Abstract
CMV reactivation is a major complication after allogeneic stem cell transplantation (SCT). Immune reconstitution of CMV-specific CTLs (CMV-CTLs) is essential for virus control. During CMV-CTL monitoring using mutated HLA/CMV tetramers selectively detecting high-avidity T cells, we observed coappearance of CMV-CTLs with low (CMV tetlow CTLs) and high tetramer binding (CMV tethigh CTLs) in 53/115 CMV IgG+ patients stem cell transplanted from CMV IgG+ donors. However, the relevance of these coappearing differentially tetramer binding (“dual”) CMV-CTLs was unclear. In this study, we investigated the kinetics, properties, and clinical impact of coappearing CMV tetlow and tethigh CTLs after allogeneic SCT. Patients with dual CMV-CTLs had more CMV tethigh than tetlow CTLs. Chimerism analysis of isolated CMV tetlow and tethigh CTLs revealed their exclusive donor origin. CMV tetlow and tethigh CTLs had an identical effector memory CD45RA−CCR7− and CD45RA+CCR7− T cell distribution, equal differentiation, senescence, and exhaustion marker expression and were negative for regulatory CD8+ T cell markers. Isolated CMV tetlow and tethigh CTLs were equally sensitive to CMV peptides in IFN-γ release and cytotoxicity assays. However, CMV tethigh CTLs proliferated more in response to low CMV peptide concentrations than tetlow CTLs. TCR repertoire analysis revealed that CMV tetlow and tethigh CTLs use different TCRs. Finally, dual CMV-CTLs were not associated with CMV antigenemia. In conclusion, these data show for the first time, to our knowledge, that both CMV tetlow and tethigh CTLs are functional effector T cells differing by proliferation, numbers in peripheral blood, and probably by their precursors without increasing the CMV reactivation risk after allogeneic SCT.
Introduction
Cytomegalovirus reactivation is an important complication after allogeneic stem cell transplantation (SCT), leading to considerable morbidity and, if untreated, mortality. Reactivation of the latent virus can lead to severe CMV disease including CMV colitis, retinitis, and pneumonia (1–3). CMV seropositivity of the recipient, seronegativity of the stem cell donor, T cell depletion of the stem cell graft, and graft-versus-host disease (GvHD) are major risk factors for CMV reactivation (3, 4). Monitoring of patients for CMV antigenemia together with preemptive treatment with antiviral drugs has dramatically improved the outcome of patients reactivating CMV. However, particularly recurrent CMV reactivations still lead to considerably prolonged hospitalization of patients after allogeneic SCT. Therefore, prevention of CMV reactivations remains a major challenge in the posttransplant period.
CMV-specific CTLs (CMV-CTLs) play an important role in controlling the virus. Reconstitution of CMV-CTLs after allogeneic SCT is frequently impaired by intensive immunosuppression. Immune monitoring of CMV-CTLs with multimers, such as tetrameric HLA/CMV peptide complexes (tetramers), helps to assess the anti-CMV immunity after allogeneic SCT (5–8). The capacity of CMV-CTLs to control the virus results from their quantity and functionality (8, 9). The functionality of CMV-CTLs has been linked to phenotypic markers describing the level of T cell differentiation, senescence, and exhaustion and to the TCR avidity. In detail, CMV-CTLs in patients after allogeneic SCT comprise effector memory CD45RA−CCR7− T cells (TEMs) and effector memory CD45RA+CCR7− T cells (TEMRAs) capable of rapid release of IFN-γ and perforin granules upon Ag stimulation (10–12). CMV TEMRA CTLs with late differentiation phenotype characterized by the absence of CD27 and CD28 cell surface expression provide protection against recurrent reactivations (13). High expression of the inhibitory programmed cell death-1 (PD-1) molecule indicating T cell exhaustion on CMV-CTLs has been associated with complicated CMV reactivations (14). Finally, transfer of early differentiated CD57− memory CMV-CTLs from the donor provided protection against CMV reactivation (15). The tetramer staining intensity reflecting TCR avidity has been described as a marker for functional T cells. In detail, high tetramer binding T cells have been shown to be associated with high levels of cytotoxicity (16), high sensitivity to activation-induced cell death (17, 18), senescence (19), and exhaustion (20, 21). In contrast, low tetramer binding CMV TEMRA CTLs have been shown to accumulate in elderly CMV-seropositive individuals and to be associated with low effector function (22). In addition, low CMV tetramer binding CD8+ T cells coexisting with high CMV tetramer binding CD8+ T cells specific for the minor histocompatibility Ag HA-1 have been described to exert regulatory T cell function in vitro and to mediate graft tolerance in human kidney transplantation (23).
HLA-mutated tetramers with restricted binding to high-avidity T cells reduce the background staining and are thus increasingly used for monitoring of CMV-specific immune responses (6, 24, 25). During routine CMV-CTL monitoring after allogeneic SCT using HLA-mutated tetramers, we observed the coexistence of differential CMV tetramer binding CTLs in a considerable number of patients. In this study, we analyzed the kinetics, phenotype, donor/recipient origin, function, TCR usage, and clinical relevance of CTLs with differential CMV tetramer staining intensity.
Materials and Methods
Patient cohort
A total of 115 adult patients who underwent allogeneic SCT at the bone marrow transplantation unit of Hannover Medical School between 2011 and 2016 were included in this study. Patients were treated according to SCT protocols approved by the Institutional Review Board of the Hannover Medical School. Patients gave written informed consent in accordance with the declaration of Helsinki. The analysis was performed with approval of the Institutional Review Board of the Hannover Medical School (1886-2013 and 2934-2015). Only patients evaluable for the first 100 d were included in this analysis. All patients were typed on high-resolution level for exons 2 and 3 for HLA-A, -B, and -C and for exon 2 for HLA-DRB1 and -DQB1 according to the current European Federation for Immunogenetics guidelines. Patient characteristics and transplant procedures are shown in Table I. Acute GvHD was graded according to the Glucksberg score (26). PBMCs were separated from whole blood by gradient centrifugation on Biocoll (Biochrom, Berlin, Germany) and cryopreserved in liquid nitrogen after supplementation of 80% RPMI 1640 (Lonza, Basel, Switzerland), 10% FCS (Sigma-Aldrich, St. Louis, MO), and 10% DMSO (Sigma-Aldrich).
Monitoring for CMV infection
Blood samples were monitored for pp65 Ag at least once weekly as previously described (27). CMV reactivation was defined as more than five pp65 Ag-positive cells per 4 × 105 leukocytes. CMV reactivation was preemptively treated at first line with ganciclovir and at second line with foscarnet.
Monitoring of CMV-specific T cell reconstitution
Peripheral blood samples were monitored after allogeneic SCT for CMV-specific immune responses on days 25 (+/10), 50 (±10), 100 (±35), 200 (±35), and 300 (±35) or weekly in case of CMV reactivation. Whole blood was stained for CD3 (PE-Cy7; clone: UCH-T1), CD8 (FITC; clone: SFCI21Thy2D3; all from Beckman Coulter, Marseille, France) expression and with six commercially available PE-labeled tetrameric HLA/CMV epitope complexes (HLA-A*01:01 pp50-VTEHDTLLY, HLA-A*02:01 pp65-NLVPMVATV, HLAA*24:02 pp65-QYDPVAALF, HLA-B*07:02 pp65-TPRVTGGGAM, HLA-B*08:01 IE1-ELRRKMMYM, HLA-B*35:01 pp65-IPSINVHHY; MBL International, Woburn, MA). The tetramers contained a mutation (A245V) in the HLA class I H chain α3 domain, which reduces interaction of tetramer with CD8, and thereby restrains tetramer binding to high-avidity T cells (24). Tetramer staining was performed at room temperature (RT) for 30 min followed by erythrocyte lysis as previously described (28). Sample acquisition was performed on a FC500 (Beckman Coulter) after standardization using calibration beads. PE-labeled negative tetramer (MBL International) was used as a negative control for tetramer staining. Absolute numbers of CD8+ T cells were determined using fluorescent beads (FlowCount Fluorospheres; Beckman Coulter). Only samples containing >50 CD8+ T cells/μl blood were included into this study. For enumeration of absolute numbers of tetramer-binding cells, the following formula was used: [(% specific tetramer-positive cells − % control tetramer-positive cells) × absolute number of CD8+ cells]. The CMV-CTL levels were calculated as mean of CMV-CTL counts per microliter peripheral blood obtained for each tetramer used. The FACS analysis was performed using CXP software (Beckman Coulter).
