Purification of PBMCs from leukapheresis obtained from voluntary healthy donors was undertaken by density gradient centrifugation using Ficoll. T cells were isolated from PBMCs by negative selection of CD3^pos cells using the Dynabeads Untouched Human^TM T cell Kit (Life Technologies) according to the manufacturer’s instructions. HFF were isolated from foreskin of circumcised donors after mechanic disruption by enzymatic digestion with 5 mg/ml Collagenase D, 25 U/ml Dispase, and 0.05% Trypsin/EDTA and were cultured in R-10 medium, consisting of RPMI GlutaMAX^TM (Life Technologies) supplemented with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin. Isolated HFF were used for experiments until passage 20. Written informed consent was obtained from every voluntary donor and each study protocol was approved by the local ethics committee (Friedrich-Alexander-Universität Erlangen-Nürnberg no. 2247, Medizinische Universität Wien no. 514/2011). Phoenix and 293T cells were cultured in DMEM (Life Technologies) supplemented with 10% FCS and 100 U/ml penicillin and 100 μg/ml streptomycin, and the hybridoma cell line “gB 27-287” (a kind gift from M. Mach, Institute of Virology, Universitätsklinikum Erlangen, Germany) was kept in R-10.

T cells were stimulated with either Human T-Activator CD3/CD28 Dynabeads (Life Technologies) according to the manufacturer’s instructions or with peptide-loaded matured Dendritic cells (DC). Briefly, monocytes were enriched by adherence in RPMI GlutaMAX^TM supplemented with 2% human serum for 2 h and cultivated afterward in R-10 supplemented with 500 U/ml IL-4 and 500 U/ml IL-4 GM-CSF (both from Peprotech). On day 5, DCs were maturated by the addition of 250 U/ml IL-4, 250 U/ml IL-4 GM-CSF, 10 ng/ml IL-1β (R&D Systems), 10 ng/ml TNF (Miltenyi), 1000 U/ml IL-6 (Peprotech), and 1 μg/ml Prostaglandin E₂ (Sigma). Mature DCs were loaded on day 7 with 5 μg/ml of the HLA-A^∗0201 restricted EBV-BMLF1 peptide GLCTLVAML (Genscript). Peptide-pulsed DCs were mixed with T cells at a ratio of 1:4 for stimulation of EBV-specific T cells and seeded at a density of 0.5 × 10⁶ T cells/ml RPMI GlutaMAX^TM supplemented with 10% human serum, 100 U/ml IL-2 (Peprotech), 50 μM β-mercaptoethanol, 100 U/ml penicillin and 100 μg/ml streptomycin, 1 mM sodium pyruvate, 1x vitamins and 1x non-essential amino acids (Gibco). Stimulation of T cells with mature peptide-pulsed DCs was performed weekly. After the second stimulation CD8^pos T cells were positively selected using Dynabeads FlowComp Human CD8 Kit (Life Technologies) according to the manufacturer’s protocol and were cultured after the third antigen-specific stimulation in medium supplemented with 10% FCS instead of 10% HS. CD3/CD28-activated T cells were kept in R-10 supplemented with 100 U/ml IL-2, 50 μM β-mercaptoethanol, 100 U/ml penicillin and 100 μg/ml streptomycin, 1 mM sodium pyruvate, 1x vitamins and 1x non-essential amino acid. EBV-specific as well as CD3/CD28-activated T cells were used between days 25 and 32 for experiments.

Human cytomegalovirus strain AD169 encoding GFP was provided by M. Marschall (Universitätsklinikum Erlangen, Erlangen, Germany; Marschall et al., 2000), GFP-recombinant strains AD169 ΔUS2-11 (Falk et al., 2002) and Towne ΔUS1-12 (Marchini et al., 2001) were kind gifts from B. Plachter (Universitätsmedizin, Johannes Gutenberg-Universität Mainz, Mainz, Germany) and H. Zhu (Rutgers New Jersey Medical School, Newark, USA), respectively. Infectious HCMV supernatants were obtained by infection of semi-confluent HFF with an MOI 0.1 and collected after 11–14 days, centrifuged, and stored at -80°C until further use. Titration of HCMV stocks was performed by using the limiting dilution method according to Reed and Munch. In the experiments, confluent HFF were infected with MOI 5 unless otherwise specified and used at the indicated time points after infection.

