Discovery of immunodominant T-cell epitopes reveals penton protein as a second immunodominant target in human adenovirus infection

The current study has been approved by the Internal Review Board of Hannover Medical School. Following written informed consent peripheral blood was obtained from 64 healthy platelet donors from the Hannover Medical School (MHH) Institute for Transfusion Medicine and 26 pediatric patients after HSCT with detectable HAdV-DNA in blood and/or stool. Healthy donors had no prior history of blood transfusion and no signs of acute infection. All donors and patients were typed for HLA class I and class II alleles at the four-digit level by sequence-based typing [32]. Informed consent was obtained from all donors and patients as approved by the Ethics Committee of Hannover Medical School, and trial subject data were treated as confidential information protected by medical confidentiality.

HAdV hexon and penton protein sequences restricted to the types 1, 2, 3, 5, and 31 were obtained from the SwissProt database (http://​www.​uniprot.​org). Epitope prediction programs SYFPEITHI (http://​www.​syfpeithi.​de) [28, 29], BIMAS (http://​www.​bimas.​cit.​nih.​gov) [30], and NetChop (http://​www.​cbs.​dtu.​dk) [31] were used to predict nonamers capable of binding to HLA- A*01:01, A*02:01, A*03:01 and B*08:01 molecules (Fig. 1). Epitope candidates were only selected if identified by all programs according to their predictive scores (Table 1). The N_ETMHC_STAB [33], N_ETMHC, and N_ETMHC_cons (Fig. 1) prediction algorithms provided by the Center for Biological Sequence Analysis (CBS, http://​www.​cbs.​dtu.​dk) was used to predict the stability of pMHC complexes for all database-available HLA types.

Peptides of the 19 top-scoring epitope candidates (Table 1) were synthesized (China Peptides, Shanghai, China; ProImmune, Oxford, UK) and used for pre-screening and T-cell immunoassays (Fig. 1). The overlapping peptide pools of hexon (HAdV5Hexon_pp, Miltenyi Biotec, Bergisch Gladbach) and penton (HAdV5Penton_pp, Miltenyi Biotec) were further used as stimuli of antiviral memory T cells. The HLA-A*01-restricted hexon-derived peptide TDLGQNLLY (A01Hexon_TDLG, ProImmune, Table 1) was used as a positive control. Peptide binding assays were performed using two additional HLA-restricted peptides from phosphoprotein 65 (pp65) of the human cytomegalovirus (YSEHPTFTSQY: A01pp65_YSEH and NLVPMVATV: A02pp65_NLVP, ProImmune) as positive controls.

The T2 peptide binding assay was performed with the HLA-A*01- and HLA-A*-02-restricted peptide candidates as described previously [34]. To determine peptide binding to HLA-A*01:01 molecules, T2 cells were transfected to express membrane-bound HLA-A*01:01 [35]. Briefly, 1 × 10⁶ T2 cells/ml were pulsed with 50 μg/ml peptide (Table 1) and 5 μg/ml beta-2 microglobulin (β2 m, Sigma, St Louis, MO, US) in serum-free medium for 15–18 h at 37 °C. T2 cells incubated without peptide served as controls. HLA expression levels were determined by flow cytometry (FACSCanto II and FACSDiva V6.1.2 software, BD Biosciences, San Jose, CA) using the monoclonal antibodies (mAb) HLA-ABC fluorescein (FITC, w6/32, AbD Serotec, Ltd-Kidlington, UK) and anti-HLA-A*02 phycoerythrin (PE, BB7.2, Biolegend, San Diego, CA, US).

The IFN-γ EliSpot assay was used for peptide screening to enumerate HAdV-specific IFN-γ-producing T cells [25, 36]. Briefly, peripheral blood mononuclear cells (PBMCs) were plated at a density of 2.5 × 10⁵ cells/well in triplicate wells and incubated overnight in the presence of the investigated peptides (10 µg/ml) and peptide pools (1 µg per peptide/ml). PBMCs cultured with medium alone or in the presence of 1 µg/ml staphylococcal enterotoxin B (SEB, Sigma-Aldrich, Hamburg, Germany) served as negative and positive controls, respectively. Spots of IFN-γ-positive cells were counted and analyzed and the results were expressed as the number of spots per well (spw). The mean number of spots in the negative control was subtracted from the mean number of spots in the antigen wells. The cut-off value for a positive response was ≥5 spw in healthy donors and ≥2 spw in HAdV-infected patients.