CMV-CTL phenotyping, sorting, and tetramer comparisons
PBMCs rested overnight after thawing in IMDM (Lonza) supplemented with 10% human serum (HS; Sigma-Aldrich) at 37°C and 5% CO2. PBMCs were stained with a live/dead marker (Alexa Fluor 750; Life Technologies, Carlsbad, CA) and anti-CCR7 (PE-CF594; clone: 150603) at 37°C for 15 min. Subsequently, PBMCs were stained with PE-labeled tetramers as described earlier in combination with anti-CD3 (Alexa Fluor 700; clone: UCH-T1; Beckmann Coulter), anti-CD8 (PE-Cy7; clone: SFCI21Thy2D3; Beckmann Coulter), and anti-CD45RA (PerCP-Cy5.5; clone: HI100; BioLegend, San Diego, CA) at RT for 30 min. For phenotypic analysis, PBMCs were additionally stained with anti-CD28 (allophycocyanin; clone: CD28.2; BioLegend), anti–PD-1 (BV421; clone: EH12.2H7; BioLegend), and anti-CD57 (FITC; clone: NK-1; BD Biosciences, Franklin Lakes, NJ). Activated and regulatory CD8+ T cells were identified by staining with anti-CD69 (FITC; clone: FN50), anti–CTLA-4 (allophycocyanin; clone: L3D10), anti–glucocorticoid-induced TNFR (anti-GITR, BV421; clone: 108-17) (all from BioLegend). Gates were set according to isotype controls and negative tetramer staining. Samples were acquired and CD8+ CMV tetramer+ cells were isolated on FACSAria II (BD Biosciences) into tubes for in vitro expansion and chimerism analysis, or into plates for telomere length analysis. Postsorting analysis of purified subsets revealed >98% purity. For the comparisons of mutated and nonmutated tetramers, the following nonmutated tetramers were used: A*01:01 pp50-VTEHDTLLY PE-labeled (Glycotope Biotechnology, Heidelberg, Germany) and HLA-A*02:01 pp65-NLVPMVATV allophycocyanin-labeled (kindly provided by Prof. E. Goulmy, Leiden University Medical Center, the Netherlands). Flow cytometry data analysis was performed on FlowJo version 10.1 (Tree Star, Ashland, OR).
Telomere length analysis
Absolute telomere length was measured by real-time PCR with a preamplification step for application to small cell numbers (29, 30). Fifty cells were sorted in triplicates directly into 4 μl of lysis buffer per well of a V-bottom 96-well plate and frozen for subsequent PCR analysis at −20°C. Preamplification was performed on the lysate using the telomere primers [forward: 5′-(TTAGGG)14-3′ and reverse 5′-AGCAAGTGGGAAGGTGTAATCCGTCTCCACAGACAAGGCCAGGACTCGTTTG-3′] and the single copy reference gene 36B4 primers (forward: 5′-CAGCAAGTGGGAAGGTGTAATCC-3′, reverse: 5′-CAGCAAGTGGGAAGGTGTAATCCGTCTCCACAGACAAGGCCAGGCCAGGACTCGTTG-3′) with the previously described reaction conditions (29). The preamplification product was purified with Zymo PCR clean and purification kit (Zymo Research, Irvine, CA). The final PCR product was eluted in 44 μl of elution buffer and used in the subsequent real-time PCR with the same primers as described earlier in this section and reaction conditions described previously (29). The methodology was validated for established tumor cell lines of known telomere length and T cell clones by Southern blot analysis at the Department of Human Genetics, Hannover Medical School (Hannover, Germany) (31).
Chimerism analysis
Genomic DNA was extracted from 3–5 × 103 sorted cells using the QIAamp DNA Micro kit (Qiagen, Hilden, Germany). The DNA samples were analyzed with short tandem repeat PCR as previously described (32). Fragment analysis was performed on the ABI 310 genetic analyzer (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. Percentage of donor chimerism was calculated based on the peak height from informative short tandem repeat regions.
Expansion of CMV-CTLs
FACS-sorted (1–10 × 103 cells per well) CMV-CTLs were cultured in round-bottom 96-well plates (BD Biosciences) in IMDM supplemented with 10% HS, 1% LeucoA (PHA 1 μg/ml; Sigma-Aldrich), 1% penicillin/streptomycin (Life Technologies), gentamicin (5 mg/ml; Life Technologies), Fungizone (0.5 mg/ml; Life Technologies), and 1 × 105 cells per well of 30-d irradiated autologous PBMCs. A total of 120 IU/ml rIL-2 (ImmunoTools, Friesoythe, Germany) was added every 3 d. Restimulation was performed with 1% LeucoA and irradiated autologous feeder cells every 7–10 d. T cells were frozen after one or two rounds of restimulation.
Phenotypical analysis of expanded CMV-CTLs
TCR or tetramer staining analysis of expanded CMV-CTLs was performed on the FACS LSR II (BD Biosciences) after staining with anti-CD3 (Alexa Fluor 700; clone: UCH-T1), anti-CD8 (PE-Cy7; clone: SFCI21Thy2D3; Beckmann Coulter), and live/dead marker (Alexa Fluor 750; Life Technologies) together with anti-TCRα/β Ab (PE; clone: IP-26; BioLegend) for 30 min at 4°C or with tetramer for 30 min at RT.
Proliferation assay
After resting overnight in 10% HS/IMDM, 2 × 104 per well autologous PBMCs (stimulators) irradiated with 100 Gy were incubated for 60 min at 37°C in a 96-well flat-bottom plate (BD Biosciences) in the presence of the relevant peptide (HLA-A*01:01 pp50-VTEHDTLLY, HLA-A*02:01 pp65-NLVPMVATV, HLAA*24:02 pp65-QYDPVAALF, HLA-B*07:02 pp65-TPRVTGGGAM, HLA-B*08:01 IE1-ELRRKMMYM; thinkpeptides, Oxford, U.K.) in the concentrations range 102–10−3 μg/ml. Subsequently, 1 × 104 per well expanded CMV-CTLs (effectors) cultured for 3 d in 10% HS/IMDM with 120 IU/ml of IL-2 were added. After 48 h, 10 μM BrdU was added. After 24 h at 37°C, BrdU incorporation was detected by using the BrdU ELISA kit (Roche, Basel, Switzerland) according to the manufacturer’s instructions. All experiments were performed in triplicates.
Cytotoxicity assay
CD4+ enriched PHA blasts (PHAbs) with the relevant HLA type were used as target cells. PHAbs were thawed, cultured overnight at 37°C in the presence of 120 IU/ml IL-2, and labeled with 3 μM CFSE (Life Technology) in 1 ml 10% HS/IMDM for 10 min at 37°C. The reaction was stopped by 2 ml of 10% HS/IMDM, followed by 2-min incubation at 4°C. After washing in PBS, 5 × 103 CFSE-labeled PHAbs in 50 μl of 10% HS/IMDM per well were added to a U-bottom 96-well microtiter plate (BD Biosciences). Subsequently, the relevant peptide concentration (in the range 102–10−4 μg/ml) in 50 μl of 10% HS/IMDM was added and incubated at 37°C for 60 min. Finally, 1.5 × 104 CMV-CTLs in 50 μl of 10% HS/IMDM were added. The plate was centrifuged at 2000 rpm for 1 min without break and incubated for 4 h at 37°C. Subsequently, cells were stained at 4°C for 30 min with anti-CD4 (BV421, clone: OKT4; BioLegend) to exclude intercellular CFSE transfer. Subsequently, 7-aminoactinomycin D (Beckmann Coulter) was added and incubated for 10 min. Acquisition was performed on BD LSR II. Targets were identified by positivity for CFSE and anti-CD4. Dead targets (Td) were identified by positivity for 7-aminoactinomycin D. Percentage of Td was calculated as: % Td = (Td/targets) × 100. Specific lysis was calculated as follows: % specific lysis = % Td with effector cells − % Td without effector cells, as previously published (33). All experiments were performed in duplicates.
ELISPOT assay
IFN-γ ELISPOT analysis was performed as previously described (34). In brief, a 96-well MultiScreen plate (Millipore, Billerica, MA) was coated with 5 μg/ml mouse anti-human IFN-γ Ab (1-D1K; Mabtech, Cincinnati, OH) overnight at 4°C. The plate was blocked with 5% HS/IMDM for 1 h at 37°C. Autologous PBMCs (1 × 104/well) rested overnight in 10% HS/IMDM at 37°C were added after irradiation with 75 Gy. Subsequently, the relevant CMV peptide was titrated into the wells in concentrations range 101–10−6 μg/ml. After incubation for 1 h at 37°C, 5 × 103 CMV-CTLs per well were added. After 19-h incubation at 37°C, plates were incubated with 0.3 μg/ml biotin-labeled anti–IFN-γ Ab (7-B6-biotin; Mabtech) and a streptavidin-alkaline phosphatase complex (Sigma-Aldrich). Staining was performed with the BCIP-NBT substrate (Sigma-Aldrich). Spots were counted using an automated ELISPOT reader (CTL ImmunoSpot Analyzer, Bonn, Germany). The number of spots from the unstimulated control was subtracted from the stimulated sample. All experiments were performed in duplicates.
