|Home | About | Journals | Submit | Contact Us | Français|
Persistence of T cells engineered with chimeric antigen receptors (CARs) has been a major barrier to use of these cells for molecularly targeted adoptive immunotherapy. To address this issue, we created a series of CARs that contain the TCR-ζ signal transduction domain with the CD28 and/or CD137 (4-1BB) intracellular domains in tandem. After short-term expansion, primary human T cells were subjected to lentiviral gene transfer, resulting in large numbers of cells with >85% CAR expression. In an immunodeficient mouse xenograft model of primary human pre-B-cell acute lymphoblastic leukemia, human T cells expressing anti-CD19 CARs containing CD137 exhibited the greatest anti-leukemic efficacy and prolonged (>6 months) survival in vivo, and were significantly more effective than cells expressing CARs containing TCR-ζ alone or CD28-ζ signaling receptors. We uncovered a previously unrecognized, antigen-independent effect of CARs expressing the CD137 cytoplasmic domain that likely contributes to the enhanced antileukemic efficacy and survival in tumor bearing mice. Furthermore, our studies revealed significant discrepancies between in vitro and in vivo surrogate measures of CAR efficacy. Together these results suggest that incorporation of the CD137 signaling domain in CARs should improve the persistence of CARs in the tumor microenvironment and hence maximize their antitumor activity.
With the advent of efficient gene transfer technologies, such as murine retroviral and HIV-derived lentiviral vectors, it has become feasible to confer novel antigenic specificity to T cells by transfer of chimeric antigen receptors (CARs) with stable, long-term expression. This technology has been used to generate T cells specific for HIV and several human tumor antigens, and some of these engineered T cells have been tested in Phase I/II studies in humans demonstrating the feasibility and relative safety of this approach [1–3]. One study has demonstrated anti-tumor activity in patients with neuroblastoma given a single CAR infusion .
CARs combine the antigen recognition domain of antibody with the intracellular domain of the TCR-ζ chain or FcγRI protein into a single chimeric protein that are capable of triggering T-cell activation in a manner very similar to that of the endogenous TCR [5, 6]. Several studies demonstrate that the addition of co-stimulatory domains, particularly the intracellular domain of CD28 can significantly augment the ability of these receptors to stimulate cytokine secretion and enhance antitumor efficacy in pre-clinical animal models using both solid tumors and leukemia that lack the expression of the CD28 receptor ligands CD80 and CD86 [7–9]. Inclusion of domains from receptors such as the TNF receptor family members, CD134(OX-40) and CD137 (4-1BB) into CARs has also been shown to augment CAR-mediated T-cell responses [10, 11]. Gene transfer approaches using these engineered CARs may therefore provide significant improvements over current adoptive immunotherapy strategies that must rely upon the endogenous TCR specificities, for which significant issues of TCR repertoire limitation and impaired tumor MHC class I expression may exist.
In this study, we have addressed the issue of limited in vivo persistence of CARs by defining the relative contributions of TCR-ζ, CD137 and CD28 signaling domains in mice engrafted with hematopoietic malignancies. We chose the human CD19 antigen as our initial target for several reasons: 1) CD19 displays a pattern of expression that is highly restricted to B-cells and B-cell progenitor cells , 2) CD19 does not appear to be expressed by hematopoietic stem cells permitting the targeting of the B-cell lineage without affecting other hematopoietic lineages , and 3) CD19 is widely expressed by malignant cells that are derived from the B cell lineage including most lymphomas and lymphocytic leukemias . After optimizing the generation of CARs with an efficient T-cell culture process, in vitro studies indicate that incorporation of either CD28 or 4-1BB signaling domains enhances activity over TCR–ζ, confirming previous studies. In contrast, compared to CARs that contain CD28, our in vivo studies indicate that CARs containing CD137 have superior antileukemic efficacy and improved persistence in a primary human acute lymphoblastic leukemia xenograft model. Furthermore, we also find that CARs expressing CD137 signaling domains can provide significant activity that appears to be antigen independent and may contribute to the efficacy of CARs in vivo.
