Strong and sustained effector function of memory- versus naïve-derived T cells upon T-cell receptor RNA transfer: Implications for cellular therapy


  • Simone Thomas,

    Corresponding author
    1. Department of Medicine III, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    • Full correspondence Dr. Simone Thomas, Department of Medicine III Hematology, Oncology, and Pneumology, University Medical Center of Johannes Gutenberg-University Mainz, Langenbeckstrasse 1, 55101 Mainz, Germany

      Fax: +49-6131-17-6678


    Search for more papers by this author
  • Sebastian Klobuch,

    1. Department of Medicine III, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    Search for more papers by this author
  • Katrin Besold,

    1. Institute of Virology, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    Current affiliation:
    1. Head of Laboratory, Cell Substrate Analytics, Novartis Pharma AG, Basel, Switzerland
    Search for more papers by this author
  • Bodo Plachter,

    1. Institute of Virology, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    Search for more papers by this author
  • Jan Dörrie,

    1. Department of Dermatology, Universitätsklinikum Erlangen, Erlangen, Germany
    Search for more papers by this author
  • Niels Schaft,

    1. Department of Dermatology, Universitätsklinikum Erlangen, Erlangen, Germany
    Search for more papers by this author
  • Matthias Theobald,

    1. Department of Medicine III, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    Search for more papers by this author
  • Wolfgang Herr

    1. Department of Medicine III, University Medical Center of Johannes Gutenberg-University Mainz, Mainz, Germany
    Search for more papers by this author


Current protocols used to select CMV-specific T cells for adoptive immunotherapy focus on virus-specific memory T cells from seropositive donors. However, this strategy is not feasible in patients undergoing allogeneic haematopoietic stem-cell transplantation (HSCT) from CMV-seronegative donors. Here, we redirected T cells of CMV-seronegative donors with a human genetically engineered TCR recognizing an HLA-A*0201-binding peptide epitope of CMVpp65. To facilitate clinical translation of this approach, we used a non-viral expression system based on in vitro transcribed RNA and electroporation. Although memory and naïve-derived T-cell subsets were both efficiently transfected by TCR-RNA, memory-derived T cells showed much stronger levels of HLA-A*0201-restricted cytolytic activity to CMV-infected fibroblasts and maintained acquired function for 5–10 days. In addition to redirection of CD8+ cytotoxic T cells, TCR-RNA transfection was capable of redirecting CD4+ T cells into potent Ag-specific Th cells that efficiently triggered maturation of DCs. Our data suggest that memory rather than naïve-derived T cells are the preferred subset for transient TCR expression by RNA electroporation, providing more efficient and sustained virus-specific CD4+ and CD8+ T-cell function. CMV TCR-RNA may represent a suitable therapeutic ‘off-the-shelf’ reagent to be used in severe CMV infections of HSCT patients when endogenous CMV-specific T-cell immunity is insufficient.


Reactivation of latent human CMV infection is a frequent complication in patients undergoing allogeneic haematopoietic stem-cell transplantation (HSCT). Although effective antiviral drugs are available for the prophylaxis and treatment of CMV infection, progression to CMV disease still occurs and remains a major cause of morbidity and mortality in these patients. Risk factors for CMV reactivation include donor seronegativity in about 50% of transplantations, as well as the use of T-cell depleted allografts, transplants from unrelated or HLA-mismatched donors, and graft-versus-host disease (GvHD) [1].

CMV control essentially requires the re-establishment of protective antiviral T-cell immunity in the host [2]. Thus, the adoptive transfer of virus-specific T lymphocytes into HSCT patients offers a reasonable alternative to antiviral drugs. Since treatment of reactivated CMV infection with unselected donor lymphocytes has been associated with GvHD [3], current efforts focus on the selection of CMV-specific CD4+ and CD8+ T cells for subsequent therapeutic use (reviewed in [4]). Specifically, direct sorting of virus-specific memory T cells from donor PBMCs using MHC-peptide multimers appears attractive [5]. However, this approach is presently restricted by focussing on CD8+ T cells and the availability of HLA-class I multimers. It has also limitations, if the donor is CMV-seronegative or contains only low numbers of CMV-specific memory T cells.

Genetic transduction of non-reactive T cells from CMV-seronegative donors by virus-Ag-specific TCR may be an alternative means to transfer CMV-specific T-cell function into HSCT recipients. The gold standard approach for TCR gene transfer is based on the use of viral vectors [6]. This technology results in stable integration of TCR and vector genes into the host genome, thereby carrying the risks of insertional mutagenesis, as well as the elimination of transduced T cells by antiviral host immunity [7]. Nevertheless, genetic transduction by viral vectors has already been successfully applied in clinical trials with tumour Ag-specific TCR and chimeric Ag receptors [8-10]. Herein, we investigated an alternative technology based on in vitro transcribed (IVT) RNA of a TCR recognizing the HLA-A*0201 (A2.1)-binding CMVpp65(495-503) peptide epitope [11] for the electroporation of T lymphocytes of CMV-seronegative donors. This strategy resulted in high transfection efficiency and transient expression of the transferred TCR chains for about 1 week. It allowed the in vitro generation of fully functional CMV-specific CD4+ and CD8+ T cells, as demonstrated by CD4+-mediated maturation of DCs, as well as CD4+- and CD8+-mediated cytokine secretion and cytotoxicity against CMV-infected fibroblasts. We also show that the transfer of TCR-RNA into purified memory-derived CD8+ and CD4+ T cells substantially increased the level and persistence of CMV-specific effector functions compared with that achieved with the use of sorted naïve-derived T cells and entire PBMCs. Altogether, we describe an effective and rapid procedure to genetically manipulate T cells of seronegative donors into CMV-specific CD4+ and CD8+ T cells. As stable insertion of genes into host genome is avoided, the safety profile of this approach appears particularly attractive for clinical translation.


