Human proT-cells generated in vitro facilitate hematopoietic stem cell-derived T-lymphopoiesis in vivo and restore thymic architecture

Hematopoietic stem cells (HSCs), which reside in the bone marrow (BM), are the ultimate source of all blood cell lineages. Although most hematopoietic lineages develop in the BM, T-cells are unique in that they must complete their development in the thymus. The thymus is composed of a 3-dimensional network of epithelial cells organized into distinct thymic epithelial cell (TEC) compartments with cortical TECs and medullary TECs located on the outer and inner thymus, respectively.¹  These distinct stromal niches provide specialized environmental cues that support thymocytes undergoing different stages of maturation.^2,3 

Under steady state conditions, the thymus does not contain a self-renewing cell, and the production of T-cells throughout life must be maintained by the continued recruitment of blood-borne progenitors arriving from the BM.^4,5  Human HSCs are found within the lineage negative (Lin⁻) CD34⁺CD38⁻ compartment. Although HSCs do not directly seed the thymus,⁶  CD34⁺ cells are present in human thymus. CD7 is one of the earliest markers to appear during human T-cell ontogeny and analysis of human fetal and postnatal organs by Haddad et al^7,8  showed that an early T-lineage progenitor in humans corresponds to a CD34⁺CD45RA⁺CD7⁺ population. A separate study by Hao et al⁹  has demonstrated that a rare population of Lin⁻CD34⁺CD7⁻ thymocytes is detectable and thus may correspond to an earlier intrathymic progenitor. However, the exact nature of the thymus-seeding cell arriving from the BM remains unknown.¹⁰  Nevertheless, the presence of a Lin⁻CD34⁺CD38⁺CD10⁺CD45RA⁺CD7^−/low common lymphoid progenitor population in human BM with T-cell potential has been described,¹¹  as well as an umbilical cord blood-derived CD34⁺CD38⁻CD7⁺ common lymphoid progenitor population¹²  and a CD34⁺CD45RA^hiCD7⁺ population from UCB and fetal BM possessing T-cell potential.^7,8  Thus, candidate populations with T-cell potential have been reported from UCB and BM.

HSC transplantation (HSCT) is a mainstay for the treatment of hematologic malignancies, however, an extended delay in immune-reconstitution after transplantation leads to cases of morbidity and mortality and remains a clinical challenge.¹³  The increased susceptibility to opportunistic infections is specifically caused by poor T-cell recovery and the absence of de novo T-cell generation from HSC-derived progenitors found in the BM.^14,15  Thus, the development of strategies to enhance T-lymphopoiesis remains an important task.¹⁶  One such strategy is the adoptive transfer of T-cell precursors to rapidly restore the T-cell compartment and T-cell–mediated immunity after HSCT. Zakrzewski et al¹⁷  demonstrated that infusion of in vitro–derived mouse CD4⁻CD8⁻ double negative (DN) cells together with HSCs led increased T-cell reconstitution in both the thymus and periphery of mice that had undergone allogeneic HSCT. More importantly, in vitro–derived T-cells exhibited early graft-versus-tumor activity in HSCT recipients.

Previous work from our laboratory and others^18,19  showed the generation of cells with a CD34⁺CD45RA⁺CD7⁺⁺ thymus-colonizing phenotype derived from CD34⁺ UCB HSCs.¹⁸  These cells fulfilled the properties of a progenitor T (proT)-cell, as they were able to home to, settle, and differentiate in the thymus of recipient nonobese diabetic (NOD)/severe combined immunodeficiency (SCID)/γc^null (NSG) and RAG2^−/−γc^−/− mice. It was unclear, however, whether distinct subsets within the CD34⁺CD45RA⁺CD7⁺⁺ population exhibited differential engraftment capacity in vivo. Additionally, it remained to be tested whether in vitro–derived proT-cells were capable of promoting human HSC-derived T-lymphopoiesis after HSCT.