The proliferative capacities of HAdV-specific T cells induced by the various peptide candidates (Table 1, highlighted in italic) were analyzed by carboxyfluorescein succinimidyl ester (CFSE) dilution assay (Invitrogen, Darmstadt, Germany). 5 × 10⁵ CFSE-labeled PBMCs were stimulated with 10 µg/ml peptide for 7 days. Unstimulated T cells were used as negative controls. Spontaneous T-cell proliferation in unstimulated controls was subtracted from the specific values in peptide-stimulated cultures. Cells were stained with allophycocyanin (APC)-conjugated anti-CD8 mAb and PE-Cy7-conjugated anti-CD3 mAb (all BD Biosciences). The distribution of viable and dead cells was analyzed by 7-amino-actinomycin D (7-AAD) staining (BD Biosciences). At least 50,000 events were acquired in the 7-AAD⁻ gate. CFSE^+/−CD3⁺ and CFSE^+/−CD8⁺ T-cell populations were gated based on the scatter properties of viable 7-AAD⁻CD3⁺ or 7-AAD⁻CD3⁺CD8⁺ T lymphocytes.

A comprehensive analysis of the phenotype and specificity of HAdV-specific T cells was performed with freshly isolated PBMCs and in vitro expanded HAdV-specific T cells. PBMCs from healthy donors were stimulated for 7 days with 10 µg/ml peptide (Table 1, highlighted in italic), and restimulated for another 7 days with irradiated autologous peptide-loaded PBMCs. Flow cytometric analysis of the CD8⁺ T-cell phenotype was performed using cells stained with the mAbs anti-CD3 peridinin chlorophyll protein (PerCP), anti-CD8 APC (BD Biosciences), anti-CD19 FITC, anti-CD62L APC-Cy7, and anti-CD45RA PE-Cy7 (BioLegend) to assess the frequencies of naïve T cells (T_N; CD62L⁺ CD45RA⁺), central memory T cells (T_CM; CD62L⁺ CD45RA⁻), effector memory T cells (T_EM; CD62L⁻ CD45RA⁻), and terminally differentiated effector memory T cells (T_EMRA; CD62L⁻ CD45RA⁺).

In addition, specificities and frequencies of CD8⁺ T cells against the peptide candidates A01Penton_STDV, A02Hexon_TLLY, and the positive control A01Hexon_TDLG (Table 1) were detected by pMHC multimer staining using R-PE-conjugated multimeric Pro5 pentamers (ProImmune). At least 100,000 events were acquired in the lymphocyte gate, which was set based on the light scatter properties scatter properties of lymphocytes and on CD3⁺ T-cell populations. To be considered positive, the sample had to (1) be a well-defined cell population and/or (2) contain ≥0.3 % pentamer⁺CD8⁺ T cells.

The cytotoxicity of in vitro expanded peptide-specific T cells (day 14) was assessed in a non-radioactive flow cytometric assay using autologous CFSE-labeled peptide-loaded PBMCs as target cells. In order to exclude alloreactivity, unloaded CFSE-labeled PBMCs were further used as target cells, in which the basal cytotoxic activity of effector T cells against the unloaded target cells was subtracted from the specific cytotoxic values. Briefly, effector T cells were incubated with target cells at effector to target (E:T) ratios of 10:1, 30:1 and 60:1. Target cell lysis was assessed after 5 h using 7-AAD staining.

In addition, antiviral T-cell degranulation was determined as a surrogate marker of cytotoxicity by testing for CD107a cell surface expression by flow cytometry. In vitro expanded HAdV-specific T cells were incubated with 10 µg/ml peptide and PE-Cy7-conjugated anti-CD107a mAb (BioLegend) at 37 °C and 5 % CO₂. After 1 h of incubation, monensin (1:1000, BioLegend) was added and cells were incubated for further 4 h before staining with anti-CD3 PerCP and anti-CD8 APC.

The enrichment efficiency of in vitro expanded HAdV peptide-specific T cells (day 14) was assessed using the cytokine secretion assay (Miltenyi Biotec). Aliquots of cell fractions before enrichment (“Origin”) and after enrichment (“Eluate”) were used for detailed analysis of IFN-γ-secreting viable HAdV-specific T-cell subsets by multicolor flow cytometry. In addition to anti-IFN-γ PE (Miltenyi Biotec), cells were stained with anti-CD45 APC-Cy7, anti-CD56 PE-Cy7, anti-CD3 FITC, anti-CD8-APC mAbs, and 7-AAD (all BD Biosciences). At least 10,000 events were acquired in the viable CD45⁺7AAD⁻ leukocyte gate.

