Memory T cells expressing stem cell–like properties have been described recently. The capacity of self-renewal and differentiation into various memory/effector subsets make them attractive for adoptive T cell therapy to combat severe virus infections and tumors. The very few reports on human memory stem T cells (TSCM) are restricted to analyses on polyclonal T cells, but extensive data on Ag-specific TSCM are missing. This might be due to their very low frequency limiting their enrichment and characterization. In this article, we provide functional and phenotypic data on human viral-specific TSCM, defined as CD8+CD45RA+CCR7+CD127+CD95+. Whereas <1% of total T cells express the TSCM phenotype, human CMV–specific TSCM can be detected at frequencies similar to those seen in other subsets, resulting in ∼1/10,000 human CMV–specific TSCM. A new virus-specific expansion protocol of sort-purified TSCM reveals both upregulation of various T cell subset markers and preservation of their stem cell phenotype in a significant proportion, indicating both self-renewal and differentiation potency of virus-specific T cells sharing their TCR repertoire. Furthermore, we describe a simplified culture protocol that allows fast expansion of virus-specific TSCM starting from a mixed naive T/TSCM pool of PBLs. Due to the clinical-grade compatibility, this might be the basis for novel cell therapeutic options in life-threatening courses of viral and tumor disease.
Adoptive T cell therapy has emerged as an effective treatment option in viral diseases after solid organ and stem cell transplantation (1–4). However, T cell engraftment and longevity of the efficacy is incidentally limited in chronically immunocompromised patients due to insufficient persistence of T cells after infusion and reduced proliferation (3, 5, 6). The ultimate objective is to adoptively transfer in those patients a long-lived memory T cell population with stem cell–like behavior, as a capacity to self-renew and the ability to differentiate to facilitate potent antigenic cytotoxicity. The underlying mechanisms to gain a T cell population that facilitates profound T cell engraftment and longevity of the cell efficacy remain to be fully elucidated.
According to the diverse expression levels of the lymph node homing molecules CCR7, CD62L, and the leukocyte common Ag (CD45) isoforms RA and RO, the memory/effector T cell compartment can be divided at least into four main subsets (7, 8): memory stem T cells (TSCM) have now joined the well-established central memory (TCM), effector memory (TEM), and terminally differentiated effector (TEMRA) T cell subsets (8). Human TCM uniformly express CD62L; however, TEM are heterogeneous for CD62L expression. Moreover, TCM coexpress CCR7 and characteristically lose CD45RA during naive → memory transition (9). CCR7 expression is rapidly downregulated upon antigenic restimulation, along with the differentiation of TCM into TEM (9). TEMRA are terminally differentiated and lose CCR7, CD62L, and CD45RO expression but re-express CD45RA. They poorly proliferate both in vitro and in vivo. Phenotypically, TSCM belong to the so-called naive-like T cell (TN) compartment (CD45RO−CD45RA+CCR7+), although a small subset of these expresses large amounts of CD95 and IL-2Rβ (8).
TCM are thought to have long-lived behavior and show superior engraftment capacities compared with other memory T cell subsets in first preclinical trials (8–11). Although some in vitro data suggest a loss of TCM phenotype during expansion (3, 12), we recently demonstrated that partial inhibition of the IL-2R signaling pathway supports expansion of TCM without loss of CD62L/CCR7 (11). The introduction of TSCM that resemble an earlier stage of memory T cell differentiation opens new opportunities for transfer of stem cell–like cells for improved long-term efficacy. TSCM are capable of reconstituting the full diversity of memory and effector T lymphocytes on serial transplantation in mice, indicating that these cells are endowed with the stem cell–like attributes of self-renewal capacity and multipotency (13). These qualities make TSCM a particularly attractive subset to use in adoptive T cell therapies and promise long-term efficacy. Although the instructive signals guiding the formation of TSCM have just begun to be investigated, clinical-grade–compliant protocols to facilitate translation of TSCM into clinical studies remain undefined. The published data from human studies are restricted to analyses on total polyclonal T cell populations, and extensive data on Ag-specific TSCM are still missing. This might be due to the low frequency of those specific TSCM cells making their access limited.
