Human CD8 T cells generated in vitro from hematopoietic stem cells are functionally mature

Background T cell development occurs within the highly specialized thymus. Cytotoxic CD8 T cells are critical in adaptive immunity by targeting virally infected or tumor cells. In this study, we addressed whether functional CD8 T cells can be generated fully in vitro using human umbilical cord blood (UCB) hematopoietic stem cells (HSCs) in coculture with OP9-DL1 cells. Results HSC/OP9-DL1 cocultures supported the differentiation of CD8 T cells, which were TCR/CD3hi CD27hi CD1aneg and thus phenotypically resembled mature functional CD8 single positive thymocytes. These in vitro-generated T cells also appeared to be conventional CD8 cells, as they expressed high levels of Eomes and low levels of Plzf, albeit not identical to ex vivo UCB CD8 T cells. Consistent with the phenotypic and molecular characterization, upon TCR-stimulation, in vitro-generated CD8 T cells proliferated, expressed activation markers (MHC-II, CD25, CD38), secreted IFN-γ and expressed Granzyme B, a cytotoxic T-cell effector molecule. Conclusion Taken together, the ability to direct human hematopoietic stem cell or T-progenitor cells towards a mature functional phenotype raises the possibility of establishing cell-based treatments for T-immunodeficiencies by rapidly restoring CD8 effector function, thereby mitigating the risks associated with opportunistic infections.


Background
T lymphocytes develop within the thymus, undergoing distinct differentiation events based on phenotype, while following spatial-temporal cues. The most immature human thymocytes are triple negative for CD3, CD4 and CD8, but express CD34. CD7 is one of the earliest markers to be expressed during human T cell ontogeny, followed by the upregulation of CD1a, which marks T cell commitment. Following this stage, human thymocytes progress to a CD4 immature single positive (CD4ISP) stage, at which point CD4 is expressed in the absence of CD8. Thereafter, a subset of CD4ISP cells are thought to complete TCRβ rearrangement leading to β-selection and differentiation to the CD4 + CD8 + double positive (DP) stage. Finally, following successful TCRα rearrangement, TCRαβ expressing DP thymocytes undergo positive and negative selection events, which result in the production of CD4 + CD8and CD4 -CD8 + single positive (SP) T cells, which then emigrate to the periphery [1].
Cytotoxic CD8 SP T cells play a critical role in adaptive immunity by destroying virally infected or malignant cells. Cytoablative regimens to treat malignancies or HIV infection results in the destruction of T cells, leading to a period of immunodeficiency and a concomitant increase in opportunistic infections [2]. While hematopoietic stem cell (HSC) transplants can reconstitute the entire immune system, T cells are the slowest population to recover or may never recover at all [3,4]. The robust differentiation of multiple sources of HSCs [5][6][7][8][9] into T-progenitors using the OP9-DL1 coculture system suggests that these progenitor cells could be harvested in vitro to rapidly restore the T cell compartment in vivo and thereby reduce the incidence of opportunistic infection and ultimately improve therapeutic outcomes [10]. While we have demonstrated that specific populations of T-progenitor cells generated in vitro can reconstitute the thymus of immunodeficient mice and generate CD8 single positive (SP) T cells in vivo [8], it was not clear whether functional maturation of mature CD8 T cells could take place entirely in vitro starting from umbilical cord blood (UCB) HSCs in OP9-DL1 cell cocultures.
Mature CD8 T cells were recently reported to be generated following coculture of early thymocyte progenitors (ETPs) on OP9-DL1 monolayers [11]. The absence of a clinically viable source from which to isolate low frequency ETPs, makes it difficult to envision a therapeutic approach to treat T-immunodeficiency based on harvested postnatal thymocytes (PNT). Therefore, it is important to determine whether accessible sources of HSCs, such as umbilical cord blood, directed to differentiate in vitro can generate functionally mature CD8 T cells. Two elegant studies have reported the generation of cytotoxic human T cells from HSCs genetically modified to express tumor-specific TCRs cocultured with OP9-DL1 stromal cells [12,13]. Whether in vitrogenerated CD8 T cells derived from unmanipulated HSCs could also give rise to mature functional T cells and respond to signals delivered through an endogenously generated TCR remained to be tested. Here we demonstrate that long-term coculture of non-transduced human UCB HSCs on OP9-DL1 cells generates αβ-TCR/CD3 + CD8 + SP T cells that exhibit the molecular and cellular signatures corresponding to functionally mature T cells.

