Genetically modified human CD4⁺ T cells can be evaluated in vivo without lethal graft‐versus‐host disease

Summary

Adoptive cell immunotherapy for human diseases, including the use of T cells modified to express an anti‐tumour T‐cell receptor (TCR) or chimeric antigen receptor, is showing promise as an effective treatment modality. Further advances would be accelerated by the availability of a mouse model that would permit human T‐cell engineering protocols and proposed genetic modifications to be evaluated in vivo. NOD‐scid IL2rγ ^null (NSG) mice accept the engraftment of mature human T cells; however, long‐term evaluation of transferred cells has been hampered by the xenogeneic graft‐versus‐host disease (GVHD) that occurs soon after cell transfer. We modified human primary CD4⁺ T cells by lentiviral transduction to express a human TCR that recognizes a pancreatic beta cell‐derived peptide in the context of HLA‐DR4. The TCR‐transduced cells were transferred to NSG mice engineered to express HLA‐DR4 and to be deficient for murine class II MHC molecules. CD4⁺ T‐cell‐depleted peripheral blood mononuclear cells were also transferred to facilitate engraftment. The transduced cells exhibited long‐term survival (up to 3 months post‐transfer) and lethal GVHD was not observed. This favourable outcome was dependent upon the pre‐transfer T‐cell transduction and culture conditions, which influenced both the kinetics of engraftment and the development of GVHD. This approach should now permit human T‐cell transduction protocols and genetic modifications to be evaluated in vivo, and it should also facilitate the development of human disease models that incorporate human T cells.

Abbreviations

APC

allophycocyanin

CHO

Chinese hamster ovary

FBS

fetal bovine serum

GAD

glutamic acid decarboxylase

GFP

green fluorescent protein

GVHD

graft‐versus‐host disease

HA

haemagglutinin

hIFN‐γ

human interferon‐γ

IMGT

ImMunoGeneTics

NSG

NOD‐scid IL2rγ ^null

PBMC

peripheral blood mononuclear cells

PE

phycoerythrin

rhIL‐7

recombinant human interleukin‐7

TCR

T‐cell receptor

Introduction

Personalized medicine, until relatively recently merely an optimistic catchphrase reflecting clinical hopes for the future, is gradually becoming a reality. Adoptive cell immunotherapy, using autologous T cells genetically engineered to recognize and aid in the elimination of specific infectious agents or tumour types, represents one important advance.1 Human T cells have been modified by retroviral transduction to express clinically relevant T‐cell receptors (TCRs), resulting in anti‐viral and anti‐tumour activity in patients.2 The modification of human T cells with chimeric antigen receptors is another strategy that has yielded clinical benefit.3 Further advances in these technologies would be accelerated by the availability of a mouse model that would permit human T‐cell transduction protocols and proposed genetic modifications to be evaluated in vivo in terms of long‐term receptor expression, T‐cell survival, and immunological activity.

NOD‐scid IL2rγ ^null (NSG) mice allow a high degree of engraftment of human cells and tissues, as they lack both adaptive immunity and natural killer cell activity.4 Engraftment of NSG mice with human haematopoietic stem cells supports the development of a human immune system and has facilitated clinically relevant investigations in the fields of infectious disease5 and transplantation,6 among others. NSG mice also accept human peripheral blood mononuclear cells (PBMC)7 and mature T cells.8 However, in these cases, long‐term studies have been hampered by xenogeneic graft‐versus‐host disease (GVHD) that can occur soon after cell transfer.7, 8, 9 Furthermore, in the case of TCR‐transduced cells, the murine host would ideally express the human MHC molecule that restricts the TCR under investigation.

Given the growing use of genetically modified T cells in the treatment of human disease,1, 2, 3 we worked to develop a system that would permit the activity of such T cells to be evaluated in vivo. Mature human CD4⁺ T cells were modified by lentiviral transduction to express the well‐characterized human TCR 164, which recognizes a peptide derived from the type 1 diabetes autoantigen glutamic acid decarboxylase (GAD) in the context of HLA‐DR4.10 The minimal epitope recognized by TCR 164 (GAD_555–567) is present in both GAD65 and GAD67, is identical in sequence in both humans and mice, and is part of a larger naturally processed and presented peptide.11 The TCR‐transduced T cells were transferred to NSG‐Ab⁰ DR4 mice,8 which express HLA‐DR4 and are deficient in murine class II MHC molecules. The latter modification was incorporated into our model as it has been shown to reduce xenogeneic GVHD mediated by CD4⁺ T cells.8 Here we report the long‐term survival (up to 3 months) of transduced primary human CD4⁺ T cells in NSG‐Ab⁰ DR4 mice in the absence of lethal xenogeneic GVHD. This favourable outcome was dependent upon the pre‐transfer T‐cell transduction and culture conditions, which influenced both the kinetics of engraftment and the development of GVHD.

Materials and methods

Cell lines and cell culture

Priess cells,12 a human B lymphoblastoid cell line expressing both HLA‐DR4 and HLA‐A2,13 were cultured in RPMI‐1640 medium (Invitrogen, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Hyclone, Logan, UT), 1 mm sodium pyruvate, 28 μm 2‐mercaptoethanol, and 1× non‐essential amino acids. Jurkat/MA cells (provided by E. Hooijberg) are a human TCR‐β chain‐deficient Jurkat derivative modified to express human CD8α and to contain a luciferase reporter gene controlled by nuclear factor of activated T cells.14 Jurkat/MA cells were maintained in Iscove's modified Dulbecco's medium (Invitrogen) supplemented with 10% FBS. Chinese hamster ovary (CHO) cells and CHO cells stably transfected to express murine DEC‐20515 (CHO/mDEC‐205; provided by C.G. Park) were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% FBS and 1× non‐essential amino acids. CHO/mDEC‐205 cells were maintained in 500 μg/ml Geneticin (Invitrogen). 293T cells16 were cultured in Dulbecco's modified Eagle's medium containing 10% FBS and 0·6 mm sodium pyruvate and used for lentiviral production at no more than 10 passages of a stock obtained from the American Type Culture Collection (Manassas, VA).

