Glioblastoma stem cell (GSC) is the major cause of glioblastoma multiforme (GBM) chemotherapy failure. Hypoxia is one of the determinants of GSC. NF-κB plays a pivotal link between hypoxia and cancer stem cells (CSCs). Disulfiram, an antialcoholism drug, has very strong NF-κB–inhibiting and anti-CSC activity. In this study, the in vitro anti-GSC activity of disulfiram and in vivo anti-GBM efficacy of poly lactic–co-glycolic acid nanoparticle-encapsulated disulfiram (DS-PLGA) were examined. We attempt to elucidate the molecular network between hypoxia and GSCs and also examined the anti-GSC activity of disulfiram in vitro and in vivo. The influence of GSCs and hypoxia on GBM chemoresistance and invasiveness was studied in hypoxic and spheroid cultures. The molecular regulatory roles of NF-κB, hypoxia-inducible factor-1α (HIF1α), and HIF2α were investigated using stably transfected U373MG cell lines. The hypoxia in neurospheres determines the cancer stem cell characteristics of the sphere-cultured GBM cell lines (U87MG, U251MG, U373MG). NF-κB is located at a higher hierarchical position than HIF1α/HIF2α in hypoxic regulatory network and plays a key role in hypoxia-induced GSC characters. DS inhibits NF-κB activity and targets hypoxia-induced GSCs. It showed selective toxicity to GBM cells, eradicates GSCs, and blocks migration and invasion at very low concentrations. DS-PLGA efficaciously inhibits orthotopic and subcutaneous U87MG xenograft in mouse models with no toxicity to vital organs.
Glioblastoma multiforme (GBM) is one of the most lethal and aggressive forms of the malignant adult brain tumor. Despite multimodal treatment including surgical resection with concurrent and adjuvant radiotherapy and chemotherapy, the prognosis of GBM is poor with the median survival time of only 14.6 months and less than 5% of patients with GBM surviving for more than 2 years (1).
GBM manifests high intratumoural heterogeneity with a very small population of cells expressing cancer stem cell (CSC) markers, for example, aldehyde dehydrogenase (ALDH), CD44, CD133, Sox2, Oct4, and Nanog, with epithelial-to-mesenchymal transition (EMT) trend (2, 3). This population of cells has been clarified as glioblastoma stem cells (GSCs), which are quiescent, chemoresistant, and metastatic (4, 5). Studies indicate that GSCs, lodged in hypoxic niches, are in a microenvironment-inducible transient and reversible state rather than a permanent entity (6). As a heterogeneous tumor with an extensively hypoxic environment, GBM provides hypoxic niches for fostering and maintaining GCSs. Hypoxia-inducible factors (HIFs) are the master transcriptional regulators induced by hypoxia (7). NF-κB is a bridge linking inflammation, hypoxia, and cancer (8, 9). NF-κB is an important chemoresistance-related transcription factor upregulating multiple genes related to inflammation, hypoxic response, EMT, and antiapoptotic signaling (10). Many human cancer cells and drug-resistant cancer cell lines possess high constitutive NF-κB activity, which can be further induced by hypoxia (9). Ectopic overexpression of NF-κB has been shown to result in inducing genes related to cancer cell stemness, EMT, and antiapoptosis (11–14). The relationship between HIFs and NF-κB in coordinating the hypoxia-induced GSC characters remains obscure. Understanding the molecular mechanisms behind hypoxia-induced EMT and GSCs may facilitate development of new drugs for GBM therapeutics.
The demand of new drugs for GBM treatment is of clinical urgency, but new drug development is a highly time- and cash-consuming procedure. There is a global trend toward the repositioning of known drugs for new indications (15). Disulfiram, a long-established alcohol-reversal drug, possesses excellent anticancer activity against a wide range of cancers including GBM with low toxicity to normal cells (13, 16–19). Disulfiram also presents a synergistic activity and potentiates the cytotoxicity of conventional chemotherapy drugs and protects normal cells in kidney, gut, and bone marrow in vivo while increasing the therapeutic index (13, 19, 20).
