RNA Interference for Glioblastoma therapy: Innovation ladder from the bench to clinical trials

Abstract

Glioblastoma multiforme (GBM) is the most common and deadliest type of primary brain tumor with a prognosis of 14 months after diagnosis. Current treatment for GBM patients includes “total” tumor resection, temozolomide-based chemotherapy, radiotherapy or a combination of these options. Although, several targeted therapies, gene therapy, and immunotherapy are currently in the clinic and/or in clinical trials, the overall survival of GBM patients has hardly improved over the last two decades. Therefore, novel multitarget modalities are urgently needed. Recently, RNA interference (RNAi) has emerged as a novel strategy for the treatment of most cancers, including GBM. RNAi-based therapies consist of using small RNA oligonucleotides to regulate protein expression at the post-transcriptional level. Despite the therapeutic potential of RNAi molecules, systemic limitations including short circulatory stability and low release into the tumor tissue have halted their progress to the clinic. The effective delivery of RNAi molecules through the blood-brain barrier (BBB) represents an additional challenge. This review focuses on connecting the translational process of RNAi-based therapies from in vitro evidence to pre-clinical studies. We delineate the effect of RNAi in GBM cell lines, describe their effectiveness in glioma mouse models, and compare the proposed drug carriers for the effective transport of RNAi molecules through the BBB to reach the tumor in the brain. Furthermore, we summarize the most important obstacles to overcome before RNAi-based therapy becomes a reality for GBM treatment.

Graphical abstract

Figure 1. This diagram summarizes the most important steps in the development of RNAi-based therapies for GBM treatment. The first step involves the identification of potential targets for the RNAi approach; and the evaluation of the efficacy of the RNAi-based therapy in GBM cell lines and animal models. The second step, involves the design of nanocarriers for RNAi delivery into the brain. These carriers, when administered systemically, should be able to cross the blood brain barrier (BBB) and reach GBM tumor tissue. In this step, RNAi-nanocarriers should be characterized in vivo and evaluated for tissue distribution, safety, and efficacy using additional animal models. Finally, clinical trials evaluating overall survival, progression free survival, and safety RNAi doses should be performed.

I. Introduction

Glioblastoma multiforme (GBM) is an aggressive central nervous system (CNS) disorder that affects about 2 to 3 out of 100,000 adults per year and is responsible for more than 14,000 deaths annually in the United States (National Cancer Institute). GBMs are fast growing tumors that normally form in the cerebral white matter without showing visible symptoms until the tumor has become large [1]. It is the deadliest of all malignant primary brain tumors with a mean survival rate of 14 months with standard of care treatment (American Brain Tumor Association, 2016). Compelling evidence indicates that in GBM cells, as in many tumor types, multiple oncogenic and tumor-suppressor pathways are altered, and multi-targeted combined therapy is recommended [2,3]. The current standard therapy for GBM patients is tumor resection (surgery) followed by radiotherapy (XRT) and/or Temozolomide (TMZ)-based chemotherapy [4,5]. TMZ is an FDA approved oral alkylating drug that crosses the blood-brain barrier (BBB) and once in the nucleus of the cells, transfers a methyl group to the purine bases in the double-stranded DNA inducing methyl-DNA adducts [6]. Such DNA adducts induce nicks in the DNA leading to cell cycle arrest and apoptosis [7,8]. Over-activation of the DNA repair enzyme, O⁶-methylguanine-DNA methyltransferase (MGMT) can lead to resistance to TMZ in GBM patients [7,9]. In fact, one great disadvantage of TMZ-chemotherapy is that about 90% of GBM patients acquire resistance and do not respond to a second round of TMZ treatment [7]. Reports have shown that TMZ increases overall survival of GBM patients only by 2.5 months [10]. Other studies have shown that the current standard of care, surgery followed by TMZ and radiotherapy, combined with other drugs such as bevacizumab have not shown significant improvement in overall survival of patients compared with control group cohorts [11,12]. Therefore, there is an urgent need to develop new therapeutic modalities for the treatment of GBM patients that can improve their overall survival.

RNA interference (RNAi) has emerged as a novel treatment modality for different human diseases including cancer. RNAi consists of using small oligonucleotides (21–45 base pairs) of single or double stranded RNA molecules to inhibit protein synthesis. In one RNAi-based therapy modality, a 21–27 base pair double stranded small interfering RNA (siRNA) is introduced into cells where it binds to its specific complementary messenger RNA (mRNA) sequence and inhibits protein synthesis (effect commonly called RNA silencing) [13]. Such siRNAs are designed to target a single gene, which is generally overexpressed in cancer cells compared to normal cells. The second RNAi modality consists in targeting microRNAs (miRNAs) with either miRNA inhibitors or mimics. MicroRNAs (miRNAs) are naturally occurring endogenous small non-coding RNAs (18–22 nucleotides) that bind preferentially to the 3’-Untranslated Regions (3’-UTR) of their cognate messenger RNAs regulating gene expression at the post-transcriptional level [14]. MiRNA binding to the 5’-Untranslated Regions (5’UTR), and to coding sequences have also been observed [15,16].

Despite the potential of RNAi-based therapy, clinical limitations include short circulatory stability, rapid clearance from the body, and inadequate delivery to the brain tumor tissue [17]. An additional limitation for the development of RNAi as a real modality for GBM treatment is the restriction of the blood brain barrier (BBB). The BBB is an anatomical and physiological barrier that functions by controlling what enters and exits the brain to maintain stable neural function. The BBB maintains proper ion levels, brain nutrition, and protects the brain tissue from chemical insult and potential damage induced by neurotoxic agents [18]. Endothelial cells, which compose most of the lining of the BBB’s capillaries, remain tightly sealed mainly through tight junctions. These tight junctions are composed of proteins that form the barrier such as claudins, occludins, junction adhesion molecules, and cytoplasmic accessory proteins [19]. These interactions only allow molecules with specific characteristics to pass into the brain through the blood. BBB also precludes the entry of more than 98% of potential therapeutics [20]. For this reason, the BBB is considered the rate limiting factor for the development of all new therapeutics, including RNAi therapies, for the treatment of GBM; and other neurological disorders [21,22].