In this study, we evaluated the efficiency of 2 distinct in vitro–generated proT subsets (CD5⁻ proT1 and CD5⁺ proT2 CD34⁺CD45RA⁺CD7⁺⁺) to engraft the thymus of NSG mice. Our findings reveal that both subsets colonize the thymus of immunodeficient mice. However, through competitive repopulation assays, we show that proT2 cells have an advantage in thymus reconstitution compared with the proT1 subset. Our findings also revealed that in vitro–generated proT2-cells are capable of enhancing thymus reconstitution from HSC-derived progenitors after transplantation into irradiated NSG mice. Recipients receiving both HSC and proT2-cells exhibited accelerated HSC-derived chimerism and higher thymus cellularity compared with HSC-only–injected mice. Finally, we addressed a potential mechanism by which proT2 cells were able to promote earlier T-cell reconstitution from BM-derived progenitors by showing that in vitro–generated proT-cells restored the thymic architecture in NSG mice.

Taken together, these findings support the use of OP9-DL cells for the generation of human proT-cells with the potential for adoptive cell transfer in the clinical setting during periods of immunodeficiency after HSCT.

Human UCB samples were obtained and HSC containing fractions were purified as previously described¹⁸  (details are provided in the supplemental Methods, available on the Blood Web site). Sorted human HSCs (Lin⁻ CD34⁺ CD38⁻) were >99% pure, as determined by post-sort analysis. Cord blood samples were collected by following approved guidelines established by the Research Ethics Board of Sunnybrook Health Sciences Centre.

NOD.cg-Prkdc^scidIL2rg^tm/Wjl/Sz (NSG) mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed and bred in a pathogen-free facility. All animal procedures were approved by the Sunnybrook Health Science Centre Animal Care Committee.

OP9-DL1 cell²⁰  cocultures were performed as previously described,¹⁸  and details are provided in the supplemental Methods.

Fluorophore-conjugated antibodies were commercially purchased (BD Biosciences or eBioscience), and used as previously described.^18,21-23  Details are included in the supplemental Methods.

Lin⁻ CD34⁺CD38^−/lo HSCs from an HLA-A2⁺–typed UCB and HLA-A2⁻ CD34⁺CD7⁺⁺CD5⁺ proT2-cells sorted from day 10 to 12 of HSC/OP9DL1 coculture were resuspended in a mixture containing recombinant human interleukin 7 (rhIL-7) (2.5 μg) and an IL-7 antibody M25 (0.5 μg) (provided by Dr C. Surh, Scripps Research Institute, La Jolla, CA) and 0.2 × 10⁵ (HSCs): 2 × 10⁵ (proT2)-cells injected (30 μL/mouse) intrahepatically into irradiated (120 rads) 2- to 5-day-old neonates. As controls, NSG mice were injected with either HSCs or proT2-cells alone. Mice were boosted with an IL-7/M25 mixture every 5 days. Thymus, spleen, and BM were harvested at either 3, 6, 9, or 12 weeks after intrahepatic transplant.

For competitive transfer experiments, CD34⁺CD38^−/lo HSCs from an HLA-A2⁻-typed and HLA-A2⁺-typed CB was sorted and placed into separate (HLA-A2⁻ vs HLA-A2⁺) and individual OP9-DL1 cell cocultures and grown for 10-12 days, as described above. Cocultures were sorted for CD34⁺CD7⁺⁺CD5⁻ proT1 and CD34⁺CD7⁺⁺CD5⁻ proT2 cells and 1 × 10⁵ of each subset were coinjected as a mixture with IL-7/M25 into 2-5 day old neonates, as indicated above.

Preparation of complementary DNA samples from thymuses, amplification conditions, and primers used are detailed in the supplemental Methods.

Preparation of thymuses sections and staining procedures are detailed in the supplemental Methods.

The data and error bars are presented as mean ± standard error of the mean. To determine statistical significance, a two-tailed unpaired Student t test was used using Prism software. Statistical significance was determined as P < .05.

We have previously reported that 2 in vitro–derived subsets, CD34⁺CD7⁺⁺CD5⁻ (proT1) and CD34⁺CD7⁺⁺CD5⁺ (proT2), generated from human HSC/OP9-DL1 cocultures display a differential capability to engraft mouse fetal thymus lobes in vitro.¹⁸  Whether they possessed a differential ability to home and reconstitute the thymus in vivo remains to be assessed. To address this, we injected NSG neonatal mice with either proT1- or proT2-cells derived from a day 10 coculture (Figure 1A) and analyzed them for thymus engraftment. Figure 1B shows a representative flow cytometric profile of human CD45-gated CD4, CD8, and CD3 cell surface expression in the thymus of mice injected with proT1- or proT2-cells. Although the proT1-injected mouse possessed a similar percentage of CD4⁺CD8⁺ double positive (DP) T-cells (93%) compared with proT2-cells (96%), they had lower levels of CD3 expression and only 6.9% of thymocytes expressed high levels of CD3 compared with 20% in the proT2-injected mice. Thus, the more advanced differentiation state of proT2-cells at the time of injection is highlighted by the more mature status of cells in the thymus of the proT2-injected mouse compared with the proT1-injected mouse. When decreasing numbers of each subset were injected into separate mice, proT1- and proT2-cells displayed a similar frequency of thymus-engraftment (Table 1 ). Engraftment was still observed for both subsets when as low as 1 × 10⁴ cells/mouse were injected, however, a failure to engraft was seen with administration of 0.5 × 10⁴ cells/mouse.