Gattinoni et al. (8) and Cieri et al. (13) identified human CMV (HCMV)–specific TSCM within the TN compartment and demonstrated that most HCMV-tetramer binding TN of HCMV-seropositive donors highly express CD95. However, it remains to be elucidated whether these HCMV-tetramer binding CD95+ TN (defined as TSCM) can exhibit rapid effector T cell functions upon short-term activation, which denotes a hallmark of T cell memory (14).
In this article, we describe the precise phenotype of TSCM specific to viral Ags. Further, we demonstrate that the generation of virus-specific TSCM from the whole-blood naive (CD45RA+CCR7+) T cell pool is feasible and making the procedure more feasible for clinical-grade application. We found that TSCM, defined by the expression of CD45RA+, CCR7+, CD127+, and CD95+, produce effector cytokines after short-term ex vivo virus-antigenic stimulation. In our hands, specifically expanded TSCM differentiated into distinct memory/effector subsets while partially preserving their TSCM phenotype. This indicates self-renewal capacities upon antigenic stimulation and supports our hypothesis that TSCM should be considered a phenotypic and functional distinct population. By further improvement of the virus-specific expansion procedure, we finally developed a protocol that will enable us to rapidly translate the generation of virus-specific TSCM from peripherally collected mixed CD45RA+CCR7+ naive T/TSCM blood lymphocytes to clinical applications.
Materials and Methods
All cell cultures and assays were performed using complete media (RPMI 1640 supplemented with 2 mM l-glutamine, penicillin [100 IU/ml], and streptomycin, all from Biochrom), containing 10% FBS (PAA), in humidified incubators at 37°C and 5% CO2.
For all experiments, blood samples were collected from healthy volunteers. PBMCs were isolated by lymphopreparation density gradient centrifugation (PAA). The Charité ethics committee (Institutional Review Board) approved the study protocol, and all blood donors provided written, informed consent.
T cell isolation and sorting
PBMCs were enriched for CD3+ T cells with the Pan15), and memory CD3+ T cells (TMEM) on BD FACSAriaII SORP (BD Biosciences) populations. Postsorting analysis of purified subsets revealed >98% purity.N, TN-CD95depleted, TSCM (
Cell stimulation and expansion
Sorted T lymphocytes were activated at a ratio of 1:10 by irradiated (30 Gy) CD3-depleted autologous PBMCs pulsed with HCMVpp65/IE1 overlapping peptide-pool, NLVPMVATV (hereafter referred to as NLVP) single peptide (known to be immunodominant in HLA-A2+ donors) in HLA-A2 or EBVEBNA1/EBNA2/EBNA3c/LMP1/LMP2/BLZF overlapping peptide pool at 1 μg/ml each peptide (all JPT Peptide Technologies). Cells were cultured in the presence of recombinant human IL-7 and IL-15 (Cellgenix) at 10 ng/ml each, followed by restimulation at day 7 with freshly isolated peptide-pool pulsed and irradiated CD3-depleted autologous PBMCs at ratio of 1:10. Peptide pools consisted of 15-mer peptides overlapped by 11 aa and were reconstituted in DMSO (16). The medium was changed every 3–4 d during culture.
Allocation of APCs
Expanded T lymphocytes were analyzed for effector functions by their ability to recognize peptide-loaded target cells, which constituted autologous lymphoblastoid B cell lines (LCL) transformed with B95-8 EBV. LCL were generated as described previously (17, 18).
Flow cytometry, Abs, and intracellular staining
TCR next-generation sequencing
CDR3 sequencing was performed on the ImmunoSEQ platform at Adaptive Biotechnologies as previously described by Robins et al. (19) and Sherwood et al. (20). Sequences that did not match CDR3 sequences were removed from the analysis. For further analysis, the standard algorithm as developed by Adaptive Biotechnologies and previously described by Yousfi Monod et al. (21) was used. Genomic DNA was isolated using QIAamp DNA Mini Kit from Qiagen according to manufacturer’s instructions.