Flow cytometric analysis of the T cell phenotype was performed using the following mAbs: CD3 (clone SK7), CD4 (clone OKT4,), CD8 (clone RPA-T8), CD27 (clone M-T271), CD28 (clone CD28.2), CD56 (clone NCAM1.2), CD62L (clone DREG-56). CAR-expression in T cells was detected with a biotinylated anti-human IgG mab (clone JDC-10) continued by incubation with PE-conjugated streptavidin. Cell surface expression of HCMV-gB in viable HFF (propidium iodide^neg or 7-AAD^neg) was detected by incubation with supernatants from hybridoma gB 27-287 and a PE- or APC-conjugated anti-mouse antibody. HLA-A^∗02 surface expression in viable HFF was determined by using an APC-labeled HLA-A^∗02 antibody (clone BB7.2). Non-infected and HCMV-infected HFF were blocked with human serum from a HCMV-seronegative individual before staining with antibodies. Intracellular expression of perforin and granzymes in T cells or UL37x1 in HFF, respectively, was analyzed after fixation with formaldehyde and permeabilization with saponin followed by incubation with monoclonal antibodies directed against perforin (clone dG9), granzyme A (clone CB9), granzyme B (clone GB11), granzyme K (clone G3H69), granzyme M (clone 4B2G4), and the myc-Tag (clone Sc-40), respectively. Cells were counted using Accu chek counting beads (Life Technologies), whereby propidium iodide or Annexin V-FITC was added for exclusion of dead cells. BD LSR Fortessa was used for measurements and BD FacsDiva and FlowJo were used for data analysis.

DNA templates for in vitro transcription of mRNA were generated by linearization of plasmids pGEM4Z encoding the CARs directed against HCMV-gB, CEA, and chNKG2D (Full et al., 2010; Lehner et al., 2012). The mRNA encoding for UL36 was generated from a PCR product amplified from pLV-EF1α-MCS-UL36-IRES-puro kindly provided by E. Mocarski (Emory University School of Medicine, Atlanta, USA; McCormick et al., 2010) using two specific primer pairs for amplification of UL36 exons (gcttacgtctgctgtcaggag, cgtgaggaatttcgacatttaatacgactcactatagggttccatttcaggtgtcgtgacgataccgtcgagattaattaaatttcagttgttcatgtaaacgtgtg, tcctgacagcagacgtaagcaccttgcgaacagacggtg) followed by Gibson assembly (NEB). The fusion construct of the gB-ectodomain and EpCAM transmembrane/cytoplasmic proportion was transcribed from a purified PCR product, which was amplified from a bacmid encoding a respective recombinant murine cytomegalovirus. In vitro mRNA transcription was performed with the mMessage mMachine T7 Ultra Kit (Ambion) according to the manufacturer’s instructions followed by RNA purification with the RNeasy Kit (Qiagen). For electroporation of the mRNA, T cells and 293T cells were resuspended in phenol-free Opti-MEM at a density of 8 × 10⁷/ml (T cells) or 0.5 × 10⁷/ml (293T cells). HFF were detached with Trypsin/EDTA and resuspended at a density of 0.5 × 10⁶/400 μl in GenePulser^® Electroporation Buffer Reagent (Bio-Rad). Electroporation was performed with 100 μl cell suspension (T cells and 293T cells) or 400 μl cell suspension (HFF) in a 4 mm electroporation cuvette after addition of 10 μg mRNA using the GenePulser square wave protocol (Gene Pulser Bio-Rad, conditions: 500 V and 5 ms for T cells and HFF, 500 V and 3 ms for 293T cells).

Lenti- or retroviral particles containing supernatants were produced by transfection of 293T or Phoenix cells, respectively. The lentiviral vector pWPI encoding EpCAM-gB was co-transfected with psPAX (Addgene 12260) and MD2.G (Addgene 12259) in a ratio of 4:3:1 in 293T cells using 3.5 μg/ml PEI. Phoenix cells were transfected with the retroviral vector pLNCX UL37x1 or pLNCX GFP [kindly provided by McCormick et al. (2005)] and pMD2.G (4:1 ratio) by addition of 3.5 μg/ml PEI. Viral supernatants were harvested after 48 h, cleared of cell debris, concentrated using Spin-X^® UF Concentrator (100,000 Dalton; Corning), aliquoted and stored at -80°C until further use. HFF were spinoculated three times with retroviral supernatants diluted 1:1 with fresh media (1500 × g, 33°C, 4 h). Transduced cells were either selected by addition of 4 μg/ml Geneticin (pLNCX-vector) or by flow cytometric cell sorting using BD FACSAria, respectively.