Expression of CMVpp65(495-503)-specific TCR-RNA in human PBMCs

We used a pp65(495-503)A2.1-specific human WT TCR (WT TCRpp65) of the subfamilies AV18 and BV13.1, and electroporated anti-CD3 stimulated PBMCs with IVT-RNA coding for the TCRα and TCRβ chains. On day 1 after transfection, about 65% of PBMCs stained positive for the Vβ13.1 subfamily domain of this TCR in flow cytometry (MFI CD8+ 3.7/CD8 4.0; Supporting Information Fig. 1A). PBMCs electroporated without RNA (mock) showed endogenous Vβ13.1 expression (CD8+ 1.5%/CD8 4.8%, MFI CD8+ 1.1/CD8 1.6). Due to the transient presence of IVT-RNA in transfected cells, WT TCRpp65 expression was only marginally detectable at day 3 after electroporation (MFI CD8+ 1.1/CD8 1.8). Next, we increased TCR expression by improving the translation efficacy of the IVT-RNA molecules via optimisation of codon usage in the TCR sequences (codon-optimised TCRpp65; co TCRpp65). After transfection of this co TCRpp65 into PBMCs, Vβ13.1 staining was slightly increased when compared with that of the WT receptor at day 1 (MFI CD8+ 4.3/CD8 4.7). However, staining on day 3 was similarly low as for the WT TCR. Since individual highly affine pp65(495-503)A2.1-specific TCR of the same subfamilies have polymorphisms in their CDR3-sequences, we aimed at improving the affinity of the codon-optimised TCR by amino acid substitutions identical to a pp65(495-503)A2.1-specific TCR that was shown to dominate long-term T-cell memory of an healthy individual after CMV infection [12]. In detail, we changed Arginine at position 114 to Serine (R114S) within the CDR3α domain as well as Serine114 to Proline (S114P) and Asparagine120 to Histidine (N120H) within the CDR3β region. PBMCs transfected with this ‘mutated’ TCR (co Mut TCRpp65) showed the strongest Vβ13.1 staining (MFI at day 1: CD8+ 5.2 / CD8 5.7) in comparison to the previously used receptors (Supporting Information Fig. 1A). When we stained PBMCs with pp65(495-503)A2.1 tetramers at day 1 after RNA transfection, tetramer binding of CD4 (presumably CD8+) T cells was more intense for co Mut TCRpp65 (MFI 3.7) than for co TCRpp65 (MFI 2.4) or WT TCRpp65 (MFI 1.6), respectively (Supporting Information Fig. 1B). Interestingly, CD4+ T cells transfected with co Mut TCRpp65 RNA were also able to bind specific tetramers (MFI 1.0), albeit less efficient than their CD4 counterpart (MFI 3.7). Thus, TCRpp65 was transiently expressed in CD8+ as well as CD4+ T cells upon RNA electroporation of pre-stimulated PBMCs.

Figure 1.

IFN-γ secretion in peptide titration assays and cytolysis of CMV-infected fibroblasts by TCRpp65-RNA-transfected PBMCs. (A) IFN-γ spot production by mock, WT TCRpp65, co TCRpp65, and co Mut TCRpp65-RNA-transfected PBMCs in response to pp65(495-503) peptide-pulsed A2.1+ T2 targets at the indicated peptide concentrations analysed at day 1, day 3 and day 5 after RNA electroporation. ELISPOT assay used a CD8+ to target cell ratio of 0.3:1 and data are shown as mean + SD of duplicates from one experiment representative of three performed. (B) Co Mut TCRpp65- and GFP-transfected PBMCs were tested at the indicated CD8+ to target cell ratios for cytolysis on CMV-infected (RVKB6: CMVpp65+ or RVKB15: CMVpp65) fibroblasts at day 1 and day 3 after RNA electroporation. The percentages of lysis are shown as means ± SD of duplicate samples and are shown for one representative 51Chromium-release assay out of three experiments. PBMCs used herein were derived from CMV-seronegative donors and were pre-stimulated with OKT3 before RNA electroporation. Proportion of CD8+ T cells was measured by flow cytometry immediately prior to use in functional assays.