Disulfiram specifically inhibits the activity of ALDH, a functional marker of CSCs with reactive oxygen species (ROS) scavenging activity (21). It effectively eliminates CSCs and reverses chemoresistance (20–22). The cytotoxicity of disulfiram depends on copper (II; Cu) and other divalent transition metal elements (23, 24). Disulfiram-metal complexes block the nuclear translocation of NF-κB by inhibiting proteasome activation and IkBα degradation (17). The functional component of disulfiram is diethyldithiocarbamate (DDC), which strongly chelates Cu to generate ROS and form the end product DDC-Cu. Both ROS and DDC-Cu induce cancer cell apoptosis (23, 24). Therefore, the intact sulfhydryl group is essential for the reaction between disulfiram and Cu. Only oral disulfiram is currently available in clinic. Oral disulfiram is quickly reduced to DDC in the gastrointestinal system. The sulfhydryl groups in DDC are instantly methylated and glucuronidated when disulfiram is enriched in the liver. The methylated and glucuronidated DDC loses its Cu chelating function and cytotoxicity (23, 24). The half-life of disulfiram is less than 1 minute in serum (16). This explains why anticancer activity of disulfiram has been known for more than 30 years without successful and reproducible cancer therapeutic outcomes. We have shown that using a nano-delivery system to protect the sulfhydryl group in disulfiram can improve the anticancer efficacy of disulfiram in breast, liver, and lung cancer models (16, 25).
Here, we demonstrate that NF-κB plays pivotal roles in hypoxia-induced GSC traits. Disulfiram inhibits the NF-κB pathway, targets GSCs, and reverses chemoresistance in vitro. A strong in vivo anti-GBM efficacy was achieved by intravenous administration of a long-half-life circulating poly lactic–co-glycolic acid (PLGA) encapsulated disulfiram (DS-PLGA) in both intracranial and subcutaneous mouse GBM models. Further study may translate DS-PLGA into GBM treatment.
Materials and Methods
Cell lines and reagents
The following human GBM cell lines U87MG ATCC (RRID:CVCL_0022), U251MG (RRID:CVCL_0021), and U373MG ATCC (RRID:CVCL_2219) were purchased from ECACC. The U373MG ATCC cell line is a derivative of the U251MG cell line. All human cell lines have been authenticated using short tandem repeat (STR) profiling within the last 3 years, and all experiments were performed with Mycoplasma-free cells. Disulfiram, temozolomide, copper chloride (CuCl2), copper gluconate (CuGlu), crystal violet, dichloromethane, poly-2-hydroxyethyl methacrylate (poly-HEMA), and BSA were from Sigma-Aldrich. DMEM medium and FCS were supplied by Lonza. Antibodies were purchased from Cell Signaling Technology (Ki67 and BAX), Abcam (NF-κBp65, IκBa, Bax, Bcl2, MMP-2, BMP4, BMP9, Sox2, Oct4, Nanog, E-cadherin, N-cadherin, Vimentin, and ALDH1), BD Biosciences, or Miltenyi Biotec (CD44-FITC and CD133-FITC). The Hypoxyprobe-1 Plus kit was from Hypoxyprobe. ALDEFLUOR was from Stemcell Technologies. Matrigel and cell culture inserts were purchased from Fisher. The DS-PLGA was developed in our laboratory (16).
Normoxic and hypoxic cell culture
The GBM cell lines were cultured under normoxia (20% oxygen) or hypoxia (1% oxygen) and maintained in DMEM supplemented with 10% FCS, 2 mmol/L L-Glutamine, 50 units/mL penicillin, and 50 μg/mL streptomycin. Cells were cultured under normoxia or hypoxia in parallel for 5 days before further experiments. The hypoxic culture was performed in a Panasonic Hypoxia Incubator (Panasonic Biomedical).
In vitro spheroid culture
Cells were cultured in ultralow adherence poly-HEMA–coated T25 flasks at a cell density of 20,000 cells/mL. Neurosphere (NS) was grown in neural stem cell medium [serum-free DMEM-F12 supplemented with B27 and N2 serum replacement (Invitrogen), 0.3% glucose (Sigma-Aldrich), 10 ng/mL epidermal growth factor (Sigma-Aldrich), 10 ng/mL basic fibroblasts growth factor (R&D Systems), 20 mg/mL insulin (Sigma-Aldrich), 2 mg/mL heparin (Sigma-Aldrich)]. The group of suspension (SUS) culture was performed in normal DMEM supplemented with 10% serum. The cells were incubated at 37°C for 7 days, before further analysis, with necessary media replenishment every 3 days.
