The versatile role of HuR in Glioblastoma and its potential as a therapeutic target for a multi-pronged attack

Abstract

Glioblastoma (GBM) is a malignant and aggressive brain tumor with a median survival of ~ 15 months. Resistance to treatment arises from the extensive cellular and molecular heterogeneity in the three major components: glioma tumor cells, glioma stem cells, and tumor-associated microglia and macrophages. Within this triad, there is a complex network of intrinsic and secreted factors that promote classic hallmarks of cancer, including angiogenesis, resisting cell death, proliferation, and immune evasion. A regulatory node connecting these diverse pathways is at the posttranscriptional level as mRNAs encoding many of the key drivers contain adenine- and uridine rich elements (ARE) in the 3’ untranslated region. Human antigen R (HuR) binds to ARE-bearing mRNAs and is a major positive regulator at this level. This review focuses on basic concepts of ARE-mediated RNA regulation and how targeting HuR with small molecule inhibitors represents a plausible strategy for a multi-pronged therapeutic attack on GBM.

Graphical Abstract

1. Introduction

Glioblastoma (GBM) is one of the most aggressive and invasive cancers of central nervous system (CNS) and comprises about 50% of all malignant brain tumors in adults [1]. Despite many clinical trials of multimodal therapeutic interventions, the prognosis of GBM remains poor with less than 5% of patients surviving beyond 5 years, a statistic that has not changed over 3 decades [2]. Major obstacles to the development of effective treatment arise from the incredible adaptability of the tumor to thrive under adverse conditions. Resistance to therapy arises from a small population of cells within GBM called glioma stem cells (GSC) which are defined by their ability to self-renew and generate heterogeneous tumors after transplantation [3, 4]. GSCs can thrive in harsh conditions such as hypoxia and acidosis, resist apoptosis and maintain tumorigenicity. The malignant phenotype is further supported by a favorable tumor microenvironment (TME). The TME is shaped by a complex glioma cell secretome that draws in tumor-associated microglia/macrophages (TAMs) and other immune cells to support proliferation, survival, parenchymal infiltration, and evasion of anti-tumor immune responses [5, 6]. Up to 40% of the GBM tumor mass is comprised of TAMs [5]. Effective therapies therefore need to target the multiple cellular and molecular components in GBM. A common regulatory thread for many of the intracellular and secreted factors, including cytokines, chemokines, anti-apoptotic factors, growth factors, and cell cycle regulators, that drive these tumor-promoting pathways is at the posttranscriptional level [7]. The mRNAs encoding these factors contain adenine- and/or uridine-rich elements (classically and historically referred to as the ARE) in the 3’ untranslated region (UTR) that govern RNA stability and translational efficiency [8–11]. These dual levels of regulation enable rapid changes in protein expression programs to maintain homeostasis and viability following cell damage [12]. Prime triggers of cell damage/cellular stress in the TME include hypoxia, acidosis, and glucose deficiency [3, 13]. A key factor in the posttranscriptional adaptive response of GBM to resist treatment and sustain growth under adverse conditions is the ARE binding protein (ARE-BP) HuR. In this review, we will describe the basic elements of ARE-mediated posttranscriptional regulation, the relevant mRNA targets, and how targeting HuR represents a plausible strategy for a multi-pronged therapeutic attack on GBM.

2. ARE-mediated posttranscriptional regulation

The field of ARE-mediated RNA regulation emerged more than three decades ago when Shaw and Kamen first observed that the normally stable β-globin mRNA transcript underwent rapid degradation simply by inserting an ARE into its 3’ UTR [11]. The ARE was historically defined by the presence of a core pentameric AUUUA motif, sometimes expanded to the nonamer UUAUUUAUU [14], but many other U- and AU-rich sequences, as well as sequences rich in other nucleotides, have been found to confer altered mRNA stability [15–17]. AREs are found in noncoding portions of mRNAs (predominately 3’ UTRs) in up to 85% of human genes, and in introns of more than 50% of human genes [18, 19]. A list of GBM-relevant mRNAs that are regulated by AREs is shown in Table 1. From the scope of function of the encoded proteins, one can appreciate the potentially broad impact that ARE-mediated regulation has on the biology of this tumor and other cancers. Through an interplay with trans-acting factors, including ARE-BPs and ncRNAs (miRNAs and lncRNAs), the ARE modulates two connected but distinct levels of gene regulation, RNA stability and translational efficiency (Fig. 1) [9, 20, 21]. RNA stability affects the overall levels of mRNA present in the cytoplasm, whereas translational efficiency reflects the actual utilization of mRNA for protein synthesis.

