The development of novel therapeutics that exploit alterations in the activation state of key cellular signaling pathways due to mutations in upstream regulators has generated the field of personalized medicine. These first-generation efforts have focused on actionable mutations identified by deep sequencing of large numbers of tumor samples. We propose that a second-generation opportunity exists by exploiting key downstream “nodes of control” that contribute to oncogenesis and are inappropriately activated due to loss of upstream regulation and microenvironmental influences. The RNA-binding protein HuR represents such a node. Because HuR functionality in cancer cells is dependent on HuR dimerization and its nuclear/cytoplasmic shuttling, we developed a new class of molecules targeting HuR protein dimerization. A structure–activity relationship algorithm enabled development of inhibitors of HuR multimer formation that were soluble, had micromolar activity, and penetrated the blood–brain barrier. These inhibitors were evaluated for activity validation and specificity in a robust cell-based assay of HuR dimerization. SRI-42127, a molecule that met these criteria, inhibited HuR multimer formation across primary patient-derived glioblastoma xenolines (PDGx), leading to arrest of proliferation, induction of apoptosis, and inhibition of colony formation. SRI-42127 had favorable attributes with central nervous system penetration and inhibited tumor growth in mouse models. RNA and protein analysis of SRI-42127–treated PDGx xenolines across glioblastoma molecular subtypes confirmed attenuation of targets upregulated by HuR. These results highlight how focusing on key attributes of HuR that contribute to cancer progression, namely cytoplasmic localization and multimerization, has led to the development of a novel, highly effective inhibitor.

Significance:

These findings utilize a cell-based mechanism of action assay with a structure–activity relationship compound development pathway to discover inhibitors that target HuR dimerization, a mechanism required for cancer promotion.

HuR belongs to the ELAV family of mRNA-binding proteins, and since our seminal observation of its upregulation in tumors of the central nervous system (CNS) and colon cancer, it has become a clear chemotherapeutic target for many types of cancer (1, 2). Increases in HuR expression, abnormal cytoplasmic localization, and molecule dimerization are reported for most cancer types and aberrantly regulate thousands of transcripts involved in tumor genomic instability, cell-cycle progression, immune resistance, and the suppression of apoptosis (3–5). Several approaches have been taken in the past decade to inhibit HuR function in cancer cells. These approaches include inhibition of (i) HuR/mRNA interaction, (ii) HuR dimerization/multimerization, (iii) HuR nuclear/cytoplasmic shuttling, and (iv) HuR expression. To date, the following key compounds/scaffolds have been identified as potential inhibitors of HuR function: MS-444 as a blocker of HuR dimerization and nuclear/cytoplasmic shuttling (6); CMLD1, CMLD2, quercetin, azaphilone derivatives, dihydrotanshinone-I (DHTS), suramin, and NSC# 84126 as the modulators of HuR/mRNA interaction (4, 7–12); and okicenone, trichostatin, and 5-aza-2′-deoxycytidine as the inhibitors of HuR shuttling (6, 13). The IC50s of these HuR inhibitors range from 300 nmol/L to 10 μmol/L in the direct biochemical assays and from 1 to 100 μmol/L in in vitro assays. The inhibition of xenograft tumor growth in vivo has been reported for colorectal, lung, melanoma, glioma, and pancreatic cancer primarily using si- or miRNA constructs (14–20). Despite the plethora of HuR inhibitors that have been proposed, none target the specific oncogenic-driven changes related to HuR of subcellular localization and dimerization in in vivo tumor models.

In a recent article, we reported on a strategy to optimize searches for inhibitors of HuR cytoplasmic multimerization in transformed cells (21). In the current article, we utilized a combination of medicinal chemistry with a cell-based HuR-specific biochemical assay to identify a new class of molecules that disrupt HuR cytoplasmic dimerization. This approach of developing a detailed understanding of the biological role with the cell-based assays capable of allowing readout of cancer-promoting mechanism of action with concurrent medicinal chemistry modifications augments the chance of success in targeted therapy development. We confirmed that the discovered HuR inhibitors are cytotoxic and suppress glioma growth and progression using in vitro and in vivo models. A detailed analysis of cell signaling pathways affected by these inhibitors has been performed on patient-derived glioma xenolines (PDGx) cell lines revealing disruption of HuR-dependent cell signaling pathways essential for the tumor cell proliferation, survival, invasion, DNA damage repair, and antitumor immune resistance, thereby providing molecular biomarkers for patient outcomes and future drug development.

Cell lines and cell culture

XD456 (ProNeural), JX6 (Classical), JX10 (Classical), JX12p (Classical), JX22p (Mesenchymal), X14p (Classical), and X1524 (ProNeural) primary PDGx with molecular subgroup were previously established (18); neurospheres were formed and maintained as previously described in the Neurobasal-A medium (Gibco) supplemented with a B-27 supplement without vitamin A (Gibco), N-2 supplement (Gibco), 2 mmol/L l-Glutamine (Mediatech), 100 U/mL penicillin/streptomycin (Mediatech), and the basal growth factors EGF, 20 ng/mL, and bFGF, 20 ng/mL (Thermo Fisher Scientific; ref. 22). U251 (Sigma-Aldrich), U87 (Sigma-Aldrich), and LN229 (ATCC) glioma cell lines were maintained in DMEM/F12 medium (Thermo Fisher Scientific) supplemented with 10% FBS (Gibco), 2 mmol/L l-Glutamine (Mediatech), and 100 U/mL penicillin/streptomycin (Mediatech). Primary human neurons with neuronal medium and poly-l-lysine were purchased from ScienCell Research Laboratories and maintained according to the company's protocol on the poly-l-lysine–coated plates. Primary human astrocytes with astrocyte medium were purchased from ScienCell Research Laboratories and were retained according to the company's protocol. Stable cell lines expressing the doxycycline-inducible Fluc, HuR-Nluc, plus HuR-Cluc constructs were developed and maintained as previously described (21); the clone stability was ensured by utilizing Hydromycin B and G418 Sulfate (Mediatech) cell-selection antibiotics in the DMEM/F12 medium (Thermo Fisher Scientific) supplemented with 10% FBS (Gibco), 2 mmol/L l-Glutamine (Mediatech), and 100 U/mL penicillin/streptomycin (Mediatech; ref. 21).

Split firefly luciferase assays

The split firefly luciferase assays and data analysis were performed as previously described (21). Briefly, the assays were performed in 96-well plates with a clear bottom (Corning Inc.). Thirty thousand to 50,000 cells were plated per well; the constructs were induced by doxycycline (0.75 μg/mL) for 48 hours. Compounds (or vehicles as the control) were administered for 6 hours. The luminescence signals were detected by using InfinitiM200 plate-reader (TECAN) in the presence of the luciferase substrate luciferin (D-luciferin, potassium salt) from GoldBio. A multichannel pipette (XL 3000ITM, Denville Scientific) was used in all procedures to ensure the experiment quality. The established inhibitors of HuR dimerization (MS444 and DHTS) were utilized as a positive control (Supplementary Fig. S1A–S1C).

Drug synthesis and medicinal chemistry

All reactions were carried out in an oven- or flame-dried glassware under argon atmosphere using the standard gas-tight syringe, cannula, and septa. All the reactions were performed using anhydrous solvents (DMF, THF, CH2Cl2, 1,4-Dioxane, 1-Butanol, CHCl3, and DME) purchased from Sigma-Aldrich. Microwave reactions were carried out using CEM discover Labmate System with Intelligent Technology for Focused Microwave Synthesizer (Explorer 48). The precise details of compound synthesis and compound characteristics were provided (Supplementary Data File S1).

Computational docking study

Computational docking study was performed to investigate the binding mode of SRI-42127 compound at HuR (details were included in Supplementary Data File S1). SRI-42127 was docked to HuR via the induced-fit docking protocol (23).

Cell-cycle analysis

Cell-cycle distribution was evaluated using a standard staining procedure with propidium iodide (Sigma-Aldrich) followed by flow cytometry as described (24).

Cell viability assay

PrestoBlue cell viability reagent (Thermo Fisher Scientific) was used for cell viability assays as previously described (21). The inhibitory dose–response data were normalized to the control (treatment with vehicles); the inhibitory dose–response curves, which were fitted by using the Boltzmann equation, and the corresponding IC50s were generated by using OriginPro software (OriginLab Corporation; ref. 19).

Immunohistochemistry

Neurons and astrocytes were fixed in 3.7% paraformaldehyde (pH 7.4) for 13 minutes at room temperature. Triton X-100 (0.3%) in PBST buffer (0.1% Tween 20) was used for cell permeabilization for 30 minutes at room temperature. Blocking buffer (3% BSA, 22.52 mg/mL glycine in PBS) was used to block unspecific antibody binding for 30 min at room temperature. Primary cleaved caspase-3 antibody (Cell Signaling Technology) was utilized at 1:200 dilution in PBS (1% BSA) buffer overnight at 4°C. The next day, cells were incubated with secondary antibody (Alexa Fluor 594 Goat anti-rabbit IgG, Invitrogen) in PBS (1% BSA) at 1:2,000/5,000 dilution for 1 hour at room temperature in the dark. DAPI was utilized for nuclear staining; images were obtained with EVOSFl (Life Technologies) imaging system.

The immunostaining on the brain tissue (fixed in paraformaldehyde, 4%) was performed as described previously (18). Briefly, tissue was permeabilized with 0.5% Triton X-100 in PBST buffer for 30 minutes, followed by 3 times wash with PBST and blocking with the Universal blocking buffer for 30 minutes. Then BEAT Blocker kit from Zymed Laboratories was utilized according to the manufactory protocol to block unspecific antibody binding to endogenous mouse IgG. The primary antibodies anti-HuR from Santa Cruz Biotechnology (1:100), anti-Bcl2 from Santa Cruz Biotechnology (1:50), and anti-Mcl1 (1:100) from Cell Signaling Technology were utilized overnight at 4°C in PBS (1% BSA) buffer. The corresponding secondary antibodies Alexa Fluor 594 Goat anti-rabbit IgG (1:1,000) from Invitrogen or Alexa Fluor 594 Donkey anti-mouse IgG (1:1,500) from Life Technologies (Eugene) were utilized in PBS (3% BSA) buffer for 1 hour at room temperature. The fluorescence signal was calibrated based on the secondary-only and primary-only control staining.

