Glioma Stem Cell Research for the development of Immunotherapy

Abstract

Glioma, especially high-grade glioblastoma multiforme (GBM), is the most common and aggressive type of brain tumor, accounting for about half of all the primary brain tumors. Despite continued advances in surgery, chemotherapy and radiotherapy, the clinical outcomes remain dismal. The two-year survival rate of GBM is <30%. Better understanding of GBM biology is desirable to develop novel therapies. Recent studies have demonstrated the existence of a small subpopulation of cells with stem like features cancer stem cells otherwise known as (CSC). These GBM CSCs are self-renewable and highly tumorigenic. They are not only chemo-radio- resistant, but also often multi-drug resistance genes and drug transporter genes. These characteristic enable GBM CSCs to survive standard cytotoxic therapies. Among GBM CSCs, CD133+ cells are a well-defined population and are prospectively isolated by their cell-surface marker. There are increasing data that CD133+ CSC presence highly correlates with patient survival. This makes it an ideal immunotherapy target population. In this article, we will review recent studies related with GBM CSCs, particularly CD133+ CSCs as well as the novel therapeutic strategies targeting these cells.

Introduction

Human brain tumors are a diverse group of diseases characterized by the abnormal growth of brain cells contained within the skull afflicting both adult and children. According to National Cancer Institute data, there are about 20,000 new cases and 13,000 deaths each year in the US. In children, brain tumors are the leading cause of solid tumor cancer death; all forms of glioma make up about 1/5 of all childhood cancers (www.cancer.gov). In adults, the most common malignant brain tumor, glioblastoma multiforme (GBM), also the most malignant primary tumor of the brain and associated with one of the worst 5-year survival rates among all human cancers [1, 2]. The median survival time is 14.6 months after first diagnosis [3, 4]. Despite the advances in conventional treatments, comprised of surgical resection, local radiotherapy and systemic chemotherapy, the incidence and mortality rates for gliomas have changed little in the past decade. With greater understanding of the cellular and molecular mechanisms cancer initiation and propagation, the cancer stem cell (CSC) hypothesis presents new insights for developing novel treatments that target this group of cells. In this chapter, we will discuss the CSC hypothesis and its application to develop treatments for glioma.

CSC BTSC and CD133 CSCs

The first conclusive evidence for CSCs came from studies of acute myeloid leukemia (AML) [5, 6]. Bonnet and Dick isolated a subpopulation of AML cells that were capable of iniating AML in immunodeficident NOD/SCID mice. These leukemia cells (leukemia stem cells, LSCs) express similar cell surface markers to normal hematopoietic stem cells (HSCs). The AML established from these LSCs recapitulates the morphologic and immunophenotypic heterogeneity of the original tumor. These seminal studies opened the door for the field of CSC study. Besides the properties shared with normal stem cells (self-renewal and the ability to differentiate into other cells), to be considered as CSCs, candidate cells must present the following properties: (a) the unique ability to engraft; (b) the ability to recapitulate the tumor of origin both morphologically and immunophenotypically in xenografts and (c) the ability to be serially transplanted [7]. These criteria are the standard to identify other CSCs not only in hematopoietic tumors, but also in solid tumors.

The first solid tumor CSCs were identified from breast cancer by isolating CD44⁺/CD24^−/low cells from primary tumor cells [8]. The isolated CSCs can recapitulate the original breast cancer with same morphologic and immunophenotypic features CSCs could be isolated from these grafts and serially transplanted. For gliomas, several groups isolated brain tumor stem cells (BTSCs) from primary tumors based on the above criteria and the ability to form the neuro-spheres as normal neural stem cells do [9–16]. In our study, as few as 100 of these BTSCs could recapitulate the heterogeneity of GBM in immunocompromised rodents [15]. In addition to primary gliomas, we also isolated cancer stem-like cells from the commercial rat gliosarcoma cell line, 9L [17]. This cell line has been cultured in the lab over a long period under neurosphere conditions used for neural stem cell expansion. Similar results were reported by Kondo et al.. For rat the GBM cell line C6 [18]. These data indicate that glioma cell lines may retain the capacity for a stem-like phenotype even after years of in vitro culture. CSCs have also been identified in variety of other malignant primary tumors as well as cancer cell lines by using different cell surface markers (summarized in table 1).