Quantitative real-time PCR
RNA of HCMV specifically expanded TMEM and TSCM
Statistical analysis and calculations
GraphPad Prism version 6 was used for graphs and bar charts. To test for normal Gaussian distribution, we performed Kolmogorov–Smirnov test. Normally distributed data were evaluated using the Student t test. If data were normally distributed, Student paired t test was used for analysis. All t tests were two-tailed unless otherwise stated. The p values ≤0.05 were considered statistically significant, and significance is denoted as *p <0.05. Linear regression analysis was performed using GraphPad Prism; correlations were assessed by Pearson and Spearman rank. T cell subset fold expansion expresses T cell subset numbers (assessed by FACS) in relation to total cell numbers of the indicated days post initial stimulation and day 1.
Functional viral-specific CD95+ TSCM can be identified after short-term ex vivo stimulation
Recent studies revealed that most HCMV-multimer binding TN highly express CD95 (8, 13), described as TSCM. To assess whether TSCM can acquire rapid antiviral memory-like effector T cell functionality, we analyzed the HCMV-specific T cell response ex vivo. To this end, human PBMCs were stimulated ex vivo with HCMVpp65/IE1 overlapping peptide pools over 12 h. According to CCR7 and CD45RA expression, T cells were subdivided into four memory subsets: TEMRA, TCM, TEM, and TN (Fig. 1A–D) (7). TSCM can be found within the TN compartment; most importantly, of these, a small subset expresses large amounts of CD95. According to the phenotype published by Gattinoni et al. (8), we defined TSCM as CD45RO−, CD45RA+, CCR7+, CD127+, and CD95+ T cells within both the CD4+ and the CD8+ T cell populations (Fig. 1E).
Upon activation, we found a gradual virus-specific cytokine formation capacity in distinct T cell subsets with the lowest in TN, intermediate in TCM, and the highest in TEM/TEMRA (Fig. 1B–D). Intriguingly, a substantial fraction of the CD8+ TSCM, but only very few CD4+ TSCM produced IFN-γ and TNF-α after HCMV-specific stimuli, and their frequency tended to correlate with that of the average total CD3+ T cell population (Fig. 1E–K). By contrast, TSCM of HCMV-seronegative donors like other memory T cell subsets failed to respond to HCMV-antigenic stimuli (data not shown). Furthermore, ex vivo HCMVNLVP-multimer staining of CD8+ T cells within PBMCs confirmed the intracellular cytokine formation data and revealed similar gradual viral specificity in frequencies within the T cell subsets and, importantly, multimer positivity of CD8+ TSCM (Supplemental Fig. 1A). TN is the most abundant population within CD4+ and CD8+ T cells with negligible capacity to secrete effector cytokines after ex vivo stimulation (Fig. 1B–D). Intriguingly, the absolute number of TSCM responding to HCMV stimuli correlates with that of the few responding T cells within the TN compartment (Fig. 1H, 1K), indicating that most IFN-γ– and TNF-α–producing T cells within the TN compartment are rather Ag-specific TSCM than TN (Fig. 1H, 1K). Further phenotypic analysis of T cell differentiation markers important to define TSCM (8) revealed that CD4+ and CD8+ TSCM express, compared with TN cells, higher levels of LFA-1 (α- and β-chain), IL18Rα, and CXCR3, similar levels of Bcl-2 and CD161, and lower levels of CXCR4, CD31, and CD38 (Fig. 1L; mean fluorescence intensity of each marker is presented in Supplemental Fig. 1B).
In summary, TSCM are phenotypically assigned to the memory compartment and acquire a memory-like effector T cell cytokine formation pattern after Ag rechallenge, allowing detection of functional antiviral-specific TSCM by their effector cytokine production after short-term ex vivo stimulation.
