Induction of WT1-specific human CD8⁺ T cells from human HSCs in HLA class I Tg NOD/SCID/IL2rgKO mice

The immune system prevents infectious disease initiation and progression and functions in multiple homeostatic processes. However, dysfunctional immunity is observed in patients with malignancies, contributing to neoplastic progression. Therefore, reconstitution of immunity by allogeneic stem cell transplant or activation of specific and nonspecific immunity-targeting diseases improves clinical outcomes in patients with solid cancers and in those with hematologic malignancies.^1,2  Such treatment can be carried out by vaccination and by adoptive immunotherapy.

Vaccinations aim to elicit antigen-specific effector cell–mediated immune responses in vivo.³  Among several candidates, peptide vaccines and dendritic cell (DC) vaccines were 2 widely selected protocols. In the last 2 decades, however, administration of these vaccines has not significantly improved the prognosis of patients with solid cancers including melanoma and other types of solid tumors.^4,5  Although several recent trials reported encouraging clinical outcomes using glycoprotein 100 peptides in combination with interleukin (IL)-2 for the treatment of melanoma,⁶  or patient-derived antigen-presenting cells (APCs; sipuleucel-T) for the treatment of prostate cancer,⁷  cancer vaccination appears to require modifications based on increased understanding of in vivo biology of human APCs and T cells.

In contrast, immunotherapy based on adoptive transfer of ex vivo expanded tumor-reactive T cells has achieved promising results. In metastatic melanoma, adoptive transfer of tumor-infiltrating lymphocytes in combination with chemotherapy or irradiation has improved cure rates up to 20% to 40%.⁸  Because the antitumor effect of tumor-infiltrating lymphocytes has not been confirmed in malignancies other than melanoma, genetically engineered T cells that express tumor antigen–specific T-cell receptor (TCR) genes or chimeric antigen receptors have been developed.⁹  Recent clinical trials showed improved clinical outcomes in patients treated with genetically engineered T cells,^10-13  whereas adverse effects were observed immediately after the transfusion of T cells expressing chimeric antigen receptors.^14,15 

In several clinical trials of vaccination therapies for hematologic malignancies, promising responses were observed using various antigens, including proteinase 3¹⁶  and Wilms tumor 1 (WT1)^16,17  for acute myeloid leukemia (AML), breakpoint cluster region/Abelson murine leukemia for chronic myelogenous leukemia,¹⁸  and patient-specific idiotypes derived from malignant B-cell clones for follicular lymphoma.¹⁹  In particular, for patients with poor prognostic factors, development of immunotherapy targeting minimal residual disease or leukemia stem cells (LSCs) should play an essential role in achieving long-term patient survival.

We recently reported that WT1, a transcription factor expressed in variety of malignant tissues, is highly expressed by CD34⁺CD38⁻ AML cells.²⁰  WT1 is considered one of the best antigens to be used for immunotherapy against malignancies, based on multiple criteria such as therapeutic function, immunogenicity, and specificity.²¹  Using WT1 peptide or full-length messenger (m)RNA for WT1, clinical trials against hematologic malignancies detected increased frequencies of WT1-specific CD8⁺ T cells in patient blood after the treatment.^16,17,22,23  Nevertheless, to accomplish significant improvement in clinical outcomes of AML patients, we need to better understand the biology of the human immune system leading to efficient activation of human acquired immunity against tumor antigens.

In the present study, we aimed to develop an in vivo system for induction of antigen-specific, HLA-restricted human CD8⁺ T cells after vaccination. HLA class I–expressing NOD/SCID/IL2rgKO (NSG) mice supported the development of human T cells and APCs after engraftment with human cord blood (CB) HSCs. We detected high frequencies of WT1-specific CD8⁺ T cells in the bone marrow (BM) and spleen of HLA class I–expressing NSG mice after vaccination. Moreover, we confirmed the development of WT1-specific CD8⁺ T cells in vivo after engraftment of human HSCs transduced with WT1-specific TCR (WT1-TCR) Vα and Vβ genes. The antigen-specific human CD8⁺ T cells expanded in response to WT1 antigen and were functional both in cytokine production and cytotoxicity. Development of immunotherapy protocols in HLA class I–expressing NSG mice may facilitate the development and optimization of antigen-specific immunotherapy against malignancies.