Table 1.

HuR-regulated mRNAs in glioblastoma.

ARE Target	Role in Glioblastoma			Posttranscriptional Regulation		References
	Glioma Cell	GSC	TME	↑ mRNA Stability	↓ mRNA Stability
Arg1			↑ TAM Recruitment ↑ Immune suppression		X	[108, 128]
Bcl2	↓ Apoptosis	↓ Apoptosis		X		[94, 97, 129, 130]
Bcl-XL	↓ Apoptosis ↑ Migration ↑ Invasion	↑ Proliferation ↑ Angiogenesis		X		[94, 97, 131]
CCL2			↑ TAM Recruitment ↑ Immune Suppression ↑ Angiogenesis	X		[132–135]
CCL5	↑ Migration ↑ Invasion		↑ TAM Recruitment		X	[108, 136, 137]
CCNB2	↑ Proliferation			X		[54, 138]
CCND1	↑ Proliferation			X		[54, 139]
CCNE1	↑ Proliferation ↓ Apoptosis			X		[54, 140]
c-FOS	↑ Proliferation ↓ Apoptosis			X		[54, 141]
cIAP2	↓ Apoptosis	↑ Self-Renewal ↓ Apoptosis		X		[97, 142]
c-Myc	↑ Proliferation	↑ Proliferation ↑ Self-Renewal		X		[54, 143, 144]
COX2	↑ Proliferation		↑ Immune Suppression ↑ Angiogenesis	X		[86, 145–147]
CXCL1	↑ Proliferation		↑ TAM Recruitment ↑ Immune Suppression	X		[108, 135, 148, 149]
CXCL2			↓ Immune Destruction ↑ TAM Recruitment	X		[108, 149]
CXCL3			↓ Immune Destruction	X		[149, 150]
CXCL8 (IL-8)	↑ Proliferation	↑ Proliferation	↑ TAM Recruitment	X		[94, 97, 151–153]
	↑ Migration ↑ Invasion	↑ Self-Renewal
CXCL1 0	↑ DNA Replication ↑ Proliferation		↑ TAM Recruitment ↑ Immune Suppression	X		[108, 154]
CXCR4	↑ Invasion ↑ Metastasis	↑ Self-Renewal	↑ Angiogenesis	X		[155–157]
DR5	↑ Apoptosis				X	[97, 158]
EGF	↑ Proliferation ↑ Metastasis		↑ Angiogenesis	X		[54, 159, 160]
GM-CSF	↑ Proliferation	↑ Proliferation	↑ TAM Recruitment ↑ Angiogenesis	X		[54, 161–163]
HIF1α	↑ Invasion	↑ Proliferation ↑ Self-Renewal	↑ Angiogenesis			[97, 164]
IL-1β	↑ Proliferation	↑ Proliferation ↑ Self-Renewal		X		[97, 108, 135, 165]
IL-4			↑ Immune Destruction ↓ Apoptosis	X		[166, 167]
IL-6	↑ Migration ↑ Invasion	↑ Migration ↑ Invasion ↑ Self-Renewal		X		[97, 108, 135]
IL-10	↑ Proliferation ↑ Migration ↑ Invasion		↑ Immune Suppression		X	[108, 168]
IL-13	↑ Apoptosis	↑ Apoptosis		X		[169, 170]
iNOS	↑ Migration ↑ Invasion	↑ Migration ↑ Invasion ↑ Self-Renewal	↑ Immune Suppression	X		[97, 171]
Mcl-1	↓ Apoptosis			X		[94, 172]
MMP-2	↑ Invasion			X		[108]
MMP-9	↑ Invasion			X		[108, 173]
MMP12	↑ Invasion			X		[135, 149]
PDCD4	↓ Proliferation ↑ Apoptosis			X		[174–176]
PD-L1			↑ Immune Suppression	X		[97, 108, 116, 177]
TGF-β1			↑ Immune Suppression ↑ Angiogenesis	X		[97, 108, 178]
TGF-β2		↑ Proliferation ↑ Self-Renewal	↑ Immune Suppression ↑ Angiogenesis	X		[97, 178]
TNF-α	↑ Migration ↑ Invasion	↑ Migration ↑ Invasion		X		[97, 108, 135]
VEGF-A			↑ Angiogenesis	X		[97, 108, 179]

Abbreviations

ARE, AU and/or U-rich element; GSC, glioma stem cell; HuR, Human Antigen R; TAM, Tumor-associated macrophages/microglia; TME, Tumor microenvironment

Figure 1: General principles of ARE-mediated posttranscriptional regulation and the role of HuR.