Statistical analysis was performed using the Prism v5.02 software (GraphPad, San Diego, California, USA). The results are displayed as mean ± standard deviation (SD). Generated data were analyzed using non-parametric Mann–Whitney U test. Significance levels were calculated and expressed as p values (*p < 0.05, **p < 0.01, ***p < 0.001).

A total of 947 hexon- and penton-derived epitope candidates restricted to HLA-types A*01, A*02, A*03, and B*08 were identified by the computer algorithms SYFPEITHI, BIMAS, and NetChop (Fig. 1). Nineteen (2 %) of the highest scoring sequences (six HLA-A*01-, five HLA-A*02-, two HLA-A*03-, and six HLA-B*08-resticted peptides; Table 2, [13, 20‐23]) were synthesized. All three algorithms yielded comparably high prediction scores for seven of the 19 peptide epitopes (A02Hexon_TLLY, B08Hexon_GLRY, A01Penton_STDV, A01Penton_SSDI, A02Penton_ILHT, B08Penton_NTKY, and B08Penton_DSKK): the SYFPEITHI score was 23–35, the BIMAS rank 1–2, and NetChop rating “strong binder (SB)” (Table 2). In comparison to the known immunodominant peptide epitopes of HAdV (A01Hexon_TDLG) and of cytomegalovirus (A01pp65_YSEH, and A02pp65_NLVP), we detected (Additional file 1: Figure S1A) the highest binding affinity for A01Penton_STDV (FI 0.13), A01Penton_SSDI (FI 0.11), A02Penton_ILHT (FI 1.29) and A02Penton_ALGI (FI 1.29) by in vitro T2 peptide binding assay. The peptide binding results generated by computer and in vitro analyses were comparable. Because immunodominant peptides bound to HLA molecules were shown to be more stable than non-immunogenic peptides [33, 37], we further investigated the peptide-HLA class I stability of peptides of the database-available HLA types A*01, A*02, and A*03 using the CBS NetMHC_stab algorithm (Table 2). HLA class I peptide complexes with a predicted half-life (t_1/2) > 2 h are defined as very stable [33]. This criterion was met by 7/14 epitopes (50 %) restricted to HLA-A*01, -A*02, and -A*03, including two highly stable (HS, t_1/2 ≥ 6 h) and five weakly stable (WS, t_1/2 ≥ 2 h) epitopes.

Our analysis of specific memory T-cell responses to the control antigens HAdV5Hexon_pp, HAdV5Penton_pp, and A01Hexon_TDLG revealed high response rates in healthy donors (Fig. 2a) and HAdV-infected HSCT recipients (Fig. 2b). A total of 19 peptide candidates were selected and analyzed by IFN-γ EliSpot for their capacity to induce HAdV-specific T-cell responses. Seven elicited a specific T-cell response, defined by a cut-off of ≥5 spw (Table 1). The response rates ranged from 16.7 to 65.8 % (Fig. 2a). Pre-classification of peptide candidates into non-immunodominant or immunodominant was performed according to the number of responders (≥20 %).

The A01Hexon_TNDQ was pre-classified as non-immunodominant (16.7 % responder), whereas peptides B08Hexon_DLQD, A02Penton_ILHT, B08penton_DSKG, B08Penton_DSKK (n = 4, Table 1) were pre-classified as low immunodominant. T cells from ≥50 % of healthy donors recognized the peptides A02Hexon_TLLY and A02Penton_STDV and were therefore pre-classified as highly immunodominant. The mean number of IFN-γ⁺ spots ranged from 6 to 17 spw, and was highest for A02Hexon_TLLY (17 spw) and A01Penton_STDV (mean 10 spw).

The response to the hexon control antigens (HAdV5Hexon_pp: 81.3 % responder, 37 spw; A01Hexon_TDLG: 55.6 % responder, 38 spw) was comparably high in healthy donors. For HAdV5Penton_pp we found a positive T-cell response in 75 % of donors (14 spw) and T-cell frequencies which were not significantly higher compared to the new A01Penton_STDV epitope.