This review discusses the most relevant information regarding the use of RNAi-based therapy in GBM cell lines and GBM mouse models; we also discuss the major devices used for RNAi delivery into the brain.

II. RNAi-based strategies

RNAi is a natural biological system that uses single or double stranded RNA (dsRNA) to induce sequence-specific posttranscriptional gene silencing [13]. Working with the nematode Caenorhabditis elegans (C. elegans), Craig Mello and Andrew Fire discovered that double-stranded RNA is more effective than single-stranded RNA at interfering endogenous mRNA in cells. For their work, Mello and Fire received the Nobel Prize in Physiology or Medicine on 2006 [23]. Inside cells, hairpin RNA or long double stranded RNA molecules are processed in an RNase III-containing protein complex (DICER) generating short double-stranded siRNAs. This siRNA duplex is then incorporated into the RNA-induced silencing complex (RISC). Once the duplex is unwound, one of the chains binds to its complementary mRNA preventing protein synthesis. On the other hand, the combined work of Ambros and Ruvkun laboratories, both working also in C. elegans [24,25], discovered the mechanism by which naturally occurring small non-coding RNAs (miRNA) inactivate gene expression through imperfect base-pair complementarity with its cognate mRNAs. MiRNA molecules are initially processed in the nucleus by Drosha (an RNase III enzyme) to form a ~70 nucleotide precursor pre-microRNA. This precursor is transported to the cytoplasm where it is processed in a similar mechanism as siRNA molecules [26]. A single miRNA can potentially bind to thousands of different mRNAs, and a particular mRNA can be regulated by more than one miRNA [10]. Multiple studies involving various types of human cancers have demonstrated that aberrantly expressed miRNAs can contribute to cancer cell growth, proliferation, and tumor maintenance [27–30]. Intercellular miRNA-mediated communication adds an additional layer of complexity to the multiple roles of miRNAs in cancer biology. Soon, after finding that miRNAs regulate more than 50% of human genes, it became apparent that targeting dysregulated miRNAs could be a molecular and therapeutic tool to manipulate gene expression at the posttranscriptional level.

In the laboratory, synthetic small oligonucleotides (21–45 base pairs) of single or double stranded RNA molecules are used in different ways. In one RNAi-based modality, a 21–27 base pair double-stranded RNA (siRNA) is introduced into cells where it binds to its specific complementary mRNA sequence to induce short-term silencing (2–4 days) of protein-coding genes [13]. SiRNAs are designed to target a single gene which is generally overexpressed in cancer cells compared to normal cells. SiRNAs can also be stable transfected in cell lines or administered in animal models using DNA constructs encoding a dsRNA molecule with a hairpin loop (shRNAs) which give the advantage of studying the chronic effect of the treatment [31,32]. On the other hand, dysregulated miRNAs are targeted with either miRNA oligonucleotide inhibitors or mimics. For miRNAs aberrantly increased in cancerous cells, oligonucleotide miRNA inhibitors (anti-MiRs) are used. When miRNAs are abnormally decreased, oligonucleotide miRNA mimics are desirable. Since a single miRNA potentially regulates multiple mRNAs at the same time, miRNA-targeted therapies could induce more beneficial effects than siRNA-based therapies [33,34]. However, miRNA-based therapies should be carefully designed for several reasons: (i) a single miRNA may regulate genes with different or even opposite biological functions and may lead to somewhat unpredictable effects [35]; (ii) the expression levels of a miRNA are generally tissue specific and these levels could change depending on the differentiation stage of a cell in a particular tissue [35,36]; (iii) as the mRNA levels in a cell can change over time, miRNAs will bind to the most abundant transcripts and the biological effect may not produce the expected results [35,36]; and (iv) the occurrence of competing endogenous RNAs (ceRNAs), also known as miRNA “decoys” or miRNA “sponges”, that decrease the availability of certain miRNAs [37–39]. CeRNAs include protein coding mRNAs, pseudogenes, and long non-coding RNAs (lncRNAs) that compete with other transcripts for the binding to the same miRNA via base-pairing with miRNA recognition/response elements (MREs) [37,38].

Several dysregulated miRNAs have been identified in GBM cell lines and GBM patients [3,40]. Particularly, upregulation of miR-21, miR-10b, miR-92b, and miR-27a; and down-regulation of miR-7, miR-145, miR-34a, and miR-128 have been identified in GBM cell lines and/or GBM tumor tissues [41–46]. The use of specific inhibitors have been effective; for instance, the use of a miR-27a inhibitor reduced cell proliferation and induced apoptosis in CRL-1690 GBM cells [45], miR-21 inhibition led to a decreased migration and invasion of A172 GBM cells [47], and miR-10b inhibition reduced tumor growth, invasion, and angiogenesis of U87-MG GBM cells [48]. Dysregulation of other miRNAs has been associated with drug resistance of GBM cells. For example, Sun et al. showed that overexpression of miR-181b sensitizes U87-MG glioma cells to teniposide by targeting MDM2 [49]. In another study, Munoz et al. demonstrated that high miR-9 expression in GBM cells results in Temozolomide chemoresistance due to an increase in drug transporters MDR1 and ABCG2 downstream of miR-9 [50].

The critical role of miRNAs in the maintenance of GBM stem cell populations has also been studied. Inhibition of miR-10b with an antisense oligonucleotide inhibitor (ASO) reduced cell viability and induced apoptosis of glioblastoma stem cells (GSC), and reduced tumor growth and progression of established intracranial GBM tumors in mouse models [51]. Taken together, targeting dysregulated miRNAs could reduce cell growth and proliferation, induce apoptosis, reduce cell invasion and migration, sensitize GBM cells to chemotherapeutic agents; and destroy GBM stem cell populations. Therefore, some of these miRNAs have been proposed as targets for GBM therapy.

