In vitro methods for generating CD8⁺ T-cell clones for immunotherapy from the naïve repertoire

Abstract

Innovations in gene discovery and the analysis of gene expression are facilitating the identification of a growing number of antigens that could potentially be targeted for immunotherapy of tumors. Methods to reliably generate antigen-specific T-cell responses in vitro would be useful not only to screen candidate antigens for immunogenicity prior to embarking on in vivo vaccination trials, but also to generate T-cell lines or clones that could be used directly for adoptive immunotherapy approaches. Although many techniques have proven successful for expanding ex vivo effector cells from antigen-specific memory CD8⁺ cells that have been primed in vivo, methods to reliably generate high-avidity CTL clones from the naïve repertoire have not been well described. Various methods for the induction and expansion of antigen-specific CD8⁺ CTL clones from healthy A2⁺ donors were compared, using WT1 as a model tumor-associated antigen for which there is a low frequency of precursor T cells in naïve individuals. In contrast to the well-studied Melan-A/MART-1 (Melan-A) A2-restricted response, for which the CD8⁺ T-cell precursor frequency in the naïve repertoire is unusually high, successful expansion of WT1-specific CD8⁺ T cells appeared to be more dependent upon cell culture conditions. In particular, primary stimulation with autologous peptide-loaded monocyte-derived DC generated in 48 h (DC2d) was more effective in expanding WT1-reactive populations of CTL than stimulation with DC generated using the more standard week-long protocol (DC7d). Adding supplemental IL-7 2 to 3 days after initiation of a stimulation cycle expanded antigen-specific cells within CTL lines more efficiently than including the cytokine from the beginning of the cycle. Following primary stimulation with peptide-loaded mature DC, subsequent restimulation with peptide-loaded PBMC as the stimulators was more effective at expanding antigen-specific cells than repeated stimulation with mature DC. Using these techniques, high-avidity CTL clones specific for an A^⁎0201-restricted epitope of WT1 have been generated from nearly all normal A2⁺ donors tested. Such clones have been demonstrated to be capable of recognizing and lysing leukemic cells, and will soon be tested for therapeutic activity in clinical trials of adoptive immunotherapy in patients with relapsed leukemia after transplantation.

Introduction

Enhancing or supplementing the body's own immune responses against cancer is a goal of the field of immunotherapy. Although non-specific approaches such as the administration of cytokines can generally augment immune responses, the major current focus for manipulating the cellular immune response to tumors is to target specific antigens associated with malignant cells with vaccines or the transfer of antigen-specific T cells. The tools used for identifying candidate tumor-specific or tumor-associated antigens have been rapidly evolving. One of the first described approaches to identify antigens that can induce T-cell responses has used tumor-reactive T cells isolated from cancer patients as the reagent for screening target cells transfected with a tumor-derived expression library (Boon et al., 1994, Rosenberg, 1999). Currently, analyses of differential gene expression, for example using cDNA microarrays or serial analysis techniques, are the most widely used methods of identifying candidate tumor antigens (Lal et al., 1999, Polyak and Riggins, 2001). With these methods, selection is based on unique or overexpression of the antigen in malignant versus normal cells; however, there is no direct evidence of immunogenicity.

Developing effective methods for evaluating the identified candidate antigens to determine those with the most clinical promise is becoming more important. One strategy is to immunize cancer patients with the antigen and assess immunologic and clinical responses. A potential advantage of this vaccine approach is that the patient's own immune system does the screening for immunogenicity and the work of generating the appropriate immune effector cells; however, there are many candidate antigens to enter into clinical trials with limited basis for prioritization, reproducibly effective vaccine vectors that can induce strong responses have not yet been developed, and, if clear anti-tumor responses are not achieved following vaccination to the selected antigen, defining the qualitative or quantitative reasons for failure and how to proceed can be difficult. Thus, using an in vitro system to screen candidate antigens for immunogenicity before proceeding to in vivo studies has significant advantages.

An alternative approach to vaccination, the adoptive transfer of in vitro generated and expanded T-cell clones, overcomes the obstacle to generating a strong response that can hinder vaccination, as the number and type of cells transferred can be adjusted so that very high in vivo frequencies can be achieved. Thus, adoptive transfer of T cells can potentially be used to determine if an antigenic target should be further pursued and to set goals for vaccine strategies. However, general application of this approach requires the development of efficient in vitro methods for generating and expanding T-cell lines and clones specific for the target antigens of interest.