TCRpp65-RNA-transfected PBMCs mediate Ag-specific effector functions

In IFN-γ ELISPOT assays, TCRpp65-RNA-transfected PBMCs specifically recognized pp65(495-503) peptide-loaded T2 cells down to a peptide concentration of 0.1 nM at day 1 and 1 nM at day 3 after electroporation (Fig. 1A). Controls including T2 cells pulsed with an irrelevant A2.1-binding MDM2 peptide, as well as mock (w/o RNA) electroporated PBMCs did not trigger specific IFN-γ secretion. On day 5 after TCR-RNA transfection, specific reactivity was no more detectable. Consistent with the observed differences in the intensity of tetramer binding, PBMCs transfected with the co Mut TCRpp65 produced the highest amount of IFN-γ upon pp65(495-503)A2.1 recognition among the different receptors especially at day 3 (Fig. 1A). Based on these results, we decided to use the co Mut TCRpp65 in all subsequent experiments. To assess the ability of TCR-RNA-transfected PBMCs to recognize the naturally processed CMVpp65(495-503) epitope, we infected A2.1+ fibroblasts with a pp65+ CMV strain (RVKB6) or a pp65 deletion mutant (RVKB15). Both virus constructs lacked expression of the immunoevasins gpUS2/US3/US6/US11, which has been shown to improve in vitro recognition of infected fibroblasts by CD8+ CTL when compared with WT CMV strains [13]. PBMCs of CMV-seronegative donors redirected with the co Mut TCRpp65 efficiently killed CMV-infected (RVKB6) A2.1+ human fibroblasts at day 1 at a CD8+-to-target ratio of 2:1 or higher and to a lesser extent at day 3 in 51Chromium-release assay, while ignoring RVKB15 (CMV w/o pp65)-infected A2.1+ fibroblasts (Fig. 1B). GFP-RNA-transfected PBMCs were unable to mediate lysis on CMV-infected targets. We concluded that the RNA transfer of pp65(495-503)A2.1-specific co Mut TCRpp65 into PBMCs of CMV-seronegative donors turned them into highly efficient virus-specific cytolytic effector cells.

CD4+ T cells can be efficiently redirected into pp65(495-503)A2.1-specific Th cells

Similar to previous data with CD8-independent tumour-Ag-specific TCR [14, 15], the ability of co Mut TCRpp65-transfected CD4+ T cells to bind pp65(495-502) A2.1 tetramers (Supporting Information Fig. 1B) suggested that they have been redirected into pp65(495-503) peptide-specific Th cells. Indeed, we observed that purified and pre-stimulated co Mut TCRpp65-transfected CD4+ Th cells were able to specifically produce IFN-γ, IL-2 and TNF-α upon recognition of CMV-RVKB6-infected A2.1+ fibroblasts, but not RVKB15 (CMV w/o pp65)-infected counterparts (Fig. 2A). Because RNA-transfected T cells express the introduced TCR only transiently, memory pool formation and sustained virus control appears impossible when applying this approach in a clinical setting. However, to trigger the activation and priming of endogenous naïve T cells in vivo, direct and indirect interactions of CD8+ T cells, CD4+ T cells and professional APC are necessary. Physiologically, after cognate class II pMHC recognition, CD4+ Th cells deliver essential maturation signals to DCs through CD40-CD40L interaction and IFN-γ release [16]. This results in up-regulation of Ag-presentation and co-stimulatory molecules by the DCs, which license them to prime naïve CD8+ T cells. To analyse whether interaction of unstimulated co Mut TCRpp65-transfected CD4+ Th cells with pp65(495-503) peptide-loaded A2.1+ immature DCs (iDCs) triggers DC maturation, cells were co-incubated over 2 days and stained thereafter for maturation markers. We observed up-regulated expression of CD40, CD80, CD83, CD86 and HLA-ABC only after loading DCs with the pp65(495-503) epitope, that matched well that of DCs which were matured by a cytokine cocktail (Fig. 2B). Furthermore, we analysed the interaction of unstimulated co Mut TCRpp65-transfected CD4+ Th cells with A2.1+ DCs that were pulsed with CMV-infected A2.1 fibro-blasts, which could not be recognised by TCRpp65-redirected CD4+ Th cells themselves (Fig. 2B). In addition to a peptide/A2.1-specific up-regulation of DC maturation markers (Fig. 2B and C), TCRpp65+ CD4+ Th cells induced specific IL-12p70 secretion by DCs (Fig. 2D). DC stimulation was not detected when using RVKB15 (CMV w/o pp65)-infected A2.1 fibroblasts. These results demonstrated that CD4+ Th cells expressing pp65-specific TCR-RNA were redirected into Ag-specific A2.1-restricted cytokine-secreting Th cells. Moreover, these cells effectively induced maturation of Ag-cross-presenting iDCs.

Figure 2.

