Less is more: lymphodepletion followed by hematopoietic stem cell transplant augments adoptive T-cell-based anti-tumor immunotherapy

Abstract

Adoptive T-cell immunotherapy combined with non-myeloablative lymphodepletion has emerged as the most effective immunotherapy treatment for patients with metastatic melanoma (objective response rates of 50%). The mechanisms underlying this major advance in the field of immunotherapy include the elimination of regulatory elements and increased access to activating cytokines. This results in the activation of low-affinity T cells, enabling them to destroy tumors. We propose that a more complete depletion of the patient’s immune system followed by transplantation with hematopoietic stem cells, which can be genetically modified, would be more effective in the treatment of metastatic cancer.

Introduction

In the past decade, our understanding of antigen-specific tumor-cell recognition and destruction has grown greatly. Much of the recent work in the field of immunotherapy has concentrated on the development of active immunization using newly discovered antigens, but the results in animal and human trials have disappointed many of the most active advocates of cancer vaccines [1]. Prophylactic vaccines for infectious diseases are highly effective, but vaccines are not currently therapeutic in the treatment of established infectious disease, such as HIV, chronic hepatitis, tuberculosis or influenza. Similarly, therapeutic vaccines for the treatment of solid tumors of non-hematopoietic cell origin have not yet been effective, although the reasons for this remain in the realm of speculation. It is plausible that the current anti-tumor vaccines do not sufficiently activate tumor-reactive T cells to a state in which they are capable of tumor destruction. Regulatory elements and poor access to activating cytokines and co-stimulatory molecules might control the activities of anti-tumor T cells in a host with established cancer. Additionally, tumor antigens are poor ‘targets’ because they are often self-antigens with low-affinity epitopes. In many cases, these epitopes are undefined [2]. Thus, tumor vaccines might fail to generate appropriate numbers of highly activated T cells that are capable of mediating a therapeutic response in patients with established metastatic tumors.

The most promising current approach for the treatment of metastatic melanoma in immunotherapy is adoptive T-cell transfer (ACT), which is given to lymphodepleted patients (see also the review by W Overwijk, in this themed issue; [3]) [1,4,5]. In ACT, autologous tumor-infiltrating lymphocytes (TILs), are extracted, expanded ex vivo and re-administered into the patient where they specifically destroy antigen-expressing tumor cells. Lymphodepletion can create an environment in the patient where the T cells against low affinity epitopes — whether of defined or unknown specificity — can be activated sufficiently to destroy tumor cells [4]. Autologous tumor-reactive T cells transfused into a lymphodepleted patient experience great exposure to activating cytokines, they are sensitized to recognize low affinity antigens and are less susceptible to suppression by regulatory elements.

This review will discuss the reasons why partial lymphodepletion in combination with ACT can be effective, and why complete ablation with hematopoietic stem cell transplant might be even more beneficial in adoptive immunotherapy.

T-cell activation in a lymphopenic host

In animal models, homeostatic expansion results in the activation of T cells. Naïve T cells transferred into a lymphopenic host undergo homeostatic proliferation and thereby acquire a memory phenotype, as manifested by the expression of memory T-cell markers (high CD44, Ly6VC and CD122) [6–9]. This proliferation is accompanied by T-cell activation, as demonstrated by enhanced in ex vivo IFN-γ release and cytolysis [6,8,9]. In vivo, T-cell activation in the lymphodepleted setting is revealed by the inhibition of tumor growth when polyclonal autologous T cells are transfused into sub-lethally irradiated mice [10]. King et al. [11] found that lymphopenia triggered autoimmunity through homeostatic proliferation of T cells in nonobese diabetic mice (NOD) and proposed that IL-21 was a homeostatic cytokine of importance in the development of activated T cells in a lymphopenic host [11]. In the lymphocytic choriomeningitis virus (LCMV) model, a significant reduction of the viral titers was achieved by transferring LCMV TCR-specific CD8⁺ cells into a lymphodepleted host. [12] These examples demonstrate the therapeutic potential of T cells undergoing homeostatic proliferation and concomitant activation, even in the absence of antigen-specific vaccination.

Partial ablation enhances success in immunotherapy

Recent work in our laboratory has revealed that partial lymphodepletion can significantly augment the anti-tumor efficiency of transferred T cells when combined with a tumor-antigen-specific vaccination. Mice with large B16 tumor burden were treated with transgenic T cells (called pmel-1) specific for the self and tumor antigen gp100 given in combination with IL-2 and virus expressing a modified form of gp100. This tripartite treatment resulted in significant tumor reduction [13]. Non-myeloablative lymphodepletion greatly enhanced the tumor treatment effect (L Gattinoni et al., unpublished).