Western blotting, antibodies, nuclear, and cytoplasmic protein fractionations

Western blotting was performed as previously described (21). Nuclear and cytoplasmic protein fractionations were performed by using NE-PER Nuclear and Cytoplasmic extraction reagents (Thermo Fisher Scientific; ref. 21). Anti-lamin A/C and anti-Tubulin alpha antibodies were utilized to verify nuclear and cytoplasmic fractions, respectively. Antibodies against lamin A/C, cleaved PARP, cleaved caspase-3, SOX2, and Mcl1 were purchased from Cell Signaling Technology. Anti-Tubulin alpha antibody was purchased from Sigma-Aldrich. Antibodies against HuR, actin, and Bcl2 were purchased from Santa Cruz Biotechnology. A polyclonal anti-HuR antibody for HuR/mRNA co-immunoprecipitation (IP) was purchased from MBL-International.

HuR/mRNA co-IP assay and Taqman data

The HuR/mRNA co-IP was performed by utilizing RiboCluster Profiler RIP assay Kit (MBL-International; see Materials and Methods in Supplementary Materials). The quantifications of Bcl2, Mcl1, and 18S transcripts were performed by using Taqman technique with the following gene-specific probes: Hs00153350_m1, Hs03003631_g1, and Hs00766187_m1 (Thermo Fisher Scientific) for Bcl2, 18S, and Mcl1, respectively (18).

mRNA isolation from adherent cell culture

mRNA isolation and purification from adherent cells were performed as previously described by using the RNeasy Mini Kit and QIAshredder columns (QIAGEN; ref. 18).

Colony formation assays

The soft agar colony formation assay was performed by using 0.9% and 0.45% agarose for the bottom and the top layers, respectively; cells were incorporated in the top layer. SeaPlaque low-melting temperature agarose was from Lonza. Five hundred cells per well were utilized in both assays, which have been performed in 6-well plates; cells were treated with vehicles or with the desired drugs, which were administered twice per week for 3 weeks. The Crystal violet solution (0.1%) was utilized for colony staining. Colonies were counted after 3 weeks of treatment, and images were obtained by using Amersham Imager-600 reader.

Illumina global RNA sequencing data

Illumina global RNA-sequencing data were generated by utilizing PDGx neurospheres, which have been treated with SRI-42127, 3 μmol/L, for 12 hours, or with vehicles as the control (Supplementary Data File S2). RNA was isolated by using TRIzol reagent (Invitrogen). The RNA samples were processed in the UAB Sequencing Core facility (see Materials and Methods in Supplementary Materials).

Proteomic data

The cell pellets from PDGx neurospheres, which have been treated with SRI-42127, 3 μmol/L, for 18 hours or with vehicles as the control, were processed in the UAB CCC Mass Spectrometry/Proteomics shared facility. The proteomic data were generated by using standard procedures (see Materials and Methods in Auxiliary Supplementary Materials and Supplementary Data Files S3 and S4).

Kinase profiling assays

The kinase inhibitory potentials of SRI-42127 and SRI-41664 compounds were evaluated by Thermo Fisher Scientific's SelectScreenTM Kinase Profiling Service (the detailed reports and the 10-point titration results for inhibition of NTRK1, AAK1, IRAK1, PIM1, and cMET kinases were presented in Supplementary Data File S5).

In vivo mouse glioma model

The athymic nude mice and PDGx XD456 cells transduced with the firefly luciferase and EGFP constructs were utilized for the in vivo mouse glioma model as previously described (see Materials and Methods in Auxiliary Supplementary Materials; ref. 18).

Pharmacokinetic assessment

The in vivo pharmacokinetic (PK) study for SRI-42127 was performed by Pharmaron; PKs were evaluated in the C57Bl/6 mice; the detailed report is provided in Supplementary Data File S6.

Statistical analysis

Statistical interpretations were achieved by using a Student t test (when only two groups of data were analyzed) and one-way ANOVA with Turkey's post hoc test (when multiple data groups were analyzed). Correlation analysis was performed using Pearson's correlation. The results are shown as the mean ± SD; P values less than 0.05 were considered to be statistically significant.

Data and materials availability

All data associated with this study are shown in the article, Supplementary Figures, Supplementary Table, and Auxiliary Supplementary Materials of the Cancer Research journal website, and Gene Expression Omnibus repository.

HuR expression is associated with poor prognosis for patients with glioma

We analyzed the clinical outcomes of patients with glioma harboring low or high expression of ELAVL1 (HuR) by utilizing R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl). Low expression levels of ELAVL1 were associated with favorable prognosis (REMBRANDT Madhavan - 550 MAS.5.0-u133p2 study; Fig. 1A); the expression levels of the ELAVL1 and ACTB (actin) as a control in the glioma samples (REMBRANDT Madhavan - 550 MAS.5.0-u133p2 study) compared with the normal brain samples (N Brain 44 Harris study) are illustrated in Fig. 1B. Note the significant ELAVL1 overexpression in the glioma group compared with the normal brain group. Next, we analyzed the influence of the ELAVL1 expression on the outcome of patients with different glioma grades. ELAVL1 expression significantly affected the prognoses of patients harboring World Health Organization (WHO) grades 1, 2, and 3 tumors and did not significantly affect the outcome of patients with WHO grade 4 (Fig. 1C). The ELAVL1 expression normalized to the ACTB expression in the tumors of different grades compared with the normal brain is shown in Fig. 1D. Note the increase in average ELAVL1 expression with higher tumor grade. In our next step, we performed a mini ontology analysis of gene sets, which exhibited significant positive or negative correlations with the ELAVL1 expression, for normal brain and for each tumor grade based on R2: platform data (Fig. 1D; Supplementary Fig. S1). We found that the low-grade (1 and 2) and the high-grade (3 and 4) tumors harbored different ELAVL1-dependent gene sets and exhibited a different degree of ELAVL1 dependence. The increase in tumor grade was associated with an enhancement in the number of ELAVL1-dependent cell signaling pathways (Fig. 1E). Note that the expression of the gene set, which determines cell-cycle progression, was not significantly correlated with ELAVL1 expression in low-grade tumors; however, it exhibited a strong positive correlation with ELAVL1 expression in grades 3 and 4 (Supplementary Fig. S2). The gene set representing DNA repair cell signaling pathways was homogeneously negatively correlated with ELAVL1 expression in low-grade tumors; however, it harbored two distinguishing groups of genes with positive and negative correlations with ELAVL1 expression in grade 4 tumors.

Figure 1.

Correlations of HuR (ELAVL1) expression in brain tumors with clinical outcomes. A, Kaplan–Meier overall survival curves (x-axis represents the time since brain tumor diagnosis) demonstrate statistically significant differences (P = 1.9 × 10–16, long-rank test) between “high ELAVL1 expression” versus “low ELAVL1 expression” (median cut-off modus) for grouping of all brain tumors. B, Plots illustrate ELAVL1 and ACTB (as a control) mRNA expression in the brain tumor samples versus normal brain samples. C, Kaplan–Meier overall survival curves according to the tracks “high ELAVL1 expression” versus “low ELAVL1 expression” (median cut-off modus) are shown for the following brain tumor subsets: grades 1, 2; grade 3; and grade 4. Kaplan–Meier overall survival curves exhibit statistically significant differences according to the tracks “high ELAVL1 expression” versus “low ELAVL1 expression” for grade 1 and 2 subsets (P = 1.2 × 10–2, long-rank test) and grade 3 subset (P = 5.4 × 10–4, long-rank test). Grades 1 and 2 consist of astrocytoma (n = 65), glioblastoma multiforme (n = 2), oligodendroglioma (n = 30), and mixed (n = 4). Grade 3 consists of astrocytoma (n = 58), oligodendroglioma (n = 23), mixed (n = 3), and unknown (n = 1). Grade 4 consists of glioblastoma multiforme (n = 134). D, The graph illustrates the enhancement of ELAVL1/ACTB mRNA ratios with the increase of the brain tumor grade. Results are shown as mean ± SD; statistical significance was determined by Student t test. E, The plot illustrates the gene sets, which exhibited significant positive or negative correlations with the ELAVL1 expression for normal brain and grade 1 through 4 brain tumors.

Figure 1.

Correlations of HuR (ELAVL1) expression in brain tumors with clinical outcomes. A, Kaplan–Meier overall survival curves (x-axis represents the time since brain tumor diagnosis) demonstrate statistically significant differences (P = 1.9 × 10–16, long-rank test) between “high ELAVL1 expression” versus “low ELAVL1 expression” (median cut-off modus) for grouping of all brain tumors. B, Plots illustrate ELAVL1 and ACTB (as a control) mRNA expression in the brain tumor samples versus normal brain samples. C, Kaplan–Meier overall survival curves according to the tracks “high ELAVL1 expression” versus “low ELAVL1 expression” (median cut-off modus) are shown for the following brain tumor subsets: grades 1, 2; grade 3; and grade 4. Kaplan–Meier overall survival curves exhibit statistically significant differences according to the tracks “high ELAVL1 expression” versus “low ELAVL1 expression” for grade 1 and 2 subsets (P = 1.2 × 10–2, long-rank test) and grade 3 subset (P = 5.4 × 10–4, long-rank test). Grades 1 and 2 consist of astrocytoma (n = 65), glioblastoma multiforme (n = 2), oligodendroglioma (n = 30), and mixed (n = 4). Grade 3 consists of astrocytoma (n = 58), oligodendroglioma (n = 23), mixed (n = 3), and unknown (n = 1). Grade 4 consists of glioblastoma multiforme (n = 134). D, The graph illustrates the enhancement of ELAVL1/ACTB mRNA ratios with the increase of the brain tumor grade. Results are shown as mean ± SD; statistical significance was determined by Student t test. E, The plot illustrates the gene sets, which exhibited significant positive or negative correlations with the ELAVL1 expression for normal brain and grade 1 through 4 brain tumors.