To more clearly determine whether one subset or another exhibited better thymus engraftment capacity, proT1- and proT2-cells were placed in a competitive repopulation assay, and the cell of origin traced based on differences in the expression of the major histocompatibility complex (MHC) class I molecule HLA-A2. In vitro–generated HLA-A2⁻ proT1- and HLA-A2⁺ proT2-cells were injected in a 1:1 ratio (1 × 10⁵ of each subset) into neonatal mice and analyzed 17 days after injection. Figure 2A shows the presence of proT1- and proT2-derived cells in the thymus, as indicated by CD45⁺HLA-A2⁻ (5.5%) and CD45⁺HLA-A2⁺ (34.5%) expression, respectively. When examined, we observed that 93% of proT1-cells acquired CD5 on the cell surface. Furthermore, the thymus of mice injected with proT1-cells also possessed 38% CD4⁺CD8⁻ immature single positive (SP) T-cells and >50% of cells had progressed to the DP stage. In contrast, >80% of cells found within the thymus of the proT2-injected mouse contained CD4⁺CD8⁺ T-cells. More importantly, when the ratio of proT1- vs proT2-cells was analyzed in the thymus, proT2-cells were present at a significantly higher ratio (P = .049; n = 7) compared with proT1-cells (Figure 2B).

Both present and previous work from our laboratory have demonstrated that in vitro–generated proT-cells reconstitute the thymus of immunodeficient mice. However, whether proT-cells obtained from HSC/OP9-DL1 cocultures could develop into functional human T-cells in a physiological setting remained to be explored. A representative flow cytometric analysis of the thymus of an NSG mouse at 8-weeks postinjection of proT2-cells is shown in supplemental Figure 1A. Human CD45 expression was observed and these cells were CD4⁺CD8⁻ and CD4⁻CD8⁺ SP T-cells, with few DN and DP thymocytes. These SP T-cells coexpressed T-cell receptor (TCR)αβ and CD3, and appeared to be mature naïve T-cells, as indicated by CD27 and CCR7 coexpression (not shown).

We tested whether these phenotypically mature thymocytes were responsive to stimulation by anti-CD3/CD28 stimulation. Prior to activation, thymocytes obtained from proT2-injected mice were loaded with carboxyfluorescein diacetate succinimidyl ester (CFSE) to assess their proliferative ability. Shown in supplemental Figure 1B, stimulated thymocytes exhibited an increase in side scatter and forward scatter (not shown) compared with nonstimulated cells. We also observed that ∼70% of cells found within the stimulated wells displayed an upregulation of the activation marker CD25 and loss of CFSE, indicating proliferation. This was in contrast to the lack of CD25 upregulation and maintenance of high CFSE levels remaining in nonstimulated human thymocytes. In addition, stimulated thymocytes downregulated CD27 and upregulated CD38 on the cell surface indicating functional responsiveness. Supplementary Figure 1C shows that human T-cells were also detected at low levels (1% CD45) in the spleen of mice at 8-weeks postinjection of proT2-cells, and similar to the thymocytes, these cells were also capable of proliferation and activation marker upregulation (86% CD25⁺CFSE⁻) after TCR/CD28 stimulation.

A major concern after HSCT is the slow recovery of T-cells in the thymus leading to an increase in opportunistic infections and mortality in patients.¹⁴  Different strategies aimed at enhancing T-cell reconstitution are being developed and tested in clinical trials.^24-27  Previous work by Zakrzewski et al²⁷  demonstrated that DN proT-cells generated from mouse hematopoietic progenitors cocultured on OP9-DL1 cells were able to enhance T-lymphopoiesis at coinjection after HSCT. However, it was unknown whether human proT-cells coinjected with HSCs during transplantation were capable of facilitating T-cell reconstitution from HSC-derived progenitors.

