Immune therapeutic targeting of glioma cancer stem cells

Introduction

Brain tumor immunotherapy holds unrealized therapeutic promise that must overcome the challenge of tumor-mediated immunosuppression, in which the cancer stem cell (CSC) may play a key role. The CSCs are a heterogeneous group of undifferentiated tumor cells defined by their capacity for self-renewal, multipotency, high tumorigenicity at low cell numbers, and capacity to recapitulate the phenotypic and histological characteristics of the original tumor [1]. CSCs have been identified in a wide variety of hematological and solid malignancies, including lung, breast, prostate, colon, and brain cancers [2-8], and are believed to play important roles in tumor initiation, proliferation, progression, invasion, metastasis, recurrence, and resistance to therapy. The identifying features of the CSCs isolated from human gliomas (gCSCs) [9, 10] include the formation of nonadherent neurospheres in vitro, the capacity for multilineage differentiation along astrocytic, oligodendroglial, and neuronal lines [11], and the ability to induce tumors in vivo that recapitulate the features of human malignant gliomas. The expression of stem cell markers such as CD133 [12] and nestin [13], are also commonly observed, but are not definitive determinants of “stem cells”.

The current standard of care for glioblastoma multiforme (GBM) consists of surgical resection, if feasible, temozolomide administration, and irradiation [14]--modalities that are effective against proliferating and differentiated tumor cells. The gCSCs, however, perhaps as a result of their undifferentiated and quiescent state, slow cell cycling, enhanced DNA repair capability, and expression of antiapoptosis genes [15], are highly resistant to these treatment modalities [16, 17]. This failure of current treatments to eliminate the gCSC population may explain the grim near-inevitability of recurrence and progression in patients with GBM, as the surviving gCSCs are capable of reconstituting the original tumor. Thus, it would seem that successful long-term treatment of GBM will require the elimination or suppression of gCSCs and that a strategy of specifically targeting gCSCs may hold promise as a paradigm for the development of future novel therapies for GBM.

Immunological approaches against GBM have been attempted in the past, but to date, none have been curative [18-30]. Although GBMs express tumor-associated and tumor-specific antigens [31] that are recognized by the host’s immune system [32, 33], they are capable of mediating a profound degree of immunosuppression, both systemically and within the tumor microenvironment [34-36], allowing tumor cells to escape host immune surveillance, and thwarting current attempts to use immunotherapies against GBM. Thus, to successfully employ immune therapy against gCSCs, two conditions must be met. First, the host immune system must be able to recognize and distinguish gCSCs in order to mount an effector response against them. Second, the immunosuppression mediated by the GBM and/or gCSCs must be circumvented. It is therefore necessary to define the antigens preferentially expressed within gCSCs relative to normal cells [37] and to determine the immunologic properties of gCSCs. This article will describe recent data concerning the immunological properties of gCSCs and explore potential strategies for immunotherapeutic approaches targeting gCSCs and reversing GBM-mediated immunosuppression.

The glioma cancer stem cell hypothesis

When first formulated, the glioma cancer stem cell hypothesis proposed a model in which the gCSC population within the GBM was responsible for tumor perpetuation and identifiable by the expression of stem cell markers such as CD133 [8, 38]. However, further investigation demonstrated that CD133 negative GBM cells can also give rise to tumors and can possess the hallmark features of gCSCs [39]. Instead, gCSCs are defined empirically by their capacity for self-renewal, pluripotency, tumorigenicity at low cell numbers, and recapitulation of the phenotypic features of primary GBM [40]. This clarifying distinction can serve to resolve disparate research findings between groups such as ours that have characterized these cells based on this more stringent criterion and found them to be immune suppressive [41, 42] and others that have relied only on the property of in vitro neurosphere formation and demonstrated immune eradication of the gCSC. At our present state of knowledge, neither neurosphere formation [43] nor CD133 expression, nor any other in vitro property, alone or in combination, should be viewed as necessary or sufficient for classification as a gCSC.

The stability of the gCSC phenotype also remains to date undetermined. The original gCSC hypothesis viewed the gCSC as a fixed phenotype of a specific population of cells whose eradication would entail the eventual destruction of the entire tumor. More recent evidence suggests that factors such as hypoxia can induce glioma cells to de-differentiate and acquire stem-like properties [44]. This raises the possibility that “stemness” is a dynamic property that many glioma cells may potentially adopt, depending on circumstance. If this is true, targeting gCSCs in isolation should not be considered a panacea for GBM, since even after successful eradication of gCSCs, other glioma cells may acquire gCSC properties and reconstitute a population of gCSCs.

