Induction of γδT cells from HSC‐enriched BMCs co‐cultured with iPSC‐derived thymic epithelial cells

Abstract T cells bearing γδ antigen receptors have been investigated as potential treatments for several diseases, including malignant tumours. However, the clinical application of γδT cells has been hampered by their relatively low abundance in vivo and the technical difficulty of inducing their differentiation from hematopoietic stem cells (HSCs) in vitro. Here, we describe a novel method for generating mouse γδT cells by co‐culturing HSC‐enriched bone marrow cells (HSC‐eBMCs) with induced thymic epithelial cells (iTECs) derived from induced pluripotent stem cells (iPSCs). We used BMCs from CD45.1 congenic C57BL/6 mice to distinguish them from iPSCs, which expressed CD45.2. We showed that HSC‐eBMCs and iTECs cultured with IL‐2 + IL‐7 for up to 21 days induced CD45.1+ γδT cells that expressed a broad repertoire of Vγ and Vδ T‐cell receptors. Notably, the induced lymphocytes contained few or no αβT cells, NK1.1+ natural killer cells, or B220+ B cells. Adoptive transfer of the induced γδT cells to leukemia‐bearing mice significantly reduced tumour growth and prolonged mouse survival with no obvious side effects, such as tumorigenesis and autoimmune diseases. This new method suggests that it could also be used to produce human γδT cells for clinical applications.


| INTRODUC TI ON
Adoptive transfer of in vitro-expanded patient-derived tumour antigen-specific αβT cells was one of the earliest immune therapies for cancer. Since then, approaches using adoptive transfer of αβT cells bearing engineered T-cell receptors (TCRs) have achieved some success but have other limitations. 1,2 More recently, interest has renewed in the use of γδT cells for various clinical applications, including cancer. 3,4 γδT cells display potent cytotoxicity but differ from classical αβT cells in several aspects. They develop earlier than αβT cells, account for only 2%-3% of T cells in lymphoid organs, 5,6 and are predominant in the intestine, skin and mucosa. 7 Moreover, γδTCRs predominantly recognize mycobacterial isopentenyl pyrophosphate and related prenyl pyrophosphate derivatives in a major histocompatibility complex-independent manner 8 and have functional properties that overlap those of innate immune cells and adaptive αβT cells. 9,10 Accordingly, γδT cells play crucial roles not only in the immune response to infection but also in anti-tumour immunity, 11 although the precise mechanisms underlying their antitumour activity remain unclear.
Because of the relative scarcity of γδT cells in vivo, boosting their production has been difficult in immunosuppressed patients who may benefit from increased numbers of γδT cells, such as in patients who have undergone chemoradiation therapy, have received immunosuppressive drug treatment or have acquired/congenital immune deficiency syndromes. Whilst several methods have been described for the large-scale ex vivo and in vitro expansion of αβT cells, similar attempts to expand γδT cells have met with little success.
However, recent advances in regenerative medicine have increased our ability to manipulate hematopoietic stem cells (HSCs) and induce pluripotent stem cells (iPSCs) to generate T cells in vitro, although current methods are usually complex and incompletely reliable.
T cells develop naturally in the thymus, and their differentiation from HSCs is critically dependent on thymic epithelial cells (TECs). 12,13 Indeed, thymus transplantation has shown benefit in the treatment of several intractable diseases, including autoimmune diseases and malignant tumours. [14][15][16] However, obtaining sufficient thymus tissue for clinical application is difficult, and the function is the highest in neonates and decreases with age. 17,18 Therefore, differentiation of TECs from iPSCs in vitro has been proposed as a possible approach to overcome this supply issue. Embryonic stem cells (ESCs) and tissue stem cells (TSCs) 19 have been successfully employed to generate mouse and human TECs. However, ethical problems are associated with the use of human ESCs and obtaining sufficient TSCs from clinical specimens is difficult. Therefore, we considered the potential use of iPSCs to generate TECs in vitro. 20 In the current report, we describe a new method to generate γδT cells in vitro through co-culture of HSCs enriched from bone marrow cells (HSC-eBMCs) with induced TECs (iTECs) differentiated from iPSCs. We demonstrate that the generated γδT cells express a broad TCRγδ repertoire and have anti-tumour activity in a mouse leukemia model but do not lead to autoimmunity or de novo tumorigenesis in vivo. Our findings suggest that this method, which employs readily available BMCs and iPSCs, could also be applied to generate therapeutic anti-tumour γδT cells in humans.