To address this question, we generated HLA-A2⁻ proT2 cells by HSC/OP9-DL1 coculture, and coinjected them with HLA-A2⁺CD34⁺CD38^–/lo UCB-HSCs into irradiated newborn NSG mice. The engraftment of HSCs in these mice was confirmed by examining the BM (Figure 3A) and spleen (not shown) for the presence of human CD45⁺ HLA-A2⁺ cells at 3 to 12 weeks after injection (Figure 3A and not shown). The percentage of HSC-derived cells in the BM did not differ from the coinjected mice at any of the time-points analyzed. Furthermore, these human cells correspond to CD19⁺ B cells and CD33⁺ myeloid cells, with the exception of the week 3 time-point when myeloid cells were absent (data not shown).

Figure 3B shows the flow cytometric analysis of human CD45⁺HLA-A2⁺ cells (40%) in the BM of 1 mouse injected with HSCs with 85% of these corresponding to CD19⁺ B cells and a lower frequency corresponding to CD33⁺ myeloid cells. A similar profile was observed in mice coinjected with HSCs, along with proT2-cells (middle panels), confirming that proT2-cells do not interfere with HSC engraftment in the BM and also validating the presence of HSC-derived cells in these mice. The coinjected mice displayed very low levels of human engraftment originating from proT2-cells (<0.5%), as in vitro generated proT2-cells are inefficient at engrafting the BM when injected into neonatal immunodeficient mice. A representative flow cytometric analysis of a mouse injected with only proT2-cells is also shown (right panel). A similar reconstitution profile was observed in the spleen of these mice (data not shown).

In contrast to the profile observed in the BM of the HSC-only injected mouse, Figure 4A shows the near absence of HSC-derived cells in the thymus of these mice. The absence of T-cells at 3- to 6-weeks post-HSC–injection in neonatal mice was not surprising as T-cell reconstitution is not usually observed in these recipients as early as 6 weeks. In striking contrast, at examination of the thymuses of HSC/proT2-coinjected mice, the mice clearly possessed a CD45⁺HLA-A2⁺ population, thus indicating thymus reconstitution from an incoming HSC-derived progenitor. As expected, these mice contained proT2-derived cells (18% and 71%) in the thymus, which were CD45⁺HLA-A2⁻. The middle panel of Figure 4A highlights the more immature status of HSC/BM-derived progenitors (HLA-A2⁺-gated) compared with proT2-derived cells (HLA-A2⁻-gated) indicated by the larger proportion of DN cells in the HSC-derived compartment, as compared with the near absence of this population in the proT2-generated compartment. Nonetheless, DP cells were emerging from HSC-derived cells in the thymus of coinjected mice. As previously shown,¹⁸  flow cytometric analysis of the thymus of proT2-only injected mice revealed the presence of proT-derived cells and no HLA-A2⁺ cells. Figure 4B highlights the significant difference in the percentage of HSC-derived T-lineage cells present in the thymus when coinjected, as compared with HSC-only–injected mice. Of note, a significant facilitation of HSC/BM-derived engraftment is seen at early time-points (3-6 weeks), whereas only a trend to enhanced engraftment is seen at later time-points (9-12 weeks).

We sought to further examine T-lymphopoiesis originating from HSC-derived progenitors when coinjected with proT2-cells at earlier time-points, after only 3 weeks, a period when T-lymphopoiesis after HSCT is not typically observed. Figure 5A demonstrates the absence of HSC-derived CD45⁺HLA-A2⁺ cells in the thymus of mice receiving HSCs alone. In contrast, we observed the presence of HSC-derived human thymocytes in mice that received in vitro–generated proT2-cells and HSCs at this early time-point. Further facilitation of HSC-derived T-lymphopoiesis was still evident when mice were analyzed after 12 weeks (Figure 5B). Additionally, mice receiving both HSCs with proT2-cells displayed a significant difference in thymus cellularity when compared with mice receiving HSCs alone (Figure 5C). Table 2 illustrates the overt improvement in T-cell reconstitution from HSC-derived progenitors when combined with proT2 coinjection. The frequency of thymus-engraftment (based on CD45⁺HLA-A2⁺ expression) from a BM-derived progenitor increased significantly (P < .05) when coinjected mice are analyzed at weeks 3 to 6 and 9 to 12.