The lytic potential of T cells was analyzed in a non-radioactive Eu-release assay (Blomberg et al., 1996) or by determination of apoptotic target cells using Annexin V by flow cytometry. Eu release assay: target cells were detached with EDTA and labeled with Eu as previously described (Lehner et al., 2007). 2500 Eu-labeled target cells were co-incubated with effector T cells at the indicated effector:target(E:T)-ratio for 4 h at 37°C. Eu-release by target cells into supernatant was analyzed by addition of 200 μl Enhancement Solution (Perkin Elmer) to 25 μl co-culture supernatant using time-resolved fluorometry (TRF; Victor, Wallac). Specific target cell lysis was calculated using the following formula: specific lysis [%] = (Eu release in co-culture – spontaneous Eu release)/(maximum Eu release – spontaneous Eu release) ^∗100. Flow cytometric apoptosis assay: Target cells were seeded at a density of 10,000 target cells/96-well after detachment. T cells were labeled with 10 μM SNARF-1^® (carboxylic acid acetate succinyl ester; Life Technologies) in PBS for 15 min and added to the target cells at the indicated E:T-ratios. After 4 or 20 h of co-culture at 37 °C target cells were harvested with Trypsin/EDTA, washed once with PBS 0.2% HA, 0.02% Na-Azide and resuspended in 1x Annexin binding buffer (eBioscience). Cells were analyzed after addition of 5 μl Annexin V and 5 μl 7-AAD in a flow cytometric analysis. Apoptotic cells were determined by exclusion of T cells and gating on Annexin V^pos target cells. Only GFP^pos cells were considered in the case of HCMV-infected HFF. Apoptosis in UL37x1-transduced HFF fraction was calculated after flow cytometric analysis of intracellular UL37x1-myc expression in HFF (60.4–94.8%) using the formula: Annexin V^pos cells [%] = 100/myc^pos HFF [%] ^∗ (Annexin V^pos UL37x1-HFF detected by FACS [%]–X); X = (100 – myc^pos [%]) ^∗ Annexin V^pos mock-HFF detected by FACS [%]/100.

Flow cytometric analysis of CD8^pos T cell degranulation was performed by extracellular staining of CD107a as previously described (Betts et al., 2003). Briefly, 50,000 T cells were co-incubated with 50,000 target cells in the presence of PE-labeled CD107a in RPMI supplemented with 10% FCS and 10% HS. Monensin was added to a final concentration of 5 μM after 1 h of co-incubation at 37°C. After 4 h, the T cells were harvested and additionally stained with antibodies directed against CD3, CD4 and CD8.

After co-culturing 60,000 T cells with 30,000 target cells in 200 μl RPMI1640 for 4 h, the amounts of secreted IFN-γ and TNF were determined using ELISA kits from eBioscience and Mabtech, respectively.

Statistical significance was calculated using the paired two-tailed Student’s t-test (^∗∗∗ = p < 0.001, ^∗∗ = p < 0.01, ^∗ = p < 0.05).

We previously demonstrated that our HCMV-gB specific CAR is capable of mediating efficient lysis of gB transfected 293T cells and of inducing cytokine release from the CAR-T cells in response to HCMV-infected HFF (Full et al., 2010). In previous unpublished experiments with anti-CD3 activated T cells, however, we observed weak or absent lytic activity of CAR-T cells against HCMV-infected HFF 3 days after infection (data not shown), although gB was strongly expressed on the surface of infected cells at that time point (Supplementary Figure S1A).