| RT-PCR analysis
Total RNA was extracted from cells using TRIzol (Thermo Fisher Scientific), and aliquots of 1 µg RNA were reverse-transcribed using ReverTra Ace (Toyobo, Osaka, Japan). PCR was performed using Tks Gflex DNA polymerase (Takara Bio, Shiga, Japan) with reaction conditions of 30 cycles of denaturation at 94°C for 1 min, annealing at 55°C for 15 s and elongation at 68°C for 30 s. PCR primers are listed in Table S1. TCR Vγ or Vδ sense primers were used together with common γ or common δ antisense primers, respectively, as previously described. 24

| Co-culture of HSC-eBMCs and iTECs to produce induced lymphocytes (iLs)
HSC-eBMCs (1 × 10 5 ) were added to culture dishes containing iTECs produced as described above for 14 days. The medium was changed to fresh SF-3 supplemented with 0.1 mM 2ME with 20 ng/ml IL-2 (PeproTech Inc., Rocky Hill, NJ, USA) and 10 ng/ml IL-7 (PeproTech Inc.). For experiments lacking iTECs, HSC-eBMCs were cultured in 10-cm type IV collagen-coated dishes. The cultured cells containing iLs were collected after 7, 14 and 21 days in culture, passed through a 70µm nylon cell strainer and analysed.

| Histology
Mouse tissues were fixed in 10% formaldehyde and embedded in paraffin. Four-micrometre-thick sections were prepared and stained with hematoxylin and eosin (H&E).

| Statistical analyses
Non-parametric analyses (Mann-Whitney U test and log-rank test) were performed using StatView (Abacus Concepts, Berkley, CA, USA). p < 0.05 was considered statistically significant.

| Induction of TECs from iPSCs
To determine whether γδT cells could be generated in vitro from HSC-eBMCs and iTECs, we first investigated the culture conditions required for induction of iTECs. Using a method modified from an earlier report, 20 iPSCs were cultured in medium containing activin A and LiCl to induce mesendodermal differentiation. After 4 days, the medium was exchanged to contain FGF8 and activin A to induce mesendodermal to endodermal transition and early differentiation of epithelial cells. On day 7, the medium was exchanged to contain FGF7, FGF10 and BMP4 to induce the proliferation of epithelial cells.
Light microscopy revealed small masses of cells by day 4 ( Figure 1A), followed by the emergence of small numbers of spindle epithelial-like cells by day 7, increasing the close proliferation of the cells by day 14 and the maintenance of the dense structure between days 14 and 21. The number of the cells increased between day 7 and day 14 and plateaued thereafter ( Figure 1B).
We examined the expression of several critical transcription factors associated with haematopoiesis by RT-PCR analysis of iPSCs and iPSC-derived cells on days 7, 14 and 21 ( Figure 1C). Expression of Nanog, which is a pluripotency marker, 26 was apparent on day 0, as expected, and then decreased on day 7 and was virtually undetectable thereafter. Conversely, expression of the epithelial marker E-cadherin (Cdh1) and Hoxa3, a marker of initial thymus formation, 27 was detectable at all time points. Expression of the FGF receptor Fgfr2IIIb, which binds to FGFs and signals for induction of TECs, 28 gradually increased between day 0 and day 21. The medullary TEC marker keratin 5 (Krt5) was detected only on day 14, whilst the cortical TEC marker Krt8 was detected at all time points. Pax1 and Pax9, which lie downstream of Hox3 during thymus formation, 27,29,30 and Plet1, a marker for early TEC progenitors, 31 were detected after day 14, albeit at low levels. Autoimmune regular (Aire), which is expressed in medullary TECs and regulates the expression of tissue-specific antigens, 32 was detected from day 0 to day 21. Finally, Foxn1, a marker of mature TECs, 33 was expressed on day 14 and, to a lower extent, on day 21. The T-cell marker CD3 was barely detectable on all days analysed. These findings, which were similar to those obtained in a previous study of the generation of iTECs from iPSCs in vitro, 20 suggest that TECs can be induced from iPSCs, and that iTECs are virtually terminally differentiated after day 14.