Detection of hypoxia in cell culture
The Hypoxyprobe Kit was used following the manufacturer's protocol. For immunocytochemistry assay, the attached cells (ATTs) were cultured at normoxic or hypoxic condition for 72 hours and cocultured with Hypoxyprobe reagent overnight. The NS and SUS cells were cocultured with Hypoxyprobe overnight and cytospined at 800 rpm for 3 minutes to spread the spheres onto Polylysine-coated slides (VWR). The cells were fixed and stained with FITC-conjugated anti-hypoxyprobe MAbs and imaged using a confocal microscope. For quantification of hypoxic cells, the cells were stained with Hypoxyprobe and subjected to flow cytometric analysis. The hypoxic population was detected using a FACS Calibur flow cytometer with a 488-nm blue laser and a standard FITC 530/30-nm bandpass filter.
Flow cytometric analysis of CSC markers
The trypsinized cells (2.5 × 105) were analyzed with corresponding CSC markers. ALDH activity was measured by staining with ALDH substrate. The CD133 and CD44 cell surface marker were detected by FITC-conjugated anti-CD133 and anti-CD44 antibodies. The cells were examined using BD FACS Calibur and Cell Quest Pro software.
In vitro cytotoxicity assay
The methodology of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay for the normoxia cultured cells has been described previously (22). To determine the effect of hypoxia on drug sensitivity, the cells were cultured in 1% oxygen at a cell density of 5 × 103 cells/well in 96-well plate for 4 days and exposed to anticancer drugs for another 72 hours before MTT assay with a parallel MTT assay in normoxia. The NS and SUS spheres were trypsinised and recultured in poly-HEMA–coated 96-well flat bottom plates in their corresponding culture medium at a cell density of 5 × 103 cells/well overnight. After exposure to drugs for 72 hours, the cells were subjected to a standard MTT assay. The NS and SUS cells were centrifuged at 800 rpm for 5 minutes before MTT assay.
Western blot analysis of proteins
The whole and nuclear protein was extracted from normoxia, hypoxia, NS, and SUS cultured GBM cells with RIPA buffer, quantified and separated by SDS PAGE. The proteins were blotted onto polyvinylidene difluoride (PVDF) membrane (Thermo Fisher Scientific), blocked with 5% nonfat milk, and incubated with appropriate primary and secondary antibodies of target proteins.
Stable transfection of U373MG cell line with NF-κBp65, HIF1α, and HIF2α
U373MG cells (2.5 × 105/well) were cultured in 35-mm dishes until 70% confluent, and Lipofectamine 2000 Superfect (Qiagen) was used to transfect pcDNA3.1(+; Invitrogen), pcDNA3.1/Hygro/NF-κBp65, pCMV6-Neo/HIF1α, pCMV6-Neo/HIF2α (Origene). The successfully transfected clones were selected in relevant antibiotics.
In vitro migration and invasion assay
The migration and invasion assays were performed using cell culture inserts (Thermo Fisher Scientific) coated with or without Matrigel (BD Biosciences). Cells (5 × 104) were resuspended in 200 μL of serum-free DMEM and placed in the upper chamber of the insert. The lower chamber was filled with medium containing 10% FCS. After 16 hours incubation at 37°C, migrated cells were fixed by methanol and stained in 0.5% crystal violet. Cells were imaged and counted under a microscope.
qRT-PCR was performed using Taqman assays. RNA was extracted using an RNA purification kit (Norgen) following the manufacturer's instructions and reverse transcribed into cDNA (Multiscribe Reverse Transcriptase, Applied Biosystems). A QuantStudio 6 Flex Real-Time PCR system (Applied Biosystems) was used for qPCR experiments. Gene expression levels were normalized to control genes (GAPDH and HPRT1) and calculated according to the ΔΔCT method.
Luciferase reporter gene assay
The cells (1 × 104/well) were cultured in 96-well plates overnight. The NF-κB (pNF-κBTal-Luc, BD Biosciences) and hypoxia response element (HRE) (pGL4.42-luc2P/HRE, Promega) luciferase reporter vectors and pGL3-Basic (Promega) were co-transfected with pSV40-Renilla (Promega) DNA. Forty-eight hours after transfection, the luciferase activity was determined using Dual Luciferase Assay reagents (Promega) according to the manufacturer's instructions. The relative luciferase activity was calculated (25).