To investigate whether the observed absence of lysis of infected cells expressing the target antigen was due to low T cell efficacy or a peculiarity of HCMV infection, we decided to thoroughly investigate this issue by conducting experiments with effector T cells activated (A) in an antigen-independent manner by αCD3/αCD28 antibody-coated beads, or (B) enriched from a preexisting memory pool of Epstein-Barr virus (EBV)-specific effector T cells. For the latter we employed a previously established culture system relying on weekly restimulation with peptide-loaded DC, which resulted in efficient expansion of antigen-specific cytotoxic memory T lymphocytes (CTLs; routinely above 50%; Supplementary Figure S1B; Lehner et al., 2007). The cultures were depleted of NK cells, and if necessary sorted for CD8^pos T cells, finally resulting effector cell populations consisting of 56.3–98.8% CD8^pos T cells (αCD3/αCD28-activated T cells) or 92.2–98.2% CD8^pos T cells (EBV-specific CTLs, not shown), and < 1.0% NK cells. EBV-specific effectors were additionally chosen because of a more advanced differentiation stage (loss of CD62L and CD45RA; upregulation of CD45RO; Supplementary Figures S1C,D). A detailed analysis of the expression profile of the cytotoxic effector molecules perforin and granzymes A, B, K, and M, however, showed a pattern which was very similar to that of αCD3/αCD28-activated T cells (Supplementary Figure S1E).

The polyclonally activated and EBV-specific T cells were transiently transfected with either the gB-specific CAR or an irrelevant CAR specific for the CEA by electroporation of CAR-encoding mRNA (CAR expression shown in Figure 1A; Hombach et al., 2001). Lytic activity of these CAR-T cells was determined 1 day after electroporation using HCMV-infected HFF and gB-expressing 293T as target cells (gB-expression is shown in Figure 1B). Unexpectedly, neither type of effector T cells achieved substantial lysis of HFF 4 days after infection (Figure 1C), not even with an effector:target(E:T)-ratio of 25:1. However, the same effectors strongly lysed gB-expressing 293T cells, although gB-expression in these target cells was lower than on infected HFFs (Figure 1D). Notably, non-specific killing, which was frequently observed with non-infected fibroblasts, also seemed to be inhibited by HCMV, since it never occurred in infected fibroblasts (Figure 1C).

We further determined the level of T cell degranulation and cytokine production in order to exclude that the observed lack of lysis of HCMV-infected HFF is due to inhibition of T cell activation. Figure 2A illustrates that the gB-specific CAR triggers degranulation in both effector populations in co-culture with infected HFF, whereas the T cells did not degranulate in the control conditions. Similarly, IFN-γ production was triggered by the gB-specific CAR in response to infected HFF, whereas there was no or low-level production under control conditions (Figure 2B); analysis of TNF showed the same pattern (data not shown). Notably, the readout of these effector functions was congruent between the αCD3/αCD28-activated T cells and the EBV-specific CTLs, and showed a high specificity for gB-recognition, in contrast to cytotoxicity with frequently also unspecific lysis of non-infected HFF (Figure 1C).

Together, these data clearly show that both polyclonally expanded as well as antigen-specific memory T cells, when engrafted with a gB-specific CAR, cannot lyse infected HFF despite detectable CAR-mediated specific activation of the T cells.

In the next step we wanted to exclude that a possible defective function of our gB-specific CAR was responsible for the observed lack of lysis. For this reason we again employed EBV BMLF1-peptide-specific CTLs, which this time, however, were directed via their TCR to HLA-A^∗0201 matched fibroblasts loaded with a saturating concentration of the same BMLF1-peptide that was used for T cell enrichment. In addition, we used a recombinant variant of the HCMV-strain Ad169 (Ad169ΔUS2-11), which contained extended deletions in the US-gene region encoding the known viral inhibitors of antigenic peptide presentation, in order to prevent down-regulation of MHC class I molecules during infection (Supplementary Figure S2A; Jones et al., 1995; Besold et al., 2009).

In this CAR-independent experimental setup HCMV-infection again resulted in a striking resistance of the HFF to lysis by the T cells (Figure 3A). Lysis-inhibition of the peptide-loaded HFF gradually increased in the course of infection and became significant already 1 day after infection at the lowest E:T-ratio of 1:1. Despite the strong gradual increase in inhibition, residual lysis obtained with an E:T of 25:1, however, was still significant even 4 days after infection. Importantly, the observed inhibition was not accompanied by any inhibition of T cell activation, as quantified by degranulation and IFN-γ secretion (Figures 3B,C). The levels of secreted IFN-γ thereby were reproducibly markedly higher than obtained with gB-CAR-mediated recognition (Figure 2B), likely reflecting enhanced T cell activation, which would explain the observed increased cytotoxicity.