Close modal

Therefore, extending results from previous publications (18, 21, 25, 26), our data analysis verifies that ELAVL1 overexpression favors glioma progression, and its action is grade-dependent. Consequently, ELAVL1 targeting in high-grade gliomas offers a favorable benefit/risk ratio with identifiable gene sets for monitoring its activity.

Discovery of a new inhibitor class targeting HuR cytoplasmic dimerization

In our previous report, we showed that HuR multimerization is essential for glioma progression and developed a cell-based high-throughput split firefly luciferase–HuR assay for the search and optimization of HuR dimerization inhibitors (21). Screening compounds using the split luciferase–HuR assay and established inhibitors of HuR dimerization (as positive control) led us to identify the commercially available compound A-92 (an inhibitor of the general control nonderepressible-2 kinase) as a new inhibitor of the HuR dimerization. Figure 2A illustrates an inhibitory dose response for the A-92 compound in a split firefly luciferase–HuR assay. Specific and robust inhibition of the HuR dimerization (HuR-Cluc+HuR-Nluc) compared with the control (Fluc-full length luciferase) occurred following cell treatment for 6 hours with A-92 (IC50 = 4.5 ± 0.5 μmol/L; n = 8). The maximum inhibition of HuR dimerization was 93% ± 2% (n = 8) after 6 hours of cell treatment. The luminescence signal from cells overexpressing the control Fluc construct decreased by 32% ± 7% (n = 8) following treatment with A-92; this value represents the cytotoxic effect of A-92 compound over a period of 6 hours.

Figure 2.

A new class of inhibitors of HuR dimerization prevents HuR oligomerization in cell-based split firefly luciferase–HuR assay. A, The graph represents the dose response of inhibition of HuR dimerization for A-92 compound, IC50 = 4.5 ± 0.5 μmol/L (n = 8); A-92 structure is shown beside the graph. Red-filled squares represent luminescence signal from HuR dimer reporter (U251 cells coexpressing HuR-Nluc and HuR-Cluc constructs); blue-filled circles represent luminescence signal from control reporter (U251 cells expressing Fluc construct). B, The graph represents the dose response of inhibition of HuR dimerization for SRI-41664 compound, IC50 = 2.4 ± 0.2 μmol/L (n = 6); SRI-41664 structure is shown beside the graph. Red-filled squares, signal from HuR dimer reporter; blue circles, signal from the control reporter. C, The graph represents the dose response of inhibition of HuR dimerization for SRI-42127 compound, IC50 = 1.2 ± 0.2 μmol/L (n = 6); modifications with resulting SRI-42124 and SRI-42127 compound structures are shown beside the graph. Red squares, signal from HuR dimer reporter; blue circles, signal from the control reporter. Results are shown as mean ± SD in all graphs. All dose responses were fitted by the Boltzmann function in OriginPro software. Note that the half inhibitions were not reached for A-92, SRI-41664, and SRI-42127 compounds in the control reporter assay. The IC50s for A-92, SRI-41664, and SRI-42117 compounds [4.5 ± 0.5 μmol/L (n = 8), 2.4 ± 0.2 μmol/L (n = 6), and 1.2 ± 0.2 μmol/L (n = 6), respectively, in the HuR dimerization reporter assay] were significantly different, P < 0.05, Student t test.

Figure 2.

A new class of inhibitors of HuR dimerization prevents HuR oligomerization in cell-based split firefly luciferase–HuR assay. A, The graph represents the dose response of inhibition of HuR dimerization for A-92 compound, IC50 = 4.5 ± 0.5 μmol/L (n = 8); A-92 structure is shown beside the graph. Red-filled squares represent luminescence signal from HuR dimer reporter (U251 cells coexpressing HuR-Nluc and HuR-Cluc constructs); blue-filled circles represent luminescence signal from control reporter (U251 cells expressing Fluc construct). B, The graph represents the dose response of inhibition of HuR dimerization for SRI-41664 compound, IC50 = 2.4 ± 0.2 μmol/L (n = 6); SRI-41664 structure is shown beside the graph. Red-filled squares, signal from HuR dimer reporter; blue circles, signal from the control reporter. C, The graph represents the dose response of inhibition of HuR dimerization for SRI-42127 compound, IC50 = 1.2 ± 0.2 μmol/L (n = 6); modifications with resulting SRI-42124 and SRI-42127 compound structures are shown beside the graph. Red squares, signal from HuR dimer reporter; blue circles, signal from the control reporter. Results are shown as mean ± SD in all graphs. All dose responses were fitted by the Boltzmann function in OriginPro software. Note that the half inhibitions were not reached for A-92, SRI-41664, and SRI-42127 compounds in the control reporter assay. The IC50s for A-92, SRI-41664, and SRI-42117 compounds [4.5 ± 0.5 μmol/L (n = 8), 2.4 ± 0.2 μmol/L (n = 6), and 1.2 ± 0.2 μmol/L (n = 6), respectively, in the HuR dimerization reporter assay] were significantly different, P < 0.05, Student t test.

Close modal

Next, optimization of A-92 was performed. First, we resynthesized the A-92 compound and renamed it SRI-41964 (Supplementary Data File S1). SRI-41964 exhibited identical chemical and HuR inhibitory properties (IC50 = 4.5 μmol/L) as the original compound A-92. We employed medicinal chemistry and the cell-based split firefly luciferase–HuR assay for lead compound optimization, resulting in the generation of SRI-41664 (Fig. 2B; Supplementary Data File S1). SRI-41664 exhibited improved HuR inhibition potency as well as improved aqueous solubility compared with the compound SRI-41964 due to the following structural modifications: (i) the excision of a pyrazole ring from SRI-41964 yielding SRI-42125 with improved aqueous solubility but HuR inhibition potency similar to that of the original compound, (ii) a scaffold hop of the triazolopyrimidine template of SRI-42125 to an equivalent 5–6-fused imidazopyridazine template. SRI-41664 inhibited HuR oligomerization with an IC50 = 2.4 ± 0.2 μmol/L (n = 6) in cell-based assay after 6 hours of treatment (Fig. 2B). The maximum inhibition of HuR dimerization was 93% ± 1% (n = 6) after 6 hours of cell treatment with the SRI-41664 compound. The luminescence signal from cells overexpressing the control Fluc construct decreased by 38% ± 5% (n = 6) after cell treatment with SRI-41664 for 6 hours.

In our next step of optimizing lead compounds (SRI-41964 and SRI-41664), a reduction of kinase inhibition potential was performed. This was achieved by the introduction of a methyl group on the pyrazolylamino and a pyranylamino NH (hydronitrogen group), resulting in compounds SRI-42124 and SRI-42127, respectively (Fig. 2C and Supplementary Data File S1). Compound SRI-42124 had poor aqueous solubility that precluded compound evaluation in the cell-based assay. Compound SRI-42127 inhibited HuR dimerization in the cell-based assay with IC50 = 1.2 ± 0.2 μmol/L (n = 6), and after 6 hours of treatment, the maximum inhibition of the HuR dimerization was 90% ± 1% (n = 6; Fig. 2C). The similar results were achieved in the four independent reporter cell clones overexpressing HuR-Cluc+HuR-Nluc constructs (Supplementary Fig. S3). The luminescence signal from cells overexpressing the control Fluc construct decreased by 26% ± 9% (n = 6) after cell treatment with SRI-42127 for 6 hours. Compound SRI-42127 had reduced kinase inhibition potential compared with the parent SRI-41664 compound (Supplementary Fig. S4 and Supplementary Data File S5). Note that SRI-42127 retained its HuR inhibition potential in the presence of all evaluated kinase inhibitors (Supplementary Fig. S5). Hence, the new lead compound SRI-42127 was established.

Structural modifications that we pursued in the optimization of the lead compounds and the HuR inhibition potencies of the compounds are presented in Supplementary Data File S1. A schematic summary of the key structure–activity relationship (SAR) finding that emerged from our study is presented in Fig. 3A. Structural changes that led to compounds with HuR inhibition IC50 values below 10 μmol/L are presented in Fig. 3B.

Figure 3.

Schematic illustration of optimization of the lead HuR dimerization inhibitor scaffolds. A, Schematic illustration of key SAR findings. B, A chart is listing structural changes that yielded compounds with HuR dimerization with IC50s below 10 μmol/L. The following accumulative information is presented in the chart for each compound: a molecular structure (column one), an identification number (column two), a molecular weight (column three), a maximum compound concentration utilized in the assay (column four), a maximum value of inhibition of HuR dimerization (column five), a maximum value of inhibition of the control reporter (column six), and an IC50 of inhibition of HuR dimerization (column seven). Columns five through seven represent the mean data of at least three experiments for each compound. Modifications for all compounds with resulting compound structures and HuR inhibition potentials are shown in Supplementary Table S2.

Figure 3.

Schematic illustration of optimization of the lead HuR dimerization inhibitor scaffolds. A, Schematic illustration of key SAR findings. B, A chart is listing structural changes that yielded compounds with HuR dimerization with IC50s below 10 μmol/L. The following accumulative information is presented in the chart for each compound: a molecular structure (column one), an identification number (column two), a molecular weight (column three), a maximum compound concentration utilized in the assay (column four), a maximum value of inhibition of HuR dimerization (column five), a maximum value of inhibition of the control reporter (column six), and an IC50 of inhibition of HuR dimerization (column seven). Columns five through seven represent the mean data of at least three experiments for each compound. Modifications for all compounds with resulting compound structures and HuR inhibition potentials are shown in Supplementary Table S2.