Table 1.

identified cancer stem cells from different primary tumors and tumor cell lines

Tumor	type	isolation markers	Reference
AML	primary tumors	CD34+CD38−	[5, 6, 77]
Breast	primary tumors	CD44+CD24−/LOW	[8]
Brain	primary tumors	CD133+	[9, 10, 13, 27, 32, 78, 79]
	Cell lines	CD133+\sephere formation	[17, 28, 80]
	Cell lines	side population (SP)	[18]
Colon	primary tumors	CD133+	[60, 81, 82]
	Primary tumors	CD133+cd44+	[83]
	Cell lines	CD133+	[84]
Laryngeal	cell lines	CD133+	[85]
Leukemia	primary tumors	CD34+CD10−	[86]
Liver	primary tumors/
	cell line/blood	CD90+CD44+	[87]
	Cell lines	CD133+	[88–91]
Lung	primary tumors	ALDH1	[92]
	primary tumors	CD133+	[93]
Melanoma	primary tumors	ABCB5+	[57]
	primary tumors	CD133+ABCG2+	[94]
Ovarian	primary tumors	CD133+	[95]
Pancreas	primary tumors	CD133+	[96, 97]
	Cell lines	CD133+	[98]
Prostate	primary tumors	CD133+	[99]

Among the CSCs associated markers, CD133 (prominin-1) is the one of the most important and studied. It is a 120kDa five transmembrane domain glycoprotein (5-TM) with two cytoplasmic loops, two glycosylated extracellular domains and a cytoplasmic C terminal domain [19–22]. Despite mounting evidence show, that CD133 is an important marker for both somatic stem cell and CSCs, its physiologic function is not known. Some studies suggested that CD133 is involved in neural-retinal development and phototransduction [23, 24]. Due to its interaction with plasma membrane cholesterol and enrichment in cholesterol-based membrane microdomains, it may play some role in membrane toplogy [25]. A published study also demonstrated that CD133+ progenitor cells could promote the healing of diabetic ischemic ulcer through stimulating angiogenesis and activating the Wnt pathway [26]. This observation may suggest a role for CD133+ CSCs in tumor angiogenesis and in related signaling pathways.

Glioma CSCs and clinical treatment

CSCs are often resistant to conventional chemotherapy and radiation therapy. Glioma CSCs are resistant to radiotherapy and chemotherapy. CD133+ glioma CSCs could preferentially activate the DNA damage checkpoint response under irradiation. The activation is Chk1 and Chk2 checkpoint kinase dependent [27]. Maki and colleague also confirmed that CD133+ glioma cells are more radiation resistant than CD133- cells [28]. This study also reported that CD133 expression is up-regulated 1.6 fold under 2% O₂ hypoxic conditions. Similar results had been reported by other groups as well [29, 30]. Because hypoxia conditions exist in most solid tumors including gliomas, this up-regulation of CD133 expression provides enhancement for specific targeting of glioma CSCs rather than NSCs.

In vitro study showed that CD133+ glioblastoma CSCs are more resistant to multiple chemotherapeutic agents treatment than CD133- counterparts [31]. Our group demonstrated that CD133+ glioma CSCs express higher levels of drug transporter gene BCRP combined with up-regulation of the DNA repair protein MGMT mRNA, as well as higher mRNA levels of other genes that inhibit apoptosis, including FLIP, Bcl-2, Bcl-X and some IAP family genes. These cells were significantly resistant to chemotherapeutic agents compared to autologous CD133- cells [32].

Glioma CSCs possess an additional property to escape from conventional therapies -- migration. Our group and others reported that over expression of chemokine receptors, such as CXCR4, is a common mechanism related to CSCs migration [32–34]. As reviewed by Lefranc and colleagues, glioma cell migration is a complex combination of multiple molecular processes, including the alteration of tumor cell adhesion to a modified extracellular matrix, the secretion of proteases by the cells, and modifications to the actin cytoskeleton. Intracellular signaling pathways involved in the acquisition of resistance to apoptosis by migrating glioma cells include PI3K, Akt, mTOR, NF-kappaB, and autophagy (programmed cell death type II)[35].

Targeting signaling pathway in CSCs

Signaling pathways including Wnt, hedgehog, notch, Hox family member, Bmi-1, PTEN, telomerase, efflux transporters are involved in balancing self-renewal and differentiation of NSCs as well as CSCs [36–39]. Recent studies also show that Notch, Hedgehog and BMP pathways are involved in controlling CD133+ CSCs functions in glioma [40–42]. Bao and colleagues recently showed that glioma CSCs generate vascular tumors through over expression of vascular endothelial growth factor (VEGF) [43]. Since VEGF is a validated therapeutic target for glioma therapy [44–46], this finding may indicate more favorable targeting of CSCs in glioma therapy.