Redirection of purified CD4+ T cells by transfection with pp65(495-503)A2.1-specific TCR-RNA. (A) IFN-γ, IL-2 and TNF-α secretion of mock- or co Mut TCRpp65-transfected anti-CD3/CD28 pre-stimulated CD4+ Th cells in response to RVKB6 (CMVpp65+)- or RVKB15 (CMVpp65)-infected A2.1+ allogeneic fibroblasts at day 1 after TCR-RNA transfection. Transfected CD4+ T cells were co-incubated overnight with fibroblasts at a 0.6:1 ratio and were then analysed by ELISA at 30,000 T cells/well. Data are shown as mean + SD of duplicates from one experiment representative of three performed. (B) Maturation of DCs was analysed by staining with mAb to CD40, CD80, CD83, CD86 and HLA-ABC after co-incubation of 3 × 105 iDCs pulsed with indicated A2.1-binding peptides and 1 × 106 TCRpp65-RNA-transfected unstimulated autologous CD4+ T cells for 2 days, or after addition of maturation cytokines IL-1β, IL-6, TNF-α and PGE2. DC maturation was also measured after co-incubating 3 × 105 iDCs with 1 × 106 TCRpp65-RNA-transfected autologous CD4+ T cells and 3 × 105 CMV-infected (RVKB6 or RVKB15) HLA-A2 allogeneic fibroblasts for 2 days. Results are representative of seven independent experiments. (C) Fold increases of MFI of maturation markers expressed by DCs after co-incubation with TCRpp65-RNA-transfected unstimulated autologous CD4+ T cells (n = 5–7 donors) and RVKB6- or RVKB15-infected HLA-A2 allogeneic fibroblasts. Each symbol represents an individual donor and horizontal bars represent means and SEM. (D) Secretion of IL-12p70 by DCs after co-incubation with TCRpp65-RNA-transfected unstimulated autologous CD4+ T cells (n = 7 donors) and RVKB6- or RVKB15-infected HLA-A2 allogeneic fibroblasts. As a control, IL-12p70 secretion was measured after co-incubation of pp65 or gp100 peptide-pulsed iDCs with TCRpp65-RNA-transfected CD4+ T cells. Data are shown as mean +SEM of 14 samples and are pooled from seven independent experiments. MACS-selected CD4+ T cells used in the described experiments had a purity of >95% before TCR transfection.

More efficient and sustained effector functions upon transfection of memory T cells

Next, we sorted CD8+ T cells from PBMCs of CMV-seronegative donors into naïve (TN), central memory (TCM) and effector memory (TEM) subsets according to expression of CD28 and CD95/Fas by flow cytometry (Fig. 3A). The phenotype of these subsets analysed before and after in vitro expansion matched very well that of TN, TCM and TEM cells isolated by alternative sorting strategies that use CD45RA, CD45RO, CD62L and CCR7 as markers for T-cell differentiation (Supporting Information Fig. 2) [17]. Expanded cell populations were transfected with co Mut TCRpp65 RNA and were monitored for TCR expression during seven subsequent days. In contrast to bulk PBMCs (Supporting Information Fig. 1), nearly 90% of co Mut TCRpp65-transfected naïve-derived T cells (MFI 16.9) and more than 90% of transfected memory-derived T-cell subsets (MFI TCM 23.6, TEM 24.7) bound tetrameric pp65(495-503)A2.1 complexes at day 1 after electroporation (Fig. 3A). Again, tetramer binding decreased over time, but was still detectable at day 7 on memory-derived T cells (MFI TN 1.0, TCM 1.4, TEM 1.8). Because in vitro expansion rates of individual CD8+ subsets after electroporation were very similar (mean 1.5-fold ±0.48 for TN, and 1.3-fold ±0.31 for TM over 1 week; data not shown), differences in TCR expression kinetics were not a mere function of T-cell proliferation. GFP-RNA-transfected TN (Fig. 3A), TCM- and TEM-derived subsets (data not shown) did not show significant pp65-tetramer staining.

Figure 3.

Tetramer binding and cytotoxicity of TCR-RNA-transfected CD8+ T-cell subsets. (A) Immunomagnetically selected CD8+ T cells were stained with mAb to CD28 and CD95/Fas to identify and sort CD28+Faslow TN, CD28+Fashigh TCM and CD28Fashigh TEM fractions by flow cytometry. After in vitro expansion by anti-CD3/CD28 beads over 12 days and subsequent co Mut TCRpp65 (or GFP) RNA transfection, CD8+ TN-, TCM- and TEM-derived subsets were monitored for pp65(495-503)A2.1 tetramer binding during an additional period of 7 days. Data shown are representative of four independent experiments performed. (B) GFP or co Mut TCRpp65 expressing CD8+ T-cell subsets were analysed for cytolytic activity to pp65(495-503) peptide-pulsed T2 cells in a 51Chromium-release assay at a CD8+ to target ratio of 7:1 at the indicated time points after RNA electroporation. An A2.1-binding p53 peptide was used as a negative control. The percentage of lysis is shown as mean ± SD from duplicates and data are from a single representative assay of four independent experiments.