Close modal

Newly identified inhibitors of HuR dimerization suppressed glioma progression in vitro

The glioma-suppressive potentials of the new class of HuR multimerization inhibitors were evaluated by utilizing SRI-41664 and SRI-42127 compounds in several cell-based assays. First, we confirmed that the SRI-41664 and SRI-42127 compounds are cytotoxic for established and PDGx cell lines. Figure 4A illustrates the inhibitory dose responses of SRI-41664 and SRI-42127 compounds in the established U251, LN229, and the PDGx XD456 glioma cell lines. The IC50s for the SRI-41664 compound were 3.4 ± 0.5 μmol/L (n = 5), 3.8 ± 0.5 μmol/L (n = 5), and 4.8 ± 1.1 μmol/L (n = 5); the IC50s for compound SRI-42127 were 2.8 ± 0.6 μmol/L (n = 5), 3.2 ± 0.6 μmol/L (n = 5), and 4.0 ± 0.6 μmol/L (n = 5) for U251, LN229, and XD456 cell lines, respectively, after 48 hours of treatment. We employed a HuR-overexpression approach to confirm SRI-42127 inhibitory potency against HuR dimerization. In the U251 glioma cells with dox-inducible HuR dimerization reporters, an increase in HuR dimerization by 2-fold led to a significant increase in the SRI-42127 IC50 for inhibition of HuR dimerization from 1.2 ± 0.1 μmol/L (n = 4) to 1.9 ± 0.1 μmol/L (n = 4), 6 hours of treatment (P < 0.05, Student t test); the IC50 of SRI-42127 for inhibition of cell viability, 48 hours after cell treatment with SRI-42127, increased from 2.8 ± 0.4 μmol/L (n = 4) to 8.0 ± 0.8 μmol/L (n = 4; P < 0.05, Student t test) after the enhancement of HuR dimer formation by 2-fold (Supplementary Fig. S6A). The overexpression of the control reporter Fluc construct by 2-fold did not significantly alter IC50s of SRI-42127 (Supplementary Fig. S6B). Similar results have been achieved in XD456 cells; the enhancement of HuR dimer formation approximately by 2-fold increased an IC50 of SRI-42127 for inhibition of cell viability, after 48 hours of treatment, by 1.9 ± 0.2 (n = 4) fold (P < 0.05, Student t test). Supplementary Fig. S7A–S7E illustrates HuR expression in established and PDGx glioma cell lines and the IC50s of SRI-42127 compound to inhibit cell viability after 48 hours of treatment [data include parental temozolomide (TMZ)-resistant and stem cell lines].

Figure 4.

Lead compounds suppress glioma progression in vitro. A, Graphs illustrate cell viability dose responses for lead compounds in established and PDGx cell lines (48 hours of treatment). Results are shown as mean ± SD, n = 5 for all cell lines. B, Flow cytometry plots illustrate cell-cycle distribution in cell lines after SRI-42127 treatment (10 μmol/L, 18 hours) versus control (vehicle treatment). G1, S, and G2 phase values are shown in percent (%), and data were analyzed by FlowJo v10. The bar graph represents the average numbers of cells (mean ± SD, %) in each phase of the cell cycle after treatment with SRI-42127 compound normalized to the corresponding control values (vehicle treatment). The alterations in cell numbers in G1 and S phases after treatment with SRI-42127 versus control were significant for all cell lines, P < 0.05, Student t test. The alterations in G2 phases after treatment with SRI-42127 versus control were significant only for U251 and LN221 cell lines, P < 0.05, Student t test. C, Phase-contrast images illustrate primary human neurons after SRI-42127 treatment (48 hours) versus control (vehicle treatment). Scale bar, 100 μm. D, The graph illustrates that neuron numbers were not affected by the SRI-42127 treatment (48 hours) versus control (vehicle treatment). Results are shown as mean ± SD (n = 3). E, The graph illustrates the metabolic alterations in primary human neurons after SRI-42127 treatment (48 hours) versus control. Results are shown as mean ± SD (n = 3). F, The graph illustrates the percentage of cells with cleaved caspase-3 immunostaining in primary human neurons and astrocytes after SRI-42127 treatment (up to 25 μmol/L) for 48 hours. Results are shown as mean ±SD of at least three experiments for each cell type. Representative images of cleaved caspase-3 immunostaining after SRI-42127 treatment versus control (vehicle) are shown for neurons. Scale bar, 200 μm. DAPI marks the cell nucleus. *, P < 0.05, Student t test.

Figure 4.

Lead compounds suppress glioma progression in vitro. A, Graphs illustrate cell viability dose responses for lead compounds in established and PDGx cell lines (48 hours of treatment). Results are shown as mean ± SD, n = 5 for all cell lines. B, Flow cytometry plots illustrate cell-cycle distribution in cell lines after SRI-42127 treatment (10 μmol/L, 18 hours) versus control (vehicle treatment). G1, S, and G2 phase values are shown in percent (%), and data were analyzed by FlowJo v10. The bar graph represents the average numbers of cells (mean ± SD, %) in each phase of the cell cycle after treatment with SRI-42127 compound normalized to the corresponding control values (vehicle treatment). The alterations in cell numbers in G1 and S phases after treatment with SRI-42127 versus control were significant for all cell lines, P < 0.05, Student t test. The alterations in G2 phases after treatment with SRI-42127 versus control were significant only for U251 and LN221 cell lines, P < 0.05, Student t test. C, Phase-contrast images illustrate primary human neurons after SRI-42127 treatment (48 hours) versus control (vehicle treatment). Scale bar, 100 μm. D, The graph illustrates that neuron numbers were not affected by the SRI-42127 treatment (48 hours) versus control (vehicle treatment). Results are shown as mean ± SD (n = 3). E, The graph illustrates the metabolic alterations in primary human neurons after SRI-42127 treatment (48 hours) versus control. Results are shown as mean ± SD (n = 3). F, The graph illustrates the percentage of cells with cleaved caspase-3 immunostaining in primary human neurons and astrocytes after SRI-42127 treatment (up to 25 μmol/L) for 48 hours. Results are shown as mean ±SD of at least three experiments for each cell type. Representative images of cleaved caspase-3 immunostaining after SRI-42127 treatment versus control (vehicle) are shown for neurons. Scale bar, 200 μm. DAPI marks the cell nucleus. *, P < 0.05, Student t test.

Close modal

HuR is one of the key regulators of cell-cycle progression in cancer cells, particularly high-grade glioma (27); therefore, the arrest of cell-cycle progression by the newly identified compounds would be supportive of HuR-targeted inhibition. Figure 4B illustrates representative histograms of cell-cycle progression in the established U251, LN229, and the PDGx XD456 glioma cell lines in the control (treatment with vehicles) and after treatment with SRI-42127 compound. The bar graph represents the average numbers of cells in each phase of the cell cycle after treatment with the SRI-42127 compound normalized to the corresponding control values (vehicle treatment). Note a significant accumulation of cells in the G1 and the reduction in the S phases after treatment with SRI-42127, 10 μmol/L, for 18 hours. The average accumulations of cells in the G1 phase increased by 97% ± 33% (n = 3), 56% ± 28% (n = 3), and 210% ± 40% (n = 3) in U251, LN229, and XD456 cell lines, respectively, after treatment with SRI-42127 compared with the control. The average reductions of cells in the S phase were 65% ± 4% (n = 3), 60% ± 3% (n = 3), and 70% ± 3% (n = 3) in U251, LN229, and XD456 cell lines, respectively, after treatment with SRI-42127 compared with the control.

Primary human neuronal cells were evaluated as the control and exhibited no morphologic changes after treatment with SRI-42127 at up to 15 μmol/L for 48 hours compared with the control (vehicle treatment); the cell numbers and cell sprouting remained unchanged (Fig. 4C and D). The cellular metabolism declined by about 44% after treatment with SRI-42127, 25 μmol/L, for 48 hours (Fig. 4E). There were no significant alterations in patterns of cleaved caspase-3 immunostaining after 48 hours of treatment with SRI-42127 at concentrations up to 15 μmol/L compared with the vehicle control (Fig. 4F). Similar results were obtained in primary human astrocytes; we did not observe significant alterations in cleaved caspase-3 activation after treatment with SRI-42127 up to 15 μmol/L (Fig. 4F). Therefore, we conclude that compound SRI-42127 inhibited glioma cell growth and had low neuronal and astrocyte cytotoxicity at concentrations below 15 μmol/L.

Novel HuR inhibitors induce apoptosis and reduce tumorigenicity in established and PDGx glioma cell lines

HuR protein plays a significant role in stabilizing and positively regulating mRNAs of the antiapoptotic Bcl2-family (18, 28) that contributes to the protection of genetically unstable glioma cells from apoptosis. Therefore, we evaluated the Bcl2-family in the established and PDGx cell lines after cell treatment with SRI-41664 and SRI-42127 compounds and observed a significant reduction in the expression of the Bcl2-family at the mRNA and protein levels (Fig. 5A and B). After 24 hours of treatment with 10 μmol/L of SRI-41664, there were a 94% to 96% reduction in Bcl2 mRNA and an 87% to 96% reduction in Mcl1 mRNA for U251, U87, LN229, and PDGx XD456 glioma cell lines compared with control. Likewise, for SRI-42127, there were an 86% to 96% reduction for Bcl2 mRNA and an 87% to 95% reduction for MCl1 mRNA versus control. The average decreases of Bcl2/18S and Mcl1/18S mRNA ratios were 96% ± 2% and 95% ± 3%, 95% ± 2% and 86% ± 7%, 95% ± 2% and 87% ± 8%, and 96% ± 2% and 90% ± 9% based on three experiments in the established U251, U87, LN229, and PDGx XD456 cell lines, respectively, after cell treatment with SRI-42127, 10 μmol/L, for 24 hours compared with the control. A significant reduction in the interaction of the HuR protein with Bcl2 and Mcl1 mRNAs after cell treatment with SRI-42127 compound compared with the control was confirmed in the HuR/mRNA co-IP assay performed in U251 cells (Supplementary Fig. S8). Consequently, protein levels of Bcl2 and Mcl1 were evaluated in U251 and PDGx XD456 glioma cells after vehicle (control) or inhibitor (SRI-41664 or SRI-42127) treatment for 48 hours. The representative Western blots in Fig. 5B illustrate a dramatic reduction of Bcl2 and Mcl1 protein levels in the cytoplasmic fractions of both cell lines after treatment with SRI-41664 or SRI-42127 compounds compared with the control. The reduction was associated with a significant mitochondria clusterization and depolarization of the mitochondrial membrane potential evaluated by using JC-1 dye (Supplementary Fig. S9A and S9B) and was accompanied by the sign of apoptosis: cleaved caspase-3 and cleaved PARP (Fig. 5C). Similar results were observed in three independent experiments. SRI-41664 and SRI-42127 compounds exhibited stronger glioma inhibitory potency than the A-92 compound; Supplementary Fig. S10A–S10C represents A-92 compound glioma inhibitory potentials, in vitro, in comparison with lead compound SRI-41664.

