. CD4+ T cells that had divided ranged between 4% and 63% of total CD4+ T-cells (mean ± SD = 19 ± 11%; median = 17%), and CD8+ T cell proliferation ranged between 2% and 88% of total CD8+ T cells (18 ± 15%, 15%). We further calculated the precursor frequency of alloreactive T cells present in the CD4+ and CD8+ subset using ModFit LT®. This software generates histograms based on CFSE intensity (Figure 1C) and applies deconvolution algorithms to calculate the precursor frequency of cells that divided, reported as% of parent population. As shown in Figure 2B, the overall CD4+ and CD8+ subsets contained comparable frequencies of alloreactive T cells: 3.9 ± 3.5% (range = 0.9–20.4%; median = 3%) and 2.5 ± 3.7% (range = 0.1–25%; 1.6%), respectively. These data underscore the heterogeneity of the human alloimmune response and are consistent with alloreactive T-cell precursor frequencies reported by others (2–4). In addition, our results clearly indicate that CD4+ and CD8+ populations proliferate equally in response to allostimulation.

Figure 2.

Alloantigen-induced proliferative response and alloreactive precursor frequencies of human T cells. (A) T-cell proliferation in response to allogeneic PBMC was measured in the one-way MLR by CFSE dilution, and the proportion of CD4+ and CD8+ T cells that had proliferated was calculated as % of total CD4+ or CD8+ T cells, respectively. (B) Alloreactive precursor frequencies present in the CD4+ and CD8+ populations were calculated by ModFit LT® software based on histogram plots similar to that shown in Figure 1C. Horizontal lines are of mean values (n = 41 responder–stimulator pairs).

We then analyzed the effect of responder–stimulator HLA disparities on CD4+ and CD8+ T-cell alloresponses. All responder–stimulator pairs had at least 3 antigen mismatches except for one pair that had one antigen mismatch restricted to the HLA-DR (class II) locus and another that had one class I and one class II HLA mismatches. We also analyzed the response of any given individual to multiple HLA-mismatched stimulators. The overall results suggest that cumulative differences in HLA mismatches do not influence T-cell proliferation in the one-way MLR when at least one HLA-DR mismatch is present and underscore the importance of the ‘quality’, rather than ‘quantity’, of the HLA mismatch in defining the strength of the alloimmune response (Supporting Figure S1A and B).

Assessment of type-1 T-cell effector molecule expression using the flow cytometry-based MLR

Unlike traditional immunological assays that measure one parameter at a time, modern flow cytometry provides the opportunity to measure multiple parameters simultaneously. Therefore, we employed the flow cytometry-based MLR method to investigate whether T cells that proliferated in response to alloantigens also acquired effector molecules necessary for their function. We chose to measure IFNγ, granzyme B and perforin because these molecules were showed to contribute to allograft destruction (15,20). A representative experiment that illustrates IFNγ measurement relative to cell proliferation (CFSE dilution) is shown in Figure 3A for both CD4+ and CD8+ T cells. Allogeneic re-stimulation induced IFNγ production only in those responder cells that had already undergone multiple proliferation cycles (Figure 3A), indicating that they had differentiated into effector T lymphocytes. Cumulative analysis of the data revealed that the CD8+ T-cell subset had a significantly higher proportion of IFNγ-producing cells than the CD4+ subset following either allospecific or polyclonal re-stimulation (Figure 3B). Moreover, IFNγ production did not correlate with the magnitude of proliferation of either CD4+ or CD8+ T cells (Figure 3C). The latter result indicates that measuring proliferation alone, as has been traditionally done in MLR assays, does not guarantee accurate inferences about T-cell effector functions.

Figure 3.

IFNγ production by alloantigen-activated T cells. (A) IFNγ-producing CD4+ and CD8+ T cells were quantified by flow cytometry at the end of the 5-day one-way MLR after 4 h of restimulation with anti-CD3 plus anti-CD28, CD3-depleted allogeneic PBMC or media alone. Number displayed in the left upper quadrant of each dot plot represents the proportion (%) of proliferating (CFSElow) CD4+ or CD8+ T cells that produced IFNγ. Results are of one representative experiment. (B) Proportion of IFNγ-producing cells, measured as in (A), is greater in the CD8+ than that in CD4+ T-cell population (n = 11). (C) Lack of correlation between cell proliferation and IFNγ production in either the CD4+ or CD8+ T-cell population. Each symbol represents a single responder–stimulator pair (n = 11).