Lentiviral vectors can transfer genes with into activated CD4+ and CD8+ human T cells with high efficiency but expression of the vector-encoded transgene depends on the internal promoter that drives its transcription. Therefore, successful CAR expression and gene therapies with CAR-expressing T cells rely on the ability of T cells to maintain adequate receptor expression over long periods of time. We tested several promoters to identify the one with the highest stable expression in both primary CD4+ and CD8+ T cells. Transduction was performed at limiting dilution to ensure that the cells have a single integrated vector per cell (data not shown). While the CMV promoter exhibited high levels of expression of a GFP transgene early after transduction, expression decreased to < 25% of the initial expression after 10 days of culture (Fig. 1b). The distribution of CMV-driven GFP expression was also quite variable compared with the other promoters tested (Suppl. Fig. S1). In contrast, the EF-1α promoter not only induced the highest level of GFP expression but also optimally maintained it in both CD4 cells and CD8 cells (Fig. 1b). These findings confirmed and extended other studies in primary human T cells . The EF-1α promoter was therefore selected for all future studies using CARs. By using lentiviral vectors and transductions at an MOI of 5, the different CARs could be expressed high expression in > 85% primary human T cells (Fig. 1c). Western blotting under both reducing and non-reducing conditions demonstrated that the CARs are present as both covalent dimers and monomers within T cells (Suppl. Fig. S2). Using the artificial bead-based APC system previously described by our laboratory , > 50-fold expansion of CAR+ T cells could be achieved over the course of transduction and growth in ~10 days (Fig. 1d).
To enhance the functionality of the immunoreceptor, we introduce the signal transduction domains of CD28 or CD137 in the TCR-ζ containing CAR (Fig. 1a). Similar to data reported by other groups [11, 17], the introduction of costimulatory domains into CARs does not improve the antigen-specific cytotoxicity triggered by these receptors (Fig. 2b). Lytic activity of transduced T cells against K562 target cells expressing CD19 correlated with the transduction efficiency of the T cells (data not shown). CAR-triggered cytotoxicity is antigen-specific with only negligible lysis of wild-type K562 cells that lack expression of the CD19 antigen (Fig. 2a). CAR+ T cells are also able to efficiently kill primary pre-B ALL cells that express physiologic levels of CD19 (Fig. 2b). Of note, these primary ALL cells lack expression of endogenous CD80 or CD86 (data not shown).
Following CAR activation with CD19+ K562 cells, CD4+ T cells expressing CARs produced abundant quantities of IL-2 and IFN-γ (Fig. 3) comparable to cells stimulated via the endogenous TCR and CD28 receptors (data not shown). T cells expressing CD28 and CD137 domain containing CARs produced greater quantities of IL-2 when compared with cells expressing the αCD19-ζ receptor (Fig. 3). The production of the type 2 cytokines, IL-4 and IL-10, by CD4+ T cells was also stimulated by all of the CARs tested; however, the levels of these cytokines were much lower, consistent with the Th1-like phenotype of T cells generated by anti-CD3 and CD28 stimulatory beads . It was notable that the incorporation of the CD137 domain into CARs decreased the production of these type 2 cytokines, consistent with previous reports of the 4-1BB signaling pathway in natural T cells . All CARs stimulated IFN-γ production by CD8+ T cells. These findings confirm that the addition of co-stimulatory domains into CARs modulates cytokine secretion in a manner that is dependent on the type of costimulatory domain [7, 8, 17, 19]. However, it is less well appreciated that the pattern of cytokine expression is altered by incorporation of different signal transduction domains into the CARs. These differences may have important consequences for the functionality of T cells engineered to express CARs.