| iPSCs express the CD45.2 allele exclusively
Next, we examined the ability of iTECs to induce T-cell development from BMCs. Because iPSCs can also differentiate into T cells, we exploited the fact that the iPSCs expressed the CD45.2 allele whilst the BMCs were prepared from CD45.1 + congenic mice ( Figure S1).
Thus, BMC-derived cells and iPSC-derived cells could be distinguished by expression of CD45.1 and CD45.2, respectively.

| Generation of iLs by co-culture of HSC-eBMCs and iTECs
Next, we examined induction of lymphocyte differentiation by co-  Figure 1B). The number of iL in the presence of iTECs increased between day 0 and day 7 and waned thereafter (Figure 2A, 2B).
Notably, large-scale expansion (approximately 30-fold) of iLs was dependent on the presence of iTECs, IL-2 and IL-7, although a small number of iLs emerged transiently in co-cultures of HSC-eBMCs and iTECs in the absence of cytokines (Figure 2A, 2B). Thus, iLs can be readily generated by 7-day co-culture of HSC-eBMCs and iTECs in the presence of appropriate cytokines. Contrary, iL increased little in the absence of iTECs with or without the cytokines.

| Analysis of the origin and phenotype of in vitro-generated iLs
To assess the origin and subtypes of iLs derived from HSC-eBMC +iTEC co-cultures, we analysed the cells by flow cytometry for the expression of CD45.1, CD45.2, TCRαβ and TCRγδ, as well as T-cell subset, B cell and NK cell markers. The iLs generated in vitro expressed CD45.1, but not CD45.2 ( Figure 3A), which indicates that they originated from the HSC-eBMCs, and not from the iPSCs.
After gating of iLs on CD45.1 + cells, the preferential differentiation of TCRγδ + cells from co-cultures was even more striking.
Thus, γδT cells and αβT cells represented 54.1% and 0.1% of CD45.1 + cells, respectively, in the presence of cytokines and 14.4% and 0.4%, respectively, in the absence of cytokines on day 7 ( Figure 3C).
Moreover, CD4 + , CD8 + , NK1.1 + and B220 + cells were virtually absent in the 7-day co-cultures ( Figure 3C). Similar results were obtained when co-cultures were examined on day 14 (data not shown). These data demonstrate that co-culture of HSC-eBMCs and iTECs with IL-2 + IL-7 preferentially results in the differentiation of TCRγδ cells, with relatively few mature αβT, NK or B cells.

| Long-term effects of transferred iLs in mice
One potential drawback to the therapeutic use of iPSCs and activated lymphocytes is their ability to induce autoimmunity and/or tumorigenesis. Therefore, we next investigated the effects of iLs injected into CD45.2 + congenic mice that had undergone autologous BMT. RT-PCR analysis of the iLs prior to injection confirmed that most cells were CD45.1 + iLs with a small fraction of residual CD45.2 + iPSCs or iPSC-derived iTECs ( Figure 5A). On day 14 after iL transfer, CD45.1 + cells were detectable in the recipient spleen, peripheral blood and small intestinal lymph nodes, but they were scarce or absent in the thymus and bone marrow ( Figure 5B). Histological analysis of the organs on day 31 after iL transfer revealed essentially normal organs, with no specific findings in the liver, lung, kidney or small intestine ( Figure 5C). In addition, these mice remained healthy and survived for >1 year (data not shown). These findings indicate that transfer of in vitro-generated iLs, containing predominantly γδT cells with some residual iPSCs/iPSC-derived cells, did not have any overt detrimental effects in vivo. Collectively, these findings demonstrate that in vitro-generated γδT cells have potent anti-tumour effects in vivo.