In HAdV-infected HSCT patients, the immunogenicity of both peptide epitopes A02Hexon_TLLY and A01Penton_STDV was further assessed to confirm their clinical relevance (Fig. 2b). Specific T-cell responses against the control antigens HAdV5Hexon_pp (45 spw) and A01Hexon_TDLG (17 spw) were found in 100 % of patients. All HLA-A*02-positive patients (5/5) developed a HAdV-specific T-cell response against A02Hexon_TLLY (≥2 spw, 5 spw). For A01Penton_STDV we found a positive T-cell response in 5/6 (83.3 %) of patients (9 spw), while 24/26 (92.3 %) of all patients developed a HAdV5Penton_pp-specific T-cell response (21 spw).

Comparing both overlapping peptide pools, T-cell responses to hexon were significantly higher (healthy donors: 2.6-fold, patients: 2.1-fold) than those to penton. Furthermore, T cells from most donors and patients recognized HAdV5Hexon_pp and A01Hexon_TDLG at higher frequencies than A02Hexon_TLLY, but showed no significant frequency differences between HAdV5Penton_pp and the peptide candidate A01Penton_STDV.

To find the optimal concentrations of the six pre-classified immunodominant peptides (Table 1, highlighted in italic) for T-cell immunoassays, we determined the sensitizing peptide concentration required to elicit 50 % of maximal T-cell responses (SD₅₀) using the IFN-γ EliSpot assay (Fig. 1; Additional file 1: Figure S1B). The SD₅₀ ranged from 5 to 10 µg/ml for B08Hexon_DLQD (n = 2) and B08Penton_DSKG (n = 2), and from 10 µg/ml to 50 µg/ml for A02Hexon_TLLY (n = 3), A01Penton_STDV (n = 2), A02Penton_ILHT (n = 2), and B08Penton_DSKK (n = 2). Therefore, T-cell immunoassays were performed at a final peptide concentration of 10 µg/ml.

The function of in vitro expanded HAdV-specific T cells was evaluated by CSA in order to determine how quickly and strongly HAdV-specific T cells secreted IFN-γ in response to A02Hexon_TLLY and A01Penton_STDV, respectively. IFN-γ⁺ T-cell populations in healthy donors were investigated by multicolor flow cytometry before and after enrichment (Fig. 5; Additional file 3: Figure S3). According to the peptide-specific CD3⁺ T-cell frequency in the CSA fraction before enrichment (“Origin”), donors were classified as either weak responders (WR < 0.3 % IFN-γ⁺CD3⁺ T cells, A02Hexon_TLLY: n = 2, A01Penton_STDV: n = 3) or strong responders (SR, > 0.3 % IFN-γ⁺CD3⁺ T cells, n = 2). The “Origin” CSA fraction contained 0.21 ± 0.13 % (WR) and 1.89 ± 0.78 % (SR) IFN-γ⁺CD3⁺ T cells, which could be enriched with a mean purity of 12.6 ± 8.22 % in WR and 76.7 ± 8.47 % in SR, including 11.0 ± 12.3 % (WR) to 86.6 ± 4.59 % (SR) of IFN-γ secreting CD3⁺CD8⁺ T cells (Fig. 5). The mean frequency of A01Penton_STDV-specific IFN-γ⁺CD3⁺ T cells before (0.17 ± 0.12 % WR and 0.85 ± 0.35 % SR) and after enrichment (3.97 ± 1.43 % WR and 54.58 ± 11.30 % SR) was 1.5-fold lower than for A02Hexon_TLLY, whereas the frequency of IFN-γ⁺CD3⁺CD8⁺ T cells was comparable (2.60 ± 0.62 % WR and 90.78 ± 2.67 % SR) to A02Hexon_TLLY. The low enrichment efficiency of donors classified as weak responders reflects the low precursor frequency of antigen-specific CTLs (A02Hexon_TLLY: 0.19 %, A01Penton_STDV: 0.30 %), as determined by pentamer staining, which may lead to a low binding affinity of IFN-γ-secreting T cells in the CSA. In this context, starting with an antigen-specific CD8⁺ T-cell frequency >0.3 % (A02Hexon_TLLY: 0.82 %, A01Penton_STDV: 5.37 %, SR) resulted in the enrichment of highly pure IFN-γ-secreting T cells.

Here, we were focused on the identification of peptide epitopes from HAdV penton, a major capsid protein. The penton base interacts with the other major capsid proteins (hexon and fiber) and is known to be involved in adenovirus cell entry [38]. Despite the variability of clinically relevant HAdV types, the penton sequence is composed of highly conserved regions and short hypervariable loops, similar to hexon [39].