Lentiviral vector production

164 is a CD4⁺ T‐cell clone specific for HLA‐DR4/GAD_555–567 (NFFRMVISNPAAT) that was isolated from a person at risk for the development of type 1 diabetes.10 According to the nomenclature of the International ImMunoGeneTics (IMGT) Information System (http://www.imgt.org), the TCR‐α and TCR‐β chain gene usage of TCR 164 is TRAV19*01/TRAJ56*01 and TRBV5‐1*01/TRBJ1‐6*01/TRBD2*01, respectively. The TCR‐β and TCR‐α chain cDNA sequences from 164 were linked by the coding sequence for the 2A peptide from porcine teschovirus‐1 (P2A), followed by coding sequences for the 2A peptide from Thoseassigna virus (T2A) and green fluorescent protein (GFP), and then cloned into a lentiviral transfer construct regulated by the spleen focus‐forming virus promoter17 (Fig. 1a). Lentiviral vectors were produced by calcium phosphate transfection of 293T cells as previously described.17 Briefly, the transfer vector encoding TCR 164 and GFP was co‐transfected into 293T cells with a packaging construct expressing the gag and pol genes, as well as constructs expressing rev and the VSV‐G envelope. The culture medium was replaced 16 hr after transfection and the lentivirus‐containing supernatant was collected 24 and 48 hr later and filtered. Lentivirus was concentrated by ultracentrifugation, resuspended in sterile PBS, and frozen at −80° in aliquots until use. The transfer construct encoding the control HLA‐A2‐restricted TCR 1.9 A2B,18 specific for HLA‐A2/HIV‐1 p17gag_77–85 (SLYNTVATL; SL9), was provided by O. Yang. The TCR 164 and 1.9 A2B transfer constructs were codon‐optimized for expression in human cells.19

Figure 1.

Lentivector design and titring. (a) The promoter and coding regions of the glutamic acid decarboxylase (GAD) T‐cell receptor (TCR) lentivector are depicted to scale. Expression of TCR 164 is controlled by the spleen focus‐forming virus promoter (SFFV). The TCR β‐ and α‐chains are linked by the 2A peptide from porcine teschovirus‐1 (P2A), followed by the 2A peptide from Thoseassigna virus (T2A) and the coding sequence for green fluorescent protein (GFP). (b) Lentivirus was titred using transduction of Jurkat/MA cells with 10‐fold serial dilutions of virus and monitoring of transduction efficiency by GFP expression. The titre was calculated from the viral dilution (1 : 10⁷ in the example shown) yielding GFP expression in 1–10% of cells.

Jurkat/MA cell transduction and lentiviral titreing

Jurkat/MA cells14 were transduced in the presence or absence of 4 μg/ml polybrene (as specified) in 24‐well plates (10⁵ cells/well in 500 μl). After addition of lentivirus sufficient to obtain > 95% transduction, plates were centrifuged for 30 min at 1350 g and then incubated for 16 hr at 37°, after which 500 μl fresh medium without polybrene was added. Transduced cells were cultured an additional 3 days before analysis by flow cytometry. Jurkat/MA cells were also employed for lentiviral titreing. Cells were plated in six‐well plates (10⁵ cells/well) and transduced with 10‐fold serial dilutions of lentivirus. Transduction efficiency was determined by flow cytometric analysis of GFP expression. Titre was calculated using the viral dilution for which GFP expression was between 1 and 10%.17

Lentiviral transduction of primary human CD4⁺ T cells

Ficoll density gradient centrifugation was used to isolate PBMC from HLA‐DR4⁺ leucopacks (New York Blood Center). HLA‐DR4 expression was determined by flow cytometry using mouse monoclonal antibody NFLD20 (provided by S. Drover) and allophycocyanin‐ (APC‐) conjugated anti‐mouse IgG (H+L) secondary antibody (eBioscience, San Diego, CA). CD4⁺ T cells were isolated from PBMC by positive selection with anti‐human CD4 magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and cultured in complete RPMI‐1640 medium. Before transduction, CD4⁺ T cells were treated either with 30 ng/ml soluble anti‐CD3 (OKT3; eBioscience) and 1 μg/ml soluble anti‐CD28 (CD28.2; BD Biosciences, San Jose, CA) or with 10 ng/ml recombinant human interleukin 7 (rhIL‐7; Peprotech, Rocky Hill, NJ) and 5 ng/ml rhIL‐15 (Peprotech) for 2 days and 50 U/ml of rhIL‐2 (Peprotech, Rocky Hill, NJ) for 1 day. Cells were transduced with lentivirus (10–20 μl/ml) in 24‐well plates (5 × 10⁵ cells/well in 500 μl) in the presence or absence of 8 μg/ml polybrene (as specified) while being centrifuged at 1350 g for 1 hr. The next day, an equal volume of complete medium with 50 U/ml IL‐2 was added and cells were cultured for an additional 3 days before evaluation by flow cytometry.

Peptide/MHC tetramers

HLA‐DR4 (HLA‐DRA1*01 : 01/HLA‐DRB1*04 : 01) monomers were produced in Drosophila S2 cells, biotinylated, loaded with peptide for 48–72 hr, and tetramerized by incubation with phycoerythrin‐ (PE‐) labelled streptavidin as described previously.21 For detection of TCR 164‐transduced cells, a PE‐labelled HLA‐DR4/GAD_{555–567 (557I)} (NFIRMVISNPAAT) tetramer was used. The irrelevant HLA‐DR4 tetramer used was loaded with an influenza haemagglutinin peptide (HA_307–319; PKYVKQNTLKLAT).