In the nucleus, HuR binds to ARE regions of some newly transcribed pre-mRNAs and participates in splicing and mRNA maturation. HuR can help translocate to the cytoplasm some ARE-bearing mRNAs, and generally promotes the stabilization and translation of bound mRNAs. Factors that promote mRNA decay, which include microRNAs and ARE-binding proteins TTP, KSRP and AUF1, are blocked from binding to the mRNA. If HuR is retained in the nucleus, has impaired RNA binding, or is sequestered by non-coding RNAs, RNA destabilizers and/or the microRNA/RISC (miRISC) complex gain access to the mRNA which is then transported to exosomes or P bodies for degradation.

3. The discovery of HuR and the link to AREs

HuR (also referred to as ELAVL1 or HuA) belongs to the Embryonic-Lethal Abnormal Vision (ELAV) family of RNA-binding proteins and was cloned based on sequence homology to the neuron-specific homologue, HuD, which was first identified as a target of anti-Hu paraneoplastic autoantibodies in patients with small cell lung cancer [22–25]. The basic structure of the ELAV family, which also includes neuronal members Hel-N1 (HuB or ELAVL2) and HuC (ELAV3), is characterized by the presence of three RNA-recognition motifs (RRMs), a hinge region between RRMs 2 and 3, and a unique N-terminal region (Fig. 2) [23, 26]. Sequence homology between HuR and neuronal ELAV members is high in the RRMs (70–90%) but drops considerably in the other regions (~15–40%) and this difference is underscored by the significantly lower immunoreactivity of HuR to anti-Hu autoantibodies [26, 27]. The broader expression pattern of HuR, beyond neural tissues and small cell lung cancer, was initially established by RNA studies [22, 25]. The link between the ELAV family and ARE binding was first made with Hel-N1 using a random RNA selection procedure which identified a binding predilection for U-rich sequences flanked by adenines and cytidines (C) [28]. RNA-binding assays then confirmed that Hel-N1 could bind to AREs in the 3’ UTRs of c-Myc, c-Fos and ID1 [26, 28]. HuR was found to have RNA binding preferences, including C- and U-rich sequences [22, 29]. Subsequent transcriptome-wide binding analyses with photoactivatable ribonucleoside enhanced crosslinking and immunoprecipitation (PAR-CLIP) with HuR confirmed these initial binding preferences and identified multiple HuR binding sites outside of the 3’ UTR including intronic sequences and 5’ UTRs [15, 30].

Figure 2: Structural features of HuR.

Schematic domain organization of HuR. The N-terminal RRM and central RRM 2 domains are separated by a short linker, while C-terminal RRM3 is connected to the tandem domains by a long intrinsically disordered hinge region (upper panel). Each of the RRMs has a cognate topology of β1α1β2β3α2β4, where the antiparallel β sheets are arranged against 2 α helices that are perpendicularly oriented (lower left panel). The N-terminal cysteine (cys13) residue, and/or C-terminal tryptophan residues (trp261, trp261 and trp271) are essential for multimerization of HuR which fine-tunes its function (lower right panel) [32, 36–38].

Detailed analyses of HuR-RNA interactions show that the N-terminal RRM1 and central RRM2 domains are essential for ARE recognition, whereas the C-terminal RRM3 interacts with poly-A and short U-rich sequences of target mRNAs [31–33]. ARE binding with RRM1 induces a conformational change in HuR, allowing further interactions with the inter-domain linker and RRM2 which enhances the ARE-binding affinity of HuR [34]. However, a recent report suggests RNA recognition involves all three RRMs of the full-length HuR [33]. Nevertheless, RRM3 discriminates between various ARE motifs and preferably binds to poly-U stretches [32, 35]. RRM3 also acts as multifunctional domain, helping in HuR multimerization and binding RNA targets using two different surfaces on opposite sides of the domain [32, 35]. In addition, multimerization was shown to be dependent on the N-terminal cysteine (cys¹³) residue, and/or C-terminal tryptophan residues (trp²⁶¹, trp²⁶¹ and trp²⁷¹) of RRM3. Multimerization is pivotal for fine-tuning HuR function (Fig. 2) [32, 36–38]. The intrinsically disordered hinge region is also important for multimerization and RNA-binding activity [38–40]. The hinge and RRM3 were found to regulate cooperative assembly of HuR oligomers on the AREs [40]. In the same study, multimerization of HuR was shown to be dependent on the density of AREs in the target transcripts [40]. Further, a HuR nucleocytoplasmic shuttling sequence (HNS) resides in the hinge region which facilitates transport of HuR back to the nucleus [41].