We then asked whether increased and prolonged tetramer binding of memory-derived T-cell (TM) subsets translated into improved CMV-specific cytolytic activity. Indeed, CD8+ co Mut TCRpp65+ TEM- and TCM-derived cells showed strong cytolysis against T2 cells pulsed with subnanomolar quantities of pp65(495-503) peptide at day 1 after electroporation. Although tetramer staining was absent beyond day 7 after electroporation (data not shown), specific lysis was detectable up to 10 days (Fig. 3B). In contrast, cytolytic activity of the co Mut TCRpp65-transfected TN-derived subset fell behind that of TM-derived subsets at day 4 after transfection, and was no longer detectable at subsequent days. Cytotoxicity was not observed against an irrelevant p53 peptide or induced by GFP-transfected T-cell subsets, respectively (Fig. 3B). Moreover, CD8+ memory-derived CTL equipped with the co Mut TCRpp65 efficiently lysed CMV-infected (RVKB6) A2.1+ fibroblasts over a time period of 6 days, while ignoring RVKB15 (CMV w/o pp65)-infected A2.1+ cells (Fig. 4). CTL also produced strong levels of IFN-γ upon recognition of A2.1+ CMV-infected (RVKB6) fibroblasts up to day 6 after transfection (data not shown). In contrast, TN-derived CD8+ cells mediated less cytolytic activity to CMV-infected fibroblasts at day 1 and failed to mediate specific lysis at day 4 and day 6 after TCR transfection (Fig. 4). Because naïve T cells usually require APC stimulation before acquiring full effector function, we also used an alternative protocol from Hinrichs et al. including allogeneic PBMCs, anti-CD3 mAb and IL-2 for in vitro stimulation before TCR transfection [18]. We observed that naïve-derived T cells developed significant cytolytic function against CMV-infected fibroblasts after one to two weekly courses of APC stimulation (data not shown). However, the level of cytotoxicity was still below that of memory-derived T cells in the absence of APC stimulation. In contrast to our standard APC-free stimulation protocol using anti-CD3/CD28 beads and recombinant cytokines (i.e. IL-2, IL-7, IL-15), survival rates of T cells upon TCR-RNA electroporation were lower (20–40% versus 50–70%) and TCR expression disappeared more rapidly (day 3 versus day 5 upon transfection).

Figure 4.

TCRpp65-RNA-transfected CD8+ T-cell subsets mediate specific lysis to CMV-infected fibroblasts. Cytolytic activity of pp65(495-503)A2.1 reactive TCRpp65-RNA- or GFP-RNA-transfected CD8+ TN-, TCM- and TEM-derived subsets in response to CMV-infected (RVKB6: CMVpp65+ or RVKB15: CMVpp65) A2.1+ fibroblasts at the indicated CD8+ to target ratios and the indicated time points after RNA electroporation. The percentage of lysis is shown as mean ± SD from duplicates and data are from a single representative 51Chromium-release assay of three independent experiments performed with three individual T-cell donors.

To investigate whether findings with CD8+ T-cell subsets could be applied to CD4+ counterpart fractions as well, we sorted CD4+ T cells from PBMCs of CMV-seronegative donors into CD28+Faslow TN and CD28+Fashigh TM subsets (Fig. 5A). Both sub-populations expressed the differentiation markers CD45RA, CD45RO, CD62L and CCR7 similar as has been reported for CD8+ TN and TM cells (Supporting Information Fig. 3) [17]. After transfection of co Mut TCRpp65 RNA into CD4+ subsets, about 87% (MFI 47.1) of TN- and 95% (MFI 54.6) of TM-derived cells stained positive for the TCR-Vβ13.1 chain at day 1 (Fig. 5A). Again, the majority of CD4+ TN and TM-derived cells bound pp65(495-503)A2.1 tetramers at day 1 after electroporation (MFI TN 69.5, MFI TM 75.6). In line with CD8+ subsets, specific TCR expression and tetramer binding decreased over time, but was consistently superior in subsets derived from TM compared with TN at day 3 and day 5 after transfection. Mock-transfected subsets stained only positive for endogenous TCR-Vβ13.1 expression and did not bind specific pp65-tetramers. Next, we analysed co Mut TCRpp65-transfected CD4+ TN- and TM-derived subsets for recognition of pp65(495-503)-peptide pulsed T2 cells, as well as CMV-RVKB6-infected A2.1+ fibroblasts in IFN-γ ELISPOT assay. Here, CD4+ TM-derived cells showed significant CMV reactivity over 5 days after TCR transfection that was clearly superior to that of TN counterpart cells (Fig. 5B). Recognition was not observed for the irrelevant MDM2 peptide, RVKB15 (CMV w/o pp65)-infected A2.1+ fibroblasts and mock-transfected TN- and TM-derived subsets. We concluded that TCRpp65-RNA-transfected CD8+ and CD4+ TM-derived subsets are stronger CMV-specific effector cells than their naïve counterpart cells and maintain acquired function for 5–10 days after electroporation.

Figure 5.

TCR expression, tetramer binding and IFN-γ production of CD4+ T-cell subsets upon TCR-RNA transfection. (A) Immunomagnetically selected CD4+ T cells were stained with mAb to CD28 and CD95/Fas to identify and sort CD28+Faslow TN and CD28+Fashigh TM fractions by flow cytometry. Following in vitro expansion (12 days) and co Mut TCRpp65-RNA transfection, CD4+ TN- and TM-derived subsets were monitored for TCR-Vβ13.1 expression and pp65(495-503)A2.1 tetramer binding during subsequent 5 days. Data shown are representative of three independent experiments. (B) Mock- or co Mut TCRpp65-transfected CD4+ T-cell subsets were analysed for IFN-γ secretion in response to pp65(495-503) or MDM2(81-88) (10−6 M) peptide-pulsed T2 cells and RVKB6 (CMVpp65+)- or RVKB15 (CMVpp65)-infected HLA-A2+ allogeneic fibroblasts in an ELISPOT assay at a CD4+ to target ratio of 0.3:1 at the indicated time points after RNA electroporation. Data are shown as mean + SD of duplicates and are from one assay representative of three performed.