The generation of a robust and sustained anti-tumor immune response requires not only triggering of cytotoxicity and cytokine production but also stimulation of T-cell proliferation. To assess the relative contribution of different costimulatory domains to proliferative signals delivered by CARs, we engineered primary human T cells to express CARs in conjunction with GFP to permit evaluation of both CAR+ and CAR− T cells in the same culture. Following T-cell re-stimulation with CD19+ K562 (K562-CD19 cells), T cells expressing the αCD19-28-ζ receptor exhibited proliferation comparable to that obtained with full stimulation of the endogenous TCR complex with K562 cells loaded with anti-CD3 and CD28 antibodies, a condition shown previously to support long-term expansion of primary human T cells (KT32-BBL)  (Fig. 4a[v]). The αCD19-28-BB-ζ triple receptor also stimulated CD19 driven proliferation (Fig. 4a[iv]), but to a lesser extent that the αCD19-28-ζ double costimulatory receptor. No significant proliferation was observed when these same T cells were stimulated with wild-type K562 cells lacking the CD19 antigen (K562 wt). As previously shown by other investigators [17, 19, 21, 22], T cells expressing the αCD19-ζ receptor showed little proliferation upon exposure to the surrogate CD19 antigen (Fig. 4a[ii]), demonstrating the dependence of CAR-driven proliferation on co-stimulatory signals.
Unexpectedly, T cells containing the αCD19-BB-ζ double costimulatory domain CAR had significantly increased proliferative capacity during in vitro expansion independently of receptor ligation with the surrogate CD19 antigen (Fig. 4a[iii] and 4b). This increased proliferation was observed in both CD4+ and CD8+ T cells (data not shown), and it was associated with a prolonged blast phase after the initial stimulation and transduction, as revealed by a longer maintenance of an elevated mean cellular volume (Fig. 4c), a parameter that correlates well with log phase proliferation of T cells . These findings suggest that incorporation of the CD137 intracellular domain mediates antigen-independent activity that is similar to that provided by the natural 4-1BB receptor in T cells following ligation . As a result of the enhanced proliferation observed following the initial activation of T cells via αCD3/αCD28-coated beads used to enhance T-cell transduction [24, 25] (Fig 4b), CAR+ T cells expressing the αCD19-BB-ζ receptor had relatively low CAR-driven proliferation (Fig. 4a[iii]).
Other than the antigen-independent proliferation of the 4-1BB containing CAR, the above in vitro findings, in aggregate, suggested that the αCD19-28-ζ CAR would be the most effective receptor for generating a sustained anti-leukemic T cell response in vivo. We evaluated the in vivo efficacy of αCD19 CARs in vivo model of ALL (Fig. 5a) in which primary human pre-B ALL cells are engrafted into immunodeficient mice. In this model system, intravenous injection of primary ALL cells leads to development of progressive leukemia with significant involvement of the bone marrow, spleen and blood including leptomeningeal involvement (Fig. 5b and 5c) that eventually leads to the death of the animal [26, 27]. Primary human T cells also readily engraft within these animals, and engraftment of mock transduced human CD4+ and CD8+ T cells in leukemia-bearing animals has little or no impact on the development of leukemia (Fig. 5d and 5e). As little as 5 × 106 CAR+ T cells can significantly delay leukemia in most mice injected with ALL 2 weeks prior (Fig. 5d and 5e, p=0.008) compared with mock-transduced T cells. A dose dependent affect is also apparent with as little as 5×105 CAR+ cells showing an effect on development of leukemia (Fig. 5d). All of the CARs demonstrated potent anti-leukemic activity when 2×106 CAR+ T cells were injected two weeks after establishing leukemia in the mice (Fig. 5f). The treatment effect was significant for the αCD19-ζ CAR (p<0.05) and for CARs that expressed costimulatory domains (p<0.01).