4. The role of HuR in ARE-mediated RNA regulation

Although HuR is predominantly nuclear, its nuclear function is not well understood. It has been found to bind to and modulate splicing of some pre-mRNAs [42] and promote cytoplasmic export of some mature mRNAs (Fig. 1). In the cytoplasm, HuR remains bound to the mature mRNA and generally protects ARE mRNAs from degradation and promotes translation by facilitating polysomal association [9, 21, 43–45]. This shift is often in response to cellular stresses or mitogens [42]. Other ARE-BPs such as Tristetraprolin (TTP) and KH-like splicing protein (KSRP) compete for binding to the same AREs but instead promote RNA degradation by relocating the mRNA to exonucleases located within P bodies or exosomes in the cytoplasm [20, 44, 46–49]. In other cases, HuR prevents degradation and promotes translation by preventing the repressive function of microRNAs and the RNA-induced silencing complex (RISC) [50]. Thus, the state of ARE mRNA stability and translational efficiency reflects a balance between negative and positive ARE-BPs and miRNAs available for binding. In GBM, the equilibrium shifts toward HuR due to its high overall expression and increased cytoplasmic localization [38, 51]. The latter is considered necessary for HuR-driven oncogenesis and is in part driven by p38 MAPK, a kinase that is active in GBM and correlates with high tumor grade, invasion, and poor survival [52–54]. Phosphorylation by this kinase can enhance HuR translocation to the cytoplasm [55–57]. The equilibrium is further shifted toward mRNA stabilization as p38 MAPK-driven phosphorylation of TTP reduces its availability and functionality as a destabilizer [38, 44, 51, 55, 58–60]. Based on our analysis of primary GBM tissue and glioma tumor cells (GTC), TTP is extensively phosphorylated and p38 MAPK plays a major role in that modification [59]. With ectopic expression of TTP in GTCs, the equilibrium could be pushed toward destabilization of transcripts including VEGF and IL-8 mRNAs resulting in downregulation of VEGF and IL-8, attenuated proliferation and increased apoptosis [59]. Mutation of key serines to alanines prevented TTP phosphorylation, retained TTP activity and increased the susceptibility of GTCs to temozolomide [61]. Therapeutic strategies to target HuR will be discussed later in this review, but center around shifting the equilibrium toward RNA destabilization by blocking HuR’s ability to bind AREs and/or translocate to the cytoplasm.

5. Functional interplay between HuR and microRNAs

In addition to phosphorylation, subcellular localization, and interaction with other RNA-binding proteins, HuR function is richly controlled through microRNAs. This control is elicited in two main ways: through microRNA-mediated changes in HuR abundance in the cell, and through cooperative and competitive actions of HuR and microRNAs to modulate expression of other proteins (Fig. 1).

Several prominent microRNAs control HuR abundance by binding the ELAVL1 mRNA (which encodes HuR) in response to different stimuli [62–66] and modulate HuR production post-transcriptionally. On the one hand, miR-519-, miR-125a-, miR-16- and miR-34a-mediated repression of HuR synthesis reduces cancer cell proliferation and chemotherapeutic resistance and enhances apoptosis, thereby suppressing the oncogenic function of HuR [62, 64–66]. On the other hand, miR-125b-mediated repression of HuR synthesis has been associated with pro-oncogenic phenotypes, supporting a tumor suppressor function of HuR in this paradigm [63], while promotion of HuR translation by miR-155–5p was reported to support colon cancer metastasis (Al-Haidari et al., 2018). In addition to the regulation of HuR synthesis by binding ELAVL1/HuR mRNA, signal-dependent posttranslational mechanisms of protein degradation, such as ubiquitination and caspase-mediated proteolytic cleavage, can also control HuR abundance [63, 67–73].