Given the enhanced thymic recruitment of BM-derived progenitors at co-transplantation with in vitro–derived proT-cells, we sought to determine a plausible mechanism by which proT-cells could be facilitating this occurrence. The critical interplay between thymocytes and TECs in the generation of a fully formed thymic microenvironment is well accepted.^28,29  Thus, we sought to assess whether proT-cells could be acting on stromal elements within the NSG thymus, positively affecting thymic function. As shown in Figure 6A, proT-cells from a day 12 coculture expressed RANK ligand (RANKL), a tumor necrosis factor receptor superfamily ligand that interacts with RANK on TECs to influence their functional maturation.²⁸  However, RANKL expression was detected at low levels on the starting HSCs used to generate the proT-cells (top plot). Of note, while both proT-cell populations expressed high levels of RANKL, the proT2 subset expressed 2.5× higher levels than the proT1-cells (bottom plot).

Next, we examined whether the expression of critical TEC-derived chemokines, such as CC19, CCL21, and CCL25, known to recruit BM and blood-borne progenitors,^6,30,31  was affected by the presence of proT-cells. First, we confirmed the engraftment of proT-cells, as indicated by the presence of human CD45 (PTPRC) transcripts within the thymuses of pro-T–injected mice vs noninjected mice (Figure 6B). With the examination of Ccl25 and Ccl21 transcript levels, our analyses revealed a marked upregulation of both chemokines in thymuses obtained from proT-injected mice as compared with noninjected mice after 3 weeks. Finally, consistent with the role of RANKL-expressing cells in directly influencing TEC maturation, the expression of Cytokeratin-5 (Krt5) was increased in the thymuses of proT-injected mice compared with noninjected littermates. Of note, thymic expression of P-selectin (Selp), RANK (Tnfrsf11a), and MHC class II (H2-Ab1) were not affected in proT-cell–injected mice (Figure 6B). The expression levels for RANK in either of the NSG mouse thymuses were about half of that observed from wild-type (WT) thymus (data not shown), whereas RANK expression was not detectable in CD45⁺ thymocytes, as previously reported.³²  Taken together, in vitro–derived proT-cells appear to induce TEC maturation and the expression of key chemokines involved in progenitor recruitment to the thymus.

In light of these findings, we sought to evaluate whether injection of proT2-cells were also facilitating the recruitment of BM-derived progenitors through structural alteration of the thymic microenvironment. To this end, immunohistologic analyses were performed on the thymuses from control (noninjected) mice or proT-injected mice. Shown in Figure 7, the thymus of an adult WT mouse, control, or proT-injected (CD34⁺CD7⁺⁺) NSG mouse were sectioned and stained with anti-Cytokeratin 5 (K5) and anti-Cytokeratin 8 (K8) expressed on thymic medullary and cortical epithelial cells, respectively. As expected, the thymus of the WT mouse possessed a cortex, medulla, and a defined corticomedullary boundary shown by anti-Cytokeratin staining. In contrast, the control NSG mouse exhibited highly disorganized thymic architecture lacking a cortex and medulla. Rather, a dense network of disorganized epithelial cells was observed. Strikingly, when human in vitro–generated proT2-cells are injected into NSG mice, a reorganization of the cortex and medulla (with a clear boundary between the 2 stromal niches) was observed. These findings suggest that a critical effect of proT-mediated enhanced thymic reconstitution may be due to the RANKL-induced differentiation and reorganization of the thymic architecture that likely leads to more effective recruitment of BM-derived T-lymphocyte progenitors.

The identification of different progenitors possessing T-cell potential from CB and BM is well established.^9,10,33,34  Previously, we characterized a CD34⁺CD7⁺⁺ T-progenitor subset generated from UCB-HSC cocultures with OP9-DL1 cells.¹⁸  Additionally, this population was capable of homing to and engrafting the thymus of immunodeficient mice. The CD34⁺CD7⁺⁺ can be fractionated into CD5⁻ and CD5⁺ subsets, both capable of reconstituting host thymus lobes in vitro.¹⁸  However, we noted that CD34⁺CD7⁺⁺CD5⁺ cells (proT2) possessed a threefold higher engrafting capacity in vitro, as compared with the CD5⁻ counterpart (proT1). Our present findings indicate that while both subsets displayed thymus-settling ability in vivo, proT2-cells possessed increased thymus engraftment function when placed in competition with proT1-cells. Although it is possible that both subsets are similarly capable of homing, the proT2 subset may have a proliferative advantage leading to a higher frequency of proT2-derived cells in the thymus of NSG mice. Nonetheless, given their more differentiated state and expression of homing molecules,¹⁸  it is simply more likely that proT2-cells are more effective at migration and/or entering the host thymus.