Intracellular granzyme B and perforin were measured in responder CD4+ and CD8+ T cells at the end of the 5-day MLR without further re-stimulation (Figure 4A). As with IFNγ, only T cells that had proliferated in the MLR expressed granzyme B and/or perforin. Cumulative analysis of the data demonstrated that the proportion of cells that acquired either granzyme B or both granzyme B and perforin expression was significantly higher among CD8+ than CD4+ T cells (Figure 4B). We also observed an inverse correlation between cell proliferation and granzyme B/perforin double expression in the CD8+, but not in the CD4+ T-cell subset (Figure 4C). These results further underscore the advantage of the flow cytometry-based MLR over other assays in that it simultaneously measures proliferation and functional attributes of alloreactive T cells. They also clearly illustrate that a strong proliferative T-cell response in the MLR does not necessarily imply a potent effector response. This may explain why MLR methods that only measure proliferation have failed to correlate with or predict clinical outcomes in transplant recipients (21–23).

Figure 4.

Granzyme B and perforin expression by alloantigen-activated T cells. (A) Granzyme B and/or perforin-expressing CD4+ and CD8+ T cells were quantified by flow cytometry at the end of the 5-day one-way MLR. Granzyme B and perforin expression are shown for both undivided (CFSEhigh) and proliferating (CFSElow) cells. Numbers displayed in the quadrants of each dot plot represent the proportion (%) of CD4+ or CD8+ T cells expressing granzyme B, perforin, both or neither. (B) Overall % of granzyme B, perforin or both granzyme B and perforin-expressing cells (n = 14). (C) Correlation between cell proliferation and granzyme B/perforin double expression in the CD4+and CD8+ T-cell population. Each symbol represents a single responder–stimulator pair (n = 14).

Relative contribution of naïve and memory T-cell populations to the normal human alloimmune response

Limiting dilution assays performed more than 15 years ago provided the first evidence that alloreactive lymphocytes are not restricted to the naïve T cell pool but include an equal number of memory T cells (12,13). Since these publications appeared, newer markers that differentiate naïve from memory T cells have been developed, and evidence that distinct memory T-cell subpopulations exist in humans has emerged: the TCM subset lacks immediate effector function and homes to secondary lymphoid organs where they proliferate and differentiate into effectors upon proper antigenic re-challenge. In contrast, TEM cells are found mostly in the peripheral inflamed tissues and readily secrete effector molecules in response to antigeneic stimulation. The TEMRA subset has similar functional properties as TEM; however, they are more potent effectors and are more susceptible to death (24,25). We therefore purified CD3+ cells from leukapheresis products of nine healthy subjects and separated them into TN, TCM, TEM and TEMRA (24,25). The phenotype and frequency of each subset are shown in Figure 5A. The T-cell subsets were then tested individually in the flow cytometry-based MLR. The results shown in Table 1 indicate that alloreactive T-cell precursors are present in all four populations with the naïve and memory subsets contributing equally. While TN and TCM cells proliferated the most in response to allostimulation, the TEM and TEMRA subpopulations trended toward higher proportion of IFNγ, and significant higher granzyme B/perforin expressing CD8+ T cells (Figure 5B). CD4+ T cell proliferation was comparable to that of CD8+ T cells, but their lytic molecules expression was significantly lower (Figure 5B). These results indicate that healthy subjects harbor alloreactive T cells within both the naïve and memory lymphocyte pools, with TEM and TEMRA subpopulations containing alloreactive T cells that readily express IFNγ and lytic molecules upon allostimulation.

Figure 5.

Contribution of naïve versus memory T-cell subsets to human alloreactivity. (A) Naïve and memory T-cell populations present in the peripheral blood of nine healthy subjects were identified and quantitated (% of total T cells) by flow cytometry. TN= naïve; TEM= effector memory; TCM= central memory; and TEMRA = terminally differentiated effector memory. (B) Proliferation, IFNγ production and granzyme B/perforin double expression of CD4+ and CD8+ purified naïve and memory T-cell subsets were determined in the 5-day one-way MLR. IFNγ data shown is after 4 h restimulation with anti-CD3 plus anti-CD28 mAbs. *One-tail Student's t-test analysis.

Table 1.  Precursor frequency of alloreactive T cells in the naïve and memory populations
 Precursor frequency (%)
  1. AMean ± SD; TEM= effector memory; TCM= central memory; TEMRA = terminally differentiated effector memory; TN= naïve.