Due to the common pathways and cell surface markers shared by NSCs and CSCs, it is important to develop CSCs specific therapies that avoid potential toxicities to NSCs. Selective targeting of AML CSCs performed by Jordan's group demonstrated the possibility of such selectivity. They showed that LSCs, but not normal hematopoietic stem cells (HSCs), were susceptible to the apoptotic effects of the proteasome inhibitor MG-132 combined with the anthracycline idarubicin through NF-κB activity [47]. NF-κB inhibitors could induce LSCs apoptosis but spare normal HSCs [48]. In a subsequent study, same group also showed that 4-benzyl, 2-methyl, 1, 2, 4-thiadiazolidine, 3, 5 Dione (TDZD-8) treatment could induce oxidative stress and selectively kill LSCs in vitro but not HSCs [49]. Other studies demonstrated that AML is phophatase and tensin homologue (PTEN) pathway dependent. Rapamycin, a PI3K/PTEN signaling pathway inhibitor, could dramatically decrease leukemia burden [50]. In addition, more importantly, this treatment appeared to be specific for the LSCs since normal HSCs were unaffected.

When selective targeting of CSCs becomes possible, another strategy to target CSCs is forcing them to differentiate and become more sensitive to conventional chemo-radiotherapies. Differentiation therapy is based on this concept and a number of agents had been tested in recent years [51, 52]. All-trans-retinoic acid (ATRA) is the most studied differentiation therapy molecule. Sell et al. reported that about 90% of newly diagnosed patients with acute promyelocytic leukemia (APL) achieve complete remission and over 70% are cured by ATRA therapy [53]. Differentiation with ARTA had been also reported in early-stage mouse embryonic stem cells [54], rat C6 glioma cells [55], human embryonic NSCs [56]. These studies raised the possibility of using ARTA to induce differentiation of glioma CSCs as a therapy. Besides ATRA, other agents have also been tested for this approach of differentiation therapy. Piccirillo et al.. Have shown that treating CSCs with differentiation factors can effectively deplete CSCs in human glioma [42]. In this study, researchers reported that bone morphogenic proteins (BMPs), especially BMP4, activate their receptors (BMPRs) and trigger the Smad signaling cascade in cells isolated from human glioblastomas. This activated signaling pathway lead to a reduction in proliferation and increased expression of differentiated neural markers in both CD133+ CSCs and normal glioma cells. When xenotransplanted BMP4 pretreated glioma CSCs were transplanted into mice, there was no invasive glioma detected. These data provided evidence that differentiation therapy is a promising noncytotoxic strategy to deplete CSCs.

Targeting CSCs using passive immunotherapy

Antibody therapy (passive immunotherapy) directed against CSCs have resulted in several experimental therapeutic success. Schatton et al.. Identified melanoma CSCs with chemoresistance mediator ABCB5+ expression [57]. Treatment with anti-ABCB5 antibody for xenografted melanomas resulted in significant reduction of tumor size. Moreover, this direct targeting of CSC antigen induced tumor cell death through antibody-dependent cell-mediated cyotxicity. Another encouraging result of antibody therapy reported by Dick's group [58]. In their study, CD44 had been identified as an AML CSC surface marker. Although the same marker is also expressed on normal bone marrow HSCs at a lower level, treatment with anti-CD44 antibody before transplant can selectively block engraftment of AML LSCs but not normal HSCs. Treatment of previously engrafted AML with the same antibody led to a significant reduction in disease burden by 83–100%. In vivo treated AML CSCs resulted in lower engraftment, suggesting that anti-CD44 antibody treatment directly altered CSC fate by either inducing differentiation or by inhibiting their repopulation ability. This study provide evidence that passive immunotherapy with antibodies targeting CSCs antigen could be effective even when the same antigen is shared with NSCs. Concurrent with the above study, Krause and colleagues also reported that CD44 is required on leukemic cells that initiate chronic myeloid leukemia (CML)[59]. Anti-CD44 antibody treatment attenuated induction of CML-like leukemia in recipients, suggesting that CD44 blockade may be beneficial in autologous transplantation in CML.

Passive immunotherapy targeting solid CSCs has also been reported. Smith et al. demonstrated antibody-drug conjugates (ADCs) could be used for both hepatocellular and gastric cancers. When an anti-CD133 antibody was conjugated to a potent cytotoxic drug, monomethyl auristatin F (MMAF), this conjugate could effectively inhibited the growth of Hep3B hepatocellular and KATO III gastric cancer cells in vitro by inducing apoptosis in CD133+ CSCs. In vivo administration this ADC also resulted in significant delay of tumor growth in SCID mice.