TCR sequencing
CD8+ tetramer sorted cells were subjected to cDNA-based CDR3-region amplification by RACE-PCR, allowing assessment of the entire TCR β repertoire as previously described by Van Heijst et al. (35). CDR3 amplicons were then sequenced using the Illumina MiSeq platform. Reads were annotated by the international ImMunoGeneTics information system database (http://IMGT.org); further bioinformatics analyses included VDJtools and the tcR R-package as described by Ravens et al. (36).
TCR CDR3 spectratyping
Total RNA was isolated from 1 × 105 cells using the RNeasy Micro Kit (Qiagen), and cDNA was generated. cDNA was synthesized in 20 μl of PCR with an anchored oligo-(dT)-primer using the Transcriptor First Strand Synthesis Kit (Roche). Amplification of 22 TCR β-chain variable (BV) gene subfamilies across the CDR3 was performed with 24 BV subfamily-specific primers previously published by Peggs et al. (37) together with an unlabeled β-chain constant (BC) region-specific primer (5′-CCGGCTGCTCCTTGAGGGGCTGCG-3′). The subsequent amplifications consisted of a TCR PCR to amplify BV fragments and a runoff PCR for fluorochrome labeling of the amplified fragments. In detail, the TCR PCR was performed in 18 μl of total volume containing PCR buffer, 0.3 mM each 2'-deoxynucleoside 5'-triphosphate, 1.5 units HotStarTaq polymerase, 0.5 μM each BV primer, 1 μM unlabeled BC primer, and 1.4 μl cDNA. PCR conditions were 95°C for 15 min, 34 cycles starting at 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s, followed by a final extension cycle at 72°C for 7 min. The runoff PCR was conducted in 10 μl of total volume of PCR buffer, with 0.2 mM each 2'-deoxynucleoside 5'-triphosphate, 0.25 unit HotStarTaq polymerase (Qiagen), 1 μM FAM-6–labeled BC primer, and 4 μl of the TCR PCR product. PCR conditions were 95°C for 15 min for denaturation followed by nine cycles of 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s, followed by a final extension at 72°C for 15 min. The runoff PCR product was analyzed on an ABI Genetic Analyzers 3730xI automated sequencer (Applied Biosystems) in the presence of ROX500 size standard (Thermo Fisher Scientific, Waltham, MA). Peak Scanner 2 software (Thermo Fisher Scientific) was used to analyze the data.
Statistics
Statistical analysis was performed using GraphPad Prism 5 (GraphPad Software, La Jolla, CA). A p value <0.05 was considered statistically significant.
Results
Differentially CMV tetramer binding CTLs emerging after allogeneic SCT
CMV-CTLs were monitored in the peripheral blood by whole blood tetramer staining in a cohort of 115 patients with a CMV R+/D+ serostatus on days 25 (±10), 50 (±10), 100 (±35), 200 (±35), and 300 (±35) after allogeneic SCT. One hundred three of 115 patients had detectable CMV-CTL responses against CMV peptides restricted by at least one of the six investigated HLA alleles (HLA-A01, -A02, -A24, -B07, -B08, -B35). The number of tetramers used was dependent on the HLA typing of the recipient and donor, and on the availability of CMV tetramers for the individual HLA alleles. In 53/103 patients with CMV-CTLs, coappearing CMV-CTLs with low tetramer binding (CMV tetlow CTLs) and CMV-CTLs with high tetramer binding (CMV tethigh CTLs) were detected by histogram analysis (Fig. 1). None of the studied patients characteristics, that is, gender, disease, stem cell source, conditioning, in vivo T cell depletion, GvHD prophylaxis, and donor matching, were associated with the coappearance of these differentially CMV tetramer binding (dual) CMV-CTLs (Table I). In 32/53 patients with dual CMV-CTLs, additional CMV-CTLs were detected in at least one additional HLA specificity. Twenty-eight of these 32 patients with dual CMV-CTLs had only single CMV-CTLs in one (n = 19), two (n = 8), or three (n = 1) additional HLA specificities. Four of these 32 patients with dual CMV-CTLs had dual CMV-CTLs in maximally a second HLA specificity. None of the patients with CMV-CTLs binding to CMV-HLA-A24 tetramer showed dual CMV-CTLs. Median onset for detection of CMV-CTLs was on day 48 (range 21–121 d) in patients having only CMV-CTLs with one tetramer staining intensity (i.e., “single” CMV-CTLs) and on day 47 (range 23–166) in patients with dual CMV-CTLs at least once during the observation period. Median onset of dual CMV-CTLs was on day 54 (range 23–217 d). The appearance of dual CMV-CTLs was classified according to three following scenarios (of note, four patients had dual CMV-CTLs in two HLA specificities): dual CMV-CTLs were: 1) preceded by a single CMV tetlow CTL population (5/53 patients), 2) preceded by a single CMV tethigh CTL population (9/53 patients), and 3) present from the first day of CMV-CTL detection onward (43/53 patients) (Fig. 1A). Longitudinal analysis of CMV-CTLs revealed that patients with dual CMV-CTLs had significantly higher CMV-CTL counts per microliter peripheral blood than those with consistently only single CMV-CTLs on days +50 and +100 and +300 (p = 0.02, p = 0.01, and p = 0.02, respectively, Mann-Whitney U test), but no significant differences were observed at days +25 and +200 (Fig. 1B). In patients with dual CMV-CTLs, comparison of CMV tetlow and tethigh CTL numbers revealed that CMV tethigh CTLs were present at significantly higher levels than CMV tetlow CTLs at days +50 and +100 (p = 0.02 and p = 0.03, respectively, Mann–Whitney U test) without significant differences beyond day 100 (Fig. 1C). Taken together, these data show that the presence of dual CMV-CTLs is frequent in patients after allogeneic SCT, with CMV tethigh CTLs being the predominant population.
Coappearance of CMV tetlow and tethigh binding CTLs after allogeneic SCT. (A) The coexistence of CMV tetlow and tethigh CTLs in the peripheral blood on different time points after allogeneic SCT is shown for three representative patients according to three observed scenarios. Dual CMV-CTLs were: 1) preceded by single CMV tetlow CTLs, 2) preceded by single CMV tethigh CTLs, and 3) present from the first day of CMV-CTL detection. x-axis: CMV tetramer staining; y-axis: CD8 staining. CMV tetlow are in blue, CMV tethigh CTLs are in red, and CD3+ CMV tetramer− cells are in green. (B) The course of the mean CMV-CTLs counts over all HLA alleles in the peripheral blood of 53 patients with dual CMV-CTLs (△) and 50 patients with consistently only single CMV-CTLs (♦) after allogeneic SCT is presented. x-axis: days after allogeneic SCT; y-axis: mean number of CMV-CTLs. Data points represent mean ± SEM; p values were calculated with Mann–Whitney U test, *p < 0.05. (C) The course of the absolute number of CMV tetlow and tethigh CTLs in the peripheral blood of 53 patients after allogeneic SCT is shown. x-axis: days after allogeneic SCT; y-axis: absolute number of CMV tetlow (○) and tethigh (▪) CTLs. Data points represent mean ± SEM; p values were calculated with Wilcoxon matched-pairs test, *p < 0.05.
Coappearance of CMV tetlow and tethigh binding CTLs after allogeneic SCT. (A) The coexistence of CMV tetlow and tethigh CTLs in the peripheral blood on different time points after allogeneic SCT is shown for three representative patients according to three observed scenarios. Dual CMV-CTLs were: 1) preceded by single CMV tetlow CTLs, 2) preceded by single CMV tethigh CTLs, and 3) present from the first day of CMV-CTL detection. x-axis: CMV tetramer staining; y-axis: CD8 staining. CMV tetlow are in blue, CMV tethigh CTLs are in red, and CD3+ CMV tetramer− cells are in green. (B) The course of the mean CMV-CTLs counts over all HLA alleles in the peripheral blood of 53 patients with dual CMV-CTLs (△) and 50 patients with consistently only single CMV-CTLs (♦) after allogeneic SCT is presented. x-axis: days after allogeneic SCT; y-axis: mean number of CMV-CTLs. Data points represent mean ± SEM; p values were calculated with Mann–Whitney U test, *p < 0.05. (C) The course of the absolute number of CMV tetlow and tethigh CTLs in the peripheral blood of 53 patients after allogeneic SCT is shown. x-axis: days after allogeneic SCT; y-axis: absolute number of CMV tetlow (○) and tethigh (▪) CTLs. Data points represent mean ± SEM; p values were calculated with Wilcoxon matched-pairs test, *p < 0.05.