The argument that gCSCs are resistant to immunological targeting

The argument that gCSCs can be targeted with immunotherapy

Reconciliation of disparate research findings

We have observed that passage number markedly affects the immunosuppressive properties of perpetuated gCSCs. Specifically, the secretion of immunosuppressive cytokines by gCSCs and the ability of gCSC-conditioned medium to induce M2 macrophages diminish markedly over time with multiple passages, suggesting a progressive loss of immunosuppressive capability as gCSCs are maintained in tissue-culture conditions (Fig. 3). Indeed, Pellegata et al. [86] reported that neurospheres derived from the GL261 cell line had increased expression of immune stimulatory cell surface markers such a CD80 and CD86, whereas we have found the expression of these two markers in gCSCs recently isolated from human tumor specimens to be low [42]. Because gCSCs are a heterogeneous cell population, and immunosuppressive properties provide no survival benefit in a tissue-culture environment, selective pressures may favor the loss of immunosuppressive phenotypes over time. An alternate hypothesis would be that factors specific to the in vivo environment of gCSCs that are not present in vitro are necessary for maintaining immunosuppressive phenotypes over the long term. In either case, the instability of the immunosuppressive phenotype in gCSCs may mean that animal model systems using established glioma cell lines do not accurately replicate the immune interactions pertinent to tumors arising de novo and that a vaccination strategy effective against the former may not be so with the latter. Furthermore, some of these vaccination strategies have been conducted in immunocompromised murine model systems, and this would artificially influence the complex interaction between the adaptive and innate immune system and potentially confound the data.

As recurrence and persistence are hallmark features of gliomas, it is clear that the intrinsic immune system of the patient, unaided, is usually unable to eradicate the gCSCs that are the progenitors giving rise to recurrence and progression. The gCSCs may be more than capable of immunosuppression sufficient to allow escape from the host’s natural immune surveillance, but this does not necessarily mean that an immune response properly primed by an external therapeutic manipulation will not be able to overcome this immunosuppression and still successfully target and kill the gCSCs. Several clinical trials of a peptide vaccine targeting the EGF receptor variant III (EGFRvIII) have suggested therapeutic efficacy in patients with malignant glioma [21, 24]. Because EGFRvIII expression may be enriched in CD133+ gCSCs [88], EGFRvIII-targeted therapies may help glioma patients to mount a more specific immune response against gCSCs.

Mechanisms for inhibiting the immunosuppressive properties of gCSCs

Ultimately, within malignancies there exists a balance between immune stimulation by tumor antigens and immune suppression mediated by tumor cells that promotes escape from immune surveillance. Tumorigenesis and recurrence occur in a setting in which the latter overrides the former. If tumor-mediated immunosuppression can be successfully inhibited, then the balance may tipped back in favor of immune stimulation, allowing the host’s immune system to more easily eradicate the tumor, either on its own or in conjunction with immunostimulatory therapies.

A simple way to reduce tumor-mediated immunosuppression is by a gross-total resection of tumor, which eliminates the bulk of the immunosuppressive tumor microenvironment in which immunosuppressive cytokines are most concentrated, as well as most of the immunosuppressive tumor-infiltrating cells, such as Tregs, and M2-polarized macrophages/microglia. Elimination of the tumor mass effect also allows for withdrawal of immunosuppressive adjuvant treatments such as steroids. An example of the benefit of surgery in the context of immunotherapy has been shown in GBM patients without a bulky and actively progressing disease who responded better to dendritic cell based vaccines [22]. It is for this reason that many recent clinical trials of immunotherapies in GBM patients are conducted on patients who first receive a complete surgical resection of their tumor.