| DISCUSS ION
In this study, we described a new method for the induction of γδT cells from HSC-eBMCs and iPSC-derived iTECs in vitro. We showed that iTECs and cytokines were critical for the differentiation of iLs The results of the present study are consistent with a previous report on the induction of TECs from iPSCs in vitro. 20 We observed comparable development of spindle/oval epithelial-like cells, and the transcription factor expression patterns and growth rates were similar in both studies. These findings indicated that TECs differentiated from iPSCs and had matured for a period of 2 weeks, after which they achieved a steady state. We used the iTECs of the 14-day period for co-culture with HSC-eBMCs, because the thymus shows the highest function at the newborn stage. 17,18 Using the CD45 congenic system, we found that γδT cells were induced from HSC-eBMCs, and not iPSCs, and that IL-2 and IL-7 were required for the expansion of CD3 + γδT cells with a broad TCRγδ repertoire. Thus, whilst the presence of iTECs is critical for the development of γδT cells in vitro, the process is promoted by the addition of the appropriate cytokines.
In addition to T cells, other lymphocyte types develop or are present in the thymus, albeit in small numbers. 40,41 Despite this, we observed predominant induction of γδT cells in culture, with the production of few cells expressing TCRαβ, CD4, CD8, B220 or NK1.1.
The reasons for the near-exclusive induction of γδT cells are not yet clear. It is possible that additional activation and/or inhibition signals may be needed to generate other lymphocyte subsets, 42,43 especially because γδT cells develop before αβT cells in vivo. 7,44 Alternatively, it may be necessary to further modify the iTEC culture system to enable αβT cell development, such as changes to the three-dimensional structure 13,45 and/or addition of distinct cortical/ medullary iTEC types.
Generally, Vγ4-6 + cells develop and migrate into the dermis and reproductive epithelium before birth, whereas Vγ1, 2, 4 and 7 + cells are induced predominantly in lymphoid organs and the intestinal epithelium after birth in mice. 37 Our co-culture system showed that the induced γδT cells expressed Vγ1-3, 2, 4 and 6, and Vδ4-7 + between days 7 and 14 in culture. This is likely to be due to differences between the generation of thymic environment in vivo and our system, such as the use of iTECs, which are postulated to be equivalent to newborn TECs, together with HSC-eBMCs purified from adult mice. Nonetheless, the method described here is significant because it at least partly reproduces the environment necessary to obtain γδT cells from two types of stem cell-derived cells in vitro.
Analysis of the fate of iLs after transfer to mice revealed their presence in the spleen, peripheral blood and intestinal lymph nodes, but not the thymus or bone marrow, of recipient mice. These findings indicate that the transferred cells are sufficiently mature to survive and migrate to peripheral hematopoietic organs. In addition, the recipient mice survived for >1 year and showed no apparent pathological effects of iLs, including the tumorigenesis and autoimmune disease. This important observation suggests that the transfer of γδT cells is not harmful, even if they contain low numbers of iPSCs or differentiated iTECs.
Another important conclusion from this study is that γδT cells are effectively able to suppress leukaemia growth in vivo, as has been shown in clinical trials for some hematopoietic malignancies. 39 Leukemic mice transplanted with iLs showed significantly reduced tumour growth and prolonged survival compared with untreated mice or those transplanted with γδT cell-depleted iLs. These findings suggest that γδT cells played the dominant role in inhibiting leukemic cell growth in vivo. Further studies will be necessary to determine the detailed mechanism of action of the γδT cells.
iPSCs have been used to generate differentiated cells for use in regenerative medicine in humans, and patient-derived iPSCs have also proven extremely useful for in vitro function/toxicity screening during drug development. 46,47 In this study, we used iPSC-derived cells to support the differentiation of specific target cells, thereby identifying a third possible use for iPSCs. Of note, although we observed no autoimmunity or malignancy in animals injected with iL populations containing low levels of residual iPSCs/iPSC-derived cells, it will be important to monitor this as part of the safety/efficacy evaluation of iPSC-derived cells as potential therapeutics for use in humans.
In summary, we present a method for the induction of γδT cells using iPSC-derived iTECs co-cultured with HSC-eBMCs. The method may be useful for developing treatments for several intrac- University, for help with the study conception and method, and Ms. K. Nishimura for secretarial assistance.

CO N FLI C T O F I NTE R E S T
All authors declare that no conflicts of interest exist.

DATA AVA I L A B I L I T Y S TAT E M E N T
Supporting data are available from the corresponding author upon reasonable request.