Detailed experimental methods are described in supplemental Methods, available on the Blood Web site.

For phenotypical analysis, cells labeled with monoclonal antibodies (mAbs) (supplemental Table 1) were analyzed using FACSAria and FACSCanto II instruments (BD Biosciences).

In the DC vaccine group, HSC-reconstituted HLA class I transgenic (Tg) NSG recipients were injected with WT1 peptide–loaded DCs via retro-orbital plexus on day 0, followed by intraperitoneal and subcutaneous immunization with 300 μg of WT1_126-134 or WT1_235-243 synthetic peptide mixed with 150 μg of polyinosinic:polycytidylic acid (poly I:C; InvivoGen) on days 0, 10, and 20. In the vaccine group without DCs, the same doses of WT1_126-134 or WT1_235-243 synthetic peptide mixed with poly I:C were administered intraperitoneally and subcutaneously on days 0, 10, and 20. One week after the last immunization, mice were necropsied for evaluation of human immune cells in peripheral blood (PB), BM, spleen, lymph nodes, and thymus.

Human cytotoxic T-lymphocytes (CTLs) were expanded from HLA class I Tg NSG humanized mice as described previously, with some modifications.²⁴  Splenocytes, thymocytes, and BM cells from recipient mice were stimulated with irradiated autologous lymphoblastoid cell lines (LCLs) pulsed with 40 μg/mL of WT1_126-134 or WT1_235-243 synthetic peptide at a responder-to-stimulator ratio of 1:1.

Purified CB CD34⁺CD38⁻ cells were incubated with concentrated lentiviral supernatant at a multiplicity of infection of 100 to 200. After 5 days of incubation, green fluorescent protein–expressing (GFP⁺) cells were sorted using a FACSAria cell sorter (BD Biosciences) and transplanted into irradiated HLA class I Tg NSG newborn mice.

WT1-specific T-cell responses were quantified by the interferon (IFN)-γ enzyme-linked immunospot (ELISpot; Mabtech) assay as described previously.²⁴  Pictures of each well were captured using AlphaImager Image Analysis System (Alpha Innotech), and individual spots were counted per well. Using the same protocol as for the IFN-γ assay, the production of human tumor necrosis factor α by human WT1-specific CD8⁺ T cells was also confirmed by ELISpot assay.

Cytotoxic activity of WT1-specific CTLs was evaluated by chromium 51 (⁵¹Cr) release assay as described previously.²⁴  To examine the HLA-restricted cytotoxicity, target cells were incubated with anti-HLA class I framework mAb (w6/32; American Type Culture Collection) or anti-HLA-DR mAb (L243; American Type Culture Collection) at 10 μg/mL for 1 hour, followed by adding the effector cells. The percentage of specific lysis was calculated as (experimental release cpm − spontaneous release cpm)/(maximal release cpm − spontaneous release cpm) × 100.

HLA-A24–expressing K562 cells²⁵  were cultured with or without WT1-specific CTLs for 5 hours. After harvesting all the cultured cells, the cells were inoculated subcutaneously into female NSG mice. WT1-specific CTLs were adoptively transferred weekly. At 24 to 28 days postinoculation, the mice were euthanized to evaluate tumor size and weight.