In vitro methods employing various forms of antigen and stimulator cells as antigen-presenting cells (APC) have been shown to be effective at expanding ex vivo memory T cells (e.g. viral-specific) that have been primed in the host by previous in vivo exposure to the antigen. However, generating and expanding primary immune responses from naïve precursors in vitro has generally been much more difficult, as naïve T cells have more stringent activation and costimulatory requirements. Additionally, since most of the target antigens in tumors are derived from normal self-proteins, the majority of potentially reactive cells will have been deleted during development or maturation in the periphery; thus, with the remaining reactive naïve precursor T cells being rare and predominantly of lower avidity, it is more difficult to isolate T-cell clones of the desired specificity and avidity.

Dendritic cells (DC) are the critical APC for inducing primary T-cell responses, but they represent a diverse and heterogeneous population of cells. Because of their ease of generation, most clinical studies have used DC generated by culture of monocytes in GM-CSF and IL-4; however, multiple protocols exist using varying types of media, serum and maturation signals/cytokines to generate such DC (Jonuleit et al., 1997, Reddy et al., 1997, Bennett et al., 1998, Ridge et al., 1998, Schoenberger et al., 1998, Cella et al., 1999, Lee et al., 2002, Dauer et al., 2003a). Although most “traditional” protocols for generating monocyte-derived DC include a week of culture, recent studies have described the use of mature DC generated from monocytes in only 48 h (“FastDC”) (Dauer et al., 2003a, Obermaier et al., 2003, Xu et al., 2003). Such DC express typical maturation markers, but morphologically appear distinct from the more classical DC, and may have distinct APC functions. The advantage of using DC generated with such a protocol would be significant savings in time and expense (related to cytokine cost) for DC production, as well as potentially distinct stimulatory capacities.

The nature and timing of cytokine exposure will likely impact the outcome of T-cell priming. IL-7 (a cytokine important for naïve T-cell survival, T-cell homeostasis and memory cell expansion) and IL-2 (the canonical T-cell growth factor) are commonly used as supplements for the in vitro generation of T-cell lines (Lynch and Miller, 1994, Walter et al., 1995, Schluns et al., 2000). However, the optimal timing for addition of these cytokines has not been defined. In theory, delaying the addition of such cytokines to later in the stimulation cycle might decrease the “non-specific” expansion of bystander CD8⁺ cells reactive with irrelevant antigens that have been transiently activated, and this could have an impact on expansion of the desired population.

Finally, while the efficiency of antigen presentation and level of costimulation provided by mature DC are likely critical for efficient priming of naïve T cells, it is less clear if repeated stimulations with DC are either required or beneficial for the efficient expansion of previously primed CD8⁺ cells. In fact, repeated stimulations with mature DC have been shown to potentially lead to not only deletion of high affinity T cells, but also the broader disappearance of virtually all antigen-specific CD8⁺ cells, possibly by activation-induced cell death (AICD) (Alexander-Miller et al., 1996, Mehrotra et al., 2003). Whether alternative types of stimulator cells might better expand previously primed antigen-specific CD8⁺ cells requires further investigation.

To address these issues, we examined a variety of culture conditions and cellular compositions to identify a reliable strategy for in vitro priming and subsequent expansion of antigen-specific CD8⁺ T cells, including the type of APC used for priming and restimulation, and the timing of addition and nature of supplemental cytokines. Normal healthy donors were used as the source of responder cells for this analysis of primary in vitro responses to minimize the potential that naïve precursors had been exposed to the tumor antigen in vivo. WT1 was chosen as the model tumor-associated antigen to examine, since this recently identified antigen currently is being evaluated in several vaccine trials as a therapeutic target. WT1 is a transcription factor that is normally expressed in several embryonic tissues and detected at low levels in adult cells including renal mesangial cells and hematopoietic stem cells (Little et al., 1999), but is overexpressed by many solid tumors and most leukemias (including AML, ALL and CML) with higher levels of expression correlating with worse prognosis (Bergmann et al., 1997). For comparison, responses to the heteroclitic peptide analogue of an HLA-A^⁎0201-restricted epitope (aa26–35) of the Melan-A melanoma antigen, to which primary in vitro responses have been well described, were also studied. This was done, in part, because there is an unusually high frequency of naïve CD8⁺ T-cell precursors due to positive selection in the thymus (Pittet et al., 1999, Oelke et al., 2000), which greatly simplifies the generation of detectable primary responses.