In melanoma patients the most dramatic treatment response was achieved in an ACT therapy protocol combined with partial ablation via chemotherapy. Autologous T cells were expanded ex vivo and transfused into a lymphodepleted patient. The objective clinical response rate exceeded 50% in partially ablated patients, using strict World Health Organization (WHO) criteria in a group of heavily pretreated (surgery, chemotherapy, radiotherapy and other biological therapies) patients with stage IV melanoma. An additional 30% of patients experienced partial or mixed responses [4,5]. Thus, homeostatic proliferation expands and activates autologous polyclonal anti-tumor T cells regardless of their epitope specificity and differentiation stage.

The importance of T-cell receptors in homeostatic proliferation

The signal strength needed to trigger proliferation and activation of T cells in a lymphodepleted host is considerably reduced — even low-affinity antigens such as self-antigens become sufficient triggering signals [6,8,9,14–18]. Although controversy remains, some authors have even claimed that homeostatic proliferation might occur in the absence of MHC presentation [19,20]. Most anti-tumor T cells target self-antigens for which they have low affinities. The lymphopenic environment facilitates the activation and proliferation of these poorly reactive T cells. Nevertheless, there is evidence that homeostatic proliferation of T cells is regulated by clonal competition and is dependent on the avidity of the self-peptide–MHC complex [21–23]. Clones that can expand most rapidly dominate the replete T-cell repertoire, whereas the other clones die out [24]. Consistent with these findings, it has also been reported that naïve monoclonal T cells expand in number when transferred into a TCR transgenic host of differing clonotype, but do not expand in a host of identical clonotype [25,26].

This is of special interest for adoptive T-cell immunotherapists because the long-term survival of tumor-reactive clones is desired to protect from tumor recurrence. Whether there exists a special phenotype or certain tumor-specific epitopes that might help for long-term survival remains an interesting field of research. Adoptive transfer after lymphodepletion in patients with objective responses leads to the persistent skewing of the T-cell repertoire towards tumor-reactive T cells [27•]. An adult patient has strongly reduced thymic output and therefore the repleting T cells are mostly from transferred T cells or the few remaining, non-depleted endogenous T cells. This steers the patient to tumor immunity through tumor-reactive T cells and to autoimmunity against self or tumor antigens [4].

The role of co-stimulatory molecules and cytokines in homeostatic expansion

At least two signals — TCR stimulation and a co-stimulatory signal — are required for proliferation and activation of T cells. Homeostatic T-cell proliferation differs from antigen-driven proliferation with respect to its requirements for co-stimulatory molecules: neither the CD28–B7 interaction nor the CD40–CD40L interaction [16,28] are required for homeostatic expansion. Although 4-1BB is involved in the allogeneic T-cell response, it is not required for homeostatic proliferation [29]. The only co-stimulatory signal that was recently shown to be necessary for optimal homeostatic proliferation of CD4⁺ and CD8⁺ cells was through CD24 [30]. The findings described above suggest that signals through co-stimulatory molecules are less critical for T-cell activation in a lymphodepleted setting.

By contrast, key cytokines (especially IL-7 and IL-15) are crucial for the homeostatic proliferation of naïve and memory T cells. IL-7 is required for the homeostatic expansion of naïve and mature T cells, where IL-7 initiates T-cell proliferation [31–35], as well as for the survival of naïve T cells in a replete host [31,33]. Additionally, exogenously administered IL-7 can promote antigen-independent proliferation of T cells in a lymphoreplete host [36,37•,38]. Conversely, IL-7 receptor blocking leads to a decrease in T-cell numbers [39,40]. Thus, low levels of IL-7 mediate T-cell survival, whereas higher levels of IL-7 induce proliferation. Higher levels of IL-7 can be achieved by exogenous administration of IL-7 or, in lymphodepleted settings, by increased cytokine availability through decreased competition for endogenously produced IL-7.