After in vitro expansion, WT1-specific CTLs were purified by fluorescence-activated cell sorting (FACS). As a control, 2.4 × 10⁴ to 1.5 × 10⁵ human CD45⁺CD3⁺CD8⁺ cells were purified from spleens of NSG mice engrafted with nontransduced human HSCs. Total RNA was extracted using TRIzol reagent (Invitrogen) and reverse transcribed for synthesis of first-strand complementary DNA using the SMARTer RACE cDNA Amplification Kit (Clontech). Both universal mix primer and primers specific for the TCRα or TCRβ constant region sequence were used for 5′ end–specific polymerase chain reaction (PCR) amplifications, resulting in both TCRα and TCRβ PCR products of high purity that were then sequenced using MiSeq (Illumina). All reads of the TCRα and TCRβ repertoire sequences were analyzed using Perl scripts based on the USEARCH algorithm (http://drive5.com/usearch/). V-region consensus sequences in each cluster were searched on the ImMunoGeneTics sites (www.imgt.org/IMGT_vquest/share/textes/).

We developed immunodeficient NSG mice expressing HLA class I antigens by backcrossing HLA-A*0201²⁴  or HLA-A*2402 transgenes onto the NSG background. Both transgenes also encode covalently bound human β-2-microglobulin. By reverse-transcription PCR, HLA transgenes were detected in the BM, spleen, and CD45⁻EpCAM⁺ thymic epithelial cells derived from both NSG-HLA-A2/HHD mice and NSG-HLA-A24/HHD mice (Figure 1A). We then transplanted 4.5 × 10³ to 5.6 × 10⁴ purified human HLA-matched CD3⁻CD4⁻CD8⁻CD34⁺CD38⁻ CB HSCs into HLA class I Tg NSG newborn mice. At 15 to 37 weeks posttransplant, we confirmed engraftment of human (h)CD45⁺ leukocytes in the PB, BM, spleen, mesenteric lymph nodes (MLNs), axillary lymph nodes, and thymus (Figure 1B-C; Table 1; PB, 72.7% ± 6.9%; spleen, 91.5% ± 2.0%; BM, 72.4% ± 9.2%; MLN, 97.3% ± 0.7%; and thymus, 97.6% ± 1.2% [n = 11]; axillary lymph nodes, 96.7% ± 1.4 % [n = 10]). Human CD3⁺ T cells were 65.9% ± 7.2% of engrafted hCD45⁺ leukocytes in the recipient spleen. CD4⁺CD8⁺ double-positive T cells accounted for 34.2% ± 12.5% in the thymus (n = 6), whereas CD4⁺ or CD8⁺ single-positive T cells were 26.2% ± 5.9% or 31.7% ± 11.3%, respectively, of engrafted human CD3⁺ T cells. Within the human CD8⁺ fraction, we identified CD45RA⁺CCR7⁺ naïve, CD45RA⁻CCR7⁺ central memory (CM), CD45RA⁻CCR7⁻ effector memory (EM), and CD45RA⁺CCR7⁻ terminally differentiated (EMRA) T-cell subsets in the recipient spleen (Figure 1D-E; naïve, 20.8% ± 10.0%; CM, 13.7% ± 4.9%; EM, 60.9% ± 10.3%; and EMRA, 4.9% ± 1.8% [n = 10]). Cytoplasmic expression of granzyme A was the highest in EM CD8⁺ T cells (Figure 1F; supplemental Figure 1; naïve, 10.0% ± 2.1%; CM, 42.0% ± 8.9%; EM, 75.4% ± 10.7%; and EMRA, 55.1% ± 10.3% [n = 4]).