IHC and hematoxylin and eosin staining
The sections from paraffin-embedded tumor and normal tissues were stained with primary antibodies [Ki67, Bax, NF-κBp65, and ALDH1 (1:200)] then biotinylated secondary antibody and followed by incubation in ABC reagent (DAKO Labs). For hematoxylin and eosin (H&E) staining, the paraffin-embedded sample slides were stained with H&E, and the slides were mounted with coverslips using Permount (Thermo Fisher Scientific).
Terminal deoxyribonucleotidyl transferase–mediated dUTP nick- end labeling assay
Tumor tissues were paraffin-embedded and stained according to the manufacturer's instructions (Roche). Briefly, the slides were incubated with TdT Enzyme, Stop/Wash Buffer, antidigoxigenenin, and then stained with peroxidase substrate and incubated in ABC reagent. Finally, the slide was mounted with 3,3′-diaminobenzidine and visualized under a light microscope.
Pharmacokinetic study and in vivo anti-GBM xenograft experiments
The animal experiments were reviewed and approved by the Ethical Committee of Third Military Medical University (Chongqing, China) and University of Wolverhampton (Wolverhampton, UK). Rats (five per group) were injected with disulfiram or DS-PLGA (250 mg/kg) through a tail vein. Blood samples were collected at 0.25, 0.5, 1, 2, 4, and 8 minutes for the disulfiram group and at 5, 10, 20, 40, 60, 120, and 180 minutes for the DS-PLGA group. The plasma (500 μL) was separated and mixed with methanol (250 μL). After vigorously shaking and centrifuging (10,000 rpm, 15 minutes), the disulfiram concentration in the supernatant was determined by high-performance liquid chromatography (HPLC) as described elsewhere (16). The intracranial GBM model was developed as previously described (26). Briefly, the 5-week-old female BALB/c Nu/Nu mice were anesthetized. The U87MG-luciferase-GFP cells (2 × 105/5 μL PBS) were injected into the brain using a 25-μL microsyringe. After 10 days, the animals were treated with empty PLGA or DS-PLGA (10 mg/kg i.v.) plus CuGlu (6 mg/kg orally) three times per week for 4 weeks. For the subcutaneous GBM model, U87MG cells (5 × 106) were injected into the front flank of BABL/c Nu/Nu mice. When the tumor volume reached approximately 200 mm3, the tumor-bearing mice were subjected to the above treatment. The xenograft size was recorded twice per week. The tumor volume was calculated by the following formula: V = (L × W2) × 0.5, where L is the length and W is the width of the tumor. After 4 weeks, the animals were sacrificed. The tumors were removed, photographed, and subjected to further analysis.
Student t test and one-way ANOVA were used in this study. P < 0.05 and <0.01 were considered statistically significant and highly significant, respectively.
Data availability statement
The data generated in this study are available within the article and its supplementary data files.
Hypoxia, stemness, NF-κB activation, and temozolomide resistance were detected in both NS and SUS cultured GBM cells
Characteristic spheres were formed in both serum-free NS and serum-rich SUS culturing conditions (Fig. 1A). All the sphere cells expressed high stem cell (Fig. 1B) and embryonic stem cell markers (Fig. 1C). Overexpression of N-cadherin, Vimentin, and downregulation of E-cadherin were detected in NS and SUS cells indicating the presence of EMT in the sphere cells (Fig. 1D). Significant temozolomide resistance was detected when the sphere cells were exposed to temozolomide for 72 hours and subjected to sphere reformation and MTT cytotoxicity assay (Fig. 1A; Supplementary Tables S1 and S2).
Hypoxyprobe staining showed hypoxic population in the core region of both NS and SUS (Fig. 1E). A high proportion of hypoxic cells were detected in NS and SUS cells using flow cytometry (Fig. 1F). Further evidence of the hypoxic condition in NS and SUS cells was shown by the overexpression of HIF1α and HIF2α proteins (Fig. 1G). Overexpression of NF-κBp65 was detected in the sphere cells (Fig. 1H).