Analysis of transduced cells by flow cytometry

Transduced Jurkat/MA and primary human CD4⁺ T cells were stained for 30 min on ice with an antibody to human TCR‐αβ (T10B9.1A‐31; BD Biosciences) as well as an antibody to detect the β‐chain of TCR 164 (anti‐Vβ5.1‐PE; IM2285; Beckman Coulter, Brea, CA). The α‐chain of TCR 164 was detected using mouse anti‐Vα12.1 (nomenclature according to Arden22 rather than the International IMGT Information System; clone 6D6.6; Thermo Fisher, Rockford, IL) and anti‐mouse IgG‐APC (H+L) secondary antibody (eBioscience). Transduced cells were stained with HLA‐DR4/GAD_{555–567 (557I)} or HLA‐DR4/HA_307–319 tetramer‐PE (0·5 mg/ml) at 1 μl per 50 μl of cells for 1–3 hr at 37° in the dark. DAPI (Sigma‐Aldrich, St Louis, MO) was added immediately before acquisition on a BD LSRII flow cytometer, allowing for the exclusion of dead cells from further analysis. Data were analysed with flowjo software (Treestar, Ashland, OR).

Luciferase assay

Priess cells12 were loaded with peptides for 1 hr at room temperature. An equal number of transduced Jurkat/MA cells (> 95% GFP⁺ as determined by flow cytometry) were added and cells were co‐cultured for 16 hr at 37°. In some experiments, cells were non‐specifically stimulated with 100 ng/ml PMA and 200 ng/ml ionomycin. The cells were washed with PBS, then lysed with Reporter Lysis Buffer (Promega, Madison, WI), and frozen at −80° and then thawed to complete lysis. Luciferase was quantified using the Promega Luciferase Assay System. Luminescence was measured using a Victor plate reader.

Human interferon‐γ ELISPOT

Priess cells12 were plated at 5 × 10³ cells/well in 50 μl complete RPMI in a 96‐well multiscreen filter plate (Millipore, Billerica, MA) pre‐coated with anti‐human interferon‐γ (hIFN‐γ) antibody (MAB285; R&D Systems, Minneapolis, MN) and blocked with 1% BSA, and loaded with the indicated peptides for 1 hr at room temperature. Transduced human CD4⁺ T cells or mock‐transduced cells were added at 2 × 10³ cells/well in 50 μl complete RPMI and incubated for 40 hr. In some experiments, cells were non‐specifically stimulated with 100 ng/ml PMA and 200 ng/ml ionomycin. The IFN‐γ was detected with a second, biotinylated anti‐hIFN‐γ antibody (BAF285; R&D Systems), and spots were developed using streptavidin‐alkaline phosphatase (Zymed Laboratories, Carlsbad, CA) and 5‐bromo‐4‐chloro‐3‐indolyl‐phosphate/nitro‐blue tetrazolium substrate (Sigma‐Aldrich). Spots were counted using an automated ELISPOT reader system (Autoimmun Diagnostika, Strassberg, Germany).

Adoptive transfer of transduced human primary CD4⁺ T cells to immunodeficient hosts

Before lentiviral transduction, purified CD4⁺ T cells were treated either with 30 ng/ml soluble anti‐CD3 (OKT3; eBioscience) and 1 μg/ml soluble anti‐CD28 (CD28.2; BD Biosciences) or with 10 ng/ml rhIL‐7 (Peprotech) and 5 ng/ml rhIL‐15 (Peprotech) for 2 days and 50 U/ml of rhIL‐2 (Peprotech) for 1 day. Cells were transduced with lentivirus (10–20 μl/ml) in 24‐well plates (5 × 10⁵ cells/well in 500 μl) in the absence of polybrene while being centrifuged at 1350 g for 1 hr. The next day, an equal volume of complete medium with 50 U/ml IL‐2 was added and cells were cultured for an additional 3 days before transfer to immunodeficient murine hosts.

NOD.Cg‐Prkdc ^scid  Il2rg ^tm1Wjl H2‐Ab1 ^tm1Gru Tg(HLA‐DRB1)31Dmz/SzJ (NSG‐Ab⁰ DR4) mice8 were obtained from The Jackson Laboratory (Bar Harbor, ME) and maintained with the HLA‐DR4 transgene in the hemizygous state. Transduced human HLA‐DR4⁺ CD4⁺ T cells (4 × 10⁶ to 5 × 10⁶) were combined with similarly cultured CD4⁺ T‐cell‐depleted PBMC from the same donor (1 × 10⁶ to 8 × 10⁶, depending on availability) and transferred through the tail vein into NSG‐Ab⁰ DR4 mice. Blood was taken weekly from the tail and analysed for the presence of hCD45⁺ cells by flow cytometry. At the end of each experiment (week 5 for cells cultured with IL‐7 and IL‐15; weeks 11–12 for cells cultured with anti‐CD3 and anti‐CD28), blood, spleen and pancreas were analysed for engraftment by flow cytometry. Cells were stained with anti‐human CD45‐V450 (HI30; BD Biosciences), anti‐human CD4‐PE (RPA‐T4; BD Biosciences), anti‐human CD4‐Peridinin chlorophyll protein‐Cy5.5 (RPA‐T4; BD Biosciences), anti‐human CD8‐APC (RPA‐T8; BD Biosciences), PE‐labelled HLA‐DR4/GAD_{555–567 (557I)} or HLA‐DR4/HA_307–319 tetramer, anti‐mouse CD45‐PE‐Cy7 (30‐F11; BD Biosciences), and Live/Dead Fixable Yellow (Invitrogen) before fixation with 1% paraformaldehyde.