To elucidate whether human APCs developed in the HLA class I Tg NSG mice, we analyzed the expression of CD33 and other myeloid-associated antigens using flow cytometry. In the NSG-HLA-A2/HHD mice (n = 4) and NSG-HLA-A24/HHD mice (n = 6), we found that CD14⁺CD33⁺HLA-DR⁺ monocytes, CD123^highCD11c⁻ plasmacytoid DCs (pDCs), and CD33⁺CD11c⁺HLA-DR⁺ conventional DCs (cDCs) differentiated in the BM and spleen of HLA class I Tg NSG recipients (Figure 1G-H; frequency in human CD45⁺ cells: BM [n = 8]: monocytes, 3.4% ± 0.8%; pDCs, 2.6% ± 0.7%; and cDCs, 5.2% ± 1.6%; spleen [n = 7]: monocytes, 0.9% ± 0.5%; pDCs, 0.4% ± 0.2%; and cDCs, 1.0% ± 0.4%). BDCA1⁺ DCs and BDCA3⁺ DCs were detected within the cDC fraction (Figure 1I; frequency in human CD33⁺ cells: BM [n = 8]: BDCA1⁺ DCs, 9.2% ± 4.7%; and BDCA3⁺ DCs, 1.0% ± 0.3%; spleen [n = 7]: BDCA1⁺ DCs, 7.6% ± 4.4%; and BDCA3⁺ DCs, 0.9% ± 0.2%). Consequently, subsets of human CTLs and human myeloid APCs, required for the establishment of acquired immunity after vaccination, were present for the long-term in HLA class I Tg NSG recipients.

On the basis of the previous report showing that WT1 is highly expressed in AML stem cells,²⁰  we examined the expression levels of WT1 in the purified AML CD34⁺CD38⁻ cells and normal BM CD34⁺CD38⁻ cells using quantitative reverse-transcription PCR. Expression of WT1 in the CD34⁺CD38⁻ stem cell fraction was higher in AML samples compared with BM samples from healthy donors (supplemental Figure 2A; [relative mRNA expression of WT1 corrected by glyceraldehyde-3-phosphate dehydrogenase] healthy donors, 2.84 ± 2.02 × 10⁻⁴ [n = 7]; AML patients, 1.07 ± 0.25 × 10⁻² [n = 38]; P = .0002 by 2-tailed Student t test). Western blot analysis confirmed the expression of WT1 in several leukemia cell lines and CD34⁺CD38⁻ cells from AML patient samples at the protein level (supplemental Figure 2B).

We then performed immunization experiments to evaluate the induction of antigen-specific human CTLs in the HLA class I Tg NSG mice. To do so, we chose WT1 as an antigen and poly I:C as an adjuvant. In addition, we prepared autologous or HLA class I–matched allogeneic WT1 peptide–loaded DCs. In immunized HLA class I Tg recipients (Table 1), we found a significant increase of WT1-specific tetramer-positive (tetramer⁺) cells in recipients’ PB after immunization (Figure 2A-B; 0.044% ± 0.017% and 0.399% ± 0.110% before and after immunization, respectively [n = 7]; P < .05 by 2-tailed paired Student t test). Compared with unvaccinated mice, the frequencies of WT1 tetramer⁺ CTLs in spleen were increased significantly in the recipients of the vaccine groups both with and without WT1 peptide–loaded DCs (Figure 2C-D; no immunization control, 0.041% ± 0.007% [n = 10]; immunized without DCs, 0.170% ± 0.043% [n = 10]; P = .0146; immunized with DCs, 0.230% ± 0.044% [n = 9]; P = .0027). In the recipients vaccinated with WT1 peptide–loaded DCs, WT1-specific CTLs were also increased in the PB, BM, and MLN; however, the increases in these organs were not significant in the group immunized without DCs (Figure 2D; WT1 tetramer⁺/CD8⁺ T cells (%): PB: no immunization control, 0.038% ± 0.014% [n = 10]; immunized without DCs, 0.072% ± 0.024% [n = 7]; immunized with DCs, 0.361% ± 0.088% [n = 9]; P = .0061; BM: no immunization control, 0.090% ± 0.043% [n = 7]; immunized without DCs, 0.429% ± 0.136% [n = 5]; immunized with DCs, 0.658% ± 0.203% [n = 5]; P = .0473; MLN: no immunization control, 0.013% ± 0.006% [n = 6]; immunized without DCs, 0.071% ± 0.032% [n = 6]; immunized with DCs, 0.146% ± 0.030% [n = 7]; P = .0040).