In addition to directly targeting CSC surface antigens, antibody therapy has also been used as sensitizing agents combined with chemotherapy. Todaro et al. showed that treatment of CD133+ colon CSCs with anti-IL-4 antibody before treatment with oxiplatin, 5-FU or TRAIL resulted in increased cell death [60]. In vivo direct injection of IL-4 neutralizing antibodies followed by oxiplatin could effectively reduced tumor burden.

Targeting CSCs using active immunotherapy

Active immunotherapy is designed to generate vaccines that could stimulate the host's intrinsic immune response to the tumor. Early stage active immunotherapy vaccines for glioma treatment utilized irradiated whole tumor cell inoculation, either engineered to secrete cytokines [61]or combined with cytokine secreting cells [62] or cytokine itself [63]. Although promising data has been obtained from those tumor-cell based vaccination strategies, the success of this approach was limited by the poor inherent antigen-presenting capacity of glioma cells themselves. The use of professional antigen-presenting cells, like dendritic cells (DCs), to initiate tumor-specific T-cell responses may be a more promising strategy for cancer vaccination. Emerging evidence showed that DC-mediated antigen presentation might be more effective than using irradiated tumor cells, as DCs abundantly express many of the co-stimulatory molecules that are essential for appropriate activation of naive T cells. Also, they have the ability to efficiently process and present antigenic peptides in combination with cell-surface MHC [64–69]. For glioma immunotherapy with DC vaccines, different tumor-associated antigens, including specific tumor-associated peptides, tumor RNA and cDNA, tumor cell lysate or apoptotic tumor cells all have been tested in various studies [reviewed in 70].

In our phase I study using DC vaccines in patients with newly diagnosed high-grade glioma [71], DC vaccine was generated with patients' PBMC derived DCs pulsed ex vivo with autologous tumor cell surface peptides isolated by means of acid elution. Following surgical resection and external beam radiotherapy, nine patients were given DC vaccination intradermally very other week over a six-week period. Four patients, who showed disease progression, underwent repeat surgery after receiving the third DC vaccination. By examining the harvested tumor tissue, two of the four patients samples demonstrated robust infiltration with CD8⁺ and CD45RO⁺ T cells, which was not apparent in the same patients' tumor specimens prior to the vaccination. More encouragingly, the median survival for the study group was 455 days, which was longer than the 257 days for the matched control population. Given the promising results without observed destructive autoimmune responses, this study was expanded into a phase II trial.

In another Phase I study by using DCs pulsed with tumor lysate as antigen [72], 14 patients with malignant glioma were given three vaccinations over a 6 week period and followed with immuno-monitor assay using an HLA-restricted tetramer staining protocol. Four patients showed that at least one or more tumor-associated antigen (TAA)-specific CTL was activated against specific glioma antigens, including melanoma antigen-encoding gene-1, gp-100 and human epidermal growth factor receptor-2 (HER-2). The median survival of the study group was significantly longer than the control group of recurrent glioblastoma patients by means of 133 weeks vs. 30 weeks.

In a study by Liau and colleagues, 12 glioma patients were treated with DC vaccination by using autologous DCs pulsed with acid-eluted autologous tumor peptides. [73] Results showed six patients generated peripheral tumor-specific CTL post-vaccination without major adverse events and autoimmune reactions. The patients who developed systemic antitumor cytotoxicity had longer survival times compared with negative response patients. And all the patients who had stable disease generated a positive CTL response, whereas those with active progressive disease did not show statistically significant CTL response.

With encouraging data generated from these DC vaccine clinical trails, current studies are attempting to further improve the efficacy of this strategy by not only inducing glioma specific CTL, but by also depleting inhibitory Treg cells [74, 75]. Two European group studies show that depletion of Tregs before DC vaccination could boost antiglioma immune response leading to tumor rejection and long-term immunity. Those studies suggested combination of Treg depletion and DC vaccination is more effective to generate anti-glioma immunity.

Summary

With emerging evidence showing that glioma CSCs play an important role in tumor initiation, escape from conventional surgical, chemotherapies, targeting glioma CSCs with different therapeutic strategies provide new hope for better glioma therapies. Current immunotherapy targeting glioma studies achieved promising results. But with the complex and divergent mechanisms with which glioma evade immune surveillance, and the genetic instability of CSCs [76], a combination of therapies with two or more immunotherapy strategies may provide more benefit to eliminate gliomas. The advancement with understanding of stem cell biology, especially CSC biology, glioma CSC specific immunotherapy based on the new discovery combined with other therapeutic strategies may eventually provide new approaches to treat gliomas.