Similar to IL-7, IL-15 is a cytokine that signals through the common γ-chain (γC, also called CD132). Although IL-15 is not required for in vivo expansion of naïve T cells, memory CD8⁺ T cells clearly benefit from IL-15 signals [33,34,41,42]. In our own model of adoptively transferred T cells in the treatment of large, established tumors (pmel-1), we found that the absence of IL-15 in the lymphodepleted state significantly compromised tumor treatment [43• •]. Another common γC cytokine, IL-21, is a candidate homeostatic factor, and IL-21 receptor knockout mice have impaired CD8⁺ expansion and cytotoxicity [44•]. These observations suggest that ACT therapy benefits not only from the homeostatic cytokine accumulation that occurs in a lymphopenic host but also from exogenous administration of cytokines such as IL-7, IL-15 and IL-21. These findings are indeed what recent experiments in our laboratory have revealed — each of these cytokines given alone can augment ACT therapy in the pmel-1 mouse system, even more so when given in combination [44•].

The impact of lymphodepletion on CD4⁺CD25⁺ T regulatory cells

As described above, low (basal) levels of homeostatic cytokines in a lymphoreplete host prevent the activation and proliferation of self- or tumor-reactive T cells. Another mechanism that also protects the host from self-or tumor-reactive T cells is the activity of CD4⁺CD25⁺ T regulatory cells (Tregs) [45–47]. Depletion of Tregs augments tumor and autoimmunity, whereas the adoptive transfer of Tregs suppress anti-tumor T-cell responses as well as autoimmunity in a variety of models [48]. Treg cells are also likely to be important in the anti-tumor immune response in humans [49••,50••,51–53].

It seems, therefore, that the enhanced recognition of low-affinity antigens by T cells in lymphopenic hosts is triggered by higher levels of cytokines and by reduced inhibitory elements leading to the proliferation and activation of self- or tumor-specific T cells.

Complete ablation, ACT and HSC transfer: a way to fight tumors expressing unknown antigens?

Low-affinity tumor antigens are poorly recognized in a lymphoreplete host that possesses strong regulatory elements and low basal levels of homeostatic cytokines. Under these ost conditions, adoptively transferred tumor-reactive T cells can only have anti-tumor activity under the influence of a strong ‘danger’ stimulus in vivo, such as a modified epitope ligand expressed by a recombinant viral vector [2,13]. In a partially ablated host (5 Grey, total body irradiation, non-myeloablative), regulatory elements are greatly reduced and cytokine levels are greatly increased. These conditions help favor T-cell reactivity to low-affinity self- or tumor antigens, but a subset of host cells survives partial ablation, and functions as regulatory elements and ‘sinks’ for activating homeostatic cytokines (CA Klebanoff et al., unpublished; see Update). In our own pmel-1 model, vaccination is still required under the condition of partial ablation for the treatment of large established tumors (L Gattinoni et al., unpublished).

Complete ablation (9 Grey, total body irradiation, myeloablative) further eradicates elements that suppress tumor-reactive transferred T cells and is likely to further reduce host cells that consume homeostatic cytokines (Figure 1). Complete ablation also harms hematopoietic stem cells (HSCs) in the bone marrow, thus requiring HSC reconstitution [54]. These cells proliferate strongly and differentiate into the cellular elements that reconstitute the host. HSCs might also have an impact on the remaining host cell population as well as on co-administered adoptively transferred T cells.

Figure 1.

In a lymphoreplete host, low basal cytokine levels and strong inhibitory elements are present. Therefore, strong antigen stimulation combined with a co-stimulatory signal is needed to activate T cells. Partial ablation reduces regulatory elements as well as other cytokine consuming host cells and leads to elevated cytokine levels. Low affinity antigens are sufficient to induce T-cell activation even with weaker co-stimulatory signals. In complete ablation with hematopoietic stem cell (HSC) transplants, host cells are further diminished and additional factors are released or induced by, for example, HSCs, cytokine levels are presumably even more elevated. Therefore, the need for co-stimulatory signals and T-cell receptor stimulation might be further reduced.

Collectively, complete ablation together with HSC transfer reduces regulatory elements and elevates cytokine levels to such an extent that transferred T cells gain the ability to recognize self- or tumor-reactive antigens (Figure 2), and the need for an additional antigen-over-expressing vaccine is virtually eliminated (C Wrzesinski et al., unpublished). As tumor-reactive T cells are mostly polyclonal and often of unknown epitope specificity, the elimination in vivo vaccination would be of profound importance to the clinic [1].

Figure 2.