TCR gene transfer is a very effective approach to generate virus-specific T cells from seronegative donor PBMCs in vitro [19]. However, the use of retroviral vectors for stable TCR expression carries the risk of insertional mutagenesis [20], even though serious genotoxic events have been reported to occur less frequent in mature T cells than in haematopoietic progenitor cells [21]. Additionally, the introduction of novel TCR α/β genes can lead to the formation of mixed dimers between naturally expressed and transferred TCR α/β chains. ‘Off-target’ reactions induced by these TCR hybrids might be harmful to the patient, particularly if stable transduced T cells cannot be completely eliminated after adoptive transfer [22, 23].

To avoid potential limitations of stable viral gene transfer, we used a transient expression system based on electroporation with IVT-RNA. This comparably fast and easy procedure resulted in high transfection efficiency of a CMV-specific A2.1-restricted TCR in CD8+ as well as CD4+ T cells [24, 25]. Although the formation of mixed TCR dimers and therefore ‘off-target’ reactions after adoptive T-cell transfer can also not be excluded in this strategy, potential adverse effects such as GvHD would be most likely temporary due to transient TCR-RNA expression. Delivery of a WT TCRpp65 into PBMCs led to peptide/A2.1-specific redirection and subsequent effector activity of both CD8+ and CD4+ T cells. Expression level and effector function of the WT TCR at day 3 after transfection was not impressive, but could be significantly improved by DNA codon-optimisation and amino acid substitutions within the variable TCRα and TCRβ domains. The latter was in line with studies from Bennett et al., in which the authors fine-tuned TCR affinities by novel combinations of polymorphisms occurring in public clonotype TCR [26]. Additional improvements in TCR expression might be achieved by the introduction of a second inter-chain disulfide bond [27], the chimerisation of the human TCR with murine constant domains [28], and the genetic linkage of the TCR genes by a short self-cleaving 2A element [29].

The success of adoptive T-cell therapy concerning the induction of durable CMV control in HSCT recipients depends, among other issues, on the long-term persistence of transferred T cells and their potential for memory pool formation. This cannot be achieved with T cells only transiently expressing the CMV-specific TCR by themselves. Alternatively, the induction of an endogenous de novo T-cell response through the activation and priming of naïve T cells should be protective as well [30]. For this, we demonstrated here that expression of the co Mut TCRpp65 in CD4+ T lymphocytes turned them into pp65-specific A2.1-restricted Th cells that were capable of triggering the maturation of monocyte-derived iDCs in vitro. Cross-presentation of viral Ags in DCs would then be essential for priming CMV-specific CD8+ T cells, as recently highlighted in acute CMV infection in mice [31]. While our experimental set-up is certainly not adequate to mimic an in vivo situation of cross-presentation and cross-priming in the patient, it could provide the basis for further experiments in an A2.1-Tg NOD/SCID/γcnull mouse model, as previously shown for the in vivo priming of EBV-specific human T cells [32]. These mice might allow, after engraftment with human CD34+ stem cells and infection with a ‘humanized’ pp65(495-503)A2.1 epitope-expressing murine CMV-strain [33], to analyse the single and combined effects of TCRpp65-RNA-transfected CD4+ and CD8+ T cells on their efficacy to facilitate priming of naïve T cells and formation of a CMV-specific memory T-cell pool.

We also analysed purified T-cell subsets to identify a cell population that allow for most effective and durable TCR-RNA expression. We first observed that sorting TN and TM cells according to CD28 and CD95/Fas expression, a strategy previously established in macaques [34, 35], appears very suitable for humans as well: The CD28/CD95-isolated CD4+ and CD8+ subsets matched the phenotype of TN and TM (for CD8 also TCM and TEM) cells defined by the differentiation markers CD45RA, CD45RO, CD62L and CCR7, and largely maintained the characteristic expression profile after in vitro expansion for 12 days. Thus, the CD28/CD95 sorting approach may be a reliable alternative to strategies based on the aforementioned conventional differentiation markers. For optimal separation of CD4+ TCM and TEM cells co-staining with CD62L can be helpful [34].

In TM-derived populations of either central or effector memory type, we observed significant CMV-specific cytolytic activity for 5–10 days after TCR-RNA electroporation, which represents a clear prolongation of effector function in comparison to previous publications that used unselected T cells rather than purified TM cells [24, 25, 36]. Memory-derived T cells expressed TCR at higher levels and for a longer timer period than naïve-derived T cells did. This very consistent observation could not be explained by differences in cell proliferation, because TN- and TM-derived cells showed a similar in vitro expansion rate upon transfection. However, a possible explanation could be that our in vitro pre-stimulation conditions using anti-CD3/CD28 beads and IL-2/IL-7/IL-15 provided improved transfection susceptibility for TM-derived cells. Alternatively, the more robust TCR expression in TM-derived cells might relate to less mixed TCR dimer formation due to lower TCR diversity [37], as well as by stronger TCR nanocluster formation in TM cells [38]. Aside from increased TCR expression, the higher level of CMV-specific cytolytic activity might also be explained by more preformed granzyme and perforin granules in TM cells (data not shown). An additional reason could be that CD4+ and CD8+ TM cells were selected based on strong expression of CD95/Fas, which is co-expressed with its ligand upon T-cell activation and plays a key role in the Fas/FasL-mediated cytolytic T-cell response [39].