Factor . | All Patients (N = 115) . | Patients with Single CMV-CTLs (n = 50) . | Patients with Dual CMV-CTLs (n = 53) . | p Value . |
---|---|---|---|---|
Median patient age (range), y | 53 (18–70) | 53 (18–67) | 54 (23–70) | |
Median donor age (range), y | 37 (17–69) | 38.5 (17–63) | 36 (21–69) | |
Recipient gender, n (%) | 0.55 | |||
Male | 65 (57) | 27 (54) | 32 (60) | |
Female | 50 (43) | 23 (46) | 21 (40) | |
Disease, n (%) | 0.30 | |||
AML | 64 (56) | 31 (62) | 27 (50) | |
ALL | 12 (10) | 6 (12) | 4 (8) | |
NHL/CLL | 9 (8) | 4 (8) | 3 (6) | |
MM | 9 (8) | 5 (10) | 4 (8) | |
MDS/MPN | 5 (4) | 1 (2) | 3 (6) | |
MPN incl. CML | 5 (4) | 1 (2) | 4 (8) | |
Othersa | 11 (10) | 2 (4) | 8 (14) | |
Stem cell source, n (%) | 1.00 | |||
PBSC | 106 (92) | 47 (94) | 49 (92) | |
BMb | 9 (8) | 3 (6) | 4 (8) | |
Conditioning, n (%) | 0.38 | |||
RIC | 81 (70) | 34 (68) | 41 (77) | |
MAC | 34 (30) | 16 (32) | 12 (23) | |
In vivo TCD, n (%) | 0.74 | |||
No TCD | 13 (11) | 5 (10) | 4 (8) | |
ATG/Thymo | 102 (89) | 45 (90) | 49 (92) | |
GvHD prophylaxis, n (%) | 0.54 | |||
CSA/MMF | 81 (71) | 36 (72) | 40 (75) | |
CSA/MTX | 27 (23) | 12 (24) | 9 (17) | |
Othersc | 7 (6) | 2 (4) | 4 (8) | |
Donor matching, n (%) | 0.70 | |||
MRD | 30 (26) | 10 (20) | 14 (26) | |
MUD | 64 (56) | 32 (64) | 28 (53) | |
MMUD | 17 (15) | 7 (14) | 9 (17) | |
MMRD | 4 (3) | 1 (2) | 2 (4) |
Factor . | All Patients (N = 115) . | Patients with Single CMV-CTLs (n = 50) . | Patients with Dual CMV-CTLs (n = 53) . | p Value . |
---|---|---|---|---|
Median patient age (range), y | 53 (18–70) | 53 (18–67) | 54 (23–70) | |
Median donor age (range), y | 37 (17–69) | 38.5 (17–63) | 36 (21–69) | |
Recipient gender, n (%) | 0.55 | |||
Male | 65 (57) | 27 (54) | 32 (60) | |
Female | 50 (43) | 23 (46) | 21 (40) | |
Disease, n (%) | 0.30 | |||
AML | 64 (56) | 31 (62) | 27 (50) | |
ALL | 12 (10) | 6 (12) | 4 (8) | |
NHL/CLL | 9 (8) | 4 (8) | 3 (6) | |
MM | 9 (8) | 5 (10) | 4 (8) | |
MDS/MPN | 5 (4) | 1 (2) | 3 (6) | |
MPN incl. CML | 5 (4) | 1 (2) | 4 (8) | |
Othersa | 11 (10) | 2 (4) | 8 (14) | |
Stem cell source, n (%) | 1.00 | |||
PBSC | 106 (92) | 47 (94) | 49 (92) | |
BMb | 9 (8) | 3 (6) | 4 (8) | |
Conditioning, n (%) | 0.38 | |||
RIC | 81 (70) | 34 (68) | 41 (77) | |
MAC | 34 (30) | 16 (32) | 12 (23) | |
In vivo TCD, n (%) | 0.74 | |||
No TCD | 13 (11) | 5 (10) | 4 (8) | |
ATG/Thymo | 102 (89) | 45 (90) | 49 (92) | |
GvHD prophylaxis, n (%) | 0.54 | |||
CSA/MMF | 81 (71) | 36 (72) | 40 (75) | |
CSA/MTX | 27 (23) | 12 (24) | 9 (17) | |
Othersc | 7 (6) | 2 (4) | 4 (8) | |
Donor matching, n (%) | 0.70 | |||
MRD | 30 (26) | 10 (20) | 14 (26) | |
MUD | 64 (56) | 32 (64) | 28 (53) | |
MMUD | 17 (15) | 7 (14) | 9 (17) | |
MMRD | 4 (3) | 1 (2) | 2 (4) |
Statistical analysis was performed to compare patient characteristics between patients with single CMV-CTLs and patients with dual CMV-CTLs. Comparisons of recipient gender, stem cell source, conditioning, and in vivo TCD were performed using Fisher exact test. Comparisons of disease, GvHD prophylaxis, and donor matching were performed using χ2 test.
Myelodysplastic syndrome (n = 4), aplastic anemia (n = 4), primary immune deficiencies (n = 1), Hodgkin’s lymphoma (n = 1), and MM/AML (n = 1).
Includes BM+PBSC (n = 1).
CSA/MMF/MTX (n = 3) and cyclophosphamide/CSA/MMF (n = 4).
ATG, antithymocyte globulin; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; BM, bone marrow; CLL, chronic lymphoid leukemia; CML, chronic myeloid leukemia; CSA, cyclosporine A; MAC, myeloablative conditioning; MDS/MPN, myelodysplastic/myeloproliferative syndrome; MM, multiple myeloma; MMF, mycophenolate mofetil; MMRD, HLA-mismatched related donor; MMUD, HLA-mismatched unrelated donor; MPN, myeloproliferative neoplasm; MRD, HLA-matched related donor; MTX, methotrexate; MUD, HLA-matched unrelated donor; NHL, non-Hodgkin’s lymphoma; RIC, reduced intensity conditioning; TCD, T cell depletion; Thymo, thymoglobulin.
CMV tethigh CTLs have a tetramer staining intensity similar to single CMV-CTLs
Subsequently, median fluorescence intensities (MFIs) of all dual CMV-CTLs were quantified on the days of the clearest separation of CMV tetlow and tethigh CTLs and compared with samples selected during the highest CMV-CTL counts of patients with only single CMV-CTLs during the monitoring period (Table II). This analysis revealed that CMV tethigh CTLs and single CMV-CTLs had comparable tetramer staining intensities (Table II).
HLA . | n . | MFI Tetlow . | MFI Tethigh . | Tetlow versus Tethigh . | n . | MFI Single Tet . | Tetlow versus Single Tet . | Tethigh versus Single Tet . |
---|---|---|---|---|---|---|---|---|
A01 | 13 | 3269 (1,176–11,360) | 33,896 (18,810–43,249) | p = 0.0002 | 10 | 26,627.5 (7,962–44,587) | p < 0.0001 | p = 0.10 |
A02 | 14 | 4600 (524–17,284) | 32,218.5 (18,432–61,496) | p = 0.0001 | 27 | 32,218 (6,543–62,757) | p < 0.0001 | p = 0.79 |
B07 | 17 | 8578 (1,225–23,202) | 45,811 (25,595–61,914) | p < 0.0001 | 7 | 394,606 (25,944–64,918) | p < 0.0001 | p = 0.47 |
B08 | 11 | 3534 (1,234–11,245) | 33,327 (14,252–57,277) | p = 0.001 | 7 | 30,109 (12,832–46,593) | p = 0.0006 | p = 0.47 |
B35 | 2 | 4956 (2,934, 6978) | 22,113 (15,510, 28,716) | —a | 7 | 38,159 (27,387–61,914) | —a | —a |
HLA . | n . | MFI Tetlow . | MFI Tethigh . | Tetlow versus Tethigh . | n . | MFI Single Tet . | Tetlow versus Single Tet . | Tethigh versus Single Tet . |
---|---|---|---|---|---|---|---|---|
A01 | 13 | 3269 (1,176–11,360) | 33,896 (18,810–43,249) | p = 0.0002 | 10 | 26,627.5 (7,962–44,587) | p < 0.0001 | p = 0.10 |
A02 | 14 | 4600 (524–17,284) | 32,218.5 (18,432–61,496) | p = 0.0001 | 27 | 32,218 (6,543–62,757) | p < 0.0001 | p = 0.79 |
B07 | 17 | 8578 (1,225–23,202) | 45,811 (25,595–61,914) | p < 0.0001 | 7 | 394,606 (25,944–64,918) | p < 0.0001 | p = 0.47 |
B08 | 11 | 3534 (1,234–11,245) | 33,327 (14,252–57,277) | p = 0.001 | 7 | 30,109 (12,832–46,593) | p = 0.0006 | p = 0.47 |
B35 | 2 | 4956 (2,934, 6978) | 22,113 (15,510, 28,716) | —a | 7 | 38,159 (27,387–61,914) | —a | —a |
Median and range of MFIs of CMV tetramer (tet) staining in patients with dual and single CMV-CTLs are shown according to the HLA specificity. The p values were calculated with Mann–Whitney U test.