Reduced lysis of HCMV-infected HFF by peptide-specific CTLs is well-known, however, this has been attributed to impaired peptide-presentation and thus reduced TCR signaling (Allart et al., 2003; Manley et al., 2004; Besold et al., 2007, 2009; Kim et al., 2011; Ameres et al., 2014). Our data obtained with the US-gene-deleted HCMV-strain Ad169ΔUS2-11 circumvents this known mechanism experimentally and strongly suggest an additional immune escape mechanism of HCMV.

To further substantiate our finding, we conducted experiments with the HCMV Towne strain, which, like Ad169, was available with deleted US-gene region. This circumstance allowed the direct comparison of strains Towne and Ad169 both in our CAR-based system and in our experimental EBV peptide specific system dependent on genuine TCR-mediated recognition. Additionally, we tested the possibility that cell death could be efficiently triggered, but progresses with slower kinetics, by prolonging the co-culture-period.

For this set of experiments we again used EBV-BMLF1-specific CTLs and directed them either via their own TCR or by the transiently expressed gB-specific CAR against the HFF. Induction of apoptosis in the HFF (non-infected or 3 days after infection) was determined after co-culture with T cells for 4 and 20 h by flow cytometry. Flow cytometry was chosen, because it allowed us to monitor cells after extended periods of co-culture, which was not possible with the Eu-release assay.

The lysis of peptide-loaded HFF by EBV-specific CTLs during 4 h of co-culture was significantly inhibited 3 days after infection with Ad169ΔUS2-11 (compared to non-infected HFF, Figure 4A), which is in accordance with our previous experiments (Figure 3A). Infection of the HFF with TowneΔUS1-12 also resulted in lower levels of lysis by the EBV-CTLs, however, this reduction (compared to non-infected HFF) did not reach statistical significance. When we extended the co-culture period to 20 h, inhibition of lysis of Ad169ΔUS2-11-infected HFF (compared to non-infected HFF) was still present (Figure 4A). However, peptide-specific lysis of the HFF (i.e., lysis of peptide pulsed versus non-pulsed HFF) generally increased and became significant for Ad169ΔUS2-11-infected HFF also at the E:T of 5:1.

When we directed, in parallel experiments, the EBV-specific CTLs by transfection of the gB-specific CAR against the infected HFF, we could not find significant CAR-specific cell death induction in any condition, not even after 20 h of co-culture (Figure 4B). This is in line with a lower degree of activation of the EBV-CTLs, when they were directed to the infected HFF via the gB-CAR, compared to activation of the EBV-CTLs in response to HFF loaded with saturating amounts of BMLF1-peptide. These differences in activation correspond to the observed different levels of degranulation (Figures 4C vs. 4D); in the case of gB-CAR-directed CTLs this seems to correlate with the levels of gB-expression on the infected HFF (Supplementary Figure S2B).

Taken together, these data indicated that lysis-inhibition is a general function of the Betaherpesvirus prototype HCMV as it is not restricted to a single laboratory strain, and it can also be demonstrated under conditions of prolonged co-culture.

The genome of HCMV encodes several anti-apoptotic proteins and untranslated RNAs, which are thought to delay suicide of infected host cells (Brune, 2011). Among them, the proteins UL36 and UL37x1 (vMIA) are known to block death receptor-mediated apoptosis and to inhibit cell death induction at different levels (Goldmacher et al., 1999; Skaletskaya et al., 2001; McCormick et al., 2005). We hypothesized that these proteins might also contribute to inhibition of T cell cytotoxicity.

Skaletskaya et al. (2001) already speculated about a role for UL36, based on its ability to block activation of caspase-8, in the inhibition of T cell cytotoxicity (Skaletskaya et al., 2001). However, the HCMV Ad169 strains used in our study encode an inactive mutant of UL36 (sequencing data not shown) but still mediated resistance to lysis by T cells, indicating that further viral proteins are involved. Furthermore, cytotoxic T cells can induce target cell death independent from caspases by releasing granzymes (Chowdhury and Lieberman, 2008). Nevertheless, we considered it important to directly investigate the ability of UL36 to block T cell cytotoxicity. Furthermore, we also considered the possibility that functional redundancy of multiple factors could prevent loss of function by inactivation of only one factor, as, e.g., UL36. Combined inactivation of these factors in HCMV result in early host cell death (McCormick et al., 2005), and was thus not an option for studying the influence on T cell cytotoxicity in our experiments. Instead, we conducted co-culture experiments of T cells with engineered non-infected HFF that expressed these HCMV-proteins separately and in combination.