Hypoxia induces stemness, NF-κB activation, EMT, and temozolomide resistance in GBM cells
To confirm the link between hypoxia and GSC phenotypes, we cultured monolayer GBM cells under hypoxia. The cellular hypoxia was confirmed by Hypoxyprobe staining (Fig. 2A and B) and overexpression of HIF1α and HIF2α proteins (Fig. 2C). The hypoxia culture induces high levels of GSC and embryonic stem cell markers (Fig. 2D and E). The hypoxic GBM cells are significantly resistant to temozolomide (Fig. 2F). The overexpression of Bcl2 and downregulation of Bax indicate that hypoxia activates antiapoptotic pathways (Fig. 2G). Hypoxia-induced EMT was confirmed by the overexpression of CD44, Vimentin, N-cadherin, downregulation of E-cadherin (Fig. 2D and H), and overexpression of mRNA from EMT- and metastasis-related genes (Supplementary Fig. S1). Increased migration and invasion abilities were manifested in the hypoxia-cultured GBM cells (Fig. 2I and J; Supplementary Table S3). Together, the above findings indicate that hypoxia induces stemness, EMT, and chemoresistance in GBM cells. Similar to spheroid cultures, the high expression of NF-κBp65 protein and its nuclear translocation were also detected in the hypoxia-cultured cells (Fig. 2K).
The regulatory relationship of NF-κB, HIF1α, and HIF2α in GBM cells
HIFs and NF-κB are hypoxia-induced genes responsible for EMT and CSC traits (14, 27). The regulatory relationship between HIFs and NF-κB is still not clear. To establish which one of these three factors is on the central regulatory position in hypoxia response, we stably transfected U373MG cells with NF-κBp65, HIF1α, and HIF2α. The successful transfected clones were confirmed by Western blot and luciferase reporter gene assay (Fig. 3A–F). The NF-κBp65–transfected clones express high levels of HIF1α and HIF2α proteins (Fig. 3G) and mRNAs (Supplementary Fig. S2) and possess high hypoxia response element (HRE) transcriptional activity (Fig. 3H). In contrast, HIF1α and HIF2α transfected clones did not show significant NF-κB upregulation at mRNA (Supplementary Fig. S2), protein (Fig. 3I and K), and transcription (Fig. 3J and L) levels. The expression of HIF2α mRNA was upregulated by HIF1α transfection but not vice versa (Supplementary Fig. S2). Therefore, NF-κB may be located at a higher hierarchical position than HIF1α and HIF2α in hypoxia response gene network.
The effect of NF-κB and HIFs on stemness, EMT, temozolomide resistance and invasiveness in GBM cells
Next, we examined the effect of NF-κB and HIFs on EMT and stemness in GBM cells. The NFκBp65, HIF1α, and HIF2α transfected clones expressed higher levels of GSC (Fig. 4A) and embryonic stem cell (Fig. 4B) markers. The mRNA and protein of EMT-related genes were detected in the transfected clones (Fig. 4C and D). The NF-κBp65 and HIF2α but not HIF1α transfected clones are significantly resistant to temozolomide (Fig. 4E). We also examined the migration and invasion abilities, one of the most common features of EMT (28), in the transfected cell lines. All of the NF-κBp65, HIF1α, and HIF2α transfected cell lines possess significantly higher migration and invasion activity (Fig. 4F–H; Supplementary Table S4). Some metastasis-related genes are overexpressed in these cell lines (Fig. 4D). These data indicate that all of these three genes promote the invasiveness of GBM, but only NF-κB and HIF2α are responsible for temozolomide resistance.
Disulfiram inhibits hypoxia-induced GSC and EMT phenotypes in vitro
Hypoxic cells remain sensitive to disulfiram (Fig. 5A), although they are significantly resistant to temozolomide. Disulfiram and Cu block the sphere reformation in NS and SUS cultures. The disulfiram or Cu alone had no effect on sphere reformation indicating the essential requirement of Cu in the function of disulfiram against GSCs (Fig. 5B and C; Supplementary Table S2). The GSC markers induced by spheroid and hypoxia culture were blocked by disulfiram/Cu (Fig. 5D and E; Supplementary Figs. S3 and S4). Disulfiram/Cu inhibited the migration and invasion of GBM cell lines at an extremely low concentration (25 nmol/L; Fig. 5F and G; Supplementary Table S3). NF-κB is a key player in GSC and EMT. The expression of NF-κBp65 protein in hypoxia and sphere cultured GBM cells was inhibited by disulfiram/Cu (Fig. 5H). Disulfiram/Cu treatment enhanced and inhibited the expression of Bax and Bcl2, respectively (Fig. 5H).