Phenotypic characterization of human CD8 SP T-cells derived from HSCs cultured with OP9-DL1 cells
We have previously reported the ability to generate CD4 + CD8 + T-lineage cells as well as CD4 and CD8 single positive (SP) cells from human cord blood-derived CD34 + CD38 -/low HSC/OP9-DL1 cocultures [7], however the maturational, molecular and functional status of these CD8 SP T-cells were not assessed.
To address this, we analyzed long-term cocultures that were allowed to differentiate for an extended period of time. Figure 1 shows the presence of both DP and SP subsets from a day 65 coculture. Examination of multiple cocultures demonstrated the generation of CD8 SP T cells (5.3% ± 1.2% (mean ± SE, n = 6)) after 60-70 days of culture. Additionally, Table 1 displays an analysis for the number of CD8 T cells obtained from individual OP9-DL1 cultures initiated with different numbers of HSCs, which provides an indication of the expected CD8 SP cell yields per input HSC. We further examined the CD8 SP subset present in these cultures for the expression of cell surface markers CD3 and CD27 [14,15], typically expressed on mature T-cells. Amongst the SP CD8 cells (SP8s) found in late cocultures, about 50-60% (Figures 1 for 50% and Figures 3a for 60%) expressed CD3/αβTCR. Of note, the majority (57% ± 10%, n = 3) of CD3 + SP8s were found to co-express CD27. In addition, CD27 + CD3 + CD8 + SPs were also found to lack CD1a expression, which is indicative of functional maturity [16]. This is in contrast to the subset of CD27 -CD3 + CD8 + SPs that continued to express CD1a, which is characteristic of the preceding stage in T-cell differentiation [16] and consistent with reports showing that the acquisition of functional maturity by CD3 + thymocytes following positive selection is associated with the expression of CD27 and down-regulation of CD1a expression [15,16].

Molecular analysis of in vitro derived human CD8 SP T-cells from UCB HSCs
To gain further insight into the phenotypically characterized CD8 SP T cells derived in vitro, we examined the expression of genes known to influence the differentiation or effector status of CD8 T cells by quantitative real-time PCR. We examined the expression of Th-POK (Zbtb7b), Eomes, PLZF (Zbtb16) and Gata-3 transcript levels in SP8 T cells obtained from CD34 + CD38 -/low HSCs cocultured with OP9-DL1 cells for approximately 60-70 days. CD4 + CD8 -CD3 + (Lin + SP4), CD4 -CD8 + CD3 + (Lin + SP8) T cells and CD33 + myeloid cells obtained from the mature lineage-positive fraction of UCB were used as controls. Consistent with the role of the zinc finger transcription factor Th-POK in CD4 lineage-commitment [17], Th-POK transcripts were barely detectable in coculture-derived SP8 T cells compared to Lin + SP4 T cells ( Figure 2). Also, a low, but detectable, level of Th-POK expression was observed in UCB-ex vivo Lin + SP8 T cells. In contrast, the T-box transcription factor Eomes, which is a critical regulator for CD8 T cell differentiation and effector function [18], was highly expressed in coculture-derived SP8 T cells at levels that were lower but comparable to that of peripheral SP8 T cells obtained from Lin + UCB. Interestingly, we also observed a high level of this transcription factor in UCB-derived CD4 T cells, which is consistent with a recent report showing the expression of Eomes in a subset of naïve peripheral CD4 T cells [19]. In addition, the NKT cell/innate T cell transcription factor PLZF [20] was expressed at low levels in HSC/OP9-DL1 coculturederived CD8 T cells similar to UCB CD8 peripheral T cells. Lastly, we observed high levels of Gata-3 transcripts in coculture-derived SP8 T cells, likely owing to the Delta-like-induced signaling of Notch in these cells.