Besides being regulated by microRNAs, HuR can also regulate the function of several microRNAs either competitively or cooperatively. The competitive regulation of microRNA function by HuR primarily includes examples in which HuR binding to the 3’UTR “hides” a microRNA-binding site and prevents it from being recognized by the miRISC [74]. For example, under amino acid deprivation, HuR translocates from nucleus to cytoplasm and derepresses the translation of CAT1 mRNA by antagonizing the action of miR-122 in hepatic carcinoma cells [50]. HuR also translocates in response to UVC irradiation and inhibits miR-125b-mediated translation repression of p53 mRNA by dissociating miRISC from the p53 mRNA 3’UTR in breast carcinoma cells [75]. A similar mechanism has been observed between HuR and different miRNAs such as miR-548c, miR-494, miR-16, and miR-200b in response to different stimuli (reviewed by Srikantan et al. [76]) (Fig. 1). In addition to a competitive function, HuR has been found to cooperate with miRNAs. For instance, HuR represses the expression of c-myc at a posttranscriptional level by recruiting let-7-loaded RISC to the 3’UTR of c-myc mRNA [77]. Further evidence suggests that HuR can also directly interact with miRNAs and regulate their function. For example, HuR interacts with miR-16, miR-29 and miR-1192, and thereby derepresses the translation of COX-2, A20 and HMGB1 mRNAs respectively [78–80]. HuR also translocates from nucleus to cytoplasm upon LPS stimulation and sequesters miR-21 as “microRNA sponge”, thereby inhibiting miR-21-mediated translational suppression of PDCD4 mRNA [81] (Fig. 1). Finally, in yet another example of HuR-mediated microRNA regulation, HuR binding to miR-122 was implicated in the export of miR-122 into extracellular vesicles under conditions of metabolic stress [82].

6. HuR and the link to GBM

The potential relevance of HuR to cancer was surmised early on as ELAV family members bound mRNAs encoding proto-oncogenes including c-Myc, c-Fos, and GM-CSF [28]. One of the first ties to cancer was in GBM, where HuR was found to be overexpressed, particularly in peri-necrotic regions within the tumor [51]. These regions represent hypoxic zones and have marked upregulation of angiogenic proteins including COX-2, VEGF, and IL-8, all of which are encoded by ARE-bearing mRNAs [83–85]. This observation was underscored by Dixon et al. in colon cancer where HuR was also found to be overexpressed and linked to increased expression of oncogenic COX-2 through mRNA stabilization of its ARE-bearing transcript [86]. Since the original study of HuR and GBM, more extensive analyses of malignant glioma and normal brain tissue samples (TCGA and Gtex) have confirmed that HuR levels are significantly elevated and increase with tumor grade (Fig. 3A). Overall survival is also markedly worse (hazard ratio of 3.8) in patients with malignant gliomas having high HuR expression (Fig. 3B). The role of HuR in promoting malignant features of cancer and its correlation with worse clinical outcome extends to multiple extracranial cancers including breast, lung and pancreatic (reviewed in [87]).

Figure 3: HuR is upregulated in malignant glioma and correlates with poor overall survival.

(A) Comparison of HuR mRNA levels in GBM and lower grade (LG) glioma tissue with normal control brain tissue. The sample numbers are shown in parentheses. * P < 10–9. (B) Overall survival in patients with malignant glioma (GBM and LG gliomas) with high HuR mRNA levels (upper 50%, n = 338) compared to those with low HuR mRNA levels (lower 50%, n = 338). Hazard ratio for the high group is 3.8. 95% CI are shown. Data obtained from GEPIA (Gene Expression Profiling Interactive Analysis) (cancer-pku.cn).