Viral-specific TSCM proliferate, show effector T cell functionality, differentiate, and self-renew on antigenic stimuli
To investigate whether HCMV-specific TSCM can be isolated and expanded from peripherally collected PBMCs, we performed sequential cell sorting followed by HCMV-specific T cell activation. To this end, CD3+ T cells were enriched from freshly isolated PBMCs by magnetic bead sorting. Next, TSCM (CD3+CD45RA+CCR7+CD95+) and TMEM (CD3+CD45RO+) subsets were further purified by polychromatic FACS (FACSorting) (Fig. 2A, 2B) (15). The average yield of TSCM reached ∼1% from presorted CD3+ T cells (3.3 × 105 ± 1.5 × 105 TSCM per 3 × 107 total T cells; Fig. 2C). Next, TSCM and TMEM were stimulated with autologous, lethally irradiated CD3-depleted PBMCs pulsed with HCMVpp65/IE1 peptide pool in the presence of low-dose IL-7 and IL-15 (Fig. 3A) (13).
Cell cultures were assessed for specific expansion of TMEM-derived [T(TMEM)] and TSCM-derived [T(TSCM)] cells. Both subsets showed rapid expansion with T(TSCM) resulting in a higher and more robust yield than T(TMEM) (Fig. 3B, 3C). Interestingly, these culture conditions induced predominant proliferation of CD8+ T cells, whereas CD4+ T cell expansion was essentially absent (Fig. 3D; data not shown for CD4+ T cells). The expression of CD95, a hallmark for memory T cells (14), was detectable on every subset analyzed at any time point measured during the course of expansion (data not shown). We confirmed presence of the TSCM population in all healthy donors measured in this study (mean = 3.14% of naive-like CD3+ T cells, SD 1.21; Fig. 2C).
We next assessed the effector cytokine response of expanded TMEM and TSCM upon HCMV-specific activation. Intriguingly, specific HCMV activation induced IFN-γ and TNF-α cytokine formation in both expanded CD8+ TMEM and expanded CD8+ TSCM (Fig. 3E, 3F). Consequently, owing to the higher overall fold expansion rate of T(TSCM) compared with T(TMEM) (Fig. 3B), HCMV-specific TSCM expansion was significantly higher (Fig. 3H, 3J). Furthermore, HCMV specifically expanded CD8+ TMEM and expanded CD8+ TSCM were capable of becoming cytotoxic effectors expressing Granzyme B (Fig. 3G). Similar results were observed when this protocol was used for the expansion of EBV-specific TSCM. T(TSCM) expanded with EBV-antigenic stimuli produce effector cytokines and upregulate activation markers after EBV stimulation (Supplemental Fig. 1C). To confirm Ag specificity, we stimulated HCMV specifically expanded T(TSCM) with unrelated antigenic stimuli (EBV Ags). As assessed by FACS analysis, activation by non-HCMV Ags did not induce cytokine formation in HCMV specifically expanded T(TSCM) (Supplemental Fig. 1D).
Specific HCMV activation induced effector cytokine formation in both expanded TMEM and TSCM. However, it remains unclear whether both subsets show overlapping TCR usage indicating comparable HCMV epitope specificities. To this end, we performed TCR repertoire analyses by next generation sequencing to assess the clonal overlap between TMEM and TSCM. These two differentially expanded cell products of three different donors (HCMVpp65/IE1 peptide-pool specifically expanded TMEM and TSCM) were analyzed. Intriguingly, comparison of unique nucleotide sequences revealed a high clonal overlap between the two populations within the same donor (0.436–0.721). In contrast, comparison of clonal repertoires between different individuals showed almost no overlap (0.001–0.072), confirming the specificity of the findings (Fig. 3K). In conclusion, expanded HCMV-specific TMEM and TSCM share strongly the TCR repertoire usage indicating the majority of those cells is derived from the same mother clones with identical viral epitope specificity.