Too few samples for proper statistical analysis.
CMV tetlow and tethigh CTLs are donor derived
Subsequently, dual CMV-CTLs were analyzed in relation to their presence in patients or donors before allogeneic SCT. The analysis was performed in 13 patient and donor pairs with detectable CMV-CTLs both in the patient and the donor before allogeneic SCT. Four scenarios were observed after allogeneic SCT. Dual CMV-CTLs were: 1) not present in the donor or patient (3/13), 2) only present in the donor (4/13), 3) only present in the patient (3/13), and 4) present in the patient and the donor before allogeneic SCT (2/13; Fig. 2A). These data suggested that dual CMV-CTLs after allogeneic SCT might be partially patient derived. To determine the donor or recipient origin of the dual CMV-CTLs after allogeneic SCT, we highly purified CMV tetlow and tethigh CTLs from 10 patients after allogeneic SCT by flow cytometry and analyzed them for donor chimerism in at least two patients per observed scenario. Both CMV tetlow and tethigh CTLs were of 100% donor origin in every scenario (Fig. 2B). Thus, dual CMV-CTLs are solely donor derived and may be either transferred from the donor or originate in the patient after allogeneic SCT.
Dual CMV-CTLs are of donor origin. (A) Presence of dual CMV-CTLs in the peripheral blood of four representative patients are shown comprising four scenarios. Dual CMV-CTLs were: 1) not present in the donor or patient before allogeneic SCT, 2) only present in the donor, 3) only present in the patient before allogeneic SCT, and 4) present in the patient and in the donor before allogeneic SCT. Samples of patients before SCT, HLA-matched related donors, and patients after SCT were analyzed on lysed whole blood. Samples of HLA-matched unrelated donors (indicated by ¥) were available only as frozen PBMCs and analyzed after dead cell exclusion. Number sign (#) indicates dot plots with enlarged dot size for better visibility. (B) Donor chimerism of CMV tetlow and tethigh CTLs purified from the peripheral blood of 10 patients comprising at least two patients per scenario. ○, scenario 1; □, scenario 2; ▴, scenario 3; ▼, scenario 4.
Dual CMV-CTLs are of donor origin. (A) Presence of dual CMV-CTLs in the peripheral blood of four representative patients are shown comprising four scenarios. Dual CMV-CTLs were: 1) not present in the donor or patient before allogeneic SCT, 2) only present in the donor, 3) only present in the patient before allogeneic SCT, and 4) present in the patient and in the donor before allogeneic SCT. Samples of patients before SCT, HLA-matched related donors, and patients after SCT were analyzed on lysed whole blood. Samples of HLA-matched unrelated donors (indicated by ¥) were available only as frozen PBMCs and analyzed after dead cell exclusion. Number sign (#) indicates dot plots with enlarged dot size for better visibility. (B) Donor chimerism of CMV tetlow and tethigh CTLs purified from the peripheral blood of 10 patients comprising at least two patients per scenario. ○, scenario 1; □, scenario 2; ▴, scenario 3; ▼, scenario 4.
CMV tetlow and CMV tethigh CTLs are phenotypically similar
Next, CMV tetlow and tethigh CTLs were phenotyped to gain insights into the level of differentiation, activation status, or a potential regulatory function (the gating strategy is shown in Fig. 3A). Cell surface expressions of CD45RA and CCR7 markers were analyzed to subdivide CMV-CTLs into CD45RA+CCR7+ naive T cells, CD45RA−CCR7+ central memory T cells, TEMs, and TEMRAs. Flow cytometric analysis of prospectively collected archived clinical samples of 10/53 patients with dual CMV-CTLs revealed that CMV tetlow and tethigh CTLs had a similar TEM and TEMRA distribution (Fig. 3B). CMV tetlow and tethigh CTLs were TEMRAs by mean 71.3% (±29.2 SD) and 59.9% (±28.6 SD), respectively (p = 0.38, Wilcoxon matched-pairs test; Fig. 3B). Percentages of highly differentiated TEMs and TEMRAs identified by the absence of CD28 did not differ significantly between CMV tetlow and tethigh cells (p = 0.65 and p = 0.63, respectively, Wilcoxon matched-pairs test; Fig. 3C). Equally, percentages of senescent TEMs and TEMRAs identified by anti-CD57 staining were similar for CMV tetlow and tethigh CTLs (p = 0.65 and p = 0.91, respectively, Wilcoxon matched-pairs test; Fig. 3C). Furthermore, there were no significant differences in the expression of the exhaustion marker PD-1 on TEMs and TEMRAs between CMV tetlow and tethigh CTLs (p = 0.08 and p = 0.3, respectively, Wilcoxon matched-pairs test; Fig. 3C). CD69 indicating recent T cell activation was present only on mean 3.9% (±5.6 SD) CMV tetlow CTLs and on mean 4.1% (± 5.1 SD) CMV tethigh CTLs (p = 0.43, Wilcoxon matched-pairs test; data not shown). Next, a potential regulatory T cell phenotype was determined in dual CMV-CTLs that was previously described in low tetramer staining CD8+ T cells specific for the minor histocompatibility Ag HA-1 (23). However, both CMV tetlow and tethigh CTLs were negative for the markers CTLA-4 and GITR (data not shown) previously described to associate with regulatory properties of CD8+ T cells (38–40). Finally, the replicative history of dual CMV-CTLs was investigated. Telomere length analysis directly performed on CMV-CTLs highly purified by flow cytometry revealed no significant difference between CMV tetlow and tethigh CTLs (Fig. 3D). Taken together, CMV tetlow and tethigh CTLs are phenotypically similar with a comparable replicative history.
Analysis of the phenotype and telomere length in CMV tetlow and tethigh CTLs. (A) Gating strategy for the assessment of the phenotypical markers. Both CMV tetlow and tethigh CTLs were separated into TEMs and TEMRAs based on CD45RA and CCR7 expression. The resulting subsets were separately analyzed for the expression of CD28, CD57, and PD-1 (shown in histogram plots; x-axis: fluorescence intensity; y-axis: events). (B) Depicted is the TEM and TEMRA subset distribution within CMV tetlow and tethigh CTLs in the peripheral blood of 10 patients after allogeneic SCT. (C) The percentage of cells negative for CD28 and positive for CD57 and PD-1 within CMV tetlow and tethigh TEMRAs and TEMs are shown. (D) Absolute telomere length (kilobases) in purified CMV tetlow and tethigh CTLs isolated from 10 patients is shown. CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪); each data point represents one patient. Horizontal lines indicate the mean value. The p values were calculated with Wilcoxon matched-pairs test. ns, not significant.
Analysis of the phenotype and telomere length in CMV tetlow and tethigh CTLs. (A) Gating strategy for the assessment of the phenotypical markers. Both CMV tetlow and tethigh CTLs were separated into TEMs and TEMRAs based on CD45RA and CCR7 expression. The resulting subsets were separately analyzed for the expression of CD28, CD57, and PD-1 (shown in histogram plots; x-axis: fluorescence intensity; y-axis: events). (B) Depicted is the TEM and TEMRA subset distribution within CMV tetlow and tethigh CTLs in the peripheral blood of 10 patients after allogeneic SCT. (C) The percentage of cells negative for CD28 and positive for CD57 and PD-1 within CMV tetlow and tethigh TEMRAs and TEMs are shown. (D) Absolute telomere length (kilobases) in purified CMV tetlow and tethigh CTLs isolated from 10 patients is shown. CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪); each data point represents one patient. Horizontal lines indicate the mean value. The p values were calculated with Wilcoxon matched-pairs test. ns, not significant.