Figure 5.

Induction of apoptosis and reduction of tumorigenicity of established and PDGx glioma cell lines after treatment with lead compounds. A, Graphs illustrate the percent reduction of Bcl2/18S and Mcl1/18S mRNAs following cell treatment with lead compounds versus control (vehicle treatment). Results are shown as mean ± SD (n = 3), differences are statistically significant for all cell lines with P < 0.005, Student t test. B, Representative Western blots confirm the reduction of Bcl2 and Mcl1 proteins in cytoplasmic fractions of U251 and XD456 cell lines following treatment with lead compounds, 10 μmol/L, for 48 hours. SOX2 reduction in nuclear fractions indicates a decrease in tumorigenicity of cell lines after treatment with lead compounds. C, Representative Western blots illustrate appearances of cleaved PARP and cleaved caspase-3 in glioma cell lines after treatment with lead compounds, 10 μmol/L, for 48 hours. D, Graphs illustrate a significant reduction of colony formations in U251 and XD456 cell lines after SRI-42127 (3 and 10 μmol/L treatments) versus control (vehicle treatment). Results are shown as mean ± SD, n = 3 and 4) for attached and anchorage-independent assays, respectively; *, P < 0.005, Student t test for all assays. Examples of colony formations are shown in the U251 cell line.

Figure 5.

Induction of apoptosis and reduction of tumorigenicity of established and PDGx glioma cell lines after treatment with lead compounds. A, Graphs illustrate the percent reduction of Bcl2/18S and Mcl1/18S mRNAs following cell treatment with lead compounds versus control (vehicle treatment). Results are shown as mean ± SD (n = 3), differences are statistically significant for all cell lines with P < 0.005, Student t test. B, Representative Western blots confirm the reduction of Bcl2 and Mcl1 proteins in cytoplasmic fractions of U251 and XD456 cell lines following treatment with lead compounds, 10 μmol/L, for 48 hours. SOX2 reduction in nuclear fractions indicates a decrease in tumorigenicity of cell lines after treatment with lead compounds. C, Representative Western blots illustrate appearances of cleaved PARP and cleaved caspase-3 in glioma cell lines after treatment with lead compounds, 10 μmol/L, for 48 hours. D, Graphs illustrate a significant reduction of colony formations in U251 and XD456 cell lines after SRI-42127 (3 and 10 μmol/L treatments) versus control (vehicle treatment). Results are shown as mean ± SD, n = 3 and 4) for attached and anchorage-independent assays, respectively; *, P < 0.005, Student t test for all assays. Examples of colony formations are shown in the U251 cell line.

Close modal

Note that short-term treatment with SRI-42127 (3 and 10 μmol/L, 8–12 hours of treatment), utilized in the HuR dimerization reporter assays, did not alter the total HuR protein levels (Supplementary Fig. S11A). However, we observed a decrease in HuR cytoplasmic levels after prolonged treatment with A-92, SRI-41664, and SRI-42127 compounds versus control and an accumulation of nuclear HuR with the following HuR proteolysis and disintegration (Fig. 5B; Supplementary Fig. S10B). The experiments in the presence of cycloheximide (an inhibitor of protein synthesis) alone or in combination with SRI-42127 (3 and 10 μmol/L) suggested that that the disruption of the HuR dimerization by SRI-42127 might facilitate HuR degradation (Supplementary Fig. S11B). Also, the activation of cleaved caspase-3 (Fig. 5C) associated with the prolonged treatment with lead compounds might contribute to HuR proteolysis and disintegration in agreement with previous reports (29, 30).

Colony formation assays (attached and anchorage-independent) were performed in U251 and XD456 glioma cells and showed a significant reduction in glioma cells' ability to form colonies after treatment with SRI-42127 (3 and 10 μmol/L) for 2 weeks compared with vehicle treatment (Fig. 5D). The number of colonies significantly decreased from 337 ± 31 (n = 3) to 49 ± 28 (n = 3) and from 402 ± 38 (n = 3) to 34 ± 12 (n = 3) in U251 and XD456 cell lines, respectively, in the attached cell colony formation assay with SRI-42127, 3 μmol/L. The number of colonies significantly decreased from 436 ± 28 (n = 4) to 90 ± 70 (n = 4) and from 283 ± 62 (n = 4) to 39 ± 6 (n = 4) in U251 and XD456 cell lines, respectively, in the soft agar colony formation assay with SRI-42127, 3 μmol/L. No significant numbers of colonies were detected in the presence of SRI-42127, 10 μmol/L, in U251 and XD456 cell lines in both assays.

Hence, our data confirmed the impairment of the in vitro growth and reduction of tumorigenicity of both established and patient-derived glioma cell lines following treatment with the new class of HuR dimerization inhibitors.

Analysis of the cell signaling pathways altered in patient-derived glioma neurospheres by SRI-42127

The Illumina global RNA sequencing was performed on PDGx-derived glioma neurospheres representing the different molecular subtypes (classical, neuronal, and mesenchymal) after treatment with vehicle (control) or SRI-42127 (3 μmol/L) for 12 hours (Supplementary Data File S2). First, analysis of direct HuR mRNA targets and their corresponding cell signaling pathways was performed using previously published gene sets of direct HuR targets (11). Supplementary Table S1 illustrates the evaluated gene sets, representing known direct HuR mRNA targets, and their corresponding cell signaling pathways. The gene set average values, which exhibited significant alterations following treatment with the SRI-42127 compound as compared with the control, are highlighted in green. Note that the gene sets corresponding to cell-cycle progression, MAPK signaling pathways, TP53-dependent transcriptional pathways, stress-response cell signaling pathways, RHO GTPase–dependent pathways, membrane trafficking pathways, cellular senescence and developmental pathways, and cell signaling pathways associated with the Epstein–Barr virus infection are significantly downregulated in at least four of five evaluated neurosphere cell lines.

Next, we utilized the Enrichr software for overall gene ontology enrichment for gene sets significantly altered in PDGx neurospheres by SRI-42127 compound compared with the control (Fig. 6). Gene sets that exhibited significant upregulation or downregulation of more than 50% across all the evaluated cell lines after treatment with SRI-42127 were selected for the enrichment analysis. Figure 6A illustrates the enrichment for downregulated cell signaling pathways and targeted subcellular structures. The detailed cluster plots of the enriched downregulated genes and corresponding cell signaling pathways and targeted subcellular structures are shown in Fig. 6A and Supplementary Data File S7. The networks of downregulated transcriptional factors and associated pathways are shown in Supplementary Fig. S12. Significantly downregulated pathways (associated with glioma progression) included (i) cell-cycle progression, (ii) DNA replication, (iii) transport mature transcript to the cytoplasm, (iv) DNA strand elongation, (v) capped intron-containing pre-mRNA processing, (vi) DNA repair, (vii) the unwinding of DNA, (viii) alpha-linolenic (omega 3) and linoleic (omega 6) metabolism, (ix) chromosome maintenance, and (x) messenger RNA splicing. We and others have previously reported that HuR may control the centrosome and microtubule-organizing center in cancer cells (24, 31, 32). Therefore, without surprise, the enrichment analysis confirmed that the centrosome, mitotic spindle, pericentriolar materials, microtubule-organizing center, and telomeric region of chromosome are among the subcellular structures affected by SRI-42127 treatment.

Figure 6.

Gene ontology enrichment analysis for gene sets significantly altered in PDGx neurospheres by SRI-42127, 3 μmol/L, 12 hours of treatment compared with control (vehicle treatment). A, Enrichment of the significantly downregulated cell signaling pathways/processes (chart on the left) and targeted subcellular structures (chart on the right) for significantly downregulated gene sets after SRI-42127 treatment compared with control (vehicle treatment). B, Enrichment of the significantly upregulated cell signaling pathways/processes (chart on the left) and targeted subcellular structures (chart on the right) for significantly upregulated gene sets after SRI-42127 treatment compared with control (vehicle treatment). Genes that exhibited the significant upregulation or downregulation on more than 50% across all of the evaluated PDGx neurosphere cell lines (JX12p, X14P, XD456, X1524, and JX22p) after treatment with SRI-42127 versus control were selected for the enrichment analysis; statistical significance was determined by Student t test. The bar graphs illustrating the relative combined scores are attached to each chart on the right side. The results are generated based on RNA sequencing data.

Figure 6.

Gene ontology enrichment analysis for gene sets significantly altered in PDGx neurospheres by SRI-42127, 3 μmol/L, 12 hours of treatment compared with control (vehicle treatment). A, Enrichment of the significantly downregulated cell signaling pathways/processes (chart on the left) and targeted subcellular structures (chart on the right) for significantly downregulated gene sets after SRI-42127 treatment compared with control (vehicle treatment). B, Enrichment of the significantly upregulated cell signaling pathways/processes (chart on the left) and targeted subcellular structures (chart on the right) for significantly upregulated gene sets after SRI-42127 treatment compared with control (vehicle treatment). Genes that exhibited the significant upregulation or downregulation on more than 50% across all of the evaluated PDGx neurosphere cell lines (JX12p, X14P, XD456, X1524, and JX22p) after treatment with SRI-42127 versus control were selected for the enrichment analysis; statistical significance was determined by Student t test. The bar graphs illustrating the relative combined scores are attached to each chart on the right side. The results are generated based on RNA sequencing data.