In vitro observations suggested that the CD28 intracellular domain should permit CAR+ T cell proliferation in vivo when transduced with CARs that contain this domain. We therefore compared the in vivo efficacy of T cells expressing the αCD19-ζ, αCD19–28-ζ and αCD19-BB-ζ CARs by injecting 10 million bulk T cells (adjusted to 50% CAR+ T cells in order to follow the fate of CAR+ vs. CAR- cells) three weeks after establishment of leukemia in NOD-SCID-γ−/− mice. The T cells were engineered to express GFP as well as the CAR by using a vector that encodes these two genes separated by the 2A ribosomal skipping sequence to allow monitoring of CAR+ T cells. All CAR+ T cells retain significant anti-leukemic efficacy compared with mock-transduced T cells when limiting numbers of T cells necessary for engraftment are transferred (Fig. 5f); however, studies using higher numbers of CAR+ T cells reveal significant differences in the engraftment and persistence of the CAR+ T cells bearing different costimulatory domains (Fig. 6a-d). The total T cell counts were highest in mice after injection with αCD19-BB-ζ CAR+ T cells (Fig. 6a), and the T cells were comprised of CD4+ and CD8+ CAR+ T cells (Fig. 6b). After injection into leukemic animals, the proportion of α CD19-BB-ζ CAR+ T cells was significantly higher than αCD19-ζ CARs+ T cells (p<0.01), whereas it was notable that the proportion of αCD19-28-ζ CAR+ T cells were not higher than the αCD19-ζ only CARs. Interestingly, the enhanced engraftment and/or persistence of the αCD19-BB-ζ CAR+ CD4 and CD8 T cells was CD19 antigen independent, because it was also observed in animals that were not injected with ALL cells (Fig. 6c, p<0.05).
It is notable that the T cells expressing the αCD19-28-ζ receptor also did not exhibit greater anti-tumor efficacy compared with T cells expressing αCD19-ζ (Fig. 5f and and6d).6d). It is possible that this is because there is low level expression of CD86 on the pre-B ALL cells (Fig S3). In contrast, αCD19-BB-ζ expressing T cells demonstrated a significant enhancement in anti-leukemic efficacy compared with T cells expressing either the αCD19-ζ or αCD19-28-ζ receptors. Median leukemia free survival was increased by 7 weeks (Fig. 6d, p=0.009). Based upon an approximate doubling time of 2.7 days for pre-B ALL cells (derived by fitting the leukemic blast counts in untreated animals to an exponential growth model), this 7-week delay in onset of leukemia corresponds to a reduction in leukemia burden of >105–fold following T cell injection when compared with the burden present in animals receiving either the αCD19-ζ or α CD19-28-ζ modified T cells.
The above experiments suggested that the CD137 signaling domain confers an antigen independent effect to enhance the survival and/or proliferation of CAR T+ cells in vivo, and the results are consistent with the in vitro effects shown above in Fig. 4b. To further characterize this effect, a long term competitive engraftment experiment was carried out as shown in Fig. 7a and described in detail in the Supplementary Methods and Table S1. After establishing leukemia in the mice, a 1:1 mixture of T cells expressing either the αCD19-ζ or αCD19-BB-ζ CAR was injected three weeks later. Mice were injected with CAR T cells at 1, 5 and 20 × 106 cells per mouse; and the mice were bled and/or sacrificed at intervals between 5 weeks and 6 months after establishment of leukemia. There was a consistent enrichment for αCD19-BB-ζ in the spleens of the mice (Fig. 7b, p=0.0001) and other organs (not shown). The enrichment was independent of the level of engraftment in that the bias of the log αCD19-BB-ζto αCD19-ζ was consistent throughout a 3log10 range of engraftment (Fig. 7b).