Clearly, therapies that target and/or reverse gCSC immunosuppression are needed. The existence of multiple overlapping and redundant mechanisms of immunosuppression suggests that focusing on any single mode of immunosuppression alone may not be effective and probably accounts for many failures of targeted immunotherapy strategies in the past. More likely, an effective strategy will need to be able to reverse multiple modes of immunosuppression simultaneously. For this reason, the STAT3 pathway garners interest. As an important regulatory hub underlying a wide range of immunosuppressive pathways in both the adaptive and innate immune systems, inhibition of p-STAT3 holds the promise of being able to simultaneously target multiple modes of immunosuppression. Furthermore, because pSTAT-3 is also directly involved in tumorigenesis and proliferation, STAT3 inhibition can also target tumor growth directly. Inhibition of STAT3 has been shown to reverse glioma-associated immunosuppression and to inhibit tumor growth [89]. Specifically, WP1066, a small molecular inhibitor of STAT3, can be administered orally, has excellent central nervous system penetration, and can activate effector T cells, induce T-cell stimulating cytokines [83], and inhibit Tregs [90, 91]. Recently, we showed that STAT3 inhibition by WP1066 and siRNA in gCSCs can diminish gCSC induction of p-STAT3 in immune cells, reduce Treg induction, restore T-cell function, and induce T-cell proliferation [41]. The pSTAT3 blockade in gCSCs also reverses gCSC-mediated inhibition of macrophage phagocytosis, gCSC-mediated induction of IL-10 secretion in macrophages, and it diminishes the degree of gCSC-mediated M2 polarization by increasing the activity of the M1 phenotype regulator, STAT1 [30].

Furthermore, STAT3 inhibitors may exert direct anti-gCSC activity. Sherry et al. [84] showed that blocking STAT3 in gCSCs leads to inhibition of growth and neurosphere formation, but did not result in apoptosis, probably because gCSCs are relatively resistant to apoptosis. In addition, loss of expression of the CSC markers, nestin and olig2, was observed upon inhibition of STAT3. These investigators suggested that STAT3 inhibition suppresses the self-renewal properties of gCSCs in culture but does not fully execute the differentiation of these cells. This would indicate that STAT3 blockade also has the potential to reverse resistance to chemotherapy and radiation therapy mediated by the gCSC.

Future directions

Studies revealing the immune properties of gCSCs are limited. A greater understanding of the immune biology of gCSCs and their interactions with the host immune system will lead to the elucidation of novel targets and new treatment regimens to enhance the efficacy of the current immunotherapeutic approaches. Detailed studies are warranted to identify new tumor-associated antigens expressed by gCSCs to be targeted for immunological approaches as well as to clarify the dependence of gCSCs on specific signaling pathways in contrast to other stem cell populations. Vaccination strategy with dendritic cells loaded with gCSCs has been shown to be effective in animal tumor model systems with different cell lines, but it will require more investigations to better understand the mechanism of effect before being translated to human studies. Therapeutic approaches targeting the key factors allowing the tumor to evade the immune system have been studied in an attempt to reverse the immunosuppression and treat the glioma. Although the STAT3 inhibitors WP1066 [41], STA-21, and S31-201 have been shown to be effective in reserving immunosuppression mediated by gCSCs, other inhibitor molecules such as JSI-124 [92], NSC 74859 [93], and AZD1480 [94] can be tested for efficacy against gCSCs. In addition to the STAT3 pathway, other signaling pathways, including Wnt, hedgehog, Notch, the HOX family, Bmi-1, PTEN, telomerase, and efflux transporters, are involved in self-renewal and differentiation of gCSCs [95-97]; thus, these molecules may also be favorable targets for suppression of gCSCs by glioma immunotherapy. In addition, treatment strategies inducing the differentiation of CSCs can make them more vulnerable to these therapies [98]. Also, recent studies have demonstrated an association between human cytomegalovirus (HCMV) infection and malignant glioma [99-101] and have shown that HCMV nucleic acids and proteins expressed in malignant gliomas promote glioma cell invasiveness [102]. The role of HCMV in malignant glioma biology and its possible interaction with gCSCs should be further investigated. Targeting HCMV could be a new treatment strategy to eliminate cancer stem cells in glioma patients. Finally, since our knowledge of the immunologic properties of gCSCs is increasing, novel approaches can continue to be identified and improved when they are administered with other immunotherapeutic modalities such as chemotherapy or radiation therapy.

In our view, targeting gCSCs holds the most promise as one component of a multi-modal treatment strategy. The aim is not simply to just eradicate the gCSC population as if it were some mythic Achilles’ heel that would bring down the whole tumor with it, but also to suppress the tumor supportive functions, such as radiation resistance, chemoresistance, and immune suppression, provided by the gCSCs, and if necessary, to prevent or reverse the adoption of gCSC phenotypes by other glioma cells. This in turn would hopefully render the remaining tumor cells more vulnerable to other concurrent treatments, leading ultimately to the successful suppression of the malignancy.