TEM4.6 ± 1.4 (n = 7)3.1 ± 1.5 (n = 8)
TCM4.9 ± 2.2 (n = 5)5.1 ± 2.3 (n = 7)
TEMRA4.4 ± 3.0 (n = 2)3.2 ± 1.1 (n = 4)
TN3.7 ± 2.6 (n = 6)4.0 ± 1.9 (n = 6)

Contribution of virus-specific CD8+ T cells to the human T-cell alloimmune response

The presence of IFNγ- and granzyme B/perforin-producing alloreactive CD8+ T cells in the memory pools of all healthy subjects studied points to the possibility that these are cross-reactive T cells generated by prior exposure to common pathogens. To test directly the contribution of virus-specific CD8+ T cells to the human alloimmune response, we took advantage of tetramer technology to track Ag-specific CD8+ T cells in the flow cytometry-based MLR assay. Burrows et al. had demonstrated that HLA-B08+/EBV+ adults express T-cell receptors specific to the EBV-epitope EBNA-3A that cross-react with HLA-B44 (9). Accordingly, we investigated the response of PBMC from two HLA-B08+/EBV+ healthy controls against irradiated PBMC from HLA-B44+ mismatched individuals or HLA-B44 allogeneic controls. EBNA-3A tetramer+ cells constituted 1.8% of circulating CD8+ T cells (96% displaying memory phenotype) at baseline (Figure 6A). Their frequency increased dramatically to 39% 7 days after allostimulation with HLA-B44+ PBMC but remained at 1.6% when allostimulated with HLA-B44 PBMC. Importantly, EBNA-3A tetramer+ T cells represented approximately 70% of CD8+ T cells that had divided in response to HLA-B44+ allogeneic cells. Interestingly, a significant proportion of these cross-reactive EBNA-3A tetramer+ CD8+ T cells produced IFNγ in response to the anti-CD3/anti-CD28 polyclonal stimulation (data not shown) and to the EBV-peptide (Figure 6B), but failed to do so in response to HLA-B44+ allostimulation (Figure 6B). These cells also expressed granzyme B and perforin (Figure 6C), and were cytotoxic against HLA-B08+ PHA blasts that carry the EBNA-3A peptide and against allogeneic PHA blasts that share HLA-B44 with the stimulator PBMC, but not against HLA-B08HLA-B44 PHA blasts (Figure 6D). Similar EBV-specific CD8+ T cell proliferation, cytotoxicity and IFNγ expression were observed in a second HLA-B08+/EBV+ subject (Supporting Figure S2). These results demonstrate that virus-specific T cells contribute to the human alloresponse if the appropriate responder–stimulator HLA combination is present. The response to alloantigens by these cross-reactive CD8+ T cells, however, may be qualitatively different from their response to the viral epitope when cytotoxicity and IFNγ production are measured.

Figure 6.

EBV-specific CD8+ T cells cross-react with human alloantigens. (A) Frequency of EBV-specific (EBNA-3A tetramer+) CD8+ T cells in PBMC of an HLA-B08+ EBV+ individual before (Day 0) and after (Day 7) allostimulation with HLA-B44+ PBMC in the one-way MLR. Control MLR (Day 7, lower panels) using HLA-B44 allostimulators is also shown. (B) Virus-specific (EBNA-3A tetramer+) CD8+ T cells that divided in response to HLA-B44+ allostimulation produce IFNγ when rechallenged with viral peptide (EBNA-3A) but not with CD3-depleted HLA-B44+ allostimulators or an irrelevant viral peptide (HIV) (C) and express granzyme B/perforin. (D) HLA-B08+ T cells allostimulated with HLA-B44+ cells lyse syngeneic targets loaded with EBV EBNA-3A peptide or HLA-B44+ allogeneic targets from three separate individuals, but do not lyse HLA-B44 targets. Cytotoxicity was measured by the 51Cr-release assay. Results are shown as % of lysis at 50:1 effectors:target ratio.


T-cell alloimmunity is considered one of the dominant factors mediating graft injury after transplantation. However, the precise contribution of naïve versus memory T-cell subsets to human alloimmunity has remained elusive. Here we established an accurate and specific flow cytometry-based ex vivo assay that measures simultaneously proliferation, precursor frequencies and effector molecule (IFNγ and granzyme B/perforin) expression of alloreactive CD4+ and CD8+ T cells present in the peripheral blood of human subjects. We then utilized this assay to assess the contributions of naïve and memory T-cell subsets to the alloimmune response.