During the first expansion phase, T(TSCM) expression of CD45RA and CCR7 decreased while CD45RO expression was upregulated, a phenotype traditionally associated with memory T cells (Fig. 3I). Notably, expanded TSCM reacquired CD45RA surface expression after 28 d of expansion, which further increased over time in frequency (Fig. 3I, 3L) and in absolute numbers (Fig. 3M). At day 35, a substantial percentage of the CD45RA+ T(TSCM) subset displayed CCR7 expression (Fig. 3I, 3N). The CD45RA− T cells expanded from TSCM, however, were negative for CCR7 (data not shown). Moreover, a considerable proportion of the expanded TSCM, but not of the expanded TMEM, expressed a TCM phenotype after 35 d of culture (Fig. 3O). We further aimed to show HCMVNLVP-multimer staining of CD8+ T(TSCM) and T(TMEM) to demonstrate the changes in the phenotypic markers over time in culture. Likewise, a substantial percentage of the multimer+ T(TSCM) subset displayed CCR7 expression at day 35, whereas multimer+ T(MEM) were negative for CCR7 (Supplemental Fig. 2A). Interestingly, further cultivation of FACS-separated expanded CD45RA+CD45RO− and CD45RA−CD45RO+ T(TSCM) on day 35 revealed a stable phenotypic differentiation pattern over the period of expansion up to 56 d after initial culture (Supplemental Fig. 2B, 2C).
This selective ability of progeny from expanded TSCM to revert to CD45RA expression is intriguing. Gattinoni and colleagues (8) noticed a low level of HNRPLL mRNA expression encoding heterogeneous nuclear ribonucleoprotein L-like, a key regulator of the alternative splicing of the CD45 pre-mRNA required for efficient CD45RO expression in TSCM (22). In this regard, we compared the mRNA expression of this gene and related differentially expressed genes of T(TSCM) and T(TMEM) (8). To this end, we performed quantitative real-time PCR to delineate naive-associated gene expression and effector-associated gene expression in T(TSCM) and T(TMEM) at distinct time points after initial stimulation. As expected, T(TSCM) expressed lower transcript levels of HNRPLL compared with T(TMEM) in the course of expansion (Fig. 4A). We further evaluated CD45RA gene expression by analyzing PTPRC. To this end, we used an assay spanning exons 3 and 4, whereas the latter encodes for section A and is being removed in all other CD45 splice variants and therefore is specific for the CD45 isoform RA (23). Strikingly, beginning at day 28, CD45RA (PTPRC) mRNA expression was increased in T(TSCM), but not in T(TMEM) (Fig. 4B). In line with the capacity of TSCM to self-renew, KLRG1 as a marker of T cell senescence (24) was decreased in T(TSCM) compared with T(TMEM) (Fig. 4C). Conversely, the expression of transcription factors that inhibit T cell activation and differentiation such as ceramide synthase 6, which promotes cellular quiescence by regulating intracellular ceramide levels (25), was elevated in T(TMEM) (Fig. 4D). mRNA expression of the effector-associated genes encoding for cytotoxic molecules like Perforin 1 and Granzyme A was similar in T(TSCM) and T(TMEM) until day 35 after expansion; however, it increased in T(TMEM) on day 42, indicating their terminal differentiation (Fig. 4E, 4F). Actinin α 1 and Lymphoid Enhancer–binding Factor 1 as naive-associated genes (8, 26) were not differentially expressed in either probes in the course of expansion (data not shown).
The expression of the IL-7Rα (CD127) on Ag-specific T cells is associated with T cell engraftment after in vivo adoptive transfer experiments, and therefore is a marker for T cell fitness and persistence (27). To further define the phenotype of expanded TMEM and TSCM, we assessed CD127 expression of both subsets after culturing (Fig. 4G–I). Intriguingly, expanded TSCM exhibited profound CD127 expression compared with TMEM (Fig. 4G, 4H). Further dissection of expanded CD45RA+ and CD45RO+ TSCM revealed higher CD127 expression on the former subset (Fig. 4I).
Importantly, weekly antigenic restimulation resulted in TSCM, but not in TMEM, in a continuous expansion of >10 wk without signs of culture contraction. Furthermore, the resting period was indispensable for the expanded TSCM to reacquire their initial phenotype (data not shown).
Altogether, the data suggest that a significant part of expanded viral-specific TSCM, in contrast with conventional defined viral-specific memory T cells, keep a cell-surface marker profile, which is similar to the phenotype ex vivo, indicating a potent self-renewal potential on antigenic stimuli. Moreover, it appears that Ag-expanded TSCM become able to differentiate into all T cell lineages in vitro.