A robust, dose dependent antileukemic treatment effect was observed in the mice given the mixture of T cells expressing the αCD19-ζ or αCD19-BB-ζ CARs (Fig. 7c). Peripheral blood analyzed between days 35 and 70 after establishment of leukemia showed that the mice treated at the 5 and 20 × 106 dose levels controlled the leukemia, while there was only partial control in the mice given 1 × 106 CAR T cells. However, even animals given the low dose of CARs had a significant treatment effects on day 57 and day 70, comparing blast counts in mice engrafted with the CAR+ T cell mixture to mice engrafted with equivalent number of mock transduced T cells (p<0.01). Long term engraftment of the CAR+ T cells was observed in the animals with controlled leukemia, as 3 of 7 mice examined more than 6 months after T cell transfer were still engrafted in the spleen (Supplementary Table S2) and other organs (not shown). There was no evidence of a change in the ratio of αCD19-ζ to αCD19-BB-ζ over time, or as a function of T cell dose (Supplementary Table S2). Finally, animal 245 (Supplementary Table S2) was necropsied on day 198, and found to be free of leukemia and to harbor CAR+ T cells in the spleen that had constitutive surface expression of the scFv at readily detectable levels (Fig. 7d).
Together, these results demonstrate that the αCD19-BB-ζ modified T cells persisted longer and had more vigorous anti-leukemic effects than CAR+ T cells that expressed CD28 signaling domains. In addition, while T cells expressing CARs that contain the TCR-ζ only or the CD28 costimulatory domain along with TCR-ζ are capable of killing ALL cells in vitro, their survival is significantly shorter than that of T cells with CARs expressing the CD137 signaling domain. Finally, there was no evidence for transformation of the CAR+ T cells that expressed the CD137 signaling domains over the course of the 6 month experiment shown in figure 7 and Supplementary Table S2.
Artificial chimeric immunoreceptors offer the possibility of reprogramming T cells for efficient targeting of tumors in an HLA-independent fashion. However, while initial clinical studies demonstrate feasibility with the retargeted T cells, poor in vivo persistence and low expression of the transgene have been documented, and these limitations have reduced potential clinical activity [2, 3, 28]. To address these issues, our studies have used a robust pre-clinical model, and we demonstrate that a single infusion of as few as 2 million engineered T cells could control and in some cases, eliminate pre-established disseminated leukemia. Surprisingly, expression of the CD137 signaling domain rather than the CD28 domain was most correlated with reprogramming T cells for persistence in vivo.
Previous in vitro studies have characterized the incorporation of CD137 domains into CARs [10, 11, 29]. Our results represent the first in vivo characterization of these CARs and uncover several important advantages of CARs that express CD137 that were not revealed by the previous in vitro studies. We demonstrated that CARs expressing the CD137 signaling domain could survive for at least 6 months in mice bearing tumor xenografts. This may have significant implications for immunosurveillance, as well as for tumor eradication. For example, in a mouse prostate cancer xenograft model, survival of CAR+ T cells for at least a week was required for tumor eradication .
Long term survival of the CARs did not require administration of exogenous cytokines, and these results significantly extend the duration of survival of human T cells expressing CARs shown in previous studies [17, 31]. To our knowledge, this is the first report demonstrating elimination of primary leukemia xenografts in a pre-clinical model using CAR+ T cells. Furthermore, complete eradication was achieved in some animals in the absence of further in vivo therapy, including prior chemotherapy or subsequent cytokine support.
The long-term control of well established tumors by immunotherapy has rarely been reported. Most pre-clinical models in a therapeutic setting have tested tumors that have been implanted for a week or less before initiation of therapy . After establishing leukemia two to three weeks before T cell transfer, we found that many animals had long-term control of leukemia for at least 6 months. The effect was specific for the chimeric receptor, as animals injected with an equivalent number of unmodified T cells had high blast counts within 3weeks (Fig. 5). The efficacy of targeted, adoptive immunotherapy in this xenograft model of primary human ALL compares favorably to our prior experience testing the antileukemic efficacy of single cytotoxic [27, and unpublished data] or targeted agents , where we have observed extension of survival but not cure of disease. Additionally, we have not previously observed the ability to control xenografted ALL for a period of as long as 6 months.