Targeting oncogenic and cell survival pathways with siRNA-based strategies have also demonstrated great therapeutic potential. In a recent study, Wichmann et al. compared the effect of targeting the epidermal growth factor receptor (EGFR) and the human epidermal growth factor receptor 2 (HER2) with antibodies (cetuximab and trastuzumab, respectively) and siRNA-mediated EGFR and HER2 silencing in GBM cells. They found that the antibodies had no significant effect on the phenotype of the parameters measured. However, HER2 targeted siRNA significantly reduced cell growth, cell migration, and cell clonogenicity; and increased the radiosensitivity of these cells [52]. This study highlights the potential of RNAi therapy over other therapies that are currently in the clinic.

III. RNAi-based therapies using GBM mouse models

Mice are ideal models for cancer biology studies and to test the effectiveness of novel chemotherapeutic agents. Mice can be manipulated genetically, easily bred, used to test same drug administration routes as in humans, and they are mammals with a similar genome to the human. Mouse models that have been used to test novel drugs for GBM treatment include transgenic, xenograft, orthotopic, syngeneic, and chemically induced models. In the next sections we discuss the most relevant studies using xenograft, orthotopic, syngeneic, and spontaneous GBM mouse models for RNAi therapy.

Subcutaneous xenograft mouse model

One of the most used GBM mouse models is the subcutaneous xenograft model. It involves the subcutaneous injection of GBM cells into one flank of the mouse for evaluation of tumor growth and drug efficacy. Various protocols describing this technique have been published [53,54]. The major advantages of this model include its low technical complexity, no specialized equipment is necessary for tumor implantation, tumor growth is easily monitored, and intra-tumoral injections are easily performed. He et al. used this model to evaluate the anti-tumoral effect of a conjugated EGFR-targeted siRNA to a cyclic arginineglycine-aspartic acid (cRGD) peptide. The cRGD peptide possesses high affinity for αvβ3 integrins which are highly expressed in GBM cells compared to normal cells [55]. Intravenous injections of this conjugate (siRNA-cRGD) in U87-MG-bearing subcutaneous xenograft mice decreased the mRNA and protein levels of EGFR; and reduced tumor growth. Furthermore, they observed higher specificity of the siRNA-EGFR conjugate for the tumor tissue compared with other organs [56]. In another study, U87-MG cells overexpressing miR-1908 were injected subcutaneously into the flank of athymic mice; and increases in tumor volume, tumor weight, and cellular proliferation (Ki67-marked cells) were observed. These studies confirmed the oncogenic role of miR-1908 in GBM cells [57].

However, there are considerable disadvantages of using this model to test novel anticancer agents, especially miRNA-based therapies, because the biological role of most miRNAs are tissue specific [58]. Moreover, since the tumor is grown and treated outside the organ of origin (brain) the tumor microenvironment is not comparable. Evidence has shown that for most tumor types, the intercellular communication of tumor cells with non-neoplastic surrounding cells (tumor microenvironment) activate signaling pathways that promote tumor growth [59–61]. Dendritic cells, macrophages, and even neurons surrounding the GBM tumor have been demonstrated to release growth factors that promote tumor growth [62–64]. The effect of the tumor microenvironment in therapy efficacy has been studied by Quail et al. where they observed that insulin-like growth factor-1 (IGF-1) present in brain tumor surrounding macrophages binds to the IGF-1 receptor of GBM tumor cells. This interaction activated the phospho inositide-3 kinase (PI3K) pathway and led to therapy resistance via a molecular blocker of the colony stimulating factor 1 receptor (CSF-1R) [60]. Other studies have also shown the impact of the tumor microenvironment on tumor progression and drug efficacy [63,65].

Interestingly, exosome miRNA trafficking by neighboring cells has also been observed as part of the complex intercellular communication of tumors including GBM [62,63]. Thus, the positive effects of miRNA-based therapies using subcutaneous GBM models may not be reproduced when intracranial models are used. In addition, the BBB is not contemplated by using subcutaneous GBM mouse models. Therefore, this is not the most appropriate clinically translatable model to test novel anticancer agents, particularly RNAi. Obvious differences in the composition and characteristics of the cells and molecules present in the tumor microenvironment should be considered. Therefore, the subcutaneous model could be a first step after in vitro studies that should be further validated using other models such as intracranial orthotopic xenografts for more reliable pre-clinical conclusions of the effectiveness of RNAi-based therapies.

Orthotopic xenograft mouse model

Orthotopic xenograft mouse models involve the implantation of human GBM cells directly into the brain (organ of origin) of immunocompromised (nude) or severe combined immunodeficient (SCID) mice, reflecting a more clinically relevant tumor setting compared to the subcutaneous model. For most GBM orthotopic experiments cell injections are made in the striatum of a frontal lobe of the brain [51,53]. This model has a higher impact during the discovery phase of novel drugs (including RNAi) since the tumor grows in its organ of origin reflecting a more legitimate situation to that of GBM patients. Additionally, it has become a modality to inject fluorescent or luminescent-labeled cells to visualize the tumor growth and its distribution noninvasively with the suitable instrumentation.