Functional viral-specific TSCM can be expanded from CD95+ TN
Phenotypically, TSCM are assigned to the TN compartment. To facilitate ex vivo expansion of HCMV-specific TSCM for translation to clinical protocols, we aimed to expand these cells from a mixed TN compartment. Moreover, we investigated whether CD95 expression on TN allows discrimination between Ag-specific TSCM and Ag-inexperienced truly naive T cells. Accordingly, we sort-purified CD45RA+CCR7+ and CD45RA+CCR7+CD95depleted TN by FACS to a purity >98% (Fig. 5A, 5B). In parallel, TSCM were FACSorted from the same donors. CD45RA+CCR7+ TN, CD45RA+CCR7+CD95depleted TN, and CD95+ TSCM were activated by an initial stimulation with autologous, lethally radiated HCMVpp65/IE1 peptide-pool pulsed CD3-depleted PBMCs and low doses of IL-7 and IL-15 (Fig. 5A). We then defined the expansion potential of the naive T cell–derived [T(TN)], expanded naive CD95depleted T cell–derived [T(TN-CD95–)], and T(TSCM) populations. All subsets were sensitive to the peptide-pool and cytokine-mediated stimuli, with enriched TSCM expanding more robustly than both TN populations (Fig. 5C, 5D). Expanded TN-CD95– showed a significantly reduced expansion potential compared with total TN (Fig. 5C, 5D). In line with the observed CD8 prevalence in Ag specifically expanded TSCM, these culture conditions predominantly induced proliferation of TN and TN-CD95– CD8+ T cells, but less CD4+ T cells (data not shown). Intriguingly, depletion of CD95 from CD8+ TN before Ag-driven expansion abolished specific effector T cell cytokine formation (IFN-γ and TNF-α) and upregulation of CD137 upon HCMV-specific activation (Fig. 5E, 5F). Owing to the higher overall fold expansion rate of T(TN) and T(TSCM) compared with T(TN-CD95–) (Fig. 5C), absolute numbers were increased in HCMV-specific T(TN) and T(TSCM) (Fig. 5G). HCMV-specific T(TN) and T(TSCM) expansion was significantly higher compared with T(TN-CD95–) (Fig. 5H).
Finally, we tested whether the culture conditions can induce naive T cell priming of HCMV-seronegative donors. We activated CD45RA+CCR7+ naive T cells and PBMCs of an HCMV-seronegative donor with the same experimental approach. Both naive T cells and PBMCs failed to induce cytokine production after HCMV-specific stimuli, excluding the possibility of naive T cell priming (Supplemental Fig. 3). Hence we reason that functional HCMV-specific TSCM can be expanded from the “mixed” TN compartment containing both naive and TSCM cells.
The presence and origin of long-lived memory T cells with stem cell–like abilities have been deliberated for some time (8, 28, 29). In humans, TSCM have been recently identified (8). Phenotypically, TSCM can be found within the TN compartment, where they were not identified for a long time. TSCM are defined to exhibit the phenotypic profile CD45RO−CD45RA+CCR7+CD62L+CD28+CD127+CD95+ and IL-2Rβ+ (8). Another study deemed the characterization of TSCM to be adequate defined by the incremental gating of CD45RO−→CD45RA+→CD127+→CCR7+→CD95+ (15). Their distinct functionality and unique homeostatic properties could be presented in a subsequent study in a nonhuman primate animal model during the course of an SIV infection (30). Inter alia, those studies present HCMV-tetramer binding TSCM lymphocytes; however, functional data of HCMV-specific TSCM are missing (8, 13). In this study, we could demonstrate that T cells phenotypically resembling TSCM (8, 15) can acquire a memory-like effector T cell cytokine formation pattern after short-term viral antigenic stimuli. Hanley and colleagues (31) could demonstrate that Ag-specific T cells targeting viral epitopes could be primed and expanded from cord blood T cells with a naive phenotype. Notwithstanding TSCM being phenotypically TN, rapid effector functionality evidences that TSCM are truly Ag-experienced memory T cells (32). Notably, TSCM lymphocytes of HCMV-seronegative donors with also no detectable memory T cells specific to HCMV epitopes failed to secrete effector cytokines in response to HCMV-antigenic stimuli. Importantly, Gattinoni and colleagues (8, 26) generated TSCM from TN in the presence of the glycogen synthase kinase-3β inhibiting the Wnt–β-catenin pathway. However, we aimed to isolate TSCM naturally generated in vivo according to their specific surface marker profile and expand viral Ag-specific T cells from this population without further addition of GSK-3β inhibitor because we do not intend to generate TSCM from TN, but rather expand ex vivo native TSCM in an Ag-specific manner.