Close modal

Figure 6B illustrates the enrichment of the significantly upregulated cell signaling pathways and targeted subcellular structures. The detailed cluster plots of the enriched upregulated genes and corresponding cell signaling pathways and targeted subcellular structures are shown in Fig. 6B and Supplementary Data File S7. There was significant upregulation of the cellular pathways responsible for (i) SRP-dependent cotranslational protein targeting to the membrane and ER, (ii) ribosomal biogenesis, (iii) ncRNA processing, and (iv) nuclear-transcribed mRNA processing, and decay components of the ribosome and polysome were the main targeted subcellular structures. The ribosomal reorganization is the common adaptive response to translational and environmental stress (33); therefore, the ribosome-reorganizing pathways are considered the cell's main compensatory response pathways. Also, importantly, the data show an upregulation of the Tubb2A transcript by more than 2-fold in four of five neurosphere cell lines (Supplementary Data File S2). Because TUBB2a is associated with neuronal protection, this pattern is consistent with the absence of toxicity in neurons treated with SRI42127 (Fig. 4; ref. 34).

Next, quantitative proteomics data were generated using glioma neurospheres of different molecular subtypes (classical, neuronal, and mesenchymal) in control (treatment with vehicles) and after treatment with SRI-42127, 3 μmol/L, for 18 hours (Supplementary Data Files S3 and S4). The enrichment of proteins, which exhibited a significant downregulation by more than 50% after cell treatment with the SRI-42127, revealed that all neurosphere cell lines uniformly had downregulation of RNA-binding function and sign of apoptosis (Fig. 6; Supplementary Fig. S13A and S13B). In addition, we found a significant downregulation of: (i) protein sets responsible for protein transporter activity, small protein activating enzyme activity, cadherin binding, double-stranded DNA binding, DNA dependent ATPase activity, nucleoside-triphosphatase activity, and ADP binding in the neurospheres of the classical subtype; (ii) the protein sets involved in transcriptional factor activity, transcription initiation factor activity, and nuclear localization sequence binding in the neurospheres of the proneuronal subtype (Supplementary Fig. S13A and Supplementary Data File S4).

Thus, the data analysis suggested that cell signaling pathways, which require HuR direct mRNA targets, were mostly downregulated in neurospheres after treatment with the SRI-42127 compound. However, some gene set modifications were specific to the glioblastoma molecular subtypes.

SRI-42127 crosses the blood–brain barrier and suppresses glioma growth

The PK evaluation of SRI-42127 in a mouse model revealed a significant penetration into the CNS with a brain-to-plasma concentration ratio of 0.42 ± 0.05 (n = 9; Supplementary Fig. S14A–S14C; and Supplementary Data File S6). The half-life of compound SRI-42127 was short 0.16 hour using a dose of 10 mg/kg by IP. Next, we evaluated the impact of SRI-42127 on intracranial glioma growth in a mouse model. Because the half-life of SRI-42127 was relatively short, we used twice a day dosing at 15 mg/kg, administered for 3 weeks (see Materials and Methods). There was a reduction of tumor progression in the mouse group treated with SRI-42127 compared with the control group (treated with vehicles) based on the detection of tumor reporter luminescence signal (Fig. 7A). Also, a reduction in tumor mass was confirmed in brain slices from the mouse group treated with compound SRI-42127 compared with the control group based on the detection of tumor reporter EGFP signal. Immunostaining for HuR, Bcl2, and MCl1 confirmed a reduction of HuR, Bcl2, and Mcl1 expression in the tumors from the mouse group treated with SRI-42127 versus the control group (Fig. 7B,D).

Figure 7.

Glioma growth inhibition by SRI-42127, in vivo, and computational modeling of SRI-42127 binding to HuR. A–D, SRI-42127 suppresses glioma progression in vivo. Mice were randomly divided into two groups after four and half days of the intracranial tumor establishment and treated with vehicle (control group) or SRI-42127 compound (SRI-42127 group), 15 mg/kg, twice a day for 3 weeks via i.p. injection. Representative scans illustrate the tumor reporter images superimposed with mouse images; luminescence/color scales are beside the corresponding scans. Graphs illustrate luminescence signals from intracranial tumors on the second (top) and third (bottom) weeks of treatment; results are shown as mean ± SD (radiance, p/sec/cm2/sr); the differences between groups were not statistically significant, P = 0.07 (n = 5) for the second week, P = 0.069 (n = 5) for the third week, Student t test. The weight changes in group treated with SRI-42127 compound were not significant compared with the control group (P > 0.05, Student t test). Immunostaining for HuR (B), Bcl2 (C), and MCl1 (D) on tumor brain tissue from control and treated with SRI-42127 compound mouse groups. E, Computational docking of SRI-42127 at HuR. SRI-42127 compound and key residues of HuR are represented with green and purple carbons, respectively. Hydrogen bonds, hydrophobic contacts, and pi–pi stacking (or pi–cation interactions) are indicated by cyan, orange, and green dash lines, respectively. Gray and blue ribbons represent RRM1 and RRM2 domains, respectively.

Figure 7.

Glioma growth inhibition by SRI-42127, in vivo, and computational modeling of SRI-42127 binding to HuR. A–D, SRI-42127 suppresses glioma progression in vivo. Mice were randomly divided into two groups after four and half days of the intracranial tumor establishment and treated with vehicle (control group) or SRI-42127 compound (SRI-42127 group), 15 mg/kg, twice a day for 3 weeks via i.p. injection. Representative scans illustrate the tumor reporter images superimposed with mouse images; luminescence/color scales are beside the corresponding scans. Graphs illustrate luminescence signals from intracranial tumors on the second (top) and third (bottom) weeks of treatment; results are shown as mean ± SD (radiance, p/sec/cm2/sr); the differences between groups were not statistically significant, P = 0.07 (n = 5) for the second week, P = 0.069 (n = 5) for the third week, Student t test. The weight changes in group treated with SRI-42127 compound were not significant compared with the control group (P > 0.05, Student t test). Immunostaining for HuR (B), Bcl2 (C), and MCl1 (D) on tumor brain tissue from control and treated with SRI-42127 compound mouse groups. E, Computational docking of SRI-42127 at HuR. SRI-42127 compound and key residues of HuR are represented with green and purple carbons, respectively. Hydrogen bonds, hydrophobic contacts, and pi–pi stacking (or pi–cation interactions) are indicated by cyan, orange, and green dash lines, respectively. Gray and blue ribbons represent RRM1 and RRM2 domains, respectively.

Close modal

Hence, we conclude that the new class of inhibitors of HuR dimerization (particularly compound SRI-42127) exhibited encouraging, however not yet statistically significant, glioma-inhibitory potential in vivo.

Putative binding model of SRI-42127 at HuR

A computational docking study was performed to investigate the binding mode of lead SRI-42127 compound at HuR (details are included in Supplementary Data File S1). The apo structure of HuR was prepared from a crystal structure of RNA/HuR complex (PDB ID: 4ED5) using the Protein Preparation Wizard (23). Figure 7E illustrates the docked pose of SRI-42127 at HuR. According to the putative binding mode, SRI-42127 is able to interact with four functionally important residues Y26, I103, R97, and R153 of HuR. Molecular mechanics-generalized Born surface area (MM-GBSA), an in silico binding-free energy scoring approach, was used to estimate the binding affinity of SRI-42127 at HuR. Specifically, MM-GBSA scoring was performed on the fixed conformation of the docked pose of SRI-42127 at HuR using VSGB solvation model and OPLS3e force field. The same scoring was performed on the UMP (U8 from HuR cocrystal structure), which resides in the same region as SRI-42127 to obtain a reference value. SRI-42127 showed a stronger MM-GBSA binding-free energy score than the UMP (-57.6 kcal/mol vs. -22.7 kcal/mol), indicating that SRI-42127 is indeed capable of forming energetically favorable interactions at the RM1–RM2 putative binding site on HuR.

Therefore, one of the possible modes of action could be that the binding of SRI-42127 to RM1–RM2 induces a conformational change of HuR monomer that disrupts the optimal interface needed for HuR dimerization and mRNA binding.

Glioma is one of the deadly tumors of the CNS and accounts for about 80% of all malignant brain tumors (35). Standard therapy with surgery, radiotherapy, and chemotherapy with TMZ represent the first line of treatment for glioblastoma, the most malignant, over the past decade (36, 37). However, due to high tumor heterogeneity, genomic instability, their diffusely infiltrating nature, and the development of drug resistance, glioma remains incurable and basically poorly responsive to treatment (37, 38).

HuR plays a crucial role in glioma development and transition to the higher grade (1, 18, 21, 32, 38–40). HuR's role is unique for each tumor grade and depends on integrated transcriptome and proteome, including cellular kinome, which is involved in the HuR posttranslational modifications and is grade-specific (41, 42). Glioma HuR dependence is progressive with tumor grade. HuR overexpression in low-grade glioma correlated with an overall decrease of transcriptional factor activity, downregulation of cell signaling pathways responsible for DNA repair, and, therefore, an accumulation of genomic instability. The overexpression of HuR in higher-grade tumors results in unleashed cell-cycle progression, chemoresistance, immune resistance, aberrant cell signal transduction, suppression of differentiation, and apoptosis. In our work, we evaluated the impact of a new class of HuR inhibitors on the progression of high-grade primary and established glioma cell lines. Uniformly, we observed an arrest of cell-cycle progression, the appearance of apoptosis, a reduction of tumor stemness, and a decrease of colony formation after glioma cell treatment. First, enrichment of cell signaling pathways indicated that centrosomes and cell division apparatuses were significantly affected. These data are in agreement with prior reports from our laboratory and elsewhere showing that HuR colocalizes with centrosomes in a phosphorylation-dependent manner, controls protein synthesis in the near proximity to centrosomes, affects the cytoskeleton architecture of the mitotic spindles, and influences the centrosome-dependent aspects of tumor cell division (24, 31, 32, 43, 44). Second, there was a significant decrease in the expression of HuR-dependent mRNAs of cyclins, which are essential for cell-cycle progression (12, 27, 32, 45). Third, there was a marked downregulation of Bcl2 family members (Fig. 5) consistent with our prior reports with HuR knockdown or inhibition (18, 40). This family of proteins plays a critical role in the prevention of apoptosis in tumor cells (18, 28, 45), and the appearance of apoptotic markers (cleaved PARP and cleaved caspase-3) after SRI-42127 treatment (Supplementary Fig. S13B) is consistent with their suppression. Taken together, the molecular and cellular effects of SRI-42127 on glioma cells are consistent with HuR-targeted inhibition.