In a recent study, Fareh et al. used an orthotopic mouse model to study the potential beneficial effects of cell-based therapy for GBM. They intracranially implanted patient-derived glioblastoma stem-like cells (GSC) both naïve and ectopically expressing miR-302–367, a miRNA cluster with a tumor suppressive biological role [66]. MiRNA-containing exosomes were naturally released from the GSC cells and the miRNA molecules were taken up by the naïve neighboring GSC cells; which provoked a reduction in tumor growth. They concluded that these miRNAs are transferred from cell-to-cell by exosomes affecting the stemness, proliferation, and tumorigenicity of both the exosome-releasing and exosome-acquiring cells. The authors of this study suggest that using cell-based therapies could be an effective method to deliver miRNA-containing exosomes as a new therapeutic alternative for GBM treatment [66]. In another study, Teplyuk et al. used the orthotopic xenograft mouse model to evaluate the therapeutic effect of a miR-10b antisense oligonucleotide (ASO) in patient-derived GSCs. They compared three different administration routes: direct intratumoral administration, continuous delivery with an osmotic pump [implanted subcutaneously (s.c.) and connected by a cannula to the mouse brain]; and systemic intravenous (i.v.) injections. A comparable reduction in tumor growth was observed in all three routes used [51]. These studies support the advantages of using orthotopic GBM mouse models to test RNAi-based therapies.

Although orthotopic xenograft mouse models have high clinical relevance for initial drug screening, some disadvantages of this model include: (1) the lack of spontaneous tumor formation reduces the opportunity to study tumorigenesis, (2) the use of immunocompromised mice affects tumor immunology, an important aspect of tumor progression and drug response, (3) the exogenous cell implantation may affect microenvironment communication, and (4) the difficulty of multiple intracranial injection cycles for evaluation of multiple rounds of therapy.

Some of the above described studies have used patient-derived primary cells which increase the clinical relevance at the discovery phase of novel chemotherapeutic agents. Established cell lines have been shown to acquire genetic and morphologic changes when maintained in vitro for long periods of time. In fact, the cell culturing process itself has been shown to induce DNA methylation, consequently altering gene expression which might affect the tumor progression and the response to therapy [67]. The use of established cell lines to test the effectiveness of novel chemotherapeutic agents probably is one of the main reasons for the low percentage of studies that advance from pre-clinical studies to clinical trials and ultimately to the clinic. The use of patient-derived primary GBM cells could reduce this problem. Patient-derived primary cells can be continuously maintained using GBM xenograft mice [53].

Syngeneic mouse models

Syngeneic models, also called allograft models, consist of the implantation of mouse cells of same genetic background as the host. These models are used in GBM and other diseases to assess the value of a specific treatment in an immunocompetent environment. Some studies have used this model as a proof of concept after using an orthotopic xenograft model. For example, Teplyuk et al. used an orthotopic xenograft model to test the therapeutic effects of miR-10b antisense oligonucleotide (ASO) in GBM tumor growth, and further validated their findings by using a syngeneic mouse model. They used Black 6 albino immunocompetent mice and implanted intracranial tumors of mouse GL261 GBM cells. When these mice were treated continuously with miR-10b ASO or control ASO via osmotic pumps for two weeks, a reduction in tumor growth in the treatment group was observed compared to the control group [51]. Similarly, Priester et al. used a syngeneic model to evaluate how STAT3 expression affects GBM tumor growth in vivo. Here, they intracranially implanted Tu-2449 mouse GBM cells into immunocompetent B6C3F1 mice. The mouse cells contained a control vector or a short hairpin RNA (shRNA) for STAT3. A survival analysis showed that mice with shRNA-mediated STAT3 silencing lived longer compared with control-shRNA mice [68].

Spontaneous GBM mouse model

Although the above described mouse models have been used to study many aspects of GBM growth and to test the effectiveness of novel drugs, the creation of mouse models to better recapitulate GBM tumorigenesis and its natural response to drugs are necessary. In this sense, spontaneous mouse models have been generated. For example, Jijiwa et al. bred a spontaneous tumor mouse model of non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice heterozygous for a mutation in the Sonic hedgehog receptor patched 1 (Ptc) with NOD/SCID mice heterozygous for the P53 gene (Ptc +/−, p53 +/−). The progeny formed spontaneous intracranial malignant tumors enriched with multipotent brain tumor stem cells. They measured the expression of the CD44 cell surface protein variant isoform 6 (CD44v6) on the brain tumor stem-like cells in this mouse model. They found that the CD44v6 was highly expressed in the tumor stem-like cells of these mice compared to normal stem cells [69]. They proposed CD44v6 as a potential candidate for GBM stem-like progenitor cell therapy, as this protein is a marker of Temozolomide resistance [70]. Therefore, spontaneous tumor models can better imitate tumorigenesis, origin, and progression of GBM tumors, as well as their natural response to chemotherapeutic agents.

IV. Assessing RNAi delivery for GBM treatment

The development of effective RNAi-based therapies for cancer treatment has been halted due to major limitations that include low stability of the RNAi molecules and rapid renal clearance. These limitations lead to a short circulatory half-life of the RNAi molecules, inefficiency in penetrating plasma membranes, potential toxicity, and poor delivery to tumor tissue [71–73]. Generally, the BBB precludes brain entry of 100% of large molecular therapeutics and more than 98% of small molecules [74,75]. Therefore, drugs that weight more than 400–500 Daltons (Da) are unable to reach the brain [74]. The BBB contains transport mechanisms (carrier mediated transport, adsorptive endocytosis, and receptor-mediated transcytosis) that enable the passage of molecules to the brain parenchyma [21,76]. Although the BBB is compromised in GBM patients, it still represents the major limitation for the development of new drugs against this disease. Therefore, an effective RNAi-based GBM treatment needs to be able to increase RNA stability, to effectively be internalized into tumor cells, and to bypass or deceive the BBB. In summary, poor stability of the RNAi molecule; and administrative, vascular, cellular, and immunological barriers should be overcomed to improve the therapeutic efficiency of RNAi-based therapies [77]. Below, we describe some approaches aimed to improve both the half-life and the delivery of RNAi molecules into the brain. Some relevant RNAi-based therapies under investigation for GBM treatment are summarized in Table 1.

Table 1.

Summary of relevant studies using RNAi-based therapies for GBM.