For assessment of insulitis, pancreata were fixed with Bouin's solution for 24 hr, embedded in paraffin, sectioned at non‐overlapping intervals, and then stained with aldehyde fuchsin and haematoxylin & eosin. A minimum of 17 islets per mouse were scored as described:23 0, no insulitis; 1, local insulitis without infiltration of islet itself; 2, < 25% infiltration; 3, 25–75% infiltration; or 4, > 75% infiltration. An insulitis index was calculated by adding the scores of all islets and dividing by four times the number of islets scored.

Animal procedures were approved by the Institutional Animal Care and Use Committee of Albert Einstein College of Medicine.

Peptide‐linked anti‐DEC‐205 antibodies

The constructs encoding the heavy and light chains of anti‐DEC‐205/SL9 and anti‐DEC‐205/GAD_{555–567 (557I)} were prepared as described,24 except that oligonucleotides encoding SLYNTVATL or NFIRMVISNPAAT were used, respectively. Peptide‐linked antibodies were produced in 293T cells by transient transfection as described.24 Binding of the peptide‐linked anti‐DEC‐205 antibodies to DEC‐205 was verified upon incubation with CHO/mDEC‐205 cells15 followed by anti‐mouse IgG‐APC (H+L) secondary antibody (eBioscience) and analysis by flow cytometry. For experiments examining the ability of the peptide‐linked anti‐DEC‐205 antibodies to stimulate transduced human CD4⁺ T cells, HLA‐DR4‐expressing dendritic cells were isolated from the spleens of NOD × NSG‐Ab⁰ DR4 mice using anti‐CD11c magnetic beads (Miltenyi), treated with lipopolysaccharide, and used as antigen‐presenting cells in a human IFN‐γ ELISPOT assay (2 × 10⁴ dendritic cells and 2 × 10⁴ T cells per well).

Statistical analysis

Statistical analyses were performed using graphpad prism 6 (GraphPad Software, San Diego, CA).

Results

Establishment of TCR expression in the TCR‐deficient human T‐cell line Jurkat/MA using lentiviral transduction

Given the expanding use of genetically engineered T cells in the treatment of human disease,1, 2, 3 we worked to develop systems that would permit the activity of such T cells to be evaluated first in vitro and then in vivo. To this end, we developed a VSV‐G‐pseudotyped, HIV‐1‐based, third‐generation lentivirus encoding the well‐characterized human TCR 164,25 which is specific for HLA‐DR4/GAD_555–567. In this lentiviral vector, expression of TCR 164 was controlled by the spleen focus‐forming virus promoter (Fig. 1a). The TCR‐β and TCR‐α chains were linked by the 2A peptide from porcine teschovirus‐1 (P2A) to permit equimolar expression of both chains,26 followed by the 2A peptide from Thoseassigna virus (T2A) and the coding sequence for GFP. Lentivirus was titred using transduction of Jurkat/MA cells with 10‐fold serial dilutions of virus and monitoring of transduction efficiency by GFP expression (Fig. 1b). The titres were calculated from the viral dilutions yielding GFP expression in 1–10% of cells and ranged from 3·6 × 10⁹ to 25 × 10⁹ transducing units/ml.

Jurkat/MA is a TCR‐β‐deficient variant of the human T‐cell line Jurkat that has been modified to express a luciferase reporter gene controlled by nuclear factor of activated T cells.14 Because Jurkat/MA does not express an endogenous TCR on its surface, it can be used not only for lentiviral titring, but also to monitor both the expression and function of a lentivector‐encoded TCR. As shown in Fig. 2(a), the transduction efficiency, as measured by GFP expression, was nearly 100%. Staining of the transduced Jurkat/MA cells with a pan‐human TCR antibody or with TCR Vα12.1 and TCR Vβ5.1 antibodies revealed excellent expression of the GAD TCR (Fig. 2a). To ensure that the TCR could bind its cognate antigen, transduced cells were stained with an HLA‐DR4/GAD_{555–567 (557I)} tetramer or an irrelevant HLA‐DR4/HA_307–319 tetramer. The GAD tetramer showed binding to a subset of the transduced Jurkat/MA cells (Fig. 2a), with the tetramer‐negative population probably reflecting pairing of the transduced TCR‐β chain with the endogenous Jurkat/MA TCR‐α chain. Combining data from multiple experiments, 95·7 ± 3·0% (n = 5) of all cells were positive for GFP. Of the GFP‐positive cells, 99·8 ± 0·1% (n = 3) were positive for TCR Vβ5.1, 77·9 ± 15·4% (n = 4) were positive for TCR Vα12.1, and 75·7 ± 11·1% (n = 2) were tetramer‐positive. Finally, to ensure that the lentiviral‐encoded GAD TCR was capable of transducing a signal, HLA‐DR4‐expressing Priess cells were pulsed with the GAD_555–567 peptide or the superagonist GAD_{555–567 (557I)}, which shows enhanced MHC binding,27 and then co‐cultured with GAD TCR‐transduced Jurkat/MA cells. T‐cell activation was monitored by measuring luciferase activity. GAD TCR‐transduced cells produced luciferase in response to the GAD peptides, whereas cells transduced with the irrelevant 1.9 A2B TCR,18 specific for HLA‐A2/HIV‐1 p17gag_77–85 (SL9), did not (Fig. 2b). Note that as Priess cells express HLA‐A2 in addition to HLA‐DR4,13 the HIV TCR‐transduced cells were activated to produce luciferase in the presence of SL9 (Fig. 2b). These results demonstrate that functional lentiviral‐encoded TCR 164 can be efficiently expressed in Jurkat/MA cells, as demonstrated by its ability to bind its cognate peptide/MHC tetramer and to mediate T‐cell activation in the presence of antigen.

Figure 2.