Identification of immunodominant T-cell epitopes from these conserved regions represents an attractive approach to induce antiviral T cells that are cross-reactive with several potentially lethal HAdV types. Previous studies have demonstrated sequential or concomitant co-infections with different HAdV types in immunocompromised patients, which seem to result in the generation of recombinant HAdV types [6, 40, 41]. In the present and a previous study we identified HAdV types 1, 2, and 31 as the predominant pathogens in most pediatric patients with severe HAdV disease [7]. In our present study, 10 of 26 post-transplant patients were infected with type 31 of HAdV species A, and three of them were co-infected with type 2 of HAdV species C. The peptide epitopes identified in this study are shared among HAdV types and may thus facilitate viral clearance even in patients suffering from co-infection with different strains.

For the first time we identified three low immunodominant peptide epitopes (A02Penton_ILHT, B08Penton_DSKG, B08Penton_DSKK) and one high immunodominant peptide epitope (A01Penton_STDV) of the major HAdV capsid protein penton. Sequences of all identified immunodominant epitopes were highly conserved among the clinically relevant HAdV species C, with the exception of A31-derived B08Penton_DSKG (DSKGRSYNL), which only differs in one amino acid from the peptide B08Penton_DSKK (DSKKRSYNL) but is composed of the same amino acids at the conserved anchor positions P2 and P9. The sequence of the newly identified A01Penton_STDV is located between amino acid positions 76 and 84 of the penton protein of HAdV types 1, 2, and 5. Immune responses to the whole adenovirus, including the hexon protein, were dominated by memory CD4⁺ T cells. Interestingly, A01Penton_STDV was identified as a strong inducer of functional CD8⁺ CTLs (11-fold increase in A01Penton_STDV-specific CD8⁺ CTLs), which could be efficiently enriched to a high frequency (90.4 % recovery), comparable to the HAdV5Penton_pp peptide pool (96.9 % recovery). In vitro expanded A01Penton_STDV-specific CD8⁺ T cells showed a differentiated CD45RA^neg T-cell phenotype, with T_EM proved to be essential for effective clearance of and protection against HAdV infections, while T_CM demonstrated to be involved in preventive long-term immunity [13]. The immunogenicity of all four pre-classified immunodominant penton epitopes was verified by antiviral T-cell monitoring and in vitro T-cell immunoassay in healthy donors and patients.

In healthy donors, coordinate responses of hexon-specific CD4⁺ and CD8⁺ T cells were demonstrated to contribute to the control of HAdV infections, which persist as memory T cells [21]. Most hexon-derived epitopes are CD4 restricted [20, 22, 23]. We aimed to identify novel hexon-derived CD8⁺ T-cell epitopes by evaluating the immunogenic potential of six predicted CD8⁺ T-cell epitopes. Overall, we detected HAdV-specific T-cell responses to 3/6 predicted peptide candidates in healthy donors. B08Hexon_DLQD was thus pre-classified as a low immunodominant (≥20 % response) and A02Hexon_TLLY [13, 22, 23] as a high immunodominant CTL epitope (≥50 % response). The immunogenicity of both candidates was verified by T-cell monitoring and in vitro T-cell immunoassay. In accordance to already published data (Table 1), we found no or only minimal T-cell responses to the peptide candidates A01Hexon_TNDQ [21], A02Hexon_TLAV [13], and B08Hexon_GLRY [20]. In vitro expanded functional A02Hexon_TLLY-specific CTLs showed a high proliferative and cytotoxic capacity and displayed a CD45RA^neg differentiated phenotype, which could be efficiently enriched to a significantly high frequency (85 % recovery), comparable to the HAdV5Hexon_pp peptide pool (95.5 % recovery).

Olive et al. demonstrated that additional three residues at each end of the immunodominant peptide Hexon_TLLY (15-mer peptide) resulted in a stronger CD4⁺ T-cell response. In addition, a lower peptide concentration for in vitro T-cell activation was required compared to the nonamer, indicating that the 15-mer peptide is an optimal CD4⁺ T-cell epitope [22, 23]. Guether et al. identified the decamer of the known immunodominant HLA-A01-restricted hexon-derived peptide (A01Hexon_TDLG, LTDLGQNLLY) as a naturally presented T-cell epitope that induces peptide-specific T cells with a higher proliferative, cytotoxic, and IFN-γ-producing capacity than T cells specific for the shorter peptide TDLGQNLLY [19]. Therefore, we will evaluate the immunogenic potential of longer peptides from the novel identified low immunodominant B08Hexon_DLQD peptide in a future study.