The oncogenic role of HuR in GBM can best be understood by the relevance of the mRNA targets it regulates within the three major components of the tumor: GTCs, GSCs, and TAMs (Fig. 4 and Table 1). In this triad, there is a complex network of secreted and intrinsic factors regulated by HuR that promotes classic hallmarks of cancer, including angiogenesis, resisting cell death (survival), invasion, proliferation, genome instability and immune evasion [88]. In the initial studies linking HuR to GBM, in vitro RNA binding assays indicated that HuR bound with high affinity to ARE-rich 3’ UTRs of mRNAs encoding growth factors that support these cancer hallmarks including VEGF, IL-6, IL-8, and TGF-β [51, 89–92]. The positive effect of HuR on the expression of these and other oncogenic targets such as the Bcl-2 family of anti-apoptotic factors (Bcl-2, Mcl-1, and Bcl-XL) and the mTORC2 subunit Rictor, was demonstrated in GTCs [93–95]. Mechanisms for this positive effect include enhanced RNA stabilization and/or translation efficiency. The number of oncogenic mRNA targets negatively impacted in GTCs by HuR inhibition has expanded recently with the advent of small molecule inhibitors and has provided additional insight into the role of HuR in promoting tumor progression [96, 97]. These include immune evasion (PD-L1, iNOS, TGF-β1 and 2), angiogenesis (HIF-1α, COX-2), survival (cIAP2, XIAP) and GSC maintenance (HIF-1α, IL-6, TGF-β1 and 2, c-Myc), invasion (MMP-2, and 9, CCL5, ITG-A4), proliferation (Rictor, GM-CSF, CCNA1, CCNB2, CCND1, CCNE1). Interestingly, HuR associates with centromeres in GTCs driven by cyclin-dependent kinase 5 (CDK5)-mediated phosphorylation of HuR at a proline-directed serine 202. This leads to dissociation of CCNA1 (cyclin A) mRNA and its reduced translation, and defects in centromere duplication and cell cycle progression [98]. The ultimate impact on GTCs is promotion of mitotic abnormalities, generation of aneuploid cells, and centrosome amplification with resultant chromosomal instability, another hallmark of cancer and tumor heterogeneity.

Figure 4: HuR positively regulates ARE mRNAs in GSCs, GTCs and TAMs to drive tumor progression.

(A) Immunohistochemistry of biopsies obtained from the edge and core of a GBM tumor showing extensive HuR immunostaining in tumor cells and Iba1+ TAMs both in the tumor (Tu) and the surrounding brain parenchyma (Br). There is extensive translocation of HuR in Iba1+ cells as indicated by a merged signal with HuR (yellow). Dashed line indicates approximate border between Br and Tu. Scale bars, 100 μm. (B) HuR functions within each of the three major cellular components of GBM to promote intrinsic and secreted factors that drive GBM progression. GSCs, glioma stem cells; GTC,s glioma tumor cell; TAMs, tumor-associated microglia and macrophages; CTLs, Cytotoxic T lymphocytes; PMN-MDSCs, polymorphonuclear myeloid-derived suppressor cells;

GSCs represent the second component of the GBM triad supported by HuR. This small population of cells which drives treatment resistance and GBM progression can be enriched in cell culture through selection with the GSC marker, CD133 [99, 100]. Using three patient-derived GBM xenografts, we found that CD133⁺ cells were especially sensitive to HuR inhibition with viability assays indicating a ~ 2- to 10-fold lower IC50 than the parent cell line after treatment with the HuR inhibitor, MS-444 [45, 97]. Sublethal doses of MS-444 completely inhibited neurosphere formation, a hallmark of stemness. HuR inhibition also attenuated invasive properties of GSCs in culture. Since GSCs are major drivers of invasiveness [101], this may partially explain prior work showing a significantly reduced infiltration of GBM tumors intracranially when HuR was silenced [94]. The increased sensitivity of GSCs to HuR inhibition may relate to the marked suppression of IL-6 (by 75%) which is required for their maintenance through autocrine and paracrine pathways [97, 102]. Other factors that support GSC maintenance that are suppressed by HuR inhibition include TGF-β1 and 2, iNOS, and HIF1α [97, 103–106].

TAMs represent the third component of the GBM triad that relies on proper HuR function. TAMs, which are under the paracrine influence of GTCs and GSCs, are essential for creating a TME favorable for tumor progression [5, 107]. HuR positively regulates many of the factors secreted by TAMs that provide trophic support for GSCs and GTCs, modify the ECM for invasion and angiogenesis, and promote evasion of the immune system (Fig. 4 and Table 1). When HuR was selectively deleted in TAMs using CX3CR1-driven Cre recombinase, there was a significant attenuation of tumor growth and a prolongation of survival after intracranial injection of syngeneic GTCs [108]. Assessment of the immune microenvironment showed an overall reduction in TAMs but with an increase in the proportion of M1-like polarized cells (MHCII^hi, CD86⁺) and a decrease in M2-like polarization (MHCII^lo, CD206⁺). There was also an attenuation of polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs). PMN-MDSCs function as suppressors of T cell function and correlate with glioma grade and treatment resistance [109, 110]. In contrast, there was an overall increase infiltration of CD4⁺ T cells including Th1 and cytotoxic granzyme B⁺ and IFNγ⁺ effector cells which can directly kill cancer cells [111]. Interestingly, this cytotoxic activity is dependent on MHCII expression which is typically suppressed in GBM [112, 113]. However, in mice bearing microglia with ablated HuR, MHCII was increased not only in TAMs but also in GTCs [108]. The altered immune environment in these mice could relate to changes in molecular targets of HuR in TAMs including attenuation of CXCL1 and 2 (major chemokines for PMN-MDSCs), and an increase in CXCL10 which boosts recruitment of CD4⁺ cells and M1-like polarization [114]. Production of PD-L1, a transmembrane protein in TAMs (and GTCs) that suppresses anti-cancer T cell function, was also attenuated [107, 108, 115]. PD-L1 mRNA contains AREs in the 3 ‘UTR [116] and is markedly induced in TAMs by conditioned media from GTCs [108]. Both chemical inhibition and genetic deletion of HuR suppressed this induction [97, 108]. Other features of the TME that might be affected by HuR deletion in TAMs include the ECM (suppressed MMP2), angiogenesis (suppressed VEGF), and trophic support for GTCs and GSCs (suppressed VEGF, iNOS, IL-1β, IL-6, IL-8, and TGF-β1).