We could detect Ag-specific TSCM lymphocytes in the peripheral blood of healthy humans that permits access to TSCM lymphocytes specific for viral Ags for analysis and adoptive transfer. We successfully generated functional virus-specific CD8+ T cell lines from FACSort enriched TSCM lymphocytes. Expanded TSCM lines were characterized by their differentiation capacity into distinct memory/effector T cell subsets, robust effector functions, high expansion potency, and self-renewal abilities in vitro. Importantly, we have found a high TCR repertoire overlap between HCMV-expanded T(TMEM) and T(TSCM), implying their closed relationship at the clonal level.
Remarkably, CD4+ TSCM cell expansion and cytokine production after specific stimuli were essentially absent. Hence the phenotypic marker profile of CD4+ TSCM may not be defined yet.
Expanded TSCM show divergent surface marker expression of T cell differentiation over the duration of expansion, characterized by high expression of CD45RA, CCR7, CD127, and CD95 at culture initiation, followed by the acquisition of CD45RO accompanied by the loss of CD45RA, CCR7, and CD127 with proliferation. Intriguingly, a substantial amount of expanded TSCM reacquired CD45RA, CCR7, and CD127 surface expression after 4 wk of expansion, implying self-renewal potency (33). Importantly, expanded virus-specific TMEM failed to upregulate CD45RA, CCR7, and CD127 after expansion. We assessed the status of CD45RO (HNRPLL) mRNA expression and related differentially expressed genes in virus specifically expanded TSCM and TMEM. Noticeably, TSCM upregulate CD45RA mRNA expression and failed to upregulate CD45RO mRNA expression in the course of expansion, indicating cell-intrinsic transcriptional programs of self-renewal. Furthermore, the expression of transcription factors that inhibit T cell activation and differentiation such as ceramide synthase 6 and a decreased mRNA expression of KLRG1 as a marker of T cell senescence indicate that TSCM should be considered a phenotypic and functional distinct population. Conferring Cieri and colleagues (13), in the first 2 wk of expansion, we could detect CD45RA/CD45RO double positivity on TSCM, however, not on TMEM. This unique feature of TSCM being able to express both CD45 splice variant needs further elucidation and cannot be answered by our studies.
Evidently, TMEM have a selective advantage over TSCM competing for instructive factors that is guiding their formation and expansion in vitro. Furthermore, the early reacquisition of CD127 expression of TSCM after antigenic stimuli may give selective advantage for susceptibility to IL-7 to promote TSCM outgrowth. Although IL-7 is not limited in our system, specific proliferation of TSCM from the bulk T cell pool with the addition of IL-7 could not be observed in this study. This observation is in agreement with Cieri et al. (13) reporting that the depletion of CD45RO+ memory T cells or the isolation of naive T cells before initial stimulation is indispensable for the enrichment of TSCM-like lymphocytes in vitro. The IL-7/CD127 interplay appears to be a critical determinant for TSCM generation and maintenance from TN (34). However, the detailed cytokine requirements for Ag-specific TSCM expansion have yet to be elucidated.
Our culture conditions induced expanded virus-specific TSCM that become all T cell lineages at different time points of expansion. Interestingly, we observed a divergent T cell differentiation pattern of expanded TSCM separated into CD45RA+CD45RO− and CD45RA−CD45RO+ T cells after 5 wk of expansion. The reacquisition after loss of CD45RA in CD45RA+CD45RO− T(TSCM) resulted in further contraction and expression of CCR7 to finally reacquire the phenotype of TSCM ex vivo, whereas T cells derived from CD45RA−CD45RO+ T(TSCM) retained their phenotype. This observation supports the model of an asymmetric division to a diverse T cell population in agreement with other reports (29, 35).