To evaluate the antigen-specific proliferation of CTLs, 2 to 10 × 10⁶ splenocytes of recipients were further stimulated with WT1 peptide–pulsed autologous LCLs in vitro. After 4 weeks, the frequency of WT1 tetramer⁺ CTLs derived from splenocytes was higher in the vaccinated recipients than in unvaccinated mice (Figure 2E-F; WT1 tetramer⁺/CD8⁺ (%) before and after in vitro expansion: no immunization group, 0.050% ± 0.006% and 0.168% ± 0.042%, respectively [n = 3]; immunized group, 0.194% ± 0.032% and 0.511% ± 0.095%, respectively [n = 12]; P = .0057). The degree of increase in the number of WT1-specific CTLs after expansion was also higher in the immunized recipients (Figure 2G; fold increase of WT1 tetramer⁺CD8⁺ cell counts before and after in vitro expansion: no immunization group, 4.06 ± 2.34 [n = 3]; immunized group, 42.13 ± 12.1 [n = 12]; P = .0097).

As another way of inducing antigen-specific human CTLs in vivo, we evaluated the differentiation of human T cells from antigen-specific TCR-transduced HSCs. We constructed a lentiviral vector encoding WT1-TCR Vα and Vβ genes. We transduced the TCR Vα and Vβ genes into purified human CB HSCs using the lentiviral vector (Figure 3A; supplemental Figure 4A), and purified CD34⁺GFP⁺ cells by FACS. Both HLA-matched and -mismatched CBs were used as HSC sources (Table 2). Then we transplanted GFP-labeled gene-transduced CD34⁺ cells into NSG-HLA-A24/HHD newborn mice. After 4 to 6 months, multilineage human hematopoietic cells developed (Figure 3B-C; supplemental Figure 4B; spleen [n = 7 nontransduced and GFP control recipients]: B cells, 56.9% ± 12.2% and T cells, 34.4% ± 13.1%; [n = 15 WT1-TCR gene–transduced HSC recipients]: B cells, 71.5% ± 4.3% and T cells, 9.8% ± 4.0%; thymus [n = 6 recipients of nontransduced or control vector–transduced HSCs]: B cells, 12.2% ± 4.9% and T cells, 53.1% ± 9.2%; and [n = 15 WT1-TCR gene–transduced HSC recipients]: B cells, 5.2% ± 2.3% and T cells, 28.0% ± 6.0%). CD4⁺ and CD8⁺ T cells were identified in the spleen of control NSG recipients and NSG recipients transplanted with WT1-TCR-transduced HSCs. In the thymus, double-positive T cells and single-positive T cells were found in both recipients (Figure 3D). Engraftment levels of human CD45⁺ cells were not statistically different between control mice and TCR-transduced mice (Table 2).

At the day of euthanization, WT1 tetramer⁺CD8⁺ cells were present at frequencies of 0.316% ± 0.070% (n = 9), 1.158% ± 0.513% (n = 11), and 0.565% ± 0.200% (n = 3) in the spleen, thymus, and BM of WT1-TCR-transduced HSC recipients, respectively (Figure 4).

Human CTLs developed in the NSG-HLA-A24/HHD mice were stimulated in vitro using HLA-A*2402(⁺) WT1 peptide–pulsed LCLs for further expansion (supplemental Figure 5). When HLA-A*2402-mismatched CB HSCs were transplanted into NSG-HLA-A24/HHD newborn mice, WT1 peptide–pulsed, HLA-A*2402-matched allogeneic CB LCLs were used as stimulators to expand WT1-specific CTLs in vitro. After 2 to 3 weeks, the frequency of WT1 tetramer⁺ CTLs derived from splenocytes of the recipients was significantly increased (Figure 5A-B; WT1 tetramer⁺/CD8⁺ T cells (%): BM [n = 5]: before, 0.626% ± 0.140% and after, 11.57% ± 10.23%; P = .3880; spleen [n = 9]: before, 0.316% ± 0.070% and after, 10.12% ± 7.781%; P = .0197; thymus [n = 4]: before, 1.533% ± 1.282% and after, 23.95% ± 14.10%; P = .0561). The cell numbers of WT1 tetramer⁺ CTLs derived from BM, spleen, and thymus were significantly increased after in vitro expansion (Figure 5C; cell numbers of WT1 tetramer⁺CD8⁺ T cells: BM [n = 5]: before, 8.1 × 10¹ ± 4.5 × 10¹ and after, 1.5 × 10⁶ ± 1.5 × 10⁶; *P = .0253; spleen [n = 9]: before, 2.7 × 10² ± 1.1 × 10² and after, 9.4 × 10⁵ ± 8.9 × 10⁵; P = .0013; thymus [n = 4]: before, 4.0 × 10² ± 2.4 × 10² and after, 1.9 × 10⁶ ± 1.8 × 10⁶; P = .0102).