Taken together, HuR is an essential component of GBM progression through its regulation of intrinsic and secreted factors in the three major cellular components of GBM. The secreted factors govern a complex interplay in this triad and enable tumor cells to thrive in a hostile environment.

7. HuR inhibition as a therapeutic strategy in GBM

HuR as a potential therapeutic target was initially considered based on its upregulation in GBM and colon cancer, and its positive regulation of angiogenic factors [51, 86, 93]. Definitive proof of concept, however, was obtained from the discovery that shRNA-mediated HuR silencing in GTCs led to significant attenuation of tumor growth and invasion in vivo using an intracranial model [94]. The in vivo effect of HuR knockdown was observed by other laboratories [95]. Investigation of GTCs in culture show that HuR silencing reduces anchorage-dependent growth and increases apoptotic induction by chemotherapies used in the treatment of GBM including etoposide, topotecan and cisplatin [94]. Studies with xenograft-derived GTCs provide evidence that the extrinsic pathway of apoptosis is activated after HuR inhibition, with cleavage of caspase 8 and upregulation of death receptor 5 (DR5) [97]. A similar response was observed in HuR-silenced pancreatic cancer cells where HuR was thought to function as a stress-related translational repressor through binding to the 5’ UTR [117]. This role is distinct from ARE-mediated translational effects in the 3’ UTR and underscores the important concept that HuR function is not limited to AREs in the 3’ UTR. For example, HuR has been shown to bind certain internal ribosome entry site (IRES) in the 5’ UTR either to repress (e.g. insulin-like growth factor receptor and the anti-proliferative CDK inhibitor p27) or promote translation (e.g. the anti-apoptotic factor XIAP) [118–121].

In addition to the anti-cancer effects of HuR silencing in GTCs, ectopic overexpression of HuR enhances chemoresistance and expression of the anti-apoptotic factor Bcl-2 [94]. Taken together, these lines of evidence supporting an oncogenic role for HuR have fueled the search for small molecule inhibitors [87, 122]. Strategies have focused on two critical properties of HuR that are required for its regulatory function: RNA binding and cytoplasmic translocation (Fig. 5A). These two properties can be altered physiologically by posttranslational modifications to HuR, including phosphorylation, ubiquitination, methylation and proteolytic cleavage, at different amino acid residues within the RRMs and hinge region between RRMs 2 and 3 (reviewed in [42]). The first HuR inhibitor was discovered in a high-throughput screen using a fluorescence anisotropy-based assay to identify molecules that could disrupt HuR-ARE complexes [45]. The prototype inhibitors that emerged, including MS-444, disrupt HuR homodimerization, which appears to be a prerequisite for RNA-protein complex formation and ultimately translocation to the cytoplasm. While ARE binding can enhance multimerization, it does not appear to be an absolute requirement [38, 123]. In GBM, Filippova et al. observed extensive multimerization of cytoplasmic HuR in GBM clinical samples and postulated that this is an essential feature of HuR-driven oncogenesis [38]. They developed an elegant split-luciferase assay to detect and quantify HuR multimerization in GTCs (Fig. 5B). They fused the C- or N-terminal half of the firefly luciferase to HuR in two separate plasmids and transfected them into GTCs. Upon HuR dimerization, reconstitution of the luciferase protein produces luminescence with the addition of luciferin. With this assay they confirmed that MS-444 is a potent inhibitor of multimerization (Fig. 5B). Moreover, 15,16-dihydrotanshinone-I (DHTS), a compound identified by Provenzani et al. to inhibit complex formation between HuR and the TNF-α ARE [124], also disrupted multimerization (Fig. 5B) and suppressed GTC viability in cell culture [38]. Our preliminary studies indicate that DHTS suppresses glioma growth in mice (unpublished). These findings combined, with our other work [97], have implicated HuR multimerization as a critical step in glioma oncogenesis and a suitable target for developing therapies. Using this assay, Filippova et al. performed a high throughput screen of a synthetic library and identified a new family of small molecule inhibitors of HuR multimerization, with SRI-42127 being the prototype (Fig. 5B) [96]. Computational docking analysis predicts that SRI-42127 interacts with 4 key residues located in RRMs 1 and 2, suggesting that inhibition of multimerization results from a conformational change. Although the modeling predicts a disruption in RNA binding, this has not yet been tested. ARE binding and HuR multimerization appear to be closely linked, but it remains unclear which comes first.