CMV tethigh CTLs show stronger CD3 downregulation after tetramer binding than CMV tetlow CTLs
To characterize the dual CMV-CTLs functionally, we sorted CMV tetlow and tethigh CTLs from peripheral blood of 10 patients by flow cytometry and expanded them in vitro on autologous feeder cells to obtain sufficient cell numbers for functional analysis. The differences in tetramer staining intensity between CMV tetlow and tethigh CTLs in patient samples largely persisted after in vitro expansion of CMV-CTLs isolated from all patients (exemplified in Fig. 4A), and they remained statistically significant (p = 0.002, Wilcoxon matched-pairs test; data not shown). In vitro–expanded CMV tethigh CTLs showed slightly lower TCRα/β staining than CMV tetlow CTLs without reaching statistical significance (p = 0.06, Wilcoxon matched-pairs test) (Fig. 4B). Thus, the differences in tetramer staining between CMV tetlow and tethigh CTLs are not due to differences in TCR expression. Subsequently, the CD3/TCR expression upon TCR engagement by tetramer binding was analyzed on CMV tetlow and tethigh CTLs in primary peripheral blood samples and after in vitro expansion. Analysis of CD3 expression in primary peripheral blood samples of 10 patients revealed a significantly lower CD3 expression on CMV tethigh than on tetlow CTLs after CMV tetramer staining (p = 0.002, Wilcoxon matched-pairs test; Fig. 4C). After in vitro expansion, CMV tetlow and tethigh CTLs showed largely equal CD3 expression (p = 0.85, Wilcoxon matched-pairs test; Fig. 4D). However, addition of CMV tetramers revealed a significantly stronger CD3 downregulation on in vitro–expanded CMV tethigh compared with CMV tetlow CTLs (p = 0.002, Wilcoxon matched-pairs test; Fig. 4D).
Tetramer staining and CD3/TCR expression on CMV tetlow and tethigh CTLs in peripheral blood samples and after in vitro expansion. (A) Tetramer staining intensities of CMV tetlow (white histograms) and tethigh (black histograms) CTLs in peripheral blood and in vitro–expanded samples are shown for one representative individual. Gray histograms represent the staining intensity of the negative tetramer (tet). (B) TCRα/β MFIs on in vitro–expanded CMV tetlow and tethigh CTLs isolated from 10 patients are shown. (C) CD3 MFIs on CMV tetlow and tethigh CTLs in peripheral blood samples of 10 patients upon CMV tetramer staining. (D) CD3 MFIs on in vitro–expanded CMV tetlow and tethigh CTLs from 10 patients stained with either negative (−) or CMV tetramer (+). CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪). Horizontal lines show the mean value. The p values were calculated with Wilcoxon matched-pairs test, **p < 0.001. ns, not significant.
Tetramer staining and CD3/TCR expression on CMV tetlow and tethigh CTLs in peripheral blood samples and after in vitro expansion. (A) Tetramer staining intensities of CMV tetlow (white histograms) and tethigh (black histograms) CTLs in peripheral blood and in vitro–expanded samples are shown for one representative individual. Gray histograms represent the staining intensity of the negative tetramer (tet). (B) TCRα/β MFIs on in vitro–expanded CMV tetlow and tethigh CTLs isolated from 10 patients are shown. (C) CD3 MFIs on CMV tetlow and tethigh CTLs in peripheral blood samples of 10 patients upon CMV tetramer staining. (D) CD3 MFIs on in vitro–expanded CMV tetlow and tethigh CTLs from 10 patients stained with either negative (−) or CMV tetramer (+). CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪). Horizontal lines show the mean value. The p values were calculated with Wilcoxon matched-pairs test, **p < 0.001. ns, not significant.
CMV tethigh CTLs show a higher proliferative sensitivity to peptide than CMV tetlow CTLs
Subsequently, in vitro–expanded CMV tetlow and tethigh CTLs from nine patients with sufficient cell numbers were analyzed for IFN-γ release, cytotoxicity, and proliferation in response to CMV peptides titrated to target cells positive for the relevant HLA allele. In the IFN-γ ELISPOT assay, sensitivity of both CMV tetlow and tethigh CTLs to titrated CMV peptides was dependent on the relevant HLA allele (sensitivity in B07 > A01 = A02 > B08) (Fig. 5). Thus, to compare the responses of CMV tetlow and tethigh CTLs independently from the relevant HLA allele, we grouped the results according to: 1) the lowest peptide concentration leading to a response for at least one of the two CMV tetramer binding subsets (indicating sensitivity), and 2) the highest peptide concentration used in the respective assays (indicating the overall response). CMV tetlow and tethigh CTLs showed equal sensitivities to peptides in the IFN-γ ELISPOT (Fig. 6A) and in the cytotoxicity assay (Fig. 6B). However, there was significantly higher peptide sensitivity and overall proliferative response in the BrdU assay for the CMV tethigh compared with CMV tetlow CTLs (p = 0.04 and p = 0.01, respectively, Wilcoxon matched-pairs test; Fig. 6C).
Sensitivity of CMV tetlow and tethigh CTLs to titrated CMV peptides is dependent on the relevant HLA allele. IFN-γ release analysis of expanded CMV tetlow (○) and tethigh CTLs (▪) from nine patients in response to the titrated concentrations of CMV peptides presented in the relevant HLA allele: (A) HLA B07, (B) HLA A01, (C) HLA A02, (D) HLA B08. x-axis: spot count; y-axis: concentration of the peptide (μg/ml). Data points represent mean ± SEM. The differences in IFN-γ release between CMV tetlow and tethigh CTLs were not significant for all HLA alleles and all peptide concentrations.
Sensitivity of CMV tetlow and tethigh CTLs to titrated CMV peptides is dependent on the relevant HLA allele. IFN-γ release analysis of expanded CMV tetlow (○) and tethigh CTLs (▪) from nine patients in response to the titrated concentrations of CMV peptides presented in the relevant HLA allele: (A) HLA B07, (B) HLA A01, (C) HLA A02, (D) HLA B08. x-axis: spot count; y-axis: concentration of the peptide (μg/ml). Data points represent mean ± SEM. The differences in IFN-γ release between CMV tetlow and tethigh CTLs were not significant for all HLA alleles and all peptide concentrations.
IFN-γ release, cytotoxicity, and proliferation in CMV tethigh and tetlow CTLs. IFN-γ release (A), cytotoxicity (B), and BrdU uptake (C) on CMV tetlow CTLs and CMV tethigh CTLs isolated from nine patients were analyzed at titrated concentrations of relevant CMV peptides. Because the responses of CMV tetlow and tethigh CTLs to titrated CMV peptides were dependent on the relevant HLA allele, results were shown as the response: 1) to the lowest peptide concentration effective for at least one of the two CMV tetramer binding subsets (indicating sensitivity, left panels), and 2) to the highest peptide concentration used in the respective assays (indicating the overall response, right panels). CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪). Horizontal lines show mean. The p values were calculated with Wilcoxon matched-pairs test, *p < 0.05. ns, not significant.
IFN-γ release, cytotoxicity, and proliferation in CMV tethigh and tetlow CTLs. IFN-γ release (A), cytotoxicity (B), and BrdU uptake (C) on CMV tetlow CTLs and CMV tethigh CTLs isolated from nine patients were analyzed at titrated concentrations of relevant CMV peptides. Because the responses of CMV tetlow and tethigh CTLs to titrated CMV peptides were dependent on the relevant HLA allele, results were shown as the response: 1) to the lowest peptide concentration effective for at least one of the two CMV tetramer binding subsets (indicating sensitivity, left panels), and 2) to the highest peptide concentration used in the respective assays (indicating the overall response, right panels). CMV tetlow CTLs are indicated as open circles (○) and CMV tethigh CTLs as filled squares (▪). Horizontal lines show mean. The p values were calculated with Wilcoxon matched-pairs test, *p < 0.05. ns, not significant.