Lentivector‐mediated expression and function of a human T‐cell receptor (TCR) specific for HLA‐DR4/GAD _555–567 in Jurkat cells. (a) Jurkat/MA cells were transduced with the TCR 164 lentivector encoding both green fluorescent protein (GFP) and the human TCR 164, which is specific for HLA‐DR4/GAD _555–567. Cells were stained with the indicated phycoerythrin (PE) or allophycocyanin (APC) ‐labelled antibodies or PE‐labelled HLA‐DR4 tetramers. (b) Priess cells, which express HLA‐DR4 and HLA‐A2, were pulsed with GAD _555–567 (GAD), GAD _{555–567 (557I)} (GAD (I)), or HIV‐1 p17gag_77–85 (SL9), and co‐cultured with GAD TCR‐ or HIV TCR‐transduced Jurkat/MA cells. T‐cell activation was measured by luciferase activity. Graph depicts mean + SEM of two independent experiments; two replicates were performed in one experiment and three were performed in the other. ***P < 0·005, ****P < 0·0001.

Endowment of primary human T cells with reactivity against GAD_555–567 using lentiviral transduction

We next turned our attention to the transduction of primary human T cells, a more clinically relevant system. For these experiments, PBMC were isolated from HLA‐DR4‐positive leucopacks by Ficoll‐Paque centrifugation. CD4⁺ T cells were then magnetically purified by positive selection, cultured with anti‐CD3 and anti‐CD28 and IL‐2, and transduced using the TCR 164 lentivector. The viability of the cells immediately preceding transduction was 89·0 ± 0·4% (n = 3), and the viability at the end of the post‐transduction culture period was 96·1 ± 2·4% (n = 4). As shown in Fig. 3(a), the primary cells were efficiently transduced, as monitored by GFP expression. Expression of the lentivector‐encoded TCR was demonstrated by staining of the transduced cells with TCR Vα12.1 and TCR Vβ5.1 antibodies (Fig. 3a). Binding to the HLA‐DR4/GAD_{555–567 (557I)} tetramer, but not the irrelevant tetramer, was also observed (Fig. 3a). Combining data from multiple experiments, 68·3 ± 2·3% (n = 7) of all cells were positive for GFP. Of the GFP‐positive cells, 98·1 ± 0·5% (n = 4) were positive for TCR Vβ5.1, 77·0 ± 8·8% (n = 3) were positive for TCR Vα12.1, and 53·3 ± 1·6% (n = 3) were tetramer‐positive. Furthermore, the GAD TCR‐transduced cells released IFN‐γ in response to peptide‐pulsed Priess cells (Fig. 3b), demonstrating the successful endowment of these primary human T cells with reactivity against GAD_555–567.

Figure 3.

Lentivector‐mediated expression and function of T‐cell receptor (TCR) 164 in primary human CD4⁺ T cells. (a) Primary human CD4⁺ T cells were activated with anti‐CD3 and anti‐CD28, mock‐transduced or transduced with the TCR 164 lentivector, and stained with the indicated phycoerythrin (PE) ‐labelled antibodies or HLA‐DR4 tetramers. (b) GAD TCR‐ or HIV TCR‐transduced primary human CD4⁺ T cells were incubated with HLA‐DR4‐ and HLA‐A2‐expressing Priess cells pulsed with GAD _{555–567 (557I)} (GAD (I)), GAD _555–567 (GAD), or HA _307–319 (HA). Interferon‐γ (IFN‐γ) production was detected by ELISPOT. Graph depicts mean + SEM of duplicates. The results shown are representative of six experiments. *P < 0·05, ***P < 0·005. (c) Anti‐DEC‐205/GAD _{555–567 (557I)} was produced in 293T cells by transient transfection and purification from the culture supernatant. SDS–PAGE analysis was performed under reducing conditions. Lane 1, anti‐DEC‐205/GAD _{555–567 (557I)} after purification on a Protein G column; lane 2, bovine γ‐globulin. (d) Chinese hamster ovary (CHO) cells or CHO cells expressing DEC‐205 were stained with anti‐DEC‐205/GAD _{555–567 (557I)} followed by an allophycocyanin (APC) ‐labelled secondary antibody. (e) GAD TCR‐ or mock‐transduced primary human CD4⁺ T cells were incubated with HLA‐DR4‐expressing murine dendritic cells treated with lipopolysaccharide and GAD _{555–567 (557I)} (GAD (I)), anti‐DEC‐205/GAD (I), anti‐DEC‐205/SL9, or HA _307–319 (HA). IFN‐γ production was detected by ELISPOT. Graph depicts mean + SEM of duplicates. The results shown are representative of two experiments. *P < 0·05, ***P < 0·005.

We next tested the ability of GAD TCR‐transduced primary human CD4⁺ T cells to recognize GAD_555–567 when presented by dendritic cells that had processed the epitope from an exogenously added protein. We developed and characterized an anti‐mouse DEC‐205 antibody in which GAD_{555–567 (557I)} was fused to the C‐terminus of the antibody heavy chain. DEC‐205 is an endocytic receptor that is present on the CD8α ⁺ subset of murine dendritic cells28 and on all CD11c⁺ human dendritic cells.29 When analysed by SDS–PAGE under reducing conditions, the molecular weights of the heavy and light chains of the peptide‐linked anti‐DEC‐205 antibody were as expected (51 000 and 25 000, respectively) (Fig. 3c, lane 1). To verify that the fusion of the peptide with the heavy chain did not interfere with binding of the antibody to its target, the ability of the peptide‐linked antibody to bind to CHO cells expressing murine DEC‐205, but not to DEC‐205‐negative cells, was confirmed (Fig. 3d). When GAD TCR‐transduced primary human CD4⁺ T cells were incubated with HLA‐DR4‐expressing murine dendritic cells (isolated from NOD × NSG‐Ab⁰ DR4 mice) treated with lipopolysaccharide and anti‐DEC‐205/GAD_{555–567 (557I)}, IFN‐γ production was detected by ELISPOT (Fig. 3e), confirming the ability of the genetically modified T cells to respond to the GAD epitope processed by dendritic cells from an exogenously added protein.