Hematopoietic stems cells (HSCs) transferred into a completely ablated host have a high proliferative capacity. The factors required for this proliferation are provided by the ablated host or by the HSC themselves. These factors not only influence the HSCs but also impact on both the remaining host cells and adoptively transferred T cells, leading to their proliferation or survival. For tumor treatment, transferred tumor antigen-specific T cells need to be provided not only in sufficient numbers, but they must also be activated. In this lymphopenic environment, the mechanisms shown in Figure 1 could activate T cells. Alternatively, T cells transferred after in vitro activation maintain their state of activation in the lymphopenic environment.

Autologous transplant with gene-modified hematopoietic stem cells

Post-transplantation, HSCs proliferate and differentiate into a variety of blood cell types, including T cells, B cells and dendritic cells, involved in the long-term immune response. Therefore, the modification of transferred HSCs could be exploited to maximize the therapeutic effect of any HSC-derived cell. By choosing an appropriate promoter, the modification could be limited to only a certain cell type (Figure 3).

Figure 3.

Hematopoietic stem cell (HSC) transplants in completely ablated patients give rise to a variety of different immunologically important blood cells. Several modifications of HSCs that are of therapeutic interest are plausible. By choosing an adequate promoter, transgene expression can be limited to a certain cell type. For example, the tumor-antigen-specific T or B cell receptors could be expressed in T or B cells, the genes encoding tumor antigens could be inserted into dendritic cells (DCs) or regulatory T cells could be eliminated through suicide genes expressed under the control of the T regulatory cell-specific Foxp3 promoter.

Dendritic cells derived from HSCs can be modified to express tumor-associated antigens in a stable and long-lasting manner. In combination with a systemic dendritic cell activating agent (anti-CD40) and mature T cells, antigen-specific T-cell expansion and activation was achieved and resulted in tumor treatment [55].

The induction of tumor-specific T cells through insertion of the T-cell receptor into HSCs is also promising. T cells with tumor specificity are rare and must be greatly expanded to achieve sufficient numbers for successful ACT therapy of metastatic tumors. This expansion process results in terminally differentiated T cells that have lost key phenotypic and functional characteristics, resulting in reduced anti-tumor activity (L Gattinoni et al., unpublished). Furthermore, the persistence of tumor-antigen-specific T cells correlates with their anti-tumor efficacy after adoptive transfer [27•]. The transduction of tumor-antigen-specific T-cell receptors into HSCs might lead to a high precursor frequency of tumor-reactive T cells as well as a lifelong persistence of these T cells, thus solving two critical problems in current ACT-based immunotherapies. Similar procedures with B-cell receptors could result in the production of monoclonal antibodies with desired specificity, such as a complement-fixing antibody with specificity for a tumor-associated antigen. T or B lymphocytes could also be modified to produce cytokines, co-stimulatory molecules or factors that block inhibition, controlled by suicide genes, to result in the constitutive activation of immunity.

Treg cells can diminish the effectiveness of the anti-tumor immune response, especially when that response is directed against self-antigens [56,57] (PA Antony et al., unpublished; see Update). Foxp3 is a trans-activating gene product that is exclusively expressed by Treg cells [58]. A suicide gene under the control of the Foxp3 promoter could be introduced into HSCs, which could then be transferred into a completely ablated host. Activation of this suicide gene would result in the complete depletion of Tregs in the patient directly before immunotherapy.

Finally, the endothelial cells that comprise the neo-vasculature of tumors are substantially derived from HSCs. Suicide genes have been directly delivered into the tumor neo-vasculature using lentiviral modified HSCs. This reportedly resulted in tumor killing followed by slower tumor growth [55].

Conclusions

Lymphodepletion before the adoptive transfer of anti-tumor T cells is the largest single advance in tumor immunotherapy in a decade. Chemotherapy in humans and total body irradiation in mice has been used primarily to induce non-myeloablative lymphodepletion. In mice, we have found that total ablation enhances ACT therapy significantly more than partial ablation. In humans, partial ablation is associated with an objective response rate of approximately 50%. A more complete lymphodepletion with autologous HSC transplant might further augment current ACT-based therapies resulting in a higher response rate in patients with metastatic cancer.

Update

A recently published paper shows for the first time that CD4⁺ CD25⁻ T cells (T helper cells) co-transferred with tumor-reactive CD8⁺ T cells help to break the tolerance to tumor self antigens in an IL-2-dependent mechanism, but only in the absence of CD4⁺ CD25⁺ T cells (T regulatory cells) [59• •]. Therefore, for an optimal tumor treatment with adoptive T-cell transfer the absence of inhibitory effects of T regulatory cells is required.