RNAi type	Description of the study	Effect	Reference
siRNA/EGFR	Mouse xenografts. Intravenous administration of siRNAs conjugated to cRGD peptide.	Decreased the EGFR mRNA and protein levels; and reduced tumor growth	[56]
AntimiR/miR-10b	Mouse orthotopic xenografts. Intratumoral, continuous delivery by osmotic pump, and intravenous administration of antimiRs.	Reduced tumor growth was observed with each of three routes used for administration.	[51]
AntimiRs/miR-10b	Mouse syngeneic model. Intratumoral administration of antimiRs.	Reduced tumor growth	[51]
shRNA/STAT3	Mouse syngeneic model using shRNA-containing GBM cells.	Increased overall survival	[68]
AntimiR/miR-92b	Mouse xenografts. Intratumoral administration of naked antimiR-92b.	Decreased tumor burden.	[78]
AntimiR/miR-21	In vitro studies in U251 human GBM cells. Treatment with antimiR-R8 CPPs.	Reduced the expression of miR-21-target genes; and inhibited cell migration.	[137]
siRNA/stathmin	Mouse xenograft. Local administration of siRNA-PP75.	Reduced Stathmin expression and inhibited tumor growth.	[138]
siRNA/PLK1	Mouse orthotopic xenogafts. Intravenous administration of liposome- encapsulated siRNA	Reduced PLK1 expression and tumor growth.	[170]
siRNA/EGFR	Mouse orthotopic xenografs. Intra venous administration of siRNAs encapsulated in transferrin targeted liposomes.	Downregulated EGFR gene. Prolonged the overall survival of GBM bearing mice.	[171]
AntimiR/miR-21	Mouse syngeneic model. Intratumoral administration of antimiRs encapsulated in CTX labeled lipid particles.	Enhanced apoptosis and decreased tumor size when administered with sunitinib.	[172]
siRNA/Bcl2L12	Mouse orthotopic xenografts. Intravenous administration of gold SNAs.	Reduced Bcl2L12 mRNA levels, increased apoptosis, and decreased tumor burden.	[178]
siRNA/Bcl2L12 (UN-0129)	Clinical trial Phase I (NCT03020017). Intravenous administration of UN-0129 prior to surgery.	Ongoing	[179]
miR-182 duplexes	Mouse orthotopic xenografts. Intravenous administration of gold SNAs.	Reduced tumor size, increased overall survival.	[180]
miR-7 mimic	Mouse orthotopic xenografts. Intravenous administration of microRNAs mimics entrapped in an integrin targeted polymeric nanoparticle.	Reduced angiogenesis and tumor proliferation	[181]
miR-146b	Rat intracranial primary brain tumor model. Intratumoral administration of exosomes containing this miRNA.	Rapid reduction in tumor volume following drug administration.	[184]

EGFE, Epidermal Growth Factor Receptor; cRGD peptid, cyclic arginine glycine-aspartic acid (cRGD) peptide; STAT3, signal transductor and activator of transcription 3, R8, octoarginine; CPP, Cell penetrating peptides; PLK1, polo-like kinase 1; CTX, Chlorotoxin; Bcl2L12, Bcl2 like protein 12; SNAs, Spherical nucleic acids.

Naked RNAi for GBM Therapy

The use of naked oligonucleotide RNA molecules has shown promising results for GBM therapy in vitro and in vivo. For example, Zhe Bao Wu et al. showed that local administration of antimir-92b in a subcutaneous U87-MG xenograft mouse model led to the downregulation of Smad3 (a miR-92b target gene) and a decrease in tumor burden [78]. However, the presence of ribonucleases greatly reduces the duration of the RNAi knockdown effects. Normally, after systemic administration in mice, siRNAs have demonstrated to have a serum half-life from minutes up to 1 hour [79] and a circulatory half-life of less than 5 minutes [80,81]. Therefore, to achieve extended beneficial effects, oligonucleotide RNAi molecules should be chemically modified.

Chemical modifications in the siRNA/miRNA strands are basically the same performed to antisense oligonucleotides (ASOs) (recently revised by Khvorova and Watts) [82,83]. The most common modification in siRNA strands is the substitution of one or two oxygen to sulfur in the non-bridge phosphate groups to produce phosphorothioates (PS) and phophodithionate (PS2), respectively. In fact, CPG-28, an immunostimulating CpG oligodeoxynucleotide (CpG-ODN) synthesized with a phosphorothioate backbone, has been evaluated in phase I and II clinical trials for recurrent and de novo GBM patients [84–86]. However, experimental evidence has shown that more than 50% of the PS modified sense siRNA strands lead to severe toxic effects [87,88].

Similarly, substitution of one oxygen in the phosphate group by BH₃ (boranophosphate) increased siRNA serum stability and silencing effects [35,89,90]. On the other hand, modifications in the 2’-OH position of the ribose ring are highly desirable since this OH group is not required for siRNA recognition by the RISC complex [91]. Thus, the 2′-OH in the ribose moiety is commonly substituted by 2′-O-methyl (2′-OMe), 2'-O-methoxy-ethyl (2'-O-MOE), fluorine (2′-F), and hydrogen (2′-H). All of these modifications increased their thermal stability and were more resistant to enzymatic digestion [35,92]. Additionally, the substitution of 2′-OH by 2′-F in both antisense and sense siRNA strands increased serum stability and the siRNA binding affinity to the mRNAs [93]. By contrast, full modification of the 2′-OH with 2′-OMe in the sense, antisense, or both siRNA strands reduced and/or abolished the silencing effect [35,94]. Combinations of 2′-OMe, 2′-F, and PS are commonly performed. Generally, the guide strand (antisense) is modified with 2′-F while the sense strand with 2′-OMe [95,96]. Nevertheless, the number of purines and pyrimidines modified in each or both strands induce different levels of toxicity, silencing, and immune responses [94,97,98].