To elucidate whether the GFP⁺ WT1 tetramer⁺CD8⁺ T cells retain specific TCRs, we performed a repertoire analysis of TCRs using MiSeq. The result of repertoire sequences demonstrated that FACS-purified GFP⁺ WT1 tetramer⁺CD8⁺ T cells displayed skewing of both Vα and Vβ repertoires and that the majority of Vα and Vβ chains of the T cells were the transduced Vα and Vβ chains (Figure 6). After the expansion of WT1-specific CTLs for 4 to 5 weeks, we found that the majority of GFP⁺CD8⁺ T cells were CD45RA(⁺)CCR7(⁻) effector T cells (supplemental Figure 6).

We also evaluated the antigen-specific cytokine production by the CTLs with the ELISpot assay. Recipient-derived WT1-specific CTLs produced IFN-γ against WT1 peptide–pulsed, but not peptide-unpulsed, HLA-A*2402(⁺) LCLs. The addition of anti-HLA class I antibody significantly reduced the number of IFN-γ-producing cells against WT1 peptide–pulsed LCLs (Figure 5D-E). WT1-specific CTLs also produced tumor necrosis factor α (supplemental Figure 7).

Using ⁵¹Cr release assay, we evaluated cytotoxicity by the amplified CTLs derived from a WT1-TCR-transduced NSG-HLA-A24/HHD recipient BM (A24-TCR-2) (Figure 5F-G). Because this donor CB sample used for the transplant of WT1-TCR-transduced HSC was HLA-A*2402 mismatched (Table 2; CB23-1), allogeneic HLA-A*2402-matched LCLs were used for in vitro stimulation. Cytotoxic activities were observed against allogeneic WT1 peptide–pulsed HLA-A*2402-matched LCLs (Figure 5F; LCL#1, derived from CB23-3), but not against peptide-unpulsed LCL#1 and autologous HLA-A*2402-mismachted LCLs (Figure 5F, upper; LCL#2, derived from CB23-1) with or without peptide loading. WT1(⁺) HLA-A*2402(⁺) KAZZ leukemia cell lines were lysed by these CTLs. Although HLA-A*2402(⁻) K562 leukemia cell lines were not lysed, HLA-A*2402 gene–transduced K562 cells became susceptible to lysis by these CTLs (Figure 5G, left). The cytotoxic activities of CTLs against WT1 peptide–pulsed LCL#1 and HLA-A*2402-expressing LCLs were inhibited by anti-HLA class I mAb, but not by anti-HLA class II mAb (Figure 5F, lower, and G, right). We further aimed to determine the effect of WT1-specific CTLs on the in vivo growth of leukemic cells. To this end, we used as effector cells WT1-specific CTLs obtained from spleens of NSG-HLA-A24/HHD recipients engrafted with human HSCs transduced with WT1-TCR genes. HLA-A24-expressing K562 cells were used as target cells. We cultured the target K562 cells with WT1-specific CTLs ex vivo, and then inoculated the cultured K562 cells subcutaneously into NSG mice. At 3 to 7 days after 3 weekly transfers of WT1-specific CTLs, we found inhibition of subcutaneous tumor formation (Table 3; supplemental Figure 8). These results indicate that WT1-TCR-transduced human HSCs can induce specific and functional CTLs and multilineage immune subsets in HLA class I Tg NSG recipients.