It is likely that several mechanisms account for the enhanced efficiency of the redirected T cells observed in the present report. First, previous studies have generally used T cells after a culture for a month or longer [2, 3, 28]. In the present work we have used an efficient bead based artificial APC, which shortens the culture to approximately 10 days, and permits the use of the T cells early at a time when we have shown previously that the average telomere length of the cultured T cells is actually longer than at the start of culture . We attribute this to the previous demonstration that the anti-CD28 driven culture system induces telomerase activity . Furthermore, the addition of CD28 to culture conditions promotes transduction of central memory T cells . CD28 bead-based cell expansion has the capacity to routinely generate >1010 CAR+ T cells in approximately 10 days using FDA compliant manufacturing procedures already in use for clinical trials in humans .
Second, previous studies have generally used murine retroviruses or electroporation to introduce the chimeric receptor [2, 3, 28]. We have used lentiviral gene transfer which permits highly efficient engineering of T cells with >85% successful gene transfer [24, 36]. As shown in our study, CAR expression was maintained for at least 6 months in vivo with no evidence of silencing using the EF-1α promoter. In comparison, murine retroviral vectors have been shown to exhibit significant silencing of gene expression over time, despite the incorporation of elements such as chromatin-insulator sequences.
Similar to other groups who have evaluated CARs incorporating costimulatory domains [9, 10, 17, 37−41], we confirmed that the addition of the CD28 intracellular domain into CARs enhances the in vitro proliferation and cytokine production of T cells stimulated through these receptors. Interestingly, the αCD19-BB-ζ CAR appears to antagonize the production of the type II cytokines, IL-4 and IL-10, but it remains unclear whether this was due to direct signaling or to selective outgrowth of Th1-like cells.
While the CD28 cytoplasmic domain has been reported to significantly enhance the anti-tumor efficacy of CAR-expressing T cells in a number of models, the present study is the first to show that incorporation of the CD137 cytoplasmic domain into a CAR augments in vivo efficacy. In other studies using CAR+ T cells and trans costimulation of CD28 and CD137 through genetic expression of CD80 and 4-1BBL, redirected T cells were also found to have potent antitumor effects . The significantly enhanced anti-leukemic activity in vivo is associated with the improved persistence of the CAR+ T cells. Our results indicate that enhanced survival and/or proliferation of CAR+ T cells contribute to the increased antitumor effects.
CD137 plays an important role in T cell proliferation and survival, particularly for T cells within the memory T cell pool . CD137 mediates its effects on T cell survival and proliferation through activation of the AKT/mTOR pathway  and the upregulation of the anti-apoptotic genes, Bcl-XL and BFL-1 . Surprisingly, the αCD19-28-ζ modified T cells failed to show a significant improvement in anti-leukemic efficacy in vivo using our primary pre-B ALL model compared with the αCD19-ζ modified T cells. This is in contrast to other models using solid tumors including B cell-derived tumors and a recent study by our laboratory using a mesothelin-directed CAR where CD28 domain-containing CARs show enhanced anti-tumor efficacy . Together, these studies suggest that the optimal signals produced by CARs may be dependent upon the particular tumor being targeted and/or the nature of the particular scFv antibody.
Our studies are the first to reveal antigen-independent effects of the αCD19-BB-ζ receptor on T cells. This receptor, while capable of triggering cytotoxicity in an antigen-dependent fashion, also significantly prolonged the initial blast-phase of T cell activation. There are several possible mechanisms by which the CARs could deliver antigen-independent signals. CARs, like some natural receptors, may deliver tonic ligand-independent signals. Impairment of the regulatory mechanisms that normally extinguish receptor signaling such as the SHP-1 and PTPH1 phosphatase that dephosphorylate the TCR-ζ ITAMs might be impaired, leading to the antigen-independent effects observed in this study. While CARs appear to exist predominantly as homodimers, these artificially-constructed receptors might also spontaneously aggregate into oligomers, especially at the high levels of expression possible with the EF-1α promoter.