Although the ability to migrate and occupy the thymus provide a key initiating step to the establishment of the T-cell compartment, it was significant to investigate whether in vitro–generated proT-cells, after completing their maturation in a host thymus, have the ability to restore functional T-cell competence. To this end, we observed that human thymocytes and peripheral cells (splenocytes) displayed hallmarks of being functionally competent T-cells by their ability to proliferate and upregulate activation markers with TCR stimulation. This result is consistent with Shultz et al³⁵  that previously reported T-cell cytotoxic ability in NSG mice reconstituted with CD34⁺ HSCs. One constraint regarding the injection of in vitro–derived proT-cells is that although maturation based on phenotypic differentiation is readily observed in the thymus of NSG mice, maintenance in the periphery in the absence of other human cells is inefficient, resulting in low numbers of human T-cells in the spleen. Although rhIL-7/M25 is periodically provided, human MHC:Ag contact and additional cell types of human origin are likely required to support the maintenance and survival of peripheral T-cells in these mice. Indeed, Shultz generated HLA class I transgenic NSG mice that support human T-cell selection in an HLA-restricted manner.³⁶  T-cells generated in these mice expressed cytotoxic molecules, such as Granzyme A and Granzyme B, and more importantly they exhibited an antiviral response to Epstein-Barr virus.³⁶ 

Similar to our previous findings,¹⁸  Reimann et al³⁷  recently demonstrated that UCB CD34⁺ cells cultured on immobilized Δ-like 4 generated T-cell progenitors were able to engraft NSG mice. Here, we extend these results by demonstrating that coinjection of a defined subset of in vitro–generated proT-cells dramatically enhances and accelerates HSC-derived thymopoiesis after HSCT. Additionally, we provide a plausible mechanism for the facilitated entry of HSC/BM-derived cells into the thymus of host mice. By using a low number of HSCs, a lack of T-cell generation at multiple time-points in irradiated neonatal mice receiving CD34⁺CD38^−/lo cells was typically observed. This was not due to lack of HSC engraftment, as all mice displayed significant levels of human chimerism in the BM and spleen, consisting mostly of CD19⁺ B cells and some CD33⁺ myeloid cells. In contrast, when proT2-cells are coinjected with HSCs, thymus reconstitution is readily attained.

Based on the low number of cells used for engraftment, sometimes proT2-cells failed to engraft; however, it is important to note that T-lymphopoiesis originating from HSC/BM-derived progenitors only occurred in coinjected mice when there was evidence that proT2-cells also engrafted the thymus. This observation underscores the importance of proT-cells for HSC-derived T-lymphopoiesis in these mice. Our findings also revealed the appearance of HSC-derived thymocytes in mice as early as 3 weeks after coinjection. This phenomenon may or may not be due to HSCs directly “piggy-backing” with proT2-cells, as these cells home directly to the thymus after transplantation. This hypothesis may be tested by examining the thymuses of coinjected mice as early as 24 hours after injection. Although the appearance of HSC-derived cells in mice as early as 3 weeks does not allow us to discern whether HSCs are “piggy-backing” or not, it does indicate that proT2-cells are capable of facilitating HSC-derived T-cell reconstitution at a much earlier time as compared with mice receiving HSCs alone.