Evoking immune responses against specific antigens requires HLA-restricted interaction between APCs and effector cells. In this study, we examined whether active immunization with vaccines can prime HLA-restricted antigen-specific human CTLs in mice expressing HLA class I molecules HLA-A*0201 and HLA-A*2402, the most common alleles in Caucasian²⁶  and Japanese populations, respectively.²⁷  First, we confirmed HLA class I expression in murine thymic epithelial cells from the previously reported NSG-HLA-A2/HHD^24,28  and from the newly developed NSG-HLA-A24/HHD strains. In HLA class I Tg NSG mice transplanted with HLA class I–matched CB HSCs, human memory T cells expressing cytotoxic molecules and human HLA-DR-positive antigen-presenting myeloid cells, including BDCA1⁺ or BDCA3⁺ cDCs, were detected in the recipient BM and spleen. After vaccination using adjuvant mixed 9-mer WT1 peptides, both with and without the support of ex vivo prepared antigen-loaded autologous DCs, antigen-specific CTLs increased in vivo. The frequencies of WT1-specific tetramer⁺CD8⁺ T cells in the vaccinated recipients were comparable to those in the AML patients who received vaccination using WT1 antigen.^16,17,22,23  When we compared the peptide-vaccination and DC-vaccination groups, we found more efficient induction of WT1-specific CD8⁺ T cells with the support of autologous DCs, which might be accounted for by the incomplete DC development in the lymph nodes of HSC-engrafted NSG mice.

Toll-like receptor 3 (TLR3)-expressing human BDCA3⁺ cDCs are regarded as a functional equivalent of murine CD8a⁺ DCs, which play crucial role in the induction of CTL responses against cancer cells and viruses.^29-32  The TLR3 agonist poly I:C has been shown to function as an adjuvant, and induces cellular immune responses in a mouse model³³  and in clinical trials.³⁴  Because we confirmed the expression of TLR3 in DCs in human HSC-engrafted NSG mice, we chose poly I:C as an adjuvant. The frequency of BDCA3⁺ DCs was increased after vaccination of poly I:C mixed with WT1 peptides (supplemental Figure 3).

Although adoptive immunotherapy with peripheral T cells that express antigen-specific TCR genes has been evaluated in clinical trials,⁹  there are potential problems such as limited ex vivo expansion of peripheral T cells³⁵  and a risk of generating autoreactive T-cell clones due to mispairing between Tg and endogenous TCR chains.³⁶  Recently, using a retroviral vector encoding small interfering RNAs targeting endogenous TCR genes,³⁷  we transduced WT1-TCR genes into peripheral T cells and demonstrated their in vivo cytotoxicity against leukemia cells.³⁸ 

In the current study, we aimed to transfer WT1-TCR genes into HSCs to generate Tg T cells in vivo for long-term function. It was reasoned that these T cells would undergo negative selection resulting in the elimination of autoreactive T cells.³⁹  During a median follow-up of 26 weeks, the recipients of WT1-TCR gene–transduced HSCs did not show any signs of severe graft-versus-host-disease such as weight loss, skin inflammation, and diarrhea. The recipient-derived WT1-specific CTLs retained the capacity to proliferate in vitro and exerted antigen-specific cytokine response and lytic activity, indicating that the HSCs generated functional Tg CTLs in vivo.

Our results supported previous reports describing the development of Tg CTLs specific for MART-1 (melanoma antigen recognized by T cells).^40-42  Tg MART-1-specific CD8⁺ T cells were detected in BM/liver/thymus humanized mice⁴³  transplanted with TCR gene–transduced fetal liver–derived CD34⁺ HSPCs⁴⁰  or in HLA-A*0201 Tg NSG mice transplanted with TCR-transduced CB CD34⁺ HSPCs.⁴²  Through the TCR Vβ repertoire analysis of MART-1-specific tetramer⁺ T cells in the recipients, both groups demonstrated that the allelic exclusion by exogenous TCRs suppressed the rearrangement of endogenous human TCR genes during thymocyte differentiation and reduced the risk of generating autoreactive CTL clones.