Polybrene inhibits lentiviral transduction and expansion of primary human CD4⁺ T cells

The HIV‐1‐based, third‐generation lentiviral vector used for our work was originally described to be used for transduction in the presence of the cationic polymer polybrene.17 Hence, we included polybrene in our transduction protocol for all of the in vitro experiments described above. However, before transitioning our studies to the transfer of transduced cells to mice, we considered reports that polybrene can have deleterious effects on certain cell types30, 31 and investigated whether it could be eliminated without compromising transduction efficiency. Importantly, we found that elimination of polybrene enhanced both transduction efficiency (Fig. 4a) and T‐cell expansion (Fig. 4b). Reactivity of the GAD TCR‐transduced cells to GAD_555–567 was also verified (Fig. 4c). Given these findings, polybrene was eliminated from subsequent experiments.

Figure 4.

Polybrene inhibits lentiviral transduction and expansion of primary human CD4⁺ T cells. Primary human CD4⁺ T cells were activated with anti‐CD3 and anti‐CD28, mock‐transduced or transduced with the T‐cell receptor (TCR) 164 lentivector in the absence or presence of polybrene, and evaluated after six total days of culture for (a) transduction efficiency, (b) fold expansion, and (c) interferon‐γ (IFN‐γ) production in response to peptide stimulation. (a) Transduction efficiency was monitored by green fluorescent protein (GFP) expression. Representative FACS plots and a summary graph depicting mean + SEM of two independent experiments are shown. *P < 0·05, **P < 0·01. (b) Fold expansion was determined by cell counting. Graph depicts mean + SEM of two independent experiments. (c) Primary human CD4⁺ T cells were mock‐ or GAD TCR‐transduced in the absence of polybrene and then incubated with HLA‐DR4‐expressing Priess cells pulsed with GAD _{555–567 (557I)} (GAD (I)), GAD _555–567 (GAD), or HA _307–319 (HA). IFN‐γ production was detected by ELISPOT. Graph depicts mean + SEM of triplicates. The results shown are representative of two experiments. *P < 0·05, **P < 0·01, ***P < 0·005.

Long‐term survival of transduced primary human CD4⁺ T cells in NSG‐Ab⁰ DR4 mice and absence of lethal xenogeneic GVHD

The evaluation of genetically modified human T cells in a mouse model will require long‐term survival in vivo. To pursue this goal, PBMC were isolated from HLA‐DR4⁺ leucopacks and CD4⁺ T cells were purified by positive selection and transduced with the GAD TCR lentivector after stimulation with anti‐CD3 and anti‐CD28. As it has previously been reported that positively selected human CD4⁺ T cells do not engraft as well as PBMC in NSG recipients,8 we cultured the CD4⁺ T‐cell‐depleted PBMC from the same donor with anti‐CD3 and anti‐CD28 and co‐transferred them with the CD4⁺ cells to NSG‐Ab⁰ DR4 mice. Also, previous work established that injection of 10 × 10⁶ or more PBMC resulted in engraftment of all NSG recipients with human CD45⁺ cells.7 Hence, inclusion of CD4⁺ T‐cell‐depleted PBMC would help to address limiting cell numbers and enable engraftment of more mice.

Upon transfer to NSG‐Ab⁰ DR4 mice, both mock‐transduced and GAD TCR‐transduced cell populations were readily detectable in the blood beginning at 5–6 weeks post‐transfer and were still present at 11–12 weeks when the experiments were terminated (Fig. 5a,d). The quantity of hCD4⁺ cells, as a percentage of all hCD45⁺ cells, remained steady throughout this time period (Fig. 5b). Furthermore, mice remained active and healthy in appearance for the duration of the experiments and showed no evidence of hair loss, weight loss, or any other symptoms of xenogeneic GVHD. This is in sharp contrast to the finding that NSG recipients show signs of GVHD mediated by the transfer of unmanipulated PBMC between 20 and 45 days post‐transfer.7, 9 Importantly, GFP‐positive (i.e. transduced) cells were readily detectable up to 11–12 weeks post‐transfer (Fig. 5c,d). The long‐term survival of the transduced cells, coupled with the lack of xenogeneic GVHD, demonstrated a successful strategy for the evaluation of genetically modified human CD4⁺ T cells in vivo.

Figure 5.

Long‐term survival of transduced primary human CD4⁺ T cells in NSG‐Ab⁰ DR4 mice. Primary human CD4⁺ T cells were activated with anti‐CD3 and anti‐CD28, mock‐transduced or transduced with the T‐cell receptor (TCR) 164 lentivector in the absence of polybrene, combined with CD4⁺ T‐cell‐depleted peripheral blood mononuclear cells (PBMC) from the same donor, and transferred to NSG‐Ab⁰ DR4 hosts. (a–c) Blood was sampled at the indicated weeks post‐transfer and spleen and pancreas were harvested at weeks 11–12 and engraftment was evaluated by flow cytometry. Two mice received mock‐transduced cells and three received GAD TCR‐transduced cells. Graphs depict mean + SEM of all mice. A fourth mouse that received GAD TCR‐transduced cells had a splenic hCD45⁺ cell frequency of < 1% at weeks 11–12 and was excluded from analysis. Mice were pooled from two separate transfer experiments. (d) FACS plots from a mouse receiving GAD TCR‐transduced cells. Top panels, hCD45⁺ cell engraftment (% of total cells) is shown. Bottom panels, engraftment of GFP ⁺ (GAD TCR‐transduced) CD4⁺ cells (% of hCD45⁺ hCD4⁺ cells) is shown. (e) Pancreata of mice receiving either GAD TCR‐ or mock‐transduced cells were fixed, sectioned and stained with aldehyde fuchsin. Islets were scored and insulitis indices were determined for each mouse as described in Materials and methods. (f) Examples of islets from mice receiving GAD TCR‐transduced cells.