Although the adoptive transfer of T cells transfected with RNA encoding chimeric Ag receptors have already demonstrated robust anti-tumour effects in mouse models when repetitively infused [40, 41], the biological potential of TCR-RNA-transfected T cells in humans has not been addressed so far. The effective treatment of CMV reactivation or disease in patients certainly requires more than one T-cell dose, for example, weekly infusions according to the length of TCR expression. In contrast to the use of TCR-RNA-grafted T cells in an autologous setting, an emerging issue in patients after allogeneic HSCT would be the induction of alloreactivity through the repeated application of TCR-transfected donor lymphocytes recognising mismatched minor and major histocompatibility Ags. One approach to reduce this problem might be the transfer of pp65-specific TCR into lymphocytes with single or few Ag specificities, for instance EBV-specific memory T-cell lines. As soon as the introduced TCR chains disappear, only T cells with the endogenous EBV specificity would still be functional. An alternative TCR recipient cell population could be the entire memory T-cell subset, because of its reduced alloreactivity compared with the naïve fraction counterpart as has been shown by several groups in vitro [42, 43]. Memory T cells offer the potential additional advantage that they are readily activated in vivo if their naturally expressed TCR recognise circulating pathogen Ags (e.g. from EBV). This natural activation might complement in vitro pre-activation of T cells and thus might ensure rapid reactivity mediated by the introduced TCR.

In conclusion we show here that RNA electroporation of an HLA-class I restricted CMV-specific TCR effectively redirects CD4+ and CD8+ T cells from seronegative donors into potent CMV-specific effector cells. These TCR-transfected T cells lyse CMV-infected fibroblasts and trigger maturation of monocyte-derived DCs upon recognition of the CMV epitope. The latter function might increase the likelihood that endogenous priming of CMV-specific T-cell immunity is initiated or facilitated in patients upon transfer of TCR-RNA-transfected T cells. We also demonstrate here that focusing the TCR-RNA approach to purified memory rather than naïve T-cell subsets increases the overall magnitude of CMV-specific effector function and allows its maintenance for about 1 week. This would be a major advantage in the clinical setting when repeated doses of transduced T cells must be scheduled. Since the introduced TCR is transiently expressed, the method avoids major problems of TCR gene therapy based on stable transduction systems. It may be also used for the pre-evaluation of immediate side effects induced by a therapeutic TCR in humans, even if future trials with stable TCR gene vectors are planned. We believe that the described ‘TCR-RNA-memory-T cell’ approach has the potential to be clinically developed for allogeneic HSCT patients with severe CMV infections, when endogenous CMV-specific T-cell immunity is insufficient.

Materials and methods

Peptides, antibodies and HLA-peptide tetramers

A2.1-binding peptide epitopes pp65(495-503), gp100(280-288), MDM2(81-88) and p53(264-272) were synthesised by Biosyntan (Berlin, Germany). mAbs were anti-human CD4-FITC, CD8-FITC, CD28-PC5, CD45RA-PE, CD45RO-PE, CD62L-PE/-PC5, CD80-PE, CD86-PE, Vβ13.1-PE (Beckman Coulter, Krefeld, Germany), CD45RA-allophycocyanin, CD45RO-allophycocyanin, CD8-V450-Horizon, CD95-PE, CD40-PE (BD Biosciences, Heidelberg, Germany), CD83-PE (Miltenyi Biotec, Bergisch Gladbach, Germany), and CCR7-FITC (R&D Systems, Minneapolis, MN, USA). PE-labelled pp65(495-503) A2.1 tetramers were synthesised by Beckman Coulter.


PBMCs were isolated from buffy coats of A2.1+ or A2.1 and CMV-seronegative donors by standard Ficoll centrifugation. TAP-deficient T2 cells and primary human foreskin fibroblasts were used as A2.1+ target cells. A2.1+ monocyte-derived iDCs were generated and matured by exogenous cytokines as previously described [44].

Sorting and pre-stimulation of T cells

PBMCs were stimulated in vitro with 30 ng/mL anti-CD3 mAb (OKT3, Janssen-Cilag, Neuss, Germany) for 5–7 days before electroporation. MACS technology was used to obtain pure CD4+ and CD8+ T cells (Miltenyi Biotec). For isolating TN and TM subsets, pure CD8+ and CD4+ T cells isolated from unstimulated PBMCs were stained with anti-CD28 and anti-CD95 mAb [34, 35] and sorted using FACSAria BD cell sorter (BD Biosciences). Before electroporation, sorted TN and TM cells were expanded over 12 days in vitro by anti-CD3/CD28 Dynabeads (Invitrogen, Darmstadt, Germany) at 3–5 μL/1 × 106 cells and 50–100 U/mL IL-2 (Proleukin, San Diego, CA, USA), 5 ng/mL IL-7 and 5 ng/mL IL-15 (both R&D Systems) in RPMI-1640 medium supplemented with 10% heat-inactivated human serum (RPMI/10%).