Recently, it has been reported that a correlation exists between a patient's gender and response to glioblastoma therapy as well as overall survival [specifically for glioblastoma multiforme (GBM) of mesenchymal, proneural, and neural subtypes; ref. 46]. Standard therapy is more effective in female patients compared with male, which exhibited upregulation of the HuR-dependent transcriptome essential for cell-cycle progression. In this regard, HuR inhibitors may serve as an adjuvant treatment modality for male patients with upregulated cell-cycle progression. The female group, which was associated with the activation of the integrin signaling components following standard therapy, might also benefit from the HuR inhibitors due to the fact that several members of the integrin signaling components (to include ITGB4, ITGA4, ITGB1, ITGA3, ICAM1, VCAM1, and VIM) and chemokine family (to include CCL2, CCL8, CCL13, CXCL1, CXCL2, CXCL3, CXCL5, and CCL20) are direct HuR mRNA targets. Therefore, sex differences in GBM biology may be useful to predict suppression of tumor invasion and angiogenesis in the female group and suppression of tumor proliferation in the male group by the HuR inhibitors. A detailed transcriptome analysis of the therapeutics targets combined with the profiling of patient transcriptome is essential for the development of personalized medicine. Our work profiled the transcriptomics and proteomic alterations of the different glioma subtypes after treatment with the new class of HuR inhibitors; we plan to extend our analysis using in vivo mouse glioma models to further understand sex-, age-, and subtype-specific benefits of HuR inhibitors.

Overall, the most unfavorable HuR-dependent transformations for patients harboring gliomas are (i) cell-cycle progression and (ii) development of the tumor heterogeneity with the consequent drug resistance (3, 18, 24, 25, 27, 40, 47, 48). HuR-dependent cell-cycle progression is well documented (10, 15, 23, 27, 40); the mechanisms of drug resistance are still under investigation and consist of the HuR-dependent activation of the ATP-binding cassette drug transporters, the antiapoptotic pathways related to the mitochondria function, and the HuR-dependent intercellular gene transfer (13, 18, 28, 47). Similar to gliomas, colon cancer cell viability and mitochondria function exhibited strong dependence from the antiapoptotic Bcl2-family (7), and transcripts of Bcl2 and Mcl1 are established HuR/mRNA targets. Depolarization of mitochondria membrane potential with an induction of apoptosis has been observed in colon cancer cells after treatment with the HuR inhibitor, CMLD-2 (7). Therefore, we predict that colon cancer (as well as breast, lung, bladder, oral, ovarian, and prostate cancers) will likely be sensitive to HuR inhibitors; however, the detailed benefits of treatment and the most patient-sensitive group remain to be determined.

Twenty-six compounds, which exhibited a HuR-inhibitory potential with IC50s less than 10 μmol/L in the cell-based HuR dimerization assay, have been synthesized and evaluated in vitro. Six recently developed compounds have IC50s below 1 μmol/L. With new HuR dimerization inhibitors, we confirmed the disruption of HuR-related oncogenic characteristics in glioma cell lines similar to disruptions induced by MS444 and DHTS, established inhibitors of HuR (6, 10, 11, 21, 47). All inhibitors of HuR dimerization (like A-92, SRI-41664, SRI-42127, MS444, and DHTS) were cytotoxic for glioma cell lines, and the reductions of cell viability were associated with the appearance of apoptosis and a reduction in the Bcl2-family at the mRNA and protein levels (10, 47). The arrest of cell-cycle progression and uniform decrease in cell transition to the S phase of the cell cycle were observed for all the above HuR dimerization inhibitors. A reduction in cancer cell line stemness following treatment with MS444, DHTS, and SRI-42127 compounds was confirmed in our experiments and was reported by others as well (10, 47). The comparison of RIP-ChIP profiling of HeLa cells treated with DHTS (11) and glioma cells treated with SRI-42127 revealed common alterations in HuR-dependent gene sets and cell signaling pathways. However, MS444 and DHTS compounds have unknown blood–brain barrier (BBB) permeability, require a high concentration for activity, and are challenging to create soluble solutions for administration. SRI-42127 compound exhibited superior BBB permeability, a stronger potency for inhibition of HuR dimerization compared to MS444, and is readily soluble. The in vivo compound assessment, in conjunction with future lead compound optimization, will focus on the improvement in the PK characteristics as our next steps. In conclusion, our work has identified a new class of HuR multimerization inhibitors and detailed the lead compound–inhibitory potentials on glioma progression in vitro and in vivo.

N. Filippova and X. Yang report grants from NIH during the conduct of the study; in addition, they have a patent for U.S. Provisional Patent Application No. 63/013,451 issued. S. Ananthan reports grants from NIH during the conduct of the study; in addition, S. Ananthan has a patent for U.S. Provisional Patent Application No. 63/013,451 pending. V. Pathak reports a patent for 63/013,451 pending. L.B. Nabors reports grants from NIH during the conduct of the study and personal fees from Karyopharm and BTG Pharma outside the submitted work; in addition, L.B. Nabors has a patent for U.S. Provisional Patent Application No. 63/013,451 issued and serves on a Data Safety and Monitoring Board for the University of Pennsylvania glioma clinical trial. No disclosures were reported by the other authors.

N. Filippova: Conceptualization, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. X. Yang: Data curation, investigation, methodology. S. Ananthan: Data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing. J. Calano: Investigation. V. Pathak: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft. L. Bratton: Conceptualization, formal analysis, validation, investigation, methodology. R.H. Vekariya: Formal analysis, investigation, visualization, methodology. S. Zhang: Data curation, formal analysis, validation, visualization, methodology. E. Ofori: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. E.N. Hayward: Data curation, investigation, methodology. D. Namkoong: Data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing. D.K. Crossman: Software, formal analysis, investigation, methodology. M.R. Crowley: Conceptualization, resources, formal analysis, validation, investigation, methodology, writing–original draft. P.H. King: Conceptualization, resources, formal analysis, supervision, validation, investigation, methodology, writing–review and editing. J. Mobley: Resources, formal analysis, investigation, visualization, methodology. L.B. Nabors: Data curation, formal analysis, validation, visualization, methodology.