Generation of functional human CD8 T cells from HSCs cultured with OP9-DL1 cells
To address the functional status of in vitro-generated SP8s, we sorted the CD3/TCR-expressing subset ( Figure  3A) and examined whether these cells had the capacity to up/down-regulate downstream differentiation markers, proliferate, express cytolytic effector-function molecules, and secrete γ-interferon (IFNγ) following stimulation with immobilized anti-CD3/anti-CD28 mAbs for 5 days. As shown in Figure 3B, a blast-like appearance based on Forward-light Scatter (FSC) is seen in stimulated (Stim) cells as compared to non-stimulated (NS) cells, which displayed significantly different (p = 0.004) FSC mean values when analyzed over multiple experiments (NS, 460 ± 12; Stim, 524 ± 10). Additionally, stimulated cells up-regulated CD45RO, CD38 and MHCclass II expression, and down-regulated CD27 expression, as compared to non-stimulated cells ( Figure 3C). As shown in Figure 3D, a similar trend was observed when UCB ex vivo CD8 T cells were assayed. However, a more prominent upregulation of CD45RO and CD38, and less prominent downregulation of CD27, was observed. Nevertheless, further analysis of CD45RO MFI and the percentage of CD27 expression ( Figure 3E) demonstrated a significant difference between non-stimulated and stimulated in vitro-derived CD8 T cells. Indeed, this was also observed for UCB-ex vivo CD8 T cells ( Figure 3F).
Furthermore, to address the extent of cellular proliferation induced by TCR-stimulation, sorted CD3 + CD8 + T-cells were loaded with CFSE and incubated for 5 days with or without TCR-stimulation. Figure 4A shows that stimulated cells undergo many rounds of cell division as indicated by the loss of CFSE (left side) compared to non-stimulated cells and by the increase in cellularity (lower right-side). A similar loss of CFSE and increased cellularity was observed in UCB-ex vivo T cells ( Figure  4B, left-side and lower right-side). Like UCB-ex vivo CD8 T cells, we observed that in vitro-derived proliferating cells also displayed marked up-regulation of CD25 expression ( Figure 4A and 4B, left side). The average percentage of CD25 on the cell surface of stimulated cells was 78% ± 5%, which was significantly higher than on non-stimulated cells ( Figure 4A, upper right-side).
To determine whether these TCR-responsive in vitroderived CD8 SP T-lineage cells can be induced to express cytotoxic/effector-function molecules, the expression of Granzyme-B and IFNγ were assessed. Intracellular Granzyme-B expression was detected in approximately 38% of stimulated CD3 + CD8 + T-cells, as compared to non-stimulated cells that expressed much lower levels of Granzyme-B ( Figure 4C). As demonstrated in Figure 4D, when in vitro-derived CD8 T cells were gated on Granzyme-B + cells, these corresponded to CD25 + proliferating cells. Finally, supernatants from wells containing in vitro-generated CD8 SPs were analyzed for the presence of IFNγ following stimulation. As shown in Figure 4E, supernatants from cells stimulated with either 2 μg/mL or 10 μg/mL of anti-CD3/anti-CD28 mAbs showed a significant increase in the amount of IFNγ, as compared to non-stimulated cells. In a separate set of experiments ( Figure 4F), in vitroderived and UCB-ex vivo CD8 T cells were directly compared, and showed similar levels of IFNγ production Table 1 Generation of TCR/CD3 + CD8 + cells from CD34 + CD38 -/low hematopoietic progenitors in OP9-DL1 cocultures Experiment CD34 + CD38 -/low (× 10 5 ) TCR/CD3 + CD8 + (× 10 5 ) a after TCR/CD28 stimulation. Taken together, these results show that in vitro-derived CD8 SP T cells are functionally mature. Lastly, we also noted the appearance of CD3 + CD4 + T-cells ( Figure 1A and 3A), which could be selected by human MHC class II-expressing cells, which may be a T-lineage cell or a non T-lineage cell such as a dendritic cell. Indeed, when late day cultures were analyzed for the presence of MHC class II expressing cells, MHC-II + CD34 + CD7 + progenitor T cells and MHC-II + CD7non-T-lineage cells were detected as shown in Figure  5A. Figure 5B, demonstrates the generation of CD33 + CD7myeloid cells, which could provide MHC-II molecules to developing T cells. Interestingly, in contrast to the CD8 SPs, in vitro-generated CD4 SP T cells do not show the hallmarks of being functionally mature T-cells. Experiments examining their proliferative ability revealed a lack of responsiveness to CD3 stimulation (data not shown). Also confirming these findings was the lack of CD27 expression on CD4 SP T-lineage cells compared to CD8 T cells in the same culture ( Figure 6). Since this molecule is upregulated upon positive selection and present on mature naive T cells, the CD4 SP generated in these cultures may represent transitional cells that require additional differentiation signals that are not readily available in these cultures.