Other chemical modifications that have gained popularity are the intramolecular linkage of 2′-oxygen to 4′-carbon in the ribose ring via methylene (Locked nucleic acid, LNA) or ethylene (Ethylene-bridged nucleic acid, ENA) bridges. Reports showed that both, LNA and ENA increased serum stability, reduced immune response, and improved silencing effects [99–102]. These modifications are highly position-sensitive since undesirable steric and conformational changes could abolish RNAi activity [35,82,103]. Similarly, phosphorodiamidate morpholino oligomers (PMO) with a morpholine ring instead of the ribose and with phosphoroamidate intersubunit linkages have shown promising therapeutic potential (revised by Evers, Toonen and Roon-Mom) [104]. Although PMOs are resistant to nuclease and protease degradation they are rapidly cleared from the blood [105,106]. PMO can be further modified with peptides and other molecules to increase the cellular uptake and pharmacokinetics [107].Unfortunately, toxic effects in mouse models have been reported for both LNA and morpholinos [108–110].

Modifications at the 5’-phosphate of the RNAi strands, essential for RISC complex recognition, are desirable [111,112]. Critically, after systemic administration, this terminal is quickly removed by phosphatases resulting in biologically inactive siRNA accumulation [82]. For example, Ionis pharmaceutical has substituted the 5′-terminal phosphate of single-stranded siRNA (ss-siRNA) by an E-vinyl phosphonate (5′-E-VP) [113]. Other studied modifications at the 5′-terminal phosphate include the 5′-methyl phosphonate and 5′-C-methyl phosphate [114]. In vivo studies are necessary to clarify which of these modifications induce better therapeutic effects without toxic effects.

To reduce immune response, off-target effects, and increase stability and therapeutic potential, other chemical and structural modifications have been evaluated. For example, long synthetic 27-mer duplex RNA without overhangs were more efficient in gene silencing than the corresponding traditional 21-mer siRNA [115]. Also, a synthetic short hairpin RNA (shRNA composed of 29 base-paired stems with 2 nt 3′-overhangs and 4 nt loops) was processed by DICER and improved RNAi potency [116]. Asymmetric RNA duplexes (aiRNA) with short 15 nt sense strand having both 3′ and 5′ antisense overhangs induced efficient gene silencing and reduced off-target effects [117]. Small internally segmented interfering RNA (sisiRNA) (with an intact antisense strand and sense strand divided into two segments), and a fork shaped siRNA (1–4 nt mismatch at the 3′-end of sense strand) also showed reduced off-target effects and maintained their silencing activity [118,119]

More sophisticated siRNA-containing nanostructures have been proposed. For example, endless dumbbell-shaped circular siRNAs [120,121], complexes of PEI and short oligomers of complementary overhang siRNAs (sticky siRNA) [122], linear PEI complexes with several cross-linked siRNAs molecules [123], 3D siRNA microhydrogels [124], RNAi microsponges [125], chimeric packaging RNA (pRNA) derived from the bacteriophage phi29 DNA [126,127], and RNA nanorings functionalized with siRNAs [128,129]. RNAi-aptamers, RNAi attached to cell penetrating peptides, and RNAi CNS-ligand conjugates have been explored for GBM therapy and are discussed in the next sections.

RNAi-Aptamers for GBM

Aptamers are short DNA or RNA molecules selected by an in vitro process known as Systematic Evolution of Ligands by Exponential Enrichment (SELEX) system [130]. They are chemically synthesized oligonucleotides that have shown to be stable upon thermal and pH changes, with high affinity and specificity for targeted cell receptors, and with low immunogenicity and toxicity [130–132]. They have been investigated for a wide range of applications including imaging [133] and cancer therapeutics [131]. For example, AS1411, a Nucleolin specific aptamer, was the first oligodeoxynucleotide to reach phase I and II clinical trials for the treatment of cancers [134], including acute myeloid leukemia (NCT01034410) and renal cell carcinoma (NCT00740441). Additionally, Ye Cheng et al. demonstrated in vitro and in vivo, the therapeutic potential of this aptamer for GBM. Their results showed that treating U87-MG, U251, and SHG44 GBM cells with AS1411 reduced cell migration [134]. Local administration of AS1411 led to decreased tumor growth and prolonged overall survival in a subcutaneous U87-MG xenograft mouse model [134]. Bioconjugation of aptamers to siRNA, miRNAs mimics, and anti-miR molecules have shown promising results in GBM cell lines [135,136]. However, more feasibility studies are necessary to assess their effectiveness as carriers of RNAi for GBM treatment.

RNAi-cell penetrating peptides

RNAi can also be conjugated to cell penetrating peptides (CPP), which can enhance RNAi internalization into cells. CPPs are small peptides containing 6–30 amino acids that are capable of membrane translocation and internalization by adsorptive mediated endocytosis [137]. RNA oligonucleotides can form complexes with peptides such as octoarginine (R8) which can aid in their cellular internalization and endosomal escape in comparison to unmodified RNAi molecules [137]. Consequently, CPPs could increase the RNAi therapeutic effect. For example, Y. Zhang et al. showed that antimir-21/R8 treatment of U251 human GBM cells led to the downregulation of the well-known miR-21-regulated genes, PDCD4 and SERPINB5; as well as inhibition of cell migration by 25% in comparison to the antimir/R8 negative control group [137]. Sariah Khormaee et al. evaluated the effect of covalently attaching stathmin-targeted siRNAs to a PP75 cell penetrating peptide [138]. Stathmin is a microtubule-regulating protein shown to be involved in GBM resistance to nitrosourea chemotherapy [139]. In this study, PP75-Stathmin siRNAs were locally administered into subcutaneous U251 xenograft mouse models. Their results showed that local treatment of PP75-Stathmin siRNAs was effective decreasing stathmin mRNA (60% decrease) and stathmin protein (50% decrease) levels in comparison with control groups (Stathmin siRNA without PP75 labeling) [138]. Although CPP is an option to enhance RNAi uptake by cells, additional carriers that cross the BBB and reach the tumor tissue should be used.