The enhanced growth effects of the αCD19-BB-ζ receptor are consistent with the antigen-independent growth effects that are observed in T cells stimulated through the natural CD137 receptor by agonist monoclonal antibody [23, 46]. Normally, CD137 expression is tightly regulated on T cells with expression limited to a window of a few days following T cell activation or following IL-15 treatment . The 4-1BB/4-1BBL interaction has been proposed as one mechanism by which IL-15 mediates its effect on memory T cells under limiting CD137 expression . The antigen-independent signals derived from the CD137 domain within the CAR may be critical to the anti-leukemic effects observed in our study, analogous to the continued presence of a 4-1BB agonist antibody, and ectopic trans expression of 4-1BBL, both of which have been shown to promote antitumor effects in vivo . Although previously unreported, these antigen-independent effects of CARs have important implications for the clinical use of these receptors.
Lentiviral vectors that encode a mouse CD8-human CD28 chimeric protein and eGFP separated by the EMCV internal ribosomal entry sequence (IRES) under the transcriptional control of either the cytomegalovirus IE gene (CMV), elongation factor-1a (EF-1α), ubiquitin C (UbiC) or phosphoglycerokinase (PGK) promoter were generated by replacing the CMV promoter within the 3rd generation self-inactivating lentiviral vector plasmid, pRRL-SIN-CMV-eGFP-WPRE (Cell Genesys, So. San Francisco, CA) . The EF-1α promoter (derived from pTracer™-CMV2, Invitrogen, Carlsbad, CA), the UbiC promoter (derived from pHUG-1 lentiviral vector, a kind gift of Dr. Eric Brown, University of Pennsylvania) and the PGK promoter (derived from pRRLsin.sppt.PGK.GFP.pre, Cell Genesys) were all cloned into pRRL-SIN-CMV-eGFP-WPRE using PCR and standard molecular biology techniques.
Fig. 1a shows schematic diagrams of the CARs used in this study. All CARs contain an scFv that recognizes the human CD19 antigen. The cDNA for the CARs that contain a truncated form of the TCR-ζ intracellular domain (αCD19-Δζ), a full-length TCR-ζ domain (αCD19-ζ) or a TCR-ζ domain in cis with the intracellular domain of the 4-1BB receptor (αCD19-BB-ζ) were generated at St. Jude’s Childrens Research Hospital . These complete CAR sequences were amplified directly from the provided plasmids by PCR. Constructs containing the CD28 transmembrane and intracellular domain alone (αCD19-28-ζ) or in combination with the 4-1BB intracellular domain (αCD19-28-BB-ζ) were generated by the procedure of splicing by overlap extension (Supplementary Table S3). A plasmid encoding a mCD28-huCD28 chimeric protein  and the above constructs were used as templates for PCR. The resulting PCR fragments containing the complete CARs were then cloned into pELPS 3′ of the promoter using standard molecular biology techniques. pELPS is a derivative of the third generation lentiviral vector pRRL-SIN-CMV-eGFP-WPRE in which the CMV promoter was replaced with the EF-1α promoter as described above and the central polypurine tract of HIV was inserted 5′ of the promoter. CAR-expressing lentiviral vectors in which the CAR sequences were preceded in frame by an eGFP sequence followed by the 2A ribosomal skipping sequence from FMDV were also generated. These vectors permit dual expression of GFP and the CARs from a single RNA transcript. All constructs were verified by sequencing.
Xenograft studies were performed as previously described [26, 27]. Briefly, 6–12 week old mice NOD-SCID-γc−/− or NOD-SCID-beta;2−/− mice were obtained from JAX (Bar Harbor, ME) or bred in-house under an approved IACUC protocol and maintained under pathogen-free conditions. Animals were injected with 5×105 to 2×106 viable human ALL cells via the tail vein in a volume of 0.3 ml. T cells were injected into animals 9–21 days following ALL injection as indicated via the tail vein. Animals were closely monitored for signs of graft-vs-host disease as evidenced by >10% weight loss, loss of fur and/or diarrhea. Peripheral blood was obtained by retro-orbital bleeding, and ALL and T cell engraftment were determined by flow cytometry using BD TruCount tubes as described in the manufacturer’s instructions. CD19, CD3, CD4 and/or CD8 expression was detected by staining with fluorescently-conjugated monoclonal antibodies. CAR+ T cells were identified by GFP expression using lentiviral vectors in which the CAR was linked to eGFP-2A as described above.