We showed that virus-specific TSCM can be obtained by the enrichment and expansion of FACSorted TSCM. For an effective translation into clinical trials, the procedure of Ag-specific TSCM enrichment and expansion has to be easy and good manufacturing practice compliant. We could demonstrate the expansion of functional virus-specific CD8+ T cell lines from the TN-like compartment, which make the process more feasible and revealed higher yield. Intriguingly, the depletion of CD95+ cells from TN compartment obliterated the expansion of viral-specific T cells, giving strong evidence that the resulting T cell population is indeed derived from the TSCM compartment (8, 13). Furthermore, this approach appears good manufacturing practice feasible using current state-of-the-art techniques because naive T cells comprising TSCM can be enriched by clinical-grade untouched magnetic separation (13, 36). Moreover, the growth factors IL-7 and IL-15 that we apply in our protocol for antiviral TSCM expansion that are certainly decisive for their expansion are being used in approved clinical protocols (11, 37, 38). Besides, CD3-depleted PBMCs as APCs were sufficient for the expansion of TSCM, indicating that the Ag presentation requirements are fulfilled and that an enrichment or the generation of professional APCs like dendritic cells is not necessary per se.
Novel cancer-eradicating therapeutic approaches have been developed in the recent years (39–41). T cells genetically modified by the incorporation of engineered receptors directed against cancer Ags have emerged to be a significant tool to combat cancer in first clinical trials (41, 42). The use of viral-specific TSCM or TN comprising TSCM might be particularly attractive to be applied in adoptive immunotherapeutic approaches to cancer patients by incorporating chimeric Ag receptors (CARs) to direct TSCM against defined tumor Ags (41, 43). Virus-specific TCM have been selectively transduced with a CAR directed against the CD19 Ag to facilitate their use in clinical trials to examine their superiority compared with other T cell subsets (44). TSCM or TN comprising TSCM appear to be promising candidates to be tested in similar clinical protocols to define their therapeutic potential in clinical trials targeting cancer. Specifically, viral-specific TSCM or TN comprising TSCM enriched to oligoclonality combined with the genetic incorporation of a CAR may be superior to currently used protocols to be better suited to function in the tumor environment by coupling anticancer efficacy with T cell persistence by inducing response to latent viral infections, which may be boosted by vaccination (13). To this end, our findings enable direct access to human Ag-specific TSCM from PBLs targeting viral Ags and pave the way for rapid broad clinical application.
M.S.-H., P.R., and H.-D.V. have a patent pending on the expansion protocol (Viral-antigen specific memory stem T cell preparations for adoptive immunotherapy, European patent application EP14196519.4, 2014). The other authors have no financial conflicts of interest.
We thank Dr. Jason Millward (Institute of Medical Immunology, Charité University Medicine Berlin) for statistical advice. We would like to acknowledge the assistance of the Berlin-Brandenburg Center for Regenerative Therapies Flow Cytometry Core Lab, Dr. D. Kunkel, A. Maluck, and Dr. S. Meier. We thank Dr. B. Sawitzki (Institute of Medical Immunology, Charité University Medicine Berlin) for critical discussions and reading of the manuscript, M. Streitz (Institute of Medical Immunology, Charité University Medicine Berlin) for technical flow-cytometry assistance, and Dr. Chantip Dang-Heine for technical advice regarding preparations of DNA/RNA.
The work was supported by the Deutsche Forschungsgemeinschaft (DFG-SFB-TR36-project A3) and the German Federal Ministry of Education and Research (Berlin-Brandenburg Center for Regenerative Therapies grant).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- chimeric Ag receptor
- forward scatter
- human CMV
- lymphoblastoid B cell line
- central memory T cells
- effector memory T cells
- terminally differentiated effector T cells
- memory CD3+ T cell
- naive-like T cell
- memory stem T cells
- TMEM-derived cells
- naive T cell–derived
- naive CD95depleted T cell–derived
- TSCM-derived cells.
- Received August 13, 2014.
- Accepted March 27, 2015.
- Copyright © 2015 by The American Association of Immunologists, Inc.