The source of TCR 164 was a CD4⁺ T‐cell clone isolated from a DR4⁺ individual at‐risk for the development of type 1 diabetes.10 The minimal epitope recognized by TCR 164 (GAD_555–567) is present in GAD65 and GAD67 in both humans and mice and is part of a larger naturally processed and presented peptide.11 When HLA‐DR4‐transgenic C57BL/6‐Ab⁰ Rag2^null mice were also made transgenic for TCR 164, pancreatic islet infiltration (insulitis) was observed.25 This suggested a potential biological read‐out for the evaluation of GAD TCR‐transduced cells in vivo. Indeed, the transfer of GAD TCR‐transduced cells was associated with islet infiltration (Fig. 5e,f). As a group, there was a trend toward greater insulitis in mice receiving GAD TCR‐transduced cells when compared with those receiving mock‐transduced cells, though this did not reach statistical significance.

Pre‐transfer T‐cell culture conditions influence in vivo engraftment and the development of GVHD

Lentiviral vectors have been used to transduce human T cells either with TCR activation or instead upon treatment with homeostatic cytokines including IL‐7 and IL‐15.32, 33 We investigated whether this second approach would lead to further improvements in in vivo outcome compared with cells treated with anti‐CD3 and anti‐CD28. As expected,32, 33 we found that transduction efficiency was greatly reduced when cells were not activated through their TCR before transduction (compare Figs 4a and 6a; transduction efficiency with anti‐CD3 and anti‐CD28 = 69·0 ± 2·6%; transduction efficiency with IL‐7 and IL‐15 = 8·3 ± 3·5%). Fold‐expansion of the cells also varied depending on the culture treatment (fold‐expansion with anti‐CD3 and anti‐CD28 = 4·5 ± 0·5; fold‐expansion with IL‐7 and IL‐15 = 1·5 ± 0·7). The CD4⁺ T‐cell‐depleted cell population was cultured similarly with IL‐7 and IL‐15 and transferred along with the mock‐transduced or GAD TCR‐transduced CD4⁺ cells to NSG‐Ab⁰ DR4 recipients. Under these conditions, engraftment of hCD45⁺ cells was rapid, with human cells detectable in blood as early as 2 weeks post‐transfer (Fig. 6b,c,e). At 5 weeks post‐transfer, all mice showed signs of xenogeneic GVHD, including hunched posture and limited activity. Two of seven recipients died before week 5 (Fig. 6g). Despite this outcome, GFP‐positive cells were detectable in mice receiving GAD TCR‐transduced cells up to 5 weeks post‐transfer (Fig. 6d,e). Splenocytes from the mice released IFN‐γ independently of antigen (Fig. 6f), probably because of the ongoing GVHD process. These results nonetheless serve to document the function of transferred cells up to 5 weeks post‐transfer. These findings also demonstrate that treatment with anti‐CD3 and anti‐CD28 before transduction is the preferred approach for the in vivo evaluation of genetically modified human T cells, because their survival and biological activity can be monitored for up to 12 weeks post‐transfer.

Figure 6.

Primary human CD4⁺ T cells cultured with interleukin‐7 (IL‐7) and IL‐15 show reduced transduction efficiency but rapid engraftment of NSG‐Ab⁰ DR4 hosts. (a) Primary human CD4⁺ T cells were cultured with IL‐7 and IL‐15, transduced with the T‐cell receptor (TCR) 164 lentivector in the absence of polybrene, and transduction efficiency was monitored by green fluorescent protein (GFP) expression. (b–d) Primary human CD4⁺ T cells were cultured with IL‐7 and IL‐15, mock‐transduced or transduced with the TCR 164 lentivector in the absence of polybrene, combined with CD4⁺ T‐cell‐depleted peripheral blood mononuclear cells (PBMC) from the same donor, and transferred to NSG‐Ab⁰ DR4 hosts. Blood was sampled at the indicated weeks post‐transfer and spleen and pancreas were harvested at week 5 and engraftment was evaluated by flow cytometry. Three mice received mock‐transduced cells and four received GAD TCR‐transduced cells. Graphs depict mean + SEM of all mice surviving at a given time‐point, except that mice with a hCD45⁺ cell percentage < 0·5% for a given time‐point were excluded from subsequent analysis for that time‐point. Mice were pooled from two separate transfer experiments. (e) FACS plots from a mouse receiving GAD TCR‐transduced cells. Top panels, hCD45⁺ cell engraftment (% of total cells) is shown. Bottom panels, engraftment of GFP ⁺ (GAD TCR‐transduced) CD4⁺ cells (% of hCD45⁺ hCD4⁺ cells) is shown. (f) Splenocytes (10⁵/well) from mice receiving mock‐ or GAD TCR‐transduced T cells were pooled and incubated with Priess cells (10⁴/well) pulsed with GAD _{555–567 (557I)} (GAD (I)), GAD _555–567 (GAD), or HA _307–319 (HA). IFN‐γ production was detected by ELISPOT. Graph depicts mean + SEM of technical replicates. (g) Survival curve of mice receiving mock‐ or GAD TCR‐transduced T cells.