Figure 5A shows the results of control experiments that confirm the protective function of the proteins, as previously reported, by blocking death-receptor mediated apoptosis triggered by an agonistic anti-Fas-antibody in combination with CHX (as previously reported by (Goldmacher et al., 1999; Skaletskaya et al., 2001; McCormick et al., 2005). As expected, UL36 strongly inhibited this type of apoptosis; apoptosis was inhibited to a lesser extent also by UL37x1, and most potently by the combination of both proteins.

In order to investigate whether UL36 and/or UL37x1 can inhibit the cytotoxic function of T cells, we used CAR-T cells directed against NKG2D ligands. Experiments with the gB-CAR were not possible, since it was technically not feasible to generate sufficient numbers of non-infected HFFs co-expressing HCMV-gB, UL36, and UL37x1. Since non-infected HFFs in cell culture do express NKG2D-ligands (Supplementary Figure S3B), we could make use of our previously generated NKG2D-based CAR which is highly potent in inducing NKG2D-ligand mediated killing (Lehner et al., 2012). We considered this option straight forward, since the key question of whether and which viral proteins block cytotoxic T cell functions is not a matter of how efficiently those cytotoxic functions are released.

Figure 5B shows apoptosis induction in the same non-infected HFFs as used in the control experiments, now targeted by CAR-T cells instead of the Fas agonistic antibody (CAR expression is shown in Supplementary Figure S3A). Indeed, combined expression of UL36 and UL37x1, and to a lesser extent of UL37x1 alone, significantly inhibited CAR-T cell cytotoxicity (p = 0.0177 and p = 0.0360, respectively). Notably, the potency of either UL36 or UL37x1 to inhibit T cell cytotoxicity was reversed compared to Fas triggered apoptosis (Figure 5A), as the inhibitory effect of UL37x1 was strong, and of UL36 was smaller and – in this series of experiments – not significant. As previously analyzed, αCD3/αCD28-activated T cells express perforin and granzymes A, B, K, and M (Supplementary Figure S1E). Blockade of the perforin/granzyme-mediated pathway in the T cells by adding a saturating concentration of CMA, which inhibits acidification of lytic granules and thereby accelerates degradation of perforin (Kataoka et al., 1996), showed a predominant contribution of this pathway (Figure 5B). Figure 5C compares the extent of UL36/UL37x1-mediated inhibition of HFF apoptosis (gray bars) with the proportion of apoptosis not inhibited by CMA (black bars), i.e., the death-receptor and not perforin/granzyme mediated proportion of cell death triggered by the CAR-T cells. The large differences between the gray and black bars illustrate that within the total of UL36/UL37x1-mediated inhibition of CAR-T cell death (gray bars) there is a substantial proportion of perforin/granzyme-mediated cell death. If CMA-blockade of the perforin/granzyme pathway in these experiments would have been incomplete, then the actual proportions of death receptor-triggered apoptosis (black bars) would even be smaller and the resulting differences (i.e., UL36/UL37x1-inhibited perforin/granzyme-apoptosis) would even be larger.

When we repeated these experiments with EBV-specific CTLs, which were directed via their TCR to peptide-loaded HFF, expression of UL36/UL37x1 in the HFF had a pronounced inhibitory effect with the T cells from one donor, which interestingly secreted the lowest levels of IFN-γ wherease the T cells from the donor with the highest IFN-γ secretion could not be blocked (Supplementary Figures S3D and S3E). Together, the experiments with three donors showed no statistically significant inhibitory effect of UL36/UL37x1. This was not unexpected, since activation of EBV-specific CTLs in response to peptide-loaded HFF possibly triggered maximum cytotoxicity and resulted in significant lysis even 4 days after infection (Figures 3A and 4A), whereas the same EBV-CTLs redirected via the gB-CAR were activated less strongly (reflected in less degranulation; Figures 4D vs. 4C) and could not trigger specific lyis (Figure 4B). The fact that HCMV-infection significantly inhibits lysis even under these conditions, whereas expression of UL36 and UL37 in engineered non-infected HFF does not, suggests the existence of yet unidentified further contributing anti-apoptotic factors.

The finding that combined expression of UL36 and UL37x1 in HFF could almost completely block cell death induction by CAR-T cells is in line with our hypothesis that HCMV encoded proteins not only block host cell suicide but also cytotoxic effector functions of T cells.