CMV tetlow and tethigh CTLs show differential TCR repertoire
Next, we analyzed ex vivo the TCRβ repertoire of directly sorted CMV tetlow and tethigh CTLs from four patients by TCR sequencing. The sequencing of the CDR3 region yielded 2.4 × 104 to 5.7 × 104 productive reads. For validation of the method, we verified that independently processed duplicates of the same samples yielded >95% identical sequencing data (data not shown). The results of one representative patient are shown in Fig. 7. The analysis revealed highly expanded oligoclonal clonotypes of both CMV tetlow and tethigh CTLs. The CDR3 length patterns and sequences differed between CMV tetlow and tethigh CTLs (Fig. 7A). The numbers of identified clonotypes in CMV tetlow or tethigh CTLs ranging from 4272 to 8639 did not significantly differ (p = 0.89, Mann–Whitney U test). The number of shared clonotypes between CMV tetlow and tethigh CTLs was low (Fig. 7B) and ranged from 0.72 to 2.82% of all identified clonotypes for individual patients. However, the abundance of these few shared clonotypes accounted for the majority of the repertoire as the median frequency of shared clonotypes in CMV tetlow or tethigh CTLs was 70.48% (17.85–77.00) (exemplified for one representative patient in Fig. 7C). The frequencies of the top five shared clonotypes differed by >30% between CMV tetlow and tethigh CTLs (Fig. 7C) for all four patients. Thus, CMV tetlow and tethigh CTLs varied considerably in their TCR repertoires. The differences in TCR usage between in vitro–expanded CMV tetlow and tethigh CTLs isolated from eight patients were confirmed by TCR spectratyping (Supplemental Fig. 1).
CMV tetlow and tethigh CTLs have different TCR repertoires. Ex vivo TCR sequencing analysis of CMV tetlow and tethigh CTLs isolated from one representative individual. (A) Spectratype showing the frequency proportion of clonotypes (y-axis) by their CDR3 length (x-axis). Top 20 clonotypes are color-coded, and the other clonotypes are displayed in gray. (B) Venn diagram showing the total number of shared (overlapping circles) and nonoverlapping clonotypes. (C) Frequencies of clonotypes (y-axis) shared (nongray color) and nonoverlapping (gray) between CMV tetlow and tethigh CTLs. Top five shared clonotypes (violet, yellow, red, green, and blue) are shown individually.
CMV tetlow and tethigh CTLs have different TCR repertoires. Ex vivo TCR sequencing analysis of CMV tetlow and tethigh CTLs isolated from one representative individual. (A) Spectratype showing the frequency proportion of clonotypes (y-axis) by their CDR3 length (x-axis). Top 20 clonotypes are color-coded, and the other clonotypes are displayed in gray. (B) Venn diagram showing the total number of shared (overlapping circles) and nonoverlapping clonotypes. (C) Frequencies of clonotypes (y-axis) shared (nongray color) and nonoverlapping (gray) between CMV tetlow and tethigh CTLs. Top five shared clonotypes (violet, yellow, red, green, and blue) are shown individually.
Dual CMV-CTLs are not associated with CMV reactivation
Patients with dual CMV-CTLs or consistently only single CMV-CTLs were studied for the occurrence of CMV reactivations. Median onset of the first CMV reactivation in patients with dual or with single CMV-CTLs was on day 41 (range 18–273) or on day 35 (range 18–240) after allogeneic SCT, respectively. The incidence of CMV reactivations (p = 0.38, Fisher exact test) did not differ in patients with dual or single CMV-CTLs. Subsequently, the capacity of CMV reactivations to induce separation of dual CMV-CTLs out of single CMV-CTLs was investigated. CMV reactivation preceded appearance of an additional CMV-CTL population in 5/14 patients with pre-existing single CMV-CTLs. Overall, neither presence nor onset of dual CMV-CTLs was associated with CMV reactivation.
Discussion
In this study, we for the first time, to our knowledge, describe the coappearance of CMV-CTLs with differential tetramer binding after allogeneic SCT. In addition, we show differences between the CMV tetlow and tethigh CTLs in their proliferative sensitivity to low CMV peptide concentrations. Although dual CMV-CTLs could be detected in more than half of the patients after allogenic SCT, a similar finding has not been reported previously to our knowledge. Reasons could be the unequal distribution of dual CMV-CTL populations, which often leads to classification of two distinctive populations as one, the sequence of staining with tetramers and anti-CD3/CD8 Abs in different staining protocols, or the special CMV tetramer type used in our study. Indeed, the simultaneous staining with tetramer and the used anti-CD3/CD8 Abs (as applied in our study) resulted similar to the tetramer preceded by anti-CD3/CD8 Ab staining in an optimal separation of CMV tetlow and tethigh CTLs (Supplemental Fig. 2). Moreover, the special CMV tetramer type used in our study may have facilitated the detection of dual CMV-CTL populations. Namely, the CMV tetramers used in our study contain a mutation (A245V) in the HLA class I α3 domain with CD8, which leads to a restrained binding to high-avidity T cells and a reduced background staining (24). Previous studies had shown that the strength of interaction between tetramers and the TCR is mainly dependent on the expression levels of the TCR, the CD8 coreceptor binding and the TCR affinity (41–43). In our study, the TCR expression levels on in vitro–expanded CMV tetlow and tethigh CTLs did not significantly differ. Because CD8 binding stabilizes the interaction between HLA–peptide complexes and the TCR, the reduction of CD8 binding mediated by the A245V mutation in the tetramers used in our study is more critical for low- than high-avidity T cells (44). Consequently, the differences in CMV tetramer binding in our study are at high avidity levels and might have been undetectable in most previous studies using nonmutated tetramers (5, 45). The restricted binding of mutated tetramers to high-avidity T cells is usually associated with a low background staining. Therefore, mutated tetramers are increasingly applied for monitoring of CMV-specific immune responses (6, 24, 25). Also in our study, staining with A245V tetramers lead to more distinct CMV tetlow and tethigh CTLs populations compared with nonmutated tetramers (Supplemental Fig. 3) based on a reduction of the background staining and scattering of fluorescence intensities of the individual populations. Thus, further improvements of the discrimination of CMV tetlow and tethigh CTLs might be achieved by 227/8KA tetramers totally abrogating the interaction with CD8 (46). Finally, CMV tetlow and tethigh CTLs may differ in the affinity of their TCRs to the peptide–HLA complex. The direct proof for the differences in TCR affinity requires cloning of the TCRs from CMV tetlow and CMV tethigh CTLs and the subsequent measurement of the absolute affinity of the purified TCRs to the HLA–CMV peptide complex without impact of cellular context (47).
The coexistence of two CMV-CTL populations against the same CMV epitope after allogeneic SCT raised the question whether these two CMV-CTL populations differ by their origin or by their cellular functions. Evidently, in the CMV R+/D+ setting, both the recipient before transplant and the donor can have a detectable CMV-CTL response (7, 48). In addition, previous studies have shown that recipient-derived CMV-CTLs can survive the conditioning and contribute to the CMV-specific immunity after T cell–depleted allogeneic SCT (32, 49, 50). In our study, chimerism analysis in patients with pre-existing CMV-CTLs in the recipient and in the donor revealed the exclusive donor origin of dual CMV-CTLs. Of note, dual CMV-CTLs were detectable already in some donors and some recipients before allogeneic SCT. Thus, coexistence of CMV tetlow and tethigh CTLs per se is not the result of transplantation.