A plausible explanation for the observed facilitated entry is through the restoration of thymic architecture and function, due to inductive lympho-stromal interactions. The importance of thymocytes in the maintenance of the thymic microenvironment has been underscored by numerous studies showing that mice possessing an early block in T-cell development display abnormalities of thymic organization, with the degree of disorganization correlating with how early the block in T-cell development occurs.^38,39  In this study, we demonstrate that the highly disorganized thymic structure of NSG mice was alleviated with injection of in vitro–derived proT-cells, inducing the organization of cortical and medullary environments important for T-lymphopoiesis. Although DN thymocytes influence cortex formation, studies have demonstrated that DP and SP thymocytes induce the medulla formation and maturation of medullary TECs.^39,40  Specifically, positively selected thymocytes have been shown to regulate the development of the medulla through TNF superfamily receptor/ligand interactions, such as RANK/RANKL.³²  Indeed, the high expression levels of RANKL on in vitro–derived proT-cells and the known cross-reactivity of human RANKL with mouse RANK,⁴¹  strongly support our hypothesis that the changes in epithelial cell organization that we observed are due to lympho-stromal interaction between in vitro–derived progenitors and TECs. Lending further support to our findings is the increased expression of CCL21 and CCL25 by thymus stromal cells in mice injected with proT-cells. However, the presence of proT-cells did not alter the expression of P-selectin, which is known to be important for entry to the thymus.⁴²  Taken together, it is plausible that RANKL-expressing in vitro–generated proT2-cells (that directly settle within the thymus) alter the stromal microenvironment found in immunodeficient mice, thereby promoting the creation of new chemokine rich niches for immigrating HSC-derived progenitors. This in turn favors the migration or settling of these cells arriving from the BM to newly-enabled stromal signals and niches.

In summary, we have characterized 2 subsets of in vitro–derived progenitor T-cells for their ability to colonize the thymus of immunodeficient recipients, and differentiate into functionally mature T-cells. We noted a superior engraftment capacity by proT2-cells, and showed that this subset is able to promote T-lymphopoiesis from HSC-derived progenitors after HSCT. One caveat to these experiments is the use of immunodeficient neonatal mice, which provide a permissive environment for xenogeneic engraftment. Of higher relevance would be the use of adult mice for these studies. We have tested to determine whether in vitro–derived progenitors are capable of thymus settling in adult mice, and we observed a low frequency of mice with thymus engraftment (data not shown). The precise reasons for the difference in engraftment between the neonatal and the adult is unknown, however, the increase in the neonatal vasculature compared with adults,⁴³  severe thymus involution and continued atrophy in adult NSG mice may be factors influencing human proT-cell engraftment in neonatal vs adult mice.

The use of in vitro–generated progenitors for clinical use is underway; Delaney et al ²⁶  cultured CD34⁺ cells from UCB on immobilized Δ-like 1 (delta1^ext-IgG) that were infused with an unmanipulated UCB unit into patients undergoing HSCT.²⁶  Patients receiving 1 expanded cord and 1 unmanipulated cord displayed greater myeloid recovery compared with patients receiving 2 unmanipulated cords, in addition to shorter time to neutrophil recovery. Thus, the use of Notch-ligands as an approach to generate large numbers of proT-cells^17-19,37  have the potential to become clinically relevant, as these could be used to accelerate T-cell reconstitution after HSCT.

The authors thank Gisele Knowles and Courtney McIntosh for their expertise in cell sorting; Sean Oh, Roxanne Holmes, John Xu, and Sanam Taheri for their technical support; and Dr Rose Kung and staff of the Women and Babies program at Sunnybrook Health Sciences Centre (Toronto, ON, Canada) for their ongoing support by providing umbilical cord blood.

This work was supported by grants to J.C.Z.-P. from The Krembil Foundation and the Canadian Institutes of Health Research (CIHR MOP-119538 and HOP-83070).

Contribution: G.A. performed and designed research, analyzed data, and wrote the paper; J.S., M. Mohatshami, M. Malm, and R.N.L.M.-M. performed experiments; E.H. provided crucial materials; P.M.B. and P.S. provided intellectual insight; M.R.v.d.B. contributed to the experimental approach; and J.C.Z.-P. designed research and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current affiliation for G.A. is McEwen Centre for Regenerative Medicine, University Health Network, Toronto, ON, Canada; for M. Malm is Vaccine Research Center, Tampere, Finland; for R.N.L.M.-M. is Wellstat Therapeutics Inc., Gaithersburg, MD; and for P.S. is Centre Esther Koplowitz, Hospital Clinic de Barcelona, Barcelona, Spain.

Correspondence: Juan Carlos Zúñiga-Pflücker, Sunnybrook Research Institute, 2075 Bayview Ave, Room A3-31, Toronto, ON M4N 3M5, Canada; e-mail: jczp@sri.utoronto.ca.