RNAi and Central Nervous System (CNS) Specific Ligands

Another approach to improve RNAi delivery into the brain is by using CNS targeting ligands. Here, either small or large molecules can bind to specific receptors on the surface of endothelial cells and cross the BBB by transcytosis mechanisms [22]. Several CNS targeting ligands such as apolipoprotein E [21,140,141], Angiopep-2 [142,143], transferrin [144,145], and Rabies virus glycoprotein (RVG) peptide [20,146,147] have shown to improve drug delivery to the brain through the BBB. For example, mouse tail vein injection of siRNA molecules attached to RVG peptides (siRNA-RVG) showed a significant increase of RNA delivery to the brain parenchyma [20]. The authors of this study suggest that although siRNA transport across the BBB was increased, their stability and effectiveness could be further enhanced by using chemical modifications or nanoparticle vehicles [20].

Nanoparticles for RNAi delivery

Conjugation of oligonucleotide RNA molecules to peptides and other ligands have shown improvements in delivery, internalization, and treatment efficiency. However, the major inconveniences of stability and renal clearance have not been solved. Currently, nanoparticles (NPs) are preferred over other carriers for systemic administration of RNAi molecules and other drugs. Nanoparticles of different materials ranging from 1 to 300 nm in diameter, decorated on their surface with several ligands are under consideration for diverse applications including imaging, diagnosis, and brain tumor treatment [71,148]. Furthermore, NPs can transport hydrophilic or lipophilic molecular cargos [22]. The most attractive characteristics of NPs include their high surface-to-volume ratio that enables the loading of large amounts of drug into tumor tissues [149], their ability to increase the longevity of the treatment in circulation, controlled release upon physiological changes such as pH [150], and their ability to accumulate in solid tumor tissue by the enhanced permeability retention (EPR) effect [151]. In GBMs, EPR occurs by virtue of the highly angiogenic nature of GBMs which leads to the imperfect formation of new vascular vessels, and partial BBB disruption [152]. In this way, nanoparticles of sizes between 1–100 nm can passively cross the BBB and accumulate in brain tumor tissue [153]. Additionally, the tumor environment can have a pH as low as 6 in comparison to normal tissue of 7.4 [154]. This phenomenon gives predilection to NPs to release their payload in response to pH changes. Finally, NPs can be multi-functionalized or loaded not only with RNAi molecules but also with chemotherapeutic agents and CNS-specific ligands, raising their probability to reach brain tumor tissue increasing treatment efficiency. The most used nanoparticles for drug delivery include liposomes and gold nanoparticles.

RNAi-Lipid based-nanoparticles for GBM

Nanoliposomes are small nanometric vesicles (30–100 nm) mainly composed of a phospholipid bilayer that forms upon the exposure of a dried lipid film to water [72]. They are composed of an aqueous cavity and a hydrophobic membrane, which allows the incorporation of lipophilic and hydrophilic drugs. Generally, the lipids used for liposomes are biocompatible, biodegradable and of low toxicity [155]. A great number of DNA/RNA drug delivery methods are being developed with these formulations [156]. Additionally, the phospholipid portions of the liposomes can be linked with polyethylene glycol (PEG) or polyethyleneimine (PEI) molecules to increase liposome blood stability. PEG and PEI are also functionalized through amine, carboxylic acids or maleimide groups to peptides, antibodies or any other ligands for active targeting [71,157,158]. Several liposomal formulations for cancer therapy have been approved by the Food and Drug Administration (FDA) [159,160], and have shown promising results in preclinical and clinical studies for drug delivery [150,161,162]. One example is a clinical trial phase I study for the delivery of a siRNA-containing liposomal formulation against EphA2 (Ephrin type-A receptor 2) to solid tumors (NCT01591356) [163]. The Ephrin-A2 receptor has shown to be involved in tumorigenesis and tumor progression of many types of solid tumors, including ovarian cancer [164–166]. Presently, two phase II clinical trials for recurrent GBM patients are using liposomes as treatment delivery platforms: [a] liposomal-Rhenium 186 (NCT01906385) [167], and [b] SGT-53 in combination with oral temozolomide (NCT02340156) [168]. SGT-53 is a nanoliposome formulation, decorated with transferrin receptor ligands, and containing p53 cDNA. Systemic administration of SGT-53 showed positive antitumor effects in preclinical studies and was safe in phase I clinical trials [169]. RNAi delivery platforms involving the use of nanoliposomes have been evaluated in preclinical studies. H.-M. Liu et al. used a malate dehydrogenase liposomal formulation with Polo-like kinase 1-targeted siRNA (siPLK1). Systemic administration of the siPLK1-liposomes was able to enhance siRNA uptake, inhibit of PLK1 protein expression, and reduce tumor growth in intracranial GBM mouse models [170]. Lin Wei et al. used a formulation of EGFR-targeted siRNAs (siEGFR) encapsulated in protamine/chondroitin sulfate cationic liposomes (LPC) decorated with T7 (a transferrin specific peptide) to treat an orthotopic U87-MG xenograft mouse model by tail vein injection [171]. Their results showed that systemic administration of T7-LPC/siEGFR accumulated preferably in the brain tumor tissue, increased downregulation of EGFR, and extended the overall survival of mice compared to the non-targeted formulation [171]. Lbcl2ipid particles have also been used to intracranially deliver anti-miR oligonucleotides (miR-inhibitor) in GBM mouse models. Pedro Costa et al. systemically administered anti-miR-21 lipid particles, labeled with chlorotoxin (CTX, a glioma-specific peptide) [172]. This treatment led to anti-miR-21 accumulation in brain tumor tissue, effective silencing of miR-21, and upregulation of RhoB (miR-21 target mRNA) in a syngeneic GL261 mouse model [172]. In addition, administration of this formulation in combination with sunitinib (a tyrosine kinase receptor inhibitor) decreased tumor size and enhanced apoptosis in GBM intracranial tumor-bearing mice [172].