DS-PLGA extends the half-life of disulfiram in vivo and suppresses GBM in orthotopic and subcutaneous GBM mouse models
To overcome the very short half-life of disulfiram in the bloodstream (16, 29), we used a DS-PLGA formulation (16) to deliver disulfiram intravenously in a mouse model. The half-life of DS-PLGA in rat was 13.4 minutes (Supplementary Table S5) while the free disulfiram was immediately undetectable after injection.
The in vivo efficacy of DS-PLGA was assessed in both intracranial and subcutaneous models. After 4 weeks of treatment, DS-PLGA/Cu significantly reduced the sizes of the intracranial and subcutaneous tumors and the tumor weight in subcutaneous injected mice (Fig. 6A–C). These results indicate that the PLGA encapsulation delivers the intact disulfiram to GBM tissues. The TUNEL staining indicated apoptosis induced by DS-PLGA/Cu in vivo (Fig. 6D). DS-PLGA/Cu treatment inhibited the expression of NF-κBp65, ALDH, and Ki67 expression in the GBM xenografts (Fig. 6E). No toxic effect on vital organs was observed and mouse body weight was maintained (Fig. 6F and G).
Investigation and unveiling of molecular mechanisms of GSC-induced chemoresistance and invasiveness (4, 5) will likely facilitate the development of novel and efficacious therapies for patients with GBM. Since Brent A. Reynolds and Samuel Weiss successfully isolated normal neural stem cells using sphere-forming culture in growth factor-supplemented serum-free medium (30), this system has been widely adapted for in vitro culture of GSCs. It has been generally believed that serum-free and growth factor-rich medium is essential for maintenance of anchorage-independent proliferation and stemness. Our study manifested that GBM cells form NSs in both serum-rich and serum-free media, which express comparable levels of GSC markers (Fig. 1B and C) and are significantly resistant to temozolomide (Fig. 1A; Supplementary Table S1). Similar findings were reported previously (25, 31, 32). EMT and CSCs are inextricably linked microenvironment-determined transient epigenetic phenomena (33), which were detected in NS and SUS cells (Fig. 1B and D). Hypoxia is a central determinant of the stem cell niche (34), which is relevant to GBM, as these tumors have significantly lower intratumoral oxygen concentration than normal brain tissues (35). In line with our previous findings in breast cancer (25), the core regions of NSs contain a high proportion of hypoxic cells (Fig. 1E and F), which express high levels of HIF1α, HIF2α, and NF-κBp65 (Fig. 1G and H), the key factors induced by inflammation and hypoxia (7, 8). The GBM cells cultured in hypoxia produce epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF); ref. 36. Our results suggest that the GSC traits may be determined by the hypoxic microstructure of the spheroid culture rather than the medium, and the selective medium may not be essential for the in vitro GSC culture. In contrast, the conventional serum-containing medium may better mimic the in vivo pathophysiologic conditions and may be more suitable for in vitro GSC study.
The interconversion of CSCs and non-CSCs is a common phenomenon driven by environmental stimuli (2, 3). As such, the GSC and non-GSC cells are mutable cellular traits determined by oxygen levels in the microenvironment (34). Hypoxia-cultured cells grown as a monolayer expressed high levels of GSC and embryonic stem cell markers and were markedly resistant to temozolomide (Fig. 2D–F). A high proportion of both sphere- and hypoxia-cultured cells overexpressed CD44, N-cadherin, Vimentin, and some key EMT markers, whereas E-cadherin was downregulated (Figs. 1B, D, and 2H; Supplementary Fig. S1). These results indicate hypoxia induces EMT in these cells. EMT is responsible for cancer cell migration, invasion, and metastasis (28). The hypoxia-cultured cells showed higher migration and invasion abilities (Fig. 2I and J; Supplementary Table S3).
Overexpression of HIF1α, HIF2α, and NF-κBp65 was detected in both sphere- and hypoxia-cultured GBM cells (Figs. 1G and H, 2C and K). These transcription factors are related to dedifferentiation and maintenance of GSCs (37). The stably transfected U373MG models were used to elucidate the hierarchical relationship among these three genes. NF-κBp65 transfection induced HIF1α and HIF2α protein, mRNA expression and HRE activity, but not vice versa (Fig. 3G–M; Supplementary Fig. S2). NF-κB-, HIF1α-, and HIF2α-transfected cells express some key EMT transcription factors (Fig. 4C and D). NF-κB may manipulate the hypoxia-induced activation of HIF1α and HIF2α via binding to the κB sites on their promoter regions (38). HIF1α-transfected clones overexpressed HIF2α but not vice versa (Supplementary Fig. S2). Although all of the HIF1α-, HIF2α-, and NF-κBp65–transfected clones overexpress GSC and EMT markers, only NF-κBp65 and HIF2α-transfected cells are resistant to temozolomide (Fig. 4E). This is consistent with previous reports that HIF1α mainly regulates acute hypoxia-induced cell responses. In contrast, HIF2α is responsible for chronic hypoxia, which induces GSC and EMT traits (39, 40). HIF1α may counterbalance the drug resistant effect of HIF2α (41). All of these three genes induce the expression of key EMT- and metastasis-related genes and promote GBM cell migration and invasion (Fig. 4C, D, F–H; Supplementary Table S4).