The work was funded by the NIH Grant R01 CA200624 and the University of Alabama at Birmingham O'Neal Comprehensive Cancer Center Neuro-oncology Research Acceleration Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Nabors
LB
,
Gillespie
GY
,
Harkins
L
,
King
PH
. 
HuR, a RNA stability factor, is expressed in malignant brain tumors and binds to adenine- and uridine-rich elements within the 3′ untranslated regions of cytokine and angiogenic factor mRNAs
.
Cancer Res
2001
;
61
:
2154
61
.
2.
Dixon
DA
,
Tolley
ND
,
King
PH
,
Nabors
LB
,
Mcintyre
TM
,
Zimmerman
GA
, et al
Altered expression of the mRNA stability factor HuR promotes cyclooxygenase-2 expression in colon cancer cells
.
J Clin Invest
2001
;
108
:
1657
65
.
3.
Schultz
CW
,
Preet
R
,
Dhir
T
,
Dixon
DA
,
Brody
JR
. 
Understanding and targeting the disease-related RNA binding protein human antigen R (HuR)
.
Wiley Interdiscip Rev RNA
2020
;
11
;
e1581
.
4.
Wang
J
,
Guo
Y
,
Chu
H
,
Guan
Y
,
Bi
J
,
Wang
B
. 
Multiple functions of the RNA-binding protein HuR in cancer progression, treatment responses and prognosis
.
Int J Mol Sci
2013
;
14
:
10015
41
.
5.
Abdelmohsen
K
,
Gorospe
M
. 
Posttranscriptional regulation of cancer traits by HuR
.
Wiley Interdiscip Rev RNA
2010
;
1
:
214
29
.
6.
Meisner
N-C
,
Hintersteiner
M
,
Mueller
K
,
Bauer
R
,
Seifert
J-M
,
Naegeli
H-U
, et al
Identification and mechanistic characterization of low-molecular-weight inhibitors for HuR
.
Nat Chem Biol
2007
;
3
:
508
15
.
7.
Muralidharan
R
,
Mehta
M
,
Ahmed
R
,
Roy
S
,
Xu
L
,
Aubé
J
, et al
HuR-targeted small molecule inhibitor exhibits cytotoxicity towards human lung cancer cells
.
Sci Rep
2017
;
7
:
9694
.
8.
Chae
M-J
,
Sung
HY
,
Kim
E-H
,
Lee
M
,
Kwak
H
,
Chae
CH
, et al
Chemical inhibitors destabilize HuR binding to the AU-rich element of TNF-α mRNA
.
Exp Mol Med
2009
;
41
:
824
.
9.
Kaur
K
,
Wu
X
,
Fields
JK
,
Johnson
DK
,
Lan
L
,
Pratt
M
, et al
The fungal natural product azaphilone-9 binds to HuR and inhibits HuR-RNA interaction in vitro
.
PLoS One
2017
;
12
:
e0175471
.
10.
D'Agostino
VG
,
Lal
P
,
Mantelli
B
,
Tiedje
C
,
Zucal
C
,
Thongon
N
, et al
Dihydrotanshinone-I interferes with the RNA-binding activity of HuR affecting its post-transcriptional function
.
Sci Rep
2015
;
5
:
16478
.
11.
Lal
P
,
Cerofolini
L
,
D'Agostino
VG
,
Zucal
C
,
Fuccio
C
,
Bonomo
I
, et al
Regulation of HuR structure and function by dihydrotanshinone-I
.
Nucleic Acids Res
2017
;
45
:
9514
27
.
12.
Kakuguchi
W
,
Nomura
T
,
Kitamura
T
,
Otsuguro
S
,
Matsushita
K
,
Sakaitani
M
, et al
Suramin, screened from an approved drug library, inhibits HuR functions and attenuates malignant phenotype of oral cancer cells
.
Cancer Med
2018
;
7
:
6269
80
.
13.
Hostetter
C
,
Licata
LA
,
Costantino
CL
,
Witkiewicz
A
,
Yeo
C
,
Brody
JR
, et al
Cytoplasmic accumulation of the RNA binding protein HuR is central to tamoxifen resistance in estrogen receptor positive breast cancer cells
.
Cancer Biol Ther
2008
;
7
:
1496
506
.
14.
Blanco
FF
,
Preet
R
,
Aguado
A
,
Vishwakarma
V
,
Stevens
LE
,
Vyas
A
, et al
Impact of HuR inhibition by the small molecule MS-444 on colorectal cancer cell tumorigenesis
.
Oncotarget
2016
;
7
:
74043
58
.
15.
Liu
Y
,
Chen
X
,
Cheng
R
,
Yang
F
,
Yu
M
,
Wang
C
, et al
The Jun/miR-22/HuR regulatory axis contributes to tumourigenesis in colorectal cancer
.
Mol Cancer
2018
;
17
:
11
.
16.
Muralidharan
R
,
Babu
A
,
Amreddy
N
,
Srivastava
A
,
Chen
A
,
Zhao
YD
, et al
Tumor-targeted nanoparticle delivery of HuR siRNA inhibits lung tumor growth in vitro and in vivo by disrupting the oncogenic activity of the RNA-binding protein HuR
.
Mol Cancer Ther
2017
;
16
:
1470
86
.
17.
Moradi
F
,
Berglund
P
,
Linnskog
R
,
Leandersson
K
,
Andersson
T
,
Prasad
CP
. 
Dual mechanisms of action of the RNA-binding protein human antigen R explains its regulatory effect on melanoma cell migration
.
Transl Res
2016
;
172
:
45
60
.
18.
Filippova
N
,
Yang
X
,
Wang
Y
,
Gillespie
GY
,
Langford
C
,
King
PH
, et al
The RNA-binding protein HuR promotes glioma growth and treatment resistance
.
Mol Cancer Res
2011
;
9
:
648
59
.
19.
Brody
JR
,
Dixon
DA
. 
Complex HuR function in pancreatic cancer cells
.
Wiley Interdiscip Rev RNA
2018
;
9
:
e1469
.
20.
Levidou
G
,
Kotta-Loizou
I
,
Tasoulas
J
,
Papadopoulos
T
,
Theocharis
S
. 
Clinical significance and biological role of HuR in head and neck carcinomas
.
Dis Markers
2018
;
2018
:
4020937
.
21.
Filippova
N
,
Yang
X
,
Ananthan
S
,
Sorochinsky
A
,
Hackney
JR
,
Gentry
Z
, et al
Hu antigen R (HuR) multimerization contributes to glioma disease progression
.
J Biol Chem
2017
;
292
:
16999
7010
.
22.
Filippova
N
,
Yang
X
,
An
Z
,
Nabors
LB
,
Pereboeva
L
. 
Blocking PD1/PDL1 interactions together with MLN4924 therapy is a potential strategy for glioma treatment
.
J Cancer Sci Ther
2018
;
10
:
190
7
.
23.
Sherman
W
,
Day
T
,
Jacobson
MP
,
Friesner
RA
,
Farid
R
. 
Novel procedure for modeling ligand/receptor induced fit effects
.
J Med Chem
2006
;
49
:
534
53
.
24.
Filippova
N
,
Yang
X
,
King
P
,
Nabors
LB
. 
Phosphoregulation of the RNA-binding protein Hu antigen R (HuR) by Cdk5 affects centrosome function
.
J Biol Chem
2012
;
287
:
32277
87
.
25.
Holmes
B
,
Benavides-Serrato
A
,
Freeman
RS
,
Landon
KA
,
Bashir
T
,
Nishimura
RN
, et al
mTORC2/AKT/HSF1/HuR constitute a feed-forward loop regulating Rictor expression and tumor growth in glioblastoma
.
Oncogene
2017
;
37
:
732
43
.
26.
Gurgis
FMS
,
Yeung
YT
,
Tang
MXM
,
Heng
B
,
Buckland
M
,
Ammit
AJ
, et al
The p38-MK2-HuR pathway potentiates EGFRvIII–IL-1β-driven IL-6 secretion in glioblastoma cells
.
Oncogene
2014
;
34
:
2934
42
.
27.
Wang
W
,
Caldwell
M
,
Lin
S
,
Furneaux
H
,
Gorospe
M
. 
HuR regulates cyclin A and cyclin B1 mRNA stability during cell proliferation
.
EMBO J
2000
;
19
:
2340
50
.
28.
Mcdonald
FE
,
Ironside
JW
,
Gregor
A
,
Wyatt
B
,
Stewart
M
,
Rye
R
, et al
The prognostic influence of bcl-2 in malignant glioma
.
Br J Cancer
2002
;
86
:
1899
904
.
29.
Janakiraman
H
,
House
RP
,
Talwar
S
,
Courtney
SM
,
Hazard
ES
,
Hardiman
G
, et al
Repression of caspase-3 and RNA-binding protein HuR cleavage by cyclooxygenase-2 promotes drug resistance in oral squamous cell carcinoma
.
Oncogene
2016
;
36
:
3137
48
.
30.
Mazroui
R
,
Marco
SD
,
Clair
E
,
Roretz
CV
,
Tenenbaum
SA
,
Keene
JD
, et al
Caspase-mediated cleavage of HuR in the cytoplasm contributes to pp32/PHAP-I regulation of apoptosis
.
J Cell Biol
2008
;
180
:
113
27
.
31.
Filippova
N
,
Yang
X
,
Nabors
L
. 
Growth factor dependent regulation of centrosome function and genomic instability by HuR
.
Biomolecules
2015
;
5
:
263
81
.
32.
Mukherjee
J
,
Ohba
S
,
See
WL
,
Phillips
JJ
,
Molinaro
AM
,
Pieper
RO
. 
PKM2 uses control of HuR localization to regulate p27 and cell cycle progression in human glioblastoma cells
.
Int J Cancer
2016
;
139
:
99
111
.
33.
Rendleman
J
,
Cheng
Z
,
Maity
S
,
Kastelic
N
,
Munschauer
M
,
Allgoewer
K
, et al
New insights into the cellular temporal response to proteostatic stress
.
Elife
2018
;
7
:
e39054
.
34.
Wheeler
HE
,
Wing
C
,
Delaney
SM
,
Komatsu
M
,
Dolan
ME
. 
Modeling chemotherapeutic neurotoxicity with human induced pluripotent stem cell-derived neuronal cells
.
PLoS One
2015
;
10
:
e0118020
.
35.
Louis
DN
,
Perry
A
,
Reifenberger
G
,
Deimling
AV
,
Figarella-Branger
D
,
Cavenee
WK
, et al
The 2016 World Health Organization classification of tumors of the central nervous system: a summary
.
Acta Neuropathol
2016
;
131
:
803
20
.
36.
Perry
JR
,
Laperriere
N
,
Mason
W
. 
Radiation plus temozolomide in patients with glioblastoma
.
N Engl J Med
2017
;
376
:
2195
7
.
37.
Tamimi
AF
,
Juweid
M
. 
Epidemiology and outcome of glioblastoma
.
Glioblastoma
2017
;:
143
53
.
38.
Parker
NR
,
Khong
P
,
Parkinson
JF
,
Howell
VM
,
Wheeler
HR
. 
Molecular heterogeneity in glioblastoma: potential clinical implications
.
Front Oncol
2015
;
5
:
55
.
39.
Wang
J
,
Leavenworth
JW
,
Hjelmeland
AB
,
Smith
R
,
Patel
N
,
Borg
B
, et al
Deletion of the RNA regulator HuR in tumor-associated microglia and macrophages stimulates anti-tumor immunity and attenuates glioma growth
.
Glia
2019
;
67
:
2424
39
.
40.
Bolognani
F
,
Gallani
A-I
,
Sokol
L
,
Baskin
DS
,
Meisner-Kober
N
. 
mRNA stability alterations mediated by HuR are necessary to sustain the fast growth of glioma cells
.
J Neurooncol
2011
;
106
:
531
42
.
41.
Eberhardt
W
,
Doller
A
,
Pfeilschifter
J
. 
Regulation of the mRNA-binding protein HuR by posttranslational modification: spotlight on phosphorylation
.
Curr Protein Pept Sci
2012
;
13
:
380
90
.
42.
Grammatikakis
I
,
Abdelmohsen
K
,
Gorospe
M
. 
Posttranslational control of HuR function
.
Wiley Interdiscip Rev RNA
2016
;
8
.
43.
Lei
W
,
Wang
ZL
,
Feng
HJ
,
Lin
XD
,
Li
CZ
,
Fan
D
. 
Long non-coding RNA SNHG12promotes the proliferation and migration of glioma cells by binding to HuR
.
Int J Oncol
2018
;
53
:
1374
84
.
44.
Hitti
E
,
Bakheet
T
,
Al-Souhibani
N
,
Moghrabi
W
,
Al-Yahya
S
,
Al-Ghamdi
M
, et al
Systematic analysis of AU-rich element expression in cancer reveals common functional clusters regulated by key RNA-binding proteins
.
Cancer Res
2016
;
76
:
4068
80
.
45.
Muralidharan
R
,
Babu
A
,
Amreddy
N
,
Basalingappa
K
,
Mehta
M
,
Chen
A
, et al
Folate receptor-targeted nanoparticle delivery of HuR-RNAi suppresses lung cancer cell proliferation and migration
.
J Nanobiotechnol
2016
;
14
:
47
.
46.
Yang
W
,
Warrington
NM
,
Taylor
SJ
,
Whitmire
P
,
Carrasco
E
,
Singleton
KW
, et al
Sex differences in GBM revealed by analysis of patient imaging, transcriptome, and survival data
.
Sci Transl Med
2019
;
11
:
eaao5253
.
47.
Wang
J
,
Hjelmeland
AB
,
Nabors
LB
,
King
PH
. 
Anti-cancer effects of the HuR inhibitor, MS-444, in malignant glioma cells
.
Cancer Biol Ther
2019
;
20
:
979
88
.
48.
Filippova
N
,
Nabors
LB
. 
ELAVL1 role in cell fusion and tunneling membrane nanotube formations with implication to treat glioma heterogeneity
.
Cancers
2020
;
12
:
3069
.

Supplementary data