RNAi-Gold Nanoparticles for GBM

Gold nanoparticles (AuNP) are biologically inert, easily synthesized, commercially available and highly stable composites [173,174]. Over the last decade, they have gained great popularity in the biomedical field. AuNPs have shown to possess feasible characteristics for diagnostic, imaging, and therapy. The surface of AuNPs can enable multiple coupling with drugs, CNS-specific ligands and oligonucleotides [175,176]. In fact, thiolated nucleic acids can be covalently attached to the AuNPs and therefore facilitate their delivery to brain tumors. Nucleic acids attached to AuNPs are known as spherical nucleic acids (SNAs), and they have shown to cross the BBB and the brain tumor barrier (BTB) through targeting class A scavenger receptors [177]. Jensen et al. prepared a formulation (SNA) of a thiolated siRNAs against the oncogene Bcl2L12 (siBcl2L12) bound to 13 nm AuNPs, and systemically injected them into an orthotopic U87-MG xenograft mouse model [178]. Results showed that SNA administration led to an effective reduction of Bcl2L12 mRNA, increased apoptosis of GBM cells, and decreased tumor burden [178]. BCl2L12 specific SNAs, currently known as UN-0129, are undergoing a clinical trial (NCT03020017) where UN-0129 is intravenously injected into GBM and gliosarcoma patients prior to surgery [179]. Moreover, SNAs are being considered for the delivery of miRNA-based therapies. For example, Kouri et al. evaluated the tumor suppressive role of miR-182 in vivo by using SNA AuNPs with covalently attached miR-182 miRNA duplexes. Results showed that systemic administration of this formulation was able to cross the extravascular glioma parenchyma [180]. In addition, the miR-182-containing formulation led to tumor reduction and extended animal survival, making the miR-182-SNAs a potential strategy for GBM intervention [180].

Other carriers for RNAi delivery to GBM tumors

Many other nanoparticle systems, including polymeric particles and exosomes, are being considered as vehicles for RNAi delivery into the brain. In a study performed by N. Babae et al., a miR-7 (negative regulator of angiogenesis) mimic entrapped in an integrin-targeted biodegradable polymeric nanoparticle system was systemically administered in orthotopic U87-MG xenograft mouse models [181]. Promising outcomes were obtained since the treatment led to a reduction in angiogenesis and U87-MG tumor proliferation [181]. In addition, exosomes have emerged as a new modality for RNAi delivery. Exosomes are vesicles of 40–100 nm in diameters secreted by many types of cells into the body fluids [182,183]. Katakowski et al. used mesenchymal stromal cell (MSCs) exosomes as carriers for RNAi delivery. They transfected MSCs with a miR-146b expression plasmid which led to the secretion of exosomes containing this miRNA. Then, the isolated exosomes were locally administered into glioma tumor tissue of an intracranial rat model of primary brain tumors [184]. Results of this study showed that the miR146b-exosomes decreased tumor volume 5 days after treatment [184]. The use of MSCs as a delivery method for the intracranial administration of a miR-124 mimic into GBM xenograft mouse models was also evaluated by A H. Lee et al. Here, bone marrow mesenchymal stem cells (BM-MSCs) were transfected with Cy3-labeled miR-124 mimics, and then administered adjacent to a GBM tumor of a GFP-U87 xenograft mouse model [65]. Results showed that BMMSCs were able to deliver Cy3-labeled miR-124 mimics to orthotopic U87-MG xenograft tumors. The therapeutic potential of this formulation was not evaluated in this study [65]. Of note, exosomes are “dirty” vesicles in the sense that they are produced according to the cell’s needs, phase, and stimuli [185] with an heterogeneous mixture of different nucleic acids [186]. However, they have been demonstrated to be good carriers of RNAi molecules [65,184,187]. More studies are needed to determine if any of these nanoparticle formulations are able to cross the BBB, deliver their RNAi cargo specifically into the GBM tumor tissue, reduce/eliminate the tumor, and increase the overall survival of the subjects.

V. Concluding Remarks

Currently, there is no optimal treatment for GBM patients and multi-targeted therapies that eliminate the tumor cells, prolong life, and improve the quality of life for GBM patients are urgently needed. Since the discovery of RNAi as an efficient, precise, and stable regulation system it’s immense applicability in biology and medicine has become evident. First, standard chemotherapeutic agents cannot discriminate between molecules differentially abundant in cancer vs. non-cancer cells. Many of these differentially abundant molecules are proteins that are considered “undruggable” due to their planar surface which obstructs the design of small molecule inhibitors. In this case, RNAi-based therapy is recommended. Second, the necessity to target multiple dysregulated cellular pathways in GBM cancer cells could be achievable by targeting miRNAs, where a single miRNA can regulate multiple targets. Unfortunately, the in vivo delivery of RNAi-based therapies is a challenge due to instability (short half-life), inability to traverse the plasma membrane, possible toxicity, and activation of the immune system. An ideal RNAi-based therapy should be non-toxic, non-invasive, and without off-target effects. In addition, when administered systemically, this therapy should be able to cross the BBB and reach specifically the GBM tumor cells. To overcome these obstacles, nanocarriers for RNAi have been proposed. Liposomes and AuNPs have shown great potential as delivery systems for RNAi-based therapies. They can both be functionalized with different ligands which enable them to cross the BBB and bind specifically to cancerous cells in the brain tumor. The development of delivery platforms with these characteristics can open new avenues for the treatment of brain tumors and other neurodegenerative diseases using RNAi-based molecules. Finally, the evaluation of the therapeutic efficacy of RNAi-based therapies and other drugs requires the selection of the proper animal model. To obtain reliable data that can be translated into the clinic the chosen model should closely imitate what occurs in GBM patients. In addition, researchers should consider using more than one model to confirm their results and display reproducibility.