In line with our previous studies (16, 22, 24, 25), disulfiram targets NF-κB and reverses hypoxia-induced cancer metastasis/chemoresistance in a Cu-dependent manner (Fig. 5A–C, F, and G). Disulfiram/Cu abolished GSC markers and completely blocked NS reformation ability (Fig. 5B–E). Intracranial metastasis is the major cause of GBM recurrence (42, 43). Disulfiram blocked GBM cell migration and invasion at an extremely low concentration (25 nmol/L). In combination with radiotherapy, disulfiram may effectively improve the current therapeutic outcomes of GBM.
The very short half-life of disulfiram in the bloodstream is the key limit for its efficacy as cancer therapeutics. The DS-PLGA has a significantly longer half-life (Supplementary Table S5; ref. 16) and can successfully deliver intact disulfiram to cancer tissues. The low molecular weight (296) and high lipophilicity facilitate disulfiram to penetrate GBM tissues. The DS-PLGA/Cu showed very strong in vivo anti-GBM efficacy in both intracranial and subcutaneous mouse models. It inhibited NF-κB and ALDH expression in GBM tissues. No systemic toxic effect of DS-PLGA/Cu on vital organs (kidney, liver, lung, spleen, and brain) and no mouse body weight loss was observed. The intracranial metastasis is the major cause of postoperative relapse of GBM, although local drug delivery using 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU)/wafer marginally improved the survival time, the vulnerability of brain tissue to BCNU hinders its success in GBM treatment (44). Disulfiram showed selective toxicity in cancer cells (25). The migration/invasion inhibiting concentration (25 nmol/L) is significantly lower than its toxic concentration. DS-PLGA may potentially be locally delivered using convection-enhanced delivery (45) or wafer formulation in GBM treatment. Nevertheless, our results suggest that repurposing disulfiram through a novel type of delivery system that prolongs its plasma half-life offers new possible therapeutic options for the treatment of GBM.
V. Kannappan reports personal fees from Disulfican Ltd. during the conduct of the study and personal fees from Disulfican Ltd. outside the submitted work. W. Wang reports grants from Innovate UK, European Regional Developmetal Fund, and Smart Concept Fund during the conduct of the study; nonfinancial support from Disulfican Ltd. outside the submitted work; and has a patent for US10695299B2 issued. No disclosures were reported by the other authors.
V. Kannappan: Data curation, formal analysis, investigation, methodology, writing–original draft. Y. Liu: Data curation, formal analysis, validation, investigation, methodology. Z. Wang: Data curation, formal analysis, validation, investigation, methodology. K. Azar: Data curation, formal analysis, validation, investigation, methodology. S. Kurusamy: Data curation, formal analysis, validation, investigation. R.S. Kilari: Investigation, methodology. A.L. Armesilla: Supervision. M.R. Morris: Supervision, writing–review and editing. M. Najlah: Investigation, methodology. P. Liu: Investigation, methodology. X.-W. Bian: Conceptualization, resources, supervision, investigation, project administration. W. Wang: Conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration.
We would like to express our great thanks to Prof. Andrew Pollard (Chairman/Director, Disulfican Ltd.) for his significant contribution to Innovate UK- Jiangsu grant application and project management. We are also immensely grateful to Prof. Patrick Ball and Dr. Hana Morrissey for careful reading and very constructive comments on an earlier version of this manuscript.
This study was supported by Research Institute in Healthcare Science/University of Wolverhampton PhD studentship and Innovate UK Jiangsu- UK Industrial Challenge Programme (104022). Some key equipment was jointly funded by European Regional Developmental Fund, Smart Concept Fund, University of Wolverhampton, and Disulfican Ltd.
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.