Patients with dual CMV-CTLs had considerably higher CMV-CTL numbers on most time points than patients with only single CMV-CTLs. The higher overall number of tetramer-binding cells in patients with dual CMV-CTLs might only be partially explained by the presence of CMV tetlow CTLs, which quantitatively add on to the CMV tethigh CTLs. However, CMV tetlow CTLs might also provide additional help to CMV tethigh CTLs via the recently described CD40/CD40L-dependent APC maturation pathway induced by CD8+ T cells or directly by cytokines supporting CMV tethigh CTL proliferation (51). Additional studies are required to elucidate the mechanisms leading to higher overall CMV-CTL numbers in patients with dual CMV-CTLs. Among the dual CMV-CTLs, CMV tethigh CTLs were the quantitatively predominant population over CMV tetlow CTLs. However, analysis of the MFI of dual CMV-CTLs showed that CMV tethigh CTLs and single CMV-CTLs had comparable tetramer staining intensities. The dominance of CMV tethigh CTLs in our study is in accordance with a previous report showing that CMV-specific responses comprise CTL clones with high avidity (19). Moreover, it has been shown that low TCR avidities are associated with low senescence (19) and exhaustion levels (20), recent T cell activation (52, 53), regulatory properties (23, 54), and low effector function (16, 22). Therefore, we studied whether differential tetramer staining intensities of CMV-CTLs after allogeneic SCT are linked to distinctive properties in phenotype and function. Phenotypical analysis revealed no differences between CMV tethigh CTLs and CMV tetlow CTLs regarding the distribution into TEM and CD45RA+CCR7− (TEMRA) subsets. In our cohort, both CMV tethigh and tetlow CTLs equally showed a highly differentiated phenotype [CD28 negativity (55)] with equal senescence [CD57 expression (56)] and exhaustion [PD-1 expression (57)] levels. Accordingly, the telomere length analysis indicated a comparable replicative history of CMV tethigh and tetlow CTLs. Previous studies in a murine model demonstrated that T cell activation results in transient TCR downregulation and altered topographical organization of TCR and coreceptors on the T cell surface, which cause a decrease in tetramer binding (52, 53). However, CD69 as marker for recent T cell activation (58) was largely undetectable both in CMV tetlow and in tethigh CTLs, suggesting that they do not differ by their activation status. Prior studies linked low tetramer binding to a regulatory function of CD8+ T cells (23). However, CTLA-4 and GITR previously shown to indicate regulatory properties of CD8+ T cells (38–40) were not expressed on dual CMV-CTLs. This indicates that not only CMV tethigh CTLs, but also CMV tetlow CTLs are effector T cells. These findings are supported by the response of in vitro–expanded CMV tetlow and tethigh CTLs to CMV-loaded target cells in both cytotoxicity and IFN-γ release assays. Moreover, both assays did not reveal differences in the sensitivity of CMV tetlow and tethigh CTLs to titrated peptide concentrations. These data are in accordance with previous studies that showed no direct relationship between the tetramer staining intensity of effector CTLs and sensitivity to peptides in cytotoxicity assays (39, 59). In contrast with the frequent link between cytotoxic functions and IFN-γ secretion (60), cytotoxic and proliferative responses are not directly related in virus-specific CTLs (61). Accordingly, and in contrast with our cytotoxicity and IFN-γ release results, CMV tethigh CTLs showed a higher sensitivity to CMV peptide in proliferation assays and a higher overall proliferative capacity than CMV tetlow CTLs. Moreover, CMV tethigh CTLs downregulated CD3 considerably more than CMV tetlow CTLs after tetramer binding both in peripheral blood samples and in isolated and in vitro–expanded CMV-CTLs. Thus, CMV tethigh CTLs are more sensitive to TCR internalization upon interaction with the HLA–peptide complex than CMV tetlow CTLs both in vivo and in vitro. Nevertheless, it remains unclear why the considerable differences in binding of mutated tetramers to CMV tetlow and tethigh CTLs resulted in only relatively small differences in proliferative sensitivities of the isolated CMV-CTLs to the relevant peptide in vitro. Although separation of CMV tetlow from tethigh CTLs was dependent on the abrogation of CD8 binding in the tetramer monitoring, CD8 coreceptor binding was possible in functional in vitro assays. Thus, CD8 coreceptor binding in functional in vitro assays may partially compensate for potential TCR affinity differences between CMV tetlow and tethigh CTLs. In addition, it needs to be considered that in vitro expansion of CMV-CTLs may influence their function. However, confirmation of functional results obtained with in vitro–expanded CMV-CTLs directly on ex vivo samples is limited by the fact that T cell stimulation downregulates the TCRs and reduces the unique identification of CMV tetlow and tethigh CTLs (data not shown). Of note, the higher sensitivity of CMV tethigh CTLs to CMV peptides in proliferation assays correlates well with the higher in vivo CTL numbers compared with CMV tetlow CTLs. Indeed, a preliminary analysis of six patients with available data on dual CMV-CTL numbers around the day of CMV reactivation showed by trend in five of six patients a higher expansion rate of CMV tethigh CTLs compared with CMV tetlow CTLs in response to CMV reactivation (Supplemental Fig. 4). Nevertheless, additional studies are required to confirm the postulated superior expansion capacity of CMV tethigh CTLs in vivo. Overall, CMV tethigh CTLs differ from CMV tetlow CTLs by their higher expansion capacity in response to low peptide concentrations in vitro and their higher quantities in vivo.
Because of the differences in tetramer binding and CMV peptide sensitivity in proliferation analysis, we questioned whether CMV tetlow and tethigh CTLs may originate from different precursors. In the majority of patients, differentially tetramer binding CTLs appeared simultaneously in the peripheral blood early after allogeneic SCT, and only in a few patients were they preceded by single CMV tetlow or CMV tethigh CTLs. To investigate whether the dual CMV-CTLs have the same clonal origin, we compared TCR usage of CMV tetlow or CMV tethigh CTLs. Directly sorted CMV-CTLs showed only low diversities of CDR3 lengths patterns both within CMV tetlow and tethigh CTLs, which is in accordance with the previously shown restricted diversity of T cells directed at individual CMV epitopes in HLA class I (37, 62, 63). Only a few of the identified TCR clonotypes in CMV tetlow and tethigh CTLs were identical. However, the most frequent clonotypes in both CMV tetlow and tethigh CTLs were overlapping. Nevertheless, the frequencies of these shared TCR clonotypes differed considerably between CMV tetlow and tethigh CTLs. Thus, CMV tetlow and tethigh CTLs used considerably different TCR repertoires. Similar results were found for in vitro–expanded CMV-CTLs by TCR spectratyping. These data suggest that CMV tetlow and CMV tethigh CTLs targeting the same CMV epitope in HLA class I are diverse pools of CTL clones with different precursors. Apart from thymic selection, the clonotypic structure of different CD8+ T cells specific for the same epitope is mainly based on the avidity-mediated competition between clones (19, 64), as shown for the clonal selection of CMV-CTLs in CMV-seropositive healthy individuals (19, 65). In fact, higher peptide sensitivity in CMV tethigh CTLs may lead to their preferential expansion also in our study. However, the mechanisms allowing the long-term coexistence of CMV tetlow and tethigh CTLs in patients after allogeneic SCT is still unclear.
Next, we questioned whether the onset of CMV reactivation after allogeneic SCT may cause the coappearance of CMV tetlow and tethigh CTLs. However, CMV reactivation preceded the appearance of dual CMV-CTLs only in a minority of patients with pre-existing single CMV-CTLs. Thus, CMV reactivation most likely does not cause the emergence of dual CMV-CTLs. A previous study had shown an association of dual CD8+ T cells specific for the minor histocompatibility Ag HA-1 with kidney graft tolerance (23). Therefore, we assumed that dual CMV-CTLs may indicate an increased risk for CMV reactivation. However, presence of dual CMV-CTLs was not associated with a higher incidence of CMV reactivations in our study. These opposite findings regarding CTLs with differential tetramer binding might be explained by the in vitro regulatory properties of low HA-1 tetramer binding CD8+ T cells in the report of Cai et al. (23) and the effector properties of CMV tetlow CTLs in our experiments. Our study indicates that CMV tethigh CTLs need to be considered as the functionally and quantitatively superior population in patients after allogeneic SCT. CMV reactivation frequencies need to be compared in patients with only CMV tethigh or only CMV tetlow CTLs to prove the clinical importance of CMV tethigh CTLs. However, these studies are difficult to perform, because most CMV-CTL–positive patients show CMV-CTLs in different HLA specificities that can differ in their CMV tetramer staining intensities.
In conclusion, our phenotypic, functional, and clinical data suggest that both CMV tethigh and CMV tetlow CTLs are functional effector T cells capable of contributing to the anti-CMV immunity after allogeneic SCT. However, CMV tethigh and CMV tetlow CTLs differed by their peptide sensitivity in proliferation and in vivo expansion after allogeneic SCT. Dual CMV-CTLs may originate from distinct donor-derived precursors and are not a risk factor for CMV reactivations.
Acknowledgements
We thank all the physicians and nurses at the Hannover Medical School transplant unit and the outpatient clinic for dedicated work. We thank Immo Prinz and Solaiman Raha (both from Institute of Immunology, Hannover Medical School, Hannover, Germany) for advice on the TCR sequencing work.
Footnotes
This work was supported by the Marie Curie Initial Training Networks (Project Number 315963, Improving HSCT by Validation of Biomarkers and Development of Novel Cellular Therapies to E.M.W.), the German Center for Infection Research (Grant DZIF; TTU IICH; 07.804 to E.M.W), and the Deutsche Forschungsgemeinschaft (Grant DFG SFB900-B08 to C.K.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- BC
β-chain constant
- BV
β-chain variable
- CMV-CTL
CMV-specific CTL
- CMV tethigh CTL
CMV-CTL with high tetramer binding
- CMV tetlow CTL
CMV-CTL with low tetramer binding
- GITR
glucocorticoid-induced TNFR
- GvHD
graft-versus-host disease
- HS
human serum
- MFI
median fluorescence intensity
- PD-1
programmed cell death-1
- PHAb
PHA blast
- RT
room temperature
- SCT
stem cell transplantation
- Td
dead targets
- TEM
effector memory CD45RA−CCR7− T cell
- TEMRA
effector memory CD45RA+CCR7− T cell.
References
Disclosures
The authors have no financial conflicts of interest.