HCMV mutants RVKB6 and RVKB15 were generated by bacterial artificial chromosome (BAC) mutagenesis of the HCMV-BAC pAD/Cre using Red recombination in Escherichia coli strain EL250, as described before [45]. For generation of BAC-pKB6, the open reading frames of the immune evasins US2, US3, US6 and US11 were sequentially deleted [13]. The recombinant RVKB15 strain is a pp65-deletion mutant of RVKB6, in which the UL83 region of RVKB6, encoding pp65, was deleted by inserting a tetracycline resistance cassette (unpublished data). Functional assays were performed using human foreskin fibroblasts infected with HCMV mutants over 1 day at a MOI of 5 as target cells.

Molecular cloning of TCR genes

Codon optimisation of the previously described TCR [46] was performed by Geneart (Regensburg, Germany). TCRα R114S and TCRβ S114P/N120H mutants were constructed from pp65(495-503)-specific codon-optimised TCRα and β chain DNA by indicated aa residue replacements according to IMGT numbering [47] and was performed according to the Quickchange mutagenesis protocol of Stratagene (La Jolla, USA). The sequences of the mutagenic forward primer for TCRα R114S was 5′-ACCTACCTGTGCGCCAGCAACACCGGCAACCAG-3′ and for TCRβ S114P/N120H 5′-TGCGCCAGCAGCCCCGTGACCGGCACCGGCCACTACGGCTACACC-3′, respectively. Sequences of CDR3 regions were CARNTGNQFYFG for WT-TCRα (AV18), and CASSSVTGTGNYGYTF for WT-TCRβ (BV13S1) according the nomenclature of Arden [48]. The coding sequences of GFP or the TCR gene constructs were first amplified by PCR and inserted into the pGEM4Z-5′UTR-sig-MAGE-A3-DC.LAMP-3′UTR vector [49] via XbaI/XhoI restriction sites, thereby replacing the MAGE-A3-DC.LAMP DNA.

In vitro transcription of TCR-RNA

pGEM4Z-TCRα, pGEM4Z-TCRβ and pGEM4Z-eGFP DNA were linearised by SpeI restriction, purified with standard phenol/chloroform extraction and ethanol precipitation and used as DNA-templates [50]. IVT was performed with T7 RNA-polymerase using the mMESSAGE mMACHINE T7 Ultra kit (Applied Biosystems/Ambion, TX, USA). After enzymatic polyadenylation and DNaseI digestion (Applied Biosystems/Ambion), the IVT-RNA was purified using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and controlled by agarose gel electrophoresis.

RNA electroporation of T cells

Pre-stimulated PBMCs or T cells were washed twice and then re-suspended in OptiMEM without phenol red (Invitrogen) at 25–50 × 106 cells/mL. Electroporation was performed in a 4-mm cuvette with the GenePulser Xcell system (Bio-Rad, Munich, Germany) applying a square wave pulse of 500 V, 5 ms, to 3–10 × 106 cells in the presence of 5–10 μg of IVT-RNA [25]. Immediately after electroporation, cells were transferred to fresh RPMI/10% medium supplemented with 50 U/mL IL-2.

Flow cytometry and T-cell assays

TCR expression was determined by flow cytometry on a FACS-Canto II device (BD Biosciences). Standard 4-h 51Chromium-release assay and 20-h IFN-γ ELISPOT assay were performed in duplicate wells as described [42]. ELISAs for IFN-γ, IL-2, TNF-α (BD Biosciences) and IL-12p70 (eBioscience, San Diego, CA, USA) were performed in duplicates according to the manufacturer's instructions. For induction of DC maturation and DC-derived IL-12p70 secretion, 106 unstimulated and TCR-RNA-transfected CD4+ T cells were co-cultured for 2 days with 3 × 105 A2.1+ iDCs and 3 × 105 CMV (RVKB6 or RVKB15)-infected A2.1 human fibroblasts (6 days after infection) or 3 × 105 peptide-loaded (10−6M) A.2.1+ iDCs.


We thank Dr. Mirjam H.M. Heemskerk (University Medical Center, Leiden, NL) for kindly providing the CMVpp65-specific WT TCR, Dr. Ralf-Holger Voss (University Medical Center, Mainz, Germany) for assistance concerning TCR codon-optimization, and Dr. Thomas Shenk (Princeton University, New Jersey, USA) for HCMV-BAC mids. This work was supported by grants KFO183-TP10 and TP8 from the Deutsche Forschungsgemeinschaft to S.T., W.H., M.T. and B.P., the MAIFOR and Science Transfer programs from the University Medical Center Mainz to S.T. and grant SCHA1247/1-1 to N.S.

Conflict of interest

The authors declare no financial or commercial conflict of interest.


graft-versus-host disease


haematopoietic stem-cell transplantation


in vitro transcribed