Discussion

NSG mice, lacking both adaptive immunity and multiple important aspects of innate immunity, are a preferred host for the engraftment of human cells and tissues.4 The translational uses of this model have been further increased in recent years by the transgenic expression of human class I34, 35 or class II MHC molecules.8 However, in the case of adoptive transfer of human T cells or PBMC to NSG mice, lethal xenogeneic GVHD has historically made long‐term studies difficult.7, 8, 9 Recently, advances have been made in this area. In a model of accelerated xenogeneic GVHD in which NSG mice are lightly irradiated before transfer of human PBMC, lethal GVHD was slowed (though not prevented) in recipients deficient for either murine class I (NSG‐β2m ^null mice) or class II (NSG‐Ab⁰ mice) MHC.9 Here we took advantage of some of these recent improvements to develop a system for the long‐term in vivo study of genetically modified human T cells.

We found that lentivirally transduced human CD4⁺ T cells could be detected up to 12 weeks post‐transfer in NSG‐Ab⁰ DR4 recipients, and that lethal GVHD was not observed. However, this outcome only occurred when cells were activated with anti‐CD3 and anti‐CD28 in vitro, but not when they were instead cultured with IL‐7 and IL‐15. Stimulation with anti‐CD3 and anti‐CD28 also led to improved lentiviral transduction efficiency and cell expansion in vitro. When cells instead were pretreated with IL‐7 and IL‐15, human cell engraftment in NSG‐Ab⁰ DR4 recipients occurred more rapidly. However, this led to lethal GVHD as early as 2 weeks post‐transfer, limiting the utility of this approach for long‐term studies. Our findings suggest that treatment in culture influences the subsequent survival, trafficking patterns, and/or activity of the T cells in vivo. For example, as has been suggested by others,32 in vitro T‐cell expansion upon TCR stimulation could limit the subsequent in vivo proliferative potential of adoptively transferred T cells or alter their immunological activity. It is also possible that treatment with IL‐7 and IL‐15 may favour the survival of cell populations that are most important in mediating xenogeneic GVHD. Given the potential clinical importance of our findings, further investigations should be undertaken to identify the underlying mechanism(s).

Our work should find utility as a model system for the long‐term examination of TCR‐modified2 or chimeric antigen receptor‐modified human CD4⁺ T cells,3 potentially combined with patient‐derived cancer xenografts, also accepted by NSG hosts.36 This is timely, as recent reports have suggested the potential clinical benefit of adoptive CD4⁺ T‐cell immunotherapy for cancer,37, 38, 39 whereas most adoptive cell therapy for cancer has concentrated on CD8⁺ T cells.40 We have previously reported that human CD8⁺ T cells modified by lentiviral transduction to express an autoreactive TCR can be detected in HLA‐A2‐transgenic NSG hosts up to 5 weeks post‐transfer.41 Those experiments were not carried out further because of concerns about xenogeneic GVHD. However, given our findings reported here, it is possible that longer‐term evaluation of genetically modified human CD8⁺ T cells in NSG hosts may also be possible. Utilization of murine class I MHC‐deficient hosts (NSG‐β2m ^null mice, for example) will probably facilitate such studies.9 For our in vivo experiments, we tracked transduced cells based on GFP expression for simplicity. This should be a good surrogate for expression of the lentivirally encoded TCR, given our in vitro results that of the GFP‐positive cells, 98·1 ± 0·5% were positive for TCR Vβ5.1, 77·0 ± 8·8% were positive for TCR Vα12.1, and 53·3 ± 1·6% were tetramer‐positive. Because GFP is linked to the TCR by a ‘self‐cleaving’ T2A peptide, rather than an internal ribosome entry site, translation of GFP is dependent on translation of the TCR. Nonetheless, in our future in vivo experiments, we will monitor the expression of the lentivirally encoded TCR more directly.

In addition to the development of a mouse model to permit human T‐cell transduction protocols and genetic modifications to be evaluated in vivo, our current efforts are also motivated by an interest in the creation of mouse models for human diseases, such as type 1 diabetes, that incorporate human T cells. Here we show that the adoptive transfer of GAD TCR‐transduced cells was associated with insulitis in some NSG‐Ab⁰ DR4 recipients. Though promising, diabetes has not yet been observed. Also, although there was a trend for mice receiving GAD TCR‐transduced cells to exhibit greater islet infiltration compared with those receiving mock‐transduced cells, statistical significance was not reached. Whether the infiltration observed in mice receiving mock‐transduced cells represents true islet reactivity, or instead tissue infiltration as a consequence of sub‐clinical xenogeneic GVHD, will require further study. Further improvements that will facilitate the in vivo evaluation of the function of GAD TCR‐transduced cells will be investigated in the future. For example, the expression of an exogenous TCR in mature human T cells may be reduced due to mispairing between endogenous and exogenous TCR chains.42 In our case, only some of the TCR 164‐transduced T cells were able to bind the HLA‐DR4/GAD_{555–567 (557I)} tetramer, suggesting the occurrence of some degree of mispairing. At least for some TCRs, replacement of the human TCR α‐ and β‐chain constant regions with their murine counterparts has been shown to reduce mispairing and enhance the activity of the transduced human T cells;42 this is one strategy that will be explored in our model. Administration of anti‐DEC‐205/GAD_{555–567 (557I)} under immunizing conditions to mice receiving GAD TCR‐transduced cells will also be examined. We will also work to determine whether conditions can be developed that will allow more rapid repopulation of NSG hosts than observed with our current anti‐CD3/anti‐CD28 stimulation protocol, while at the same time still avoiding lethal GVHD. For example, transient elimination of mouse macrophages with clodronate liposomes has been shown to enhance engraftment of human cells in immunodeficient mice43 and will be explored in this regard. Future investigations to identify the mechanism(s) underlying our findings will likely also suggest additional improvements to further enhance the translational utility of our model.

Disclosures

M.A.B. and D.L.G. receive grant support and are consultants for The Jackson Laboratory. The remaining authors have no financial or commercial conflicts of interest to disclose.