Our results provide new insights into the human T-cell alloimmune response. First, we found that CD4+ and CD8+ T cells proliferate equally in response to allostimulation and contain comparable alloreactive precursor frequencies determined by CFSE dilution (approximately 4%), not detected previously by thymidine incorporation MLR. In addition, allostimulated CD4+ T cells acquired effector mediators, including lytic molecules, albeit at lower levels than CD8+ T cells. Prior attempts at quantitating alloreactive CD4+ T-cell precursors by limiting dilution assays and IL-2 measurement resulted in much lower frequencies (<0.2%) (26,27), suggesting that alloreactivity of the CD4+ compartment may have been underestimated. Second, our results extended the observation made more than 15 years ago that the human alloresponse is not ‘truly primary’ but is a response mediated by both naïve and memory T cells (12,13). By applying the flow cytometry-based MLR method to cells sorted from nine different healthy subjects, we established that CD4+ and CD8+ T cells that proliferate and produce IFNγ and granzyme B/perforin upon allostimulation reside in the TN, TCM, TEM and TEMRA populations. The memory populations contained comparable alloreactive CD4+ and CD8+ T-cell precursor frequency as the naïve population. Importantly, a trend toward higher proportion of IFNγ and significant higher proportion of granzyme B/perforin-expressing T cells was present in the TEM and TEMRA populations, underscoring the potential threat these T cells may pose to transplanted organs. The latter concept is consistent with prior evidence that heightened lytic molecule expression in the urine of renal transplant recipients correlates with acute cellular rejection episodes (20). Third, by analyzing the response of EBV-specific CD8+ T cells to allogeneic stimulators, we validated that the presence of alloreactive T cells in the memory populations of healthy subjects not previously exposed to alloantigens is likely the result of TCR cross-reactivity. This finding is congruent with previous molecular analyses demonstrating the alloreactivity of virus-specific human T cells (9,10). However, our results indicate subtle qualitative differences between the response of these T cells to their cognate viral antigen and their response to the allogeneic MHC molecule with which they cross-react. In an effort to detect high-risk candidates on the waiting list and to predict at the time of patient's enrolment the likelihood of the allograft to suffer major immunologic injuries, screening for pretransplant donor-specific memory T cells has been performed by measuring IFNγ responses in the ELISPOT assay (28). This approach was reported to identify candidates with increased risk for acute rejection and poor late outcomes (28–30). As this approach still awaits clinical validation, our results suggest the importance of a multiparameter monitoring, namely granzyme B and perforin, proliferation and precursor frequencies in addition to IFNγ, as screening tools for detecting T-cell alloreactivity and, therefore, as a strategy to optimize pretransplant risk assessment.

Flow cytometry-based MLR assays have been previously employed to measure alloimmune responses in transplanted patients (22,23,31). Tanaka et al. quantified CD4+ and CD8+ T cell proliferation in living-donor liver transplant recipients by this method in response to either donor or third-party allogeneic cells, but no significant correlation between proliferation and acute rejection could be found (22). Precursor frequencies and cytokine or lytic molecule expression were not measured in this study. Using a similar assay, Kreijveld et al. calculated CD4+ and CD8+ alloreactive T-cell precursor frequency in renal transplant recipients who had been withdrawn from tacrolimus therapy. They found that frequencies were stable over time and did not correlate with the occurrence of acute rejection. IFNγ and granzyme B/perforin expression by alloreactive T cells were not determined (23). These reports, therefore, suggest that measuring T-cell proliferation alone does not provide an adequate reflection of human alloimmunity. Our analysis of PBMC from healthy subjects provided direct proof that proliferation does not correlate with the functional potential of alloreactive CD4+ or CD8+ T cells (Figures 3C and 4C). In fact, we found an inverse relation between granzyme B/perforin production and CD8+ T cell proliferation (Figure 4C). Therefore, we propose that simultaneous monitoring of donor-specific alloreactivity using multiple functional parameters is a superior approach to single parameter measurements and may allow an accurate prediction of graft outcomes posttransplantation. In addition, this multiparameter analysis could be slightly modified to use allogeneic PBMC lysates instead of allogeneic irradiated PBMC stimulation to also quantify the indirect pathway of allorecognition.

In summary, the data presented in this manuscript provide direct evidence for the importance of memory T cells in the normal human alloimmune response. Specifically, we have identified the CD8+ TEM and TEMRA subsets as likely threats to transplanted organs because of their heightened expression of effector molecules upon stimulation. Monitoring these T-cell subsets in pre- and postsolid organ or bone marrow transplantation may prove useful for predicting adverse immunological events such as allograft rejection and graft-versus-host disease.


We thank Dr Igor Dvorchik, PhD, and Dr Hongmei Shen, PhD, for their valuable advice with the establishment and validation of the flow cytometry-based MLR assay. We also acknowledge the NIH Tetramer Facility (Emory University, Atlanta, GA) for generating the PE-HLA-B08 EBNA-3A tetramer. This work was supported in part by NIH grant AI049466 (FGL).