Figure 5: Development of small molecules that block HuR multimerization.

(A) Based on studies with primary GBM tissues, there is extensive multimerization of cytoplasmic HuR which is an essential component to its oncogenic effect [38, 96]. (B) A split-luciferase assay was developed by Filippova and Nabors to detect HuR multimerization [38]. The N- and C-terminal halves of firefly luciferase were fused to HuR in separate plasmids which were then transfected into malignant glioma cells. Upon HuR dimerization, there is reconstitution of the luciferase protein as indicated by luminescence after addition of luciferin to the culture media. MS-444, a previously identified inhibitor of HuR dimerization [180] and lethal to malignant glioma cells [97], potently attenuates luminescence (shown to the right). A control plasmid expressing full-length luciferase is not affected. Likewise DHTS, which was shown by others to inhibit HuR-TNF-α ARE complex formation [124], also suppresses dimerization. The assay was then used in a high-throughput screen where a new class of dimerization inhibitors was discovered with SRI-42127 being the prototype [96].

The prototype inhibitor, SRI-42127, shows excellent penetration into the CNS (a barrier for new therapeutic molecules that often prevents advancement to the clinic) and suppresses GBM growth in vivo [96]. Interestingly, SRI-42127 potently suppresses HuR translocation in activated microglia and the production of key GBM-promoting factors associated with TAMs including IL-1β, IL-6, IL-8, iNOS, TNF-α, CCL2, CXCL1 and 2 [106]. Thus, the drug targets both GBM and the TME, adding potency to its anti-cancer effects. Because of the high amino acid sequence homology in the RRMs between HuR and neuronal members of ELAV [25], SRI-42127 may impact the function of neuronal ELAV members. DHTS for example, inhibits HuD-ARE complex formation [124]. It should be noted that other HuR inhibitors have been discovered that disrupt ARE complex formation or HuR trafficking but have not been tested in GBM and go beyond the scope of this review. The reader is directed to a recent review which nicely details the different candidate inhibitors, their putative mechanisms of action, and their anti-cancer effects [87].

8. Conclusions and perspectives

GBM remains a highly lethal tumor with little advancement in treatment options over the past three decades. Current strategies for treatment of glioblastoma rely on the modalities of surgery, radiation therapy, and primarily alkylating agent-based chemotherapy [125]. Although the Cancer Genome Atlas (TCGA) has identified potentially targetable oncogenes in GBM, such as EGFR, the extreme heterogeneity of the tumor substantially limits the clinical benefit of targeted therapies [126]. As covered in this review, HuR promotes multiple pro-oncogenic pathways within the major cellular components of GBM by increasing the stability and translational efficiency of mRNA transcripts encoding key molecular drivers of these pathways. Therapeutic targeting of HuR with a small molecule inhibitor such as SRI-42127 would therefore have a multipronged effect in suppressing GBM growth, invasion and resistance to therapy. This impact provides a strong rationale for advancing the discovery and refinement of small molecule HuR inhibitors for treatment of GBM and other cancers. The brain permeability of SRI-42127 overcomes a large obstacle in the treatment of GBM and holds promise for this class of inhibitors as adjuvant therapy to more conventional treatment modalities. Newer delivery methods using nanotechnology, convection-enhanced drug delivery, or intra-tumoral injections may further enhance this treatment approach [127].