DNA was extracted from either cultured T cells or 1 mm3 of spleen tissue using a QiAmp or PureGene kit (Qiagen, Valencia, CA), respectively, according to the manufacturer. Quantitative real-time PCR was performed on 500 ng of tissue-derived DNA with 4 replicates for each DNA sample using the ABI 2X Taqman Universal Master Mix with AmpErase UNG (Applied Biosystems). The αCD19-ζ specific primers and probe were designed to amplify the junction region between the CD8 transmembrane region and the ζ signaling chain:
CD19 Zeta F primer: 5′-TCC TTC TCC TGT CAC TGG TTA TCA A-3′
CD19 Zeta R primer: 5′-G GTT CTG GCC CTG CTG GTA-3′
CD19 Zeta MGB probe: 5′-FAM CTT TAC TGC AGA GTG AAG T-3′
and the αCD19-BB-ζ specific primers and probe were designed to amplify the junction region between the 4-1BB and ζ signaling chains:
CD19 41BB F primer: 5′-TGC CGA TTT CCA GAA GAA GAA GAA G-3′
CD19 41BB R primer: 5′-GCG CTC CTG CTG AAC TTC-3′
CD19 41BB MGB probe: 5′-VIC ACT CTC AGT TCA CAT CCT C-3′
PCR with real time fluorescence detection was performed on a 384 well HT7900 real time PCR thermocycler (Applied Biosystem). The numbers of copies for each vector, expressed as copies/500 ng DNA, was determined by comparison of the measured cycle threshold (Ct) for each well to the Ct of a standard curve prepared by dilution of a receptor-encoding plasmid in 500 ng pooled genomic DNA (Bioline USA, Inc) per well. Water added to 500ng genomic DNA was used as the negative control, which was run in 2–3 wells per plate. Each sample was also evaluated in duplicate following the spiking of 20 copies of plasmid DNA for each CAR receptor into 500 ng of sample to evaluate for the presence of a PCR inhibitor. The assay was qualified to a limit of quantitation of 5 copies per 500 ng genomic DNA, which correlates to 5 copies in approximately 75, 000 cells; the precision of values below 5 copies per well was not evaluated. Variation between runs during assay qualification was minimal and operator independent, with a %CV among two operators and four runs ranging from 5.31 to 1.47, an R2 value of >0.995 and a slope ranging from −3.05 to −3.65. Known spike controls ranging from 5 × 103 to 1 × 106 copies per well were also included in validation runs and were typically within 90% of the expected value.
Statistical analyses were performed as indicated using STATA version 10 (StataCorp LP, College Station, TX). In analysis where multiple groups were compared, a one-way analysis of variance (ANOVA) was performed with a threshold F-test p value of 0.05 prior to performance of post-hoc analysis by the Scheffe F-test. Absolute peripheral blood T cell counts, ALL blast counts and αCD19-ζ:αCD19-BB-ζ copy number ratios were log-transformed prior to analysis. Survival curves were compared using the log-rank test.
We thank Wei-Ting Hwang for statistical analysis, Bruce Levine for helpful discussions, the Human Immunology Core for obtaining and purifying the primary T cells used in this study, Martin Carroll for providing the leukemia samples, and John Scholler, Treasa Smith, Ronghua Liu, Xiaochuan Shan, Junior Hall and Anthony Secreto for expert technical assistance. This work was supported by NIH grants 1R01CA105216, RO1AI057838, and R01113482, the Alliance for Cancer Gene Therapy, the Weinberg and Foerderer-Murray Funds at Children’s Hospital, and the Leukemia & Lymphoma Society. J.L.R. and C.H.J. have patent applications in some of the technology described in the manuscript.