Clinical relevance of the identified high immunodominant CTL epitopes A02Hexon_TLLY and A01Peton_STDV was further assessed in HAdV-infected HSCT recipients, compared to healthy donors and correlated to T-cell responses induced by HAdV5Hexon_pp, HAdV5Penton_pp, and the immunodominant A01Hexon_TDLG. As expected, the majority of donors and patients had HAdV-specific T cells against hexon and penton, and the frequency of hexon-specific T cells was higher than that of penton-specific T cells in both cohorts. More than 50 % of all tested donors and patients had HAdV-specific T cells against A02Hexon_TLLY, but at an up to 2.2 (donors) to ninefold (patients) lower frequency than the investigated control antigens HAdV5Hexon_pp and A01Hexon_TDLG. For this reason, A02Hexon_TLLY can be classified as a high immunodominant CTL epitope but, as mentioned before, it has a lower immunogenic potential compared to the known HAdV5Hexon_pp and A01Hexon_TDLG, as underlined by the higher immunogenicity of the 15-mer peptide DEPTLLYVLFEVFDV [22]. Interestingly, the immunogenic potential of A01Penton_STDV, the novel penton-derived CTL epitope, was comparable to that of HAdV5Penton_pp in healthy donors and patients. This finding indicates that A01Penton_STDV is a major immunodominant penton-derived CTL epitope. In infected HSCT recipients we detected higher T-cell responses to the A01Penton_STDV than to the hexon-derived A02Hexon_TLLY, which was the reverse in healthy donors. These data indicate that penton-specific T cells might be essential for viral control. Future studies including monitoring of T-cell functionality and specificity should be done to get more inside into the interplay between hexon- and penton-specific T cells in the first line of defense and/or long-term protective immunity against HAdV.

Low frequencies of circulating HAdV-specific T-cell precursors in peripheral blood make it difficult to detect HAdV-specific T cells, which can lead to false-negative results [36]. We were able to detect specific T-cell responses to the peptide epitopes A02Hexon_TLLY and A01Penton_STDV directly in freshly isolated PBMCs from healthy donors by IFN-γ EliSpot, albeit at quite low frequencies (A02Hexon_TLLY: 17 spw, A01Penton_STDV: 10 spw). Interestingly, the number of responders after peptide-restimulation was unchanged, indicating that no false-negative results were obtained in freshly isolated PBMCs compared to peptide-restimulated cells and that all tested donors were correctly identified as responders and non-responders. However, evaluation of A02Hexon_TLLY and A01Penton_STDV by pMHC multimer staining indicates the need for antigen-dependent T-cell expansion to avoid overlooking effector function of peptide-induced CTLs with very low precursor frequencies. Pentamer staining of freshly isolated PBMCs yielded false-negative results in 14.3 % A02Hexon_TLLY-responders and 9.1 % of A02Penton_STD-responders. Pentamer staining with in vitro stimulated cells resulted in more accurate identification of positive responders as a clearly defined peptide-specific CD8⁺ T-cell population with up to 3.3-fold higher frequencies. In this context, the highly sensitive IFN-γ EliSpot assay is a suitable technology for the detection of low precursor frequencies of circulating HAdV-specific T cells in peripheral blood. Conversely, for unambiguous identification of responders and non-responders by pentamer staining, we recommend short in vitro stimulation. The enrichment efficiency of antigen-specific T cells via CSA was further affected by the percentage of HAdV-specific T cells. The enrichment efficiency of A02Hexon_TLLY- and A02Hexon_TLLY-specific CD3⁺ T cells was lower (purity < 13 %) when the starting frequency was <0.3 %, whereas a high enrichment efficiency was obtained by using starting frequencies >0.3 % (<67 % purity). Therefore, short-term expansion protocols for the generation of higher numbers of functionally active HAdV-specific T cells, as described by Geyeregger et al., are a suitable option to generate sufficient numbers of HAdV-specific T cells for adoptive immunotherapy [12, 13]. In addition, manufacturing of clinical-grade HAdVPenton_pp- and HAdVHexon_pp-specific T cells in combination might be a promising option to provide sufficient numbers of effective HAdV-specific T cells for the adoptive T-cell transfer in immunocompromised patients.