Discussion
The HSC/OP9-DL1 cocultures not only serve to characterize progenitor function or early events in human T-cell development [8], but may also provide a simple method for the generation of functional human T-cells in vitro. This approach may be applicable to cell-based immunotherapies that presently capitalize on T-cell effector-function to induce/enhance anti-tumor eradicating immunity [21]. Thus far, adoptive cell transfer (ACT) for tumor eradication has employed the use of autologous and allogeneic T cells obtained from blood or tumor-infiltrating lymphocytes, followed by expansion protocols to generate millions of cells, with the efficacy of ACT depending on the ability to generate large cell numbers of T cells that maintain effector function. One caveat to stromal-based cultures of HSCs into CD8 T cells is that clinically relevant numbers of CD8 + T cells are not readily obtained (Table 1) following one round of TCR-stimulation. However, one could attempt to overcome this by promoting expansion of CD8 T cells with artificial-APCs and cytokines with several rounds of stimulation [22,23]. Additionally, the number of CD8 T cells obtained in these cultures is directly related to the number of input HSCs; thus, expansion of HSCs prior to differentiation into CD8 + T cells would provide another strategy for the large-scale generation Figure 2 Gene expression analysis of CD8 + SP T cells obtained from HSCs cultured on OP9-DL1 cells. Gene expression analysis by QRT-PCR from coculture-derived CD8 SP T cells (CD8 + CD4 -CD3 + ). CD8 SP, CD4 SP T cells and CD33 + myeloid cells were purified from the Lin + fraction of UCB samples and served as controls. Transcript levels for the indicated genes were normalized to human β-actin, and these data are representative of 3 independent experiments, with the standard error bars shown corresponding to values obtained from triplicate wells within an individual experiment.
of mature T-lineage cells [24,25]. Nonetheless, we now provide clear evidence for the maturation of functionally responsive CD8 T cells generated from human UCB HSCs in coculture with OP9-DL1 cells.
Although some of the gene transcript levels were not identical to UCB-ex vivo CD8 T cells, this may be reflect the fact that coculture-derived T cells are more akin to human post-natal thymocytes, rather than T cells found within the periphery. In addition, some differences in cell surface expression may be due to different temporal kinetics of up/downregulation between in vitro-generated versus UCB-ex vivo CD8 T cells. Recently, van Coppernolle et al. reported the development of CD8 SP αβT cells and large numbers of γδT cells following OP9-DL1 coculture from PNT. While γδT cells can be detected in UCB-derived OP9-DL1 cocultures, we find them at very low frequencies (data not shown). This suggests that the CD34 + PNT, compared to UCB HSCs, have received instructive signals that are Notch independent, which appear to affect specific T-lineage outcomes.
The observation that functionally mature CD8 SP Tlineage cells are generated from HSCs on OP9-DL1 stroma, raises the question of which cell-type, within the cultures, is mediating the MHC-dependent positive selection of this population. It is unlikely that the OP9 cells, which express mouse MHC class I, which is not effectively recognized by human CD8 molecules [26], would supply the required positive selection signals. Rather, a human MHC class I-expressing UCB-derived cell, which may or may not be a T-lineage cell, is likely to be the conveyor of these signals. Although van Coppernolle et al. describe the generation of mature CD8 T cells from postnatal thymocytes cocultured on OP9-DL1, and excluded mouse MHC-I as a mode of selection due to its absence on OP9-DL1 cells, our analysis showed that OP9-DL1 cells are in fact MHC class Iexpressing cells (data not shown), suggesting that this possibility remains available for supporting CD8 positive selection. Indeed, the high transcript levels of Eomes, while low for PLZF, suggest that the in vitro-derived CD8 SP cells are likely conventional T-lineage cells and not cells with innate-like function and phenotype, which are typically the product of T-to-T selection [27], such as NKT cells, γδ-T cells and innate CD8 T cells [28].

Conclusions
Taken together, our findings demonstrate that HSC/ OP9-DL1 cocultures produce functional CD8 SP T-cells, which are CD3 + CD27 + CD1a -, and following TCR-stimulation up-regulate CD25, CD45RO, CD38 and MHC class II, while down-regulating CD27 expression. This complex phenotype is unique to activated human T cells [29,30] and consistent with full effector maturation and effective cytolytic capability [31,32], as seen with the induced expression of Granzyme-B and IFNγ
Quantitative real-time reverse-transcriptase polymerase chain reaction (QRT-PCR) Total RNA was isolated in Trizol-reagent and reverse transcribed using Superscript-III and Oligo(dT) 12-18 primers (Invitrogen, Burlington, ON). Diluted cDNA samples from sorted coculture-derived CD8 + and from the UCB Lin + fraction that was sorted for CD4 -CD8 + CD3 + T cells, CD4 + CD8 -CD3 + T cells and CD33 + cells were used as templates for QRT-PCR reactions. Detection of the QRT-PCR was performed with the SYBR Green PCR master mix according to manufacturer's instructions (Qiagen, Mississauga, ON or Bio-Rad, Hercules