The frequency of WT1-specific Tg CTLs in our recipients was relatively lower compared with that of MART-1-specific CTLs detected in those reports.^40-42  Several differences in experimental settings might explain this disparity. First, we isolated CD34⁺CD38⁻ CB HSCs for lentiviral TCR transfer, and transduction marker–positive (GFP⁺) cells for transplant, to exclude the possibility of residual preformed human T cells in the graft. Purified WT1-TCR-transferred HSCs still retained the multilineage differentiation capacity. Second, the variances of gene expression patterns between WT1 and MART-1 might affect their immunogenicity. WT1 is expressed in normal tissues, including podocytes of normal kidney,⁴⁴  Sertoli cells of testis, and granulosa cells of ovary,⁴⁵  whereas the expression of MART-1 is localized to melanocytes of skin and retina.^46,47 

Without expression of human HLA class II molecules, appropriate CD4 T-cell help might be lacking in our recipients, causing a failure of T-cell memory formation, as well as incomplete priming of CD8⁺ cells in vivo.⁴⁸  In addition, current xenograft recipient lymph nodes do not fully recapitulate human lymph node architecture (supplemental Figure 9). In the future, we aim to humanize the lymph node microenvironment in the recipient and support the development of lymph node architecture through Tg expression of human cytokines and a more physiological repopulation of human immune cells.

As previously reported,²⁰  we confirmed high expression of WT1 in the CD34⁺CD38⁻ LSC fraction from patients with AML at mRNA and protein levels. As a marker of minimal residual disease, WT1 mRNA expression in patients with AML has been assessed for the prediction of relapse after chemotherapy⁴⁹  and HSC transplant.⁵⁰  Our results might provide some motivation to perform immunotherapeutic targeting of WT1 against persistent LSCs in hematologic remission. Regardless of the HLA-A locus genotype of the donor, we detected functional WT1-specific CTLs in the BM^51,52  and spleen of the recipients transplanted with TCR-transduced CB HSCs, suggesting the application for a wide variety of donor-graft combinations in the clinical setting of HSC transplant.

In the current study, we were able to evaluate 2 different options of human immunotherapy in a mouse model. On the basis of these results, the HLA class I Tg NSG xenograft system might serve as a preclinical tool for immunotherapy against human malignancies.

The authors thank David Serreze for providing NOD/Lt mice transgenically expressing HLA-A24, and Kiyotaka Kuzushima for providing HLA-A24-expressing K562 cells.

This study was supported by the Basic Science and Platform Technology Program for Innovative Biological Medicine; the Project for Development of Innovative Research on Cancer Therapeutics from Ministry of Education, Culture, Sports, Science, and Technology; The Japan Agency for Medical Research and Development; The Maine Cancer Foundation; and National Institutes of Health National Cancer Institute grants CA034196 and CA171983 and Office of the Director grant R24OD018259.

Contribution: Y.N. designed and performed the experiments, analyzed the data, and wrote the manuscript; M.T.-M. performed the sorting experiments and T-cell functional assays; T.O. and H.F. performed the ⁵¹Cr releasing assay and analyzed the data; R.O. analyzed the data; T.W. analyzed the genomic data for T cells; N.S. performed the sorting experiments and analyzed the myeloid differentiation; and Y.S., O.O., L.D.S., M.Y., and F.I. designed the research, analyzed the data, and wrote the manuscript.

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

Correspondence: Fumihiko Ishikawa, Laboratory for Human Disease Models, RIKEN Center for Integrative Medical Sciences, 1-7-22 Suehiro-cho Tsurumi-ku, Yokohama 230-0045, Japan; e-mail: fumihiko.ishikawa@riken.jp.